Ecological Speciation

Home , Astragalus whitneyi, Ecological facilitation, Ecological selection, Ecological speciation, Exema canadensis, Nonecological speciation

International Journal of Ecology

Guest Editors: Marianne Elias, Rui Faria, Zachariah Gompert, and Andrew Hendry Ecological Speciation International Journal of Ecology

Ecological Speciation

This is a special issue published in “International Journal of Ecology.” All articles are open access articles distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Editorial Board

Mariana Amato, Italy Jean-Guy Godin, Canada Panos V. Petrakis, Greece Madhur Anand, Canada David Goldstein, USA Daniel I. Rubenstein, USA Joseph R. Bidwell, USA Shibu Jose, USA Herman H. Shugart, USA L. M. Chu, Hong Kong Chandra Prakash Kala, India Andrew Sih, USA Jean Clobert, France Pavlos Kassomenos, Greece R.C. Sihag, India Michel Couderchet, France Thomas H. Kunz, USA C. ter Braak, The Netherlands Ronald D. Delaune, USA Bruce D. Leopold, USA John Whitaker, USA Andrew Denham, Australia A. E. Lugo, USA Walter Whitford, USA Mark A. Elgar, Australia Patricia Mosto, USA J. J. Wiens, USA Jingyun Fang, China Mats Olsson, Australia Xiaozhang Yu, China Contents

Factors Inﬂuencing Progress toward Ecological Speciation, Marianne Elias, Rui Faria, Zachariah Gompert, and Andrew Hendry Volume 2012, Article ID 235010, 7 pages

The Role of Parasitism in Adaptive RadiationsWhen Might Parasites Promote and When Might They Constrain Ecological Speciation?, Anssi Karvonen and Ole Seehausen Volume 2012, Article ID 280169, 20 pages

Parallel Ecological Speciation in Plants?, Katherine L. Ostevik, Brook T. Moyers, Gregory L. Owens, and Loren H. Rieseberg Volume 2012, Article ID 939862, 17 pages

From Local Adaptation to Speciation: Specialization and Reinforcement, Thomas Lenormand Volume 2012, Article ID 508458, 11 pages

Testing the Role of Habitat Isolation among Ecologically Divergent Gall Wasp Populations, Scott P. Egan, Glen R. Hood, and James R. Ott Volume 2012, Article ID 809897, 8 pages

Underappreciated Consequences of Phenotypic Plasticity for Ecological Speciation, Benjamin M. Fitzpatrick Volume 2012, Article ID 256017, 12 pages

Ecological Adaptation and Speciation: The Evolutionary Signiﬁcance of Habitat Avoidance as a Postzygotic Reproductive Barrier to Gene Flow,Jeﬀrey L. Feder, Scott P. Egan, and Andrew A. Forbes Volume 2012, Article ID 456374, 15 pages

Larval Performance in the Context of Ecological Diversiﬁcation and Speciation in Lycaeides Butterﬂies, Cynthia F. Scholl, Chris C. Nice, James A. Fordyce, Zachariah Gompert, and Matthew L. Forister Volume 2012, Article ID 242154, 13 pages

Divergent Selection and Then What Not: The Conundrum of Missing Reproductive Isolation in Misty Lake and Stream Stickleback,KatjaRas¨ anen,¨ Matthieu Delcourt, Lauren J. Chapman, and Andrew P. Hendry Volume 2012, Article ID 902438, 14 pages

How Facilitation May Interfere with Ecological Speciation, P. Liancourt, P. Choler, N. Gross, X. Thibert-Plante, and K. Tielborger¨ Volume 2012, Article ID 725487, 11 pages

Use of Host-Plant Trait Space by Phytophagous Insects during Host-Associated Diﬀerentiation: The Gape-and-Pinch Model,StephenB.Heard Volume 2012, Article ID 192345, 15 pages

Habitat Choice and Speciation, Sophie E. Webster, Juan Galindo, John W. Grahame, and Roger K. Butlin Volume 2012, Article ID 154686, 12 pages The Role of Environmental Heterogeneity in Maintaining Reproductive Isolation between Hybridizing Passerina (Aves: Cardinalidae) Buntings, Matthew D. Carling and Henri A. Thomassen Volume 2012, Article ID 295463, 11 pages

Pollinator-Driven Speciation in Sexually Deceptive Orchids, Shuqing Xu, Philipp M. Schluter,¨ and Florian P. Schiestl Volume 2012, Article ID 285081, 9 pages

Synergy between Allopatry and Ecology in Population Diﬀerentiation and Speciation, Yann Surget-Groba, Helena Johansson, and Roger S. Thorpe Volume 2012, Article ID 273413, 10 pages

Of “Host Forms” and Host Races: Terminological Issues in Ecological Speciation, Daniel J. Funk Volume 2012, Article ID 506957, 8 pages

Learning the Hard Way: Imprinting Can Enhance Enforced Shifts in Habitat Choice, Niclas Vallin and Anna Qvarnstrom¨ Volume 2011, Article ID 287532, 7 pages

Sympatric Speciation in Threespine Stickleback: Why Not?, Daniel I. Bolnick Volume 2011, Article ID 942847, 15 pages

Ecological Divergence and the Origins of Intrinsic Postmating Isolation with Gene Flow, Aneil F. Agrawal, JeﬀreyL.Feder,andPatrikNosil Volume 2011, Article ID 435357, 15 pages Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 235010, 7 pages doi:10.1155/2012/235010

Editorial Factors Inﬂuencing Progress toward Ecological Speciation

Marianne Elias,1 Rui Faria,2, 3 Zachariah Gompert,4 and Andrew Hendry5

1 CNRS, UMR 7205, Muséum National d’Histoire Naturelle, 45 Rue Buffon, CP50, 75005 Paris, France 2 CIBIO/UP—Centro de Investigaçao˜ em Biodiversidade e Recursos Genéticos, Universidade do Porto, Campus Agrario´ de Vairao,˜ R. Monte-Crasto, 4485-661 Vairao,˜ Portugal 3 IBE—Institut de Biologia Evolutiva (UPF-CSIC), Universitat Pompeu Fabra, PRBB, Avenue Doctor Aiguader N88, 08003 Barcelona, Spain 4 Department of Botany, 3165, University of Wyoming, 1000 East University Avenue Laramie, WY 82071, USA 5 Redpath Museum and Department of Biology, McGill University, 859 Sherbrooke Street West Montreal, QC, Canada H3A 2K6

Correspondence should be addressed to Marianne Elias, [email protected]

Received 28 March 2012; Accepted 28 March 2012

Copyright © 2012 Marianne Elias et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction why there are so many species [13] and also why there are so few species [14]. Ecological speciation occurs when adaptation to divergent This special issue on ecological speciation puts snapshots environments, such as different resources or habitats, leads of progress toward speciation sharply in focus and then to the evolution of reproductive isolation [1, 2]. More specif- investigates this topic from several angles. First, several ically, divergent (or disruptive) selection between environ- papers provide conceptual or theoretical models for how to ments causes the adaptive divergence of populations, which consider progress toward ecological speciation (Funk; Heard; leads to the evolution of reproductive barriers that decrease, Lenormand; Liancourt et al.; Agrawal et al.). Second, several and ultimately cease, gene flow [3, 4]. Supported by a papers highlight the noninevitability of ecological speciation growing number of specific examples, ecological speciation is through investigations where ecological speciation seems to thought to be a primary driving force in evolutionary diver- be strongly constrained (Ras¨ anen¨ et al.; Bolnick) or at least sification, exemplified most obviously in adaptive radiations lacking definitive evidence (Ostevik et al.; Scholl et al.). Some [5–8]. of these papers also uncover specific factors that seem par- As acceptance of the importance of ecological speciation ticularly important to ecological speciation, such as the has grown, so too has the recognition that it is not all power- combination of geographic isolation and habitat differences ful. Specifically, a number of instances of nonecological (Surget-Groba et al.), the strength of disruptive selection and speciation and nonadaptive radiation seem likely [9], and assortative mating (Bolnick), and host-plant adaptation colonization of different environments does not always lead (Scholl et al.). Third, several particularly important factors to speciation [10, 11]. This latter point is obvious when emerge as a common theme across multiple papers, par- one recognizes that although essentially all species are com- ticularly parasites/pollinators (Xu et al.; Karvonen and See- posed of a number of populations occupying divergent envi- hausen), habitat choice (Webster et al.; Feder et al.; Carling ronments [12], only a fraction of these ever spin off to be- and Thomassen; Egan et al.), and phenotypic plasticity come full-fledged species. Instead, populations occupying (Fitzpatrick; Vallin and Qvarnstrom).¨ divergent environments or using different resources show Here we highlight the most important aspects of these varying levels of progress toward ecological speciation—and contributions and how they relate to three major topics: (i) this variation provides the substrate to study factors that models for progress toward ecological speciation; (ii) vari- promote and constrain progress along the speciation contin- able progress toward ecological speciation in nature; and (iii) uum. By studying these factors, we can begin to understand factors affecting progress toward ecological speciation. 2 International Journal of Ecology

2. Models for Progress toward a theoretical and statistical framework for testing this model Ecological Speciation and applies it to insects using goldenrod plants. Local adaptation is often the first step in ecological spe- Terminological issues have long bedevilled communication ciation, and so factors influencing local adaptation will be among researchers working on speciation. D. J. Funk ad- critical for ecological speciation. Local adaptation can either dresses this topic by first clarifying the relationship between increase over time (if more specialized alleles spread), even- sympatric speciation (whereby reproductively isolated pop- tually leading to speciation, or it can decrease over time (if ulations evolve from an initially panmictic population) and more generalist alleles spread). T. Lenormand reviews the ff ecological speciation (whereby reproductive isolation evolves conditions that favor these di erent scenarios and empha- as a consequence of divergent/disruptive natural selection). sizes the role of three positive feedback loops that favor These are orthogonal concepts [15]. First, even if disruptive increased specialization. In the demographic loop, local selection is a common way of achieving sympatric speciation, adaptation results in higher population density, which in this can also be caused by other factors, such as changes in turn favors the recruitment of new locally adapted alleles. chromosome number. Second, ecological speciation can In the recombination loop, locally adapted alleles are more readily occur in allopatry [16, 17]. Funk then introduces four likely to be recruited in genomic regions already harbouring new concepts aiming to reduce confusion in the literature. loci with locally adapted alleles, thereby generating genomic regions of particular importance to local adaptation. In the Sympatric race is a generalisation of host race (usually used reinforcement loop, local adaptation selects for traits that for herbivores or parasites) and refers to any sympatric popu- promote premating isolation (reinforcement), which in turn lations that experience divergent selection and are partly but increases the recruitment and frequency of locally adapted incompletely reproductively isolated. Envirotypes are popu- alleles. Lenormand then details the mechanisms involved lations that differ due to phenotypic plasticity. Host forms in reinforcement, particularly assortative mating, dispersal, are populations that exhibit host-associated variation, but for and recombination. He highlights that these characteristics which the nature of variation (e.g., envirotype, host race, represent the three fundamental steps in a sexual life cycle cryptic species) has not yet been diagnosed. Ecological forms (syngamy, dispersal, and meiosis) and that they promote are a generalization of host forms for nonherbivore or para- genetic clustering at several levels (within locus, among in- sitic taxa. The two latter concepts acknowledge the fact that dividuals, among loci). His new classification is orthogonal one has an incomplete understanding of speciation. To to, and complements, the traditional one- versus two-allele overcome the problem of overdiagnosing host races, Funk distinction [14]. Overall, the rates of increased specialization introduces five criteria, based on host association and choice, and reinforcement determine progress toward ecological coexistence pattern, genetic differentiation, mate choice, speciation. gene flow, and hybrid unfitness. Funk’s maple and willow One of the major constraints on ecological speciation associated phytophagous populations of Neochlamisus beb- is the establishment of self-sustaining populations in new/ bianae leaf beetle meet all these criteria and can, therefore, marginal environments, because the colonizing individuals be considered as host races. are presumably poorly adapted to the new conditions. This Another phytophagy-inspired conceptual model for how difficulty might be eased through facilitation, the ameliora- an insect species initially using one plant species might diver- tion of habitat conditions by the presence of neighbouring sify into multiple insect species using different host plants is living organisms (biotic components) [18]. According to this presented by S. Heard. This effort explicitly links variation in process, the benefactor’s “environmental bubble” facilitates host plant use within insect species or races to the formation the beneficiary’s adaptation to marginal conditions, which of different host races and species. In this proposed “gape- can result in ecological speciation if gene flow from the core and-pinch” model, Heard posits four stages (or “hypothe- habitat is further reduced. At the same time, however, faci- ses”) of diversification defined in part by overlap in the plant litation might hinder further progress toward ecological spe- trait space used by the insect races/species. In the first stage ciation by maintaining gene flow between environments and “adjacent errors,” some individuals within an insect species by preventing reinforcement in secondary contact zones. P. using one plant species might “mistakenly” use individuals Liancourt, P. Choler, N. Gross, X. Thibert-Plante, and K. of another plant species that have similar trait values to their Tielborg¨ consider these possibilities from the beneficiary spe- normal host plant species. In the next stage “adjacent oligo- cies perspective, through a spatially and genetically explicit phagy,” populations formed by the insects that shifted plant modelling framework that builds on earlier models [19, 20]. species then experience divergent selection—and undergo They find that ecological speciation is more likely with adaptive divergence—leading to a better use of that new host. larger patch (facilitated versus harsh) sizes. Liancourt and In the third stage “trait distance-divergence,” competition coauthors further suggest that facilitation can play another and reproductive interactions cause character displacement important role in evolution by helping to maintain a genetic between the emerging insect races or species so that they diversity “storage” in marginal habitats, a process with some become specialized on particularly divergent subsets of the parallel to niche conservationism. A deeper understanding of trait distributions of the two plant species. In the final stage the role of facilitation in diversification is needed (both theo- “distance relaxation,” the new species become so divergent retically and empirically), and the authors suggest that stress- that they no longer interact, and can then evolve to use trait ful environmental gradients would be useful study systems values more typical of each plant species. Heard provides for this endeavour. International Journal of Ecology 3

Intrinsic postzygotic isolation, a fundamental contribu- of selection and assortative mating measured in lake popula- tor to speciation, is often caused by between-locus genetic tions in nature are too weak to cause sympatric speciation. incompatibilities [21, 22]. The origin of these incompat- Instead, lake stickleback appears to respond to disruptive ibilities, particularly in the face of gene flow, remains an selection through alternative means of reducing competition, outstanding question. A. F. Agrawal, J. L. Feder, and P. Nosil such as increased genetic variance, sexual dimorphism, and use two-locus two-population mathematical models to ex- phenotypic plasticity. plore scenarios where loci subject to divergent selection also Another classic system for studying ecological speciation, affect intrinsic isolation, either directly or via linkage dis- or more generally adaptive radiation, is Anolis lizards of the equilibrium with other loci. They quantified genetic differ- Caribbean. In particular, many of the larger islands contain entiation (allelic frequencies of loci under selection), the repeated radiations of similar “ecomorph” species in similar extent of intrinsic isolation (hybrid fitness), and the overall habitats [8]. Contrasting with this predictable and repeatable barrier to gene flow (based on neutral loci). They find that diversity on large islands, smaller islands contain only a few divergent selection can overcome gene flow and favors the species. Y. Surget-Groba, H. Johansson, and R. S. Thorpe evolution of intrinsic isolation, as suggested previously [23]. studied populations of Anolis roquet from Martinique. This Counterintuitively, intrinsic isolation can sometimes weaken species contains populations with divergent mitochondrial the barrier to gene flow, depending on the degree of linkage lineages, a consequence of previous allopatric episodes, and between the two focal loci. This occurs because intrinsic is distributed over a range of habitats. It can, therefore, be isolation sometimes prevents differentiation by divergent used to address the relative importance of past allopatry, selection. present ecological differences, and their combination in determining progress toward ecological speciation. Using microsatellite markers, the authors find that geographic isolation alone does not result in significant population dif- 3. Variable Progress toward ferentiation, habitat differences alone cause some differenti- Ecological Speciation in Nature ation, and geographic isolation plus habitat differences cause the strongest differentiation. The authors conclude that Threespine stickleback fish, with their diverse populations speciation is likely initiated in allopatry but is then completed adapted to different habitats, had provided a number of following secondary contact only through the action of adap- examples of how adaptive divergence can promote ecological tation to different habitats. speciation [24, 25]. Indeed, work on this group has funda- Even though ecological differences are clearly important mentally shaped our modern understanding of ecological in the diversification of both plants and animals [1, 30], it speciation [1–3]. At the same time, however, three-spine remains uncertain as to whether the process is fundamentally stickleback also provides evidence of the frequent failure of the same or different between them. Part of the reason is divergent selection to drive substantial progress toward eco- that typical methods for studying ecological speciation differ logical speciation [25]. This special issue provides two of between the two groups. In an effort to bridge this method- such examples. In one, K. Ras¨ anen,M.Delcourt,L.J.¨ ological divide, K. Ostevik, B. T. Moyers, G. L. Owens, and Chapman and A. P. Hendry report that, despite strong diver- L. H. Rieseberg apply a common method of inference from gent selection, lake and stream stickleback from the Misty animals to published studies on plants. In particular, ecolog- watershed do not exhibit positive assortative mate choice in ical speciation is often inferred in animals based on evidence laboratory experiments. These results are in contrast to the that independently derived populations show reproductive strong assortative mating observed in similar studies of other isolation if they come from different habitats but not if they stickleback systems, such as benthic versus limnetic [26]and come from similar habitats: that is, parallel speciation [31, marine versus fresh water [27]. In addition to providing 32]. Ostevik and coauthors review potential examples of eco- potential explanations for this discrepancy, Ras¨ anen¨ et al. logical speciation in plants for evidence of parallel speciation. conclude that the apparent conundrum of limited gene flow They find that very few plant systems provide such evidence, but no obvious reproductive barriers could be very infor- perhaps simply because not many studies have performed mative about the factors that constrain progress toward the necessary experiments. Alternatively, plants might differ ecological speciation. fundamentally from animals in how ecological differences The second paper on three-spine stickleback, by D. I. drive speciation, particularly due to the importance of be- Bolnick, considers the opposite conundrum: reproductive haviour in animals. barriers are seemingly present but gene flow is not limited. A current topic of interest in ecological speciation is Particularly, even though ecologically driven sympatric spe- whether strong selection acting on a single trait (strong selec- ciation does not always occur in sticklebacks, its theoretically tion) or relatively weak selection acting on a greater number necessary and sufficient conditions seem often to be present of traits (multifarious selection) is more common and more in nature. First, some populations experience strong com- likely to complete the speciation process [11]. Using another petition for resources that causes extreme phenotypes to well-studied model of ecological speciation, butterflies of the have higher fitness [28]. Second, assortative mating based genus Lycaeides,C.F.Scholl,C.C.Nice,J.A.Fordyce,Z. on diet and morphology is present in some of these same Gompert, and M. L. Forister compared host-plant associated lakes [29]. So how to solve this new conundrum? Using a larval performance of butterflies from several populations simulation model, Bolnick demonstrates that the strengths of L. idas, L. melissa, and a species that originated through 4 International Journal of Ecology hybridization between the two. By conducting a series of recognized. This habitat isolation is determined by habitat reciprocal rearing experiments, they found little to no evi- choice (preference or avoidance), competition, and habitat dence for local adaptation to the natal hosts. By putting these performance (fitness differences between habitats) [36]. S. E. results into the context of the other previously studied eco- Webster, J. Galindo, J. W. Grahame, and R. K. Butlin propose logical traits (e.g., host and mate preference, phenology, and a conceptual framework to study and classify traits involved egg adhesion [33, 34]), the authors constructed a schematic in habitat choice, based on three largely independent criteria: representation of the diversification within this butterfly (1) whether habitat choice allows the establishment of a species complex. They conclude that no single trait acts as stable polymorphism maintained by selection without in- a complete reproductive barrier between the three taxa and terfering with mating randomness or if it also promotes thatmosttraitsreducegeneflowonlyasymmetrically.The assortative mating; (2) whether it involves one-allele or two- authors suggest the need for further study of multiple traits allele mechanisms of inheritance; (3) whether traits are of and reproductive barriers in other taxa. single or multiple effect [37], the latter when habitat choice is simultaneously under direct selection and contributes to 4. Factors Affecting Progress toward assortative mating. The combination of these three criteria underlies ten different scenarios, which the authors visit Ecological Speciation using previously published empirical data. They argue that 4.1. The Role of Pollinators/Parasites. In many cases of ecolog- the speed and likelihood of ecological speciation depends ical speciation, we think of the populations in question col- on the mechanism of habitat choice and at which stage of onizing and adapting to divergent environments/resources, the process it operates, with scenarios of one-allele and/or ff such as different plants or other food types. However, envi- multiple-e ect traits being more favorable. While these ronments can also “colonize” the populations in question scenarios have rarely been distinguished in empirical studies, that might then speciate as a result. Colonization by different Webster et al. reason that such distinctions will help in the pollinators and subsequent adaptation to them, for example, design of future studies and enable more informative com- is expected to be particularly important for angiosperms. A parisons among systems. In practice, however, the identifica- particularly spectacular example involves sexually deceptive tion of the mechanisms involved and discriminating among Orchids, where flowers mimic the scent and the appearance different scenarios may sometimes be difficult, as exemplified of female insects and are then pollinated during attempted by the case of the intertidal gastropod Littorina saxatilis,a copulation by males. In a review and meta-analysis of two model system for ecological speciation. Orchid genera, S. Xu, P. M. Schluter,¨ and F. P. Schielst find Hybrids resulting from the crosses between individuals floral scent to be a key trait in both divergent selection and from populations with different habitat preferences will tend reproductive isolation. Other traits, including flower colour, to show interest in both parental habitats. This will increase morphology and phenology, also appear to play an impor- gene flow between parental species, inhibiting reproductive tant role in ecological speciation within this group. The isolation. Inspired by host-specific phytophagous insects, J. authors also conclude that although sympatric speciation L. Feder, S. P. Egan, and A. A. Forbes ask, what if indi- is likely rare in nature, it is particularly plausible in these viduals choose their habitat based on avoidance rather than Orchids. preference? According to the authors, hybrids for alleles in- Parasites can be thought of as another instance of differ- volved in avoidance of alternate parental habitats may ex- ent environments “colonizing” a focal species and then caus- perience a kind of behavioral breakdown and accept none ing divergent/disruptive selection and (perhaps) ecological of the parental habitats, generating a postzygotic barrier to speciation. As outlined in the contribution by A. Karvonen gene flow. Feder and collaborators determine the reasons ff and O. Seehausen, di erences in parasites could contribute why habitat avoidance is underappreciated in the study of to ecological speciation in three major ways. First, divergent ecological speciation (theoretical and empirical), and try to parasite communities could cause selection against locally improve this issue. They propose new theoretical models and adapted hosts that move between those communities, as do not find strong theoretical impediments for habitat avoid- well as any hybrids. Second, adaptation to divergent parasite ance to evolve and generate hybrid behavioral inviability even communities could cause assortative mating to evolve as a for nonallopatric scenarios. They also suggest a physiological pleiotropic by-product, such as through divergence in MHC mechanism to explain how habitat specialists evolve to prefer genotypes that are under selection by parasites and also a new habitat and avoid the original one. Feder et al. also influence mate choice (see also [35]). Third, sexual selection document empirical support for this theory. Accumulated might lead females in a given population to prefer males that data on Rhagoletis pomonella and preliminary results on are better adapted to local parasites and can thus achieve bet- Utetes lectoides strongly suggest that avoidance has evolved ter condition. The authors conclude that although suggestive in these species, contributing to postzygotic reproductive evidence exists for all three possibilities, more work is needed isolation. A literature survey in phytophagous insects reveals before the importance of parasites in ecological speciation at least ten examples consistent with habitat avoidance, and can be confirmed. three cases of behavior inviability in hybrids consistent with this mechanism. The authors also present suggestions and 4.2. The Role of Habitat Choice. The importance of habi- cautionary notes for design and interpretation of results tat (or host) isolation in ecological speciation is widely when it comes to experiments on habitat choice. International Journal of Ecology 5

Hybrid zones are particularly useful systems for deter- displaced from their preferred habitat due to competition mining whether differences in habitat preference or habitat- with the dominant collared flycatchers. Cross-fostering ex- associated adaptation contribute to reproductive isolation. periments showed that rearing environment matters to M. D. Carling and H. A. Thomassen investigate the effect of recruits’ habitat choice more than does the environment of environmental variation on admixture in a hybrid zone bet- the genetic parents: pied flycatcher fledglings whose parents ween the Lazuli Bunting (Passerian amoena) and the Indigo were displaced to pine habitats were more likely to return to Bunting (P. cyanea). They find that differences in environ- nest in pine habitats. Thus, competition-mediated switches ment explain interpopulation differences in the frequency between habitats can cause a change of habitat choice and genetic composition of hybrids. This is not the first study through learning, which might then enhance reproductive to document an effect of environmental variation on the isolation via ecological segregation. This role of plasticity and production or persistence of hybrids [38, 39] but Carling learning in habitat choice is also acknowledged in the con- and Thomassen were also able to associate this pattern with tribution of Webster and collaborators. specific environmental variables, particularly rainfall during the warmest months of the year. They discuss possible, complementary mechanistic explanations for these patterns, 5. Unanswered Questions and Future Directions including habitat avoidance or preference in hybrids and habitat-dependent fitness. Their results indicate that inher- Although it is widely recognized that ecological speciation ent (i.e., non-geographic) barriers to gene flow between P. can occur without gene flow between diverging groups of amoena and P. cyanea are environment dependent, which individuals [43], the recognition of its importance has grown means these barriers could be ephemeral and vary in space becauseofrecentevidenceforspeciationwith gene flow [44]. and time. If gene flow commonly occurs during divergence, some S.P.Egan,G.R.HoodandJ.R.Ottpresentoneofthe mechanism, such as divergent selection must also occur fre- first direct tests of the role of habitat (host) isolation driven quently to counteract the homogenizing effect of gene flow. by host choice. Different populations of the gall wasp Belo- The manuscripts in this special issue, and a plethora of nocnema treatae feed on different oak species. Egan et al. first other recent publications [45–50], have made great strides confirmed that B. treatae prefer their native host plant, with in advancing our understanding of ecological speciation. a stronger preference for females. They then demonstrated These allow us to identify several key factors that affect assortative mating among host populations, which was en- progress toward ecological speciation, such as habitat choice hanced by the presence of the respective host plants. This (preference and avoidance), phenotypic plasticity, role of enhancement was due to the fact that females usually mate pollinators/parasites, complex biological interactions such as on their host and that males also prefer their natal host plant. facilitation, as well as geographical context. However, for Therefore, host preference is directly responsible for rep- most cases, our understanding is still incomplete. For in- roductive isolation in B. treatae, by decreasing the pro-babi- stance, the circumstances under which plasticity favors or lity of encounter between individuals from different host inhibits adaptation, mate choice, and consequently ecolog- populations. The mechanism revealed here likely applies to ical speciation are still largely unknown. Further insights will many host/phytophagous or host/parasite systems. certainly arise from a multitude of empirical and theoretical studies, but certain areas of research are particularly likely to yield important results. For example, whereas we can 4.3. The Role of Phenotypic Plasticity. Phenotypic plasticity, rarely observe the time course of speciation in a single spe- the ability of a single genotype to express different pheno- cies, we can learn about factors affecting progress toward types under different environmental conditions, has long ecological speciation by studying and contrasting pairs of been seen as an alternative to genetic divergence, and there- related populations at different points along the speciation fore as potential constraint on adaptive evolution [40, 41]. continuum. Similarly, the study of parallel speciation may be More recently, however, adaptive phenotypic plasticity has highly informative. Such studies exist (e.g., [25, 26, 51, 52]), been rehabilitated as a factor potentially favoring divergent but we need many more systems where we can examine evolution by enabling colonizing new niches, where diver- variation in progress toward ecological speciation. It is im- gent selection can then act on standing genetic variation [42]. portant that we also investigate instances where speciation B. M. Fitzpatrick reviews the possible effects of phenotypic fails, as these cases will advance our understanding of factors plasticity on the two components of ecological speciation: that constrain and enhance progress toward speciation. Fur- local adaptation and reproductive isolation. He finds that thermore, recent advances in DNA sequencing and statistical both adaptive and maladaptive plasticity can promote or analysis offer an unprecedented opportunity to study the constrain ecological speciation, depending on several factors, genetic basis and evolution of reproductive isolation during and concludes that many aspects of how phenotypic plastic- ecological speciation. The application of these new methods ity acts have been underappreciated. and models to ecologically well-studied systems have been Several other papers in the special issue also provide and will be particularly informative [53, 54]. Finally, more potential examples of the role of plasticity in ecological spe- studies using experimental manipulations to study the effects ciation. For instance, N. V. Vallin and A. Qvarnstrom¨ studied of key parameters on ecological speciation are badly needed, habitat choice in two hybridizing species of flycatchers. When especially if they can be combined with an understanding of the two species occur in sympatry, pied flycatchers are natural populations. 6 International Journal of Ecology

Acknowledgments [17] T. H. Vines and D. Schluter, “Strong assortative mating between allopatric sticklebacks as a by-product of adaptation to We would like to express our gratitude to all the authors different environments,” Proceedings of the Royal Society B, vol. and reviewers that contributed to the successful completion 273, no. 1589, pp. 911–916, 2006. of this special issue. R. Faria research is financed by the [18] M. D. Bertness and R. Callaway, “Positive interactions in com- Portuguese Science Foundation (FCT) through the program munities,” Trends in Ecology and Evolution,vol.9,no.5,pp. COMPETE (SFRH/BPD/26384/2006 and PTDC/BIA-EVF/ 191–193, 1994. 113805/2009). [19] M. Kirkpatrick and N. H. Barton, “Evolution of a species’ range,” American Naturalist, vol. 150, no. 1, pp. 1–23, 1997. Marianne Elias [20] J. R. Bridle, J. Polechova,´ M. Kawata, and R. K. Butlin, “Why Rui Faria is adaptation prevented at ecological margins? New insights Zachariah Gompert from individual-based simulations,” Ecology Letters, vol. 13, Andrew Hendry no. 4, pp. 485–494, 2010. [21] T. Dobzhansky, Genetics and the Origin of Species, Columbia References University Press, New York, NY, USA, 1st edition, 1973. [22] H. J. Muller, “Isolating mechanisms, evolution, and tempera- [1] D. Schluter, The Ecology of Adaptive Radiation,OxfordUniver- ture,” Biology Symposium, vol. 6, pp. 71–125, 1942. sity Press, Oxford, UK, 2000. [23] M. Turelli, N. H. Barton, and J. A. Coyne, “Theory and specia- [2] P. Nosil, Ecological Speciation,OxfordUniversityPress, tion,” Trends in Ecology and Evolution, vol. 16, no. 7, pp. 330– Oxford, UK, 2012. 343, 2001. [3] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology [24] J. S. McKinnon and H. D. Rundle, “Speciation in nature: the Letters, vol. 8, no. 3, pp. 336–352, 2005. threespine stickleback model systems,” Trends in Ecology and [4] K. Ras¨ anen¨ and A. P. Hendry, “Disentangling interactions Evolution, vol. 17, no. 10, pp. 480–488, 2002. between adaptive divergence and gene flow when ecology [25] A. P.Hendry, D. I. Bolnick, D. Berner, and C. L. Peichel, “Along drives diversification,” Ecology Letters, vol. 11, no. 6, pp. 624– the speciation continuum in sticklebacks,” Journal of Fish Bio- 636, 2008. logy, vol. 75, no. 8, pp. 2000–2036, 2009. [5] D. Lack, Darwin’s Finches, Cambridge University Press, Cam- [26] H. D. Rundle, L. Nagel, J. W. Boughman, and D. Schluter, bridge, UK, 1947. “Natural selection and parallel speciation in sympatric stick- [6] G. G. Simpson, The Major Features of Evolution, Columbia lebacks,” Science, vol. 287, no. 5451, pp. 306–308, 2000. University Press, New York, NY, USA, 1953. [27] J. S. McKinnon, S. Mori, B. K. Blackman et al., “Evidence for [7]P.R.GrantandB.R.Grant,How and Why Species Multiply: ecology’s role in speciation,” Nature, vol. 429, no. 6989, pp. The Radiation of Darwin’s Finches, Princeton University Press, 294–298, 2004. Princeton, NJ, USA, 2008. [28] D. I. Bolnick and L. L. On, “Predictable patterns of disruptive [8]J.B.Losos,Lizards in an Evolutionary Tree: Ecology and selection in stickleback in postglacial lakes,” American Natu- Adaptive Radiation of Anoles, University of California Press, ralist, vol. 172, no. 1, pp. 1–11, 2008. Berkeley, Calif, USA, 2009. [29] L. K. Snowberg and D. I. Bolnick, “Assortative mating by diet [9] R. J. Rundell and T. D. Price, “Adaptive radiation, nonadaptive in a phenotypically unimodal but ecologically variable popu- radiation, ecological speciation and nonecological speciation,” lation of stickleback,” American Naturalist, vol. 172, no. 5, pp. Trends in Ecology and Evolution, vol. 24, no. 7, pp. 394–399, 733–739, 2008. 2009. [30] J. M. Sobel, G. F. Chen, L. R. Watt, and D. W. Schemske, “The [10] A. P. Hendry, “Ecological speciation! Or the lack thereof?” biology of speciation,” Evolution, vol. 64, no. 2, pp. 295–315, Canadian Journal of Fisheries and Aquatic Sciences, vol. 66, no. 2010. 8, pp. 1383–1398, 2009. [31] D. J. Funk, “Isolating a role for natural selection in speciation: [11] P. Nosil, L. J. Harmon, and O. Seehausen, “Ecological explana- host adaptation and sexual isolation in Neochlamisus beb- tions for (incomplete) speciation,” Trends in Ecology and Evo- bianae leaf beetles,” Evolution, vol. 52, no. 6, pp. 1744–1759, lution, vol. 24, no. 3, pp. 145–156, 2009. 1998. [12] J. B. Hughes, G. C. Daily, and P. R. Ehrlich, “Population [32] L. Nagel and D. Schluter, “Body size, natural selection, and diversity: its extent and extinction,” Science, vol. 278, no. 5338, speciation in sticklebacks,” Evolution, vol. 52, no. 1, pp. 209– pp. 689–692, 1997. 218, 1998. [13] G. E. Hutchinson, “Homage to Santa Rosalia or why are there [33] J. A. Fordyce, C. C. Nice, M. L. Forister, and A. M. Shapiro, so many kinds of animals?” The American Naturalist, vol. 93, “The significance of wing pattern diversity in the Lycaenidae: pp. 145–159, 1959. mate discrimination by two recently diverged species,” Journal [14] J. Felsenstein, “Skepticism towards Santa Rosalia, or why are of Evolutionary Biology, vol. 15, no. 5, pp. 871–879, 2002. there so few kinds of animals?” Evolution, vol. 35, pp. 124–138, [34] J. A. Fordyce and C. C. Nice, “Variation in butterfly egg ad- 1981. hesion: adaptation to local host plant senescence characteris- [15] U. Dieckmann, J. A. J. Metz, M. Doebeli, and D. Tautz, tics?” Ecology Letters, vol. 6, no. 1, pp. 23–27, 2003. “Introduction,” in Adaptive Speciation, U. Dieckmann, J. A. J. [35] C. Eizaguirre and T. L. Lenz, “Major histocompatability com- Metz, M. Doebeli, and D. Tautz, Eds., pp. 1–17, Cambridge plex polymorphism: dynamics and consequences of parasite- University Press, Cambridge, UK, 2004. mediated local adaptation in fishes,” Journal of Fish Biology, [16] R. B. Langerhans, M. E. Gifford, and E. O. Joseph, “Ecological vol. 77, no. 9, pp. 2023–2047, 2010. speciation in Gambusia fishes,” Evolution,vol.61,no.9,pp. [36] J. A. Coyne and H. A. Orr, Speciation, Sinauer, Sunderland, 2056–2074, 2007. Mass, USA, 2004. International Journal of Ecology 7

[37] C. Smadja and R.K. Butlin, “A framework for comparing processes of speciation in the presence of gene flow,” Molecular Ecology, vol. 20, pp. 5123–5140, 2011. [38] D. M. Rand and R. G. Harrison, “Ecological genetics of a mosaic hybrid zone: mitochondrial, nuclear, and reproductive differentiation of crickets by soil type,” Evolution, vol. 43, no. 2, pp. 432–449, 1989. [39] T. H. Vines, S. C. Kohler,¨ M. Thiel et al., “The maintenance of reproductive isolation in a mosaic hybrid zone between the fire-bellied toads Bombina bombina and B. Variegata,” Evolu- tion, vol. 57, no. 8, pp. 1876–1888, 2003. [40] S. Wright, “Evolution in Mendelian populations,” Genetics, vol. 16, pp. 97–159, 1931. [41] D. A. Levin, “Plasticity, canalization and evolutionary stasis in plants,” in Plant Population Ecology,A.J.Davy,M.J. Hutchings, and A. R. Watkinson, Eds., pp. 35–45, Blackwell Scientific Publications, Oxford, UK, 1988. [42] D. W. Pfennig, M. A. Wund, E. C. Snell-Rood, T. Cruickshank, C. D. Schlichting, and A. P. Moczek, “Phenotypic plasticity’s impacts on diversification and speciation,” Trends in Ecology and Evolution, vol. 25, no. 8, pp. 459–467, 2010. [43] D. Schluter, “Evidence for ecological speciation and its alternative,” Science, vol. 323, no. 5915, pp. 737–741, 2009. [44] C. Pinho and J. Hey, “Divergence with gene flow: models and data,” Annual Review of Ecology, Evolution, and Systematics, vol. 41, pp. 215–230, 2010. [45] P. Nosil, D. J. Funk, and D. Ortiz-Barrientos, “Divergent selection and heterogeneous genomic divergence,” Molecular Ecology, vol. 18, no. 3, pp. 375–402, 2009. [46] J. L. Feder and P. Nosil, “The efficacy of divergence hitchhiking in generating genomic islands during ecological speciation,” Evolution, vol. 64, no. 6, pp. 1729–1747, 2010. [47] J. L. Feder, G. Gejji, S. Yeaman, and P. Nosil, “Establishment of new mutations under divergence and genome hitchhiking,,” Philosophical Transactions of the Royal Society B, vol. 367, pp. 461–474, 2012. [48] J. L. Feder and P. Nosil, “Chromosomal inversions and species differences: when are genes affecting adaptive divergence and reproductive isolation expected to reside within inversions?” Evolution, vol. 63, no. 12, pp. 3061–3075, 2009. [49]A.P.Michel,S.Sim,T.H.Q.Powell,M.S.Taylor,P.Nosil,and J. L. Feder, “Widespread genomic divergence during sympatric speciation,” Proceedings of the National Academy of Sciences of the United States of America, vol. 107, no. 21, pp. 9724–9729, 2010. [50]R.Faria,S.Neto,M.A.F.Noor,andA.Navarro,“Roleof natural selection in chromosomal speciation,” in Encyclopedia Life Sciences, Chichester, UK, 2011. [51] D. Berner, A. C. Grandchamp, and A. P. Hendry, “Variable progress toward ecological speciation in parapatry: stickleback across eight lake-stream transitions,” Evolution, vol. 63, no. 7, pp. 1740–1753, 2009. [52]R.M.Merrill,Z.Gompert,L.M.Dembeck,M.R.Kronforst, W. O. Mcmillan, and C. D. Jiggins, “Mate preference across the speciation continuum in a clade of mimetic butterflies,” Evolution, vol. 65, no. 5, pp. 1489–1500, 2011. [53]P.A.Hohenlohe,S.Bassham,P.D.Etter,N.Stiffler,E.A. Johnson, and W. A. Cresko, “Population genomics of parallel adaptation in threespine stickleback using sequenced RAD tags,” PLoS Genetics, vol. 6, no. 2, Article ID e1000862, 2010. [54] P. Nosil and J. L. Feder, “Genomic divergence during speciation: causes and consequences,” Philosophical Transactions of the Royal Society B, vol. 367, pp. 332–342, 2012. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 280169, 20 pages doi:10.1155/2012/280169

Review Article The Role of Parasitism in Adaptive Radiations—When Might Parasites Promote and When Might They Constrain Ecological Speciation?

Anssi Karvonen1 and Ole Seehausen2, 3

1 Department of Biological and Environmental Science, Centre of Excellence in Evolutionary Research, University of Jyvaskyl¨ a,¨ P.O. Box 35, 40014, Finland 2 Department of Fish Ecology and Evolution, Eawag: Centre of Ecology, Evolution and Biogeochemistry, Seestrasse 79, 6047 Kastanienbaum, Switzerland 3 Division of Aquatic Ecology & Macroevolution, Institute of Ecology and Evolution, University of Bern, Baltzerstrasse 6, 3012 Bern, Switzerland

Correspondence should be addressed to Ole Seehausen, [email protected]

Received 16 August 2011; Revised 28 November 2011; Accepted 2 January 2012

Academic Editor: Andrew Hendry

Copyright © 2012 A. Karvonen and O. Seehausen. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Research on speciation and adaptive radiation has ﬂourished during the past decades, yet factors underlying initiation of reproductive isolation often remain unknown. Parasites represent important selective agents and have received renewed attention in speciation research. We review the literature on parasite-mediated divergent selection in context of ecological speciation and present empirical evidence for three nonexclusive mechanisms by which parasites might facilitate speciation: reduced viability or fecundity of immigrants and hybrids, assortative mating as a pleiotropic by-product of host adaptation, and ecologically-based sexual selection. We emphasise the lack of research on speciation continuums, which is why no study has yet made a convincing case for parasite driven divergent evolution to initiate the emergence of reproductive isolation. We also point interest towards selection imposed by single versus multiple parasite species, conceptually linking this to strength and multifariousness of selection. Moreover, we discuss how parasites, by manipulating behaviour or impairing sensory abilities of hosts, may change the form of selection that underlies speciation. We conclude that future studies should consider host populations at variable stages of the speciation process, and explore recurrent patterns of parasitism and resistance that could pinpoint the role of parasites in imposing the divergent selection that initiates ecological speciation.

1. Introduction often referred to as “ecological speciation” [4–8]. Ecological speciation research has now begun to integrate ecological Since the publication of the Darwin’s “Origin of species” one and genomic research towards the identification of genes and a half centuries ago, processes and mechanisms by which that are important at the onset of ecological speciation in new species arise have fascinated evolutionary biologists. a few systems [9–13]. However, at the same time, some It is increasingly apparent that the rich biodiversity found of the most basic questions, such as what factors initiate on our planet has, at least partly, evolved in bursts of and drive the emergence of reproductive isolation between adaptive diversification, associated with the quick origin diverging populations, remain unanswered for all but a of new species, referred to as adaptive radiation [1, 2]. handful of systems. Traditionally, research on ecological The intensive research on speciation of the past 20+ years, speciation has focused on habitat and trophic specialization initiated perhaps by the publication of “Speciation and its and on the role of resource competition, as drivers of consequences” [3], has produced much support for the divergence and reproductive isolation within and between hypothesis of speciation through divergent natural selection, populations [14, 15]. Some of the recent empirical evidence 2 International Journal of Ecology supports the role of these mechanisms (reviewed in [6, 8]). some important tests on the theory of parasite-mediated Moreover, predation has classically been considered as an ecological speciation which are currently lacking. Essentially, important potential driver of divergence [16], and this idea these concern the initial stages of the speciation process, that has recently been explored in a number of papers (e.g., is, at which stage of the speciation continuum do parasite [17, 18]). infections become divergent among the host populations, Parasitism is a predominant biological interaction in the and do they importantly restrict the gene flow between wild [19, 20], but it has received relatively little attention in host populations? Also, we contrast the role of diversity speciation research. Parasites live on the expense of other of a parasite community with the role of single parasite organisms by taking some or all of the energy they need species in driving parasite-mediated speciation, conceptually from their host. Because of this peculiar life style, parasites linking this to discussion on multifariousness of selection have significant ecological and evolutionary consequences and the strength of selection. Finally, we discuss how for hosts and host populations [21–24]. Potentially, infec- different types of infections that, for example, alter host tions might also initiate, facilitate, or reinforce speciation behaviour or visual abilities, could influence the process of by imposing selective pressures that differ in form and speciation, or its reversal. We limit this review to meta- strength from those imposed by the abiotic environment. zoan and microparasite (protozoans, bacteria, and viruses) Parasites may also impose a range of interrelated effects infections, while acknowledging that reproductive isolation on host appearance, behaviour, condition, and, importantly, and speciation may occur also in other fascinating parasitic defence system. Classical papers have identified parasites as interactions. These include, for example, brood parasitism in important sources of divergent selection [25, 26] and there birds [42], where the interaction differs from “traditional” is strong evidence to support their role as mediators of host-parasite systems as the parasite is not physically attached species coexistence [27, 28]. However, while this has led some to the host, and symbiotic bacteria-host interactions, where authors to make far-reaching statements about the role of mating preference can develop as a side effect of host parasites in driving host diversification, evidence for speci- adaptation to the environment [43]. We also restrict our ation driven by parasites is very limited (though evidence review of empirical evidence to the zoological literature, may be strong for intraspecific genetic host diversity). The but acknowledge that there is a larger body of evidence for reasons for this lack of evidence are several: most studies to speciation in plants driven by coevolution with pathogens date are correlational and cannot separate cause and effect and predators (see [33] as a classical starting point). There regarding diversity in parasites and hosts. For example, while is also a wealth of recent literature on speciation in microbial some studies conclude that parasite diversity is a result of systems, such as bacteria-phage interactions (e.g., [44, 45]), host diversity (e.g., [29, 30]), others have concluded the which is not considered here. We provide examples mainly opposite even with the same data sets [31]. Coevolution that from fishes and birds where some of the best case studies commonly prevails in host-parasite interactions is predicted of ecological speciation and adaptive radiation exist and to generate diversity at least in some constellations [32], significant progress has been made in testing predictions and there is wide-spread empirical support for parasites from models of parasite-mediated speciation. diverging in response to host speciation ([33–35], [36]for a model). Such speciation may be ecological but is mediated 2. Prerequisites for Parasite-Mediated by resource specialization and not by parasites. Yet, in cases Divergent Selection of cospeciation, it can be difficult to interpret which one (if any) of the coevolving partners actually triggered the There are three main prerequisites for parasite-mediated speciation in the other one. Moreover, divergence in parasite divergent selection to operate in natural host populations. infections is commonly associated with divergence in food First and the most obvious is that infections should differ regimes and habitat [37–39]. This makes it difficult to infer within or between the host populations. This can happen parasite-mediated host divergence when there is coincident in allopatric host populations experiencing differences in multivariate divergent selection between niches. It is also diversity or magnitude of infections, but also in sympatric possible, and supported by some data, that parasites may or parapatric populations where heterogeneities in ecological actuallypreventhostspeciation[40, 41]. (the extent of exposure) or genetic (susceptibility) predis- In the present paper, we review and discuss the role that position to infection create subgroups or subpopulations parasites might have in ecological speciation and adaptive that have different infection levels. Overall, heterogeneities radiation of their hosts. We go through the existing literature in infections within a host species inhabiting different on the theory of parasite-mediated selection and discuss geographical areas represent one of the best known phe- mechanisms that could lead to reproductive isolation in nomena in host-parasite interactions, and basically lay the allopatric, parapatric or sympatric host populations, and foundations for investigating parasite-mediated divergent the prerequisites for these mechanisms to operate. We selection. For example, it is well known that ecological then review the empirical literature on parasite-mediated factors such as differences in host population structure or speciation with an emphasis on fishes and birds. Hoping not in environmental factors may generate variation in infection to miss recent publications, we viewed all papers published among populations of one host species (e.g., [46–48]). in the past two years (May 2009 to July 2011) that were Typically, this is seen as a decrease in similarity of parasite retrieved from Web of Science using the combination of species composition with increasing geographical distance search terms “parasite” and “speciation.” We also point out among the host populations [49, 50] or even among different International Journal of Ecology 3 locations within one host population [51]. Overall, such parasites and hosts, but empirical evidence for it was lacking. heterogeneities of infections could generate highly variable Since then, new empirical evidence has been gathered, and conditions for parasite-mediated selection. some of it supports the hypothesis of parasite-mediated The second prerequisite for parasite-mediated divergent ecological speciation, yet overall the empirical support is still selection is that differences in infections should remain rea- scant. Some of the best data to test the hypothesis come sonably constant among the host populations through time, from freshwater fish and from birds. Progress has recently thus maintaining the direction and perhaps also the strength been made in some of these key systems in identifying of the divergent selection. For example, infections could differences in infections among populations, ecotypes and/or be highly predictable with the same species composition sister species (the first prerequisite for parasite-mediated and more or less similar infection intensities occurring in selection), and connecting these to possible mechanisms hosts every year, or show high levels of stochastic year-to- initiating, facilitating, or maintaining host population diver- year variation among the host populations causing parasite- gence and speciation. Table 1 summarizes some of the best mediated selection to fluctuate in strength and direction studied examples. Here, we first review the existing literature and making consistent divergent selection unlikely. Similarly, on divergent parasite faunas in ecotypes of freshwater fishes spatial repeatability of infections across replicated host where much new data have been gathered recently. Second, populations can be important when evaluating the role of we bring up examples of studies that have gone further parasites in speciation. In particular, such repeatability could into testing predictions of mechanisms of parasite-mediated reveal patterns of parallel ecological speciation, which is speciation and discuss these under the three categories of discussed in more detail below. Moreover, if host divergence mechanisms: reduced immigrant and hybrid viability or is more likely across populations when certain parasite fecundity, pleiotropy, and ecologically based sexual selection. species are present (or absent), this can support the role of For this second part of the review, we do not restrict ourselves these parasites in host divergence. We come back also to this to fish. topic later in this paper. The third important prerequisite for parasite-mediated divergent selection is that infections impose fitness conse- 3.1. Divergent Parasite Infections. Despite the wealth of the quences for the hosts and that these are sufficiently strong literature on heterogeneities in parasite infections across to overrule possible conflicting fitness consequences of other host species and populations, surprisingly few empirical factors. This is required for parasites to actually impose net studies have investigated differences in parasite species divergent selection between host populations. Such fitness composition in sympatric and parapatric host ecotypes or consequences are generally assumed because parasites take sister species in the context of parasite-mediated divergent the energy they need from the hosts which may result selection and speciation. In fishes, such systems include in reduced host condition and reproduction. Testing it, salmonid and three-spine stickleback populations in the however, requires empirical measurement of fitness in nature northern hemisphere, as well as cichlid fishes in East African or in reciprocal transplants that simulate natural conditions, great lakes (Table 1). For example, parapatric lake and river whereas measurement of infection-related fitness compo- populations of sticklebacks in northern Germany differ nents is insufficient. An important feature of host-parasite in parasite species composition so that lake populations interactions is that wild hosts are typically infected with a harbour a significantly higher diversity of infections [54–56]. range of parasite species at the same time. For example, in Differences in parasitism have also been reported between aquatic systems, individual fish hosts are commonly infected marine and freshwater ecotypes of stickleback [57], as well with dozens of parasite species simultaneously (e.g., [46, as between sympatric stickleback species specializing on 47]). This is important in terms of direction and magnitude benthic and limnetic environments in lakes of Western of selection. Under such circumstances, parasite-mediated Canada [58]. In all of these systems, divergent patterns divergent selection could be driven by a single parasite of infection are most likely explained by differences in species having major impact on host viability or reproduc- parasite transmission between different environments or by tion. Alternatively, selection could represent joint effects of adaptation of the immune defence to these habitats [54, multiple parasite species, each with unique types of effects on 56]. Other systems in the northern hemisphere also include the host and possibly opposing effects in terms of divergent whitefish and Arctic charr in lakes in Norway, where ecotypes selection (e.g., see recent discussion in Eizaguirre and Lenz and species inhabiting pelagic versus benthic habitats, and [52] on selection on MHC polymorphism). Separating such profundal versus benthic/pelagic habitats, respectively, show effects in natural host populations is a demanding task, significant differences in parasite infections [59, 60]. Similar which is discussed more below. differences in infections have also been reported from four ecotypes of Arctic charr in a large lake in Iceland (Figure 1); ecotypes inhabiting littoral areas are more heavily infected 3. Mechanisms and Empirical Evidence of with parasites transmitted through snails while the pelagic Parasite-Mediated Host Speciation ecotypes harbour higher numbers of cestode infections transmitted trophically through copepods [61]. Moreover, In a review on this subject eight years ago, Summers we have recently observed differences in parasitism between et al. [53] concluded that theory suggests that parasite- whitefish populations and species reproducing at different host coevolution might enhance speciation rates in both depths in Swiss prealpine lakes [62]. 4 International Journal of Ecology

Divergent parasite infections have also been described We consider three nonexclusive categories of mecha- from cichlid fish in the lakes of East Africa, especially Lake nisms (Figure 2). Malawi and Lake Victoria. These systems are particular as they harbour a tremendous diversity of hundreds of cichlid (1) Direct natural selection: reproductive isolation due fish species each that have emerged in the lakes in a few ten to parasite-mediated reduction of immigrant and thousand to one or two million years [63–65], representing hybrid viability or fecundity [68]. spectacular examples both of biodiversity and adaptive (2) Pleiotropy: direct natural selection operates on the radiation, and of the high rates with which these can emerge. genes of the immune system, and the latter pleiotrop- Recently, Maan et al. [66] described divergent parasite species ically affectmatechoice[69]. composition in the closely related sister species Pundamilia (3) Ecologically based sexual selection: reproductive iso- pundamilia and P. nyererei of Lake Victoria. These differences lation due to parasite-mediated divergent sexual were caused mainly by larval nematodes in the internal selection [8]. organs and ectoparasitic copepods associated with feeding more benthically in shallower water or more limnetically and The first two categories of mechanisms could be considered slightly deeper. Similarly, heterogeneous infections have been byproduct speciation mechanisms, although the first one in reported in Lake Malawi, where the closely related species particular may require reinforcement selection for comple- Pseudotropheus fainzilberi and P. emmiltos show divergent tion of speciation. The third mechanism could be considered parasite species composition particularly in terms of certain reinforcement-like speciation [8]. ectoparasitic and endoparasitic infections [67]. Overall, such differences in infections fulfil the first 3.2.1. Tests of Parasite-Mediated Viability or Fecundity Loss in prerequisite of parasite-mediated divergent selection and Immigrants and Hybrids. In theory, adaptation to habitat- support the idea of a possible role of parasites in ecological specific parasite challenges in ecotypes experiencing diver- speciation.However,itisstilldifficult to evaluate the gent parasite infections could facilitate reproductive isolation generality of these findings. This is first because the number between the ecotypes through parasite-mediated selection of empirical studies describing divergent parasitism among against immigrants that acquire higher infection load outside host ecotypes is still quite limited and examples only come their habitat, or hybrids that show nonoptimal resistance from few relatively well-known systems. Second, it is possible against the parasites and higher infection in either habitat that there is an ascertainment bias in the literature so that (Figure 2). Selection against immigrants was recently investi- studies reporting nonsignificant differences in infections gated in marine and freshwater sticklebacks in Scotland and tend to not get published. This would be particularly likely in Canada [70]. In these systems, anadromous marine fish, with hosts in early stages of the speciation continuum ancestral populations to the freshwater ecotypes, regularly if infections are not yet significantly divergent. However, migrate to freshwater to breed, but are still reproductively we point to the necessity of such data in detail below. isolated from the resident, sympatric freshwater ecotypes. Overall, differences in parasite infections between diverging Using transplant experiments of lab-raised fish to simu- hosts alone do not reveal mechanisms underlying speciation, late dispersal and antihelminthic treatment, MacColl and which we will discuss next. Chapman [70] demonstrated that ancestral-type marine sticklebacks contract higher burdens of novel parasites when introduced to freshwater, than in saltwater and suffer a 3.2. Mechanisms of Parasite-Mediated Host Speciation. Speci- growth cost as a direct result. Susceptibility to parasites and ation is a complex process, typically characterized by simul- their detrimental effect in freshwater was less in derived, taneous operation of several factors and a cascade of events freshwater fish from evolutionarily young populations, pos- from initiation to completion. One of the most challenging sibly as a result of selection for resistance. MacColl and problems in speciation research is to determine the relative Chapman [70] concluded that differences in infections importance to initiating, stabilizing, and completing the could impose selection against migrants from the sea into process of the many factors that typically vary between freshwater populations, but they did not test for selection populations, incipient and sister species. Given that specia- against migrants in the opposite direction. Similar evidence tion is most readily defined as the evolutionary emergence comes from mountain white-crowned sparrows (Zonotrichia of intrinsic reproductive barriers between populations, the leucophrys oriantha), a passerine bird where immigrant males most central question in speciation research is which factors were more heavily infected with bloodborne Haemoproteus drive its emergence, and what is the sequence in which parasites and had lower mating success [71]. The authors they typically play? In this paper we are concerned with the suggested that immigrant birds may be immunologically mechanisms by which parasites could initiate the emergence disadvantaged, possibly due to a lack of previous experience of reproductive isolation, or facilitate or reinforce it after it with the local parasite fauna, resulting in low mating success. had been initiated by other (ecological) factors. This also A related mechanism by which direct natural selection leads to a key question: at which stage of the speciation could act in generating reproductive isolation is reduced process do infections become divergent and begin to reduce fitness in hybrids, that is, offspring of two divergently gene flow between the host populations? In other words, do adapted individuals from environments or habitats that host divergence and the initiation of reproductive isolation differ in parasite infections suffer reduced viability or follow divergence in parasite infections, or vice versa? fecundity, for example, because their intermediate resistance International Journal of Ecology 5

300

250

200

150

Number of parasites Number 100

0 SB LB PL PI Figure 1: Top left: three of the four ecotypes and species of Arctic charr (Salvelinus alpinus species complex) found in Thingvallavatn, Iceland (top left). S. thingvallensis, a small benthic ecotype (uppermost) lives mainly in lava crevices, the pelagic ecotype, one morph of S. murta (middle) feeds mainly on plankton in open water, and the large piscivore ecotype, another morph of S. murta (lowermost) preys upon smaller fishes. Bottom left: Diplostomum metacercariae in an eye lens of fish (all photos Anssi Karvonen). These widespread and abundant parasites cause cataracts and have significant fitness consequences for the fish. Species of the same genus are also found in the vitreous humour of the fish eye, like in the Icelandic ecotypes of charr. Right: average total sum of cestodes (white bars) and trematodes (grey bars) in the four ecotypes of arctic charr in Thingvallavatn (SB: small benthic S. thingvallensis, LB: large benthic S. sp., PL: planktivorous S. murta, PI: piscivorous S. murta). The result illustrates the extent of variation in parasite infections between the sympatric and parapatric ecotypes and interactions between different parasite taxa. Figure produced with permission of John Wiley & Sons Inc. from data in Frandsen et al. [61]. profiles do not match with either of the environments. the mountain white-crowned sparrows. Studying an out- There is a wealth of empirical literature on parasitism in bred population for which it was known that parasites animal species hybrids, and a small number of studies reduce fitness, MacDougall-Shackleton et al. [75]found report higher infection rates of hybrid individuals, reviewed that haematozoan parasite load was significantly negatively by Fritz et al. [72] and Moulia [73]. However, most of correlated with two complementary measures of microsatel- these studies deal with only distantly related species and lite variability. The authors suggested that heterozygote interpretation in the context of speciation is problematic. advantage in terms of parasite load may counteract the In the context of the present paper, we are interested in high parasitism of immigrants (see above), who are likely examples involving ecotypes, sibling species, or young sister to produce the most heterozygous offspring (Figure 3). A species. We go through recent examples from sticklebacks, similar situation has also been reported in a population mountain white-crowned sparrows and other birds, all of of song sparrows (Melospiza melodia), a species in which which actually speak against the hypothesis of parasite- females often display strong preferences for local male song, driven hybrid inviability and infecundity. Rauch et al. [55] and that is thought to undergo speciation in parts of its range studied hybrids between stickleback from lake populations [76]. Here too immigrants were less likely than residents harbouring high parasite infections and river populations to breed, but the outbred offspring of these immigrants with fewer infections in Northern Germany. Hybrids with had higher survivorship [77]. Perhaps the best evidence for intermediate defence profiles in terms of MHC did not suffer parasite-induced loss of hybrid viability comes from studies higher parasite infections in reciprocal infection trials in on hybrid zones between eastern and western house mice either lake or river environments. Similar evidence against [72, 73]. This is considered as classical tension zone where the hypothesis of parasite-mediated selection against hybrids allopatric lineages with well-divergent genomes meet and has also been presented from collared and pied flycatchers, hybridize such that some hybrid genotypes suffer intrinsic where individuals living in the hybrid zone of these two sister incompatibilities. House mice F1 hybrids enjoy reduced species showed intermediate prevalence of Haemoproteus parasite susceptibility but hybrid breakdown is apparent in blood parasites, as well as intermediate immune responses higher generation hybrids. to infection [74](Figure 3). Overall, it is difficult to draw general conclusions on Even stronger evidence against the hypothesis of parasite- the role of parasite-induced hybrid inviability or infecundity mediated selection against hybrids comes from work on in speciation processes as evidence for parasite-mediated 6 International Journal of Ecology

1 It is important to note that coevolution in host-parasite interactions may either facilitate hybridisation and gene flow or isolation and speciation, depending on the dynamics of coevolution (reviewed in [53]). For example, locally adapted 2 parasites should have higher success in their resident hosts, providing an advantage to immigrants and hybrids in the hosts, whose genetic profile cannot be matched by the X locally adapted parasites (i.e., the enemy release hypothesis in invasion biology). On the other hand, if local hosts are (a) well adapted to their local parasites, and parasites are con- sequentially not locally adapted, resident hosts should have equal or higher resistance than immigrants and hybrids. In theory, the situation where parasites are ahead of their hosts, should favour speciation between parasite populations but (b) constrain speciation between host populations, whereas the reverse should facilitate speciation between host populations [53]. Few empirical studies of parasite-mediated speciation have explicitly looked at this. We also point out that the above coevolutionary scenarios (c) between parasites and hosts could commence not only at the level of different parasite species compositions, but also Figure 2: Schematic presentation of models of parasite-mediated at parasite genotype compositions. Under such circum- speciation. (a) Reproductive isolation due to reduced viability or stances, different coevolutionary dynamics driving divergent fecundity of immigrants and hybrids. (1) Immigrants from host ff populations (a) (dark grey) and (b) (light grey) suffer higher parasite-mediated selection between di erent environments infection levels in the habitat of the other population resulting in could take place with seemingly identical parasite species reduced survival or fecundity. (2) Hybrids (middle grey) between assemblies that are “cryptically divergent” showing different divergently adapted parent populations (a) and (b) have higher genotype composition between the environments. Concep- infection levels and reduced survival or fecundity in either of tually, this can be seen as an extension to the hypothesis on the parental habitats as their intermediate defence profiles cannot divergent selection between contrasting environments. match with the parasite pressure of the parental habitats. (b) Moreover, it is important to note that host-parasite Reproductive isolation due to pleiotropic effects of MHC on mate interactions commonly show high levels of genetic polymor- choice. Divergence of parasite infections between host populations phism that could fuel speciation potential in parasites and/or (indicated by darker and lighter grey background) with initially hosts. In general, such variation could be maintained by similar MHC profiles leads to divergent adaptation in MHC profiles different combinations of genotype by environment inter- to the particular infection conditions (dark grey and light grey). × × × × Reproductive isolation between the populations increases in the actions (G G, G EorG G E) [79, 80], for example, course of the process through the pleiotropic effects of MHC as a consequence of parasite-parasite interactions within a on mate choice. (c) Ecologically based sexual selection.Twohost coinfecting parasite community [81], or because of effects populations that differ in parasite infections because of habitat or of environment on host susceptibility [82]. However, genetic diet, diverge in their use of mating cues because different cues polymorphism does not necessarily lead to emergence of new better signal heritable resistance to the different infections (here species if factors mediating divergent selection between host red and blue). Initially they are weakly reproductively isolated populations are absent. with frequent occurrence of hybrid individuals (purple). Sexual selection for individuals that better resist parasites in a given environment (bright blue and red) over more heavily infected 3.2.2. Tests of Pleiotropy: MHC and Mate Choice. Reproduc- individuals (pale blue and red) facilitates divergent adaptation and tive isolation can also emerge as a byproduct of parasite- results in reproductive isolation between the populations. mediated divergent evolution at the genes of the immune system that pleiotropically affectmatechoice(Figure 2). This includes the highly polymorphic family of genes in the major histocompatibility complex (MHC) that encode selection against hybrids in animal systems is mainly antigen-presenting molecules and have an important role in restricted to hybrids between old and genetically very distinct identification of non-self-particles and activation of adaptive host species [78]. Such data speak little to the role of parasites immunity. They are also often involved in mate choice in speciation just like studies on resource partitioning [69], thus having a pleiotropic role in parasite resistance between old coexisting species do not inform us about the and reproductive behaviour. The role of MHC in immune possible role of resource competition in speciation. More defence and mate choice has recently been reviewed by controlled experimental studies are needed to tackle effects Eizaguirre and Lenz [52]. In theory, MHC-mediated mate of parasitism in recently diverged host species in the natural choice may lead to assortative mating in host populations, ecological context. Hybrid zones and sympatric hybridising albeit under restricted conditions [83]. This has recently ecotypes would be good places to do such studies. received empirical support in some systems [54](Figure 4). International Journal of Ecology 7

0.8

0.7

0.6

Heterozygosity 0.5

0.4 Uninfected Infected Figure 3:Topleft:inthehybridzonebetweenpiedflycatcher(Ficedula hypoleuca) and collared flycatcher (Ficedula albicollis) hybrids showed intermediate prevalence of Haemoproteus blood parasites compared to the parental species, and also intermediate immune responses to infection [74](photoshowsamalehybridF. hypoleuca × F. albicollis, courtesy of Miroslav Kral).´ Bottom left: haematozoan blood parasites can be agents of severe selection in birds (photo shows Haemoproteus multipigmentatus infecting red blood cells of endemic Galapagos´ doves [84], courtesy of Gediminas Valkiunas). Right panels: in the mountain white-crowned sparrows (Zonotrichia leucophrys oriantha) heterozygote advantage in terms of reduced parasite load may counteract elevated parasitism of immigrants, who are likely to produce the most heterozygous offspring [75] (photo courtesy of Bob Steele). Figure reproduced with permission of the Royal Society of London from MacDougall-Shackleton et al. [75].

1.8

1.6

1.4 sp. (log) sp. 1.2

0.8 Gyrodactylus 0.6

0.4

Number ofNumber 0.2

0 Without G With G

Figure 4: Top left: a nuptial male of a lake ecotype of the Three-spined stickleback (Gasterosteus aculeatus species complex, photo Ole Seehausen). Bottom left: a breeding male stickleback and a series of Schistocephalus cestodes that were found in its body cavity (photo courtesy of Kay Lucek). These tapeworms change the behaviour of stickleback and effectively castrate them. Right: mean number of ectoparasitic Gyrodactylus monogenean parasites (log transformed) on Schleswig Holstein lake (grey bars) and river (white bars) sticklebacks, with or without an MHC haplotype G. Occurrence of the haplotype coincides with higher resistance against the parasite in the river ecotype while there is a tendency for the opposite pattern in the lake ecotype. Figure produced with permission from data in Eizaguirre et al. [54]. 8 International Journal of Ecology , ] 91 ] 92 – ] ] , , 54 70 70 58 56 [ [ [ Refs 142 [ Parapatric Parapatric Parapatric The geography of divergence Reinforcement after secondary contact Yes: pleiotropy: pre-existing tendency for MHC com- plementarity causes female discrimination against too dissimilar males, that is, assortative mating. indirect selection: female choice for locally resistant males Is divergent mate choice based on MHC- odours? Not tested Not tested Is divergent mate choice based on a revealing signal? Yes, it is at least partly based on male breeding dress Yes, thepopulation lake has larger MHC allele diversity Not known Not tested Not tested Not known Not tested Not tested Is there divergence in a potential revealing signal or MHC? Yes, the limnetic species display redder male breeding dress and have lower MHC allele diversity Yes Not known Not known Is there assortative mating? Yes er ff increased susceptibility to parasites No, intermediates do not su No data Is there parasite- mediated selection against hybrids? No data erence in ff Yes, selection against marine immigrants operates in freshwaters; the reciprocal direction is not known Likely, river ecotypes acquire higher parasite load in lake exposure compared to lake ecotype, but no di reciprocal river exposure Yes, selection against marine immigrants operates in freshwaters; the reciprocal direction is not known Is there parasite- mediated selection against immigrants? Not known tion of the lake and the stream clade Not known Not known, but unlikely given the wide geographical distribu- Does parasite- driven divergence precede other divergence? Not known No data No data Divergence in MHC profiles, immuno- logical parameters and habitat specific resistance Freshwater residents are less susceptible than marines to infestation with freshwater parasites, the reverse is not known Divergence in resistance traits? Divergence in MHC profiles Freshwater residents are less susceptible than marines to infestation with freshwater parasites, the reverse is not know 1: Empirical data speaking to evidence for a role of parasites in ecological speciation. erent ff Table Yes, freshwater residents are heavily infested by a cestode Yes, lake populations harbor larger numbers of parasites compared to river populations Not known (but likely between the Sea and freshwaters) Divergence in parasite assemblage? Yes, limnetic species have much more cestodes, fewer mollusks and di trematodes No data, but very likely Lake and stream populations are no direct sister taxa but belong to geographically more widespread reciprocally monophyletic clades No data, but very likely Evidence for phylogenetic sister taxa? Yes in Paxton lake, no in Priest lake Parapatric populations, perhaps quite old Parapatric populations, perhaps species Reproductively isolated species Parapatric populations, perhaps species Three- spine stickleback parapatric lake and river populations in Schleswig Holstein (Germany) Three- spine stickleback marine versus freshwater species in Scotland Three- spine stickleback sympatric species pairs in Vancouver lakes (Canada) Three- spine stickleback marine versus freshwater populations on Vancouver coast (Canada) Taxon System Speciesstatus Fish International Journal of Ecology 9 , , 66 143 ], ] ] ] , ] , 57 59 10 67 60 [ 101 Selz et al. Manuscript [ 144 [ [ [ Refs Parapatric Parapatric Geographically parapatric Parapatric Geographically sympatric, ecologically parapatric The geography of divergence ects of ff Weakly so, and labile to environmental conditions Olfaction is not required to maintain assortative mating Indirect evidence. Olfactory plays a role in mate choice which could mediate e divergent MHC profiles Is divergent mate choice based on MHC- odours? Not tested Not known preference for male nuptial coloration, which appears to be a revealing signal at least in one of the species Female Is divergent mate choice based on a revealing signal? erent P. P. nyererei ff Yes, di MHC allele frequencies between the populations Divergent MHC profiles between the populations Yes, malesof pundamilia show bright blue coloration and males of show bright red coloration Not known Not tested Not tested Not known No tested Not tested Is there divergence in a potential revealing signal or MHC? Yes Yes Yes Yes Yes Is there assortative mating? Is there parasite- mediated selection against hybrids? 1: Continued. No data No data Is there parasite- mediated selection against immigrants? Table Not known No data No data Not known No data No data Not known but perhaps not likely given that other traits diverge very early in the process Does parasite- driven divergence precede other divergence? erent erent ff ff MHC profiles between the populations MHC profiles between the populations No data Di Di No data Not known No data No data No data Not known No data No data Divergence in resistance traits? erent erent ff ff erent ff parasite taxa, but data available only for the complete speciation stage in speciation continuum Yes Yes, di parasite taxa are abundant in di environments Yes, profundal ecotype harbours significantly fewer infections compared to littoral/pelagic ecotype Yes, divergence in infections corresponds with the divergence in diet of the ecotypes Divergence in parasite assemblage? Yes, di Probably not direct sister species Yes, from microsatellite and AFLP data No data No data Evidence for phylogenetic sister taxa? No data Reproductively isolated species with perhaps mild gene flow Complete speciation continuum, but parasites only studied at the complete speciation stage Conspecific parapatric populations Conspecific parapatric populations Parapatric populations P. Lake Malawi cichlids Pseudotro- pheus fainzilberi and emmiltos Lake Victoria cichlids Pundamilia pundamilia and Pundamilia nyererei Alpine charr in Norway Benthic and pelagic ecotypes of whitefish in two lakes in northern Norway Three- spine stickleback, marine versus freshwater populations in Atlantic Canada Taxon System Speciesstatus 10 International Journal of Ecology ] ] 75 145 ] , , ] 74 71 62 40 [ [ [ Refs [ Single population Geographically sympatric, ecologically parapatric Parapatric The geography of divergence Allopatric/ parapatric Not known Not tested Is divergent mate choice based on MHC- odours? MHC could act as a homogenizing mechanism counteracting speciation among the populations although mechanisms is not tested Not known Is divergent mate choice based on a revealing signal? It is based on song which might be a revealing signal Not tested erent ff No data No data Not tested Not tested Not known, but resident males sing a local dialect Is there divergence in a potential revealing signal or MHC? Yes, di MHC allele frequencies between the populations Yes, inferred from microsatellite Fst in sympatry Yes Is there assortative mating? Yes, partially No, hybrids had intermediate infection levels and immune responses compared to parent populations No, hybrids were at a selective advantage Is there parasite- mediated selection against hybrids? 1: Continued. No data No data Is there parasite- mediated selection against immigrants? Yes, immigrant individuals had higher infection rates and lower reproductive success Table erentiation ff Not known No data Does parasite- driven divergence precede other divergence? Not known but divergence of parasite assemblage increases with genetic di Not known No data No data Not known erent erence in ff ff No di immune responses between the species Divergence in resistance traits? No data Di MHC allele frequencies among the populations, individuals carrying certain allele had fewer parasites erent ff erences in erently ff ff infection rate between the species Yes, di parasite taxa Divergence in parasite assemblage? Di No data No data Not known Yes, populations were di infected with Gyrodactylus monogeneans Evidence for phylogenetic sister taxa? Probably, recently diverged species No data, but very likely Yes, from microsatellite and AFLP data No data erent ff Complete speciation (or speciation reversal) continuum Distinct sympatric species with a hybrid zone Conspecific populations rivers with no reproductive isolation Allopatric/ parapatric populations from di Collared and pied flycatchers Mountain white- crowned sparrows Whitefish in Swiss prealpine lakes Populations of guppies in Trinidad Bird Bird TaxonSystem Speciesstatus International Journal of Ecology 11

200 200 150 150 100 100

50 50 40 40 30 30 20 20 Total parasite load parasite Total Total parasite load parasite Total

40 40 30 30 20 20

10 10

5 5 Median parasite load parasite Median load parasite Median

0 10 20 30 40 50 0246810 Red score (% body coverage) Territory size (m2) Figure 5: Left: the cichlid fish sister species Pundamilia pundamilia and P. nyererei from Lake Victoria differ in their parasite assemblages, dominated by larval nematodes in the internal organs and ectoparasitic copepods on the gills, respectively (photo of P. pundamilia Ole Seehausen, photo of P. nyererei courtesy of Martine Maan, drawings courtesy of Jeanette Bieman). Right: males with bright red (shown) or bright blue (not shown) colouration and males with larger territories were found to carry lower parasite infections than males with duller coloration [66, 101] and females use male nuptial colouration in intra- and interspecific mate choice [102, 103]. Figure reproduced with permission from Maan et al. [101].

On the one hand, MHC-mediated mate selection is known Much of the empirical work on the interactions between to often favour outbreeding and associated disassortative parasitism and diversity of MHC genes has been conducted mating between and within host populations [69, 85]. MHC in contrasting infection environments. Recent progress has disassortative mating preferences can in principle increase been made especially in ecotype and species pairs of three- fitness of choosy parents because a disproportionate number spine sticklebacks in northern Germany and Canada. Lake of offspring would be high fitness MHC heterozygotes [86]. and river ecotypes of sticklebacks in Germany harbour There is also evidence to suggest that intermediate MHC significantly different parasite communities so that the diversity results in higher fitness for an individual since low lake populations are infected with a higher diversity of diversity allows some parasites to escape immune detection parasite species [54, 90]. This difference is known to be and very high diversity increases the risk of autoimmunity linked positively with the diversity of MHC genes; more [87]. Under such circumstances, divergent optimality of heavily infected lake populations show higher diversity MHC allele frequencies in contrasting environments that of MHC compared to river populations [54]. Moreover, differ in parasite exposure may lead to or facilitate eco- this is accompanied by variation in resistance between logical speciation through mate choice for well-adapted ecotypes where the less-infected river ecotype shows reduced MHC profiles in each environment [88, 89]. For example, immunocompetence [56]. Such between-habitat variation experimental work with stickleback suggests that female in the pools of MHC alleles suggests operation of parasite- sticklebacks use evolutionarily conserved structural features mediated selection, although it does neither imply divergent of MHC peptide ligands to evaluate MHC diversity of their selection, nor exclude simultaneous action of genetic drift prospective mating partners [88]. It should be noted here [54]. Somewhat contrasting results on MHC diversity in that MHC-mediated divergent sexual selection does not sticklebacks, however, come from Canadian benthic-limnetic require individuals to be infected with the parasite species species pairs. In this system, the limnetic ecotype carries that has driven the evolutionary divergence in MHC profiles, higher number of parasites, especially those species likely or with any other parasite. This is in contrast to situations to impose selection, than the benthic ecotype [58], but still of direct natural selection, for example, when the fitness of harbour fewer MHC alleles [91]. immigrants or hybrids is reduced by the actual infection Outside the stickleback systems, evidence for divergent (see above). However, suboptimal or superoptimal MHC MHC profiles in ecotypes or young sister species is scarce. profiles of hybrids could still cause higher parasite infection One of the few fish examples is the study by Blais et al. on and reduced fitness, adding to reproductive isolation through sympatric cichlid species in Lake Malawi [67]. These authors viability selection as suggested by theoretical models [89]. demonstrated high polymorphism in MHC, divergence in 12 International Journal of Ecology

MHC allele frequencies, and differences in parasite infections environmental conditions and are not necessarily underlying between these closely related fish species, supporting the the propensity towards assortative mating [57]. idea of parasite-mediated divergent selection. We expect Evidence for a counteracting role of MHC in host speci- that more tests will be conducted in ecotypes and young ation, on the other hand, comes from Trinidad guppies [40]. sister species of fish, and other host taxa in the near future. These authors found significantly lower divergence in MHC This will be important in establishing the generality of among guppy populations than expected from divergence relationships and unravelling the somewhat contradictory at neutral loci and concluded that stabilizing selection on results obtained in different systems even with the same MHC and its pleiotropic role in mate choice could act as host taxon. a homogenizing mechanism among the populations [40]. The above examples demonstrate heterogeneous MHC Evidence for stabilizing selection on some and divergent profiles between ecotypes and young species. However, very selection on other MHC loci was observed in Alpine trout few studies have taken the necessary next step towards testing populations adapting to steep thermal gradients [97]. It the MHC-pleiotropy speciation hypothesis by looking into is clear from these contrasting results that more examples MHC-mediated assortative mating among ecotypes. Again, from different systems are needed to address the generality experimental work here includes that in two stickleback of parasite-mediated divergent selection on MHC and the systems. Mate choice trials by Eizaguirre et al. [92]insem- role of MHC cues in mate choice among diverging host inatural enclosures revealed that female sticklebacks from populations. lake populations in northern Germany show preferences for males with an intermediate MHC diversity, and for males carrying an MHC haplotype that provides protection against 3.2.3. Tests of Parasite-Mediated Divergent Sexual Selection. a locally common parasite. Subsequently, the same authors Reproductive isolation in this scenario emerges by sexual extended this approach to lake and river ecotypes and found, selection for direct benefits (i.e., healthy mates) or for using a flow channel design, that females preferred the odour heritable fitness (i.e., parasite resistance) as an indirect of their sympatric males [54]. They concluded from these consequence of adaptation to different parasite challenges studies that parasite-induced divergent selection on MHC (Figure 2). In theory, divergent sexual selection is an effective diversity and for local adaptation could act as a mechanism mechanism of reproductive isolation [98], while parasites are of speciation through the pleiotropic role of MHC in mate considered mediators of mate choice [99]. Several aspects choice. However, actual assortative mating between the of parasitism may lead to population divergence under lake and stream populations remains to be demonstrated. these circumstances (discussed recently in [8]). For example, Other authors found in common garden experiments that infections are commonly related to habitat and diet, and they assortative mating between lake and stream sticklebacks may impose selection on signal design, maintain genetic variation often not evolve despite divergent selection on ecological and honesty of sexual signalling through the cost they impose traits ([93]; Ras¨ anen¨ et al. this volume [94]). However, it is onsignalproductionandmaintenance,aswellashavehealth nevertheless possible that such assortative mating occurs in consequences for their hosts which may filter down to mate nature owing to environmental influences. attraction and selection [8]. There are several alternative Assortative mating mediated by MHC has also been scenarios for how this may lead to population divergence, studied in saltwater versus freshwater sticklebacks in the but most of the studies so far have been conducted on St. Lawrence River in Canada [57], where the populations interactions between parasitism and mating preferences. differed significantly in the frequency of MHC alleles and These are related to the seminal paper by Hamilton and in the communities of helminth parasites. Strong signatures Zuk [99] on the tradeoffs between individuals’ ability to of natural selection on MHC genes were inferred in the resist parasite infections and produce extravagant sexual freshwater, but not in the marine population. Relationships ornamentation by which sexual selection for healthier mates between parasite load and MHC diversity were indicative of ensures heritable resistance to offspring. Such selection could balancing selection, but only within the freshwater popula- also promote speciation if ecological factors that lead to tion. The latter result is in accordance with other studies divergent parasite infections result in reproductive isolation on sticklebacks suggesting maximisation of host fitness at through selection for parasite resistance [8]. intermediate rather than maximal MHC diversity in some Empirical evidence comes from freshwater fishes. For environments [95, 96]. Mating trials found signals of MHC- example, Skarstein et al. [100] showed that individual mediated mate choice to be weak and significantly influenced Arctic charr (Salvelinus alpinus) within one population had by environmental conditions (salinity; [57]). By allowing marked differences in habitat and diet, and this correlated full mating contact to the fish, these authors demonstrated also with parasite infection and breeding colouration. Such differences between the ecotypes in the importance of variation could facilitate niche-specific adaptation in hosts MHC-mediated mate choice, and very strong environment and set the initial stages for speciation through divergent dependence where mating preferences with regard to MHC sexual selection. Similarly, in cichlid fish, males with bright were sometimes inversed depending on whether fish were red (Pundamilia nyererei)orbrightblue(P. pundamilia) tested in their own or a different salinity environment. The colouration were found to carry lower parasite infections authors concluded that MHC probably plays an impor- than males with duller coloration [66, 101]. At the same time, tant role when individuals evaluate prospective mates, but females use male nuptial colouration in intra- and interspe- that MHC-mediated mate choice decisions depend on the cific mate choice [102, 103](Figure 5). However, what role International Journal of Ecology 13 the divergent parasite infections have in speciation in this lake cichlid fish, and from sticklebacks, charr, and whitefish system is unclear because parasites have been investigated species in postglacial lakes. only in the well-advanced stage of speciation, and because the At the same time, studies should consider spatiotemporal differences in infections coincide with differentiation in diet, variation, or consistency, in parasite-induced selective pres- microhabitat, and visual system [66],whichsupportsthe sures. Investigating infections over replicated pairs of host idea that parasite-mediated divergent selection is often just populations could reveal if infection patterns are consistently one component of multifarious divergent selection between different, for example, among hosts inhabiting two distinct habitats [104]. Overall, it remains to be tested in all of the environments. Under such circumstances, infections could systems if the preference of females for more resistant males drive parallel ecological speciation in different populations actually results in production of offspring that are more resis- [106–108]. In reality, infections almost always differ to some tant to parasites found in each particular environment [8]. extent among populations in terms of species composition and infection intensities because of heterogeneities in local conditions for parasite transmission. For example, asso- 4. Future Directions: Missing Tests of ciations between host diet and infection from trophically Parasite-Mediated Divergent Selection transmitted parasites could show habitat-specific variation. However, if the “core” of parasites (including one or The empirical examples reviewed above demonstrate recent ff several species) that underlies selective pressures and host progress in identifying di erences in parasite infections divergence remains more or less the same, parallelism in host between host populations that occupy contrasting environ- divergence among different populations could be observed. ments and have diverged in phenotypic and genetic traits, ff Approaches that capture parallelism along a continuum of and in linking these di erences to assortative mating and host speciation could tackle not only the importance of speciation (Table 1). However, it is also clear that many more infections in speciation process per se, but the significance empirical tests of the role of parasite-mediated divergent of individual parasites species as well (see below). selection in these and other taxa are needed before the Similarly, temporal stability of infections on the time generality of some of the more trenchant findings can scale of host generations would indicate if the strength of be assessed. Next we will discuss three categories of tests selection is stable enough to result in divergence of host of parasite-mediated speciation that are currently lacking: populations. So far, the extent of spatiotemporal variation parasite-mediated divergent selection along a speciation or consistency is unknown in most systems that include continuum, the strength of selection versus multifariousness a limited number of host populations and/or a narrow of selection, and measuring the relative rates of adaptation in temporal window for observations (but see Knudsen et parasites and in hosts. al. [39]forlong-termdataonArcticcharrinNorway). Attempts to relate such temporal patterns to variation in 4.1. Divergent Selective Pressures along a Continuum of the progression of speciation have also been few [109]. In Speciation. We ought to understand at which point of the general, what is needed are long-term data on spatiotemporal speciation process parasite assemblages become sufficiently consistency of divergent parasite infections in replicated pairs divergent to reduce gene flow between host populations. of host populations that are at different stages of speciation. This is particularly true because divergent infections are This is a demanding task, but could help in answering usually associated with divergent habitat and/or diet, and fundamental questions such as whether speciation takes hence rarely come isolated from other sources of diver- place only in populations where costly infections do not gent selection. At present, it remains unknown whether fluctuate randomly. Most importantly, patterns of parasitism divergence in parasitism could itself initiate the divergence must be investigated in an integrated speciation perspective of host populations, or rather follows the divergence of alongside various other ecological and genetic factors some host populations initiated by other ecological factors. In of which would typically interrelate with parasitism. This the second case, it is unknown in most cases whether should include effective combinations of field surveys and divergence in parasitism is consequential or inconsequential experimental approaches to generate real insights into the to speciation. Answering these questions requires approaches question of how parasites affect the speciation process in that capture the entire continuum of speciation (see [104, relation to other factors initiating or promoting ecological 105]). This should include replicated populations occupying speciation. For example, divergent parasitism may initiate contrasting habitats, but showing no apparent divergence, speciation on its own right, or it may be a necessary force and extending to well-differentiated ecotypes and incipient in the transition from early stages of divergent adaptation species and eventually all the way to fully isolated sister to assortatively mating species, but it may also latch onto species [104, 105]. However, all empirical investigations of other mechanisms of divergent evolution without being parasite effects known to us have dealt each with just one consequential for the process of speciation. stage in the continuum of the speciation process, and in fact often with species that are already divergent in many different traits (Table 1). Nevertheless, there are several potential 4.2. Strength versus Multifariousness of Parasite-Mediated systems where the speciation continuum approach could be Selection. Divergent selection may act on a single trait, a applied. In fishes, the strongest candidate populations would few traits or many traits at the same time, and in each of most likely come from African and Central American great these cases selection may have a single or many sources. In 14 International Journal of Ecology a recent review, Nosil et al. [5] related this to hypotheses some tools to tackle the relative importance of strong and about “stronger selection” when the completion of speciation multifarious selection. depends on the strength of selection on a single trait, or “multifarious selection” when completion of speciation 4.3. Measuring Relative Rates of Adaptation in Parasites and is more likely driven by independent selection on several Hosts. Measuring relative rates of adaptation in parasites and traits [5]. A predominant feature of natural host-parasite their hosts could allow empirical testing of the theoretical interactions is that hosts harbour a community of parasite predictions about when parasite-host coevolution should species that are interconnected through the use of the facilitate and when it should actually constrain speciation common host as a resource, interspecific parasite-parasite in host populations. Theoretical considerations predict that interactions, and direct and indirect effects of host immunity parasite-host coevolution should facilitate speciation in host (e.g., [110, 111]). Parasite-mediated divergent selection can populations when host populations can adapt to the parasite vary between host populations both in strength of selection community that infects them. In this situation, gene flow imposed by one parasite species, in the number of different from nonadapted host populations could be maladaptive, parasite species, and in the number of traits or genes and assortative mating between host populations may evolve affected. In other words, it is essential to understand the by either of the three mechanisms reviewed above. On relative importance to speciation of shifts in the strength the other hand, when parasites adapt to their local host of selection exerted by one parasite or in the number population (reviews in [113–115]), gene flow into the host of parasite species together exerting multifarious selection. population from outside could be adaptive because it would However, strength and dimensionality of parasite-mediated provide genetic variants not known to the local parasites selection in the context of host divergence and speciation that escape parasitism. Parasite-host coevolution should is currently largely unknown. Empirical examples typically constrain speciation in host populations in this scenario [53]. report multiple parasite species infections in the diverging Given that most parasites have faster generation times than host populations, some of which show differences between their hosts, it seems that the second scenario could apply the populations. Hence, current data tend to suggest that quite often. Conditions favourable to host speciation would diverging ecotypes are often exposed to divergent parasite include situations where hosts are ahead of their parasites species assemblages (Table 1). The role of individual species in the coevolutionary race, for example, because parasite and especially their joint effects, however, are poorly known. adaptation is genetically constrained. They may also entail In sticklebacks, parasite diversity tends at least sometimes situations where host-parasite coevolution happens in one- to be positively associated with host diversity at MHC loci to-many or many-to-many constellations. In such situations, [54, 112] suggesting that multiple parasite species could be adapted hosts may cope well with the local community of driving host population divergence at multiple MHC loci. diverse parasites without coevolving closely with any one of However, typically some parasite species impose stronger them. Very few existing data speak to these theoretical predic- selection, for example, because they are more numerous, tions. One interesting corollary of the above is that the odds have larger body size, or infect a more vital organ of of speciation in hosts and those in parasites should often be the host. For example in German sticklebacks, resistance negatively correlated. Very high speciation rates in parasites against Gyrodactylus parasites is determined by a certain compared to their hosts have been described for example MHC haplotype in river populations, whereas that same in monogenean platyhelminthes infecting freshwater fishes haplotype when occurring in the lake population results [116]. To the extent that these parasites evolve host-specific in slightly higher parasite numbers [54](Figure 4). Given adaptations, they might facilitate outbreeding and constrain that MHC-mediated resistance correlates with reproductive speciation in host populations. success [92], the opposite outcomes of infection in different environments could facilitate speciation [54]. This supports the role of this individual parasite species in divergence of the 5. Reversal of Speciation host populations. However, variable outcomes of selection in different combinations of parasite species are nevertheless Some of the best examples of adaptive radiation come likely; one species could drive the selection on its own in from freshwater fishes, and in several of these the frequent some populations favouring the idea of strong selection, reversal of speciation has also been described particularly while effects from multiple species predominate in other as a consequence of human activity. The most compelling systems supporting multifarious selection. It is also generally evidence comes from cichlid fishes in Lake Victoria, stickle- unknown if selection imposed by different parasite species back in Western Canada, and whitefish and ciscoes in central takes the same direction with regard to fitness effects of Europe and North America [117, 118]. Rapid adaptive migration and gene flow, and hence speciation. For example, radiation in Lake Victoria has produced a magnificent diver- Keller et al. [97] found evidence for stabilizing selection sity of hundreds of cichlid species. One likely mechanism on one and divergent selection of another locus in the driving speciation involves evolution in the visual system MHC gene family in Alpine trouts. On the scale of entire and of visual signals in response to heterogeneous light parasite communities, such interactions may be complex and conditions, and its effects on mate choice and speciation difficult to resolve. However, search for recurring patterns through mechanisms related to sensory drive [10]. However, of infections and resistance in replicated pairs of similar increasing eutrophication of the lake from the 1920s and ecotypes, or along a speciation continuum, could provide subsequent turbidity of the water have resulted in relaxation International Journal of Ecology 15 of the diversity-maintaining mate selection and collapse of be needed to link the probability of infection with the degree species diversity in some parts of the lake [119]. Similar of host population divergence observed in nature. collapses of species diversity through loss of reproductive Also other infections could lead to alterations in mate isolation following anthropogenic impacts have occurred in choice and sexual selection, possibly resulting in reduced ciscoes of the Laurentian Great lakes, in whitefish of Swiss probability of host divergence or in reversal of speciation. prealpine lakes [118], and in Canadian sticklebacks [120]. For example, several parasite species alter the behaviour or appearance of their hosts as a side effect of infection or 5.1. Role for Parasites? The above observations relate to by actively manipulating the host to improve transmission the hypothesis of parasite-mediated speciation in two ways. [21, 23]. Depending on the prevalence of such infections First, investigating mechanisms and processes of speciation in a host population, this could dramatically change sexual reversal may also shed light onto those promoting ecological selection. For example, it has been suggested that effects speciation. In general, reversal of speciation may (re)create of Schistocephalus infection on the growth of sticklebacks a speciation continuum, where formerly distinct sympatric directly affect assortative mating which is based at least partly species become admixed to variable degrees, with variation on size [126]. Coinfections with several parasite species that in time, space, or both. Capturing and studying such changes commonly prevail in wild hosts may also change circum- as they happen in nature could provide effective tools to stances for sexual selection in hosts. For example, whereas evaluate the possible role of parasites in divergence of natural some parasite species use hosts as transmission vehicles host populations. For example, environmental changes such to the next host by manipulating their behaviour, other as pollution and eutrophication may affect not only the species may use the same host individuals for reproduction host community, but could change biomass and reduce the and completion of the life cycle. Two parasite species with diversity of parasite species as well, with species having complex life cycles can also use the same host individual as a multiple host life cycles probably being among the most vehicle to different definitive host taxa, such as fish and bird. prone to extinction [47, 121]. Changes that relax parasite- These situations can result in conflicts between the opposing mediated divergent selection among host populations could parasite interests (e.g., [127, 128]), when the outcome for lead to gradual loss of reproductive isolation, but also reveal host behaviour and sexual selection may depend on which the role of specific parasite species that had maintained the of the parasite species dominates the coinfection situation. divergence. Additionally, experimental work is also needed The important point emerging from this is that the variety to address the possible role of parasites in breaking down of interrelated mechanisms by which parasitic infections speciation or preventing it, something that is known very can reduce host fitness (e.g., depletion of energy, changes little about. in appearance or behaviour, increase in susceptibility to Second, it is possible that certain parasite species may predation) can set up very different conditions for sexual directly drive reversal or prevention of speciation. In systems selection depending on how common different types of where divergence of species is maintained by divergent (co)infections are. adaptation in the visual system, and visually mediated sexual selection, such as African cichlids and sticklebacks, parasites 5.2. Speciation by Hybridization. Hybridisation between that influence host vision might have effects comparable to species can also give rise to new species [129–131]. Hybrid loss of visibility due to eutrophication. One group of such speciation is particularly likely when the hybrid population parasites are those of the genus Diplostomum that infect is able to colonize a niche that is not occupied by either eyes of a range of freshwater fish species around the world parental species and makes it spatially isolated from both causing cataracts and partial or total blindness (Figure 1; the parental species [132, 133]. A scenario that would be [122]). Infection has dramatic effects on fish [123–125]and relevant in the context of our review is if hybrids are able to it could also impair the ability of fish to visually select mates. colonize a novel niche where infections differ from those in For example, females with impaired visual ability could be the parental niches. In principle it is then even possible that less choosy as they cannot properly assess the quality of the parasite-mediated reversal of speciation, happening in parts male colouration or courtship. Impaired vision of males, of the larger range of the parental species, could lead to the on the other hand, could affect their ability to compete local emergence of a new hybrid species. There currently is a with other males, court females or, in case of sticklebacks, growing number of empirical examples of hybrid speciation build nests. The risk of eye fluke infection is typically both from plant and animal systems (reviewed in [131, 134]), variable in both space and time, which can result even in and it will be interesting to see if future studies will pick up total absence of infection from some lakes (Karvonen et al., a signature of parasite-mediated selection in some of these unpublished). This could set up very different conditions for cases. host divergence mediated by vision-based mate choice within allopatric host populations. For example, it is tempting to ask 5.3. Parasitism, Phenotypic Plasticity, and Speciation. The if divergence of the hosts under such circumstances would be above scenarios of infections that influence the condition or more likely in populations that are only moderately, or not behaviour of their host may also be relevant in the context at all, infected with eye flukes, conceptually linking this to of old and recent discussions on the role of plasticity in the effect of eutrophication on the loss of diversity of cichlid generating reproductive isolation that precedes any adaptive fishes in Lake Victoria [119]. Comparative experimental divergence (e.g., [135, 136]; Fitzpatrick, this issue [137]). In approaches could shed light onto these associations and will host-parasite interactions, such plasticity could be observed, 16 International Journal of Ecology for example, if reduction or change in host condition [5] P. Nosil, L. J. Harmon, and O. Seehausen, “Ecological or behaviour as a result of infection leads to condition- explanations for (incomplete) speciation,” Trends in Ecology dependent habitat choice and thus promotes reproductive and Evolution, vol. 24, no. 3, pp. 145–156, 2009. isolation between individuals (see [138]). Phenotypic vari- [6] D. Schluter, “Evidence for ecological speciation and its ation in individual’s ability to avoid infections by shifting alternative,” Science, vol. 323, no. 5915, pp. 737–741, 2009. to another habitat away from the infection source (e.g., [7] J. M. Sobel, G. F. Chen, L. R. Watt, and D. W. Schemske, “The biology of speciation,” Evolution, vol. 64, no. 2, pp. 295–315, [139]) could also contribute to segregation. Moreover, such 2010. behaviours could be further shaped or reinforced by the [8]M.E.MaanandO.Seehausen,“Ecology,sexualselectionand earlier infection experience and “learning” of an individual speciation,” Ecology Letters, vol. 14, no. 6, pp. 591–602, 2011. [140, 141]. Overall, there are several ways how plastic [9] P. F. Colosimo, K. E. Hosemann, S. Balabhadra et al., responses of hosts to a parasite infection could contribute to “Widespread parallel evolution in sticklebacks by repeated reproductive isolation, but they wait for empirical tests. fixation of ectodysplasin alleles,” Science, vol. 307, no. 5717, pp. 1928–1933, 2005. [10] O. Seehausen, Y. Terai, I. S. Magalhaes et al., “Speciation 6. Conclusions through sensory drive in cichlid fish,” Nature, vol. 455, no. 7213, pp. 620–627, 2008. Empirical evidence available today has just begun to unravel [11] D. Schluter and G. L. Conte, “Genetics and ecological mechanisms of parasite-mediated selection and how these speciation,” Proceedings of the National Academy of Sciences affect the course of ecological speciation. Evidence for direct of the United States of America, vol. 106, pp. 9955–9962, 2009. natural selection is equivocal because reduced viability or [12] C. Salazar, S. W. Baxter, C. Pardo-Diaz et al., “Genetic fecundity of immigrants may be compensated by hybrid evidence for hybrid trait speciation in Heliconius butterflies,” vigour. Evidence for pleiotropy through effects of MHC PLoS Genetics, vol. 6, no. 4, article e1000930, 2010. on assortative mating is mixed too because divergent and [13] P. Nosil and D. Schluter, “The genes underlying the process stabilizing selection both occur, sometimes even in the same of speciation,” Trends in Ecology and Evolution, vol. 26, no. 4, populations, and because MHC may not be an overriding pp. 160–167, 2011. mate choice cue. Evidence for parasite-mediated divergent [14] D. Schluter, The Ecology of Adaptive Radiation,Oxford University Press, 2000. sexual selection is scarce and incomplete, with best examples [15] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology perhaps from Arctic charr and cichlid fish. The future Letters, vol. 8, no. 3, pp. 336–352, 2005. should see more detailed investigations of parasitism and [16] T. E. Reimchen, “Living in the ecological theatre,” in host resistance at all stages of speciation. Importantly, Tinbergen’s Legacy in Behaviour. Sixty Years of Landmark such investigations must become part and parcel of an Stickleback Papers, F. A. von Hippel, Ed., pp. 385–392, Brill, integrated analytical and experimental research framework The Netherlands, 2010. on ecological speciation. Particular emphasis should be [17] S. M. Vamosi, “The presence of other fish species affects placed on studying the entire speciation continuum and not speciation in threespine sticklebacks,” Evolutionary Ecology just the beginning and end. This is the only way to determine Research, vol. 5, no. 5, pp. 717–730, 2003. whether, how and at what stage parasites begin to influence [18] P. Nosil and B. J. Crespi, “Experimental evidence that preda- a divergence process that actually has ecological speciation as tion promotes divergence in adaptive radiation,” Proceedings of the National Academy of Sciences of the United States of its end product. Further, empirical testing of the theoretical America, vol. 103, no. 24, pp. 9090–9095, 2006. predictions about when parasite-host coevolution should [19] R. Poulin, “How many parasite species are there: are we close facilitate and when it should constrain speciation in host to answers?” International Journal for Parasitology, vol. 26, no. populations has great potential to make major contributions 10, pp. 1127–1129, 1996. to an integration of the currently still disparate literature. [20] D. A. Windsor, “Most of the species on Earth are parasites,” Clearly expectations should depend on the relative rates of International Journal for Parasitology, vol. 28, no. 12, pp. and constrains to evolutionary adaptation in parasites and 1939–1941, 1998. their hosts, and it is likely that the current lack of data on [21] R. Poulin, “’Adaptive’ changes in the behaviour of parasitized this explains some of the variable and contradictory results of animals: a critical review,” International Journal for Parasitol- empirical studies. Finally, many more speciating taxa ought ogy, vol. 25, no. 12, pp. 1371–1383, 1995. to be studied to identify generalities. [22]P.J.Hudson,A.P.Dobson,andD.Newborn,“Preventionof population cycles by parasite removal,” Science, vol. 282, no. 5397, pp. 2256–2258, 1998. References [23] J. Moore, Parasites and the Behaviour of Animals,Oxford University Press, 2002. [1] G. G. Simpson, The Major Features of Evolution, Columbia [24] J. Jokela, M. F. Dybdahl, and C. M. Lively, “The maintenance University Press, New York, NY, USA, 1953. of sex, clonal dynamics, and host-parasite coevolution in a [2] D. Jablonski, “Biotic interactions and macroevolution: exten- mixed population of sexual and asexual snails,” American sions and mismatches across scales and levels,” Evolution, vol. Naturalist, vol. 174, no. 1, pp. S43–S53, 2009. 62, no. 4, pp. 715–739, 2008. [25] J. B. S. Haldane, “Disease and evolution,” La Ricerca Scien- [3] D. Otte and J. A. Endler, Speciation and Its Consequences, tifica, vol. 19, pp. 68–76, 1949. Sinauer Associates, Sunderland, Mass, USA, 1989. [26] P. W. Price, M. Westoby, B. Rice et al., “Parasite mediation [4]J.A.CoyneandH.A.Orr,Speciation, Sinauer Associates, in ecological interactions,” Annual Review of Ecology and Sunderland, Mass, USA, 2004. Systematics, vol. 17, pp. 487–505, 1986. International Journal of Ecology 17

[27] W. J. Freeland, “Parasites and the coexistence of animal host Proceedings of the National Academy of Sciences of the United species,” American Naturalist, vol. 121, no. 2, pp. 223–236, States of America, vol. 107, no. 46, pp. 20051–20056, 2010. 1983. [44] A. Buckling and P. B. Rainey, “The role of parasites in [28] M. D. Pagel, R. M. May, and A. R. Collie, “Ecological aspects sympatric and allopatric host diversification,” Nature, vol. of the geographical distribution and diversity of mammalian 420, no. 6915, pp. 496–499, 2002. species,” American Naturalist, vol. 137, no. 6, pp. 791–815, [45] M. A. Brockhurst, P. B. Rainey, and A. Buckling, “The effect 1991. of spatial heterogeneity and parasites on the evolution of host [29] B. R. Krasnov, G. I. Shenbrot, I. S. Khokhlova, and A. A. diversity,” Proceedings of the Royal Society B, vol. 271, no. Degen, “Relationship between host diversity and parasite 1534, pp. 107–111, 2004. diversity: flea assemblages on small mammals,” Journal of [46]G.W.Esch,A.O.Bush,andJ.M.Aho,Eds.,Parasite Biogeography, vol. 31, no. 11, pp. 1857–1866, 2004. Communities: Patterns and Processes, Chapman and Hall, [30] C. L. Nunn, S. Altizer, W. Sechrest, K. E. Jones, R. A. London, UK, 1990. Barton, and J. L. Gittleman, “Parasites and the evolutionary [47] E. T. Valtonen, J. C. Holmes, and M. Koskivaara, “Eutroph- diversification of primate clades,” American Naturalist, vol. ication, pollution, and fragmentation: effects on parasite 164, no. 5, pp. S90–S103, 2004. communities in roach (Rutilus rutilus)andperch(Perca [31] C. L. Fincher and R. Thornhill, “A parasite-driven wedge: fluviatilis) in four lakes in central Finland,” Canadian Journal infectious diseases may explain language and other biodiver- of Fisheries and Aquatic Sciences, vol. 54, no. 3, pp. 572–585, sity,” Oikos, vol. 117, no. 9, pp. 1289–1297, 2008. 1997. [32] J. B. Yoder and S. L. Nuismer, “When does coevolution [48] R. Poulin, Evolutionary Ecology of Parasites, Princeton Uni- promote diversification?” American Naturalist, vol. 176, no. versity Press, NJ, USA, 2007. 6, pp. 802–817, 2010. [49] R. Poulin, “The decay of similarity with geographical dis- [33] P.R. Ehrlich and P.H. Raven, “Butterflies and plants—a study tance in parasite communities of vertebrate hosts,” Journal in coevolution,” Evolution, vol. 18, pp. 586–608, 1964. of Biogeography, vol. 30, no. 10, pp. 1609–1615, 2003. [50] J. Soininen, R. McDonald, and H. Hillebrand, “The distance [34] M. Bueno-Silva, W. A. Boeger, and M. R. Pie, “Choice decay of similarity in ecological communities,” Ecography, matters: incipient speciation in Gyrodactylus corydori vol. 30, no. 1, pp. 3–12, 2007. (Monogenoidea: Gyrodactylidae),” International Journal for Parasitology, vol. 41, no. 6, pp. 657–667, 2011. [51] A. Karvonen, G. H. Cheng, and E. T. Valtonen, “Within-lake dynamics in the similarity of parasite assemblages of perch [35] N. Nishimura, D. C. Heins, R. O. Andersen, I. Barber, and W. (Perca fluviatilis),” Parasitology, vol. 131, no. 6, pp. 817–823, A. Cresko, “Distinct lineages of Schistocephalus parasites in 2005. threespine and ninespine stickleback hosts revealed by DNA sequence analysis,” PLoS One, vol. 6, no. 7, article e22505, [52] C. Eizaguirre and T. L. Lenz, “Major histocompatability 2011. complex polymorphism: dynamics and consequences of parasite-mediated local adaptation in fishes,” Journal of Fish [36] I. Emelianov, A. Hernandes-Lopez, M. Torrence, and N. Biology, vol. 77, no. 9, pp. 2023–2047, 2010. Watts, “Fusion-fission experiments in Aphidius: evolution- [53] K. Summers, S. McKeon, J. Sellars et al., “Parasitic exploita- ary split without isolation in response to environmental tion as an engine of diversity,” Biological Reviews of the bimodality,” Heredity, vol. 106, pp. 798–807, 2010. Cambridge Philosophical Society, vol. 78, no. 4, pp. 639–675, [37] R. Knudsen, P. A. Amundsen, and A. Klemetsen, “Arctic 2003. charr in sympatry with burbot: ecological and evolutionary [54] C. Eizaguirre, T. L. Lenz, R. D. Sommerfeld, C. Harrod, M. consequences,” Hydrobiologia, vol. 650, no. 1, pp. 43–54, Kalbe, and M. Milinski, “Parasite diversity, patterns of MHC 2010. II variation and olfactory based mate choice in diverging [38] R. Knudsen, R. Primicerio, P. A. Amundsen, and A. Klemet- three-spined stickleback ecotypes,” Evolutionary Ecology, vol. sen, “Temporal stability of individual feeding specialization 25, pp. 605–622, 2010. may promote speciation,” Journal of Animal Ecology, vol. 79, [55] G. Rauch, M. Kalbe, and T. B. H. Reusch, “Relative impor- no. 1, pp. 161–168, 2010. tance of MHC and genetic background for parasite load in a [39] R. Knudsen, A. Siwertsson, C. E. Adams, M. Garduno-Paz,˜ J. field experiment,” Evolutionary Ecology Research, vol. 8, no. 2, Newton, and P. A. Amundsen, “Temporal stability of niche pp. 373–386, 2006. use exposes sympatric Arctic charr to alternative selection [56] J. P. Scharsack, M. Kalbe, C. Harrod, and G. Rauch, “Habitat- pressures,” Evolutionary Ecology, vol. 25, pp. 589–604, 2010. specific adaptation of immune responses of stickleback [40] B. A. Fraser and B. D. Neff, “Parasite mediated homogenizing (Gastetosteus aculeatus) lake and river ecotypes,” Proceedings selection at the MHC in guppies,” Genetica, vol. 138, no. 2, of the Royal Society B, vol. 274, no. 1617, pp. 1523–1532, 2007. pp. 273–278, 2010. [57] R. J.S. McCairns, S. Bourget, and L. Bernatchez, “Putative [41] R. E. Ricklefs, “Host-pathogen coevolution, secondary sym- causes and consequences of MHC variation within and patry and species diversification,” Philosophical Transactions between locally adapted stickleback demes,” Molecular Ecol- of the Royal Society B, vol. 365, no. 1543, pp. 1139–1147, 2010. ogy, vol. 20, no. 3, pp. 486–502, 2011. [42] M. D. Sorenson and R. B. Payne, “A single ancient origin of [58] A. D. C. MacColl, “Parasite burdens differ between sympatric brood parasitism in African finches: implications for host- three-spined stickleback species,” Ecography, vol. 32, no. 1, parasite coevolution,” Evolution, vol. 55, no. 12, pp. 2550– pp. 153–160, 2009. 2567, 2001. [59]R.Knudsen,R.Kristoffersen, and P. A. Amundsen, “Parasite [43] G. Sharon, D. Segal, J. M. Ringo, A. Hefetz, I. Zilber- communities in two sympatric morphs of Arctic charr, Rosenberg, and E. Rosenberg, “Commensal bacteria play Salvelinus alpinus (L.), in northern Norway,” Canadian a role in mating preference of Drosophila melanogaster,” Journal of Zoology, vol. 75, no. 12, pp. 2003–2009, 1997. 18 International Journal of Ecology

[60] R. Knudsen, P. A. Amundsen, and A. Klemetsen, “Inter- and [77] A. B. Marr, L. F. Keller, and P. Arcese, “Heterosis and out- intra-morph patterns in helminth communities of sympatric breeding depression in descendants of natural immigrants to whitefish morphs,” Journal of Fish Biology,vol.62,no.4,pp. an inbred population of song sparrows (Melospiza melodia),” 847–859, 2003. Evolution, vol. 56, no. 1, pp. 131–142, 2002. [61] F.Frandsen,H.J.Malmquist,andS.S.Snorrason,“Ecological [78] J. Wolinska, B. Keller, K. Bittner, S. Lass, and P. Spaak, parasitology of polymorphic Arctic charr, Salvelinus alpinus “Do parasites lower Daphnia hybrid fitness?” Limnology and (L.), in Thingvallavatn, Iceland,” Journal of Fish Biology, vol. Oceanography, vol. 49, no. 4, pp. 1401–1407, 2004. 34, no. 2, pp. 281–297, 1989. [79] S. L. Nuismer and S. Gandon, “Moving beyond common- [62] A. Karvonen, B. Lundsgaard-Hansen, J. Jokela, and O. garden and transplant designs: insight into the causes of local Seehausen, “Differentiation in parasitism among ecotypes of adaptation in species interactions,” American Naturalist, vol. whitefish segregating along depth gradients,” Oikos. In press. 171, no. 5, pp. 658–668, 2008. [63] T. D. Kocher, “Adaptive evolution and explosive speciation: [80] S. Gandon and S. L. Nuismer, “Interactions between genetic the cichlid fish model,” Nature Reviews Genetics, vol. 5, no. 4, drift, gene flow, and selection mosaics drive parasite local pp. 288–298, 2004. adaptation,” American Naturalist, vol. 173, no. 2, pp. 212– [64] M. J. Genner and G. F. Turner, “The mbuna cichlids of Lake 224, 2009. Malawi: a model for rapid speciation and adaptive radiation,” [81] O. Seppal¨ a,¨ A. Karvonen, E. T. Valtonen, and J. Jokela, “Inter- Fish and Fisheries, vol. 6, no. 1, pp. 1–34, 2005. actions among co-infecting parasite species: a mechanism [65] O. Seehausen, “African cichlid fish: a model system in maintaining genetic variation in parasites?” Proceedings of the adaptive radiation research,” Proceedings of the Royal Society Royal Society B, vol. 276, no. 1657, pp. 691–697, 2009. B, vol. 273, no. 1597, pp. 1987–1998, 2006. [82] O. Seppal¨ a,¨ A. Karvonen, M. Haataja, M. Kuosa, and J. Jokela, [66] M. E. Maan, A. M. C. Van Rooijen, J. J. M. Van Alphen, and O. “Food makes you a target: disentangling genetic, physio- ff Seehausen, “Parasite-mediated sexual selection and species logical, and behavioral e ects determining susceptibility to divergence in Lake Victoria cichlid fish,” Biological Journal of infection,” Evolution, vol. 65, no. 5, pp. 1367–1375, 2011. the Linnean Society, vol. 94, no. 1, pp. 53–60, 2008. [83] S. L. Nuismer, S. P. Otto, and F. Blanquart, “When do [67] J. Blais, C. Rico, C. van Oosterhout, J. Cable, G. F. Turner, and host-parasite interactions drive the evolution of non-random L. Bernatchez, “MHC adaptive divergence between closely mating?” Ecology Letters, vol. 11, no. 9, pp. 937–946, 2008. related and sympatric African cichlids,” PLoS One, vol. 2, no. [84] G. Valkiunas, D. Santiago-Alarcon, I. I. Levin, T. A. Iezhova, 8, article no. e734, 2007. andP.G.Parker,“AnewHaemoproteus species (Haemo- sporida: Haemoproteidae) from the endemic galapagos [68] P. Nosil, T. H. Vines, and D. J. Funk, “Perspective: reproduc- dove Zenaida galapagoensis, with remarks on the parasite tive isolation caused by natural selection against immigrants distribution, vectors, and molecular diagnostics,” Journal of from divergent habitats,” Evolution, vol. 59, no. 4, pp. 705– Parasitology, vol. 96, no. 4, pp. 783–792, 2010. 719, 2005. [85] C. Landry, D. Garant, P. Duchesne, and L. Bernatchez, [69] M. Milinski, “The major histocompatibility complex, sexual “’Good genes as heterozygosity’: the major histocompati- selection, and mate choice,” Annual Review of Ecology, bility complex and mate choice in Atlantic salmon (Salmo Evolution, and Systematics, vol. 37, pp. 159–186, 2006. salar),” Proceedings of the Royal Society B, vol. 268, no. 1473, [70] A. D.C. MacColl and S. M. Chapman, “Parasites can pp. 1279–1285, 2001. cause selection against migrants following dispersal between [86] L. Bernatchez and C. Landry, “MHC studies in nonmodel environments,” Functional Ecology, vol. 24, no. 4, pp. 847– vertebrates: what have we learned about natural selection in 856, 2010. 15 years?” Journal of Evolutionary Biology,vol.16,no.3,pp. [71] E. A. MacDougall-Shackleton, E. P. Derryberry, and T. Hahn, 363–377, 2003. “Nonlocal male mountain white-crowned sparrows have [87] B. Woelfing, A. Traulsen, M. Milinski, and T. Boehm, “Does lower paternity and higher parasite loads than males signing intra-individual major histocompatibility complex diversity local dialect,” Behavioural Ecology, vol. 13, pp. 682–689, 2002. keep a golden mean?” Philosophical Transactions of the Royal [72] R. S. Fritz, C. Moulia, and G. Newcombe, “Resistance of Society B, vol. 364, no. 1513, pp. 117–128, 2009. hybrid plants and animals to herbivores, pathogens, and [88] M. Milinski, S. Griffiths, K. M. Wegner, T. B. H. Reusch, A. parasites,” Annual Review of Ecology and Systematics, vol. 30, Haas-Assenbaum, and T. Boehm, “Mate choice decisions of pp. 565–591, 1999. stickleback females predictably modified by MHC peptide [73] C. Moulia, “Parasitism of plant and animal hybrids: are facts ligands,” Proceedings of the National Academy of Sciences of and fates the same?” Ecology, vol. 80, no. 2, pp. 392–406, the United States of America, vol. 102, no. 12, pp. 4414–4418, 1999. 2005. [74] C. Wiley, A. Qvarnstrom,¨ and L. Gustafsson, “Effects of [89] C. Eizaguirre, T. L. Lenz, A. Traulsen, and M. Milinski, hybridization on the immunity of collared ficedula albicollis “Speciation accelerated and stabilized by pleiotropic major and pied flycatchers f. hypoleuca, and their infection by histocompatibility complex immunogenes,” Ecology Letters, haemosporidians,” JournalofAvianBiology,vol.40,no.4,pp. vol. 12, no. 1, pp. 5–12, 2009. 352–357, 2009. [90] M. Kalbe, K. M. Wegner, and T. B. H. Reusch, “Dispersion [75] E. A. MacDougall-Shackleton, E. P. Derryberry, J. Foufopou- patterns of parasites in 0+ year three-spined sticklebacks: a los, A. P. Dobson, and T. P. Hahn, “Parasite-mediated cross population comparison,” Journal of Fish Biology, vol. 60, heterozygote advantage in an outbred songbird population,” no. 6, pp. 1529–1542, 2002. Biology Letters, vol. 1, no. 1, pp. 105–107, 2005. [91] B. Matthews, L. J. Harmon, L. M’Gonigle, K. B. Marchinko, [76]M.A.Patten,J.T.Rotenberry,andM.Zuk,“Habitatselec- and H. Schaschl, “Sympatric and allopatric divergence of tion, acoustic adaptation, and the evolution of reproductive MHC genes in threespine stickleback,” PLoS One, vol. 5, no. isolation,” Evolution, vol. 58, no. 10, pp. 2144–2155, 2004. 6, Article ID e10948, 2010. International Journal of Ecology 19

[92] C. Eizaguirre, S. E. Yeates, T. L. Lenz, M. Kalbe, and [108] K. L. Ostevik, B. T. Moyers, G. L. Owens, and L. H. Riese- M. Milinski, “MHC-based mate choice combines good berg, “Parallel ecological speciation in plants?” International genes and maintenance of MHC polymorphism,” Molecular Journal of Ecology, vol. 2012, Article ID 939862, 2012. Ecology, vol. 18, no. 15, pp. 3316–3329, 2009. [109] B. A. Fraser, I. W. Ramnarine, and B. D. Neff,“Temporal [93] J. A. M. Raeymaekers, M. Boisjoly, L. Delaire, D. Berner, K. variation at the MHC class IIb in wild populations of the Ras¨ anen,¨ and A. P. Hendry, “Testing for mating isolation guppy (Poecilia reticulata),” Evolution, vol. 64, no. 7, pp. between ecotypes: laboratory experiments with lake, stream 2086–2096, 2010. and hybrid stickleback,” Journal of Evolutionary Biology, vol. [110] R. Poulin, “Interactions between species and the structure of 23, no. 12, pp. 2694–2708, 2010. helminth communities,” Parasitology, vol. 122, pp. S3–S11, [94] K. Ras¨ anen,¨ M. Delcourt, L. J. Chapman, and A. P. Hendry, 2001. “Divergent selection and then not: the puzzle of reproductive [111] A. Karvonen, O. Seppal¨ a,¨ and E. Tellervo Valtonen, “Host isolationin Misty lake and stream stickleback,” International immunization shapes interspecific associations in trematode Journal of Ecology. In press. parasites,” Journal of Animal Ecology, vol. 78, no. 5, pp. 945– [95]K.M.Wegner,M.Kalbe,J.Kurtz,T.B.H.Reusch,andM. 952, 2009. Milinski, “Parasite selection for immunogenetic optimality,” [112] K. M. Wegner, T. B. H. Reusch, and M. Kalbe, “Multiple Science, vol. 301, no. 5638, p. 1343, 2003. parasites are driving major histocompatibility complex polymorphism in the wild,” Journal of Evolutionary Biology, vol. [96] J. Kurtz, M. Kalbe, P. B. Aeschlimann et al., “Major histo- 16, no. 2, pp. 224–232, 2003. compatibility complex diversity influences parasite resistance [113] O. Kaltz and J. A. Shykoff, “Local adaptation in host-parasite and innate immunity in sticklebacks,” Proceedings of the Royal systems,” Heredity, vol. 81, no. 4, pp. 361–370, 1998. Society B, vol. 271, no. 1535, pp. 197–204, 2004. [114] T. J. Kawecki and D. Ebert, “Conceptual issues in local [97] I. Keller, A. Taverna, and O. Seehausen, “Evidence of neutral adaptation,” Ecology Letters, vol. 7, no. 12, pp. 1225–1241, and adaptive genetic divergence between European trout 2004. populations sampled along altitudinal gradients,” Molecular [115] M. A. Greischar and B. Koskella, “A synthesis of experimental Ecology, vol. 20, no. 9, pp. 1888–1904, 2011. work on parasite local adaptation,” Ecology Letters, vol. 10, [98] M. G. Ritchie, “Sexual selection and speciation,” Annual no. 5, pp. 418–434, 2007. Review of Ecology, Evolution, and Systematics, vol. 38, pp. 79– [116] M. S. Zietara and J. Lumme, “Speciation by host switch 102, 2007. and adaptive radiation in a fish parasite genus Gyrodactylus [99]W.D.HamiltonandM.Zuk,“Heritabletruefitnessand (Monogenea, Gyrodactylidae),” Evolution, vol. 56, no. 12, pp. bright birds—a role for parasites?” Science, vol. 218, no. 4570, 2445–2458, 2002. pp. 384–387, 1982. [117] O. Seehausen, “Conservation: losing biodiversity by reverse [100] F. Skarstein, I. Folstad, and H. P. Rønning, “Spawning speciation,” Current Biology, vol. 16, no. 9, pp. R334–R337, colouration, parasites and habitat selection in Salvelinus 2006. alpinus: initiating speciation by sexual selection?” Journal of [118] P. Vonlanthen, D. Bittner, A. G. Hudson et al., “Eutrophi- Fish Biology, vol. 67, no. 4, pp. 969–980, 2005. cation causes speciation reversal in whitefish adaptive radi- [101] M. E. Maan, M. Van Der Spoel, P. Q. Jimenez, J. J. M. ations,” Nature, vol. 482, no. 7385, pp. 357–362, 2012. Van Alphen, and O. Seehausen, “Fitness correlates of male [119] O. Seehausen, J. J. M. Van Alphen, and F. Witte, “Cichlid coloration in a Lake Victoria cichlid fish,” Behavioral Ecology, fish diversity threatened by eutrophication that curbs sexual vol. 17, no. 5, pp. 691–699, 2006. selection,” Science, vol. 277, no. 5333, pp. 1808–1811, 1997. [102] O. Seehausen and J. J. M. Van Alphen, “The effect of [120] E. B. Taylor, J. W. Boughman, M. Groenenboom, M. male coloration on female mate choice in closely related Sniatynski, D. Schluter, and J. L. Gow, “Speciation in reverse: Lake Victoria cichlids (Haplochromis nyererei complex),” morphological and genetic evidence of the collapse of Behavioral Ecology and Sociobiology, vol. 42, no. 1, pp. 1–8, a three-spined stickleback (Gasterosteus aculeatus)species 1998. pair,” Molecular Ecology, vol. 15, no. 2, pp. 343–355, 2006. [121] D. J. Marcogliese, “Implications of climate change for [103] M. E. Maan, O. Seehausen, L. Soderberg¨ et al., “Intraspecific parasitism of animals in the aquatic environment,” Canadian sexual selection on a speciation trait, male coloration, in the Journal of Zoology, vol. 79, no. 8, pp. 1331–1352, 2001. Lake Victoria cichlid Pundamilia nyererei,” Proceedings of the ¨ ¨ Royal Society B, vol. 271, no. 1556, pp. 2445–2452, 2004. [122] A. Karvonen, O. Seppala, and E. T. Valtonen, “Eye fluke- induced cataract formation in fish: quantitative analysis [104] O. Seehausen, “Progressive levels of trait divergence along using an ophthalmological microscope,” Parasitology, vol. a ’speciation transect’ in the Lake Victoria cichlid fish 129, no. 4, pp. 473–478, 2004. Pundamilia,” in Ecological Reviews: Speciation and Patterns of [123] A. E. Crowden and D. M. Broom, “Effects of the eyefluke, Diversity, R. Butlin, J. Bridle, and D. Schluter, Eds., pp. 155– Diplostomum spathaceum, on the behaviour of dace (Leucis- 176, Cambridge University Press, 2009. cus leuciscus),” Animal Behaviour, vol. 28, no. 1, pp. 287–294, [105] A. P. Hendry, D. I. Bolnick, D. Berner, and C. L. Peichel, 1980. “Along the speciation continuum in sticklebacks,” Journal of [124] O. Seppal¨ a,¨ A. Karvonen, and E. T. Valtonen, “Manipulation Fish Biology, vol. 75, no. 8, pp. 2000–2036, 2009. offishhostbyeyeflukesinrelationtocataractformation [106] D. Schluter and L. M. Nagel, “Parallel speciation by natural and parasite infectivity,” Animal Behaviour,vol.70,no.4,pp. selection,” American Naturalist, vol. 146, no. 2, pp. 292–301, 889–894, 2005. 1995. [125] A. Karvonen and O. Seppal¨ a,¨ “Effect of eye fluke infection [107] H. D. Rundle, L. Nagel, J. W. Boughman, and D. Schluter, on the growth of whitefish (Coregonus lavaretus)—an exper- “Natural selection and parallel speciation in sympatric imental approach,” Aquaculture, vol. 279, no. 1-4, pp. 6–10, sticklebacks,” Science, vol. 287, no. 5451, pp. 306–308, 2000. 2008. 20 International Journal of Ecology

[126] A. D. C. MacColl, “Parasites may contribute to ’magic [143] M. Plenderleith, C. Van Oosterhout, R. L. Robinson, and G. trait’ evolution in the adaptive radiation of three-spined F. Turner, “Female preference for conspecific males based on sticklebacks, Gasterosteus aculeatus (Gasterosteiformes: Gas- olfactory cues in a Lake Malawi cichlid fish,” Biology Letters, terosteidae),” Biological Journal of the Linnean Society, vol. 96, vol. 1, no. 4, pp. 411–414, 2005. no. 2, pp. 425–433, 2009. [144] C. J. Allender, O. Seehausen, M. E. Knight, G. F. Turner, [127] S. P. Brown, “Cooperation and conflict in host-manipulating and N. Maclean, “Divergent selection during speciation of parasites,” Proceedings of the Royal Society B, vol. 266, no. Lake Malawi cichlid fishes inferred from parallel radiations 1431, pp. 1899–1904, 1999. in nuptial coloration,” Proceedings of the National Academy of [128] F. Cezilly, A. Gregoire, and A. Bertin, “Conflict between co- Sciences of the United States of America, vol. 100, no. 2, pp. occurring manipulative parasites? An experimental study of 14074–14079, 2003. the joint influence of two acanthocephalan parasites on the [145] A. G. Hudson, P. Vonlanthen, and O. Seehausen, “Rapid behaviour of Gammarus pulex,” Parasitology, vol. 120, no. 6, parallel adaptive radiations from a single hybridogenic pp. 625–630, 2000. ancestral population,” Proceedings of the Royal Society B, vol. [129] M. L. Arnold, Natural Hybridization and Evolution,Oxford 278, no. 1702, pp. 58–66, 2011. University Press, USA, 1997. [130] O. Seehausen, “Hybridization and adaptive radiation,” Trends in Ecology and Evolution, vol. 19, no. 4, pp. 198–207, 2004. [131] J. Mallet, “Hybrid speciation,” Nature, vol. 446, no. 7133, pp. 279–283, 2007. [132] C. A. Buerkle, R. J. Morris, M. A. Asmussen, and L. H. Rieseberg, “The likelihood of homoploid hybrid speciation,” Heredity, vol. 84, no. 4, pp. 441–451, 2000. [133] C. A. Buerkle, D. E. Wolf, and L. H. Rieseberg, “The origin and extinction of species through hybridization,” in Population Viability in Plants: Conservation, Management, and Modeling of Rare Plants,C.A.BrighamandM.W. Schwartz, Eds., pp. 117–141, Springer, Heidelberg, Germany, 2003. [134] O. Seehausen, G. Takimoto, D. Roy, and J. Jokela, “Speciation reversal and biodiversity dynamics with hybridization in changing environments,” Molecular Ecology,vol.17,no.1,pp. 30–44, 2008. [135] D. W. Pfennig, M. A. Wund, E. C. Snell-Rood, T. Cruick- shank, C. D. Schlichting, and A. P. Moczek, “Phenotypic plasticity’s impacts on diversification and speciation,” Trends in Ecology and Evolution, vol. 25, no. 8, pp. 459–467, 2010. [136] X. Thibert-Plante and A. P. Hendry, “The consequences of phenotypic plasticity for ecological speciation,” Journal of Evolutionary Biology, vol. 24, no. 2, pp. 326–342, 2011. [137] B. M. Fitzpatrick, “Underappreciated consequences of phenotypic plasticity for ecological speciation,” International Journal of Ecology, vol. 2012, Article ID 256017, 2012. [138] P. Edelaar, A. M. Siepielski, and J. Clobert, “Matching habitat choice causes directed gene flow: a neglected dimension in evolution and ecology,” Evolution, vol. 62, no. 10, pp. 2462– 2472, 2008. [139] A. Karvonen, O. Seppal¨ a,¨ and E. T. Valtonen, “Parasite resistance and avoidance behaviour in preventing eye fluke infections in fish,” Parasitology, vol. 129, no. 2, pp. 159–164, 2004. [140] C. T. James, K. J. Noyes, A. D. Stumbo, B. D. Wisenden, and C. P. Goater, “Cost of exposure to trematode cercariae and learned recognition and avoidance of parasitism risk by fathead minnows Pimephales promelas,” Journal of Fish Biology, vol. 73, no. 9, pp. 2238–2248, 2008. [141] M. R. Servedio, S. A. Sæther, and G. P. Sætre, “Reinforcement and learning,” Evolutionary Ecology, vol. 23, no. 1, pp. 109– 123, 2009. [142] E. B. Taylor and J. D. McPhail, “Historical contingency and ecological determinism interact to prime speciation in sticklebacks, Gasterosteus,” Proceedings of the Royal Society B, vol. 267, no. 1460, pp. 2375–2384, 2000. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 939862, 17 pages doi:10.1155/2012/939862

Review Article Parallel Ecological Speciation in Plants?

Katherine L. Ostevik,1 BrookT.Moyers,1 Gregory L. Owens,1 and Loren H. Rieseberg1, 2

1 Department of Botany, University of British Columbia, 3529-6270 University Boulevard, Vancouver, BC, Canada V6T 1Z4 2 Biology Department, Indiana University, 1001 E Third Street, Bloomington, IN 47405, USA

Correspondence should be addressed to Katherine L. Ostevik, [email protected]

Received 3 August 2011; Revised 14 October 2011; Accepted 18 November 2011

Academic Editor: Andrew Hendry

Copyright © 2012 Katherine L. Ostevik et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Populations that have independently evolved reproductive isolation from their ancestors while remaining reproductively cohesive have undergone parallel speciation. A speciﬁc type of parallel speciation, known as parallel ecological speciation, is one of several forms of evidence for ecology’s role in speciation. In this paper we search the literature for candidate examples of parallel ecological speciation in plants. We use four explicit criteria (independence, isolation, compatibility, and selection) to judge the strength of evidence for each potential case. We ﬁnd that evidence for parallel ecological speciation in plants is unexpectedly scarce, especially relative to the many well-characterized systems in animals. This does not imply that ecological speciation is uncommon in plants. It only implies that evidence from parallel ecological speciation is rare. Potential explanations for the lack of convincing examples include a lack of rigorous testing and the possibility that plants are less prone to parallel ecological speciation than animals.

1. Introduction studies linking intrinsic genetic incompatibilities with divergent ecological selection; and (5) tests of parallel ecological The past two decades have witnessed a dramatic shift in stud- speciation, which is the process in which related lineages ies of speciation from an emphasis on stochastic and other independently evolve similar traits that confer shared repro- nonecological processes to a focus on ecological mechanisms ductive isolation from their ancestral populations [6]. of speciation. Indeed, the proverbial pendulum has swung In plants, widespread application of the first two methods so far toward ecology that some authors have argued that listed above points to an important role for ecology in essentially all plausible types of speciation involve ecological speciation. For example, there is a long tradition of reciprocal processes [1]. Despite the pervasive role of natural selection transplant studies since early in the 20th century, and there is in evolution, evidence that ecologically based divergent nat- abundant evidence of immigrant inviability among recently ural selection is the primary cause of reproductive isolation diverged populations or species [7, 8]. Because habitats often (ecological speciation sensu Schluter, 2001 [2]) is often weak are spatially segregated, divergent habitat adaptation results or incomplete in case studies [3]. in ecogeographic isolation, which is considered by many To more critically evaluate the importance of ecological botanists to be the most important reproductive barrier in speciation in nature, several authors have suggested methods plants [1, 9]. Likewise, studies that examine the relative for reliably inferring ecological speciation [2–5]. These importance of different components of reproductive isola- include (1) direct measurements of divergent ecological tion in plants indicate that ecologically based reproductive selection on parental genotypes (e.g., immigrant inviability) barriers often play a key role in the early stages of plant or hybrids (i.e., extrinsic postzygotic isolation) in the differ- speciation [8]. However, evidence of extrinsic postzygotic ent environments; (2) natural selection studies showing that isolation in plants is surprisingly weak, possibly because phenotypic differences underlying premating reproductive of heterosis [8, 10]. Also, few studies have explicitly tested barriers are a consequence of divergent ecological selection; for isolation by adaptation in plants [1]orforecological (3) molecular marker studies of selection against immigrants causes of hybrid incompatibilities (reviewed in [11]). The (i.e., isolation by adaptation); (4) molecular evolutionary evidence for parallel ecological speciation is perhaps least 2 International Journal of Ecology clear because while “recurrent” formation of plant species multiple independent origins (reviewed in [35]), in which and races is thought to be common [12], the evidence independently derived polyploid lineages are reproductively underlying these apparent examples has not been examined isolated from their common ancestor but not from one systematically. another. Additionally, a high proportion of homoploid On the other hand, parallel speciation is regularly cited hybrid species studied arose in parallel [36]. Although there as evidence for ecological speciation in animals (e.g., [6, 13– is evidence that natural selection is important in polyploid 22]), and the evidence for many individual cases of parallel andhybridspeciation[10, 37, 38], the genomic changes ecological speciation is strong. For example, the threespine that accompany polyploidization and hybridization reduce stickleback has undergone several well-documented parallel our ability to show that parallel ecological selection was the transitions between environments. The most well-known primary driver of reproductive isolation. This differentiates case is likely the independent origin of “benthic” and “lim- parallel polyploid and hybrid speciation from parallel eco- netic” ecotypes in at least five British Columbian lakes [23, logical speciation described above and represented in the 24]. Another well-studied system is the marine snail Littorina animal cases listed previously. saxatilis, which has repeatedly evolved pairs of ecotypes on In cases of parallel ecological speciation, the independent the rocky coasts of Northwestern Europe [25]. Numerous descendent populations are found in a new environment other strong candidates for parallel ecological speciation are where they experience new and shared ecological selection found in animals, including but not limited to lake whitefish that causes speciation. However, not all cases in which mul- [26], cave fish [27], walking sticks [20], scincid lizards [21], tiple transitions to a new environment are associated with lamprey [28], electric fish [29], horseshoe bats [30], and repeated speciation events represent true parallel ecological possibly even in the genetic model organism Drosophila speciation. Several possible patterns exist and are shown melanogaster [31], though not all of these examples are fully in Figure 1. In parallel ecological speciation, ancestral and validated with the criteria described below. descendent groups each represent single compatible groups Here we use explicit criteria to evaluate the strength (Figure 1(a)). However, one can also envision several other of evidence for parallel ecological speciation in plants. We patterns, in which either the ancestral or descendent groups evaluate plant systems using criteria that are more often used (or both) represent multiple compatibility groups (Figures to evaluate animal systems because comparable evaluations 1(b)–1(d)). across taxa are important for determining general patterns The pattern in which the descendent groups are incom- of speciation. We find that evidence for parallel ecological patible with one another (Figure 1(b))canbecausedby speciation in plants is surprisingly rare in comparison to mutation-order speciation in which the same selective animals and provide potential explanations for this finding. pressure leads to different genetic changes in the multiple populations [4]. The isolation between ancestral groups could also be the result of drift. The third pattern (Figure 2. Studying Parallel Ecological Speciation 1(c)) has been called “replicated ecological speciation” [39] and is made up of multiple distinct speciation events. Parallel speciation is the process in which related lineages Studying similarities and differences between these replicate independently evolve similar traits that confer shared repro- speciation events can help identify general patterns of ductive isolation from their ancestral populations [6]. It speciation [39]. Finally, the last pattern (Figure 1(d))isto is good evidence that selection drove the evolution of our knowledge novel, and we do not know of any empirical reproductive isolation, as it is unlikely that the same barriers examples. would arise independently by chance [6]. Schluter and Nagel In this survey, however, we are only interested in parallel [6] listed three criteria for parallel speciation: (1) related ecological speciation (Figure 1(a)), which tells us something lineages that make up the new descendent populations more specific than other patterns. In particular, parallel are phylogenetically independent; (2) descendent popula- ecological speciation indicates that all of the new barriers tions are reproductively isolated from ancestral populations; present are predominantly if not entirely due to natural (3) independently derived descendent populations are not selection, whereas in the other cases, other forces may have reproductively isolated from each other. They add that an been at play along with natural selection. For example, if the adaptive mechanism must be identified to show that natural descendent populations are reproductively isolated from one selection drove the evolution of reproductive isolation [6]. another (Figures 1(b) and 1(c)), it is plausible that changes, Together, these four conditions are the evidence necessary driven by processes other than ecological adaptation, caused to demonstrate the process we refer to as parallel ecological the isolation between descendent populations as well as speciation. We choose to use the term “parallel” over the the isolation across habitats. This means that testing the alternative term “convergent” because of the initial similarity compatibility of descendent populations is essential for of the independent lineages [32–34] and because this documenting parallel ecological speciation. However, this vocabulary is consistent with the original description of the test is not necessary for demonstrating all forms of evidence process [6]. for ecological speciation. If we apply these criteria to well-known examples of When studying parallel ecological speciation, it is also parallel or “recurrent” speciation in plants [12], it is clear that useful to recognize that evidence of parallelism may or may botanists and zoologists are mainly studying different things. not extend across multiple levels of biological organization. For example, many auto- and allopolyploid species have Although the individuals of the descendent species must have International Journal of Ecology 3

(a) (b)

Figure 1: An illustration of several patterns in which multiple independent transitions to a new environment are coupled with speciation. The yellow and green boxes represent two different environments, the arrows represent multiple independent transitions between the environments, and the closed shapes represent reproductively compatible groups. Only panel (a) represents parallel ecological speciation, the process we are interested in. the same isolating trait, this common trait is not necessarily Levin [12] reviewed a number of potential cases of parallel governed by the same mutation, gene, or even pathway in ecological speciation in plants. Although his paper was not the different replicates. Even when the same mutations are strictly about parallel ecological speciation, we consider all responsible for parallel transitions in a given isolating trait, of his ecological examples in our table in order to revisit these mutations can be either the result of recurrent de novo their validity and discuss new evidence for each case. We mutation or standing variation. While independent genetic used explicit criteria to determine the strength of evidence changes simplify the task of reconstructing population for parallel ecological speciation. Specifically, we judged and trait histories, parallel changes from standing genetic the strength of evidence for each of the aforementioned variation can be useful for pinpointing regions of the four criteria: independence, isolation, compatibility, and genome responsible for ecological selection and reproductive selection. To merit inclusion, we required that each newly isolation [40]. identified case has at minimum weak direct evidence for repeated adaptation to similar environments (independence) or multiple origins of an isolating trait (isolation) and 3. Literature Survey indirect evidence for the other condition. Evidence that an example failed to meet any of the four criteria resulted in For this survey, we searched the scientific literature for its exclusion. As a result, several promising systems were not evidence of parallel ecological speciation in plants. We used included in our table or appendix (e.g., Frankenia ericifolia combinations of the words “parallel”, “recurrent”, “multi- [41]andHeteropappus hispidus ssp. Leptocladus [42]). This ple”, “convergent” and “speciation”, “evolution”, “origins”, was because Frankenia ericifolia has only indirect evidence of “reproductive barriers/isolation” as well as “paraphyly” as both independence and isolation and Heteropappus hispidus search terms in Web of Science and Google Scholar. We ssp. Leptocladus has no evidence of isolating traits. Lastly, we also examined all papers citing candidate examples and evaluated a few of the frequently cited examples of parallel searched for additional papers about the candidate species. ecological speciation in animals for comparison. The list of 4 International Journal of Ecology animal examples is not exhaustive and does not necessarily speciation [43](e.g.,[44]). Furthermore, because explicitly include all of the best cases. testing that isolation is genetically based was rare in our The following sections discuss each criterion and the candidate studies, we only required the genetic basis of types and strengths of evidence we considered. isolation to be confirmed for cases to have strong direct evidence. 3.1. Independence. Parallel speciation requires that replicate lineages be phylogenetically independent, so that the “shared 3.3. Compatibility. As we briefly discuss above, the lack of traits responsible for reproductive isolation evolved sepa- reproductive barriers between descendent populations is a rately” [6]. Estimating phylogenetic independence can be key criterion distinguishing true parallel ecological speci- difficult for recently diverged taxa because of inadequate ation from other forms of replicated ecological speciation resolution, hybridization and introgression, and deep coa- (Figure 1). For strong evidence of reproductive compatibility lescence. These issues are exacerbated when phylogenies are we require that descendent populations show little or no based on a small number of loci or if the loci employed barriers to reproduction when experimentally crossed and have little phylogenetic information content (e.g., isozymes). minimal ecological divergence as demonstrated by recip- Additionally, it is not strictly necessary for populations to rocal transplant or manipulative ecological experiments. If be phylogenetically independent for reproductive barriers to only one of these two components of compatibility were evolve separately, if different genes, mutations, or genetic demonstrated, we consider the evidence to be weak. We also pathways are exploited by selection in populations that consider substantial genetic analysis showing little genetic remain connected by gene flow. differentiation at neutral markers between the descendent Therefore, in this survey, strong evidence for indepen- populations to be weak direct evidence for compatibility, dent evolution includes: (1) phylogenetic analyses support- although this might weaken the case for the criterion of inde- ing independence with multiple, phylogenetically informa- pendence. Lastly, if the evolution of environmental specificity tive loci or (2) direct evidence that a shared isolating trait has or mating system is such that descendant populations are evolved independently (e.g., the trait is a result of different indistinguishable in these parallel traits and have no other mutations in different populations). Phylogenetic analyses phenotypic differences, we consider this indirect evidence for with a single locus, or with loci having little information con- compatibility. tent, are deemed weak direct evidence. Note that we consider It is difficult to ascertain the strength of evidence genetic information from completely linked markers (e.g., for isolation and compatibility when multiple independent multiple genes sequenced from the chloroplast genome) origins of self-fertilization (autogamy) have occurred. This as information from one locus. Finally, indirect evidence is because replicate populations of selfers are likely to be for independence could include the improbability of long- as strongly isolated from one another as they are from the distance colonization or gene flow between geographically ancestral outcrossing populations. For this reason, replicated separated but ecologically similar habitats, or phenotypes selfing lineages were excluded from consideration unless that are similar but not identical suggesting different genetic accompanied by the evolution of other reproductive barriers. bases. 3.4. Selection. Without evidence of selection, parallel spe- 3.2. Isolation. As with any test of incipient or recent speciation cases can tell us nothing about ecology’s role in ciation, reproductive isolation must have evolved between speciation. Because of this, we searched for evidence of descendent and ancestral populations, though not necessar- the adaptive mechanism(s) underlying parallel isolation. ily to completion. For strong evidence of isolation, we require For strong evidence, we include reciprocal transplants experimental evidence for strong reproductive barriers that showing strong local adaptation, manipulative experiments are genetically based, such as substantial differences in relating adaptive traits to extrinsic fitness, and/or signatures flowering time or pollination syndrome in a common garden of selective sweeps at loci underlying putatively adaptive and/or F1 hybrid inviability or sterility. Experimentally traits. Similarly, reciprocal transplants showing weak local demonstrated weak but statistically significant reproductive adaptation, manipulative experiments showing a weak rela- barriers between diverging populations (including selection tionship between traits and extrinsic fitness, or common against immigrants and hybrids), or genetic divergence garden experiments comparing QST to FST are deemed weak between locally diverging populations despite the oppor- evidence. Finally, we consider correlations between novel tunity for gene flow, are considered weak evidence for traits and environments or habitats to be indirect evidence isolation. We view systems with apparent immigrant or for the role of selection. hybrid inviability (e.g., serpentine adaptation), long-term persistence of divergent populations in sympatry, or strong 4. Results of the Survey divergence in mating system as indirect evidence that barriers likely exist. Although we would not consider isolation that The most striking result of this survey is that very few plant has no genetic basis as evidence for parallel ecological cases have strong evidence for two or more criteria of parallel speciation, we acknowledge that phenotypic plasticity can ecological speciation, and most have only weak or indirect facilitate or impede the evolution of reproductive barriers evidence for any of the criteria (Table 1; Appendix A). Only and is an important consideration for studies of ecological 3 of the 15 examples discussed by Levin [12]meetour International Journal of Ecology 5

Table 1: Candidate examples of parallel ecological speciation in plants showing the strength of evidence for each case. The details of each example can be found in Appendix A. The strength of evidence for each case was coded as follows: ∗∗∗: strong direct evidence, ∗∗: weak direct evidence, ∗: indirect evidence, —: no evidence, †: evidence against, and NA: no data.

Min. number Species Independence Isolation Compatibility Selection Major parallel trait(s) References of origins Alopecurus myosuroidesLX Many ∗∗ NA NA ∗∗∗ Herbicide tolerance [45–51] Agrostis capillariesLX Many ∗∗∗∗∗Edaphic tolerance [52–54] Agrostis stoloniferaLX NA NA ∗∗∗∗Edaphic tolerance [55–57] Agrostis tenuisL NA ∗∗∗NA ∗∗ Edaphic tolerance [58, 59] Armeria maritimaLX NA ∗∗∗ † ∗∗Edaphic tolerance [60–63] Chaenactis spp.LX 2 ∗∗∗ † ∗∗ ∗ Flower colour [64, 65] Cerastium alpinum 2 ∗∗ ∗ NA ∗∗∗ Edaphic tolerance [66] Chenopodium albumLX NA ∗ NA NA ∗ Herbicide Tolerance [67–70] Crepis tectorumLX 2 ∗∗ NA NA ∗∗ Leaf morphology [71, 72] Deschampsia caespitosaL 2 ∗∗ ∗ ∗ ∗ Edaphic tolerance [73, 74] Eucalyptus globulus 3 ∗∗∗ ∗∗ ∗ ∗ Dwarfed morphology [75] Habitat type; Geonoma macrostachys 3 ∗∗ ∗∗ ∗ ∗ [76–78] reproductive strategy Hemerocallis citrina var. 3 ∗∗ ∗ ∗ NA Nocturnal flowering [79] vespertina Hieracium umbellatumLX NA ∗∗∗∗Habitat type (dunes) [80] Lasthenia californica 2 ∗∗∗ ∗∗ ∗∗ ∗∗ Edaphic tolerance [81–87] Habitat type Microseris lanceolataL 2 ∗∗∗ ∗ ∗∗ ∗ [88–90] (elevation) Petunia axillaris 6 ∗∗ ∗∗∗ ∗ — Flower colour [91, 92] Plantago maritimaLX 2 ∗ NA NA NA Spike morphology [93] Poa annuaLX NA ∗ — ∗∗Herbicide tolerance [94, 95] Pollination syndrome, Schizanthus grahamii 2 ∗∗ ∗∗ NA ∗ [96] self-compatibility Silene dioicaLX NA ∗∗ NA NA — Edaphic tolerance [97, 98] Silene vulgarisLX 2 ∗∗NA ∗∗∗ Edaphic tolerance [99–102] Streptanthus glandulosus Many ∗∗ ∗ ∗ ∗∗ Edaphic tolerance [103–107] L : Levin example. X: Example that would not have been included in this table had they not been reviewed in Levin [12]. minimum requirements for inclusion in the table. We did place. Hopefully, future studies in these systems will test identify 8 new candidate systems. These were Cerastium for evidence of compatibility between suspected examples alpinum, Eucalyptus globulus, Geonoma macrostachys, Heme- of parallel species. On the other hand, the criterion with rocallis citrina var. vespertina, Lasthenia californica, Petunia the most evidence is the independent evolution of lineages axillaris, Schizanthus grahamii, and Streptanthus glandulosus. that appear to be diverging in parallel. This is unsurprising Some of the new candidate systems are quite promising, but given the widespread application of molecular phylogenetic none have strong evidence for all criteria. These cases and a methods in plants. review of Levin’s [12] examples are summarized in Table 1 One of the best candidates of parallel ecological and Appendix A. The animal cases used as a comparison speciation in plants to date is Eucalyptus globulus [75] are summarized in Table 2 and Appendix B. Although few (Appendix A). In this case, three populations of E. globulus of the animal cases have strong evidence for all the criteria that inhabit granite headlands have a dwarfed morphology. either, these well-studied animal systems are more strongly Data from several nuclear and chloroplast markers show that supported than the plant examples. each of the dwarfed populations is more closely related to its Many of the putative plant cases lack evidence of com- nearest tall population than to other dwarfed populations. patibility among independent parallel populations. In fact, Furthermore, there are two lines of evidence for isolation. none of the examples have strong evidence for compatibility. First, the dwarfed populations flower earlier than the tall This is unfortunate given the effectiveness of this criterion for populations. Second, there is no evidence of pollen flow demonstrating that ecological selection was the main cause from the tall populations to the dwarfed populations despite of reproductive isolation in these systems—the primary a thorough examination of variation at microsatellite loci. reason for studying parallel ecological speciation in the first However, the genetic basis of isolation, compatibility among 6 International Journal of Ecology

Table 2: Examples of parallel ecological speciation in animals showing the strength of evidence for each case. This list is not exhaustive and instead is a selection of well-studied cases serving as a comparison to the plant examples. The strength of evidence for each case was coded as follows: ∗∗∗: strong direct evidence, ∗∗: weak direct evidence, ∗: indirect evidence, —: no evidence, †: evidence against, and NA: no data.

Min. number Species Independence Isolation Compatibility Selection Major parallel trait(s) References of origins Pigmentation, eye Astyanax cave fishes 2 ∗∗∗ ∗∗ ∗∗ ∗∗∗ [27] functionality Body size and shape, Coregonus clupeaformis 6 ∗∗∗ ∗∗ ∗ ∗∗∗ [13, 108–115] foraging niche Body size and shape, Coregonus lavaretus 3 ∗∗∗ ∗ ∗ ∗∗∗ [14, 115–117] foraging niche Gasterosteus aculeatus Body size and shape, [3, 15, 16, 23, 5 ∗∗∗ ∗∗ ∗∗ ∗∗ benthic/limnetic foraging niche 24, 118–122] Gasterosteus aculeatus Body size and shape, [3, 23, 123– 6 ∗∗∗ ∗∗ ∗ ∗∗ lake/stream foraging niche 130] Gasterosteus aculeatus Body size and shape, [3, 17, 23, anadromous/stream many ∗∗∗ ∗∗ ∗∗ ∗∗∗ life history 131–133] resident Shell size and shape, Littorina saxatilis-Galician [18, 19, 134– 3 ∗∗∗ ∗∗∗ ∗ ∗∗∗ microhabitat Spain 141] adaptation Color pattern, host Timema cristinae NA ∗∗∗ ∗∗ ∗∗ ∗∗ [20, 142–144] plant adaptation dwarfed populations, and the selective advantage of being typically do not do the necessary tests. Perhaps this is because dwarfed still need to be confirmed in this example. botanists have never doubted the importance of ecology in Other promising cases include Lasthenia californica [81] speciation [1, 7, 9] or because other methods of inference (Appendix A)andSchizanthus grahamii [96](Appendix A). have been successful. However, we believe that it is useful Lasthenia californica has strong evidence for parallel evolu- to examine parallel ecological speciation explicitly in plants tion of two races that have different flavonoid pigments and given how fruitful such studies have been in animals. In are found in different edaphic environments using phylo- the present paper, this has allowed us to not only identify genetic analyses of ribosomal and chloroplast sequence and the key information that is missing in most potential case allozyme variation [81–84].RaceAismoretolerantofNa+ studies of parallel ecological speciation in plants, but also and Mg2+ and race C flowers earlier and produced more seed to recommend the experimental tests that are likely to be heads under drought conditions [85, 86]. However, isolation most profitable. For example, immigrant inviability is likely and compatibility between and within the races need to be an important barrier in cases of parallel ecological speciation confirmed. Conversely, there are clear reproductive barriers in plants. Therefore, reciprocal transplants between the new between Schizanthus grahamii and Schizanthus hookeri as habitat types (as a test of isolation) and among sites in a they have different primary pollinators and experimental single habitat type (as a test of compatibility) are crucial. interspecific crosses produced no seeds [96]. However, it A second essential set of tests that should be conducted is is not certain that these barriers arose multiple times crosses between ancestral and descendent populations and independently because only chloroplast sequence data has crosses among populations in each habitat. It is surprising been analyzed. that these data are lacking given that both an explicit There are a few commonalities among the candidate framework for studying parallel ecological speciation [6]and examples. In many of the candidate systems ancestral and a list of possible cases [12] have been available for many years. descendent forms have evolved particular edaphic tolerances Furthermore, the signature of parallel speciation is and/or specific changes in reproductive phenotype or mating easily lost. For example, if gene flow occurs between inde- system. Under divergent ecological selection, these traits are pendently derived populations, the signals of phylogenetic likely to make ecological speciation relatively easier because independence may be lost. Conversely, if descendent lineages they can cause assortative mating as a byproduct of divergent are geographically isolated (i.e., allopatric) but otherwise selection. reproductively compatible, they will likely eventually evolve reproductive isolation from one another even if they were not originally isolated. Thus, the window of time in which 5. Why Is Parallel Ecological Speciation parallel speciation can be detected may be relatively narrow. Seemingly Rare in Plants? Interestingly, some of the strongest animal cases (Littorina, threespine stickleback, and whitefish) are no more than 5.1. Lack of Data. The lack of strong examples of parallel 40,000 years old (postglacial) [23, 108], and some are ecological speciation in plants is probably because botanists thought to be as young as 10,000 years [19]. However, we International Journal of Ecology 7 see no reason why this window would be narrower in plants well as provide guidance regarding the kinds of studies that than animals. Thus, while the narrow window of detectability should be performed in each system. may account for the overall paucity of convincing examples of parallel ecological speciation in either the plant or animal Appendices kingdoms, it cannot explain why there are fewer examples in plants than in animals. A. Descriptions of Potential Examples of 5.2. Plants Are Different. It is also possible that parallel Parallel Ecological Speciation in Plants ecological speciation is truly rare in plants. Considerable Note that superscript “L” (L) indicates that the example was work would need to be done to validate this conjecture. reviewed in Levin [12]. However, should this pattern exist, there are several potential explanations. First, it is possible that the types of habitat A.1. Alopecurus myosuroidesL. The black-grass Alopecurus distributions that promote parallel speciation in animals myosuroides is an agricultural weed that has evolved resis- are more rare for plant populations. Many but not all of tance to herbicides in many locations, possibly indepen- the animal examples involve adaptation to systems such dently [45]. Independent evolution may be occurring even as lakes or streams which are common, offer geographical on very local scales: Cavan et al. [46] used microsatellite isolation, and provide relatively homogeneous ecological data to show that four patches of resistant black grass environments. Perhaps these kinds of ecological opportuni- in two neighbouring fields were independently derived ties are less frequent for plants? We think this explanation from nonresistant plants. Herbicide resistance occurs either unlikely given that patchy environments (especially edaphic through plant metabolism, often polygenic, or via mutant environments) are common in terrestrial ecosystems and ACCase alleles, and seven mutant resistance alleles have parallel adaptation into those habitats occurs frequently. been identified [47–49]. A study of herbicide resistance in It is also possible that plants have certain characteristics populations across Europe concluded that the same mutant that make parallel ecological speciation unlikely or lack ACCase alleles have appeared repeatedly [50]. No work has characteristics that promote parallel ecological speciation. been done on reproductive barriers between resistant and This potential difference between plants and animals may be nonresistant plants, although AFLP analysis shows little dif- in part because behavior is not particularly relevant to plants. ferentiation between resistant and nonresistant populations Behavior, especially behaviorally based mate preference, may [51]. An important consideration in this case is that selection be an important component of parallel ecological speciation is human mediated and therefore unlikely to remain constant in animal systems (though pollinator behavior in flowering long enough to allow for speciation. plants may act analogously). Perhaps plants have no trait equivalent to body size in animals, which can act as a L “magic trait” [145] to serve in both assortative mating and A.2. Agrostis capillaris . The corrosion of zinc-galvanized ecological adaptation. However, flowering time could be electricity pylons in South Wales has created repeated patches such a trait, and there are many examples of flowering time of zinc-contaminated soil that have been colonized by the changing in new edaphic environments (e.g., Lord Howe grass Agrostis capillaries. Zinc tolerance levels for A. capil- Island palms [146]). On the other hand, flowering time laries plants vary from low to high across multiple pylons may be quite constrained because partitioning flowering [52]. Tolerance appears to be polygenic and dependent on time requires narrower windows of flowering, which can standing genetic variation [53]. Jain and Bradshaw [54] have strong negative fitness consequences. Other potential determined that seed and pollen dispersal is limited beyond traits are floral morphology and edaphic tolerances. Floral 5 m, suggesting that tolerance is evolving independently morphology may adapt to attract different pollinators and, at each pylon, although the still relatively small distances consequently, lead to pollinator isolation. Similarly, the between pylons (300 m) do not rule out occasional pylon evolution of edaphic tolerance often leads to selection against to pylon gene flow. Further work should establish if the immigrants. populations are truly independent and measure barriers to The lack of evidence for parallel ecological speciation in gene flow between tolerant and nontolerant neighbouring plants is a mystery that may represent a key to understanding populations. how species arise in plants. If parallel ecological speciation is more common than our survey suggests, then we can bolster A.3. Agrostis stoloniferaL. Metal refining in Prescot, UK, existing evidence that ecology plays an important role in caused considerable copper contamination to surrounding plant speciation. On the other hand, if parallel ecological soil, and the grass Agrostis stolonifera has since then colonized speciation is determined to be rare, we can conclude that a number of contaminated sites. Older sites were found to speciation may be less repeatable and more complicated have more complete ground cover and a greater proportion than sometimes believed. We do not intend to imply that of resistant individuals [55], suggesting that the evolution ecological speciation does not happen in plants. In fact, of tolerance is ongoing at younger sites. Morphological we believe it to be common. However, evidence of parallel and isozyme analyses suggest, counterintuitively, that there ecological speciation in plants is not yet as convincing as it is is a reduction in clone number in uncontaminated sites for animal examples. We hope our study will spur additional compared to contaminated ones. All sites are centered on investigation of the promising systems identified here, as a single copper refinery, so the independence of the sites 8 International Journal of Ecology is questionable. Agrostis stolonifera has also evolved salt gene flow is possible although it may be limited by differences tolerance in multiple inland and coastal sites, possibly inde- in edaphic preference. Future work should use molecular pendently [56, 57]. The strength of evidence for independent tools to verify the cytological data and quantify gene flow or parallel evolution in either of these cases is quite weak. between species.

A.4. Agrostis tenuisL. Copper-tolerant populations of Agro- A.8. Chenopodium albumL. TheagriculturalweedChenopo- stis tenuis, from the UK and Germany, were shown to dium album has developed resistance to triazine herbicides have different responses to copper using regression anal- in multiple locations [67]. Early work showed that different ysis [58]. This variability indicates that multiple genes or resistant populations have distinct isozyme patterns in alleles are responsible for copper tolerance in different France [68] with at least two resistant genotypes in Canada populations, although crossing experiments could further [69]. Although a single amino acid mutation in the psbA support this inference. Other work has shown asymmetric gene has been linked to herbicide resistance [70], further gene flow between tolerant and nontolerant populations molecular analysis should be performed to establish if this with moderate levels of copper tolerance found in sites mutation has evolved independently. No work has been done downwind of colonized mine tailings [59]. Although adult on gene flow between populations. plants from mine sites showed high degrees of tolerance, seeds were not tested, so it is unknown if gene flow from A.9. Crepis tectorumL. The degree of leaf dissection, a trait nontolerant populations is reduced pre- or postzygotically. with fitness consequences, varies among populations of The adaptation to copper-contaminated soil at multiple sites Crepis tectorum [71]. In the Baltic region, populations vary may be parallel, or may be the result of the transmission in leaf shape, with two island populations exhibiting more of tolerant genotypes between mines. Without stronger deeply lobed leaves than those from the mainland. Andersson evidence for independence, this case is very weak. [72] used a crossing experiment to demonstrate that while deep lobes on one island are caused by a single dominant A.5. Armeria maritimaL. Populations of Armeria maritima locus, on the other island they are caused by multiple loci, across Europe vary in metal tolerance: those living on metal- which suggest an independent origin of the trait on each liferous soil are tolerant while those on uncontaminated soil island. Further work is needed to confirm this independence, are not [60]. Isozyme and nuclear marker data suggest that to elaborate the adaptive value of leaf dissection in this tolerance has evolved multiple times independently [61–63]. system and to establish if there is any reproductive isolation Furthermore, this tolerance is maintained even in the face other than geographic between the deeply lobed and less of substantial gene flow between neighbouring nontolerant lobed forms. populations, indirect evidence for the strength of selection on the metalliferous soil. This does not appear to represent A.10. Deschampsia cespitosaL. In the 1970s, this perennial parallel speciation, however, as pollen fertility and gene flow grass colonized metal-contaminated soil at two locations ff are not reduced between populations with di erent levels of in Southern Ontario, Canada. Isozyme analysis of the tolerance [61]. populations at both contaminated sites, as well as uncontaminated sites to the south, found reduced variability in A.6. Cerastium alpinum. Enzyme phenotypes suggest that the metal-contaminated populations [73]. Unique alleles Northern Europe was colonized by two postglacial lineages in each contaminated site suggested that each had an of Cerastium alpinum [64]. The two lineages are found on independent origin. However, a more recent genetic marker both serpentine and nonserpentine soils but a principle analysis with the same populations has produced equivocal components analysis of enzyme phenotype does not reveal results, indicating that although there are two origins for any clustering by soil type, suggesting that serpentine the contaminated site populations, one population at one tolerance evolved independently in each lineage [64]. Further site shares its origin with all populations at the other experiments manipulating Ni and Mg concentrations show contaminated site [74].Noworkhasbeendoneonbarriers that serpentine populations have higher tolerance to Ni and to gene flow between populations or on the mechanisms of Mg [65].Nobarrierstoreproductionhavebeendocumented heavy metal adaptation. in this system although selection against immigrants seems likely. A.11. Eucalyptus globulus. Three populations of Eucalyptus globulus that inhabit exposed granite headlands in south- A.7. Chaenactis spp.L. Three closely related species of pin- eastern Australia have a dwarfed morphology and flower cushion are found in California: Chaenactis glabriuscula, C. earlier than their tall ancestors [75]. Relatedness analyses stevioides, and C. fremontii. C. glabriuscula is found in mesic using several nuclear and chloroplast markers show that the habitat, has yellow flowers, and has n = 6 chromosomes dwarfed populations are more closely related to the nearest while both C. stevioides and C. fremontii are found in desert population of the tall ecotype than to each other [75]. habitat, have white flowers, and have n = 5 chromosomes. Observations of progeny allele frequency show no evidence Cytological analysis indicated that C. stevioides and C. of pollen-mediated gene flow from the much more abundant fremontii arose from independent aneuploid reductions [66]. tall ecotypes to the dwarf ecotypes [75]. This suggests that Frequent natural hybrids between the species indicate that there have been at least three independent transitions to International Journal of Ecology 9 dwarfism in the novel exposed granite headland habitat be used to confirm the phylogenetic independence of these (barring, of course, a long history of introgression after a populations, and further work needs to characterize the single origin and dispersal). This case is quite promising, as adaptive mechanisms underlying the ecotypic characters and it has strong evidence for both independence and isolation the extent of reproductive isolation between and within from ancestral populations. What remains is to demonstrate ecotypes. the compatibility of the dwarf populations with each other, and to more clearly elucidate the adaptive value of dwarfism A.15. Lasthenia californica. The common goldfield, Lasthe- in this system. nia californica, grows in a variety of habitats and has two flavonoid pigment races that strongly correlate with edaphic A.12. Geonoma macrostachys. Lowland forests in Peru are tolerance. Race A grows on ionically extreme habitats such home to two subspecies of the palm Geonoma macrostachys as coastal bluffs, alkaline flats, vernal pools, and serpentine that are alternately more abundant in flood plain versus soil, while race C is found on ionically benign and drier tierra firme habitat [76]. The two subspecies differ in leaf locations such as pastures and oak woodlands. Phylogenetic shape and are reproductively isolated by phenology, flow- analyses using ribosomal and chloroplast sequences along ering activity, and pollinator spectrum [77]. However, ISSR with allozyme variation indicate two cryptic clades within variation strongly partitions among sympatric populations the species with representatives of both races in each [81– of both subspecies rather than between the subspecies, and 84], suggesting a parallel origin of each race. Greenhouse subspecific genetic classification is not possible [78]. In experiments indicate that race A plants, regardless of phy- three different forests, Roncal [78] found consistently strong logenetic clade, have greater tolerance to Na+ and Mg2+ microhabitat preferences for each of the two subspecies, and in drought conditions race C plants flower earlier and which, along with the genetic data, suggest an independent producemoreflowerheads[85, 86]. Preliminary data shows origin of the subspecies in each environment. Alternate reduced seed set between different races of the same clade hypotheses of a history of local gene flow among subspecies and greater pollination success between populations of the or phenotypic plasticity must be ruled out before this case same race during interclade crossing, although these data can be considered parallel speciation, and further work on have not been formally published after being presented in reproductive isolation and the mechanisms of microhabitat Rajakaruna and Whitton [87]. This case has great potential, adaptation is warranted. but further conclusions await stronger published evidence.

A.13. Hemerocallis citrina var. vespertina. On Japanese A.16. Microseris lanceolataL. Australia is home to two eco- archipelagos, there appear to be three independent origins types of Microseris lanceolata: a “murnong” ecotype found of nocturnal flowering and associated changes in floral below 750 m elevation which produces tubers, and an morphology in Hemerocallis citrina var.vespertinafrom the “alpine” ecotype found above 1000 m elevation which whole-day flowering H. flava. Data from three chloroplast reproduces vegetatively in addition to having a significantly markers place H. citrina var. vesperina within three different later flowering time [88]. Phylogenetic analyses based on geographically distinct subspecies of H. flava from mainland chloroplast markers show three geographically correlated Asia, despite persistent morphological and phenological clades within M. lanceolata that all include individuals of differences between the two species [79]. This could be the both ecotypes, suggesting parallel independent origins [89]. result of introgression leading to chloroplast capture, or Nuclear AFLP markers also support this hypothesis, as incomplete lineage sorting of ancestral variation, and little is genetic distance among populations correlates strongly with known about reproductive barrier strengths within the three geographic distance rather than ecotype identity [90]. This clades of H. citrina var. vespertina or between the two species. pattern may be explained by a single origin and dispersal Further study is needed to differentiate these hypotheses of each ecotype followed by significant local hybridization and elucidate reproductive isolation in the system, as well between ecotypes, but Vijverberg et al. [90] emphasize that as the adaptive mechanism underlying variation in floral these populations have managed to maintain their ecotypic phenology. characteristics even in the face of gene flow. Given this and evidence that crosses between and within ecotypes are viable, A.14. Hieracium umbellatumL. Possibly the first observed it seems likely that selection is acting in parallel to maintain ff case of parallel ecotypic differentiation, Swedish Hieracium or recreate fixed di erences between these populations. umbellatum, was described by Turesson in 1922 [80]. His study found that dune inhabiting plants produced more A.17. Petunia axillaris. Petunia axillaris has likely repeatedly prostrate stems and thicker leaves than those in open wood- evolved white flowers from ancestral colored flowers, as lands and that these differences were heritable. Furthermore, indicated by sequence data showing 6 different loss-of- although these ecotypes shared many morphological traits, function mutations of the ANTHOCYANIN2 (AN2) gene they also retained some leaf characters more like those of in wild P. axillaris populations [91, 92]. It is possible that neighbouring populations of a different ecotype than distant AN2 was downregulated a single time and that the loss of populations of the same ecotype. Ecotypes also differed in function mutations occurred subsequently, but P. axillaris flowering time, an early-acting reproductive barrier. This does not exhibit the low expression of AN2 that would case is promising, but modern population genetics should be expected if the AN2 promoter was inactivated [92]. 10 International Journal of Ecology

Furthermore, pollination experiments using introgression A.21. Silene dioicaL. The red campion, Silene dioica, has the lines and transgenic flowers have shown that functional and ability to colonize both serpentine and nonserpentine habi- nonfunctional AN2 alleles have a large effect on pollinator tats. Although Westerbergh and Saura [97] demonstrated visitation, which is likely a strong reproductive barrier in using isozymes that serpentine populations tended to group this system [92]. However, we do not yet have evidence that with neighbouring nonserpentine populations, indicating these floral colour transitions have been driven by natural multiple origins of serpentine tolerance or possibly ongoing selection, and there is only weak evidence for directional gene flow, later work showed that all populations, regardless selection at these loci [92]. It is certainly possible that this is of soil type, had serpentine soil tolerance [98]. Thus, a case of repeated adaptation to a new pollination syndrome, serpentine tolerance in Swedish S. dioica is likely constitutive but this remains to be tested. and not parallel.

L A.18. Plantago maritima . This widespread plant grows on A.22. Silene vulgarisL. AtminesitesacrossEurope,Silene inland, coastal, and salt marsh habitats across North America vulgaris from two subspecies (ssp. maritima in coastal and and Europe. In eastern North America, salt marsh plants ssp. vulgaris in continental Europe) has acquired tolerance have relatively lax spikes when compared to plants in rocky to high levels of zinc and copper. Complementation tests habitats. Similarly, British coastal plants also have lax spikes between sites indicate that zinc tolerance is governed by relative to inland populations, although in this case the spikes two loci, both acting in highly tolerant populations of both are less dense than North American salt march plants [93]. subspecies [99]. In one mildly tolerant population, zinc tol- Although these traits appear to have evolved independently erance appears to be controlled by only one of the tolerance and in parallel on two different continents, little is known alleles, and intolerant populations in both subspecies have about the genetic basis of these traits or their effects on neither. Similarly, copper tolerance is controlled at two loci: reproductive isolation or local adaptation. one common across all tolerant populations and a second found only in Imsbach, Germany where plants are extremely A.19. Poa annuaL. Annual bluegrass (Poa annua)isan tolerant [100]. The presence of populations with a variable agricultural weed with populations known to be resistant to genetic basis for tolerance in two subspecies at multiple the triazine herbicides [94]. Isozyme work has found equiv- sites across Europe may represent parallel adaptation, but alent levels of variability between resistant and nonresistant the phylogenetic independence of these populations has not populations, suggesting ongoing gene flow after the founding been confirmed, and no studies of reproductive isolation of resistance [95]. Although herbicide resistance is a relatively in the system have been completed. At minimum, popu- simple trait to evolve (often requiring a single amino lations of both subspecies lack strong postzygotic barriers, acid change) there is no evidence to suggest independent as complementation tests are possible. Further work to evolution of resistance in this species beyond the geographic understand the population genetics of metal tolerance (from distance between resistant populations and no evidence that ancestral variation, repeated novel mutations, or gene flow triazine resistance is involved in reproductive isolation. between metalliferous sites) should also be done. Additional populations of S. vulgaris have colonized naturally metalliferous (serpentine) soils in Switzerland and differently A.20. Schizanthus grahamii. Two closely related Andean contaminated mine sites in Canada and Europe [101, 102], butterfly flowers are taxonomically differentiated by polli- which may indicate the ease of evolving metal tolerance in nation syndrome, floral morphology, and mating system: this species. Schizanthus hookeri is purple flowered, bee pollinated, and highly outcrossing, as are other species in the genus, while S. grahamii is capable of self-fertilization, primarily A.23. Streptanthus glandulosus. The Streptanthus glandulosus hummingbird pollinated, and exhibits several color morphs. complex contains several subspecies endemic to serpentine The two taxa are rarely found growing sympatrically despite outcrops in California. Although a majority of populations overlapping elevational ranges (with S. grahamii generally are found on serpentine soil, nonserpentine populations at higher elevations). Within one sympatric population, are also present. Kruckeberg [103] tested serpentine and experimental interspecific crosses produced no seed set, nonserpentine populations of S. glandulosus on serpentine while intraspecific seed set was 63–72% [96]. Chloroplast soil and found that nonserpentine populations were ser- sequence data support two independent parallel origins of pentine intolerant, although this study only qualitatively the S. grahamii morphotype: a southern clade character- examined growth rate due to technical problems. Later ized by red flowers that shares haplotypes with southern studies used cpDNA restriction site data and ITS sequence populations of S. hookeri, and a northern clade with yellow to show that the species is structured into several roughly or pink flowers that shares haplotypes with the northern- geographically based subspecies [104–106]. Nonserpentine most S. hookeri populations [96]. However, this pattern populations occur in multiple subspecies and are more could be explained by historical hybridization followed by closely related to nearby serpentine populations rather chloroplast capture, and further work needs to be done than further nonserpentine populations. This suggests that to rule out this possibility and characterize gene flow serpentine intolerance, as well as perhaps greater competitive and reproductive barriers between the two S. grahamii ability on nonserpentine soil, has occurred multiple times in clades. this species complex. Crossing experiments in this complex International Journal of Ecology 11 found that hybrid fertility is inversely related to geographic and morphology have been identified as under divergent distance, suggesting that nonserpentine populations would selection between the ecotypes [111, 112]. Further research be more compatible with neighbouring serpentine popu- has demonstrated parallel changes in gene expression among lations than distant nonserpentine ones [107]. More study independent sympatric ecotype pairs [113], while changes at is needed to determine if the change in edaphic tolerance the genetic level between normal and dwarf populations are is associated with a change in compatibility, a condition more weakly correlated among lakes [112, 114]. Changes in necessary for parallel ecological speciation. Additionally, the expression in at least two candidate genes are also replicated serpentine intolerance of nonserpentine populations should in the closely related European whitefish system ([115], see be reevaluated in a more quantitative manner. below). In a series of Northern European lakes, European whitefish (C. lavaretus)hasdifferentiated into two ecotypes: B. Descriptions of Frequently Cited Examples of a “sparsely rakered”, larger-bodied, benthic form, and a Parallel Speciation in Animals “densely rakered” smaller limnetic form. Populations of this species form three ancient mitochondrial clades, which do B.1. Astyanax cave fishes. In northeastern Mexico, fish of the not correlate with and are more ancient than gill raker Astyanax species complex have repeatedly adapted to cave divergence [116]. The divergent traits in these populations environments. A recent study incorporating mitochondrial are highly bimodally distributed, and are strongly correlated and nuclear markers supports at least two independent with habitat use and diet [14, 117]. These morphological origins of cave-adapted Astyanax [27]. One genetic cluster relationships contrast with genetic relationships—where is associated with older cave populations characterized by ecotypes within a single lake cluster more closely with highly reduced eyes and pigmentation, while another is each other than with fish of similar morphology in other shared by many surface populations and putatively more lakes [14]. These lakes are less than 15,000 years old. recent cave-adapted populations with less extreme pheno- Although phenotypic and genetic differences in both the types. Nevertheless, all cave populations will interbreed in North American and European species complex are well the laboratory and share many adaptations to a subterranean characterized, evolving in parallel, and show signatures of environment, including an increase in taste bud number, divergent selection, comparatively little is known about improved lateral line sense, and greater fat storage ability the strength and nature of reproductive barriers in both as well as reduced pigmentation and eyes. Although surface systems. and cave fish will also cross in the laboratory, there is no genetic evidence of recent hybridization between the two B.3. Gasterosteus aculeatus. Threespine sticklebacks seem groups in most populations. In one location, surface fish are particularly prone to parallel evolution, with several well- even regularly swept into a cave by flooding—yet this cave documented parallel transitions between environments population shows very little genetic admixture, and only two and foraging niches. Throughout the Northern Hemi- intermediate forms have ever been found despite repeated sphere (and in Japan), large-bodied anadromous threespine sampling [27]. In another cave with frequent introductions stickleback have repeatedly evolved into smaller stream- of surface fish, fish without a cave-adapted phenotype have resident fish [17, 131]. These freshwater transitions involve been observed starving to death and being eaten by fish with the parallel fixation via natural selection of low-armor, cave-adapted phenotypes [27]. Yet in lighted conditions in reduced pigmentation, and pelvic loss alleles in multiple the laboratory, surface fish outcompete cave fish for food. independent populations [131–133]. In the case of pelvic Taken together, the evidence is quite strong for at least loss, these populations exhibit at least three different two independent parallel adaptations to caves by Astyanax, mutations at the same locus—incontrovertible evidence and although reproductive isolation in the system may be for the independent origin of this adaptation [133]. In primarily extrinsic it is reciprocal and appears quite effective. addition, microsatellite data reveals that the majority of genetic variation is partitioned regionally rather than B.2. Coregonus spp. Whitefish is potentially undergoing between ecotypes [17]. In the lab, stream-resident fish several parallel speciation events. The North American lake are more than twice as likely to mate with other stream- whitefish, C. clupeaformis, is present in at least six lakes resident fish than with anadromous fish (and vice versa), in both a “dwarf” limnetic form and a larger-bodied, even when from populations as distant as Iceland and Japan “normal” benthic form [108]. Geographical isolation during [17]. the last glaciation is reflected by three ancient mitochondrial Stickleback have also transitioned in parallel between lineages, likely without much morphological divergence [13, lake habitats and local streams at least 6 times independently 109]. The data suggest that subsequent secondary contact in British Columbia and likely elsewhere, as demonstrated (<15,000 years) of these lineages gave rise to parallel, inde- by a number of genetic analyses [123–125]. This transition pendent sympatric populations of the two ecotypes [108], involves a substantial shift in prey availability and abiotic at least some of which exhibit strong intrinsic and extrinsic environment, and the ecotypes differ in body size, shape, postzygotic reproductive isolation [110]. In addition to body and foraging behavior [124–126]. In one British Columbian size, these ecotypes differ in gill-raker number, age at matura- lake-stream system, common garden experiments confirmed tion, relative fecundity, growth rate, and swimming behavior the genetic basis of these traits, a reciprocal transplant [108]. A number of quantitative trait loci for growth rate experiment showed a weak reduction in growth rate for lake 12 International Journal of Ecology and inlet fish enclosed in the other habitat, and release- locidiscoveredinanFST outlier screen appear to be under recapture studies demonstrated a bias towards recapture of divergent selection between ecotypes in multiple populations inlet fish in the inlet [126]. Taken together, this is relatively [139]. In a reciprocal transplant experiment, each ecotype weak evidence for isolation and selection, but other lake- survived at much higher rates in its native microhabitat stream systems may have evolved stronger barriers between than either the other ecotype or hybrids between them ecotype pairs [124]. [134]. A later reciprocal transplant demonstrated a strong Possibly the best-studied case of potential parallel eco- correlation between the divergent morphological characters logical speciation is the independent origin of “benthic” and and survival in each environment, further evidence for the “limnetic” ecotypes in at least five British Columbian lakes role of selection in maintaining ecotypic differences [140]. [15, 24]. These ecotypes specialize in foraging niche, with These ecotypes exhibit assortative mating by body size that, a number of morphological and behavioral differences, and along with immigrant inviability and habitat preferences, exhibit both prezygotic behavioral and extrinsic postzygotic acts to reduce but not eliminate gene flow between ecotypes barriers to gene flow [15, 16, 118]. Genetic and biogeo- [18, 141]. graphic data support the independent colonization of each lake [15, 119, 120]. Each ecotype is more likely to spawn B.5. Timema cristinae. These western North American with fish of the same ecotype than of the other ecotype, walking-stick insects specialize on two host plant species regardless of lake of origin [15]. Limnetic backcross hybrid and exhibit two different color morphs that are reciprocally fish grew twice as fast as benthic backcross fish in a limnetic more cryptic on different host plants [142]. The two color environment, and vice versa for benthic backcrosses in a morphs are found in higher proportions on the host plant benthic environment [118]. species on which they are most cryptic, although this is In all of these threespine stickleback transitions, repro- quite variable for individual plants, and crypsis helps the ductive isolation appears to have evolved via assortative insects to avoid strong predation pressure by birds and lizards mating by body size acting concurrently with divergent [142, 143]. The color morphs also differ in average body size, selection on body size [3, 16, 23]. However, parallel cases host preference, and cryptic resting behavior, although again vary widely in the strength of reproductive isolation and this is quite variable for individual populations [20, 143]. In genetic differentiation between local ecotypes (e.g. [125, one study, each color morph copulated more readily with 127]). In some cases, evidence for reproductive isolation is the same color morph than the other morph, regardless of conflicting despite evidence for genetic differentiation [128– the population of origin [20]; however, there is no clear 130]. Researchers have been able to manipulate mating pref- relationship between morph-specific divergent characters erences in stickleback by rearing juveniles with individuals and reproductive isolation [143]. Phylogenetic analyses show of the other ecotype, indicating that reproductive isolation that the color morphs are not monophyletic [20, 144], is at least partly extrinsic [121]. Additionally, at least one indicating possible multiple origins of at least one of the independent case of “benthic” and “limnetic” ecotypes has morphs, although this pattern could also be explained by a collapsed back into a single panmictic pool, possibly due single diversification event followed by ongoing local gene to the human-mediated introduction of an exotic crayfish flow between the morphs. One of these studies also demon- [122]. strated using an outlier approach that a small number of loci are under divergent selection between the color morphs B.4. Littorina saxatilis. The marine snail Littorina saxatilis [144]. Consequently, divergent selection appears to be acting has repeatedly evolved pairs of ecotypes on the rocky coasts in parallel in many independent populations. Indeed, as of Northwestern Europe [19]. Although we only discuss discussed by researchers in this system, these walking-stick one here, there are several regional cases of this divergence, insects seem to be experiencing a heterogeneous balance of with ecotypes adapted to different microhabitats created by gene flow and divergent selection, and it is unclear whether tidal and substrate variation. The best-studied ecotype pair this process will ultimately result in speciation between the has evolved at least three times independently in Galician currently weakly isolated morphs. Spain, where the two types specialize in and prefer either the high intertidal barnacle belt or the low intertidal mussel Acknowledgments belt and occasionally hybridize in the intermediate area between those environments [134, 135].Themainaxesof The authors would like to thank Andrew Hendry, Dolph morphological differentiation are in shell size and shape, Schluter, Kieran Samuk, Gina Conte, Sebastien´ Renaut, and which aid in resistance either to dislodging by wave action two anonymous reviewers for helpful comments and discus- (thin, small shells with large apertures for the muscular foot) sions. Funding for this work was provided by the Natural or to crab predation and desiccation stress (thick, large shells Sciences and Engineering Research Council of Canada (DG with small apertures), and there are additional differences in 327475 to L. H. Rieseberg). shell ornamentation [19]. These characters are heritable, and genetic variation partitions primarily among independent References beaches rather than between ecotypes, although the ecotypes do appear genetically divergent on a local scale [136, 137]. [1] J. M. Sobel, G. F. Chen, L. R. Watt, and D. W. Schemske, “The Mitochondrial data similarly support multiple independent biology of speciation,” Evolution, vol. 64, no. 2, pp. 295–315, origins of the ecotypes [138]. In contrast, several candidate 2010. International Journal of Ecology 13

[2] D. Schluter, “Ecology and the origin of species,” Trends in [22]E.P.Palkovacs,K.B.Dion,D.M.Post,andA.Caccone, Ecology and Evolution, vol. 16, no. 7, pp. 372–380, 2001. “Independent evolutionary origins of landlocked alewife [3] A. P. Hendry, “Ecological speciation! Or the lack thereof?” populations and rapid parallel evolution of phenotypic Canadian Journal of Fisheries and Aquatic Sciences, vol. 66, traits,” Molecular Ecology, vol. 17, no. 2, pp. 582–597, 2008. no. 8, pp. 1383–1398, 2009. [23] J. S. McKinnon and H. D. Rundle, “Speciation in nature: the [4] D. Schluter, “Evidence for ecological speciation and its threespine stickleback model systems,” Trends in Ecology and alternative,” Science, vol. 323, no. 5915, pp. 737–741, 2009. Evolution, vol. 17, no. 10, pp. 480–488, 2002. [5] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology [24] J. L. Gow, S. M. Rogers, M. Jackson, and D. Schluter, Letters, vol. 8, no. 3, pp. 336–352, 2005. “Ecological predictions lead to the discovery of a benthic- [6] D. Schluter and L. M. Nagel, “Parallel speciation by natural limnetic sympatric species pair of threespine stickleback in selection,” The American Naturalist, vol. 146, no. 2, pp. 292– Little Quarry Lake, British Columbia,” Canadian Journal of 301, 1995. Zoology, vol. 86, no. 6, pp. 564–571, 2008. [7] J. Clausen, Stages in the Evolution of Plant Species,Cornell [25] K. Johannesson, “Inverting the null-hypothesis of speciation: University Press, Ithaca, NY, USA, 1951. a marine snail perspective,” Evolutionary Ecology, vol. 23, no. [8] D. B. Lowry, J. L. Modliszewski, K. M. Wright, C. A. Wu, and 1, pp. 5–16, 2009. J. H. Willis, “The strength and genetic basis of reproductive [26] K. Østbye, P.A. Amundsen, L. Bernatchez et al., “Parallel evo- isolating barriers in flowering plants,” Philosophical Transac- lution of ecomorphological traits in the European whitefish tions of the Royal Society B: Biological Sciences, vol. 363, no. Coregonus lavaretus (L.) species complex during postglacial 1506, pp. 3009–3021, 2008. times,” Molecular Ecology, vol. 15, no. 13, pp. 3983–4001, [9] G. L. Stebbins, Variation and Evolution in Plants,Cornell 2006. University Press, Ithaca, NY, USA, 1950. [27] U. Strecker, B. Hausdorf, and H. Wilkens, “Parallel speciation in Astyanax cave fish (Teleostei) in Northern Mexico,” [10] C. Lexer, R. R. Randell, and L. H. Rieseberg, “Experimental Molecular Phylogenetics and Evolution, vol. 62, no. 1, pp. 62– hybridization as a tool for studying selection in the wild,” 70, 2011. Ecology, vol. 84, no. 7, pp. 1688–1699, 2003. [28] R. Espanhol, P. R. Almeida, and M. J. Alves, “Evolutionary [11] L. H. Rieseberg and B. K. Blackman, “Speciation genes in history of lamprey paired species Lampetra fluviatilis (L.) plants,” Annals of Botany, vol. 106, no. 3, pp. 439–455, 2010. and Lampetra planeri (Bloch) as inferred from mitochondrial [12] D. Levin, “The recurrent origin of plant races and species,” DNA variation,” Molecular Ecology, vol. 16, no. 9, pp. 1909– Systematic Botany, vol. 26, no. 2, pp. 197–204, 2001. 1924, 2007. [13] D. Pigeon, A. Chouinard, and L. Bernatchez, “Multiple [29] M. E. Arnegard, S. M. Bogdanowicz, and C. D. Hopkins, modes of speciation involved in the parallel evolution of sym- “Multiple cases of striking genetic similarity between alter- patric morphotypes of lake whitefish (Coregonus clupeaform- nate electric fish signal morphs in sympatry,” Evolution, vol. is, Salmonidae),” Evolution, vol. 51, no. 1, pp. 196–205, 1997. 59, no. 2, pp. 324–343, 2005. [14] A. G. Hudson, P. Vonlanthen, and O. Seehausen, “Rapid [30] T. Kingston and S. J. Rossiter, “Harmonic-hopping in Wal- parallel adaptive radiations from a single hybridogenic lacea’s bats,” Nature, vol. 429, no. 6992, pp. 654–657, 2004. ancestral population,” Proceedings of the Royal Society B: [31] R. Yukilevich and J. R. True, “African morphology, behavior Biological Sciences, vol. 278, no. 1702, pp. 58–66, 2011. and phermones underlie incipient sexual isolation between [15] H. D. Rundle, L. Nagel, J. W. Boughman, and D. Schluter, us and Caribbean Drosophila melanogaster,” Evolution, vol. “Natural selection and parallel speciation in sympatric 62, no. 11, pp. 2807–2828, 2008. sticklebacks,” Science, vol. 287, no. 5451, pp. 306–308, 2000. [32] J. Arendt and D. Reznick, “Convergence and parallelism [16] J. W. Boughman, H. D. Rundle, and D. Schluter, “Parallel reconsidered: what have we learned about the genetics of evolution of sexual isolation in sticklebacks,” Evolution, vol. adaptation?” Trends in Ecology and Evolution, vol. 23, no. 1, 59, no. 2, pp. 361–373, 2005. pp. 26–32, 2008. [17] J. S. McKinnon, S. Mori, B. K. Blackman et al., “Evidence for [33] J. B. Losos, “Convergence, adaptation, and constraint,” ecology’s role in speciation,” Nature, vol. 429, no. 6989, pp. Evolution, vol. 65, no. 7, pp. 1827–1840, 2011. 294–298, 2004. [34] D. B. Wake, M. H. Wake, and C. D. Specht, “Homoplasy: [18] P. Conde-Pad´ın, R. Cruz, J. Hollander, and E. Rolan-Alvarez,´ from detecting pattern to determining process and mech- “Revealing the mechanisms of sexual isolation in a case anism of evolution,” Science, vol. 331, no. 6020, pp. 1032– of sympatric and parallel ecological divergence,” Biological 1035, 2011. Journal of the Linnean Society, vol. 94, no. 3, pp. 513–526, [35] D. E. Soltis and P. S. Soltis, “Molecular data and the dynamic 2008. nature of polyploidy,” Critical Reviews in Plant Sciences, vol. [19] K. Johannesson, M. Panova, P. Kemppainen, C. Andre,´ E. 12, no. 3, pp. 243–273, 1993. Rolan-Alvarez, and R. K. Butlin, “Repeated evolution of [36] B. L. Gross and L. H. Rieseberg, “The ecological genetics of reproductive isolation in a marine snail: unveiling mecha- homoploid hybrid speciation,” The Journal of Heredity, vol. nisms of speciation,” Philosophical Transactions of the Royal 96, no. 3, pp. 241–252, 2005. Society B: Biological Sciences, vol. 365, no. 1547, pp. 1735– [37] J. A. Coyne and H. A. Orr, Speciation, Sinauer Associates, 1747, 2010. Sunderland, Mass, USA, 2004. [20] P. Nosil, B. Crespi, and C. Sandoval, “Host-plant adaptation [38] J. Mallet, “Hybrid speciation,” Nature, vol. 446, no. 7133, pp. drives the parallel evolution of reproductive isolation,” 279–283, 2007. Nature, vol. 417, no. 6887, pp. 440–443, 2002. [39] E. B. Rosenblum and L. J. Harmon, “”Same same but [21]J.Q.RichmondandT.W.Reeder,“Evidenceforparallel different”: replicated ecological speciation at white sands,” ecological speciation in scincid lizards of the Eumeces Evolution, vol. 65, no. 4, pp. 946–960, 2011. skiltonianus species group (Squamata: Scincidae),” Evolution, [40] T. L. Turner, E. C. Bourne, E. J. von Wettberg, T. T. Hu, vol. 56, no. 7, pp. 1498–1513, 2002. and S. V. Nuzhdin, “Population resequencing reveals local 14 International Journal of Ecology

adaptation of Arabidopsis lyrata to serpentine soils,” Nature rapid evolution of copper tolerance in Agrostis stolonifera,” Genetics, vol. 42, no. 3, pp. 260–263, 2010. Heredity, vol. 34, no. 2, pp. 165–187, 1975. [41] C. Brochmann, W. Lobin, P. Sunding, and O. Stabbetorp, [56] N. Hannon and A. D. Bradshaw, “Evolution of salt tolerance “Parallel ecoclinal evolution and taxonomy of F’rankenia in two coexisting species of grass,” Nature, vol. 220, no. 5174, (Frankeniaceae) in the Cape Verde Islands, W Africa,” Nordic pp. 1342–1343, 1968. Journal of Botany, vol. 15, no. 6, pp. 603–623, 1995. [57] A. V. Venables and D. A. Wilkins, “Salt tolerance in pasture [42] T. Yokoo, S. Kobayashi, K. Oginuma et al., “Genetic structure grasses,” New Phytologist, vol. 80, no. 3, pp. 613–622, 1978. among and within populations of the serpentine endemic [58] M. K. Nicholls and T. McNeilly, “The possible polyphyletic Heteropappus hispidus ssp. leptocladus (Compositae),” Bio- origin of copper tolerance in Agrostis tenuis (Gramineae),” chemical Systematics and Ecology, vol. 37, no. 4, pp. 275–284, Plant Systematics and Evolution, vol. 140, no. 2-3, pp. 109– 2009. 117, 1982. [43] X. Thibert-Plante and A. P. Hendry, “The consequences of [59] S. S. Karataglis, “Gene flow in Agrostis tenuis (Poaceae),” phenotypic plasticity for ecological speciation,” Journal of Plant Systematics and Evolution, vol. 134, no. 1-2, pp. 23–31, Evolutionary Biology, vol. 24, no. 2, pp. 326–342, 2011. 1980. [44] N. Vallin and A. Qvarnstrom,¨ “Learning the hard way: imprinting can enhance enforced shifts in habitat choice,” [60] E. Simon and C. Lefebvre,` “Aspects of heavy-metal tolerance International Journal of Ecology, vol. 2011, Article ID 287532, in Agrostis tenuis Sibth, Festuca ovina and Armeria maritima 7 pages, 2011. (Mill) Willd,” Oecologia Plantarum, vol. 12, no. 2, pp. 95–110, [45] S. R. Moss and J. H. Clarke, “Inheritance of herbicide resis- 1977. tanceinblack-graa(Alopecurus myosuroides) and responses [61] X. Vekemans and C. Lefebvre,“Ontheevolutionofheavy-` of the weed to a range of herbicides,” Tech. Rep. 116, Home metal tolerant populations in Armeria maritima: evidence Grown Cereals Authority, London, UK, 1995. from allozyme variation and reproductive barriers,” Journal [46] G. Cavan, P. Biss, and S. R. Moss, “Localized origins of Evolutionary Biology, vol. 10, no. 2, pp. 175–191, 1997. of herbicide resistance in Alopecurus myosuroides,” Weed [62] H. Baumbach and F. H. Hellwig, “Genetic variation within Research, vol. 38, no. 3, pp. 239–245, 1998. and among metal-tolerant and non-tolerant populations of [47] C. Delye,´ E.´ Calmes,andA.Mat` ejicek,´ “SNP markers Armeria maritima (Mill.) Willd. s.l. (Plumbaginaceae) in for black-grass (Alopecurus myosuroides Huds.) genotypes Central and Northeast Germany,” Plant Biology, vol. 5, no. resistant to acetyl CoA-carboxylase inhibiting herbicides,” 2, pp. 186–193, 2003. Theoretical and Applied Genetics, vol. 104, no. 6-7, pp. 1114– [63] H. Baumbach and F. H. Hellwig, “Genetic differentiation 1120, 2002. of metallicolous and non-metallicolous Armeria maritima [48] C. Delye,´ X. Q. Zhang, C. Chalopin, S. Michel, and S. (Mill.) Willd. taxa (Plumbaginaceae) in Central Europe,” B. Powles, “An isoleucine residue within the carboxyl- Plant Systematics and Evolution, vol. 269, no. 3-4, pp. 245– transferase domain of multidomain acetyl-coenzyme A car- 258, 2007. boxylase is a major determinant of sensitivity to aryloxyphe- [64] A. B. N. Berglund and A. Westerbergh, “Two postglacial noxypropionate but not to cyclohexanedione inhibitors,” immigration lineages of the polyploid Cerastium alpinum Plant Physiology, vol. 132, no. 3, pp. 1716–1723, 2003. (Caryophyllaceae),” Hereditas, vol. 134, no. 2, pp. 171–183, [49] C. Delye,´ X. Zhang, S. Michel, A. Matejicek,´ and S. B. 2001. Powles, “Molecular bases for sensitivity to acetyl-coenzyme [65] A. B. Nyberg Berglund, S. Dahlgren, and A. Westerbergh, A carboxylase inhibitors in black-grass,” Plant Physiology, vol. “Evidence for parallel evolution and site-specific selection 137, no. 3, pp. 794–806, 2005. of serpentine tolerance in Cerastium alpinum during the [50] C. Delye,´ S. Michel, A. Berard´ et al., “Geographical variation colonization of Scandinavia,” New Phytologist, vol. 161, no. in resistance to acetyl-coenzyme A carboxylase-inhibiting 1, pp. 199–209, 2004. herbicides across the range of the arable weed Alopecurus [66] D. W. Kyhos, “The independent aneuploid origin of two myosuroides (black-grass),” New Phytologist, vol. 186, no. 4, species of Chaenactis (Compositae) from a common ances- pp. 1005–1017, 2010. tor,” Evolution, vol. 19, no. 1, pp. 26–43, 1965. [51] Y. Menchari, C. Delye,´ and V. Le Corre, “Genetic variation and population structure in black-grass (Alopecurus [67] J. Gasquez, A. A. Mouemar, and H. Darmency, “Triazine her- myosuroide Huds.), a successful, herbicide-resistant, annual bicide resistance in Chenopodium album L.: occurrence and grass weed of winter cereal fields,” Molecular Ecology, vol. 16, characteristics of an intermediate biotype,” Pest Management no. 15, pp. 3161–3172, 2007. Science, vol. 16, no. 4, pp. 392–396, 1985. [52] S. A. Al-Hiyaly, T. McNeilly, and A. D. Bradshaw, “The effects [68] J. Gasquez and J. P. Compoint, “Isoenzymatic variations of zinc contamination from electricity pylons—evolution in in populations of Chenopodium album L. resistant and a replicated situation,” New Phytologist, vol. 110, no. 4, pp. susceptible to triazines,” Agro-Ecosystems,vol.7,no.1,pp. 571–580, 1988. 1–10, 1981. [53] S. A. K. Al-Hiyaly, T. McNeilly, A. D. Bradshaw, and A. M. [69] S. I. Warwick and P. B. Marriage, “Geographical variation in Mortimer, “The effect of zinc contamination from electricity populations of Chenopodium album resistant and susceptible pylons. Genetic constraints on selection for zinc tolerance,” to atrazine. I. Between- and within- population variation Heredity, vol. 70, no. 1, pp. 22–32, 1993. in growth and response to atrazine,” Canadian Journal of [54] S. K. Jain and A. D. Bradshaw, “Evolutionary divergence Botany, vol. 60, no. 4, pp. 483–493, 1982. among adjacent plant populations I. The evidence and its [70] P. Bettini, S. McNally, M. Sevignac et al., “Atrazine resistance theoretical analysis,” Heredity, vol. 21, no. 3, pp. 407–441, in Chenopodium album: low and high levels of resistance to 1966. the herbicide are related to the same chloroplast psbA gene [55]L.Wu,A.D.Bradshaw,andD.A.Thurman,“Thepotential mutation,” Plant Physiology, vol. 84, no. 4, pp. 1442–1446, for evolution of heavy metal tolerance in plants. III. The 1987. International Journal of Ecology 15

[71] S. Andersson, “Phenotypic selection in a population of Crepis [87] N. Rajakaruna and J. Whitton, “Trends in the evolution of tectorum ssp. pumila (Asteraceae),” Canadian Journal of edaphic specialists with an examples of parallel evolution Botany, vol. 70, no. 1, pp. 89–95, 1992. in the Lasthenia californica complex,” in Plant Adaptation: [72] S. Andersson, “Differences in the genetic basis of leaf Molecular Genetics and Ecology, Q. C. B. Cronk, J. Whitton, R. dissection between two populations of Crepis tectorum H. Ree, and I. E. P. Taylor, Eds., pp. 103–110, NRC Research (Asteraceae),” Heredity, vol. 75, no. 1, pp. 62–69, 1995. Press, Ottawa, Ontario, Canada, 2004. [73] E. J. Bush and S. C. H. Barrett, “Genetics of mine invasions [88] K. Vijverberg, L. Lie, and K. Bachmann, “Morphological, by Deschampsia cespitosa (Poaceae),” Canadian Journal of evolutionary and taxonomic aspects of Australian and Botany, vol. 71, no. 10, pp. 1336–1348, 1993. New Zealand Microseris (Asteraceae),” Australian Journal of [74] K. K. Nkongolo, A. Deck, and P. Michael, “Molecular and Botany, vol. 50, no. 1, pp. 127–143, 2002. cytological analyses of Deschampsia cespitosa populations [89] K. Vijverberg, T. H. M. Mes, and K. Bachmann, “Chloroplast from Northern Ontario (Canada),” Genome,vol.44,no.5, DNA evidence for the evolution of Microseris (Asteraceae) in pp. 818–825, 2001. Australia and New Zealand after long-distance dispersal from [75] S. A. Foster, G. E. McKinnon, D. A. Steane, B. M. Potts, and Western North America,” American Journal of Botany, vol. 86, R. E. Vaillancourt, “Parallel evolution of dwarf ecotypes in no. 10, pp. 1448–1463, 1999. the forest tree Eucalyptus globulus,” New Phytologist, vol. 175, [90]K.Vijverberg,P.Kuperus,J.A.J.Breeuwer,andK.Bach- no. 2, pp. 370–380, 2007. ff mann, “Incipient adaptive radiation of New Zealand and [76] J. Roncal, “Habitat di erentiation of sympatric Geonoma Australian Microseris (Asteraceae): an amplified fragment macrostachys (Arecaceae) varieties in Peruvian lowland length polymorphism (AFLP) study,” Journal of Evolutionary forests,” JournalofTropicalEcology, vol. 22, no. 4, pp. 483– Biology, vol. 13, no. 6, pp. 997–1008, 2000. 486, 2006. [91] F. Quattrocchio, J. Wing, K. van der Woude et al., “Molecular [77] C. Listabarth, “Pollination in Geonoma macrostachys and analysis of the anthocyanin2 gene of Petunia and its role in three congeners, G. acaulis, G. gracilis and G. interrupta,” the evolution of flower color,” The Plant Cell, vol. 11, no. 8, Botanica Acta, vol. 106, no. 6, pp. 496–506, 1993. pp. 1433–1444, 1999. [78] J. Roncal, Molecular phylogenetics of the palm tribe Geonom- eae,anddifferentiation of Geonoma macrostachys Western [92] M. E. Hoballah, T. Gubitz,¨ J. Stuurman et al., “Single gene- Amazonian varieties, Ph.D. thesis, Florida International Uni- mediated shift in pollinator attraction in Petunia,” Plant Cell, versity, Miami, Fla, USA, 2005. vol. 19, no. 3, pp. 779–790, 2007. [79] J. Noguchi and H. De-yuan, “Multiple origins of the Japanese [93] J. Gregor, “Experimental Taxonomy IV. Population differen- nocturnal Hemerocallis citrina var. vespertina (Asparagales: tiation in North American and European sea plantains allied Hemerocallidaceae): evidence from noncoding chloroplast to Plantago maritima L,” New Phytologist,vol.38,no.4,pp. DNA sequences and morphology,” International Journal of 293–322, 1939. Plant Sciences, vol. 165, no. 1, pp. 219–230, 2004. [94] I. M. Heap, “Multiple herbicide resistance in annual blue- [80] G. Turesson, “The genotypical response of the plant species grass (Poa annua),” WSSA Abstracts, vol. 42, p. 56, 1995. to the habitat,” Hereditas, vol. 3, no. 3, pp. 211–350, 1922. [95] H. Darmency and J. Gasquez, “Interpreting the evolution [81]N.Rajakaruna,B.G.Baldwin,R.Chan,A.M.Desrochers, of a triazine-resistant population of Poa annua L,” New B. A. Bohm, and J. Whitton, “Edaphic races and phyloge- Phytologist, vol. 95, no. 2, pp. 299–304, 1983. netic taxa in the Lasthenia californica complex (Asteraceae: [96] F. Perez,´ “Discordant patterns of morphological and genetic Heliantheae): an hypothesis of parallel evolution,” Molecular divergence in the closely related species Schizanthus hookeri Ecology, vol. 12, no. 6, pp. 1675–1679, 2003. and S. grahamii (Solanaceae),” Plant Systematics and Evolu- [82] R. Chan, B. G. Baldwin, and R. Ornduff, “Goldfields tion, vol. 293, no. 1–4, pp. 197–205, 2011. revisited: a molecular phylogenetic perspective on the evo- [97] A. Westerbergh and A. Saura, “The effect of serpentine on lution of Lasthenia (Compositae: Heliantheae sensu lato),” the population structure of Silene dioica (Caryophyllaceae),” International Journal of Plant Sciences, vol. 162, no. 6, pp. Evolution, vol. 46, no. 5, pp. 1537–1548, 1992. 1347–1360, 2001. [98] A. Westerbergh, “Serpentine and non-serpentine Silene [83] R. Chan, B. G. Baldwin, and R. Ornduff, “Cryptic goldfields: dioica plants do not differ in nickel tolerance,” Plant and Soil, a molecular phylogenetic reinvestigation of Lasthenia Califor- vol. 167, no. 2, pp. 297–303, 1994. nica sensu lato and close relatives (Compositae: Heliantheae sensu lato),” American Journal of Botany,vol.89,no.7,pp. [99] H. Schat, R. Vooijs, and E. Kuiper, “Identical major gene loci for heavy metal tolerances that have independently evolved in 1103–1112, 2002. ff [84] A. M. Desrochers and B. A. Bohm, “Biosystematic study of di erent local populations and subspecies of Silene vulgaris,” Lasthenia californica (Asteraceae),” Systematic Botany, vol. 20, Evolution, vol. 50, no. 5, pp. 1888–1895, 1996. no. 1, pp. 65–84, 1995. [100] H. Schat, E. Kuiper, W. M. Ten Bookum, and R. Vooijs, “A [85] N. Rajakaruna, M. Y. Siddiqi, J. Whitton, B. A. Bohm, general model for the genetic control of copper tolerance in andA.D.M.Glass,“Differential responses to Na+/K+ and Silene vulgaris: evidence from crosses between plants from ff Ca2+/Mg2+ in two edaphic races of the Lasthenia californica di erent tolerant populations,” Heredity,vol.70,no.2,pp. (Asteraceae) complex: a case for parallel evolution of physi- 142–147, 1993. ological traits,” New Phytologist, vol. 157, no. 1, pp. 93–103, [101] M. Bratteler, M. Baltisberger, and A. Widmer, “QTL analysis 2003. of intraspecific differences between two Silene vulgaris eco- [86] N. Rajakaruna, G. Bradfield, B. Bohm, and J. Whitton, types,” Annals of Botany, vol. 98, no. 2, pp. 411–419, 2006. “Adaptive differentiation in response to water stress by [102] G. Paliouris and T. C. Hutchinson, “Arsenic, cobalt and nickel edaphic races of Lasthenia californica (Asteraceae),” Interna- tolerances in two populations of Silene vulgaris (Moench) tional Journal of Plant Sciences, vol. 164, no. 3, pp. 371–376, Garcke from Ontario, Canada,” New Phytologist, vol. 117, no. 2003. 3, pp. 449–459, 1991. 16 International Journal of Ecology

[103] A. R. Kruckeberg, “Intraspecific variability in the response [118] H. D. Rundle, “A test of ecologically dependent postmating of certain native plant species to serpentine soil,” American isolation between sympatric sticklebacks,” Evolution, vol. 56, Journal of Botany, vol. 38, no. 6, pp. 408–419, 1951. no. 2, pp. 322–329, 2002. [104] M. S. Mayer and P. S. Soltis, “The evolution of serpentine [119] E. B. Taylor and J. D. McPhail, “Evolutionary history of an endemics: a chloroplast DNA phylogeny of the Streptanthus adaptive radiation in species pairs of threespine sticklebacks glandulosus complex (Cruciferae),” Systematic Botany, vol. (Gasterosteus): insights from mitochondrial DNA,” Biological 19, no. 4, pp. 557–574, 1994. Journal of the Linnean Society, vol. 66, no. 3, pp. 271–291, [105] M. S. Mayer and P. S. Soltis, “Intraspecific phylogeny analysis 1999. using ITS sequences: insights from studies of the Streptanthus [120] E. B. Taylor and J. D. McPhail, “Historical contingency glandulosus complex (Cruciferae),” Systematic Botany, vol. and ecological determinism interact to prime speciation in 24, no. 1, pp. 47–61, 1999. sticklebacks, Gasterosteus,” Proceedings of the Royal Society B: [106] M. S. Mayer and L. Beseda, “Reconciling taxonomy and Biological Sciences, vol. 267, no. 1460, pp. 2375–2384, 2000. phylogeny in the Streptanthus glandulosus complex (Brassi- [121] G. M. Kozak, M. L. Head, and J. W. Boughman, “Sexual caceae),” Annals of the Missouri Botanical Garden, vol. 97, no. imprinting on ecologically divergent traits leads to sexual 1, pp. 106–116, 2010. isolation in sticklebacks,” Proceedings of the Royal Society B: [107] A. R. Kruckeberg, “Variation in fertility of hybrids between Biological Sciences, vol. 278, no. 1718, pp. 2604–2610, 2011. isolated populations of the serpentine species, Streptanthus [122] E. B. Taylor, J. W. Boughman, M. Groenenboom, M. glandulosus,” Evolution, vol. 11, no. 2, pp. 185–211, 1957. Sniatynski, D. Schluter, and J. L. Gow, “Speciation in reverse: [108] L. Bernatchez, S. Renaut, A. R. Whiteley et al., “On the morphological and genetic evidence of the collapse of origin of species: insights from the ecological genomics of a three-spined stickleback (Gasterosteus aculeatus)species lake whitefish,” Philosophical Transactions of the Royal Society pair,” Molecular Ecology, vol. 15, no. 2, pp. 343–355, 2006. B: Biological Sciences, vol. 365, no. 1547, pp. 1783–1800, 2010. [123] A. P. Hendry and E. B. Taylor, “How much of the variation in adaptive divergence can be explained by gene flow? An [109] L. Bernatchez and J. J. Dodson, “Allopatric origin of sym- evaluation using lake-stream stickleback pairs,” Evolution, patric populations of lake whitefish (Coregonus clupeaformis) vol. 58, no. 10, pp. 2319–2331, 2004. as revealed by mitochondrial-DNA restriction analysis,” [124] D. Berner, A. C. Grandchamp, and A. P. Hendry, “Variable Evolution, vol. 44, no. 6, pp. 1263–1271, 1990. progress toward ecological speciation in parapatry: stickle- [110] S. M. Rogers and L. Bernatchez, “The genetic basis of back across eight lake-stream transitions,” Evolution, vol. 63, intrinsic and extrinsic post-zygotic reproductive isolation no. 7, pp. 1740–1753, 2009. jointly promoting speciation in the lake whitefish species [125] D. Berner, M. Roesti, A. P. Hendry, and W. Salzburger, complex (Coregonus clupeaformis),” Journal of Evolutionary “Constraints on speciation suggested by comparing lake- Biology, vol. 19, no. 6, pp. 1979–1994, 2006. stream stickleback divergence across two continents,” Molec- [111] S. M. Rogers, V. Gagnon, and L. Bernatchez, “Genetically ular Ecology, vol. 19, no. 22, pp. 4963–4978, 2010. based phenotype-environment association for swimming [126] A. P. Hendry, E. B. Taylor, and J. D. McPhail, “Adaptive behavior in lake whitefish ecotypes (Coregonus clupeaformis divergence and the balance between selection and gene flow: Mitchill),” Evolution, vol. 56, no. 11, pp. 2322–2329, 2002. lake and stream stickleback in the Misty system,” Evolution, [112] M. Rogers and L. Bernatchez, “The genetic architecture vol. 56, no. 6, pp. 1199–1216, 2002. of ecological speciation and the association with signa- [127] R. Kaeuffer,C.L.Peichel,D.I.Bolnick,andA.P.Hendry, tures of selection in natural lake whitefish (Coregonus sp. “Parallel and nonparallel aspects of ecological, phenotypic, Salmonidae) species pairs,” Molecular Biology and Evolution, and genetic divergence across replicate population pairs of vol. 24, no. 6, pp. 1423–1438, 2007. lake and stream stickleback,” Evolution,vol.66,no.2,pp. [113] N. Derome, P. Duchesne, and L. Bernatchez, “Parallelism in 402–418, 2012. gene transcription among sympatric lake whitefish (Core- [128] J. D. McPhail, “Speciation and the evolution of reproductive gonus clupeaformis Mitchill) ecotypes,” Molecular Ecology, isolation in the sticklebacks (Gasterosteus)ofsouth-western vol. 15, no. 5, pp. 1239–1249, 2006. British Columbia,” in The Evolutionary Biology of the Three- [114] S. Renaut, A. W. Nolte, S. M. Rogers, N. Derome, and L. spine Stickleback, M. A. Bell and S. A. Foster, Eds., pp. 399– Bernatchez, “SNP signature of selection on standing genetic 437, Oxford University Press, Oxford, UK, 1994. variation, and association with adaptive phenotypes in lake [129] J. A. M. Raeymaekers, M. Boisjoly, L. Delaire, D. Berner, K. whitefish species pairs (Coregonus spp.),” Molecular Ecology, Ras¨ anen,¨ and A. P. Hendry, “Testing for mating isolation vol. 20, no. 3, pp. 545–559, 2011. between ecotypes: laboratory experiments with lake, stream [115] J. Jeukens, D. Bittner, R. Knudsen, and L. Bernatchez, andhybridstickleback,”Journal of Evolutionary Biology, vol. “Candidate genes and adaptive radiation: insights from 23, no. 12, pp. 2694–2708, 2010. transcriptional adaptation to the limnetic niche among [130] R. Katja, D. Matthieu, L. J. Chapman, and A. P. Hendry, coregonine fishes (Coregonus spp., Salmonidae),” Molecular “Divergent selection and then what not: the puzzle of missing Biology and Evolution, vol. 26, no. 1, pp. 155–166, 2009. reproductive isolation in Misty lake and stream stickleback,” [116] K. Østbye, L. Bernatchez, T. F. Næsje, K.-J. M. Himberg, and International Journal of Ecology. In press. K. Hindar, “Evolutionary history of the European whitefish [131] P. F. Colosimo, K. E. Hosemann, S. Balabhadra et al., Coregonus lavaretus (L.) species complex as inferred from “Widespread parallel evolution in sticklebacks by repeated mtDNA phylogeography and gill-raker numbers,” Molecular fixation of ectodysplasin alleles,” Science, vol. 307, no. 5717, Ecology, vol. 14, no. 14, pp. 4371–4387, 2005. pp. 1928–1933, 2005. [117] P.-A. Amundsen, T. Bøhn, and G. H. Vaga,˚ “Gill raker [132] C. T. Miller, S. Beleza, A. A. Pollen et al., “cis-regulatory morphology and feeding ecology of two sympatric morphs of changes in Kit ligand expression and parallel evolution of European whitefish (Coregonus lavaretus),” Annales Zoologici pigmentation in sticklebacks and humans,” Cell, vol. 131, no. Fennici, vol. 41, no. 1, pp. 291–300, 2004. 6, pp. 1179–1189, 2007. International Journal of Ecology 17

[133] Y. F. Chan, M. E. Marks, F. C. Jones et al., “Adaptive evolution of pelvic reduction in sticklebacks by recurrent deletion of a Pitx1 enhancer,” Science, vol. 327, no. 5963, pp. 302–305, 2010. [134] E. Rolan-Alvarez,´ K. Johannesson, and J. Erlandsson, “The maintenance of a cline in the marine snail Littorina saxatilis: the role of home site advantage and hybrid fitness,” Evolution, vol. 51, no. 6, pp. 1838–1847, 1997. [135] R. Cruz, C. Vilas, J. Mosquera, and C. Garc´ıa, “Relative contribution of dispersal and natural selection to the maintenance of a hybrid zone in Littorina,” Evolution, vol. 58, no. 12, pp. 2734–2746, 2004. [136] K. Johannesson, B. Johannesson, and E. Rolan-Alvarez,´ “Morphological differentiation and genetic cohesiveness over a microenvironmental gradient in the marine snail Littorina saxatilis,” Evolution, vol. 47, no. 6, pp. 1770–1787, 1993. [137] E. Rolan-Alvarez,´ M. Carballo, J. Galindo et al., “Nonal- lopatric and parallel origin of local reproductive barriers between two snail ecotypes,” Molecular Ecology, vol. 13, no. 11, pp. 3415–3424, 2004. [138] H. Quesada, D. Posada, A. Caballero, P.Moran,´ and E. Rolan-´ Alvarez, “Phylogenetic evidence for multiple sympatric ecological diversification in a marine snail,” Evolution, vol. 61, no. 7, pp. 1600–1612, 2007. [139] J. Galindo, P. Moran,´ and E. Rolan-Alvarez,´ “Comparing geographical genetic differentiation between candidate and noncandidate loci for adaptation strengthens support for parallel ecological divergence in the marine snail Littorina saxatilis,” Molecular Ecology, vol. 18, no. 5, pp. 919–930, 2009. [140] R. Cruz, C. Vilas, J. Mosquera, and C. Garc´ıa,“Thecloserela- tionship between estimated divergent selection and observed differentiation supports the selective origin of a marine snail hybrid zone,” Journal of Evolutionary Biology, vol. 17, no. 6, pp. 1221–1229, 2004. [141] R. Cruz, M. Carballo, P. Conde-Pad´ın, and E. Rolan-´ Alvarez, “Testing alternative models for sexual isolation in natural populations of Littorina saxatilis: indirect support for by-product ecological speciation?” Journal of Evolutionary Biology, vol. 17, no. 2, pp. 288–293, 2004. [142] C. P.Sandoval, “The effects of the relative geographic scales of gene flow and selection on morph frequencies in the walking- stick Timema cristinae,” Evolution, vol. 48, no. 6, pp. 1866– 1879, 1994. [143] P. Nosil and B. J. Crespi, “Does gene flow constrain adaptive divergence or vice versa? A test using ecomorphology and sexual isolation in Timema cristinae walking-sticks,” Evolu- tion, vol. 58, no. 1, pp. 102–112, 2004. [144] P. Nosil, S. P. Egan, and D. J. Funk, “Heterogeneous genomic differentiation between walking-stick ecotypes: “isolation by adaptation” and multiple roles for divergent selection,” Evolution, vol. 62, no. 2, pp. 316–336, 2008. [145] S. Gavrilets, Fitness Landscapes and the Origin of Species, Princeton University Press, Princeton, NJ, USA, 2004. [146] V. Savolainen, M. Anstett, C. Lexer et al., “Sympatric speciation in palms on an oceanic island,” Nature, vol. 441, no. 7090, pp. 210–213, 2006. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 508458, 11 pages doi:10.1155/2012/508458

Review Article From Local Adaptation to Speciation: Specialization and Reinforcement

Thomas Lenormand

CEFE-UMR 5175, 1919 Route de Mende, 34293 Montpellier Cedex 5, France

Correspondence should be addressed to Thomas Lenormand, [email protected]

Received 29 July 2011; Revised 21 November 2011; Accepted 12 December 2011

Academic Editor: Marianne Elias

Copyright © 2012 Thomas Lenormand. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Local adaptation is the first step in the process of ecological speciation. It is, however, an unstable and dynamic situation. It can be strengthened by the occurrence of alleles more specialized to the different habitats or vanish if generalist alleles arise by mutations and increase in frequency. This process can have complicated dynamics as specialist alleles may be much more common and may maintain local adaptation for a long time. Thus, even in the absence of an absolute fitness tradeoff between habitats, local adaptation may persist a long time before vanishing. Furthermore, several feedback loops can help to maintain it (the reinforcement, demographic, and recombination loops). This reinforcement can occur by modifying one of the three fundamental steps in a sexual life cycle (dispersal, syngamy, meiosis), which promotes genetic clustering by causing specific genetic associations. Distinguishing these mechanisms complements the one- versus two-allele classification. Overall, the relative rates of the two processes (specialization and reinforcement) dictate whether ecological speciation will occur.

1. Introduction unfitness can be caused by alternative adaptive peaks in the same habitat, whereas local adaptation usually occurs with The debate over speciation is not new [1–7] and is a com- a single peak that differs in different habitats, which is quite plex subject as speciation represents a “cluster of theories different, even if not always recognized as such [1, 8, 11]. Of woven from many strands” [7] that can be approached course, both phenomenon are not exclusive and evaluating from different angles: sympatry versus allopatry, intrinsic their relative importance is often controversial [12], as both versus extrinsic selection, premating versus postmating isola- produce indistinguishable spatial patterns [13]. We will see tion, isolation versus adaptation, one-allele versus two-allele that the route from local adaptation to speciation has been mechanisms, primary versus secondary contact, genic versus explored in a somewhat separate theoretical corpus that still genomic, and so forth. This diversity of possible approaches has to be fully incorporated to a microevolutionary view of can even lead to a synthesis based on the idea of a menu with speciation. different possible options for each course [8]. Environmental heterogeneity is pervasive and local ad- One view has reemphasized Darwin’s view that speci- aptation is the direct selective consequence of different ation was about adaptation to different habitats or niches selection pressures occurring in different places [14]. Local [2, 4, 9, 10]. At a microevolutionary scale, this process often adaptation—the greater average fitness of local individuals starts within species as local adaptation. This can be the compared to immigrants—occurs whenever the “grain” beginning of future divergence and eventually speciation. of habitat is sufficiently coarse compared to the scale of However, the basic process of local adaptation is often seen dispersal [15, 16]. It has various consequences ranging from as too “preliminary” to stand at the core of a theory for niche evolution, evolution of specialization, to ecological speciation. Indeed, “mere differential adaptation (...)does speciation. In the context of speciation, local adaptation is not constitute species” [1] and hybrid unfitness is often the a selective context that is universal: it applies to sexual or main selective scenario envisioned to be necessary. Hybrid asexual species alike, contrary to selection pressures caused 2 International Journal of Ecology by hybrid incompatibilities. However, it also has specific scope for the long term maintenance of locally specialized features. First it is a dynamic and gradual process. Local genotypes. Overall, locally adapted genotypes are unlikely adaptation can constantly strengthen if new alleles better to arise “on every bush” [21], yet are expected to arise adapted to local conditions arise and increase in frequency, or frequently. disappear if generalist genotypes spread or if differentiation is swamped by gene flow. In contrast, a situation of secondary 2.2. Local Adaptation and the Origin of Tradeoffs. One basic contact is not a gradual process, but often the abrupt ex- process by which local adaptation arises is when, within a posure of genetic incompatibilities that have accumulated large enough habitat, an allele increases in frequency that is independently in allopatry. It results in tension zones that beneficial inside but deleterious/neutral outside this habitat. can be stable over long periods of time [17]. Second, The condition for this increase in frequency is determined hybrids between two locally adapted parental genotypes are by the strength of selection inside and outside, the size of often not the worst genotype in any of the environments. the pocket relative to the scale of dispersal and by possible With local adaptation, there is not selection against hybrids gene flow asymmetries between habitats [15, 22]. Any allele per se: there is selection against alleles or genotypes that with a sufficient benefit inside will increase in frequency are not locally favorable. This distinction has important no matter if it presents strong deleterious effects outside. consequences understanding how reinforcement works in This process does not necessarily favor alleles exhibiting the context of local adaptation compared to the case of little antagonistic effect across habitats; anything goes that reinforcement to avoid producing unfit hybrids. Here, I use is sufficiently favorable locally. At the same locus, allele reinforcement sensu lato to mean the evolution of premating replacement can occur in both directions, favoring either isolation resulting from selection against hybrids or locally stronger or weaker specialization [23]. Another possibility is maladapted genotypes [18]. In this paper, I will focus on that neutral mutations drift at high frequency locally despite these two topics and try to show that their outcomes are being deleterious elsewhere [14, 24, 25]. However, even if this often less straightforward than early models have suggested, can occur [26], it requires very limited gene flow and may be and that the theory of local adaptation has to be more fully globally less conducive to strong local adaptation. integrated into a global view of speciation and diversification. This process has no reason to stop and lead to a process of “amelioration” [27] whereby new favorable alleles replace 2. Topic 1—The Dynamics of Local Adaptation previous ones at a given locus [23, 28–30], modifier alleles at new loci evolve to correct for deleterious side effects Because niche specialization can occur in the presence of of previous ones [31–33], duplications and new functions gene flow in primary contact and does not require time in can arise [34–36], and so forth. The question is whether isolation to evolve genetic incompatibilities, it results in a this amelioration will lead to the evolution of generalist more dynamical process, where further niche specialization genotypes that can accommodate all habitats or whether may or may not occur, with direct consequences on rein- local adaptation will strengthen and lead to specialized forcement towards more isolation. ecotypes that have diverged at a large number of loci. The answer is not straightforward. The first approach is to build a model imposing a trade-off curve between habitats. Whether 2.1. Habitats. Models of local adaptation and ecological specialists or generalists evolve depends then on the shape speciation most often start by assuming that there are of this trade-off curve [37, 38]. Globally speaking, more different habitats exchanging migrants (with the rate of mi- concave curves facilitates the evolution of specialists, whereas gration corresponding to the cases of sympatry, parapatry more convex ones favor generalists, and sometimes both can and allopatry [3]). However, the definition of habitat can coexist [37, 38]. However, there is no clear ultimate reason be problematic. Field ecologists would certainly define it by for choosing one curve over another or not allowing these a combination of biotic/abiotic conditions in a landscape. trade-off curves to evolve as well. Another approach that has For instance, when thinking about land plants, they would been much less explored would be to introduce a distribution categorize soil type, moisture, temperature, light, slope, of mutation effects, where specialist mutations are much disturbance regime, and so forth. From an ecological genetics more common than generalist mutations (having to solve perspective, a first difficulty is to define these variations at the problem of a single habitat) and are constantly appearing the relevant scale, that is, at the scale of dispersal of the at different loci maintaining local adaptation. In the latter focal species [15, 16]. A second difficulty is to account for situation, and if no other constraint is involved, the outcome distance. Because dispersal is most often distance limited, in the very long run would be nearly perfect adaptation to the the spatial configuration makes a difference in terms of different habitats. There are however three positive feedback habitat definition. This is well known in complex or mosaic loops that are likely to interfere with this outcome and favor habitats [19], but is true even in very simple landscapes increased specialization. [20]. A given ecological condition will be less prone to local adaptation if close to a habitat boundary than if surrounded by identical conditions. Boundaries may even favor the 2.3. The Demographic Feedback Loop. The first feedback local evolution of generalists, something that contradicts is demographic. As far as local adaptation causes a local the simple competition exclusion principle [20]. Finally, increase in density, it will also make life easier for more conditions also vary through time, which strongly limit the specialist alleles to increase in frequency. This is due to International Journal of Ecology 3 the fact that density differences cause asymmetric gene Reciprocally, strong local adaptation increases the selection flow, which gives an advantage to alleles favored in the pressure to reinforce it, so that both phenomena can act in denser habitat [14, 22].Becauseofgeneflow,analleletoo concert in a positive feedback loop [9, 37, 38, 48]. detrimental outside the habitat where it is favored may be unable to increase in frequency. Despite having the potential 2.6. The Relative Dynamics of Local Adaptation and Rein- to contribute to specialization if the habitat was isolated, it forcement. Ultimately, ecological speciation will result only if remains at mutation-selection balance. Such alleles may be reinforcement occurs quickly enough compared to the evo- very common; I term them “contending” alleles. If density lution of generalists and the breakdown of local adaptation. becomes higher in this habitat, contending alleles may now The different feedback loops mentioned above will tend to be able to increase in frequency and contribute to local favor this outcome, but may not be strong enough to lead adaptation. This increase in density is likely to occur at least to speciation. For instance, the evolution of habitat choice, in some cases when local adaptation takes place. The positive reduced dispersal, selfing, and so forth, can strongly reduce feedback loop occurs because a local increase in density the chance that a generalist allele would spread, but the actual causes more and more alleles that are locally beneficial to outcome depends on the relative rates of reinforcement and be recruited, which strengthens local adaptation, increases loss of specialization. This dynamical issue is not something local density and facilitates further the increase in frequency that has been fully appreciated in the context of ecological of other locally beneficial alleles (the reverse can also occur, speciation (but see [38] in the context of a fixed tradeoff). which is known as migration meltdown [14, 39]). Because It differs from the situation of reinforcement in a tension the ratio of density is as effective as the square of selection zone by the fact that it is less stable and very sensitive ratio inside versus outside the habitat [14, 22], this effect to several feedbacks. More work is certainly needed to is likely to be strong in natural populations. Conversely clearly delineate the conditions favoring speciation in this adapting to a sink habitat makes it very difficult for the same context. reason as shown by niche expansion models [40–42]. 3. Topic 2—The Reinforcement of 2.4. The Recombination Feedback Loop. The second feedback Local Adaptation loop is due to indirect selection among loci directly involved in local adaptation. When a locally beneficial allele increases “Individuals that have parents selected in the same in frequency, it will favor the spread at closely linked loci habitat and that stay in that habitat are more of other locally beneficial alleles. This is due to the fact likely to have genes appropriate to that environ- that dispersal generates linkage disequilibrium between loci ment than another randomly chosen individual. that share a similar frequency variation across habitats, as Thus, mating locally and staying in the same expected for two loci involved in local adaptation to the same habitat is always favored from the point of view habitat. This linkage disequilibrium translates into indirect of a continually evolving genome”[25]. selection that mutually benefits the locally adapted alleles In the early 70s, several models have addressed the at the two loci [43, 44]. For instance, a contending allele problem of reinforcement of local adaptation. The first was could start to increase in frequency if it becomes sufficiently proposed by Antonovics [49]. This model showed that, linked to another locus involved in the local adaptation. This because mating at random is risky in the context of local phenomenon generates a positive feedback loop within the adaptation, evolution favors that like mates with like and genome where locally adapted alleles are more likely to be that selfing is even safer to maintain local adaptation. Then recruited in genomic regions already harboring a previous Balkau and Feldman [50] showed that, in the context of locus involved in the local adaptation. It can generate local adaptation, migrating, or sending offspring elsewhere “genomic islands” of local adaptation that extend further and is likely to decrease an individual’s fitness or that of further [45, 46], a specific process that can gradually produce its offspring. With a modifier model, they showed that strong genetic divergence at many contiguous loci in linkage this effect caused indirect selection to reduce dispersal as disequilibrium [1, 47]. much as possible. Finally, Slatkin [43] suggested and D. Charlesworth and B. Charlesworth confirmed [51] that sex 2.5. The Reinforcement Feedback Loop. The third feedback and recombination is likely to break combination of genes loop is due to reinforcement, that is, the evolution of that have been locally selected for and should be selected traits promoting premating isolation between differentially against in the context of local adaptation. These findings locally adapted genotypes. Reinforcement tends to make have since been constantly reported or given as examples of life easier for locally beneficial alleles: it allows more alleles “one-allele” mechanisms that are likely to drive the evolution contributing to local adaptation to be recruited and locally of isolation in parapatry. (The one-allele versus two-allele beneficial alleles to reach higher frequencies. For instance, classification refers to cases where a single or two different contending alleles could be recruited if habitat choice started alleles spread at the modifier locus to promote genetic to evolve. In effect, habitat choice minimizes the possible clustering [21]. This classification is very useful despite negative fitness that an allele can have in habitats that leading to some complications (see [52]andTable 1 note are different from the habitat where it is favored. Thus, 7)). In this section, I will reconsider these conclusions in the reinforcement is likely to promote increased specialization. light of more recent models on the evolution of assortative 4 International Journal of Ecology

Table 1: Classification of reinforcement mechanisms based on the life stage modified and their consequences on genetic associations involved in genetic clustering. This classification is orthogonal to the one- versus two-allele classification.

Life stage modified Dispersal Syngamy Meiosis Prominent biological Habitat choice Mate choice Sex/asex example Increased frequency Genetic clustering at local differences between Increased homozygosity Increased linkage disequilibrium adaptation loci through1 habitats Usual population genetic parameter measuring Fst (differentiation) Fis(departure from Hardy-Weinberg) D (linkage disequilibrium) clustering Primary effect of modifier Change in frequency Change in within-locus associations2 Change in between-loci associations on local adaptation loci Primary modifier C C C association3 ma,Ø ma,a mab,Ø Increased differentiation Directly Indirectly4 Indirectly4 Notable complication Kin selection Recessivity6 Negative epistasis preventing clustering5 An allele reducing dispersal An allele causes assortment based on An allele causes a reduction in “One-allele” examples or causing preference to self-similarity (Figure 1) recombination (Figure 4) natal habitat (Figure 2) Allele 1 causes preference to Allele 1 (inversion) causes linkage Allele 1 cause preference to phenotype “Two-allele” examples habitat 1 and allele 2 to in group of genes 1 and allele 2 1 and allele 2 to phenotype 27 habitat 2 (noninversion) in group of genes 2 1 This would also apply to loci involved in genetic incompatibilities in a secondary contact. 2Unless mating is selective and causes a direct advantage to locally adapted alleles (i.e., it changes frequency at the local adaptation loci), as foundinmodel involving sexual selection [8, 103]. 3Notation as in [104, 105], where m is the modifier locus and a, b the local adaptation loci. “Primary” association refers to the fact that the phenotypic effect of the modifier causes first a direct change on the genetic composition of the population (on frequency, within or between loci associations), which may then change the efficacy of selection. Eventually, a modifier promoting clustering will end up associated to the beneficial allele locally (Cma,Ø > 0). See Figures 1, 2 and 4 for examples. 4Indirectly by increasing the variance in fitness and the efficacy of selection. 5Besides possible direct costs relative to the strategy used (e.g., cost of finding a mate or the right habitat). Different traits are exposed to a variety of other selective effects (see text). 6Which generates inbreeding depression. 7Phenotype 1 and 2 may result from alleles at the adaptation locus or to another unrelated marker trait. Similarly in a “one-allele” model, self-similarity may be evaluated in reference to a marker trait at another locus than the modifier or the local adaptation locus. In both cases, the marker trait has to diverge in the two incipient species, which is essentially a two-allele mechanism. Thus, with three locus like this, the one- versus two-allele distinction is made more complicated by the fact that the marker trait must diverge (two-allele), but the modifier of the strength of assortment need not (it can be one- or two-allele) [52]. Another complication of the one- and two-allele classification arises when the locus exposed to postzygotic selection also causes premating isolation (as seen in so- called “magic trait” models). This can be thought as the limit where the loci causing prezygotic isolation and postzygotic selection become confounded. mating [53], dispersal [54, 55], and recombination [11, 45, 3.1. The Evolution of Selfing and Assortative Mating. The 56] in the context of local adaptation. Contrary to what is evolution of nonrandom mating has been extensively studied commonly thought [8, 21, 57], I will show that these one- in the context of reinforcement and reproductive isolation allele mechanisms do not inevitably lead to speciation, even [18, 68–70]. However, it has also been extensively studied in the absence of direct cost. I will then make a comparison of to understand the evolution of mating systems within the underlying mechanisms and propose a typology of cases species [71–73]. Interestingly, the two approaches are usually (orthogonal to the one- versus two-allele classification) that considered separately and emphasize completely opposite may prove useful understanding and modeling speciation outcomes. The first predicts the evolution of more assortative (Table 1). mating with increased outbreeding depression or hybrid Before proceeding, we note that several important find- unfitness. The second predicts the evolution of less assorta- ings have also been made regarding this process since these tive mating or selfing with increased inbreeding depression. early models. First, the role of ecological-based adaptation The models studying reinforcement include outbreeding but in speciation has received considerable support in the last not inbreeding depression [18, 70, 74–77] while the models decade [2, 4, 9, 10]. Second several empirical findings have studying mating system evolution do exactly the opposite supported the idea that reinforcement could indeed occur [71–73]. in the context of adaptation to different habitats [49, 58– Local adaptation causes outbreeding depression if differ- 67] or at least that there is often ample opportunity for ent alleles are favored in different habitats [78], which is the reinforcement [6]. reason why it is widely thought that spatially heterogenous International Journal of Ecology 5

A a

SA sA Sa (1) before sa migration SA sA Sa sa

sA (2) after Sa sa SA migration SA sA Sa sa

sa sa Sa sa Sa sa Sa Sa Sa sA sA Sa sA sA (3) syngamy Sa Sa SA sa sa SA sa sa SA SA sA SA SA sA SA sA SA sA

sa sa Sa sa Sa sa Sa Sa Sa Sa (4) after sA sA sAA sA Sa Sa selection SA sa sa SA sa sa SA SA sA SA SA sA SA sA SA sA

Figure 1: Indirect selection on a selfing/assortative mating modifier with local adaptation [53]. Sketches how a selfing or assortative mating modifier evolves in presence of local adaptation (for the sake of illustration, the S allele causes maximal assortment and s specifies random mating). Here assortment can be produced by selfing or assortative mating based on the genotype at the local adaptation locus (A mates with A and a with a). Before migration (step 1), consider two habitats with haploid individuals. On the left A allele is favored at a local adaptation locus, whereas a is favored on the right. To make things simple, we consider these alleles to be fixed where they are beneficial. At step 2, migration occurs between habitats (with m = 1/2, migrants in red). Then syngamy occurs in each habitat. The small circles represent diploid individuals. In each habitat, one can distinguish the subpopulation with full S-assortment allele (4 individuals on the left) and random mating s allele (4 individuals on the right). Importantly, at this step the S allele becomes positively associated to extreme aa and AA homozygotes (thus, variance in fitness is greater in the subpopulation with the S allele). Finally, selection occurs (favoring A on the left and a on the right, very strongly but in a codominant way in the illustration). At the end of this generation, the modifier has not changed in frequency (it is still 1/2). Yet, selection has generated LD between the random mating s allele and the locally inferior allele locally (the inferior allele is only found on the same chromosome as s in each habitat, orange dot). At the next generation, this LD will persist (it is decreased at most by one half by a round of free recombination) and cause indirect selection in favor of S. Note that if the locally beneficial allele is recessive (all heterozygotes eliminated on the right and the left), we see that direct selection occurs favoring S (its frequency rises to 2/3 on the sketch), but that less LD is generated. Exactly the opposite occurs if the local beneficial allele is dominant. selection favors the evolution of assortative mating by a one- of reinforcement have rarely considered the case of local allele mechanism in the absence of direct costs [8, 18, 21, 68, adaptation, and when they did, considered only haploid 69, 79, 80]. However, the dominance relationship of locally or diploid with particular dominance [18, 70, 82], these adapted alleles within each habitat may also cause inbreeding conclusions have remained largely overlooked. There are depression which can, in fact, prevent the evolution of numerous mechanisms of assortment [69]andeachcan premating isolation. When both phenomena act in concert, evolve slightly differently. For instance when local adaptation the outcomes vary tremendously depending on parameter is based on a conspicuous trait (shell thickness in Littorina values [53], and more assortment is not necessarily favored [65], coloration in Chrysopa [62]orHeliconius [83], etc.), even in presence of strong local adaptation. In fact, because mate choice can be cued directly on this trait, which is very a polymorphism at a locus involved in local adaptation is efficient unless the right mate is rare and difficult to find in more easily maintained when locally beneficial alleles are the population [84]. Another simple way to mate with a self- dominant, less assortment may often be favored in natural similar phenotype is to self-fertilize when hermaphrodite situations with local adaptation [53]. There are theoretical which is also very efficient, does not incur the cost of finding reasons for expecting such dominance relationships between the right mate and does not require the ability to discriminate locally adapted alleles [81]. Because theoretical models the locally adapted trait. In both cases, local adaptation 6 International Journal of Ecology

a A MA (1) before mA Ma ma migration mA MA Ma ma

Ma MA (2) after mA ma migration mA ma Ma MA

Ma mA MA ma (3) after ma MA selection mA Ma

Figure 2: Indirect selection on a migration modifier with local adaptation. Sketches how a migration modifier evolves in presence of local adaptation (for the sake of illustration the M allele causes maximal migration (1/2) and the m allele zero migration). Before migration (step 1) consider two habitats with haploid individuals. Before migration (step 1) consider two habitats with haploid individuals. On the left the A allele is favored at a local adaptation locus whereas a is favored on the right. To make things simple we consider these alleles to be fixed where they are beneficial. During migration only individuals with allele M move between habitats (step (2) migrants in red). Half of the M individuals move to the other habitat and the other half stays at home. Importantly migration directly generates LD between the M allele and the locally inferior allele (the locally inferior allele is found only with M and not with m). Finally, selection occurs favoring A on the left and a on the right very strongly in the illustration and carries the m allele with the adaptation locus because of the linkage disequilibrium generated at the previous step (the m overall frequency has raised from 1/2 to 2/3 on the illustration). Note that in a finite population kin selection by contrast favors M [54].

loci will cause indirect selection (Figure 1) and, if not co- the evolution of assortment, inbreeding depression causes a dominant, inbreeding depression. With selfing however, all selection pressure in favor of dispersal [89]. This inbreeding other loci in the genome experiencing recessive deleterious depression can be partly, but not only, caused by the loci mutations will also contribute to inbreeding depression. As responsible for the local adaptation. The second factor is kin a consequence, and unless the cost of finding a mate is high, selection (Figure 3). As soon as one considers a stochastic reinforcement may be less likely to evolve via selfing than via model for the evolution of dispersal, kin selection occurs assortment based on the local adaptation traits [53]. Two- and must be taken into account to determine how dispersal allele models and models involving sex-specific traits and evolves [90–92]. Intuitively, it is straightforward to see that a preferences also provide several alternatives [8, 52]. given allele causing zero dispersal cannot fix in a subdivided population. In other words, zero dispersal cannot be a convergent stable state as was suggested in deterministic 3.2. The Evolution of Dispersal. Individuals that have sur- models of reinforcement. As expected from this heuristic vived until reproduction have genotypes that work relatively argument, kin selection favors more dispersal than predicted well where they are. Because of environmental hetero- in a deterministic model [54]. However, this is not the only geneity, migrating or sending offspring elsewhere is likely effect as local adaptation interacts with the effect of kin to decrease fitness. Local adaptation indeed generates an selection: strong differentiation at a local adaptation locus indirect selection pressure in favor of less dispersal [50, 85– magnifies kin selection at short recombination distance. This 87](Figure 2). However, as for the evolution of nonrandom indirect kin selection can cause bistability (i.e., different mating, the evolution of dispersal has been studied in a dispersal rate can evolve depending on the initial conditions), variety of contexts and not only in reference to the process which changes qualitatively the expectation [54]. There are of the reinforcement of local adaptation or speciation and different ways to reduce dispersal, and all may not be a large number of factors interact to shape this trait [88]. equivalent even if the selection pressures at work will share However, in the context of the evolution of dispersal in strong similarities. In particular it is clearly important to presence of local adaptation there are at least two factors distinguish between dispersal and habitat choice. As we that cannot be ignored. The first is that, as in the case of have seen dispersal cannot evolve to very low rates because International Journal of Ecology 7

M m (1) before m m migration m m M m

m m (2) after m migration M M m m m

m mm M m (3) after m M mm reproduction M m M m m m m

4 alleles 4 alleles (4) after sampled sampled population at random at random regulation

Figure 3: Kin selection on a migration modifier. Sketches how a migration modifier evolves because of kin selection. As in Figure 2, the M allele causes maximal migration (1/2) and the m allele specifies zero migration). Before migration (step 1), consider two subpopulations. In one of them the M allele is frequent (1/2), but it is absent in the other. During migration, only M individuals move between habitats (step (2), the migrant is shown in red). Half of the M individuals move to the other habitat and the other half stays at home. Then reproduction occurs (step 3). All individuals produce say, two offsprings (note that all individuals have the same survival and reproduction). Finally, population regulation occurs: juveniles compete to repopulate each subpopulation with four adults and all have the same chance to get established. After this step, the M frequency has risen to (1/3 + 1/5)/2, which is greater than 1/4, the initial frequency. There is thus selection on M allele, which is traditionally explained in terms of “kin selection”: the migrating M allele sacrifices itself by competing in a more crowded population, but it leaves room behind that benefits the other M allele, which will compete in a less crowded population. The decreased chance of survival by the migrating M (1/5–1/4) is more than compensated by the increased chance that the remaining M allele will survive (1/3–1/4). This process requires only that the M alleles are concentrated in the same population at step 1 (i.e., it requires population structure or relatedness), which is easily generated by drift [54].

of kin selection. However, choosing the natal habitat (to an immediate effect on the genetic composition of the popu- maintain local adaptation) while quitting its natal patch (to lation, here genotypic frequencies at the local adaptation loci: release kins from competition) may provide the best from dispersal modification changes allelic frequencies; assortment both worlds and is therefore a more likely candidate trait changes within locus associations; recombination changes for reinforcement. Two-allele models also provide several between loci associations. This immediate effect causes a alternatives [38, 57]. frequency change at the modifier locus if there is selection on alleles, dominance, and epistasis, respectively. When the 3.3. Comparisons among Reinforcement Traits. The common modifier changes within or between loci associations, a effect in all these processes is that alleles that favor more secondary effect occurs. Increased associations generate a assortment, less dispersal or tighter linkage become associ- higher variance in fitness, more efficient selection, and thus ated with locally beneficial alleles, which in turn generates an increase of the frequency difference between habitats an indirect selection pressure in their favor. In each of at the local adaptation locus. As a consequence, modifiers these cases however, the way linkage disequilibrium is built increasing these associations (modifier increasing assortment between the modifiers and the locally beneficial alleles is or reducing recombination in our examples) become asso- distinct (compare Figures 1, 2 and 4). First, a modifier has ciated, and hitchhike, with locally beneficial alleles. There 8 International Journal of Ecology

A, B a, b RAB Rab (1) before rAB rab RAB migration Rab rab rAB

RAB Rab (2) after rAB rab migration Rab RAB rab rAB

rAB RAB RAB rAB rAB (3) syngamy RAb RAb rAB rab and meiosis RaB RaB rab rab Rab Rab rab

rAB RAB RAB rAB rAB (4) after RAb RAb rAB rab selection RaB RaB rab Rab rab Rab rab

Figure 4: Indirect selection on a recombination modifier with local adaptation. Sketches how a recombination modifier evolves in presence of local adaptation (for the sake of illustration, the R allele causes maximal recombination, and the r allele specifies zero recombination). Before migration (step 1), consider two habitats with haploid individuals. On the left, the A and B alleles are favored at two different loci; the a and b alleles are favored on the right. To simplify the illustration, we consider the alleles to be fixed where they are beneficial. After migration between habitats (step (2) with m = 1/2, migrants shown in red), strong LD is generated between the selected loci: in each habitat there are AB and ab but no Ab and aB haplotypes. Then random mating and meiosis occur (step 3). In each habitat, one can distinguish the subpopulation with the full R recombination allele and the zero r recombination allele (groups of four individuals on the left and right, resp.). Within the former, LD between selected loci has been much reduced (illustrated at zero, in fact even full recombination only halves LD at each meiosis), whereas in the latter it stayed intact. Importantly, at this step the r recombination modifier becomes positively associated to the extreme AB and ab haplotypes. Note also that in each habitat, variance in fitness is greater in the subpopulation with the r allele. Finally, selection occurs (favoring A and B on the left and a and b on the right, very strongly on the illustration) and takes the r allele along because of the linkage disequilibrium generated at the previous step (the overall frequency of r has risen from 1/2 to 2/3 in the illustration). Note that the case illustrated involves positive epistasis (only the extreme genotypes AB and ab survive, on the left and right, resp.). However, the r allele is favoured even if epistasis is zero, because selection is more efficient in subpopulations carrying the r allele since variance in fitness is greater in these subpopulations. are thus several ways to promote distinct genetic clusters the different ways reinforcement and genetic clustering can between habitats and the three examples detailed in this occur. It is orthogonal to, and complements the usual one- paper illustrate each of these cases: directly magnifying versus two-allele classification (Table 1). allelic frequency differences between populations (case illustrated by dispersal modification), promoting within locus associations (case illustrated by assortment modification), 4. Conclusion promoting between loci associations (case illustrated by recombination modification). Reinforcement may occur by The first conclusion is that the process of reinforcement and the evolution of many other traits than the ones mentioned local adaptation are intertwined and occur simultaneously. here (in particular involving two-allele mechanisms, see Whether pre- and postzygotic isolation will eventually evolve Table 1), but their impact are likely to be achieved via one of is uncertain in such a dynamic process. In particular, these effects alone or in combination. Considering the three local adaptation can collapse if generalist alleles arise and possible impacts of a modifier on the genetic composition spread. However, there are several positive feedback loops of populations (on frequencies, within locus and between that will tend to drive the system towards divergence (the loci associations) may be a useful typology to understand reinforcement, demographic, and recombination loops). International Journal of Ecology 9

From a theoretical standpoint, this process has rarely been [9] S. Via, “Sympatric speciation in animals: the ugly duckling analyzed jointly and in a dynamic way with changes in local grows up,” Trends in Ecology and Evolution,vol.16,no.7,pp. adaptation itself. In the context of mounting evidence in 381–390, 2001. favor of ecological speciation [106, 107], such an approach [10] D. Schluter, “Ecology and the origin of species,” Trends in would certainly help evaluate its likelihood, tempo, and Ecology and Evolution, vol. 16, no. 7, pp. 372–380, 2001. mechanism. [11] M. Kirkpatrick and N. Barton, “Chromosome inversions, Second, Felsenstein [21] proposed to distinguish the local adaptation and speciation,” Genetics, vol. 173, no. 1, pp. 419–434, 2006. different mechanisms for reinforcement on the idea that they [12] M. Arnold, Natural Hybridization and Evolution,Oxford involved the spread of one or two-alleles in the incipient University Press, Oxford, UK, 1997. species. This distinction is an important one, but it is not [13]L.E.B.Kruuk,S.J.E.Baird,K.S.Gale,andN.H. the only one to be made. Many one-allele mechanisms Barton, “A comparison of multilocus clines maintained by are only superficially similar as they can promote genetic environmental adaptation or by selection against hybrids,” clustering and speciation in different ways. A useful typology Genetics, vol. 153, no. 4, pp. 1959–1971, 1999. could be that they increase differentiation among popula- [14] T. Lenormand, “Gene flow and the limits to natural selections, heterozygote deficit, or linkage disequilibrium, which tion,” Trends in Ecology and Evolution, vol. 17, no. 4, pp. 183– corresponds to modifying one of the three fundamental 189, 2002. events in a sexual life cycle (dispersal, syngamy, or meiosis, [15] T. Nagylaki, “Conditions for the existence of clines,” Genetics, resp.). Furthermore, different traits may increase genetic vol. 80, no. 3, pp. 595–615, 1975. clustering, but may not contribute to reinforcement because [16] M. Slatkin, “Gene flow and selection in a cline,” Genetics, vol. they are exposed to a variety of other selective effects. 75, no. 4, pp. 733–756, 1973. Models of reinforcement based on the evolution of particular [17] N. H. Barton and G. M. Hewitt, “Analysis of hybrid zones,” traits must integrate what is known outside the speciation Annual Review of Ecology and Systematics, vol. 16, pp. 113– literature for those traits. For instance, recombination [108], 148, 1985. [18] M. R. Servedio, “The evolution of premating isolation: local mating systems [109], and dispersal [110], as discussed adaptation and natural and sexual selection against hybrids,” above, have all been intensely studied outside this context ff Evolution, vol. 58, no. 5, pp. 913–924, 2004. pinpointing a variety of selective e ects. These theoretical [19] J. Mallet, A. Meyer, P. Nosil, and J. L. Feder, “Space, sympatry developments certainly have to be merged. and speciation,” Journal of Evolutionary Biology, vol. 22, no. 11, pp. 2332–2341, 2009. [20] F. Debarre´ and T. Lenormand, “Distance-limited dispersal Acknowledgments promotes coexistence at habitat boundaries: reconsidering I thank A. Whibley, J. Mallet and two anonymous reviewers the competitive exclusion principle,” Ecology Letters, vol. 14, no. 3, pp. 260–266, 2011. for comments on this manuscript. This work was supported [21] J. Felsenstein, “Skepticism towards Santa Rosalia, or why are by the European Research Council starting grant “Quan- there so few kinds of animals?” Evolution, vol. 35, no. 1, pp. tEvol”. 124–138, 1981. [22] T. Nagylaki, “Clines with asymmetric migration,” Genetics, vol. 88, no. 4, pp. 813–827, 1978. References [23] P. Labbe,´ N. Sidos, M. Raymond, and T. Lenormand, [1] C. I. Wu, “The genic view of the process of speciation,” “Resistance gene replacement in the mosquito Culex pipiens: Journal of Evolutionary Biology, vol. 14, no. 6, pp. 851–865, fitness estimation from long-term cline series,” Genetics, vol. 2001. 182, no. 1, pp. 303–312, 2009. [2] M. Schilthuizen, “Dualism and conflicts in understanding [24] T. J. Kawecki, N. H. Barton, and J. D. Fry, “Mutational speciation,” BioEssays, vol. 22, no. 12, pp. 1134–1141, 2000. collapse of fitness in marginal habitats and the evolution of [3] S. Gavrilets, “Perspective: models of speciation: what have ecological specialisation,” Journal of Evolutionary Biology, vol. we learned in 40 years?” Evolution, vol. 57, no. 10, pp. 2197– 10, no. 3, pp. 407–429, 1997. 2215, 2003. [25] M. C. Whitlock, “The red queen beats the jack-of-all-trades: [4] C. D. Jiggins and J. Mallet, “Bimodal hybrid zones and the limitations on the evolution of phenotypic plasticity speciation,” Trends in Ecology and Evolution, vol. 15, no. 6, and niche breadth,” The American Naturalist, vol. 148, pp. 250–255, 2000. supplement, pp. S65–S77, 1996. [5] J. Mallet, “A species definition for the modern synthesis,” [26] M. C. Hall, D. B. Lowry, and J. H. Willis, “Is local adaptation Trends in Ecology and Evolution, vol. 10, no. 7, pp. 294–299, in Mimulus guttatus caused by trade-offs at individual loci?” 1995. Molecular Ecology, vol. 19, no. 13, pp. 2739–2753, 2010. [6] J. Mallet, “Hybridization as an invasion of the genome,” [27] F. M. Cohan, E. C. King, and P. Zawadzki, “Amelioration of Trends in Ecology and Evolution, vol. 20, no. 5, pp. 229–237, the deleterious pleiotropic effects of an adaptive mutation in 2005. Bacillus subtilis,” Evolution, vol. 48, no. 1, pp. 81–95, 1994. [7] N. H. Barton and B. Charlesworth, “Genetic revolutions, [28] T. Guillemaud, T. Lenormand, D. Bourguet, C. Chevillon, founder effects, and speciation,” Annual Review of Ecology N. Pasteur, and M. Raymond, “Evolution of resistance in and Systematics, vol. 15, pp. 133–164, 1984. Culex pipiens: allele replacement and changing environment,” [8] M. Kirkpatrick and V. Ravigne,´ “Speciation by natural and Evolution, vol. 52, no. 2, pp. 443–453, 1998. sexual selection: models and experiments,” The American [29] J. B. S. Haldane, The Causes of Evolution,Harper,NewYork, Naturalist, vol. 159, supplement 3, pp. S22–S35, 2002. NY, USA, 1932. 10 International Journal of Ecology

[30] P. Labbe, T. Lenormand, and M. Raymond, “On the world- [49] J. Antonovics, “Evolution in closely adjacent plant popula- wide spread of an insecticide resistance gene: a role for local tions—V. Evolution of self-fertility,” Heredity,vol.23,no.2, selection,” Journal of Evolutionary Biology,vol.18,no.6,pp. pp. 219–238, 1968. 1471–1484, 2005. [50] B. J. Balkau and M. W. Feldman, “Selection for migration [31] R. A. Fisher, “The possible modification of the response of modification,” Genetics, vol. 74, no. 1, pp. 171–174, 1973. the wild type to recurrent mutations,” The American Natu- [51] D. Charlesworth and B. Charlesworth, “Selection of recom- ralist, vol. 62, no. 679, pp. 115–126, 1928. bination in clines,” Genetics, vol. 91, no. 3, pp. 581–589, 1979. [32] R. E. Lenski, “Experimental studies of pleiotropy and epis- [52] M. R. Servedio, “The role of linkage disequilibrium in the tasis in Escherichia coli—II. Compensation for maladaptive evolution of premating isolation,” Heredity, vol. 102, no. 1, effects associated with resistance to virus T4,” Evolution, vol. pp. 51–56, 2009. 42, no. 3, pp. 433–440, 1988. [53] G. Epinat and T. Lenormand, “The evolution of assortative [33] J. A. McKenzie, Ecological and Evolutionary Aspects of Insecti- mating and selfing with in- and outbreeding depression,” cide Resistance, R. G. Landes, Austin, Tex, USA, 1996. Evolution, vol. 63, no. 8, pp. 2047–2060, 2009. [34] P. Labbe,´ A. Berthomieu, C. Berticat et al., “Independent [54] S. Billiard and T. Lenormand, “Evolution of migration under duplications of the acetylcholinesterase gene conferring in- kin selection and local adaptation,” Evolution, vol. 59, no. 1, secticide resistance in the mosquito Culex pipiens,” Molecular pp. 13–23, 2005. Biology and Evolution, vol. 24, no. 4, pp. 1056–1067, 2007. [55] P. R. Armsworth, “Conditional dispersal, clines, and the evo- [35] P. Labbe,´ C. Berticat, A. Berthomieu et al., “Forty years of lution of dispersiveness,” Theoretical Ecology, vol. 2, no. 2, pp. erratic insecticide resistance evolution in the mosquito Culex 105–117, 2009. pipiens,” PLoS Genetics, vol. 3, no. 11, Article ID e205, pp. [56] K. V. Pylkov, L. A. Zhivotovsky, and M. W. Feldman, “Migra- 2190–2199, 2007. tion versus mutation in the evolution of recombination [36] T. Lenormand, T. Guillemaud, D. Bourguet, and M. Ray- under multilocus selection,” Genetical Research, vol. 71, no. mond, “Appearance and sweep of a gene duplication: adap- 3, pp. 247–256, 1998. tive response and potential for new functions in the mosquito [57] J. D. Fry, “Multilocus models of sympatric speciation: bush Culex pipiens,” Evolution, vol. 52, no. 6, pp. 1705–1712, 1998. versus rice versus felsenstein,” Evolution, vol. 57, no. 8, pp. [37] M. Egas, U. Dieckmann, and M. W. Sabelis, “Evolution 1735–1746, 2003. restricts the coexistence of specialists and generalists: the role [58] R. Hopkins and M. D. Rausher, “Identification of two of trade-off structure,” The American Naturalist, vol. 163, no. genes causing reinforcement in the Texas wildflower Phlox 4, pp. 518–531, 2004. drummondii,” Nature, vol. 469, pp. 411–414, 2011. [38] V. Ravigne,´ U. Dieckmann, and I. Olivieri, “Live where you [59] J. T. Anderson, J. H. Willis, and T. Mitchell-Olds, “Evolution- thrive: joint evolution of habitat choice and local adapta- ary genetics of plant adaptation,” Trends in Genetics, vol. 27, tion facilitates specialization and promotes diversity,” The no. 7, pp. 258–266, 2011. American Naturalist, vol. 174, no. 4, pp. E141–E169, 2009. [60] J. Hollander, M. Lindegarth, and K. Johannesson, “Local [39] O. Ronce and M. Kirkpatrick, “When sources become sinks: adaptation but not geographical separation promotes assor- migrational meltdown in heterogeneous habitats,” Evolution, tative mating in a snail,” Animal Behaviour,vol.70,no.5,pp. vol. 55, no. 8, pp. 1520–1531, 2001. 1209–1219, 2005. [40] R. Gomulkiewicz, R. D. Holt, and M. Barfield, “The effects [61] T. McNeilly and J. Antonovics, “Evolution in closely adjacent of density dependence and immigration on local adaptation plant populations—IV. Barriers to gene flow,” Heredity, vol. and niche evolution in a black-hole sink environment,” 23, no. 2, pp. 205–218, 1968. Theoretical Population Biology, vol. 55, no. 3, pp. 283–296, [62] C. A. Tauber and M. J. Tauber, “Sympatric speciation based 1999. on allelic changes at three loci: evidence from natural [41] N. Barton, “Adaptation at the edge of a species’ range,” populations in two habitats,” Science, vol. 197, no. 4310, pp. in Integrating Genetics and Ecology in a Spatial Context,J. 1298–1299, 1977. Antonovics and J. Silvertown, Eds., pp. 365–392, Blackwell, [63] M. Macnair and P. Christie, “Reproductive isolation as a London, UK, 2001. pleiotropic effect of copper tolerance in Mimulus guttatus?” [42] M. Kirkpatrick and N. H. Barton, “Evolution of a species’ Heredity, vol. 50, no. 3, pp. 295–302, 1983. range,” The American Naturalist, vol. 150, no. 1, pp. 1–23, [64] S. Dubois, P. O. Cheptou, C. Petit et al., “Genetic structure 1997. and mating systems of metallicolous and nonmetallicolous [43] M. Slatkin, “Gene flow and selection in a two locus system,” populations of Thlaspi caerulescens,” New Phytologist, vol. Genetics, vol. 81, no. 4, pp. 787–802, 1975. 157, no. 3, pp. 633–641, 2003. [44] T. Lenormand, T. Guillemaud, D. Bourguet, and M. Ray- [65] K. Johannesson, “Evolution in Littorina: ecology matters,” mond, “Evaluating gene flow using selected markers: a case Journal of Sea Research, vol. 49, no. 2, pp. 107–117, 2003. study,” Genetics, vol. 149, no. 3, pp. 1383–1392, 1998. [66] J. W. Grahame, C. S. Wilding, and R. K. Butlin, “Adaptation [45] T. Lenormand and S. P. Otto, “The evolution of recombina- to a steep environmental gradient and an associated barrier tion in a heterogeneous environment,” Genetics, vol. 156, no. to gene exchange in Littorina saxatilis,” Evolution, vol. 60, no. 1, pp. 423–438, 2000. 2, pp. 268–278, 2006. [46] S. Yeaman and M. C. Whitlock, “The genetic architecture [67] J. Antonovics, “Evolution in closely adjacent plant of adaptation under migration-selection balance,” Evolution, populations—X: long-term persistence of prereproductive vol. 65, no. 7, pp. 1897–1911, 2011. isolation at a mine boundary,” Heredity,vol.97,no.1,pp. [47] C. I. Wu and C. T. Ting, “Genes and speciation,” Nature 33–37, 2006. Reviews Genetics, vol. 5, no. 2, pp. 114–122, 2004. [68] J. Coyne, H. Orr et al., Speciation, Sinauer Associates, [48] U. Dieckmann and M. Doebeli, “On the origin of species by Sunderland, Mass, USA, 2004. sympatric speciation,” Nature, vol. 400, no. 6742, pp. 354– [69] S. Gavrilets, Fitness Landscape and the Origin of Species, 357, 1999. Princeton University Press, Princeton, NJ, USA, 2004. International Journal of Ecology 11

[70] M. R. Servedio, “Reinforcement and the genetics of nonran- [90] W. D. Hamiltion and R. M. May, “Dispersal in stable habi- dom mating,” Evolution, vol. 54, no. 1, pp. 21–29, 2000. tats,” Nature, vol. 269, no. 5629, pp. 578–581, 1977. [71] R. Lande and D. W. Schemske, “The evolution of self- [91] S. A. Frank, “Dispersal polymorphisms in subdivided pop- fertilization and inbreeding depression in plants—I. Genetic ulations,” Journal of Theoretical Biology, vol. 122, no. 3, pp. models,” Evolution, vol. 39, no. 1, pp. 24–40, 1985. 303–309, 1986. [72]D.Charlesworth,M.T.Morgan,andB.Charlesworth,“In- [92] P. D. Taylor, “An inclusive fitness model for dispersal of breeding depression, genetic load, and the evolution of offspring,” Journal of Theoretical Biology, vol. 130, no. 3, pp. outcrossing rates in a multilocus system with no linkage,” 363–378, 1988. Evolution, vol. 44, no. 6, pp. 1469–1489, 1990. [93] N. Barton and B. O. Bengtsson, “The barrier to genetic ex- [73] M. K. Uyenoyama and D. M. Waller, “Coevolution of change between hybridising populations,” Heredity, vol. 57, self-fertilization and inbreeding depression—I. Mutation- no. 3, pp. 357–376, 1986. selection balance at one and two loci,” Theoretical Population [94] N. H. Barton and B. Charlesworth, “Why sex and recombi- Biology, vol. 40, no. 1, pp. 14–46, 1991. nation?” Science, vol. 281, no. 5385, pp. 1986–1990, 1998. [74] M. Kirkpatrick, “Reinforcement and divergence under assor- [95] S. P. Otto and T. Lenormand, “Resolving the paradox of sex tative mating,” Proceedings of the Royal Society B, vol. 267, no. and recombination,” Nature Reviews Genetics, vol. 3, no. 4, 1453, pp. 1649–1655, 2000. pp. 252–261, 2002. [75] M. Kirkpatrick and M. R. Servedio, “The reinforcement of [96] A. I. Khan, D. M. Dinh, D. Schneider, R. E. Lenski, and T. F. mating preferences on an island,” Genetics, vol. 151, no. 2, Cooper, “Negative epistasis between beneficial mutations in pp. 865–884, 1999. an evolving bacterial population,” Science, vol. 332, no. 6034, [76] S. P. Otto, M. R. Servedio, and S. L. Nuismer, “Frequency- pp. 1193–1196, 2011. dependent selection and the evolution of assortative mating,” [97] H. H. Chou, H. C. Chiu, N. F. Delaney, D. Segre,` and C. Genetics, vol. 179, no. 4, pp. 2091–2112, 2008. J. Marx, “Diminishing returns epistasis among beneficial [77] M. R. Servedio and M. Kirkpatrick, “The effects of gene flow mutations decelerates adaptation,” Science, vol. 332, no. 6034, on reinforcement,” Evolution, vol. 51, no. 6, pp. 1764–1772, pp. 1190–1192, 2011. 1997. [98]D.R.Rokyta,P.Joyce,S.B.Caudle,C.Miller,C.J.Beisel, and H. A. Wichman, “Epistasis between beneficial mutations [78] H. D. Rundle and M. C. Whitlock, “A genetic interpretation and the phenotype-to-fitness map for a ssDNA virus,” PLoS of ecologically dependent isolation,” Evolution, vol. 55, no. 1, Genetics, vol. 7, no. 6, Article ID e1002075, 2011. pp. 198–201, 2001. [99] G. Martin, S. F. Elena, and T. Lenormand, “Distributions of [79] J. A. Endler, Geographic Variation, Speciation and Clines, epistasis in microbes fit predictions from a fitness landscape Princeton University Press, Princeton, NJ, USA, 1977. model,” Nature Genetics, vol. 39, no. 4, pp. 555–560, 2007. [80] J. Maynard Smith, “Sympatric speciation,” The American [100] N. H. Barton and S. P.Otto, “Evolution of recombination due Naturalist, vol. 100, no. 916, pp. 637–650, 1966. to random drift,” Genetics, vol. 169, no. 4, pp. 2353–2370, [81] S. P. Otto and D. Bourguet, “Balanced polymorphisms and 2005. the evolution of dominance,” The American Naturalist, vol. [101] G. Martin, S. P. Otto, and T. Lenormand, “Selection for 153, no. 6, pp. 561–574, 1999. recombination in structured populations,” Genetics, vol. 172, [82] S. Gavrilets, “The Maynard Smith model of sympatric no. 1, pp. 593–609, 2006. speciation,” Journal of Theoretical Biology, vol. 239, no. 2, pp. [102] D. Roze and N. H. Barton, “The Hill-Robertson effect and 172–182, 2006. the evolution of recombination,” Genetics, vol. 173, no. 3, pp. [83]C.D.Jiggins,R.E.Naisbit,R.L.Coe,andJ.Mallet, 1793–1811, 2006. “Reproductive isolation caused by colour pattern mimicry,” [103] M. A. R. de Cara, N. H. Barton, and M. Kirkpatrick, “A Nature, vol. 411, no. 6835, pp. 302–305, 2001. model for the evolution of assortative mating,” The American [84] M. Kirkpatrick and S. L. Nuismer, “Sexual selection can con- Naturalist, vol. 171, no. 5, pp. 580–596, 2008. strain sympatric speciation,” Proceedings of the Royal Society [104] N. H. Barton and M. Turelli, “Natural and sexual selection B, vol. 271, no. 1540, pp. 687–693, 2004. on many loci,” Genetics, vol. 127, no. 1, pp. 229–255, 1991. [85] S. Karlin and R. B. Campbell, “The existence of a protected [105] M. Kirkpatrick, T. Johnson, and N. Barton, “General models polymorphism under conditions of soft as opposed to hard of multilocus evolution,” Genetics, vol. 161, no. 4, pp. 1727– selection in a multideme population system,” The American 1750, 2002. Naturalist, vol. 117, no. 3, pp. 262–275, 1981. [106] D. Schluter, “Evidence for ecological speciation and its [86] P. Wiener and M. W. Feldman, “The effects of the mating alternative,” Science, vol. 323, no. 5915, pp. 737–741, 2009. system on the evolution of migration in a spatially hetero- [107] J. Mallet, “Hybridization, ecological races and the nature geneous population,” Evolutionary Ecology, vol. 7, no. 3, pp. of species: empirical evidence for the ease of speciation,” 251–269, 1993. Philosophical Transactions of the Royal Society B, vol. 363, no. [87] J. H. Gillespie, “The role of migration in the genetic struc- 1506, pp. 2971–2986, 2008. ture of populations in temporally and spatially varying [108] S. P. Otto and N. H. Barton, “The evolution of recombina- environments—III. Migration modification,” The American tion: removing the limits to natural selection,” Genetics, vol. Naturalist, vol. 117, no. 3, pp. 223–233, 1981. 147, no. 2, pp. 879–906, 1997. [88] O. Ronce, “How does it feel to be like a rolling stone? [109] K. Holsinger, “Pollination biology and the evolution of mat- Ten questions about dispersal evolution,” Annual Review of ing systems in flowering plants,” Evolutionary Biology, vol. 29, Ecology, Evolution, and Systematics, vol. 38, pp. 231–253, no. 5, pp. 107–149, 1996. 2007. [110] O. Ronce, I. Olivieri, J. Clobert, and E. Danchin, “Perspective [89] D. Roze and F. Rousset, “Inbreeding depression and the evo- on the study of dispersal evolution,” in Dispersal,J.Clobert, lution of dispersal rates: a multilocus model,” The American E. Danchin, A. Dhondt, and J. Nichols, Eds., pp. 341–357, Naturalist, vol. 166, no. 6, pp. 708–721, 2005. Oxford University Press, Oxford, UK, 2001. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 809897, 8 pages doi:10.1155/2012/809897

Research Article Testing the Role of Habitat Isolation among Ecologically Divergent Gall Wasp Populations

Scott P. Egan,1, 2 Glen R. Hood,1 and James R. Ott3

1 Department of Biological Sciences, University of Notre Dame, Notre Dame, IN 46556, USA 2 Advanced Diagnostics and Therapeutics, University of Notre Dame, Notre Dame, IN 46556, USA 3 Population and Conservation Biology Program, Department of Biology, Texas State University—San Marcos, San Marcos, TX 78666, USA

Correspondence should be addressed to Scott P. Egan, [email protected]

Received 17 October 2011; Accepted 9 January 2012

Academic Editor: Marianne Elias

Copyright © 2012 Scott P. Egan et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Habitat isolation occurs when habitat preferences lower the probability of mating between individuals associated with differing habitats. While a potential barrier to gene flow during ecological speciation, the effect of habitat isolation on reproductive isolation has rarely been directly tested. Herein, we first estimated habitat preference for each of six populations of the gall wasp Belonocnema treatae inhabiting either Quercus virginiana or Q. geminata. We then estimated the importance of habitat isolation in generating reproductive isolation between B. treatae populations that were host specific to either Q. virginiana or Q. geminata by measuring mate preference in the presence and absence of the respective host plants. All populations exhibited host preference for their native plant, and assortative mating increased significantly in the presence of the respective host plants. This host-plant-mediated assortative mating demonstrates that habitat isolation likely plays an important role in promoting reproductive isolation among populations of this host-specific gall former.

1. Introduction Rhagoletis fruit flies [8], Littorina snails [9], Neochlamisus leaf beetles [10], Gasterosteus aculeatus sticklebacks [11], Ecological speciation describes the process by which repro- Gambusia fishes [12], Timema walking sticks [13], Mimulus ductive isolation evolves as a consequence of divergent monkeyflowers [14], and cynipid gall wasps [15]). Moreover, natural selection between environments [1, 2]. Studies of these studies have documented that a diversity of both ecological speciation seek to associate the origin of specific prezygotic and postzygotic reproductive barriers can arise as reproductive isolating barriers that reduce gene flow with a result of divergent ecological adaptation [3, 16], including sources of divergent selection [3]. Throughout the modern temporal isolation [17], sexual isolation [18], cryptic iso- synthesis, biologists described a central role of ecological lation [19], and extrinsic (ecological) postzygotic isolation adaptation in the speciation process [4–6]; however, it was [20, 21]. not until a recent renaissance of empirical study that spe- The study of ecological speciation has been especially cific ecological barriers have been experimentally shown informed by studies of herbivorous insects [22, 23]. The inti- to contribute to reproductive isolation. In strong support mate interactions between herbivorous insects and their host of the role of ecology in speciation, a comparative study plants suggest a strong role for divergent natural selection across many plant and animal taxa provided evidence that in promoting diversification. Herbivorous insects tend to be ecological adaptation generally contributes to the evolution highly specialized in their use of host plant taxa [24], and of reproductive isolation [7]. specialized insect herbivores can exhibit pronounced geo- Recent studies of the role of ecology in speciation have graphic variation in, and rapid evolution of, host plant pref- documented a central role of divergent natural selection in erence and performance traits (e.g., [25, 26]). The increased the speciation process among a diverse set of taxa (e.g., rates of speciation associated with herbivory among insects 2 International Journal of Ecology

Table 1: Locations and host plant associations of the six B. treatae populations from central Florida used in the present study.

Host Population Latitude Longitude association Near Avon Park (AP) Q. geminata 27◦ 36 00 N81◦ 30 42 W Scrub field (S) Q. geminata 27◦ 30 48 N81◦ 20 16 W Archbold Biological Station (ABS)∗ Q. geminata 27◦ 10 57 N81◦ 21 08 W Near Hickory Hammock Natural Area (HH)∗ Q. virginiana 27◦ 24 09 N81◦ 06 42 W Gatorama (GR) Q. virginiana 26◦ 55 30 N81◦ 18 44 W Near Koreshan State Park (KSP) Q. virginiana 26◦ 26 04 N81◦ 48 56 W ∗ Denotes populations used in the mate preference tests. provides evidence that host plant ecology may generally significant increase in assortative mating due to host plant contribute to the speciation process [27, 28]. presence. Walsh [29] was one of the first to associate phenotypic In the present study, we use a combination of habitat variation among insects with the host plants upon which preference and mate preference assays among ecologically they were found and Bush [30] was one of the first to argue divergent populations of the gall wasp Belonocnema treatae for a direct role of host-plant-associated selection in the (Hymenoptera: Cynipidae) to test for (a) variation among genesis of new insect species. Continued work has since host-associated populations in habitat (i.e., host plant) highlighted the role of divergent selection due to host plant preference and (b) an explicit role for habitat isolation in useamongtaxawheregeneflowispossible(e.g.,[8, 10, overall reproductive isolation. We test these hypotheses using 13, 17, 31, 32]). A critical barrier to gene flow among spe- populations of B. treatae that inhabit two sister species cialist herbivore insect taxa is “habitat isolation” [8, 16, of live oak, Quercus virginiana and Q. geminata,which 30]. Habitat isolation for host-specific phytophagous insect geographically overlap in the southeastern United States. species describes the process by which the differing habitat The habitat of each oak differs slightly, with Q. virginiana preferences of insect populations associated with alternative occurring in moister, more nutrient rich, and higher pH host plants reduces the frequency of encounters and thus sites than Q. geminata [37], and the oaks themselves differ the likelihood of mating between individuals from the in leaf morphology and flowering times [38]. Populations of differing host-associated populations. For example, Nosil B. treatae that inhabit these oak species exhibit significant et al. [33] examined 27 populations of Timema cristinae differences in root gall structure and adult body size that are walking sticks feeding on Ceanothus or Adenostoma host associated with host use, and gall wasp populations exhibit plants. Populations of walking sticks on different host plants host-associated assortative mating [15]. expressed stronger divergence in host plant preference than populations on the same host plant. These differences likely 2. Methods result in reduced encounters among individuals preferring different hosts. Similar inferences regarding the role of host 2.1. Study System and Sampling. Belonocnema treatae is a plant preference in speciation have been made for leaf beetles host-specific gall former [39] that exhibits regional speci- [10], pea aphids [32], ladybird beetles [34], Rhagoletis fruit ficity (Ott and Egan, personal observation) on species of live flies [8, 35], and Eurosta galling flies [36]. oak, Quercus, within the Virentes series of the genus [40]. However, rarely has the effect of observed differences in Belonocnema treatae exhibits a heterogonous life cycle with host plant preference on reproductive isolation been tested temporally segregated sexual and asexual generations [39]. directly [23]. Field studies of the apple and hawthorn host The asexual generation develops within single-chambered, races of Rhagoletis pomonella found evidence that host plant spherical galls on the undersides of leaves during the summer preference could generate habitat isolation [8]. Here, the and fall and emerges in the winter. The sexual generation apple and hawthorn host races returned to their natal plant develops within multichambered galls on the root tissue, and species when released in the presence of both apple and males and females emerge during the spring. We collected hawthorn trees. Because these host races mate on their host root galls containing the sexual generation from six allopatric plant it is likely that host preference translates into host- populations in central Florida in April 2010 (three Q. associated assortative mating that restricts gene flow between geminata and three Q. virginiana populations; see Table 1 for the ecologically divergent populations. In a direct laboratory- location information). Galls were husbanded under common ◦ based test of habitat isolation, Funk [10] performed mating laboratory conditions (12 : 12 light : dark, 23 C), and upon assays among ecologically divergent host forms of the leaf emergence adults were sorted by sex and population for host beetle Neochlamisus bebbianae. To isolate the role of the preference and mating assays, which took place within 48 host plant on overall sexual isolation, the host plant of each hours of emergence. individual was included in half of the mating assays. Results from Funk [10] were mixed, with one of the four different 2.2. Host Preference Assays. Trials took place within 25 × host comparisons of N. bebbianae populations exhibiting a 8 cm clear-plastic cups stocked with a cutting of each host International Journal of Ecology 3

1 0 0.9 22 0.1 0.8 18 15 0.2 18 0.7 0.3

20 geminata Q. 0.6 19 0.4

Q. virginianaQ. 18 0.5 0.5 13 13 0.4 0.6 0.3 0.7 20 22 16 0.2 0.8

0.1 0.9 for preference Relative

Relative preference for preference Relative 0 1 GR HH KSP Scrub ABS AP Q. virginiana Q. geminata

Females Males

Figure 1: Host preference of individual B. treatae gall wasps expressed for each of three Q. virginiana and three Q. geminata host-associated populations during choice tests that paired each wasp’s native/natal host with the alternative oak species. Illustrated is the mean (±SE) of the proportion of time spent on the native host plant by each sex within each population. The dashed line highlights no preference (defined as 50% of time spent on each host). Note each population differed significantly from 50% (P<0.0001). Numbers above the SEs are the number of replicates. Note the reversed left and right y-axis.

plant (Q. virginiana and Q. geminata). A single B. treatae 1 was aspirated into each cup and then observed at ﬁve-minute N = 75 N = 72 intervals for one hour for a total of 12 observations. At each time point, we recorded the location (on Q. virginiana,onQ. 0.8 geminata, or on the cup) of each individual. Both sexes were tested. Host preference was calculated for each individual as N = 73 the relative time spent on one host plant species divided by the total time spent on both host plants during the trials 0.6 (e.g., individual preference for Q. virginiana = (number of observations on Q. virginiana)/(number of observations on N = 71 Q. virginiana +numberofobservationsonQ. geminata)). 0.4 We performed a total of 214 preference assays distributed Copulation frequency Copulation across the six B. treatae populations (see Figure 1 for sample sizes per population). 0.2

2.3. Assays of Sexual Isolation with an Explicit Test of Habitat Isolation. No-choice mating trials were conducted to test 0 for assortative mating as a function of population of Population: SameDifferent Same Different origin between one population (ABS—Archbold Biological Host plant: Absent Present Station, FL) of B. treatae reared from galls that developed on Q. geminata and one population (HH—near Hickory Hammock State Natural Area, FL) reared from galls on Q. Figure 2: Mean (±SE) of copulation frequency among B. treatae virginiana (N = 291 total; see Figure 2 for sample sizes individuals from their own host-associated population or the per treatment). Trials again took place within 25 × 8cm alternative host-plant-associated population when host plants were absent or present. Sexual isolation is indicated by a difference clear-plastic cups. In half of the trials we placed a small, between copulation frequency when paired with an individual from defoliated, dried twig for the wasps to walk on as a control. the “same” host plant population versus a “different” host plant Alternatively, in half the trials a small leaf-bearing section population. The additional effect of habitat isolation is indicated by of stem of the species of oak representing each individual’s a shift in the magnitude of the difference between same and different host plant was added. One male and one female were then host pairings when the host plant is present during the mating assay. aspirated into each cup (replicate). Each pair was observed at Numbers above the SEs are the number of replicates. 4 International Journal of Ecology

five-minute intervals for one hour for a total of 12 observa- Table 2: ANOVA: sources of variation in relative host plant pref- tions. At each survey we recorded each individual’s location erence of individual B. treatae assessed from no-choice preference (on Q. virginiana,onQ. geminata, or test arena) and whether assays. the pair was copulating. Copulations were defined as males Source df SS FP having mounted the female with abdomens in contact. An < additional estimate of host plant preference was calculated Population 5 2124.1 29.88 0.0001 during these mating trials based on the proportion of time Sex 1 10.1 0.71 0.4009 (n/12 observation periods) that wasps of each sex were Population × sex 5 585.7 8.24 <0.0001 observed on each host plant. Estimates were then converted Error 203 5803.1 to a relative value of host preference as previously described. For interpopulation pairings, the average host preference of males and females was compared to the probability of a degree of preference for the native host plant. The KSP successful hybrid mating (i.e., copulation). and HH populations exhibited stronger preferences than the GR population as shown by Tukey’s HSD test following the 2.4. Statistical Analyses. To test for differences in host plant ANOVA (KSP = 0.71 ± 0.02SE, HH = 0.70 ± 0.02SE, GR = > P< . preferences of individual wasps between the sexes, among 0.60 ± 0.02SE; Tukey’s HSD test: [KSP = HH] GR, 0 05; populations, and their interaction, we conducted an ANOVA Figure 1). The three populations of B. treatae associated with ff on individual relative preference for Q. virginiana (1- Q. geminata did not di er in their degree of native host preference for Q. geminata) followed by Tukey’s HSD test to preference (ABS = 0.33 ± 0.02SE, S = 0.28 ± 0.03SE, AP P compare means among populations. Population was treated = 0.29 ± 0.03SE; Tukey’s HSD test: ABS = S = AP], . as a random effect; sex was treated as a fixed effect. A similar 0 05; Figure 1). Females consistently expressed stronger host analysis of host preference expressed by individual wasps preference than did males for their native host plant across all from the two population sources when the sexes were paired six populations (Figure 1). The significant population × sex for the mating assays was also performed. We also compared interaction term in the ANOVA (Table 2) demonstrates that ff each population’s relative preference for its native host to a the degree of di erence between males and females in host value of 0.5 by means of a t-test. The value 0.5 indicates equal preference varied by population. For example, males and time spent on each of the two host plants and characterizes females from KSP preferred their native host Q. virginiana “no preference.” similarly, but males and females from HH varied by over To test for assortative mating, we used logistic regression 30%, with females expressing strong host preference for Q. to examine the effects of male host plant, female host plant, virginiana and males spending a similar amount of time on the presence/absence of the host plant, and their interactions each host plant (Figure 1). on copulation frequency in the mating assays. The two- way interaction term, female host plant × male host plant, 3.2. Host Plant Effects on Mating Preference. Patterns of tests for overall assortative mating whereas the three-way B. treatae copulation frequency (number of copulations/ interaction term, female host plant × male host plant × host number of mating trials) revealed strong evidence of host- plant present/absent, tests the effect of habitat preference on associated sexual isolation between individual gall wasps assortative mating. To examine the role of habitat isolation from the HH Q. virginiana and the ABS Q. geminata source on sexual isolation further, we compared the host preference populations. Importantly for the hypothesis that habitat expressed by the male and female in each interpopulation isolation drives reproductive isolation, B. treatae gall wasps mating assay when host plants were present between those were more likely to copulate when paired with individuals assays that resulted in a “hybrid” copulation and those that from the same host plant than from the alternative host plant did not by means of a standard t-test. as shown by the significant interaction term, female host × male host (Table 3, Figure 2). Moreover, the magnitude of 3. Results sexual isolation significantly increased in the presence of the host plant (three-way interaction: female host × male host 3.1. Habitat Plant Preference. Geographic variation in rela- × host plant; Table 2). This result is explained in part by tive host plant preference among the six B. treatae popula- the 47% decrease in the frequency of between-host matings tions is evidenced by the significant population term in the when host plants were present during the mating assays ANOVA of host preference assays (Table 2). The difference (Figure 2). This result is likely due to the intrinsic effect of in preferences among the populations is clearly associated habitat isolation arising from host preference that is effective with the host plant from which the B. treatae populations even within the confines of the small enclosures used for the were collected (Figure 1). Each population preferred its mating assays. native host, as shown by the highly significant difference The average host preference expressed by paired male between relative preference for native host and the no-choice and female B. treatae during mating trials was similar to expectation of equal time (t-test of population mean versus that revealed by the testing of individuals shown in Figure 1. 0.5: KSP tdf=32 = 6.31, HH tdf=41 = 6.76, GR tdf=36 = 5.42, However, if a mating did occur between B. treatae from AP tdf=29 = 8.11, ABS tdf=38 = 5.45, S tdf=33 = 10.11; different host plants, the average host preference expressed P<0.0001 for all comparisons). There was also significant by a pair of individuals was associated with the degree of variation among Q. virginiana associated populations in the sexual isolation among them. Interestingly, this appeared to International Journal of Ecology 5

1 1 A P<0.0001 B N.S.

0.8 0.8

0.6 0.6

0.4 0.4 Male host preference Male Female host preference Female

0.2 0.2

0 0 Yes No Yes No Interpopulation copulation? Interpopulation copulation?

Figure 3: Mean (±SE) of relative host preference expressed by individual (A) males and (B) females in the interpopulation (Q. geminata × Q. virginiana) mating trials as a function of copulation success. The greater observed host preference of males in those mating trials that did not result in a copulation further support the role of habitat isolation as an important premating reproductive barrier among gall wasp populations.

Table 3: Logistic regression analysis of sources of variation in differing selective regimes is antagonistic to local adaptation, mating preferences of male and female B. treatae assayed from no- population differentiation, and speciation [42]. Habitat iso- choice mating trials. This analysis evaluates the influence of the lation arising from the evolution of habitat preferences can male and female B. treatae source population and the presence reduce dispersal between contrasting habitats and promote versus the absence of the host plant during mating trials on the = adaptive divergence [8, 10, 33, 36, 43]. Perhaps nowhere is probability of mating. Foliage present or absent; male and female this more evident than among herbivorous insect specialists, host = host plant species from which each sex was collected (Q. geminata or Q. virginiana). who tend to oviposit, feed, rest, develop, and mate on their host plants [23]. In the present study, we demonstrated Likelihood that multiple populations of the gall former B. treatae, Source df P ratio χ2 each inhabiting either of two closely related oak species Foliage 1 1.76 0.1841 [37, 38] each express strong preferences for their natal host Male host 1 5.40 0.0202 plant species. This preference was especially apparent among females as shown in Figures 1 and 3. Our results are consis- Female host 1 0.06 0.7946 tent with partial habitat isolation in gall wasps evolving as a × Foliage male host 1 0.40 0.5256 byproduct of adaptation to different hosts, as has now been Foliage × female host 1 1.87 0.1710 demonstrated in a number of plant-insect systems [8, 10, 20]. Male host × female host 1 60.76 <0.0001 Our experimental assays of host preference demonstrated Male host × female host × foliage 1 6.48 0.0109 that (a) each host-associated population accepts its native oak more than the alternative sister species of oak, (b) populations vary in the degree of host preference exhibited, be driven by variation of the male preference (t-test: t = and (c) individuals within host-associated populations vary 3.875, df = 62, P<0.0003; Figure 3(A)) rather than the in the strength of their preference, a proxy for ecological female preference (t-test: t = −0.49, df = 63, P = 0.9614; specialization. In light of the spatial distribution of the Figure 3(B)). two species of oaks throughout central Florida, our results suggest an active role for habitat isolation in the ongoing 4. Discussion evolution of reproductive isolation in this species of gall former. Given that the expansive geographic range of B. 4.1. Host Preference, Mating Preference, and Population Differ- treatae (Florida to Texas) spans the geographic ranges of entiation. Spatially divergent selection among populations the six species of closely related live oaks that constitute the leading to local adaptation is recognized as being central to series Virentes [37, 38, 40], our results hint at the as yet initiating the divergence of incipient species [41]. However, unexplored prospect of replicated regional differentiation in dispersal and gene flow among populations experiencing host preference within B. treatae. 6 International Journal of Ecology

An underlying assumption of our current measurement 4.2. Habitat Preference and Speciation. Divergent habitat of host preference is that time spent on a host plant is (e.g., host plant) preferences can promote the speciation correlated with mating and oviposition decisions, as has been process in two ways: (a) directly by reducing encounters shown in other host-associated and ecologically divergent between potential mates and driving assortative mating insect populations [44]. However, we consider residence and (b) indirectly by generally reducing gene flow, which time in our study to be a conservative measure of host facilitates overall adaptive divergence and increases the preference, as in two other species that form host-associated opportunity for postzygotic barriers to arise. Habitat pref- populations, E. solidaginis and R. pomonella, the insect is erence is considered to act directly as a form of reproductive more likely to sit on the alternate host plant than to mate isolation; however, to date, only a modest number of studies or oviposit on it [8, 44]. have actually demonstrated that habitat preference results in Through mating assays we showed that individual B. assortative mating and reduced gene flow. Cage experiments treatae prefer to mate with individuals from their natal oak show increased assortative mating between host-associated population rather than individuals from the alternative oak populations of Eurosta solidaginis gall flies when host plants species. Most important to the present study was our explicit are present relative to when they are absent [36]. Mark- test of the role of host preference, an estimate of habitat recapture studies of hawthorn and apple host races of isolation, on the degree of sexual isolation among these Rhagoletis pomonella flies suggest that the tendency of flies to ecologically divergent host-associated populations. Here, reproduce on the same host species used in earlier life cycle comparing sexual isolation with and without host plants stages strongly reduces gene flow between the races [8, 10]. being present in the experimental arena does just this [10, Most convincingly, a combination of field and molecular 23]. We found that the presence of the host plant during data indicates that variation in host plant choice reduces mating trials increased the degree of sexual isolation among gene flow between clover- and alfalfa-adapted populations of B. treatae reared from different host plants by reducing Acyrthosiphon pisum pea aphids [32]. between-host matings by 47%. Moreover, we found this Habitat preferences can also indirectly contribute to effect to be associated with variation in the host plant the speciation process by (a) reducing the constraining preference exhibited by males as most matings occurred effects of gene flow on adaptive divergence in ecologically on the female’s host plant, regardless of the type of cross important traits [13, 33], (b) promoting postzygotic isolation (interpopulation or intrapopulation). through less fit hybrids [46–48], and (c) increasing the Our mating assays constituted “no-choice” conditions opportunity for Dobzhansky-Muller postzygotic barriers to where females were paired with males from one of two pos- arise [49]. Thus, when divergent habitat preference acts as a sible populations. An alternative approach can involve barrier to gene flow, additional prezygotic and postzygotic “choice” experiments, in which individuals can choose be- reproductive barriers can evolve via the byproduct model tween inhabitants from each population. “Choice” tests of ecological speciation [1, 10, 23, 50]. Under this model, can offer a different perspective of mate choice, commonly reproductive barriers evolve as an indirect consequence of observing stronger preferences than “no-choice” tests [45]. reduced gene flow rather than as a direct result of selection. In our assays, we imagined the biologically relevant scenario to be one in which an individual on its native host plant 4.3. Conclusions. While habitat isolation is thought to play a encountered a single migrant in a “no-choice-” type scenario. critical role in premating reproductive isolation among All adults assayed in both the host and mate preference herbivorous insect populations and, in general, among all trials were reared directly from their native host hence we taxa undergoing ecological divergence and speciation, very cannot rule out host environment as a contributing factor to rarely is the role of habitat selection directly tested [23]. the observed patterns of host and mate preferences. Recipro- In the present study, we used a combination of habitat cal transplant experiments, repeated across populations, will be needed to distinguish genetic and environmental contri- preference and mate preference assays among ecologically butions to the observed differences in habitat and mating divergent populations of the gall wasp Belonocnema treatae to preferences and assess the adaptive nature of these traits document variation among populations in habitat (i.e., host through measurements of the fitness of each population on plant) preference and examine the role of habitat isolation to the two host species. However, the observed differences in overall reproductive isolation. Overall, all populations exam- host preference, even if due to host environment, would ined showed habitat fidelity and habitat preference decreased still be expected to contribute to divergence of B. treatae the probability of mating between individuals from alter- populations, as parental generation migrant wasps would native host plants. The increase in the degree of assortative be predicted to be averse to settling on, or mating with mating due to the presence of the host plant during mate individuals from, the alternative host plant. Thus, our ex- choice provides an example of the importance of habitat perimentssupportacriticalroleforhostplantuseinpro- isolation in promoting reproductive isolation between host- moting reproductive isolation among populations of B. plant-associated populations of herbivorous insects. treatae regardless of the underlying basis for preference variation. Future work will test additional populations in Acknowledgments mate choice assays to assess the generality of the current support for the hypothesis that habitat isolation directly The authors thank M. Deyrup and Archbold Biological contributes to sexual isolation during mate choice. Station for providing laboratory space and help with host International Journal of Ecology 7 plant identification, J. L. Greene for help with experiments, [20] S. P. Egan and D. J. Funk, “Ecologically dependent postmating and E. Silverfine for editorial comments. isolation between sympatric host forms of Neochlamisus bebbianae leaf beetles,” Proceedings of the National Academy of Sciences of the United States of America, vol. 106, no. 46, pp. References 19426–19431, 2009. [21] C. S. Mcbride and M. C. Singer, “Field studies reveal strong [1] D. Schluter, The Ecology of Adaptive Radiation,OxfordUniver- postmating isolation between ecologically divergent butterfly sity Press, Oxford, UK, 2000. populations,” PLoS Biology, vol. 8, no. 10, Article ID e1000529, [2] D. Schluter, “Ecology and the origin of species,” Trends in Ecol- 2010. ogy and Evolution, vol. 16, no. 7, pp. 372–380, 2001. [22] S. H. Berlocher and J. L. Feder, “Sympatric speciation in phy- [3] H. D. Rundle and P.Nosil, “Ecological speciation,” Ecology Let- tophagous insects: moving beyond controversy?” Annual ters, vol. 8, no. 3, pp. 336–352, 2005. Review of Entomology, vol. 47, pp. 773–815, 2002. [4] E. Mayr, “Ecological factors in speciation,” Evolution, vol. 1, [23] D. J. Funk, K. E. Filchak, and J. L. Feder, “Herbivorous insects: pp. 263–288, 1947. model systems for the comparative study of speciation ecol- [5] T. Dobzhansky, Genetics and the Origin of Species, Columbia ogy,” Genetica, vol. 116, no. 2-3, pp. 251–267, 2002. University Press, New York, NY, USA, 3rd edition, 1951. [24] E. A. Bernays and R. F. Chapman, Host–Plant Selection by Phy- [6] G. G. Simpson, The Major Features of Evolution, Columbia tophagous Insects, Chapman and Hall, London, UK, 1994. University Press, New York, NY, USA, 1953. [25] M. C. Singer and C. D. Thomas, “Evolutionary responses of [7] D. J. Funk, P. Nosil, and W. J. Etges, “Ecological divergence a butterfly metapopulation to human- and climate-caused exhibits consistently positive associations with reproductive environmental variation,” American Naturalist, vol. 148, pp. isolation across disparate taxa,” Proceedings of the National S9–S39, 1996. Academy of Sciences of the United States of America, vol. 103, [26] S. P. Carroll, H. Dingle, and S. P. Klassen, “Genetic differenti- no. 9, pp. 3209–3213, 2006. ation of fitness-associated traits among rapidly evolving pop- [8]J.L.Feder,S.B.Opp,B.Wlazlo,K.Reynolds,W.Go,andS. ulations of the soapberry bug,” Evolution, vol. 51, no. 4, pp. Spisak, “Host fidelity is an effective premating barrier between 1182–1188, 1997. sympatric races of the apple maggot fly,” Proceedings of the [27] C. Mitter, B. Farrell, and B. Wiegmann, “The phylogenetic National Academy of Sciences of the United States of America, study of adaptive zones: has phytophagy promoted insect vol. 91, no. 17, pp. 7990–7994, 1994. diversification?” American Naturalist, vol. 132, no. 1, pp. 107– [9] E. Rolan-Alvarez,´ K. Johannesson, and J. Erlandsson, “The 128, 1988. maintenance of a cline in the marine snail Littorina saxatilis: [28] B. D. Farrell, “‘Inordinate fondness’ explained: why are there the role of home site advantage and hybrid fitness,” Evolution, so many beetles?” Science, vol. 281, no. 5376, pp. 555–559, vol. 51, no. 6, pp. 1838–1847, 1997. 1998. [10] D. J. Funk, “Isolating a role for natural selection in speciation: [29] B. D. Walsh, “On phytophagic varieties and phytophagic spe- host adaptation and sexual isolation in Neochlamisus beb- cies,” Proceedings of the Entomological Society of Philadelphia, bianae leaf beetles,” Evolution, vol. 52, no. 6, pp. 1744–1759, vol. 3, pp. 403–430, 1864. 1998. [30] G. L. Bush, “Sympatric host race formation and speciation in [11]H.D.Rundle,L.Nagel,J.W.Boughman,andD.Schluter, frugivorous flies of genus Rhagoletis (Diptera, Tephritidae),” “Natural selection and parallel speciation in sympatric stick- Evolution, vol. 23, pp. 237–251, 1969. lebacks,” Science, vol. 287, no. 5451, pp. 306–308, 2000. [31] T. P. Craig, J. D. Horner, and J. K. Itami, “Hybridization stud- [12] R. B. Langerhans, M. E. Gifford, and E. O. Joseph, “Ecological ies on the host races of Eurosta solidaginis: implications for speciation in Gambusia fishes,” Evolution,vol.61,no.9,pp. sympatric speciation,” Evolution, vol. 51, no. 5, pp. 1552–1560, 2056–2074, 2007. 1997. [32] S. Via, “Reproductive isolation between sympatric races of pea [13] P. Nosil, “Divergent host plant adaptation and reproductive aphids. I. Gene flow restriction and habitat choice,” Evolution, isolation between ecotypes of Timema cristinae walking vol. 53, no. 5, pp. 1446–1457, 1999. sticks,” American Naturalist, vol. 169, no. 2, pp. 151–162, 2007. [33] P.Nosil, C. P.Sandoval, and B. J. Crespi, “The evolution of host [14] D. B. Lowry, R. C. Rockwood, and J. H. Willis, “Ecological preference in allopatric vs. parapatric populations of Timema reproductive isolation of coast and inland races of Mimulus cristinae walking-sticks,” Journal of Evolutionary Biology, vol. guttatus,” Evolution, vol. 62, no. 9, pp. 2196–2214, 2008. 19, no. 3, pp. 929–942, 2006. [15]S.P.Egan,G.R.Hood,J.L.Feder,andJ.R.Ott,“Divergent [34] H. Katakura, M. Shioi, and Y. Kira, “Reproductive isolation host plant use promotes reproductive isolation among cynipid by host specificity in a pair of phytophagous ladybird beetles,” gall wasp populations,” Biology Letters, in press. Evolution, vol. 43, no. 5, pp. 1045–1053, 1989. [16] J. A. Coyne and H. A. Orr, Speciation, Sinauer, Sunderland, [35] C. Linn Jr., J. L. Feder, S. Nojima, H. R. Dambroski, S. H. Mass, USA, 2004. Berlocher, and W. Roelofs, “Fruit odor discrimination and [17] T. K. Wood and M. C. Keese, “Host-plant-induced assortative sympatric host race formation in Rhagoletis,” Proceedings of the mating in Enchenopa treehoppers,” Evolution, vol. 44, no. 3, National Academy of Sciences of the United States of America, pp. 619–628, 1990. vol. 100, no. 20, pp. 11490–11493, 2003. [18] L. Nagel and D. Schluter, “Body size, natural selection, and [36] T. P. Craig, J. K. Itami, W. G. Abrahamson, and J. D. Horner, speciation in sticklebacks,” Evolution, vol. 52, no. 1, pp. 209– “Behavioral evidence for host-race formation in Eurosta 218, 1998. solidaginis,” Evolution, vol. 47, no. 6, pp. 1696–1710, 1993. [19] P. Nosil and B. J. Crespi, “Ecological divergence promotes the [37] J. Cavender-Bares, K. Kitajima, and F. A. Bazzaz, “Multiple evolution of cryptic reproductive isolation,” Proceedings of the trait associations in relation to habitat differentiation among Royal Society of London B, vol. 273, no. 1589, pp. 991–997, 17 Floridian oak species,” Ecological Monographs, vol. 74, no. 2006. 4, pp. 635–662, 2004. 8 International Journal of Ecology

[38] J. Cavender-Bares and A. Pahlich, “Molecular, morphological, and ecological niche differentiation of sympatric sister oak species, Quercus virginiana and Q. geminata (Fagaceae),” American Journal of Botany, vol. 96, no. 9, pp. 1690–1702, 2009. [39] J. N. Lund, J. R. Ott, and R. J. Lyon, “Heterogony in Belonoc- nema treatae Mayr (Hymenoptera: Cynipidae),” Proceedings of the Entomological Society of Washington, vol. 100, no. 4, pp. 755–763, 1998. [40] J. Cavender-Bares, A. Gonzalez-Rodriguez, A. Pahlich, K. Koehler, and N. Deacon, “Phylogeography and climatic niche evolution in live oaks (Quercus series Virentes) from the tropics to the temperate zone,” Journal of Biogeography, vol. 38, no. 5, pp. 962–981, 2011. [41] T. J. Kawecki and D. Ebert, “Conceptual issues in local adaptation,” Ecology Letters, vol. 7, no. 12, pp. 1225–1241, 2004. [42] P. Edelaar, A. M. Siepielski, and J. Clobert, “Matching habitat choice causes directed gene flow: a neglected dimension in evolution and ecology,” Evolution, vol. 62, no. 10, pp. 2462– 2472, 2008. [43] D. I. Bolnick, L. K. Snowberg, C. Patenia, W. E. Stutz, T. Ingram, and O. L. Lau, “Phenotype-dependent native habitat preference facilitates divergence between parapatric lake and stream stickleback,” Evolution, vol. 63, no. 8, pp. 2004–2016, 2009. [44] W. G. Abrahamson and A. E. Weis, Evolutionary Ecology across Three Trophic Levels: Goldenrods, Gallmakers, and Natural Enemies, Monographs in Population Biology, Princeton Uni- versity Press, Princeton, NJ, USA, 1997. [45] J. A. Coyne, S. Elwyn, and E. Rolan-Alvarez,´ “Impact of experimental design on Drosophila sexual isolation studies: direct effects and comparison to field hybridization data,” Evolution, vol. 59, no. 12, pp. 2588–2601, 2005. [46] C. E. Linn Jr., H. R. Dambroski, J. L. Feder, S. H. Berlocher, S. Nojima, and W. L. Roelofs, “Postzygotic isolating factor in sympatric speciation in Rhagoletis flies: reduced response of hybrids to parental host-fruit odors,” Proceedings of the National Academy of Sciences of the United States of America, vol. 101, no. 51, pp. 17753–17758, 2004. [47] H. R. Dambroski, C. Linn, S. H. Berlocher, A. A. Forbes, W. Roelofs, and J. L. Feder, “The genetic basis for fruit odor discrimination in Rhagoletis flies and its significance for sympatric host shifts,” Evolution, vol. 59, no. 9, pp. 1953–1964, 2005. [48] J. L. Feder, S. P. Egan, and A. A. Forbes, “Ecological adaptation and speciation: the evolutionary significance of habitat avoidance as a postzygotic reproductive barrier to gene flow,” International Journal of Ecology. In press. [49] H. A. Orr and D. C. Presgraves, “Speciation by postzygotic isolation: forces, genes and molecules,” BioEssays, vol. 22, no. 12, pp. 1085–1094, 2000. [50] P. Nosil, B. J. Crespi, and C. P. Sandoval, “Host-plant adaptation drives the parallel evolution of reproductive isolation,” Nature, vol. 417, no. 6887, pp. 440–443, 2002. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 256017, 12 pages doi:10.1155/2012/256017

Review Article Underappreciated Consequences of Phenotypic Plasticity for Ecological Speciation

Benjamin M. Fitzpatrick

Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN 37996, USA

Correspondence should be addressed to Benjamin M. Fitzpatrick, benﬁ[email protected]

Received 25 July 2011; Revised 22 October 2011; Accepted 12 December 2011

Academic Editor: Andrew Hendry

Copyright © 2012 Benjamin M. Fitzpatrick. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Phenotypic plasticity was once seen primarily as a constraint on adaptive evolution or merely a nuisance by geneticists. However, some biologists promote plasticity as a source of novelty and a factor in evolution on par with mutation, drift, gene flow, and selection. These claims are controversial and largely untested, but progress has been made on more modest questions about effects of plasticity on local adaptation (the first component of ecological speciation). Adaptive phenotypic plasticity can be a buffer against divergent selection. It can also facilitate colonization of new niches and rapid divergent evolution. The influence of nonadaptive plasticity has been underappreciated. Non-adaptive plasticity, too can interact with selection to promote or inhibit genetic differentiation. Finally, phenotypic plasticity of reproductive characters might directly influence evolution of reproductive isolation (the second component of ecological speciation). Plasticity can cause assortative mating, but its influence on gene flow ultimately depends on maintenance of environmental similarity between parents and offspring. Examples of plasticity influencing mating and habitat choice suggest that this, too, might be an underappreciated factor in speciation. Plasticity is an important consideration for studies of speciation in nature, and this topic promises fertile ground for integrating developmental biology with ecology and evolution.

1. Introduction of speciation” [13–15]isaspecialcaseofecologicalspecia- tion, and I review the subject by breaking down the effects Phenotypic plasticity has often been seen primarily as an of plasticity on the two components of ecological speciation: alternative to genetic divergence and a feature making popu- adaptive divergence and the evolution of reproductive isola- lations less responsive to natural selection [1–3]. For exam- tion [28]. I close with a few suggestions for future work. ple, those studying adaptation and speciation have often used By any definition, speciation requires genetic divergence. phrases like “merely plastic” to contrast environmentally Therefore, integration of ecological developmental biology ff induced variation against geographic or species di erences with the well-developed body of fact and theory on the with strong genetic bases [4–8]. However, others suggested genetics of speciation [29–31]willbemoreproductivethan that phenotypically plastic traits can promote adaptive evo- attempting to replace this population genetic foundation. lution and the origin of species [9–16]. The general issue of Recent reviews and models support this perspective [27, 32– plasticity and adaptation has been reviewed extensively in the 37]. last decade (e.g., [15, 17–26]). Adaptive plasticity’s impact on speciation was recently reviewed by Pfennig et al. [27], and I do not attempt to duplicate their efforts. Instead I make a few 2. Definitions points that have not been emphasized in the recent literature. In particular, nonadaptive plasticity and environmentally Understanding the relationship between environmental induced barriers to gene flow deserve greater attention. induction and speciation requires a set of consistent def- After making explicit my working definitions of key initions. Terms like “environment,” “speciation,” “plastic- terms, I argue that the “developmental plasticity hypothesis ity,” and “natural selection” are sometimes assumed by 2 International Journal of Ecology different workers to have different definitions, and this can To put it another way, natural selection exists whenever affect communication [38]. The definitions that follow are phenotypic variation causes covariance between phenotype intended to clarify what I mean by certain words and phrases and fitness [43, 44]. Fitness is metaphorical shorthand for within this paper; they are not intended to challenge or the ability to survive and reproduce. It is important to replace alternative definitions. I believe I have followed recent emphasize that natural selection, under this definition, can convention in all cases [39, 40]. exist without genetic variation and can recur over many generations without causing evolution [15, 17]. Adaptation. Genetic change in response to natural selection and resulting in organisms with improved performance with Plasticity. The ability of a single genotype to express different respect to some function or feature of the environment. phenotypic values or states under different environmental conditions, that is, in response to environmental induction. Countergradient Variation. Pattern of geographic variation Plasticity can include developmental plasticity, physiological in which genetic differences between populations affect acclimation, or behavioral flexibility. Plasticity might be their phenotypes in the opposite way from environmental adaptive or not. Adaptive plasticity is a tendency for a differences between populations. For example, if the mean genotype to express a phenotype that enhances its ability phenotype of population i is given by the sum of genetic to survive and reproduce in each environment. Nonadaptive and environmental effects Pi = Gi + Ei, and the effect of plasticity includes any response to environmental induc- environment 2 tends to increase the phenotypic value relative tion that does not enhance fitness (including maladaptive ff to environment 1 (E2 >E1), countergradient variation would responses). Noisy plasticity is e ectively unpredictable phe- exist if G2 anagenesis or ff Environmental Induction. Any e ect of environment on trait phyletic speciation [48–50]. In any case, speciation is usually expression is environmental induction. Environmental a gradual, continuous process of genetic divergence resulting induction usually refers to an event (developmental outcome in a discontinuous pattern of variation (species taxa). Ecolog- or process) caused by an environmental condition, whereas ical speciation is the evolution of reproductive isolation as a plasticity (see below) refers to the propensity of an organism consequence of divergent ecological adaptation [28, 51, 52]. or trait to respond to environmental change. Without reproductive isolation, this is local adaptation. This definition is based on the biological species concept [53], Genetic Assimilation. Evolutionary reduction in the degree which emphasizes genetically based reproductive isolation as of plasticity such that a character state or trait value that was the primary explanation for the existence of distinct kinds once conditionally expressed depending on the environment of organisms (i.e., those recognized as species taxa). I am becomes expressed constitutively (unconditionally, regard- not making a recommendation about taxonomic practice. less of environment). Rather, from the perspective of evolutionary biology, the evolution of reproductive isolation is what distinguishes spe- Natural Selection. Natural selection refers to any nonrandom ciation from more general phenomena of genetic divergence difference between entities in survival or reproduction. [30, 40]. International Journal of Ecology 3

3. The Developmental Plasticity Hypothesis of Speciation

Matsuda [54] hypothesized that phenotypic plasticity was a crucial first step in the adaptive evolution of distinct, ecologically specialized lineages. As an example, Matsuda (a) [54] suggested that major differences in life history, such as presence or absence of metamorphosis prior to reproduction in salamanders (Figure 1), likely began as nongenetic polyphenisms and evolved via genetic assimilation in habi- (b) tat specialists. Widespread generalists such as Ambystoma tigrinum and A. velasci show conditional expression of metamorphosis from aquatic larva to terrestrial adult in small temporary ponds versus a fully aquatic life cycle with (c) no metamorphosis in more permanent water bodies. This Figure 1: The tiger salamander radiation exemplifies the develop- polyphenism likely represents the ancestral state of the tiger mental plasticity hypothesis of ecological speciation [13–15, 54]. salamander clade [55]. In several isolated lakes in Mexico, (a) Environmentally induced alternative phenotypes of the tiger permanently aquatic endemics such as A. mexicanum and salamander (Ambystoma tigrinum) include terrestrial metamorphs A. dumerilii no longer express metamorphosis in nature and aquatic paedomorphs. (b) The California tiger salamander owing to genetic changes in the thyroid hormone system (A. californiense) is an obligate metamorph derived from a devel- [54–56]. opmentally plastic common ancestor with A. tigrinum [55]. (c) The Mexican axolotl (A. mexicanum) is one of several obligate Along the same lines, West-Eberhard [13–15] proposed paedomorphs, independently derived from plastic ancestors within a generalized “developmental plasticity hypothesis of speci- the last few million years [55, 57, 58]. ation” in which the evolution of ecologically distinct forms in different environments depends on the initial appearance of those distinct forms as alternative phenotypes in a phenotypically plastic ancestor. She argued that when adaptive 4. Phenotypic Plasticity and phenotypic plasticity results in strong associations between Ecological Divergence phenotypes and environments, rapid speciation could occur Phenotypic plasticity can slow or enhance genetic divergence. in three steps. First, alternative phenotypes become fixed in ff different populations owing to environmental differences, How plasticity a ects divergence depends to some extent on whether plasticity is adaptive or not. but with little or no genetic change. Then, genetic assimilation and/or other adaptive modifications of each phenotype occur owing to divergent selection. Finally, reproductive 4.1. Adaptive Plasticity. Adaptive plasticity can dampen or isolation evolves as a byproduct of adaptive divergence or eliminate divergent selection. If any individual can express via reinforcement if there is contact between the diverging a nearly optimal phenotype in whatever environment it populations. finds itself, then there is little or no variation in the ability Clearly, West-Eberhard’s [13–15] hypothesis is a kind of to survive and reproduce, hence little or no divergent ecological speciation in which developmental plasticity pro- selection [64]. This has long been an intuitive reason to motes genetic divergence in response to ecologically based regard plasticity as a constraint on genetic evolution and selection. In the absence of plasticity, divergence might be to discount the evolutionary potential of environmentally prevented entirely if the single expressed phenotype cannot induced variation [1]. Models have supported the prediction that adaptive plasticity can effectively take the place of genetic establish a viable population in the alternative environment divergence between environments [65–67]. And a large [22, 27, 36] or might be much slower if phenotypic diver- number of empirical studies are consistent with increased gence must await new mutations and their gradual fixation plasticity in species with high dispersal rates [68]. However, [13–15]. the extent to which plasticity prevents or slows genetic West-Eberhard’s developmental plasticity hypothesis of divergence depends on several factors explored by Thibert- speciation is focused on adaptive phenotypic plasticity and Plante and Hendry [36] in individual-based simulations. its influence on one component of ecological speciation: First, is development sufficiently flexible that an individ- the evolutionary response to divergent selection. How- ual can express traits near either environmental optimum? ever, nonadaptive plasticity might be equally if not more Given alternative environments or niches with divergent influential in promoting an evolutionary response [18]. fitness functions, the only way environmental induction can Further, the other component of ecological speciation, the completely eliminate divergent selection is to cause the mean evolution of reproductive isolation [28, 62, 63], also can expressed trait of a single gene pool to match the optimum be directly influenced by phenotypic plasticity. In the next in each environment [22] (Figure 2). If adaptive plasticity is sections, I examine how plasticity can interact with these two less than perfect, divergent selection might still exist. Then components of ecological speciation. the question is whether plasticity quantitatively dampens 4 International Journal of Ecology

to demographic factors [74–76]. Therefore the impact of adaptive plasticity on the potential for genetic divergence depends on how it affects the tension between divergent Ancestral environment selection and gene flow [36, 77, 78]. Moreover, plasticity itself is adaptive only to the extent that individuals have a reasonable chance of experiencing alternative environments. If the Trait x populations expressing alternative phenotypes are isolated Fitness in their respective environments, the ability to express the Frequency alternative phenotype is likely to be lost owing to selection for efficient development or simply because loss of function (a) mutations are likely to accumulate neutrally in genes that are never expressed [79]. This process of a conditionally expressed trait becoming constitutively expressed is genetic Zero plasticity: assimilation [11, 16, 80]. High chance of extinction Third, are the systems sensing the environment and in new environment regulating trait expression sufficiently accurate that the best phenotype is reliably expressed in each environmental Trait x context? There are two components to this, first is simply the (b) question of how well developmental or behavioral systems are able to sense and react to environmental stimuli [81]. Again, if adaptive plasticity is less than perfect, divergent Moderate plasticity: selection can exist. Second is the question of whether the tim- Facilitates colonization ing of key developmental and life history events is such that and divergent selection future environmental conditions can be correctly predicted in new environment [36]. For example, if individuals disperse and settle before completing development (e.g., seeds or planktonic larvae), Trait x they might be able to accurately tune their adult phenotypes (c) to the environment in which they settle. However, if a substantial number of individuals disperse after completing development (e.g., animals with extended parental care Strong plasticity: [82]), then developmental plasticity would do little to help Prevents divergent them accommodate new environmental challenges because selection their phenotypes are adjusted to their natal habitat rather than their new habitat. In this case, there might be strong Trait x selection against immigrants before any genetic differences (d) arise between populations [36]. Finally, is there any cost to plasticity? Several modeling Figure 2: Effects of adaptive plasticity on colonization and studies have confirmed the idea that plasticity is less likely adaptation to a new niche. A population well adapted to its niche to evolve if there are fitness costs to maintaining multiple is illustrated in (a) by coincidence of the mean trait value (black developmental pathways or changing expression during line) with the peak of the fitness function (dashed line). If the ff development [36]. Empirical tests for costs of plasticity itself fitness function is dramatically di erent in a new environment, a are rare [83], but it is conceivable that some pathways might population with trait values favored in the old environment (b) ff ffi might have such low fitness as to have little chance of survival. have inherent tradeo sbetweene ciency and plasticity If environmental induction produces a shift in trait values toward [84, 85], and adaptive plasticity probably always comes with higher fitness phenotypes (c), the population might persist but still some potential for error; that is, the best developmental experience selection. If phenotypic plasticity results in a perfect “decision” might not be made every time [64, 81]. When match between mean trait value and fitness optimum (d), then there plasticity is costly enough to outweigh its fitness benefits, is no effect of selection on the population mean. possible alternative outcomes are the evolution of a single “compromise” or generalist phenotype, evolution of a simple genetic “switch” enabling coexistence of alternative specialist phenotypes [14, 15, 86], or divergent evolution of specialist the fitness tradeoff enough to substantially slow or prevent populations (local adaptation) [45, 67, 87]. In general we divergent evolution [36]. know very little about the prevalence or influence of costs Second, how much dispersal occurs between environ- of plasticity in nature. ments? In the absence of gene flow, almost any amount The potential for adaptive plasticity to evolve as a re- of divergent selection will eventually cause evolutionary sponse to ecological tradeoffs instead of genetic divergence is divergence. When there is gene flow between populations, well supported. The dampening effect of plasticity is reduced the effect of divergent selection depends on the relative but not eliminated by reduction in the extent and precision magnitudes of selection and gene flow [1, 69–73], in addition of plasticity, reduction in gene flow between environments, International Journal of Ecology 5

Cool-adapted Optimum Warm-adapted genotype

Gene expression Optimum

Warm-adapted Cool-adapted genotype

Temperature Temperature (a) (b)

Figure 3: Adaptive and nonadaptive reaction norms. These hypothetical examples suppose that the optimum expression level for some gene increases with temperature and that increased temperature induces increased expression of the gene. (a) Plasticity is adaptive when it keeps expression closer to the optimum than it would be if expression, were constant across temperatures. (b) Plasticity is nonadaptive (maladaptive, in this case) when induced changes in a new environment take expression further from the optimum than it would be if it had remained constant. (b) is an example of countergradient variation, in which genetic diﬀerences cause cool-adapted genotypes to have higher gene expression than warm-adapted genotypes at the same temperature. The result is overexpression by cool-adapted genotypes transplanted to warm environments.

and increased costs of plasticity [36]. However, plasticity can developmental evolutionary biology [14, 15, 17, 88]and actually promote genetic divergence under some conditions. treated extensively in recent reviews and models [22, 27, In particular, when development is completed after dispersal 32, 36]. The effects of nonadaptive phenotypic plasticity (e.g., sessile organisms), adaptive plasticity might make have received less attention. However, any environmental successful colonization of new environments more likely effect on phenotypes can affect the strength and direction of (Figure 2). In West-Eberhard’s [14, 15]conceptualmodel selection in addition to the genetic variances and covariances and Thibert-Plante and Hendry’s [36] mathematical model, of important traits [18, 22, 42]. Some kinds of nonadaptive individuals are able to colonize a radically new environment environmental induction might affect the probability of by adjusting developmentally, behaviorally, and/or physi- ecological speciation. In particular, suboptimal develop- ologically. This adaptive plasticity allows a population to ment or noisy plasticity [45] in stressful environments persist in the new environment, continually exposed to could inhibit adaptation by decreasing the fitness of local divergent selection. If instead all individuals entering the new relative to immigrant individuals. However, it would also environment die or leave, there is no divergent selection. increase the strength of selection and potentially result Thus, without adaptive plasticity, there might simply be in cryptic adaptive divergence (countergradient variation) suitable and unsuitable environments, with little opportunity [18, 41]. for divergent evolution. Successful colonization of a new Countergradient variation is negative covariance be- environment can initiate divergent selection, not only on tween genetic and environmental effects on phenotype [41]. the plastic trait, but possibly also on other traits, which Classic examples are poikilotherms such as fish and molluscs might then cause ecological speciation. This to some extent [89], flies [90], or frogs [91] that grow more slowly in cold reconciles the conflicting effects of plasticity. In other cir- climates, but cold-adapted populations have higher growth cumstances, when dispersal occurs after development (e.g., rates than warm-adapted populations when raised at the animals with extended parental care), individuals settling in same temperature. The best explanation for this pattern is new environments are especially likely to express suboptimal that genetic differences have evolved to compensate for diver- phenotypes, which might accentuate the effects of selection gent effects of environmental induction, resulting in popula- and spatial separation once a new habitat has been colonized tions that appear similar when measured each in their native [36]. This effect is similar to effects of nonadaptive plasticity habitat but show maladaptive plasticity when transplanted discussed below. (Figure 3). Note that negative covariance between genetic and environmental effects is not necessarily maladaptive, but 4.2. Nonadaptive Plasticity. The potential for adaptive plas- environmental effects will tend to be maladaptive if the opti- ticity to promote colonization and adaptation to alter- mum phenotype is roughly constant across the environment native environments has been promoted by advocates of range. 6 International Journal of Ecology

When countergradient variation exists, we expect immi- m = 0.5 (19% speciation) grants to have a fitness disadvantage owing to under- or overexpression of an environmentally sensitive trait relative m = 0.375 (51% speciation) to a local optimum (Figure 3). This might well promote evolution of restricted or nonrandom dispersal; hence intrinsic m = 0.25 (74% speciation) barriers to gene exchange as a result of divergent ecological selection. However, countergradient variation will evolve Flowering week only if selection is stronger than gene flow and if reaction ff norms are genetically constrained (otherwise, we might Figure 4: E ects of phenotypic plasticity of flowering time on ecological speciation. The lines illustrate flowering periods for expect adaptive evolution to flatten the reaction norm). At plants on two soil types (dashed and solid) based on initial least in the early stages of colonization of a challenging new conditions in the simulations of Gavrilets and Vose [59]for habitat, nonadaptive plasticity (such as stunted growth or ecological speciation based on the Howea palm tree study [60]. suboptimal metabolic rates) might make resident individuals I illustrate the case where flowering time is affected by 8 loci, less viable and fecund than healthy immigrants from a and the environmental effectofsoiltypeiseither0,2,or4 less stressful habitat [22]. This potential fitness asymmetry weeks difference. In the simulations [59], changing the number of could offset effects of adaptive genetic changes on the loci has confounding effects on the initial environmental variance relative fitness of immigrants and residents. That is, offspring component and the fitness effects of mutations, so a single genetic of immigrants might have lower fitness than offspring of scenario is illustrated here for simplicity. I calculated the initial rate native genotypes with locally adaptive alleles. However, of gene flow between soil types as m = 0.5 (% time overlap), based offspring of immigrants might nevertheless outnumber on the assumption that any tree is equally likely to receive pollen offspring of natives if immigrants come into the stressful from any other currently flowering tree. %speciation is based on the number of simulations ending in speciation, given in Table 2 of habitat with substantial viability and fertility advantages Gavrilets and Vose [59]. from being raised in a higher quality habitat. Parental care and physiological maternal effects could further extend those environmentally induced advantages to the offspring. The net effect could be a tendency for locally adapted genotypes to be replaced (“swamped”) by immigrants owing Environmental similarity can be a byproduct of geographic to a negative covariance between environmental and genetic isolation or might be promoted by environmentally induced effects. This effect of nonadaptive plasticity is synergistic with variation in habitat choice, phenology, or other aspects the potential for demographic swamping, a well-known con- of mating behavior. Whether or not ecological speciation straint on local adaptation to novel habitats [74, 76, 87, 92]. ensues depends on whether genetic reproductive barriers For now, it appears that the impact of nonadaptive plas- evolve and whether that evolution can be attributed to ticity on the probability of ecological speciation cannot be divergent selection. predicted without additional detailed knowledge. Just as gene flow can constrain or facilitate local adaptation [74, 76], and A seemingly commonplace example of environmentally adaptive plasticity can inhibit or promote adaptive genetic induced barriers to gene flow is flowering time in plants. divergence [22, 27, 36], nonadaptive plasticity might impede Flowering time is often accelerated or delayed when a given ff genetic divergence by accentuating fitness advantages of genotype is grown on di erent substrates [34]. For example, immigrants and/or promote divergence by increasing the grasses and monkey flowers colonizing contaminated soils intensity of selection. around mines show environmentally induced shifts in flowering [95, 96], and palm trees on Lord Howe Island show soil-dependent flowering times [60]. These examples are also 5. Phenotypic Plasticity and the Evolution widely recognized cases of ecological speciation. Gavrilets of Reproductive Isolation and Vose [59] used simulations to explore the impact of environmentally induced shifts in flowering time on the Plasticity of a different sort might directly affect repro- probability of adaptive divergence and ecological speciation ductive compatibility between populations developing in and confirmed that this instantaneous barrier to gene flow different environments. Environmental induction might between habitats can markedly increase the probability and generate differences in preference, reproductive phenology, rate of divergence (Figure 4). or expression of secondary sexual characteristics. Coinci- A similar effect arises owing to behavioural imprinting dence of environmentally induced reproductive barriers and [97, 98]. Many animals, such as birds and anadromous fishes, potentially divergent selection can be genetically equivalent are known for imprinting on their natal habitat [99–101]. to a geographic barrier between divergent environments When this is a direct matter of memorizing where home is [59, 93, 94]. A key element of the developmental plasticity (as might be the case in Ficedula flycatchers [101]), it simply model of ecological speciation is the establishment of a accentuates the relationship between geography and gene consistent relationship between the environment of parents flow. Slightly different implications emerge from imprinting and that of their offspring. Similarity of parent and offspring on a kind of habitat, host, or resource because then the environments maintains shared environmental effects on environmental influence on gene flow is independent of phenotype, consistency of selection, and reduces gene flow. geography. Some phytophagous insects imprint on their International Journal of Ecology 7 host plant species based on chemical cues [102], and nest- Environmentally induced differences in habitat choice parasitic indigobirds imprint on their hosts [103]. In these reduce gene flow when individuals are more likely to mate examples, phenotypic plasticity helps maintain similarity with other individuals using the same habitat, and offspring between maternal and offspring environments, which is are more likely to grow up in habitats similar to those of important both for maintaining phenotypic similarity and their parents. If habitat choice is entirely determined by consistency of selection on parents and offspring. However, the individual’s environment (i.e., if there is no tendency it is not always obvious that habitat or resource imprinting for the offspring of immigrants to return to their parent’s should directly affect mating preferences. In the Vidua original habitat), then the effect is genetically identical to a indigobirds, there is an imprinting effect of host song, such geographic barrier. Either way we can describe the system that nest parasites raised by the same host species tend to in terms of populations of individuals or gametes with some mate assortatively owing to learned elements of their own probability (m) of “moving” from their natal population to songs and preferences [103, 104]. In many phytophagous breed in a different population. In the simple case of two insects, mating occurs on or near the host plant [102, 105], environments and nonoverlapping generations, the expected making plant choice a “magic trait” simultaneously effecting frequency of an allele in generation t in environment i is a ecological and sexual differentiation [31]. As with genetically weighted average of the frequencies in environments i and j determined traits, phenotypic plasticity of traits directly in generation t − 1[113]: linked to both ecological adaptation and assortative mating pi t = (1 − m)pi t− + mpj t− . (1) is most likely to contribute to ecological speciation. ( ) ( 1) ( 1) Environmental effects on traits directly involved in sex- It makes no difference whether m is determined by geography ual selection are not unusual. Expression of pigments, or environmental induction as long as there are not heritable pheromones, and other displays can depend on diet, condi- differences in m among individuals within a habitat. More tion, or experience [19, 106]. For example, premating isola- generally, geographic or spatial covariance is a special case tion is induced by larval host plant differences in Drosophila of environmental similarity, and the extensive knowledge mojavensis, because the chemical properties of their cuticular from decades of conceptual and mathematical modeling hydrocarbons (important contact pheromones) are strongly of gene flow’s effects on adaptation and speciation [30, influenced by diet [107, 108]. Sharon et al. [109] recently 31, 87, 114, 115] can be extended directly to include showed that mate choice in D. melanogaster can be modified this kind of phenotypic plasticity. In particular, we might by symbiotic gut bacteria. Flies raised on a high-starch expect environmentally induced restrictions on gene flow to medium had microbiota dominated by Lactobacillus plan- facilitate the evolution of postzygotic incompatibilities (both tarum, which was only a minor constituent of the microbiota environment-dependent and -independent selection against of flies raised on a standard cornmeal-molasses medium. hybrids) and genetic assimilation of behavioral reproductive Sharon et al. [109] found that the differences in bacterial barriers (habitat and mate choice), but also to lessen the composition can affect cuticular hydrocarbon levels, provid- potential for selective reinforcement of assortative mating ing a probable mechanism affecting mate choice. (just as adaptive plasticity lessens divergent selection on In Sockeye Salmon (Oncorhynchus nerka), postmating ecological phenotypes). isolation might be caused by sexual selection on diet-derived coloration. Anadromous sockeye sequester carotenoids from 6. Conclusions and Future Directions crustaceans consumed in the ocean and use them to express their brilliant red mating colors. The nonanadromous Opinions still seem to outnumber data about the impact morph (kokanee) expresses equally bright red mating color of environmental induction and plasticity on evolution, but on a diet with much less carotenoids, an example of substantial progress has been made in the last 20 years countergradient variation [110]. Anadromous morphs and [23, 64]. Plasticity appears to be a common if not universal hybrids raised in the freshwater habitat of the kokanee (low- feature of developmental systems and should not be ignored. carotenoid diet) underexpress the red pigment and probably Plasticity and environmental effects were once black boxes, suffer reduced mating success as a consequence [111, 112]. ignored by some, uncritically promoted as threats to evo- The actual effect of phenotypic plasticity on gene flow lutionary theory by others. But theoretical and empirical depends on environmental similarity between parents and investigations have increasingly shed light on how plasticity offspring. Environmentally induced mate discrimination evolves and interacts with natural selection. Whether devel- will have little or no hindrance on gene flow unless it opmental plasticity rivals mutation as a source of quantitative also affects the phenotypes and/or environments of the or qualitative change (the evolution of “novelty”) remains offspring. For example, imagine a phytophagous insect contentious [17, 80, 116]. However, to the extent that with environmentally induced contact pheromones causing speciation is defined by genetic divergence, genes will not be perfect assortative mating between individuals raised on the displaced from their central role in the study of speciation. same host plant. If there are no differences in host choice, Phenotypic plasticity can promote or constrain adaptive then the offspring of each mating type are equally likely evolution and ecological speciation. The effects of plasticity to grow up on each plant and therefore have no tendency in a particular case, and whether there is an overall trend, are to develop the same pheromone profile as their parents empirical questions. (Figure 5). In this hypothetical case, there is free gene flow Important theoretical challenges for understanding the despite assortative mating of phenotypes. importance of plasticity for adaptation and speciation 8 International Journal of Ecology

Host 1 Host 2 A. 1st generation: Larvae on different host plants develop environmentally induced mating cues represented by red and 100% AA 100% aa blue.

B. Random dispersal results in equal frequencies on each host plant. p q Association between genotype and p AA, q aa AA, aa mating phenotype persists.

C. 2nd generation: Larvae produced by assortative mating deviate from Hardy- p q p AA, q aa Weinberg, but develop environmentally AA, aa induced mating cues.

D. Random dispersal maintains equal frequencies. There is no association p[p AA, q aa] p[p AA, q aa] between genotype and mating + q[p AA, q aa] + q[p AA, q aa] phenotype

E. 3rd generation: Hardy-Weinberg p p2 pq q2 p p2 pq q2 genotype proportions are expected for [ AA, 2 Aa, aa] [ AA, 2 Aa, aa] larvae on each host despite assortative + q[p2 AA, 2pq Aa, q2 aa] + q[p2 AA, 2pq Aa, q2 aa] mating by environmentally induced = p2 AA, 2pq Aa, q2 aa = p2 AA, 2pq Aa, q2 aa phenotype.

Figure 5: Free gene flow despite assortative mating. If mating cues are environmentally induced but there is no habitat choice, host-associated populations will not be genetically differentiated, and it takes two generations to establish Hardy-Weinberg equilibrium. If the host densities are p and 1 − p = q, a locus that is fixed for different alleles on different hosts has allele frequencies p and q. At first, genotypes are perfectly associated with the host-induced mating cue, but random dispersal eliminates that association in one generation if there are no other factors maintaining covariance between parent and offspring phenotype (e.g., maternal effects or divergent selection). Once genotype frequencies are equalized between phenotypes, mating within phenotypes establishes Hardy-Weinberg genotype proportions [61]. Although imagining a locus with complete differentiation makes for the simplest illustration, the result is completely general for any allele frequency. If we somehow know what alleles have ancestors in each habitat in generation 0, the result is that it takes just two generations to completely randomize that ancestry, regardless of whether or not the alleles are actually different by state.

include the extension of models (such as [36]) to incorporate given to the relationship between plasticity and postzygotic nonadaptive plasticity and countergradient variation, further isolation. Key questions include how might plasticity affect investigation of how costs of plasticity affect genetic assim- ecologically based selection on hybrids, and are highly ilation and reproductive isolation, and careful examination plastic developmental pathways good or bad candidates for of what kinds of environmental effects on phenotype can involvement in postzygotic developmental incompatibilities? be considered equivalent to geographic restrictions of gene Addressing empirical challenges might best begin by flow (environmental effects on dispersal) [59, 93, 94, 114]. establishing criteria for recognizing plasticity as a causal fac- Not all environmental effects on mate choice will affect gene tor in speciation or adaptive radiation. West-Eberhard [15] flow (Figure 5). However, maternal effects and factors that and Pfennig and McGee [117] suggested that associations promote environmental similarity between relatives seem between intraspecific plasticity and species diversity support particularly likely to reduce gene flow and promote adaptive a role for plasticity in promoting adaptive radiation. Exam- divergence. Incorporation of such effects into the classic ples like the tiger salamander radiation (Figure 1) and oth- models of local adaptation [87] and ecological speciation ers [117], where an intraspecific polyphenism parallels a [31] might bring substantial clarification to the subject. repeated pattern of interspecific divergence, are consistent Moreover, how environmentally induced barriers to gene with adaptive plasticity as an origin of ecological divergence flow might affect the evolution of constitutive genetic bar- between species. More confirmatory evidence for a causal riers has not been explored. Should we expect environmental role of plasticity in adaptive divergence (the first component effects on assortative mating to slow the evolution of genetic of ecological speciation) would come from testing the barriers to gene flow? Finally, virtually no attention has been strengths of tradeoffs, costs of plasticity, and selection against International Journal of Ecology 9 immigrants and hybrids. Effects of plasticity on reproductive [7] U. Dieckmann and M. Heino, “Probabilistic maturation isolation (the second component of ecological speciation) reaction norms: their history, strengths, and limitations,” are illustrated in studies like those of parasitic indigobirds Marine Ecology-Progress Series, vol. 335, pp. 253–269, 2007. [103, 104]andDrosophila [107–109] on environmentally [8] M. G. Palacios, A. M. Sparkman, and A. M. Bronikowski, induced barriers to gene flow. We presently know little of “Developmental plasticity of immune defence in two life- the prevalence of this phenomenon or whether it is often history ecotypes of the garter snake, thamnophis elegans ff - a common-environment experiment,” Journal of Animal strong enough to facilitate substantial genetic di erentiation. Ecology, vol. 80, no. 2, pp. 431–437, 2011. Field studies are needed to document the prevalence and [9] J. M. Baldwin, “A new factor in evolution,” American strength, in nature, of costs of plasticity, evolvability of Naturalist, vol. 30, pp. 441–451, 1896. ff reaction norms, and environmental e ects on reproductive [10] C. H. Waddington, “Canalization of development and the isolation. The principle illustrated in Figure 5 should be inheritance of acquired characters,” Nature, vol. 150, no. considered in the design and interpretation of experiments. 3811, pp. 563–565, 1942. In order for plasticity of mating behaviour to directly impact [11] C. H. Waddington, “Genetic assimilation of acquired charac- gene flow, there must be some factor maintaining similarity ters,” Evolution, vol. 7, no. 4676, pp. 118–126, 1953. of environmental effects across generations. Finally, devel- [12] Schmalhausen II, Factors of Evolution, Blakiston, Philadel- opmental genetic studies are needed to assess whether genes phia, Pa, USA, 1949. causing postzygotic or prezygotic isolation are often involved [13] M. J. West-Eberhard, “Alternative adaptations, speciation, in highly plastic developmental pathways. and phylogeny,” Proceedings of the National Academy of The theoretical and empirical foundations of speciation Sciences of the United States of America,vol.83,no.5,pp. research are very strong in terms of genetics and geography 1388–1392, 1986. [29–31, 115]. The roles of ecology, environment, and [14] M. J. West-Eberhard, “Phenotypic plasticity and the origins of diversity,” Annual review of Ecology and Systematics, vol. development are prominent among the remaining frontiers 20, pp. 249–278, 1989. [27, 36, 52]. Further research integrating genetics, ecology, [15] M. J. West-Eberhard, Developmental Plasticity and Evolution, and development promises great gains in understanding the Oxford University Press, Oxford, UK, 2003. origins of biological diversity. [16] C. D. Schlichting, “The role of phenotypic plasticity in diversification,” in Phenotypic Plasticity: Functional and Con- ceptual Approaches, T. J. DeWitt and S. M. Scheiner, Eds., pp. Acknowledgments 191–200, Oxford University Press, 2004. The author thanks A. Hendry and two reviewers for their [17] M. J. West-Eberhard, “Developmental plasticity and the origin of species differences,” Proceedings of the National very helpful comments on an earlier draft of this paper. His Academy of Sciences of the United States of America, vol. 102, knowledge and ideas on the subject, and their presentation no. 1, pp. 6543–6549, 2005. herein, were greatly improved by discussions with J. Fordyce [18] G. F. Grether, “Environmental change, phenotypic plasticity, and X. Thibert-Plante. Research relevant to this paper and genetic compensation,” The American Naturalist, vol. has been supported by the United States National Science 166, no. 4, pp. E115–123, 2005. Foundation (DEB-0516475 and DEB-1011216). [19] T. D. Price, “Phenotypic plasticity, sexual selection and the evolution of colour patterns,” Journal of Experimental Biology, vol. 209, no. 12, pp. 2368–2376, 2006. References [20] E. Crispo, “The Baldwin effect and genetic assimilation: revisiting two mechanisms of evolutionary change mediated [1] S. Wright, “Evolution in Mendelian populations,” Genetics, by phenotypic plasticity,” Evolution, vol. 61, no. 11, pp. 2469– vol. 16, no. 2, pp. 97–159, 1931. 2479, 2007. [2] S. C. Stearns, “The role of development in the evolution of [21] E. Crispo, “Modifying effects of phenotypic plasticity on life histories,” in Evolution and Development,J.T.Bonner,Ed., interactions among natural selection, adaptation and gene pp. 237–258, Springer, Berlin, Germany, 1982. flow,” Journal of Evolutionary Biology, vol. 21, no. 6, pp. 1460– [3] D. A. Levin, “Plasticity, canalization and evolutionary stasis 1469, 2008. in plants,” in Plant Population Ecology,A.J.Davy,M.J. Hutchings, and A. R. Watkinson, Eds., pp. 35–45, Blackwell [22] C. K. Ghalambor, J. K. McKay, S. P. Carroll, and D. N. Scientific, 1988. Reznick, “Adaptive versus non-adaptive phenotypic plasticity [4] J. Clausen, “Principles for a joint attack on evolutionary and the potential for contemporary adaptation in new problems,” in Proceedings of the 6th International Congress of environments,” Functional Ecology, vol. 21, no. 3, pp. 394– Genetics, D. F. Jones, Ed., vol. 2, pp. 21–23, Brooklyn Botanic 407, 2007. Garden, Ithaca, NY, USA, 1932. [23] S. F. Gilbert and D. Epel, Ecological Developmental Biology: [5] R. K. Clements, J. M. Baskin, and C. C. Baskin, “The com- Integrating Epigenetics, Medicine, and Evolution, Sinauer parative biology of the two closely-related species Penstemon Associates, 2009. tenuiflorus Pennell and P. hirsutus (L.) Willd. (Scrophu- [24] G. Fusco and A. Minelli, “Phenotypic plasticity in devel- lariaceae, section Graciles): I. Taxonomy and geographical opment and evolution: facts and concepts,” Philosophical distribution,” Castanea, vol. 63, no. 2, pp. 138–153, 1998. Transactions of the Royal Society B, vol. 365, no. 1540, pp. [6] P. B. Reich, I. J. Wright, J. Cavender-Bares et al., “The 547–556, 2010. evolution of plant functional variation: traits, spectra, and [25] T. Schwander and O. Leimar, “Genes as leaders and followers strategies,” International Journal of Plant Sciences, vol. 164, in evolution,” Trends in Ecology and Evolution, vol. 26, no. 3, no. 3, pp. S143–S164, 2003. pp. 143–151, 2011. 10 International Journal of Ecology

[26] T. Uller and H. Helantera, “When are genes ‘leaders’ or [48] G. G. Simpson, Tempo and Mode in Evolution, Columbia ‘followers’ in evolution?” Trends in Ecology and Evolution, vol. University Press, New York, NY, USA, 1944. 26, no. 9, pp. 435–436, 2011. [49] S. J. Gould, The Structure of Evolutionary Theory,Harvard [27] D. W. Pfennig, M. A. Wund, E. C. Snell-Rood, T. Cruick- University Press, Cambridge, Mass, USA, 2002. shank, C. D. Schlichting, and A. P. Moczek, “Phenotypic [50] T. F. Stuessy, G. Jakubowsky, R. S. Gomez et al., “Anagenetic plasticity’s impacts on diversification and speciation,” Trends evolution in island plants,” Journal of Biogeography, vol. 33, in Ecology and Evolution, vol. 25, no. 8, pp. 459–467, 2010. no. 7, pp. 1259–1265, 2006. [28] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology [51] A. P. Hendry, “Ecological speciation! or the lack thereof?” Letters, vol. 8, no. 3, pp. 336–352, 2005. Canadian Journal of Fisheries and Aquatic Sciences, vol. 66, [29] M. Turelli, N. H. Barton, and J. A. Coyne, “Theory and no. 8, pp. 1383–1398, 2009. speciation,” Trends in Ecology and Evolution, vol. 16, no. 7, [52] D. Schluter, “Evidence for ecological speciation and its pp. 330–343, 2001. alternative,” Science, vol. 323, no. 5915, pp. 737–741, 2009. [30] J. A. Coyne and H. A. Orr, Speciation, Sinauer Associates, [53] E. Mayr, Systematics and the Origin of Species from the Sunderland, Mass, USA, 2004. Viewpoint of a Zoologist, Columbia University Press, New [31] S. Gavrilets, Fitness Landscapes and the Origin of Species, York, NY, USA, 1942. [54] R. Matsuda, “The evolutionary process in talitrid amphipods Princeton University Press, Princeton, NJ, USA, 2004. and salamanders in changing environments, with a discus- [32] T. D. Price, A. Qvarnstrom, and D. E. Irwin, “The role of phe- sion of “genetic assimilation” and some other evolutionary notypic plasticity in driving genetic evolution,” Proceedings of concepts,” Canadian Journal of Zoology, vol. 60, no. 5, pp. the Royal Society B, vol. 270, no. 1523, pp. 1433–1440, 2003. 733–749, 1982. [33] R. Lande, “Adaptation to an extraordinary environment by [55] H. B. Shaffer and S. R. Voss, “Phylogenetic and mechanistic evolution of phenotypic plasticity and genetic assimilation,” analysis of a developmentally integrated character complex: Journal of Evolutionary Biology, vol. 22, no. 7, pp. 1435–1446, alternate life history modes in ambystomatid salamanders,” 2009. American Zoologist, vol. 36, no. 1, pp. 24–35, 1996. [34] D. A. Levin, “Flowering-time plasticity facilitates niche shifts [56] S. R. Voss and H. B. Shaffer, “Adaptive evolution via a in adjacent populations,” New Phytologist, vol. 183, no. 3, pp. major gene effect: paedomorphosis in the Mexican axolotl,” 661–666, 2009. Proceedings of the National Academy of Sciences of the United [35] R. Svanback,¨ M. Pineda-Krch, and M. Doebeli, “Fluctuating States of America, vol. 94, no. 25, pp. 14185–14189, 1997. population dynamics promotes the evolution of phenotypic [57] H. B. Shaffer, “Evolution in a paedomorphic lineage. I. plasticity,” American Naturalist, vol. 174, no. 2, pp. 176–189, An electrophoretic analysis of the Mexican ambystomatid 2009. salamanders,” Evolution, vol. 38, pp. 1194–1206, 1984. [36] X. Thibert-Plante and A. P. Hendry, “The consequences of [58] H. B. Shaffer and M. L. Mcknight, “The polytypic species phenotypic plasticity for ecological speciation,” Journal of revisited: genetic differentiation and molecular phylogenetics Evolutionary Biology, vol. 24, no. 2, pp. 326–342, 2011. of the tiger salamander Ambystoma tigrinum (Amphibia: [37] M. D. Herron and M. Doebeli, “Adaptive diversification Caudata) complex,” Evolution, vol. 50, no. 1, pp. 417–433, of a plastic trait in a predictably fluctuating environment,” 1996. Journal of Theoretical Biology, vol. 285, pp. 58–68, 2011. [59] S. Gavrilets and A. Vose, “Case studies and mathematical [38] D. J. Funk, “Of ‘host forms’ and host races: terminological models of ecological speciation. 2. Palms on an oceanic issues in ecological speciation,” International Journal of island,” Molecular Ecology, vol. 16, no. 14, pp. 2910–2921, Ecology, vol. 2012, Article ID 506957, 8 pages, 2012. 2007. [60] V. Savolainen, M. C. Anstett, C. Lexer et al., “Sympatric [39] J. K. Conner and D. L. Hartl, A Primer of Ecological Genetics, speciation in palms on an oceanic island,” Nature, vol. 441, Sinauer Associates, Sunderland, Mass, USA, 2004. no. 7090, pp. 210–213, 2006. [40] D. J. Futuyma, Evolution, Sinauer Associates, Sunderland, [61] G. H. Hardy, “Mendelian proportions in a mixed popula- Mass, USA, 2nd edition, 2009. tion,” Science, vol. 28, no. 706, pp. 49–50, 1908. [41] D. O. Conover and E. T. Schultz, “Phenotypic similarity and [62] D. Schluter, “Ecology and the origin of species,” Trends in the evolutionary significance of countergradient variation,” Ecology and Evolution, vol. 16, no. 7, pp. 372–380, 2001. Trends in Ecology and Evolution, vol. 10, no. 6, pp. 248–252, [63] K. Ras¨ anen¨ and A. P. Hendry, “Disentangling interactions 1995. between adaptive divergence and gene flow when ecology [42] J. A. Fordyce, “The evolutionary consequences of ecological drives diversification,” Ecology Letters, vol. 11, no. 6, pp. 624– interactions mediated through phenotypic plasticity,” Journal 636, 2008. of Experimental Biology, vol. 209, no. 12, pp. 2377–2383, [64] T. J. DeWitt and S. M. Scheiner, “Phenotypic variation 2006. from single genotypes: a primer,” in Phenotypic Plasticity: [43] G. R. Price, “Selection and covariance,” Nature, vol. 227, no. Functional and Conceptual Approaches,T.J.DeWittandS.M. 5257, pp. 520–521, 1970. Scheiner, Eds., pp. 1–9, Oxford University Press, New York, [44] J. A. Endler, Natural Selection in the Wild, Princeton Univer- NY, USA, 2004. sity Press, Princeton, NJ, USA, 1986. [65] S. Via and R. Lande, “Genotype-environment interaction and [45] S. Via, “The evolution phenotypic plasticity: what do we the evolution of phenotypic plasticity,” Evolution, vol. 39, no. really know?” in Ecological Genetics,L.A.Real,Ed.,pp.35– 3, pp. 505–522, 1985. [66] L. Zhivotovsky, M. Feldman, and A. Bergman, “On the evo- 57, Princeton University Press, 1994. lution of phenotypic plasticity in a spatially heterogeneous [46] G. L. Bush, “Reply [to M. Claridge] from G. L. Bush,” Trends environment,” Evolution, vol. 50, no. 2, pp. 547–558, 1996. in Ecology & Evolution, vol. 10, no. 1, p. 38, 1995. [67] S. Sultan and H. G. Spencer, “Metapopulation structure [47] M. Claridge, “Species and speciation,” Trends in Ecology & favors plasticity over local adaptation,” American Naturalist, Evolution, vol. 10, no. 1, p. 38, 1995. vol. 160, no. 2, pp. 271–283, 2002. International Journal of Ecology 11

[68] J. Hollander, “Testing the grain-size model for the evolution [88] C. D. Schlichting and M. Pigliucci, Phenotypic Evolution: A of phenotypic plasticity,” Evolution, vol. 62, no. 6, pp. 1381– Reaction Norm Perspective, Sinauer Associates, 1998. 1389, 2008. [89] T. H. Bullock, “Compensation for temperature in the meta- [69] J. B. S. Haldane, “A mathematical theory of natural and bolism and activity of poikilotherms,” Biological Reviews, vol. artificial selection—part VI. Isolation,” Proceedings of the 30, pp. 311–342, 1955. Cambridge Philosophical Society, vol. 26, pp. 220–230, 1930. [90] R. Levins, “Thermal acclimation and heat resistance in [70] M. G. Bulmer, “Multiple niche polymorphism,” American Drosophila species,” American Naturalist, vol. 103, pp. 483– Naturalist, vol. 106, pp. 254–257, 1972. 499, 1969. [71] M. Slatkin, “Gene flow and the geographic structure of [91]K.A.Berven,D.E.Gill,andS.J.Smithgill,“Countergradient natural populations,” Science, vol. 236, no. 4803, pp. 787–792, selection in the green frog, rana clamitans,” Evolution, vol. 33, 1987. pp. 609–623, 1979. [72] W. R. Rice and E. E. Hostert, “Laboratory experiments on [92] M. Kirkpatrick and N. H. Barton, “Evolution of a species’ speciation: what have we learned in 40 years?” Evolution, vol. range,” American Naturalist, vol. 150, no. 1, pp. 1–23, 1997. 47, no. 6, pp. 1637–1653, 1993. [93] B. M. Fitzpatrick, J. A. Fordyce, and S. Gavrilets, “What, if [73] S. Gavrilets, “Models of speciation: what have we learned in anything, is sympatric speciation?” Journal of Evolutionary 40 years?” Evolution, vol. 57, no. 10, pp. 2197–2215, 2003. Biology, vol. 21, no. 6, pp. 1452–1459, 2008. [74] R. D. Holt, “On the evolutionary ecology of species’ ranges,” [94] B. M. Fitzpatrick, J. A. Fordyce, and S. Gavrilets, “Pattern, Evolutionary Ecology Research, vol. 5, no. 2, pp. 159–178, process and geographic modes of speciation,” Journal of 2003. Evolutionary Biology, vol. 22, no. 11, pp. 2342–2347, 2009. [75] D. Garant, S. E. Forde, and A. P. Hendry, “The multifarious [95] J. Antonovics, “Evolution in closely adjacent plant popula- effects of dispersal and gene flow on contemporary adaptations. X. Long-term persistence of prereproductive isolation tion,” Functional Ecology, vol. 21, no. 3, pp. 434–443, 2007. at a mine boundary,” Heredity, vol. 97, no. 1, pp. 33–37, 2006. [76] T. J. Kawecki, “Adaptation to marginal habitats,” Annual [96] M. C. Hall and J. H. Willis, “Divergent selection on flowering Review of Ecology, Evolution, and Systematics, vol. 39, pp. 321– time contributes to local adaptation in Mimulus guttatus 342, 2008. populations,” Evolution, vol. 60, no. 12, pp. 2466–2477, 2006. [77] S. Via, “Genetic constraints on the evolution of phenotypic [97] J. B. Beltman and P. Haccou, “Speciation through the plasticity,” in Genetic Constraints on Adaptive Evolution,V. learning of habitat features,” Theoretical Population Biology, Loeschcke, Ed., pp. 47–71, Springer, Boston, Mass, USA, vol. 67, no. 3, pp. 189–202, 2005. 1987. [98] J. Beltman and J. A. Metz, “Speciation: more likely through a [78] S. Via, R. Gomulkiewicz, G. de Jong, S. M. Scheiner, C. D. genetic or through a learned habitat preference?” Proceedings Schlichting, and P. H. Van Tienderen, “Adaptive phenotypic of the Royal Society B, vol. 272, no. 1571, pp. 1455–1463, 2005. plasticity: consensus and controversy,” Trends in Ecology and [99] J. M. Davis and J. A. Stamps, “The effect of natal experience Evolution, vol. 10, no. 5, pp. 212–217, 1995. on habitat preferences,” Trends in Ecology and Evolution, vol. [79] A.RomeroandS.M.Green,“Theendofregressiveevolution: 19, no. 8, pp. 411–416, 2004. examining and interpreting the evidence from cave fishes,” [100] A. P. Hendry, V. Castric, M. T. Kinnison, and T. P. Quinn, Journal of Fish Biology, vol. 67, no. 1, pp. 3–32, 2005. “The evolution of philopatry and dispersal: homing versus [80] M. Pigliucci, C. J. Murren, and C. D. Schlichting, “Phe- straying in salmonids,” in Evolution illuminated: Salmon and notypic plasticity and evolution by genetic assimilation,” their relatives, A. P. Hendry and S. C. Stearns, Eds., pp. 52–91, Journal of Experimental Biology, vol. 209, no. 12, pp. 2362– Oxford University Press, Oxford, UK, 2004. 2367, 2006. [101] N. Vallin and A. Qvarnstrom, “Learning the hard way: [81] R. B. Langerhans and T. J. DeWitt, “Plasticity constrained: imprinting can enhance enforced shifts in habitat choice,” over-generalized induction cues cause maladaptive pheno- International Journal of Ecology, vol. 2011, Article ID 287532, types,” Evolutionary Ecology Research, vol. 4, no. 6, pp. 857– 7 pages, 2011. 870, 2002. [102] E. A. Bernays and R. F. Chapman, Host-Plant Selection by [82] F. C. James, “Environmental component of morphological Phytophagous Insects, Chapman and Hall, New York, NY, differentiation in birds,” Science, vol. 221, no. 4606, pp. 184– USA, 1994. 186, 1983. [103] M. D. Sorenson, K. M. Sefc, and R. B. Payne, “Speciation by [83] T. Steinger, B. A. Roy, and M. L. Stanton, “Evolution host switch in brood parasitic indigobirds,” Nature, vol. 424, in stressful environments II: adaptive value and costs of no. 6951, pp. 928–931, 2003. plasticity in response to low light in Sinapis arvensis,” Journal [104] C. N. Balakrishnan, K. M. Sefc, and M. D. Sorenson, of Evolutionary Biology, vol. 16, no. 2, pp. 313–323, 2003. “Incomplete reproductive isolation following host shift in [84] P. H. Van Tienderen, “Evolution of generalists and specialists brood parasitic indigobirds,” Proceedings of the Royal Society in spatially heterogeneous environments,” Evolution, vol. 45, B, vol. 276, no. 1655, pp. 219–228, 2009. no. 6, pp. 1317–1331, 1991. [105] S. H. Berlocher and J. L. Feder, “Sympatric speciation in [85] T. J. DeWitt, A. Sih, and D. S. Wilson, “Costs and limits of phytophagous insects: moving beyond controversy?” Annual phenotypic plasticity,” Trends in Ecology and Evolution, vol. Review of Entomology, vol. 47, pp. 773–815, 2002. 13, no. 2, pp. 77–81, 1998. [106] H. Knuttel¨ and K. Fiedler, “Host-plant-derived variation in [86] T. B. Smith and S. Skulason, “Evolutionary significance of ultraviolet wing patterns influences mate selection by male resource polymorphisms in fishes, amphibians, and birds,” butterflies,” Journal of Experimental Biology, vol. 204, no. 14, Annual Review of Ecology and Systematics, vol. 27, pp. 111– pp. 2447–2459, 2001. 133, 1996. [107] M. D. Stennett and W. J. Etges, “Premating isolation is deter- [87] T. Lenormand, “Gene flow and the limits to natural selec- mined by larval rearing substrates in cactophilic Drosophila tion,” Trends in Ecology and Evolution, vol. 17, no. 4, pp. 183– mojavensis. III. Epicuticular hydrocarbon variation is deter- 189, 2002. mined by use of different host plants in Drosophila mojavensis 12 International Journal of Ecology

and Drosophila arizonae,” Journal of Chemical Ecology, vol. 23, no. 12, pp. 2803–2824, 1997. [108] W. J. Etges, C. L. Veenstra, and L. L. Jackson, “Premating isolation is determined by larval rearing substrates in cactophilic Drosophila mojavensis. VII. Eﬀects of larval dietary fatty acids on adult epicuticular hydrocarbons,” Journal of Chemical Ecology, vol. 32, pp. 2629–2646, 2006. [109] G. Sharon, D. Segal, J. M. Ringo, A. Hefetz, I. Zilber- Rosenberg, and E. Rosenberg, “Commensal bacteria play a role in mating preference of Drosophila melanogaster,” Proceedings of the National Academy of Sciences of the United States of America, vol. 107, no. 46, pp. 20051–20056, 2010. [110] J. K. Craig and C. J. Foote, “Countergradient variation and secondary sexual color: phenotypic convergence promotes genetic divergence in carotenoid use between sympatric anadromous and nonanadromous morphs of sockeye salmon (Onchorhynchus nerka),” Evolution, vol. 55, pp. 380– 391, 2001. [111] C. J. Foote, G. S. Brown, and C. W. Hawryshyn, “Female colour and male choice in sockeye salmon: implications for the phenotypic convergence of anadromous and nonanadromous morphs,” Animal Behaviour, vol. 67, no. 1, pp. 69–83, 2004. [112] J. K. Craig, C. J. Foote, and C. C. Wood, “Countergradient variation in carotenoid use between sympatric morphs of sockeye salmon (Oncorhynchus nerka) exposes nonanadromous hybrids in the wild by their mismatched spawning colour,” Biological Journal of the Linnean Society, vol. 84, no. 2, pp. 287–305, 2005. [113] D. L. Hartl and A. G. Clark, Principles of Population Genetics, Sinauer Associates, Sunderland, Mass, USA, 3rd edition, 1997. [114] M. Kirkpatrick and V. Ravigne,´ “Speciation by natural and sexual selection: models and experiments,” American Naturalist, vol. 159, pp. S22–S35, 2002. [115] D. I. Bolnick and B. M. Fitzpatrick, “Sympatric speciation: models and empirical evidence,” Annual Review of Ecology, Evolution, and Systematics, vol. 38, pp. 459–487, 2007. [116] G. de Jong, “Evolution of phenotypic plasticity: patterns of plasticity and the emergence of ecotypes,” New Phytologist, vol. 166, no. 1, pp. 101–117, 2005. [117] D. W. Pfennig and M. McGee, “Resource polyphenism increases species richness: a test of the hypothesis,” Philosoph- ical Transactions of the Royal Society B, vol. 365, no. 1540, pp. 577–591, 2010. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 456374, 15 pages doi:10.1155/2012/456374

Review Article Ecological Adaptation and Speciation: The Evolutionary Signiﬁcance of Habitat Avoidance as a Postzygotic Reproductive Barrier to Gene Flow

Jeffrey L. Feder,1 Scott P. Egan,2 and Andrew A. Forbes3

1 Department of Biological Sciences, University of Notre Dame, Notre Dame, IN 46556, USA 2 Department of Biological Sciences, Advanced Diagnostics and Therapeutics, Environmental Change Initiative, University of Notre Dame, Notre Dame, IN 46556, USA 3 Department of Biology, University of Iowa, 434A Biology Building, Iowa City, IA 52242, USA

Correspondence should be addressed to Jeﬀrey L. Feder, [email protected]

Received 14 August 2011; Accepted 16 November 2011

Academic Editor: Rui Faria

Copyright © 2012 Jeﬀrey L. Feder et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Habitat choice is an important component of most models of ecologically based speciation, especially when population divergence occurs in the face of gene ﬂow. We examine how organisms choose habitats and ask whether avoidance behavior plays an important role in habitat choice, focusing on host-speciﬁc phytophagous insects as model systems. We contend that when a component of habitat choice involves avoidance, there can be repercussions that can have consequences for enhancing the potential for specialization and postzygotic reproductive isolation and, hence, for ecological speciation. We discuss theoretical and empirical reasons for why avoidance behavior has not been fully recognized as a key element in habitat choice and ecological speciation. We present current evidence for habitat avoidance, emphasizing phytophagous insects, and new results for parasitoid wasps consistent with the avoidance hypothesis. We conclude by discussing avenues for further study, including other potential roles for avoidance behavior in speciation related to sexual selection and reinforcement.

1. Introduction barriers due to fitness trade-offs can form that, if strong enough, can promote speciation. Thelasttwodecadeshaveseenarenewalofinterestinthe Two aspects of ecological adaptation are particularly im- ecological context of speciation [1–5]. This interest has been portant for speciation-with-gene-flow. One concerns hab- spawned, in large part, by a greater acceptance that speciation itat-related performance traits and the other habitat choice can progress in the face of gene flow [6–10]. Ecological [13–17]. Habitat performance involves traits or genes that adaptation is a central component of many speciation- increase the survivorship of their progeny in one habitat but with-gene-flow models [8]. Specifically, divergent natural se- have negative fitness consequences in alternative habitats. lection imposed by different habitats or environments can Such tradeoffs have obvious consequences for reducing serve as an important “extrinsic” barrier to gene flow to effective gene flow between populations due to the reduced initiate speciation [11]. When fitness tradeoffs exist, popula- fitness of migrants and hybrids, as discussed previously. tions may not always be capable of simultaneously adapting Questions have been raised as to how common and strong phenotypically and genetically to the contrasting selection performance tradeoffs are between populations in nature pressures imposed by different habitats to be jack-off-all- [18–20]. However, instances of strong host plant-related trades generalists. Consequently, migrants moving between fitness tradeoffs for plant-eating insects have been found [8, habitats or hybrids formed by mating between individuals 21]. Moreover, it is clear that certain ecological dimensions from different populations will not thrive compared to of habitats will likely impose strong divergent selection resident parental types [12]. Thus, ecological reproductive pressures. For example, morphological aspects of resource 2 International Journal of Ecology processing such as the beaks of birds [22–24], the proboscis consequently an interest for both habitats. These hybrids length of insects [25], and the pharyngeal gills of fish [26] could therefore serve as effective bridges or conduits of gene will not be ideal for all food items and will incur a cost in flow between populations in different habitats and impede handling time to effectively use a subset of resources versus specialization. For this reason, Mayr [31, 32] argued that the others. Also, seasonal differences in resource availability genetics of habitat preference was a problem for ecological cannot always be bridged by a single life history strategy for speciation-with-gene-flow. The more preference genes that a short-lived organism that, for instance, has only a single affect habitat choice in the genome, the harder it would be for generation per year [27]. In addition, aspects of pollinator pure habitat choice for one parental habitat versus another to attraction [28] and crypsis in relation to the environment sort itself out in individuals of mixed ancestry. Thus, it would [29] can also have consequences for fitness. Thus, there be difficult for ecological specialization to drive speciation in are reasons to suspect that several ecological dimensions of the face of gene flow unless habitat choice was concentrated habitats can impose strong and divergent selection pressures in just a few loci of strong effect with consistent dominance on populations to most effectively utilize them. interactions for choice of a particular habitat. The second component of ecological adaptation con- Habitat choice may not be solely determined by positive cerns habitat choice. When habitat-related performance dif- preference, however. It has long been argued that phy- ferences exist, individuals that prefer to reside in the habitat tophagous insect specialists, for example, use positive cues to which they are best adapted benefit from a fitness advan- from hosts and negative cues from nonhost plants when tage [15]. When habitat choice affects mate choice, then evaluating their environment [33–35]. Consequently, as well the result is an ecologically based barrier to gene flow. This as possessing genes to prefer specific habitats, phytophagous most easily occurs when individuals mate in their preferred insects may also have loci or alleles at certain loci that cause habitats (e.g., if populations form leks or court in different them to avoid nonnatal, alternate habitats. habitats or at different times of the year). The resulting Theoretical models have been constructed that have assortative mating can facilitate ecological speciation-with- considered habitat avoidance as a factor in ecological spe- gene-flow because it accentuates selection for performance ciation but these have been restrictive in that they tend to traits that in turn increase selection for habitat fidelity [15– model avoidance additively [36, 37]. However, avoidance 17]. By mating in preferred habitats, individuals that possess (as well as preference) may commonly have nonadditive traits increasing their survivorship in a particular habitat will phenotypic and fitness effects with important consequences tend to mate with other individuals possessing the same suite for ecological speciation that go beyond the implications of of ecological adaptations, increasingly favoring those that habitat preference. For example, alleles resulting in increased remain in their “natal” habitats. Thus, positive correlations preference for one habitat may have pleiotropic conse- can evolve between traits affecting habitat choice (e.g., quences simultaneously increasing an individual’s disdain for host plant quality decision for oviposition by phytophagous other nonnatal habitats that do not possess preferred char- insects) and offspring survival in habitats [30]. acteristics [17]. Also, if avoidance genes exist, then hybrids Here, we are interested in how organisms choose habitats. possessing alleles to avoid alternate parental habitats may Specifically, we ask the question of whether avoidance be behaviorally conflicted and accept no habitat. Thus, behavior plays an important role in habitat choice, focusing hybrids could suffer a degree of postzygotic reproductive on host-specific phytophagous insects as model systems. We isolation, being incapable of finding suitable habitats to contend that when a component of habitat choice involves feed and mate in (i.e., hybrids would incur a degree of avoidance, there can be repercussions that can have conse- behavioral inviability/sterility). Moreover, in contrast to quences for enhancing the potential for specialization and preference genes, the more genes there are that strongly and postzygotic reproductive isolation and, hence, for ecological independently affect avoidance, the stronger the barrier to speciation. gene flow between populations, as it would become harder and harder to segregate out a parental behavioral phenotype that would be willing to reside in any one habitat [17]. 2. Habitat Preference versus Avoidance Thus, rather than the genetics of habitat choice providing a bridge, avoidance could create a greater reproductive chasm In general, when we think about how organisms decide to fostering ecological speciation-with-gene-flow. reside in one habitat versus alternatives, we think in terms of Avoidance could also have consequences for ecological positive preference. Specifically, preferences exist that result speciation not only occurring in the face of gene flow in in an individual deciding to live in one habitat that it likes the sympatry or parapatry but also occurring in allopatry in best over others. If we think about preferences being affected geographically isolated populations. If the evolution of in- by some number of genetic loci, then a greater proportion of creased preference has pleiotropic effects that increase avoid- alleles across these preference genes favoring the acceptance ance for alternative habitats, then the independent evolution of one habitat versus another would result in an organism of increased natal habitat choice in allopatry might cause being relatively more likely to reside in that habitat [16]. both pre- and postzygotic isolation if populations were ever Hybridization between populations with preferences for to come back together in secondary contact. In this case, differenthabitatswillresultinoffspring of mixed ancestry the evolution of increased preference in allopatry could that will, in the absence of genetic dominance, tend to stem, for example, from selective pressures for individuals have no strong preference for either parental habitat and to more efficiently and accurately judge variation in their International Journal of Ecology 3 natal habitat quality and avoid suboptimal natal resources with one another when brought together for the first time [38–40]. Alternatively, avoidance could also initially evolve in in the genome of an individual of mixed ancestry ([48] response to local nonnatal habitats that become significant for review). Genetic architectures of this form are known following secondary contact when these alternate habitats as Dobzhansky-Muller (D-M) incompatibilities and allow share characteristics in common with other, previously populations to evolve reproductive isolation without passage allopatric, nonnative habitats inhabited by a related taxon. through a fitness-valley that would be opposed by selection. The issue of why habitat avoidance is not a more generally Alleles for preference for the natal habitat that inadvertently appreciated element of ecological adaptation is a central cause avoidance to alternative, nonnatal habitats accumulate topic of this paper. We argue that part of the answer may independently in allopatric populations separated by geo- be theoretical and part due to experimental design. For graphic barriers that preclude migration and, consequently, avoidance to cause postzygotic reproductive isolation in sym- no negative fitness consequences in hybrids. If these physical patry or parapatry invokes the classic theoretical problem barriers happened to dissipate and populations were to come that a new mutation causing inviability or sterility cannot back into secondary contact and interbreed, then hybrids establish in the face of gene flow [41]; any new muta- would suffer behavioral inviability/sterility due to possessing tion reducing fitness in heterozygotes, including behavioral alleles causing them to avoid aspects of all suitable parental avoidance incompatibilities in hybrids, should be rapidly habitats. Thus, avoidance can easily be envisioned to play selected against and eliminated from populations. In this an important role in generating ecological reproductive paper, we propose several ways in which this problem can isolation in allopatry. be alleviated and that avoidance behavior can evolve in the We note that in allopatry, a similar argument can also face of gene flow despite negative fitness consequences in be made with respect to mate choice. Sexual selection and hybrids. This includes a discussion on the potential genetic assortative mating might involve both a preference for a given and physiological changes that can result in the evolution of trait(s) in mates and avoidance of individuals lacking the habitat avoidance, focusing on the olfactory system of host trait(s). Hence, avoidance aspects of mate choice can result plant specific phytophagous insects. in both pre- and postzygotic reproductive isolation following The second problem with avoidance is empirical. secondary contact in a similar manner as ecological habitat Although the hypothesis of habitat avoidance has been artic- choice. Moreover, the same considerations with respect to ulated, in part, by several leading entomologists including the number of loci underlying habitat avoidance would apply Bernays and Chapman [33], Jermy and Szentesi [42], and to mate avoidance; the greater the number of genes that Frey and Bush [43], the idea has overall gained only modest independently affect avoidance, the potentially stronger the experimental traction (see [34]). We suggest that this may reproductive barrier to gene flow. The role of avoidance stem from studies of habitat use generally not being designed behavior in sexual selection and assortative mating is an area or analyzed with the possibility of avoidance behavior in warranting further theoretical study. mind. Rather, most host choice experiments are focused on The theoretical difficulty with habitat avoidance there- preference and whether hybrid individuals will accept one fore rests primarily on whether similar behavioral inviabil- parental habitat (or host plant) versus the other. Evidence for ity/sterility barriers can also evolve in sympatry or parapatry avoidance can go undetected or unexamined in such tests. in the face of gene flow, as is possible in the absence of Here, we discuss several studies of insect host plant choice migration in allopatry. Here, the argument is that it is not in the literature and new results from a wasp parasitoid feasible because unlike the situation in allopatry, the negative Utetes lectoides that are consistent with the evolution of host consequences of a new mutation causing avoidance would be avoidance. We conclude by examining several aspects of host immediately exposed to selection against it in heterozygotes. plant choice studies that may mask the presence of avoidance However, there are reasons why this conventional wisdom behavior and suggest that avoidance may be a common may be incorrect. First, parapatric models have shown how component of ecological specialization and speciation, at new favored mutations that can cause genomic incompati- least for phytophagous insect specialists. bilities can alternately arise in the nonoverlapping portions of two taxon’s ranges experiencing low levels of migration and elevate to high frequencies (e.g., see [49]). When these 3. Theoretical Models for Habitat Avoidance favored mutations spread into the contact area, hybrids suffer reduced fitness and thus a reproductive barrier to The basic theoretical issue with habitat avoidance generating gene flow exists that fosters speciation. Consequently, with behavioral inviability/sterility concerns the following conun- certain parapatric distributions of taxa, behavioral habitat drum: how is it possible, if hybrids are unfit, to evolve inviability/sterility can evolve in a similar manner as in from the high-fitness genotype of one specialist population allopatry. But in these circumstances, avoidance behavior is (species) to the high-fitness genotype of the other population not initially evolving directly in the face of gene flow. (species) without passing through a low fitness intermediate? Second, Agrawal et al. [50] in this issue show that In allopatry, the answer is relatively straight-forward and when new mutations subject to divergent natural selection is attributed to the insights of Bateson, Dobzhansky, and also affect intrinsic isolation, either directly or via linkage Muller [44–47]. Specifically, hybrid inviability or sterility disequilibrium with other loci, such alleles can overcome often arises from between-locus genetic incompatibilities: gene flow, diverge between populations, and thus result in alleles that function well within species are incompatible the evolution of intrinsic isolation. Thus, if divergent 4 International Journal of Ecology

Table 1: Three models for how genetically based behavioral incompatibilities due to conﬂicting habitat avoidance can evolve during speciation-with-gene-ﬂow that cause postzygotic isolation in hybrids (see Figures 1–3).

Model 1: intermediate null allele (o) Step 1: a+ state in ancestral population for a habitat choice locus with a recessive null (o)allelealso segregating as low a frequency polymorphism that confers no habitat preference. Step 2: new habitat becomes available and o/o genotypes from ancestral population shift and adapt to novel habitat. Step 3: new avoidance mutation to ancestral habitat (a−) arises and elevates to high frequency in primarily o null allele initial genetic background of novel habitat population. Step 4: new avoidance mutation to novel habitat (n−) arises that confers even greater fidelity to ancestarl habitat than a+ allele and elevates to high frequency in the ancestral habitat population. Step 5: behavioral inviability/sterility in hybrids when individuals move between novel and ancestral habitat and cross mate producing n−/a− heterozygotes at habitat choice locus. Model 2: epistasis between habitat preference and avoidance loci Step 1: initial a+ state exists for ancestral habitat preference at habitat preference locus in ancestral population and initial n− state for novel habitat avoidance at habitat avoidance locus. Step 2: new habitat becomes available and new preference mutation for novel habitat (n+)arisesand elevates to high frequency in novel habitat population. Step 3: new avoidance mutation to ancestral habitat (a−) arises at avoidance locus and elevates to high frequency in novel habitat population because negative effects of n−/a− heterozygosity at avoidance locus are not expressed in n+/n+ genetic background at the preference locus in the novel habitat. Step 4: behavioral inviability/sterility in hybrids when individuals move between novel and ancestral habitat and cross mate producing n−/a− heterozygotes at habitat avoidance locus in n+/a+ genetic background of hybrids at the preference locus. Model 3: epistasis between ancestral and novel habitat loci with preference and avoidance alleles Step 1: initial a+ state exists for ancestral habitat preference at ancestral habitat choice locus in ancestral population and initial n− state for novel habitat avoidance at novel habitat locus. Step 2: new habitat becomes available and new preference mutation for novel habitat (n+)arisesatnovel habitat choice locus and elevates to high frequency in novel habitat population. Step 3: new avoidance mutation to ancestral habitat (a−) arises at ancestral habitat choice locus and elevates to high frequency in n+/n+ genetic background at the novel habitat locus in the novel habitat population. Step 4: behavioral inviability/sterility in hybrids when individuals move between novel and ancestral habitat and cross mate producing a+/a− heterozygotes at ancestral habitat locus in n+/n− genetic background of hybrids at the novel habitat locus. selection for nonnatal habitat avoidance is strong enough mutation conferring a fitness advantage in one habitat could relative to migration and to its deleterious consequences establish in the face of gene flow if it directly interacted in heterozygote “hybrids”, it can evolve in the face of gene negatively with other alleles in the prestanding genetic back- flow to differentiate sympatric and parapatric populations. ground to cause postzygotic behavioral inviability/sterility. Increased habitat fidelity could be favored to restrict the However, much like traditional D-M incompatibilities [41], formation of less fit hybrids, even if it involved a degree the evolution of avoidance behavior could involve a stepwise of increased behavioral inviability/sterility in individuals sequence of substitutions of new mutations that each resulted of mixed ancestry, provided that the increase in fitness in a minimum amount of reduced fitness in heterozygotes or due to pre-zygotic isolation outweighs the postzygotic cost. hybrids until completion of the process. As a consequence, Indeed, a difficulty with reinforcement models is that with sequential habitat choice models might widen the conditions decreased hybridization due to pre-zygotic isolation, the for divergent ecological selection to generate postzygotic selection pressure for the evolution of further reinforcement reproductive. We depict three such sequential habitat choice decreases [51]. However, if the differential establishment models in Table 1 that involve either intermediate alleles of avoidance alleles in divergent populations increases not and dominance interactions at a single locus or epistatic only pre-zygotic isolation but also the degree of behavioral interactions between loci to cross through fitness valleys inviability/sterility in hybrids, then selection pressures for for habitat avoidance to evolve and generate postzygotic reinforcement could remain strong even as gene flow levels isolation. decrease. The role of avoidance behavior in reinforcement is The first model (see Figure 1 and Table 1)involvesmul- an area in need of further theoretical analysis. tiple substitutions at a single locus affecting habitat choice in Third, the divergent selection models of Agrawal et al. which one of the allelic states (o) at the choice gene is a null [50] in this issue mainly focus on whether a new favorable mutation that does not affect habitat choice. This model is an International Journal of Ecology 5

Habitat 1 Habitat 2 (1) o a+ o a+ a+ a+ o a+ a+

o o (2) a+ a+ o o o a+ a+ o o o a+ a+ o o

a+ o

a+ a− (3) − a+ a+ a a− a+ o a− − a+ a o a− a+ o a− o o

n− a− (4) − − a n − − n n− a− a

− − n− a a − n− n− a a− a− − n a−

Hybridization? n− (5) n− a− a−

n− a−

F1 hybrid rejects habitat 1 F1 hybrid rejects habitat 2 − due to a− allele due to n allele

Figure 1: Illustration of model 1 based on intermediate null allele (see Table 1 for model description). Numbered steps of model (1–5) are separated by dashed lines and ordered vertically. Individuals are represented by chromosome pairs as grey bars; avoidance locus is marked by solid black tick mark on grey chromosome. Alleles at avoidance locus are labeled as follows: a+: preference for ancestral habitat; o: recessive null allele (i.e., no preference or avoidance); a−: avoidance for ancestral habitat; n−: avoidance for novel habitat. Arrow depicts shift from ancestral habitat 1 to novel habitat 2. 6 International Journal of Ecology extension of Nylin and Wahlberg’s [52] hypothesis that host conferring different avoidance or preference responses to shifts for phytophagous insects start with a broadening of the just those habitat specific cues. In the case of model 3, an host acceptance range (e.g., ovipositing females respond to a initial mutation at the novel habitat locus from n− to n+ wider range of plant cues). Following this initial generalist can result in a degree of orientation to the novel host (Step (oligophagous) phase, sympatric speciation can occur via 2). In an n+/n+ genetic background, a new a− avoidance host race formation or allopatric speciation can occur mutation to the ancestral host occurs at the ancestral habitat via the evolution of more specialized local subpopulations locus which establishes in the novel habitat (Step 3) and is [53, 54]. For example, initially most individuals attacking only behaviorally incompatible in hybrids possessing a− and an ancestral host plant may possess a+ alleles conferring n− alleles at the ancestral and novel habitat choice genes, ancestral host recognition. However, a recessive null (o) respectively (Step 4). allele resulting in no habitat preference may also initially Feder and Forbes [17] theoretically analyzed the second be present in the ancestral population as standing genetic epistasis model through a series of computer simulations in variation (see Step 1 for Model 1 in Figure 1 and Table 1). which levels of gene flow, divergent selection, habitat choice, In the absence of an alternative host, the null allele might recombination, and hybrid behavioral incompatibility were have little negative fitness consequences to its bearers and varied. The results implied that the conditions for habitat so exists as a low-frequency polymorphism maintained by avoidance were not overly restrictive under the epistasis sce- purifying selection/mutation balance. An alternative, novel nario and that the evolution of behavioral inviability/sterility habitat then becomes available and is colonized primarily by in the face of gene flow was possible. More detailed analyses individuals with the recessive (o)allele(Step2).Aftersome estimating the probabilities of establishment of new avoid- local (performance-related) adaptation to the novel habitat, ance alleles are still needed to definitively assess how likely a new dominant avoidance allele (a−) can arise and reach epistasis model two is for generating behavioral inviabil- high frequency in the novel population to avoid the ancestral ity/sterility with gene flow. The computer simulations were habitat where migrants would have lower fitness (Step 3). based on deterministic models showing that the frequencies Following this, if a new mutation for strongly avoiding of new avoidance alleles would increase due to selection but the novel habitat (n−) happened to arise in the ancestral did not assess how probable this increase would be relative population, then it could reach high frequency in the to stochastic loss. Moreover, analyses of the first and third ancestral population (Step 4). Rather than being primarily models are still required, although preliminary deterministic selected against in an a− genetic background in which the simulations suggest that they have similar characteristics as negative fitness consequences of conflicted host choice would model two (Feder, unpublished data). Consequently, while be immediately exposed in heterozygotes, the n− allele would the evolution of genomic incompatibilities is undoubtedly mainly be selected for in the ancestral population in an easier in allopatry than in sympatry or parapatry, there is no a+ genetic background. Thus, alternative a− and n− alleles strong theoretical impediment to habitat avoidance evolving could come to differentiate populations utilizing novel versus and causing a degree of hybrid behavioral inviability/sterility. ancestral habitats that adversely interact with each other to decrease hybrid fitness (Step 5). A second model (see Figure 2 and Table 1)involvestwo 4. Physiological Basis for Habitat Avoidance separate loci, one affecting habitat preference and the other habitat avoidance, that epistatically interact. In most genetic A key question with respect to the issue of habitat avoidance backgrounds, heterozygotes for alternate habitat avoidance is how organisms that are specialized for an ancestral habitat alleles to the novel and ancestral habitats (n−/a−) at the evolve the physiological mechanisms to not only prefer a habitat avoidance locus would have impaired behavior. novel habitat but also avoid their original ancestral habitat. However, in an n+/n+ homozygous genetic background for In this regard, it would seem most likely that such a the preference locus causing individuals to orient to the novel transition, if it evolved in the face of gene flow, would habitat (Step 2), the deleterious phenotypic effects of n−/a− initially involve a large and rapid shift in behavior due to heterozygosity at the avoidance locus are not expressed (Step a small number of important alleles or loci, rather than a 3), providing a pathway to alleviate the maladaptive fitness number of small, quantitative, stepwise changes. If habitat valley associated with habitat avoidance during a shift to a choice involved a large number of small effect genes, then novel habitat (Step 4). this would imply that a relatively high degree of migration Finally, a third model (see Figure 3 and Table 1) involves would be occurring between populations during critical early epistatic interactions as in model two, except rather than periods of speciation-with-gene-flow. In this case, unless separate preference and avoidance loci, habitat choice is all new mutations for choosing (or avoiding) the novel based on one locus that influences orientation to the habitat were dominant and/or habitat-related performance ancestral habitat and a second that dictates choice of the differences were pronounced and genomically wide-spread, novel habitat. The distinction between models 2 and 3 then it would seem difficult for a number of small effect genes isthatinmodel2agivenlocusaffects either avoidance or to establish. Moreover, it would make it theoretically more preference, but not both behaviors, to alternative habitats. difficult for behavioral inviability/sterility to evolve, as the In contrast, in model 3 the different loci are directly tied small advantage in habitat choice conferred by these genes to cues associated with one habitat or the other (be they could not be offset by any appreciable negative consequences olfactory, visual, tactile, or gustatory), with different alleles in hybrids or the new mutations would not establish. Thus, International Journal of Ecology 7

Habitat 1 Habitat 2

(1) Host avoidance locus

a+ n−

Host preference locus

− (2) a+ n n+ n−

a+ n− a+ n− n+ n− n+ n− a+ n− a+ n− a+ n− a+ n−

a+ n− n+ n−

(3) a+ n− n+ a−

a+ n− a+ n− n+ a− n+ n− a+ n− n+ a− a+ n− n+ n−

a+ n− n+ a−

Hybridization? a+ n− (4)

a+ n− n+ a− n+ a−

a+ n− n+ a−

F1 hybrid rejects habitat 1 due F1 hybrid rejects habitat 2 due − − to a−/n− eﬀect unmasked in to a /n eﬀect unmasked in a+/n+ background a+/n+ background

Figure 2: Illustration of model 2 based on epistasis between habitat preference and avoidance loci. Numbered steps of model (1–4) are separated by dashed lines and ordered vertically. Individuals are represented by chromosome pairs as grey bars; preference locus (on left) and avoidance locus (on right) are marked by two pairs of vertical black tick marks on grey chromosomes. Alleles at preference locus are labeled as follows: a+: preference for ancestral habitat; n+: preference for novel habitat. Alleles at avoidance locus are labeled as follows: n−: avoidance for novel habitat; a−: avoidance for ancestral habitat. Arrow depicts shift from ancestral habitat 1 to novel habitat 2. 8 International Journal of Ecology

Habitat 1 Habitat 2

(1) Novel habitat locus

a+ n−

Ancestral habitat locus

(2) a+ n− a+ n+

a+ n− a+ n− a+ n+ a+ n− a+ n− a+ n+ a+ n− a+ n+

a+ n− a+ n+

(3) a+ n− a− n+

a+ n− a+ n− a− n+ a− n+ a+ n− a− n+ a+ n− a− n+

a+ n− a− n+

Hybridization? a+ n− (4)

a+ n− a− n+ a− n+

+ − a n a− n+

F1 hybrid rejects habitat 1 due F1 hybrid rejects habitat 2 due to a+/a− eﬀect unmasked in to a+/a− eﬀect unmasked in n+/n− background n+/n− background

Figure 3: Illustration of model 3 based on epistasis between ancestral and novel habitat loci with preference and avoidance alleles. Numbered steps of model (1–4) are separated by dashed lines and ordered vertically. Individuals are represented by chromosome pairs as grey bars; ancestral (on left) and novel (on right) habitat loci are marked by two pairs of vertical black tick marks on grey chromosomes. Alleles at ancestral locus are labeled as follows: a+: preference for ancestral habitat; a−: avoidance for ancestral habitat. Alleles at novel locus are labeled as follows: n+: preference for novel habitat; n−: avoidance for novel habitat. Arrow depicts shift from ancestral habitat 1 to novel habitat 2. International Journal of Ecology 9

Table 2: Changes to olfactory system that can potentially affect the understanding the relationship between ecological adapta- behavioral orientation of insects to plants. tion and the genesis of biodiversity [60]. The peripheral olfactory system of insects consists of che- Genetic change Effect mosensory neurons present in specialized sensory hairs Determines nonpolar called sensilla (Supplementary Figure 1) ([71] for review). Odorant binding-protein compounds carried to In many insects, olfactory sensilla are found on two pairs conformation dendrite of chemoreceptor of olfactory organs on the head, the antennae and the Influences sensitivity of maxillary palps. Each olfactory sensillum is innervated by up Change in conformation of protein cell to particular to four olfactory receptor neurons (ORNs). Odor molecules at receptor site on chemoreceptor compound from the environment pass through pores in the cuticular Alter numbers of receptor sites for walls of the sensilla where they become bound to odor Alter sensitivity of cell to particular chemical on binding proteins (OBPs). The OBPs transport the odor particular compound chemoreceptor molecules to seven-transmembrane bound odorant receptor Alter second messenger or (OR) proteins that span the cell membranes of the dendrites Altersensitivityof+or membrane properties of + or − − cell overall of ORNs. Individual ORNs express only one or a few ORs of a chemoreceptor potential large superfamily of OR genes [72–75]. Drosophila, Switch effect of ff Alter gene expression pattern of for example, possess 60 di erent OR genes in their genomes stimulation from + to − receptor proteins on + and − cells [76]. Each OR binds to a unique class of molecule or or vice versa compound, which confers specific odor response properties Switch effect of Alter wiring of sense cells or to the firing of the ORN [73–75, 77]. Output from the stimulation from + to − interneurons peripheral olfactory system ORNs is sent to two antennal or vice versa lobes that contain a number of nerve cells organized into Alter levels of one or more Alter weighting of glomeruli. In most cases, the innervated axons of ORNs ff − neuromodulators at synapses di erent + and inputs expressing the same OR converge to a single glomerulus in Alter sign of synaptic inputs in path Alter weighting of each antennal lobe [72, 78]. Thus, the number of glomeruli to controlling center different + and − inputs is approximately equal to the number of OR genes an insect possesses. Here, the ORN axons synapse with second-order neurons that project to the higher brain centers in insects: the mushroom body and the lateral horn [79]. The glomeruli are it would seem that if speciation with-gene-flow is to occur, also the locations of local interneuron synapses, which enable initial changes in host choice may often involve large effect the flow of information between glomeruli and likely play alleles. Of course, this is not an important consideration roles in organizing the input signal [79, 80]. It is generally in allopatry, where habitat choice genes of small effect thought that the negative and positive neural inputs are could accumulate independently in geographically isolated processed in an additive manner in the central nervous demes and have large deleterious behavioral consequences system of insects [81], resulting in an insect behaviorally in hybrids following secondary contact and gene flow. responding to whatever the balance of olfactory input signals Regardless, the issue remains of how a new mutation could is at a particular moment in time [35]. induce a big (or even small) shift in behavior from a like to Switches from a positive to negative response to a dislike (or vice versa). chemical input signal could conceivably occur due to changes We illustrate possible mechanisms for generating large at any one of several points in the olfactory system. (Note behavioral shifts in habitat choice using the olfactory that there is a general similarity and deep evolutionary system of phytophagous insects as a model (see Fig- homology in the sensory systems between insects and other ure 1 in supplementary material available online at doi: multicellular organisms, implying that changes in the insect 10.1155/2012/456374.). We focus on olfaction in insects can have parallels in many other life forms.) We outline a because it is well-established that many phytophagous insect few potential target points in Table 2. In essence, the central specialists are highly sensitive to particular compounds [55– component of the argument is that if areas of the insect brain 57] or characteristic mixtures of volatiles [35, 58, 59]emitted are associated with positive orientation when innervated and from their host plants. This olfactory sensitivity is normally others with avoidance, then a shift in behavior can occur viewed as an adaptation for host finding, especially when through any mechanism that changes (switches) the input the herbivore displays directed, odor-based anemotaxis over long distances [35]. In addition to being attracted by host- signals to these areas. Thus, a change in the expression specific chemicals, specialists also tend to be deterred by pattern of OBPs or ORs genesinORNscanresultinan nonhost secondary metabolites [33, 34, 42]. This suggests odorant that was formally an attractant (agonist) of behavior a possible role for olfactory avoidance in host plant choice becoming a deterrent (antagonist) or vice versa. For example, and the potential for behavioral incompatibilities in hybrids. oviposition site preference is determined by a few loci in Moreover, there are more plant-associated insects (>500,000 the fruit fly, Drosophila sechellia, where an odorant-binding described species) than any other form of Metazoan life, protein gene is involved in the specialization of the fly to making the issue of host avoidance particularly relevant for the fruit of its host plant, Morinda citrifolia [85]. Similarly, 10 International Journal of Ecology a developmental change in the connections of ORN axons to Most interestingly, F1 hybrids between apple and haw- glomeruli or of the secondary order neurons to the higher thorn flies failed to orient to the odor of either apple or brain centers would also produce such a transformation. hawthorn fruit in flight tunnel assays, consistent with behav- Thus, there are several potential mechanisms in which a new ioral conflicts generated by fruit volatile avoidance [82]. In mutation could have a major effect on chemotaxis. addition, behavioral responses to parental fruit volatiles were An outstanding issue is still why a phytophagous insect’s restored in a fair proportion of F2 and backcross hybrids (35– olfactory system should not be more sophisticated and allow 60%), suggesting that only a few major genetic differences for fine-tuned and subtle host discrimination rather than underlie the phenotype [93]. This does not mean that a the general black and white patterns of likes and dislikes large number of genes do not affect olfaction and host that we cite previously. One answer, if our hypothesis is choice. Rather, it suggests that only that a few changes in a correct, may be that the level of complex decision making relatively small number of genes might generate behavioral in olfactory (and sense) systems is context dependent on the incompatibilities. Field tests still need to be conducted to neural capacity of organisms. Bernays [35] has argued that determine if the compromised chemosensory system of neurons are energetically expensive to operate and develop, nonresponding hybrids results in greatly reduced fitness in as well as to house when body size is an issue. Thus, there nature, although it is difficult to envision how these flies do may be constraints on how many nerve cells and how large not suffer at least some disadvantage. We also caution that and interconnected many phytophagous insect’s brain can the Rhagoletis results do not directly confirm that conflicting be. As a result, many phytophagous insects may be limited avoidance behavior is the cause for the lack of behavioral to more basic agonistic (likes) versus antagonistic (dislikes) response of F1 hybrid flies in flight tunnel assays. It is possible behavioral responses to host plant cues. Such a system can that the disrupted olfactory system of hybrids could be a also make sense in how most phytophagous insects probably pleiotropic consequence of developmental incompatibilities experience the world. It may not usually be the case that due to divergent selection on other host-related traits, such an insect will have to make a choice between two or more as differences in the timing of diapause that adapts apple different potential host plants simultaneously. Rather, it and hawthorn flies to variation in when their respective host may be more common in typical spatially heterogeneous plants fruit [27, 94, 95]. It also remains to be determined environments for an individual to experience a single host at what specific aspect of the Rhagoletis olfactory system a time and have to decide whether to accept or reject it. Thus, is impaired in hybrids. Initial studies suggested that the although we do not expect that all insect olfactory responses peripheral olfactory system was altered in hybrids, as a far are black/white in nature, such decision-making system, higher percentage of single cell ORNs responded to multiple while perhaps not always optimal, may be a serviceable classes of host plant volatiles in hybrids than in parental solution for host choice for many insects given the inherent apple and hawthorn flies [96]. This implied that these ORNs evolutionary constraints. were misexpressing multiple OR genes that could disrupt behavior. However, comparison between F2 and backcross flies that both responded and failed to respond to apple 5. Empirical Studies and hawthorn fruit volatiles in the flight tunnel showed no differences in ORN firing patterns to specific compounds; Our motivation for considering the role of habitat avoidance both behavioral responders and nonresponders displayed the in ecological speciation is due, in part, to empirical discov- same altered ORN patterns that F1 hybrids did [97]. Thus, eries in the apple maggot fly, Rhagoletis pomonella,amodel altered OR gene expression could be incidental or subtly system for sympatric speciation via host shifting [1, 86– contribute to loss of behavioral orientation in F1 hybrids but 88]. In particular, the recent shift of the fly from its native may not be the only or prime cause for the disruption. In and ancestral host hawthorn (Crataegus spp.) to introduced, a similar manner, altered ORN patterns do not distinguish domesticated apple around 150 years ago in the eastern U.S. whether habitat avoidance or a pleiotropic effect of a more is often cited as an example of incipient sympatric speciation general developmental incompatibility forms the basis for in action [8]. loss of behavior in hybrids. Additional work needs to be done Volatile compounds emitted from the surface of ripening to clarify these issues. fruit have been shown to be the most important long to intermediate range cues used by R. pomonella to locate host trees [89]. Once in the tree canopy, flies use both visual 6. Additional Evidence and Problems and olfactory cues at distances of ≤1 meter to pinpoint the location of fruit for mating and oviposition. Recently, we To determine how general habitat avoidance may be and how found that not only are the ancestral hawthorn and recently common disrupted host choice behavior may be in hybrids, formed apple-infesting host races of R. pomonella attracted we surveyed the plant-associated insect literature for poten- to volatile compounds emitted from their respective natal- tial examples. Although not exhaustive, we found 10 exam- host fruit [90], but they also tend to avoid the nonnatal ples consistent with avoidance behavior (Table 3). These volatiles of the alternate fruit [64, 65]. Because R. pomonella include insects across several orders, including Coleoptera, flies mate only on or near the fruit of their respective host Hymenoptera, Hemiptera, and Diptera. A Y-tube or alterna- plants [91, 92], fruit odor discrimination results directly in tive olfactometer apparatus is often the method used to assess differential mate choice. avoidance, although no-choice field and laboratory-based International Journal of Ecology 11

Table 3: Studies showing evidence consistent with insect avoidance of nonhost plants and/or their volatiles.

Electrophysio. Species Order; Family Assessment method(s) response nonhost Reference volatiles? Dryocosmus kuriphilus Hymenoptera: Cynipidae Y-tube olfactometer Yes [61] Sitophilus zeamais Coleoptera: Curculionidae Four-way olfactometer ? [62] Brevicoryne brassicae Hemiptera: Aphididae Y-tube olfactometer Yes [63] Aphis fabae Hemiptera: Aphididae Y-tube olfactometer Yes [63] Field test; no-choice Rhagoletis pomonella Diptera: Tephritidae Yes [64, 65] ﬂight tunnel Diachasma alloeum Hymenoptera: Braconidae Y-tube olfactometer Yes [66] Pityogenes chalcographus Coleoptera: Curculionidae Field test ? [67] Tomicus piniperda Coleoptera: Curculionidae Field test Yes [68] Hylurgops palliatus Coleoptera: Curculionidae Field test ? [68] Heliothrips haemorrhoidalis Thysanoptera: Thripidae unknown ? [70]

tests have been used, as well. In many cases, electroantennal ThechoicetrialsforRhagoletis also highlight a second readings were made in addition to behavioral tests to deter- problem in testing for avoidance behavior: the use of mine whether nonhost volatiles induced electrophysiological proportional host acceptance rather than absolute values to activity. In all cases, experiments were designed to explicitly assess discrimination. If one were to analyze the relative test for avoidance behavior. proportions of acceptance of natal versus nonnatal volatiles Adifficulty with assessing avoidance behavior stems as a comparative metric in host studies, it could appear from the nature of the experimental assays of behavior. A that host discrimination goes down in multiple choice particular problem is that host studies are often done in experiments versus no choice tests. The key is to estimate choice experiments in which insects are given plant material and compare absolute acceptance behaviors in no choice from alternative hosts in a confined area and the relative trials, however, as well as the time it takes for acceptance. proportions of time they choose to use the plants used as If not, evidence for host avoidance could be masked and a measure of preference. The difficulty here can be twofold. underestimated. For example, it may be that 50% of the First, as discussed previously, insects usually do not have to time a hybrid insect may choose one plant versus another directly choose between alternate hosts in nature. Rather, in a choice test, suggesting no preference or avoidance on a more often than not, host use comes down to acceptance relative scale. However, it may be that hybrids come to reside or rejection of a given plant. Hence, no choice experiments on plants and less often than their parents do. Thus, although are a better test for preference behavior, providing that the hybrids are not showing a preference difference, on an the subjects are not overly stressed by their physiological absolute scale they are making host acceptance decisions condition or circumstances to accept lower ranking hosts, less often than parental individuals, which could reflect the regardless of plant quality. This may be especially important existence of conflicting avoidance behavior. when testing for host avoidance. Having plants too close We also surveyed the literature for plant-associated together in an arena in a choice test can result in mixed insects to look for behavioral host choice dysfunction in sensory cues confounding insects with signals not often hybrids consistent with avoidance alleles reducing hybrid fit- experienced in the wild. For example, in Rhagoletis, the ness (Table 4). Here, the current evidence mainly comes from addition of nonnatal compounds to a natal fruit volatile Rhagoletis, a leaf-feeding beetle (Neochlamisus bebbianae) blend (apple or hawthorn) does not just result in behavioral and a hymenopteran parasitoid (Leptopilina boulardi).In indifference but often an active avoidance of apple and all cases, F1 hybrids express a reduced ability to orient or hawthorn flies to the mixed blend [64, 65]. In choice respond to parental host plants or host-plant volatiles. situations between natal versus nonnatal blends, overall capture rates of apple and hawthorn flies can fall and, of great 7. New Empirical Data importance, discrimination for a fly’s natal blend declines [98]. It is likely that in these instances the one meter distance As an example of how one might implement a test for between natal versus nonnatal blends in these trials was not habitat avoidance, we recently conducted a preliminary Y- sufficient to preclude volatile plume mixing. Flies therefore tube olfactory study testing for nonnatal host avoidance for interpreted both volatile blends as nonnatal and displayed a braconid parasitoid, Utetes lectoides, attacking the fruit fly decreased orientation to the natal versus nonnatal blend Rhagoletis zephyria in snowberry fruit in the western USA than in tests conducted using the natal blend versus a blank Domesticated apples were brought by settlers to the Pacific control. One must therefore be careful in interpreting the Northwest region of the USA in the last 200 years and results from choice studies. it is believed that R. pomonella was introduced via larval 12 International Journal of Ecology

Table 4: Studies presenting data consistent with habitat avoidance alleles causing problems in hybrids. Each cross-generated hybrids displaying reduced response relative to parents in habitat choice. Host plant or locality information is in parentheses.

Taxon Hybrid cross Reduced hybrid response Reference R. pomonella (apple) × R. pomonella Rhagoletis fruit flies Upwind flight to host odor [82] (haw) R. pomonella (apple) × R. nr. pomonella Rhagoletis fruit flies Upwind flight to host odor [82] (dogwood) R. pomonella (haw) × R. nr. pomonella Rhagoletis fruit flies Upwind flight to host odor [82] (dogwood) R. pomonella (apple) × R. mendax Rhagoletis fruit flies Electroantennogram to fruit odor [43] (blueberry) N. bebbianae (maple) × N. bebbianae Egan and Funk, Neochlamisus leaf beetles Time spent on foraging on parental host plants (willow) unpubl.; [83] L. boulardi (Nasrallah) × L. boulardi Leptopilina parasitoids Probing response to fruit odor 1 [84] (Brazzaville) 1 Only in one direction of the F1 hybrid cross (Nasrallah female × Brazzaville male).

infested apples within the last 50 years. We hypothesized avoidance behavior can have important implications for the that snowberry fly origin populations of U. lectoides in the evolution of ecologically based reproductive isolation that western USA might have recently evolved to avoid the volatile go beyond the pre-zygotic barriers resulting from preference odors emanating from introduced apple fruit. A glass Y-tube alone. Specifically, contrasting avoidance behaviors can cause with each arm connected to a filtered air source was set on host choice conflicts in hybrids, resulting in postzygotic a flat surface with an attractive light source at the end of the reproductive isolation. Similar thinking about behavioral tube where the arms intersect. All wasps were individually avoidance could also apply to sexual selection and rein- tested for behavioral orientation in the system by assessing forcement, widening the consequences of these processes for whether they turn into the left-hand arm, or the right-hand postzygotic isolation during speciation. arm, or whether they fail to reach the y-intersection. This We discussed the current theoretical and empirical establishes a baseline level of response due to the system evidence for habitat avoidance, focusing on phytophagous itself, in the absence of fruit volatiles. An odor (here, emitted insects as a model system. In general there is evidence from the surface of apple or snowberry fruit) can then be supporting habitat avoidance, but more work needs to be added to just one randomly selected arm of the tube, and done to verify that avoidance conflicts in hybrids directly wasps reintroduced to the system. Avoidance (or preference) cause F1 inviability and sterility in finding habitats and is measured by the change in response to the arm of the tube mates. In allopatry, there would seem to be no theoret- containing the odor. Preliminary results for U. lectoides were ical difficulty for selection on habitat choice to generate consistent with the nonnatal habitat avoidance hypothesis. behavioral, as well as developmental conflicts, in hybrids Only 1 of 12 wasps attacking snowberry wasps that were following secondary contact. Although more difficult in tested (8.3%) oriented to the arm of the Y-tube containing sympatry or parapatry, we outlined how it is also possible apple fruit volatiles versus a blank, odorless arm compared to evolve habitat avoidance that causes postzygotic isolation to the established 42% baseline control response when both in the face of gene flow. More theoretical work needs to arms were blank (χ2 = 4.9, P = 0.026, 1 df). Additional be done in this area to produce estimates of how probable testing with larger sample sizes is needed to confirm the pilot it is for new mutations having differing effects on avoid- study results and to better establish the extent to which U. ance and behavioral incompatibility to establish between lectoides orients to snowberry volatiles (4 of 9 = 44% of diverging populations. Experimental work also clearly shows wasps did in our initial study). Nevertheless, this example that insects use certain chemical cues from nonhosts to demonstrates avoidance and how one might easily test for avoid these plants. However, it must still be clearly shown such behavior. that reduced host choice response in hybrids is due to conflicting avoidance behaviors rather than to the pleiotropic consequences of developmental incompatibilities for other 8. Conclusions traits affecting sensory systems involved in habitat choice. In this paper, we have investigated the idea that avoidance Moreover, detailed neural physiology and genetic studies behavior may play a significant role in how organisms select are needed to determine and map how avoidance evolves the habitats they reside in (see also [99] for a complementary and how habitat choice is disrupted in hybrids. Analysis of review of habitat choice). We contend that in addition to habitat choice, and avoidance behavior in particular, is still positively orienting to certain critical cues in their natal in its early stages but is an intriguing area of theoretical and habitat, organisms also actively reject alternative habitats that empirical study linking ecology adaptation and speciation contain nonnatal elements. If true, then the evolution of with physiology, development, and genetics. Our current International Journal of Ecology 13 knowledge, while incomplete, suggests that there may be [15] S. R. Diehl and G. L. Bush, “The role of habitat preference in great ecological significance and evolutionary potential for adaptation and speciation,” in Speciation and Its Consequences, the often anthropomorphically ill-viewed trait of disdain. D. Otte and J. Endler, Eds., Sinauer Associates, Sunderland, Mass, USA, 1989. [16] J. D. Fry, “Multilocus models of sympatric speciation: bush Acknowledgments versus rice versus felsenstein,” Evolution,vol.57,no.8,pp. 1735–1746, 2003. This work was supported by grants from the NSF to [17] J. L. Feder and A. A. Forbes, “Habitat avoidance and speciation J. L. Feder and A. A. Forbes and the USDA to J. L. for phytophagous insect specialists,” Functional Ecology, vol. Feder, as well as support from the Notre Dame AD&T 21, no. 3, pp. 585–597, 2007. and ECI to S. P. Egan. The authors would like to thank [18] R. Butlin, “A new approach to sympatric speciation,” Trends in Dietmar Schwarz for supplying U. lectoides wasps for Y-tube Ecology and Evolution, vol. 2, no. 10, pp. 310–311, 1987. experiments, Stuart Baird for discussions and ideas about the [19] D. J. Futuyma and T. E. Philippi, “Genetic variation and co- potential role of avoidance behavior in sexual selection and variation in responses to host plants by Alsophila pometaria ( reinforcement, and Roger Butlin, Stewart Berlocher, and two Lepidoptera: Geometridae),” Evolution, vol. 41, no. 2, pp. 269– anonymous reviewers for helpful comments on improving 279, 1987. the manuscript. [20] J. D. Fry, “The evolution of host specialization: are trade-offs overrated?” American Naturalist, vol. 148, pp. S84–S107, 1996. [21] J. Scheirs, K. Jordaens, and L. De Bruyn, “Have genetic trade- References offs in host use been overlooked in arthropods?” Evolutionary Ecology, vol. 19, no. 6, pp. 551–561, 2005. [1] J. L. Feder, C. A. Chilcote, and G. L. Bush, “Genetic differentiation between sympatric host races of the apple maggot fly [22] P. R. Grant and B. R. Grant, “Unpredictable evolution in a 30- Rhagoletis pomonella,” Nature, vol. 336, no. 6194, pp. 61–64, year study of Darwin’s finches,” Science, vol. 296, no. 5568, pp. 1988. 707–711, 2002. [23] T. L. Parchman and C. W. Benkman, “Diversifying coevolution [2] D. Schluter, “Ecological causes of speciation,” in Endless Forms: between crossbills and black spruce on Newfoundland,” Species and Speciation, D. J. Howard and S. H. Berlocher, Eds., Evolution, vol. 56, no. 8, pp. 1663–1672, 2002. pp. 114–129, Oxford University Press, Oxford, UK, 1998. [24] C. W. Benkman, “Divergent selection drives the adaptive [3] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology radiation of crossbills,” Evolution, vol. 57, no. 5, pp. 1176– Letters, vol. 8, no. 3, pp. 336–352, 2005. 1181, 2003. [4] D. J. Funk, P. Nosil, and W. J. Etges, “Ecological divergence [25] S. P.Carroll and C. Boyd, “Host race radiation in the soapberry exhibits consistently positive associations with reproductive bug: natural history with the history,” Evolution, vol. 46, no. 4, isolation across disparate taxa,” Proceedings of the National pp. 1052–1069, 1992. Academy of Sciences of the United States of America, vol. 103, [26] J. J. Willacker, F. A. Von Hippel, P. R. Wilton, and K. M. no. 9, pp. 3209–3213, 2006. Walton, “Classification of threespine stickleback along the [5]A.P.Hendry,P.Nosil,andL.H.Rieseberg,“Thespeedof benthic-limnetic axis,” Biological Journal of the Linnean Soci- ecological speciation,” Functional Ecology,vol.21,no.3,pp. ety, vol. 101, no. 3, pp. 595–608, 2010. 455–464, 2007. [27] J. L. Feder and K. E. Filchak, “It’s about time: the evidence [6] G. L. Bush, “Sympatric speciation in animals: new wine in old for host plant-mediated selection in the apple maggot fly, bottles,” Trends in Ecology and Evolution, vol. 9, no. 8, pp. 285– Rhagoletis pomonella, and its implications for fitness trade- 288, 1994. offs in phytophagous insects,” Entomologia Experimentalis et [7] S. Via, “Sympatric speciation in animals: the ugly duckling Applicata, vol. 91, no. 1, pp. 211–225, 1999. grows up,” Trends in Ecology and Evolution,vol.16,no.7,pp. [28]J.Ramsey,H.D.Bradshaw,andD.W.Schemske,“Compo- 381–390, 2001. nents of reproductive isolation between the monkeyflowers [8]S.H.BerlocherandJ.L.Feder,“Sympatricspeciationinphy- Mimulus lewisii and M. cardinalis (Phrymaceae),” Evolution, tophagous insects: moving beyond controversy?” Annual Re- vol. 57, no. 7, pp. 1520–1534, 2003. view of Entomology, vol. 47, pp. 773–815, 2002. [29] P. Nosil and B. J. Crespi, “Experimental evidence that preda- [9] M. Dres` and J. Mallet, “Host races in plant-feeding insects tion promotes divergence in adaptive radiation,” Proceedings and their importance in sympatric speciation,” Philosophical of the National Academy of Sciences of the United States of Transactions of the Royal Society B, vol. 357, no. 1420, pp. 471– America, vol. 103, no. 24, pp. 9090–9095, 2006. 492, 2002. [30] S. Gripenberg, P. J. Mayhew, M. Parnell, and T. Roslin, [10] D. I. Bolnick and B. M. Fitzpatrick, “Sympatric speciation: the- “A meta-analysis of preference-performance relationships in ory and empirical data,” Annual Review of Ecology, Evolution, phytophagous insects,” Ecology Letters, vol. 13, no. 3, pp. 383– and Systematics, vol. 38, pp. 459–487, 2007. 393, 2010. [11] D. Schluter, “Evidence for ecological speciation and its alter- [31] E. Mayr, “Ecological factors in speciation,” Evolution, vol. 1, native,” Science, vol. 323, no. 5915, pp. 737–741, 2009. pp. 263–288, 1947. [12] P. Nosil, “Divergent host plant adaptation and reproductive [32] E. Mayr, Animal Species and Evolution,HarvardUniversity isolation between ecotypes of Timema cristinae walking Press, Harvard, Mass, USA, 1963. sticks,” American Naturalist, vol. 169, no. 2, pp. 151–162, 2007. [33] E. A. Bernays and R. F. Chapman, “The evolution of deterrent [13] J. Maynard Smith, “Sympatric speciation,” American Natural- responses in plant feeding insects,” in Perspectives in Chemore- ist, vol. 100, pp. 637–650, 1966. ceptors and Behavior,R.F.Chapman,E.A.Bernays,andJ.G. [14] J. Felsenstein, “Homage to Santa Rosalia, or why are there so Stoffolano, Eds., pp. 159–173, Springer, New York, NY, USA, few kinds of animals?” Evolution, vol. 35, pp. 124–138, 1981. 1987. 14 International Journal of Ecology

[34] E. A. Bernays, S. Oppenheim, R. F. Chapman, H. Kwon, and [53] N. Janz, S. Nylin, and N. Wahlberg, “Diversity begets diversity: F. Gould, “Taste sensitivity of insect herbivores to deterrents host expansions and the diversification of plant-feeding is greater in specialists than in generalists: a behavioral test of insects,” BMC Evolutionary Biology, vol. 6, article 4, 2006. the hypothesis with two closely related caterpillars,” Journal of [54] E. Weingartner, N. Wahlberg, and S. Nylin, “Dynamics of Chemical Ecology, vol. 26, no. 2, pp. 547–563, 2000. host plant use and species diversity in Polygonia butterflies [35] E. A. Bernays, “Neural limitations in phytophagous insects: (Nymphalidae),” Journal of Evolutionary Biology, vol. 19, no. implications for diet breadth and evolution of host affiliation,” 2, pp. 483–491, 2006. Annual Review of Entomology, vol. 46, pp. 703–727, 2001. [55] P. M. Guerin and E. Stadler,¨ “Host odour perception in three [36]C.Castillo-Chavez,S.A.Levin,andF.Gould,“Physiological phytophagous Diptera: a comparative study,” in Proceedings of and behavioral adaptation to varying environments: a mathe- the 5th International Symposium on Insect-Plant Relationships, matical model,” Evolution, vol. 42, no. 5, pp. 986–994, 1988. pp. 95–106, Pudoc, Wageningen, The Netherlands, 1982. [37] M. D. Rausher, “The evolution of habitat preference: avoid- [56] E. Thibout, J. Auger, and C. Lecomte, “Host plant chemicals ance and adaptation,” in Evolution of Insect Pests: Patterns of responsible for attraction and oviposition in Acrolepiopsis Variation, K. C. Kim and B. A. McPheron, Eds., pp. 259–283, assectella,” in Proceedings of the 5th International Symposium on Wiley, New York, NY, USA, 1993. Insect-Plant Relationships, pp. 107–116, Pudoc, Wageningen, [38] N. Janz and S. Nylin, “The role of female search behaviour in The Netherlands, 1982. determining host plant range in plant feeding insects: a test [57] E. Bartlet, M. M. Blight, P. Lane, and I. H. Williams, “The of the information processing hypothesis,” Proceedings of the responses of the cabbage seed weevil Ceutorhynchus assimilis Royal Society B, vol. 264, no. 1382, pp. 701–707, 1997. to volatile compounds from oilseed rape in a linear track [39] S. P. Egan and D. J. Funk, “Individual advantages to ecological olfactometer,” Entomologia Experimentalis et Applicata, vol. 85, specialization: insights on cognitive constraints from three no. 3, pp. 257–262, 1997. conspecific taxa,” Proceedings of the Royal Society B, vol. 273, [58] J. H. Visser, “Host odor perception in phytophagous insects,” no. 1588, pp. 843–848, 2006. Annual Review of Entomology, vol. 31, pp. 121–144, 1986. [40] H. D. Rundle and J. W. Boughman, “Behavioral ecology and [59] C. R. Roseland, M. B. Bates, R. B. Carlson, and C. Y. Oseto, speciation,” in Evolutionary Behavioral Ecology, D. F. Westneat “Discrimination of sunflower volatiles by the red sunflower andC.W.Fox,Eds.,OxfordUniversityPress,Oxford,UK, seed weevil,” Entomologia Experimentalis et Applicata, vol. 62, 2010. no. 2, pp. 99–106, 1992. [41] J. A. Coyne and H. A. Orr, Speciation, Sinauer Associates, Inc., [60] V.Novotny, P.Drozd, S. E. Miller et al., “Why are there so many Sunderland, Mass, USA, 2004. species of herbivorous insects in tropical rainforests?” Science, [42] T. Jermy and A. Szentesi, “The role of inhibitory stimuli in vol. 313, no. 5790, pp. 1115–1118, 2006. the choice of oviposition site by phytophagous insects,” En- [61] G. S. Germinara, A. De Cristofaro, and G. Rotundo, “Chemical tomologia Experimentalis et Applicata, vol. 24, no. 3, pp. 458– cues for host location by the chestnut gall wasp, Dryocosmus 471, 1978. kuriphilus,” Journal of Chemical Ecology, vol. 37, no. 1, pp. 49– [43] J. E. Frey and G. L. Bush, “Impaired host odor perception in 56, 2011. hybrids between the sibling species Rhagoletis pomonella and [62] D. A. Ukeh, M. A. Birkett, T. J. A. Bruce, E. J. Allan, J. A. R. mendax,” Entomologia Experimentalis et Applicata, vol. 80, Pickett, and A. Jennifer Mordue, “Behavioural responses of no. 1, pp. 163–165, 1996. the maize weevil, Sitophilus zeamais, to host (stored-grain) and [44] W. Bateson, “Heredity and variation in modern lights,” in non-host plant volatiles,” Pest Management Science, vol. 66, no. Darwin and Modern Science, A. C. Seward, Ed., pp. 85–101, 1, pp. 44–50, 2010. Cambridge, UK, Cambridge University Press, 1909. [63] S. F. Nottingham, J. Hardie, G. W. Dawson et al., “Behavioral [45] T. Dobzhansky, Genetics and the Origin of Species, Columbia and electrophysiological responses of Aphids to host and University Press, New York, NY, USA, 1st edition, 1937. nonhost plant volatiles,” Journal of Chemical Ecology, vol. 17, [46] H. J. Muller, “Bearing of the Drosophila work on systematics,” no. 6, pp. 1231–1242, 1991. in The New Systematics, J. Huxley, Ed., pp. 185–268, Oxford, [64] A. A. Forbes, J. Fisher, and J. L. Feder, “Habitat avoidance: UK, Oxford University Press, 1940. overlooking an important aspect of host-specific mating and [47] H. J. Muller, “Isolating mechanisms, evolution, and temera- sympatric speciation?” Evolution, vol. 59, no. 7, pp. 1552– ture,” Biology Symposium, vol. 6, pp. 71–125, 1942. 1559, 2005. [48] S. Gavrilets, Fitness Landscapes and the Origin of Species, [65] C. Linn, S. Nojima, and W. Roelofs, “Antagonist effects of non- Princeton University Press, Princeton, NJ, USA, 2004. host fruit volatiles on discrimination of host fruit by Rhagoletis [49] A. S. Kondrashov, “Accumulation of Dobzhansky-Muller flies infesting apple (Malus pumila), hawthorn (Crataegus incompatibilities within a spatially structured population,” spp.), and flowering dogwood (Cornus florida),” Entomologia Evolution, vol. 57, no. 1, pp. 151–153, 2003. Experimentalis et Applicata, vol. 114, no. 2, pp. 97–105, 2005. [50] A. F. Agrawal, J. L. Feder, and P. Nosil, “Ecological divergence [66] A. A. Forbes, T. H. Q. Powell, L. L. Stelinski, J. J. Smith, and J. L. and the origins of intrinsic postmating isolation with gene Feder, “Sequential sympatric speciation across trophic levels,” flow,” International Journal of Ecology, vol. 2011, Article ID Science, vol. 323, no. 5915, pp. 776–779, 2009. 435357, 15 pages, 2011. [67] J. A. Byers, Q.-H. Zhang, F. Schlyter, and G. Birgersson, “Vol- [51] A. R. Templeton, “Mechanisms of speciation—a population atiles from nonhost birch trees inhibit pheromone response in genetic approach,” Annual Review of Ecology and Systematics, spruce bark beetles,” Naturwissenschaften, vol. 85, no. 11, pp. vol. 12, pp. 32–48, 1981. 557–561, 1998. [52] S. Nylin and N. Wahlberg, “Does plasticity drive speciation? [68] L. M. Schroeder, “Olfactory recognition of nonhosts aspen and Host-plant shifts and diversification in nymphaline butterflies birch by conifer bark beetles Tomicus piniperda and Hylurgops (Lepidoptera: Nymphalidae) during the tertiary,” Biological palliatus,” Journal of Chemical Ecology, vol. 18, no. 9, pp. 1583– Journal of the Linnean Society, vol. 94, no. 1, pp. 115–130, 2008. 1593, 1992. International Journal of Ecology 15

[69] F. Schlyter, Q.-H. Zhang, P. Anderson et al., “Electrophysio- perception and host-plant preference in Drosophila sechellia,” logical and behavioural responses of Tomicus piniperda and PLoS Biology, vol. 5, no. 5, article e118, 2007. Tomicus minor (Coleoptera: Scolytidae) to non-host leaf and [86] G. L. Bush, “The taxonomy, cytology, and evolution of the bark volatiles,” Canadian Entomologist, vol. 132, no. 6, pp. genus Rhagoletis in North America (Diptera, Tephritidae),,” 965–981, 2000. Bulletin of the Museum of Comparative Zoology, vol. 134, pp. [70] A. S. Scott Brown, Interactions of thrips and their control agents 431–562, 1966. on host plants within a glasshouse containing a diverse collection [87] G. L. Bush, “Sympatric host race formation and speciation of plant species, PhD dissertation, Birkbeck College, University in frugivorous flies of the genus Rhagoletis (Diptera; Tephri- of London, 2002. tidae),” Evolution, vol. 23, pp. 237–251, 1969. [71] A. Dahanukar, E. A. Hallem, and J. R. Carlson, “Insect [88]B.A.McPheron,D.C.Smith,andS.H.Berlocher,“Genetic chemoreception,” Current Opinion in Neurobiology, vol. 15, differences between host races of Rhagoletis pomonella,” no. 4, pp. 423–430, 2005. Nature, vol. 336, no. 6194, pp. 64–66, 1988. [72] L. B. Vosshall, A. M. Wong, and R. Axel, “An olfactory sensory [89] R. J. Prokopy and B. D. Roitberg, “Foraging behavior of true map in the fly brain,” Cell, vol. 102, no. 2, pp. 147–159, 2000. fruit flies,” American Scientist, vol. 72, pp. 41–49, 1984. [73] A. A. Dobritsa, W. Van Der Goes Van Naters, C. G. Warr, R. [90] C. Linn, J. L. Feder, S. Nojima, H. R. Dambroski, S. H. Ber- A. Steinbrecht, and J. R. Carlson, “Integrating the molecular locher, and W. Roelofs, “Fruit odor discrimination and sym- and cellular basis of odor coding in the Drosophila antenna,” patric host race formation in Rhagoletis,” Proceedings of the Neuron, vol. 37, no. 5, pp. 827–841, 2003. National Academy of Sciences of the United States of America, [74] T. Elmore, R. Ignell, J. R. Carlson, and D. P. Smith, “Targeted vol. 100, no. 20, pp. 11490–11493, 2003. mutation of a Drosophila odor receptor defines receptor [91] R. J. Prokopy, E. W. Bennett, and G. L. Bush, “Mating behavior requirement in a novel class of sensillum,” Journal of Neuro- in Rhagoletis pomonella (Diptera: Tephritidae)—I. Site of science, vol. 23, no. 30, pp. 9906–9912, 2003. assembly,” Canadian Entomologist, vol. 103, pp. 1405–1409, [75] A. L. Goldman, W. Van Der Goes Van Naters, D. Lessing, C. G. 1971. Warr, and J. R. Carlson, “Coexpression of two functional odor [92] R. J. Prokopy, E. W. Bennett, and G. L. Bush, “Mating behavior receptors in one neuron,” Neuron, vol. 45, no. 5, pp. 661–666, in Rhagoletis pomonella (Diptera: Tephritidae)—II. Temporal 2005. organization,” Canadian Entomologist, vol. 104, pp. 97–104, [76] H. M. Robertson, C. G. Warr, and J. R. Carlson, “Molecular 1972. evolution of the insect chemoreceptor gene superfamily in [93] H. R. Dambroski, C. Linn Jr., S. H. Berlocher, A. A. Forbes, Drosophila melanogaster,” Proceedings of the National Academy W. Roelofs, and J. L. Feder, “The genetic basis for fruit of Sciences of the United States of America, vol. 100, no. 24, pp. odor discrimination in Rhagoletis flies and its significance for 14537–14542, 2003. sympatric host shifts,” Evolution, vol. 59, no. 9, pp. 1953–1964, [77] E. A. Hallem, M. G. Ho, and J. R. Carlson, “The molecular 2005. basis of odor coding in the Drosophila antenna,” Cell, vol. 117, [94] J. L. Feder, T. A. Hunt, and L. Bush, “The effects of climate, no. 7, pp. 965–979, 2004. host plant phenology and host fidelity on the genetics of [78] Q. Gao, B. Yuan, and A. Chess, “Convergent projections of apple and hawthorn infesting races of Rhagoletis pomonella,” Drosophila olfactory neurons to specific glomeruli in the Entomologia Experimentalis et Applicata,vol.69,no.2,pp. antennal lobe,” Nature Neuroscience, vol. 3, no. 8, pp. 780–785, 117–135, 1993. 2000. [95] K. E. Filchak, J. B. Roethele, and J. L. Feder, “Natural selection [79] R. F. Stocker, “The organization of the chemosensory system and sympatric divergence in the apple maggot Rhagoletis in Drosophila melanogaster:areview,”Cell and Tissue Research, pomonella,” Nature, vol. 407, no. 6805, pp. 739–742, 2000. vol. 275, no. 1, pp. 3–26, 1994. [96] S. B. Olsson, C. E. Linn, A. Michel et al., “Receptor expression [80] R. F. Stocker, M. C. Lienhard, A. Borst, and K. F. Fischbach, and sympatric speciation: unique olfactory receptor neuron “Neuronal architecture of the antennal lobe in Drosophila responses in F1 hybrid Rhagoletis populations,” Journal of melanogaster,” Cell and Tissue Research, vol. 262, no. 1, pp. 9– Experimental Biology, vol. 209, no. 19, pp. 3729–3741, 2006. 34, 1990. [97] S. B. Olsson, C. E. Linn, J. L. Feder et al., “Comparing pe- [81] L. M. Schoonhoven, T. Jermy, and J. A. A. Van Loon, Insect- ripheral olfactory coding with host preference in the Rhagoletis Plant Biology, Chapman and Hall, London, UK, 1998. species complex,” Chemical Senses, vol. 34, no. 1, pp. 37–48, [82] C. E. Linn, H. R. Dambroski, J. L. Feder, S. H. Berlocher, 2009. S. Nojima, and W. L. Roelofs, “Postzygotic isolating factor [98] A. A. Forbes and J. L. Feder, “Divergent preferences of Rha- in sympatric speciation in Rhagoletis flies: reduced response goletis pomonella host races for olfactory and visual fruit cues,” of hybrids to parental host-fruit odors,” Proceedings of the Entomologia Experimentalis et Applicata, vol. 119, no. 2, pp. National Academy of Sciences of the United States of America, 121–127, 2006. vol. 101, no. 51, pp. 17753–17758, 2004. [99] S. E. Webster, J. Galindo, J. W. Grahame, and R. K. Butlin, [83] S. P. Egan, Ecological speciation in Neochlamisus bebbianae leaf “Habitat choice and speciation,” International Journal of beetles: The role of postmating isolation and the genetic basis of Ecology. In press. host use traits, PhD dissertation, Vanderbilt University, 2010. [84] E. Campan, A. Couty, Y. Carton, M. H. Pham-Delegue,` and L. Kaiser, “Variability and genetic components of innate fruit odour recognition in a parasitoid of Drosophila,” Physiological Entomology, vol. 27, no. 3, pp. 243–250, 2002. [85] T. Matsuo, S. Sugaya, J. Yasukawa, T. Aigaki, and Y. Fuyama, “Odorant-binding proteins OBP57d and OBP57e affect taste Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 242154, 13 pages doi:10.1155/2012/242154

Research Article Larval Performance in the Context of Ecological Diversiﬁcation and Speciation in Lycaeides Butterﬂies

Cynthia F. Scholl,1 Chris C. Nice,2 James A. Fordyce,3 Zachariah Gompert,4 and Matthew L. Forister1

1 Department of Biology, University of Nevada, Reno, NV 89557, USA 2 Department of Biology, Population and Conservation Biology Program, Texas State University, San Marcos, TX 78666, USA 3 Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN 37996, USA 4 Department of Botany, Program in Ecology, University of Wyoming, Laramie, WY 82071, USA

Correspondence should be addressed to Cynthia F. Scholl, [email protected]

Received 26 July 2011; Accepted 29 November 2011

Academic Editor: Rui Faria

Copyright © 2012 Cynthia F. Scholl et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The role of ecology in diversification has been widely investigated, though few groups have been studied in enough detail to allow comparisons of different ecological traits that potentially contribute to reproductive isolation. We investigated larval performance within a species complex of Lycaeides butterflies. Caterpillars from seven populations were reared on five host plants, asking if host-specific, adaptive larval traits exist. We found large differences in performance across plants and fewer differences among populations. The patterns of performance are complex and suggest both conserved traits (i.e., plant effects across populations) and more recent dynamics of local adaptation, in particular for L. melissa that has colonized an exotic host. We did not find a relationship between oviposition preference and larval performance, suggesting that preference did not evolve to match performance. Finally, we put larval performance within the context of several other traits that might contribute to ecologically based reproductive isolation in the Lycaeides complex. This larger context, involving multiple ecological and behavioral traits, highlights the complexity of ecological diversification and emphasizes the need for detailed studies on the strength of putative barriers to gene flow in order to fully understand the process of ecological speciation.

1. Introduction well-studied systems, although exceptions exist [6, 11], niche conservatism and niche evolution are often characterized by Understanding the processes underlying diversification is a a small number of ecological traits, such as habitat preference central question in evolutionary biology. Lineages diversify or physiological performance [12–14]. To understand the along multiple axes of variation, including morphologi- causes and consequences of evolution in ecological traits, cal, physiological, and ecological traits. With respect to more studies are needed of groups in which diversification is diversification in ecological traits, many recent studies have recent or ongoing and multiple ecological traits are studied. found that ecological niches can be highly conserved from a The study of multiple traits is particularly important for our macroevolutionary perspective [1–3]. In other words, closely understanding of ecological speciation. For example, it has related species tend to utilize similar resources or occupy been suggested that weak selection acting on a multifarious similar environments. In contrast, the field of ecological spe- suite of traits could be as important for speciation as strong ciation suggests that ecological traits can evolve due to selection acting on a single ecological trait [15]. disruptive selection and drive the process of diversification The butterfly genus Lycaeides (Lycaenidae) includes a [4–8]. In herbivorous insects, evolution in response to complex of taxa in North America that has been the focus habitat or host shifts is often thought to be a first step of studies investigating the evolution and ecology of host in the evolution of reproductive isolation [9, 10]. In most use, mate choice, and genitalic morphology, among other 2 International Journal of Ecology subjects [16–19]. In the context of diversification, this group context of possible reproductive isolation related to variation is interesting because hybridization has been documented in ecological traits. among multiple entities, with a variety of consequences [20, 21], including the formation of at least one hybrid species in the alpine of the Sierra Nevada mountains [22]. 2. Methods The Lycaeides taxa in western North America (specifically L. idas, L. melissa, and the hybrid species) differ in many traits, Two of our focal taxa, L. idas and L. melissa, are widely dis- some of which have been implicated in the evolution of tributed across western North America. Our study focused ecological reproductive isolation in this system. For example, on populations of these two species and the hybrid species there is variation in the strength of host preference, which is in northern California and Nevada (Figure 1). In this area, often linked to reproductive isolation in herbivorous insects L. idas is found on the west slope of the Sierra Nevada, L. that mate on or near their host plants, as Lycaeides do [17]. melissa is found on the eastern side, and the hybrid species is There are also potentially important differences in mate only found in the alpine zone. Lycaeides species use a variety preference, phenology, and egg adhesion [16, 23]. The latter of plants in the pea family, Fabaceae, as hosts, although trait is interesting with respect to the evolution of the hybrid (with few exceptions) specific Lycaeides populations gen- species, which lacks egg adhesion [22, 23]. The eggs of the erally utilize a single host plant species. The two L. idas hybrid species fall from the host plants. This is presumed to populations studied were Yuba Gap (YG), which uses Lotus be an adaptation to the characteristics of the alpine plants, for nevadensis as a host, and Leek Springs (LS) which uses the which the above-ground portions senesce and are blown by host Lupinus polyphyllus (Table 1). Both populations of the the wind away from the site of next year’s fresh growth (thus hybrid species, Mt. Rose (MR) and Carson Pass (CP), use eggs that fall off are well positioned for feeding in the spring). Astragalus whitneyi. At Washoe Lake (WL), L. melissa uses Since the eggs of lower-elevation Lycaeides taxa do adhere the native host Astragalus canadensis;atBeckwourthPass to hosts, this trait could serve as a barrier to gene flow with (BP), butterflies use both A. canadensis and alfalfa, Medicago respect to individuals immigrating from lower elevations. sativa; at Goose Lake (GLA), the only available host is alfalfa Our state of knowledge for Lycaeides is unusual for well- (Table 1). studied groups of herbivorous insects in that we know a great Lycaeides idas and the hybrid species are univoltine, deal about a diversity of traits, as discussed above, but we while L. melissa populations have at least three generations have not heretofore investigated larval performance across per year. Eggs from the univoltine populations have to be taxa in the context of ecological speciation, which is often maintained under winter conditions (i.e., cold temperatures one of the first traits studied in other insect groups [24]. This and darkness) in the lab for experiments in the following study has two goals, first to investigate larval performance spring. Females and eggs from univoltine populations were and then to put this information in the context of other collected in the summer of 2008 to be reared in the summer already-studied traits to investigate which traits might be of 2009, while L. melissa females and eggs were collected important for reducing gene flow between populations and during the 2009 summer. Females were collected from species in this system. We have focused on performance of the L. idas andhybridspeciespopulations(32fromYuba caterpillars from both L. idas and L. melissa populations Gap, 50 from Leek Springs, 45 from Carson Pass, and 40 as well as from populations of the hybrid species. Beyond from Mt. Rose) and caged individually or in small groups the inclusion of the hybrid taxon, of added interest is with host plants for a period of three days after which the fact that L. melissa has undergone a recent expansion eggs were collected. Eggs were washed with a dilute (2%) ◦ of diet, encompassing exotic alfalfa (Medicago sativa)asa bleach solution and held over the winter at 4–6 C. Eggs larval host plant across much of its range. Thus we are were removed from cold storage on May 27th, 2009, and able to investigate variation in the key ecological trait of the majority hatched within several hours. The number of larval performance across multiple levels of diversification, caterpillars hatching synchronously required that the larvae including the differentiation of L. idas and L. melissa, the be moved in groups of twenty to standard-sized petri dishes formation of a hybrid species, and a host expansion that has (100 mm diameter) with fresh plant material on the 27th and occurred within the last two hundred years [19, 22, 25]. 28th. On the 29th and 30th of May, the groups of twenty Using individuals from two L. idas, three L. melissa,and were split into three dishes each containing three to seven two hybrid species populations, we conducted reciprocal individuals. An average of 6 caterpillars was added to 9 dishes rearing experiments using all five of the host species found per treatment (plant/population combination); in some at these focal populations. We assessed larval performance by cases fewer (but not less than three) caterpillars were added examining survival, time to emergence (eclosion), and adult per dish to try to maximize the number of dishes, which is weight, and by comparing survival curves from different the unit of replication (see below). Once all the larvae in populations on the different plants. For each population, we a petri dish reached the 3rd or 4th instar they were moved contrasted larval performance on a natal host to performance to larger petri dishes (170 mm diameter). These groups of on the plants of other populations. Higher larval perfor- individuals were considered a “rearing dish,” and dish was mance on natal host plants would support the hypothesis of used as a random factor in statistical analyses (see below). local adaptation to host plant species. In the second part of Females and eggs from the three L. melissa populations the paper, these results are discussed both within the light of were collected following similar protocols (though without local adaptation in a diversifying group and also within the the necessity of overwintering). Seventeen females were International Journal of Ecology 3

L.idas

L.melissa

Leek Springs (LS) L.idas Yuba Gap (YG) Carson Pass (CP) Alpine Mt. Rose (MR) Washoe Lake (WL) Beckwourth Pass (BP) L.melissa Goose Lake (GLA)

(a) (b)

Figure 1: (a) Ranges of L. idas and L. melissa across North America, darker shaded regions correspond to ranges of overlap, which includes alpine populations of the hybrid species considered here (CP and MR). (b) Map of sampled Lycaeides populations in Northern California and Nevada, USA. Symbols correspond to populations and taxa.

Table 1: Locations of populations (see also Figure 1) and hosts associated with the seven populations studied.

Taxon Location Latitude/longitude Host Leek Springs (LS) 38◦388 N/120◦1425 W Lupinus polyphyllus L. idas Yuba Gap (YG) 39◦1924 N/120◦3560 W Lotus nevadensis Carson Pass (CP) 38◦4228 N/120◦028 W Astragalus whitneyi Hybrid species Mt. Rose (MR) 39◦1921 N/119◦5547 W Astragalus whitneyi Washoe Lake (WL) 39◦1359 N/119◦4646 W Astragalus canadensis L. melissa Beckwourth Pass (BP) 39◦4655 N/120◦423 W Astragalus canadensis and Medicago sativa Goose Lake (GLA) 41◦599 N/120◦1732 W Medicago sativa

collected from Beckwourth Pass during the last week of June, Larvae from each population were reared on all five 27 were collected from Goose Lake the third week of July, and plants, Astragalus canadensis, Astragalus whitneyi, Lotus ne- 14 were collected from Washoe Lake the last week of July. For vadensis, Lupinus polyphyllus, and Medicago sativa,with these populations, larvae were added to standard-sized petri individual rearing dishes being assigned exclusively to a sin- dishes in groups of four to six individuals as soon as the eggs gle plant throughout development. Caterpillars in the wild hatched. Again, caterpillars were transferred to a large petri consume both vegetative and reproductive tissues, but only dish once all the individuals in a dish reached the 3rd or 4th leaves were used in this study, as flowers would be difficult to instar. standardize across plants (not being available synchronously 4 International Journal of Ecology for most species). For a study of this kind, ideally all plant populations and taxa that may have inherent differences, material to be used in rearing would be collected from such as in size or in development time. Z scores were used focal locations (where butterflies are flying) or grown in a in analyses described below unless otherwise noted. common environment. However, many of these species are Dish was considered the unit of replication, thus percent not easily propagated, and moreover our focal locations are survival was calculated per dish. For analyses of adult widely dispersed geographically; these factors necessitated weight and time to emergence, dish was used as a random some compromise in collecting some of the plants. A. factor. Percent survival was analyzed using analysis of vari- canadensis cuttings were obtained at the site of the butterfly ance (ANOVA) with plant, population and the interaction populations at Beckwourth Pass and Washoe Lake and from between the two as predictor variables. Time to emergence the greenhouse (plants were grown from seeds collected at and adult weight were both analyzed with ANOVA, using Washoe Lake). Astragalus whitneyi was collected from the population, plant, the interaction between the two and site of the Mt. Rose hybrid species population and on a sex as predictor variables, along with dish as a random hillside adjacent to Carson Pass (38◦4223/120◦0023). factor nested within plant and population. For all of these All Lotus nevadensis were collected from Yuba Gap (YG). analyses, ANOVA was performed a second time without the The only case in which plant material was collected from a plant/population interaction if it was not significant at α< site where the butterfly is not found is Lupinus polyphyllus. 0.05. These analyses were performed using JMP software These plants were collected off I-80 at the Soda Springs exit version 8.0.2 (SAS Institute). (39◦1929/120◦2325) and seven miles north of Truckee Differences in survival were also investigated by generat- CA, off State Route 89 (39◦2559/120◦1213). Medicago ing and comparing survival curves. To create survival curves, sativa was obtained from Beckwourth Pass (BP) and from individual caterpillars were assumed to be alive until the plants grown in the greenhouse with seeds from BP. M. date they were found dead. Rather than analyzing survival sativa was also collected from south of Minden, NV on State curves on an individual-dish basis (where sample sizes were Route 88 (38◦4860/119◦/46/46)andoff of California small), the number of individuals surviving on a given State Route 49 in Sierra Valley, CA (39◦3835/120◦2310). day was calculated for each plant/population combination, Plant material was kept in a refrigerator and larvae were fed giving one curve per combination, as is often done in fresh cuttings whenever the plant material in petri dishes survival analysis [26]. Survival curves were generated in R was significantly reduced or wilted, which was approximately (2.12.2) using the packages splines and survival. The shapes every two to seven days. Each time caterpillars were given of the curves were investigated within population using the fresh plant material, the number of surviving caterpillars packages MASS and fitdistrplus. Weibull distributions are was recorded along with the date. All dishes were kept at commonly used to model survival using two parameters, room temperature, 20◦ to 23◦ Celsius, on lab benches. Newly shape and scale. The shape parameter measures where the emerged adults were individually weighed to the nearest inflection point occurs or practically whether individuals 0.01 mg on a Mettler Toledo XP26 microbalance and sex was are lost more at the beginning or end of a given time recorded. period, and the scale parameter characterizes the depth of the curve. We estimated the two Weibull parameters, shape and 2.1. Analyses. The strengths of our experiment were that scale, that characterized the fitted curves using maximum we reared a large number of individuals from multiple likelihood. One-thousand bootstrap replicates were then taxa across five plants, but a weakness of our design was used to generate 95% confidence intervals for the shape and that not all rearing could be done simultaneously. As scale parameters, so that they could be compared across discussed further below, flowers were not included in the plants within a given population. rearing, and plant material was collected from most but not all focal populations. Experiments were conducted in 3. Results two phases, first involving the populations of the hybrid species and L. idas, being reared together and earlier in We began the larval performance experiments with 2040 the spring, and second involving the three low-elevation L. caterpillars in 357 dishes. Average survival to eclosion across melissa populations being reared later in the summer. This all experiments was 23.4%. In general, differences in larval division into two rearing groups was largely a consequence performance among plants were greater than differences of being constrained by the total number of caterpillars that between populations, which can be seen both in Figure 2 could be handled and reared in the lab at any one time. and also by comparing variation partitioned by plants and Considering the possibility that phenological variation in population in Table 2. For example, survival was highest on plants could have implications for larval performance, we Lotus nevadensis across all populations for all three taxa, with conducted analyses separately for the three butterfly species. anaveragesurvivalof56.5%(survivalonLupinus polyphyllus Postemergence adult weight, time to emergence as adult, was comparable for two of the three L. melissa populations). and survival to adult were recorded. Mortality (reflected in Survival on alfalfa was consistently the lowest of any plant the survival data) included death associated with caterpillars across populations: only two caterpillars survived to eclosion that died while developing, individuals that pupated but (Figures 2 and 3). Because survival was so low on alfalfa, it failed to emerge, and disease; we did not distinguish between was excluded from most analyses and figures. The inferior these sources of mortality. Data were standardized (Z trans- nature of alfalfa as a host plant is consistent with previous formed) within populations to facilitate comparisons among studies, particularly when caterpillars do not have access to International Journal of Ecology 5

Survival Emergence weight Survival 2 2 0.8 a 1.5 1.5 0.4 1 1 0 b 0.5 −0.4 0.5 −0.8 0 0 −1.2 −0.5 −0.5 −1.6 −1 −1 Ac Aw Ln Lp Ac Aw Ln Lp Ac Aw Ln Lp

LS LS MR YG YG CP (a) (b) (c)

Emergence weight Survival Emergence weight 1 1.5 1 ab 0.5 a 1 a 0 0.5 b 0.5 b −0.5 ab 0 0 b −1 a −0.5 − − 1.5 −0.5 1 ab −2 −1.5 −1 −2.5 −2.5 −3 −1.5 −2 Ac Aw Ln Lp Ac Aw Ln Lp Ac Aw Ln Lp

MR WL WL CP BP BP GLA GLA (d) (e) (f)

Figure 2: Survival and emergence weights from rearing experiments: (a and b) L. idas, (c and d) hybrid species, (e and f) L. melissa. Values are standardized (Z transformed). Shading of bars indicates plant species; see legends and Figure 1 for population locations. See Table 2 for associated model details, including plant and population effects. Significant differences (P<0.05) for population effects within plants are indicated here with small letters near bars. Black dots identify natal host associations for each population. Host plant abbreviations as follows. Ac; Astragalus canadensis;Aw;Astragalus whitneyi; Ln; Lotus nevadensis; Lp; Lupinus polyphyllus. Results from alfalfa, M. sativa,are not shown here because survival was very low; see main text for details.

flowers and flower buds. When flowers have been included exotic host alfalfa, and survival on A. canadensis is interme- in performance experiments, survival of L. melissa on alfalfa diate for BP, where both A. canadensis and alfalfa are utilized. and A. canadensis was equal, although those individuals Thus host use by L. melissa populations predicts variation reared on alfalfa were significantly smaller adults [19]. in larval performance. Effects of plant and population were Plant and the interaction between plant and population generally not as pronounced for either adult weight or time were significant predictors of survival for L. idas, L. melissa, to emergence (for L. idas, the only significant predictors of and the hybrid species (Table 2). Within taxa, there were dif- adult weight were dish and sex); exceptions to this include the ferences among populations on certain plants. For example, significant population by plant interaction for adult weight survival for L. idas on Lotus nevadensis was greater for LS of L. melissa.Aswithsurvival,L. melissa performance (adult compared to YG, but the pattern was reversed for the host weight) was greater on A. canadensis for the population that Lupinus polyphyllus (Figure 2); in other words, each popula- is associated with that plant, WL (Figure 2(f)). tion had higher survival on the natal host of the other pop- Consistent with results for survival to emergence as an ulation. A different pattern can be seen across populations adult, survival curves through time also showed pronounced of L. melissa on Astragalus canadensis, where survival was differences among plants (Figure 3; Table 4). For example, highest for individuals from WL, a population whose natal the Weibull scale parameter for alfalfa was generally different host is A. canadensis. L. melissa survival on A. canadensis was compared to the other plants, reflecting early and pervasive lowest for GLA, which is a population associated with the mortality for individuals reared on that plant. Most but not 6 International Journal of Ecology

Table 2: Results from analyses of variance for the three measures of performance: percent survival, adult weight, and time to emergence. In all cases dish was used as the unit of replication. Most population/plant combinations had 9 dishes, except for the following: YG/Ac 12 dishes, YG/Ln 12 dishes, YG/Lp 12 dishes, YG/Ms 12 dishes, and all plant combinations for CP and MR had 12 dishes. The total number of dishes was 357. P SS F Ratiodf Survival L. idas < Plant 48.62 42.023,73 0.0001

Population 0.02 0.061,73 0.81

Plant × population 4.00 3.463,73 0.02 Survival hybrid species < Plant 30.93 16.333,88 0.0001

Population 0.00 0.001,88 1.00

Plant × population 7.50 3.963,88 0.01 Survival L. melissa < Plant 44.95 29.323,96 0.0001

Population 0.03 0.032,96 0.97

Plant × population 9.99 3.266,96 0.006 Adult weight L. idas

Plant 2.11 0.523,33.47 0.67

Population 0.49 0.141,19.38 0.71 < Dish (plant, population) random 47.78 4.0426,74.00 0.0001 < Sex 10.81 11.872,74.00 0.0001 Adult weight hybrid species

Plant 7.25 2.813,48.09 0.049

Population 0.53 0.6131,48.45 0.44

Dish (plant, population) random 14.90 0.7420,30.00 0.76

Sex 1.29 1.281,30.00 0.27 Adult weight L. melissa

Plant 21.22 7.253,132.13 0.0002

Population 8.62 4.542,154.10 0.01

Dish (plant, population) random 89.56 1.4484,217.00 0.02

Plant × population 11.24 15.151,217.00 0.0001

Sex 25.93 4.396,126.13 0.0005 Time to emergence L. idas < Plant 20.32 8.913,59.17 0.0001

Population 0.69 0.991,4.11 0.37

Dish (plant, population) random 19.34 0.9426,75.00 0.55

Sex 3.87 2.452,75.00 0.09 Time to emergence hybrid species

Plant 5.30 3.282,25.82 0.05

Population 0 00,28.00 1.00

Dish (plant, population) random 16.06 0.9520,28.00 0.54

Sex 1.20 1.411,28.00 0.25

Plant × population 6.04 3.612,42.94 0.04 Time to emergence L. melissa

Plant 9.36 2.413,106.23 0.07

Population 1.66 0.612,99.01 0.54 < Dish (plant, population) random 125.50 2.3484,223.00 0.0001 < Sex 24.18 37.871,223.00 0.0001 International Journal of Ecology 7

Leek Springs (LS) Yuba Gap (YG) 1 1

0.8 Ln 74.7 0.8 0.6 0.6 Ln 55.9

0.4 0.4 Survival Lp 28.2 0.2 0.2 Aw 6.7 Lp 6.6 Ac 2.9 0 Ac 2.2 0 0 1020304050 0 10 20 30 40 50 Time (days) Time (days) Ac (A, a) Lp (AB, b) Ac (A, b) Lp (A, c) Aw (B, a) Ms (B, a) Aw (A, b) Ms (B, a) Ln (A, c) Ln (A, d) (a) (b)

Carson Pass (CP) Mt. Rose (MR) 1 1

0.8 0.8

0.6 0.6 Ln 45.8 0.4

Survival 0.4 Ac 5.6 0.2 Lp 3.8 0.2 Ln 12.0 Aw 3.1 Ac 8.3 0 Ms 1.4 0 Lp 1.7 010203040 50 0 1020304050 Time (days) Time (days) Ac (B, b) Lp (B, a) Ac (A, b) Lp (A, a) Aw (B, b) Ms (B, a) Aw (B, b) Ms (AB, a) Ln (A, c) Ln (A, b) (c) (d)

Washoe Lake (WL) Beckwourth Pass (BP) 1 1 Lp 83.0 0.8 Ln 77.8 0.8 Ln 77.8 Ac 64.8 0.6 Lp 63.0 0.6

Survival 0.4 0.4 Ac 34.3 Aw 30.0 Aw 22.2 0.2 0.2

0 0 0 1020304050 0 1020304050 Time (days) Time (days) Ac (A, c) Lp (A, c) Ac (B, b) Lp (A, c) Aw (A, b) Ms (A, a) Aw (B, b) Ms (BC, a) Ln (A, c) Ln (A, c) (e) (f)

Figure 3: Continued. 8 International Journal of Ecology

Goose Lake (GLA) 1

0.8 Ln 70.4 Lp 68.5 0.6

0.4 Survival Aw 29.2 0.2 Ac 14.8 0 Ms 1.85 0 1020304050 Time (days) Ac (B, b) Lp (A, c) Aw (A, b) Ms (AB, a) Ln (A, c)

(g)

Figure 3: Survival curves for the seven populations studied. Colors indicate survival associated with a given plant; letters next to each plant in legends correspond to differences in the shape (upper case) and scale (lower case) of each curve indicated by nonoverlapping 95% confidence intervals from bootstrapped parameter values (see Table 4 for more details). Black dots indicate native host association for each population. Average, final survival is shown to the right of each graph for nonzero results. Plant abbreviations as follows. Ac; Astragalus canadensis;Aw; Astragalus whitneyi; Ln; Lotus nevadensis; Lp; Lupinus polyphyllus; Ms; Medicago sativa. all mortality was manifested quite early in development, In general, the survival that we report (23.4% throughout particularly for the alpine and L. idas populations across the experiment) could reflect the absence of flowers or other plants. Mortality was more evenly distributed through time unfavorable lab conditions, and we do not at this time have for L. melissa on all plants. In some cases, patterns of survival life history data from the field for Lycaeides with which vary among plants within populations, even when overall to compare our results. However, in interpreting results survival was low. For example, survival curves for CP drop here and elsewhere (e.g., [19]), we make the assumption much less rapidly for three plants, one of which is the natal that lab experiments are informative with respect to relative host A. whitneyi and another is the congeneric A. canadensis performance across hosts. In other words, the consumption (the third is L. nevadensis). For A. canadensis, it is interesting of M. sativa by Lycaeides caterpillars is associated with to note that across all populations there was a drop-off in sur- development into adults that are small relative to adults that vival near the end of development: many individuals made it develop on other plants. Without artifacts of lab rearing, it is to the pupal stage, but failed to emerge, perhaps suggesting possible that performance would generally be higher in the a subtle nutritional challenge for successful completion of wild, but we would predict that performance on M. sativa development presented by that plant. would still be lower relative to performance on native hosts. An alternate possibility, which we cannot test at this time, is that lab rearing has plant-specific effects (i.e., M. sativa is a 4. Discussion poor host only when used under artificial conditions). For all the performance results, it is also important to The reciprocal rearing experiment detected strong host plant note that phenological effects of changes in plant quality ff e ects and limited evidence of local adaptation to natal or suitability could be pronounced, but are not addressed host plant species in the Lycaeides species complex. For by our experimental design. In particular, as noted above, example, development on L. nevadensis (the host of the L. our rearings were conducted in two phases due to logistical idas population at YG) resulted in relatively high survival constraints: first including L. idas and populations of the throughout the experiment, while development on Medicago hybrid species and second including all three L. melissa sativa (the exotic host of L. melissa at GLA and BP) led to populations. This is not a completely unnatural situation, as extremely low survival in all cases. These plant effects that L. idas and the hybrid species are univoltine, while L. melissa transcend populations could be indicative of larval traits populations are multivoltine. Thus L. idas andhybridspecies (such as high survival on L. nevadensis) that are conserved caterpillarsaremorelikelytobeexposedtoonlytheearly in the group and are not particularly labile. Our results could spring vegetation, as in our experiment. The consideration be influenced by the use of leaves but not flowers in larval of phenological effects in plants is most relevant when rearing. Previous work has shown that survival is improved comparing performance among butterfly taxa (e.g., the on Medicago sativa for larvae that have access to flowers, performance of L. melissa versus L. idas on a particular plant) but this is not true on Astragalus canadensis [19]. We do but is less important when making comparisons within a not know if flowers are or are not important for larvae taxon (e.g., the performance of L. melissa on different plant developing on the other plants. species). International Journal of Ecology 9

In contrast to the general result of strong plant effects Egg adhesion across taxa and populations, one result suggestive of local Hybrid species adaptation is the performance (survival and adult weight) of L. melissa on the native host A. canadensis [27]. Performance was highest on the native for the population that utilizes that host in the wild and lowest for a population associated Voltinism Hybrid species with the exotic host alfalfa. Performance on A. canadensis is L. idas L. melissa intermediate for the population where both hosts are used. These results raise a number of possibilities, including a scenario in which genetic variants associated with higher Host preference Hybrid species performance on an ancestral host were lost in the transition to the exotic host, which could be a consequence of relaxed L. idas L. melissa selection or a population bottleneck in the new environment. Another explanation could involve a change in gene regulation associated with performance, rather than a loss of alleles. In any event, the transition to the novel host has apparently L. idas L. melissa Larval not been accompanied by an increase in performance on performance alfalfa. One caveat to this conclusion is that the M. sativa used Hybrid species in experiments was collected at one of the focal locations Male mate (BP), but could not, for logistical reasons, be collected discrimination from GLA. The latter population (GLA) is the population Hybrid species associated only with M. sativa, thus the conclusion that performance has not increased following the colonization of L. idas L. melissa the novel host could have been different if local plant material from that location had been used in experiments; however, we have found consistently low performance on M. sativa in L. idas L. melissa other experiments [19], suggesting generality to the result of Figure 4: Summary of hypotheses relating ecological traits to low performance on that plant. reproductive isolation between taxa based on the current and past Variation in host preference has previously been doc- studies [16, 17, 19, 22, 23]. Arrows joining two taxa correspond to umented among populations of Lycaeides butterflies, with greater gene flow and those that are faded represent gene flow that populations of the hybrid species in particular exhibiting could be prevented or reduced by a given trait. For details see text. strong preferences for their natal host, A. whitneyi [17, 22], relative to the hosts of other Lycaeides populations. However, we found low survival and low adult weights for individuals ecological trait or a single niche dimension could be of the hybrid species reared on A. whitneyi (Figure 2). It is important for initiating speciation, while the evolution of possible that laboratory conditions were a poor reflection differences along multiple ecological axes might often be of appropriate abiotic conditions for the hybrid species needed for complete reproductive isolation [15]. Multiple individuals adapted to an alpine environment. It is also ecological axes could be different aspects of, for example, possible that other factors, such as the absence of flowers resource use [29, 30], or they could be more disparate traits, in experiments or induced defenses in leaves, could be such as mate finding or predator avoidance. In either case, important in A. whitneyi, which supported poor growth for the idea is that selection along one axis might be insufficient larvae from all populations. In any event, the patterns of perfor reproductive isolation, but selection acting along multiple formance that we report are not consistent with an expected axes might confer a high, overall level of reproductive isola- preference-performance paradigm for host shifts leading to tion. the evolution of reproductive isolation [28]. Variation in Considering the potential importance of multiple traits both adult preference and larval performance is discussed in ecological speciation, Figure 4 and Table 3 summarize further in the following section considering ecological traits information from this and other studies in Lycaeides and and hypotheses relating traits to reproductive isolation. present hypotheses regarding multiple ecological and behavioral traits and how these might interact with ecological, reproductive isolating processes in this system. Specifically, 4.1. Ecology and Diversification. Although many studies of Figure 4 explores hypotheses about reduced gene flow, herbivorous insects have focused on larval performance represented by faded arrows, between the taxa due to the with respect to local adaptation and ecological speciation, ecological differences of a given trait. For example, the model populations of herbivorous insects (or of any organism) can shown for egg adhesion posits that variation in adhesion of course differ in numerous ways, some related to resource could be a barrier to gene flow going from L. idas and L. use but also to other aspects of the environment. Nosil et al. melissa populations into populations of the hybrid species. L. [15] have suggested a number of scenarios in which multiple idas and L. melissa females lay eggs that adhere to plants. As traits could be important in the evolution of reproductive discussed above, the alpine host plants senesce and are blown isolation. In particular, natural selection acting on a single from the area, thus removing any attached eggs from the site 10 International Journal of Ecology

Table 3: Details of behavioral and ecological variation among taxa, specifically as variation relates to potential barriers to gene flow; see Figure 4 for a graphical interpretation of these traits in relation to gene flow. The descriptions of male mate discrimination refer to preferences of males for females of the other taxa. Details for larval performance refer to performance on the hosts of the other taxa relative to performance of the other taxa on those same plants.

L. idas Hybrid species L. melissa Egg adhesion Yes No Yes Preference for natal host(s) Moderate High Low to high Against L. melissa and hybrid Male mate discrimination Against L. idas None species Voltinism Univoltine Univoltine Multivoltine Poor on hosts of L. melissa; Equivalent on hosts of L. idas Poor on hosts of both L. melissa Larval performance superior on host of hybrid and superior on host of hybrid and L. idas species species Further information on specific behavioral and ecological variables (other than larval performance, reported here) can be found as follows: egg adhesion [23], preference for natal hosts [17, 19, 22], male mate discrimination [16], and voltinism [23]. of next-spring’s fresh plant growth [23]. Differences in host to the gene pool) relative to the dynamic just described (an preference might also affect patterns of gene flow between the L. melissa female arriving at an L. idas population). It is species. The host of the hybrid species populations is readily possible to imagine all of the traits depicted in Figure 4 being accepted by ovipositing females from all Lycaeides examined involved in either pre- or postzygotic isolation. For example, thus far [17], thus it would likely be accepted by females from mate preference could act as just described on immigrant, L. idas and L. melissa populations arriving at a population virgin females, as in the immigrant inviability concept of of the hybrid species. In contrast, the hosts of the L. idas Nosil et al. (2005) [31]. An alternative but similar scenario populations are not preferred by females of either the hybrid could involve the offspring of an immigrant; in this case, species or L. melissa [17, 19]. The arrows pointing towards wing-pattern alleles (related to mate choice) would interact L. melissa assume the presence of only the native host A. with mate choice in the next generation. canadensis, not the exotic M. sativa (excluding the exotic is Variation in voltinism could affect gene flow from L. a simplifying assumption for Figure 4, but also appropriate melissa into the other univoltine taxa. Because L. melissa given that ecological diversification occurred before the populations are multivoltine, it is possible that an L. melissa recent introduction of alfalfa). Astragalus canadensis and A. female moving into populations of the other taxa would lay whitneyi are equally acceptable for oviposition by hybrid eggs that failed to diapause in habitats where the univoltine species individuals (Forister, unpublished data), and we strategy is superior (such as in the alpine habitat where there assume the same equivalence for L. idas (i.e., we assume L. is a short window for larval development [32]). Alternatively, idas females would readily accept A. canadensis, just as they diapause could be plastic, in which case the patterns of do with A. whitneyi, thus an arrow without a barrier pointing connectivity (hypothesized patterns of gene flow) pictured from L. idas to L. melissa in the host preference diagram). would be different. Similar to host preference, variation in male mate We can now add larval performance to the suite of discrimination potentially presents barriers only between hypotheses linking ecology and gene flow in Lycaeides.In hybrid species and L. idas populations and between L. idas generating hypotheses relating larval performance to gene and L. melissa populations but not between hybrid species flow, we have used this criterion (focusing on survival, and L. melissa populations [17]. L. melissa males will readily rather than adult weight, as the most straightforward metric approach either L. idas or females of the hybrid species, while of performance): if foreign larvae (i.e., the offspring of a L. idas males discriminate against females from the other two recently arrived female) have lower survival on the local taxa, and hybrid species males discriminate against L. idas host relative to local individuals, we hypothesize a relative females. This behavioral variation among taxa, reported in reduction in gene flow associated with performance. For Fordyce et al. [16], comes from choice tests involving dead gene flow between L. idas and L. melissa, the survival of L. and paper-model females presented in experimental arrays idas larvae on the host of L. melissa is lower than the survival in the field. It is important to note that being a less-preferred of L. melissa caterpillars on the same plant (see mean survival mate is of course not the same as not being mated. In other values in Figure 3). Interestingly in the context of hybrid words, a virgin L. melissa female that immigrated into an L. speciation, our results suggest that gene flow from both L. idas population might be a low-ranked mate for male L. idas idas and L. melissa would be unimpeded into populations of relative to local females, but it is possible that she would the hybrid species relative to the reverse, meaning that the eventually find a mate. However, the patterns of gene flow two parental species had higher survival (relative to hybrid shown in Figure 4 are meant to be hypotheses for potential individuals) on the alpine host and that hybrid individuals barriers to gene flow within a given trait. An L. melissa female had relatively inferior performance on the two parental immigrating into a population of the hybrid species would species’ hosts. Of course, this could be different if another be mated more readily (and thus be more likely to contribute trait, for example, egg adhesion, had a stronger effect or International Journal of Ecology 11

Table 4: Survival curves for each host-population combination. A weibull distribution was fitted to each combination with 1000 bootstrap replicates. We report shape and scale parameters along with bootstrapped confidence intervals. Upper case and lower letters following shape and scale values correspond to 95% confidence intervals that do not overlap (upper case letters for shape and lower case letters for scale) within populations based on the thousand bootstrapped replicates. See Figure 3 for graphical representation of survival curves.

Population Host Shape Scale CP Ac 1.47 (1.25–1.75) B 31.16 (26.22–36.37) b CP Aw 1.61 (1.35–1.97) B 23.53 (20.27–27.09) b CP Ln 0.82 (0.71–0.95) A 63.75 (42.86–110.98) c CP Lp 1.38 (1.12–2.62) B 11.73 (9.61–14.66) a CP Ms 1.67 (1.34–2.42) B 10.86 (9.41–12.76) a BP Ac 1.50 (1.17–2.01) B 51.38 (41.18–64.28) b BP Aw 1.31 (1.07–1.61) B 39.56 (31.44–50.49) b BP Ln 0.44 (0.38–0.52) A 1128.76 (307.38–7477.83) c BP Lp 0.44 (0.39–0.51) A 2152.31 (501.95–21856.68) c BP Ms 2.65 (2.00–4.07) BC 26.36 (23.90–28.98) a GLA Ac 2.154(1.76–2.69) B 40.31 (34.73–46.00) b GLA Aw 0.88 (0.74–1.07) A 39.16 (25.84–61.45) b GLA Ln 0.66 (0.53–0.86) A 236.85 (115.08–750.35) c GLA Lp 0.94 (0.68–1.50) A 134.35 (82.65–321.98) c GLA Ms 1.70 (1.44–2.10) AB 20.27 (16.98–23.59) a LS Ac 1.57 (1.21–2.86) A 10.70 (8.96–13.28) a LS Aw 2.31 (1.97–3.09) B 10.43 (9.22–11.71) a LS Ln 0.88 (0.67–1.24) A 189.07 (104.65–511.29) c LS Lp 1.58 (1.33–2.17) AB 19.51 (16.28–23.62) b LS Ms 2.70 (2.20–3.70) B 9.52 (8.62–10.63) a MR Ac 1.11 (0.97–1.23) A 24.07 (18.38–31.14) b MR Aw 1.84 (1.59–2.16) B 21.37 (18.59–24.33) b MR Ln 1.29 (1.08–1.59) A 22.56 (17.80–29.12) b MR Lp 1.788 (1.38–5.18) A 10.23 (8.83–11.96) a MR Ms 2.73 (2.26–4.42) AB 8.54 (7.71–9.45) a WL Ac 1.65 (1.03–3.29) A 82.61 (62.56–145.41) c WL Aw 1.20 (0.92–1.57) A 41.69 (31.24–56.54) b WL Ln 0.77 (0.57–1.38) A 302.19 (125.45–1912.44) c WL Lp 0.77 (0.56–1.24) A 144.13 (79.93–404.52) c WL Ms 1.43 (1.22–1.89) A 22.96 (18.23–28.27) a YG Ac 1.29 (1.17–1.45) A 22.20 (18.23–27.03) b YG Aw 1.34 (1.12–1.88) A 17.23 (13.41–22.86) b YG Ln 1.05 (0.89–1.21) A 71.74 (50.50–111.48) d YG Lp 1.07 (0.94–1.23) A 36.16 (27.35–49.30) c YG Ms 2.45 (2.12–3.32) B 9.25 (8.30–10.29) a

acted before larval performance in restricting gene flow 5. Conclusion (see [6, 11] for examples of the complexities of estimating components of reproductive isolation associated with a We conducted a performance experiment for seven popula- suite of traits). We stress that these are hypotheses that bear tions from three species within the Lycaeides species complex, further investigation, as we know that larval performance is L. idas, L. melissa, and the hybrid species, on five different complex, being affected not only by variation in host quality plants. Our primary results include large plant effects, with (i.e., the availability of flowers [19]) but also by the presence L. idas hosts being generally superior for larval development of mutualistic ants and natural enemies [33]. and the exotic host of L. melissa being extremely poor, both 12 International Journal of Ecology for L. melissa and the other taxa. In general, there is little was supported by the National Science Foundation (IOS- evidence of local adaptation in these performance data. This 1021873 and DEB-1050355 to C. C. Nice; DEB-0614223 conclusion is perhaps consistent with the fact that these and DEB-1050947 to J. A. Fordyce; DEB 1020509 and DEB butterfly taxa are associated with multiple hosts throughout 1050726 to M. L. Forister; DEB-1011173 to Z. Gompert). C. their geographic ranges. Thus gene flow could limit local ad- F. Scholl was supported by the Biology Department at the aptation to any particular plant species. As a consequence, University of Nevada, Reno. variation in larval performance across multiple hosts is unlikely to be the dominant mechanism of reproductive isolation between populations and taxa. References Our results (including some evidence for local adaptation [1]A.T.Peterson,J.Soberon,´ and V. Sanchez-Cordero,´ “Conser- among L. melissa populations for their native host, A. vatism of ecological niches in evolutionary time,” Science, vol. canadensis) together with previously published data [16, 17, 285, no. 5431, pp. 1265–1267, 1999. 19, 22, 23] were integrated to build a hypothetical model [2] A. Prinzing, W. Durka, S. Klotz, and R. Brandl, “The niche of relating ecology to reproductive isolation and diversification higher plants: evidence for phylogenetic conservatism,” Pro- in Lycaeides. The model presented in Figure 4 describes a ceedings of the Royal Society B: Biological Sciences, vol. 268, no. system that is well poised for a test of the “multifarious 1483, pp. 2383–2389, 2001. selection” hypothesis [15]. One hypothesis that can be [3] K. H. Kozak and J. J. Wiens, “Does niche conservatism generated from Figure 4 is that there might not be one promote speciation? A case study in North American salaman- single trait that could act as a barrier to gene flow between ders,” Evolution, vol. 60, no. 12, pp. 2604–2621, 2006. all three taxa, and most traits might only act to reduce [4] U. Dieckmann and M. Doebeli, “On the origin of species by gene flow asymmetrically. For example, egg adhesion could sympatric speciation,” Nature, vol. 400, no. 6742, pp. 354–357, affect gene flow from both L. idas and L. melissa into the 1999. hybrid species, but would not necessarily be effective in the [5] D. Schluter, “Ecology and the origin of species,” Trends in Ecol- opposite directions (from the hybrid species into L. idas ogy and Evolution, vol. 16, no. 7, pp. 372–380, 2001. and L. melissa). More generally in the context of ecological [6] J. Ramsey, H. D. Bradshaw, and D. W. Schemske, “Compo- speciation, a greater number of traits might increase the nents of reproductive isolation between the monkeyflowers possibility that a hybrid “falls between” the niches repre- Mimulus lewisii and M. cardinalis (Phrymaceae),” Evolution, vol. 57, no. 7, pp. 1520–1534, 2003. sented by the two adaptive peaks occupied by the species [7] D. J. Funk, P. Nosil, and W. J. Etges, “Ecological divergence ex- or incipient species [34, 35]. In a relatively simple example hibits consistently positive associations with reproductive iso- involving two traits, hybrids between populations of Mitoura ff lation across disparate taxa,” Proceedings of the National Acad- butterflies associated with di erent host plant species inherit emy of Sciences of the United States of America, vol. 103, no. 9, a maladaptive mismatch of traits: hybrid individuals have pp. 3209–3213, 2006. higher performance on one of the parental hosts, but express [8] K. W. Matsubayashi, I. Ohshima, and P.Nosil, “Ecological spe- an oviposition preference for the other host [36]. ciation in phytophagous insects,” Entomologia Experimentalis However, the importance of multiple traits for ecological et Applicata, vol. 134, no. 1, pp. 1–27, 2010. speciation in Lycaeides must wait on estimates of historical [9] M. Dres` and J. Mallet, “Host races in plant-feeding insects and contemporary gene flow between pairs of populations and their importance in sympatric speciation,” Philosophical and analyses of those estimates in light of variation in Transactions of the Royal Society B: Biological Sciences, vol. 357, ecological and behavioral traits [37]. The inclusion of such no. 1420, pp. 471–492, 2002. comparative data, particularly for a larger suite of popula- [10] P. Nosil, “Divergent host plant adaptation and reproduc- tions, would perhaps reveal the influence of a single trait tive isolation between ecotypes of Timema cristinae walking for explaining a majority of the variation in reproductive sticks,” American Naturalist, vol. 169, no. 2, pp. 151–162, 2007. isolation. It is also possible that a key trait for reproductive [11] N. H. Martin and J. H. Willis, “Ecological divergence associ- isolation remains unstudied in this system. For future studies ated with mating system causes nearly complete reproductive in this group, it will also be important to sample popu- isolation between sympatric Mimulus species,” Evolution, vol. lations widely throughout the geographic ranges of the focal 61, no. 1, pp. 68–82, 2007. butterfly species (Figure 1), as dynamics of local adaptation [12] C. Wiklund, “The evolutionary relationship between adult and diversification can be affected by geographic context, oviposition preferences and larval host plant range in Papilio machaon L,” Oecologia, vol. 18, no. 3, pp. 185–197, 1975. particularly proximity to the edge of a range and potentially marginal habitats. Beyond the details of ecological [13]S.S.WassermanandD.J.Futuyma,“Evolutionofhostplant utilization in laboratory populations of the Southern cowpea diversification in Lycaeides, our results should generally weevil, Callosobruchus maculatus fabricius (Coleoptera: Bru- stress the importance of delving deeper than the traditional chidae),” Evolution, vol. 35, no. 4, pp. 605–617, 1981. “preference-performance relationship” when investigating [14] T. P.Craig, J. D. Horner, and J. K. Itami, “Hybridization studies ecological speciation in herbivorous insects. on the host races of Eurosta Solidaginis: implications for sympatric speciation,” Evolution, vol. 51, no. 5, pp. 1552–1560, Acknowledgments 1997. [15] P. Nosil, L. J. Harmon, and O. Seehausen, “Ecological explana- The authors would like to thank Bonnie Young and Temba tions for (incomplete) speciation,” Trends in Ecology and Evo- Barber for assistance in rearing caterpillars. This work lution, vol. 24, no. 3, pp. 145–156, 2009. International Journal of Ecology 13

[16] J. A. Fordyce, C. C. Nice, M. L. Forister, and A. M. Shapiro, [33] M. L. Forister, Z. Gompert, C. C. Nice, G. W. Forister, and “The significance of wing pattern diversity in the Lycaenidae: J. A. Fordyce, “Ant association facilitates the evolution of diet mate discrimination by two recently diverged species,” Journal breadth in a lycaenid butterfly,” Proceedings of the Royal Society of Evolutionary Biology, vol. 15, no. 5, pp. 871–879, 2002. B: Biological Sciences, vol. 278, no. 1711, pp. 1539–1547, 2011. [17]C.C.Nice,J.A.Fordyce,A.M.Shapiro,andR.Ffrench- [34] H. A. Orr, “Adaptation and the cost of complexity,” Evolution, Constant, “Lack of evidence for reproductive isolation among vol. 54, no. 1, pp. 13–20, 2000. ecologically specialised lycaenid butterflies,” Ecological Ento- [35] S. Gavrilets, Fitness landscapes and the origin of species, Prince- mology, vol. 27, no. 6, pp. 702–712, 2002. ton University Press, Princeton, NJ, USA, 2004. [18] L. K. Lucas, J. A. Fordyce, and C. C. Nice, “Patterns of genitalic [36] M. L. Forister, “Independent inheritance of preference and morphology around suture zones in North American Lycaei- performance in hybrids between host races of Mitoura butter- des (Lepidoptera: Lycaenidae): implications for taxonomy and flies (Lepidoptera: Lycaenidae),” Evolution,vol.59,no.5,pp. historical biogeography,” Annals of the Entomological Society of 1149–1155, 2005. America, vol. 101, no. 1, pp. 172–180, 2008. [37] G. Lu and L. Bernatchez, “Correlated trophic specialization [19]M.L.Forister,C.C.Nice,J.A.Fordyce,andZ.Gompert, and genetic divergence in sympatric lake whitefish ecotypes “Host range evolution is not driven by the optimization of (Coregonus clupeaformis): support for the ecological speciation larval performance: the case of Lycaeides melissa (Lepidoptera: hypothesis,” Evolution, vol. 53, no. 5, pp. 1491–1505, 1999. Lycaenidae) and the colonization of alfalfa,” Oecologia, vol. 160, no. 3, pp. 551–561, 2009. [20]Z.Gompert,C.C.Nice,J.A.Fordyce,M.L.Forister,andA.M. Shapiro, “Identifying units for conservation using molecular systematics: the cautionary tale of the Karner blue butterfly,” Molecular Ecology, vol. 15, no. 7, pp. 1759–1768, 2006. [21] Z. Gompert, L. K. Lucas, J. A. Fordyce, M. L. Forister, and C. C. Nice, “Secondary contact between Lycaeides idas and L. melissa in the Rocky Mountains: extensive admixture and a patchy hybrid zone,” Molecular Ecology, vol. 19, no. 15, pp. 3171– 3192, 2010. [22] Z. Gompert, J. A. Fordyce, M. L. Forister, A. M. Shapiro, and C. C. Nice, “Homoploid hybrid speciation in an extreme habitat,” Science, vol. 314, no. 5807, pp. 1923–1925, 2006. [23] J. A. Fordyce and C. C. Nice, “Variation in butterfly egg adhesion: adaptation to local host plant senescence characteristics?” Ecology Letters, vol. 6, no. 1, pp. 23–27, 2003. [24] J. Jaenike, “Host specialization in phytophagous insects,” Annual Review of Ecology and Systematics,vol.21,no.1,pp. 243–273, 1990. [25] Z. Gompert, J. A. Fordyce, M. L. Forister, and C. C. Nice, “Recent colonization and radiation of North American Lycaei- des (Plebejus) inferred from mtDNA,” Molecular Phylogenetics and Evolution, vol. 48, no. 2, pp. 481–490, 2008. [26] J. P. Klein and M. L. Moeschberger, Survival Analysis: Tech- niques for Censored and Truncated Data,Springer,NewYork, NY, USA, 2nd edition, 2003. [27] T. J. Kawecki and D. Ebert, “Conceptual issues in local adaptation,” Ecology Letters, vol. 7, no. 12, pp. 1225–1241, 2004. [28] S. Gripenberg, P. J. Mayhew, M. Parnell, and T. Roslin, “A meta-analysis of preference-performance relationships in phytophagous insects,” Ecology Letters, vol. 13, no. 3, pp. 383– 393, 2010. [29] M. L. Forister, A. G. Ehmer, and D. J. Futuyma, “The genetic architecture of a niche: variation and covariation in host use traits in the Colorado potato beetle,” Journal of Evolutionary Biology, vol. 20, no. 3, pp. 985–996, 2007. [30] P. Nosil and C. P. Sandoval, “Ecological niche dimensionality and the evolutionary diversification of stick insects,” PLoS ONE, vol. 3, no. 4, Article ID e1907, 2008. [31] P.Nosil, T. H. Vines, and D. J. Funk, “Perspective: reproductive isolation caused by natural selection against immigrants from divergent habitats,” Evolution, vol. 59, no. 4, pp. 705–719, 2005. [32] L. Somme, “Adaptations of terrestrial arthropods to the alpine environment,” Biological Reviews: Cambridge Philosophical So- ciety, vol. 64, no. 4, pp. 367–407, 1989. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 902438, 14 pages doi:10.1155/2012/902438

Research Article Divergent Selection and Then What Not: The Conundrum of Missing Reproductive Isolation in Misty Lake and Stream Stickleback

Katja Ras¨ anen,¨ 1, 2 Matthieu Delcourt,1, 3 Lauren J. Chapman,1 and Andrew P. Hendry1

1 Redpath Museum and Department of Biology, McGill University, 859 Sherbrooke Street. W, Montreal, QC, Canada H3A 2K6 2 Department of Aquatic Ecology, Eawag and ETH-Zurich, Institute of Integrative Biology, Ueberlandstraβe 133, 8600 Duebendorf, Switzerland 3 Department of Biology and Centre for Advanced Research in Environmental Genomics, University of Ottawa, Ottawa, ON, Canada K1N 6N5

Correspondence should be addressed to Katja Ras¨ anen,¨ [email protected]

Received 1 July 2011; Accepted 28 October 2011

Academic Editor: Rui Faria

Copyright © 2012 Katja Ras¨ anen¨ et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

In ecological speciation, reproductive isolation evolves as a consequence of adaptation to different selective environments. A frequent contributor to this process is the evolution of positive assortative mate choice between ecotypes. We tested this expectation for lake and inlet stream threespine stickleback (Gasterosteus aculeatus) from the Misty system (Vancouver Island, Canada), which show strong genetically based adaptive divergence and little genetic exchange in nature. This, and work on other stickleback systems, led us to expect positive assortative mating. Yet, our standard “no-choice” laboratory experiment on common-garden fish revealed no evidence for this—despite divergence in traits typically mediating assortative mating in stickleback. These results remind us that divergent natural selection may not inevitably lead to the evolution of positive assortative mate choice. The apparent lack of strong and symmetric reproductive barriers in this system presents a conundrum: why are such barriers not evident despite strong adaptive divergence and low gene flow in nature?

1. Introduction Threespine stickleback (Gasterosteus aculeatus Linnaeus) are a useful system for investigating ecological speciation Ecological speciation occurs when divergent selection causes because they show dramatic adaptive divergence between adaptive divergence that then causes the evolution of repro- populations inhabiting different environments, hereafter ductive isolation [1]. This process now has considerable referred to as “ecotypes” [20–22]. This adaptive divergence empirical support across a wide range of taxa [1–3]. Some appears linked to PAMC in experimental studies on a of the strongest support comes from laboratory studies number of ecotype pairs: benthic versus limnetic [5, 8, 9, showing that mating is more frequent between individuals 23–25], fresh water versus anadromous [7, 26], and mud from similar environments than between individuals from versus lava [27]. By contrast, PAMC was not found in recent ff di erent environments [4–10]. However, some recent work comparisons of fresh water versus anadromous [28]andlake suggests that positive assortative mate choice (henceforth versus stream stickleback [29]. Our study will further explore PAMC) does not always evolve in response to divergent the lake/stream situation. selection [11–15]. These results suggest the importance Several characteristics make lake/stream stickleback suit- of other modifying factors that influence the evolution able for the study of ecological speciation. First, lake/stream of PAMC—even when divergent selection and adaptive population pairs have evolved independently in many dif- divergence are present [16, 17]. Insight into these modifying ferent watersheds following postglacial colonization by sim- factors can be gained from empirical systems within which ilar marine ancestors [30–33]. Second, lake/stream stick- the strength of PAMC varies among population pairs (e.g., leback often show genetically based adaptive divergence [18, 19]). in phenotypic traits related to foraging, swimming, and 2 International Journal of Ecology predator avoidance [31, 32, 34–40]. Third, lake/stream stick- between the phenotypically and genetically similar Lake and leback are often parapatric, and different population pairs Outlet fish: Inlet females should prefer Inlet males and show different levels of gene flow. In particular, the lake/ choose against both Lake and Outlet males, whereas both stream pairs that show the greatest adaptive divergence are Lake and Outlet females should prefer Lake and Outlet males also those that show the lowest gene flow at neutral markers and choose against Inlet males. [31, 33, 38, 39, 41]. This last pattern is often used to infer progress toward ecological speciation, but it has two 2. Materials and Methods limitations: (1) cause and effect are hard to disentangle [42] and (2) neutral markers are often insensitive indicators of 2.1. Collections and Rearing. In June, parental fish were progress toward ecological speciation [43–45]. It is therefore collected from the Misty system at one Lake site (site 1), important to also test for specific reproductive barriers one Inlet site (site 4, ca. 2.6 km from the lake), and one thought to have an ecological basis [16]. Outlet site (site 4, ca. 0.85 km from the lake) (for a map, see Our investigation of reproductive barriers in lake/stream [47]). Standard artificial crossing methods [52] were used to stickleback focuses on populations found in the Misty generate eight full sibling families for the Lake population, watershed, Vancouver Island, Canada. The Lake and Inlet seven for the Outlet, and four for the Inlet. The smaller stream populations in this watershed show strong genetically number of Inlet families was due to limited availability of based adaptive divergence and very low gene flow in neutral mature females during the collection period. Fertilized eggs markers [37, 38, 40, 46, 47]. They are clearly under strong were transferred to our laboratory at McGill University, divergent selection and so might be expected by the theory Montreal, where rearing proceeded in family-specific groups of ecological speciation to exhibit PAMC. Yet our first test in 20–100 L tanks (aquaria), with a given family split into for this possibility did not find any evidence for this [29]. multiple tanks randomized across four climate-controlled However, the experimental design used in [29](i.e.,pairs chambers. Fish were maintained at roughly equal densities of directly interacting males between which females could (25 fish per 100 L), and families were never mixed. Males choose) was quite different from the standard “no-choice” and females could not be visually discriminated as juveniles, designs that have so frequently revealed PAMC in stickleback and so were raised together. As fry, the fish were initially (Table S1 of the Supplementary material available online at fed live brine shrimp nauplii (Artemia sp.) and then later doi:10.1155120111902438). As experimental setup and envi- switched to a mixed diet of live brine shrimp nauplii, frozen ronmental conditions may influence the outcome of mate blood worms (Chironomid sp.), and frozen brine shrimp. choice studies in various ways (e.g., [48–50]), it is important During the conditioning period (see below), the fish were fed to repeat these tests in multiple different conditions. We a mixture of live brine shrimp nauplii and live blackworms therefore here revisit the question by analyzing mate choice (Lumbriculus sp.). experiments done with the “no-choice” design. This allows Through April of the following year, the fish were us to (1) see if the apparent lack of PAMC between Misty reared under constant “summer” conditions (16 : 8 L : D Lake and Inlet stickleback is robust to experimental methods, photoperiod, 17◦C), after which they were switched to and (2) make more direct comparisons to previous work “winter” conditions (8L : 16D, 12◦C). The different chambers on other stickleback systems. In addition, whereas the study were then switched back to “summer” conditions at various by Raeymaekers et al. [29]investigatedmatechoiceoflake times from September through October to ensure the and inlet females towards lake, inlet, and hybrid males, we availability of fish in breeding condition over an extended include a “control” comparison of Misty Lake fish versus period of time. No nesting material was provided in the Misty Outlet stream fish. These two populations show very rearing tanks, thus preventing breeding activities prior to low adaptive divergence and very high gene flow in nature the experimental trials described below. Fish were used [32, 38, 39, 41, 46, 47, 51]. This comparison can therefore in the experiment once they reached sexual maturity, as serve as a low divergence lake/stream “control” in which we indicated by nuptial color in males and distended abdomens have no reason to expect PAMC—in contrast to the high- in females. This laboratory cohort of fish has been previously divergence comparison of Misty Lake versus Misty Inlet. analyzed for morphological traits [46], male courtship [51], Some other features of our study design are important and swimming performance [40] and represent the parental to note. First, our experimental fish were reared for their generation for the fish used in the experiments of [29, 53]. entire lives in a common garden environment, as opposed to being collected from the wild. This was done to reduce 2.2. The Experiment. Eighteen experimental tanks (102 L; the chance that observed mate choice patterns result from length = 92 cm × width = 32 cm × depth = 39 cm), each environmental (nongenetic) effects, such as imprinting, filled with water to a depth of 20 cm, were set up in a shaded maternal effects, prior experience, and/or other forms of green house at McGill University. The bottom of each tank plasticity (e.g., [48–50]). Second, we presented individual was covered with dark gravel, and one end was additionally males of each type (Lake, Inlet, and Outlet) to females covered with a patch of sand (length = 15 cm × width = of all three types, which should increase inferential ability 32 cm × depth = 2.5 cm). Dead plant matter collected from by controlling for variation among individual males. Our Misty Lake, plus a small amount of green algae and grass main prediction is that if PAMC contributes to reproductive clippings, was provided as nest building material because isolation in the Misty watershed, we should see PAMC pilot studies with wild-caught fish showed that this material between the strongly divergent Lake and Inlet fish, but not was readily used for nest building by all ecotypes. All sides International Journal of Ecology 3 of the tanks were covered with brown packing paper to MS222, measured for standard length, and photographed on minimize disturbance. a grid-ruled background (Nikon coolpix 5400, Nikon Inc.). A The experiment was run under “summer” conditions total of 111 trials were conducted on 43 males. Total numbers (16L : 8D, 18◦C). A single male was placed into each tank and of trials for the Inlet (I), Outlet (O), and Lake (L) ecotype was then stimulated to build a nest by periodic exposures combinations were II = 12, IL = 13, IO = 14, LI = 12, LL = (15 minutes once or twice per day) to a gravid female. 14, LO = 13, OI = 12, OL = 11, and OO = 10 (male origin These “stimulus” females were chosen haphazardly from the indicated by first letter, female origin by the second letter). rearing tanks, were presented in a clear glass jar with a mesh lid, and were not used in the experimental trials described 2.3. Response Variables. Our three measures of female behav- below. A male was assumed to be ready for a trial when he ior were (1) head up (the female inclines her head and body showed nuptial coloration and guarded a nest. Males that upwards in a roughly 45◦ angle), (2) nest inspection (the did not build nests within a week were not used in the female places her snout in the nest opening), and (3) nest experiment. As all males faced multiple randomly chosen entry (the female enters the nest). These variables represent females during the stimulation period, it is unlikely that the progressive stages during the mating sequence. Head up indi- identity of these females would have biased male behavior cates female responsiveness, and nest inspection indicates a during the mating experiment. After a male was removed transition between female responsiveness and female choice. from a tank, the tank was cleaned, all materials were replaced, Nest entry indicates female choice and is strongly correlated and a new male was added. with spawning (pers. obs). (Note. Females can skip any of Prior to the experimental trials, the fish had been reared these steps.) Each of these variables has been used in previous under a combination of natural and full-spectrum fluores- work on stickleback (e.g., [25, 52, 54]), and they are here cent light (Vita-Lite 40 W, Duro-Test, Canada). The Misty treated as binomial variables (i.e., whether or not a given system, however, is highly tannic [37] and is dominated by female showed a given response when interacting with a red light (confirmed by spectroradiometer measurements; given male). N. P. Millar, unpubl. data). We mimicked these lighting To test whether mating is nonrandom with respect to conditions during the experimental trials by filtering full- phenotype, we analyzed female mate choice in relation to spectrum fluorescent light through a filter (rust, code 777, body size and male courtship. Body size is of particular Lee Filters, England) placed above each tank. interest as it has been suggested to be a “magic trait” driving An experimental trial was initiated by allowing a gravid ecologically mediated PAMC in stickleback through a “mate female to swim freely from a clear plastic box into a with your own size” rule [7, 8, 24]. Our two body size male’s tank. Subsequent male-female interactions were then metrics were relative size difference (male size minus female videotaped with a digital camcorder (Canon ZR90; Canon, size, henceforth RSD) and absolute size difference (larger sex see [51] for details). Video recording stopped after 45 minus smaller sex, henceforth ASD)—both indicative of size minutes or after the female entered the nest, whichever came assortative mating. (We also tested for an effect of absolute first. After each mating trial, ripeness of the eggs of each male size, but as that had very minimal effects and did not female was confirmed by gentle pressure on her belly (ripe alter any of our conclusions, we do not present the results eggs require only gentle pressure to extrude). Any trial with here.) Our two measures of male behavior were aggression a nonripe female was disregarded, and the male was tested and display toward females. Aggression and display were again with another female of the same ecotype after a lag time both composite variables calculated as the sum of the of at least three hours. frequencies of individual behaviors that males showed within Each male was sequentially exposed to three females, one each of these behavioral types during the first 15 minutes of each ecotype, at minimum intervals of three hours. The of each trial (for details see [51]). As the aim of the male sequence in which different female types were introduced courtship analyses was to estimate ecotype-specific male to a male was varied among males so as to achieve a behavior, only the first 15 minutes were used to reduce bias similar sequence distribution across the male ecotypes. Some from female behaviors on male behaviors. Both aspects of deviations from a perfectly balanced design occurred when behavior are thought to influence mate choice in stickleback gravid females of the right type were not available at the [55–57] and differ among the ecotypes within the Misty right time. Nine males were therefore tested with only two system, particularly by Lake (and Outlet) males showing female ecotypes and one male with only one female ecotype. more “aggressive” behaviors than Inlet males do [51]. Excluding these males did not influence our conclusions, and so all males are included in the analyses presented below. 2.4. Statistical Analyses. All statistical analyses were con- Each female was used in only a single trial, and males and ducted in SAS 9.2 (SAS Institute, Inc.). Female behaviors females from the same family were not combined in any trial. were analyzed with generalized linear mixed models (Proc In order to standardize the conditions experienced by glimmix) using Maximum likelihood models with a Laplace one male across all three female types, we prevented actual approximation, binomial error structure, and a logit link spawning during the experimental trials. This was done by function [58]. As an immediate test for PAMC, we first ran using long forceps to gently squeeze the female’s caudal “ecotype models” (one for each female response variable) peduncle if she entered the nest, which always precipitated that included the fixed effects of male ecotype, female her immediate departure. After the trials for a given male or ecotype, and their interaction. In these models, random female were complete, the fish was anesthetized with buffered effects included male identity (nested within male ecotype) 4 International Journal of Ecology and the intercept, but not family (the data were insufficient interactions, showed multicollinearity and reduced model to do so for binomial response variables). Also not included stability. We therefore ran four sets of covariate models, in the analyses presented here was the order of presentation each containing male ecotype, female ecotype, and female of female ecotypes to a given male because this term was ecotype × covariate interactions and two covariates: (a) never significant. Adjusted post hoc Tukey tests based on display and RSD, (b) display and ASD, (c) aggression and least square means were used to assess pairwise differences RSD, or (d) aggression and ASD, but excluding the male among different ecotypes when relevant. To strengthen our ecotype × female ecotype interaction. Although we do not conclusions, we confirmed our results for nest entry (the conduct formal model comparisons [59], but run several core measure of mating isolation) also in two other ways. models on the same data set, we report AICc scores for the First, by scoring each female response variable from 0 (none ecotype models and for the final (reduced) covariate models of the responses) to 3 (nest entry alone or together with to allow some context to their relative explanatory power headup and/or nest inspection) and conducted a generalized while discounting for model complexity [59]. Here those linear mixed model with an ordered multinomial response. models having clearly lower AICc values were considered to Second, we treated both lake and outlet fish as the same have stronger support [59]. ecotype (“lake”) and analyzed the ecotype model comparing Inlet versus “Lake”. However, as the results in both of these 2.5. Literature Survey. To allow more direct comparisons analyses were similar to our present analyses, these results are of our results with those from other studies, we also sur- not further reported. veyed the published literature for studies that have explic- We next considered divergence in traits (body size and itly tested for PAMC between divergent threespine stick- male behavior) that might influence mating patterns. In our leback. Here we included studies that used comparable previous paper [51], we analyzed male behavior in more experimental methodology (either a choice or no-choice detail, but not its influence on female preferences (which we setting) and response variables (nest inspection and nest do here). As a few males included in [51] were not included entry/spawning). The details for this survey are given in Table in our mating trials, we here confirm ecotype specific S1. trait values and their statistical main effects (Table 2 and Table S2). Body size (log-transformed) and male behaviors 3. Results (square-root-transformed) were analyzed separately. Body size divergence (individual standard length) among ecotypes 3.1. Ecotype Models: Testing for PAMC. Neither head up was analyzed within each sex with mixed model analyses nor nest inspection rates depended on male ecotype, female of variance (Proc mixed), where fixed effects were male (or ecotype, or the male ecotype × female ecotype interac- female) ecotype, and family (nested within ecotype) (nesting tion (Figures 1(a) and 1(b), Table 1(a)). Nest entry was being possible for the continuous data used here). Male marginally influenced by male ecotype, and significantly by behavioral traits were aggression (frequency of aggressive female ecotype, but not by a male ecotype × female ecotype behaviors during a given mating trial) and display (frequency interaction (Figure 1(c), Table 1). Outlet males tended to of display behaviors during a given mating trial). Finally, we have higher mating success than either Inlet or Lake males ran “covariate models” to test whether mating is nonrandom (Figure 1(c)); however, these effects were not significant after with respect to male phenotype, and whether female eco- Tukey adjustment (all pairwise P>0.1). Inlet females were types have diverged in preferences for body size and male more likely to mate than either Lake (Tukey P = 0.044) or behavior. The two body size metrics used as covariates were Outlet females (Tukey P = 0.087, Figure 1(c)). Thus, genetic RSD and ASD, and the two male behavior variables were rates differences between ecotypes may exist in their propensity to of aggression and display, as defined above. As for the ecotype solicit nest entry (males) or enter nests (females), but these models, these covariate models included male identity as a differences did not generate PAMC. random effect (subject), but did not include family effects or the order of presentation of female ecotypes to males. Two 3.2. Trait Divergence. Both for males and females, Lake modifications to the data were necessary in these covariate and Outlet fish were similar in size and larger than Inlet models. First, in analyses of head up and nest entry one fish (Table 2). In the between-type mating trials, Inlet fish outlier had to be removed to achieve convergence. Second, therefore tended to be smaller than Lake or Outlet fish a few missing values for body size (N = 3) or behavior (combinations IL, LI, IO, and OI in Figure 2). Confirming (N = 1) had to be estimated from the ecotype means. These results from Delcourt et al. [51], a significant male ecotype modifications had no effect on our conclusions. effect in aggression and display, and a near significant female Covariate model analyses started with full models that ecotype effect in display, (Table 2) was present because Lake includedfixedeffects of male and female ecotype, male phe- and Outlet males were more aggressive but displayed less notype covariates (behaviors and body size), and all two-way than Inlet males and because Inlet females tended to elicit interactions between female ecotype and male phenotype. more displays from males than did Lake or Outlet females We then generated reduced models by removing sequentially (Table 2). Neither RSD nor ASD had a significant effect on nonsignificant (P>0.1) terms involving covariates. Because male aggressive behavior (both P>0.2, Table S2), whereas display and aggression were strongly correlated and distinct RSD had a significant effect on male display behavior: males among the male ecotypes, covariate models including both displayed more towards relatively smaller females (β ± S.E.: display and aggression, or male ecotype × female ecotype 1.925 ± 0.662, P = 0.005). However, even correcting for International Journal of Ecology 5

100 100

75 75 (%)

(%)

50 50 inspection

Head

25 Nest 25

0 0 Inlet Outlet Lake Inlet Outlet Lake Female ecotype Female ecotype Male ecotype Male ecotype Inlet Lake Inlet Lake Outlet Outlet (a) (b) 100

75 (%)

50 entry

Nest 25

0 Inlet Outlet Lake Female ecotype Male ecotype Inlet Lake Outlet (c)

Figure 1: Frequencies of (a) head up posture, (b) nest inspection, and (c) nest entry for different combinations of male and female ecotype of Misty Inlet, Outlet, and Lake threespine stickleback. The bars present different male ecotypes and are means ± S.E. of binomial frequencies (yes/no). body size, the male × female ecotype interaction was never inspection rates with greater male aggression, whereas there significant (P>0.7 in all cases, Table S2). was no such trend for Inlet females (Table 3). For nest entry, females showed greater responses to males with greater 3.3. Covariate Models: Testing for Female Trait Preference. We display or aggression (Tables 1 and 3). Females also appeared here summarize the significant effects of covariates, with to more frequently enter nests in combinations where males details and nonsignificant effects appearing in Tables 1 and were relatively larger (Figures 2(b) and 2(c)), but neither of 3. Females showed the head up response more frequently the size metrics had statistically significant effects (Table 1). towards males that displayed more, and marginally more Significant female ecotype effects arose in the covariate towards males that were relatively smaller than females models for head up and nest entry: Outlet females showed (Figure 2, Tables 1 and 3). Females also inspected more fre- stronger head up response (models a, b, and c, Table 1)and quently nests of males that showed higher rates of display or Inlet females tended to have higher nest inspection rates aggression (Table 3). Moreover, there was a near significant than Outlet females (model c, Table 1) (Figures 1(a) and female ecotype × aggression interaction on nest inspection 1(b)). For nest entry, the male and female ecotype effects (Table 1): both Lake and Outlet females showed higher nest seen in the ecotype models remained qualitatively similar in 6 International Journal of Ecology

Table 1: Generalized linear mixed model results for (I) ecotype and (II) covariate models testing for effects of male ecotype (meco), female ecotype (feco), and relevant covariates on head up, nest inspection, and nest entry. Mid stands for male identity. Covariates included in the initial full models were (a) display and RSD, (b) display and RSD, (c) aggression and ASD, or (d) aggression and ASD. Ecotype model and final (reduced model) AICc are given to allow inference of relative support of different models, whereby lower (difference >2) is better [59]. See Table 3 for covariate slopes. Significant effects are highlighted in bold and marginally significant (P<0.1) in italics.

Female response Model type Head up Nest inspection Nest entry (I) Ecotype models Source AICc 130.79 144.12 135.12 Random Variance ± S.E. Variance ± S.E. Variance ± S.E. Mid (meco) 1.518 ± 1.481 0.046 ± 0.493 0.749 ± 0.999 Fixed ndf ddf F P ddf F P ddf F P Meco 2 38 0.29 0.725 38 1.55 0.226 39 2.78 0.075 Feco 2 62 0.00 1.000 62 1.18 0.314 63 3.53 0.035 Meco × Feco 4 62 0.19 0.943 62 0.19 0.942 63 0.65 0.630 (II) Covariate models (a) Display and RSD Source AICc 98.87 122.33 103.19 Random Variance ± S.E. Variance ± S.E. Variance ± S.E. Mid (meco) 3.414 ± 4.455 0.435 ± 0.774 1.586 ± 2.013 Fixed ndf ddf F P ddf F P ddf F P Meco 2 38 1.18 0.318 38 2.59 0.088 38 3.21 0.052 Feco 2 63 2.81 0.068 65 1.31 0.276 66 3.32 0.042 Display 1 63 5.19 0.026 65 8.50 0.005 66 6.65 0.012 RSD 1 63 3.02 0.087 —— (b) Display and ASD Source AICc 103.53 Same final model as in a Same final model as in a Random Variance ± S.E. Mid (meco) 1.449 ± 1.732 Fixed ndf ddf F P Meco 2 38 0.01 0.988 Feco 2 64 3.91 0.025 Display 1 64 7.98 0.006 (c)Aggression and RSD Source AICc 109.80 122.36 109.46 Random Variance ± S.E. Variance ± S.E. Variance ± S.E. Mid (meco) 0.524 ± 0.940 0.012 ± 0.611 0.185 ± 0.784 Fixed ndf ddf F P ddf F P ddf F P Meco 2 38 2.84 0.071 38 0.61 0.546 38 3.92 0.072 Feco 2 64 4.93 0.010 63 4.01 0.023 66 4.34 0.017 Aggression 1 64 9.47 0.003 63 6.50 0.013 66 10.35 0.002 Feco × Aggr. — 63 3.11 0.051 — (d) Aggression and RSD In all cases same final models as in c thecovariatemodels(i.e.,Inletfemalesweremorelikelyto Lake and Outlet males show more aggressive courtship [51] enter nests and Outlet males to solicitate more nest entries, than Inlet males, but males of all three ecotypes display Figure 1 and Table 1), though the pairwise differences were more towards relatively smaller (Inlet) females. Third, female not statistically significant after Tukey adjustment (Table 3). mate choice was nonrandom with respect to male phenotype: some covariates were highly significant and the covariate 3.4. Summary of Main Findings. Taken together our analyses models clearly had more support (much lower AICc scores) on female mating preferences, and those on male courtship than did the ecotype models. Fourth, females tended to differ behavior (see also [51]), yielded five basic conclusions. First, in trait preferences: Lake and Outlet females tended to prefer there is no evidence for PAMC among the ecotypes. Second, more aggressive males (nest inspection), but this was not International Journal of Ecology 7

LO 100 100 IO 80 OI 80 (%)

OO (%) II OI IL OL 60 60 LL frequency

frequency LI 40 OL 40 up

entry IO OO 20 II 20 Head Nest IL LO LL 0 0 −15 −10 −50 5 10 −15 −10 −50 5 10 RSD (mm) RSD (mm) (a) (b) 100

80 (%)

60 frequency LI 40 OL entry II OO IO 20 Nest LO IL LL 0 051015 ASD (mm) (c) Figure 2: The relationships between relative (RSD) or absolute (ASD) body size difference of males and females and head up (a) or nest entry (b, c) frequencies in nine ecotype combinations of Misty threespine stickleback. Each dot represents the within combination average (%) against the size difference (RSD: male length minus female length and ASD: larger sex minus smaller sex). The ecotype combination is given in letters (I: Inlet, O: Outlet, L: Lake, first letter refers to male ecotype, second to female ecotype). See Tables 1 and 3 for covariate model test statistics.

Table 2: Ecotype-specific LS means ± S.E for body size (mm, back transformed values with upper and lower S.E. in brackets) and sqrt (male behaviour) for Misty Inlet (I), Outlet (O), and Lake (L) stickleback, with ecotype main effects and Tukey adjusted pairwise comparisons. Trait Male display Male display Female size Male size Male aggression (male ecotype) (female ecotype) Inlet 59.04 (0.54/0.54) 55.77 (0.86/0.85) 1.381 ± 0.208 3.424 ± 0.270 3.074 ± 0.204 Ecotype Outlet 69.81 (0.57/0.56) 65.16 (0.93/0.92) 2.304 ± 0.206 2.413 ± 0.266 2.934 ± 0.206 Lake 68.48 (0.63/0.62) 65.20 (0.94/0.93) 2.454 ± 0.225 2.730 ± 0.293 2.559 ± 0.203 F . = 102.19 F . = 36.04 F . = 7.54 F . = 3.69 F . =2.09 Ecotype main effect 2,7 24 2,5 41 2,36 7 2,38 5 2,66 2 P <0.001 <0.001 0.002 0.034 0.059 Ivs.L ∗∗∗ ∗∗ ∗∗ ∗ ◦ Pairwise Ivs.O ∗∗∗ ∗∗ ∗∗ ns ns comparisons Lvs.Onsnsnsnsns P: ◦ < 0.1, ∗ < 0.05, ∗∗ < 0.01, ∗∗∗ < 0.001, ns > 0.1. seen for Inlet females. Fifth, there was no evidence for size in contrast to most other studies, nest entry rates where assortative mate choice: female responses were only weakly somewhat higher for different ecotype matings than same affected by gender differences in body size (effect of RSD on ecotype matings in our study. head up). The survey of other PAMC tests in stickleback (Figure 3 4. Discussion and Table S1) revealed that the mating frequencies in our study fell within the range of those in other studies, though We found no evidence of positive assortative mate choice being on the lower end for nest inspection. Moreover, (PAMC) between lake and stream stickleback from the 8 International Journal of Ecology

Table 3: Ecotype main effect LS means ± S.E. and covariate slopes ± S.E. for head up, nest inspection, and nest entry in the Misty Lake (L), Inlet (I), and Outlet (O) three spine stickleback. The values are from final (reduced) covariate models in Table 1.Forcovariateeffects, common slopes are given when there was no significant female ecotype × male trait interaction, whereas female ecotype specific slopes are given when these interactions were at least nearly significant. Models refer to models with (a) display and RSD, (b) display and ASD, (c) aggression and RSD, or (d) aggression and ASD as covariates in initial full model, with results of Tukey adjusted pairwise comparisons given below. Female response Model Head up Nest inspection Nest entry (a) Display and RSD Ecotype main effects Inlet 2.556 ± 1.317 −0.926 ± 0.458 −0.437 ± 0.353 Female ecotype Outlet 4.451 ± 1.701 −1.905 ± 0.591 −2.419 ± 0.914 Lake 0.945 ± 0.821 −1.034 ± 0.477 −2.349 ± 0.841 Ivs.L ns ns ◦ Pairwise comparisons Ivs.O ns ns ◦ Lvs.O ◦ ns ns Inlet 0.728 ± 1.076 −2.388 ± 0.745 −3.651 ± 1.318 Male ecotype Outlet 3.550 ± 1.471 −0.470 ± 0.472 0.129 ± 0.637 Lake 3.682 ± 1.724 −1.389 ± 0.523 −1.684 ± 0.755 Ivs.L ns ns ns Pairwise comparisons Ivs.O ns ◦∗ Lvs.O ns ns ns Display 1.998 ± 0.877∗ 0.931 ± 0.319∗∗ 1.622 ± 0.629∗ Slopes RSD −6.768 ± 3.898◦ —— (b) Display and ASD Ecotype main effects Inlet 0.706 ± 0.512 Final model same as in a Final model same as in a Female ecotype Outlet 4.122 ± 1.247 Lake 1.402 ± 0.656 Ivs.L ns Pairwise comparisons Ivs.O ∗ Lvs.O ∗ Slopes Display 1.373 ± 0.486∗∗ 4.977 ± 1.711∗∗ Final model same as in a (c) Aggression and RSD Ecotype main effects Inlet 0.779 ± 0.432 −0.587 ± 0.365 −0.132 ± 0.386 Female ecotype Outlet 3.560 ± 0.943 −2.029 ± 0.709 −1.574 ± 0.557 Lake 0.632 ± 0.454 −1.389 ± 0.523 −2.423 ± 0.696 Ivs.L ns ns ns Pairwise comparisons Ivs.O ∗ ns ns Lvs.O ∗∗ ns ◦ Inlet 2.863 ± 0.773 −1.104 ± 0.488 −1.091 ± 0.519 Male ecotype Outlet 1.154 ± 0.574 −1.161 ± 0.502 −0.661 ± 0.480 Lake 0.954 ± 0.516 −1.741 ± 0.542 −2.378 ± 0.674 Ivs.L ◦ ns ∗ Pairwise comparisons Ivs.O ns ns ◦ Lvs.O ns ns ns Slopes Aggression 1.047 ± 0.340∗∗ 1.135 ± 0.352∗∗ Inlet — −0.035 ± 0.436 — Female ecotype Outlet — 1.817 ± 0.787∗ — Lake — 1.023 ± 0.415∗ —— (d) Aggression and ASD Finalmodelssameasinc P: ◦ < 0.1, ∗ < 0.05, ∗∗ < 0.01, ∗∗∗ < 0.001. International Journal of Ecology 9

100 100 %) 75 75 %)

ecotype

spawning 50 50 or

inspection

Different entry Different

(nest 25 25 (nest

0 0 0 25 50 75 100 0 25 50 75 100 Same ecotype Same ecotype (nest inspection %) (nest entry or spawning %) (a) (b) Figure 3: The relationship between same ecotype and different ecotype rates of (a) nest inspection and (b) nest entry or spawning in previously published threespine stickleback studies (open or grey circles) and current study (solid symbols). Open circles: wild-caught fish, grey circles: lab reared fish. Solid triangles: LO or OL combinations, which are treated as same ecotype due to phenotypic and genetic similarity. The line represents the 1:1 line of same and different ecotype matings. The details of the included studies are given in Table S1.

Misty system. This was not surprising for Lake versus Second, laboratory mate choice experiments are always Outlet fish given their phenotypic and genetic similarity, conducted under artificial conditions and thus exclude and so we do not further discuss that result. By contrast, genotype-by-environment interactions that might be impor- the lack of PAMC between Lake and Inlet fish (which is tant in nature (e.g., [49, 60]). Regardless, our experimental also found for a different lab cohort and with a different tanks and our no-choice design were very similar to those experimental design, [29]) is surprising. That is, PAMC used in several previous studies that did detect PAMC in might be expected given (1) strong adaptive divergence in stickleback (Table S1), and we also did not detect PAMC traits typically associated with mate choice in stickleback, in using a different experimental design [29]. Our fish orig- particular, body size and male behavior, (2) very low gene inated from multiple full-sib families, but we could not flow between these populations in nature, and (3) previous account statistically for family effects in our mate choice demonstrations of PAMC between other stickleback ecotypes analyses (the data was insufficient to do so), and we can (see Introduction). In the following sections, we first discuss therefore not fully discount that family effects would have factors that might influence this negative outcome, paying influenced our analyses. However, we think that any such particular attention to experimental design and biological effects would likely have been small compared to the factors suggested to be important in theory and in stickleback ecotype differences, as supported also by the observation and compare our studies with previous studies testing for that divergence in traits typically mediating mate choice PAMC in stickleback (Figure 3; Table S1). We then discuss (behavior, body size, and body shape) remained strong even how our results might relate to pressing questions in the after accounting for family effects [46, 51]. We therefore study of ecological speciation. suggest that differences in experimental venue and design are unlikely to have caused the difference between our results and 4.1. Experimental Design. Perhaps the different outcome of those documented in other stickleback systems. our study versus previous work on other stickleback systems Third, the present study and that of Raeymaekers et (Table S1) stems from differences in experimental design. al. [29] both used common-garden fish, whereas nearly First, our experiment might have lacked the statistical power all previous stickleback studies have used wild-caught fish necessary to detect PAMC. We do not think this is the case (Table S1, Figure 3, with an exception of [48]and[52]). for three complementary reasons: (1) our statistical control When wild-caught fish are used, observed mating patterns for individual male variation should be a powerful design for could reflect environmental (nongenetic) effects, such as detecting ecotype effects, (2) our sample sizes were similar to imprinting, prior experience, or some other form of phe- a number of previous studies that did detect PAMC (Table notypic plasticity (e.g., [48–50]). Studies testing for PAMC S1), and (3) the range of observed mating frequencies in would benefit from employing both common-garden and our study fell within those seen in other studies (Figure 3). wild-caught individuals because this allows complementary Moreover, although mate choice is clearly not random across inferences: common-garden individuals inform whether or the combinations, neither alternative ways of analyzing (see not genetic divergence in mate choice has occurred, whereas statistical analyses section) nor visual inspection of the wild-caught individuals might be more reflective of actual data (Figure 1 and Figure S1) suggest any hint of positive mating patterns in nature. In addition, complementary stud- assortative mating. ies on PAMC in the wild would allow strong inferences about 10 International Journal of Ecology the true extent of PAMC. Combined experiments of this sort exaggerated version of a stimulus that causes a strong could make important contributions to the current debates response) (e.g., [74, 75]). about the importance of plasticity in promoting versus Another possible magic trait would be male reproductive constraining ecological speciation and adaptive radiation behavior, which is divergent between the ecotypes and [61–65]. influences mate choice. In particular, Lake and Outlet females tended to more frequently inspect the nests of more 4.2. The Role of Trait Divergence. In the context of ecological aggressive males, and Lake and Outlet males were the most speciation, PAMC is expected to arise as a result of adaptive aggressive. However, even this correlated divergence of male divergence in traits [66]. We consider two important factors traits and female preferences was not sufficient to generate that might contribute to this association. First, it has PAMC. It is of course possible that the role of male trait been argued that substantial progress toward ecological divergence and correlated female preference may only be speciation, whether through PAMC or other reproductive apparent in more natural conditions. For example, if a more barriers, requires either strong divergent selection on one aggressive male courtship reflects his ability to defend his ecological/trait dimension or modest-to-strong divergent territory and offspring (e.g., [54, 71]), females might favor selection along multiple ecological/trait dimensions [17]. more aggressive males in the lake environments (i.e., high Second, the evolution of mating isolation during ecological population densities and likely higher predation and egg speciation is more likely when the traits undergoing adaptive raiding risk) in nature. Or maybe a distinction is needed divergence and the traits influencing mate choice are geneti- between potential magic traits that are “trivial” versus cally associated [67, 68], so-called “magic traits” [69, 70]. “important,” depending on just how much they contribute With respect to the strength and dimensionality of to mating isolation [70]. It is also possible that interactions divergent selection, Misty Lake-Inlet stickleback should be between divergent natural and divergent sexual selection well endowed: they differ dramatically in a number of traits acting on traits, or selection acting in different directions in thought to be under divergent selection. These include body males and females, (e.g., [11, 15, 25]) may have undermined size, morphological traits related to predator defense (e.g., the evolution of PAMC. spine length and male color), foraging (e.g., mouth shape and gill raker number and length), swimming performance 4.3. Relevance for Key Questions in Ecological Speciation. (e.g., sustained and burst swimming and maneuverability), Assuming that PAMC really is absent between Misty Lake and male reproductive behavior (e.g., nest construction, and Inlet stickleback, as our work suggests, what might aggression, and courtship) [29, 37, 38, 40, 46, 51, 53]. this, together with the frequent evidence for PAMC in Moreover, common-garden fish maintain divergence in these other stickleback systems, tell us about ecological speciation? traits (see above citations), making them relevant in the One question relates to the geographical context: close present experiment. Given all of these trait differences, physical proximity between groups under divergent selection it seems unlikely that the apparent lack of PAMC is the (i.e., sympatry or broad parapatry) can both constrain result of insufficient divergent selection in either strength or and promote the evolution of assortative mating. The dimensionality. Even strong and multifarious selection will potential constraint occurs when high gene flow hinders the not inevitably lead to PAMC. It is theoretically possible that development of a genetic association between adaptive traits lack of sufficient time for the evolution of mating isolation and preferences for those traits (recombination: [43, 76]). could contribute to the lack of PAMC. However, we do not The potential promoter comes in the form of direct [77] think this is the case here, as divergence times based on or indirect (e.g., “reinforcement”; [78]) selection to avoid neutral markers in the Misty system are similar to those mating with the opposite type. in other lake/stream systems that show similar patterns of Previous work on stickleback suggests both possibilities. trait divergence [30, 32, 38]. Likewise, the time frame for On the constraint side, high gene flow has clearly con- colonization is similar in several types of stickleback ecotype strained adaptive divergence and progress toward ecological pairs, which do show PAMC (i.e., after the last ice age, [21]). speciation in some lake/stream pairs [31–33, 41]. On the With respect to magic traits, the most widely discussed promoting side, sympatry seems to enhance PAMC: PAMC possibility for stickleback is body size (Table S1 and refer- is typically present between benthic and limnetic stickleback ences therein). As noted above, this trait differs markedly (Table S1), but it is higher when they come from the same between Misty Lake and Inlet fish in the wild and in the lake [5, 71] than from different lakes [5, 9]. In our case, lab. However, the suggested “mate with your own size” the Lake and Inlet stickleback populations are parapatric, rule for stickleback (e.g., [7, 24]) was clearly absent in our but our particular inlet sampling site was relatively far experiment (Figure 2): males of all types tended to display from the lake where gene flow from the lake population is more towards smaller (Inlet) females [51] and nest entry probably very low [47]. This suggests that PAMC may be was more frequent when males were larger than females less likely to evolve in this essentially allopatric situation. (Figure 2(b)). Possible explanations for these patterns are This hypothesis could be tested in lake/stream systems by that (1) larger females might be avoided because they more conducting experiments with stream stickleback both from frequently cannibalize the eggs in male nests [71], (2) larger very near and far to the lake. males are of higher quality within both habitats (e.g., [72, Another question relates to the number and type of 73]), and (3) females have retained ancestral preferences reproductive barriers that drive ecological speciation [79, for large males or show a “supernormality” response (i.e., 80]. That is, ecological speciation might sometimes be driven International Journal of Ecology 11 by only a single barrier that renders all other barriers Schluter, Karen Faller, and the personnel of the McGill unimportant. Such a barrier would have to be strong and Phytotron. All experiments were conducted in accordance symmetric and would most likely occur early in the life cycle. with animal use protocols at McGill University. The study Putative examples of such barriers include habitat choice was funded by Swedish Research Council (to K. Ras¨ anen)¨ (e.g., [81, 82]), divergent reproductive timing (e.g., [83, 84]), and the Natural Sciences and Engineering Research Council strong viability selection against migrants (e.g., [79, 85]), of Canada (to A. P. Hendry and L. J. Chapman). or postzygotic incompatibilities (e.g., [86]). Alternatively, ecological speciation could be the result of the joint evolution References of many reproductive barriers that are strong and symmetric only in aggregate. Along this vein, a number of studies have [1] D. Schluter, The Ecology of Adaptive Radiation,OxfordUniver- documented many reproductive barriers between diverging sity Press, Oxford, UK, 2000. taxa, such as in Mimulus monkeyflowers [87]andTimema [2] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology walking sticks [18]. However, it remains unclear to what Letters, vol. 8, no. 3, pp. 336–352, 2005. extent these multiple barriers contributed to reproductive [3] D. J. Funk, P. Nosil, and W. J. Etges, “Ecological divergence isolation during the course of speciation, as opposed to exhibits consistently positive associations with reproductive accumulating incidentally or after the fact. isolation across disparate taxa,” Proceedings of the National Work on the benthic/limnetic stickleback system suggests Academy of Sciences of the United States of America, vol. 103, that multiple reproductive barriers help to maintain repro- no. 9, pp. 3209–3213, 2006. ductive isolation ([21, 22]). However, the relative importance [4] D. J. Funk, “Isolating a role for natural selection in speciation: host adaptation and sexual isolation in Neochlamisus of each of these barriers during the process of divergence bebbianae leaf beetles,” Evolution, vol. 52, no. 6, pp. 1744–1759, is uncertain. For the lake/stream stickleback system, the 1998. problem is the opposite: no strong barriers have been found [5]H.D.Rundle,L.Nagel,J.W.Boughman,andD.Schluter, despite limited gene flow in nature. Instead, the various “Natural selection and parallel speciation in sympatric stick- potential barriers investigated so far seem asymmetric and lebacks,” Science, vol. 287, no. 5451, pp. 306–308, 2000. certainly incomplete, including limited dispersal ability (dis- [6] P. Nosil, B. J. Crespi, and C. P. Sandoval, “Host-plant adap- tance and beaver dams), habitat choice [38], selection against tation drives the parallel evolution of reproductive isolation,” migrants ([38]; K. Ras¨ anen¨ and A.P. Hendry, unpubl. data), Nature, vol. 417, no. 6887, pp. 440–443, 2002. temporal isolation (J.-S. Moore and K. Ras¨ anen,¨ unpubl. [7] J. S. McKinnon, S. Mori, B. K. Blackman et al., “Evidence for data), and sexual selection against hybrids [29]. No clear ecology’s role in speciation,” Nature, vol. 429, no. 6989, pp. evidence exists either for the role of postzygotic incom- 294–298, 2004. patibilities, as fertilization success and survival to hatching [8] J. W. Boughman, H. D. Rundle, and D. Schluter, “Parallel evo- are high in hybrids ([37], Ras¨ anen,¨ K., pers. obs.). Overall lution of sexual isolation in sticklebacks,” Evolution, vol. 59, reproductive isolation might therefore reflect a combination no. 2, pp. 361–373, 2005. of many different partial reproductive barriers. [9] T. H. Vines and D. Schluter, “Strong assortative mating be- Lake/stream stickleback thus present us with a conun- tween allopatric sticklebacks as a by-product of adaptation to different environments,” Proceedings of the Royal Society B, vol. drum that could ultimately be very informative of the factors 273, no. 1589, pp. 911–916, 2006. that influence progress toward ecological speciation. Clearly [10] R. B. Langerhans, M. E. Gifford, and E. O. Joseph, “Ecological more studies, replicated over multiple systems, are needed speciation in Gambusia fishes,” Evolution,vol.61,no.9,pp. on PAMC and other potential reproductive barriers. Of 2056–2074, 2007. particular utility is the availability of many independently [11] J. Ellers and C. L. Boggs, “The evolution of wing color: male derived parapatric population pairs that show various levels mate choice opposes adaptive wing color divergence in Colias of geographical separation, adaptive divergence, and gene butterflies,” Evolution, vol. 57, no. 5, pp. 1100–1106, 2003. flow [31–33]. Lake/stream stickleback therefore provide [12] H. D. Rundle, “Divergent environments and population bot- an opportunity to quantify the relative importance and tlenecks fail to generate premating isolation in Drosophila symmetry of multiple potential reproductive barriers during pseudoobscura,” Evolution, vol. 57, no. 11, pp. 2557–2565, the early stages of divergence [79, 80]aswellasinteractions 2003. between ecology and the geographical context in the devel- [13] A. Brelsford and D. E. Irwin, “Incipient speciation despite little opment of these barriers (e.g., [15, 19, 88]). This, and studies assortative mating: the yellow-rumped warbler hybrid zone,” on other taxa with similar levels of variation and replication, Evolution, vol. 63, no. 12, pp. 3050–3060, 2009. together with tests of PAMC in wild and common garden [14] L. Kwan and H. D. Rundle, “Adaptation to desiccation fails to populations, should take us a step closer to understanding generate pre- and postmating isolation in replicate Drosophila melanogaster laboratory populations,” Evolution, vol. 64, no. 3, the key features of ecological speciation. pp. 710–723, 2010. [15] A. K. Schwartz, D. J. Weese, P. Bentzen, M. T. Kinnison, and A. Acknowledgments P. Hendry, “Both geography and ecology contribute to mating isolation in guppies,” PLoS ONE, vol. 5, no. 12, Article ID The authors thank Jean-Sebastien Moore, Martin Turcotte, e15659, 2010. Angeliina Alopaeus, and the Friends of the Marble River [16] A. P. Hendry, “Ecological speciation! Or the lack thereof?” for help in the field. Laboratory work was facilitated or Canadian Journal of Fisheries and Aquatic Sciences, vol. 66, no. aided by Jihane Giraud, Jaclyn Paterson, Kate Hudson, Dolph 8, pp. 1383–1398, 2009. 12 International Journal of Ecology

[17] P. Nosil, L. J. Harmon, and O. Seehausen, “Ecological expla- [35] G. E. E. Moodie, “Predation, natural selection and adaptation nations for (incomplete) speciation,” Trends in Ecology and in an unusual threespine stickleback,” Heredity, vol. 28, pp. Evolution, vol. 24, no. 3, pp. 145–156, 2009. 155–167, 1972. [18] P. Nosil, “Divergent host plant adaptation and reproductive [36] T. E. Reimchen, E. M. Stinson, and J. S. Nelson, “Multivariate isolation between ecotypes of Timema cristinae walking differentiation of parapatric and allopatric populations of sticks,” American Naturalist, vol. 169, no. 2, pp. 151–162, 2007. threespine stickleback in the Sangan River watershed, Queen [19] E. B. Rosenblum and L. J. Harmon, ““Same same but dif- Charlotte Islands,” Canadian Journal of Zoology, vol. 63, pp. ferent”: replicated ecological speciation at white sands,” Evo- 2944–2951, 1985. lution, vol. 65, no. 4, pp. 946–960, 2011. [37] P. A. Lavin and J. D. McPhail, “Parapatric lake and stream [20] M. A. Bell and S. A. Foster, The Evolutionary Biology of the sticklebacks on northern Vancouver Island: disjunct distribu- Threespine Stickleback, Oxford University Press, New York, NY, tion or parallel evolution?” Canadian Journal of Zoology, vol. USA, 1994. 71, no. 1, pp. 11–17, 1993. [21] J. S. McKinnon and H. D. Rundle, “Speciation in nature: the [38] A. P. Hendry, E. B. Taylor, and J. D. McPhail, “Adaptive threespine stickleback model systems,” Trends in Ecology and divergence and the balance between selection and gene flow: Evolution, vol. 17, no. 10, pp. 480–488, 2002. lake and stream stickleback in the Misty system,” Evolution, [22] A. P. Hendry, D. I. Bolnick, D. Berner, and C. L. Peichel, vol. 56, no. 6, pp. 1199–1216, 2002. “Along the speciation continuum in sticklebacks,” Journal of [39]D.Berner,D.C.Adams,A.C.Grandchamp,andA.P.Hendry, Fish Biology, vol. 75, no. 8, pp. 2000–2036, 2009. “Natural selection drives patterns of lake-stream divergence [23] M. S. Ridgway and J. D. McPhail, “Ecology and evolution of in stickleback foraging morphology,” Journal of Evolutionary sympatric sticklebacks (Gasterosteus): mate choice and repro- Biology, vol. 21, no. 6, pp. 1653–1665, 2008. ductive isolation in the Enos Lake species pair,” Canadian [40] A. P. Hendry, K. Hudson, J. A. Walker, K. Ras¨ anen,¨ and L. J. Journal of Zoology, vol. 62, no. 9, pp. 1813–1818, 1984. Chapman, “Genetic divergence in morphology-performance [24] L. Nagel and D. Schluter, “Body size, natural selection, and mapping between Misty Lake and inlet stickleback,” Journal of speciation in sticklebacks,” Evolution, vol. 52, no. 1, pp. 209– Evolutionary Biology, vol. 24, no. 1, pp. 23–35, 2011. 218, 1998. [41] J. S. Moore, J. L. Gow, E. B. Taylor, and A. P. Hendry, “Quan- [25] G. M. Kozak, M. Reisland, and J. W. Boughmann, “Sex dif- tifying the constraining influence of gene flow on adaptive ferences in mate recognition and conspecific preference in divergence in the lake-stream threespine stickleback system,” species with mutual mate choice,” Evolution,vol.63,no.2,pp. Evolution, vol. 61, no. 8, pp. 2015–2026, 2007. 353–365, 2009. [42] K. Ras¨ anen¨ and A. P. Hendry, “Disentangling interactions [26] D. E. Hay and J. D. McPhail, “Mate selection in threespine between adaptive divergence and gene flow when ecology sticklebacks,” Canadian Journal of Zoology, vol. 53, pp. 441– drives diversification,” Ecology Letters, vol. 11, no. 6, pp. 624– 450, 1975. 636, 2008. [27] G. A.´ Olafsd´ ottir,´ M. G. Ritchie, and S. S. Snorrason, “Posi- [43] S. Gavrilets and A. Vose, “Dynamic patterns of adaptive tive assortative mating between recently described sympatric radiation,” Proceedings of the National Academy of Sciences of morphs of Icelandic sticklebacks,” Biology Letters, vol. 2, no. 2, the United States of America, vol. 102, no. 50, pp. 18040–18045, pp. 250–252, 2006. 2005. [28] F. C. Jones, C. Brown, and V. A. Braithwaite, “Lack of assortative mating between incipient species of stickleback from a [44] X. Thibert-Plante and A. P.Hendry, “Five questions on ecolog- hybrid zone,” Behaviour, vol. 145, no. 4-5, pp. 463–484, 2008. ical speciation addressed with individual-based simulations,” [29] J. A. M. Raeymaekers, M. Boisjoly, L. Delaire, D. Berner, Journal of Evolutionary Biology, vol. 22, no. 1, pp. 109–123, K. Ras¨ anen,¨ and A. P. Hendry, “Testing for mating isolation 2009. between ecotypes: laboratory experiments with lake, stream [45] X. Thibert-Plante and A. P. Hendry, “When can ecological andhybridstickleback,”Journal of Evolutionary Biology, vol. speciation be detected with neutral loci?” Molecular Ecology, 23, no. 12, pp. 2694–2708, 2010. vol. 19, no. 11, pp. 2301–2314, 2010. [30]C.E.Thompson,E.B.Taylor,andJ.D.Mcphail,“Parallel [46] D. M. T. Sharpe, K. Ras¨ anen,D.Berner,andA.P.Hendry,¨ evolution of lake-stream pairs of threespine sticklebacks “Genetic and environmental contributions to the morphology (Gasterosteus) inferred from mitochondrial DNA variation,” of lake and stream stickleback: implications for gene flow and Evolution, vol. 51, no. 6, pp. 1955–1965, 1997. reproductive isolation,” Evolutionary Ecology Research, vol. 10, [31] A. P. Hendry and E. B. Taylor, “How much of the variation no. 6, pp. 849–866, 2008. in adaptive divergence can be explained by gene flow? An [47] J. S. Moore and A. P.Hendry, “Both selection and gene flow are evaluation using lake-stream stickleback pairs,” Evolution, vol. necessary to explain adaptive divergence: evidence from clinal 58, no. 10, pp. 2319–2331, 2004. variation in stream stickleback,” Evolutionary Ecology Research, [32] D. Berner, A. C. Grandchamp, and A. P. Hendry, “Variable vol. 7, no. 6, pp. 871–886, 2005. progress toward ecological speciation in parapatry: stickleback [48] G. M. Kozak, M. L. Head, and J. W. Boughman, “Sexual across eight lake-stream transitions,” Evolution, vol. 63, no. 7, imprinting on ecologically divergent traits leads to sexual pp. 1740–1753, 2009. isolation in sticklebacks,” Proceedings of the Royal Society B, [33] D. Berner, M. Roesti, A. P. Hendry, and W. Salzburger, “Con- vol. 278, no. 1718, pp. 2604–2610, 2011. straints on speciation suggested by comparing lake-stream [49] G. M. Kozak and J. W. Boughman, “Learned conspecific mate stickleback divergence across two continents,” Molecular Ecol- preference in a species pair of sticklebacks,” Behavioral Ecology, ogy, vol. 19, no. 22, pp. 4963–4978, 2010. vol. 20, no. 6, pp. 1282–1288, 2009. [34] G. E. E. Moodie, “Morphology, life history and ecology of [50] J. H. Jennings and W. J. Etges, “Species hybrids in the lab- an unusual stickleback (Gasterosteus aculeatus) in the Queen oratory but not in nature: a reanalysis of premating isolation Charlotte Islands, Canada,” Canadian Journal of Zoology, vol. between Drosophila arizonae and D. mojavensis,” Evolution, 50, pp. 721–732, 1972. vol. 64, no. 2, pp. 587–598, 2010. International Journal of Ecology 13

[51] M. Delcourt, K. Ras¨ anen,¨ and A. P. Hendry, “Genetic and Trends in Ecology and Evolution, vol. 26, no. 8, pp. 389–397, plastic components of divergent male intersexual behavior in 2011. Misty lake/stream stickleback,” Behavioral Ecology, vol. 19, no. [71] A. Y. K. Albert and D. Schluter, “Reproductive character dis- 6, pp. 1217–1224, 2008. placement of male stickleback mate preference: reinforcement [52] T. Hatfield and D. Schluter, “A test for sexual selection on or direct selection?” Evolution, vol. 58, no. 5, pp. 1099–1107, hybrids of two sympatric sticklebacks,” Evolution, vol. 50, no. 2004. 6, pp. 2429–2434, 1996. [72] S. B. M. Kraak and T. C. M. Bakker, “Mutual mate choice in [53] J. A. M. Raeymaekers, L. Delaire, and A. P. Hendry, “Genet- sticklebacks: attractive males choose big females, which lay big ically based differences in nest characteristics between lake, eggs,” Animal Behaviour, vol. 56, no. 4, pp. 859–866, 1998. inlet, and hybrid threespine stickleback from the misty system, [73] U. Candolin and H. R. Voigt, “Correlation between male size British Columbia, Canada,” Evolutionary Ecology Research, vol. and territory quality: consequence of male competition or 11, no. 6, pp. 905–919, 2009. predation susceptibility?” Oikos, vol. 95, no. 2, pp. 225–230, [54] H. D. Rundle and D. Schluter, “Reinforcement of stickleback 2001. mate preferences: sympatry breeds contempt,” Evolution, vol. [74] G. E. E. Moodie, “Why asymmetric mating preferences may 52, no. 1, pp. 200–208, 1998. not show the direction of evolution,” Evolution, vol. 36, pp. [55] J. D. McPhail and D. E. Hay, “Differences in male courtship 1096–1097, 1982. in freshwater and marine sticklebacks,” Canadian Journal of [75] W. J. Rowland, “Mate choice and the supernormality effect Zoology, vol. 61, pp. 292–297, 1983. in female sticklebacks (Gasterosteus aculeatus),” Behavioral [56] S. A. Foster, “Understanding the evolution of behavior in Ecology and Sociobiology, vol. 24, no. 6, pp. 433–438, 1989. threespine stickleback: the value of geographic variation,” [76] J. Felsenstein, “Skepticism towards Santa Rosalia, or why are Behaviour, vol. 132, no. 15-16, pp. 1107–1129, 1995. there so few kinds of animals?” Evolution, vol. 35, pp. 124–138, [57] M. Ishikawa and S. Mori, “Mating success and male courtship 1981. behaviors in three populations of the threespine stickleback,” [77] M. R. Servedio, “Beyond reinforcement: the evolution of pre- Behaviour, vol. 137, no. 7-8, pp. 1065–1080, 2000. mating isolation by direct selection on preferences and post- [58] SAS Institute, SAS/STAT 9.2 User’s Guide,SASInstitute, mating, prezygotic incompatibilities,” Evolution, vol. 55, no. Cary, NC, USA, 2008. 10, pp. 1909–1920, 2001. [59] K. P. Burnham and D. R. Andersson, Model Selection and [78] M. R. Servedio and M. A. F. Noor, “The role of reinforcement Multimodel Inference: A Practical Information-Theoretical Ap- in speciation: theory and data,” Annual Review of Ecology, proach, Springer, New York, NY, USA, 2nd edition, 2002. Evolution, and Systematics, vol. 34, pp. 339–364, 2003. [60] E. Lewandowski and J. Boughman, “Effects of genetics and [79] P.Nosil, T. H. Vines, and D. J. Funk, “Perspective: reproductive light environment on colour expression in threespine stickle- isolation caused by natural selection against immigrants from backs,” Biological Journal of the Linnean Society,vol.94,no.4, divergent habitats,” Evolution, vol. 59, no. 4, pp. 705–719, pp. 663–673, 2008. 2005. [61] M. J. West-Eberhardt, Developmental Plasticity and Evolution, [80] D. B. Lowry, J. L. Modliszewski, K. M. Wright, C. A. Wu, Oxford University Press, New York, NY, USA, 2003. and J. H. Willis, “Review. The strength and genetic basis of [62] E. Crispo, “Modifying effects of phenotypic plasticity on reproductive isolating barriers in flowering plants,” Philosoph- interactions among natural selection, adaptation and gene ical Transactions of the Royal Society B, vol. 363, no. 1506, pp. flow,” Journal of Evolutionary Biology, vol. 21, no. 6, pp. 1460– 3009–3021, 2008. 1469, 2008. [81] J. L. Feder, S. B. Opp, B. Wlazlo, K. Reynolds, W. Go, and S. [63] R. Svanback,¨ M. Pineda-Krch, and M. Doebeli, “Fluctuating Spisak, “Host fidelity is an effective premating barrier between population dynamics promotes the evolution of phenotypic sympatric races of the apple maggot fly,” Proceedings of the plasticity,” American Naturalist, vol. 174, no. 2, pp. 176–189, National Academy of Sciences of the United States of America, 2009. vol. 91, no. 17, pp. 7990–7994, 1994. [64] D. W. Pfennig, M. A. Wund, E. C. Snell-Rood, T. Cruickshank, [82] D. J. Bolnick, L. Snowberg, C. Patenia, O. L. Lau, W. E. Stutz, C. D. Schlichting, and A. P. Moczek, “Phenotypic plasticity’s and T. Ingram, “Habitat choice contributes to adaptive diver- impacts on diversification and speciation,” Trends in Ecology gence between lake and stream populations of three-spine and Evolution, vol. 25, no. 8, pp. 459–467, 2010. stickleback,” Evolution, vol. 63, pp. 2004–2016, 2009. [65] X. Thibert-Plante and A. P. Hendry, “The consequences of [83] R. D. Alexander and R. S. Bigelow, “Allochronic speciation in phenotypic plasticity for ecological speciation,” Journal of field crickets, and a new species, Acheta veletis,” Evolution, vol. Evolutionary Biology, vol. 24, no. 2, pp. 326–342, 2011. 14, pp. 334–346, 1960. [66] D. Schluter, “Ecology and the origin of species,” Trends in [84] V. L. Friesen, A. L. Smith, E. Gomez-D´ ´ıaz et al., “Sympatric Ecology and Evolution, vol. 16, no. 7, pp. 372–380, 2001. speciation by allochrony in a seabird,” Proceedings of the [67] A. S. Kondrashov and F. A. Kondrashov, “Interactions among National Academy of Sciences of the United States of America, quantitative traits in the course of sympatric speciation,” vol. 104, no. 47, pp. 18589–18594, 2007. Nature, vol. 400, no. 6742, pp. 351–354, 1999. [85] S. Via, A. C. Bouck, and S. Skillman, “Reproductive isolation [68] J. D. Fry, “Multilocus models of sympatric speciation: bush between divergent races of pea aphids on two hosts. II. Sele- versus rice versus felsenstein,” Evolution,vol.57,no.8,pp. ction against migrants and hybrids in the parental environ- 1735–1746, 2003. ments,” Evolution, vol. 54, no. 5, pp. 1626–1637, 2000. [69] S. Gavrilets, Fitness Landscapes and the Origin of Species, [86] D. C. Presgraves, L. Balagopalan, S. M. Abmayr, and H. A. Orr, Princeton University Press, Princeton, NJ, USA, 2004. “Adaptive evolution drives divergence of a hybrid inviability [70]M.R.Servedio,G.S.V.Doorn,M.Kopp,A.M.Frame,and gene between two species of Drosophila,” Nature, vol. 423, no. P. Nosil, “Magic traits in speciation: “magic” but not rare?” 6941, pp. 715–719, 2003. 14 International Journal of Ecology

[87]J.Ramsey,H.D.Bradshaw,andD.W.Schemske,“Compo- nents of reproductive isolation between the monkeyﬂowers Mimulus lewisii and M. cardinalis (Phrymaceae),” Evolution, vol. 57, no. 7, pp. 1520–1534, 2003. [88] R. K. Butlin, J. Galindo, and J. W. Grahame, “Review. Sym- patric, parapatric or allopatric: the most important way to classify speciation?” Philosophical Transactions of the Royal Society B, vol. 363, no. 1506, pp. 2997–3007, 2008. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 725487, 11 pages doi:10.1155/2012/725487

Research Article How Facilitation May Interfere with Ecological Speciation

P. Liancourt,1, 2 P. C hol er , 3, 4 N. Gross,5, 6 X. Thibert-Plante,7 and K. Tielborger¨ 2

1 Department of Biology, University of Pennsylvania, 322 Leidy Labs, Philadelphia, PA 19104, USA 2 Department of Plant Ecology, University of Tubingen,¨ Auf der Morgenstelle 3, 72072 Tubingen,¨ Germany 3 Laboratoire d’Ecologie Alpine, CNRS UMR 5553, Universit´e de Grenoble, BP 53, 38041 Grenoble, France 4 Station Alpine J. Fourier, CNRS UMS 2925, Universit´e de Grenoble, 38041 Grenoble, France 5 INRA, USC Agripop (CEBC-CNRS), 79360 Beauvoir sur Niort, France 6 CEBC-CNRS (UPR 1934), 79360 Beauvoir sur Niort, France 7 National Institute for Mathematical and Biological Synthesis (NIMBioS), University of Tennessee, 1534 White Avenue, Suite 400, Knoxville, TN 37996, USA

Correspondence should be addressed to P. Liancourt, [email protected]

Received 13 July 2011; Revised 15 November 2011; Accepted 10 December 2011

Academic Editor: Rui Faria

Copyright © 2012 P. Liancourt et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Compared to the vast literature linking competitive interactions and speciation, attempts to understand the role of facilitation for evolutionary diversification remain scarce. Yet, community ecologists now recognize the importance of positive interactions within plant communities. Here, we examine how facilitation may interfere with the mechanisms of ecological speciation. We argue that facilitation is likely to (1) maintain gene flow among incipient species by enabling cooccurrence of adapted and maladapted forms in marginal habitats and (2) increase fitness of introgressed forms and limit reinforcement in secondary contact zones. Alternatively, we present how facilitation may favour colonization of marginal habitats and thus enhance local adaptation and ecological speciation. Therefore, facilitation may impede or pave the way for ecological speciation. Using a simple spatially and genetically explicit modelling framework, we illustrate and propose some first testable ideas about how, when, and where facilitation may act as a cohesive force for ecological speciation. These hypotheses and the modelling framework proposed should stimulate further empirical and theoretical research examining the role of both competitive and positive interactions in the formation of incipient species.

1. Introduction diversification has been largely overlooked. Very recently, a few empirical studies have attempted to bridge this gap [11– Over the last 20 years, there has been increasing recognition 13]. that besides competition, facilitation, that is, the amelio- The theoretical study of positive interactions in an ev- ration of biotic and/or abiotic conditions by neighboring olutionary context has an old history (e.g., [14]), but these organisms, is a key driver controlling assembly and dynamics models only consider mutualism, a two-species interaction of communities ([1], see [2] for recent reviews). The impact that benefits both partners [15]. However, mutualism is char- of facilitation in a community ecology context has been now acterized by balanced reciprocal benefits for species involved widely accepted, but its evolutionary consequences have been in a one-to-one relationship, while facilitation encompasses rarely explored [2, 3]. There also has been an increasing a larger spectrum of positive interactions between organisms awareness that positive and negative interactions usually [2, 16]. Namely, it commonly includes one-to-many inter- cooccur in nature, and recent works have turned to examine actions where one organism ameliorates habitat conditions conditions under which the outcome of biotic interactions for several beneficiary organisms and many-to-one inter- is positive, negative, or neutral [4–6]. Yet, while there is a actions where an organism benefits from diffuse facilitation long-standing literature linking competitive interactions and by neighbors. For example, facilitation between plants speciation [7–10], the role of facilitation for evolutionary occurs when neighboring vegetation or a nurse plant (more 2 International Journal of Ecology generally called a benefactor) ameliorates environmental our arguments (Box 1). Lastly, we call for a new body of conditions directly, through the provision of additional empirical and theoretical works to test the presented hypoth- resources or shelter from extreme conditions (e.g., high or eses (Box 2). The ideas and hypotheses developed in this low temperatures), or indirectly through protection from paper are biased towards the plant kingdom (as is the facil- herbivores or via attraction of pollinators [16]. Since facili- itation literature in general, e.g., [2, 16]) but should apply tation generates local variation in environmental conditions, equally to other kingdoms. For example, there is now some it should also affect the manner in which organisms respond evidence showing the importance of facilitation for a wide and adapt to their environment. However, it is important to range of organisms such as terrestrial animals [27], marine realize that facilitation differs intrinsically from microhabitat invertebrates [28], fungi [29], bacteria [3], and viruses [30]. amelioration due to abiotic components of the system (e.g., rock, ponds, or nutrient patches) because the source 2. Facilitation and Niche Theory, Moving from of environmental heterogeneity is a living organism. For Species to Genotypes instance, the distribution of favourable microhabitats may change at the time scale of ecological responses because of the Facilitation is likely to be as common as competition along population dynamics of benefactor species [17]. Addition- environmental gradients [2]. It is generally hypothesized to ally, the net effect of benefactors can shift from facilitation increase in importance as environmental severity increases to competition due to interannual resource dynamics (e.g., ([1], but see [31]) and is likely to enhance species diversity in change in rainfall pulse [18]). harsh habitats [31–34]. It is now recognized that facilitation The growing interest for eco-evolutionary dynamics, that can expand the distribution range of a species at its harsher is, the interplay between ecological and evolutionary pro- end, or marginal habitat [35], whereas competition generally cesses [19–22], demonstrates the need to study both the reduces it at the favorable end of the environmental gradient evolutionary consequences of ecological processes and the [4, 5, 36, 37]. By widening the distribution range of certain effect of evolutionary processes on ecological responses. In a species not adapted to stressful conditions, facilitation allows recent study, Michalet et al. [17] provided one specific aspect them to persist with subordinate [4, 5, 33, 34, 36]oreven of the problem from the benefactor’s point of view. Namely, dominant status in environments, where without facilitation they could show that ecotypic variation in benefactor traits they would not be able to persist [37]. Therefore, within may modify the manner in which they affect understory a particular community, the ability of a species to be species. Viceversa, the understory also affected fitness of the facilitated is strongly linked with its own niche [4, 5, 37] benefactor. Here, we tackle this problem by looking at the in that only species deviated from their niche optima are other side of this coin and from a more general perspective, likely to be facilitated. More specifically, competitive species that is, the evolutionary consequences of facilitation on dominating favorable environments are more affected by diversification of beneficiaries. facilitation as stress increases than stress-tolerant species We aim at complementing the vast literature on compet- adapted to harsher ecological conditions. Conversely, stress- itive effects on evolutionary diversification by a conceptual tolerant species are competitively excluded as environmental model addressing the role of facilitation on distribution stress decreases [31]duetoatrade-off between competitive range, local adaptation, and genetic structure of beneficiary ability and stress-tolerance (see [5]). species. In particular, we discuss the likely effect of facil- Here, we apply these concepts developed on a species itation on microevolutionary processes and ultimately on level to variation among incipient species. Namely, we pro- ecological speciation. Ecological speciation can be defined as pose to establish a missing link between facilitation as a a particular type of speciation, where reproductive isolation force structuring communities and one affecting population evolves between populations as a result of ecologically based structure. Due to the fact that different genotypes within a divergent selection [23–25]. In this context, the interplay species may differ in their resource use and stress tolerance between biotic and abiotic factors is of paramount impor- [38, 39], they can be characterized by contrasting ecological tance and, not surprisingly, the attempts to link ecological tolerances and different optima. For example, local adapta- and evolutionary processes are the most tantalizing [3, 17, tion along environmental gradients [40]willleadtoecotype- 26]. specific “realized niches.” Note that this concept is commonly We do not present an exhaustive list of all potential ev- incorporated in models of resource competition (e.g., [41, olutionary consequences of facilitation but instead propose 42]). On the one hand, genotypes exhibiting traits linked an alternative to traditional views on the role of biotic inter- to stress tolerance should be selected for, and thus be more actions for evolutionary processes. Our overarching hypoth- common at the margin of the species’ realized niche (harsh esis is that, contrary to competition, facilitation may impede conditions). On the other hand, genotypes enabling larger ecological speciation by maintaining gene flow between competitive response ability and low stress tolerance should incipient species and preventing niche partitioning. Alter- dominate in populations in the center of the species niche, natively, we also discuss under which more specific circum- that is, more in favorable environments [43, 44]. stances facilitation could act as a stepping stone to promote Any ecological process that has the potential to signif- ecological speciation. First, we highlight the role of facili- icantly impact gene flow in marginal habitats may play a tation in the context of niche theory. Secondly, we present key role in ecological speciation [45, 46]. In that context, how facilitation is likely to interfere with microevolutionary dispersal-related processes have been well investigated [43]. processes and propose a modeling framework to illustrate Here, we argue that the likelihood that core genotypes can International Journal of Ecology 3

Divergent selection of T2 in E2 character displacement selection for T in E 1 1 and reproductive isolation (a) Fitness

E2 Trait attributeT1 Marginal Habitat E1 T2

Core Habitat

dispersal into facilitation of maladapted morphs marginal (b) maintains gene ﬂow among morphs habitat E2

Niche conservatism: (c) facilitation of maladapted morphs

+ +

habitat dynamics

Figure 1: How facilitation may interfere with ecological speciation and niche conservatism. These 3D plots sketch ongoing disruptive selection and trait divergence of incipient species. The vertical axis represents the relative fitness of genotypes in a trait-ecological space. We assumed that ecologically marginal conditions (E2) also correspond to peripheral areas. Solid arrows are for gene flow including dispersal of propagules, individuals, and sexual reproduction. Dotted lines denote reproductive isolation. A common starting point is immigration of genotypes from the favorable part of the gradient to marginal/peripheral habitats. (a) Diversifying selection without facilitation, and reproductive isolation that eventually arises as a by-product of trait divergence. (b) Maladapted forms persist in marginal habitats because of facilitation and a maintained gene flow among genotypes prevents ecological speciation. (c) Facilitation allows maladapted forms to persist in a changing environment, that is, niche conservatism.

persist in a species’ ecological margin without facilitation is are distributed. Genotypes are supposed to differ by a key particularly low. In the likely scenario of net dispersal from trait under selective pressure (e.g., a stress-tolerance-related core (source) to marginal habitats (sinks) [43], facilitation trait). Local adaptation to marginal habitats is possible if may enhance the establishment of core genotypes in habitats gene flow from core habitats is limited due to pre- or postzy- where they otherwise would not be able to persist. This gotic barriers ([47], but see [48]). For example, ecological will have crucial consequences for adaptation to marginal filtering is a prezygotic barrier that will select genotypes habitats, and in turn for species diversification along envi- best adapted to the local conditions and will reduce viability ronmental gradients, which we explain below. of the immigrants [49]. Postzygotic barriers may include reduced fitness of hybrids related to stress intolerance or to 3. Consequence of Facilitation for Gene Flow competitive displacement by more adapted forms [50, 51]. If ffi and Genetic Structure of Beneficiaries these mechanisms are su cientlystrongtoreducegeneflow between adapted and maladapted forms, trait divergence A first mechanism by which facilitation may impact adaptive and ultimately reproductive isolation between populations diversification is in enabling maladapted genotypes/forms to may be observed (Figure 1(a)). An alternative scenario persist and reproduce outside of their niche optima. Figure 1 is depicted in Figure 1(b), where maladapted forms are depicts two habitat types with contrasting environmental facilitated by neighboring vegetation, that is, benefactors conditions (core and marginal), where different genotypes allow establishment, growth, and reproductive success of 4 International Journal of Ecology nonoptimal genotypes. In this case, facilitation may cause be tested with a spatially and genetically explicit modeling allele frequencies in the sink (marginal) to be similar to framework derived from Kirkpatrick and Barton [58]and those of the source (core) habitat. The result would be stable Bridle et al. [62] (Box 1). Our simulations confirm that this coexistence between two genotypes, a core and a marginal modeling framework represents a flexible tool to explore the genotype, and thus an overall increased genetic diversity important set of parameters and scenarios that we believe and decreased inbreeding but also greater genetic swamping, shouldbeinvestigated(Box2).Wechoseheretoexplorea that is, a decreased potential to adapt to the marginal con- reduced set of parameters, that is, size of the patch created ditions. Facilitation may thus be a cohesive force that tends by the benefactors and the environmental difference between to maintain gene flow between these populations and the core and the marginal habitat (Box 1). Our simulations counteracts diversifying selection. To our knowledge, only show that, by producing a mosaic of mild conditions in a one experimental field study has tested this hypothesis and harsh environment, benefactors increase gene flow between supported our conjecture. This study was conducted on the populations from core and marginal habitats [63], that is, annual grass Brachypodium distachyon in the Middle East. B. lower FST between populations for both neutral [64]and distachyon has a broad distribution range and occurs from selected loci (Figures 2(a) and 2(b), resp.). They thus ulti- Mediterranean (core habitat) to semiarid habitats (marginal mately prevent local adaptation in the harsh environment, as habitat). Ecotypic differentiation has been observed in this expressed in lower fitness in the marginal environment with species, with the semiarid ecotypes being more stress-toler- facilitation compared to a scenario without facilitation (see ant [44, 52, 53] and the Mediterranean ecotype having a Figure 2(c) comparing without benefactors versus smaller better ability to cope with competition [44]. When both patch size benefactors). ecotypes where transplanted to the semiarid environment, both were facilitated by the presence of shrubs but the 4. Benefactors and Disruptive Selection Mediterranean ecotype could only survive to reproduction underneath the nurse [13]. Facilitation, therefore, enabled Though facilitation is likely to be a cohesive force, we would a maladapted morph to persist and reproduce in the mar- also like to touch upon possible conditions under which ginal habitat as described above. positive interactions may contribute to, rather than impede, Another scenario would consider the secondary contact ecological speciation. This is most easily illustrated with of two populations that exhibit trait differentiation. Rein- particularly stressful environments, where benefactors such forcement, that is, selection against hybridization, and in- as cushion plants in alpine ecosystems or shrubs in water- trogression are two opposing forces that operate during limited ecosystems [16], represent islands of milder condi- this secondary contact phase, the former enhancing genetic tions or islands of fertility. At the community scale, this re- divergence and the latter having the potential to erase it [23]. sults in a mosaic of microhabitats determined by the Secondary contact zones often represent marginal habitats distribution and abundance of benefactors. Such systems for incipient species (e.g., [54]). Facilitation could favor the have provided the most conclusive evidence for facilita- cooccurrence of ecotypes according to the mechanism de- tion [65], even if the outcome of biotic interactions may scribed above and ultimately lead to increased gene flow strongly depend upon the particular location of the focal between ecotypes. Facilitation may also circumvent postzy- species within this mosaic. Arguably, the patchy distribution gotic barriers by increasing the fitness of introgressed forms. of biotically engineered favorable microhabitats within a By increasing the chance of introgression and by preventing harsher matrix (e.g., shrubs creating islands of fertility in reinforcement, we suggest that facilitation could strongly im- deserts, see Figure 3) may represent a case at hand for pede further genetic differentiation between ecotypes. disruptive selection between adjacent subpopulations (see Persistence of core genotypes in marginal habitats in- [66] for the example of Bromus erectus adapting to different creases genetic diversity and thus increases the potential for Thymus vulgaris chemotypes in calcareous grasslands). The adapting to changes [46], at least if these occur towards benefactors could promote local adaptation in the under- conditions more similar to the core habitat. Higher genetic story habitats as observed in our simulation when they diversity in marginal habitats compared to core habitats, were forming large patches (Figure 2). Therefore, facilitation that is, arid versus Mediterranean environments [55], could could also pave the way to ecological speciation, particularly support this prediction if this diversity is maintained through if directional selection leads to differences in traits that affect time. The trend toward higher genetic variability may be en- the reproductive system, such as flowering time [67]. While hanced when the marginal environment varies in time. In this scenario is certainly a possibility, we suggest that it is less that case, the different genotypes inhabiting patches with likely than the opposite case due to the small spatial scale of and without facilitation could have an advantage in different environmental heterogeneity usually created by benefactors. years. Facilitation would then increase the role of temporal Namely, the size of these islands of mild conditions roughly variation in supporting stable coexistence of genotypes (e.g., corresponds to the size of the nurse’s crown or to a clump of [56]), similar to the storage effect [57] that increases diversity shrubs and is usually small compared to the dispersal kernel at a species level. and pollination distance [68]. Furthermore, in extreme Recentnumericalmodelshavebeenexploredunder habitats the nurses also function as a sink for seeds [69]. As which conditions ecological speciation is likely to occur a result, it is unlikely that gene flow will be severely restricted [51, 58–61]. The hypothesis that facilitation acts as a cohesive between populations of the understory habitats and those force and increases genetic diversity in marginal habitats can of the harsher matrix. Nonetheless, further empirical and International Journal of Ecology 5

0.03 0.8

ST 0.015 0.4 F (selected loci) (selected ST 0 F 0 0 512 1024 2048 0 512 1024 2048 Facilitator patch size Facilitator patch size Core versus marginal Core versus marginal Benefactor versus core Benefactor versus core Benefactor versus marginal Benefactor versus marginal (a) (b)

2.4

Average ﬁtness Average 1.6

0 512 1024 2048 Facilitator patch size Core Benefactor Marginal (c)

Figure 2: Facilitation acting as a cohesive force and reducing the chance of ecological speciation. (a) represents the population divergence at neutral loci as a function of the size of patch. (b) represents the same population divergence at selected loci. (c) represents the average fitness in each environment as a function of the patch size (Box 1). Consistent with our hypothesis, in the absence of benefactors (i.e., no facilitation, patch size of 0), populations were locally adapted and gene flow was reduced between the core and the marginal habitat (higher FST on both neutral and selected loci). In the presence of benefactors, adaptation to the marginal environment was reduced by the increased gene flow from the core population and especially from populations occupying the mild patches created by the benefactors. Individuals occupying these benefactor patches have an increased fitness because of the low population density in the neighboring maladapted marginal environment that inflates the logistic growth part of the fitness function. As the size of the patches increased, migration and gene flow decreased between the patches created by the benefactors and marginal environment (increased FST on both neutral and selected loci). Because of this reduction in gene flow, the average fitness in the marginal environment increased, and consequently the fitness advantage in the benefactor patches decreased. Additionally, the FST for selected loci between the core and the benefactors, even though still high, is reduced at large patch size. This is because each patch has a subpopulation that can achieve the same phenotype using different genotypic combinations. Thus, as the patch size increased the number of patches decreased, the number of subpopulation decreased, the genotypic variance within an environment decreased, and so did the FST. theoretical studies (Box 2) are needed to test for the over- one hand, positive interactions in intermediate habitats may whelming importance of facilitation as a cohesive, rather favor niche broadening and local adaptation. On the other than a disruptive, force under these particular environmental hand, facilitation may contribute to gene flow maintenance conditions. between the core and the newly colonized habitats as de- In the context of colonization of a new habitat, where scribed in Figure 2. To our opinion, this model system pro- individuals from the population of the core habitat could not vides excellent opportunities to examine how the spatial and persist away from a benefactor in the first place (Figure 4), temporal dynamics of positive interactions could interfere benefactors could first promote local adaptation at the niche with ecological speciation. margin and then in the harsh matrix. Recent studies have pointed out the role of intermediate quality habitats in 5. Facilitation and Niche Conservatism ecotypic differentiation and in the colonisation of a new niche [70]. In the Littorina saxatilis case study, it may be Linkage between facilitation and macroevolutionary pro- suggested that facilitation played a role in this process given cesses has recently arisen from experimental work on sem- the patchy distribution of mussels and barnacles (in inter- iarid Mexican plant communities [11, 71]. Here, benefac- mediate habitats) associated with the two ecotypes observed tor species were phylogenetically unrelated to beneficiary in Spain for these species (SU and RB, resp.). However, species; the former belonged to recent Quaternary lineages this hypothesis needs to be properly tested. Thus, on the and were well adapted to the present dry climate, the latter 6 International Journal of Ecology

macro-evolutionary processes [74] and about the effects of (c) evolutionary processes on facilitative interactions [17]. So far, there is no empirical study that has simultaneously examined genetic population structure of beneficiaries and facilitation for contrasting genotypes/phenotypes in natural conditions. Therefore, there is an urgent need to design novel studies that more closely associate community ecologists and evolutionary biologists to tackle these questions (Box 2). To conclude, we suggest that plant ecologists should (a) invest more into studying biotic interactions at the popu- (b) lation level. First, shifting from species to populations and from population to genotypes in the design of ecological experiments is essential to examine the role of facilitation in ecological speciation [12, 13]. Second, population models Figure 3: View of a North- (a) and a South- (b) facing slope of should gain more realism by acknowledging the balance Wadi Shuayb, Jordan. The North-facing slope corresponds to a core between positive and negative interactions along environ- habitat for many annual (low stature) species and the South-facing mental gradients and their underlying ecological mecha- slope to a mosaic of marginal habitats with patches of mild habitats created by benefactors (Box 1). (c) Facilitation by the benefactor nisms. This research agenda should help to bridge the gap Gymnocarpos decanter of annual vegetation on the South-facing between community ecology and evolutionary biology. slope. Bruno et al. [36] claimed that “including facilitation in niche theory will challenge some of our most cherished para- digms.” Ecological speciation is another paradigm that is showed closer affinities with Tertiary lineages and were worth revisiting by examining the impact of both negative drought intolerant [11]. It was hypothesized that less adapted and positive interactions on the evolutionary history of in- “Tertiary” species survive because of facilitation [11]. In this cipient species. Facilitation is certainly challenging our un- scenario, our model of facilitation acting as a cohesive force derstanding of the origin of new species. is likely to explain niche conservatism of Tertiary species and the fact that they did not become more stress-tolerant Boxes. or go extinct through evolutionary time. Transposed to the microevolutionary scale, these results suggest that facilitation Box 1: Description of the Modeling Framework. The model may cause a genotype’s niche to remain unchanged over time of Bridle et al. [62] (initially developed to study adaptation despite environmental changes (Figure 1(c)). Namely, one on an environmental gradient) was modified as follows to may imagine that core and marginal habitats are temporally explore the effect of facilitation on ecological speciation. segregated, that is, a population experiences increasingly stressful conditions with time and adaptation to the new The Environment. We used a grid of 8192 × 4096 cells conditions is hampered by the persistence of facilitated and wrapped on a torus. Without facilitation, the left half is maladapted genotypes from more favorable times. Recently, the core habitat with an optimum (Ux in Bridle et al. there has been a renewed interest in evaluating niche con- [62]) of ΘC and the right half is the marginal habitat with servatism across lineages [72], in particular by using species optimum of ΘM and reduced carrying capacity (KM = KC/2, distribution models [73]. To our knowledge, none of these the carrying capacity if defined on a circle with radius of models has properly included the balance between positive 50 cells). In the presence of benefactors, the half right- and negative interactions along gradients to address the hand side of the grid is fragmented in a checkerboard-like underlying mechanisms and the resulting patterns of niche pattern with alternating benefactor and marginal regions. conservatism. The environments produced by the presence of benefactors have an optimum intermediate between the core and the 6. Conclusion marginal habitat (ΘF = (ΘC +ΘM)/2) and the same carrying K = K We have argued that facilitation can be viewed as a cohesive capacity ( F C) as the core environment. force limiting genetic and phenotypic differentiation in marginal habitats. Facilitation may also explain the lack of adap- The Individuals. Individuals are diploid with 128 bi-allelic tive diversification over time in certain lineages. On the other unlinked loci (64 under selection and 64 neutral). Allele hand, under very specific conditions, facilitation could also values are either zero or one with a symmetric mutation operate as a stepping stone and pave the way to ecological probability between them of 10−4.Thephenotype(z) is the speciation. This calls for a new body of empirical or theo- sum of those loci and thus range from zero to 128. The fitness retical studies addressing the strength of these two possible of males and females is determined by a logistic growth and effects of facilitations on ecological speciation (Box 2). Our local adaptation focus was on the population scale to provide the linkage with mechanisms of speciation. We thus complement recent N (θ − z)2 w = max 2+r f 1 − − s ,0 ,(1) opinions about the role of various types of facilitation on K 2 International Journal of Ecology 7

0.06 0.8

ST 0.03 F 0.4 (selected loci) (selected ST 0 F 0 0 512 1024 2048 0 512 1024 2048 Facilitator patch size Facilitator patch size Core versus marginal Core versus marginal Benefactor versus core Benefactor versus core Benefactor versus marginal Benefactor versus marginal (a) (b)

2.4

Average ﬁtness Average 1.6

0 512 1024 2048 Facilitator patch size Core Benefactor Marginal (c) Figure 4: How facilitation may act as a stepping stone for ecological speciation. We looked at that question by increasing the environmental difference between core and marginal habitats. (a) represents the population divergence at neutral loci with either no benefactor or different patch sizes of benefactor in the marginal environment. (b) represents the divergence at selected loci. (c) represents the average fitness in each environment for these two scenarios. The same model and parameters as in Figure 2 were used, except that we increased environmental difference between core and marginal habitats (Box 1). Without benefactor, no population can establish in the marginal environment that is too harsh for migrants to persist. Only dispersers can be observed there, except in one occasion of actual colonization (outlier at patch size of 0). In presence of benefactors, all environments are occupied by locally adapted populations and gene flow among environments is reduced. Small patch sizes will increase dispersal between the benefactor and marginal environment as in Figure 2.

where w if the fitness, r f is the maximum rate of increase core environment, and we follow their evolution over 2000 (set to 0.8), N is the local density in a radius of 50 cells, s generations. Three patch sizes were used to simulate the is the strength of stabilizing selection (set at 0.125, for the presence of benefactors in the harsh environment, smaller twoscenariospresentedinFigures2 and 4), and K and Θ patches (512 cells, isolated benefactors in the landscape), are the carrying capacity and local optimum at the position intermediate sizes (1024 cells), and large patch sizes (2048 of the individual. The markers start with no polymorphism cells). Each parameter set was replicated ten times (the (all alleles are zero) and have a mutation rate of 10−3 that boxplots in Figures 2 and 4 reflect the ten replicates). change their value by plus or minus one, so each allele is represented by an integer. We used FST [75] to infer gene flow Example of Two Contrasting Scenarios. The two sets of sim- among populations using neutral markers and selected loci ulations that we performed, corresponding to two different separately. scenarios, are presented in Figures 2 and 4. The first scenario presents the case where facilitation inhibits ecological specia- Reproduction. To find a mate, females scan in a radius of tion (ΘC = 62, ΘF = 64, ΘM = 66, Figure 2), and the second mating distance (MD) set at 150 cells and choose randomly a scenario presents the case where facilitation promotes eco- male with probability proportional to its fitness (wMale). Each logical speciation (ΘC = 61, ΘF = 64, ΘM = 67, Figure 4). couple leaves an average of wFemale offsprings, drawn from a Following (1), for the first scenario at carrying capacity in Poisson distribution. Each offspring is placed at a distance both environments, w, the fitness for an individual locally from the mother drawn from a normal distribution of mean adapted to the core environment (z = 62) is equal to 2, zero and standard deviation D (set at 100 cells) in a direction but the same individual would have a fitness w equal to 1 randomly selected from a uniform distribution. in the marginal environment, representing the equivalent of a selection coefficient of 50%. This value is consistent with Initial Conditions. We started each simulation with 500 the average measure of local adaptation recently reported for individuals randomly located in the centre 700 × 700 of the reciprocal transplantation experiments [76]. In the second 8 International Journal of Ecology scenario, we set the environmental difference between the spatiotemporal configurations. Namely, the crucial role of core (ΘC = 61) and the marginal environment (ΘM = 67) dispersal for adaptation to marginal habitats should be such that individuals locally adapted to the core environment investigated in the presence of facilitation. For example, an (z = 61) could not maintain a viable population in the mar- important consequence of facilitation may be that it enhan- ginal environment. ces “effective” dispersal from the source to sinks, because establishment probabilities of maladapted genotypes are enhanced. Furthermore, the sensitivity of model outputs Box 2: Facilitation and Ecological Speciation: Ways Forward to parameters controlling (i) the shape of the competition Stress Gradients as a Model System. For empirically studying versus facilitation balance along ecological gradients, that the role of facilitation in ecological speciation, one should is still widely debated [6, 31], (ii) the spatial distribution focus on target species with populations widely distributed and dynamics of nurse plants in marginal habitats, (iii) the along stress gradients, and for which facilitation and trait complex interactions among benefactors and beneficiaries divergence in marginal habitats have been documented. [17], and (iv) the dispersal ability and life history traits of Common examples are gradients of water availability in focal species, has to be investigated. Ultimately, these models semiarid landscapes or gradients of temperature along altitu- should help us sharpen and phrase testable hypotheses dinal gradients. Population genetic structure and inferences about the relative importance of facilitation on adaptation about gene flow could be evaluated first descriptively by and ecological speciation. Finally, a spatially and genetically using neutral or selected markers and by QTL analyses [77]. explicit model [51, 59–61] could couple both benefactor ff The next step would be to investigate adaptive genetic var- e ect on beneficiaries’ microevolutionary responses (this ff iation between microhabitats with and without benefac- study) and the feedback e ect of beneficiaries on population tor. These should be accompanied by experiments testing and evolutionary dynamics of benefactor species (cost of whether genotypes from “favorable” habitats rely on facilita- facilitation, [17]). Such an approach would shed light on how ff tion to persist and reproduce in marginal habitats [13]. These the evolutionary consequence of facilitation a ects commu- questions can be addressed with the same experimental de- nity assembly and dynamics. signs that are carried out at the species level, that is, reciprocal transplants or sowing experiments coupled with neighbor Authors’ Contribution manipulations. P. Liancourt wrote the first draft of the manuscript, X. Which Traits to Measure? A particular challenge to exploring Thibert-Plante performed the modelling work. All authors the effects of facilitation is to identify adaptive genetic made a substantial intellectual contribution to the paper and variation in traits (1) that are under disruptive selection pres- its revision. P. Choler, N. Gross and X. Thibert-Plante had sure along the investigated gradient and (2) for which trait equal contribution to the paper. expression strongly depends on the balance between stress tolerance and success in the core habitat. Seed size, for ex- Acknowledgment ample, is a good candidate because it is related to seedling size and tolerance of hazards for young seedlings [78]. As P. Liancourt and K. Tielborger¨ were supported by the competitive ability and stress tolerances ability are related GLOWA Jordan River Project funded by the German Min- to the probability of being facilitated [5], traits linked with istry of Science and Education (BMBF), and further support the resource capture and utilization, such as leaf traits, plant was granted to K. Tielborger¨ by the German Research stature traits, and root traits, need to be considered as well Foundation (TI338/10-1). P. Choler was partly funded by a (see [79]). Marie-Curie fellowship (contract CASOAR MOIF-CT-2006- A key trait would be both under strong selective pressure 039688). X. Thibert-Plante conducted this work while a and closely related to mate choice, the so-called “magic Postdoctoral Fellow at the National Institute for Mathe- traits” [45, 51, 80]. Because the benefactor may modify the matical and Biological Synthesis, an institute sponsored by length of the favorable season, for instance, phenological the National Science Foundation, the U.S. Department of traits such as time to flowering or length of the life cycle Homeland Security, and the U.S. Department of Agricul- [44, 81] may fall in this category and are worth examining. ff ture through NSF Award no. EF-0832858, with additional Depending on how benefactors a ect these key traits, fa- support from The University of Tennessee, Knoxville, and cilitation could potentially either hinder, accelerate or even Le Fonds Queb´ ecoisdelaRecherchesurlaNatureetles´ lead to ecological speciation (see “Section 4”). Technologies (FQRNT) directly to X. Thibert-Plante. This research was also partially supported by NIH Grant GM- Theory and Simulations. Therearelargeopenavenuesfor 56693 to S. Gavrilets. The authors thank R. Glade for editing tackling these questions with spatially explicit models of pop- the paper to improve the English, J. Bridle and J. Polechova´ ulation dynamics, building on a rich set of various modeling for sharing with them details of their model, S. Gavrilets, A. frameworks [26, 51, 62, 70, 82–84]. We present here an P.Henry, R. M. Callaway, and R. Michalet for feedback on the example (Figures 2 and 4) but such models should be refined paper, and R. Faria and the anonymous reviewers for helpful to explore the specific feature of facilitation and its impli- comments on the paper and suggesting the analysis of the cation on adaptive diversification along gradients and other fixation index on the selected loci. International Journal of Ecology 9

References Trends in Ecology and Evolution, vol. 22, no. 5, pp. 250–257, 2007. [1] M. D. Bertness and R. Callaway, “Positive interactions in com- [20] J. R. Haloin and S. Y. Strauss, “Interplay between ecolog- munities,” Trends in Ecology and Evolution, vol. 9, no. 5, pp. ical communities and evolution: review of feedbacks from 191–193, 1994. microevolutionary to macroevolutionary scales,” Annals of the [2] R.W.Brooker,F.T.Maestre,R.M.Callawayetal.,“Facilitation New York Academy of Sciences, vol. 1133, pp. 87–125, 2008. in plant communities: the past, the present, and the future,” [21] D. M. Post and E. P. Palkovacs, “Eco-evolutionary feedbacks Journal of Ecology, vol. 96, no. 1, pp. 18–34, 2008. in community and ecosystem ecology: interactions between [3] T. Day and K. A. Young, “Competitive and facilitative evolu- the ecological theatre and the evolutionary play,” Philosophical tionary diversification,” BioScience, vol. 54, no. 2, pp. 101–109, Transactions of the Royal Society B, vol. 364, no. 1523, pp. 2004. 1629–1640, 2009. [4]P.Choler,B.Erschbamer,A.Tribsch,L.Gielly,andP.Taberlet, [22] T. W. Schoener, “The newest synthesis: understanding the “Genetic introgression as a potential to widen a species’ niche: interplay of evolutionary and ecological dynamics,” Science, insights from alpine Carex curvula,” Proceedings of the National vol. 331, no. 6016, pp. 426–429, 2011. Academy of Sciences of the United States of America, vol. 101, [23] D. Schluter, “Ecology and the origin of species,” Trends in no. 1, pp. 171–176, 2004. Ecology and Evolution, vol. 16, no. 7, pp. 372–380, 2001. [5] P.Liancourt, R. M. Callaway, and R. Michalet, “Stress tolerance [24] J. S. McKinnon, S. Mori, B. K. Blackman et al., “Evidence for and competitive-response ability determine the outcome of ecology’s role in speciation,” Nature, vol. 429, no. 6989, pp. biotic interactions,” Ecology, vol. 86, no. 6, pp. 1611–1618, 294–298, 2004. 2005. [25] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology [6]F.T.Maestre,R.M.Callaway,F.Valladares,andC.J.Lortie, Letters, vol. 8, no. 3, pp. 336–352, 2005. “Refining the stress-gradient hypothesis for competition and [26] M. Doebeli and U. Dieckmann, “Speciation along environ- facilitation in plant communities,” Journal of Ecology, vol. 97, mental gradients,” Nature, vol. 421, no. 6920, pp. 259–264, no. 2, pp. 199–205, 2009. 2003. [7] C. Darwin, On the Origin of Species by Means of Natural [27] C.C.Wilmers,R.L.Crabtree,D.W.Smith,K.M.Murphy,and Selection, or the Preservation of Favoured Races in the Struggle W. M. Getz, “Trophic facilitation by introduced top predators: for Life, John Murray, London, UK, 1859. grey wolf subsidies to scavengers in Yellowstone National [8] M. L. Rosenzweig, “Competitive speciation,” Biological Journal Park,” Journal of Animal Ecology, vol. 72, no. 6, pp. 909–916, of the Linnean Society, vol. 10, no. 3, pp. 275–289, 1978. 2003. [9] D. Schluter, “Experimental evidence that competition pro- [28] F. Bulleri, “Facilitation research in marine systems: state of the motes divergence in adaptive radiation,” Science, vol. 266, no. art, emerging patterns and insights for future developments,” 5186, pp. 798–801, 1994. Journal of Ecology, vol. 97, no. 6, pp. 1121–1130, 2009. [10] D. I. Bolnick, “Can intraspecific competition drive disruptive [29]M.G.A.VanDerHeijdenandT.R.Horton,“Socialisminsoil? selection? An experimental test in natural populations of the importance of mycorrhizal fungal networks for facilitation sticklebacks,” Evolution, vol. 58, no. 3, pp. 608–618, 2004. in natural ecosystems,” JournalofEcology, vol. 97, no. 6, pp. 1139–1150, 2009. [11] A. Valiente-Banuet, A. V. Rumebe, M. Verdu,andR.M.Call-´ [30]D.J.Hodgson,R.B.Hitchman,A.J.Vanbergen,R.S.Hails,R. away, “Modern quaternary plant lineages promote diversity D. Possee, and J. S. Cory, “Host ecology determines the relative through facilitation of ancient tertiary lineages,” Proceedings fitness of virus genotypes in mixed-genotype nucleopolyhe- of the National Academy of Sciences of the United States of drovirus infections,” Journal of Evolutionary Biology, vol. 17, America, vol. 103, no. 45, pp. 16812–16817, 2006. no. 5, pp. 1018–1025, 2004. [12] E. K. Espeland and K. J. Rice, “Facilitation across stress [31]R.Michalet,R.W.Brooker,L.A.Cavieresetal.,“Dobiotic gradients: the importance of local adaptation,” Ecology, vol. 88, interactions shape both sides of the humped-back model of no. 9, pp. 2404–2409, 2007. ff species richness in plant communities?” Ecology Letters, vol. 9, [13] P. Liancourt and K. Tielborger,¨ “Ecotypic di erentiation no. 7, pp. 767–773, 2006. determines the outcome of positive interactions in a dryland [32] S. D. Hacker and S. D. Gaines, “Some implications of direct annual plant species,” Perspectives in Plant Ecology, Evolution positive interactions for community species diversity,” Ecology, and Systematics, vol. 13, no. 4, pp. 259–264, 2011. vol. 78, no. 7, pp. 1990–2003, 1997. [14] J. H. Vandermeer and D. H. Boucher, “Varieties of mutualistic [33] R. M. Callaway, R. W. Brooker, P. Choler et al., “Positive interaction in population models,” Journal of Theoretical interactions among alpine plants increase with stress,” Nature, Biology, vol. 74, no. 4, pp. 549–558, 1978. vol. 417, no. 6891, pp. 844–848, 2002. [15] J. L. Bronstein, “The evolution of facilitation and mutualism,” [34] L. A. Cavieres and E. I. Badano, “Do facilitative interactions Journal of Ecology, vol. 97, no. 6, pp. 1160–1170, 2009. increase species richness at the entire community level?” Jour- [16] R. M. Callaway, Positive Interactions and Interdependence in nal of Ecology, vol. 97, no. 6, pp. 1181–1191, 2009. Plant Communities, Springer, Dordrecht, The Netherlands, [35] L. Pellissier, K. Anne Brathen,˚ J. Pottier et al., “Species distribu- 2007. tion models reveal apparent competitive and facilitative effects [17] R. Michalet, S. Xiao, B. Touzard et al., “Phenotypic variation of a dominant species on the distribution of tundra plants,” in nurse traits and community feedbacks define an alpine Ecography, vol. 33, no. 6, pp. 1004–1014, 2010. community,” Ecology Letters, vol. 14, no. 5, pp. 433–443, 2011. [36]J.F.Bruno,J.J.Stachowicz,andM.D.Bertness,“Inclusion [18] K. Tielborger¨ and R. Kadmon, “Temporal environmental of facilitation into ecological theory,” Trends in Ecology and variation tips the balance between facilitation and interference Evolution, vol. 18, no. 3, pp. 119–125, 2003. in desert plants,” Ecology, vol. 81, no. 6, pp. 1544–1553, 2000. [37] N. Gross, P. Liancourt, P. Choler, K. N. Suding, and S. Lavorel, [19] M. T. J. Johnson and J. R. Stinchcombe, “An emerging syn- “Strain and vegetation effects on local limiting resources thesis between community ecology and evolutionary biology,” explain the outcomes of biotic interactions,” Perspectives in 10 International Journal of Ecology

Plant Ecology, Evolution and Systematics, vol. 12, no. 1, pp. 9– [56] A. S. Jump and J. Penuelas,˜ “Running to stand still: adaptation 19, 2010. and the response of plants to rapid climate change,” Ecology [38] M. Abrams, “Genotypic and phenotypic variation as stress ad- Letters, vol. 8, no. 9, pp. 1010–1020, 2005. aptations in temperate tree species—a review of several case- [57] A. L. Angert, T. E. Huxman, P. Chesson, and D. L. Venable, studies,” Tree Physiology, vol. 14, pp. 833–842, 1994. “Functional tradeoffs determine species coexistence via the [39] G. E. Rehfeldt, C. C. Ying, D. L. Spittlehouse, and D. A. storage effect,” Proceedings of the National Academy of Sciences Hamilton, “Genetic responses to climate in Pinus contorta: of the United States of America, vol. 106, no. 28, pp. 11641– niche breadth, climate change, and reforestation,” Ecological 11645, 2009. Monographs, vol. 69, no. 3, pp. 375–407, 1999. [58] M. Kirkpatrick and N. H. Barton, “Evolution of a species’ [40] M. Ackermann and M. Doebeli, “Evolution of niche width and range,” American Naturalist, vol. 150, no. 1, pp. 1–23, 1997. adaptive diversification,” Evolution, vol. 58, no. 12, pp. 2599– [59] S. Gavrilets, A. Vose, M. Barluenga, W. Salzburger, and A. 2612, 2004. Meyer, “Case studies and mathematical models of ecological [41] M. Slatkin, “Frequency- and density-dependent selection on a speciation. 1. Cichlids in a crater lake,” Molecular Ecology, vol. quantitative character,” Genetics, vol. 93, no. 3, pp. 755–771, 16, no. 14, pp. 2893–2909, 2007. [60] S. Gavrilets and A. Vose, “Case studies and mathematical mod- 1979. els of ecological speciation. 2. Palms on an oceanic island,” [42] F. B. Christiansen and V. Loeschcke, “Evolution and intraspe- Molecular Ecology, vol. 16, no. 14, pp. 2910–2921, 2007. cific exploitative competition I. One-locus theory for small [61] X. Thibert-Plante and A. P. Hendry, “The consequences of additive gene effects,” Theoretical Population Biology, vol. 18, phenotypic plasticity for ecological speciation,” Journal of Ev- no. 3, pp. 297–313, 1980. olutionary Biology, vol. 24, no. 2, pp. 326–342, 2011. [43] T. J. Kawecki, “Adaptation to marginal habitats,” Annual [62] J. R. Bridle, J. Polechova,´ M. Kawata, and R. K. Butlin, “Why Review of Ecology, Evolution, and Systematics, vol. 39, pp. 321– is adaptation prevented at ecological margins? New insights 342, 2008. from individual-based simulations,” Ecology Letters, vol. 13, [44] P. Liancourt and K. Tielborger,¨ “Competition and a short no. 4, pp. 485–494, 2010. ff growing season lead to ecotypic di erentiation at the two [63] J. A. Endler, “Geographic Variation, Speciation, and Clines,” extremes of the ecological range,” Functional Ecology, vol. 23, Princeton University. no. 2, pp. 397–404, 2009. [64] X. Thibert-Plante and A. P. Hendry, “When can ecological [45] S. Gavrilets, Fitness Landscapes and the Origin of Species, speciation be detected with neutral loci?” Molecular Ecology, Princeton University Press, 2004. vol. 19, no. 11, pp. 2301–2314, 2010. [46] R. D. Holt and T. H. Keitt, “Species’ borders: a unifying theme [65] R. Tirado and F. I. Pugnaire, “Shrub spatial aggregation and in ecology,” Oikos, vol. 108, no. 1, pp. 3–6, 2005. consequences for reproductive success,” Oecologia, vol. 136, [47] T. Lenormand, “Gene flow and the limits to natural selection,” no. 2, pp. 296–301, 2003. Trends in Ecology and Evolution, vol. 17, no. 4, pp. 183–189, [66] B. K. Ehlers and J. Thompson, “Do co-occurring plant species 2002. adapt to one another? The response of Bromus erectus to the [48] H. Gonzalo-Turpin and L. Hazard, “Local adaptation occurs presence of different Thymus vulgaris chemotypes,” Oecologia, along altitudinal gradient despite the existence of gene flow in vol. 141, no. 3, pp. 511–518, 2004. the alpine plant species Festuca eskia,” Journal of Ecology, vol. [67] V. Savolainen, M. C. Anstett, C. Lexer et al., “Sympatric 97, no. 4, pp. 742–751, 2009. speciation in palms on an oceanic island,” Nature, vol. 441, [49] P.Nosil, T. H. Vines, and D. J. Funk, “Perspective: reproductive no. 7090, pp. 210–213, 2006. isolation caused by natural selection against immigrants from [68] A. Gros, H. Joachim Poethke, and T. Hovestadt, “Evolution of divergent habitats,” Evolution, vol. 59, no. 4, pp. 705–719, local adaptations in dispersal strategies,” Oikos, vol. 114, no. 3, 2005. pp. 544–552, 2006. [50] M. R. Servedio and M. A. F. Noor, “The role of reinforcement [69] J. Flores and E. Jurado, “Are nurse-protege interactions more in speciation: theory and data,” Annual Review of Ecology, common among plants from arid environments?” Journal of Evolution, and Systematics, vol. 34, pp. 339–364, 2003. Vegetation Science, vol. 14, no. 6, pp. 911–916, 2003. [70] S. Sadedin, J. Hollander, M. Panova, K. Johannesson, and S. [51] X. Thibert-Plante and A. P.Hendry, “Five questions on ecolog- Gavrilets, “Case studies and mathematical models of ecological speciation addressed with individual-based simulations,” ical speciation. 3: ecotype formation in a Swedish snail,” Mo- Journal of Evolutionary Biology, vol. 22, no. 1, pp. 109–123, lecular Ecology, vol. 18, no. 19, pp. 4006–4023, 2009. 2009. [71] A. Valiente-Banuet and M. Verdu,´ “Facilitation can increase [52] J. Aronson, J. Kigel, A. Shmida, and J. Klein, “Adaptive the phylogenetic diversity of plant communities,” Ecology phenology of desert and Mediterranean populations of annual Letters, vol. 10, no. 11, pp. 1029–1036, 2007. plants grown with and without water stress,” Oecologia, vol. 89, [72] J. J. Wiens and C. H. Graham, “Niche conservatism: inte- no. 1, pp. 17–26, 1992. grating evolution, ecology, and conservation biology,” Annual [53] J. Aronson, J. Kigel, and A. Shmida, “Reproductive allocation Review of Ecology, Evolution, and Systematics, vol. 36, pp. 519– strategies in desert and Mediterranean populations of annual 539, 2005. plants grown with and without water stress,” Oecologia, vol. [73] P. B. Pearman, A. Guisan, O. Broennimann, and C. F. Randin, 93, no. 3, pp. 336–342, 1993. “Niche dynamics in space and time,” Trends in Ecology and [54] P. Choler, R. Michalet, and R. M. Callaway, “Facilitation Evolution, vol. 23, no. 3, pp. 149–158, 2008. and competition on gradients in alpine plant communities,” [74] Z. Kikvidze and R. M. Callaway, “Ecological facilitation may Ecology, vol. 82, no. 12, pp. 3295–3308, 2001. drive major evolutionary transitions,” BioScience, vol. 59, no. [55] S. Volis, S. Mendlinger, L. Olsvig-Whittaker, U. N. Safriel, 5, pp. 399–404, 2009. and N. Orlovsky, “Phenotypic variation and stress resistance [75] B. S. Weir, Genetic Data Analysis II: Methods for Discrete Pop- in core and peripheral populations of Hordeum spontaneum,” ulation Genetic Data, Sinauer Associates, Sunderland, Mass, Biodiversity and Conservation, vol. 7, no. 6, pp. 799–813, 1998. USA, 1996. International Journal of Ecology 11

[76] J. Hereford, “A quantitative survey of local adaptation and fitness trade-offs,” American Naturalist, vol. 173, no. 5, pp. 579–588, 2009. [77] P. Nosil, D. J. Funk, and D. Ortiz-Barrientos, “Divergent selection and heterogeneous genomic divergence,” Molecular Ecology, vol. 18, no. 3, pp. 375–402, 2009. [78] A. T. Moles and M. Westoby, “Seedling survival and seed size: a synthesis of the literature,” Journal of Ecology,vol.92,no.3, pp. 372–383, 2004. [79] V. Maire, N. Gross, L. Da Silveira Pontes, C. Picon-Cochard, andJ.F.Soussana,“Trade-off between root nitrogen acquisition and shoot nitrogen utilization across 13 co-occurring pasture grass species,” Functional Ecology,vol.23,no.4,pp. 668–679, 2009. [80]M.R.Servedio,G.S.V.Doorn,M.Kopp,A.M.Frame,and P. Nosil, “Magic traits in speciation: “magic” but not rare?” Trends in Ecology and Evolution, vol. 26, no. 8, pp. 389–397, 2011. [81] M. Petru,˚ K. Tielborger,¨ R. Belkin, M. Sternberg, and F. Jeltsch, “Life history variation in an annual plant under two opposing environmental constraints along an aridity gradient,” Ecogra- phy, vol. 29, no. 1, pp. 66–74, 2006. [82] A. Hastings and S. Harrison, “Metapopulation dynamics and genetics,” Annual Review of Ecology and Systematics, vol. 25, pp. 167–188, 1994. [83]J.MolofskyandJ.D.Bever,“Anewkindofecology?” BioScience, vol. 54, no. 5, pp. 440–446, 2004. [84]J.M.J.Travis,R.W.Brooker,E.J.Clark,andC.Dytham, “The distribution of positive and negative species interactions across environmental gradients on a dual-lattice model,” Jour- nal of Theoretical Biology, vol. 241, no. 4, pp. 896–902, 2006. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 192345, 15 pages doi:10.1155/2012/192345

Research Article Use of Host-Plant Trait Space by Phytophagous Insects during Host-Associated Differentiation: The Gape-and-Pinch Model

Stephen B. Heard

Department of Biology, University of New Brunswick, P.O. Box 4400, Fredericton, NB, Canada E3B 5A3

Correspondence should be addressed to Stephen B. Heard, [email protected]

Received 22 July 2011; Accepted 6 November 2011

Academic Editor: Andrew Hendry

Copyright © 2012 Stephen B. Heard. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Ecological speciation via host shifting has contributed to the astonishing diversity of phytophagous insects. The importance for host shifting of trait differences between alternative host plants is well established, but much less is known about trait variation within hosts. I outline a conceptual model, the “gape-and-pinch” (GAP) model, of insect response to host-plant trait variation during host shifting and host-associated differentiation. I offer four hypotheses about insect use of plant trait variation on two alternative hosts, for insects at different stages of host-associated differentiation. Collectively, these hypotheses suggest that insect responses to plant trait variation can favour or oppose critical steps in herbivore diversification. I provide statistical tools for analysing herbivore trait-space use, demonstrate their application for four herbivores of the goldenrods Solidago altissima and S. gigantea, and discuss their broader potential to advance our understanding of diet breadth and ecological speciation in phytophagous insects.

1. Introduction A common theme among case studies of ecological speciation is the existence of two alternative niches—micro- The insects have long been held up as providing spectac- habitats, resources, reproductive strategies, and so forth— ular examples of rapid diversification and high standing that can be exploited by individuals of a single species, with diversity (e.g., [1–3]). Among insects, phytophagous clades the potential for disruptive selection to operate between the often undergo dramatic radiations [4], and phytophagous alternative niches. For phytophagous insects, the alternative lineages tend to be more diverse than their nonphytophagous niches are a pair of host plant species (or organs). One com- sisters [5, 6]. One likely driver of diversification among monly imagines an evolutionary sequence beginning with phytophagous insects is their tendency to specialize on host- an insect exploiting only one of the two alternative hosts. plant species or organs [7–10] and to diversify via host Perhaps via host-choice errors, some individuals occasionally or organ shifts followed by host-associated differentiation attack individuals of the second host, and if fitness penalties (HAD), the evolution of new specialist races or species [9, for doing so are not too severe, a host shift occurs and the 11–14]. Because many cases of HAD appear to have pro- insect begins to exploit both alternative hosts. (Description ceeded in sympatry [15], a great deal of theoretical and of these events as “errors” is standard in the plant-insect empirical work has focused on understanding ways in which literature, but of course this usage is teleological shorthand adaptation to different host plants can impose disruptive se- and can conceal interesting biology. For instance, it might be lection on nascent specialist forms and also reduce gene that genotypes with strong enough host preferences to avoid flow (or permit differentiation in the face of gene flow) be- “errors” would also show costly rejection of some suitable tween those forms [14–16]. Phytophagous insects, along with hosts; in this case, the occurrence of host-choice “errors” is parasitoids [17], freshwater fishes [18], seed-eating birds simply an adaptive compromise. Nonetheless, for simplicity I [19], and habitat-specialist plants [20] and lizards [21]have retain the standard usage here). Disruptive selection can now therefore been central to the development of ideas about begin to favour genotypes better adapted to each alternative ecological speciation [22]. host. If reproductive isolation arises between nascent forms, 2 International Journal of Ecology then ecological speciation can proceed, and a single (perhaps with a statistical approach for investigating trait-space use polymorphic) generalist is replaced with a pair of host- in phytophagous insects. I suggest hypotheses for temporal specialist races or species. Because reproductive isolation is changes in host trait-space use over evolutionary time, from expected to take some time to evolve, if it can evolve at all, initial host-choice errors through to the independent evolu- different insects exploiting a pair of alternative host plants are tion of a pair of well-isolated host-specialist sibling species. I expected to fall on a continuum from generalists to nascent, call the overall model the “gape-and-pinch,” or “GAP,” model poorly differentiated host forms to distinct host-specialist of trait-space use (the reason for this name will be apparent sister species [13, 23]. There will be analogous continua for after the model is described). While I outline the model for ecological speciation across other kinds of alternative niches, plants and phytophagous insects, it will apply to many other for instance, in parasitoids speciating across hosts or fish systems with some straightforward vocabulary substitutions. across depth niches (e.g., [17, 24]). The process of HAD in phytophagous insects has been 2. Conceptual “GAP” Model of Host widely discussed, both in general [14] and in the context of Trait-Space Use a few well-studied model systems (e.g., apple maggot fly [25, 26], goldenrod ball-gall fly [27, 28]). Perhaps unsurprisingly, All plant species vary intraspecifically for numerous mor- nearly all studies of HAD have emphasized insect responses phological, phenological, and chemical traits, with variation to differences in plant traits between the alternative hosts, having genetic, epigenetic, and/or environmental causes while downplaying variation in plant traits among individ- (e.g., [42–45]). This variation defines a set of phenotypes that uals within each host. Such an interspecific perspective is are available for attack by a phytophagous insect searching its obviously appropriate for studies using population-genetic environment for suitable hosts. This set of phenotypes can tools to detect host-associated forms and reconstruct their be depicted as a cloud of points in a multidimensional trait history (e.g., [13, 29–31]), but it is also near universal space, with each point representing an individual plant (or a in studies discussing ecological mechanisms by which host ramet, for clonal plants; or even a module, when important shifts and HAD proceed (e.g., [25, 26, 28, 32–38]). An alter- variation occurs within individuals [46]). It is convenient native approach would explicitly recognize within-species to consider the two-dimensional case (Figure 1), which can variation in host-plant traits and consider possible roles for represent either a system in which two plant traits show such variation in favouring or impeding host shifting and variation relevant to insect attack or a two-dimensional sum- HAD. This approach has yet to be applied in earnest to any mary of a higher-dimensional trait space (using principal system, but intriguing hints at its usefulness appear in the components to extract two dominant axes of trait variation). literature for the goldenrod ball-gall fly, Eurosta solidaginis, I use the term “available trait space” to describe this cloud and its races on the goldenrods Solidago altissima and S. of points, as combinations of plant traits falling inside it are gigantea.Forexample,Eurosta of the S. altissima race prefer available to attacking herbivores, while combinations outside the largest ramets of their host [39], and since S. gigantea are not available (i.e., they do not correspond to real plants plants tend to be shorter when Eurosta oviposits [40], this which might be attacked). This trait space can be character- preference might discourage host-choice errors by altissima ized by calculating its centroid (the point whose coordinate flies. In contrast, if gigantea flies similarly prefer taller ramets, on each axis is the mean value of that coordinate for all they could be susceptible to host-choice errors (although individuals), its size (average distance from individual plants gigantea flies’ preferences have not been assessed, and neither to the centroid), and its shape. hypothesis raised here appears to have been tested). Work on Consider first an insect interacting with a single host the phenology of insect emergence and host-plant growth species. Some plant individuals will be attacked, but oth- has similar implications. Eurosta adults emerge from S. ers will likely escape. The attacked individuals define the gigantea earlier than from S. altissima [41], and this pattern is “attacked trait space” (filled circles in Figure 1), which must correlated with availability of rapidly growing ramets of each be a subset of the available trait space (and might be host to be attacked [40]. Thus, individual S. altissima ramets expected to be a smaller subset for more specialist herbivores with earlier phenology, or S. gigantea ramets with later phe- [47, 48]). The relationship between available and attacked nology, might be more likely to be attacked by the “wrong” trait spaces will depend on active host-selection behavior by host race. There is geographic variation in the abundance of insects (insect preference), and also on whether insects can such intermediate-phenology ramets, and How et al. [40] survive on an individual plant after initiating attack (insect suggested that host shifts might be more easily initiated performance). Both preference and performance will depend where host phenology overlaps more extensively. on plant traits—sometimes the same traits, but sometimes Even for Eurosta, however, there are few plant traits for not. For simplicity, I use “herbivore attack” to denote the which insect responses have been studied on both alternative occurrence of feeding herbivores on plants, whether patterns hosts, and so we know little about how insect responses in occurrence arise from preference or performance, and to plant trait variation might relate to the ecology of host terms like “selective” attack should similarly be taken to shifting and HAD. Furthermore, there is no system for which include both preference and performance effects. The at- we can compare insect responses to plant trait variation for a tacked trait space may represent a common preference by all set of insects attacking the same plants but differing in stage herbivore individuals, or the sum of herbivore individuals’ of host shifting and HAD. I outline here a conceptual model distinct preferences in species with strong individual special- of host trait-space use during host shifting and HAD, along ization [49, 50]. International Journal of Ecology 3 Plant trait PC2 Plant trait PC2 Plant trait PC2

Plant trait PC1 Plant trait PC1 Plant trait PC1 (a) (b) (c) Figure 1: Possible patterns of host trait-space use by herbivorous insects. Available host-plant individuals form a cloud in trait space (which is likely multidimensional, but summarized here by the ﬁrst two principal component axes). Open circles denote unattacked plants, and ﬁlled circles attacked ones.

When herbivore attack is random with respect to plant trait-space relationships predicted for an insect herbivore traits (Figure 1(a)), the attacked trait space will resemble the moving through a four-step evolutionary sequence: from available trait space in centroid location, shape, and (once original specialization on one of the two hosts, through a corrected for the smaller number of attacked plants) size. For host shift, to early and late stages of HAD. a selective herbivore (one that rejects some available plants), in contrast, the attacked and available trait spaces will differ. 2.1. Stage 1: Single-Host Specialists and the Importance of Many patterns are possible, but two are particularly likely. Host-Choice Errors. An insect attacking a single host could First, the herbivore might attack typical plants (those with show virtually any pattern in the relationship between at- trait values near the population means) and reject extreme tacked and available trait spaces, but some patterns are of ones, leading to an attacked trait space that is central with special interest in the context of possible host-shifting to respect to the available trait space: the attacked space is small- an evolutionarily novel host. (By a “host shift” I mean the er than the available space, but the two spaces have similar addition of a novel plant to the herbivore species’ diet, centroids (Figure 1(b)). Such a pattern might be favoured which will normally occur without immediate abandonment by selection because (for example) typical plants are most of the old). Such host shifts are likely to begin when a common, and insects preferring them pay lower search costs few individuals attack the “wrong” (novel) host, making it and experience greater resource availability. Alternatively, possible for selection to favour the incorporation of the novel herbivore attack might be associated with extreme trait host into the insect’s host range. Importantly, the likelihood values (e.g., herbivores might perform best on the largest or of host-choice errors is likely to depend on the insect’s use least-defended individuals), leading to an attacked trait space of plant trait space. In particular, imagine an insect showing that is marginal with respect to the available trait space: the a marginal attacked trait space on the ancestral host. That attacked trait space is again restricted in size, but in this case marginal attacked trait space could be adjacent to the the two spaces have different centroids (Figure 1(c)). available trait space defined by the novel host (Figure 2(a)), Now consider a pair of plant species available for attack or could be distant from it. When it is adjacent, host-choice (Extension to larger numbers of hosts is possibly but con- errors are more likely and insects making those errors are siderably complicating.) Given a common set of measured more likely to survive on the novel host [52]. In contrast, traits, we should see two clouds of points in trait space when the ancestrally attacked trait space is distant from (Figure 2) defining a pair of available trait spaces. The the novel host, host-choice errors (and thus host shifting) distance between available trait spaces defined by different should be less likely. I call this the “adjacent errors hypothesis.” plant species might be large compared to the size of each The logic mirrors the widespread expectation that host- available trait space [38], but, for closely related pairs of choice errors and host shifts are more likely between species phenotypically variable plants, this need not be so (e.g., for that resemble each other morphologically, chemically, or Solidago altissima and S. gigantea,seeFigure 3). Two avail- phylogenetically [38] but stresses that the distance in trait able trait spaces could even be touching or interdigitated, space that needs to be crossed for a host-choice error depends especially for hybrid swarms [51]. One can again consider not only on the distance between available trait spaces but attacked trait spaces in comparison to available trait spaces also on how insect preference and performance define the on each host, but now there are many more possibilities, as attacked trait space. each attacked trait space could be nonselective, central, or marginal (and if marginal, toward or away from the other 2.2. Stage 2: Oligophagous Feeding Following Diet Expansion. host). Among possible patterns, I emphasize here a set of Following a host shift that expands diet, our focal insect 4 International Journal of Ecology Plant trait PC2 Plant trait PC2

Plant trait PC1 Plant trait PC1 (a) (b) Plant trait PC2 Plant trait PC2

Plant trait PC1 Plant trait PC1 (c) (d)

Figure 2: Some possible relationships between attacked trait spaces on two alternative host plants. Illustrated are hypothetical relationships for four stages during a host shift and subsequent host-associated differentiation. (a) Specialist on host 1, predisposed to host-choice errors due to marginal attack on trait space adjacent to the alternative host. (b) Following diet expansion, herbivore is oligophagous accepting both hosts; selection favours central attack on the combined trait space. (c) Early evolution of host-associated differentiation: use of marginal and distant trait spaces on the two hosts reduces host-choice errors and hence gene flow. (d) Reproductively isolated monophagous specialists on each host: selection favours central attack by each specialist on the trait space of its host. See text for further discussion of these scenarios. species will be oligophagous, feeding on two hosts (ancestral the ancestrally attacked trait space are more easily colonized. plus novel) rather than one. Because this stage should I call this the “adjacent oligophagy hypothesis.” follow from patterns in attack allowing host-choice errors Note that the adjacency pattern is equivalent to restricted (adjacent errors hypothesis), attacked trait space is likely but central use of an available trait space defined by the two to remain marginal on the ancestral host (Figure 2(a)). As hosts in combination (compare Figures 1(b) and 2(b)). If host preference and performance evolve to include attack disruptive selection between alternative hosts does not act or on the novel host, the attacked trait space on that host is is not powerful, oligophagy and the adjacency pattern could likely to be marginal as well, but with the two attacked trait be evolutionarily persistent. Alternatively, this stage might spaces adjacent (Figure 2(b)) because novel plants closer to be transient, persisting only until disruptive selection has International Journal of Ecology 5 time to drive HAD of insect subpopulations exploiting the In summary, the trait distance-divergence hypothesis two hosts. The contrast between these possibilities highlights holds that attainment of distance between attacked trait spa- an important fork in the evolutionary road [53], in which ces (Figure 1(c)) can be both a symptom of HAD and also disruptive selection favouring HAD is or is not sufficient to a factor permitting HAD. The larger the distance between overcome gene flow working to homogenize the herbivore attacked trait spaces, the more likely is the evolution of population and to maintain an oligophagous diet. genetic differentiation between insects on the two hosts. Since genetic differentiation can ease the evolution of dis- 2.3. Stage 3: Nascent Host-Specialist Forms and the Selection- tance between attacked trait spaces, this stage of evolution Gene Flow Tension. How might insect trait-space use favour can involve positive feedback [58–60]. In contrast, an insect or oppose the ability of disruptive selection to achieve HAD? for which attacked trait spaces remain adjacent (Figure 1(b)) Craig et al. [52] argued that persistent oligophagy is likely is likely to remain an oligophagous insect with no host- when the two available trait-spaces are very close, with HAD associated structure to its gene pool. likely when they are more distant. However, their conceptual model assumes that attacked trait spaces on the two hosts 2.4. Stage 4: Pair of Established Host Specialists. As HAD pro- remain indefinitely adjacent (their Figure 1). I suggest that ceeds and gene flow between nascent host forms declines, we there is another important possibility; a critical step in would expect the gradual accumulation of more, and more HAD may be the separation of the attacked trait spaces effective, reproductive isolating mechanisms [24, 61, 62]. (Figure 2(c)) such that insect subpopulations on the two This should continue until ecological speciation is complete, hosts (now appropriately thought of as nascent host forms) and the two host-specialist forms attain the status of full come to attack dissimilar individuals rather than similar biological species. As reproductive isolation becomes en- ones. I call this the “trait distance-divergence hypothesis”. forced by multiple, redundant mechanisms, the importance Separation of the two attacked trait spaces could arise in two of separation between attacked trait spaces should decline. ff di erent but complementary ways. Selection will then be free to mould trait-space use inde- First, distance between attacked trait spaces could arise pendently for each species, and if reinforcement earlier in simply because disruptive selection for adaptation to the HAD pushed the attacked trait spaces apart (trait distance- ff alternative hosts overpowers the homogenizing e ect of gene divergence hypothesis), this force can now relax. If selection flow between nascent host forms. Under this scenario, the favours use of central trait space on each host, for example attacked trait spaces could move to opposite ends of the (Figure 2(d); or if it favours nonselective use of trait space available trait spaces, as in Figure 1(c), or merely further on each host), the distance between the attacked trait spaces apart than adjacency, depending on the shape of the fitness should decrease. I call this the “distance relaxation hypothe- landscape—that is, fitness optima on the alternative hosts sis.” might favour central or marginal trait spaces. Under this Note that through the temporal sequence (Figures 2(a)– scenario, distance between attacked trait spaces is just a 2(d)), the overall conceptual model suggests a pair of at- symptom by which the progress of HAD can be recognized. tacked trait spaces that begin close together, move apart, Second, distance could be a product of selection to and then move back together like pincers. This movement minimize host-choice errors by each nascent host form or underlies the terminology “gape-and-pinch” model of trait- hybridization between them [22]. Host-choice errors could space use. be opposed by selection because they lead to preference- performance mismatches, or because they put larvae in competition with members of the other host form (encouraging 2.5. What about Generalists? The foregoing considered in- divergence by character displacement). Alternatively, host- sects that begin as monophagous on one of the two hosts and choice errors coupled with the tendency for phytophagous remain narrowly oligophagous or monophagous at all stages insects to mate on their host plants could lead to hybrid of HAD. However, many herbivores are broader generalists matings. Hybrid disadvantage is possible given tradeoffsin [63] for which we would not expect any of the trait- ability to exploit the alternative hosts, or if hybrids prefer space patterns shown in Figure 2. In particular, it would or are best suited for trait-value combinations falling in be very surprising if a broad generalist showed nonrandom the gap between the two available trait spaces (this gap is separation between attacked trait spaces (Figures 2(c) and shown narrow in Figure 2 butwilloftenbewider[38]). 2(d)). Instead, attacked trait space might be nonselective on Selection to reduce hybridization by widening the distance both hosts [47, 48], or restricted but marginal along a trait ff between attacked trait spaces (Figure 1(c))wouldbeaform axis orthogonal to the di erence between the two hosts (e.g., of reinforcement [54]. Under the reinforcement scenario, insects might prefer larger individuals of each host and also distance between attacked trait spaces is more than a symp- attack other, larger species). Such broad generalists are much tom of HAD; once achieved, it serves to reduce gene flow less likely than host specialists to undergo HAD because between nascent host forms and permit HAD to progress. (being already adapted to multiple hosts) they are less likely (In passing, I note that if selection simply overpowers gene to experience strongly disruptive selection for performance flow, we would expect HAD to involve genetic divergence in on one host versus another [64]. genomic islands, whereas if selection opposes hybridization, genome-wide genetic divergence should result via “isolation 2.6. Testing the Hypotheses. The four hypotheses that make by adaptation”) [55–57]. up the GAP model are logically distinct; finding that one 6 International Journal of Ecology

Table 1: Relationships between patterns in trait-space use (Figures 2(a)–2(d)), the GAP model, and statistical tests implemented for analysis of attacked trait spaces.

Attacked trait spaces Attacked trait spaces Attacked trait spaces Pattern in trait-space use Hypothesis marginal?1 restricted?1 distant? Monophagous, attacked Adjacency favours host Ancestral host: marginal trait space marginal and shifting (adjacent errors Novel host: marginal but — Attacked spaces close2 adjacent to alternative hypothesis) rare2 host (Figure 2(a)) Oligophagous, attacked Adjacency persists after Ancestral host: marginal trait spaces marginal and host shifting (adjacent — Attacked spaces close Novel host: marginal adjacent (Figure 2(b)) oligophagy hypothesis) Distance permits, and is Nascent host races, also symptomatic of, attacked trait spaces Ancestral host: marginal genetic isolation (trait — Attacked spaces distant marginal and distant Novel host: marginal distance-divergence (Figure 2(c)) hypothesis) Other isolating Pair of monophagous mechanisms reduce Ancestral host: restricted Ancestral host: not species, attacked trait importance of (central)ornonselective Attacked spaces neither marginal spaces central on each trait-space distance Novel host: restricted close nor distant Novel host: not marginal host (Figure 2(d)) (distance relaxation (central)ornonselective hypothesis) 1 More strictly, only restricted or marginal trait-space use along the PC axis (or axes) defining the difference between available hosts is directly relevant to the GAP model. 2But statistical detection is difficult because attack on the novel host is rare. hypothesis holds (or fails) implies nothing about the others. assessment of the GAP model depends more on comparative For example, for a given herbivore, the adjacent errors and tests of the hypotheses for herbivores as a class, and this adjacent oligophagy hypotheses could hold, but the distance- will require data on host trait-space use for multiple insect divergence and distance relaxation hypotheses fail, if HAD herbivores differing in host range. A particularly revealing proceeds to ecological speciation without any movement of approach will involve sets of herbivores that differ in the attacked trait spaces away from each other following the host extent of their progression through HAD (e.g., [13]), because shift. of the expectation that the trait-space relationships shown Each of the four hypotheses can also be posed, and tested, in Figures 2(a)–2(d) form a temporal evolutionary sequence. at two levels. First, we can test each hypothesis for a single The cleanest comparative tests will involve herbivores sharing herbivore. For instance, are attacked trait spaces adjacent a common pair of alternative hosts, so that attacked trait on Solidago altissima and S. gigantea for the narrowly spaces can be contrasted among herbivores (e.g., recent host oligophagous [13] gallmaker Epiblema scudderiana (adjacent forms and ancient specialist pairs, for the distance relaxation oligophagy hypothesis)? Of course, such tests focus on hypothesis) while seen against the simple backdrop of a patterns, and confirmation of a pattern need not constitute common pair of available trait spaces. a strong test of underlying mechanism. Second, and more Progress towards understanding the influence of trait- powerfully, we can test each hypothesis for herbivorous space use on the evolutionary trajectory of herbivore spe- insects as a class. For instance, are attacked trait spaces cialization, then, can be made by measuring for multiple (statistically) further apart for recently divergent and incom- insect herbivores the relationships between available and pletely isolated pairs of host races than they are for more attacked trait spaces on the alternative hosts, for comparison ancient specialist species pairs (distance relaxation hypothe- with those suggested by the four hypotheses (summarized in sis)? At this level, the hypotheses can hold strongly or weakly Table 1). In particular, we will be interested in whether (or not at all); that is, the empirical relationship between attacked trait space on each host is marginal (and in what trait-space distance and extent of reproductive isolation direction), and whether the two attacked trait spaces are could be stronger or weaker (or non-significant). adjacent or distant. Testing the adjacent errors, adjacent oligophagy, trait Unfortunately, we do not yet have sufficient comparative distance-divergence, and/or distance relaxation hypotheses data to test the GAP model. Members of my laboratory for individual herbivores will require trait-space use data for are beginning to gather such data for insect herbivores large numbers of individuals on the alternative hosts. Some attacking the goldenrods Solidago altissima and S. gigantea. tests will be difficult at the individual-herbivore level (e.g., In following sections, I provide formal statistical tools for testing the distance relaxation hypothesis for an individual analysis of such data and apply them to an illustrative data herbivore would require historical data on past trait-space set demonstrating a path towards the comparative hypothesis use, which will only rarely be available). Ultimately, though, tests that are our ultimate goal. International Journal of Ecology 7

3. Statistical Methods that the two attacked trait spaces are significantly distant. On the other hand, when the actual centroids are closer than the I developed statistical methods to test for three patterns in mean distance from randomizations, then twice the fraction host trait-space use by insect herbivores. These patterns are of randomization distances that are smaller than the actual ff predicted, in di erent combinations, by the adjacent errors, distance is a P-value, which when small supports rejection of adjacent oligophagy, distance-divergence, and distance relax- the null hypothesis in favour of the alternative that the two ation hypotheses (Table 1) and thus provide windows on the attacked trait spaces are significantly adjacent. overall GAP model. The tests share a common framework The marginal trait-space, restricted trait-space, and in that they are based on relationships between attacked and distant trait-space tests are implemented in TraitSpaces 1.20, available trait spaces for host plants of two species (Figures a program written in Microsoft Visual Basic.NET for Win- 1 and 2). Two tests pertain to the pattern of attack on a dows. The software takes as input a datafile with a row for single host, and the third to the pattern of attack on each each individual host plant, and columns for host species host relative to the other. identity, presence/absence of each herbivore, and first and First, I test for central versus marginal location of the second principal components calculated from the host trait attacked trait space on each host (“Marginal trait-space test”). matrix. (Principal components may be output from any I calculate the centroid of the available trait space (mean standard statistical package.) Extension to trait spaces of PC1 and PC2 scores for all available plants, attacked and higher dimensionality, if desired, is straightforward; one unattacked) and that of the attacked trait space (mean PC could even use an unreduced trait matrix at the cost of some scores for attacked plants only). I then calculate the distance complexity in displaying results. The analytical framework between available and attacked centroids and compare this easily accommodates data for other host/attacker systems to a null distribution of 10,000 such distances calculated and could even be applied to cases where consumers use following random shuffling of attack status across all plant variable microhabitats or food resources. The current version individuals. The fraction of randomization distances larger of the TraitSpaces package is available from the author on than the actual attacked-available distance is a P value, and request. when it is small we reject the null hypothesis that attacked and available plants have a common centroid (central or 4. Field Methods nonselective attack, Figures 1(a) and 1(b)) in favour of the alternative of marginal attack (Figure 1(c)). 4.1. Study System. The goldenrods Solidago altissima L. and When we are unable to detect marginal use of available S. gigantea Ait. are clonal perennials codistributed over much trait space, we might seek to distinguish between nonse- of eastern and central North America. Intermixed stands of lective (Figure 1(a)) and restricted but central (Figure 1(b)) the two species are common in open habitats such as prairies, alternatives. To do so, I use the “restricted trait-space test.” I old fields, roadsides, and forest edges. Individual ramets grow calculate the Euclidean distance from each attacked plant to in spring from underground rhizomes, flower in late summer the centroid of attacked trait space. The size of the attacked and fall, and die back to ground level before winter. The two trait space is given by the sum of these distances. I then species differ most obviously in pubescence [65]: S. altissima compare this trait-space size to a null distribution the sizes of stems are sparsely to densely short-hairy, especially basally, 10,000 attacked trait spaces generated by randomly shuffling while S. gigantea stems are typically glabrous. Both species attack status across all plant individuals. Note that shuffling display extensive intraspecific variation (genetic and plastic) attack status maintains the number of attacked plants, which in most traits, including ramet size, pubescence, leaf shape, is critical when calculating the size of a trait space. The size, and toothiness, and chemical profiles ([27, 66, 67], S. B. fraction of randomization attacked trait spaces smaller than Heard, unpubl. data). the actual one is a P-value, and when it is small we reject S. altissima and S. gigantea are attacked by a diverse fauna the null hypothesis that attack is nonselective. Since we are of insect herbivores [68–70], which vary in diet specializa- using the restricted trait-space test following a nonsignificant tion. Some are broad generalists that accept Solidago as part marginal trait-space test, the alternative is that herbivores of a taxonomically diverse diet (e.g., the exotic spittlebug Phi- exploit a restricted but central subset of available trait space. laenus spumarius [71]), and some are broadly oligophagous, Finally, I test whether the distance between attacked cen- feeding on Solidago among other members of the Aster- troids on the two host plants is smaller or larger than aceae (e.g., the chrysomelid Exema canadensis [72]). Others expected at random (“Distant trait-spaces test,” Figure 2(b) are more narrowly oligophagous, attacking only Solidago versus 2(c)). I first calculate the distance between attacked spp. (e.g., the tortricid stem-galler Epiblema scudderiana trait-space centroids on the two alternative hosts. This dis- [13, 73]). Finally, at least four herbivores have evolved tance is compared, in a two-tailed test, to a null distribution monophagous host races or cryptic species on S. altissima of 10,000 such distances calculated following random shuf- and S. gigantea [13, 30, 74, 75], with divergence ranging from fling of attack status across individuals of each plant species quite recent for the ball-gall fly Eurosta solidaginis (at most (separately). When the actual centroids are farther apart 200,000 years, but likely much less) to >2 × 106 years old for than the mean distance from randomizations, then twice the bunch-gall flies Rhopalomyia solidaginis/R. capitata. the fraction of randomization distances that are larger than Especially for the better-studied S. altissima, attack by the actual distance is a P-value, which when small supports various herbivores is known to vary among clones [68, 76, rejection of the null hypothesis in favour of the alternative 77], and with plant traits including ramet size [39, 78], 8 International Journal of Ecology

Philaenus Exema PC2: hairinessPC2: PC2: hairinessPC2:

PC2: size PC2: size (a) (b) Uroleucon Rhopalomyia PC2: hairinessPC2: hairinessPC2:

PC2: size PC2: size (c) (d) Figure 3: Attacked trait spaces for four goldenrod herbivores on S. altissima (triangles) and S. gigantea (circles). Filled symbols denote attacked plants, and open symbols unattacked ones. Axis labels are shorthand for the first two principal components from a 7-variable morphological dataset; full factor loadings are provided in Table 2. growth rate [79], nutritional status [80, 81] and ploidy S. gigantea exclusively diploid, so effects of ploidy on herbi- wherethisvarieslocally[82]. These trait-attack relationships vore attack [83] need not be considered here. involve both plant resistance and insect preference [27]and It is important to assess the available and attacked trait may be concordant or discordant among different herbivore spaces using traits measured before herbivore attack; other- species [68, 82]. wise, herbivore responses to plant traits could be confounded with herbivore-driven changes in the same traits. In early 4.2. Field Data. I and my field team gathered data on plant June 2004, we marked 104 S. altissima ramets and 186 traits and herbivore attack in old-field and trailside Solidago S. gigantea ramets by setting line transects through well- populations in Fredericton, NB, Canada (45◦ 57 30 N, 66◦ mixed patches of the two species and marking each ramet 37 1–20 W). Here both S. gigantea and S. altissima are touched by the line. At the time of marking, a few ramets abundant along with S. rugosa, S. juncea, S. canadensis, Eu- had already been attacked by the stem-galler Gnorimoschema thamia graminifolia, Symphyotrichum spp., and other Aster- gallaesolidaginis (Lepidoptera: Gelechiidae; galls on 4 S. aceae. S. altissima is exclusively hexaploid in the east, and altissima and 3 S. gigantea ramets), but other herbivores International Journal of Ecology 9

Table 2: Correlations among morphological variables measured for S. altissima and S. gigantea.

Ramet height Leaf length Leaf width Teeth Water content Trichomes Stem width 0.88 0.72 0.18 0.40 −0.42 0.36 Ramet height 0.75 0.17 0.43 −0.44 0.36 Leaf length 0.47 0.56 −0.23 0.12 Leaf width 0.32 0.24 −0.57 Teeth −0.18 −0.01 Water content −0.48 had yet to attack. We measured 7 morphological traits of Table 3: Factor loadings for the first two principal components our marked ramets, focusing on easily measured traits that from the morphological data matrix for S. altissima and S. gigantea. were likely to influence herbivore attack, that help distinguish Trait Loading on PC1 Loading on PC2 the two study species, or both. We measured stem trichome density by counting, in the field with a hand lens, all Stem width 0.50 0.04 trichomes in silhouette along a 10 cm length of stem just Ramet height 0.51 0.04 below the terminal bud. We measured stem width 5–10 cm Leaf length 0.47 −0.22 above ground using a caliper, and stem height (from ground Leaf width 0.14 −0.63 to the base of the terminal bud) using a measuring tape. For Teeth 0.34 −0.24 the largest leaf from each ramet, we measured leaf length, leaf Water content −0.29 −0.39 width at the widest point, and the number of teeth along Trichomes 0.34 0.58 one leaf edge. Finally, we weighed each largest leaf before and after drying to constant mass at 45-55◦Candcalculated percent water content. (Table 2). The first two principal component axes explained We surveyed marked ramets twice weekly until the end 47% and 28% of the morphological variance (75% total), of August, identifying herbivores present as specifically as while no other axis explained more than 9.3%. PC1 largely possible without disturbing them on the plant (for some reflects ramet size (heavy loadings for stem width, ramet groups, like larval Trirhabda beetles, species-level identifi- height, and leaf length; Table 3), but also leaf toothiness cations require the removal of the insects to the laboratory, (positively) and water content (negatively). PC2 contrasts and we wanted to leave plants to experience natural levels trichome counts (strong positive loading; Table 3)withleaf of herbivory). When herbivores of the same species were width (strong negative loading) but also includes leaf water present on consecutive surveys, we were usually unable to content, leaf length, and toothiness (all negative). These determine whether they were the same individuals, so rather two principal components do a good job of capturing both than count individuals we classified each ramet as attacked intraspecific and interspecific variation (Figure 3), with S. or unattacked by each herbivore over the course of the entire altissima and S. gigantea separated primarily along PC2 (the season. pubescent S. altissima with high scores, and the glabrous S. Some marked ramets were lost or damaged during the gigantea with low scores). season, leaving 92 S. altissima ramets and 175 S. gigantea ramets with comprehensive herbivory and plant-trait data. Four herbivores were identifiable to species and abundant 5.2. Herbivore Use of Phenotype Space. Attack rates by Sol- enough to give our analyses reasonable power: the xylem- idago herbivores are generally low (often 1–10% or even less), sucking spittlebug Philaenus spumarius, which is broadly with the exception of some diet generalists and outbreaking polyphagous [71]; the folivorous chrysomelid beetle Exema species in high-density years (S. Heard, unpubl. data). In canadensis, which is oligophagous with many hosts in the our dataset, even though we worked with some of the most tribe Astereae [72]; the phloem-sucking aphid Uroleucon common herbivores, only one herbivore on one host had nigrotuberculatum, which is narrowly oligophagous on Sol- an incidence above 30% (Exema canadensis on S. altissima, idago spp. [84]; and the gall-making cecidomyiid fly Rhopal- 64% of ramets attacked). Other herbivore/host combinations omyia solidaginis/R. capitata, which is a pair of monophagous had lower incidences, with several less than 10% (Uroleucon specialists (R. solidaginis on S. altissima and R. capitata on S. nigrotuberculatum on both hosts and Rhopalomyia capitata gigantea [13]). All further analyses use this reduced set of 267 on S. gigantea; Table 4). ramets and 4 herbivores. The patterns I document in trait-space use could have arisen via herbivore preference, or via performance if poor 5. Field Results and Discussion herbivore growth leads to death or departure of herbivores before surveys can detect them. For most herbivores, re- 5.1. Plant Traits. Among the 7 measured traits stem width, peated surveys allow herbivore detection shortly after attack ramet height, and leaf length were strongly intercorrelated begins, and so preference is the most likely driver of patterns (0.72

Table 4: Tests of trait-space use on Solidago altissima and S. gigantea for four Solidago herbivores. P-values in bold are signiﬁcant at α = 0.05. Distant Marginal Restricted Distance between #Attacked trait- Herbivore Host Distance from available centroid2 trait-space trait-space attacked plants1 spaces test P test P trait-spaces3 test P PC1 PC2 S. altissima 25 1.00 −0.50 <0.001 —4 Philaenus Small 0.71 S. gigantea 22 1.11 −0.52 0.001 — S. altissima 59 0.30 0.05 0.023 — Exema Small 0.14 S. gigantea 33 0.65 0.04 0.023 — S. altissima 91.39−0.10 0.007 — Uroleucon Large 0.86 S. gigantea 82.32−0.62 <0.001 — S. altissima 27 0.09 0.35 0.21 0.18 Rhopalomyia Large 0.80 S. gigantea 80.66−0.05 0.31 1.0 1 Of 92 available S. altissima and 175 available S. gigantea ramets. 2Attacked centroid minus available centroid (PC1 and PC2 components). A positive entry means that ramets with a large PC score are more likely to be attacked. 3“Small” if the two attacked trait spaces are adjacent (Figure 2(b)), and “large” if the two attacked trait spaces are distant (Figure 2(c)). 4This test is informative only when the marginal trait-space test is not signiﬁcant.

Plant genotype effects on gall induction, mismatched with [78] also have well-documented associations with larger herbivore preference, are known (for example) for Eurosta ramets, something that is common but not universal among solidaginis on S. altissima [85]. phytophagous insects [86, 87]. Such concordance across The two most generalist herbivores (Philaenus and herbivore species in the use of trait space increases the Exema) showed similar patterns in trait-space use (Figures likelihood of multiple herbivores cooccuring on a single 3(a) and 3(b); Table 4). Both showed significant evidence ramet—something very unlikely under the null hypothesis for nonrandom use of available hosts (marginal trait-space of independent occurrence, since most attack rates are low. test). For Philaenus on both hosts, attack was concentrated Herbivores cooccurring on a plant may compete directly (via on larger but less pubescent ramets (higher PC1 and lower resource consumption) or indirectly (via induced resistance) PC2), while for Exema on both hosts attack was concentrated or may even show facilitation [88] although we know little on larger ramets but did not depend on pubescence (higher about potential interactions among goldenrod herbivores PC1). For both species, the distance between attacked trait- [89–91]. However, concordance among goldenrod herbi- spaces on S. altissima and S. gigantea was slightly but not vores in use of plant trait space is far from universal [68, 82]. significantly smaller than expected under the null (distant How do the illustrative data fit with the GAP model of trait-spaces test). trait space use during host shifting and HAD laid out above? For the oligophagous Uroleucon (Figure 3(c)), attack on The tendency for most herbivores to attack larger ramets both hosts was significantly marginal, being concentrated on (larger PC1) generates pattern in trait-space use. However, larger and less pubescent ramets. The distance between the this shared tendency means that both attacked trait spaces are two attacked trait spaces was slightly, but not significantly, offset from the available spaces, in parallel and orthogonally larger than expected under the null. Because this herbivore to the contrast between alternative host plants (PC2). The had the smallest sample size (just 17 attacked ramets total), distance between attacked trait spaces is unaffected, and so these tests have much less power than for the more common this ramet-size pattern is not directly relevant to the GAP generalists. model. Three of the four herbivores analyzed (Philaenus, For the monophagous Rhopalomyia species pair Exema,andUroleucon) have host ranges broader than just the (Figure 3(d)), there was no evidence on either host for mar- S. altissima-S. gigantea pair and might therefore be expected ginal use of trait space, and the restricted trait space test to be rather unselective about traits distinguishing the two suggests nonselective rather than central use of available trait hosts. Indeed, Exema showed no offsets between attacked space (Table 4). The distance between the two attacked trait- and available trait spaces along the principal components spaces was slightly, but not significantly, larger than expected axis contrasting S. altissima and S. gigantea (PC2; Table 4). under the null. Sample size, however, was very small for R. Philaenus and Uroleucon did show offsets along this axis, but capitata on S. gigantea (8 attacked ramets), so the tests for because they were in the same direction and roughly equal that species and for the distance between attacked centroids on the two hosts, separation of attacked trait-spaces was not are likely not very powerful. significantly large for either herbivore (Table 4, distant trait spaces test). The fourth herbivore, Rhopalomyia, is a pair of 5.3. Interpretation and Prospects. The clearest pattern in relatively old monophagous species [13]. For such a species the illustrative dataset is that for three of four herbivores, pair, the distance relaxation hypothesis suggests that the use attack is significantly concentrated on larger ramets (large of distant trait spaces may no longer be an important barrier PC1). The stem gallers Eurosta [39]andGnorimoschema to gene flow (Figure 2(d)). Rhopalomyia’s use of trait space International Journal of Ecology 11

(Table 4: no evidence for marginal attack, and separation Similarly, interspecific variation in the same kinds of traits between the two attacked trait spaces no larger than expected has been widely held up as the key to the macroevolutionary at random) is consistent with this (Table 1). fate of herbivore lineages (host shifting, diversification, spe- Overall, none of the illustrative data are inconsistent with cialization, and so forth [38, 63, 64, 97]). What is surprising the GAP model, but none of the four herbivores analyzed is that the intersection of these perspectives is so little de- provides a strong test of its hypotheses. I did not find veloped: we know almost nothing about trait-space use in any examples of the patterns hypothesized for a single- systems where host shifting and HAD are suspected. This host specialist making host-choice errors, for a narrowly gap is clearly illustrated by the two best-studied cases of oligophagous species immediately following a host shift, or HAD in phytophagous insects: Eurosta solidaginis on Solidago for a pair of nascent forms early in HAD (Table 1, Figures altissima and S. gigantea and Rhagoletis pomonella on apple 2(a)–2(c)). This is not surprising, though, because species and hawthorn. For Eurosta, despite a wealth of information known to be in early stages of HAD on S. altissima/S. about how preference and performance relate to genetic and gigantea, such as the ball-gall fly Eurosta solidaginis and trait variation within S. altissima [27], few comparable data the spindle-gall moth Gnorimoschema gallaesolidaginis,were are available for flies attacking S. gigantea (except see [40]). insufficiently abundant for analysis. At the broader level of For Rhagoletis, much has been written about the importance hypothesis testing, four herbivores constitute just a small step for HAD of apple-hawthorn differences in ripening phenol- towards assessing general patterns in trait-space use through ogy [26, 98, 99] and fruit size [100]. However, data on local HAD and ecological speciation. It will take many studies like intraspecific variation in phenology appear to be unavailable mine, with herbivores on Solidago and other plants, before (although latitudinal clines have been documented [26]), we can assess the generality of patterns in trait-space use. and data on intraspecific fruit size variation appear limited to Because attack rates for most Solidago herbivores are confirming significance of interspecific differences in average low, achieving powerful hypothesis tests for any herbivore fruit size [100]. This is not to criticize work on these two will entail marking very large numbers of ramets—especially model systems, which has pioneered the study of HAD, but since ramet selection must be done before attack begins to rather to draw attention to a significant opportunity for avoid distortions of attack-space measurements if trait values progress. change under herbivore attack. In Solidago, for instance, Of course, the GAP model likely falls short of recognizing ramet biomass and height are often reduced by herbivory the full complexity of trait-space use in nature. While I have [91–93]. I am currently expanding on the illustrative study focused on snapshots of trait variation and insect use of trait with the goal of securing larger sample sizes for the herbi- space at a single site and in a single year, both available trait vores studied here and acceptable sample sizes for many more space and its use are likely to vary in space and time. This herbivores. variation could have interesting and important consequences Another obvious limitation of the illustrative dataset is for HAD. For example, intraspecific variation in Solidago that it includes measurements of only seven plant traits, phenology and the difference in average phenology between and conspicuously omits leaf-chemistry traits (and ploidy S. altissima and S. gigantea change in space and time, and [83], which varies elsewhere but not in New Brunswick). S. phenological differences are involved in host choice for at altissima and S. gigantea have complex secondary chemistry, least two Solidago herbivores undergoing HAD (Eurosta [40]; and variation in leaf chemistry is known to influence Gnorimoschema gallaesolidaginis,S.B.Heard,unpubl.data). herbivore attack [66, 67]. Unmeasured morphological traits Hawthorn phenologies show latitudinal gradients across may also be relevant to insect attack; for instance, Philaenus space favouring local adaptation rather like that required prefers plant species and individuals with wider leaf axils during Rhagoletis’ host shift to apple [26]. There are thus [94], and this trait varies among S. altissima genotypes likely to be places or times that are more conducive to host (Maddox unpubl. in [77]). Expanding the list of measured shifts and HAD than others [40, 52, 53]. Superimposed over traits, and especially incorporating leaf chemistry, is a high this variation in available trait space can be strong geographic priority for future work. variation in insect preference (e.g., [82, 101]) and thus trait- Despite the small numbers of attacked ramets and mea- space use. As a consequence, the places or times conducive sured traits that earn the “illustrative” dataset its descriptor, to host shifting for one insect herbivore might not be so my analysis of trait-space use for four herbivores establishes conducive for shifts by another. This is consistent with the that the field and analytical approach outlined here is feasible evolutionary pattern seen in the Solidago system, in which and can detect nonrandom trait-space use. Because the gold- three gallmakers have made host shifts from S. altissima to S. enrod system includes such a diverse herbivore community gigantea buthavedonesoatdifferent times [13]. attacking syntopic ramets of the alternative hosts, it offers the Thinking about intraspecific variation in plant traits, and potential for great progress in testing hypotheses about host patterns of insect attack with respect to that variation, can trait-space use during hostshifts and HAD. expand and enhance our view of ecological speciation by phytophagous insects. Testing the hypotheses I frame about 6. General Discussion trait-space use for herbivores differing in diet breadth and in progress along the evolutionary sequence of HAD (Table 1) The literature on how insect preference and performance could take us a long way towards a predictive understand- vary with intraspecific variation in host-plant genotype, ing of diet evolution and specialization in phytophagous morphology, chemistry, and phenology is immense [95, 96]. insects. Ultimately, we would like to know for which taxa 12 International Journal of Ecology host-shifting and HAD are likely, and for which taxa they are phenomenon or isolated exceptions? Evidence from a gold- not—and why [13, 38, 64, 102]. While much data collection enrod-insect community,” Evolution, vol. 59, no. 12, pp. and analysis lies ahead, the trait-space perspective promises a 2573–2587, 2005. new and powerful window on the fascinating complexity of [14] K. W. Matsubayashi, I. Ohshima, and P. Nosil, “Ecological insect-plant interactions and herbivore diversification. speciation in phytophagous insects,” Entomologia Experimen- talis et Applicata, vol. 134, no. 1, pp. 1–27, 2010. [15] S. H. Berlocher and J. L. Feder, “Sympatric speciation in Acknowledgments phytophagous insects: Moving beyond controversy?” Annual Review of Entomology, vol. 47, pp. 773–815, 2002. The author thanks the City of Fredericton for permission [16] M. Dres` and J. Mallet, “Host races in plant-feeding insects to conduct field work; William Godsoe and Susan Timmons and their importance in sympatric speciation,” Philosophical collected the field data, and William Godsoe conducted pre- Transactions of the Royal Society B, vol. 357, no. 1420, pp. liminary statistical analysis. He thanks Andrew Hendry and 471–492, 2002. two anonymous reviewers for comments on the manuscript. [17] J. O. Stireman III, J. D. Nason, S. B. Heard, and J. M. John Nason and John Stireman helped shape his view of the Seehawer, “Cascading host-associated genetic differentiation Solidago system through collaborations, and Warren Abra- in parasitoids of phytophagous insects,” Proceedings of the hamson and John Semple have provided advice and encour- Royal Society B, vol. 273, no. 1586, pp. 523–530, 2006. agement over the years. This work was supported by Dis- [18] S. Skulason,´ S. S. Snorrason, and B. Jonsson,´ “Sympatric covery Grants from the Natural Sciences and Engineering morphs, populations and speciation in freshwater fish with Research Council (Canada). emphasis on arctic charr,” in Evolution of Biological Diversity, A. E. Magurran and R. M. May, Eds., Oxford University Press, Oxford, UK, 1999. References [19] A. P. Hendry, S. K. Huber, L. F. De Leon,´ A. Herrel, and J. Podos, “Disruptive selection in a bimodal population of [1] D. Sharp, “Insects, part I,” in The Cambridge Natural History, Darwin’s finches,” Proceedings of the Royal Society B: Biologi- S. F. Harmer and A. E. Shipley, Eds., pp. 83–565, MacMillan cal Sciences, vol. 276, no. 1657, pp. 753–759, 2009. and Co., London, UK, 1895. [20] J. Antonovics and A. D. Bradshaw, “Evolution in closely [2] E. O. Wilson, The Diversity of Life, W.W. Norton, New York, adjacent plant populations. 8. Clinal patterns at a mine NY, USA, 1992. boundary,” Heredity, vol. 25, pp. 349–362, 1970. [3] P. J. Mayhew, “Why are there so many insect species? [21] E. B. Rosenblum and L. J. Harmon, “‘same same but Perspectives from fossils and phylogenies,” Biological Reviews, different’: replicated ecological speciation at white sands,” vol. 82, no. 3, pp. 425–454, 2007. Evolution, vol. 65, no. 4, pp. 946–960, 2011. [4]D.R.StrongJr.,J.H.Lawton,andR.Southwood,Insects [22] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology on Plants: Community Patterns and Mechanisms,Blackwell Letters, vol. 8, no. 3, pp. 336–352, 2005. Scientific, Oxford, UK, 1984. [23] J. Peccoud and J. C. Simon, “The pea aphid complex as a [5] C. Mitter, B. Farrell, and B. Wiegmann, “The phylogenetic model of ecological speciation,” Ecological Entomology, vol. study of adaptive zones: has phytophagy promoted insect 35, no. 1, pp. 119–130, 2010. diversification?” American Naturalist, vol. 132, no. 1, pp. 107– 128, 1988. [24] A. P. Hendry, D. I. Bolnick, D. Berner, and C. L. Peichel, [6] B. D. Farrell, “’Inordinate fondness’ explained: Why are there “Along the speciation continuum in sticklebacks,” Journal of so many beetles?” Science, vol. 281, no. 5376, pp. 555–559, Fish Biology, vol. 75, no. 8, pp. 2000–2036, 2009. 1998. [25] J. L. Feder, S. H. Berlocher, and S. B. Opp, “Sympatric [7] E. A. Bernays and R. F. Chapman, Host-Plant Selection by host-race formation and speciation in Rhagoletis (Diptera: Phytophagous Insects, Chapman and Hall, New York, NY, Tephritidae): a tale of two species for Charles D,” in Genetic USA, 1994. Structure and Local Adaptation in Natural Insect Populations, [8]P.D.N.Hebert,E.H.Penton,J.M.Burns,D.H.Janzen,and S. Mopper, Ed., pp. 408–441, Chapman and Hall, New York, W. Hallwachs, “Ten species in one: DNA barcoding reveals NY, USA, 1998. cryptic species in the neotropical skipper butterfly Astraptes [26] J. L. Feder, T. H. Powell, K. Filchak, and B. Leung, “The fulgerator,” Proceedings of the National Academy of Sciences diapause response of Rhagoletis pomonella to varying envi- of the United States of America, vol. 101, no. 41, pp. 14812– ronmental conditions and its significance for geographic and 14817, 2004. host plant-related adaptation,” Entomologia Experimentalis et [9] J. B. Joy and B. J. Crespi, “Adaptive radiation of gall-inducing Applicata, vol. 136, no. 1, pp. 31–44, 2010. insects within a single host-plant species,” Evolution, vol. 61, [27] W. G. Abrahamson and A. E. Weis, Evolutionary Ecology no. 4, pp. 784–795, 2007. Across Three Trophic Levels: Goldenrods, Gallmakers, and [10]M.A.Condon,S.J.Scheffer, M. L. Lewis, and S. M. Swensen, Natural Enemies, Princeton University Press, Princeton, NJ, “Hidden neotropical diversity: Greater than the sum of its USA, 1997. parts,” Science, vol. 320, no. 5878, pp. 928–931, 2008. [28] T. P. Craig and J. K. Itami, “Divergence of Eurosta solidaginis [11] C. Darwin, The Origin of Species, 1968 reprint, Penguin in response to host plant variation and natural enemies,” Books, London, UK, 1859. Evolution, vol. 65, no. 3, pp. 802–817, 2011. [12] B. D. Walsh, “The apple-worm and the apple-maggot,” [29] S. H. Berlocher, “Radiation and divergence in the Rhagoletis Tilton’s Journal of Horticulture and Florist’s Companion, vol. pomonella species group: inferences from allozymes,” Evolu- 2, pp. 338–343, 1867. tion, vol. 54, no. 2, pp. 543–557, 2000. [13] J. O. Stireman, J. D. Nason, and S. B. Heard, “Host-associ- [30] J. D. Nason, S. B. Heard, and F. R. Williams, “Host-associated ated genetic differentiation in phytophagous insects: General genetic differentiation in the goldenrod elliptical-gall moth, International Journal of Ecology 13

Gnorimoschema gallaesolidaginis (Lepidoptera: Gelechii- [48] N. Janz and S. Nylin, “The role of female search behaviour in dae),” Evolution, vol. 56, no. 7, pp. 1475–1488, 2002. determining host plant range in plant feeding insects: a test [31] A. M. Dickey and R. F. Medina, “Testing host-associated of the information processing hypothesis,” Proceedings of the differentiation in a quasi-endophage and a parthenogen on Royal Society B, vol. 264, no. 1382, pp. 701–707, 1997. native trees,” Journal of Evolutionary Biology, vol. 23, no. 5, [49] D. I. Bolnick, R. Svanback,¨ M. S. Araujo,´ and L. Persson, pp. 945–956, 2010. “Comparative support for the niche variation hypothesis that [32] S. P. Carroll and C. Boyd, “Host race radiation in the soap- more generalized populations also are more heterogeneous,” berry bug: natural history with the history,” Evolution, vol. Proceedings of the National Academy of Sciences of the United 46, no. 4, pp. 1052–1069, 1992. States of America, vol. 104, no. 24, pp. 10075–10079, 2007. [33] D. J. Funk, “Isolating a role for natural selection in specia- [50] D. I. Bolnick, R. Svanback,¨ J. A. Fordyce et al., “The ecology tion: host adaptation and sexual isolation in Neochlamisus of individuals: incidence and implications of individual bebbianae leaf beetles,” Evolution, vol. 52, no. 6, pp. 1744– specialization,” American Naturalist, vol. 161, no. 1, pp. 1–28, 1759, 1998. 2003. [34] S. Via, A. C. Bouck, and S. Skillman, “Reproductive isolation [51] L. M. Evans, G. J. Allan, S. M. Shuster, S. A. Woolbright, and between divergent races of pea aphids on two hosts. II. T. G. Whitham, “Tree hybridization and genotypic variation Selection against migrants and hybrids in the parental drive cryptic speciation of a specialist mite herbivore,” environments,” Evolution, vol. 54, no. 5, pp. 1626–1637, 2000. Evolution, vol. 62, no. 12, pp. 3027–3040, 2008. [35] I. Emelianov, F. Simpson, P. Narang, and J. Mallet, “Host [52] T. P. Craig, J. K. Itami, T. Ohgushi, Y. Ando, and S. Utsumi, choice promotes reproductive isolation between host races of “Bridges and barriers to host shifts resulting from host plant the larch budmoth Zeiraphera diniana,” Journal of Evolution- genotypic variation,” Journal of Plant Interactions, vol. 6, no. ary Biology, vol. 16, no. 2, pp. 208–218, 2003. 2-3, pp. 141–145, 2011. [36] I. Ohshima, “Host-associated pre-mating reproductive isola- [53] D. I. Bolnick, “Sympatric speciation in threespine stickle- tion between host races of Acrocercops transecta: mating site back: why not?” International Journal of Ecology, vol. 2011, preferences and effect of host presence on mating,” Ecological Article ID 942847, 15 pages, 2011. Entomology, vol. 35, no. 2, pp. 253–257, 2010. [54] M. R. Servedio and M. A. F. Noor, “The role of reinforcement [37] I. Ohshima, “Host race formation in the leaf-mining moth in speciation: theory and data,” Annual Review of Ecology, Acrocercops transecta (Lepidoptera: Gracillariidae),” Biologi- Evolution, and Systematics, vol. 34, pp. 339–364, 2003. cal Journal of the Linnean Society, vol. 93, no. 1, pp. 135–145, [55] P. Nosil, D. J. Funk, and D. Ortiz-Barrientos, “Divergent 2008. selection and heterogeneous genomic divergence,” Molecular [38] T. Nyman, “To speciate, or not to speciate? Resource hetero- Ecology, vol. 18, no. 3, pp. 375–402, 2009. geneity, the subjectivity of similarity, and the macroevolu- [56] P. Nosil, S. P. Egan, and D. J. Funk, “Heterogeneous genomic tionary consequences of niche-width shifts in plant-feeding differentiation between walking-stick ecotypes: “isolation insects,” Biological Reviews, vol. 85, no. 2, pp. 393–411, 2010. by adaptation” and multiple roles for divergent selection,” [39] R. Walton, A. E. Weis, and J. P.Lichter, “Oviposition behavior Evolution, vol. 62, no. 2, pp. 316–336, 2008. and response to plant height by Eurosta solidaginis Fitch [57] X. Thibert-Plante and A. P. Hendry, “When can ecological (Diptera: Tephritidae),” Annals of the Entomological Society of speciation be detected with neutral loci?” Molecular Ecology, America, vol. 83, pp. 509–514, 1990. vol. 19, no. 11, pp. 2301–2314, 2010. [40] S. T. How, W. G. Abrahamson, and T. P. Craig, “Role of [58] B. J. Crespi, “Vicious circles: positive feedback in major ev- host plant phenology in host use by Eurosta solidaginis olutionary and ecological transitions,” Trends in Ecology and (Diptera: Tephritidae) on Solidago (Compositae),” Environ- Evolution, vol. 19, no. 12, pp. 627–633, 2004. mental Entomology, vol. 22, no. 2, pp. 388–396, 1993. [59] A. P. Hendry, “Selection against migrants contributes to [41] T. P. Craig, J. K. Itami, W. G. Abrahamson, and J. D. Horner, the rapid evolution of ecologically dependent reproductive “Behavioral evidence for host-race formation in Eurosta isolation,” Evolutionary Ecology Research,vol.6,no.8,pp. solidaginis,” Evolution, vol. 47, no. 6, pp. 1696–1710, 1993. 1219–1236, 2004. [42] B. Schmid and C. Dolt, “Effects of maternal and paternal [60] K. Ras¨ anen¨ and A. P. Hendry, “Disentangling interactions environment and genotype on offspring phenotype in Sol- between adaptive divergence and gene flow when ecology idago altissima L,” Evolution, vol. 48, no. 5, pp. 1525–1549, drives diversification,” Ecology Letters, vol. 11, no. 6, pp. 624– 1994. 636, 2008. [43] J. N. Maloof, “QTL for plant growth and morphology,” Cur- [61] A. F. Agrawal, J. L. Feder, and P. Nosil, “Ecological divergence rent Opinion in Plant Biology, vol. 6, no. 1, pp. 85–90, 2003. and the origins of intrinsic postmating isolation with gene [44] M. T. J. Johnson and A. A. Agrawal, “Plant genotype and flow,” International Journal of Ecology, vol. 2011, Article ID environment interact to shape a diverse arthropod commu- 435357, 15 pages, 2011. nity on evening primrose (Oenothera biennis),” Ecology, vol. [62] J. A. Coyne and H. A. Orr, Speciation, Sinauer Associates, 86, no. 4, pp. 874–885, 2005. Sunderland, Mass, USA, 2004. [45] R. A. Rapp and J. F. Wendel, “Epigenetics and plant [63] M. S. Singer, “Evolutionary ecology of polyphagy,” in Special- evolution,” New Phytologist, vol. 168, no. 1, pp. 81–91, 2005. ization, Speciation, and Radiation: The Evolutionary Biology of [46] A. A. Winn, “Adaptation to fine-grained environmental Herbivorous Insects, K. J. Tilmon, Ed., pp. 29–42, University of variation: an analysis of within-individual leaf variation in an California Press, Berkeley, Calif, USA, 2008. annual plant,” Evolution, vol. 50, no. 3, pp. 1111–1118, 1996. [64] D. J. Funk and P. Nosil, “Comparative analyses of ecological [47] M. C. Singer, D. Ng, D. Vasco, and C. D. Thomas, “Rapidly speciation,” in Specialization, Speciation, and Radiation: The evolving associations among oviposition preferences fail to Evolutionary Biology of Herbivorous Insects,K.J.Tilmon,Ed., constrain evolution of insect diet,” American Naturalist, vol. pp. 117–135, University of California Press, Berkeley, Calif, 139, no. 1, pp. 9–20, 1992. USA, 2008. 14 International Journal of Ecology

[65] J. C. Semple and R. E. Cook, “Solidago,” in Flora of North on environmental context,” Evolution, vol. 40, pp. 863–866, America, Flora North America Editorial Committee, Ed., pp. 1986. 107–166, Oxford University Press, Oxford, UK, 2006. [82] K. Halverson, S. B. Heard, J. D. Nason, and J. O. Stireman, [66]H.M.Hull-Sanders,R.Clare,R.H.Johnson,andG.A. “Differential attack on diploid, tetraploid, and hexaploid Meyer, “Evaluation of the evolution of increased competitive Solidago altissima L. by five insect gallmakers,” Oecologia, vol. ability (EICA) hypothesis: loss of defense against generalist 154, no. 4, pp. 755–761, 2008. but not specialist herbivores,” Journal of Chemical Ecology, [83] K. Halverson, S. B. Heard, J. D. Nason, and J. O. Stireman, vol. 33, no. 4, pp. 781–799, 2007. “Origins, distribution, and local co-occurrence of polyploid [67] R. H. Johnson, R. Halitschke, and A. Kessler, “Simultane- cytotypes in Solidago altissima (Asteraceae),” American Jour- ous analysis of tissue- and genotype-specific variation in nal of Botany, vol. 95, no. 1, pp. 50–58, 2008. Solidago altissima (Asteraceae) rhizome terpenoids, and the [84] N. Moran, “The genus Uroleucon (Homoptera, Aphididae) in polyacetylene dehydromatricaria ester,” Chemoecology, vol. Michigan—key, host records, biological notes, and descrip- 20, no. 4, pp. 255–264, 2010. tions of 3 new species,” Journal of the Kansas Entomological [68] G. D. Maddox and R. B. Root, “Structure of the encounter Society, vol. 57, pp. 596–616, 1984. between goldenrod (Solidago altissima) and its diverse insect [85] S. S. Anderson, K. D. McCrea, W. G. Abrahamson, and L. M. fauna,” Ecology, vol. 71, no. 6, pp. 2115–2124, 1990. Hartzel, “Host genotype choice by the ball gallmaker Eurosta [69] R. B. Root and N. Cappuccino, “Patterns in population solidaginis (Diptera: Tephritidae),” Ecology, vol. 70, no. 4, pp. change and the organization of the insect community asso- 1048–1054, 1989. ciated with goldenrod,” Ecological Monographs, vol. 62, no. 3, [86] P. W. Price, Macroevolutionary Theory on Macroecological pp. 393–420, 1992. Patterns, Cambridge University Press, Cambridge, UK, 2003. [70]E.M.G.Fontes,D.H.Habeck,andF.SlanskyJr.,“Phy- [87] D. T. Quiring, L. Flaherty, R. Johns, and A. Morrison, tophagous insects associated with goldenrods (Solidago spp.) “Variable effects of plant module size on abundance and in Gainesville, Florida,” Florida Entomologist, vol. 77, no. 2, performance of galling insects,” in Galling Arthropods and pp. 209–221, 1994. Their Associates: Ecology and Evolution,K.Ozaki,J.Yukawa, [71] C. R. Weaver and D. R. King, “Meadow spittlebug,” Ohio T. Ohgushi, and P. W. Price, Eds., pp. 189–198, Springer, Agricultural Experiment Station Research Bulletin, vol. 741, Sapporo, Japan, 2006. pp. 1–99, 1954. [88] S. B. Heard and C. K. Buchanan, “Larval performance and [72] R. B. Root and F. J. Messina, “Defensive adaptations and association within and between two species of hackberry natural enemies of a case-bearing beetle, Exema canadensis nipple gall insects, Pachypsylla spp. (Homoptera: Psyllidae),” (Coleoptera: Chrysomelidae),” Psyche, vol. 90, pp. 67–80, American Midland Naturalist, vol. 140, no. 2, pp. 351–357, 1983. 1998. [73] W. E. Miller, “Biology and taxonomy of three gall-forming [89] J. T. Cronin and W. G. Abrahamson, “Goldenrod stem galler species of Epiblema (Olethreutidae),” Journal of the Lepi- preference and performance: effects of multiple herbivores dopterists’ Society, vol. 30, pp. 50–58, 1976. and plant genotypes,” Oecologia, vol. 127, no. 1, pp. 87–96, [74] G. L. Waring, W. G. Abrahamson, and D. J. Howard, 2001. “Genetic differentiation among host-associated populations [90] J. T. Cronin and W. G. Abrahamson, “Host-plant genotype of the gallmaker Eurosta solidaginis (Diptera: Tephritidae),” and other herbivores influence goldenrod stem galler prefer- Evolution, vol. 44, pp. 1648–1655, 1990. ence and performance,” Oecologia, vol. 121, no. 3, pp. 392– [75] C. P. Blair, W. G. Abrahamson, J. A. Jackman, and L. Tyrrell, 404, 1999. “Cryptic speciation and host-race formation in a purportedly [91] E. A. Lehnertz, Impacts of herbivory in a goldenrod/insect generalist tumbling flower beetle,” Evolution, vol. 59, no. 2, community: effects of early-season herbivores on host plants and pp. 304–316, 2005. a late-season herbivore, M.S. thesis, University of Iowa, Iowa, [76] K. D. McCrea and W. G. Abrahamson, “Variation in herbi- Canada, 2001. vore infestation: historical vs. genetic factors,” Ecology, vol. [92] D. C. Hartnett and W. G. Abrahamson, “The effects of stem 68, no. 4, pp. 822–827, 1987. gall insects on life history patterns in Solidago canadensis,” [77] G. D. Maddox and R. B. Root, “Resistance to 16 diverse Ecology, vol. 60, pp. 910–917, 1979. species of herbivorous insects within a population of gold- [93] S. B. Heard and E. K. Kitts, “Impact of attack by Gnori- enrod, Solidago altissima: genetic variation and heritability,” moschema gallmakers on their ancestral and novel Solidago Oecologia, vol. 72, no. 1, pp. 8–14, 1987. hosts,” Evolutionary Ecology. In press. [78] S. B. Heard and G. H. Cox, “Plant module size and attack [94] P.B. McEvoy, “Niche partitioning in spittlebugs (Homoptera: by the goldenrod spindle-gall moth,” Canadian Entomologist, Cercopidae) sharing shelters on host plants,” Ecology, vol. 67, vol. 141, no. 4, pp. 406–414, 2009. no. 2, pp. 465–478, 1986. [79] T. P. Craig, J. K. Itami, C. Shantz, W. G. Abrahamson, [95] S. Gripenberg, P. J. Mayhew, M. Parnell, and T. Roslin, J. D. Horner, and J. V. Craig, “The influence of host “A meta-analysis of preference-performance relationships in plant variation and intraspecific competition on oviposition phytophagous insects,” Ecology Letters, vol. 13, no. 3, pp. 383– preference and offspring performance in the host races of 393, 2010. Eurosta solidaginis,” Ecological Entomology,vol.25,no.1,pp. [96] D. Carmona, M. J. Lajeunesse, and M. T. Johnson, “Plant 7–18, 2000. traits that predict resistance to herbivores,” Functional Ecol- [80] J. D. Horner and W. G. Abrahamson, “Influence of plant ogy, vol. 25, no. 2, pp. 358–367, 2011. genotype and environment on oviposition preference and [97] N. Janz and S. Nylin, “The oscillation hypothesis of host- offspring survival in a gallmaking herbivore,” Oecologia, vol. plant range and speciation,” in Specialization, Speciation, and 90, no. 3, pp. 323–332, 1992. Radiation: The Evolutionary Biology of Herbivorous Insects, [81] G. D. Maddox and N. Cappuccino, “Genetic determination K. J. Tilmon, Ed., pp. 29–42, University of California Press, of plant susceptibility to an herbivorous insect depends Berkeley, Calif, USA, 2008. International Journal of Ecology 15

[98] J. L. Feder, T. A. Hunt, and L. Bush, “The effects of climate, host plant phenology and host fidelity on the genetics of apple and hawthorn infesting races of Rhagoletis pomonella,” Entomologia Experimentalis et Applicata,vol.69,no.2,pp. 117–135, 1993. [99] H. R. Dambroski and J. L. Feder, “Host plant and latitude- related diapause variation in Rhagoletis pomonella: a test for multifaceted life history adaptation on different stages of diapause development,” Journal of Evolutionary Biology, vol. 20, no. 6, pp. 2101–2112, 2007. [100] J. L. Feder, “The effects of parasitoids on sympatric host races of Rhagoletis pomonella (Diptera: Tephritidae),” Ecology, vol. 76, no. 3, pp. 801–813, 1995. [101] M. S. Singer, B. Wee, S. Hawkins, and M. Butcher, “Rapid natural and anthropogenic diet evolution: three examples from checkerspot butterflies,” in Specialization, Speciation, and Radiation: The Evolutionary Biology of Herbivorous Insects,K. J. Tilmon, Ed., pp. 311–324, University of California Press, Berkeley, Calif, USA, 2008. [102] D. J. Funk, “Does strong selection promote host specialisation and ecological speciation in insect herbivores? Evidence from Neochlamisus leaf beetles,” Ecological Entomology, vol. 35, no. 1, pp. 41–53, 2010. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 154686, 12 pages doi:10.1155/2012/154686

Review Article Habitat Choice and Speciation

Sophie E. Webster,1 Juan Galindo,1, 2 John W. Grahame,3 and Roger K. Butlin1

1 Department of Animal and Plant Sciences, The University of Sheffield, Western Bank, Sheffield S10 2TN, UK 2 Departamento de Bioqu´ımica, Genética e Inmunolog´ıa, Facultad de Biolog´ıa, Universidad de Vigo, 36310 Vigo, Spain 3 Institute of Integrative and Comparative Biology, Faculty of Biological Sciences, University of Leeds, Leeds LS2 9JT, UK

Correspondence should be addressed to Sophie E. Webster, sophie.webster@sheﬃeld.ac.uk

Received 2 August 2011; Revised 2 November 2011; Accepted 28 November 2011

Academic Editor: Rui Faria

Copyright © 2012 Sophie E. Webster et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The role of habitat choice in reproductive isolation and ecological speciation has often been overlooked, despite acknowledgement of its ability to facilitate local adaptation. It can form part of the speciation process through various evolutionary mechanisms, yet where habitat choice has been included in models of ecological speciation little thought has been given to these underlying mechanisms. Here, we propose and describe three independent criteria underlying ten different evolutionary scenarios in which habitat choice may promote or maintain local adaptation. The scenarios are the result of all possible combinations of the independent criteria, providing a conceptual framework in which to discuss examples which illustrate each scenario. These examples show that the different roles of habitat choice in ecological speciation have rarely been effectively distinguished. Making such distinctions is an important challenge for the future, allowing better experimental design, stronger inferences and more meaningful comparisons among systems. We show some of the practical difficulties involved by reviewing the current evidence for the role of habitat choice in local adaptation and reproductive isolation in the intertidal gastropod Littorina saxatilis,amodel system for the study of ecological speciation, assessing whether any of the proposed scenarios can be reliably distinguished, given current research.

1. Introduction of habitat choice. We define habitat choice as any behaviour that causes an individual to spend more time in one habitat The role of divergent natural selection in speciation has type than another compared with the expectation based on been widely studied in recent years [1]. There is now broad random dispersal (see “habitat selection”, p. 184, [9]). On acceptance that selection of this type can lead to the evolution the basis of this definition, a simple reduction in dispersal of reproductive isolation, even in the face of gene flow [2]. distance would not constitute habitat choice. Examples of Nevertheless, significant controversy remains. Is “ecological mechanisms that might underlie choice include active move- speciation” really distinct from other modes of speciation ment into a preferred habitat; reduced dispersal in the pre- [3]? Why does reproductive isolation remain incomplete ferred habitat relative to a nonpreferred habitat; preferential in some cases but not in others [4]? Do chromosomal re- settling of propagules after a dispersal phase or a change arrangements [5] or divergence hitchhiking [6]helptoover- in the timing of dispersal that influences the probability come the antagonism between selection and recombination? of arriving in a different habitat. Habitats may be spatially What is the role of the so-called “magic traits” [7]? separated on various scales, from abutting distributions to “Habitat isolation” is one part of the ecological barrier a fine-scale mosaic, even different parts of the same host to gene exchange between species that includes effects due to plant [10]. They also need not be separated in space at all: local adaptation, competition, and choice [8]. In this paper, temporal or seasonal separation is possible. In the case of we will focus our attention on habitat choice, discussing the seasonal separation, allochronic emergence or reproduction nature of its role in ecological speciation and the potential [11]effectively constitutes forms of habitat choice but will contribution towards reproductive isolation of various forms not be considered here. 2 International Journal of Ecology

Habitat heterogeneity can lead to ecological speciation Single effect Multiple effect Single effect in the presence of gene flow. It requires divergent selection, which results in the establishment of a multiple-niche polymorphism [1](Figure 1). This might arise in situ or on secondary contact between previously allopatric populations. A mechanism for nonrandom mating must then become 1 allele associated with this polymorphism [12]. As a result of this two-step process, habitat choice can influence the probability of speciation or the degree of reproductive isolation achieved in one of two ways; firstly, habitat choice may increase the 2 alleles range of parameters under which a stable polymorphism can be maintained by selection in a heterogeneous environment [13]. This effect can be independent of any effect on mating pattern if, for example, individuals feed in two contrasting environments but mate away from the food resources. Sec- Polymorphism Both Nonrandom mating ondly, habitat choice may cause assortative mating if it results in partially independent mating pools. Habitat choice of this Figure 1: Habitat choice and the probability of speciation. Darker type may be favoured by selection against hybrids and so colours represent increased probability of speciation. See main text for further explanation. constitutes a form of reinforcement [14]. Choice of mating habitat may cause assortment without influencing local adaptation but, clearly, habitat choice may have both effects (i.e., on polymorphism and assortative mating) simultane- Habitat choice may also evolve by either a “one-allele” or ously in some systems, strongly influencing the likelihood a “two-allele” mechanism [18], as in the “single-variation” and speed of speciation [15]. and “double-variation” models of de Meeusˆ et al. [20]. A “one-allele” scenario might involve “habitat matching” (e.g., Additionally, the trait responsible for habitat choice and [21]), for example, causing an animal to move until it is the locally adapted trait—responsible for adaptations to local cryptic against its background and then to remain stationary. conditions—must generally become associated, typically re- Indirect selection would favour the spread of such an allele quiring the build-up of linkage disequilibrium but poten- in an environment with two backgrounds where predation tially facilitated by pleiotropy (see preference-performance maintained a polymorphism for alternate cryptic colours, correlations in [16]), as is the case for other components without the need for linkage disequilibrium because the of reproductive isolation [2]. However, it is possible for matching allele is favoured in association with both colour habitat choice to be under direct selection and contribute morphs. Alternatively, a “two-allele” mechanism might in- to nonrandom mating, thus constituting a “magic trait” volve one allele favouring upward movement and another (sensu Gavrilets [17]). “Magic traits” are usually discussed in favouring downward movement. Here, the evolution of habi- the context of locally adaptive traits that also contribute to tat choice, and so reproductive isolation, relies on linkage assortative mating, such as signals and preferences, though disequilibrium between the upward movement allele and the concept can clearly be applied to any trait that promotes alleles conferring increased fitness in the high habitat and reproductive isolation. A recent attempt at clarification of the between the downward allele and the alternate alleles at “magic trait” definition suggested “a trait subject to divergent the fitness loci. Note that the “one-allele” versus “two- selection and a trait contributing to nonrandom mating that allele” distinction still holds for polygenic traits and can be are pleiotropic expressions of the same gene(s)” [7] but this made without knowing the genetic basis of the trait—the view unfortunately confounds two distinct ideas: the impact primary issue is whether the trait has to change in the same of a single trait on more than one component of reproductive direction (one-allele) or in different directions (two-allele) in isolation and the effect of a gene on more than one trait. the diverging populations. “One-allele” mechanisms increase Here, we follow the proposal by Smadja and Butlin [2]to the probability of speciation and can potentially be more distinguish “single-effect” and “multiple-effect” traits and effective in limiting gene flow between subpopulations than avoid the use of the, now confusing, term “magic.’’ two-allele traits. As with multiple-effect traits, this is because ff Multiple-e ect traits facilitate the evolution of reproduc- they remove the need for linkage disequilibrium. tive isolation by reducing or removing the need for linkage Habitat choice may be plastic, including the possibility ff disequilibrium and so avoiding the negative e ects of re- of learning, and plasticity can be an important factor in combination [2, 18]. Their contribution to speciation still speciation [22]. The extent of plastic response may vary, depends on the magnitude of their effects [19]. A trait that up to the point where no genetic difference needs to exist contributes to reproductive isolation through habitat choice between individuals showing behavioural preferences for may also contribute to reproductive isolation in other ways, contrasting environments. However, plasticity itself has a including but not exclusively through effects that lead to genetic basis; evolution can act upon the degree of plasticity, direct selection, and so may be a multiple-effect trait. Other both the ability to learn and biases in the way individuals things being equal, we expect such traits to increase the learn can evolve. Evolution of plasticity or learning may probability or speed of speciation. be considered examples of a “one-allele” mechanism, where International Journal of Ecology 3 the ability to modify the learning phenotype in a way that identifying the mechanisms involved. We illustrate some of results in advantageous habitat choice is genetically deter- these difficulties in a final section dealing with one particular mined and the same alleles, for effective learning, are ben- example, the periwinkle Littorina saxatilis. eficial everywhere. These three different criteria for classifying traits respon- 2.1. Polymorphism or Nonrandom Mating. The first of our sible for habitat choice (whether habitat choice influences categories deals with the role of polymorphism and nonran- the maintenance of polymorphism or assortative mating; dom mating in habitat choice scenarios: the establishment of whether habitat choice traits are multiple or single effect; polymorphism generated by divergent selection is a neces- whether they follow a “one-allele” or “two-allele” inheritance sary step in ecological speciation. mechanism) are largely independent. An exception arises Ecological speciation in the presence of gene flow re- because habitat choice may influence both the maintenance quires divergent selection, which results in the establishment of polymorphism and assortative mating, and this is an of polymorphism [1]. The maintenance of genetic variation example of a multiple-effect trait. If all possible combinations within populations in heterogeneous environments has been between the three criteria are considered, they may therefore widely discussed in the past, from theoretical models to lead, at least theoretically, to ten different habitat choice experimental studies (reviewed in [23]). One of the first scenarios (Figure 1). Here we will discuss the implications of models, by Levene [24], showed that the maintenance of each one of these criteria for habitat choice and speciation, polymorphism in an environment with two habitats was including examples of empirical studies and/or models. We possible. In this model there is random migration of will also review and discuss the current evidence for the role individuals into the habitats, selection favours one of the of habitat choice in local adaptation, including its possible genotypes in each habitat, each habitat contributes a constant effect on ecological speciation in an intertidal gastropod, number of individuals to the next generation, and there is the rough periwinkle Littorina saxatilis. We use this example random mating. The range of the parameters necessary to to illustrate some of the practical difficulties involved in maintain polymorphism is rather restricted; nevertheless, distinguishing among habitat choice scenarios empirically. this model was the basis for Maynard Smith’s [25] classic This system also emphasises the generality of the distinctions analysis of sympatric speciation. It is clear that under by comparison with the more widely studied insect models this model habitat choice is favoured by indirect selection in which host plants form the contrasting habitats. (local adaptation to one of the habitats will secondarily promote preference for that habitat) and that habitat choice considerably expands the range of parameters under which 2. Habitat Choice Scenarios the polymorphism is stable. Many subsequent models have shown that frequency- or density-dependent selection makes Figure 1 provides a conceptual classification of habitat choice polymorphism more likely (e.g., [26, 27]) and so extends the scenarios, illustrating the likelihood of progression towards range of situations in which an initial polymorphism will speciation under each scenario (a scenario consisting of a create conditions for the evolution of habitat choice, or where combination of criteria). It suggests that, at one extreme, preexisting habitat choice will favour the establishment of habitat choice may contribute to the maintenance of poly- multiple-niche polymorphism, reinforcing local adaptation morphism without directly contributing towards reproduc- (e.g., [20, 28]). The interplay between habitat choice and tive isolation (as indicated by the lightest area) whereas, at local adaptation has recently been reviewed by Ravigneetal.´ the other extreme, habitat choice evolution may be a primary [13, 29]. From the point of view of speciation, the main- driver of speciation (as indicated by the darkest, central tenance of local adaptation creates opportunities for the areas). For simplicity we have erected a categorical system further evolution of reproductive isolation [25]. Therefore, (our “criteria”). However, due to the dynamic nature of the if habitat choice helps to maintain polymorphism and if evolutionary process, one or more habitat choice scenarios polymorphism reinforces local adaptation, it also increases from our framework may contribute to speciation at dif- the likelihood of speciation. This is true even if habitat choice ferent stages in the process. For example, stabilisation of has no influence on mating pattern. However, if habitat multiple-niche polymorphism may often be an early stage choice also generates assortative mating, its contribution to of speciation whereas the evolution of divergent preferences speciation will be greater (Figure 1). Habitat choice of this for mating habitat could be a form of reinforcement at a late type is favoured by indirect selection when a polymorphism stage of the process. Below, we will discuss the relevance of is established and this constitutes a form of reinforcement each criterion of this classification with reference to illus- [14]. Finally, it is possible that habitat choice may apply trative examples. However, it will quickly become clear that only to the mating habitat and so influence mating patterns there is often insufficient empirical information to be certain without being directly connected to a source of divergent how individual case studies fit into our classification, partly selection. To understand the role of habitat choice in speci- because the role of habitat choice in speciation has not been ation, it is important to distinguish these categories and to addressed in a strong conceptual framework. Additionally, determine the stage in the speciation process at which choice we were unable to identify good empirical examples that evolves. satisfactorily demonstrate some of our proposed scenarios— Natural cases where both polymorphism and habitat this highlights both the limited range of the studies that choice are present in the population (but the choice does not have been undertaken in this area and the difficulty in influence mating pattern) are uncommon in the literature. 4 International Journal of Ecology

From the point of view of speciation, these examples should occurs on the plant and females oviposit in ripening fruit. represent initial stages of speciation or divergence, but Larvae complete their development in the ripe fallen fruits, the outcome of this process is highly dependent on the pupate in the soil, and undergo a facultative diapause until reigning conditions (e.g., habitat distribution, habitat size, spring. Adults congregate again on the host plant for mating. and selection coefficients) and there may actually be no Host races are differentially adapted, primarily through their progression towards greater isolation [4]. The most clear- diapause characteristics which match the timing of the life- cut examples are likely to come from species where mating cycles to differences in phenology of their hosts [34, 35]. occurs in a different habitat from the majority of the life- There are also host-associated differences in larval survival cycle, such as aquatic insects with a brief aerial mating phase. [36]. One interesting possible case concerns the aphid genus Host fidelity in Rhagoletis is partly a result of limited Cryptomyzus in which sibling species, which still occasionally dispersal but there is clear evidence for active habitat choice hybridise in the field, utilise either dead nettle (Lamium gale- involving fruit size and colour and, especially, volatile chem- obdolon) or hemp-nettle (Galeopsis tetrahit) as summer host. ical signatures [37, 38]. Because there are clear fitness differ- On these hosts they reproduce asexually, but their sexual ences and mating occurs on the host fruit, host choice clearly generations occur on redcurrant (Ribes rubrum)regardlessof contributes to both stabilisation of the coexistence of the the summer host [30]. Here, there is assortative mating but it races and to assortative mating between them. is not due to the strong preference for distinct summer hosts. Thehistoricalroleofhostchoiceislesseasilydeter- Rather, it seems to be mediated by differences in the diurnal mined. Feder [39] suggested that individuals with a genetic cycle of pheromone release by females. This situation, where preference for apples, the derived host, may have gained multiple races or species share the same primary host while an immediate selective advantage, perhaps involving the use utilising distinct summer hosts, seems to be common in of an empty niche or escape from parasitism (“apple race aphids. The scenario where habitat choice can maintain a flies have less parasitoids than hawthorn race flies because polymorphism but have no influence on nonrandom mating parasitoids use plant cues when searching for their hosts,” appears to be rare but may simply be under-represented in [40]). In this case, habitat choice may have evolved first and the literature. It is not likely to be favourable for progression facilitated subsequent host adaptation, rather than evolving to complete reproductive isolation. in response to the fitness costs of oviposition on the wrong The opposite situation, where habitat choice influences host. Habitat choice would then be a multiple-effect trait mating alone, is also not widely documented but is also since it would be under direct selection, as well as contribut- most likely where mating occurs in a distinct habitat from ing to assortative mating (see below). A further complication the majority of the life cycle. A possible example may occur to this hypothesis is the association of some host-specific in the mosquito Anopheles gambiae whose larvae develop traits with chromosomal inversions. These inversions appear in small, often temporary water bodies but whose adults to predate the introduction of apples to North America and mate in aerial swarms. The M and S molecular forms of A. may have evolved during a period of allopatry [41]. Their gambiae cooccur in many parts of Africa and show strong, presence in the population may also have facilitated the host but incomplete reproductive isolation. A major contributor shift, interacting with changes in preference. to pre-zygotic isolation appears to be the choice of distinct habitats in which to form mating swarms [31]. This scenario 2.2. Habitat Choice: “Multiple-Effect Trait” versus “Single- is also unfavourable for speciation because there is no close Effect Trait”. The Rhagoletis example nicely illustrates the connection between assortative mating and a source of distinction between multiple-effect and single-effect traits, as divergent selection. defined in Section 1. We can envisage two possible historical In the majority of cases habitat choice is likely to influ- sequences. In the first, multiple-effect, scenario, an allele ence both maintenance of polymorphism and assortative arose in the ancestral hawthorn population which increased mating, as mating usually takes place in the same habitat the likelihood of females ovipositing on apple. This habitat as the life-cycle phase in which selection occurs. A trait choice allele was favoured by direct selection because larvae responsible for habitat choice of this type may be considered developing on apple had higher survival but at the same time a multiple-effect trait because it has effects on reproductive contributed to isolation through its impact on assortative isolation both through assortative mating and through mating. This led to the establishment of a population on enhancing local adaptation. This combination may generate apple trees, which further adapted to the new habitat, a higher probability of progressing to complete reproductive including divergence in diapause timing. Johnson et al. [42] isolation than cases where habitat choice influences only modelled a scenario of this type. An alternative, single-effect, one component of isolation. The situation is typical of scenario would begin with a proportion of females oviposit- many phytophagous insects, which remain on their host to ing on apple by chance. This favoured alleles for high survival mate, and has been very widely studied in this context [32], on apple leading to the establishment of a multiple-niche including important early models (e.g., [28]) and classic polymorphism. Indirect selection could then have favoured research on the apple maggot fly, Rhagoletis pomonella. habitat choice through its impact on assortative mating Sympatric host races of the apple maggot fly in North alone, requiring linkage disequilibrium between survival and America have evolved in the last 150 years, with a host choice alleles to be established in the face of gene flow and shift occurring from hawthorn to apple around 1860 [33]. recombination (a form of reinforcement). Distinguishing Life cycles in these two host races are very similar; mating such alternatives retrospectively is likely to be very difficult. International Journal of Ecology 5

The importance of phenology in Rhagoletis suggests a considered one of the most useful ways to categorise speci- further option. Apple and hawthorn fruit are temporally ation [47]. In principle, habitat choice may evolve by either separated habitats. Therefore, any change in diapause timing mechanism [20] but making the distinction for empirical early in the evolution of the apple race would have consti- examples is not straightforward. Therefore, we begin with tuted a multiple-effect trait under direct selection, because conceptual examples to illustrate the ways in which “one- of the benefits of matching timing to the host and also in- allele” and “two-allele” mechanisms might apply to habitat fluencing habitat choice. Since the timing difference could choice before suggesting possible case studies. have generated assortative mating, this trait would have made In “one-allele” mechanisms, a single allele present in a three distinct contributions to reproductive isolation. population under divergent or disruptive selection generates Interestingly, disrupted host finding in Rhagoletis hybrids habitat preference independently of the direction of selec- also contributes to postzygotic isolation [43], a neglected tion. One possible way for this to work would be through potential contribution of habitat choice to speciation. This natal habitat preference induction (reviewed in [48]), in constitutes a distinct pathway by which a habitat choice trait which experience with a natal habitat shapes the preferences can have multiple effects on reproductive isolation, which is of individuals for that habitat. Experience with particular independent of the polymorphism versus nonrandom mat- stimuli increases subsequent preference for a habitat that ing criterion. For this reason, we consider the single-effect/ contains those same stimuli, which might help dispersing multiple-effect distinction separately, because the impact of a individuals to locate a suitable habitat quickly and efficiently. habitat choice trait on both polymorphism and nonrandom Because assessing habitat quality involves time, risk, and mating is not the only way in which it can act as a multiple- energy invested in sampling potential habitats, selection effect trait. should favour mechanisms that help individuals to select Another good example of a direct link between habitat and use habitats that best suit their phenotypes. Any allele choice and adaptation is the pea aphid, Acyrthosiphon pisum. that spreads in response to such selection would enhance In this phytophagous insect, feeding, mating, and ovipo- divergent habitat choice in a population showing local sition occur only on the host plant. In the northeastern adaptation to multiple niches. Habitat matching, for example United States, host races of the subspecies A. pisum pisum in cryptically coloured species, would have a similar effect live on alfalfa (Medicago)andclover(Trifolium), crops that without the need to condition on the natal habitat (matching are sometimes grown in adjacent fields, and some gene flow habitat choice, [21]), promoting local adaptation and even persists between the races [6]. The aphids acceptance of the leading to speciation. On the other hand, in response to host plant is one of the main reasons for assortative mating: environmental change, adaptation could involve change or pea aphids can distinguish between their preferred host and relaxation of habitat choice instead of adaptation to modified the alternate host by probing with their stylets. When they conditions, resulting in more variable habitat use [21], detect the alternate host they do not feed but will move in which could lead to breakdown of reproductive isolation. search of another plant in order to increase their probabilities Under these conditions the “two-allele” mechanism would of reproductive success [44]. Thus, habitat choice and offer greater resistance to hybridisation than the “one-allele” habitat-associated fitness are aspects of the same underlying mechanism, although “two-allele” habitat choice would be trait of host acceptance, which can be considered a multiple- less likely to initiate the process of speciation. effect trait. As with the apple maggot fly, host acceptance “Two-allele” habitat choice is more likely to involve in- also influences assortative mating, and this situation greatly nate preferences for specific habitat features, such as sub- facilitates the evolution of reproductive isolation [2]. It is strate colour or odour or the presence of particular resources. likely that the key genetic changes of host acceptance involve Divergent or disruptive selection is required to establish such the aphids chemical senses, and the recent characterisation of distinct preferences. chemosensory gene families in the pea aphid [45] opens the Differentiating between “one-allele” (e.g., habitat match- way to identification of the responsible genes. ing) and “two-allele” (distinct preferences in different sub- Hawthorne and Via [46] showed that, for the traits they populations) scenarios is not an easy task, as previously men- defined as host acceptance and host-associated performance, tioned. As an example, we consider colour morphs of another QTL mapped close together in the pea aphid genome. They phytophagous insect, the walking stick Timema cristinae (see suggest that there may be either close linkage between genes [49] for a review). Two host species with highly divergent for the different traits or alleles with pleiotropic effects on leaf colour and shape, Ceanothus spinosus and Adenostoma the traits. Following Smadja and Butlin [2], we suggest that fasciculatum, are utilised. Insects found on different hosts it is more instructive to view host acceptance as a multiple- have different cryptic colour patterns because of selection effect trait with direct effects on fitness. Of course, there may due to predation. These wingless insects feed, mate, and be other traits that also adapt the aphids to different host reproduce on the same host individual and movement environments, such as mechanisms for tackling host defen- between plants is restricted (12 m per generation, [50]). They sive compounds. show host preference [51] and partial assortative mating [52]. Immigrant inviability (selection against ecotypes from 2.3. “One-Allele” versus “Two-Allele” Mechanisms. “One- the contrasting habitat: i.e., host plant) is also an impor- allele” and “two-allele” mechanisms have been widely distant process maintaining the ecotype divergence [53]. It cussed in speciation research since the distinction was intro- is clear that habitat choice contributes to the maintenance duced by Felsenstein [18] and the distinction has been of Timema ecotypes but the mechanistic basis of this 6 International Journal of Ecology preference has not yet been characterised. The preference continuum. Imprinting causes habitat choice, through its could, for example, involve detection of different chemical influence on female choice of host nests, as well as mate compounds on the host plant surfaces that are unrelated to choice and so is a multiple-effect as well as a “one-allele” trait crypsis: a “two-allele” mechanism. Alternatively, the insects [2, 7]. may match their own colouration to the background, a In these sections, we have selected empirical examples mechanism that could fall into the “one-allele” category. that illustrate each of our classifying criteria. Figure 1 shows These possibilities could be distinguished experimentally, for the ways in which these criteria may combine to create example by allowing insects to choose between backgrounds conceptual scenarios with varying probabilities of progres- of different colour/pattern in the absence of plant material. sion to speciation. Working from the framework, how easily Preferences for divergent chemical signals are likely to can these conceptual scenarios be applied to real-world underlie host-plant preference in many phytophagous insects systems? Our brief review of the literature suggests that [54]. These are likely to involve “two-allele” mechanisms habitat choice studies pertaining to ecological speciation are where different alleles for positive or negative responses biased towards phytophagous insects. This is not surprising, [55] to particular stimuli have to be fixed in the diverging because “host race” formation seems to be a common route subpopulations and have to be associated, through linkage to speciation, which has been widely studied. However, in disequilibrium or pleiotropy, with traits that underlie local an attempt to expand the scope of habitat choice studies adaptation. Because this is less favourable to speciation than in an ecological speciation context, we discuss below the matching mechanisms, it is important to make the distinc- evidence for habitat choice in the intertidal gastropod genus tion in more case studies. Littorina. We examine the current evidence for habitat Particularly interesting possible cases of “one-allele” choice, discussing which scenarios are likely to be involved habitat choice involve learning, including imprinting. Such and the difficulties in trying to distinguish them. cases are sometimes described as “nongenetic” but clearly the ability to imprint or the strength of imprinting can have 3. Habitat Choice in Littorina a genetic basis and the spread of a single allele can then cause divergent habitat preferences by promoting imprinting Intertidal gastropods present ideal systems for studies of on different signals. Obligate brood parasitic birds represent habitat choice: the littoral zone can create extreme environ- potential examples of this process. In the continuum of mental gradients and highly heterogeneous habitats within divergence and/or speciation [4, 56, 57], different races or relatively short distances, and generally the animals are easy species pairs represent different stages in the process of to locate and manipulate for both in situ and lab-based trials. ecological speciation, and this is also the case for brood Heterogeneous habitats of this type can lead to differential parasitic birds. Indigobirds (genus Vidua) are obligate, host- survival and generate divergent selection in polymorphic specific brood parasites of firefinches (genus Lagonosticta) populations. Microhabitat use in this landscape has been and other estrildid finches (family Estrildidae). Assortative identified as strongly influencing survival in intertidal gas- mating is due to song learning and mimicry of the host tropods [65–67], so habitat choice presents itself as a likely song by the males [58, 59], and this allows a host switch trait to respond to this selection. to occur in a single generation. Females also learn the song Large-scale transplant experiments have indicated habi- of their foster parents and choose their mates and the nests tat preference behaviour in Littorina species, such as L. keenae they parasitise using song. Different degrees of divergence are [68], L. angulifera [69], and L. unifasciata [70]. All show that found depending on the species under study, for example, the snails tend to return to the approximate tidal height from some of them are morphologically indistinguishable [60, 61] which they were displaced, exhibiting directional movement and they lack genetic differentiation at neutral markers [62]. towards the shore level of origin. However, these transplant When host species of different indigobirds have overlapping experiments may also be influenced by effects of differential distributions, hybridisation can occur due to egg-laying mis- survival that are hard to separate from behavioural effects. takes(e.g.,afemalelaysanegginanestparasitisedby We will examine this problem below. another indigobird species). Individuals of different species Littorina saxatilis, the rough periwinkle, is a marine gas- would then learn the same song, through imprinting, and are tropod that is emerging as a model system for studying eco- likely to hybridise. However, in most of the cases, indigobird logical speciation. It is widely distributed across rocky shore- species have evolved several other polymorphisms, such as lines in the North Atlantic, extremely polymorphic (shell different male plumage colour and nestling mouth markings colour, shell shape, and behaviour), and prone to ecotype that match those of their respective hosts [63], thereby formation due to local adaptation because of its low average enabling young indigobirds to better compete for parental dispersal [71]. Pairs of phenotypically divergent ecotypes care in host nests. These polymorphisms represent different occupying different niches in the intertidal zone are found axes of divergence (greater “ecological niche dimensionality,” over scales of tens of metres or shorter across different shores [64]) promoting increased divergence, despite the possibility along its distribution and are maintained through divergent of accidental hybridisation in regions where host ranges over- natural selection [72]. These ecotypes of L. saxatilis have lap. This suggests that, following colonisation of a new host, been studied in detail on shores from three geographical host fidelity due to imprinting can be sufficient for divergent regions (Sweden, UK, and Spain), and a process of parallel natural and/or sexual selection on morphology, ecology, ecological speciation between them has been suggested ([72], and/or behaviour to generate progress on the speciation but see [73, 74]). However, despite displaying phenotypic International Journal of Ecology 7 divergence, the ecotypes are not completely reproductively the gradient by strengthening barriers to gene flow. In the isolated, with gene flow still occurring (Sweden: [75], UK: middle of the shore gradient in Spain, mussel and barnacle [76], and Spain: [77]). dominated patches are intermingled on a scale of a few The ecotype pairs on each of these shores are separated centimetres and the RB and SU L. saxatilis ecotypes are on a microgeographic scale, exhibiting adaptations to the associated with these patches [89]. This is strongly suggestive prevailing habitat. On Swedish shores, the habitat is com- of active habitat choice [90]. The heterogeneous nature of posed of a mosaic of cliff habitat punctuated with boulder this connecting habitat may be particularly important in fields, whereas the UK and Spanish ecotype pairs are found the maintenance of the hybrid zone and the segregation on the same shores, but at different levels of the littoral of the ecotypes, as has been demonstrated with Bombina zone. The ecotypes in the UK are known as H and M (high- toads [91]. Nevertheless, in order to determine the role of shore and midshore), those in Sweden are termed E and S habitat choice in maintaining divergent populations, in L. (exposed and sheltered), and the Spanish pair are termed RB saxatilis and in other comparable systems, it is necessary to and SU (ridged-banded and smooth-unbanded) (see [73]). utilise manipulative experiments (e.g., using mark-recapture The M, S, and RB ecotypes are morphologically congruent, approaches). exhibiting thick shells, relatively small shell apertures, and Clear evidence for home-site advantages in littorinid large body size. These features are considered to be adapta- species has been documented [83, 89, 92] along with evi- tions to an important selection pressure: predation by crabs. dence of selection on shell characters. In this context, we The H, E, and SU ecotypes from these three shores also consider a home-site advantage to be where individuals are share similar morphological characters: smaller size, thinner likely to have increased fitness in the habitat or microhabitat shells, and a larger shell aperture. In avoiding the hazards to which their ecotype may be presumed to be adapted. This of crab predation by their position on the shore (low in advantage may vary at different stages of the life history. Is Spain, high in Britain and Sweden), these ecotypes are free to there also good evidence for habitat choice in the L. saxatilis develop larger shell apertures, increasing foot area and thus ecotypes? Has L. saxatilis evolved habitat choice in response grip on the substrate to minimise dislodgement. Nonrandom to divergent selection, or did nonrandom mating and mating is also observed in each population of ecotype pairs adaptive polymorphism evolve in the presence of preexisting [78–81], primarily due to assortative mating by size (see habitat choice? Habitat choice can be an adaptive behaviour, [72] for a review). L. saxatilis lacks a pelagic larval stage, increasing fitness in the “home” habitat even when only instead exhibiting direct development where females retain a single habitat type is occupied [93] and so could have their brood internally and release fully formed young [71]. been present before ecotype differentiation began. Is habitat Many other littorinid species (such as L. littorea)producea choice a multiple-effect trait in Littorina andisitbasedon pelagic larval stage, allowing dispersal over a wide range and “one-allele” or “two-allele” genetic variation? We discuss the maintaining gene flow between populations [71]. The low evidence for the presence of habitat choice in L. saxatilis dispersal range of L. saxatilis (1–4 m, [82, 83]) limits gene and consider whether it is possible to make any of these flow, and this facilitates much greater local adaptation in this distinctions. species than in many of its congeners [72]. Work on Swedish populations indicates that morpholog- Since selection drives differential adaptation to closely ical adaptation to the contrasting environment has a strong adjacent habitats, habitat choice mechanisms could easily be genetic basis but has an element of plasticity which can imagined to play a role in population divergence. Random improve local adaptation [94, 95]. However, the E (exposed) dispersal combined with selection against less fit phenotypes ecotype displayed significantly lower levels of plasticity than may superficially look like habitat choice as the phenotypes the S (sheltered) ecotype, indicating differential plasticity are segregated into divergent habitats, as noted above. This within local populations. This leads to the question of the is particularly true where dispersal distance is short allowing effect of plasticity on the role of habitat choice: it is feasible selection to produce sharp phenotypic transitions at habitat that lower phenotypic plasticity might favour the evolution boundaries. Selection for reduced postnatal dispersal [84] of genetically based habitat choice, to increase occupation of may accentuate this effect. However, as there is no active be- the environment in which individuals are more fit. Increased havioural mechanism, this does not constitute habitat choice plasticity may decrease selection for habitat choice, since as we define it. individuals would be better able to adapt their phenotype A possible exception, where habitat choice can be in- to local conditions. This might be tested if the degree of ferred from phenotypic distributions, is where habitat het- morphological plasticity varies among regions, leading to a erogeneity is on a scale much smaller than the dispersal dis- prediction of varying habitat choice. tance. It is then not possible for selection to maintain genetic Experimental evidence for nonrandom dispersal in differentiation between patches [85] independent from habi- L. saxatilis ecotype populations has been obtained in both tat choice, although phenotypic plasticity could still result Spain and Sweden [78, 82, 83]. In the Swedish populations, in strong phenotype-habitat associations. Morphological displaced snails exhibited greater average dispersal distances and AFLP (amplified fragment length polymorphism, see than nondisplaced ones and dispersal differed between E and [86] for a review) clines have been identified, which are S ecotypes, in addition to a tendency to recapture snails in too steep to have been generated by selection alone [87, their own habitat more often than expected from random 88]. In these cases, additional mechanisms such as habitat dispersal [83]. However, this tendency to recapture snails in choice may contribute to the genesis and maintenance of their own habitat may be a function of differential survival 8 International Journal of Ecology in native and nonnative habitats. Additionally, although sur- and could be criticised for releasing snails at high densities in vival rates and migration distances were measured, direction potentially unnatural positions. The analyses do not provide of movement was not. Erlandsson et al. [82]expandedon quantitative estimates of survival, dispersal, or their ecotype- this study to determine whether the dispersal was directional habitat interactions, the sort of variables that would be in the Spanish population. They detected random dispersal needed to model likely evolutionary scenarios. A recent when snails were placed at their native shore level (with theoretical model of ecotype formation in Littorina saxatilis overall dispersal distances averaging less than 2 m), whereas [96] did not include habitat choice as a parameter. Due to when animals were transplanted to their nonnative shore the currently unknown contribution of habitat choice to level they moved further and more directionally, with reproductive isolation in this species, it would be interesting the Spanish RB morph exhibiting the greatest directional to see how the potential to evolve choice might influence response. Although this hinted at habitat choice in this model outcomes. ecotype, the recapture rate was low (<20%) and conclusions Cruz et al. [78] also discussed the possible behavioural were drawn only from recaptured individuals (which are basis of habitat choice. The observation that shell morphol- likely to be a strongly selected sample), therefore it is difficult ogy provided the best predictor of habitat-specific viability to make any meaningful conclusions about habitat choice but that sampling location best explained the pattern of from these simpler experimental studies. movement led them to suggest that shell shape and habitat On the Galician shore in Spain, Cruz et al. [78]triedto choice are genetically independent. Therefore, in the terms separate survival and habitat preference in the two ecotypes we use here, shell shape is not a multiple-effect trait in the of L. saxatilis using two reciprocal transplant experiments. sense that a change in shape alters the fitness profile but In the first experiment, sample groups of each ecotype were does not automatically alter habitat preference (as it might transplanted both to their native and nonnative shore level if snails had a pre-existing tendency to move to a habitat at each of two sites. Snails were then recaptured and their that was favourable for their shell shape). Note, however, that movement recorded from two days after transplant. The shell size does seem to be multiple-effect in that it influences study compared the recapture positions of the transplanted both differential survival and assortative mating. A “two- snails and the recapture positions of the corresponding allele” mechanism (or “double-variation”: [20, 78]) seems control snails, to correct for movements that may been more likely than a “one-allele” mechanism for the same induced via prevailing climatic conditions. In addition to reason. Movement to the optimum habitat could be a “one- the confounding effectthatsuchforcedmigrationmight allele” mechanism but would result in a strong association have had on the snails survival, transplanted snails may between shell shape and differential movement, which may have dispersed beyond the study area leading to reduced occur prior to local adaptation. An upward bias in RB and recapture.OnewayinwhichCruzetal.[78] avoided these a downward bias in SU would most likely be a two-allele complications was to argue that only directional movement mechanism, dependent on linkage disequilibrium and so could result in more than 50% of the released snails in the less tightly linked to shape. Under this assumption, habitat treatment group being recaptured in the direction of their choice could evolve after local adaptation. Lack of preference preferred habitat (up shore for RB, down shore for SU). With in hybrids [90] tends to support this conclusion, suggesting this stringent criterion, habitat choice was only observed in that habitat choice evolved after local adaptation whereas one site. a tendency toward matching position to optimal habitat The second experiment involved collection of snails from preceded local adaptation. However, direct behavioural tests five intertidal levels on each of two sites and reciprocal and genetic analysis are needed to confirm these speculations transplants across sites for release. This destroyed the cor- andwillbedifficult if habitat choice is as weak in other relation between shore level and snail phenotype (measured regions as it seems to be in Spain. as the first axis of shell shape variation). Over a period Other littorinid studies highlight the role of chemore- of two weeks, they observed the reformation of the shell- ception in influencing the behaviour of individuals (such shape cline and measured the relative contributions, for the as trail following: [97, 98]). It has been determined that cline reformation, of the snails migrating movements and L. saxatilis E ecotype males (S males were not studied) are survival. Using a clever comparison between the change in able to discriminate between mucous trails of the female average positions of all snails recaptured and the change of each ecotype [99] and show a clear preference for trails in position of those that were known to have survived, of females within the size range of the E ecotype females. they separated the contributions of differential survival and In addition to its role in assortative mating, trail following habitat choice to the changing cline. Differential movement could play a role in habitat choice. This has been studied contributed between one-third and one-half of the change in to an extent in L. littorea [100]. When chemical cues were cline at one site and hardly at all at the other site, leading removed from the “home” boulder and substrate, L. littorea to the conclusion that habitat choice was present but less displayed a significantly impaired ability to navigate back important than differential viability in the maintenance of to the boulder from which they had been displaced. In the phenotypic cline. L. saxatilis it would be interesting to separate the role This important study illustrates the difficulties associated of assortative mating from habitat choice. Trail following with demonstrating habitat choice in the field. Despite could impact dispersal experiments by making individual considerable effort and thoughtful design, the experiments movements nonindependent [78] and, if it forms the basis were still hampered by low recapture rates (around 50%) of philopatry, could represent a “one-allele” habitat choice International Journal of Ecology 9 mechanism. More studies are needed to unveil the role of L. saxatilis is an important facet in our overall understanding chemoreception in habitat choice in this species. of the processes and mechanisms leading to species forma- Did size-assortative mating evolve after ecological parti- tion. tioning and evolution of habitat choice? Or did habitat choice Describing the role of habitat choice within the concep- facilitate ecotype formation after the development of assorta- tual framework that we propose represents an important step tive mating? The model by Sadedin et al. [96] suggested that in understanding speciation. It shows how habitat choice can assortative mating may be considered ancestral. However, affect reproductive isolation in very different manners, influ- although a number of ecological factors were modelled, encing the likelihood of speciation and potentially leading to habitat choice was not included as a parameter. Dispersal different stages along the continuum of speciation. Empirical was included, but this was not directional. Dispersal was studies of habitat choice in divergent populations or closely an important consideration though: frequent long-range related species, representing different stages of speciation random dispersal eliminated spatial genetic structure and and different evolutionary scenarios, should form a focus did not lead to ecotype formation. Although the role and for future research. When analysed within such a conceptual mechanism of habitat choice in Littorina have not yet been framework, we believe these studies will give more insight explicitly modelled, we may draw inferences from models into the part that habitat choice plays in ecological speciation developed for other organisms. Early models suggested that than if they are considered in isolation. when fitness, mating, and habitat choice are all based on the same character, speciation with gene flow may result– Acknowledgments the degree of reproductive isolation is determined by the strength of assortative mating and the strength of disruptive The authors would like to thank Rui Faria, Steven Parratt, selection. For moderate selection, habitat-based nonrandom and the three anonymous referees for their useful comments mating also facilitates reproductive isolation. However, in on the paper. All authors are supported by the Natural Envi- simulations, the size-related mate choice mechanism in ronment Research Council (NERC). J. Galindo is currently L. saxatilis could not explain more than a small part of the funded by “Isidro Parga Pondal” research fellowship from sexual isolation between morphs [101]. This implies that Xunta de Galicia (Spain). size-related mate choice, although considered a multiple- effect trait, may only be important in a speciation context References if it evolves in parallel with other ecological traits, including habitat choice. [1] D. Schluter, “Evidence for ecological speciation and its alter- These studies highlight the difficulties in connecting native,” Science, vol. 323, no. 5915, pp. 737–741, 2009. [2] C. M. Smadja and R. K. Butlin, “A framework for comparing theoretical evolutionary scenarios with existing empirical processes of speciation in the presence of gene flow,” Molecu- data. However, if future habitat choice studies are carried out lar Ecology, vol. 20, no. 24, pp. 5123–5140, 2011. with an explicit conceptual framework in mind and across [3] J. M. Sobel, G. F. Chen, L. R. Watt, and D. W. Schemske, “The a wide range of study systems, they will contribute more biology of speciation,” Evolution, vol. 64, no. 2, pp. 295–315, effectively to our understanding of speciation. 2010. [4] P. Nosil, L. J. Harmon, and O. Seehausen, “Ecological expla- 4. Conclusion nations for (incomplete) speciation,” Trends in Ecology and Evolution, vol. 24, no. 3, pp. 145–156, 2009. The influence of habitat choice on ecological speciation [5] M. Kirkpatrick and N. Barton, “Chromosome inversions, clearly varies in both magnitude and mechanism, and in local adaptation and speciation,” Genetics, vol. 173, no. 1, pp. many cases we cannot be sure about the contribution it 419–434, 2006. [6] S. Via, “Natural selection in action during speciation,” Pro- makes to reproductive isolation or at what stage it evolved. ceedings of the National Academy of Sciences of the United The empirical examples discussed for some of our projected States of America, vol. 106, pp. 9939–9946, 2009. scenarios provide an indication of which evolutionary sce- [7]M.R.Servedio,G.S.V.Doorn,M.Kopp,A.M.Frame,andP. narios have been observed in natural systems. We would Nosil, “Magic traits in speciation: magic but not rare?” Trends expect those scenarios where habitat choice does not strongly in Ecology and Evolution, vol. 26, no. 8, pp. 389–397, 2011. favour progress towards speciation to be detected in studies [8]J.A.CoyneandH.A.Orr,Speciation, Sinauer, Sunderland, of within-species polymorphism, whereas those promoting Mass, USA, 2004. speciation may be more prevalent among studies of ecolog- [9] D. J. Futuyma, “Ecological specialization and generalization,” ical speciation. A more exhaustive review is needed to test in Evolutionary Ecology,C.W.Fox,D.A.Roff,andD.J. this prediction but may be premature since many case studies Fairbairn, Eds., pp. 177–189, Oxford University Press, New do not yet provide enough information to distinguish among York, NY, USA, 2001. scenarios for the evolution of habitat choice. [10] R. F. Denno, M. S. McClure, and J. R. Ott, “Interspecific interactions in phytophagous insects: competition reexamined Although there have been some valuable habitat choice and resurrected,” Annual Review of Entomology, vol. 40, pp. studies on Littorina saxatilis, there are still a lot of unan- 297–331, 1995. swered questions regarding its role in the maintenance of [11]E.Tauber,H.Roe,R.Costa,J.M.Hennessy,andC.P. both the phenotypic and genetic clines. As a candidate system Kyriacou, “Temporal mating isolation driven by a behavioral for ecological speciation, the understanding of the role of gene in Drosophila,” Current Biology, vol. 13, no. 2, pp. 140– habitat choice prior to complete reproductive isolation in 145, 2003. 10 International Journal of Ecology

[12] M. Kirkpatrick and V. Ravigne,´ “Speciation by natural and [31] A. Diabate,´ A. Dao, A. S. Yaro et al., “Spatial swarm segrega- sexual selection: models and experiments,” American Natu- tion and reproductive isolation between the molecular forms ralist, vol. 159, no. 3, pp. S22–S35, 2002. of Anopheles gambiae,” Proceedings of the Royal Society B, vol. [13] V. Ravigne,´ U. Dieckmann, and I. Olivieri, “Live where you 276, no. 1676, pp. 4215–4222, 2009. thrive: joint evolution of habitat choice and local adaptation [32] K. W. Matsubayashi, I. Ohshima, and P. Nosil, “Ecological facilitates specialization and promotes diversity,” American speciation in phytophagous insects,” Entomologia Experimen- Naturalist, vol. 174, no. 4, pp. E141–E169, 2009. talis et Applicata, vol. 134, no. 1, pp. 1–27, 2010. [14] M. R. Servedio and M. A. F. Noor, “The role of reinforcement [33]G.L.Bush,J.L.Feder,S.H.Berlocher,B.A.McPheron, in speciation: theory and data,” Annual Review of Ecology, D. C. Smith, and C. A. Chilcote, “Sympatric origins of R. Evolution, and Systematics, vol. 34, pp. 339–364, 2003. pomonella,” Nature, vol. 339, no. 6223, p. 346, 1989. [15] A. P. Hendry, P. Nosil, and L. H. Rieseberg, “The speed of [34] J. L. Feder, J. B. Roethele, B. Wlazlo, and S. H. Berlocher, ecological speciation,” Functional Ecology,vol.21,no.3,pp. “Selective maintenance of allozyme differences among sym- 455–464, 2007. patric host races of the apple maggot fly,” Proceedings of the [16] J. Jaenike and R. D. Holt, “Genetic variation for habitat pref- National Academy of Sciences of the United States of America, erence: evidence and explanations,” American Naturalist, vol. vol. 94, no. 21, pp. 11417–11421, 1997. 137, pp. S67–S90, 1991. [35] D. C. Smith, “Heritable divergence of Rhagoletis pomonella [17] S. Gavrilets, Fitness landscapes and the Origin of Species, host races by seasonal asynchrony,” Nature, vol. 336, no. 6194, Princeton University Press, Princeton, NJ, USA, 2004. pp. 66–67, 1988. [18] J. Felsenstein, “Skepticism towards Santa Rosalia, or why are [36] D. Schwarz and B. A. McPheron, “When ecological isolation there so few kinds of animals?” Evolution, vol. 35, pp. 124– breaks down: sexual isolation is an incomplete barrier to 138, 1981. hybridization between Rhagoletis species,” Evolutionary Ecol- [19]B.C.Haller,L.F.DeLeon,´ G. Rolshausen, K. M. Gotanda, ogy Research, vol. 9, no. 5, pp. 829–841, 2007. and A. P. Hendry, “Magic traits: distinguishing the important [37] C. Linn, J. L. Feder, S. Nojima, H. R. Dambroski, S. H. from the trivial,” Trends in Ecology and Evolution, vol. 27, no. Berlocher, and W. Roelofs, “Fruit odor discrimination and 1, pp. 4–5, 2012. sympatrichostraceformationinRhagoletis,” Proceedings [20] T. de Meeus,ˆ Y. Michalakis, F. Renaud, and I. Olivieri, of the National Academy of Sciences of the United States of “Polymorphism in heterogeneous environments, evolution America, vol. 100, no. 20, pp. 11490–11493, 2003. of habitat selection and sympatric speciation: soft and hard [38] C. E. Linn, H. Dambroski, S. Nojima, J. L. Feder, S. H. selection models,” Evolutionary Ecology, vol. 7, no. 2, pp. 175– Berlocher, and W. L. Roelofs, “Variability in response speci- 198, 1993. ficity of apple, hawthorn, and flowering dogwood-infesting [21] P. Edelaar, A. M. Siepielski, and J. Clobert, “Matching habitat Rhagoletis flies to host fruit volatile blends: implications choice causes directed gene flow: a neglected dimension in for sympatric host shifts,” Entomologia Experimentalis et evolution and ecology,” Evolution, vol. 62, no. 10, pp. 2462– Applicata, vol. 116, no. 1, pp. 55–64, 2005. 2472, 2008. [39] J. L. Feder, “The apple maggot fly, Rhagoletis pomonella: flies [22] D. W. Pfennig, M. A. Wund, E. C. Snell-Rood, T. Cruick- in the face of conventional wisdom about speciation?” in shank, C. D. Schlichting, and A. P. Moczek, “Phenotypic Endless Forms,D.J.HowardandS.H.Berlocher,Eds.,pp. plasticity’s impacts on diversification and speciation,” Trends 130–144, Oxford University Press, New York, NY, USA, 1998. in Ecology and Evolution, vol. 25, no. 8, pp. 459–467, 2010. [40] J. L. Feder, “The effects of parasitoids on sympatric host races [23] P. W. Hedrick, “Genetic-polymorphism in heterogeneous of Rhagoletis pomonella (Diptera: Tephritidae),” Ecology, vol. environments: a decade later,” Annual Review of Ecology and 76, no. 3, pp. 801–813, 1995. Systematics, vol. 17, pp. 535–566, 1986. [41] J. L. Feder, J. B. Roethele, K. Filchak, J. Niedbalski, and J. [24] H. Levene, “Genetic equilibrium when more than one eco- Romero-Severson, “Evidence for inversion polymorphism logical niche is available,” American Naturalist, vol. 87, pp. related to sympatric host race formation in the apple maggot 331–333, 1953. fly, Rhagoletis pomonella,” Genetics, vol. 163, no. 3, pp. 939– [25] J. Maynard Smith, “Sympatric speciation,” American Natu- 953, 2003. ralist, vol. 100, pp. 637–650, 1966. [42] P. A. Johnson, F. C. Hoppensteadt, J. J. Smith, and G. L. [26] D. Udovic, “Frequency-dependent selection, disruptive selec- Bush, “Conditions for sympatric speciation: a diploid model tion, and the evolution of reproductive isolation,” American incorporating habitat fidelity and non-habitat assortative Naturalist, vol. 116, pp. 621–641, 1980. mating,” Evolutionary Ecology, vol. 10, no. 2, pp. 187–205, [27] D. Sloan Wilson and M. Turelli, “Stable underdominance 1996. and the evolutionary invasion of empty niches,” American [43] C. E. Linn, H. R. Dambroski, J. L. Feder, S. H. Berlocher, Naturalist, vol. 127, no. 6, pp. 835–850, 1986. S. Nojima, and W. L. Roelofs, “Postzygotic isolating factor [28] S. R. Diehl and G. L. Bush, “The role of habitat preference in sympatric speciation in Rhagoletis flies: reduced response in adaptation and speciation,” in Speciation and Its Conse- of hybrids to parental host-fruit odors,” Proceedings of the quences, D. Otte and J. Endler, Eds., pp. 345–365, Sinauer National Academy of Sciences of the United States of America, Associates, Sunderland, Mass, USA, 1989. vol. 101, no. 51, pp. 17753–17758, 2004. [29] V. Ravigne,´ I. Olivieri, and U. Dieckmann, “Implications of [44] M. C. Caillaud and S. Via, “Specialized feeding behavior habitat choice for protected polymorphisms,” Evolutionary influences both ecological specialization and assortative mat- Ecology Research, vol. 6, no. 1, pp. 125–145, 2004. ing in sympatric host races of pea aphids,” American Natural- [30] J. A. Guldemond and A. F. G. Dixon, “Specificity and daily ist, vol. 156, no. 6, pp. 606–621, 2000. cycle of release of sex pheromones in aphids: a case of [45] C. Smadja, P. Shi, R. K. Butlin, and H. M. Robertson, “Large reinforcement,” Biological Journal of the Linnean Society, vol. gene family expansions and adaptive evolution for odorant 52, no. 3, pp. 287–303, 1994. and gustatory receptors in the pea aphid, Acyrthosiphon International Journal of Ecology 11

pisum,” Molecular Biology and Evolution, vol. 26, no. 9, pp. [63] R. B. Payne, “Nestling mouth markings and colors of 2073–2086, 2009. Old World finches Estrildidae: mimicry and coevolution [46] D. J. Hawthorne and S. Via, “Genetic linkage of ecological of nestling finches and their brood parasite,” Miscellaneous specialization and reproductive isolation in pea aphids,” Publications Museum of Zoology University of Michigan, vol. Nature, vol. 412, no. 6850, pp. 904–907, 2001. 194, pp. 1–45, 2005. [47] M. R. Servedio, “The role of linkage disequilibrium in the [64] P. Nosil and C. P. Sandoval, “Ecological niche dimensionality evolution of premating isolation,” Heredity, vol. 102, no. 1, and the evolutionary diversification of stick insects,” PLoS pp. 51–56, 2009. ONE, vol. 3, no. 4, Article ID e1907, 2008. [48] J. M. Davis and J. A. Stamps, “The effect of natal experience [65] E. G. Boulding, “Mechanisms of differential survival and on habitat preferences,” Trends in Ecology and Evolution, vol. growth of two species of Littorina on wave-exposed and on 19, no. 8, pp. 411–416, 2004. protected shores,” Journal of Experimental Marine Biology [49] P. Nosil, “Divergent host plant adaptation and reproductive and Ecology, vol. 169, no. 2, pp. 139–166, 1993. isolation between ecotypes of Timema cristinae walking [66] K. M. M. Jones and E. G. Boulding, “State-dependent habitat sticks,” American Naturalist, vol. 169, no. 2, pp. 151–162, selection by an intertidal snail: the costs of selecting a physi- 2007. cally stressful microhabitat,” Journal of Experimental Marine [50] C. P. Sandoval, Geographic, ecological and behavioral factors Biology and Ecology, vol. 242, no. 2, pp. 149–177, 1999. ff a ecting spatial variation in color or morph frequency in the [67] R. P. Kovach and D. A. Tallmon, “Strong influence of walking-stick Timema cristinae, Ph.D. thesis, University of microhabitat on survival for an intertidal snail, Nucella lima,” California, Santa Barbara, Calif, USA, 1993. Hydrobiologia, vol. 652, no. 1, pp. 49–56, 2010. [51] P. Nosil, C. P. Sandoval, and B. J. Crespi, “The evolution [68] L. P. Miller, M. J. O’Donnell, and K. J. Mach, “Dislodged of host preference in allopatric vs. parapatric populations but not dead: survivorship of a high intertidal snail following of Timema cristinae walking-sticks,” Journal of Evolutionary wave dislodgement,” Journal of the Marine Biological Associa- Biology, vol. 19, no. 3, pp. 929–942, 2006. tion of the United Kingdom, vol. 87, no. 3, pp. 735–739, 2007. [52] P. Nosil, B. J. Crespi, and C. P. Sandoval, “Host-plant adap- [69] D. A. Antwi and C. Ameyaw-Akumfi, “Migrational orienta- tation drives the parallel evolution of reproductive isolation,” tion in two species of littoral gastropods (Littorina angulifera Nature, vol. 417, no. 6887, pp. 440–443, 2002. and Nerita senegalensis),” Marine Biology,vol.94,no.2,pp. [53] P. Nosil, T. H. Vines, and D. J. Funk, “Perspective: reproduc- 259–263, 1987. tive isolation caused by natural selection against immigrants [70] M. G. Chapman, “Assessment of variability in responses from divergent habitats,” Evolution, vol. 59, no. 4, pp. 705– of intertidal periwinkles to experimental transplantations,” 719, 2005. Journal of Experimental Marine Biology and Ecology, vol. 236, [54] C. Smadja and R. K. Butlin, “On the scent of speciation: the no. 2, pp. 171–190, 1999. chemosensory system and its role in premating isolation,” [71] D. G. Reid, Systematics and Evolution of Littorina,RaySociety, Heredity, vol. 102, no. 1, pp. 77–97, 2008. London, UK, 1996. [55] J. Feder, S. Egan, and A. A. Forbes, “Ecological adaptation and speciation: the evolutionary significance of habitat [72] E. Rolan-Alvarez,´ “Sympatric speciation as a by-product of avoidance as a post-zygotic reproductive barrier to gene ecological adaptation in the Galician Littorina saxatilis hybrid flow,” International Journal of Ecology. In press. zone,” Journal of Molluscan Studies, vol. 73, no. 1, pp. 1–10, [56] A. P. Hendry, “Ecological speciation! Or the lack thereof?” 2007. Canadian Journal of Fisheries and Aquatic Sciences, vol. 66, [73] R. K. Butlin, J. Galindo, and J. W. Grahame, “Sympatric, no. 8, pp. 1383–1398, 2009. parapatric or allopatric: the most important way to classify [57] J. Mallet, “Hybridization, ecological races and the nature speciation?” Philosophical Transactions of the Royal Society B, of species: empirical evidence for the ease of speciation,” vol. 363, no. 1506, pp. 2997–3007, 2008. Philosophical Transactions of the Royal Society B, vol. 363, no. [74] K. Johannesson, M. Panova, P. Kemppainen, C. Andre,´ E. 1506, pp. 2971–2986, 2008. Rolan-Alvarez, and R. K. Butlin, “Repeated evolution of [58] R. B. Payne, L. L. Payne, and J. L. Woods, “Song learning in reproductive isolation in a marine snail: unveiling mecha- brood-parasitic indigobirds Vidua chalybeata: song mimicry nisms of speciation,” Philosophical Transactions of the Royal of the host species,” Animal Behaviour,vol.55,no.6,pp. Society B, vol. 365, no. 1547, pp. 1735–1747, 2010. 1537–1553, 1998. [75] M. Panova, J. Hollander, and K. Johannesson, “Site-specific [59] R. B. Payne, L. L. Payne, J. L. Woods, and M. D. Sorenson, genetic divergence in parallel hybrid zones suggests non- “Imprinting and the origin of parasite-host species asso- allopatric evolution of reproductive barriers,” Molecular ciations in brood-parasitic indigobirds, Vidua chalybeata,” Ecology, vol. 15, no. 13, pp. 4021–4031, 2006. Animal Behaviour, vol. 59, no. 1, pp. 69–81, 2000. [76] C. S. Wilding, R. K. Butlin, and J. Grahame, “Differential [60] C. N. Balakrishnan, K. M. Sefc, and M. D. Sorenson, gene exchange between parapatric morphs of Littorina sax- “Incomplete reproductive isolation following host shift in atilis detected using AFLP markers,” Journal of Evolutionary brood parasitic indigobirds,” Proceedings of the Royal Society Biology, vol. 14, no. 4, pp. 611–619, 2001. B, vol. 276, no. 1655, pp. 219–228, 2009. [77] J. Galindo, P. MorAn,´ and E. RolAn-Alvarez,´ “Comparing [61] R. B. Payne, C. R. Barlow, C. N. Balakrishnan, and M. D. geographical genetic differentiation between candidate and Sorenson, “Song mimicry of black-bellied Firefinch Lagonos- noncandidate loci for adaptation strengthens support for ticta rara and other finches by the brood-parasitic Cameroon parallel ecological divergence in the marine snail Littorina indigobird Vidua camerunensis in West Africa,” Ibis, vol. 147, saxatilis,” Molecular Ecology, vol. 18, no. 5, pp. 919–930, 2009. no. 1, pp. 130–143, 2005. [78] R. Cruz, C. Vilas, J. Mosquera, and C. Garc´ıa, “Relative [62] K. M. Sefc, R. B. Payne, and M. D. Sorenson, “Genetic contribution of dispersal and natural selection to the main- continuity of brood-parasitic indigobird species,” Molecular tenance of a hybrid zone in Littorina,” Evolution, vol. 58, no. Ecology, vol. 14, no. 5, pp. 1407–1419, 2005. 12, pp. 2734–2746, 2004. 12 International Journal of Ecology

[79] J. Hollander, M. Lindegarth, and K. Johannesson, “Local Nerita textilis,” Biological Bulletin, vol. 168, pp. 214–221, adaptation but not geographical separation promotes assor- 1985. tative mating in a snail,” Animal Behaviour,vol.70,no.5,pp. [98] M. S. Davies and P. Beckwith, “Role of mucus trails and trail- 1209–1219, 2005. following in the behaviour and nutrition of the periwinkle [80] S. L. Hull, “Assortative mating between two distinct micro- Littorina littorea,” Marine Ecology Progress Series, vol. 179, pp. allopatric populations of Littorina saxatilis (Olivi) on the 247–257, 1999. northeast coast of England,” Hydrobiologia, vol. 378, no. 1– [99] K. Johannesson, J. N. Havenhand, P. R. Jonsson, M. Linde- 3, pp. 79–88, 1998. garth, A. Sundin, and J. Hollander, “Male discrimination of [81] A. R. Pickles and J. Grahame, “Mate choice in divergent female mucous trails permits assortative mating in a marine morphs of the gastropod mollusc Littorina saxatilis (Olivi): snail species,” Evolution, vol. 62, no. 12, pp. 3178–3184, 2008. speciation in action?” Animal Behaviour,vol.58,no.1,pp. [100] C. Chapperon and L. Seuront, “Cue synergy in Littorina 181–184, 1999. littorea navigation following wave dislodgement,” Journal of [82] J. Erlandsson, E. Rolan-Alvarez,´ and K. Johannesson, “Migra- the Marine Biological Association of the United Kingdom, vol. tory differences between ecotypes of the snail Littorina 89, no. 6, pp. 1133–1136, 2009. saxatilis on Galician rocky shores,” Evolutionary Ecology, vol. [101] E. Rolan-Alvarez,´ J. Erlandsson, K. Johannesson, and R. 12, no. 8, pp. 913–924, 1998. Cruz, “Mechanisms of incomplete prezygotic reproductive [83] K. Janson, “Selection and migration in two distinct pheno- isolation in an intertidal snail: testing behavioural models in types of Littorina saxatilis in Sweden,” Oecologia, vol. 59, no. wild populations,” Journal of Evolutionary Biology, vol. 12, no. 1, pp. 58–61, 1983. 5, pp. 879–890, 1999. [84] B. J. Balkau and M. W. Feldman, “Selection for migration modification,” Genetics, vol. 74, no. 1, pp. 171–174, 1973. [85] M. Slatkin, “Gene flow in natural populations,” Annual Review of Ecology and Systematics, vol. 16, pp. 393–430, 1985. [86] S. Bensch and M. Akesson,˚ “Ten years of AFLP in ecology and evolution: why so few animals?” Molecular Ecology, vol. 14, no. 10, pp. 2899–2914, 2005. [87] N. H. Barton and G. M. Hewitt, “Analysis of hybrid zones,” Annual Review of Ecology and Systematics, vol. 16, pp. 113– 148, 1985. [88] J. W. Grahame, C. S. Wilding, and R. K. Butlin, “Adaptation to a steep environmental gradient and an associated barrier to gene exchange in Littorina saxatilis,” Evolution, vol. 60, no. 2, pp. 268–278, 2006. [89] E. Rolan-Alvarez,´ K. Johannesson, and J. Erlandsson, “The maintenance of a cline in the marine snail Littorina saxatilis: the role of home site advantage and hybrid fitness,” Evolution, vol. 51, no. 6, pp. 1838–1847, 1997. [90] M. Carballo, A. Caballero, and E. Rolan-Alvarez,´ “Habitat- dependent ecotype micro-distribution at the mid-shore in natural populations of Littorina saxatilis,” Hydrobiologia, vol. 548, no. 1, pp. 307–311, 2005. [91] C. J. MacCallum, B. Nurnberger,¨ N. H. Barton, and J. M. Szymur, “Habitat preference in the Bombina hybrid zone in Croatia,” Evolution, vol. 52, no. 1, pp. 227–239, 1998. [92] D. G. Reid, “The comparative morphology, phylogeny and evolution of the gastropod family littorinidae,” Philosophical Transactions B, vol. 324, no. 1220, pp. 1–110, 1989. [93] H.D. Rundle and J. W. Boughman, “Behavioural ecology and speciation,” in Evolutionary Behavioral Ecology,D.F. Westneat and C. W. Fox, Eds., Oxford University Press, New York, NY, USA, 2010. [94] J. Hollander and R. K. Butlin, “The adaptive value of phenotypic plasticity in two ecotypes of a marine gastropod,” BMC Evolutionary Biology, vol. 10, no. 1, article 333, 2010. [95] J. Hollander, M. L. Collyer, D. C. Adams, and K. Johannesson, “Phenotypic plasticity in two marine snails: constraints su- perseding life history,” Journal of Evolutionary Biology, vol. 19, no. 6, pp. 1861–1872, 2006. [96] S. Sadedin, J. Hollander, M. Panova, K. Johannesson, and S. Gavrilets, “Case studies and mathematical models of ecological speciation. 3: ecotype formation in a Swedish snail,” Molecular Ecology, vol. 18, no. 19, pp. 4006–4023, 2009. [97] G. Chelazzi, P. Dellasantina, and M. Vannini, “Long-lasting substrate marking in the collective homing of the gastropod Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 295463, 11 pages doi:10.1155/2012/295463

Research Article The Role of Environmental Heterogeneity in Maintaining Reproductive Isolation between Hybridizing Passerina (Aves: Cardinalidae) Buntings

Matthew D. Carling1 and Henri A. Thomassen2

1 Berry Biodiversity Conservation Center, Department of Zoology and Physiology, University of Wyoming, 1000 E. University Avenue, Laramie, WY 82071, USA 2 Institute of Evolution and Ecology, University of Tubingen,¨ Auf der Morgenstelle 28, 72076 Tubingen,¨ Germany

Correspondence should be addressed to Matthew D. Carling, [email protected]

Received 25 June 2011; Revised 8 September 2011; Accepted 27 September 2011

Academic Editor: Zachariah Gompert

Copyright © 2012 M. D. Carling and H. A. Thomassen. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Hybrid zones are useful systems in which to investigate processes important in creating and maintaining biological diversity. As they are often located in ecotones, patterns of environmental heterogeneity may influence hybridization, and may also influence the maintenance of reproductive isolation between hybridizing species. Focusing on the hybrid zone between Passerina amoena (Lazuli Bunting) and Passerina cyanea (Indigo Bunting), located in the eastern Rocky Mountain/western Great Plains ecotone, we examined the relationship between population-pairwise differences in the proportion of hybrids and environmental variation. Models including environmental variables explained more of the variation in hybridization rates across the ecotone than did models that only included the geographic distance between sampling localities as predictor variables (63.9% and 58.9% versus 38.8% and 39.9%, depending on how hybridization was quantified). In the models including environmental variables, the amount of rainfall during the warmest quarter had the greatest explanatory power, consistent with a hypothesis that P. cyanea is better adapted to the mesic environments of eastern North America and P. amoena is better adapted to the xeric habitats of western North America. These results suggest that continued reproductive isolation between these species is mediated, at least partially, by differential adaptations to local environmental conditions.

1. Introduction its potential to cause habitat isolation between closely related taxa [10–12]. When traits subject to divergent Hybrid zones formed through the interbreeding of divergent natural selection are associated or correlated with traits lineages as a result of secondary contact are powerful models that may influence reproduction, ecological speciation may for investigating the forces important in generating and result [12]. Naturally occurring hybrid zones are often maintaining biodiversity [1–5]. These “natural laboratories” located in ecotones produced by sharp transitions between [1] provide an opportunity to explore patterns of gene flow different environments in which the hybridizing species and introgression between closely related taxa, which can aid were subjected to divergent natural selection [13, 14]. Such in the identification of genes contributing to reproductive ecotones are often characterized by areas of marginal habitat isolation [6, 7]. In addition, naturally occurring hybrid zones that may not be ideal for the parental species found on allow the investigation of how and to what extent ecological either side of the environmental gradient and therefore differences between the hybridizing lineages contribute to provide an opportunity to examine the impact of habitat speciation [8, 9]. on hybridization and the evolution of reproductive isolation Divergent natural selection between different habitats [8, 11]. In this manner, hybrid zones offer a chance to can be an important force in promoting speciation through explore the environmental conditions that were potentially 2 International Journal of Ecology important to the initial divergence of the hybridizing species, sex-linked loci, relative to the pattern seen for nuclear providing insight into the importance of ecological specia- autosomal loci [23], a finding consistent with Haldane’s rule tion, mediated through divergent natural selection. [24]. Spatial environmental heterogeneity can change over Additional work has shown that introgression patterns time, with the potential to greatly alter the structure of of sex-linked loci can vary substantially [7]; maximum hybrid zones, and influence speciation and biodiversity likelihood estimates of 10 locus-specific cline widths differed dynamics [8]. If hybridizing species are differentially adapted by ∼200-fold, identifying a candidate genomic region for to the environments they primarily inhabit, changes in those reproductive isolation. Suggestively, the sex-linked locus with environments and the intervening ecotone may result in one the narrowest cline width is an intron in the very-low-density of two general outcomes: (1) reduction in species diversity— lipoprotein receptor gene (VLDLR); mutations in this gene one species may swamp the other, causing extinction of are known to reduce egg laying capabilities in chickens [25] the overwhelmed species, or the species may collapse into and it is possible that divergent P. cyanea and P. amoena a hybrid swarm wherein both parental species in essence VLDLR alleles do not function well in a heterospecific become extinct, but a new “hybrid” species is formed—or genomic background, which could cause hybrid females to (2) reduction or cessation of hybridization—if hybridization lay fewer eggs. The hypothesis that a Dobzhansky-Muller only occurs in the “hybrid” habitat found in an ecotone incompatibility [26–28] causes a reduction in egg laying in and such habitats are altered or eliminated. Alternatively, hybrid females is consistent with the field research findings environmental stability may promote the stability of a hybrid of Baker and Boylan [21] described above. zone, regardless of whether the zone is maintained through Using a species distribution modeling approach, Swen- endogenous selection against hybrids balanced by continued son [29] suggested that the ecotonal environment between migration into the zone (i.e., a tension zone) or exogenous the eastern Rocky Mountains and western Great Plains is selection creating fitness differences along the environmental important in determining the geographic location of the gradient found in the ecotone. Passerina bunting hybrid zone, with temperature differences In this paper, we quantify the relationship between having the biggest impact. This may partially explain the patterns of environmental heterogeneity and patterns of observed shift in location and width of the hybrid zone hybridization across an ecotone, using the hybrid zone that has occurred over the past 40–45 years [30]. However, formed between Lazuli (Passerina amoena) and Indigo Swenson’s [29] analysis utilized only the geographic location (Passerina cyanea) buntings as a model. We explore whether of hybrid individuals; here we quantify the proportion of differences between populations in the proportion of hybrid hybrid individuals within a population to elucidate the individuals are better predicted by the variation found along influence of environmental heterogeneity on hybridization the ecological gradient experienced across the hybrid zone dynamics. than by simple geographic distance between populations. In doing so, we investigate the potential role of ecological differences between P. amoena and P. cyanea in maintaining 2. Materials and Methods reproductive isolation between these species. Passerina amoena and P. cyanea are closely related 2.1. Genetic Data and Analyses. Population samples of P. passerines [15] that form a hybrid zone in the ecotone amoena,P.cyanea,andP. amoena × P. cyanea hybrids were located between the eastern Rocky Mountains and the collected from 21 sympatric and parapatric localities across western Great Plains of North America (Figure 1;[16–18]). the contact zone during May, June, and July 2004–2007 Behavioral studies, carried out in both the lab and the field, (Figure 1, Table 1). All collected individuals were prepared demonstrate strong patterns of assortative mating and show as voucher specimens and deposited in the Louisiana State that, when choosing mates, females of both species place University Museum of Vertebrates. Additional samples from greater emphasis on male plumage patterns than on song allopatric populations (P. amoena: WA, ID; P. cyanea:MN, characteristics [19–21]. Field-based research also suggests IL, MI) were also obtained (Figure 1, Table 1). that hybrid individuals suffer fitness consequences relative From each individual, genomic DNA was extracted from to pure individuals. Mated pairs that involve at least one pectoral muscle and used to generate sequence data from hybrid individual produce fewer nestling and fledglings than a suite of 14 loci (see Supplementary Table in Supplemen- do pairs involving only nonhybrid individuals [21]. tary Material available online at doi: 10.1155/2011/295463 More recently, coalescent analyses of the evolutionary Table S1) as previously described [7, 22]. We resolved history of P. amoena and P. cyanea suggest that speciation individual haplotypes probabilistically using PHASE [33, 34] occurred through either parapatric divergence or repeated and then identified the largest independently segregating cycles of allopatric divergence with gene flow during periods block of sequence for all loci as in Carling and Brumfield of secondary contact [22]. Regardless of the spatial context [7, 22]. Individual haplotypes with frequencies ≥0.80 in the of divergence, the current hybrid zone originated through “allopatric” P. amoena population (WA) were classified as P. secondary contact and is probably no older than ∼6500 amoena haplotypes, and haplotypes with frequencies ≥0.80 years [7, 22, 23]. Patterns of genetic introgression across the in the “allopatric” P. cyanea populations (MN, IL, MI) were hybrid zone, based on allelic data from two mitochondrial designated as P. cyanea haplotypes [7, 22]. We then generated genes, four nuclear autosomal loci, and two sex-linked loci, multilocus genotypes for each individual by coding each indicate reduced introgression of mitochondrial DNA and allele at each locus as either P. amoena or P. cyanea. International Journal of Ecology 3

1 2 24 26 25

Passerina amoena Passerina cyanea

(a)

−108.5◦ −104.5◦ −100.5◦

46.5◦ 9 21 7 3 10 4 11 5 17

13 20 15 23 ◦ 18 19 22 42.5 12 14 16 8 Elevation 3001–4000 m 2001–3000 m 1001–2000 m 6 501–1000 m 40.5◦ Contact zone 150 km 0–500 m

(b)

Figure 1: (a) Primary breeding distributions and allopatric sampling localities of Passerina amoena and Passerina cyanea. Area outlined by the black box is enlarged in (b). (b) Sympatric and parapatric sampling localities analyzed in this study. Background shading represents elevation and white box indicates approximate location of the sympatric contact zone. See Table 1 for more detailed information on sampling localities. Digital maps for each species [31] were downloaded from NatureServe [32] and modiﬁed.

Using these multilocus genotypes we quantified the and 0.10. In both scenarios, individuals whose proportion of genetic ancestry of each individual using STRUCTURE genetic ancestry assigned to cluster 1 was greater than 0.80 [35–38]. In the analyses, the allopatric populations (P. (criterion 1) or 0.90 (criterion 2) were classified as pure P. amoena: WA, ID; P. cyanea:MN,IL,MI)werecodedas amoena individuals and all other individuals were classified “learning populations” (USEPOPINFO model). For all other as pure P. cyanea individuals. individuals, we estimated the proportion of their ancestry Within each sympatric and parapatric population, we attributed to the two clusters defined by the “learning calculated the proportion of hybrid individuals, which were populations.” We used the default allele frequency model then used to generate a distance matrix of the relative similar- and ran the analysis three times, each time with a burnin ity in hybridization among populations. The “hybridization length of 100,000 steps followed by a postburnin run of distance” between two populations was calculated as 1,000,000 steps. We then calculated the mean ancestry values |(PL +0.5PH ) − (PL +0.5PH )|,(1) across the three runs for each individual from the sympatric 1 1 2 2 and parapatric populations. Individuals were classified as where PL1 is the proportion of P. amoena individuals in a “hybrid” according to two different criteria: (1) “80– population 1, PH1 is the proportion of hybrids in population 20” wherein an individual was classified as a hybrid if the 1, PL2 is the proportion of P. amoena individuals in proportion of its genetic ancestry assigned to cluster 1 (P. population 2, and PH2 is the proportion of hybrids in amoena) was between 0.80 and 0.20 and (2) “90–10” wherein population 2. This hybridization distance was calculated for an individual was classified as a hybrid if the proportion of every pair of populations, except the allopatric ones (WA, its genetic ancestry assigned to cluster 1 was between 0.90 ID, MN, IL, MI) and the resultant matrix (Table 2)formed 4 International Journal of Ecology

Table 1: Sampling localities, sample sizes, and the proportion of hybrid individuals for populations analyzed in this study.

Proportion of Proportion of Code Locality Latitude Longitude N hybrids (80–20) hybrids (90–10) 1 WA 47.8 −118.9 29 — — 2 ID 44.1 −116.2 3 — — 3 MT: Custer National Forest no. 1 45.1 −108.5 4 0.00 0.00 4 WY: Bighorn National Forest no. 1 44.8 −107.3 8 0.00 0.00 5 WY: Bighorn National Forest no. 2 44.6 −107.7 2 0.00 0.00 6 CO: Roosevelt National Forest 40.7 −105.2 7 0.14 0.14 7 MT: Custer National Forest no. 2 45.7 −106 5 0.00 0.00 8 WY: Medicine Bow National Forest 42.4 −105.3 3 0.00 0.00 9 ND: Little Missouri National Grassland 46.8 −103.5 13 0.00 0.00 10 SD: Custer National Forest 45.3 −103.2 10 0.00 0.00 11 WY: Sand Creek∗ 44.5 −104.1 14 0.36 0.50 12 NE: White River∗ 42.6 −103.5 8 0.38 0.50 13 SD: Black Hills National Forest 43.7 −103.8 5 0.20 0.20 NE: Ponderosa State Wildlife 14 42.6 −103.3 7 0.57 1.00 Management Area SD: The Nature Conservancy Whitney 15 43.3 −103.6 3 0.00 0.00 Preserve 16 NE: Nebraska National Forest 42.8 −102.9 8 0.25 1.00 17 SD: Ft. Meade National Recreation Area 44.4 −103.5 4 0.50 0.50 18 NE: Nenzel∗ 42.8 −101.1 4 0.25 0.25 19 NE: The Nature Conservancy 42.8 −100 18 0.00 0.00 Niobrara Valley Preserve 20 SD: Carpenter Game Production Area 43.7 −99.5 13 0.08 0.08 21 ND: The Nature Conservancy 46.3 −97.3 4 0.00 0.00 Pigeon Point Preserve 22 NE: Wiseman State Wildlife Area 42.8 −97.1 10 0.00 0.00 23 SD: Newton Hills Game Production Area 43.2 −96.6 13 0.00 0.08 24 MN 44.9 −93.7 10 — — 25 IL 40.5 −88.9 10 — — 26 MI 42.3 −83.7 7 — — ∗ Access to these localities generously provided by private landowners. the basis of our generalized dissimilarity modeling (see difference in daily maximum and minimum temperatures, below). As an alternative measure of “hybridization distance” averaged across the year; mean temperature of the warmest between two populations, we also calculated the absolute quarter (Bio10); mean temperature of the coldest quarter difference between mean population Q-scores, as estimated (Bio11); total annual precipitation (Bio12); precipitation by STRUCTURE [35–38]. seasonality (Bio15), which is the coefficient of variation in monthly rainfall across the year; total precipitation of the 2.2. Spatial Data and Analyses wettest quarter (Bio17); total precipitation of the warmest quarter (Bio18). 2.2.1. Environmental Data. We used a set of moderately In addition to these ground-based measurements of high-resolution climate and satellite remote sensing variables climate, we used satellite remote sensing data from both to characterize the habitat within and on both sides of the passive optical sensors (MODIS; https://lpdaac.usgs.gov/ hybrid zone (Table 3). Climate data were obtained from lpdaac/products/modis overview) and active radar scatter- the WorldClim database [39], which are spatially explicit ometers (QuickScat; http://www.scp.byu.edu/data/Quikscat/ estimates of annual means, seasonal extremes, and degrees SIRv2/qush/World regions.htm) to infer a variety of ecolog- of seasonality in temperature and precipitation based on a ical characteristics of the land surface. From the MODIS 50-year climatology (1950–2000). The climate variables that archive, we used the average monthly Normalized Difference were included in our analyses are annual mean temperature Vegetation Index (NDVI) to infer vegetation density [40]. (Bio1); mean diurnal temperature range (Bio2), which is the NDVI is sometimes also referred to as a measure of

International Journal of Ecology 5

D etnHlsGPA Hills Newton SD:

E iea SWA Wiseman NE:

D N ienPitPreserve Point Pigeon TNC ND:

D apne GPA Carpenter SD:

E N ibaaVle Preserve Valley Niobrara TNC NE:

E Nenzel NE:

D t ed NRA Meade Ft. SD:

E ersaNF Nebraska NE:

D N hte Preserve Whitney TNC SD:

E odrs SWMA Ponderosa NE:

D lc il NF Hills Black SD:

E ht River White NE:

Y adCreek Sand WY: D utrNF Custer SD: ; see text for details) between all pairs of sympatric and allopatric populations. Above the diagonal, hybrids |

)

2 D iteMsor NG Missouri Little ND:

5PH

. +0 NF Bow Medicine WY: 2

(PL − no.2 NF Custer MT:

)

1 5PH NF Roosevelt CO: . +0

1 Y ihr Fno.2 NF Bighorn WY:

(PL

Y ihr Fno.1 NF Bighorn WY: T utrN no.1 NF Custer MT: 0.00 0.00 0.00 0.07 0.001.00 0.00 1.00 0.00 1.00 0.001.00 0.93 0.32 1.00 1.00 0.25 1.00 1.00 0.10 0.93 1.00 0.50 1.00 1.00 — 1.00 0.68 1.00 0.13 0.75 1.00 0.50 0.90 0.68 0.88 0.50 0.75 1.00 1.00 0.90 0.96 0.50 0.50 1.00 0.50 1.00 1.00 0.13 0.50 1.00 — 0.50 0.04 0.13 0.00 0.00 0.00 0.04 0.00 — 0.00 0.00 2: Matrix of hybridization distances ( NE: Nebraska NFSD: Ft. Meade NRANE: NenzelNE: TNC 0.50 Niobrara 0.50Valley 0.50 PreserveSD: 0.50 Carpenter GPA 0.50ND: 0.50 TNC Pigeon Point 0.43Preserve 0.43 0.88 0.50 0.96NE: Wiseman 0.50 0.88 SWA 0.50 0.96SD: Newton Hills 0.50 0.88 GPA 0.50 0.96 0.50 0.80 1.00 0.96 0.50 0.89 0.50 0.88 1.00 0.96 0.18 0.96 0.18 0.88 1.00 0.96 0.25 0.96 0.25 0.88 0.93 0.89 0.40 0.96 0.40 0.88 1.00 0.96 0.00 0.96 0.00 0.55 1.00 0.96 0.50 0.64 0.50 0.63 1.00 0.96 0.71 — 0.00 0.78 1.00 0.96 0.86 0.38 0.38 — 0.68 0.64 0.46 0.75 0.88 0.75 0.71 0.38 0.96 0.88 0.38 0.90 0.86 0.50 0.46 0.84 0.38 0.50 0.46 0.46 0.46 0.88 — 1.00 0.96 0.50 0.09 0.88 0.50 0.46 0.50 0.13 0.04 0.88 0.50 0.46 0.50 0.09 — 0.13 0.09 0.13 0.04 0.00 0.04 0.13 0.04 0.04 0.00 0.13 0.04 0.00 0.04 0.04 — — 0.00 MT: Custer NF no. 1WY: Bighorn NF no. 1WY: Bighorn NF no. 2 —CO: 0.00 Roosevelt NFMT: Custer 0.00 0.00 NF — no. 2WY: Medicine 0.00 0.00 Bow NF 0.00ND: 0.07 Little 0.07 Missouri — NG 0.00 0.00 0.07SD: Custer 0.07 NF 0.00 0.00 0.00 0.00 0.00 0.00WY: Sand 0.07 Creek 0.00 0.00 0.00 0.00 0.00 0.00NE: White River 0.00 — 0.00 0.07 0.00 0.07 0.00SD: Black Hills 0.00 NF 0.07 0.07 0.00 0.00NE: 0.00 — Ponderosa 0.00 SWMA 0.32 0.32 0.00 0.07SD: 0.00 0.32 TNC 0.00 — Whitney 0.00 0.19 0.25 0.32 0.50Preserve 0.00 0.07 0.32 0.19 0.10 0.00 0.00 0.00 0.10 0.25 0.32 0.50 0.07 0.19 0.10 — 0.10 0.07 0.00 0.00 0.29 0.25 0.25 0.50 0.25 0.10 0.29 0.10 0.00 0.00 0.32 0.32 0.00 0.18 0.32 0.43 0.12 0.29 0.00 0.03 0.00 0.32 0.19 0.19 0.13 0.25 0.32 0.50 0.03 0.00 0.13 0.10 0.00 0.19 0.10 0.10 0.50 0.25 0.32 0.50 0.21 0.13 0.50 0.10 0.10 0.29 — 0.29 0.88 0.25 0.32 0.50 0.07 0.50 0.88 0.10 0.29 0.00 0.00 1.00 0.25 0.50 0.32 — 0.05 0.88 1.00 0.10 0.00 0.13 0.13 0.96 0.07 0.18 0.19 0.43 1.00 0.96 0.22 0.13 0.13 0.50 0.50 1.00 0.25 0.10 — 0.80 0.96 1.00 0.15 0.22 0.50 0.88 0.88 1.00 0.40 0.29 0.93 1.00 1.00 0.09 0.04 0.88 — 1.00 1.00 1.00 0.00 — 0.89 1.00 1.00 0.10 0.32 1.00 0.96 0.96 0.19 0.13 0.93 1.00 0.19 0.20 0.29 0.96 1.00 1.00 0.10 0.50 0.93 0.06 0.18 0.16 1.00 1.00 1.00 0.03 0.88 0.93 0.31 0.55 0.21 1.00 1.00 1.00 0.40 1.00 0.69 0.68 0.59 1.00 0.78 0.96 0.81 0.64 0.71 0.90 1.00 0.77 0.68 0.68 0.86 1.00 0.81 0.68 0.71 0.90 1.00 0.81 0.68 0.71 0.90 0.81 0.71 0.90 Table were identiﬁed using the “80–20” criterion; below the diagonal, hybrids were identiﬁed using the “90–10” criterion. 6 International Journal of Ecology

Table 3: List of environmental variables used in our analyses.

Data Record Instrument Ecological attributes Variables derived

Normalized Diﬀerence Vegetation Index (NDVI) Satellite-MODIS Vegetation density NDVI mean

Vegetation Continuous Field (VCF)∗§ Satellite-MODIS Forest cover + heterogeneity Tree cover Surface moisture + roughness Scatterometer-Backscatter∗† Satellite-QSCAT QSCAT aug (forest structure), seasonality

DEM∗ SpaceShuttle SRTM Topography + ruggedness SRTM Bio1, Bio2, Bio10, Bio11, WorldClim¶ Station-Network Bioclimatic variables Bio12, Bio15, Bio17, Bio18 ∗ Data at native resolutions smaller or larger than 1 km have been aggregated to 1 km. †QSCAT is based on data from 2001. §Based on MODIS data from 2001 [41]. ¶WorldClim data are based on monthly climatologies from 1950–2000 [39]. The bioclimatic variables are annual mean temperature (Bio1), mean diurnal temperature range (Bio2), mean temperature of the warmest quarter (Bio10), mean temperature of the coldest quarter (Bio11), annual precipitation (Bio12), precipitation seasonality (Bio15), precipitation of the wettest quarter (Bio17), and precipitation of the warmest quarter (Bio18).

“greenness” and is high over forested areas and low over heterogeneity and geographic distance. One advantage of areas with sparse vegetation. In addition, we used the GDM over other modeling methodologies is that it can vegetation continuous field [41] product as a measure of the fit nonlinear relationships between predictor and response percentage of tree cover. In contrast to NDVI, percent tree variables through the use of I-spline basis functions [42]. cover specifically measures the land surface area covered by Another advantage of GDM is that, because it uses pair- trees, disregarding other types of vegetation. From QuickScat wise comparisons between sites, it can explicitly take into (QSCAT), we obtained raw backscatter measurements for the account the influence of geographic distance on explaining month of August that capture attributes related to surface biological variation—a particularly important feature in our moisture [40], and, from the Shuttle Radar Topography study. The relative importance of predictor variables in a Mission (SRTM), we acquired elevation data. GDM can be assessed by means of response curves. To Time series of the remote sensing data sources were further evaluate the extent to which geographic distance acquired from 2001 for QSCAT and tree cover and means is potentially correlated with environmental differences, we over the period 2000–2004 for NDVI. Variables with native ran independent tests with the following sets of predictor resolutions higher (e.g., SRTM: 30 m) or lower (e.g., QSCAT: variables: (1) environmental variables and geographic dis- 2.25 km) than 1 km were reaggregated to a 1 km grid tance; (2) only geographic distance; (3) only environmental cell resolution. This resolution is often used in ecological variables. niche modeling at the regional or subcontinental scales and GDM is a two-step method: first, dissimilarities of a set balances resolutions from climate data, which often have of predictor variables are fitted to the genetic dissimilarities coarser native resolutions, and remote sensing data, with (the response variables). The contributions of predictor higher native resolutions. To improve interpretation, we variables to explaining the observed response variation are checked for covariance among variables and only included tested by permutations, and only those variables that are those with substantial unique variance. Various criteria were significant are retained in the final model. These procedures used to decide which layers of correlated pairs (with Pearson’s result in a function that describes the relationship between correlations on the order of 0.9 or larger) were retained environmental and response variables. Second, using the for further analysis. These included keeping layers that are function resulting from the first step, a spatial prediction more commonly used in distribution modeling (WorldClim) is made of the response variable patterns. For visualization or that exhibit larger contrast/variance over the study area purposes, classes of similar response are color coded, where (QSCAT) as well as having best data quality (NDVI). Pear- larger color differences between two localities represent son’s cross correlations of the used environmental variables larger differences in the proportion of hybrid individuals. are shown in supplementary Table S2. In order to assess the significance of the level of variation that was explained by our models, we ran additional models 2.2.2. Generalized Dissimilarity Modeling. To analyze the in which the environmental layers were substituted by relative contribution of geographic distance and envi- layers with random values for each grid cell. The resulting ronmental variables to explaining the similarity among percentage of variation explained was compared to that of the sampling locations in the percentage of hybrids, we used full model. We considered the performance of the full model generalized dissimilarity modeling (GDM, [42]). GDM is not significant if it explained an equal amount or less of a matrix regression technique that predicts biotic dissimi- the total variation than a model with random environmental larity (turnover) between sites based upon environmental variables. International Journal of Ecology 7

Figure 2: Generalized dissimilarity modeling results of the proportion of pure P. amoena individuals, plus one half of the hybrid individuals, in a population. Left two panels: “80–20” criterion; right two panels: “90–10” criterion. Top two panels: full model (including both geographic distance and environmental variables); bottom two panels: models for only environmental variables. Colors indicate the proportion of pure P. amoena individuals, plus one half of the hybrid individuals, in the population (brown/tan 100%, blue = 0%).

3. Results The most important predictor variable in the full models was geographic distance, whereas these were Bio18 (precip- 3.1. Genetic Analyses. Using multilocus genotypes, the pro- itation of the warmest quarter) and elevation for models portion of hybrid individuals in the 21 sympatric and para- using only environmental variables (Figure 3). These results patric populations ranged from 0.00 to 0.57 using the “80– suggest that geographic distance and environmental hetero- 20” criterion and 0.00 to 1.00 using the “90–10” criterion geneity along our sampling sites are partially correlated, but (Table 1). Across all sympatric and parapatric populations, that geographic distance alone does not perform as well 20 individuals (“80–20” criterion) and 33 individuals (“90– in explaining differences in the level of hybridization as 10” criterion) of the 163 sampled (12.2% and 20.2%, environmental variables or the combination of distance and resp.) were classified as hybrids. Focusing on the 20 hybrid environment. individuals classified using the “80–20” criterion, hybrids were found as far west as the Roosevelt National Forest sampling locality in Colorado and as far east as the Carpenter 4. Discussion Game Production Area along the Missouri River in South Dakota (Figure 1, Table 1). Using the more liberal hybrid Understanding the selective forces maintaining hybrid zones classification criterion (“90–10”), the presence of hybrids is an important component of understanding the evolu- extended east to the Newton Hills Game Production Area in tionary process generating and maintaining biodiversity. South Dakota; the westernmost locality where hybrids were If natural selection in response to different environments present was unchanged (Figure 1, Table 1). during periods of allopatry contributed to divergence and speciation between closely related taxa, we might expect 3.2. Spatial Analyses. Generalized dissimilarity models patterns of hybridization upon secondary contact to reflect (Figure 2) performed well as measured by the total variance the patterns of local environmental heterogeneity. In this explained and performed significantly better than models study, we have shown that patterns of hybridization between with random environmental variables. The full models Passerina amoena and Passerina cyanea are best explained explained 58.9% (“90–20” criterion) and 63.9% (“80–20” by a model that includes both patterns of environmental criterion) of the total observed variation in the similarity variation across the Rocky Mountains/Great Plains ecotone of the level of hybridization (Table 4). Models using only and geographic distance between sampling localities (Figures environmental variables explained about 6% less of the 2 and 3, Table 4). Although the distance between sampling observed variation, and models using only geographic dis- localities was clearly an important determinant of the dif- tance explained 38.8% (“90–10” criterion) and 39.9% (“80– ferences in hybridization among populations, environmental 20” criterion) of the variation. The GDM results based on the variation alone was a more powerful predictor. This result absolute difference in mean Q-scores were very similar (not held for both criteria we used to assess whether an individual shown) and are not discussed further. was a hybrid (“80–20” and “90–10”). 8 International Journal of Ecology

80–20 full model

12 10 8 6 4 Predicted response Predicted 2 0 160 180 200 220 0 24681012 120 160 200 240 350 450 550 500 1000 1500 2000 Geographic distance Bio18 Bio12 Elevation Bio10

80–20 only environment 1

0.8

0.6

0.4

0.2 Predicted response Predicted 0 500 1000 1500 2000 160 180 200 220 120 160 200 240 350 450 550 Elevation Bio10 Bio18 Bio12

90–10 full model

2 Predicted response Predicted

0 024681012 120 160 200 240 350 450 550 500 1000 1500 2000 160 180 200 220 Geographic distance Bio18 Bio12 Elevation Bio10

90–10 only environment

0.8

0.6

0.4

Predicted response Predicted 0.2

0 120 160 200 240 160 180 200 220 500 1000 1500 2000 350 450 550 Bio18 Bio10 Elevation Bio12 Figure 3: Response curves for generalized dissimilarity models of the proportion of pure P. amoena individuals, plus one half of the hybrid individuals, in a population. Results are shown for “80–20” and “90–10” criteria, and for full models (with both geographic distance and environmental variables) and models using only environmental variables. Response curves in GDM are nonlinear, but constrained to be monotonic. The slope of each function is indicative of the rate of change in the proportion of hybrids along the environmental gradient concerned. The maximum height of a response curve indicates the relative importance of that variable in explaining the observed variation in the similarity between populations in the proportion of hybrids. The y-axes indicate ﬁtted functions f (x), where x denotes the environmental variable concerned. Each of the functions is ﬁtted as a linear combination of I-spline basis functions [41]. International Journal of Ecology 9

Table 4: Generalized dissimilarity modeling results. Shown are the performance, and the third is that mate choice patterns could percentages of the total variation observed and important predictor differ across the hybrid zone. variables for “80–20” and “90–10” criteria for full models (using That the fitness of hybrid individuals may be related to both geographic distance and environmental variables), models environmental conditions is not new as it is central to the with environmental variables only, models with geographic distance environmental gradient selection hypothesis of hybrid zone only, and random models. structure [43] and its close relative, the bounded-superiority % variance hypothesis [14]. In both these hypotheses, the selection Model Variables selected∗ explained pressures faced by hybrid individuals depend on ecological dist, Bio12, Bio10, context. It is important to note that hybrid individuals may 80–20 full 63.9 elevation, Bio18 or may not actively prefer particular habitats. It is possible Bio16, elevation, Bio12, that individuals might disperse some given distance and that 80–20 environment only 57.4 Bio10 any subsequent fitness consequences in a particular habitat 80–20 distance only 39.9 dist may not be driven by choice. Our results are consistent with the possibility that the 80–20 random env 0.6 environmental variation present across the Passerina hybrid dist, Bio18, Bio10, 90–10 full 58.9 zone somehow influences hybrid fitness. It has been argued elevation, Bio12 that both the oriole and flicker hybrid zones (Icterus bullockii elevation, Bio18, Bio10, 90–10 environment only 53.4 × Icterus galbula and Colaptes auratus auratus × Colaptes Bio12 auratus cafer, resp.), which are also located in the eastern 90–10 distance only 38.8 Dist Rocky Mountains/western Great Plains ecotone, are main- 90–10 random env 0.4 tained because hybrid individuals are restricted to particular ∗ Variables are shown in order of decreasing importance (Figure 3); dist: habitats. In part due to differences in metabolic performance geographic distance; for an explanation of the bioclimatic variables, see between the two species, Rising [44] suggested that I. galbula Table 3. is better adapted to the more mesic environment characteristic of eastern deciduous forest and I. bullockii is better adapted to the more xeric environments that are more dom- This pattern supports the findings of Swenson [29], who inant in western North America. He further suggested that used a much different approach to explore the relationship hybrids could only survive in the intervening ecotone [44]. between environmental heterogeneity and hybridization in Similarly, hybridization between the red-shafted and yellow- Passerina buntings. Whereas our analyses focused on the shafted subspecies of the Northern Flicker (C. auratus) patterns of hybridization within and among populations, appears to be strongly related to moisture patterns present the unit of hybridization in Swenson’s analysis was the geo- across the same ecotone [14, 45]. Since mean temperature graphic location of an individual determined to be a hybrid. and precipitation of the warmest quarter (Bio10, Bio18) were The proportion of hybrid individuals in a population, which important contributors in the GDM results (Table 4), it is was central to our analysis, did not factor into Swenson’s possible that P. cyanea and P. amoena are differentially work. Instead, Swenson solely focused on whether or not a adapted to their respective habitats (more mesic for P. cyanea hybrid had been recorded in a particular location and did not xeric for P. amoena) similar to what Rising hypothesized explore the similarities in the proportion of hybrids among for Icterus [44]. Either through metabolic differences or populations as we did. Although the relevant environmental some other mechanism, hybrid Passerina buntings may suffer variables identified in our study (BioClim variables Bio10, greater fitness consequences on one or both sides of the Bio12, and Bio18, Table 4) were slightly different from the hybrid zone. Irrespective of the exact mechanism, environ- environmental variable with the greatest explanatory power mentally influenced differences in hybrid fitness suggest that in Swenson’s study (mean annual temperature, BioClim natural-selection-driven adaptations to different habitats variable Bio1), the variables are correlated (Supplemental during the course of divergence between P. amoena and P. Table S2). For example, the correlation between Bio18 (pre- cyanea contributed to the speciation process, a hallmark of cipitation during the warmest quarter), which had the great- ecological speciation. est explanatory power in both of our “environment only” The third possible driver of the differences in hybridiza- models (Table 4), and mean annual temperature (Bio1) tion relates to mate choice, which may depend on ecological was 0.506 (Supplementary Table S1). Despite the difference context. If sensory bias plays a role in female mate choice in methodologies between the studies, both found that [46] and the local environment influences a female ability to environmental variation predicts patterns of hybridization, distinguish between con and heterospecific mates, the rate which strongly indicates that local environments influence of hybridization may differ among populations. Although the structure of the Passerina bunting hybrid zone. natal dispersal patterns are unknown for Passerina buntings At present, our data do not allow us to determine [47, 48], if they do not differ among populations, then what is driving the relationship between hybridization and the relationship between environmental heterogeneity and environmental variation, but we consider three general, hybridization observed could potentially result from differ- not mutually exclusive possibilities. The first is that hybrid ences in female mate choice. Previous work indicated that individuals may either prefer or avoid particular habitats, the female buntings can readily, but not perfectly, distinguish second is that there may be differences in habitat-specific between con and heterospecific males [19, 20],butitisnot 10 International Journal of Ecology known whether the local environment can influence female Prepathon Funds, Louisiana State University Department of mate choice abilities. Biological Sciences, and Sigma-Xi to M. D. Carling. Recent work has shown a significant narrowing and westward shift in the structure of the Passerina hybrid zone over the past 40–45 years [30]. Although climate data from References WorldClim [38]arenotofsufficient resolution to test this hypothesis, it is intriguing to speculate that changes in [1] G. M. Hewitt, “Hybrid zones-natural laboratories for evolu- precipitation over the same time frame played an important tionary studies,” Trends in Ecology and Evolution, vol. 3, no. 7, pp. 158–167, 1988. role in the hybrid zone movement. Additionally, climate [2] R. G. Harrison, “Hybrid zones: windows on the evolutionary models suggest that annual precipitation will increase over process,” in Oxford Surveys in Evolutionary Biology,D.J. much of North America, with the exception of the southwest, Futuyma and J. Antonovics, Eds., pp. 68–129, Oxford Univer- over the next 100 years [49]. Interestingly, the majority of sity Press, Oxford, UK, 1990. the increase in annual precipitation is predicted to be driven [3] R. G. Harrison, “Hybrids and hybrid zones: historical per- by precipitation in the winter months (December, January, spective,” in Hybrid Zones and the Evolutionary Process,R.G. and February), whereas precipitation in the summer months Harrison, Ed., pp. 3–12, Oxford University Press, New York, (June, July, and August) is predicted to decrease [49]. If the NY, USA, 1993. structure of the Passerina hybrid zone is mediated through [4] M. L. Arnold, Natural Hybridization and Evolution,Oxford precipitation patterns during the warmest part of the year, as University Press, New York, NY, USA, 1997. our data suggests, the zone may actually reverse course and [5] J. Mallet, “Hybridization as an invasion of the genome,” Trends shift eastward in the future as the ecotone becomes drier. in Ecology and Evolution, vol. 20, no. 5, pp. 229–237, 2005. ff Regardless of the exact mechanism, it is clear that [6] B. A. Payseur, J. G. Krenz, and M. W. Nachman, “Di erential in order to fully understand hybridization and continued patterns of introgression across the X chromosome in a hybrid reproductive isolation between P. amoena and P. cyanea the zone between two species of house mice,” Evolution, vol. 58, no. 9, pp. 2064–2078, 2004. ecological context within which the interactions between the [7] M. D. Carling and R. T. Brumfield, “Speciation in Passerina species occur should not be ignored. Given that these species buntings: introgression patterns of sex-linked loci identify a did not diverge sympatrically [22], they are likely adapted candidate gene region for reproductive isolation,” Molecular to different environmental conditions and these differences Ecology, vol. 18, no. 5, pp. 834–847, 2009. influence hybridization patterns. Consequently, the long- [8] O. Seehausen, G. Takimoto, D. Roy, and J. Jokela, “Speciation term stability of the hybrid zone, as well as the species reversal and biodiversity dynamics with hybridization in themselves, depends at least in part on the stability of the changing environments,” Molecular Ecology, vol. 17, no. 1, pp. patterns of environmental heterogeneity found in the eastern 30–44, 2008. Rocky Mountains/western Great Plains ecotone. [9]M.E.MaanandO.Seehausen,“Ecology,sexualselectionand speciation,” Ecology Letters, vol. 14, no. 6, pp. 591–602, 2011. [10] D. Schluter, “Ecology and the origin of species,” Trends in Ecology and Evolution, vol. 16, no. 7, pp. 372–380, 2001. Acknowledgments [11] J. A. Coyne and H. A. Orr, Speciation, Sinauer Associates, The authors thank S. Birks (University of Washington Burke Sunderland, Mass, USA, 2004. [12] P. Nosil, L. J. Harmon, and O. Seehausen, “Ecological expla- Museum of Natural History and Culture), A. Capparella nations for (incomplete) speciation,” Trends in Ecology and (Illinois State University), J. Hinshaw (University of Michi- Evolution, vol. 24, no. 3, pp. 145–156, 2009. gan Museum of Zoology), and M. Westberg (University of [13] J. A. Ender, Geographic Variation, Speciation and Clines, Minnesota Bell Museum) for generously providing tissue Princeton University Press, Princeton, NJ, USA, 1977. samples for this project. Scientific collecting permits were [14] W. S. Moore, “Evaluation of narrow hybrid zones in verte- granted by the Colorado Division of Wildlife, Montana brates,” Quarterly Review of Biology, vol. 52, pp. 263–277, 1977. Fish, Wildlife and Parks, Nebraska Game and Parks, North [15] M. D. Carling and R. T. Brumfield, “Integrating phylogenetic Dakota Game and Fish, South Dakota Game, Fish and and population genetic analyses of multiple loci to test species Parks, Wyoming Game and Fish, and the United States Fish divergence hypotheses in Passerina buntings,” Genetics, vol. and Wildlife Service. Louisiana State University Institutional 178, no. 1, pp. 363–377, 2008. Animal Care and Use Committee Protocol: 08-025. The [16] C. G. Sibley and L. L. J. Short, “Hybridization in the buntings map data were provided by NatureServe in collaboration (Passerina) of the Great Plains,” Auk, vol. 76, pp. 443–463, with Robert Ridgely, James Zook, The Nature Conservancy 1959. Migratory Bird Program, Conservation International-CABS, [17] S. T. Emlen, J. D. Rising, and W. L. Thompson, “Behavioral World Wildlife Fund, USA, and Environment Canada- and morphological study of sympatry in Indigo and Lazuli buntings of the Great Plains,” Wilson Bulletin, vol. 87, pp. 145– WILDSPACE. This work was supported by grants from 179, 1975. NSF (DEB-0543562 and DBI-0400797 to R. T. Brumfield [18] R. L. Kroodsma, “Hybridization in buntings (Passerina)in at Louisiana State University, DEB-0515981 and DEB- North Dakota and eastern Montana,” Auk, vol. 92, pp. 66–80, 0814277 to I. J. Lovette at Cornell University, and DEB- 1975. 0808464 to M. D. Carling), as well as by grants from the [19] M. C. Baker and A. E. M. Baker, “Reproductive behavior of AMNH Chapman Fund, AOU, Explorer’s Club, Louisiana female buntings: isolating mechanisms in a hybridizing pair State University Museum of Natural Science Birdathon and of species,” Evolution, vol. 44, no. 2, pp. 332–338, 1990. International Journal of Ecology 11

[20] M. C. Baker, “Female buntings from hybridizing populations sample group information,” Molecular Ecology Resources, vol. prefer conspecific males,” Wilson Bulletin, vol. 108, no. 4, pp. 9, no. 5, pp. 1322–1332, 2009. 771–775, 1996. [39] R. J. Hijmans, S. E. Cameron, J. L. Parra, P. G. Jones, and A. [21] M. C. Baker and J. T. Boylan, “Singing behavior, mating Jarvis, “Very high resolution interpolated climate surfaces for associations and reproductive success in a population of global land areas,” International Journal of Climatology, vol. 25, hybridizing Lazuli and Indigo Buntings,” Condor, vol. 101, no. no. 15, pp. 1965–1978, 2005. 3, pp. 493–504, 1999. [40] R. B. Myneni, S. Hoffman, Y. Knyazikhin et al., “Global prod- [22] M. D. Carling, I. J. Lovette, and R. T. Brumfield, “Historical ucts of vegetation leaf area and fraction absorbed PAR from divergence and gene flow: coalescent analyses of mitochon- year one of MODIS data,” Remote Sensing of Environment, vol. drial, autosomal and sex-linked loci in Passerina buntings,” 83, no. 1-2, pp. 214–231, 2002. Evolution, vol. 64, no. 6, pp. 1762–1772, 2010. [41] M. C. Hansen, R. S. DeFries, J. R. G. Townshend, R. Sohlberg, [23] M. D. Carling and R. T. Brumfield, “Haldane’s rule in an C. Dimiceli, and M. Carroll, “Towards an operational MODIS avian system: using cline theory and divergence population continuous field of percent tree cover algorithm: examples genetics to test for differential introgression of mitochondrial, using AVHRR and MODIS data,” Remote Sensing of Environ- autosomal, and sex-linked loci across the Passerina bunting ment, vol. 83, no. 1-2, pp. 303–319, 2002. hybrid zone,” Evolution, vol. 62, no. 10, pp. 2600–2615, 2008. [42] S. Ferrier, G. Manion, J. Elith, and K. Richardson, “Using [24] J. B. S. Haldane, “Sex ratio and unisexual sterility in hybrid generalized dissimilarity modelling to analyse and predict animals,” Journal of Genetics, vol. 12, no. 2, pp. 101–109, 1922. patterns of beta diversity in regional biodiversity assessment,” [25] O. M. Ocon-Grove, S. Maddineni, G. L. Hendricks et Diversity and Distributions, vol. 13, no. 3, pp. 252–264, 2007. al., “Pituitary progesterone receptor expression and plasma [43] R. M. May, J. A. Endler, and R. E. McMurtrie, “Gene frequency gonado- tropic concentrations in the reproductively dysfunc- clines in the presence of selection opposed by gene flow,” tional mutant restricted ovulator chicken,” Domestic Animal American Naturalist, vol. 109, pp. 659–676, 1975. Endocrinology, vol. 32, pp. 201–210, 2007. [44] J. D. Rising, “A comparison of metabolism and evaporative [26] T. Dobzhansky, Genetics and the Origin of Species, Columbia water loss of baltimore and bullock orioles,” Comparative University Press, New York, NY, USA, 1937. Biochemistry And Physiology, vol. 31, no. 6, pp. 915–925, 1969. [27] H. J. Muller, “Bearing of the Drosophila work on systematics,” [45] W. S. Moore and J. T. Price, “Nature of selection in the North- in The New Systematics, J. B. S. Haldane, Ed., pp. 185–268, ern Flicker hybrid zone and its implications for speciation Clarendon Press, Oxford, UK, 1940. theory,” in Hybrid Zones and the Evolutionary Process,R.G. [28] H. J. Muller, “Isolating mechanisms, evolution, and tempera- Harrison, Ed., pp. 196–225, Oxford University Press, New ture,” Biological Symposia, vol. 6, pp. 71–125, 1942. York, NY, USA, 1993. [29] N. G. Swenson, “Gis-based niche models reveal unifying [46] M. J. Ryan and A. Keddy-Hector, “Directional patterns of climatic mechanisms that maintain the location of avian female mate choice and the role of sensory biases,” American hybrid zones in a North American suture zone,” Journal of Naturalist, vol. 139, pp. S4–S35, 1992. Evolutionary Biology, vol. 19, no. 3, pp. 717–725, 2006. [47] E. Greene, V. R. Muehter, and W. Davison, “Lazuli Bunting [30] M. D. Carling and B. Zuckerberg, “Spatio-temporal changes (Passerina amoena),” in The Birds of North America, A. Poole, in the genetic structure of the Passerina bunting hybrid zone,” Ed., Cornell Lab of Ornithology, Ithaca, NY, USA, 1996. Molecular Ecology, vol. 20, no. 6, pp. 1166–1175, 2011. [48] R. B. Payne, “Lazuli Bunting (Passerina amoena),” in The Birds [31] R. S. Ridgely, T. F. Allnut, T. Brooks et al., Digital Distribution of North America, A. Payne, Ed., 2006. Maps of the Birds of the Western Hemisphere, Version 1.0, [49] S. Solomon, D. Qin, M. Manning et al., Eds., Contribution NatureServe, Arlington, Va, USA, 2003. of Working Group 1 to the Fourth Assessment Report of [32] NatureServe, “NatureServe explorer: an online encyclopedia of the Intergovernmental Panel on Climate Change, Cambridge life, version 6.1,” 2010, http://www.natureserve.org/explorer. University Press, New York, NY, USA, 2007. [33] M. Stephens, N. J. Smith, and P. Donnelly, “A new statistical method for haplotype reconstruction from population data,” American Journal of Human Genetics, vol. 68, no. 4, pp. 978– 989, 2001. [34] M. Stephens and P. Donnelly, “A comparison of Bayesian methods for haplotype reconstruction from population genetic data,” American Journal of Human Genetics, vol. 73, no. 5, pp. 1162–1169, 2003. [35] J. K. Pritchard, M. Stephens, and P. Donnelly, “Inference of population structure using multilocus genotype data,” Genetics, vol. 155, no. 2, pp. 945–959, 2000. [36]D.Falush,M.Stephens,andJ.K.Pritchard,“Inferenceof population structure using multilocus genotype data: linked loci and correlated allele frequencies,” Genetics, vol. 164, no. 4, pp. 1567–1587, 2003. [37] D. Falush, M. Stephens, and J. K. Pritchard, “Inference of population structure using multilocus genotype data: dominant markers and null alleles,” Molecular Ecology Notes, vol. 7, no. 4, pp. 574–578, 2007. [38] M. J. Hubisz, D. Falush, M. Stephens, and J. K. Pritchard, “Inferring weak population structure with the assistance of Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 285081, 9 pages doi:10.1155/2012/285081

Review Article Pollinator-Driven Speciation in Sexually Deceptive Orchids

Shuqing Xu,1, 2 Philipp M. Schluter,¨ 1 and Florian P. Schiestl1

1 Institute of Systematic Botany, University of Zurich,¨ Zollikerstrasse 107, 8008 Zurich,¨ Switzerland 2 Department of Molecular Ecology, Max Planck Institute for Chemical Ecology, 07745 Jena, Germany

Correspondence should be addressed to Shuqing Xu, [email protected]

Received 23 July 2011; Accepted 8 November 2011

Academic Editor: Rui Faria

Copyright © 2012 Shuqing Xu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Pollinator-mediated selection has been suggested to play a major role for the origin and maintenance of the species diversity in orchids. Sexually deceptive orchids are one of the prime examples for rapid, pollinator-mediated plant radiations, with many species showing little genetic differentiation, lack of postzygotic barriers, but strong prezygotic reproductive isolation. These orchids mimic mating signals of female insects and employ male insects as pollinators. This kind of sexual mimicry leads to highly specialised pollination and provides a good system for investigating the process of pollinator-driven speciation. Here, we summarise the knowledge of key processes of speciation in this group of orchids and conduct a meta-analysis on traits that contribute to species differentiation, and thus potentially to speciation. Our study suggests that pollinator shift through changes in floral scent is predominant among closely related species in sexually deceptive orchids. Such shifts can provide a mechanism for pollinator-driven speciation in plants, if the resulting floral isolation is strong. Furthermore, changes in floral scent in these orchids are likely controlled by few genes. Together these factors suggest speciation in sexually deceptive orchids may happen rapidly and even in sympatry, which may explain the remarkable species diversity observed in this plant group.

1. Introduction Many theoretical analyses have suggested that three factors are of major importance in the process of ecological The species diversity in the orchid family is extraordinary. speciation: (i) disruptive selection, (ii) the strength of There are more than 20 000 species in the orchid family [1], reproductive isolation, and (iii) the genetic basis of traits representing about 10 percent of angiosperm species. The underlying reproductive isolation and traits under natural remarkable floral forms found among orchids have always selection [4, 5]. For reasons detailed below and discussed fascinated botanists and evolutionary biologists. Orchid elsewhere [2], nonadaptive speciation is unlikely to play species richness and the spectacular diversity of plant- a major role in the diversification of the well-investigated pollinator interactions provide an exceptional opportunity genera of sexually deceptive orchids, Ophrys and Chiloglottis. for the study of pollinator-driven speciation. There is more In this review, we therefore focus on these three factors in and more evidence suggesting that pollinators play a major sexually deceptive orchids. role in the generation of novel floral forms and the maintenance of species diversity in orchids [2, 3]. Sexually deceptive orchids are particularly suitable for investigating pollinator- 2. Sexually Deceptive Orchids driven speciation, because of their specialised pollination system. Here, using sexually deceptive orchids as examples, Pollination by deception is very common in orchids (about we systematically review what is known about the process one third of orchids are deceptive) [1, 6], with sexual of pollinator-driven speciation, examine the key factors that deception as an especially intriguing variety. Sexually decep- are essential for speciation processes, and discuss the possible tive orchids mimic the mating signals of female insects modes of speciation in this orchid pollination system. In and employ male insects as pollinators, inducing them to particular, we focus on the scenario of ecological speciation engage in mating behaviour and pseudocopulation [7–9]. with gene flow. The chemical mimicry of sex pheromones of pollinator 2 International Journal of Ecology females has been suggested to play a major role in this selection imposed by them may lead to different adaptations process [8–13]. Because insect mating signals are usually very [21, 49–51]. Floral traits involved in plant-pollinator interspecific [14], pollinator attraction by sexual deception is also actions, such as phenology, scent, colour, and morphology, very specific, each orchid only attracting one or very few are likely under pollinator-mediated selection, since they insect species [15–17]. This specific pollinator attraction can are directly associated with pollination success (Figure 1) therefore act as a reproductive barrier and prevent gene flow [2]. However, only one study has hitherto quantified pheno- among species [7, 17–22]. typic selection in a sexually deceptive orchid [52]. Another Pollination by sexual deception is mostly found in interesting yet noninvestigated aspect is density-dependent orchids from Australia, Europe, South Africa, and South selection, a potential driver of pollinator shifts in this America [23–27]. However, sexual deception has recently group of plants. Below, we summarise the traits likely been reported for the daisy Gorteria diffusa, the first under selection and highlight important targets for future confirmed case outside of Orchideceae [28], indicating investigations. that this pollination mechanism may be more common Pollinator-mediated selection on floral phenology can be than previously thought [29]. Among the different sex- seen as the need for synchronisation between flowering time ually deceptive orchids, the two best-studied genera are and emergence time of pollinators [23, 48]. Because of their Chiloglottis (Diurideae) in Australia, and Ophrys (Orchideae) pollinators’ ability to learn to avoid deceptive flowers [53], in Europe. In Chiloglottis, there are more than 30 species most sexually deceptive orchids are probably predominantly that are pollinated mostly by thynnine wasps of the genus pollinated by na¨ıve male insects [54]. This may impose Neozeleboria [15, 30–32]. In Ophrys, most of the more than strong selection on the flowering time of orchids. In other 200 species are pollinated by species representing different words, the flowering time of different orchid species can be genera of solitary bees, while a few species are pollinated by regarded as an adaptation to the optimal time of pollinator species of solitary wasps, flies, and beetles [17, 33, 34]. activity. From our meta-study, about 8–10% of closely Although sexually deceptive orchids have been suggested related species pairs of sexually deceptive orchids did not to be a spectacular example of ecological speciation (with or overlap in floral phenology (counted from Table S1 available without gene flow), few studies have systematically investi- online at doi:10.1155/2012/285081), suggesting disruptive gated key factors like selection and reproductive isolation. selection on floral phenology among closely related orchid Instead, most studies have focused on mechanistic aspects of species. However, most species pairs do overlap in flowering floral mimicry [8, 9, 35]orphylogeneticpatterns[36, 37]. time, the average degree of flowering time overlap being 59% Here, we critically review the process of ecological speciation and 47%, in Ophrys and Chiloglottis,respectively(Table 1). with gene flow for two orchid systems with sexual mimicry. In most sexually deceptive orchids, floral scent, which Additionally, we conduct a meta-analysis focusing on traits mimics the sex pheromones of female pollinators, has been that differ among closely related orchid species. For our suggested to be the main attractant of pollinators [8, 9, meta-study, we extracted and combined data from previous 20]; it is therefore likely to be under pollinator-mediated studies on orchid phylogeny [22, 37], pollinator informa- selection. Indeed, in both Ophrys and Chiloglottis, floral scent tion [34], and phenotypic traits [33, 38]. We considered composition is driven towards their pollinators’ preferences only species for which all three types of information is in both quantity and quality [8, 9, 11–13, 15, 20, 55]. available. Pairwise species comparisons of phenotypic traits Furthermore, direct scent manipulation experiments showed and geographic distribution [33, 38] within monophyletic that changing floral scent can dramatically reduce or increase groups, both within Ophrys and Chiloglottis [22, 37], were pollinator visitation ([22] and Xu et al. unpublished data). performed, because reproductive isolation and selection is Selection on floral scent is likely to be divergent among ff required to prevent gene flow within these groups. Since closely related species, which attract di erent pollinators ff the potential for speciation through hybridisation has been with di erent floral scent compositions. In Chiloglottis, 95% ff reviewed previously [35], we will not discuss this aspect here. of closely related species pairs produce a di erent type of floral scent and are pollinated by different pollinators (Table 1). ThesameistrueforOphrys, where 98% of closely related 3. Pollinator-Mediated Selection in species pairs attract different pollinators (Table 1), most Sexually Deceptive Orchids likely by the use of different floral odour bouquets [12, 17– 20, 55–59]. In angiosperms, floral and species diversity is thought to be Despite the importance of floral scent, floral colour driven by pollinator-mediated selection because of pollen and/or floral display may also play a role for effective pol- limitation [39]. Pollen limitation has been shown in most linator visitation. In many species of Ophrys and Chiloglottis, flowering plants [3, 39–42]. In orchids, many pollination the floral labellum (a modified petal) resembles the visual studies suggest that pollen limitation is widespread in the pattern of pollinator females, indicating that pollinators whole family [24, 43–45]. This is especially true for deceptive may use this visual signal to locate the flowers at short orchids [46], because pollinator learning may reduce polli- distance [38, 48]. Although it appears different to human nator visitation [47, 48]. Therefore, pollinators can impose eyes, the labellum of the Australian sexually deceptive strong selection on floral traits by mediating reproductive orchid Cryptostylis is effectively identical in coloration to its success and outcrossing [49]. Since morphology, sensory pollinator females [60]. Many Ophrys labella may mimic the preferences, and behaviour differ among pollinator insects, wings of a pollinator female with their shiny surface [16]. International Journal of Ecology 3

Floral trait divergence among closely related species (%) 0 20406080100 C. cunicularius O. exaltata

I II I II XII III XII III XI IV XI IV X V X V Floral phenology IX VI IX VI VIII VII VIII VII in pollination d R R Floral scent R R

Floral colour Data not available

Floral size Sequence of ﬂoral traits involve

Chiloglottis Ophrys

C. cunicularius landed on O. exaltata Figure 1: Summary of floral traits involved in plant-pollinator interaction. The left flow-chart shows the sequence of floral traits involved in pollination, whereas the right side shows the divergence of these traits between closely related species pairs in sexually deceptive orchids. Dark squares refer to Ophrys; grey squares to Chiloglottis. Floral trait data were extracted from Table 1 and Table S1; error bars represent standard error.

Table 1: Summary of comparisons of floral traits, distribution, and pollinators among closely related species pairs in Ophrys and Chiloglottis. Data on floral labellum, flowering time, and altitude range distribution are presented as mean (± standard deviation). Data on distribution, floral scent, and pollinators are presented as mean value.

Number of Overlap in labellum Overlap in Overlap in habitat: Overlap in Overlap in ﬂoral Overlap in Genus species-pairs length (%) ﬂowering time (%) Altitude (%) habitat: Area (%) scent (%) pollinators (%) compared [33, 38] [33, 38] [33, 38] [33, 38] [22] [34] Ophrys 213 42.6 (±31.7) 59.0 (±31.2) 83.0 (±19.2) 53 N/A 2 Chiloglottis 38 42.9 (±32.1) 47.4 (±34.6) 35.1 (±32.5) 74 5 6

In long-horned bee-pollinated Ophrys species, the perianth After the pollinator has landed on a flower, floral colour and/or contrast with background can affect the short- morphology such as labellum shape, size, and texture may range detectability of Ophrys flowers to their pollinators [61– have a strong influence on pollinator behaviour. In orchids, 63]. However, this does not appear to be the case in a Colletes precisely removing and delivering pollinia is highly depen- cunicularius-pollinated Ophrys species [57], suggesting that dent on the match of shape and/or size of the pollinators’ the role of colour and/or display on pollinator attraction may body and the floral labellum [48]. The significant correlation vary among different pollinators and visual systems. In our between pollinator body length and labellum length found metastudy, we did not include floral colour, because most in different Ophrys species clearly suggests that pollinator- descriptions of floral coloration are based on observations mediated selection on floral labellum size is strong [48, by human eyes and not on insect vision models. 52, 64]. Selection imposed on floral labellum size may 4 International Journal of Ecology be disruptive, since the body size of pollinators may vary hybridisation using molecular markers, are needed [73]. considerably from each other. Indeed, the average overlap Furthermore, studies comparing neutral markers versus in floral labellum length among closely related species is markers under disruptive selection will be important to shed only 43% in both Ophrys and Chiloglottis (Table 1), which light on the detailed mechanisms of speciation. suggests divergent selection on floral labellum size among Floral isolation among closely related species of sexually closely related species. deceptive orchids is either due to different pollinator attraction (ethological isolation) or different floral phenology (temporal isolation). Although mechanical isolation among 4. Reproductive Isolation in species of the Ophrys sections Pseudophrys and Ophrys is Sexually Deceptive Orchids evident [67, 74, 75], these groups are not closely related to each other and thus mechanical (morphological) isolation Reproductive isolation among closely related sexually decep- is unlikely to play an important role in their radiation tive orchid species is mainly due to floral isolation, a form (see discussion below). The key trait involved in ethological of pollinator-mediated prepollination reproductive isolation, isolation is floral scent [21, 22, 56]. while the postpollination reproductive barriers tend to be effectively absent or weak [21, 49](butsee[13]). For Ophrys flowers produce a complex mixture of more example, the closely related sympatric species O. exaltata, than 100 chemical compounds. Among these compounds, O. garganica,andO. sphegodes attract different solitary saturated and unsaturated hydrocarbons, that is, alkanes and bees, Colletes cunicularius, Andrena pilipes, and Andrena alkenes, are responsible for pollinator attraction in many nigroaenea, respectively, and the lack of pollinator sharing species [9, 11, 12, 18, 76]. Their specificity is due to quantita- ff leads to a highly effective barrier to gene flow in this case, tive variation in various alkenes with di erent double-bond although interspecies hand pollination experiments yielded positions and carbon chain lengths [9, 18, 19, 76, 77]. How- normal seed set [18]. Interestingly, this apparent lack of ever, there are also some Ophrys species that employ only postpollination reproductive barriers appears to be common a few unusual chemical compounds for specific pollinator in both Ophrys and Chiloglottis [22, 65, 66]. attraction. For example, in O. speculum, pollinator attraction can be achieved by the mixture of only eight compounds, Since reproductive isolation in sexually deceptive orchids in which two enantiomers of 9-hydroxydecanoic acid act is mainly (if not only) mediated by floral isolation, the as key substances [78]. Such signalling with few, specific strength and stability of floral isolation are critical for chemical compounds is more common in Chiloglottis.Here, pollinator-driven speciation. Although the hypothesis that specific pollinator attraction is achieved by a single unusual floral isolation may effectively prevent gene flow between compound (called “chiloglottone”) [8] or simple blends of closely related species and thus drive speciation was already chemically related chiloglottones [8, 22]. suggested half a century ago [16, 67], it was only recently that the strength of floral isolation was quantitatively measured Besides floral scent, floral phenology and floral mor- in situ between species of the O. sphegodes group [18]. The phology may also play a role in floral isolation among results indicated that floral isolation among closely related closely related species. Some species pairs show no overlap species is very strong, with an isolation index higher than in flowering time in sympatry, (e.g., O. iricolor and O. 0.98 [18]; a similar pattern was also found in another, mesaritica on Crete [69, 79]). Among such species pairs, unrelated Ophrys group (Gervasi and Schiestl, unpublished floral phenology may act as a strong reproductive barrier. ∼ data), indicating that strong floral isolation is probably However, as most closely related species pairs ( 90%) of common in Ophrys.InChiloglottis, pollinator behavioural sexually deceptive orchids do overlap in their flowering times tests on several sympatric species in different populations to a certain degree, floral phenology alone is usually not the ff showed no cross-attraction, likewise suggesting strong floral primary reproductive barrier. Di erences in floral labellum isolation in this system [30, 66]. Furthermore, population length between closely related species pairs may contribute genetic studies in both Ophrys and Chiloglottis suggest that to mechanical reproductive isolation as well. However, due floral isolation can effectively prevent gene flow between to pollinator movements during mating attempts, variation species in sympatry [18, 19, 66, 68–70]. in labellum morphology alone might not provide a strong barrier to gene flow. To better understand the evolutionary Other studies based on genetic and phylogenetic analysis patterns of the respective reproductive barriers in sexually showed low genetic divergence among species in Ophrys and deceptive orchids, direct quantitative measurements of the concluded that floral isolation might not be strong enough contributions of different floral traits to reproductive isola- to prevent gene flow between species in sympatry [37, 71, tion are needed. 72]. However, low genetic divergence among species can be explained by different scenarios, such as recent radiations resulting in the retention of ancestral polymorphism [70]. 5. The Genetic Basis of Reproductive Therefore, low genetic divergence among species cannot be “Barrier Traits” that Are under Pollinator- taken as strong evidence of weak floral isolation per se. Mediated Divergent Selection For a better understanding of the strength and variability of floral isolation, in situ measurements of floral isolation Understanding the genetic basis and complexity of divergent or systematic pollinator preference tests in more groups traits involved in reproductive isolation is important for and populations over several years, or rigorous tests for understanding the process of pollinator-driven speciation, International Journal of Ecology 5 since they basically constitute “automatic magic traits” [80]. odour signals for pollinator attraction in sexually deceptive Flowering time and flower morphology are likely to be com- orchids makes floral scent the prime candidate for mediating plex genetic traits controlled by multilayered developmental pollinator shifts. Disruptive selection on odour phenotypes genetic pathways that are still poorly understood in orchids would lead to species divergence following an ecological [81]. Floral odour and coloration are likely formed by speciation process [84, 85]. Since such odour traits may have biosynthetic pathways that appear to be relatively conserved a simple genetic basis [21, 22, 77, 81], this may also be a [21, 81]. In both Ophrys and Chiloglottis, the pollinator- process of speciation with gene flow (or, “genic speciation”) attractive scent compounds are probably derived from fatty [86–88].Here,divergenceoccursinspiteofgeneflow, acid intermediates [35, 77, 82]. Scent is a key trait for specific while selection on a few loci in the genome is responsible pollinator attraction and thus may be both important for for differences in pollinator attraction among species. The reproductive isolation and subject to divergent selection, as creation of population subdivision by pollinator-mediated changes in scent may directly cause pollinator shifts. selection may then lead to a buildup of larger islands of The number of genes underlying species differences in divergence in the genome brought about by an effective flower colour and odour has been suggested to be low [21]. reduction of recombination, a process termed divergence For example, pollinator specificity among some species of hitchhiking [89–91]. At the same time, it should be noted Ophrys is linked to the double-bond position in a series that many orchid species have not reached the point where of straight-chain alkenes, the bioactive pseudopheromone significant postmating barriers have accumulated [18, 65]. produced by the plants [18, 19].Thealkenesarelikely Overall, speciation by pollinator shift in sexually deceptive synthesised from few saturated (e.g., C16 and C18) fatty acids, orchids may frequently be consistent with ecological specia- and the introduction of the double-bond in these precursor tionprocesseswithgeneflow. compounds by stearoyl-ACP desaturase (SAD) enzymes The possibility for pollinator shifts is limited by the is thought to lead to an orchestrated change in double- local availability of potential pollinators, which is subject to bond position in the downstream biosynthetic products geographical variation, often referred to as the geographical [77]. Indeed, the alkene double-bond position in different pollinator mosaic [3]. The pollinator mosaic provides a route Ophrys species is associated with changes in expression to allopatric and parapatric divergence scenarios [3], which and possibly also activity of desaturases. For example, areasapplicabletosexuallydeceptiveorchidsastheyare higher expression levels of SAD2 lead to the presence of to other plants. However, unlike in many other pollination higher amounts of 9- and 12-alkenes in Ophrys sphegodes systems, sympatric speciation appears plausible in sexually [77]. Likewise, in sexually deceptive orchids with simple deceptive systems [17, 19, 20, 22, 35, 69], the credibility bioactive volatile chemistry such as Chiloglottis, a simple of which is strengthened by the combination of strong genetic basis underlying changes in scents and pollinator floral isolation, divergent selection, and possibly a simple attraction has been hypothesised [22]. This seems especially genic basis for species differences in this system. Molecular plausible since most chiloglottones differ only in minor data suggesting gene flow among sympatric species (e.g., chemical modifications of a common structure. Thus, it [72]) have been taken as evidence against pollinator-driven seems that changes at few regulatory or structural genes are speciation in sympatry [3]. However, (i) it is often difficult likely sufficient for changing traits that lead to changes in to distinguish between current gene flow and ancestral pollinator attraction in sexually deceptive orchids, favouring polymorphism [92, 93], and (ii) gene flow itself is consistent the view that changes in only a few genes of large effect rather with genic and ecological speciation processes [86–88]. This than many genes of small effect can make speciation more also highlights the difficulty researchers face to conclusively likely. It should be noted, however, that although studies of demonstrate sympatric speciation, requiring the combined molecular gene function may support this view, experiments study of population genetics (or genomics), phylogeny, to quantitatively assess the size of a gene’s phenotypic effect biogeography, and data on the strength and genetic basis of in the field are necessary to test this hypothesis. For example, reproductive barriers between sibling species. trait manipulation in the field or direct manipulation of gene Ecological speciation processes with gene flow are not expression via virus-induced gene silencing (VIGS) [81, 83] only consistent with sympatric modes of speciation, but would allow testing the effect of a candidate gene on plant- also the local nature of species divergence implies that spe- pollinator interactions. Moreover, another caveat is that the ciation processes may often result in progenitor-derivative apparent simplicity of a candidate trait based on existing species patterns rather than strict sister species relationships. evidence may be misleading due to the inherent detection Progenitor-derivative speciation (or “quantum speciation”) bias of current approaches. denotes the evolution of a new lineage (the derivative) from a source population (the progenitor) without affecting the progenitor [94–99]. Since pollinator shifts are not 6. The Process of Pollinator-Driven expected to change pollinator specificity in the source pop- Speciation in Orchids ulation, progenitor-derivative patterns are a likely outcome of pollinator-driven speciation in sexually deceptive orchids, A key element of pollinator-driven speciation is that changes with some genetic data supporting such scenarios [69, 70]. in pollinators are linked to reproductive isolation. While Despite the potential for sympatric and progenitor-derivative such pollinator shifts can in principle be brought about by speciation, it remains to be seen whether these modes a wide variety of floral trait changes, the importance of of speciation are common in sexually deceptive orchids, 6 International Journal of Ecology our current knowledge being limited by the resolution of [8] F. P. Schiestl, R. Peakall, J. G. Mant et al., “The chemistry phylogenetic reconstructions among closely related sexually of sexual deception in an orchid-wasp pollination system,” deceptive orchids (e.g., [37]). Science, vol. 302, no. 5644, pp. 437–438, 2003. [9]F.P.Schiestl,M.Ayasse,H.F.Paulusetal.,“Orchid pollination by sexual swindle,” Nature, vol. 399, no. 6735, pp. 7. Conclusions and Future Directions 421–422, 1999. [10] M. Ayasse, F. P. Schiestl, H. F. Paulus, D. Erdmann, and Sexually deceptive orchids provide an exceptional study W. Francke, “Chemical communication in the reproductive system for pollinator-driven speciation in plants. Our critical biology of Ophrys sphegodes,” Mitteilungen Der Deutschen review and meta-analysis of traits in closely related sexually Gesellschaft Fur¨ Allgemeine Und Agewandte Entomologie, vol. deceptive orchid species pairs suggest that pollinator shifts 11, no. 1–6, pp. 473–476, 1997. through changes in floral traits, especially floral scent, may [11] J. Stokl,¨ R. Twele, D. H. Erdmann, W. Francke, and M. Ayasse, be the main mechanism for speciation in this plant group. “Comparison of the flower scent of the sexually deceptive Such speciation through pollinator shift may happen rapidly orchid Ophrys iricolor and the female sex pheromone of its and even in sympatry, if floral isolation is strong and floral pollinator Andrena morio,” Chemoecology,vol.17,no.4,pp. trait changes are controlled by few genes. Therefore, to 231–233, 2007. understand the detailed process of speciation in sexually [12] J. Gogler,¨ J. Stokl,¨ A. Sramkova et al., “The role of pol- deceptive orchids, it is necessary to identify the genetic basis linator attracting scent in the sexually deceptive orchids Ophrys chestermanii, O. normanii and O. tenthredinifera,” of key floral traits and quantify their contributions to plant- Mitteilungen Der Deutschen Gesellschaft Fur¨ Allgemeine Und pollinator interaction. Furthermore, to better understand Angewandte Entomologie, vol. 16, pp. 175–178, 2008. the speciation patterns in this plant group, it would be [13] J. Gogler,¨ J. Stokl,¨ A. Sramkova et al., “Menage´ a` trois- important to evaluate their likelihood in a theoretical model two endemic species of deceptive orchids and one pollinator under biologically plausible conditions based on empirical species,” Evolution, vol. 63, no. 9, pp. 2222–2234, 2009. data [100, 101]. Such an integrative theoretical framework [14] M. Ayasse, R. J. Paxton, and J. Tengo,¨ “Mating behavior is still missing but is urgently needed to truly understand and chemical communication in the order Hymenoptera,” pollinator-driven speciation. Annual Review of Entomology, vol. 46, pp. 31–78, 2001. [15] C. C. Bower and G. R. Brown, “Pollinator specificity, cryptic species and geographical patterns in pollinator responses Acknowledgments to sexually deceptive orchids in the genus Chiloglottis:the The authors are grateful for the financial support by ETH Chiloglottis gunnii complex,” Australian Journal of Botany, vol. 57, no. 1, pp. 37–55, 2009. Zurich¨ (TH0206-2 to FPS) and Swiss National Science [16] B. Kullenberg, “Studies in Ophrys pollination,” Zoologiska Foundation (SNF 31003A 130796 to PMS). The authors Bidrag Fran˚ Uppsala, vol. 34, pp. 1–340, 1961. thank R. Faria and two anonymous reviewers for their [17] H. F. Paulus and C. Gack, “Pollinators as prepollinating constructive comments on the paper. isolation factors: evolution and speciation in Ophrys (Orchi- daceae),” Israel Journal of Botany, vol. 39, no. 1-2, pp. 43–79, References 1990. [18]S.Xu,P.M.Schluter,¨ G. Scopece et al., “Floral isolation is [1] L.R.Dressler,The Orchids: Natural History and Classification, the main reproductive barrier among closely related sexually Harvard University Press, Cambridge, Mass, USA, 1981. deceptive orchids,” Evolution, vol. 65, no. 9, pp. 2606–2620, [2] F. P. Schiestl, “Animal pollination and speciation in plants: 2011. general mechanisms and examples from the orchids,” in [19] J. Mant, R. Peakall, and F. P.Schiestl, “Does selection on floral Evolution of Plant Pollinator Interactions,S.Patiny,Ed., odor promote differentiation among populations and species Cambridge University Press, New York, NY, USA. of the sexually deceptive orchid genus Ophrys?” Evolution, vol. 59, no. 7, pp. 1449–1463, 2005. [3] S. D. Johnson, “Pollinator driven speciation in plants,” in Ecology and Evolution of Flowers,L.D.HarderandS.C.H. [20] F. P. Schiestl and M. Ayasse, “Do changes in floral odor cause Barrett, Eds., pp. 295–310, Oxford University Press, Oxford, speciation in sexually deceptive orchids?” Plant Systematics UK, 2006. and Evolution, vol. 234, no. 1–4, pp. 111–119, 2002. [21] F. P. Schiestl and P. M. Schluter,¨ “Floral isolation, specialized [4] N. M. Waser and D. R. Campbell, “Ecological speciation pollination, and pollinator behavior in orchids,” Annual in flowering plants,” in Adaptive Speciation, U. Dieckmann, Review of Entomology, vol. 54, pp. 425–446, 2009. M. Doebeli, M. J. Metz, and D. Tautz, Eds., pp. 264–277, [22] R. Peakall, D. Ebert, J. Poldy et al., “Pollinator specificity, flo- Cambridge University Press, Cambridge, UK, 2004. ral odour chemistry and the phylogeny of Australian sexually [5]J.A.CoyneandH.A.Orr,Speciation, Sinauer Associates, deceptive Chiloglottis orchids: implications for pollinator- Sunderland, Mass, USA, 2004. driven speciation,” New Phytologist, vol. 188, no. 2, pp. 437– [6] S. Cozzolino and A. Widmer, “Orchid diversity: an evolu- 450, 2010. tionary consequence of deception?” Trends in Ecology and [23] A. Dafni, “Mimicry and deception in pollination,” Annual Evolution, vol. 20, no. 9, pp. 487–494, 2005. Review of Ecology and Systematics, vol. 15, pp. 259–278, 1984. [7] R. Peakall, “Responses of male Zaspilothynnus trilobatus [24] L. A. Nilsson, “Orchid pollination biology,” Trends in Ecology Turner wasps to females and the sexually deceptive orchid and Evolution, vol. 7, no. 8, pp. 255–259, 1992. it pollinates,” Functional Ecology, vol. 4, no. 2, pp. 159–167, [25] K. E. Steiner, V. B. Whitehead, and S. D. Johnson, “Floral and 1990. pollinator divergence in two sexually deceptive South African International Journal of Ecology 7

orchids,” American Journal of Botany, vol. 81, no. 2, pp. 185– [43] S. D. Johnson and W. J. Bond, “Evidence for widespread 194, 1994. pollen limitation of fruiting success in Cape wildflowers,” [26] R. B. Singer, “The pollination mechanism in Trigonidium Oecologia, vol. 109, no. 4, pp. 530–534, 1997. obtusum Lindl (Orchidaceae: Maxillariinae): sexual mimicry [44] J. D. Ackerman and A. M. Montalvo, “Short- and long-term and trap-flowers,” Annals of Botany, vol. 89, no. 2, pp. 157– limitations to fruit production in a tropical orchid,” Ecology, 163, 2002. vol. 71, no. 1, pp. 263–272, 1990. [27]R.B.Singer,A.Flach,S.Koehler,A.J.Marsaioli,andM.D. [45] L. M. O’Connell and M. O. Johnston, “Male and female C. E. Amaral, “Sexual mimicry in Mormolyca ringens (Lindl.) pollination success in a deceptive orchid, a selection study,” Schltr. (Orchidaceae: Maxillariinae),” Annals of Botany, vol. Ecology, vol. 79, no. 4, pp. 1246–1260, 1998. 93, no. 6, pp. 755–762, 2004. [46] G. Scopece, S. Cozzolino, S. D. Johnson, and F. P. Schiestl, [28] A. G. Ellis and S. D. Johnson, “Floral mimicry enhances “Pollination efficiency and the evolution of specialized pollen export: the evolution of pollination by sexual deceit deceptive pollination systems,” The American Naturalist, vol. outside of the Orchidaceae,” The American Naturalist, vol. 175, no. 1, pp. 98–105, 2010. 176, no. 5, pp. E143–E151, 2010. [47] J. B. Ferdy, P.H. Gouyon, J. Moret, and B. Godelle, “Pollinator [29] F. P. Schiestl, “Pollination: sexual mimicry abounds,” Current behavior and deceptive pollination: learning process and Biology, vol. 20, no. 23, pp. R1020–R1022, 2010. floral evolution,” The American Naturalist, vol. 152, no. 5, pp. 696–705, 1998. [30] C. C. Bower, “Demonstration of pollinator-mediated repro- [48] H. F. Paulus, “Deceived males—Pollination biology of the ductive isolation in sexually deceptive species of Chiloglottis Mediterranean orchid genus Ophrys (Orchidaceae),” Journal (Orchidaceae: Caladeniinae),” Australian Journal of Botany, Europaischer¨ Orchideen, vol. 38, no. 2, pp. 303–353, 2006. vol. 44, no. 1, pp. 15–33, 1996. [49] K. M. Kay and R. D. Sargent, “The role of animal polli- [31] C. C. Bower, “Specific pollinators reveal a cryptic taxon in nation in plant speciation: integrating ecology, geography, the bird orchid, Chiloglottis valida sensu lato (Orchidaceae) and genetics,” Annual Review of Ecology, Evolution, and in south-eastern Australia,” Australian Journal of Botany, vol. Systematics, vol. 40, pp. 637–656, 2009. 54, no. 1, pp. 53–64, 2006. [50] G. L. Stebbins, “Adaptive radiation of reproductive charac- [32] J. Mant, R. Peakall, and P. H. Weston, “Specific pollina- teristics in angiosperms, I: pollination mechanisms,” Annual tor attraction and the diversification of sexually deceptive Review of Ecology, Evolution, and Systematics,vol.1,no.1,pp. Chiloglottis (Orchidaceae),” Plant Systematics and Evolution, 307–326, 1970. vol. 253, no. 1–4, pp. 185–200, 2005. [51] R. D. Sargent and S. P. Otto, “The role of local species [33] P. Delforge, Orchids of Europe, North Africa, and the Middle abundance in the evolution of pollinator attraction in East, A&C Black, London, UK, 3rd edition, 2006. flowering plants,” The American Naturalist, vol. 167, no. 1, [34] A. C. Gaskett, “Orchid pollination by sexual deception: pp. 67–80, 2006. pollinator perspectives,” Biological Reviews,vol.86,no.1,pp. [52] S. Benitez-Vieyra, A. M. Medina, and A. A. Cocucci, “Vari- 33–75, 2011. able selection patterns on the labellum shape of Geoblasta [35] M. Ayasse, J. Stokl,¨ and W. Francke, “Chemical ecology and pennicillata, a sexually deceptive orchid,” JournalofEvolu- pollinator-driven speciation in sexually deceptive orchids,” tionary Biology, vol. 22, no. 11, pp. 2354–2362, 2009. Phytochemistry, vol. 72, no. 13, pp. 1667–1677, 2011. [53] M. Ayasse, F. P. Schiestl, H. F. Paulus et al., “Evolution [36] M. Soliva, A. Kocyan, and A. Widmer, “Molecular phyloge- of reproductive strategies in the sexually deceptive Orchid netics of the sexually deceptive orchid genus Ophrys (Orchi- Ophrys sphegodes: how does flower-specific variation of odor daceae) based on nuclear and chloroplast DNA sequences,” signals influence reproductive success?” Evolution, vol. 54, Molecular Phylogenetics and Evolution, vol. 20, no. 1, pp. 78– no. 6, pp. 1995–2006, 2000. 88, 2001. [54] H. F. Paulus, “Co-evolution and unilateral adaptation in [37] D. S. Devey, R. M. Bateman, M. F. Fay, and J. A. Hawkins, flower-pollinator systems: pollinators as pacemakers in the “Friends or relatives? Phylogenetics and species delimitation evolution of flowers,” Verhandlungen Der Deutschen Zoolo- in the controversial European orchid genus Ophrys,” Annals gischen Gesellschaft, vol. 81, pp. 25–46, 1988. of Botany, vol. 101, no. 3, pp. 385–402, 2008. [55] J. Stokl,¨ P. M. Schluter,¨ T. F. Stuessy et al., “Speciation in sexually deceptive orchids: pollinator-driven selection [38] D. L. Jones, ACompleteGuidetoNativeOrchidsofAustralia maintains discrete odour phenotypes in hybridizing species,” Including the Island Territories, Reed New Holland, Sydney, Biological Journal of the Linnean Society,vol.98,no.2,pp. Australia, 2006. 439–451, 2009. [39] T. M. Knight, J. A. Steets, J. C. Vamosi et al., “Pollen [56] F. P. Schiestl, “On the success of a swindle: pollination by limitation of plant reproduction: pattern and process,” deception in orchids,” Die Naturwissenschaften, vol. 92, no. Annual Review of Ecology, Evolution, and Systematics, vol. 36, 6, pp. 255–264, 2005. pp. 467–497, 2005. [57] N. J. Vereecken and F. P. Schiestl, “On the roles of colour [40] M. Burd, “Bateman’s principle and plant reproduction: the and scent in a specialized floral mimicry system,” Annals of role of pollen limitation in fruit and seed set,” Botanical Botany, vol. 104, no. 6, pp. 1077–1084, 2009. Review, vol. 60, no. 1, pp. 83–139, 1994. [58] A. K. Borg-Karlson and J. Tengo,¨ “Odor mimetism? Key [41] T. L. Ashman, T. M. Knight, J. A. Steets et al., “Pollen substances in Ophrys lutea-Andrena pollination relationship limitation of plant reproduction: ecological and evolutionary (Orchidaceae: Andrenidae),” Journal of Chemical Ecology, vol. causes and consequences,” Ecology, vol. 85, no. 9, pp. 2408– 12, no. 9, pp. 1927–1941, 1986. 2421, 2004. [59] N. J. Vereecken, S. Cozzolino, and F. P.Schiestl, “Hybrid floral [42] L. D. Harder and M. A. Aizen, “Floral adaptation and diver- scent novelty drives pollinator shift in sexually deceptive sification under pollen limitation,” Philosophical Transactions orchids,” BMC Evolutionary Biology, vol. 10, no. 1, article 103, of the Royal Society B, vol. 365, no. 1539, pp. 529–543, 2010. 2010. 8 International Journal of Ecology

[60] A. C. Gaskett and M. E. Herberstein, “Colour mimicry (Orchidaceae),” Annals of Botany, vol. 104, no. 3, pp. 497– and sexual deception by Tongue orchids (Cryptostylis),” 506, 2009. Naturwissenschaften, vol. 97, no. 1, pp. 97–102, 2010. [76] J. Mant, C. Brandli,¨ N. J. Vereecken, C. M. Schulz, W. Francke, [61] M. Streinzer, T. Ellis, H. F. Paulus, and J. Spaethe, “Visual dis- and F. P. Schiestl, “Cuticular hydrocarbons as sex pheromone crimination between two sexually deceptive Ophrys species of the bee Colletes cunicularius and the key to its mimicry by a bee pollinator,” Arthropod-Plant Interactions, vol. 4, no. by the sexually deceptive orchid, Ophrys exaltata,” Journal of 3, pp. 141–148, 2010. Chemical Ecology, vol. 31, no. 8, pp. 1765–1787, 2005. [62] M. Streinzer, H. F. Paulus, and J. Spaethe, “Floral colour sig- [77] P. M. Schluter,¨ S. Xu, V. Gagliardini et al., “Stearoyl-acyl nal increases short-range detectability of a sexually deceptive carrier protein desaturases are associated with floral isolation orchid to its bee pollinator,” Journal of Experimental Biology, in sexually deceptive orchids,” Proceedings of the National vol. 212, no. 9, pp. 1365–1370, 2009. Academy of Sciences of the United States of America, vol. 108, [63] J. Spaethe, W. H. Moser, and H. F. Paulus, “Increase of no. 14, pp. 5696–5701, 2011. pollinator attraction by means of a visual signal in the [78] M. Ayasse, F. P. Schiestl, H. F. Paulus, F. Ibarra, and W. sexually deceptive orchid, Ophrys heldreichii (Orchidaceae),” Francke, “Pollinator attraction in a sexually deceptive orchid Plant Systematics and Evolution, vol. 264, no. 1-2, pp. 31–40, by means of unconventional chemicals,” Proceedings of the 2007. Royal Society B, vol. 270, no. 1514, pp. 517–522, 2003. [64] N. J. Vereecken, “Deceptive behavior in plants. I. Pollination [79] H. Kretzschmar, G. Kretzschmar, and W. Eccarius, Orchideen by sexual deception in orchids: a host-parasite perspective,” auf Kreta, Kasos und Karpathos. Ein Feldfuhrer¨ durch die in Plant-Environment Interactions, F. Baluska, Ed., pp. 203– Orchideenflora der zentralen Insel der Sud¨ ag¨ ais¨ , Selbstverlag 222, Springer, Heidelberg, Germany, 2009. H. Kretzschmar, Bad Hersfeld, Germany, 2002. [65] G. Scopece, A. Musacchio, A. Widmer, and S. Cozzolino, [80]M.R.Servedio,G.S.VanDoorn,M.Kopp,A.M.Frame,and “Patterns of reproductive isolation in Mediterranean decep- P. Nosil, “Magic traits in speciation: “magic” but not rare?” tive orchids,” Evolution, vol. 61, no. 11, pp. 2623–2642, 2007. Trends in Ecology & Evolution, vol. 26, no. 8, pp. 389–397, [66] J. Mant, C. C. Bower, P. H. Weston, and R. Peakall, “Phylo- 2011. geography of pollinator-specific sexually deceptive Chiloglot- [81] P. M. Schluter¨ and F. P. Schiestl, “Molecular mechanisms of tis taxa (Orchidaceae): evidence for sympatric divergence?” floral mimicry in orchids,” Trends in Plant Science, vol. 13, Molecular Ecology, vol. 14, no. 10, pp. 3067–3076, 2005. no. 5, pp. 228–235, 2008. [67] B. Kullenberg, “Investigations on the pollination of Ophrys [82] S. Franke, F. Ibarra, C. M. Schulz et al., “The discovery of species,” Oikos, vol. 2, no. 1, pp. 1–19, 1950. 2,5-dialkylcyclohexan-1,3-diones as a new class of natural [68] P. M. Schluter,¨ P. M. Ruas, G. Kohl, C. F. Ruas, T. F. products,” Proceedings of the National Academy of Sciences of Stuessy, and H. F. Paulus, “Reproductive isolation in the the United States of America, vol. 106, no. 22, pp. 8877–8882, Aegean Ophrys omegaifera complex (Orchidaceae),” Plant 2009. Systematics and Evolution, vol. 267, no. 1–4, pp. 105–119, [83] T. M. Burch-Smith, J. C. Anderson, G. B. Martin, and S. 2007. P. Dinesh-Kumar, “Applications and advantages of virus- [69] P. M. Schluter,¨ P. M. Ruas, G. Kohl, C. F. Ruas, T. F. Stuessy, induced gene silencing for gene function studies in plants,” and H. F. Paulus, “Genetic patterns and pollination in Ophrys Plant Journal, vol. 39, no. 5, pp. 734–746, 2004. iricolor and O. mesaritica (Orchidaceae): sympatric evolution [84] T. J. Givnish, “Ecology of plant speciation,” Taxon, vol. 59, no. by pollinator shift,” Botanical Journal of the Linnean Society, 5, pp. 1329–1366, 2010. vol. 159, no. 4, pp. 583–598, 2009. [85] D. Schluter and G. L. Conte, “Genetics and ecological [70] P. M. Schluter,¨ P. M. Ruas, G. Kohl, C. F. Ruas, T. F. speciation,” Proceedings of the National Academy of Sciences Stuessy, and H. F. Paulus, “Evidence for progenitor-derivative of the United States of America, vol. 106, pp. 9955–9962, 2009. speciation in sexually deceptive orchids,” Annals of Botany, [86] C. I. Wu, “The genic view of the process of speciation,” vol. 108, pp. 895–906, 2011. Journal of Evolutionary Biology, vol. 14, no. 6, pp. 851–865, [71] D. S. Devey, R. M. Bateman, M. F. Fay, and J. A. Hawkins, 2001. “Genetic structure and systematic relationships within the [87] C. Lexer and A. Widmer, “The genic view of plant specia- Ophrys fuciflora aggregate (Orchidaceae: Orchidinae): high tion: recent progress and emerging questions,” Philosophical diversity in Kent and a wind-induced discontinuity bisecting Transactions of the Royal Society B, vol. 363, no. 1506, pp. the Adriatic,” Annals of Botany, vol. 104, no. 3, pp. 483–495, 3023–3036, 2008. 2009. [88] C. I. Wu and C. T. Ting, “Genes and speciation,” Nature [72] M. Soliva and A. Widmer, “Gene flow across species bound- Reviews Genetics, vol. 5, no. 2, pp. 114–122, 2004. aries in sympatric, sexually deceptive Ophrys (Orchidaceae) [89] S. Via, “Natural selection in action during speciation,” species,” Evolution, vol. 57, no. 10, pp. 2252–2261, 2003. Proceedings of the National Academy of Sciences of the United [73] R. Peakall, C. C. Bower, A. E. Logan, and H. I. Nicol, “Con- States of America, vol. 106, pp. 9939–9946, 2009. firmation of the hybrid origin of Chiloglottis x pescottiana [90] S. Via and J. West, “The genetic mosaic suggests a new role (Orchidaceae: Diurideae). 1. Genetic and morphometric for hitchhiking in ecological speciation,” Molecular Ecology, evidence,” Australian Journal of Botany,vol.45,no.5,pp. vol. 17, no. 19, pp. 4334–4345, 2008. 839–855, 1997. [91] J. L. Feder and P. Nosil, “The efficacy of divergence [74] L. Agren,˚ B. Kullenberg, and T. Sensenbaugh, “Congruences hitchhiking in generating genomic islands during ecological in pilosity between three species of Ophrys (Orchidaceae) in speciation,” Evolution, vol. 64, no. 6, pp. 1729–1747, 2010. their hymenopteran pollinators,” Nova Acta Regiae Societatis [92] D. J. Funk and K. E. Omland, “Species-level paraphyly Scientiarum Upsaliensis, Serie V, vol. 3, no. 3, pp. 15–25, 1984. and polyphyly: frequency, causes, and consequences, with [75] P.Cortis, N. J. Vereecken, F. P.Schiestl, M. R. Barone Lumaga, insights from animal mitochondrial DNA,” Annual Review A. Scrugli, and S. Cozzolino, “Pollinator convergence and the of Ecology, Evolution, and Systematics, vol. 34, pp. 397–423, nature of species’ boundaries in sympatric Sardinian Ophrys 2003. International Journal of Ecology 9

[93] C. Pinho and J. Hey, “Divergence with gene ﬂow: models and data,” Annual Review of Ecology, Evolution, and Systematics, vol. 41, pp. 215–230, 2010. [94] D. A. Levin, “Local speciation in plants: the rule not the exception,” Systematic Botany, vol. 18, no. 2, pp. 197–208, 1993. [95] D. A. Levin, “The ecological transition in speciation,” New Phytologist, vol. 161, no. 1, pp. 91–96, 2004. [96] L. H. Rieseberg and L. Brouillet, “Are many plant species paraphyletic?” Taxon, vol. 43, no. 1, pp. 21–32, 1994. [97] D. J. Crawford, “Progenitor-derivative species pairs and plant speciation,” Taxon, vol. 59, no. 5, pp. 1413–1423, 2010. [98] L. D. Gottlieb, “Rethinking classic examples of recent speciation in plants,” New Phytologist, vol. 161, no. 1, pp. 71–82, 2004. [99] V. Grant, Plant Speciation, Columbia University Press, New York, NY, USA, 2nd edition, 1981. [100] D. I. Bolnick and B. M. Fitzpatrick, “Sympatric speciation: models and empirical evidence,” Annual Review of Ecology, Evolution, and Systematics, vol. 38, pp. 459–487, 2007. [101] D. I. Bolnick, “Sympatric speciation in threespine stickleback: why not?” International Journal of Ecology, vol. 2011, Article ID 942847, 15 pages, 2011. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 273413, 10 pages doi:10.1155/2012/273413

Research Article Synergy between Allopatry and Ecology in Population Differentiation and Speciation

Yann Surget-Groba,1, 2 Helena Johansson,1, 3 and Roger S. Thorpe1, 4

1 School of Biological Sciences, Bangor University, Bangor LL57 2UW, UK 2 Department of Ecology and Evolutionary Biology, University of California Santa Cruz, Santa Cruz, CA 95064, USA 3 Department of Biosciences, University of Helsinki, P.O. Box 65, 00014 Helsinki, Finland 4 School of Biological Sciences, College of Natural Sciences, Bangor University, Deiniol Road, Bangor, Gwynedd LL57 2UW, UK

Correspondence should be addressed to Roger S. Thorpe, [email protected]

Received 19 July 2011; Revised 23 September 2011; Accepted 26 October 2011

Academic Editor: Andrew Hendry

Copyright © 2012 Yann Surget-Groba et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The general diversity pattern of the Caribbean anole radiation has been described in detail; however, the actual mechanisms at the origin of their diversification remain controversial. In particular, the role of ecological speciation, and the relative importance of divergence in allopatry and in parapatry, is debated. We describe the genetic structure of anole populations across lineage contact zones and ecotones to investigate the effect of allopatric divergence, natural selection, and the combination of both factors on population differentiation. Allopatric divergence had no significant impact on differentiation across the lineage boundary, while a clear bimodality in genetic and morphological characters was observed across an ecotone within a single lineage. Critically, the strongest differentiation was observed when allopatry and ecology act together, leading to a sharp reduction in gene flow between two lineages inhabiting different habitats. We suggest that, for Caribbean anoles to reach full speciation, a synergistic combination of several historical and ecological factors may be requisite.

1. Introduction differentiation of populations, this model is the easiest to explain speciation, especially if the isolated populations are Speciation, the mechanism at the origin of species diversi- exposed to distinct selective environments (reviewed in [8]). fication, is one of the most studied subjects in evolutionary The possibility of speciation in the presence of recurrent biology. Despite this enormous interest, very little is known gene flow (sympatric or parapatric speciation) is much more about the factors needed for speciation to occur. For instance, debated. In this model, the divergence between populations the relative importance of ecological versus purely historical exchanging migrants is driven by ecological differentiation factors, as well as the geographic context of speciation, is (ecological speciation [9]), and/or sexual differentiation still debated, and mechanisms that are sometimes invoked to (speciation by sexual selection [10]). Because even a limited explain lack of speciation, such as observation of a species- amount of gene flow is expected to counter differentiation, area relationship [1], are themselves not fully understood. early theoretical studies rejected this speciation model [11]. The most widely recognised speciation model is the However, other theoretical and empirical work suggested allopatric model, where different populations that are geo- that nonallopatric speciation is possible under particular graphically isolated develop genetic incompatibilities, purely circumstances (reviewed in [12]). by genetic drift or founder effects [2–4], by adapting A group in which both the geographic context of spe- to different habitats (by-product speciation [5]), by sex- ciation, and the role of ecological speciation, is debated is ual selection [6], or by fixation of incompatible mutations the Caribbean anoles. The anole radiation is a highly diverse through adaptation to a similar habitat (mutation-order species group useful for understanding the mechanisms speciation [7]). Because there is no gene flow to counter the at the origin of diversification. Anolis is one of the most 2 International Journal of Ecology speciose vertebrate genera with more than 400 described very heterogeneous, both topographically and ecologically. species. Of these, ca. 150 are found in the Caribbean islands, The mountains in the North are exposed to the trade winds and are thought to originate from just two independent and receive a very high amount of precipitation all year colonizations from the mainland. This species diversity is round. At the opposite, the northern Caribbean coast is in accompanied by a very high morphological and ecological the “rain shadow” of these mountains and is much drier diversity. For these reasons, the anole radiation in the and seasonal. Hence, the habitat changes dramatically from Caribbean has been the focus of intense studies in the last a cool montane rainforest to a hot xeric scrubland in just decades (reviewed in [13]). a few kilometres. Previous studies [18] have shown that One of the striking patterns in Anolis diversity is the con- the divergent selective forces along this habitat gradient trast between the Greater and the Lesser Antilles. Whereas lead to significant morphological and genetic differentiation numerous species coexist within the large islands of the between coastal and mountain populations of A. roquet.On Greater Antilles, islands of the Lesser Antilles have at most the neighbouring island of Dominica, a similar situation two naturally occurring species with most of the islands appears to occur with its endemic anole, Anolis oculatus having only one. Furthermore, if most of the diversification [32], and indeed, in the single case to date where contact in the Greater Antilles occurred within island, most of the zones have been studied in the Greater Antilles, a significant species pairs in the Lesser Antilles are not sister species, reduction in gene flow between divergent lineages is only suggesting that these species pairs did not diverge within observed in an area with a steep ecological gradient [33]. island but rather came together after dispersal from another In this paper we reanalysed data from a previous study island, with one possible exception in Saint Vincent [13]. [18] and added a new transect of A. roquet populations The Caribbean anoles show a well-documented species- where the ecotone and the lineage boundary overlap. We area relationship [14, 15]; however, the reason why so many compared the population structure of this new transect speciation events happened in the Greater Antilles while to the two previously published ones, one between two almost none happened in the Lesser Antilles is still specu- lineages within a single habitat, and the other between two lative. Losos and Schluter [14] proposed two explanations habitats within a single lineage. We studied the population for this observation. First, larger islands could offer more structure and admixture rates along these different transects opportunities of geographic fragmentation and hence lead to investigate the effects of geographic isolation, ecological to the formation of more species by allopatric speciation. isolation, and the combination of these two factors on anole Second, larger islands may have more habitat diversity, population differentiation and speciation. We observe that and hence populations submitted to divergent selective population differentiation is at its highest when both factors pressure in different habitats could lead to the formation act simultaneously and suggest that a possible explanation of of more species by ecological speciation. The observation the species-area relationship observed in Caribbean anoles is that some of the largest islands in the Lesser Antilles, like that the probability of both allopatric and ecological factors Dominica, Guadeloupe, or Martinique, have a very high acting in synergy increases with island size. habitat diversity and hence should be able to support several species if ecological speciation was the driving force in anole diversification led Losos [13] to suggest that nonallopatric 2. Material and Methods modes of speciation driven by ecological speciation are not supported in anoles. 2.1. Sampling. The distribution of anoles in Martinique is Several studies suggest the opposite. For instance, bec- more or less continuous. Hereafter we use the term “popu- ause of its complex geologic history [16], the island of lation” to refer to a discrete sampling site, not to genetically ff Martinique has offered plenty of opportunities for allopatric or ecologically di erentiated entities. We sampled three speciation to occur in its endemic anole, Anolis roquet. distinct transects in northern Martinique (Figure 1). First, Present day Martinique was once formed of separate proto- the “lineage transect” (transect II in [18]) consisted of eight islands where distinct mtDNA lineages of A. roquet evolved populations sampled in similar habitat (transitional forest) in allopatry for millions of years before coming back into across the lineage boundary between the North-West (NW) secondary contact [17]. However, high gene flow is observed and the Central (C) lineages. Second, the “habitat transect” between previously allopatric lineages showing that this long (populations 1 to 6 from transect IV in [18]) consisted of six geographic isolation did not lead to complete speciation populations sampled within the NW lineage and across an [18]. Similarly, deep mtDNA lineages are observed in several ecotone between coastal scrubland and montane rainforest. species, both in the Lesser Antilles [19–23], and in the Finally, the “combined transect” (new data) consisted of Greater Antilles [24–31]. In the cases where it has been eight populations sampled across the lineage boundary studied, no restriction of gene flow has been detected between the NW and the C lineages and across the ecotone between these previously allopatric lineages [18, 32, 33], with between coastal scrubland and montane rainforest. For each one possible exception in North-Eastern Martinique [18]. population, tail tips from 48 individuals were sampled, and All this suggests that even if allopatric speciation probably quantitative traits measurements (see below) were collected occurred in some situations, geographic isolation alone is not on ten adult males. sufficient to reach speciation in anoles. Instead, ecological gradients seem to be driving popu- 2.2. Genetic Analyses. For each individual, genotypes were lation differentiation in Martinique anoles. This island is scored at nine microsatellite loci (AAE-P2F9, ABO-P4A9, International Journal of Ecology 3

method implemented in GESTE v2.0 [41]. GESTE employs a hierarchical Bayesian framework to estimate population 1 specific Fst (representing the differentiation of a given population relative to all other populations) and uses a generalized linear model in order to test the contribution of biotic or 6 nonbiotic factors to genetic structuring. For our analyses 8 8 we used the default settings (sample size of 10000, thinning 1 interval 20), and in accordance with guidelines we allowed ten pilot runs to estimate means and variances for the required input parameters [41]. The analysis was repeated five times for each transect to ensure that the results were 1 consistent. Three different factors were considered. First 19 bioclimatic variables were obtained for each population from the WorldClim database. These variables were annual mean temperature, mean diurnal range, isothermality, tempera- N ture seasonality, maximum temperature warmest month, W E minimum temperature coldest month, temperature annual range, mean temperature wettest quarter, mean temperature S 0 2.5 51015 20 driest quarter, mean temperature warmest quarter, mean temperature coldest quarter, annual precipitation, precipita- (kilometres) tion wettest month, precipitation driest month, precipitation seasonality, precipitation wettest quarter, precipitation driest Lineage transect Lineage boundary quarter, precipitation warmest quarter, and precipitation Ecotone Habitat transect coldest quarter. A principal component analysis (PCA) of Combined transect the log transformed variables was performed using the Figure 1: Map of the sampled populations. The lineage transect ade4 package [42] in the R environment [43]. The first is indicated in red, the habitat transect in blue, and the combined component explained more than 75% of the variation and transect in yellow. The first and last sites of each transect are was used as a composite bioclimatic variable to describe the numbered on the map. environmentamongeachsite.Thisfirstaxisdescribedthe variation from hot and seasonally dry coastal sites to cool and wet montane sites. Second, the geographic connectivity AEX-P1H11, ARO-HJ2 [34], ARO-035, ARO-062, ARO-065, of the populations was estimated as described in [44]to ARO-120 [35], and ALU-MS06 [36]). Mitochondrial DNA include a measure of geographic isolation for each individual lineage was inferred by amplifying the complete Cyt-b gene, population (mean geographical distance between a given digesting the PCR product with the restriction enzymes population and all other populations). Third, each site was SspI and DraI (New England Biolabs) whose cutting pattern assigned to a lineage (lineage transect), a habitat (habitat allows to distinguish the different lineages present in this transect), or both (combined transect) as follows. For the species. More details on these molecular techniques can be lineage transect, sites 1–3 were assigned to the NW lineage, found in [18]. Only individuals successfully genotyped at and sites 4–8 were assigned to the C lineage. For the habitat least at four loci were kept for analyses. Three sets of analyses transect, sites 1–3 were assigned to coastal habitat, and sites were conducted. 4–6 were assigned to montane habitat. For the combined (i) Admixture Analysis. We estimated the genetic struc- transect, sites 1–3 were assigned to NW/coastal group, while ture within each transect using the Bayesian clustering sites 4–8 were assigned to C/montane group. method implemented in STRUCTURE v2.3.2 [37]. First, we (iii) Transects Comparison. To determine if the genetic estimated the most likely number of clusters (k) by running structure was different among transects, we computed and the analysis with k ranging from 1 to the real number of plotted the pairwise Fst’ values between populations on populations. For each k value, the analysis was run 10 times each side of the contact zone within each transect. This using the admixture model, with a burn-in of 100,000 steps allowed us to compare the level of across habitat/lineage for a total run length of 500,000 steps. The optimal number differentiation among transects. Because the observations of clusters (K) was inferred following the method outlined are nonindependent and hence violate the assumption of in [38], choosing the number of clusters corresponding to both parametric and nonparametric tests, these plots provide the highest rate of change of the log probability of data qualitative information that has not been tested statisti- between successive K values. We considered an individual as cally. admixed if less than 80% of its genome was assigned to a single cluster. We also estimated the standardised pairwise Fst 2.3. Combined Analysis of Genetic and Quantitative Data. We (Fst’ [39]) between adjacent populations along each transect conducted a modified version of the Discriminant Analysis of using RecodeData v0.1 [39] and FSTAT v2.9.3 [40]. Principal Components (DAPC, [45]) to describe the global (ii) We then estimated the influence of different poten- structure of populations within each transect using both tially important factors on the genetic structure using the genetics and morphology. 4 International Journal of Ecology

First different morphological characters were recorded as is no obvious trend in the pairwise Fst’ that vary between previously described [22, 46]. These included body dimen- 0.059 and 0.112 (Figure 2(b)). The combined dataset shows sions (jaw length, head length, head depth, head width, a much higher level of variation than on the lineage upper leg length, lower leg length, dewlap length), scalation transect (almost twice as high), and a marked differentiation (number of postmental scales, scales between supraorbitals, between populations 1-2 and 4–6, with population 3 being ventral scales, and dorsal scales), colour pattern (number of intermediate (Figure 3(b)). The population specific Fst are dorsal chevrons, chevron intensity, occipital “A” mark, black higher than in any site of the lineage transect (ranging from dorsal reticulation, white spots), and trunk background 0.0148 to 0.0522, Table 1) confirming the higher genetic colour (percentage red, green and blue hue on the posterior structuring along this transect. The best model to explain trunk). These were combined to six independent hues this genetic structure only incorporates a constant, but (UV 330–380 nm, UV/violet 380–430 nm, blue 430–490 nm, the second and third best models have a nonnegligible green 520–590 nm, yellow/orange 590–640 nm, and red 640– probability and incorporate, respectively, the habitat type 710 nm) extracted from the spectrum of the anterior and and the bioclimatic variable (Table 1); these two factors posterior dewlap by a multiple-group eigenvector procedure have a combined probability of 0.175 and 0.134, respectively [23]. This combined data set was subjected to a principal (Table 2). component analysis (PCA) using the ade4 package in R. Then, the microsatellite data were also subjected to a PCA 3.3. Combined Transect. Here again two genetic clusters were using the adegenet package [47] in R. Finally, the compo- identified (cluster 1: populations 1–3; cluster 2: populations nents from the genetic (14 components) and quantitative 4–8, Figure 2(c)), but with a much higher differentiation. (4 components) datasets were combined and subjected to a This genetic division corresponds both to the habitat and linear discriminant analysis using the MASS package [48]in lineage boundaries, the ecotone being situated between R, using the population of origin as the grouping factor. populations 3 and 4, as well as the lineage boundary (Figure 4). On this transect, the admixture rate is much lower 3. Results (range 4–27%) than on the other transects. It is somewhat higher in population 4, situated at the lineage boundary 3.1. Lineage Transect. Two genetic clusters were identified and at the ecotone, but even in this population, it is lower in this transect, with individuals from populations 1 to 3 than in any of the lineage transect’s populations and than being assigned in majority to the first cluster and individuals in most of the habitat transect’s populations (Figure 2(c)). from populations 4 to 8 to the second cluster (Figure 2(a)). There is also a marked increase of the pairwise Fst’ at the This separation corresponds to what was observed with contact zone, with a value of 0.169 between populations 3 mitochondrial DNA, populations 1 to 3 being in majority and 4, while the other values are much lower (range: 0.052– from the NW lineage while populations 4 to 8 are in majority 0.0760, Figure 2(b)), suggesting the existence of a barrier to from the C lineage [18, 46]. However, the proportion of gene flow at the lineage/habitat boundary. The combined admixed individuals is very high and relatively constant all dataset shows a similar level of variation than in the habitat along this transect (between 35 and 56%), and there is transect, with a strong differentiation between coastal (1- no trend in the pairwise Fst’ that vary between 0.072 and 2) and montane (5–8) populations, with the populations 3 0.101 (Figure 2(a)). Morphological and genetic data analysed and 4 being intermediate. Along this transect, the population separately (Figure S1) show the same trend that is magnified specific Fst are similar or higher to what is observed ff in the combined analyses. We observe a slight di erentiation in the habitat transect (Table 1). The genetic structure is between populations 1 to 3 and populations 4 to 8, but significantly associated with the habitat type/lineage factor, there is an overlap between these groups (Figure 3(a)). The and the second best model includes the bioclimatic variable population specific Fst estimated with GESTE (Table 1) that with a nonnegligible probability (Table 2). represents the level of differentiation of one population relative to all other [41] are very low (ranging from 0.008 to 0.013, Table 1) and do not correlate with any of the 4. Discussion factors investigated (geographic connectivity, environment, ff lineage), underlining again the lack of genetic structure along In this paper, we describe the di erentiation between two this transect (Table 2). lineages of Anolis roquet living in contrasting habitats and coming into secondary contact at the ecotone between these 3.2. Habitat Transect. Two genetic clusters were also iden- habitats. As a comparison, we reanalysed two previously tified in this transect (cluster 1: populations 1–3; cluster 2: published transects, a “lineage transect” where two lineages populations 4–6, Figure 2(b)). This genetic division corre- meet within a same habitat, and a “habitat transect” where sponds to the habitat division, the ecotone being situated different populations of the same lineage live in contrasting between populations 3 and 4 [18]. The admixture rates are environments and are in contact at the ecotone between globally lower than on the lineage transect (range: 15–55%). these habitats. This design allowed to investigate the effects It is relatively low at the two opposite sides of the transect of allopatry, habitat, and of the combination of these two (29% in population 1 and 15% in population 6) and gets factors on anole population differentiation and speciation. higher in the centre of the transect, around the ecotone (55% Amarkeddifference could be observed in the population in population 3 and 45% in population 4). Here again, there structure along these three transects (Figure 5), suggesting International Journal of Ecology 5

50 0.2  40 0.15 30 0.1 20 0.05 10 0 Fst Pairwise

Admixed ind. (%) ind. Admixed 12345678 Population (a)

50 0.2  40 0.15 30 0.1 20 0.05 10 0 Fst Pairwise Admixed ind. (%) ind. Admixed 123456 Population (b)

50 0.2  40 0.15 30 0.1 20 0.05 10 0 Fst Pairwise Admixed ind. (%) ind. Admixed 12345678 Population (c)

Figure 2: Population structure and admixture rates along the three transects. Solid lines represent admixture rates, and broken lines represent pairwise Fst values.

4 4 4

2 2 2

0 0 0

−2 −2 −2

DAPC axis 1 DAPC −4 −4 −4

−6 −6 −6

−8 −8 −8

12345678 123456 12345678 Population Population Population (a) (b) (c)

Figure 3: DAPC (discriminant analysis of principal components) based on the combination of genetic and quantitative trait characters (see supplementary Figure S1 available online at doi: 10.1155/2012/273413 for the separate analyses). 6 International Journal of Ecology

Table 1: Bayesian estimates of Fst values for each population. The mean value and the lower and upper limits of the 95% highest probability density interval are indicated. These Fst values measure the diﬀerentiation of a given population relative to all other populations.

Population Lineage transect Habitat transect Combined transect 1 0.0078 [0.0038; 0.0123] 0.0522 [0.0357; 0.0695] 0.0721 [0.0524; 0.0934] 2 0.0104 [0.0055; 0.0155] 0.0464 [0.0301; 0.0627] 0.0513 [0.0357; 0.0674] 3 0.0054 [0.0019; 0.0093] 0.0330 [0.0209; 0.0468] 0.0507 [0.0350; 0.0671] 4 0.0097 [0.0049; 0.0151] 0.0148 [0.0071; 0.0228] 0.0148 [0.0075; 0.0223] 5 0.0061 [0.0022; 0.0101] 0.0219 [0.0122; 0.0323] 0.0131 [0.0064; 0.0206] 6 0.0132 [0.0075; 0.0193] 0.0233 [0.0142; 0.0339] 0.0134 [0.0071; 0.0205] 7 0.0090 [0.0044; 0.0139] 0.0122 [0.0060; 0.0190] 8 0.0101 [0.0052; 0.0153] 0.0136 [0.0066; 0.0210]

Table 2: Environmental factors determining the genetic structure of populations. (a) Sum of posterior probabilities of models including a given factor. (b) Posterior probability of the 8 models considered. The best model is indicated in bold. (c) Estimates of the regression parameters for the best model (mean value and lower and upper limits of the 95% highest probability density interval are indicated).

Lineage transect Habitat transect Combined transect (a) Connectivity (G1) 0.069 0.107 0.081 Environment (G2) 0.070 0.184 0.458 Lineage/habitat (G3) 0.079 0.228 0.572 (b) Constant 0.800 0.554 0.068 Constant, G1 0.057 0.067 0.012 Constant, G2 0.058 0.134 0.319 Constant, G3 0.006 0.175 0.431 Constant, G1, G2 0.067 0.017 0.029 Constant, G1, G3 0.006 0.020 0.031 Constant, G2, G3 0.006 0.030 0.100 Constant, G1, G2, G3 0.001 0.003 0.010 (c) α0 −3.80 [−4.28; −3.34] −3.49 [−4.17; −2.78] −3.77 [−4.24; −3.24] α1 0.75 [−1.22; −0.29] σ20.39 [0.10; 0.84] 0.70 [0.13; 1.63] 0.40 [0.10; 0.89] that they are at different stages of the speciation continuum present strongly discontinuous variation, but there is still a [49, 50]. According to Hendry et. al [49], four stages can significant level of admixture at the contact zone suggesting be distinguished along this continuum: “(1) continuous that the reproductive isolation is not complete despite a variation within panmictic populations, (2) partially dis- strong reduction in gene flow. This would correspond to continuous variation with minor reproductive isolation, (3) State 3 of the speciation continuum. The situation of this strongly discontinuous variation with strong but reversible last transect, where the ecotone and the lineage boundary reproductive isolation and (4) complete and irreversible overlap, is unique on this island, and hence this result could reproductive isolation.” The weak structure observed in the not be replicated. lineage transect, and the high admixture level correspond In terms of association between observed genetic struc- the State 1 of the continuum. This pattern is similar to ture and biotic and abiotic factors, no significant factors were what was observed in a previous study [18]forallbut detected on the lineage transect, all the models other than one (transect I in [18]) transects sampled across lineages the one incorporating only a constant having a very low (transects II, III, IV, V, VI, VII, VIII in [18]). In the probability. To the contrary, for the combined transect both habitat transect, the variation is clearly bimodal, but the the habitat/lineage category and the bioclimatic composite high level of admixture in the contact zone between the variable are the best at explaining the genetic structure, two habitats suggests that the reproductive isolation is still suggesting the important driving force of the environment limited between the ecotypes, corresponding to a stage into population differentiation. The situation in the habitat between State 2 and State 3 of the continuum. Here again, transect is not so clear. The best model only incorporates a this pattern is similar to the other transect sampled across constant, but the combined probability of the habitat type habitats (transect III in [18]). Finally, the combined transects and the environment factor are not negligible as they were on International Journal of Ecology 7

1 in accordance with an extreme environment gradient, the divergence we demonstrate in this study with these neutral 0.8 markers is undoubtedly considerably less than the divergence of the traits and loci under selection. 0.6 Since the habitat along the lineage transect is very homogeneous we do not expect ecological speciation to currently play a role in this area. However, this does not mean 0.4 it has always been the case. When the populations from the two lineages were isolated on different proto-islands, 0.2 it is possible that they were submitted to different envi- Proportion ofProportion NW individuals ronmental pressures. In such a situation, several studies 0 have demonstrated that isolated populations can rapidly 12345678 evolve partial reproductive isolation as a byproduct of local Population adaptation [57–60]. Such a mechanism could also explain Figure 4: Proportion of individuals from the NW (black) and C the differences observed between the three transects. When (white) mitochondrial lineages along the combined transect. populations from the C and NW lineages came back into secondary contact, they did so in two very different contexts: either along a very sharp environmental gradient (combined 0.4 transect), where any preexisting reproductive isolation could be maintained or strengthened by current disruptive selection, or within an homogeneous habitat (lineage transect) where any preexisting reproductive isolation may have been

 0.3 lost. Breakdowns of reproductive barriers associated with ecological changes have been recently described in various species [61, 62]. Therelativeroleofgeographyandecologyinspeciation Pairwise Fst Pairwise 0.2 remains a subject of debate. The main discussion relates to whether or not ecological factors can drive speciation in the presence of gene flow. Several convincing empirical studies suggest that it is indeed possible to reach full 0.1 speciation in sympatry by ecological speciation (e.g., [63– 67]), while others emphasize the combined role of historical Combined Habitat Lineage and ecological factors in shaping species diversity [68–70]. Transect For instance, in Trinidadian guppies, ecological speciation played a role in premating isolation either in allopatry Figure 5: Box plot of the pairwise Fst’ values between populations (byproduct speciation) or in parapatry (to avoid maladaptive across the lineage and/or habitat contact zone for the three tran- matings) [69]. For Martinique anoles, it is clear that for sects. the populations that reached secondary contact very little evidence of the role of geographic isolation exists, while ecological factors seem to play a more important role. the lineage transect. The lack of significance of these factors However, without conducting mate-choice experiments, it is could be due to the lack of power of this method when the not possible to determine conclusively the role of ecological number of populations analysed is low. Indeed, simulations speciation in allopatry. showed that this method failed to identify the true model Despite the large number of studies on the Caribbean with five or seven populations [41]. However, the observed anole radiation, very little is known about the factors at trend suggests that environment factors may play a role in the origin of their diversity. Several papers have described the genetic structuring of the populations along this transect the diversity patterns, demonstrating a correlation between too. island size and species diversity in large islands, while no As expected, we find the strongest signal of divergence speciation events were recorded on islands below a threshold in morphological data that are known to react quickly to size [13]. Recent work demonstrated that within island selection [51–54] (Figure S1). The genetic data is similar in diversity could be the result of ecological opportunities and pattern but smaller in magnitude. Recent simulation studies that net speciation rate decreased with time as opportunities suggest that neutral markers, that is, markers that are not decreased to reach an equilibrium at the island carrying linked with the traits under selection, are not very sensitive capacity [15]. However, these studies do not explain the to detect ecological speciation [55, 56]. The authors conclude mechanisms at the origin of these diversity patterns and only that this would lead to false negatives (failure to detect propose several hypotheses to explain these observations. ecological speciation) rather than false positives. Taking this In line with previous studies, we demonstrate that indeed into account, and the fact that we found a clear signal of gene environmental factors have a strong effect on the genetic flow reduction both in the habitat and the combined transect structure of Martinique anole populations, with a reduction 8 International Journal of Ecology of gene flow at the ecotone between coastal and montane [13] J. B. Losos, Lizards in an Evolutionary Tree, University of habitats and that it does not seem sufficienttoleadtofull California Press, Berkeley, Calif, USA, 2009. reproductive isolation. Furthermore, we refine our under- [14] J. B. Losos and D. Schluter, “Analysis of an evolutionary standing of divergence in these anoles by demonstrating species-area relationship,” Nature, vol. 408, no. 6814, pp. 847– that when allopatric lineages come into secondary contact 850, 2000. on an ecotone, the differentiation is much stronger, with a [15] D. L. Rabosky and R. E. Glor, “Equilibrium speciation dynamics in a model adaptive radiation of island lizards,” significant reduction in gene flow. The absence of replication Proceedings of the National Academy of Sciences of the United of the combined transect does not allow the generalisation States of America, vol. 107, no. 51, pp. 22178–22183, 2010. of these findings, but we hypothesize that to reach full [16] D. Westercamp, P. Andreieff,P.Bouysse,S.Cottez,andR. speciation anoles need first to evolve in allopatry and then Battistini, Notice Explicative, Carte Géol. France (1/50 000), come into secondary contact in an area where divergent Feuille MARTINIQUE, Bureau de Recherches Geologiques´ et natural selection will allow them to stay separate and further Minieres,` Orleans,´ France, 1989. reinforce their divergence. Because this combination of [17] R. S. Thorpe and A. G. Stenson, “Phylogeny, paraphyly and factors is more likely to be found on large islands, it could be ecological adaptation of the colour and pattern in the Anolis the mechanism at the origin of the species-area relationship roquet complex on Martinique,” Molecular Ecology, vol. 12, no. observed in Caribbean anoles. 1, pp. 117–132, 2003. [18] R. S. Thorpe, Y. Surget-Groba, and H. Johansson, “Genetic tests for ecological and allopatric speciation in Anoles on an Acknowledgments Island archipelago,” PLoS Genetics, vol. 6, no. 4, Article ID e1000929, 2010. The authors wish to thank the DIREN Martinique for pro- [19] A. G. Stenson, R. S. Thorpe, and A. Malhotra, “Evolutionary viding permits to collect samples, and W. Grail for assistance differentiation of bimaculatus group anoles based on analyses with molecular work. This work was funded by a Marie Curie of mtDNA and microsatellite data,” Molecular Phylogenetics Intra-European Fellowship to Y. Surget-Groba, a BBSRC and Evolution, vol. 32, no. 1, pp. 1–10, 2004. grant to R. S. Thorpe, and a NERC studentship to H. [20] R. S. Thorpe, A. G. Jones, A. Malhotra, and Y. Surget-Groba, “Adaptive radiation in Lesser Antillean lizards: molecular Johansson. phylogenetics and species recognition in the Lesser Antillean dwarf gecko complex, Sphaerodactylus fantasticus,” Molecular References Ecology, vol. 17, no. 6, pp. 1489–1504, 2008. [21] R. S. Thorpe and A. Malhotra, “Molecular and morphological [1] Y. Kisel and T. G. Timothy, “Speciation has a spatial scale that evolution within small islands,” Philosophical Transactions of depends on levels of gene flow,” American Naturalist, vol. 175, the Royal Society B: Biological Sciences, vol. 351, no. 1341, pp. no. 3, pp. 316–334, 2010. 815–822, 1996. [2] E. Mayr, Systematics and the Origin of Species, Columbia [22] R. S. Thorpe, Y. Surget-Groba, and H. Johansson, “The relative University Press, New York, NY, USA, 1942. importance of ecology and geographic isolation for speciation in anoles,” Philosophical Transactions of the Royal Society B: [3] E. Mayr, Animal Species and Speciation,HarvardUniversity Biological Sciences, vol. 363, no. 1506, pp. 3071–3081, 2008. Press, Cambridge, Mass, USA, 1963. [23] R. S. Thorpe, “Analysis of color spectra in comparative evolu- [4] M. Nei, T. Maruyama, and C. Wu, “Models of evolution of tionary studies: molecular phylogeny and habitat adaptation reproductive isolation,” Genetics, vol. 103, no. 3, pp. 557–579, in the St. Vincent anole (Anolis trinitatis),” Systematic Biology, 1983. vol. 51, no. 4, pp. 554–569, 2002. [5] E. Mayr, “Ecological factors in speciation,” Evolution, vol. 1, [24] T. Jezkova, M. Leal, and J. A. Rodr´ıguez-Robles, “Living pp. 263–288, 1947. together but remaining apart: comparative phylogeography [6] T. M. Panhuis, R. Butlin, M. Zuk, and T. Tregenza, “Sexual of Anolis poncensis and A. cooki, two lizards endemic to the selection and speciation,” Trends in Ecology and Evolution, vol. aridlands of Puerto Rico,” Biological Journal of the Linnean 16, no. 7, pp. 364–371, 2001. Society, vol. 96, no. 3, pp. 617–634, 2009. [7] P. Nosil and S. M. Flaxman, “Conditions for mutation- [25] J. A. RodrIguez-Robles,´ T. Jezkova, and M. Leal, “Climatic order speciation,” Proceedings of the Royal Society B: Biological stability and genetic divergence in the tropical insular lizard Sciences, vol. 278, no. 1704, pp. 399–407, 2011. Anolis krugi, the Puerto Rican “Lagartijo Jardinero de la [8]J.A.CoyneandH.A.Orr,Speciation, Sinauer Associates, Montana”,”˜ Molecular Ecology, vol. 19, no. 9, pp. 1860–1876, Sunderland, Mass, USA, 2004. 2010. [9] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology [26] T. R. Jackman, D. J. Irschick, K. De Queiroz, J. B. Losos, and Letters, vol. 8, no. 3, pp. 336–352, 2005. A. Larson, “Molecular phylogenetic perspective on evolution [10] G. F. Turner and M. T. Burrows, “A model of sympatric of lizards of the Anolis grahami series,” Journal of Experimental speciation by sexual selection,” Proceedings of the Royal Society Zoology, vol. 294, no. 1, pp. 1–16, 2002. B: Biological Sciences, vol. 260, no. 1359, pp. 287–292, 1995. [27] R. E. Glor, J. J. Kolbe, R. Powell, A. Larson, and J. B. Lossos, [11] J. Felsenstein, “Skepticism towards Santa Rosalia, or why are “Phylogenetic analysis of ecological and morphological diver- there so few kinds of animals,” Evolution, vol. 35, pp. 124–138, sification in hispaniolan trunk-ground anoles (Anolis cybotes 1981. group),” Evolution, vol. 57, no. 10, pp. 2383–2397, 2003. [12]M.R.Servedio,G.S.V.Doorn,M.Kopp,A.M.Frame,and [28]R.E.Glor,J.B.Losos,andA.Larson,“OutofCuba: P. Nosil, “Magic traits in speciation: “magic” but not rare?” overwater dispersal and speciation among lizards in the Anolis Trends in Ecology and Evolution, vol. 26, no. 8, pp. 389–397, carolinensis subgroup,” Molecular Ecology,vol.14,no.8,pp. 2011. 2419–2432, 2005. International Journal of Ecology 9

[29]J.J.Kolbe,R.E.Glor,L.R.Schettino,A.C.Lara,A.Larson, [46] H. Johansson, Y. Surget-Groba, and R. S. Thorpe, “The roles and J. B. Losos, “Genetic variation increases during biological of allopatric divergence and natural selection in quantitative invasion by a Cuban lizard,” Nature, vol. 431, no. 7005, pp. trait variation across a secondary contact zone in the lizard 177–181, 2004. Anolis roquet,” Molecular Ecology, vol. 17, no. 23, pp. 5146– [30] J. A. Rodr´ıguez-Robles, T. Jezkova, and M. A. Garc´ıa, “Evo- 5156, 2008. lutionary relationships and historical biogeography of Anolis [47] T. Jombart, “Adegenet: a R package for the multivariate desechensis and Anolis monensis, two lizards endemic to small analysis of genetic markers,” Bioinformatics, vol. 24, no. 11, pp. islands in the eastern Caribbean Sea,” Journal of Biogeography, 1403–1405, 2008. vol. 34, no. 9, pp. 1546–1558, 2007. [48] W. Venables and B. Ripley, Modern Applied Statistics with S, [31]R.E.Glor,M.E.Gifford, A. Larson et al., “Partial island Springer, New York, NY, USA, 2002. submergence and speciation in an adaptive radiation: a [49] A. P. Hendry, D. I. Bolnick, D. Berner, and C. L. Peichel, multilocus analysis of the Cuban green anoles,” Proceedings of “Along the speciation continuum in sticklebacks,” Journal of the Royal Society B: Biological Sciences, vol. 271, no. 1554, pp. Fish Biology, vol. 75, no. 8, pp. 2000–2036, 2009. 2257–2265, 2004. [50] P. Nosil, L. J. Harmon, and O. Seehausen, “Ecological expla- [32] A. G. Stenson, A. Malhotra, and R. S. Thorpe, “Population nations for (incomplete) speciation,” Trends in Ecology and ff di erentiation and nuclear gene flow in the Dominican anole Evolution, vol. 24, no. 3, pp. 145–156, 2009. (Anolis oculatus),” Molecular Ecology, vol. 11, no. 9, pp. 1679– [51] A. Malhotra and R. S. Thorpe, “Experimental detection of 1688, 2002. rapid evolutionary response in natural lizard populations,” ff [33] J. Ng and R. E. Glor, “Genetic di erentiation among popula- Nature, vol. 353, no. 6342, pp. 347–351, 1991. tions of a Hispaniolan trunk anole that exhibit geographical [52] M. Leal and L. J. Fleishman, “Differences in visual signal variation in dewlap colour,” Molecular Ecology, vol. 20, no. 20, design and detectability between allopatric populations of pp. 4302–4317, 2011. Anolis lizards,” American Naturalist, vol. 163, no. 1, pp. 26–39, [34] J. L. Gow, H. Johansson, Y. Surget-Groba, and R. S. Thorpe, 2004. “Ten polymorphic tetranucleotide microsatellite markers iso- [53] R. S. Thorpe, J. T. Reardon, and A. Malhotra, “Common lated from the Anolis roquet series of Caribbean lizards,” garden and natural selection experiments support ecotypic Molecular Ecology Notes, vol. 6, no. 3, pp. 873–876, 2006. ff ffi di erentiation in the Dominican anole (Anolis oculatus),” [35] R. Ogden, T. J. Gri ths, and R. S. Thorpe, “Eight microsatel- American Naturalist, vol. 165, no. 4, pp. 495–504, 2005. lite loci in the Caribbean lizard, Anolis roquet,” Conservation [54] R. Calsbeek, J. H. Knouft, and T. B. Smith, “Variation in scale Genetics, vol. 3, no. 3, pp. 345–346, 2002. numbers is consistent with ecologically based natural selection [36] H. Johansson, Y. Surget-Groba, J. L. Gow, and R. S. Thorpe, acting within and between lizard species,” Evolutionary Ecol- “Development of microsatellite markers in the St Lucia anole, ogy, vol. 20, no. 4, pp. 377–394, 2006. Anolis luciae,” Molecular Ecology Resources,vol.8,no.6,pp. [55] X. Thibert-Plante and A. P.Hendry, “Five questions on ecolog- 1408–1410, 2008. ical speciation addressed with individual-based simulations,” [37] J. K. Pritchard, M. Stephens, and P. Donnelly, “Inference Journal of Evolutionary Biology, vol. 22, no. 1, pp. 109–123, of population structure using multilocus genotype data,” 2009. Genetics, vol. 155, no. 2, pp. 945–959, 2000. [38] G. Evanno, S. Regnaut, and J. Goudet, “Detecting the number [56] X. Thibert-Plante and A. P. Hendry, “When can ecological of clusters of individuals using the software STRUCTURE: a speciation be detected with neutral loci?” Molecular Ecology, simulation study,” Molecular Ecology, vol. 14, no. 8, pp. 2611– vol. 19, no. 11, pp. 2301–2314, 2010. 2620, 2005. [57] D. J. Funk, “Isolating a role for natural selection in speci- [39] P. G. Meirmans, “Using the amova framework to estimate a ation: host adaptation and sexual isolation in Neochlamisus standardized genetic differentiation measure,” Evolution, vol. bebbianae leaf beetles,” Evolution, vol. 52, no. 6, pp. 1744–1759, 60, no. 11, pp. 2399–2402, 2006. 1998. [40] J. Goudet, “FSTAT (Version 1.2): a computer program to [58] T. H. Vines and D. Schluter, “Strong assortative mating calculate F-statistics,” Journal of Heredity, vol. 86, pp. 485–486, between allopatric sticklebacks as a by-product of adaptation ff 1995. to di erent environments,” Proceedings of the Royal Society B: [41] M. Foll and O. Gaggiotti, “Identifying the environmental Biological Sciences, vol. 273, no. 1589, pp. 911–916, 2006. factors that determine the genetic structure of populations,” [59] R. B. Langerhans, M. E. Gifford, and E. O. Joseph, “Ecological Genetics, vol. 174, no. 2, pp. 875–891, 2006. speciation in Gambusia fishes,” Evolution,vol.61,no.9,pp. [42] J. Thioulouse, D. Chessel, S. Doledec,´ and J. M. Olivier, “ADE- 2056–2074, 2007. 4: a multivariate analysis and graphical display software,” [60]A.P.Hendry,P.Nosil,andL.H.Rieseberg,“Thespeedof Statistics and Computing, vol. 7, no. 1, pp. 75–83, 1997. ecological speciation,” Functional Ecology,vol.21,no.3,pp. [43] R Development Core Team, R: A Language and Environment 455–464, 2007. for Statistical Computing, R Foundation for Statistical Com- [61] E. B. Taylor, J. W. Boughman, M. Groenenboom, M. Sni- puting, Vienna, Austria, 2011. atynski, D. Schluter, and J. L. Gow, “Speciation in reverse: [44] O. E. Gaggiotti, D. Bekkevold, H. B. H. Jørgensen et al., morphological and genetic evidence of the collapse of a “Disentangling the effects of evolutionary, demographic, and three-spined stickleback (Gasterosteus aculeatus) species pair,” environmental factors influencing genetic structure of natural Molecular Ecology, vol. 15, no. 2, pp. 343–355, 2006. populations: atlantic herring as a case study,” Evolution, vol. [62]O.Seehausen,G.Takimoto,D.Roy,andJ.Jokela,“Speciation 63, no. 11, pp. 2939–2951, 2009. reversal and biodiversity dynamics with hybridization in [45] T. Jombart, S. Devillard, and F. Balloux, “Discriminant anal- changing environments,” Molecular Ecology, vol. 17, no. 1, pp. ysis of principal components: a new method for the analysis 30–44, 2008. of genetically structured populations,” BMC Genetics, vol. 11, [63] E. Rolan-Alvarez, M. Carballo, J. Galindo et al., “Nonallopatric article 94, 2010. and parallel origin of local reproductive barriers between two 10 International Journal of Ecology

snail ecotypes,” Molecular Ecology, vol. 13, no. 11, pp. 3415– 3424, 2004. [64] V. Savolainen, M. C. Anstett, C. Lexer et al., “Sympatric speciation in palms on an oceanic island,” Nature, vol. 441, no. 7090, pp. 210–213, 2006. [65] M. Barluenga, K. N. Stolting,¨ W. Salzburger, M. Muschick, and A. Meyer, “Sympatric speciation in Nicaraguan crater lake cichlid fish,” Nature, vol. 439, no. 7077, pp. 719–723, 2006. [66] A. S.T. Papadopulos, W. J. Baker, D. Crayn et al., “Speciation with gene flow on Lord Howe Island,” Proceedings of the National Academy of Sciences of the United States of America, vol. 108, no. 32, pp. 13188–13193, 2011. [67] E. B. Rosenblum and L. J. Harmon, ““same same but different”: replicated ecological speciation at white sands,” Evolu- tion, vol. 65, no. 4, pp. 946–960, 2011. [68] P. Nosil, “Ernst Mayr and the integration of geographic and ecological factors in speciation,” Biological Journal of the Linnean Society, vol. 95, no. 1, pp. 26–46, 2008. [69] A. K. Schwartz, D. J. Weese, P. Bentzen, M. T. Kinnison, and A. P. Hendry, “Both geography and ecology contribute to mating isolation in guppies,” PLoS ONE, vol. 5, no. 12, Article ID e15659, 2010. [70] M.A.Frey,“Therelativeimportanceofgeographyandecology in species diversification: evidence from a tropical marine intertidal snail (Nerita),” Journal of Biogeography, vol. 37, no. 8, pp. 1515–1528, 2010. Hindawi Publishing Corporation International Journal of Ecology Volume 2012, Article ID 506957, 8 pages doi:10.1155/2012/506957

Review Article Of “Host Forms” and Host Races: Terminological Issues in Ecological Speciation

Daniel J. Funk

Department of Biological Sciences, Vanderbilt University, Nashville, TN 37235-1634, USA

Correspondence should be addressed to Daniel J. Funk, [email protected]

Received 31 July 2011; Accepted 26 October 2011

Academic Editor: Zachariah Gompert

Copyright © 2012 Daniel J. Funk. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Successful communication and accurate inferences in science depend on the common understanding and consistent usage of the terms we apply to concepts of interest. Likewise, new terminology is required when important concepts have gone unnamed. Here, I focus on terminological issues about biological variation and ecological speciation, especially in herbivorous insects but also more generally. I call for the more restricted use of concepts that have sometimes been misapplied, and thus caution against synonymizing ecological speciation with sympatric speciation and the unwarranted invocation of “host races” to describe herbivorous insect differentiation. I also call for the qualified application of terms for different kinds of biological variation and for host range when confronting uncertainty. Among other “missing terms” introduced here is “host form,” a generic term describing any case of host-associated differences for which current evidence does not allow diagnosis of the specific kind of variation. Embracing the use of host form should free host race from its current overapplication. Finally, I present a case study in which Neochlamisus leaf beetle populations previously described as host forms are hereby declared to be host races, based on accumulated evidence supporting each of the associated criteria.

1. Introduction what organism two people might be discussing, even if they differed in nationality or native tongue. Arosebyanyothernamemayindeedsmellassweet.How- The Linnaean system of binomials continues to be use- ever, if a rose vendor started advertising his product as “rag- fully applied today. However, the typological view of species weed,” he would find himself selling fewer flowers. And if a on which it was based primarily retains its utility for the biologist unfamiliar with ragweed purchased his wares for taxonomist. Following Darwin’s Origin [4], the typologist’s analysis, they would come to different conclusions than had notion of species as largely invariant and discretely differ- they collected actual ragweed in the field. Words, after all, ent entities gave way to a population-oriented view that are imbued with the meanings we assign to them. In science, embraced continuous intraspecific variation and invoked effective communication entails that terms have well-defined reproductive barriers to genetic exchange to explain how and consistently used meanings and that new terms be species remained differentiated from each other [5]. This created as needed to describe unlabelled concepts. In a field “population thinking” and the associated biological species as concept-driven as evolutionary biology [1, 2], maintaining concept promoted the recognition of various kinds of a one-to-one relationship between terms and meanings is biological variation within and around the species level and especially critical. Indeed, our rose merchant’s application the consideration of how they relate to species diagnosis and of a different “common name” to his roses need not have the process of speciation [5]. For example, the entomologist confused our biologist if their scientific name was also on Benjamin Walsh recognized quite early that causal relation- display. This is because Linnaeus’s stratagem of applying a ships likely existed among the host plant associations of her- unique Latin binomial to every species [3]providedacom- bivorous insects, their patterns of phenotypic variation, their mon and specific language that eliminated confusion about opportunities for interbreeding, and their species status. 2 International Journal of Ecology

He thus spoke of “phytophagic varieties” and “phytophagic and comparative evidence for the association of ecological species” [6]. In so doing, Walsh began a long tradition of traits with reproductive isolation accumulated, ecological using herbivorous insects as exemplars of different kinds of speciation gained support as an evolutionarily important biological variation [7–10] and for the study of speciation [2, process and has induced little controversy [18, 19] but see 22, and 34]. Most recently, insect herbivores have provided [20]. many of the best-documented examples of what has become Critically, ecological speciation is not dependent on geo- known as ecological speciation [11, 12]. graphic context. So long as reproductive isolation is evolving Here I address outstanding terminological issues that as a consequence of divergent adaptation, ecologically speci- affect our understanding of biological variation and ecolog- ating populations could be completely allopatric on separate ical speciation and that are currently in need of attention. islands or completely sympatric, living on different plant In illustrating such issues I focus largely on herbivorous species intermingling in the same field. Similarly, sympatric insects for the reasons given above. That said, I additionally speciation may depend on ecological differentiation, or introduce new terms that are applicable to all taxa. A major alternatively on sexual selection or polyploidy [21]. Thus, goal is to demonstrate that, far from being merely semantic ecological speciation and sympatric speciation incompletely in nature, terminological shortcomings have important overlap in the cases of speciation they circumscribe [22]. consequences for how we view central issues in speciation. I Nonetheless, the intermingling plant scenario exemplifies thus further call for the careful and specified use and creation the reliance of many models of sympatric speciation on the of terms. I do not, however, chronicle and count up cases ecological divergence of newly formed subpopulations. And of terminological misuse in the literature to support my if indeed sympatric speciation is rarer than allopatric spe- assertions of their existence. I do not wish to unfairly single ciation and most commonly involves ecological divergence, out particular authors for misnomers common to others. then most cases of sympatric speciation will also represent And I do not believe that a more systematic evaluation of ecological speciation, while only a minority of ecological misuse frequencies is necessary to argue the value of a more speciation scenarios will involve sympatric speciation. stringent use of terms. Rather I hope that by considering Despite this asymmetry, ecological speciation is some- the points made here the reader will develop a more critical times equated (conflated) with sympatric speciation in eye in reading and writing the associated literature, to the the literature. Sometimes an association between ecological betterment of speciation studies. divergence and reproductive isolation is taken as de facto evidence for sympatric speciation in the absence of support for 2. Ecological Speciation =/ Sympatric Speciation sympatric origins. This presumably occurs because the commonly ecological basis of sympatric speciation has forged a To begin let us consider a recurring misuse of “ecological false equivalency between ecological causes of reproductive speciation” itself. First I give some background. isolation and sympatric speciation in the minds of some For much of the 20th century a principal focus of spe- workers. In the extreme, the term ecological speciation itself ciation studies was determining the geographic context in is sometimes coopted as a synonym for sympatric speciation which speciation occurred and the relative efficacy of dif- and even erroneously contrasted with allopatric speciation. ferent contexts for facilitating this process [13]. An issue of This terminological misuse is important as it contributes special concern was how readily populations could speciate to misunderstanding about the relative frequency of specia- in the absence of geographic isolation (allopatry), with the tion mechanisms. Ecological speciation per se is theoretically view becoming dominant that nonallopatric, “speciation simple and straightforward, and abundant empirical evi- with gene flow,” modes were difficult and rare [13]. Sym- dence now supports it [18], while conditions for sympatric patric speciation received special scrutiny as it required that speciation are theoretically restrictive and evidence com- reproductively isolated populations evolve from within an pellingly supporting it—regardless of how common it may initially panmictic population in the face of opportunities actually be—is difficult to obtain [14]. Thus, conceptually for homogenization by the countervailing forces of gene flow equating these terms makes sympatric speciation appear to and recombination. Thus, while enthusiasm for sympatric be more broadly supported than it actually is. speciation has continued to wax and wane, it is generally accepted to be a theoretically more onerous mode than 3. The Kinds of Biological Variation allopatric speciation and thus to require quite particular conditions for its occurrence [14, 15]. Speciation is a population-level process with many stages, Toward the end of the 20th century, debates about geo- and species are heterogeneous entities. Thus, their study graphic modes had to make room for an emerging interest requires a lexicon of terms describing variation at the indi- in the causes of reproductive isolation per se, and especially vidual, population, and species levels, one that is sufficient in the potential contributions of ecological adaptation and to describe any case of biological variation. In herbivorous sexual selection. The term “ecological speciation” was coined insects, these kinds of variation are often host specific. They to describe the evolution of reproductive isolation as an may also be cryptic, as host adaptation may be behaviorally incidental consequence of divergent adaptation to alternative or physiologically rather than morphologically based. Not environments, due to the pleiotropic effects of selected genes, for nothing were herbivorous insects focal subjects for early or the direct effects of genes tightly linked to them, on traits discussions of sibling species [7, 8] and later evaluations involved in reproductive isolation [16, 17]. As experimental of intraspecific ecological heterogeneity [23]. These many International Journal of Ecology 3 sources of variation led to terminological confusion and a induced differences between conspecific herbivores as a func- need for clear distinctions [10]. The following list of terms is tion of the different host plants on which they developed. a bit entomocentric and is not comprehensive (missing, e.g., various infraspecific botanical terms), but should nonethe- Morphs (sensu [13]). Discrete phenotypic variants that seg- less allow for the assignment of differences between two regate within a population. organisms of any taxon to one or more kinds of biological variation. Polymorphisms (sensu [27]). Multiple genetic variants, often segregating within a population (includes morphs). Species. A contentious term with many suggested definitions (i.e., concepts; see [21]). These generally describe a set of Biotypes (sensu [10]). The set of herbivorous insect individ- populations that is reproductively, historically, genetically, uals that share the same genotype at given loci of interest and and/or ecologically distinct from other such populations. thus phenotypically differ from other biotypes. Most often The biological species concept, whereby species are groups used in reference to different insect herbivore genotypes, of actually or potentially interbreeding populations that are each adapted to use a different host plant genotype. reproductively isolated from other such groups, remains the most frequently adopted definition by evolutionary biolo- Sexual dimorphism . Phenotypic differentiation between the gists. sexes within a population. Sibling species (sensu [13]) = cryptic species. Biological spe- Developmental variation . Differences in phenotype reflect- cies that are approximately anatomically identical to each ing different stages of individual development or of a com- other. Herbivore sibling species are commonly associated plex life cycle. with different host plants. Host forms. A “missing term” (coined here) for groups of Hybrids (sensu [24]). The products of interbreeding between herbivore individuals or populations exhibiting host-associ- differentiated populations or separate species. ated biological variation, but where the kind of variation has not yet been diagnosed. Geographic races = subspecies (cf. [24]). Geographically distinct populations exhibiting genetically based phenotypic Ecological forms. A “missing term” (coined here) completely differences. Subspecies are essentially geographic races with analogous to host forms but appropriately applied to mem- formal taxonomic status. bers of any taxon—namely, nonparasitic ones—that exhibit habitat-or resource-associated differentiation. Ecotypes (sensu [25]) (related to the above). Spatially distinct populations exhibiting divergent adaptation to alternative Excepting host races, discussed below, means of diagnos- environments. Herbivores adapted to different host plants in ing these kinds of variation are not described here. different areas are an example. 4. Nominal Host Ranges Hostraces(sensu[26]). Sympatric populations that are incompletely reproductively isolated but remain ecologically For insect herbivores another sort of variation is critical for differentiated in the face of gene flow due to divergent se- understanding ecological speciation: host range, that is, the lection on populations using alternative hosts. Generally ap- number, identity, and relatedness of host plant taxa [28]. plied to herbivores, but sometimes to other parasites. The Variation in host range is described by its own suite of terms. diagnosis of host races is discussed below. A “narrow” host range includes few and/or closely related hosts, while a “broad” host range includes many and/or distantly related host taxa. The narrower its host range, the Sympatric races. A “missing term” (coined here) completely more “specialized” a herbivore is said to be, with insects us- analogoustohostracesbutappliedtopopulationsofany ing a single host taxon being described as “monophagous.” taxon—namely, nonparasitic ones—using analogous crite- The broader the host range, the more “generalized” or “pol- ria. yphagous” the herbivore is said to be. Herbivores using several host taxa are referred to as “oligophagous.” Populations (sensu [13]) . Ideally, panmictic groups of indi- An accurate understanding of a herbivore’s host range viduals. More commonly used to refer to spatially coherent is crucial to studying host-associated ecological speciation. groups of potentially interbreeding individuals that exhibit Notably, most such examples involve highly specialized pop- reduced migration and thus potential genetic differentiation. ulations that are divergently adapted to alternative host plant species [11, 12]. Thus, one might predict a greater Envirotypes. A “missing term” (coined here) referring to tendency towards specialization in a herbivore group to populations or individuals that differ in phenotype due to be associated with a greater tendency towards ecological differences in environmental rather than genetic factors, that speciation, via host shift and subsequent adaptation to the is, reflecting phenotypic plasticity. Examples are provided by new host. One might also predict a herbivore taxon whose 4 International Journal of Ecology species tend to specialize on distantly related host plants sufficientevidencetoruleoutalternativesrisksanerroneous to be more prone to ecological speciation than one whose diagnosis while compromising inferences premised on the species all use closely related hosts. This is because host shifts kind of variation at hand. For example, if different envi- between more biologically divergent hosts will likely incur rotypes are instead presumed to be host races or separate more strongly divergent selection, resulting in concomitantly species, associated conclusions about speciation processes large increases in reproductive isolation [29]. Neochlamisus will be erroneous. As in the case of host range, a problem that leaf beetles, my focal study system, provide an example promotes such misinterpretations is a current vocabulary of both of these tendencies. Species in this genus usually that is insufficient to describe uncertainties. And a lack of specialize on a single host genus or species, with these hosts terms indicating ambiguity may encourage workers to be less broadly scattered across eudicot phylogeny [29]. Following circumspect and make a best guess as to the kind of variation the above predictions, the only Neochlamisus species that at hand in the absence of sufficient evidence. As suggested for regularly uses multiple (six) different and distantly related host range, one could simply qualify uncertain assignments host genera/species, N. bebbianae, exhibits pronounced host- as “nominal.” A more conservative approach might, however, associated ecological divergence and reproductive isolation be advantageous, especially in the frequent situations where between populations affiliated with different hosts [29]. In variation along host plant lines has been documented, but this example, what was described as an oligophagous species current evidence is limited and remains consistent with [30] was experimentally proven not to be composed of pop- multiple kinds of variation. The term “biotype” used to be ulations with similar oligophagous tendencies, but rather of rather haphazardly applied to diverse kinds of variation [10]. a complex of divergently host-specialized populations, each But this owed to a lack of consensus on the meaning of that accepting its native host (i.e., the one used in nature) more term rather than to a conscious effort to indicate uncertainty. readily than the five plants used by other populations [31]. In recent years, workers have begun writing about “host- ThecaseofN. bebbianae not only illustrates why accu- associated differentiation (HAD),” a more general and rately assessing host range is vital to understanding ecological somewhat variably applied term recently defined by Dickey speciation but also why many herbivore host ranges are likely and Medina [34] as, “the formation of genetically divergent to be inaccurately or incompletely documented. Identifying host-associated sub-populations.” However, this term is not the particular suite of hosts used by a given species or popu- broad enough to span all kinds of biological variation and lation requires intensive fieldwork. This includes the inspec- describes the relationship between populations rather than tion of presumed nonhosts cooccurring with documented the populations themselves. hosts, to determine if host range is broader than presumed. It It seems then that the field is missing what would be further requires evidence that a “documented” host is indeed a valuable term, one that allowed workers to acknowledge actively used as resource rather than simply being a place the existence of host-associated variation, while remaining of temporary rest by an insect “tourist,” in which case the neutral on the kind of variation it represents. The creation actual host range is narrower than presumed. Moreover, as of such a term would eliminate the necessity of guesswork with N. bebbianae, experimental work is needed to identify while recognizing a phenomenon of interest and pointing it possible cryptic variation in host use, that is, whether a out as needing additional study. nominal species is actually subdivided into a complex of Thus, host forms = groups of individuals or populations populations or sibling species with divergent host ranges. exhibiting host-associated biological variation, but where the Thus, determining host range requires determining the kind kind of variation has not yet been diagnosed. of biological variation one is studying. For herbivores that I am formally introducing this term for the first time have not been studied in this comprehensive manner, the here. That fact notwithstanding, I have used it in prior publi- potential for ambiguity might be usefully highlighted by cations when referring to the populations of N. bebbianae leaf speaking of a “nominal” host range. Acknowledging such beetles associated with their six different host plants [11, 29, ambiguity is critical as host range bears not only on ecolo- 31, 35–42]. As my prior use of “host form” in papers and talks gical speciation but also on the evolution of ecological spe- has not led to its broader adoption by the field, I hope that cialization, the macroevolutionary dynamics of host use, this paper will more effectively encourage such usage (!). The crop pest control strategies, and so forth (e.g., [32, 33]). coining of this term was motivated by three things: first, the Note that these terms and issues may analogously be demonstration of cryptic host-associated variation within N. applied to the host range of nonherbivore parasites or to the bebbianae, namely, patterns of relative acceptance of the six “ecological range” of nonparasitic taxa. N. bebbianae hosts that varied according to the native host of the test population [31]. Second, insufficient data to infer what kind of biological variation was represented by beetles 5. “Host Form”: A Missing Term in natively associated with each of the six hosts. Third, the the Vocabulary of Biological Variation lack of an existing term suitable for describing this real yet underdetermined kind of host-associated variation. Thus, I Just as it is important to recognize when one has an in- coined “host form” as an evolutionarily neutral and needed complete understanding of host range, it is vital to assess missing term in the vocabulary of biological variation. I have confidence in one’s assessment of the kind of biological var- also introduced “ecological form” as an analogously neutral iation being addressed and to acknowledge the lack of same. term to be applied to underinvestigated nonparasitic taxa Presuming the kind of variation one is dealing with without exhibiting some form of ecological differentiation. International Journal of Ecology 5

6. Host Races: Ever Intriguing, Overdiagnosed mate choice and (4b) undergo appreciable gene flow. (5a) They have higher fitness on natal than alternative hosts and Among the different kinds of biological variation, host races (5b) produce hybrids that are less fit than parental forms. might be argued to be the most intriguing, at least to students (Note that analogous criteria must be met for populations of of speciation. The reasons are twofold (see [21, 26, 43], for nonherbivorous parasites to be appropriately assigned host discussion of the following issues). First, host races represent race status or for nonparasites to be assigned sympatric race intermediate points along the speciation continuum and status.) The authors apply these criteria to the literature in thus provide unusual opportunities for studying speciation search of study systems that qualify as host races. Notably, as a process. This is because host races exhibit considerable this search turned up only three such systems, and even these but incomplete reproductive isolation. This reproductive are listed as uncertain for criterion (5b). This number is isolation, in concert with divergent selection, can facilitate strikingly low. And even if the ten years since this paper’s or maintain ecological differentiation between host races publication has seen the documentation of more such cases, and genomic differentiation at regions under sufficiently it seems unlikely that compelling evidence for host races is strong divergent selection. The gene flow that does occur nearly as common as their invocation in the literature would prevents the fixation of alternative alleles even at selected loci, imply. Surely this term is being misapplied. Why so? while homogenizing the majority of the genome between Three related reasons come to mind. First, consider again populations(see, e.g., [44]). However, whether host races will that sympatric speciation is theoretically onerous and thus continue along the path to speciation, remain indefinitely at often argued to be rare and also that understanding patterns an equilibrium level of differentiation, or ultimately collapse of diversification (e.g., the spatial coexistence of related back into a single population is never known. species) would be much simpler should this speciation Second, because host races demonstrate that adaptive mode be plausibly frequent. For these reasons, evidence sup- differentiation and partial reproductive isolation can be porting sympatric speciation is rightly greeted with special maintained in the face of local gene flow, they have long excitement. Consider again too that evidence for host races been offered as evidence for the plausibility of sympatric is sometimes considered de facto evidence for sympatric speciation (e.g., [9]). This is important because, as noted speciation. Due to this connection, then, work claiming host earlier, the theoretical conditions under which sympatric race documentation might also be viewed as especially note- speciation can occur are generally restrictive. However, worthy, above and beyond the intrigue rightly accorded to although the existence of host races is consistent with sym- host races per se. Because journals seek to publish important patric speciation, it does not imply sympatric speciation. and broadly interesting findings, a selection process may Indeed, the most theoretically onerous stages of sympatric thus be inadvertently imposed that favors the publication of speciation are the early ones in which reproductive isolation such work and thus the making of such claims. Somewhat must evolve from within an initially panmictic population conversely, a second reason may be that as the plausibility [14]. Thus, unless sympatric origins have been demon- of sympatric speciation has become more widely accepted in strated, host races may alternatively reflect secondary contact recent years ([46], e.g., announces its comeback) so has the and the maintenance of differentiation between populations plausibility of host races, again by virtue of their perceived whose initial divergence occurred allopatrically. Nonetheless, equivalence. Presuming the veracity of this view, there may an erroneous bond between these terms can be found in the be every good reason, from the perspective of likelihood, to literature whereby claims of host race status are implicitly or suspect that a case of host-associated differentiation could explicitly linked with an inference of sympatric speciation. potentially be a case of host race formation. Thus, data This bond may have promoted an overestimation of how consistent with, if not specifically demonstrative of, host amply sympatric speciation has been documented. This is races may more likely be used as the basis for a paper focusing somewhat similar to how the overlap between ecological on hopeful host races. The third reason is based on what speciation and sympatric speciation has led to inappropriate sometimes appears to be the invocation of host races without equating of these concepts and an overestimate of the latter. any apparent justification. Here, it might be suspected that To take this argument back a step, a greater problem workers have simply not familiarized themselves with the is a considerable overdiagnosis of host races themselves criteria for diagnosing host races and that their frequent that runs boldly through the literature. This despite a long invocation has, via a positive feedback loop, resulted in such history of delineating the criteria for host race status [10, haphazard use as to make this term almost synonymous 13, 45], culminating in an especially well-detailed and argued with host-associated variation per se. That is, it appears characterization of these criteria [26]. According to Dres` and that in some circles this most intriguing and empirically Mallet, host races are sets of populations with the following onerous term is now serving the purpose of the least characteristics. (1a) They use different host taxa in the wild restrictive of terms, that of the here-proposed “host form.” and (1b) consist of individuals that exhibit host fidelity to This explanation might account for my own experiences their respective hosts. (2) They coexist in sympatry in at least of having N. bebbianae populations that I have consistently part of their range. (3a) They are genetically differentiated referred to as “host forms” nevertheless described in paper at more than one locus and (3b) are more genetically and grant reviews as “host races.” Hopefully, the introduction differentiated from each other in sympatry than either is of host form into the vocabulary of biological variation will from some geographically distant populations on the same help restore host race to its proper place as an inherently fas- host. (4a) They display a correlation between host choice and cinating phenomenon and a more reluctantly invoked term. 6 International Journal of Ecology

7. From Host Forms to Host Races: mating more readily with individuals of the same host form The Maple- and Willow-Associated [11, 35]. Populations of N. bebbianae (4b) Population trees based on putatively neutral loci group the sympatric populations together as more geneti- Readers to this point may suspect that I hold views anti- cally similar to each other than either is to allopatric pop- thetical to the notions of host races and sympatric speciation. ulations of the same host form, as expected if gene flow has Rather, I am fairly agnostic on the issue of how evolutionarily homogenized neutral loci in sympatry [44]. frequent and important these phenomena are. Indeed, re- (5a) Each host form grows faster and survives better on cent population genomic work may herald the common its native host than on the host of the other host form [35, documentation of adaptive genomic differentiation in the 41, 47]. face of gene flow [42]. However, just as Dres` and Mallet (5b) Hybrids between these host forms exhibit reduced [26] have provided criteria for documenting host races growth rate and survivorship than the parental types do on that seldom seem to have been demonstrated, so too does their native hosts [41, 47]. sympatric speciation theory lay out predicted parameters N. bebbianae thus presents one of the very few examples that have rarely been estimated [15]. Part of the problem for which data meet all the criteria of host races. They also is undoubtedly that obtaining adequate data for any one more generally represent an increasingly well-understood system requires a great amount of work. And sealing a strong case of potentially ongoing ecological speciation. case for sympatric speciation requires information—on the historical distributions of populations/species—that is rarely available. The latter point may forever prohibit claims about 8. Conclusion sympatric speciation (pro or con) in my focal study system, As does any form of rigorous inquiry, the accurate, informed Neochlamisus leaf beetles. However, work on two of the six study of ecological speciation depends crucially on termino- N. bebbianae host forms, those associated with maple and logical issues. It requires that terms be used in a manner con- willow, has by now given me sufficient confidence to cease sistent with their definitions. It requires that terms be quali- my use of “host forms” to describe these populations and fied as necessary to accurately convey degree of confidence. It declare, here, that they represent proper host races. Following requires that “missing” terms be created as needed to fill con- the criteria of Dres` and Mallet [26], the findings on which ceptual gaps. It requires that inappropriate synonymizing, this claim is based are as follows. loose usage, and simple terminological ignorance be avoided. (1a) These host forms are associated with divergent host As argued here, deviations from these prescriptions seem taxa in nature, using red maple (Acer rubrum, Aceraceae) and to have led to overestimation of the evidence supporting Bebb’s willow (Salix bebbiana, Salicaceae), respectively [35]. host races and sympatric speciation, and the diffusion of the (1b) Choice experiments involving6” host cuttings in former term’s special status as an important evolutionary salad containers and leaf fragments in Petri dishes demon- phenomenon. This loss may owe to the prior lack of an strate that each sex of both host forms spends the majority of alternative term to convey the existence of host-differentiated its time on its native host, demonstrating host fidelity [35]. entities whose evolutionary status cannot yet be determined. (2) These host forms coexist in sympatry and syntopy Such a term, “host form,” is proposed here as an important in the same disturbed and riparian habitats across a broad conceptual gap-filler. By contrast, the proper application of region of northeastern North America (New England, New the term host race is illustrated here by describing how each York, and southeastern Canada), where they commonly of its criteria are fulfilled by maple- and willow-associated grow intermixed with each other. For example, I have populations of N. bebbianae leaf beetles. Although the focus documented host-associated differentiation in sympatric of this paper has been on herbivorous insects, the appropriate populations from Cumberland Co., ME, Kennebec Co., ME, application of (sometimes new) terms and concepts across Middlesex Co., MA, Exeter Co., NH, Oswego Co., NY (from diverse taxa is also highlighted. 2different sites), and Caledonia Co., VT. (3a) Genome scan data reveal multiple loci that are much more highly differentiated than expected under drift [44]. References Many of these putatively divergently selected “outlier” loci are only differentiated in comparisons of populations using [1] E. F. Keller and E. A. Lloyd, Keywords in Evolutionary Biology, different hosts, implicating them in divergent host adapta- Harvard University Press, Cambridge, Mass, USA, 1992. tion. Notably, this differentiation is observed between sym- [2] E. Mayr, This is Biology: the Science of the Living World,Har- patric populations of the host forms from a site in northern vard University Press, Cambridge, Mass, USA, 1997. Vermont. [3] C. Linnaeus, Systema Naturae, Stockholm, Sweden, 10th edi- (3b) Population trees based on these host-associated out- tion, 1758. [4] C. Darwin, On the Origin of Species by Means of Natural Selec- lier loci show these sympatric host forms to be more dif- tion or the Preservation of Favored Races in the Struggle for Life, ferentiated from each other than either is to allopatric pop- Murray, London, UK, 1859. ulations of the same host form [44]. [5] E. Mayr, Systematics and the Origin of Species, Columbia Uni- (4a) The host fidelity of these host forms is associated versity Press, New York, NY, USA, 1942. with mating fidelity, in that positively assortative mating [6] B. D. Walsh, “On phytophagic varieties and phytophagic spe- (= sexual isolation) is observed between them, with each one cies, with remarks on the unity of coloration in insects,” International Journal of Ecology 7

Proceedings of the Entomological Society of Philadelphia, vol. 5, Neochlamisus leaf beetles,” Ecological Entomology, vol. 35, no. pp. 194–215, 1865. 1, pp. 41–53, 2010. [7] W. J. Brown, “Sibling species in the Chrysomelidae,” in Pro- [30] J. P. Karren, “A revision of the subfamily Chlamisinae north ceedings of the 10th International Congress of Entomology, vol. of Mexico (Coleoptera: Chrysomelidae),” University of Kansas 1, pp. 103–109, 1958. Science Bulletin, vol. 49, pp. 875–988, 1972. [8] W. J. Brown, “Taxonomic problems with closely related spe- [31] D. J. Funk, “Investigating ecological speciation,” in Speciation cies,” Annual Review of Entomology, vol. 4, pp. 77–98, 1959. and Patterns of Diversity, R. K. Butlin, J. Bridle, and D. Schluter, [9] G. L. Bush, “Sympatric host-race formation and speciation in Eds., pp. 195–218, Cambridge University Press, Cambridge, frugivorous flies of the genus Rhagoletis (Diptera, Tephriti- UK, 2009. dae),” Evolution, vol. 23, pp. 237–251, 1969. [32] J. Jaenike, “Host specialization in phytophagous insects,” An- [10] S. R. Diehl and G. L. Bush, “An evolutionary and applied per- nual Review of Ecology and Systematics, vol. 21, no. 1, pp. 243– spective of insect biotypes,” Annual Review of Entomology, vol. 273, 1990. 29, pp. 471–504, 1984. [33] D. J. Futuyma, M. C. Keese, and D. J. Funk, “Genetic con- [11] D. J. Funk, K. E. Filchak, and J. L. Feder, “Herbivorous insects: straints on macroevolution: the evolution of host affiliation in model systems for the comparative study of speciation ecol- the leaf beetle genus Ophraella,” Evolution,vol.49,no.5,pp. ogy,” Genetica, vol. 116, no. 2-3, pp. 251–267, 2002. 797–809, 1995. [12] K. W. Matsubayashi, I. Ohshima, and P.Nosil, “Ecological spe- [34] A. M. Dickey and R. F. Medina, “Testing host-associated dif- ciation in phytophagous insects,” Entomologia Experimentalis ferentiation in a quasi-endophage and a parthenogen on na- et Applicata, vol. 134, no. 1, pp. 1–27, 2010. tive trees,” Journal of Evolutionary Biology, vol. 23, no. 5, pp. [13] E. Mayr, Animal Species and Evolution,HarvardUniversity 945–956, 2010. Press, Cambridge, Mass, USA, 1963. [35] D. J. Funk, “Isolating a role for natural selection in specia- [14] D. J. Futuyma and G. C. Mayer, “Non-allopatric speciation in tion: host adaptation and sexual isolation in Neochlamisus beb- animals,” Systematic Zoology, vol. 29, pp. 254–271, 1980. bianae leaf beetles,” Evolution, vol. 52, no. 6, pp. 1744–1759, [15] D. I. Bolnick and B. M. Fitzpatrick, “Sympatric speciation: 1998. models and empirical evidence,” Annual Review of Ecology, Ev- [36] C. G. Brown and D. J. Funk, “Aspects of the natural history olution, and Systematics, vol. 38, pp. 459–487, 2007. of Neochlamisus (Coleoptera: Chrysomelidae): fecal case-asso- [16] D. Schluter, The Ecology of Adaptive Radiation,OxfordUniver- ciated life history and behavior, with a method for studying sity Press, Oxford, UK, 2000. insect constructions,” Annals of the Entomological Society of [17] D. Schluter, “Ecology and the origin of species,” Trends in Eco- America, vol. 98, no. 5, pp. 711–725, 2005. logy and Evolution, vol. 16, no. 7, pp. 372–380, 2001. [37] P.Nosil, T. H. Vines, and D. J. Funk, “Perspective: reproductive [18] H. D. Rundle and P.Nosil, “Ecological speciation,” Ecology Let- isolation caused by natural selection against immigrants from ters, vol. 8, no. 3, pp. 336–352, 2005. divergent habitats,” Evolution, vol. 59, no. 4, pp. 705–719, [19] D. J. Funk, P. Nosil, and W. J. Etges, “Ecological divergence ex- 2005. hibits consistently positive associations with reproductive iso- [38] S. P. Egan and D. J. Funk, “Individual advantages to ecological lation across disparate taxa,” Proceedings of the National Acad- specialization: insights on cognitive constraints from three emy of Sciences of the United States of America, vol. 103, no. 9, conspecific taxa,” Proceedings of the Royal Society B, vol. 273, pp. 3209–3213, 2006. no. 1588, pp. 843–848, 2006. [20] A. P. Hendry, “Ecological speciation! Or the lack thereof?” [39] C. S. Chaboo, C. G. Brown, and D. J. Funk, “Faecal case archi- Canadian Journal of Fisheries and Aquatic Sciences, vol. 66, no. tecture in the gibbosus species group of Neochlamisus Karren, 8, pp. 1383–1398, 2009. 1972 (Coleoptera: Chrysomelidae: Cryptocephalinae: Chlam- [21] J. A. Coyne and H. A. Orr, Speciation, Sinauer, Sunderland, isini),” Zoological Journal of the Linnean Society, vol. 152, no. MA, 2004. 2, pp. 315–351, 2008. [22] U. Dieckmann, J. A. J. Metz, M. Doebeli, and D. Tautz, “In- [40] D. J. Funk and P. Nosil, “Comparative analyses and the study troduction,” in Adaptive Speciation, U. Dieckmann, J. A. J. of ecological speciation in herbivorous insects,” in Specializa- Metz, M. Doebeli, and D. Tautz, Eds., pp. 1–17, Cambridge tion, Speciation, and Radiation: the Evolutionary Biology of Her- University Press, Cambridge, UK, 2004. bivorous Insects, K. Tilmon, Ed., pp. 117–135, University of [23] L. R. Fox and P. A. Morrow, “Specialization: species property California Press, Berkeley, Calif, USA, 2008. or local phenomenon,” Science, vol. 211, no. 4485, pp. 887– [41] S. P. Egan and D. J. Funk, “Ecologically dependent postmating 893, 1981. isolation between sympatric host forms of Neochlamisus beb- [24] D. J. Futuyma, Evolutionary Biology, Sinauer Associates, Sun- bianae leaf beetles,” Proceedings of the National Academy of derland, Mass, USA, 1998. Sciences of the United States of America, vol. 106, no. 46, pp. [25] M. Allaby, The Concise Oxford Dictionary of Ecology,Oxford 19426–19431, 2009. University Press, Oxford, UK, 1994. [42] P. Nosil, D. J. Funk, and D. Ortiz-Barrientos, “Divergent selec- [26] M. Dres` and J. Mallet, “Host races in plant-feeding insects tion and heterogeneous genomic divergence,” Molecular Ecol- and their importance in sympatric speciation,” Philosophical ogy, vol. 18, no. 3, pp. 375–402, 2009. Transactions of the Royal Society B, vol. 357, no. 1420, pp. 471– [43] S. H. Berlocher and J. L. Feder, “Sympatric speciation in phy- 492, 2002. tophagous insects: moving beyond controversy?” Annual Re- [27] R. Lincoln, G. Boxshall, and P. Clark, A Dictionary of Ecology, view of Entomology, vol. 47, pp. 773–815, 2002. Evolution and Systematics, Cambridge University Press, Cam- [44] S. P. Egan, P. Nosil, and D. J. Funk, “Selection and genomic bridge, UK, 1998. differentiation during ecological speciation: Isolating the con- [28] D.R.Strong,J.H.Lawton,andR.Southwood,Insects on Plants, tributions of host-association via a comparative genome scan Harvard University Press, Cambridge, Mass,USA, 1984. of Neochlamisus bebbianae leaf beetles,” Proceedings of the [29] D. J. Funk, “Does strong selection promote host specialisation Entomological Society of Philadelphia, vol. 62, pp. 1162–1181, and ecological speciation in insect herbivores? Evidence from 2008. 8 International Journal of Ecology

[45] J. Jaenike, “Criteria for ascertaining the existence of host races,” American Naturalist, vol. 117, pp. 830–834, 1981. [46] S. Via, “Sympatric speciation in animals: the ugly duckling grows up,” Trends in Ecology and Evolution,vol.16,no.7,pp. 381–390, 2001. [47] S. P. Egan, E. M. Janson, C. G. Brown, and D. J. Funk, “Post- mating isolation and genetically variable host use in ecologically divergent host forms of Neochlamisus bebbianae leaf beetles,” Journal of Evolutionary Biology, vol. 24, no. 10, pp. 2217– 2229, 2011. Hindawi Publishing Corporation International Journal of Ecology Volume 2011, Article ID 287532, 7 pages doi:10.1155/2011/287532

Research Article Learning the Hard Way: Imprinting Can Enhance Enforced Shifts in Habitat Choice

Niclas Vallin and Anna Qvarnstrom¨

Animal Ecology, Department of Ecology and Genetics, Evolutionary Biology Centre, Uppsala University, Norbyvagen¨ 18D, 752 36 Uppsala, Sweden

Correspondence should be addressed to Niclas Vallin, [email protected]

Received 29 June 2011; Revised 30 August 2011; Accepted 16 September 2011

Academic Editor: Zachariah Gompert

Copyright © 2011 N. Vallin and A. Qvarnstrom.¨ This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We investigated the potential importance of learning in habitat choice within a young hybrid zone of two closely related species of birds. Pied flycatchers (Ficedula hypoleuca) are being excluded from deciduous habitats into a mixed forest type by collared flycatchers (F. albicollis). We investigated whether this enforced habitat shift influenced reproductive isolation between the two species, and, by cross-fostering nestlings, we tested whether learning may lead to a corresponding shift in habitat choice in consecutive generations. Our results show that the majority of the recruits, even if translocated across different habitat types, return to breed in the area where they were fostered. As male pied flycatchers were more likely to hybridize in the originally preferred habitat, we argue that early imprinting on an alternate habitat can play an important role in increasing reproductive isolation and facilitate regional coexistence between species experiencing secondary contact.

1. Introduction response have higher fitness than those that do not. In particular, moderate levels of plasticity appear to enhance In the past decade, a revived interest in the role of ad- evolution in novel environments by placing populations ff aptation to di erent environments in speciation, that is, under directional selection towards new adaptive peaks ecological speciation, has emerged [1–4]. One important [17]. Thibert-Plante and Hendry [16] found that plasticity mechanism underlying ecological speciation is the buildup also could counteract the buildup of reproductive isolation of reproductive isolation caused by habitat segregation (i.e., between populations after colonization of new environ- habitat isolation, e.g., [5]). Theoretical models of sympatric ments. This is because selection against immigrants can be speciation generally include disruptive selection in resource mitigated if dispersal occurs before the plastic adjustment to ff use, which by extension can lead to di erences in habitat the environment. Thus, plasticity may have a large impact choice and thereby reduced interbreeding [6–12]. Adaptive on the speciation process. Still, an underlying assumption ff di erences in habitat use can moreover evolve among in many models on the evolution of habitat isolation is that allopatric populations, resulting in habitat isolation at sec- habitat choice is a genetically determined trait. However, ondary contact [13, 14]. Habitat isolation between species learned habitat choice occurs in several animal species can also evolve at secondary contact, where reinforcement [20] and may, together with other forms of learning (e.g., and competition are plausible drivers [15]. sexual imprinting; [21]), be of importance in relation to One factor suggested to potentially both promote and speciation [6, 16, 18, 22, 23]. Beltman and Metz [24]found reduce the likelihood of ecological speciation is phenotypic that speciation is more likely to occur through a learned plasticity [16]. Phenotypic plasticity can be defined as than through a genetic habitat preference, unless the cost changes in an individual’s behavior, morphology, and physi- of learning is high, which, together with recent empirical ology directly induced by different environmental conditions findings (reviewed by [20, 25]), indicates that the relevance [17–19] and is adaptive if individuals showing a plastic of learning in speciation may have been underestimated. 2 International Journal of Ecology

In this study, we use a combination of long-term breeding data and an experimental approach to investigate Mainland mechanisms behind divergence in habitat choice within a Sweden 1 young hybrid zone of two closely related, hybridizing, and competing species of Ficedula flycatchers. Collared (Ficedula albicollis) and pied (F. hypoleuca) flycatchers cooccur in central and eastern Europe and on the Baltic isles of Oland¨ and Gotland in Sweden [26]. The two species started to diverge 1-2 million years ago and were probably periodically isolated in separate glacial refuges during the Pleistocene, ¨ before expanding their breeding ranges northward [27]. Both Oland 2 species are migratory and winter in Africa in separated, but probably slightly overlapping, areas [28]. Pied flycatchers 137 km migrate along the Iberian Peninsula and winter in western to central Africa, and collared flycatchers winter in southeastern Africa, taking a more eastern migration route through Italy or further east [28, 29]. Males of the two species compete fiercely over natural tree holes or nest boxes after arrival at the breeding grounds in Europe [30, 31], and the two species overlap in timing of breeding [32] and feeding habits Baltic sea [33] suggesting strong interspecific competition. Collared 3 flycatcher offspring have a higher growth potential under favorable environmental conditions [32, 34], but pied flycatchers are relatively more robust to harsh environments Lantmateriet¨ Gavle¨ 2011. and towards the seasonal decline in food availability, pro- Medgivande I 2011/0100 viding them an opportunity to prolong coexistence with Figure 1: Landscape types (1 = mixed or coniferous forest, 2 = the otherwise more aggressive collared flycatchers [32, 34]. = Pied flycatchers prefer to breed in deciduous forests [29]but deciduous forest, 3 agricultural land) and distribution of Ficedula ¨ flycatcher nest-box areas (blue circles) on the Swedish island of are often found in coniferous forests on Oland [35]. The Oland.¨ The large circle represents nine nest-box areas surrounding proportion of deciduous trees in pied flycatcher territories the town of Lottorp¨ where collared flycatchers started to colonize has declined in recent years as the number of collared the island (and where pied flycatchers were already present) in the flycatchers has increased [36] suggesting that competition- late 1950s–early 1960s. Collared flycatchers now outnumber pied mediated habitat segregation is currently taking place. Here, flycatchers in this area (∼90% collared flycatchers). The relative we investigate potential mechanisms enhancing divergent proportion of collared flycatchers gradually declines southward. habitat choice in these two ecologically similar Ficedula fly- Two of the nest-box areas in the north of the island are dominated catcher species. More specifically, we test the role of learning by coniferous forest and so far only inhabited by pied flycatchers. in maintaining initially enforced shifts in habitat choice. is monitored every third day to assess the starting day of 2. Material and Methods egg laying. All breeding birds and their offspring are marked individually with numbered metal rings, and a small amount 2.1. Study System. Collared flycatchers started to colonize of blood is sampled for genetic analyses. Yearly records Oland,¨ where pied flycatchers were already present, in the late are kept on onset of egg laying, clutch size, hatching date, 1950s–early 1960s, and since then the relative proportion of hatching, and fledging success. Morphological characters are pied flycatchers has quickly declined in the most favorable measured in a standardized manner [30]. Adult females are breeding sites [32]. There is species assortative mating in caught when incubating the eggs, and adult males are caught the mixed study population on Oland,¨ but 5% of all pairs when feeding the nestlings. Nestlings are weighed at day 7 are heterospecific, and approximately 2–7% of the breed- and 13, and tarsus length is measured at day 13, a couple ing flycatchers in different mixed populations are hybrids of days before fledging. Monitoring of the nest boxes has (reviewed in [26]). Male hybrids have lowered fitness, mainly been performed during the periods 1981–1985 and from due to disadvantages in competition over mates, while female 2002 onwards. The breeding habitat was measured as the hybrids are sterile in accordance with Haldane’s rule [37]. On abundance of each tree species 360◦ around the nest boxes Oland,¨ the two species breed in 21 nest-box areas situated using a “relascope” (see [35]),whichisascaleheldatafixed across the island (Figure 1). The landscape on Oland¨ is distance from the eye. By looking through an opening in the characterized by a limestone plain covered by a thin soil layer. scale, individual trees are assigned into three categories based The southern part of the island is dominated by agricultural on trunk size and distance from the nest box. To investigate land, the middle part by deciduous forest, and the very north whether the habitat choice of pied flycatchers influences the of the island contains mixed or coniferous dominated forests risk of hybridization, we compared the habitat composition (Figure 1). Females lay one egg each day, and every nest box of con- and heterospecifically paired male pied flycatchers. International Journal of Ecology 3

CF 177 25 1

CF PF 0.8

(i) (ii) 0.6 (a) CF PF 0.4 Proportion of trees

0.2

0 (i) (ii) PF PFCF (b) Coniferous Figure 2: Cross-fostering of nestlings between collared (CF) Deciduous and pied flycatchers (PF) on the Swedish island of Oland¨ using Figure 3: Proportion of coniferous versus deciduous tree species in two different approaches. (a) Complete clutches of eggs swapped the breeding territories of pied flycatcher pairs (PF) compared to between nests in three steps. (b) Partial cross-fostering of three day the breeding territories of male pied flycatchers paired to a female old nestlings between two nests. The left side (i) refers to the nests collared flycatcher (PFCF) on the Swedish island of Oland,¨ 2002– before swapping, and the right side (ii) refers to the same nests after 2010. Male pied flycatchers have a lower risk of hybridizing in swapping (collared flycatchers in white and pied flycatchers in grey). habitats with coniferous forest. Sample sizes are given above the bars. 2.2. Cross-Fostering Experiment. We investigated the role of early learning in breeding habitat choice by comparing the general habitat composition (proportion of deciduous patterns of recruitment of birds that had been reared trees) influenced the chance of pairing with a con- or by foster parents. To increase the likelihood of recruiting heterospecific female among male pied flycatchers using nestlings reared in a new environment, we swapped complete a generalized linear model adjusted for overdispersion with clutches of eggs three days before the expected hatching date binomial errors and a logit link. We found that hybridizing between different nests; the genetic parents nest (i.e., the male pied flycatchers had a significantly higher proportion nest of origin) and the foster parents nest (i.e., the nest of deciduous trees in their breeding territories than males of fledging). The cross-fostering experiment was performed paired to a conspecific female (N = 202, df = 1, χ2 = 14.19, between three nests with the same clutch size and date of egg P<0.001, Figure 3). laying: from collared flycatcher to collared flycatcher (to be We used cross fostering experiments to investigate used as a control in another study; Vallin et al. unpublished whether learning may influence habitat choice. In total, manuscript), from collared flycatcher to pied flycatcher with we included 264 nests subject to cross-fostering between the pied flycatcher eggs thereafter being translocated back the years 2002–2009 in these analyses: 96 nests where eggs to the original collared flycatcher nest (Figure 2(a)). In were translocated and 168 nests where nestlings were cross- addition, we also included recruits stemming from earlier fostered. Many of the study plots are located far apart cross-fostering experiments in our analysis. In these cross- (maximum distance ∼85000 m, Figure 1) with pronounced fostering experiments, broods with coinciding hatching dates differences in habitat composition [32]. We compared the were split in half and nestlings were swapped between nests at general habitat composition (proportion of deciduous trees) the age of three days (e.g., [34, 38], Figure 2(b)). In one study among the pied and collared flycatcher territories that [32], cross-fostering was performed with an additional brood nestlings were swapped between using a generalized linear size manipulation to simulate good and harsh environments. model adjusted for overdispersion with binomial errors All nestlings were individually marked by clipping their and a logit link. We found that the habitat composition toenails to enable individual measurements of growth and of pied and collared flycatchers included in the experi- to ensure correct identification of species and nest of origin. ment was significantly different (N = 48, χ2 = 10.62, JMP 8 (SAS Institute, Cary, NC, USA) was used to analyze df = 1, P = 0.001), with a higher proportion of coniferous the data. trees in pied flycatcher territories (Figure 4). Thus, in general nestlings that were swapped into a new nest experienced a ff 3. Results di erent rearing environment than they would have done if they had been reared by their genetic parents. Pied flycatchers are currently being excluded from the pre- We identified 49 recruits stemming from the experimen- ferred deciduous habitats on Oland,¨ and we compared how tal nests. The average recruitment rate was 4.3% (40 recruits 4 International Journal of Ecology

24 24 1 Nest of genetic parents Nest of foster parents

0.8 Mean distance = Mean distance = 22754 m 1022 m

0.6 Own breeding nest consecutive year

Figure 5: Yearling birds recruited from cross-fostering experiments 0.4 between pied and collared ﬂycatchers on the Swedish island of ¨ Proportion of trees Oland 2002–2009 subsequently settled to breed signiﬁcantly closer totheirfosterparentsthantotheirgeneticparents. 0.2

P<0.001). The mean distance between the foster parents 0 CF PF nest and the nest where the recruit subsequently settled was 1022 meters (N = 24, range = 33–4256 m, se = 247 m, Coniferous Figure 5), and the mean distance between the genetic parents nest and the nest of recruitment was 22754 meters (range = Deciduous 130–64481 m, se = 5413 m, Figure 5). Figure 4: Proportion of coniferous versus deciduous tree species Lastly, we tested whether recruited offspring that had in the breeding territories of collared flycatchers (CF) and pied been translocated between habitat types returned to breed flycatchers (PF) subject to experimental cross- fostering of their in a habitat different from their genetic parents. The pied fly- nestlings on Oland,¨ Sweden, 2002–2009. The territories of collared catcher recruits had all been swapped between different nest- flycatchers had a relatively higher proportion of deciduous trees box areas and had a significantly higher mean proportion of as compared to the territories of pied flycatchers. Sample sizes are deciduous trees in their breeding territory as compared to given above the bars. their genetic parents (two-tailed Fisher’s exact test, N = 8, P = 0.009, Figure 6(a)). Similarly, an analysis with a subset from 933 confirmed fledged offspring, i.e., data on fledging of collard flycatcher recruits that had been swapped between success were missing from the nests of 9 recruits). There different nest-box areas revealed that also juvenile collared was no significant difference in recruitment between nests flycatchers shifted their breeding habitat choice when they depending on whether eggs or nestlings were swapped (N = hadbeenrearedinadifferent environment as compared to 218, χ2 = 0.704, df = 1, P = 0.40), and no significant their genetic parents (two-tailed Fisher’s exact test, N = 28, difference in recruitment between the species (N = 218, P = 0.0240, Figure 6(b)). χ2 = 0.386, df = 1, P = 0.53); pied flycatchers recruited 21 juveniles from a total of 431 confirmed fledged offspring 4. Discussion and collared flycatchers recruited 19 juveniles from a total of 502 confirmed fledged offspring. Of the 49 recruits from Pied flycatchers are currently being rapidly excluded from the cross-fostering experiments, 30 (61.2%) were recruits their preferred deciduous habitats on Oland¨ and are instead that had been swapped into a different rearing nest than the increasingly found breeding in mixed or coniferous forests one attended by their genetic parents, and 24 of those were [32, 36]. We tested if the breeding habitat of male pied recruits from swaps between pied and collared flycatchers: flycatchers influenced the risk of hybridizing and found that 20 collared flycatchers (11 males and 9 females) and 4 pied male pied flycatchers were more likely to pair with a het- flycatchers (1 male and 3 females). Thus, collared flycatchers erospecific female in deciduous habitats, that is, the observed had a significantly higher chance of recruiting back to the shift in habitat occupancy increases premating isolation population compared to pied flycatchers when being raised between the species. We investigated whether learning may by foster parents (two-tailed Fisher’s exact test: N = 24, speed up segregation in habitat choice. By cross-fostering P = 0.03). nestling Ficedula flycatchers, we found that the recruits, We applied a matched pairs t-test to test whether that is, one-year old birds returning to breed themselves, nestlings cross-fostered between pied and collared flycatchers settled closer to the territory of their foster parents than returned closer to the breeding site of their genetic parents or to the territory of their genetic parents (Figure 5). In fact, to the breeding site of their foster parents. GPS coordinates the majority of the recruits, even if translocated into a (Swedish grid system RT90) of the nest-box locations were different habitat type, returned as adults to breed in the same taken and the Euclidean distance in meters between them woodlot as they were fostered into. A similar pattern was calculated using Pythagoras’ formula. found for both species but the comparatively low proportion The distance between the nest where the recruits of pied flycatchers returning to breed as adults after being returned to breed and the nest of the foster parents was raised by foster parents (4 pied flycatchers versus 20 collared significantly shorter than the distance between the nest flycatchers) means that we cannot investigate if there is a where they bred as recruited and the nest of origin (mean difference in the strength of learned habitat choice between difference = 21732 m, N = 24, t = 3, 92722, df = 23, the two species. The difference in recruitment per se is likely International Journal of Ecology 5

4 4 4 in coniferous habitats as compared to deciduous habitats 1 [35, 40, 41], and allopatric populations of pied flycatchers experience higher reproductive success in deciduous habitats 0.8 [29, 42]. Coniferous habitats are therefore unlikely to be the originally preferred habitat of pied flycatchers and in the young hybrid zone on Oland¨ competition with collared 0.6 flycatchers appears to be responsible for a rapid shift in habitat occupancy [36]. Our present study provides a 0.4 mechanism by which a competition-mediated switch into a

Proportion of trees suboptimal habitat could result in a changed habitat choice 0.2 through learning. We found that being reared in the initially less preferred habitat even makes collared flycatchers return to breed in this habitat as young adults. By extension, the 0 fact that young birds develop a preference for breeding in the PF recruit Foster parents Genetic parents same type of habitat as they were reared in means that an enforced habitat shift due to competition quickly can result Coniferous in divergence in habitat choice and strengthen premating Deciduous isolation. (a) One may argue that a possible alternative explanation to our findings is that flycatchers exhibit a genetically fixed 14 14 14 1 strategy to return to breed within a certain distance from their parents’ nest. However, several previous findings are not compatible with this alternative view. On the Swedish 0.8 mainland, pied flycatcher nestlings and their parents do not stay in close vicinity to the nest site once the offspring are 0.6 fledged and the nestlings do not return to breed within the same areas as their parents [29]. On the islands of Oland¨ and Gotland, such movements are more limited. Moreover, 0.4 aviary experiments performed in the older hybrid zone in

Proportion of trees central Europe suggest that pied flycatchers have developed 0.2 a preference to feed in coniferous trees [43], and these birds were caught in a coniferous habitat as the process of habitat segregation has reached a later stage in this older hybrid zone 0 [44]. CF recruit Foster parents Genetic parents For how long do the fledglings stay in their natal area, Coniferous and when does imprinting occur? Van Balen [45] studied postfledging dispersal in a Dutch pied flycatcher population Deciduous and found that juvenile birds tend to stay within 600 m of (b) their hatching site up to about 50 days. Berndt and Winkel [46] transferred both eggs and fledglings (about 36 days Figure 6: Habitat composition in the breeding territories of cross- old) of pied flycatchers between areas located 250 km apart fostered yearling pied (a) and collared flycatchers (b) as compared to the breeding territories of their foster parents and genetic parents in northern Germany. They found that all returning birds on Oland,¨ Sweden, 2002–2009. Both pied and collared flycatcher came back to the area where they had been transferred, recruits returned to breed in a habitat different from their genetic showing that imprinting can take place even after three weeks parents. Only recruits that were swapped between different nest-box postfledging. A recent study on flycatchers reveals that it is areas are included in this comparison, sample sizes are given above possible to translocate also adult birds between environments the bars. [47]. Apart from opening up new possibilities to study the fitness consequences of breeding in different habitats, this finding also indicate that the importance of plasticity to reflect the fact that pied flycatchers are having problems and learning in general might have been underestimated in establishing territories in areas inhabited by collared flycatch- studies of avian habitat choice. By showing that birds that ers [39]. Another potential source contributing to the biased have been experimentally reared in a different environment, pattern of recruitment might be that collared flycatcher as compared to their genetic parents, subsequently return to nestlings sharing nests with pied flycatcher nestlings have a the area of their foster parents, our results provide additional competitive advantage in achieving food from the parents support for the importance of early learning in habitat [38], which could increase the likelihood of reaching a choice, also on a small geographical scale. threshold weight for successful recruitment (e.g., [29]). In a few cases, the birds recruiting from our experiment Several studies indicate a poorer quality in terms of avail- settled to breed relatively far away from their foster area. ability of the food that flycatchers feed their nestlings with Although we cannot make any inferences from those few 6 International Journal of Ecology cases, future studies could try to pinpoint detailed differ- Acknowledgments ences in dispersal strategies and habitat choice in relation to, for example, population density, relative frequency of The authors would like to thank Zoologiska stiftelsen (NV), competitors, and variation in microhabitat quality. Natal the Swedish Research Council (AQ), and the European conditions are likely to affect dispersal propensity in a Research Foundation (AQ) for financial support. number of ways. For example, high-quality habitats may produce more individuals prone to disperse as compared to References low-quality habitats [48]. Lens and Dhondt [49]founda link between habitat condition and natal dispersal in crested [1] D. Schluter, “Ecology and the origin of species,” Trends in tits (Parus cristatus), where individuals from higher quality Ecology and Evolution, vol. 16, no. 7, pp. 372–380, 2001. habitats dispersed more readily than individuals from lower- [2] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology quality habitats. Heritable differences in dispersal propensity Letters, vol. 8, no. 3, pp. 336–352, 2005. in coupling with other behaviors (e.g., aggression) have [3] G. S. Van Doorn, P. Edelaar, and F. J. Weissing, “On the origin of species by natural and sexual selection,” Science, vol. 326, been suggested to be especially important during natural no. 5960, pp. 1704–1707, 2009. range expansions or hybrid zone movements in birds [50]. [4] A. M. Rice, A. Rudh, H. Ellegren, and A. Qvarnstrom,¨ “A guide However, as Davis and Stamps [20] point out: if dispersers to the genomics of ecological speciation in natural animal have a strong innate preference for their natal habitat type, populations,” Ecology Letters, vol. 14, no. 1, pp. 9–18, 2011. the degree of similarity between patches will influence the [5] W. R. Rice and G. W. Salt, “The evolution of reproductive rate of dispersal such that dispersal into novel habitats is isolation as a correlated character under sympatric conditions: prohibited. On the other hand, recent bird studies indicate a experimental evidence,” Evolution, vol. 44, no. 5, pp. 1140– strong role for habitat-related cultural transmission through 1152, 1990. early learning, for example, in the feeding strategy of great [6] W. R. Rice, “Disruptive selection on habitat preference and tits [51], natal dispersal of pied flycatchers [52], and habitat the evolution of reproductive isolation: a simulation study,” selection in the warbler finches of Galapagos,´ where females Evolution, vol. 38, no. 6, pp. 1251–1260, 1984. choose to breed on islands with habitats similar to their natal [7] P. A. Johnson, F. C. Hoppensteadt, J. J. Smith, and G. L. environment [53]. Bush, “Conditions for sympatric speciation: a diploid model Tonnis et al. [53] found a positive correlation between incorporating habitat fidelity and non-habitat assortative mating,” Evolutionary Ecology, vol. 10, no. 2, pp. 187–205, genetic distances in the warbler finches of Galapagos´ and ff 1996. di erences in maximum elevation among islands and con- [8] T. J. Kawecki, “Sympatric speciation driven by beneficial cluded that habitat selection may have helped to initiate mutations,” Proceedings of the Royal Society B, vol. 263, no. the speciation process. The genetic component of habitat 1376, pp. 1515–1520, 1996. preferences might be larger in more specialized populations, [9] T. J. Kawecki, “Sympatric speciation via habitat specialization and speciation may occur as assortative mating evolves driven by deleterious mutations,” Evolution,vol.51,no.6,pp. [24]. The effectiveness of learned habitat preferences in 1751–1763, 1997. speciation could be due to being a one-allele mechanism [10] A. S. Kondrashov and F. A. Kondrashov, “Interactions among [54], whereas speciation through a genetic habitat preference quantitative traits in the course of sympatric speciation,” has been suggested to be a two-allele mechanism [24]. In Nature, vol. 400, no. 6742, pp. 351–354, 1999. other words, when habitat choice is genetically determined, [11] U. Dieckmann and M. Doebeli, “On the origin of species by genetic recombination can lead to a breakdown of the sympatric speciation,” Nature, vol. 400, no. 6742, pp. 354–357, associations between the alleles underlying habitat choice 1999. and the alleles underlying ecological adaptation, whereas [12] J. D. Fry, “Multilocus models of sympatric speciation: bush versus rice versus felsenstein,” Evolution,vol.57,no.8,pp. this is not a problem with a learned habitat preference. 1735–1746, 2003. Song learning is another example of a culturally transmit- [13] E. Mayr, Systematics and the Origin of Species, Columbia ted trait in birds suggested to operate in a similar way University Press, New York, NY, USA, 1942. [55]. [14] T. Dobzhansky, Genetics and the Origin of Species, Columbia In summary, we have shown that an enforced shift in University Press, New York, NY, USA, 3rd edition, 1951. breeding habitat choice can be further enhanced by early [15] J. A. Coyne and H. A. Orr, Speciation, Sinauer Associates, imprinting on the natal area. From a theoretical point of Massachusetts, Mass, USA, 2004. view, learned habitat choice might be a powerful mechanism [16] X. Thibert-Plante and A. P. Hendry, “The consequences of of generating assortative mating [24]. When one species phenotypic plasticity for ecological speciation,” Journal of shifts breeding habitat due to interspecific competition, Evolutionary Biology, vol. 24, no. 2, pp. 326–342, 2011. segregation in habitat choice will increase (and premating [17] T. D. Price, A. Qvarnstrom,¨ and D. E. Irwin, “The role of isolation as a side effect) when recruitment into the alter- phenotypic plasticity in driving genetic evolution,” Proceedings of the Royal Society B, vol. 270, no. 1523, pp. 1433–1440, 2003. native environment exceeds recruitment into the originally [18] M. J. West-Eberhard, Developmental Plasticity and Evolution, preferred environment due to learning (i.e., imprinting on Oxford University Press, New York, NY, USA, 2003. the natal environment). Thus, learned habitat choice can play [19] M. J. West-Eberhard, “Developmental plasticity and the origin an important role both in enhancing reproductive isolation of species differences,” Proceedings of the National Academy of and in facilitating regional coexistence at secondary contact Sciences of the United States of America, vol. 102, no. 1, pp. between diverging populations. 6543–6549, 2005. International Journal of Ecology 7

[20] J. M. Davis and J. A. Stamps, “The effect of natal experience on [39] N. Vallin, A. M. Rice, H. Arntsen, K. Kulma, and A. habitat preferences,” Trends in Ecology and Evolution, vol. 19, Qvarnstrom,“Combinede¨ ffects of interspecific competition no. 8, pp. 411–416, 2004. and hybridization impede local coexistence of Ficedula fly- [21] D. E. Irwin and T. Price, “Sexual imprinting, learning and catchers,” Evolutionary Ecology. In press. speciation,” Heredity, vol. 82, no. 4, pp. 347–354, 1999. [40] L. Gezelius, M. Grahn, H. Kallander, and J. Karlsson, “Habitat- [22] W. H. Thorpe, “The evolutionary significance of habitat related differences in clutch size of the pied flycatcher Ficedula selection,” Journal of Animal Ecology, vol. 14, pp. 67–70, 1945. hypoleuca,” Annales Zoologici Fennici, vol. 21, no. 3, pp. 209– [23] J. Maynard Smith, “Sympatric speciation,” American Natural- 212, 1984. ist, vol. 100, pp. 637–650, 1966. [41] T. Eeva, S. Ruuskanen, J. P. Salminen et al., “Geographical [24] J. B. Beltman and J. A. J. Metz, “Speciation: more likely trends in the yolk carotenoid composition of the pied through a genetic or through a learned habitat preference?” flycatcher (Ficedula hypoleuca),” Oecologia, pp. 1–11, 2010. Proceedings of the Royal Society B, vol. 272, no. 1571, pp. 1455– [42] A. Jarvinen, “Clutch-size variation in the pied flycatcher 1463, 2005. Ficedula hypoleuca,” Ibis, vol. 131, no. 4, pp. 572–577, 1989. [25] T. Slagsvold and K. L. Wiebe, “Social learning in birds and its [43] P. Adam´ık and S. Bures,ˇ “Experimental evidence for species- role in shaping a foraging niche,” Philosophical transactions of specific habitat preferences in two flycatcher species in their the Royal Society of London Series B, vol. 366, no. 1567, pp. hybrid zone,” Naturwissenschaften, vol. 94, no. 10, pp. 859– 969–977, 2011. 863, 2007. [26] A. Qvarnstrom,¨ A. M. Rice, and H. Ellegren, “Speciation in [44] G. P. Sætre, M. Kral,´ S. Bures,andR.A.Ims,“Dynamicsofaˇ Ficedula flycatchers,” Philosophical Transactions of the Royal clinal hybrid zone and a comparison with island hybrid zones Society B, vol. 365, no. 1547, pp. 1841–1852, 2010. of flycatchers (Ficedula hypoleuca and F. Albicollis),” Journal of [27] G. P. Sætre, T. Borge, J. Lindell et al., “Speciation, introgressive Zoology, vol. 247, no. 1, pp. 53–64, 1999. hybridization and nonlinear rate of molecular evolution in [45] J. H. Van Balen, “Observations on the post-fledging dispersal flycatchers,” Molecular Ecology, vol. 10, no. 3, pp. 737–749, of the pied flycatcher, Ficedula hypoleuca,” Ardea, vol. 67, pp. 2001. 134–137, 1979. [28] T. Veen, N. Svedin, J. T. Forsman et al., “Does migration of [46] R. Berndt and W. Winkel, “Transfer-experiments on problems hybrids contribute to post-zygotic isolation in flycatchers?” of imprinting to the birthplace in the Pied Flycatcher Ficedula Proceedings of the Royal Society B, vol. 274, no. 1610, pp. 707– hypoleuca,” Journal of Ornithology, vol. 120, no. 1, pp. 41–53, 712, 2007. 1979. [29] A. Lundberg and R. V. Alatalo, The Pied Flycatcher,Poyser, [47] C. Burger and C. Both, “Translocation as a novel approach to London, UK, 1992. study effects of a new breeding habitat on reproductive output [30] T. Part¨ and A. Qvarnstrom,¨ “Badge size in collared flycatchers in wild birds,” PLoS ONE, vol. 6, no. 3, article e18143, 2011. predicts outcome of male competition over territories,” Ani- [48] M. F. Benard and S. J. McCauley, “Integrating across life- mal Behaviour, vol. 54, no. 4, pp. 893–899, 1997. history stages: consequences of natal habitat effects on disper- [31] A. Qvarnstrom, “Experimentally increased badge size sal,” American Naturalist, vol. 171, no. 5, pp. 553–567, 2008. increases male competition and reduces male parental care in [49] L. Lens and A. A. Dhondt, “Effects of habitat fragmentation the collared flycatcher,” Proceedings of the Royal Society B, vol. on the timing of crested tit Parus cristatus natal dispersal,” Ibis, 264, no. 1385, pp. 1225–1231, 1997. vol. 136, no. 2, pp. 147–152, 1994. [32] A. Qvarnstrom,¨ C. Wiley, N. Svedin, and N. Vallin, “Life- [50] R. A. Duckworth and L. E. B. Kruuk, “Evolution of genetic history divergence facilitates regional coexistence of compet- integration between dispersal and colonization ability in a ing Ficedula flycatchers,” Ecology, vol. 90, no. 7, pp. 1948–1957, bird,” Evolution, vol. 63, no. 4, pp. 968–977, 2009. 2009. [51] T. Slagsvold and K. L. Wiebe, “Learning the ecological niche,” [33] C. Wiley, N. Fogelberg, S. A. Sæther et al., “Direct benefits Proceedings of the Royal Society B, vol. 274, no. 1606, pp. 19–23, and costs for hybridizing Ficedula flycatchers,” Journal of 2007. Evolutionary Biology, vol. 20, no. 3, pp. 854–864, 2007. [52] N. Chernetsov, L. V. Sokolov, V. Kosarev et al., “Sex-related [34] A. Qvarnstrom,¨ N. Svedin, C. Wiley, T. Veen, and L. Gustafs- natal dispersal of Pied Flycatchers: how far away from home?” son, “Cross-fostering reveals seasonal changes in the relative Condor, vol. 108, no. 3, pp. 711–717, 2006. fitness of two competing species of flycatchers,” Biology Letters, [53] B. Tonnis, P. R. Grant, B. R. Grant, and K. Petren, “Habitat vol. 1, no. 1, pp. 68–71, 2005. selection and ecological speciation in Galapagos´ warbler [35]T.Veen,B.C.Sheldon,F.J.Weissing,M.E.Visser,A. finches (Certhidea olivacea and Certhidea fusca),” Proceedings Qvarnstrom,¨ and G. P. Sætre, “Temporal differences in food of the Royal Society B, vol. 272, no. 1565, pp. 819–826, 2005. abundance promote coexistence between two congeneric [54] J. Felsenstein, “Skepticism towards Santa Rosalia, or why are passerines,” Oecologia, vol. 162, no. 4, pp. 873–884, 2010. there so few kinds of animals,” Evolution, vol. 35, pp. 124–138, [36] N. Vallin, A. M. Rice, R. I. Bailey, A. Husby, and A. 1981. Qvarnstrom,¨ “Positive feedback between ecological and repro- [55] R. F. Lachlan and M. R. Servedio, “Song learning accelerates ductive character displacement in a young avian hybrid zone,” allopatric speciation,” Evolution, vol. 58, no. 9, pp. 2049–2063, Evolution. In press. 2004. [37] N. Svedin, C. Wiley, T. Veen, L. Gustafsson, and A. Qvarn- strom,¨ “Natural and sexual selection against hybrid flycatchers,” Proceedings of the Royal Society B, vol. 275, no. 1635, pp. 735–744, 2008. [38] A. Qvarnstrom,¨ J. V. Kehlenbeck, C. Wiley, N. Svedin, and S. A. Sæther, “Species divergence in offspring begging intensity: difference in need or manipulation of parents?” Proceedings of the Royal Society B, vol. 274, no. 1612, pp. 1003–1008, 2007. Hindawi Publishing Corporation International Journal of Ecology Volume 2011, Article ID 942847, 15 pages doi:10.1155/2011/942847

Review Article Sympatric Speciation in Threespine Stickleback: Why Not?

Daniel I. Bolnick

Howard Hughes Medical Institute; Section of Integrative Biology, University of Texas at Austin, One University Station C0930, Austin, TX 78712, USA

Correspondence should be addressed to Daniel I. Bolnick, [email protected]

Received 3 May 2011; Accepted 25 June 2011

Academic Editor: Andrew Hendry

Copyright © 2011 Daniel I. Bolnick. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Numerous theoretical models suggest that sympatric speciation is possible when frequency-dependent interactions such as intraspecific competition drive disruptive selection on a trait that is also subject to assortative mating. Here, I review recent evidence that both conditions are met in lake populations of threespine stickleback (Gasterosteus aculeatus). Nonetheless, sympatric speciation appears to be rare or absent in stickleback. If stickleback qualitatively fit the theoretical requirements for sympatric speciation, why do they not undergo sympatric speciation? I present simulations showing that disruptive selection and assortative mating in stickleback, though present, are too weak to drive speciation. Furthermore, I summarize empirical evidence that disruptive selection in stickleback drives other forms of evolutionary diversification (plasticity, increased trait variance, and sexual dimorphism) instead of speciation. In conclusion, core assumptions of sympatric speciation theory seem to be qualitatively reasonable for stickleback, but speciation may nevertheless fail because of (i) quantitative mismatches with theory and (ii) alternative evolutionary outcomes.

1. Introduction sympatric speciation are so well known [1, 2] that many biologists take it for granted that sympatric speciation is The feasibility and prevalence of sympatric speciation have unlikely. If so, why bother explaining a specific example in been in contention since the birth of evolutionary biology which sympatric speciation fails? The answer is that we might [1, 2]. Recent empirical work on sympatric speciation has find sympatric speciation fails for unexpected reasons. focused on documenting a few likely examples [3–6], or estimating the relative frequency of various geographic modes of speciation [7]. Concurrently, theory has focused on determining which conditions are necessary and sufficient 2. Theoretical Background for sympatric speciation to occur [8–11]. Unfortunately, Theory has revealed two key underpinnings of sympatric these two related research programs rarely intersect. Namely, speciation. First, a negative frequency-dependent process there is a need for models based on empirically measurable such as intraspecific competition is required to generate parameters, preferably tailored to the natural history of specific case studies (e.g., [12, 13]), and for empirical persistent disruptive selection [15–18], which splits the estimates of key parameters in such models. Such fusions population into phenotypically divergent groups. Second, of empirical data and theory will provide more biologically assortative mating is required to generate reproductive realistic insights into when or why sympatric speciation isolation that maintains those divergent groups in the face might succeed or fail, and thus explain its frequency. of recombination [19, 20].Thereisabroadconsensusfrom In this paper, I attempt such a fusion, by comparing existing theory that sympatric speciation is easiest when simulation results to empirically derived parameter estimates disruptive selection and assortative mating act on the same from threespine stickleback (Gasterosteus aculeatus). The trait (or closely correlated traits) [11]. Such pleiotropic traits goal is to understand why sympatric speciation is rare or have been dubbed “magic traits” for their uniquely favorable absent in this organism [14] despite a qualitative fit with role in speciation [21], because they prevent recombination some key requirements for speciation. The objections to from decoupling the target of mate choice and selection. 2 International Journal of Ecology

80 8.5 r =−0.141 P = 0.0019 60 8

7.5 40 N 15 δ

Frequency 7 20 6.5

0 6 −3 −2 −1 0 1 2 3 −32 −30 −28 −26 −24 −22 −20 −18 Size-standardized gill raker length, PCA2 δ13 C (a) (b)

1 2

1 0 0 −1 −1 Isotopic PC1 −

Egg isotopes, PC1 2 −2 r = 0.298 r = 0.35 P<0.0001 P = 0.025 −3 −3 −3 −2 −1 0 1 2 3 0.5 1 1.5 2 Size-standardized gill raker length, PCA2 Male isotopes, PC1 (c) (d)

Quadratic coeﬃcient = 0.102 3 P = 0.0005

0 Relative growth rate, RNA/DNA −2 −1 0 1 2 3 4 Size-standardized gill raker length, PCA2 (e)

Figure 1: Illustration of the evidence for phenotypic and diet variation, phenotype-diet correlation, disruptive selection, and assortative mating in stickleback. (a) Phenotypic variation in gill raker length within a population of stickleback (First Lake, 2008; [22]). (b) Diet variation within a population of stickleback (First Lake, 2008; [22]) indicated by the high variance in carbon and nitrogen stable isotope ratios δ13Candδ15N compared to the variance observed when individuals all consume the same set of prey (thick cross over the centroid represents ±2 s.d., based on lab-raised stickleback [23]). Adjusting for baseline isotopic variation (thin arrow indicates a benthic primary consumer [snails], thick arrow represents a planktivorous primary consumer [mussels]), individual stickleback fall anywhere from 0% to 100% benthic carbon. Because of the correlation between δ13Candδ15N, the first isotopic principal component axis can be used to represent the axis from benthic to limnetic diets. (c) Phenotypic and diet variation are correlated: fish with longer gill rakers tend to consume a more limnetic diet (high values of isotopic PC1 indicate lower δ13C, higher δ15N; First Lake). (d) Assortative mating by diet, indicated by a positive correlation between males’ isotopic signature and the isotopes from eggs (indicative of female diet; modified from [23]). (e) A key measure of trophic morphology, gill raker length, is subject to disruptive selection as indicated by a significant quadratic coefficient in a regression of growth rate on size-standardized gill raker length. (First Lake, 2005; modified from [24]). Growth rate was measured using a biochemical index, the ratio of RNA to DNA in muscle tissue. The relationships in each panel of this figure are repeatable across multiple solitary lake populations. International Journal of Ecology 3

Skeptics of sympatric speciation typically adopt several mates who are more (or less) phenotypically similar to them- objections to this theory [1, 11, 20, 25], arguing that (i) selves than expected by chance. The result is a phenotypic (or models often assume excessively high initial genetic variance genetic) correlation between mated males and females, with and/or high mutation rates, (ii) disruptive selection is likely respect to one or more traits expressed in both sexes. This to be dynamically unstable (populations evolve away from correlation is easily measured by quantifying some trait in fitness minima towards regions of stabilizing selection), and both sexes across multiple mated pairs. The magnitude and therefore, disruptive selection should be rare, (iii) costs to sign of this correlation provides a measure of the strength mate choice will select against assortative mating, and (iv) and direction of assortative mating. One caveat is that “magic traits” subject to both selection and assortment are assortative mating might act on multiple traits concurrently, rare. Finally, casual readers of the literature on speciation in which case multivariate canonical correlations might be theory often come away with the impression that sympatric more appropriate. Studies focusing on one or a few traits speciation is all but inevitable when disruptive selection might thus underestimate assortative mating by overlooking and assortative mating jointly act on a magic trait. In fact, one or more important traits. With these caveats in mind, speciation can only occur if these evolutionary forces are a recent meta-analysis established that there is a broad strong. Otherwise, the population may evolve higher trait tendency towards weak positive assortative mating within variance, possibly even a bimodal phenotypic distribution ostensibly panmictic populations (mean r = 0.23; [33]). with some assortative mating, but never reach full speciation Note that these positive correlations can arise from a in the sense of forming strongly reproductively isolated and variety of mate choice processes, including a preference for phenotypically nonoverlapping populations [8]. Thus, the phenotypically similar mates (as assumed in many sympatric core theoretical conflict between skeptics and opponents of speciation models), spatial sorting of phenotypes at the time sympatric speciation concerns the biological realism of core of mating, allochrony, or directional mate preferences in both assumptions, and thus can be best addressed by answering sexes. the following empirical questions. (iv) How often Do Disruptive Selection and Assortative Mating (i) Do Single Populations Typically Exhibit Substantial Genetic Affect the Same Trait? If disruptive selection and positive Variance in Ecologically Relevant Traits and/or Mating Strat- assortative mating are both moderately common, it is likely egy? There is no doubt that many populations harbor that they sometimes act concurrently in a given species, substantial variation for ecologically relevant traits such as possibly on the same (or correlated) traits. Although exam- resource use. Within many populations, discrete morpho- ples of “magic traits” have been documented (particularly in types or quantitative trait variation give rise to heritable incipient species pairs; e.g., [5, 34, 35]), we do not know their among-individual differences in prey preferences [26, 27]. general frequency in nature, particularly within populations As a result, individuals typically consume a small subset of prior to speciation. Thus, at present, we must rely on insights their population’s collective diet (across 142 measures of diet obtained from specific case studies. For example, disruptive variationfrom35species,therewasanaverageofonly45% selection and assortative mating have been shown to act on similarity between individual and population diets [28]). It the same traits in Darwin’s finches [36, 37] and in stickleback is less clear how often populations harbor heritable variation [23, 24]. In the remainder of this paper, I focus on the latter for the degree of choosiness in assortative mating though as a case study. such variation has been documented [29]. (v) When They Coincide, Are Disruptive Selection and (ii) How Common Is Disruptive Selection? One can measure Assortative Mating Strong Enough to Drive Speciation? To disruptive and stabilizing selection by quadratic regression date, this question has remained essentially unanswered for of a fitness measure against a phenotype [30]. Positive two reasons. First, there are few species for which both ffi γ quadratic coe cients ( ) imply disruptive selection if the processes are known to co-occur within a single population. fitness minimum lies within the range of extant phenotypes. Second, theoretical models are usually constructed using A meta-analysis of selection gradients in nature found hard-to-measure parameters, making it hard to evaluate ffi that positive and negative quadratic coe cients were about what constitutes (un)realistically strong parameter values. equally common and of similar magnitude (though both Threespine stickleback (Gasterosteus aculeatus, a small |γ|∼ . were weak: average 0 1[31]). This suggests that north-temperate fish) is one of the few organisms known to disruptive selection is more widespread than often believed, experience both disruptive selection and assortative mating perhaps because it is stabilized by negative frequency- on a shared trait within single populations (see also studies dependent interactions that prevent evolution away from of Darwin’s finches [36, 37]). In the remainder of this fitness valleys towards fitness peaks [15, 32]. One caveat is paper, I review recent work supporting this claim, focusing that it is not known how many cases of positive curvature on processes occurring in single-species populations of entail a real fitness minimum within the phenotypic range, stickleback that represent plausible precursors to speciation as opposed to just a monotonic but curved fitness function. (Figure 1(a); as opposed to the well known stickleback species pairs [38]). However, sympatric speciation is rare or (iii) How Common Is Assortative Mating within Populations? absent in stickleback [14], raising the question, “Why has Assortative mating is the tendency for individuals to choose sympatric speciation failed to occur?” To answer this model, 4 International Journal of Ecology

I combine the reviewed empirical data with a new numerical for exploitative competition. Accordingly, experimentally simulation to show that disruptive selection and assortative elevated stickleback density in field enclosures suppressed mating in stickleback are too weak to drive speciation. prey density, reducing stickleback stomach content mass, Furthermore, I review empirical evidence that fluctuating growth, and reproductive investment [48, 49]andaltered selection and alternative forms of diversification further growth rates in seminatural ponds [50–52]. These results inhibit speciation. Thus, many of the common objections to suggest that intraspecific competition has the potential sympatric speciation do not hold in this system, but other to significantly impact stickleback. However, further work quantitative and qualitative hurdles remain, many of which is required to provide direct proof of density-dependent are not widely incorporated into theoretical models. population regulation in unmanipulated settings.

3. Diversifying Forces within Lake Stickleback (ii) Individual Specialization. As a species, stickleback are ecological generalists, consuming a wide diversity of inverte- 3.1. Disruptive Selection. Theory predicts that stable disrup- brate prey including crustacean zooplankton, mites, aquatic tive selection can arise from frequency-dependent interac- and terrestrial insects, and even other stickleback. This tions such as intraspecific competition for heterogeneous is often characterized as a bimodal resource base, com- resources [16]. This resource heterogeneity is most pro- prised large benthic macroinvertebrate prey and smaller nounced when the population uses two distinct resource limnetic zooplankton [38]. In reality, the resource base is types (e.g., benthic and limnetic prey [39], hard and soft more complex and can be described using a continuous seeds [36, 40], etc.) but can also apply to multimodal [41] distribution of prey body sizes: within solitary stickleback or unimodally varying resources [42]. When co-occurring populations, individuals’ mean prey size conforms to a individuals use divergent subsets of the available resources, unimodal distribution [47]. Individuals also specialize more the level of competition experienced by a given individual finely on prey taxa [41],andonmicrohabitattype[53], depends on the density of like phenotypes that consume than the benthic/limnetic distinction would imply, though, similar resources, rather than the total population density. the benthic/limnetic axis remains a convenient and fairly Phenotypically average individuals experience stronger com- reasonable simplification. petition than rare extreme phenotypes, and thus suffer lower Within a population, any given individual tends to be relative fitness (disruptive selection). Note that competition relatively specialized compared to its population as a whole is only one of several ecological interactions that can generate [54]. For example, direct observation of foraging individuals the necessary frequency dependence [43]. However, some in Little Mud Lake (Vancouver Island) showed that some form of frequency dependence is necessary to stabilize individuals consistently directed more than 90% of their the disruptive selection; for example, a bimodal resource attacks at benthic prey, whereas others used 70% mid-water distribution cannot generate sustained disruptive selection prey, or else specialized on surface prey [55]. Accordingly, by itself, because the population will evolve to specialize on within a lake stomach contents reveal among-individual diet whichever resource type conveys higher fitness. Negative- variation ranging from mostly benthic to mostly limnetic frequency dependence arises because this transition to prey [41]. This stomach content variation is corroborated specializing on one resource leads to reduced availability of by a longer-term diet measure based on stable isotope that resource, favoring an evolutionary shift back towards variation among individuals, which reveals that co-occurring generalist phenotypes that (inefficiently) use both resources. individuals vary from 0% to 100% dietary benthic carbon. Here, I focus on competitive disruptive selection, because This isotopic variability is far greater than we expect when this is the most widely modeled source of frequency individuals feed on the same prey in the same proportions dependence and the best documented in stickleback. For (Figure 1(b);[54]). We observe this diet heterogeneity even this disruptive selection to occur, a population must exhibit among individuals held in small (10 m2) enclosures that (i) high population density leading to resource limitation, ensure individuals have equal access to the same prey [56]. (ii) among-individual variation in use of a diverse set of It is apparent that diet variation reflects among-individual resources (known as “individual specialization”; [26]), and differences in prey preferences and/or acquisition abilities, (iii) a measurable phenotype correlated with this resource rather than coarse-grained differences in prey availability due use variation, which is thus subject to frequency-dependent to spatial separation of foraging individuals. competition. In the following paragraphs, I review the exist- Some of the diet variation is attributed to morphological ing evidence from stickleback for each of these conditions differences among individuals (Figure 1(c)). Within any and the resulting disruptive selection. given lake, individuals with relatively long gill rakers typically have more pelagic zooplankton in their stomachs and have (i) Intraspecific Competition. Numerous laboratory studies correspondingly limnetic carbon and nitrogen stable isotope confirm that food ration strongly affects stickleback growth, signatures [23, 54, 57–59]. Gill raker number, body size, body female egg production and spawning frequency, and male shape, jaw lever ratios, hyoid length, and buccal volume have reproductive success [44–46]. Stickleback kept at approxi- also been found to be correlated with diet variation [41]. mately natural density in field enclosures reduced benthic Importantly, several of these trophic traits have been shown invertebrate density by 54% [47], implying that stickle- to exhibit heritable variation [60–63], and thus should be back exert top-down control of prey density as required capable of responding to selection. International Journal of Ecology 5

(iii) Frequency Dependence. As a result of individual specialization, not all individuals within a population use the P = 0.022 same resources. Within a typical population of stickleback length any pair of individuals only share an average of 30% of Gill raker their prey in common [41]. Because diet is correlated with morphology, this diet overlap is even lower between mor- P = 0.015

phologically divergent individuals: in one lake, diet similarity number was twice as high for the most morphologically similar Gill raker Stabilizing Disruptive individuals, compared with the most dissimilar individuals −0.1 −0.05 0 0.05 0.1 0.15 0.2 [54]. Consequently, the level of competition experienced by Quadratic selection gradient an individual depends not on total population density but on the abundance of individuals with overlapping diets (e.g., Figure 2: Lake populations exhibit a tendency towards disruptive ffi similar morphology). Intraspecific competition should thus selection (positive quadratic selection coe cients). Box plots favor individuals with rare feeding strategies over individuals represent the mean (thick vertical lines), 50th quartiles (boxes), and range (whiskers) of quadratic selection gradients for gill raker with common strategies who have proportionally more length and gill raker number from 14 populations, represented as competitors. Direct evidence for such frequency-dependent individual dots [24]. P values are provided for t-tests of whether competition comes from experiments in artificial ponds. the mean quadratic selection gradient is significantly different from Schluter [52] manipulated the relative frequency of benthic zero. The values are modified from [24],whoreportedthequadratic and limnetic species pairs and their hybrids and found that coefficients γ whereas disruptive selection should be measured as 2γ a given phenotype experienced reduced growth when it was as shown here. the more common form. Although the equivalent study has not yet been done within a single-species population of stickleback, it is reasonable to assume that similar processes operate. Within-population frequency dependence may of equal bination of these resources. Consequently, we expect course be weaker, given the lower phenotypic variance disruptive selection to be more pronounced in intermediate- compared with species-pair lakes [38]. sized lakes. Indeed, disruptive selection seems to be The joint occurrence of competition and diet varia- maximized in intermediate-sized lakes between 40 to 60 ha tion should drive frequency-dependent disruptive selection. surface area (there is a significant quadratic relationship Accordingly, a survey of multiple solitary populations of between lake area and the quadratic selection coefficient stickleback found a widespread tendency towards disruptive γ [24]). Disruptive selection is also modified by other selection on two uncorrelated trophic traits (gill raker length ecological interactions. In experimental ponds, disruptive and number; [24]). On average, phenotypically intermediate selection was exaggerated in the presence of predatory trout individuals (of either sex) grow more slowly (Figure 1(e)), [50, 64]. The effect of interspecific competition is not well attain smaller body size and invest proportionally less mass understood, but depending on the species, fish predators can in reproductive tissue than individuals at either extreme of a increase or decrease the diet variation required for disruptive given trait. The curvature of the estimated fitness landscape selection [65]. averaged γ ∼ 0.06 (Figure 2), somewhat weaker than the average quadratic curvature found in Kingsolver et al.’s meta-analysis [31]. Artificially elevated stickleback density 3.2. Assortative Mating. Positive assortative mating occurs led to stronger-than-average disruptive selection, whereas when mated pairs are phenotypically more similar than reduced density eliminated disruptive selection altogether expected by chance. There is abundant evidence that the [48]. These results suggest that intraspecific competition benthic and limnetic species pairs exhibit premating repro- generates disruptive selection in many lake populations of ductive isolation [66–70], as do some ecologically and phe- stickleback. However, an important caveat is that that none notypically divergent allopatric populations [71, 72] though of the fitness metrics (adult growth rate and variation in other divergent stickleback populations show no isolation reproductive tissue mass) represents total lifetime fitness. [73]. Remarkably, assortative mating also occurs within soli- Consequently, although intermediate phenotypes grow more tary stickleback populations. Stable isotope analyses reveal slowly as adults and invest less mass in gonads, it is possible that within a single population, males with more benthic that additional components of fitness (larval or juvenile isotope signatures had mated with more benthic females growth, survival, mating success, breeding duration, etc.) (r = 0.507; female diet inferred by isotope signature of their might overwhelm the component of selection arising from eggs; Figure 1(d);[23]). Unlike many studies of reproductive foraging success as measured in the studies cited above. isolation between divergent populations, typically studied in Becauselifetimefitnessiseffectively impossible to track laboratory aquaria, this intraspecific assortment is measured for individual stickleback of known morphology, our best in the field with wild individuals, so it represents a more available estimates of selection are likely to be incomplete. natural setting. The positive assortative mating represents Even taking these selection estimates at face value, a “magic trait” system, because individuals pair on the disruptive selection is not ubiquitous. Large or small basis of a trait (isotope ratio) which is correlated with the lakes are dominated by single habitat types (limnetic or ecomorphological traits under disruptive selection [23, 24]. benthic), whereas intermediate-sized lakes contain a more In fact, individuals’ foraging strategy (reflected in isotopes) 6 International Journal of Ecology may be the most direct target of selection, which results in taxon cycle seems unlikely, as the examples of recent species correlated selection on trophic morphology [22]. pair collapse entail anthropogenic environmental change At present, the mechanism underlying within-popula- [82, 83]. tion assortative mating in stickleback is unknown. Assor- A few additional pairs of lacustrine sympatric “morphs” tative mating may arise indirectly if different phenotypes have been described outside of British Columbia [80]. In are spatially segregated during mating, as has been shown Alaska, Benka Lake stickleback caught in different habitats for the benthic and limnetic species pairs [70, 72]. Thus, are morphologically divergent [84]. However, these ecotypes microhabitat choice could contribute to within-population do not exhibit the bimodal trait distribution seen in species assortative mating. A spatially explicit follow-up study of pair lakes, nor is it known whether they are genetically dis- isotopic assortative mating [74] (in a second lake population) tinct populations. Similar divergence patterns are observed confirmed the previous positive correlation [23]. Males and between substrate habitats within several lakes in Iceland females were distributed nonrandomly in space; isotopically [85]. This divergence can evolve rapidly: damming of a fjord more limnetic individuals tended to nest deeper and closer in Iceland in 1987 led to rapid divergence from the marine to vegetation than benthic individuals (nests ranged from ancestor, and differentiation among habitats [86]. However, 0.5 to 2 m deep) [74]. However, isotopic assortative mating these ecomorph pairs do not yet exhibit the near-complete remained even after statistically controlling for spatial and reproductive isolation and genetic divergence that typified habitat structure, implying a role for active mate preference. the British Columbian species pairs. Direct preferences for similar mates might involve morphological traits that are correlated with diet (e.g., size, body shape, and color) or cues derived directly from the prey. 4.2. How Did the Few Species Pairs Arise? Not only are Some evidence is available for the latter effect: fish exper- species pairs rare, but there is reason to doubt they arose via imentally fed one food type for two weeks, subsequently sympatric speciation. Two alternative models have been pro- exhibited a preference to shoal with conspecifics fed the posed for the geography of speciation in British Columbian same food [75]. This shoaling preference was maintained species pairs [87, 88]. The first invokes standard sympatric as long as the focal individuals received olfactory cues from speciation: marine stickleback colonized the postglacial lakes the potential social groups. Similar results seem to hold ∼12,500 years ago, after which increased lake elevation due when studying females’ mate preferences instead of just social to isostatic rebound prevented recurrent colonization [87]. affiliation [76]. Olfactory cues have also been implicated in Subsequently, disruptive selection may have driven ecological MHC-based mate choice in stickleback [71, 77]thoughitis divergence, which in turn led to reproductive isolation based unclear to what extent this might interact with diet-induced on size, color, and behavior. Reproductive isolation may have assortment. Regardless of the exact mechanism underlying been a pleiotropic effect of trait divergence [66, 68, 89], assortative pairing, the ultimate effect will be to maintain supplemented by reinforcement [69]. linkage disequilibrium among genes underlying ecologically The alternative model invokes a brief period of allopatry diverging traits, thus facilitating speciation. [87]. The initial colonization by marine fish led to rapid adaptation into a generalist that may have resembled today’s solitary populations. A subsequent brief sea level rise [90] 4. Does Sympatric Speciation was thought to have introduced a second population of Occur in Stickleback? marine stickleback, which underwent ecological and reproductive character displacement with the native generalist 4.1. Rarity of Species Pairs. Despite disruptive selection and form to produce, respectively, a limnetic and benthic species. assortative mating acting on correlated traits, stickleback This model is neither purely allopatric nor sympatric, since do not appear to undergo sympatric speciation with any some component of reproductive isolation evolved in both great frequency, if at all [14]. Stickleback occur in north contexts at different times. However, the critical initial step temperate coastal watersheds throughout the Atlantic and of divergence was allopatric. Pacific [78]. On Vancouver Island alone, they inhabit many For a brief period in the 1990s, phylogeographic data hundreds of lakes that were all likely colonized shortly based on mitochondrial DNA sequences lent support to the after deglaciation. Despite this plethora of similar-aged lake sympatric model of stickleback speciation, because species populations regionally and globally, only a handful of species pairs for each lake were typically reciprocally monophyletic pairs have been described. In British Columbia, species [88]. However, subsequent analysis of microsatellites told pairs have been documented in seven lakes (Balkwill, Enos, adifferent story: each ecotype was most closely related to Hadley, Paxton, Priest, Emily, and Little Quarry Lakes) in marineformsortomorphsinotherlakes[91] though five watersheds from four islands [38, 79, 80]. Collectively, phylogenetic resolution remained poor. The monophyly of researchers studying stickleback in the American and Cana- mtDNA sequences may thus be the result of introgression dian northwest have surveyed hundreds of lakes, suggesting following secondary contact. Additional evidence for the that species pairs occur at most in a few percent of inhabited allopatric model comes from the fact that limnetic stick- lakes. That said, it is likely that additional species pairs leback are more tolerant of salinity [92] and genetically remain to be discovered: a new pair was described as recently more related to marine forms than are benthic stickleback as 2008 [79]. Also, it is possible that species pair lakes arise [91]. Also, all the species pair lakes are geographically more frequently but commonly collapse [81, 82]. Such a nearby and at similar low elevation, suggesting that some International Journal of Ecology 7 unique localized geological event facilitated speciation. Con- sequently, the case for an allopatric phase of divergence seems fairly strong. However, newer geological data does not support the required secondary sea level rise [93], making 30 it hard to explain how a secondary introduction could have happened. Thus, biological data remains most consistent with the secondary contact scenario though the mechanism 20 of secondary colonization is unclear (Schluter, pers. comm.).

The question may be unresolvable unless perhaps ancient Phenotype DNA analyses of subfossil stickleback bones in sediment 10 could reveal evidence for a double invasion.

5. It Is Not That Easy: Simulations with 0 Empirical Parameters 20 40 60 80 Generation The absence (or at best, rarity) of sympatric speciation in Figure 3: An illustration of the simulated dynamics of the lake stickleback despite disruptive selection and assortative phenotype distribution over time, using parameter values that mating raises the question, “why do stickleback apparently lead to rapid speciation. Darker shading indicates greater numbers not experience sympatric speciation?” Several of the standard of individuals for a given phenotype at a given time. Parameter theoretical objections to sympatric speciation (Section 2, values are population size = 500, number of loci = 40, initial allele γ = . r = above) do not apply, as there is substantial polymorphism in frequency = 0.5, 2 0 4, and assortative mating correlation . resource use, disruptive selection, and assortative mating on 0 7. the ecological trait, as described above. The simplest remaining hypothesis is that disruptive selection and/or assortment, while present, are too weak to complete speciation [8, 94, conducive to speciation). Note that my simulations focused 95], or that some additional evolutionary processes impose on just these parameters and omits extensive consideration additional constraints on reproductive isolation. of other potentially relevant factors such as the genetic To test whether the observed selection and assortment architecture of the trait, population size, mutation rate, and are sufficient to drive speciation in a best case scenario, costs to choosiness. I performed stochastic numerical simulations of the joint Consistent with many previous theoretical models and effect of disruptive selection and assortative mating acting on simulations [97], I found that sympatric speciation was a quantitative trait based on the additive effects of numerous possible given the joint action of strong disruptive selection small-effect loci (see the Appendix for model details). Unlike and assortative mating (Figure 3). By “strong” I mean most previous models of sympatric speciation, the model quadratic selection coefficients of γ>0.25; empirically, used here focused on the effect of empirically measur- γ follows a double exponential distribution and ranges able parameters (but see the empirically driven papers between about −1to1,withmostvaluesbetween−0.1 by Gavrilets and colleagues [12, 13]). Here, I focus on to 0.1 [31]. Speciation required strong positive assortative asking whether the quadratic selection gradients (γ)and mating, defined here as a trait correlation above about 0.6; the phenotypic correlations between mates (r)measuredin empirically r ranges from below −0.9 to above 0.9, with stickleback are sufficient to split the phenotypically unimodal ameanof0.23[33]. However, if either evolutionary force population into a bimodal distribution of two reproductively is weak or moderate (e.g., average values of γ ∼ 0.1and isolated populations. This focus on present-day parameter r ∼ 0.25), then speciation does not occur (Figure 4). Instead, values sets aside the question of whether disruptive selection populations may either remain unimodal but with inflated would drive increased assortative mating in the future. variance (compare Figure 5(a) with 5(b)), or may attain a The reinforcement-like process often examined in sympatric uniform or even bimodal distribution (Figure 5(c))inwhich speciation models [8, 9, 11, 96] remains empirically untested, assortative mating is too weak to prevent the continual and and so, it is not studied here explicitly. frequent recreation of intermediate phenotypes [8, 94, 98] Replicate simulations were run for a range of empirically but maintains some linkage disequilibrium that substantially plausible values of disruptive selection and assortative mat- inflates trait variance. Notably, however, sympatric specia- ing to determine what parameter combinations are required tion (Figure 5(d)) can be expected to occur with empirically for speciation. Speciation was said to have occurred if the reasonable (albeit strong) parameter values. initially unimodal phenotype distribution became strongly Empirically estimated values of disruptive selection and bimodal, indicated by two major peaks separated by a assortative mating in stickleback fall well within the param- trough less than 5% as abundant as the lowest mode. The eter space where speciation is never observed (Figure 4). The empirical measures from stickleback (reviewed above) were strongest observed disruptive selection would be sufficient compared with the simulation results to determine whether for speciation, if assortative mating were twice as strong as solitary stickleback populations could plausibly be expected the observed average. The strongest observed assortative to speciate sympatrically (e.g., fell within the parameter space mating would be sufficient if disruptive selection were seven 8 International Journal of Ecology

60 sd(x) = 3.14 40 0.4 20

Speciation Frequency 0 )

γ 0 10 20 30 40 0.3 No speciation (a) 50 sd(x) = 4.19 30 0.2

Frequency 10 Disruptive selection ( 0 0 10 20 30 40 0.1 (b) x = 50 sd( ) 10.41 0.2 0.4 0.6 0.8 30 Assortative mating (r)

Frequency 10 Figure 4: Speciation is only observed when there is strong 0 disruptive selection and strong assortative mating (shaded region). 0 10 20 30 40 For comparison, the crossed lines indicates the range and mean (c) (intersection) of these parameters measured in single-species lake populations of stickleback [23, 24]. Parameter values are population sd(x) = size = 500, number of loci = 40, and initial allele frequency 80 14.43 = 0.5. Disruptive selection and assortative mating were varied 40 γ = r = systematically from 0to0.5and 0 to 0.95, in increments of Frequency 0.025. This parameter space falls well within empirically justifiable 0 values based on meta-analyses [31, 33]. Ten replicate simulations 0 10 20 30 40 were run for each factorial combination of parameters. Dark Phenotypic value shading indicates regions of parameter space in which speciation was observed at least once. (d) Figure 5: An illustration of various outcomes of disruptive selection and assortative mating, depending on parameter values. times stronger than average (or more than twice the maxi- Each histogram represents the phenotype frequency distribution at mum). Even the joint effect of the empirical maximum dis- a quasiequilibrium after 1,000 generations. Longer simulation runs ruptive selection and assortative mating would be insufficient do not qualitatively change the phenotype distribution for a given > parameter combination. (a) With no disruptive selection and no for sympatric speciation (in simulations run for 50,000 assortative mating (γ = 0, r = 0), phenotypic variance follows generations). It is thus clear that solitary populations of a binomial (approximately normal) distribution whose variance stickleback are unlikely to undergo sympatric speciation in depends on the level of polymorphism at the 40 additive loci, their current ecological setting. This is a reassuring result in maintained by mutation-drift balance. (b) Using a combination that it may explain why solitary populations have remained of parameters close to the strongest disruptive selection and solitary. Of course, this does not rule out the possibility that assortment observed in stickleback (2γ = 0.15, r = 0.5), the the few stickleback species pairs arose from populations with population attains modestly higher phenotypic variance than in exceptionally high levels of one or both parameters. (a) but remains unimodal. (c) Larger parameter values (2γ = 0.3, One caveat regarding these simulation results is that r = 0.6) rapidly lead to a weakly bimodal phenotype distribution I omit the possibility that assortative mating itself may that does not resolve itself into separate species even after 50,000 evolve. Numerous models suggest that disruptive selection generations. (d) Speciation can occur when disruptive selection and assortment are near the upper end of plausible empirical values can drive a reinforcement-like process favoring the evolution (2γ = 0.4, r = 0.8). of increased assortative mating [9, 43] though this possibility remains contentious [25, 99]. Thus, it is conceivable that stickleback populations experiencing the strongest levels of disruptive selection could evolve stronger mate preferences in stickleback may or may not shift to elsewhere in parameter the future. This increased choosiness depends on a variety of space. Although I do not explicitly model the evolution of additional assumptions about the behavioral basis of assorta- assortment, it is possible to infer the result of a hypothet- tive mating, the genetic architecture of mate choice, costs to ical increase in female choice, which amounts to moving mate choice, mating system, and so on. Lacking information horizontally right in the parameter space shown in Figure 4. about these parameters, it seems more appropriate at present For populations at the highest level of disruptive selec- to evaluate whether existing parameter values are sufficient tion (Figure 4), such a rightward shift in parameter space to bifurcate populations, than to speculate over whether could conceivably bring them into parameter space where International Journal of Ecology 9 speciation is possible. However, for the majority of stickle- The most straightforward outcome of disruptive selec- back populations such reinforced assortative mating would tion is increased genetic variance within a phenotypi- still not facilitate speciation. An important subject for future cally unimodal population (Figure 5(b)). Consistent with empirical investigation is whether disruptive selection really this outcome, derived lacustrine populations of stickleback does drive the evolution of increased assortative mating. exhibit higher phenotypic variance (particularly for gill raker traits under disruptive selection) compared to marine 6. It Is Not That Easy: Additional Constraints on stickleback that represent a putative ancestral character state [116]. It is important to note, however, that these Speciation in Stickleback results reflect phenotypic variances, and it is not known to what extent among-population differences in phenotypic If reinforcement could push the most disruptive populations ff into speciation-prone parameter space, one might ask why variances are matched by underlying di erences in genetic this has not actually happened. The answer is that there may variances. be additional constraints on sympatric speciation, many of Disruptive selection due to intraspecific competition can which are not typically considered in speciation models. For also drive the evolution of ecological sexual dimorphism example, increased assortment may be inhibited if selection [115, 117, 118], which can inflate trait variance. This is variable through time, or if the population evolves an intraspecific ecological character displacement occurs under alternative solution to disruptive selection. In the last section the same theoretical conditions as sympatric speciation of this paper, I summarize evidence for both kinds of [115]. However, dimorphism can evolve more quickly than speciation-inhibiting factors in stickleback. assortative mating, reducing disruptive selection to the point where sympatric speciation is inhibited [115] though this antagonism can be relaxed if disruptive selection acts 6.1. Stability of the Fitness Landscape. Theoretical mod- along multiple trait axes [119]. Stickleback exhibit sexual els of speciation typically assume populations are subject dimorphism with respect to diet and associated trophic to persistent disruptive selection despite general evidence morphology such as gill raker length [24, 104, 120, 121]. This for temporal fluctuations in fitness landscapes [100, 101] dimorphism is maximized in intermediate sized lakes where (though some fluctuations can be due to sampling noise). disruptive selection is typically strongest [24], supporting the Of the 14 lakes studied by Bolnick and Lau [24], First notion that disruptive selection leads to dimorphism [115, and McNair Lakes were among the ones with the strongest 119]. However, controlling for lake size, there is a negative disruptive selection (Figure 2). A second sample of these two relationship between present-day disruptive selection and lakes, two years later, found significant stabilizing selection dimorphism [24], implying that once dimorphism evolves in First Lake and directional selection in McNair [22]. the fitness landscape becomes flatter, which may inhibit The reason for this interannual variation is not clear. One speciation [115]. These results seem to confirm the notion hypothesis is that fluctuating population density [102](or that sexual dimorphism can evolve quickly and mitigate dis- resource availability) alters the strength of competition and ruptive selection, thereby potentially reducing the capacity thus the fitness landscape [48]. For instance, a long-term for sympatric speciation. study of a population of Eurasian Perch (Perca fluviatilis) Another possibility is that disruptive selection favors the found shifts in the fitness landscape associated with two- evolution of phenotypic plasticity [122], allowing individuals order-of-magnitude population density cycles. At high den- to shift onto underused resources within a single generation sity, perch experienced disruptive selection, which changed rather than over evolutionary time [110]. Optimal foraging to stabilizing or directional selection during periods of theory (OFT) is predicated on the idea that individuals low density [103]. Changes in selection gradients might are able to flexibly choose prey to maximize their expected also reflect interannual variation in predation regimes [50, fitness [123], accepting lower-value resources as preferred 104–107] or the strength of interspecific competition [65]. Such variable selection regimes would undermine sympatric prey become scarce [124]. Applied to phenotypically variable speciation [108, 109] and might instead favor the evolution populations, OFT predicts that strong competition might of phenotypic plasticity [110]. In fact, temporal changes in lead to behavioral rather than evolutionary diversification the fitness landscape may explain the introgressive collapse [125]. Consistent with this prediction, experimentally ele- of formerly distinct species pairs of stickleback [83]. vated stickleback densities led to greater population niche width and greater among-individual diet variation over the spaceofameretwoweeks[41, 56]. Thus, behavioral plas- 6.2. Alternative Forms of Diversification. Frequency-dependent competition gives rise to disruptive selection, because ticity clearly does provide a mechanism for diversification in the resource distribution can support a broader phenotypic response to changing fitness landscapes. Morphological plas- variance than presently found in the population [111–113]. ticity may play a similar role albeit over a slower time scale as Evolutionwillthusfavoranyofanumberofevolutionary plasticity relies on somatic growth and reshaping. Stickleback changes that increase the population’s phenotypic variance; do exhibit diet-induced plasticity, which is greatest in solitary assortative mating is just one of several mechanisms that populations (Svanback¨ and Schluter, unpublished). can increase this variance [114]. If some other form of To summarize, stickleback concurrently exhibit several diversification evolves first and mitigates the disruptive forms of within-population diversification: increased phe- selection, sympatric speciation will be less likely [115]. notypic variance, dimorphism, behavioral plasticity, and 10 International Journal of Ecology morphological plasticity. Any or all of these may increase Appendix trophic diversity in ways that mitigate disruptive selection and can inhibit speciation. Simulation of Phenotypic Evolution under Disruptive Selection and Assortative Mating 7. Conclusions I used stochastic simulations of an individual-based model The past two decades have seen a burst of renewed research to explore the potential for sympatric speciation across a on sympatric speciation. Collectively, this body of literature range of magnitudes of disruptive selection and assortative suggests that sympatric speciation (1) probably has occurred mating. Sympatric speciation has been widely explored with in at least a few cases [3, 5, 6] but is still relatively rare both stochastic and numerical simulations and analytical [97, 126] and difficult to prove and (2) is “theoretically theory (summarized in [97]), but these models tend not to plausible” [127] but requires very specific conditions [25]. be framed in terms of empirically measurable parameters. An important next step is to fuse the theoretical and I, therefore, focus here on describing a model that can be empirical work (e.g., [12, 13]), by empirically evaluating the framed in terms of disruptive selection gradients and male- biological plausibility of the key theoretical assumptions in female trait correlations. particular model organisms [97]. This shifts empirical focus I consider a population of N haploid individuals charac- away from putative cases of sympatric speciation to a more terized by a phenotype subject to both disruptive selection general program of jointly measuring key parameters (e.g., and assortative mating. Each individual’s phenotype is a selection strength, spatial structure and gene flow, strength quantitative trait that depends on the additive effect of of assortative mating, and identifying phenotypic targets of 40 independently assorting loci of small and equal effect. mate choice). These estimates can then be used to build a Initially, each locus is assumed to be polymorphic with equal more biologically informed body of theory (e.g., [12, 13]) or frequencies of two alleles of effect size 0 or 1. Thus, the to identify those populations best suited to the various forms phenotype space ranges from a minimum of 0 to a maximum of sympatric diversification. of 40 with an initial mean of 20 and variance 10. Individuals In stickleback, the good news for theoreticians is that the were created by randomly drawing a set of 40 allele values. joint occurrence of (weak) disruptive selection and assor- The general results described here are qualitatively robust tative mating validates long-standing assumptions. Meta- to using smaller numbers of loci [63]. The traits under analyses suggest that both phenomena are quite common disruptive selection in stickleback are known to be polygenic, [31, 33],thoughweakandperhapstemporallyvariable with multiple QTL none of which have substantial effect sizes [100, 101]. While the joint action of disruptive selection [63]. Consequently, using a moderate to large number of loci and assortment is less well established, the work reviewed is probably appropriate. above indicates that the processes do co-occur in lake In most sympatric speciation models, disruptive selec- populations of stickleback. The bad news for theory is that tion is an emergent outcome of some assumed ecological both forces are typically too weak to drive speciation. As processsuchasintraspecificcompetition[129]. This adds a is often the case in ecological and evolutionary theory, degree of realism, to the extent that the ecological process model predictions are sensitive to the precise parameter is biologically justified, but makes it difficult to directly values used. Consequently, rigorous tests of theory requires control the strength of disruptive selection as measured models framed using measurable parameters coupled with empirically. Here, I ignored the ecological mechanism suitable empirical data. Theoretical models of speciation underlying selection and instead used a specified disruptive also typically ignore temporal fluctuations in selection, selection gradient to calculate the corresponding fitness and alternative forms of diversification, such as we see in for each phenotype in the population. Specifically, I first stickleback. Such constraints do not make the theory wrong standardized the population’s phenotype distribution to a so much as incomplete, echoing Gavrilets’ statement that “it standard normal (mean = 0, standard deviation = 1). Next, is not that easy” [25]. individuals’ fitness is calculated as The absence of sympatric speciation in stickleback is not w(x) = w + βx + γx2,(A.1) necessarily an indictment of sympatric speciation theory. center Indeed, it is the comparison of theory with empirical where x is an individual’s standardized phenotype, wcenter is parameter values that helps illuminate why speciation may the fitness of the average phenotype, β is the linear selection not readily occur. Furthermore, existing models have made gradient (assumed to be 0), and γ is the quadratic selection a variety of related testable predictions, such as the links gradient. Fitness was then standardized by dividing every between competition, habitat diversity, and sexual dimor- individuals’ fitness w(x) by the maximal fitness, yielding a phism, which have been empirically tested [24, 48, 56, 65, value that falls within the range 0 to 1. This fitness is used as 119]. Thus, models of sympatric speciation may be an the probability the individual survives viability selection. illustration of G. E. P. Box’s statement that, “all models This selection scheme ensures that the minimum of dis- are wrong, but some are useful” [128]. Perhaps continued ruptive selection tracks the phenotype mean, preventing the controversy over sympatric speciation will prove useful as population from evolving away from the fitness minimum. well, regardless of its prevalence in nature, particularly if The stabilized fitness minimum roughly mimics the behavior the debate spurs more extensive empirical measurement of of models of frequency-dependent competition that generate generally important evolutionary forces. stable fitness minima required for speciation [15]. Finally, International Journal of Ecology 11 a uniform random number was drawn for each individual; probability (here, 0.0001, which was chosen because it if the random number was less an individual’s w(x), the maintains fairly constant phenotypic variance in the face of individual was retained for mating. drift in simulations without selection or assortment). Following a round of selection, the surviving individ- I iterated this life cycle (selection, then rounds of mating uals select mates and reproduce. Individuals are randomly until the population size was met) for a specified number assigned a sex assuming an equal sex ratio. One female of generations. At each time step, I recorded the phenotype and one male are randomly drawn from the pool of distribution. At the end of the simulation, speciation was said available survivors, and we calculated the probability the to have occurred if the phenotype distribution was bimodal, individuals mate based on their phenotypic similarity (see and intermediate phenotypes had a frequency less than 5% next paragraph for details). If a random uniform number fell that of the lower of the two modes. I iterated through a range below this probability, they produced one offspring and were of parameter space (γ from 0 to 0.5 in increments of 0.5, and returned to the pool of breeders. If not, they were simply r from 0 to 0.95 in increments of 0.5). For each parameter returned to the breeders, and a new pair was randomly combination, I replicated a 1000 generation simulation 10 selected. Whether or not a pair mated, they were, therefore, times and recorded whether or not speciation occurred. I able to subsequently remate. This scheme allows for some also recorded the phenotypic variance of the ending trait sexual selection to occur; for instance, with strong positive distribution. Simulations were conducted in R [132], and the assortative mating, rare phenotypes may be unable to find code is available from the author on request. acceptable mates. This mate pairing was iterated until the desired population size of offspring was reached (allowing Acknowledgments a constant population size throughout the simulations), at which point all adults died and a new generation began. The authors would like to thank Chad Brock, Edgar Duenez- The typical speciation model calculates a probability the Guzman, Andrew Hendry, Jonathan Losos, Dolph Schluter, female accepts the male, given their phenotypic difference Mike Singer, Michael Turelli, and anonymous reviewers for discussions and feedback on the paper. This work was (x − x )2 supported by funding from the David and Lucille Packard P mating = exp − male female ,(A.2) σa2 Foundation, and many of the reviewed results arose from NSF Grant no. DEB-0412802. where x is the phenotypic value of the male and female 2 being evaluated and σa is the variance of the mate probability References function. This variance measures how quickly mating probability drops off with phenotypic difference. Most speciation [1]J.A.CoyneandH.A.Orr,Speciation, Sinauer Associates, 2 2 Sunderland, Mass, USA, 2004. models either used a set value of σa or allow σa to evolve [2] E. Mayr, Evolution and the Diversity of Life, Belknap Press, as a female phenotype (choosiness; e.g., [9]). However, this ffi Cambridge, Mass, USA, 1976. parameter is di cult to measure empirically in the field. [3]M.Barluenga,K.N.Stolting,¨ W. Salzburger, M. Muschick, Instead, I ran simulations with a specified male-female and A. Meyer, “Sympatric speciation in Nicaraguan crater correlation, r, and used a known relationship to convert r lake cichlid fish,” Nature, vol. 439, no. 7077, pp. 719–723, 2 to σa which we subsequently used in (A.2) to calculate the 2006. mating probability for a given male-female pair. Assuming a [4] P. G. Ryan, P. Bloomer, C. L. Moloney, T. J. Grant, and Lande-stylerelativematechoicefunction[130], the variance W. Delport, “Ecological speciation in South Atlantic island of the mate choice function and the phenotypic standard finches,” Science, vol. 315, no. 5817, pp. 1420–1423, 2007. deviation determine the resulting correlation between mated [5] V. Savolainen, M. C. Anstett, C. Lexer et al., “Sympatric pairs speciation in palms on an oceanic island,” Nature, vol. 441, no. 7090, pp. 210–213, 2006. σmσ f [6] U. K. Schliewen, D. Tautz, and S. Paabo, “Sympatric specia- r = . (A.3) σ4 σ2 σ2 tion suggested by monophyly of crater lake cichlids,” Nature, a + m f vol. 368, no. 6472, pp. 629–632, 1994. [7] T. G. Barraclough and A. P. Vogler, “Detecting the geograph- 2 See [131]. This can be rearranged to calculate a value for σa ical pattern of speciation from species-level phylogenies,” given the phenotypic variance for a trait found in both sexes American Naturalist, vol. 155, no. 4, pp. 419–434, 2000. (σx = σm = σ f ) and a desired r [8] R. Burger,¨ K. A. Schneider, and M. Willensdorfer, “The conditions for speciation through intraspecific competition,” . σ4 0 25 Evolution, vol. 60, no. 11, pp. 2185–2206, 2006. x 4 σa = − σx . (A.4) r2 [9] U. Dieckmann and M. Doebeli, “On the origin of species by sympatric speciation,” Nature, vol. 400, no. 6742, pp. 354– ff 357, 1999. When individuals did mate, I generated one o spring [10] U. Dieckmann, M. Doebeli, J. A. J. Metz, and D. Tautz, Eds., whose genotype is a random sample of alleles from the two Adaptive Speciation, Cambridge University Press, New York, ff parents. That is, for each locus the o spring receives an allele NY, USA, 2004. randomly from one parent. To maintain genetic variation, I [11] S. Gavrilets, “Perspective: models of speciation—what have allowed a mutation process that inverts the phenotypic value we learned in 40 years?” Evolution, vol. 57, no. 10, pp. 2197– of the inherited allele (0 to 1, or 1 to 0) with a specified 2215, 2003. 12 International Journal of Ecology

[12] S. Gavrilets and A. Vose, “Case studies and mathematical [31] J. G. Kingsolver, H. E. Hoekstra, J. M. Hoekstra et al., “The models of ecological speciation. 2. Palms on an oceanic strength of phenotypic selection in natural populations,” island,” Molecular Ecology, vol. 16, no. 14, pp. 2910–2921, American Naturalist, vol. 157, no. 3, pp. 245–261, 2001. 2007. [32] D. S. Wilson and M. Turelli, “Stable underdominance and the [13] S. Gavrilets, A. Vose, M. Barluenga, W. Salzburger, and A. evolutionary invasion of empty niches,” American Naturalist, Meyer, “Case studies and mathematical models of ecological vol. 127, no. 6, pp. 835–850, 1986. speciation. 1. Cichlids in a crater lake,” Molecular Ecology, vol. [33] Y. Jiang, D. I. Bolnick, and M. Kirkpatrick, “Widespread 16, no. 14, pp. 2893–2909, 2007. positive assortative mating within animal populations,” In [14] A. P. Hendry, D. I. Bolnick, D. Berner, and C. L. Peichel, revision. “Forth and back on the speciation continuum: explorations [34] O. Puebla, E. Bermingham, F. Guichard, and E. Whiteman, with stickleback,” Journal of Fish Biology,vol.75,no.8,pp. “Colour pattern as a single trait driving speciation in 2000–2036, 2009. Hypoplectrus coral reef fishes?” Proceedings of the Royal [15] P. A. Abrams, H. Matsuda, and Y. Harada, “Evolutionarily Society B, vol. 274, no. 1615, pp. 1265–1271, 2007. unstable fitness maxima and stable fitness minima of contin- [35] R. G. Reynolds and B. M. Fitzpatrick, “Assortative mating in uous traits,” Evolutionary Ecology, vol. 7, no. 5, pp. 465–487, poison-dart frogs based on an ecologically important trait,” 1993. Evolution, vol. 61, no. 9, pp. 2253–2259, 2007. [16] R. Burger¨ and A. Gimelfarb, “The effects of intraspecific [36] A. P. Hendry, S. K. Huber, L. F. De Leon,´ A. Herrel, and competition and stabilizing selection on a polygenic trait,” J. Podos, “Disruptive selection in a bimodal population of Genetics, vol. 167, no. 3, pp. 1425–1443, 2004. Darwin’s finches,” Proceedings of the Royal Society B, vol. 276, [17] M. Doebeli, “A quantitative genetic competition model for no. 1657, pp. 753–759, 2008. sympatric speciation,” Journal of Evolutionary Biology, vol. 9, [37] S. K. Huber, L. F. De Leon, A. P. Hendry, E. Bermingham, and no. 6, pp. 893–909, 1996. J. Podos, “Reproductive isolation between sympatric morphs [18] M. L. Rosenzweig, “Competitive speciation,” Biological Jour- in a bimodal population of Darwin’s finches,” Proceedings of nal of the Linnean Society, vol. 10, no. 3, pp. 275–289, 1978. the Royal Society of London B, vol. 274, pp. 1709–1714, 2007. [19] R. Burger¨ and K. A. Schneider, “Intraspecific competi- [38] D. Schluter and J. D. McPhail, “Ecological character displace- tive divergence and convergence under assortative mating,” ment and speciation in sticklebacks,” American Naturalist, American Naturalist, vol. 167, no. 2, pp. 190–205, 2006. vol. 140, no. 1, pp. 85–108, 1992. [20] J. Felsenstein, “Skepticism towards Santa Rosalia, or why are [39] B. W. Robinson and D. S. Wilson, “Character release and there so few kinds of animals,” Evolution, vol. 35, pp. 124– displacement in fishes: a neglected literature,” American 138, 1981. Naturalist, vol. 144, no. 4, pp. 596–627, 1994. [21] S. Gavrilets, Fitness Landscapes and the Origin of Species, [40] T. B. Smith, “Bill size polymorphism and intraspecific niche Princeton University Press, Princeton, NJ, USA, 2004. utilization in an African finch,” Nature, vol. 329, no. 6141, pp. [22] M. Araujo´ and D. I. Bolnick, “Diet as a primary target for 717–719, 1987. natural selection in the threespine stickleback,” Evolutionary [41] M. S. Araujo,´ J. P. R. Guimaraes, R. Svanback,¨ A. Pinheiro, Ecology Research. In press. S. F. dos Reis, and D. I. Bolnick, “Network analysis reveals [23] L. K. Snowberg and D. I. Bolnick, “Assortative mating by contrasting effects of intraspecific competition on individual diet in a phenotypically unimodal but ecologically variable vs. population diets,” Ecology, vol. 89, no. 7, pp. 1981–1993, population of stickleback,” American Naturalist, vol. 172, no. 2008. 5, pp. 733–739, 2008. [42] T. Price, “Diet variation in a population of Darwin’s finches,” [24] D. I. Bolnick and L. L. On, “Predictable patterns of disruptive Ecology, vol. 68, no. 4, pp. 1015–1028, 1987. selection in stickleback in postglacial lakes,” American Natu- [43] M. Doebeli and U. Dieckmann, “Evolutionary branching and ralist, vol. 172, no. 1, pp. 1–11, 2008. sympatric speciation caused by different types of ecological [25] S. Gavrilets, ““Adaptive speciation”—it is not that easy: a interactions,” American Naturalist, vol. 156, no. 4, pp. S77– reply to Doebeli et al,” Evolution, vol. 59, no. 3, pp. 696–699, S101, 2000. 2005. [44] R. J. Wootton, “The effectofsizeoffoodrationoneggpro- [26] D. I. Bolnick, R. Svanback,¨ J. A. Fordyce et al., “The ecology duction in the female three-spined stickleback, Gasterosteus of individuals: Incidence and implications of individual aculeatus L,” Journal of Fish Biology, vol. 5, no. 1, pp. 89–96, specialization,” American Naturalist, vol. 161, no. 1, pp. 1–28, 1973. 2003. [45] R. J. Wootton, “Effect of food limitation during the breeding [27] T. B. Smith and S. Skulason,´ “Evolutionary significance of season on the size, body components and egg production resource polymorphisms in fishes, amphibians, and birds,” of female sticklebacks (Gasterosteus aculeatus),” Journal of Annual Review of Ecology and Systematics, vol. 27, pp. 111– Animal Ecology, vol. 46, pp. 823–834, 1977. 133, 1996. [46] R. J. Wootton, “Effects of food and density on the reproduc- [28] M. A. Araujo,´ C. Layman, and D. I. Bolnick, “The ecological tive biology of the threespine stickleback with a hypothesis causes of individual specialization,” Ecology Letters, vol. 14, on population limitation in sticklebacks,” Behaviour, vol. 93, pp. 948–958, 2011. pp. 101–111, 1985. [29] R. K. Butlin, “The variability of mating signals and [47] T. Ingram, W. E. Stutz, and D. I. Bolnick, “Does intraspecific preferences in the brown planthopper, Nilaparvata sizevariationinapredatoraffectitsdietdiversityandtop- lugens(Homoptera: Delphacidae),” Journal of Insect Behavior, down control of prey?” PLOS One, vol. 6, Article ID e20782, vol. 6, no. 2, pp. 125–140, 1993. 2011. [30] R. Lande and S. J. Arnold, “The measurement of selection [48] D. I. Bolnick, “Can intraspecific competition drive disruptive on correlated characters,” Evolution, vol. 37, pp. 1210–1226, selection? An experimental test in natural populations of 1983. sticklebacks,” Evolution, vol. 58, no. 3, pp. 608–618, 2004. International Journal of Ecology 13

[49] S. M. Vamosi, T. Hatfield, and D. Schluter, “A test of [66] J. W. Boughman, “Divergent sexual selection enhances ecological selection against young-of-the-year hybrids of reproductive isolation in sticklebacks,” Nature, vol. 411, no. sympatric sticklebacks,” Journal of Fish Biology, vol. 57, no. 6840, pp. 944–948, 2001. 1, pp. 109–121, 2000. [67] T. Hatfield and D. Schluter, “A test for sexual selection on [50]H.D.Rundle,S.M.Vamosi,andD.Schluter,“Experimental hybrids of two sympatric sticklebacks,” Evolution, vol. 50, no. test of predation’s effect on divergent selection during 6, pp. 2429–2434, 1996. character displacement in sticklebacks,” Proceedings of the [68] J. S. McKinnon, S. Mori, B. K. Blackman et al., “Evidence for National Academy of Sciences of the United States of America, ecology’s role in speciation,” Nature, vol. 429, no. 6989, pp. vol. 100, no. 25, pp. 14943–14948, 2003. 294–298, 2004. [51] D. Schluter, “Experimental evidence that competition pro- [69] H. D. Rundle and D. Schluter, “Reinforcement of stickleback motes divergence in adaptive radiation,” Science, vol. 266, no. mate preferences: sympatry breeds contempt,” Evolution, vol. 5186, pp. 798–801, 1994. 52, no. 1, pp. 200–208, 1998. [52] D. Schluter, “Frequency dependent natural selection during [70] S. M. Vamosi and D. Schluter, “Sexual selection against character displacement in sticklebacks,” Evolution, vol. 57, hybrids between sympatric stickleback species: evidence no. 5, pp. 1142–1150, 2003. from a field experiment,” Evolution, vol. 53, no. 3, pp. 874– [53] L. Snowberg, K. A. Hendrix, and D. I. Bolnick, “Individual 879, 1999. diet specialization is widespread across lacustrine threespine [71] C. Eizaguirre, T. L. Lenz, R. D. Sommerfeld, C. Harrod, M. stickleback,” In preparation. Kalbe, and M. Milinski, “Parasite diversity, patterns of MHC [54] D. I. Bolnick and J. Paull, “Diet similarity declines with II variation and olfactory based mate choice in diverging morphological distance between conspecific individuals,” three-spined stickleback ecotypes,” Evolutionary Ecology, vol. Evolutionary Ecology Research, vol. 11, pp. 1217–1233, 2009. 25, no. 3, pp. 605–622, 2010. [55] L. Snowberg, K. A. Hendrix, and D. I. Bolnick, “Variable [72] T. H. Vines and D. Schluter, “Strong assortative mating variation: patterns of intraspecific niche variation across lake between allopatric sticklebacks as a by-product of adaptation populations of threespine stickleback,” In press. to different environments,” Proceedings of the Royal Society, [56] R. Svanback¨ and D. I. Bolnick, “Intraspecific competition vol. 273, no. 1589, pp. 911–916, 2006. drives increased resource use diversity within a natural [73] J. A. M. Raeymaekers, M. Boisjoly, L. Delaire, D. Berner, K. population,” Proceedings of the Royal Society, vol. 274, no. Ras¨ anen,¨ and A. P. Hendry, “Testing for mating isolation 1611, pp. 839–844, 2007. between ecotypes: laboratory experiments with lake, stream [57] D. I. Bolnick, E. J. Caldera, and B. Matthews, “Evidence for andhybridstickleback,”Journal of Evolutionary Biology, vol. asymmetric migration load in a pair of ecologically divergent 23, no. 12, pp. 2694–2708, 2010. stickleback populations,” Biological Journal of the Linnean [74] L. K. Snowberg and D. I. Bolnick, “Disentangling spatial Society, vol. 94, no. 2, pp. 273–287, 2008. segregation and assortative mating in threespine stickleback,” [58] B. Matthews, K. B. Marchinko, D. I. Bolnick, and A. Unpublished manuscript. Mazumder, “Specialization of trophic position and habitat [75] A. J. W. Ward, P. J. B. Hart, and J. Krause, “The effects of use by sticklebacks in an adaptive radiation,” Ecology, vol. 91, habitat- and diet-based cues on association preferences in no. 4, pp. 1025–1034, 2010. three-spined sticklebacks,” Behavioral Ecology, vol. 15, no. 6, [59] B. W. Robinson, “Trade offs in habitat-specific foraging pp. 925–929, 2004. efficiency and the nascent adaptive divergence of sticklebacks [76] L. Snowberg and D. I. Bolnick, “Diet-dependent mate prefer- in lakes,” Behaviour, vol. 137, no. 7-8, pp. 865–888, 2000. ences in threespine stickleback,” Unpublished manuscript. [60]D.Berner,R.Kaeuffer,A.-C.Grandchamp,J.A.M.Raey- [77] M. Milinski, “The function of mate choice in sticklebacks: maekers, K. Rasanen, and A. P.Hendry, “Quantitative genetic optimizing Mhc genetics,” Journal of Fish Biology, vol. 63, inheritance of morphological divergence in a lake-stream supplement A, pp. 1–16, 2003. stickleback ecotype pair: implications for reproductive isola- [78] M. A. Bell and S. A. Foster, The Evolutionary Biology of the tion,” Journal of Evolutionary Biology, vol. 24, no. 9, pp. 1975– Threespine Stickleback, Oxford University Press, New York, 1983, 2011. NY, USA, 1994. [61] T. Day, J. Pritchard, and D. Schluter, “A comparison of two [79] J. L. Gow, S. M. Rogers, M. Jackson, and D. Schluter, sticklebacks,” Evolution, vol. 48, no. 5, pp. 1723–1734, 1994. “Ecological predictions lead to the discovery of a benthic- [62] M. Hermida, C. Fernandez,´ R. Amaro, and E. San Miguel, limnetic sympatric species pair of threespine stickleback in “Heritability and “evolvability” of meristic characters in Little Quarry Lake, British Columbia,” Canadian Journal of a natural population of Gasterosteus aculeatus,” Canadian Zoology, vol. 86, no. 6, pp. 564–571, 2008. Journal of Zoology, vol. 80, no. 3, pp. 532–541, 2002. [80] J. S. McKinnon and H. D. Rundle, “Speciation in nature: the [63] C. L. Peichel, K. S. Nereng, K. A. Ohgi et al., “The genetic threespine stickleback model systems,” Trends in Ecology and architecture of divergence between threespine stickleback Evolution, vol. 17, no. 10, pp. 480–488, 2002. species,” Nature, vol. 414, no. 6866, pp. 901–905, 2001. [81] J. L. Gow, C. L. Peichel, and E. B. Taylor, “Ecological [64] S. M. Vamosi, “The presence of other fish species affects selection against hybrids in natural populations of sympatric speciation in threespine sticklebacks,” Evolutionary Ecology threespine sticklebacks,” Journal of Evolutionary Biology, vol. Research, vol. 5, no. 5, pp. 717–730, 2003. 20, no. 6, pp. 2173–2180, 2007. [65]D.I.Bolnick,T.Ingram,W.E.Stutz,L.K.Snowberg,O. [82] E. B. Taylor, J. W. Boughman, M. Groenenboom, M. L. Lau, and J. S. Pauli, “Ecological release from interspecific Sniatynski, D. Schluter, and J. L. Gow, “Speciation in reverse: competition leads to decoupled changes in population and morphological and genetic evidence of the collapse of individual niche width,” Proceedings of the Royal Society B, a three-spined stickleback (Gasterosteus aculeatus)species vol. 277, no. 1689, pp. 1789–1797, 2010. pair,” Molecular Ecology, vol. 15, no. 2, pp. 343–355, 2006. 14 International Journal of Ecology

[83]J.E.Behm,A.R.Ives,andJ.W.Boughman,“Breakdown [100] A. M. Siepielski, J. D. Dibattista, and S. M. Carlson, “It’s in postmating isolation and the collapse of a species pair about time: the temporal dynamics of phenotypic selection through hybridization,” American Naturalist, vol. 175, no. 1, in the wild,” Ecology Letters, vol. 12, no. 11, pp. 1261–1276, pp. 11–26, 2010. 2009. [84] W. A. Cresko and J. A. Baker, “Two morphotypes of lacustrine [101] A. M. Siepielski, J. D. Di Battista, J. A. Evans, and S. threespine stickleback, Gasterosteus aculeatus, in Benka Lake, M. Carlson, “Differences in the temporal dynamics of Alaska,” Environmental Biology of Fishes, vol. 45, no. 4, pp. phenotypic selection among fitness components in the wild,” 343–350, 1996. Proceedings of The Royal Society B, vol. 278, pp. 1572–1580, [85] B. K. Kristjansson,´ S. Skulason,´ and D. L. G. Noakes, “Mor- 2011. phological segregation of Icelandic threespine stickleback [102] R. J. Wootton and C. Smith, “A long-term study of a (Gasterosteus aculeatus L),” Biological Journal of the Linnean short-lived fish: the demography of Gasterosteus aculeatus,” Society, vol. 76, no. 2, pp. 247–257, 2002. Behaviour, vol. 137, no. 7-8, pp. 981–997, 2000. [86] B. K. Kristjansson,´ S. Skulason,´ and D. L. G. Noakes, “Rapid [103] R. Svanback¨ and L. Persson, “Population density fluctuations divergence in a recently isolated population of threespine change the selection gradient in eurasian perch,” American stickleback (Gasterosteus aculeatus L.),” Evolutionary Ecology Naturalist, vol. 173, no. 4, pp. 507–516, 2009. Research, vol. 4, no. 5, pp. 659–672, 2002. [104] T. E. Reimchen and P. Nosil, “Ecological causes of sex-biased [87] J. D. McPhail, “Ecology and evolution of sympatric stickle- parasitism in threespine stickleback,” Biological Journal of the backs (Gasterosteus): origin of the species pairs,” Canadian Linnean Society, vol. 73, no. 1, pp. 51–63, 2001. Journal of Zoology, vol. 71, no. 3, pp. 515–523, 1993. [105] T. E. Reimchen and P.Nosil, “Temporal variation in divergent [88] E. B. Taylor and J. D. McPhail, “Evolutionary history of an selection on spine number in threespine stickleback,” Evolu- adaptive radiation in species pairs of threespine sticklebacks tion, vol. 56, no. 12, pp. 2472–2483, 2002. (Gasterosteus): insights from mitochondrial DNA,” Biological [106] T. E. Reimchen and P. Nosil, “Variable predation regimes Journal of the Linnean Society, vol. 66, no. 3, pp. 271–291, predict the evolution of sexual dimorphism in a population 1999. of threespine stickleback,” Evolution, vol. 58, no. 6, pp. 1274– [89] H. D. Rundle, L. Nagel, J. W. Boughman, and D. Schluter, 1281, 2004. “Natural selection and parallel speciation in sympatric [107] T. E. Reimchen and P. Nosil, “Replicated ecological land- sticklebacks,” Science, vol. 287, no. 5451, pp. 306–308, 2000. scapes and the evolution of morphological diversity among [90] W. H. Mathews, J. G. Fyles, and H. W. Nasmith, “Postglacial Gasterosteus populations from an archipelago on the west crustal movements in southwestern British Columbia and coast of Canada,” Canadian Journal of Zoology, vol. 84, no. adjacent Washington State,” Canadian Journal of Earth 5, pp. 643–654, 2006. Science, vol. 7, pp. 690–702, 1970. [108] J. Johansson and J. Ripa, “Will sympatric speciation fail due [91] E. B. Taylor and J. D. McPhail, “Historical contingency to stochastic competitive exclusion?” American Naturalist, and ecological determinism interact to prime speciation in vol. 168, no. 4, pp. 572–578, 2006. sticklebacks, Gasterosteus,” Proceedings of the Royal Society B, [109] J. Johansson, J. Ripa, and N. Kucklander,¨ “The risk of vol. 267, no. 1460, pp. 2375–2384, 2000. competitive exclusion during evolutionary branching: effects [92] R. Kassen, D. Schluter, and J. D. McPhail, “Evolutionary his- of resource variability, correlation and autocorrelation,” The- tory of threespine sticklebacks (Gasterosteus spp.) in British oretical Population Biology, vol. 77, no. 2, pp. 95–104, 2010. Columbia: insights from a physiological clock,” Canadian [110] R. Svanback,¨ M. Pineda-Krch, and M. Doebeli, “Fluctuating Journal of Zoology, vol. 73, pp. 2154–2158, 1995. population dynamics promotes the evolution of phenotypic [93] I. Hutchinson, T. S. James, J. J. Clague, J. V. Barrie, and plasticity,” American Naturalist, vol. 174, no. 2, pp. 176–189, K. W. Conway, “Reconstruction of late Quaternary sea-level 2009. change in southwestern British Columbia from sediments in [111] M. Ackermann and M. Doebeli, “Evolution of niche width isolation basins,” Boreas, vol. 33, no. 3, pp. 183–194, 2004. and adaptive diversification,” Evolution, vol. 58, no. 12, pp. [94] D. I. Bolnick, “Waiting for sympatric speciation,” Evolution, 2599–2612, 2004. vol. 58, no. 4, pp. 895–899, 2004. [112] J. Roughgarden, “Evolution of niche width,” American [95]M.Doebeli,H.J.Blok,O.Leimar,andU.Dieckmann, Naturalist, vol. 106, pp. 683–718, 1972. “Multimodal pattern formation in phenotype distributions [113] M. Slatkin, “Frequency- and density-dependent selection of sexual populations,” Proceedings of the Royal Society B, vol. on a quantitative character,” Genetics,vol.93,no.3,pp. 274, no. 1608, pp. 347–357, 2007. 755–771, 1979. [96] S. Gavrilets, “The Maynard Smith model of sympatric [114] C. Rueffler,T.J.M.VanDooren,O.Leimar,andP.A. speciation,” Journal of Theoretical Biology, vol. 239, no. 2, pp. Abrams, “Disruptive selection and then what?” Trends in 172–182, 2006. Ecology and Evolution, vol. 21, no. 5, pp. 238–245, 2006. [97] D. I. Bolnick and B. M. Fitzpatrick, “Sympatric speciation: [115] D. I. Bolnick and M. Doebeli, “Sexual dimorphism and models and empirical evidence,” Annual Review of Ecology, adaptive speciation: two sides of the same ecological coin,” Evolution, and Systematics, vol. 38, pp. 459–487, 2007. Evolution, vol. 57, no. 11, pp. 2433–2449, 2003. [98] C. Matessi, A. Gimelfarb, and S. Gavrilets, “Long term [116] D. Berner, W. E. Stutz, and D. I. Bolnick, “Diversification in buildup of reproductive isolation promoted by disruptive phenotypic (co)variances among lacustrine populations of selection: how far does it go?” Selection, vol. 2, pp. 41–64, threespine stickleback,” Evolution, vol. 64, pp. 2265–2277, 2001. 2010. [99] J. Polechova´ and N. H. Barton, “Speciation through compe- [117] R. Shine, “Ecological causes for the evolution of sexual tition: a critical review,” Evolution, vol. 59, no. 6, pp. 1194– dimorphism: a review of the evidence,” Quarterly Review of 1210, 2005. Biology, vol. 64, no. 4, pp. 419–461, 1989. International Journal of Ecology 15

[118] T. J. M. Van Dooren, M. Durinx, and I. Demon, “Sexual dimorphism or evolutionary branching?” Evolutionary Ecology Research, vol. 6, no. 6, pp. 857–871, 2004. [119] I. Cooper, R. T. Gilman, and J. W. Boughman, “Sexual dimorphism and speciation on two ecological coins: patterns from nature and theoretical predictions,” Evolution, vol. 65, no. 9, pp. 2553–2571, 2011. [120] J. Kitano, S. Mori, and C. L. Peichel, “Sexual dimorphism in the external morphology of the threespine stickleback (Gasterosteus aculeatus),” Copeia, no. 2, pp. 336–349, 2007. [121] T. Leinonen, J. M. Cano, and J. Merila, “Genetic basis of seual dimorphism in the threespine stickleback Gasterosteus aculeatus,” Heredity, vol. 106, pp. 218–227, 2011. [122] X. Thibert-Plante and A. Hendry, “The consequences of phenotypic plasticity for ecological speciation,” Journal of Evolutionary Biology, vol. 24, pp. 326–342, 2011. [123] D. W. Stephens and J. R. Krebs, Foraging Theory, Princeton University Press, Princeton, NJ, USA, 1986. [124] E. E. Werner, “Optimal foraging and the size selection of prey by the bluegill sunfish (Lepomis macrochirus),” Ecology, vol. 55, pp. 1042–1052, 1974. [125] R. Svanback¨ and D. I. Bolnick, “Intraspecific competition affects the strength of individual specialization: an optimal diet theory method,” Evolutionary Ecology Research, vol. 7, no. 7, pp. 993–1012, 2005. [126] M. Dres and J. Mallet, “Host races in plant-feeding insects and their importance in sympatric speciation,” Philosophical Transactions of the Royal Society of London B, vol. 357, pp. 471–472, 2002. [127] M. Doebeli, U. Dieckmann, J. A. J. Metz, and D. Tautz, “What we have also learned: adaptive speciation is theoretically plausible,” Evolution, vol. 59, no. 3, pp. 691–695, 2005. [128] G. E. P. Box and N. R. Draper, Empirical Model Building and Response Surfaces, John Wiley & Sons, Boston, Mass, USA, 1987. [129] U. Dieckmann, M. Doebeli, J. A. J. Metz, and D. Tautz, Eds., Adaptive Speciation, Cambridge University Press, New York, NY, USA, 2004. [130] R. Lande, “Models of speciation by sexual selection on polygenic traits,” Proceedings of the National Academy of Sciences of the United States of America,vol.78,no.6,pp. 3721–3725, 1981. [131] D. W. Hall, M. Kirkpatrick, and B. West, “Runaway sexual selection when female preferences are directly selected,” Evolution, vol. 54, no. 6, pp. 1862–1869, 2000. [132] R Development Core Team, R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, 2007. Hindawi Publishing Corporation International Journal of Ecology Volume 2011, Article ID 435357, 15 pages doi:10.1155/2011/435357

Research Article Ecological Divergence and the Origins of Intrinsic Postmating Isolation with Gene Flow

Aneil F. Agrawal,1 Jeffrey L. Feder,2, 3 and Patrik Nosil3, 4

1 Department of Ecology & Evolutionary Biology, University of Toronto, Toronto, ON, Canada M5S 3B2 2 Department of Biological Sciences, University of Notre Dame, Notre Dame, IN 46556, USA 3 Institute for Advanced Study, Wissenschaftskolleg, 14193 Berlin, Germany 4 Department of Ecology and Evolutionary Biology, University of Boulder, Boulder, CO 80309, USA

Correspondence should be addressed to Aneil F. Agrawal, [email protected]

Received 25 April 2011; Revised 14 June 2011; Accepted 21 June 2011

Academic Editor: Zachariah Gompert

Copyright © 2011 Aneil F. Agrawal et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The evolution of intrinsic postmating isolation has received much attention, both historically and in recent studies of speciation genes. Intrinsic isolation often stems from between-locus genetic incompatibilities, where alleles that function well within species are incompatible with one another when brought together in the genome of a hybrid. It can be difficult for such incompatibilities to originate when populations diverge with gene flow, because deleterious genotypic combinations will be created and then purged by selection. However, it has been argued that if genes underlying incompatibilities are themselves subject to divergent selection, then they might overcome gene flow to diverge between populations, resulting in the origin of incompatibilities. Nonetheless, there has been little explicit mathematical exploration of such scenarios for the origin of intrinsic incompatibilities during ecological speciation with gene flow. Here we explore theoretical models for the origin of intrinsic isolation where genes subject to divergent natural selection also affect intrinsic isolation, either directly or via linkage disequilibrium with other loci. Such genes indeed overcome gene flow, diverge between populations, and thus result in the evolution of intrinsic isolation. We also examine barriers to neutral gene flow. Surprisingly, we find that intrinsic isolation sometimes weakens this barrier, by impeding differentiation via ecologically based divergent selection.

1. Introduction The most famous answer to this question is attributed to the insights of Bateson, Dobzhansky, and Muller [10–13]. Speciation is a central topic in evolutionary biology and Specifically, intrinsic isolation often arises from between- occurs as inherent barriers to gene flow evolve between locus genetic incompatibilities: alleles that function well formerly interbreeding populations. As data accumulate, it within species are incompatible with one another when is becoming more apparent that many different forms of brought together for the first time in the genome of a hybrid barriers to gene exchange (reproductive isolation) can be (see [14] for review). Genetic architectures of this form involved in speciation [1–4]. One barrier that has received are known as Dobzhansky-Muller (DM) incompatibilities much attention, both historically and in recent work on (illustrated in Table 1) and, in modern parlance, allow the genes causing reproductive isolation, that is, “speciation populations to evolve reproductive isolation without passage genes”, is low hybrid fitness due to genetic incompatibil- through a fitness valley that would be opposed by selection. ities (intrinsic postmating isolation) [5–9]. The evolution The accumulation and origin of genetic incompatibil- of intrinsic reproductive isolation has received so much ities, as well as their effects on levels of gene flow upon attention because it has been one of the key mysteries of secondary contact, have received much attention (Table 2 for speciation. How is it possible, if hybrids are unfit, to evolve review). Along these lines, it can be difficult for intrinsic from the high-fitness genotype of one species to the high- postzygotic barriers to evolve in the face of gene flow (i.e., fitness genotype of the other species without passing through in parapatry or sympatry). The logic is that in interbreed- a low-fitness intermediate? ing populations, deleterious genotypic combinations will 2 International Journal of Ecology

Table 1: Simple Dobzhanksy-Muller incompatibility. With this divergence and intrinsic isolation, independent from time type of genetic architecture, it is possible to evolve from one high (i.e., after controlling for time since species divergence using fitness genotype (AB) to another (ab) without passing through any genetic distance between species pairs) and report positive fitness valleys (by AB → aB → ab). Matings between two high correlations between ecological divergence and intrinsic iso- × fitness types yield unfit hybrids (i.e., AB ab produces the unfit lation. Finally, experimental and genetic studies in yeast also type Ab). implicate divergent adaptation in promoting the evolution of AAintrinsic incompatibilities [47, 48]. B Fit Fit Although these arguments and observations about how divergent selection might overcome gene flow to generate b Unfit Fit intrinsic incompatibilities have been around for some time and make intuitive sense, there has been little formal theoretical exploration of the specific conditions under which be created and then quickly eliminated by selection [3]. genetic incompatibilities might originate during ecological However, divergent selection in general might overcome speciation with gene flow. As noted by Coyne and Orr [3], gene flow to cause population divergence [15–20]. Thus, “we currently lack any theory telling us how much intrinsic one might imagine scenarios for the origin of intrinsic isolation can evolve under a given level of hybridization” incompatibilities in the face of gene flow where divergent (page 177). Here we aim to help fill this theoretical gap selection drives the evolution of incompatibilities, perhaps in by exploring the evolution of intrinsic isolation in the the face of gene flow. Such scenarios represent the process of face of gene flow using a combination of analytical and “ecological speciation”, where reproductive isolation evolves simulation models. We examine how migration, selection, as populations adapt to different ecological environments, and recombination interact to affect this process. We also as a simple and incidental by-product of adaptive genetic examine how the evolution of intrinsic isolation affects rates divergence [21–26]. of neutral gene flow. The results help clarify that ecological Ecological speciation is expected to result in reduced adaptation can generate intrinsic incompatibilities. fitness of immigrants into foreign environments and hybrids Our treatment differs from and extends important past (i.e., extrinsic isolation), due to an ecological mismatch models of DM incompatibilities in several ways. For example, between the phenotype of immigrants and hybrids and local classic theoretical treatments of intrinsic incompatibilities environmental conditions [4, 42]. However, the evolution (e.g., [27, 28, 33]) generally did not consider the evolutionary ofanytypeofreproductivebarriermightevolveasaby- processes driving genetic divergence in the underlying genes product of divergent adaptation. Thus, the negative gene-by- involved; allopatry is explicitly assumed or implied, and gene interactions that cause intrinsic postmating isolation incompatibilities evolve by drift or some unspecified form might also arise via adaptive genetic divergence if, for of selection. Newer models include selection, but it is example, the different alleles favored in divergent habitats are usually not divergent between environments (see Table 2 for incompatible with one another. Along these lines, intrinsic consideration of past models). Our focus on the origin of isolation may evolve in the face of gene flow, as long as the incompatibilities via divergent selection and its interaction diversifying effects of divergent selection can overcome the with gene flow sets our treatment apart from most previous homogenizing effects of gene flow. Verbal arguments along models. While many past studies involving selection were these lines were made quite some time ago, for example, in completely simulation based, our model helps to answer the the writings of Sir Ronald Fisher [43] and in more recent call for analytical treatments of speciation models [49]. reviews [44]. Our models are extensions of classical theoretical work Indeed, there is empirical evidence that ecological on geographic clines by Clarke [16]andEndler[18]and divergence can drive the evolution of intrinsic postmating of more recent models by Gavrilets [14], all of which did isolation. Consider the classic case of ecological adaptation consider divergent selection in some context. Indeed, the and intrinsic hybrid sterility in Mimulus monkey-flowers hybrid zone literature in general discussed how geographic adapted to different soil types [45]. In this example, clines generated by divergent selection could be steepened crossing experiments revealed that a yet unidentified gene by the spread of modifier loci that also contribute to conferring tolerance to copper in the soil (or a gene tightly DM incompatibilities [19, 20].Ourworkismostclosely physically linked to it) interacts with a small number of related to that of Clarke [16], though our goals are quite genes in another population to generate intrinsic hybrid different. He was interested in understanding how clines of incompatibility. Thus, genes conferring adaptation to copper key morphological characters (controlled by a single locus) contribute to postmating isolation. Evidence for a role for could be sharpened by epistatic modifiers. To our knowledge, divergent adaptation in the evolution of intrinsic reproduc- he was the first to explicitly model how epistatic modifiers tive isolation also stems from broad comparative studies can build upon differences established by divergent selection, in fish [46] and insects (fruit flies and moths/butterflies though he did not explicitly discuss reproductive isolation or [25]). These studies considered multiple species pairs and measure barriers to gene flow in other parts of the genome. examined the relationship between indices of the degree of Our work thus extends previous models by considering ecological divergence between species pairs and the extent (1) the joint evolution of extrinsic versus intrinsic isolation of intrinsic postmating isolation between species pairs. during ecological divergence, (2) the explicit effects of Both studies report positive correlations between ecological divergent ecological selection on epistatically interacting loci, International Journal of Ecology 3

Table 2: In chronological order, a summary of some major previous models of Dobzhansky-Muller Incompatibilities (DMIs). Uniform selection refers to scenarios where selection occurs (i.e., some alleles or genotypic combinations are more fit than others), but selection is not divergent between different ecological environments (i.e., no environmental variation). Uniform selection thus fits the “mutation-order speciation” model recently advanced by Schluter [26]. In contrast, divergent selection fits the ecological speciation model. IBD: Isolation- by-distance.

Process driving Migration Discrete patches Model Method Notes divergence and gene flow (populations) Divergent selection is on major locus, modifier locus Clinal affects the fitness of alleles at the major locus, but is not Divergent divergence under divergent selection itself, barrier to neutral gene [16] Yes Analytical selection rather than flow not examined. Also, focus of the model is how the discrete patches modifier, once fixed, alters the shape of the cline, rather than on how the modifier fixes in the first place. Clinal Divergent selection is on major locus, modifier locus Divergent divergence [18] Yes Analytical causes DMI but is not under divergent selection itself, selection rather than barrier to neutral gene flow not examined. discrete patches Stochastic Drift fixed new mutations in small populations; two [27] No Yes Simulation (drift) loci considered. The classic paper on the accumulation of DMIs. No [28] Unspecified No Yes Analytical assumptions are made about evolutionary causes of substitutions driving DMIs. Migration can allow the spread of a gene combination Uniform whose component alleles are not individually favored, selection, not [29, 30] Yes Yes Analytical but the conditions for this are restricted (low explicitly migration, strong advantage for particular genotypic divergent combinations). Population subdivision has no effect on time to Analytical Uniform speciation by drift. Under selection, speciation occurs [31] No Yes and selection or drift more rapidly between two large populations than simulation between many small ones. Unspecified; populations initially fixed for Yes, stepping Focused on the effect of intrinsic incompatibilities on [32] Yes Analytical different stone the barrier to neutral gene flow. incompatible alleles Adds stochasticity of molecular evolution to the results [33] Unspecified No Yes Analytical of Orr [28]. Yes, between Speciation in subdivided populations occurs most Uniform [34] Yes neighboring Simulation rapidly when there is some migration (which spreads selection patches incompatible alleles); many (250) loci are considered. Strong selection counters the inhibitory effects of gene Uniform [35] Yes Yes Simulation flow, permitting the evolution of intrinsic postmating selection isolation. Uniform Migration occurs through a single, spatially structured [36] Yes Yes, Analytical selection population and affects the accumulation of DMIs. Loci causing DMIs themselves are under selection (i.e., Uniform new alleles at these loci can be favored), but selection is [37] Yes Yes Analytical selection uniform. Barrier to gene flow is considered, including the effect of chromosomal inversions. (Gavrilets [14]) Divergent Incompatible alleles advantageous in one environment and previous selection, but neutral in other (thus wider range of scenarios models by same uniform Yes Yes Analytical examined in current study), barrier to gene flow author (e.g., selection, or examined. [38, 39]) drift Uniform No, clinal with IBD and outbreeding depression were sufficient to drive [40] Yes Simulation selection IBD speciation, in the absence of ecological differences. 4 International Journal of Ecology

Table 2: Continued. Process driving Migration Discrete patches Model Method Notes divergence and gene flow (populations) Focused on the origin of DMIs due to different, incompatible mutations diverging between populations Uniform [41] Yes Yes Simulation adapting to identical selection regimes. Gene flow selection strongly impeded strong divergence via such a “mutation-order” process (c.f., Schluter [26]) Focus on the origination of incompatibilities in the face of gene flow. Examined scenarios with and without Analytical Divergent direct, divergent selection on modifier loci which also Current study Yes Yes and selection contribute to intrinsic isolation. Examined barrier to simulation neutral gene flow caused by evolution of intrinsic isolation.

(3) the examination of the barrier to neutral gene flow caused Table 3: Fitnesses of the four haplotypes in the full model. For Case t = t = by intrinsic isolation, and (4) a focus on the initiation of 1, 1 2 0. divergence. Population 1 Population 2 Our model is similar to that of Gavrilets [32] in its exam- AaAa ination of barriers and intrinsic isolation (point (3) above). − s − s − t − t However, Gavrilets [32] assumes divergence in allopatry B 111 (1 2)(1 2)1 2 (via some unspecified process) and begins his analysis after b(1 − t1)(1 − εA)(1− t1)(1− s1)(1 + εa)(1− s2)(1 − εA)1+εa intrinsic isolation already exists. In contrast, our model focuses on how divergent ecological selection causes intrinsic isolation to originate: ecological divergence in our models establishes the initial genetic divergence upon which incom- a allele. Assuming migration levels are low relative to the patibilities can be built. As discussed below, the models yield strength of divergent selection, aB becomes the common some novel and counterintuitive predictions. For example, genotype in Population 2. Imagine that the B locus is an we find that the evolution of intrinsic isolation need not epistatic modifier of the A locus. While the B allele works always strengthen the barrier to neutral gene flow; whether well with the A allele, the b allele improves the fitness of this occurs is dependent on the effect of intrinsic isolation on ff individuals carrying an a allele. For example, the a allele the genetic di erentiation causing extrinsic isolation. may be a gene that confers tolerance to high levels of 2. Model copper, as in the monkey-flower example, but also incurs some pleiotropic cost. The b allele might mitigate this cost, If the level of migration is low compared to the strength thus improving the intrinsic fitness of individuals carrying of divergent selection, populations inhabiting different envi- an a allele. The b allele, however, could interact negatively ronments can quickly become differentiated with respect to with the A allele, regardless of the environment, uncovering genes underlying key ecological traits [3, 50, 51]. Intrinsic deleterious consequences of the A allele normally masked isolation can then arise between taxa because (1) the genes by the B allele. If AB is common in Population 1 and ab is underlying key ecological traits also generate incompatibil- common in Population 2, then recombinants between them ities themselves, or (2) subsequent mutations that are not will be less fit on average, even in the absence of copper. This necessarily integral to ecology are incompatible with the would be considered intrinsic isolation. altered genetic backgrounds generated by the initial adaptive Specifically, we assume that the A locuscontrolsakey changes. Attempting to model the evolution of reproductive ecological trait and is subject to divergent ecological selec- isolation in a spatially heterogeneous environment with tion. In Population 1, ecological selection reduces the fitness multilocus genotypes can quickly lead to an overly complex of individuals with an a allele by an amount s1;inPopulation model with numerous parameters. To minimize these diffi- 2, ecological selection reduces the fitness of individuals with culties, we initially focus on the simplest possible model that an A allele by an amount s2.TheB locus is an epistatic allows us to illustrate some key points: a two-locus haploid modifier of the A locus. Specifically, the b allele improves the model with two populations connected by migration. We intrinsic fitness of the a genotype by an amount εa,perhaps then show through simulation that the major results apply by mitigating some ecologically independent pleiotropic cost in diploid systems. associated with a.Theb allele interacts negatively with the We are interested in exploring the following scenario. A allele, reducing intrinsic fitness by εa. There may also be Consider a species in which the AB haplotype is both fit and direct ecological selection on the B locus. If so, b genotypes commoninPopulation1(Figure 1). A second population have a fitness disadvantage of t1 in Population 1 whereas B of this species (Population 2) becomes established in a new genotypes have a disadvantage of t2 in Population 2. These habitat in which there is ecological selection favoring an fitness effects are shown explicitly in Table 3. International Journal of Ecology 5

Population 1 Population 2

Time 0 AB ﬁxed AB ﬁxed

Divergent ecological selection causes divergence between populations with respecttotheA locus

Time 1 AB common aB common

The modiﬁer b evolves in Population 2 because it is beneﬁcial in combination with a. Direct ecological selection may also favour the b allele.

Time 2 AB common ab common

Figure 1: Divergent selection and the evolution of intrinsic isolation. Time 0: Two populations, connected by migration inhabit different environments. Both are initially fixed for the AB genotype. Time 1: Divergent ecological selection has caused divergence at the A locus, with the a allele evolving to high frequency in Population 2. Time 2: The modifier b evolves in Population 2, either because it ameliorates a pleiotropic cost of the a allele (Model 1) or because the B locus also experiences divergent ecological selection (Model 2).

We assume that migration occurs following selection that drift can be ignored. Recursion equations describing this within each population. The migration rate of individuals life cycle (selection, migration, reproduction) were created in into Population 1 is m1, and the rate into Population 2 the standard way and are available as supplementary material is m2. Technically, this is accomplished as follows. Both in the form of a Mathematica notebook (see Supplementary populations contribute to the migrant pool, but they may Material available online at doi: doi:10.1155/2011/435357). do so differently. A fraction c of individuals in the migrant A number of assumptions were made in order to make pool comes from Population 1, while the remainder (1 − c) analytical progress. We assume that both ecological and comes from Population 2. In each population, a fraction intrinsic selections are weak in an absolute sense but strong m of all individuals come from the migrant pool with the relative to migration. Specifically, the selection parameters 2 remainder being residents. This means m1 = m(1 − c) s1, s2, t1, t2, εA,andεa are O(ξ), and m is O(ξ )whereξ 1. and m2 = cm. A number of ecological scenarios can be We performed the analysis with both tight and loose linkage described by the appropriate choice of m and c.Forexample, (i.e., assuming r is O(ξ)orO(ξ0)). The results represented if Population 1 is much larger than Population 2 such that assume tight linkage unless otherwise stated. Differences migration is primarily from Population 1 into Population 2, between the analyses are noted when they are important then c would be close to 1. Throughout, we have assumed (see supplementary material for details). We also conducted soft selection (i.e., a deme’s contribution to the migrant pool numerical simulations for comparison with our analytical is independent of its genetic composition/fitness). results. These simulations indicate that the analytical results Following migration, mating occurs randomly within are reasonably robust to these assumptions provided that each population. The rate of recombination between the two migration does not become too strong relative to selection loci is r. The populations are assumed to be sufficiently large (see Results). 6 International Journal of Ecology

3. Results The invasion of b alters the equilibrium frequency of A in Population 2, thereby changing the degree of differentiation Our primary results are most easily seen in a simplified case between the two populations at the ecologically important in which there is no ecological selection on the epistatic mod- locus. The change in differentiation is quantified by ifier. We present this case below. The full model including Δδ = δ − δ both ecological selection on the modifier and the complete A|b present A|B fixed ff range of epistatic e ects is discussed in the supplementary εa material and presented in the numerical results below. =−m1 s1(r + s1 − εa) (6) ff r s ε ε Case 1 (Epistatic Trade-o ). In this case, the epistatic modi- + 2 + A + a 1 2 − m2 − + O ξ . fier allele b improves the fitness of individuals carrying the a (s2 + εA + ε)(r + s2 − εa) s2 allele by an amount εa but reduces the fitness of A individuals When migration is weak relative to selection, there is strong by an amount εA. There is no divergent ecological selection acting on the B locus, t = t = 0. divergence at the A locus and the evolution of the b locus 1 2 has very little effect on the magnitude of divergence (i.e., |Δδ| δ). The expression above tends to be negative, ff 3.1. Equilibrium Allele Frequencies and Di erentiation. Diver- implying that the evolution of an epistatic modifier reduces ff gent ecological selection causes the populations to di er- differentiation at the A locus. This provides our first hint that entiate with respect to the ecologically important locus A. the evolution of incompatibilities can reduce barriers to gene Assuming the b allele is initially absent, the equilibrium fre- flow (see below). quencies for A at migration-selection balance in Populations The equilibrium described above is stable under the 1 and 2, respectively, are given by assumption that migration is weak relative to selection. m When linkage is loose (r s), analysis of the largest 1 2 pA,1 = 1 − + O ξ , eigenvalues of the Jacobian matrix indicates that stability s1 (1) requires the following conditions m p = 2 O ξ2 . s ε s ε A,2 s + 1 A 1 A 2 m1 < min , , 2εA + εa εA + εa + s1 We can quantify the differentiation between the two popula- (7) (εA + εa + s2)εa tionsatthislocusas m2 < . 2(εA + εa) + s2 m m δ = p − p = − 1 − 2 O ξ2 . A|B fixed A,1 A,2 1 s s + (2) The reasons for instability are fairly intuitive. With high 1 2 migration, the b allele will often find itself in the wrong b population and with the wrong (i.e., A)allele.Itsfateis The allele can invade the system by first spreading in ε Population 2, where b is favored if the a allele is sufficiently determined by its epistatic interactions. When A is low, b common there. This will be the case provided that migration fixes in both populations because the disadvantage of Ab into this population is not too high, relative to AB in Population 1 is small compared to the advantage of ab relative to aB in Population 2. If εa is high, ab can fix in both populations because ab is more fit than s2(r + s2 + εA + εa)εa m2 < . (3) AB across the metapopulation (the average fitness over both r(s + εA + εa) 2 environments is the relevant fitness measure if populations ε ε Following invasion, the equilibrium frequencies of the B arewellmixed).If a is low and A is high, then B fixes in both allele in the two populations are populations because the disadvantage of Ab relative to AB in Population 1 is large compared to disadvantage of aB relative m r ε 1( + A) 2 to ab in Population 2. While the stability conditions above pB,1 = 1 − + O ξ , εA(r + s1 − εa) serve as a useful heuristic, they should be regarded with (4) caution as they are obtained from analyses assuming that m r ε 2( + a) 2 migration is weak relative to selection (i.e., instability occurs pB,2 = + O ξ . εa(r + s2 + εa) outside the region where the model assumptions are valid). The stability conditions outlined above assume loose At this equilibrium, b is common in Population 2. Because b linkage. Stability is also influenced by the recombination rate enhances the fitness advantage of a over A, the equilibrium r when linkage is tight (see Mathematica fileprovidedason- frequencies of A change due to the invasion of b. The leading line supplementary material). In the simulations described order approximations for the equilibrium frequencies of A at below, we see that tight linkage tends to destabilize the two- this new equilibrium are locus polymorphism. In the remaining sections we focus m r s on parameter regions where the polymorphic equilibrium is 1( + 1) 2 pA,1 = 1 − + O ξ , stable. s1(r + s1 − εa) (5) m r s ε ε 2( + 2 + A + a) 2 3.2. Intrinsic Isolation. Intrinsic isolation is usually measured pA,2 = + O ξ . (s2 + εA + εa)(r + s2 + εa) empirically by comparing the fitness of hybrids to the average International Journal of Ecology 7

fitness of parental types. Typically, such experiments are done aweakeffect on IFn (see higher order terms shown in in the lab where ecological selection is typically absent, or at supplementary material) because they affect the degree least different than in either of the environments from which of divergence between parental populations and thus the the parents are collected. Even in the lab, the ecological trait magnitude of disequilibrium in hybrids. While ecological controlled by the A locus may affect fitness. To account for selection does not have an important quantitative direct this, we model extrinsic selection for or against the a allele effect on the degree of intrinsic isolation, as measured in in the lab by the variable slab, where the value of slab may be the lab, it does have a critical role. Ecological selection negative, positive, or zero. Assuming that the intrinsic fitness initiates the divergence between populations that allows the effects represented by εA and εa are truly “intrinsic” (i.e., invasion of b and the establishment of a stable equilibrium environment independent), the fitness of the four haplotypes in which both loci are polymorphic. In principle, the effects when measured in the lab are given by wAB = (1 + slab/2), of ecology could be discerned if lab rearing conditions were waB = (1−slab/2), wAb = wAB(1−εA), and wab = waB(1+εa). to properly emulate the divergent environmental selection To calculate the mean fitness of population x (where x pressures experienced by the parental populations in nature. may represent one of the source populations or a hybrid crosssuchasF1 and F2), we must know the genotype 3.3. The Barrier to Gene Flow. In the previous section, we frequencies. These genotype frequencies can be calculated as quantified the extent of intrinsic isolation as might be mea- fAB,x = pA,x pB,x + Dx, fAb,x = pA,x(1 − pB,x) − Dx, faB,x = sured empirically. Intuitively, we would expect the evolution (1 − pA,x)pB,x − Dx,and fab,x = (1 − pA,x)(1 − pB,x)+Dx, of intrinsic isolation to serve as a barrier to gene flow and, where Dx is the linkage disequilibrium in population x. These as such, contribute to speciation [14, 32]. Here we show that values for the source populations are given above and in thisisnotalwaystrue. the supplementary material. For the F1 hybrid population, We quantify the barrier to gene flow using the metric pro- p = / p p p = / p p A,F1 1 2( A,1 + A,2)and B,F1 1 2( B,1 + B,2). The posed by Zhivotovsky and Christiansen [52]. This measure linkage disequilibrium in the F1 population is reflects the disadvantage experienced by an unlinked neutral marker that enters a focal population through migration m s ε D = 1 − r 1( 1 + A) from the other population. In the absence of differentiation F1 1 2+ 4 s1εA(r + s1 − εa) between the populations, a neutral marker that migrates into a population would have no disadvantage and the barrier m s ε ε 2( 2 + A +2 a) O ξ2 . strength would be zero. If the two populations are completely +ε r s ε ε ε s + a( + 2 + a)( A + a + 2) incompatible such that all hybrids die, then the barrier (8) strength is infinite. The barrier is quantified as ⎛ ⎞ Assuming future hybrid generations can be produced with- ∞ F wm t out selection, the allele frequencies of the n hybrid genera- B =−ln⎝ , ⎠, (11) wr t tion remain the same as those in the F1. The disequilibrium t=t0 , D = D − r n−1 of later generation hybrids is given by Fn F1 (1 ) . Intrinsic isolation is quantified by comparing the mean where wm,t is the average fitness, in generation t, of the fitness of Fn hybrids to the average of the mean fitnesses of descendants of migrants that entered the population in individuals from each of the two parental populations: generation t0. wr,t is defined analogously for a corresponding set of resident descendents. This measure of barrier strength 1 IFn = (E[wP ] + E[wP ]) − E wFn ,(9)is closely related to that suggested by Bengtsson [53] 2 1 2 provided that the focal neutral locus is unlinked to the genes where E[wx] = fi,x wi,x is the mean fitness of population responsible for the barrier [54]. x. Evaluating this expression with the equilibrium genotype The analytical approximations given below are based frequencies gives on the first five terms in the infinite series in (11). This provides a reasonably good approximation because the effec- 1 2 tiveness of the barrier quickly declines with each additional IFn = (εA + εa) − DFn + O ξ . (10) 4 generation postmigration as the unlinked neutral marker disassociates from migrant genetic background in which it This is always positive because 0 ≤ DFn < 1/4 indicating initially immigrated. The results presented below assume that hybrids are less fit than parental types. As expected, the that recombination is strong relative to selection. Details on degree of isolation depends on the intrinsic epistatic effects the calculation of the barrier, including results that allow for of the modifier, εA and εa. The modifier allele b causes the tight linkage, are provided in the supplementary material. evolution of intrinsic isolation; IFn = 0ifb is absent from Taking Population 1 as our focal population, we find that the system, but IFn > 0afterb invades. To leading order, this the barrier to gene flow of alleles from Population 2 into measure of intrinsic isolation is largely independent of any Population 1 is inadvertent ecological selection in the lab, slab, though higher order terms include interactions between ecological and 31 1 2 3 4 intrinsic selection (see supplementary material). Similarly, B1 = (s1 − εa) + r 57 − 42r +22r − 7r + r s 32 32 the true ecological selection that initiated divergence ( 1, (12) 2 s2) does not appear in (10). These parameters do have × (εA + εa) + O ξ . 8 International Journal of Ecology

Several important points emerge from this result. Both in Population 2 competing against A-bearing migrants (i.e., ecological selection (s1) and intrinsic selection (εA, εa)affect wAB versus wab in Population 2). Combined with the second the strength of the barrier. Whereas ecological selection term, which reflects the consequences of breaking down the clearly has a positive effect on the barrier, the same cannot coadapted gene complex, the intrinsic epistatic effects always be said of intrinsic epistatic selection. The second term in increase the barrier in Population 2. (12) indicates a positive effect of intrinsic epistatic selection The reason that Population 1, rather than Population 2, on barrier strength but the first term shows a negative effect. experiences conflicting effects of intrinsic isolation is because These conflicting effects can be understood as follows. of the perspective we have taken here. In discussing the Consider an unlinked neutral allele n entering Population effect of intrinsic isolation on the barrier strength, we have 1fromPopulation2inanab genotype (the common implicitly compared the barriers when b is absent (and there genotype in Population 2). For as long as the marker allele n is no intrinsic isolation) to the barriers after b invades. The remains associated with the a allele, it suffers from ecological strength of the barriers when b is absent is given by (12)and selection against a in Population 1. The b allele provides (13)withεA = εa = 0. The effects of intrinsic isolation an intrinsic advantage to a-bearing individuals (εa), thereby on each population should be interpreted with respect to reducing the net selection against the ab genotype. By that population’s current genetic composition relative to its ameliorating the selection against a, the b allele makes it composition prior to the evolution of intrinsic isolation. easier for the neutral marker n to successfully migrate into Initially, Population 1 had the appropriate epistatic allele (B), the new population, thus reducing the barrier (see first term whereas Population 2 did not. The evolution of intrinsic of (12)).However,recombinationcanseparatethea and b isolation increases the barrier strength for the population in alleles from each other such that the marker allele n may which the new modifier allele is common (Population 2). then find itself in the Ab genotype, which suffers negative However, the effect on the barrier in the other population intrinsic epistatic selection (wab − wAb ≈ εA + εa). This (Population 1) may be positive or negative, depending on breakdown of the coadapted gene complex makes it more recombination. difficult for the neutral marker to successfully introgress into the new population, thereby increasing the barrier strength 3.4. Numerical Simulations. Numerical simulations were (see second term of (12)). performed to test several of the major conclusions from The extent to which the marker will suffer from negative analytical approximations presented above. The results are effects of the coadapted gene complex breakdown depends presented in Figures 2 and 3. Divergence at the A and on the rate of recombination, r, between the A and B loci. B loci between the two populations is shown in Figures If the A and B loci are tightly linked (i.e., r is low), the 2(a), 2(b), 3(a),and3(b) forCases1and2(fullmodel), unlinked marker will have left the ab haplotype before this respectively. In each figure, the black line represents the level haplotype is destroyed by recombination. Conversely, when of divergence before the evolution of the b allele whereas the r is high, the marker has a good chance of still being on the colored lines show divergence after the introduction of this ab haplotype when recombination occurs between the A and allele. For both models, divergence between populations is B loci. This effect of recombination is reflected in the second only possible when migration is weak relative to selection term of (12) which shows that r mediates the negative effects (Figures 2(a), 2(b), 3(a) and 3(b)). Divergence at the B locus of intrinsic epistatic effects on gene flow (or, equivalently, the (i.e., evolution of the epistatic modifier b) typically has little positive effects on barrier strength). Tightly linked coadapted quantitative effect on divergence at the A locus (compare complexes serve as weaker barriers to gene flow at unlinked colored lines to black lines in Figures 2(a) and 3(a)), provided loci than loosely linked complexes. Gavrilets [32] previously polymorphism persists. However, as discussed in our analyt- showed that barrier strength increases with recombination ical section above, the introduction of the epistatic modifier for DM incompatibilities (see also Bengtsson [53]). can completely destabilize the polymorphism, causing loss In the opposite direction, the leading order approxima- of divergence at both loci. This tends to occur when the tion for the barrier entering Population 2 from Population 1 epistatic modifier confers a large benefit to the a allele (i.e., is εa large relative to s) and there is linkage between the A and B loci (compare left [r = 0.1] and right [r = 0.5] columns 31 1 B = (s + εa) + r 57 − 42r +22r2 − 7r3 + r4 in Figures 2(a) and 2(b)). Direct ecological selection on the 2 32 2 32 (13) B locus helps to maintain divergence at both loci (compare × (εA + εa) + O ξ2 . Figures 2(a) and 2(b) with Figures 3(a) and 3(b)). The strength of intrinsic isolation is shown in Figures This equation is similar to (12) except the first term 2(c) and 3(c) for Cases 1 and 2, respectively. As indicated which is different in two ways. First, the relevant ecological by the points in black, there is no intrinsic isolation prior selection is that which occurs in Population 2 (i.e., s2 rather to the evolution of the b allele. However, the introduction than s1). Second, the intrinsic epistatic effect is positive of the epistatic modifier causes the evolution of intrinsic in (13) rather than being negative as in (12). Just as the isolation in the face of gene flow. The strength of intrinsic presence of b ameliorated the ecological disadvantage of a- isolation increases with the magnitude of epistatic effects on bearing migrants competing against A-bearing residents in either background (εA and εa) (see colored lines in Figures Population 1 (causing a negative effect on B1), the b allele 2(c) and 3(c). Additional simulations (not shown) confirm exaggerates the ecological advantage of a-bearing residents that divergent ecological selection on the B locus has little International Journal of Ecology 9

r = 0.1 r = 0.5 1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

24681012 2 4 6 8 10 12 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

(a) Divergence at A locus

r = 0.1 r = 0.5 1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

24681012 2 4681012 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

(b) Divergence at B locus r = 0.1 r = . 0.07 0 5 0.06 0.015 0.05 0.04 0.01 0.03 0.005 0.02 0.01

24681012 2 4 6 8 10 12 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

0.15 0.08

0.06 0.1

0.04 0.05 0.02

24681012 2 4 6 8 10 12 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

Before b introduced εA = 0.5,s εa = 1.5s εA = 0.5,s εa = 0.5s εA = 1.5,s εa = 1.5s εA = 1.5,s εa = 0.5s

(d) Barrier into Population 1

Figure 2: Continued. 10 International Journal of Ecology

r = 0.1 r = . 0.35 0 5 0.2 0.3 0.25 0.15 0.2 0.1 0.15 0.1 0.05 0.05

2 4681012 2 4 6 8 10 12 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

Before b introduced εA = 0.5,ssεa = 1.5 εA = 0.5,s εa = 0.5s εA = 1.5,s εa = 1.5 s εA = 1.5,s εa = 0.5s

(e) Barrier into Population 2

Figure 2: Numerical simulation results for the epistatic trade-off model (Case 1). Various properties are shown as a function of the migration rate, expressed relative to the strength of ecological selection on the A locus, s. The migration rates numbered 1–12 correspond to m = ∗ s {1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2} ; migration is symmetric, that is, m1 = m2 = 1/2m(c = 1/2). The points in black represent values prior to the evolution of the modifier (i.e., assuming both populations fixed for the B allele as shown in Time 1 in Figure 1). Colored points represent values after the evolution of the modifier; different colors correspond to modifiers of varying effects (see legend). Left column: moderate linkage r = 0.1; right column: no linkage: r = 0.5. (a) Divergence at A locus. (b) Divergence at B locus. (c) Intrinsic isolation as measured by the average fitness individuals from Populations 1 and 2 compared to the average fitness of F2 hybrids. (d) Strength of the barrier into Population 1. (e) Strength of barrier into Population 2. Parameter values used: s = s1 = s2 = 0.1; t1 = t2 = 0. See text for details.

quantitative effect on strength of intrinsic isolation, provided primarily when there is linkage between genes. As predicted polymorphism persists. However, ecological selection can analytically, epistatic effects increase barriers to a greater have a critical qualitative effect by allowing divergence to extent when there is loose linkage (compare left and right occur despite gene flow; intrinsic isolation is limited by the columns in Figures 2(d) and 2(e)). If the interacting genes genetic differentiation at the interacting genes. (A and B) are closely linked, a neutral marker allele has When measured by using early generation hybrids (e.g., a better chance of escaping a migrant chromosome before F2), intrinsic isolation is larger if A and B are unlinked rather an adaptive complex (such as ab) is converted into a than linked (compare left and right columns of Figures 2(c) maladaptive complex (such as Ab). Thus, linkage between and 3(c), noting scales differ). This is because coadapted the interacting genes makes it easier for neutral markers to gene complexes are more likely to be broken down in early move between populations, as observed for past models that generation hybrids if recombination is high. However, if focused on barriers to gene flow per se [14, 32]. Finally, and intrinsic isolation is measured using very late generation unsurprisingly, it is clear that direct ecological selection on hybrids, say F100, then linkage has a smaller effect because the B locus strengthens the barrier to gene flow (compare linkage disequilibrium has been mostly eliminated in late Figures 2(d) and 2(e) with 3(d) and 3(e)). generation hybrids even if recombination is low (not shown). A comparison of the analytical approximations to sim- The strength of the barrier to gene flow into each ulation results is presented in the supplementary material population is shown in Figures 2(d), 2(e), 3(d),and3(e) (Figures S1 and S2). As expected, the analytical results for Cases 1 and 2, respectively. The black line represents the provide good approximations under the conditions under barrier strength prior to the introduction of the b allele. which they were derived (m s). The analytical approxi- Differentiation at the A locus alone creates a barrier to mations tend to underestimate the magnitude of intrinsic gene flow because of divergent ecological selection on this isolation and overestimate the barrier strengths. The devia- gene. The evolution of b always increases the strength of tions of the analytical approximations from the simulation Barrier 2 (i.e., colored lines lie above black line). Both results increase as migration becomes of the same mag- beneficial epistatic effects on the a-allele background (εa) nitude as selection. The most obvious discrepancy occurs and deleterious epistatic effects on the A-allele background when the analytical results wrongly predict the two-locus (εA) strengthen the barrier (see Figure 2(e)). As predicted polymorphism is stable. Instability is predicted to occur from the analytical approximations, the b allele can either when migration is strong relative to selection. However, increase or decrease the strength of Barrier 1; εA increases the analysis is based on a weak migration assumption, so this barrier, whereas εa decreases this barrier. For example, it is not surprising that our analytically derived stability high values of εa greatly reduce the barrier (red and blue conditions are not very good at quantitatively predicting lines in Figure 2(d)), though these negative effects occur instability. International Journal of Ecology 11

r = 0.1 r = 0.5 1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

24681012 2 4 6 8 10 12 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

(a) Divergence at A locus

r = 0.1 r = 0.5 1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

24681012 2 4 6 8 10 12 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

(b) Divergence at B locus r = 0.1 r = . 0.07 0 5 0.06 0.015 0.05 0.04 0.01 0.03 0.005 0.02 0.01

2 4 6 8 10 12 2 4 6 8 10 12 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

24681012 2 4 6 8 10 12 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

Before b introduced εA = 0.5,s εa = 1.5 s εA = 0.5,s εa = 0.5s εA = 1.5,s εa = 1.5s εA = 1.5,ssεa = 0.5

(d) Barrier into Population 1

Figure 3: Continued. 12 International Journal of Ecology

r = 0.1 r = 0.5 0.25 0.2 0.3

0.15 0.2 0.1 0.1 0.05

2 4 6 8 10 12 2 4681012 Migration rate Migration rate Values correspond to m = s∗{1/100, 1/80, 1/60, 1/40, 1/20, 1/10, 1/8, 1/6, 1/4, 1/2, 1, 2}

Before b introduced εA = 0.5,s εa = 1.5 s εA = 0.5,s εa = 0.5 s εA = 1.5,s εa = 1.5 s εA = 1.5,s εa = 0.5 s

(e) Barrier into Population 2

Figure 3: Numerical simulation results for the model with modiﬁer with both epistatic trade-oﬀ and direct, divergent selection (Case 2, full model). Parameter values used: s = s1 = s2 = 0.1; t1 = t2 = 0.5 s.SeeFigure 2 for other details.

We have also tested whether our results are sensitive once contact is reestablished in a two-patch model [32, to the assumption of haploidy by performing a series of 53]. Such DM incompatibilities only become stable when simulations for diploids. Results for these simulations are more complicated geographies, such as stepping-stone type presented in the supplementary material (Figures S(3)–S(8)). models, are invoked (e.g., see [32]). Qualitatively, all of our major results apply to diploids, Alternatively, high fitness ridges may not be completely including the counter intuitive finding that the evolution of flat so that selection can drive the local fixation of alleles intrinsic isolation can reduce barrier strength. This is most that have the potential to be involved in incompatibilities. likely to occur in diploid systems when negative epistatic However, when there is gene flow, such alleles are expected effects are recessive (compare Figures S3(d), S6(d), and to quickly spread across a population as they are initially S9(d)). This type of gene action underlies the dominance beneficial everywhere if there is no divergent ecological theory of Haldane’s Rule [55] and is common in Drosophila selection. In the face of gene flow, incompatibilities can incompatibilities [3, 7, 56]. only evolve if mutations at different interacting loci spread simultaneously in different parts of the metapopulation 4. Discussion [36] or if new alleles interact with previously established incompatibilities [37]. 4.1. The Evolution of Intrinsic Postmating Isolation. Here, we Although it is intuitively obvious that the addition of presented a theoretical analysis of the role that divergent nat- divergent ecological selection will make it easier to establish ural selection can play in generating intrinsic reproductive incompatibilities in the face of gene flow, this possibility has isolation during divergence with gene flow. Our objective was been largely ignored in terms of formal models (but see [18]). to assess the extent to which intrinsic reproductive isolation Our simplest finding was that maintaining differentiation must be a by-product of genetic divergence in allopatry. is the primary contribution of divergent selection to the The typical DM incompatibility is often depicted as evolution of intrinsic isolation. In particular, establishing shown in Table 1 where fitnesses are not ecologically depen- genetic differentiation at ecologically important loci provides dent. This representation has the characteristic property that the appropriate setting for the subsequent evolution of two high-fitness genotypes (AB and ab) are connected by a epistatically interacting modifiers that can pleiotropically ridgeofhigh-fitnessgenotypes,soitispossibletomovefrom cause intrinsic reproductive isolation. This can happen one high-fitness genotype to the other by a series of more in two ways. Either divergent ecological selection can act or less neutral substitutions. This type of representation is directly on the modifier locus, or not. common in text books (Barton et al. [57, page 643]) and Divergent ecological selection acting directly on a speciation models (see [14, 27, 32, 49], see also Table 2). modifier locus (Case 2) is the easiest way to maintain Truly allopatric populations can diverge along neutral ridges high levels of differentiation, even with strong epistatic in the adaptive landscape. However, such divergence will be effects. In the case where ecological selection does not act severely impeded if there is gene flow both because migration directly on the modifier locus (Case 1), there are limits directly homogenizes populations and because selection will to the strength of intrinsic isolation that can evolve. It oppose alleles that are incompatible with other alleles found is difficult for epistatic modifiers with strong effects to in some parts of the range. Even if divergence occurs via remain differentiated between populations, unless there is drift in allopatry, the two-locus polymorphism will collapse the right balance between the modifier’s positive effect on the International Journal of Ecology 13 a-allele background (εa) and its negative effect on the A-allele see [3, 5, 6, 58–61]). Most studies have centered on genes background (εA). Without the appropriate balance, strong underlying intrinsic postmating isolation, particularly in modifiers are either lost in both populations or fixed in hybrids between species of Drosophila. A major discovery both, in the latter case often eliminating polymorphism at emerging from this work, particularly in animal systems, is the ecologically differentiated locus (A) in the process. Loose that genes underlying intrinsic postzygotic isolation often linkage makes it somewhat easier to maintain a high level show molecular signatures of positive selection [3, 9, 62– of differentiation for strong modifiers, thus allowing for the 64]. In one case, the gene “Overdrive”, the mechanism of evolution of stronger intrinsic isolation (compare left and selection appears to be intragenomic conflict rather than right columns of Figure 2). environmental, ecologically based selection [8], and a similar In sum, divergent ecological selection allows intrinsic mechanism is suspected in other cases [7, 65]. In contrast, isolation to evolve through alleles that would otherwise genetic studies in yeast implicate ecological adaptation as not contribute to reproductive barriers in the presence of the driver of postmating isolation [47, 48]. Nevertheless, gene flow. For example, consider a modifier that did not in most cases of putative speciation genes, the mechanisms experience ecological selection and was neutral on one of selection involved are not yet known. As discussed in background but had intrinsic epistatic effects on the other. the introduction, there is empirical evidence from other It would either spread to fixation in both populations or types of studies that ecological divergence can indeed drive be completely lost, thus not contributing to reproductive the evolution of intrinsic postmating isolation. Collectively, isolation. However, when the modifier also experiences these studies suggest that the theoretical scenarios treated divergent ecological selection, populations can differentiate in our paper are plausible and that studies of speciation with respect to allele frequencies for this gene, allowing genes might consider whether the genes involved evolve via reproductive isolation to evolve. divergent selection, perhaps in the face of gene flow.

4.2. The Barrier to Neutral Gene Flow. An interesting 5. Conclusions counterintuitive and somewhat unexpected result from our Postzygotic isolation is often dichotomized as “intrinsic” analysis is that the same genes that cause intrinsic isolation or “ecological/extrinsic”. However, many cases of the spe- can reduce the barrier to gene flow. By improving the ciation process may involve both [3]. Divergent ecological intrinsic fitness of an ecologically maladaptive allele (a), selection is well documented and can allow for genetic an epistatic modifier (b) reduces the net selection against differentiation despite gene flow [14, 18]. Moreover, adap- foreign genotypes, potentially making it easier for foreign tation to particular environmental challenges may often alleles to enter the focal population. For a related reason, come at some pleiotropic cost. This creates the potential the epistatic modifiers typically have asymmetric effects on for epistatic modifiers that ameliorate such costs but that the barriers to gene flow. The barrier into the population may be incompatible on the original genetic background. containing the derived modifier allele (b) tends to be stronger Under such conditions, divergent ecological selection can than the barrier into the other population. The derived be considered the catalyst for the evolution of intrinsic modifier allele’s beneficial epistatic effect on its preferred isolation. However, the genes that contribute to intrinsic background makes it easier for genotypes from its population isolation may have mixed effects on barrier to gene flow. to invade the other population while at the same time making Thus, although divergent selection can thus promote the it more difficult for the reverse direction of gene flow. evolution of intrinsic isolation, the overall importance of this Our result about the asymmetric effects of epistatic for the process of ecological speciation will depend on (1) the modifiers on barrier strengths seems related to asymmetric degree to which this actually reduces gene flow and (2) the effects with respect to intrinsic isolation in classic DM extent and rate to which this occurs relative to commonly incompatibilities. We found that the barrier from the documented scenarios where divergent selection drives the population containing the derived modifier allele into the evolution of premating or extrinsic barriers to gene flow population with the ancestral allele tends to be weaker than [3, 24, 25, 66]. the reverse situation, whereas Orr [28] showed that derived alleles are more likely to involve (cause) incompatibilities. Acknowledgments However, the underlying reasons for these two types of “asymmetries” are very different. Our result is driven by ThisworkwassupportedbytheNaturalSciencesand the derived modifier improving the intrinsic fitness of one Engineering Research Council of Canada (Discovery Grant) genotype, thereby mitigating its ecological disadvantage in to A. F. Agrawal, as well as NSF and USDA support to J. L. one habitat and augmenting its ecological advantage in the Feder. During a period of working on this project, P. Nosil other. In contrast, Orr’s result does not involve ecological and J. L. Feder were hosted by the Institute for Advanced dependence and is driven by negative effects of the derived Study, Wissenschaftskolleg, Berlin. The authors thank S. allele in (hybrid) combinations that have gone untested by Gavrilets and other anonymous reviewers for comments on selection in allopatry. previous versions of this manuscript.

4.3. Implications for Empirical Studies of “Speciation Genes”. References Much empirical work has been dedicated to ﬁnding “speci- [1] J. A. Coyne and H. A. Orr, “Patterns of speciation in ation genes” underlying reproductive isolation (for reviews Drosophila ,” Evolution, vol. 43, pp. 362–381, 1989. 14 International Journal of Ecology

[2] J. Ramsey, H. D. Bradshaw, and D. W. Schemske, “Compo- Academy of Sciences of the United States of America, vol. 103, nents of reproductive isolation between the monkeyflowers no. 9, pp. 3209–3213, 2006. Mimulus lewisii and M. cardinalis (Phrymaceae),” Evolution, [26] D. Schluter, “Evidence for ecological speciation and its alter- vol. 57, no. 7, pp. 1520–1534, 2003. native,” Science, vol. 323, no. 5915, pp. 737–741, 2009. [3]J.A.CoyneandH.A.Orr,Speciation, Sinauer Associates, [27] M. Nei, T. Maruyama, and C. Wu, “Models of evolution of Sunderland, Mass, USA, 2004. reproductive isolation,” Genetics, vol. 103, no. 3, pp. 557–579, [4] P.Nosil, T. H. Vines, and D. J. Funk, “Perspective: reproductive 1983. isolation caused by natural selection against immigrants from [28] H. A. Orr, “The population genetics of speciation: the divergent habitats,” Evolution, vol. 59, no. 4, pp. 705–719, evolution of hybrid incompatibilities,” Genetics, vol. 139, no. 2005. 4, pp. 1805–1813, 1995. [5] C. I. Wu and C. T. Ting, “Genes and speciation,” Nature [29] M. Slatkin, “Epistatic selection opposed by immigration Reviews Genetics, vol. 5, no. 2, pp. 114–122, 2004. in multiple locus genetic systems,” Journal of Evolutionary [6] H. A. Orr, “The genetic basis of reproductive isolation: insights Biology, vol. 8, no. 5, pp. 623–633, 1995. from Drosophila,” Proceedings of the National Academy of [30] N. H. Barton, “The role of hybridization in evolution,” Sciences of the United States of America, vol. 102, no. 1, pp. Molecular Ecology, vol. 10, no. 3, pp. 551–568, 2001. 6522–6526, 2005. [31] H. A. Orr and L. H. Orr, “Waiting for speciation: the effect of [7] D. C. Presgraves, “Speciation genetics: epistasis, conflict and population subdivision on the time to speciation,” Evolution, theoriginofspecies,”Current Biology, vol. 17, no. 4, pp. R125– vol. 50, no. 5, pp. 1742–1749, 1996. R127, 2007. [32] S. Gavrilets, “Hybrid zones with Dobzhansky-type epistatic [8] N. Phadnis and H. A. Orr, “A single gene causes both male selection,” Evolution, vol. 51, no. 4, pp. 1027–1035, 1997. sterility and segregation distortion in Drosophila hybrids,” [33] H. A. Orr and M. Turelli, “The evolution of postzygotic iso- Science, vol. 323, no. 5912, pp. 376–379, 2009. lation: accumulating Dobzhansky-Muller incompatibilities,” [9] S. Tang and D. C. Presgraves, “Evolution of the Drosophila Evolution, vol. 55, no. 6, pp. 1085–1094, 2001. nuclear pore complex results in multiple hybrid incompatibil- [34] S. A. Church and D. R. Taylor, “The evolution of reproductive ities,” Science, vol. 323, no. 5915, pp. 779–782, 2009. isolation in spatially structured populations,” Evolution, vol. [10] W. Bateson, “Heredity and variation in modern lights,” in 56, no. 9, pp. 1859–1862, 2002. Darwin and Modern Science, A. C. Seward, Ed., pp. 85–101, [35] A. H. Porter and N. A. Johnson, “Speciation despite gene flow Cambridge University Press, Cambridge, UK, 1909. when developmental pathways evolve,” Evolution, vol. 56, no. [11] T. Dobzhansky, Genetics and the Origin of Species, Columbia 11, pp. 2103–2111, 2002. University Press, New York, NY, USA, 1st edition, 1973. [36] A. S. Kondrashov, “Accumulation of Dobzhansky-Muller [12] H. J. Muller, “Bearing of the Drosophila work on systematics,” incompatibilities within a spatially structured population,” in The New Systematics, J. Huxley, Ed., pp. 185–268, Oxford Evolution, vol. 57, no. 1, pp. 151–153, 2003. University Press, Oxford, UK, 1940. [37] A. Navarro and N. H. Barton, “Accumulating postzygotic [13] H. J. Muller, “Isolating mechanisms, evolution, and temera- isolation genes in parapatry: a new twist on chromosomal ture,” Biology Symposium, vol. 6, pp. 71–125, 1942. speciation,” Evolution, vol. 57, no. 3, pp. 447–459, 2003. [14] S. Gavrilets, Fitness Landscapes and the Origin of Species, [38] S. Gavrilets and A. Hastings, “Founder effect speciation: a Princeton University Press, Princeton, NJ, USA, 2004. theoretical reassessment,” American Naturalist, vol. 147, no. 3, [15] J. B. S. Haldane, “The theory of a cline,” Journal of Genetics, pp. 466–491, 1996. vol. 48, no. 3, pp. 277–284, 1948. [39] S. Gavrilets, “Waiting time to parapatric speciation,” Proceed- [16] B. Clarke, “Evolution of morph-ratio clines,” American Natu- ings of the Royal Society B, vol. 267, no. 1461, pp. 2483–2492, ralist, vol. 100, pp. 389–402, 1966. 2000. [17] J. A. Endler, “Gene flow and population differentiation,” [40] G. A. Hoelzer, R. Drewes, J. Meier, and R. Doursat, “Isolation- Science, vol. 179, no. 4070, p. 243, 1973. by-distance and outbreeding depression are sufficient to [18] J. A. Endler, Geographic Variation, Speciation, and Clines, drive parapatric speciation in the absence of environmental Princeton University Press, Princeton, NJ, USA, 1977. influences,” PLoS Computational Biology, vol. 4, no. 7, Article [19] N. H. Barton and G. M. Hewitt, “Analysis of hybrid zones,” ID e1000126, 2008. Annual review of ecology and systematics, vol. 16, pp. 113–148, [41] P. Nosil and S. M. Flaxman, “Conditions for mutation-order 1985. speciation,” Proceedings of the Royal Society B, vol. 278, pp. [20] N. H. Barton and G. M. Hewitt, “Adaptation, speciation and 399–407, 2011. hybrid zones,” Nature, vol. 341, no. 6242, pp. 497–503, 1989. [42] H. D. Rundle and M. C. Whitlock, “A genetic interpretation of [21] D. J. Funk, “Isolating a role for natural selection in speciation: ecologically dependent isolation,” Evolution,vol.55,no.1,pp. host adaptation and sexual isolation in Neochlamisus beb- 198–201, 2001. bianae leaf beetles,” Evolution, vol. 52, no. 6, pp. 1744–1759, [43] R. A. Fisher, The Genetical Theory of Natural Selection, 1998. Claredon Press, Oxford, UK, 1930. [22] D. Schluter, “Ecological causes of speciation,” in Endless Forms: [44] M. Turelli, N. H. Barton, and J. A. Coyne, “Theory and Species and Speciation, D. J. Howard and S. H. Berlocher, Eds., speciation,” Trends in Ecology & Evolution, vol. 16, no. 7, pp. pp. 114–129, Oxford University Press, Oxford, UK, 1998. 330–343, 2001. [23] D. Schluter, “Ecology and the origin of species,” Trends in [45] M. R. Macnair and P. Christie, “Reproductive isolation as a Ecology & Evolution, vol. 16, no. 7, pp. 372–380, 2001. pleiotropic effect of copper tolerance in Mimulus-guttatus,” [24] H. D. Rundle and P. Nosil, “Ecological speciation,” Ecology Heredity, vol. 50, pp. 295–302, 1983. Letters, vol. 8, no. 3, pp. 336–352, 2005. [46] D. I. Bolnick, T. J. Near, and P. C. Wainwright, “Body size [25] D. J. Funk, P. Nosil, and W. J. Etges, “Ecological divergence divergence promotes post-zygotic reproductive isolation in exhibits consistently positive associations with reproductive centrarchids,” Evolutionary Ecology Research, vol. 8, no. 5, pp. isolation across disparate taxa,” Proceedings of the National 903–913, 2006. International Journal of Ecology 15

[47] J. R. Dettman, C. Sirjusingh, L. M. Kohn, and J. B. Anderson, “Incipient speciation by divergent adaptation and antagonistic epistasis in yeast,” Nature, vol. 447, no. 7144, pp. 585–588, 2007. [48] H. Y. Lee, J. Y. Chou, L. Cheong, N. H. Chang, S. Y. Yang, and J. Y. Leu, “Incompatibility of nuclear and mitochondrial genomes causes hybrid sterility between two yeast species,” Cell, vol. 135, no. 6, pp. 1065–1073, 2008. [49] S. Gavrilets, “Perspective: models of speciation: what have we learned in 40 years?” Evolution, vol. 57, no. 10, pp. 2197–2215, 2003. [50]A.P.Hendry,P.Nosil,andL.H.Rieseberg,“Thespeedof ecological speciation,” Functional Ecology,vol.21,no.3,pp. 455–464, 2007. [51] P. Nosil, “Adaptive population divergence in cryptic color- pattern following a reduction in gene flow,” Evolution, vol. 63, no. 7, pp. 1902–1912, 2009. [52] L. A. Zhivotovsky and F. B. Christiansen, “The selection barrier between populations subject to stabilizing selection,” Evolution, vol. 49, no. 3, pp. 490–501, 1995. [53] B. O. Bengtsson, The Flow of Genes through a Genetic Barrier, Cambridge University Press, Cambridge, UK, 1985. [54] K. V. Pylkov, L. A. Zhivotovsky, and F. B. Christiansen, “The strength of the selection barrier between populations,” Genetical Research, vol. 76, no. 2, pp. 179–185, 2000. [55] M. Turelli and H. A. Orr, “The dominance theory of Haldane’s rule,” Genetics, vol. 140, no. 1, pp. 389–402, 1995. [56] D. C. Presgraves, L. Balagopalan, S. M. Abmayr, and H. A. Orr, “Adaptive evolution drives divergence of a hybrid inviability gene between two species of Drosophila,” Nature, vol. 423, no. 6941, pp. 715–719, 2003. [57] N. H. Barton, D. E. G. Briggs, J. A. Eisen, D. B. Goldstein, and N. H. Patel, Evolution, CSHL Press, Cold Spring Harbor, NY, USA, 2007. [58] H. A. Orr and D. C. Presgraves, “Speciation by postzygotic isolation: forces, genes and molecules,” BioEssays, vol. 22, no. 12, pp. 1085–1094, 2000. [59] K. Bomblies, “Doomed lovers: mechanisms of isolation and incompatibility in plants,” Annual Review of Plant Biology, vol. 61, pp. 109–124, 2010. [60] L. H. Rieseberg and B. K. Blackman, “Speciation genes in plants,” Annals of Botany, vol. 106, no. 3, pp. 439–455, 2010. [61] P. Nosil and D. Schluter, “The genes underlying the process of speciation,” Trends in Ecology & Evolution, vol. 26, pp. 160– 167, 2011. [62] D. A. Barbash, D. F. Siino, A. M. Tarone, and J. Roote, “A rapidly evolving MYB-related protein causes species isolation in Drosophila,” Proceedings of the National Academy of Sciences of the United States of America, vol. 100, no. 9, pp. 5302–5307, 2003. [63] D. A. Barbash, P. Awadalla, and A. M. Tarone, “Functional divergence caused by ancient positive selection of a Drosophila hybrid incompatibility locus,” PLoS Biology, vol. 2, no. 6, pp. 839–848, 2004. [64] J. Mallet, “What does Drosophila genetics tell us about speciation?” Trends in Ecology & Evolution,vol.21,no.7,pp. 386–393, 2006. [65] D. C. Presgraves, “Does genetic conflict drive rapid molecular evolution of nuclear transport genes in Drosophila?” BioEs- says, vol. 29, no. 4, pp. 386–391, 2007. [66] T. C. Mendelson, “Sexual isolation evolves faster than hybrid inviability in a diverse and sexually dimorphic genus of fish (Percidae: Etheostoma),” Evolution, vol. 57, no. 2, pp. 317–327, 2003.