THE YEAR IN EVOLUTIONARY BIOLOGY 2009

The Genetics and Ecology of Reinforcement Implications for the Evolution of Prezygotic Isolation in Sympatry and Beyond

Daniel Ortiz-Barrientos,a Alicia Grealy,a and Patrik Nosilb,c aSchool of Biological Sciences, University of Queensland, St. Lucia, Queensland, Australia bDepartment of Ecology and Evolutionary Biology, University of Colorado, Boulder, Colorado, USA cInstitute for Advanced Study, Wissenschaftskolleg, Berlin, Germany

Reinforcement, the evolution of prezygotic reproductive barriers by natural selection in response to maladaptive hybridization, is one of the most debated processes in spe- ciation. Critics point to “fatal” conceptual flaws for sympatric evolution of prezygotic isolation, but recent theoretical and empirical work on genetics and ecology of reinforce- ment suggests that such criticisms can be overcome. New studies provide evidence for reinforcement in frogs, fish, insects, birds, and plants. While such evidence lays to rest the argument over reinforcement’s existence, our understanding remains incomplete. We lack data on (1) the genetic basis of female preferences and the links between genet- ics of pre- and postzygotic isolation, (2) the ecological basis of reproductive isolation, (3) connections between prezygotic isolation between species and within-species sexual selection (potentially leading to a “cascade” of effects on reproductive isolation), (4) the role of habitat versus mate preference in reinforcement, and (5) additional detailed comparative studies. Here, we review data on these issues and highlight why they are important for understanding speciation.

Key words: speciation; reinforcement; reproductive isolation; female preferences; as- sortative mating; genetic incompatibilities; natural selection

Introduction patric) diverging species (Dobzhansky 1940; Howard 1993; Noor 1999). Subsequently, hy- Since Darwin, scientists have strived to un- bridizing populations can achieve full repro- derstand the evolutionary forces that drive spe- ductive isolation when alleles that confer mate ciation (Dobzhansky 1937b; Mayr 1963). In discrimination spread to the entire species particular, can the formation of new species be (Servedio & Kirkpatrick 1997). Importantly, favored by natural selection? One popular the- many types of costs to hybridization, including ory in which natural selection drives speciation both ecological and genetic dysfunctions in hy- is termed “reinforcement.” Under this mode of brid offspring that reduce their fitness, can drive speciation, natural selection strengthens prezy- reinforcement. The process of reinforcement gotic reproductive isolation (for example, mate is usually detectable before species reach full discrimination) as a response to maladaptive prezygotic isolation: because selection against hybridization between co-occurring (or sym- hybrids only occurs in sympatry, then pairs of populations in regions where hybridization oc- curs will exhibit stronger prezygotic isolation Address for correspondence: Daniel Ortiz-Barrientos, The University than pairs of populations in regions where hy- of Queensland, School of Biological Sciences, Goddard Building (8) R120, St. Lucia, QLD 4072, Australia. Voice: 61-7-3365-1767; fax: 61-7-3365- bridization is absent (allopatry). This pattern of 1655. [email protected] greater prezygotic isolation in sympatry relative

The Year in Evolutionary Biology 2009: Ann. N.Y. Acad. Sci. 1168: 156–182 (2009). doi: 10.1111/j.1749-6632.2009.04919.x c 2009 New York Academy of Sciences. 156 Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 157 to allopatry is one of the main “signatures” of reinforcement and has been termed reproduc- tive character displacement (RCD, Servedio & Noor 2003). Reinforcement suffers from several major theoretical obstacles, which have been reviewed extensively elsewhere and are therefore treated only briefly here (Felsenstein 1981; Sanderson 1989; Howard 1993; Butlin 1995; Servedio & Noor 2003; Coyne & Orr 2004). Most of these obstacles involve the “swamping” effects of gene flow. In particular, hybridization (with gene flow) between populations or species in regions of sympatry acts as a homogenizing process, which can prevent divergence via re- inforcement. In addition, an influx of “less dis- criminating” alleles from allopatry into sympa- try can weaken the strength of discrimination against heterospecific individuals, thereby con- straining reinforcement (see Coyne & Orr 2004 for a review). Finally, it can be difficult for alle- les contributing to reinforcement to spread out of regions of immediate sympatry because such alleles may be neutral or even disadvantageous Figure 1. The signatures of speciation by rein- forcement. (A) Persistence of increased prezygotic in other regions (that is, in allopatry) (Table isolation in sympatry despite ongoing gene flow be- 3, Fig. 1). Additionally, from an empirical per- tween sympatric and allopatric populations of one spective, several processes other than reinforce- species. (B) The evolution of prezygotic isolation ment can generate the pattern of RCD, and between two sympatric species (RI), could lead to ruling out these alternatives to reinforcement the evolution of sexual isolation among populations can be difficult because it relies on the presence within species (SI), despite ongoing gene flow. This would lead to the persistence of the signature of re- of stringent—and often unavailable—controls inforcement, and possibly to the origin of another (Rundle & Schluter 1998; Noor 1999). species (the “cascade effects” hypothesis, see text Despite these obstacles, there are convincing for details). m refers to migration rate among conspe- examples that reinforcement operates in na- cific sympatric and allopatric populations of a species ture (Noor 1995; Saetre et al. 1997; Rundle & undergoing reinforcement. p (mismating) is the prob- ability that a female from a given area mates with a Schluter 1998; Hoskin et al. 2005; Silvertown heterospecific male versus a conspecific male. et al. 2005; Jaenike et al. 2006; Kronforst et al. 2007; Nosil et al. 2007; Urbanelli & Porretta 2008), and this process appears to have con- rant further attention. First, the genetic basis of tributed to the origin of diverse lineages of female preferences and the link between the butterflies, fish, fruit flies, birds, walking stick genetics of prezygotic and postzygotic isola- insects, frogs, and flowering plants (e.g., Noor tion have only been dissected in a few cases of 1997; Lukhtanov et al. 2005; Kay & Schemske reinforcement. As discussed below, certain ge- 2008). Servedio and Noor (2003) reviewed the netic architectures facilitate reinforcement and empirical and theoretical evidence for rein- help overcome many of the theoretical ob- forcement and highlighted five areas that were stacles associated with the homogenizing ef- particularly poorly understood, and thus war- fects of gene flow on speciation. Second, the 158 Annals of the New York Academy of Sciences ecological basis of prezygotic and postzygotic uals may waste time or energy courting het- reproductive isolation is poorly documented. erospecifics that always reject them (Schluter If ecological sources of selection against hy- 2001), which might have further negative fitness brids are widespread, then the selection that consequences such as increasing predation risk drives reinforcement itself could be more com- (Albert & Schluter 2004). In these scenarios, mon than currently thought. Third, connec- failed reproductive attempts before the forma- tions between the evolution of prezygotic iso- tion of zygotes can also create enough selective lation between species and sexual selection pressure to promote the evolution of prezygotic within species are likely,but poorly understood. isolation (Higgie et al. 2000). For simplicity, we Fourth, habitat preferences might also be rein- discuss in subsequent sections the connections forced, but the relative roles of habitat versus between prezygotic and postzygotic reproduc- mate preference in reinforcement are relatively tive isolation, but acknowledge that prezygotic unexplored both theoretically and empirically. isolation might evolve in response to any cost In particular, it is unknown how much these to hybridization, including costs associated with different mechanisms of reinforcement might simply attempting to hybridize. facilitate or constrain one another. Fifth, com- One of the fundamental problems of the the- parative studies outside of Drosophila are scarce, ory of reinforcement is explaining how such yet are necessary if any broad generalities about genetic associations between prezygotic and patterns for the evolution of reproductive iso- postzygotic isolation are maintained when con- lation are to be inferred. Here we review and fronted with the homogenizing effects of gene synthesize recent work that addresses these un- flow (Felsenstein 1981). Certain genetic sce- resolved issues and suggest new avenues for test- narios mitigate this problem. For example, re- ing the importance of reinforcement on the ori- gions of low recombination, like sex chromo- gins of biodiversity. somes (Lemmon & Kirkpatrick 2006; Saether et al. 2007), and chromosomal inversions (Noor et al. 2001b), can help maintain linkage dise- The Genetic Basis of Reinforcement: quilibrium among genes within them (Ortiz- Inversions, Sex Chromosomes, and Barrientos et al. 2002; Butlin 2005). Thus, re- One-Allele Assortative Mating inforcement is predicted to be possible when Mechanisms genes affecting hybrid fitness and assorta- tive mating reside within regions of reduced Reinforcement in the traditional sense re- recombination. lies on genetic associations between prezy- There are other genetic scenarios in which gotic isolation (including assortative mating) linkage disequilibrium between forms of repro- and postzygotic reproductive isolation (that is, ductive isolation is maintained. One stems from low hybrid fitness). This is because such associa- the conceptual distinction between one-allele tions are required for selection against hybrids, and two-allele models described by Felsenstein to result in the evolution of reproductive prezy- (1981) (see Servedio 2009 for a detailed review gotic isolation. Selection against hybrids itself of these mechanisms). In a one-allele mecha- might arise from either extrinsic (ecological) or nism, reproductive isolation is caused by the intrinsic causes (such as genetic incompatibil- same allele fixing in both populations (for ex- ity) (Coyne & Orr 2004). However, reinforce- ample, an allele causing individuals to prefer ment in the “broad sense” (Servedio & Noor mates phenotypically similar to themselves). 2003) can arise via any type of reproductive Such alleles cause nonrandom mating regard- cost driving the evolution of prezygotic isola- less of genetic background (assuming the pop- tion, including costs that arise prior to hybrid ulations differ already in the trait that is being formation. For example, pure species individ- used as a mating cue; for instance courtship Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 159

Figure 2. The most conducive genetic architectures for the evolution of prezygotic isolation in response to maladaptive hybridization (following Lemmon & Kirkpatrick 2006). X and A denote sex and autosomal chromosomes, respectively, p the female preference, and X-A, A-A, or X-X genetic incompatibilities between species, whose interaction is denoted with an arrow. (A, B) Both preference and genetic incompatibilities have the same mode of inheritance; (C, D) X-A genetic incompatibilities are required regardless of the location of the preference. Note, however, that the second type of genetic incompatibility (X-X or A-A) is favorable depending on the location of the preference. song, body size, or pheromones might already tween diverging populations and complete the differ between the two hybridizing species). In speciation process. a two-allele mechanism, different alleles fix in Figure 2 depicts some of the most con- each population (for example, a preference al- ducive genetic architectures for reinforcing lele for large individuals in one population and prezygotic isolation (based on Lemmon & Kirk- a different preference allele for small individu- patrick 2006). When genetic incompatibilities als in the other). and mate preference genes share the same in- This distinction is important when consider- heritance (i.e., are on the same chromosome; ing the effects of recombination: recombination Fig. 2A, B), or when genetic incompatibilities in a two-allele mechanism breaks down linkage and mate preference genes are between the disequilibrium, weakening the genetic associ- sex chromosome and an autosome (Fig. 2C, ation between genes affecting pre- and post- D), prezygotic isolation usually spreads (Kelly zygotic isolation, while in a one-allele mecha- & Noor 1996; Lemmon & Kirkpatrick 2006). nism, linkage disequilibrium between pre- and This suggests a special role for the X (or Z) chro- postzygotic isolation exists automatically and mosome in reinforcement, most likely due to its cannot be reduced by recombination. Similarly, reduced recombination rate and because of its alleles that affect a trait which in turn affects pronounced effects on the heterogametic sex both post- and prezygotic isolation (or “magic (because recessive genetic incompatibilities are traits” [Gavrilets 2004]) also facilitate the evo- covered in heterozygotes, but expressed pheno- lution of prezygotic isolation in the face of gene typically in homozygotes). flow. In both the one-allele and magic-trait sce- Although few studies have looked in detail narios, linkage disequilibrium between forms of at the genetics of reinforcement, several stud- reproductive isolation is unopposed by recom- ies have looked at the genetics of hybrid steril- bination (Servedio 2009). Thus, it is possible, ity (Table 1 and references therein) or invia- for instance, that a single allele can spread be- bility (Adam et al. 1993; Barbash et al. 2000; 160 Annals of the New York Academy of Sciences Continued Araripe, et al. 2007; Tao, Masly, et2007) al. 2007) 2007; Tao, Masly, et al. 2007) et al. 2004) Orr et al. 2007) (Tao et al. 2001; Tao, (Masly & Presgraves (Tao, Araripe, et al. (Ting et al. 1998; Sun (Orr & Irving 2005; (Nakazato et al. 2007) (Moehring et al. 2007) faster-male theories are supported; Large X effect: sex-ratio distortion, dosage compensation or X inactivation result from genetic incompatibilities evolving by intragenomic conflict within species DNA sequence contributes to hybrid sterility species and suppressed systems of meiotic drive within species share genetic basis species and suppressed systems of meiotic drive within species share genetic basis involving cytoplasmic effects, cause hybrid sterility transcriptional failure in their hybrids in spermatogenesis genes may lead to Haldane’s rule: Both dominance and The genetics of hybrid sterility may Divergence in both expression and Genetic incompatibilities between Genetic incompatibilities between Genetic incompatibilities, possibly Regulatory divergence between species that also D. simulans hybrid male sterility (compared to female sterility and general hybrid inviability) and act recessively. There is a disproportional contribution of the X to hybrid sterility. retroposition from another gene on the X. Sex-ratio suppressor is an autosomal retroposon gene of the distorter. RNAi is a likely mechanism of suppression. homeobox gene suppressor of an X-linked sex-ratio distortion in causes hybrid male sterility to the same genetic locus Mendelian segregation, but a single QTL for spore viability was identified profiles between hybrids between different pairs of species. Spermatogenesis disruption appear to be postmeiotic. Most introgression segments cause X-linked distorter evolved from Identification of a rapidly evolving Identification of a dominant autosomal Sterility and segregation distortion map A large number of loci departed from Significant overlap on missexpression D. mauritiana D. mauritiana D. mauritiana D. D. simulans chromosome segments into D. sechellia a sex-ratio distortion/ suppressor system chromosome segments into D. simulans 3rd chromosome segments into X-linked regions between subspecies of pseudoobscura throughout the genome of hybrid populations of haploid lines and F2s spermatogenesis genes expression profiles between parentals and their hybrids Introgression of Mendelian segregation tests Introgression of Comparison of Introgression of Introgression experiments of Molecular characterization of Studies Examining the Genetic Basis of Hybrid Sterility sechellia races mauritiana mauritiana, D. sechellia mauritiana D. mauritiana x D. D. pseudoobscura SystemCeratopteris richardii Methodology Findings Implications Reference Drosophila simulans D. simulans x D. D. simulans x D. D. simulans x D. TABLE 1. Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 161 2005; Moyle & Nakazato 2008) 2008b) 2008) 2006a, 2006b) (Moyle & Graham (Good et al. 2008a, (Greig 2007, 2008) (Michalak & Malone (Sweigart et al. 2006) (Chen et al. 2008) (Dettman et al. 2007) (Masly et al. 2006) (Moehring et al. studies, the Drosophila contributing to pollen sterility and other hybrid dysfunctions; QTL for pollen and seed sterility are colocalized X-linked genetic incompatibilities, some polymorphic not cause yeast hybrid sterility effects may lead to hybrid sterility genetic basis of sterility is simple variability for alleles causing reproductive isolation evolution of postzygotic isolation incompatibilities, may cause hybrid sterility evolution of pigmentation genes contribute to hybrid sterility Possible large-X effect; recessive Recessive genetic incompatibilities do Imprinting and transgenerational Incipient speciation shows genetic Ecological adaptation can lead to the Mechanisms, other than typical genetic Large X-effect and possible adaptive In contrast to QTL on X contains few genes including a fast-evolving spermatogenesis gene deathofhybridgametes micro-RNAs relative to parentals contributes to patterns of female gamete abortion in hybrids incompatibilities and not underdominant effects or antirecombinational effects chromosomes in the two species. Sterility results from the absencethe of gene in the homozygous introgression line. and colocalization of hybrid sterility loci with those causing pigmentation differences between species incompatibility causes nearly complete male sterility Asymmetric X-linked hybrid sterility. No chromosomal introgression caused Hybrid testis are depleted of Three alleles from an aspartic protease Evolved sterility involves genetic Gene, JYAlpha, locates to different Disproportional contribution of the X, A two-locus dominant–recessive S. D. S. paradoxus background chromosome background X-chromosomes between 2 species cerevisiae chromosomes into profiling in parentals and their hybrids sterility genes early stages of speciation D. simulans segments into a melanogaster and deficiency mapping of sterility gene hybrid sterility loci lines genetic mapping of sterility loci Reciprocal introgression of Nearly isogenic lines mapping Pollen and seed sterility Similar number of genomic regions Micro-RNA expression Map-based cloning of hybrid Experimental evolution of the Introgression of Homozygous introgression of Backcross genetic mapping of Backcross and nearly isogenic and X.

Continued and M. domesticus habrochaites x S. lycopersicum muelleri japonica cerevisiae paradoxus santomea melanogaster M. nasutus Solanum penellii, S. X. laevis Mus musculus Oryza sativa x O. Saccharomyces S. cerevisiae x S. D. yakuba x D. SystemD. simulans x D. Methodology Findings Implications Reference Mimulus guttatus x TABLE 1. 162 Annals of the New York Academy of Sciences Continued 2005; Moyle 2007; Moyle & Nakazato 2008) Moyle 2007; Moyle & Nakazato 2009) 2001; Greenberg et al. 2003) Doi et al. 2001; Sawamura et al. 2006b) 2006) (Moyle & Graham (Takahashi et al. (Moehring et al. but weak effects Not detected discrimination and 3 for male mating success 3 Significant 3 Not detected (Shug et al. 2007; At least 3 for single locus autosomal locus with moderate to large effects; X-effects negligible Sex- Locus Number Prezygotic isolation Postzygotic isolation Yes N.A. Weak 2–6 N.A. (Moyle 2005; Yes No One major Yes No Few autosomal Single Sex- Locus Negative Possibly few 4–11 QTL N.A. Strong Possibly rare fertility morphology hydrocarbons male and female dis- crimination 1 MY Pollen and seed 1 MY Floral 0.1 MY Cuticular 0.4 MY Components of Recent Female choice Yes Yes One major ∼ > ∼ ∼ Z Concurrent Exploration of the Genetic Architecture of Prezygotic and Postzygotic Isolation lycopersicum x S. esculentum, S. habrochaites lycopersicum x S. esculentum, S. habrochaites melanogaster and M races ananassae yakuba Solanum Solanum Drosophila System D Trait incompatibilities Autosomal linked effect epistasis References System D Trait Autosomal linked effect of QTL Epistasis References D. pallidosa x D. D. santomea x D. TABLE 2. Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 163

Presgraves et al. 2003; Presgraves & Stephan 2007), and the genetics of female preferences in general (Table 2 and references therein, see Presgraves 2007) et al. 2006) et al. 2006a) also Etges & Noor 2002; Arbuthnott in press). (Masly & Many studies on the genetics of hybrid sterility have found strong genetic effects on the X chro- mosome, as well as multiple genetic incompat- ibilities that involve the sex chromosome and an autosome. Similarly, many studies tend to Common Present (Sawamura Not detected (Moehring find a disproportionate genetic effect of the sex chromosome on female mate preferences. Al- though none of these studies deals directly with a case of reinforcement, they do suggest that the overall genetic variability conducive for re- recessive recessive– recessive interactions are strong inforcement exists in nature (Fig. 2). Strong and fairly Autosomal effects Two major studies on the genetics of reinforcement illustrate how suitable genetic architectures promote speciation with gene flow. Drosophila pseudoobscura and D. persimilis are hybridizing species of fruit flies that produce least 9 sterility QTL when X is from D. santomea sterile F1 hybrids (Dobzhansky 1937a). They 1 viability and at co-occur (sympatry) in the northwest coast of the United States and western Canada, with D. pseudoobscura extending its range into central United States and Central America (allopatry). Females derived from sympatry consistently sterility QTL Postzygotic isolation copulate more readily with males from the same species than with males from the other species (that is, prezygotic isolation has evolved between the two species). Conversely, females

Single Sex-derived Locus from Negative areas where D. pseudoobscura Unknown 3 viability and 8 Unknown 2 QTL Not detected Possibly moderate Unknown 3 QTLoccurs Possibly 1 QTL alone tend to mate as readily with heterospecific as with conspecific males in a laboratory setting (Noor 1995). This type of reproductive character displacement (in the presence of maladaptive hybridization) is the signature of speciation by reinforcement. and fertility breakdown behavioral sterility

Hybrid Genetic incompatibilities causing hybrid sterility, as well as basal female preferences for conspecifics (those present between allopatric 0.3 MY Hybrid viability 0.4 MY Hybrid populations of the two species), consistently ∼ ∼ map to chromosomal inversions located on the X and 2nd chromosomes (Noor et al.

Continued 2001a, 2001b). These observations are con- sistent with Lemmon and Kirkpatrick’s (2006) predictions that the most conducive genetic D. sechellia ananassae yakuba D, divergence time; MY, million years before the present; N.A., not applicable; QTL, Quantitative Trait Loci. D. mauritiana x System D Trait incompatibilities Autosomal linked effect epistasis References D. pallidosa x D. D. santomea x D. TABLE 2. architecture for reinforcement involves 164 Annals of the New York Academy of Sciences interactions between a sex (X, or W) and contributing to the evolution of prezygotic iso- an autosomal chromosome (Fig. 2), as well lation when species exchange genes (Ortiz- as with chromosomal models of speciation Barrientos & Noor 2005). In the one-allele where the antirecombinational effects of chro- model scenario, the high mating discrimination mosomal rearrangements maintain genetic allele derived from D. pseudoobscura in sympatry associations between forms of reproductive is expected to cause assortative mating when isolation (Noor et al. 2001b; Rieseberg 2001; introgressed into a D. persimilis genetic back- Ortiz-Barrientos et al. 2002; Navarro & Barton ground. The low discrimination allele from al- 2003; Kirkpatrick & Barton 2006). All together, lopatry is expected to have a neutral effect on these data suggest that the observed genetic mating discrimination or perhaps increase the architecture in D. pseudoobscura and D. persimilis frequency of matings with heterospecifics. In a is favorable for reinforcement (Fig. 3A, B). 2-allele model, in contrast, the allele derived The genetic basis of female preferences be- from sympatric pseudoobscura is expected to con- tween sympatric and allopatric populations of fer stronger female preference for a D. pseudoob- D. pseudoobscura also reveals fascinating insights scura phenotype. As such, a D. persimilis female about reinforcement. Ortiz-Barrientos and col- carrying such allele should also prefer such phe- leagues (Ortiz-Barrientos et al. 2002, 2004) notype and thus mate more often with D. pseu- found two quantitative trait loci (QTL), Coy-1 doobscura males than if carrying an allele derived and Coy-2, that contribute to the differences in from allopatric populations of D. pseudoobscura. mating discrimination between D. pseudoobscura The D. pseudoobscura data strongly suggest that females derived from sympatry (which exhibit a one-allele model might have operated dur- high mate discrimination) and from allopatry ing reinforcement between this species and D. (low mate discrimination). These QTL con- persimilis tained as few as five genes and were located The genetic basis of reinforcement has also in two chromosomal regions. Coy-1 was located been studied in great detail in the European next to a chromosomal inversion on the right flycatcher birds, Ficedula hypoleuca and F. albicolis arm of the X-chromosome, with the antirecom- (Saetre et al. 1997). Males of both species (that, binational effects of the adjacent inversion fa- for the most part, have allopatric ranges) have cilitating its persistence (Machado et al. 2007; black and white plumage. However, in zones of Noor et al. 2007, but see; Yatabe et al. 2007). In contact, the males of the two species vastly dif- contrast, the location of Coy-2 was not near or fer in plumage coloration. Plumage differences within an inversion, which suggested that some appear to be derived characters within each of reinforcement genes might have experienced the two species, suggesting that they evolved in high levels of gene flow between the two hy- sympatry, possibly in response to maladaptive bridizing species. This implied that the effects hybridization (especially given that hybridiza- of these genes on mating discrimination could tion between these species is common and that operate regardless of genetic background, as F1 hybrids have low fitness compared to the occurs in one-allele models. Consistent with parental species). Females in sympatry display this suggestion, Ortiz-Barrientos & Noor (2005) stronger preferences for the distinct, and con- showed that alleles conferring strong or weak specific, flycatcher males, leading to assortative assortative mating in D. pseudoobscura produce mating. Using a series of experiments in the the same relative strength of assortative mating field and in the lab, Saether et al. (2007) com- when inserted into D. persimilis. This suggests pared expectations of one-allele models ver- that the same assortative mating locus occurs sus recombination–suppression models (specif- in both species (a one-allele mechanism). ically, mechanisms invoking chromosomal in- Figure 3C shows the contrasting predic- versions or sex chromosomes) in the evolution tions of one- and two-allele models for a gene of assortative mating. The authors contrasted Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 165

Figure 3. The genetics of reinforcement in D. pseudoobscura.(A) Genetic architecture of genetic incompatibilities and female preferences between D. pseudoobscura and D. persimilis, (B) genetic architecture of reinforcement in D. pseudoobscura,(C) predictions derived from one-allele versus two-allele models in D. pseudoobscura and D. persimilis. D. persimilis females carried either a chromosomal segment with discriminating alleles derived from sympatric D. pseudoobscura females, or carried one with less discriminating alleles derived from allopatric D. pseudoobscura females. In the one-allele model, the discriminating alleles favor assortative mating, and therefore D. persimilis females carrying the D. pseudoobscura sympatric alleles would tend to discriminate strongly against D. pseudoobscura males. In contrast the two-allele model predicts that D. persimilis females carrying the D. pseudoobscura sympatric alleles would have a strong preference for a trait specific to D. pseudoobscura. Therefore, such females would tend to prefer D. pseudoobscura to D. persimilis males. X and A denote sex and autosomal chromosomes respectively, p the female preference, and X-A, A-A, or X-X genetic incompatibilities between species, whose interaction is denoted with an arrow. the effects of imprinting (a one-allele model) some or paternal imprinting of preferences on with the effects of sex linkage on the Z chro- female offspring could explain sympatric fe- mosome (a recombination–suppression model) male preferences for distinct plumages. Using on female preferences. Since both genes for a combination of both behavioral and genetic genetic incompatibilities and genes for female experiments, Saether et al. (2007) found conclu- mate preferences map to the Z chromosome, sive evidence that Z-linkage of genes contribut- they are subjected to rounds of recombination ing to various forms of reproductive isolation, only in females. However, because females in- and not imprinting, facilitated the evolution of herit the Z chromosome from their fathers both prezygotic isolation in hybridizing flycatchers recombination–suppression on the Z chromo- (see also Albert 2005). 166 Annals of the New York Academy of Sciences

Together, the cases of reinforcement in Timema cristinae walking stick insects, intrinsic Drosophila and in Ficedula illustrate the genetic genetic incompatibilities in hybrids are lack- conditions under which natural selection can ing and sources of selection against hybrids ap- more easily complete speciation in the face pear primarily (if not solely) ecological. In both of gene flow: genetic incompatibilities can in- systems, there is evidence for RCD and rein- volve both the sex and autosomal chromosomes forcement (Rundle & Schluter 1998; Hatfield (Noor et al. 2001a; Ortiz-Barrientos et al. 2004; & Schluter 1999; Nosil et al. 2003, 2007). Saether et al. 2007), but regions of the genome Second, ecological differences between pop- that prevent the formation of recombinant hy- ulations might also affect prezygotic isolation. brids (like chromosomal inversions and sex- For example, during the process of “sensory chromosomes) play a key role in the strength- drive” mating signals evolve to be most de- ening of prezygotic isolation as they maintain tectable in the environment they are transmit- genetic associations between genetic incompat- ted in, and sensory/perceptual systems evolve ibilities and sexual preferences. Sexual prefer- to be tuned to these signals (Endler & Basolo ences appear to have the same mode of in- 1998 for review; Boughman 2002; Nosil, et al. heritance as genetic incompatibilities. Finally, 2003). As a consequence, fine-tuning of sig- the evolution of assortative mating in some nals to divergent environments (adaptive diver- cases involves one-allele mechanisms, but the gence) has the potential to produce differences prevalence of these in nature is unknown in mating signals, perceptual abilities, and mat- (Ortiz-Barrientos & Noor 2005). Overall, these ing preferences between populations. Repro- processes generate genomic divergence that is ductive barriers can evolve as a byproduct of heterogeneous, with some regions causing re- this adaptive divergence in mating signals. The productive isolation, and with others freely ex- contributions of sensory drive to reinforcement changed between species (Saetre et al. 1999; are poorly studied, although in some cases both Noor et al. 2001a), possibly a genetic signature processes are known to operate within the same of speciation with gene flow (Via & West 2007). system (e.g., stickleback fishes, see Boughman 2001). Third, ecology affects numerous processes The Ecological Basis of Prezygotic other than reinforcement. These other pro- and Postzygotic Reproductive cesses can either confound or promote the Isolation effects of reinforcement. One example is com- petition for resources, which can generate ac- We cover three topics here pertaining to centuated divergence between sympatric versus how ecological differences between populations allopatric taxa, a pattern referred to as “ecolog- might contribute to reinforcement. First, eco- ical character displacement” (ECD, Schluter logical differences between populations can af- 2000 for review). Separating the effects of fect postzygotic isolation; namely,ecological di- reinforcement and competition on levels of vergence can result in hybrids exhibiting low prezygotic isolation may be difficult; both fitness (for example, if an intermediate hy- occur in sympatry from the interaction of brid phenotype renders them ecologically mal- populations and can produce the same evo- adapted to either parental environment, and lutionary outcome—stronger prezygotic isola- an intermediate niche is unavailable [Rundle tion between sympatric populations (Servedio & Whitlock 2001]). Indeed there are some ex- & Noor 2003). The extent of this problem amples where reinforcement is likely driven will not be known until we determine how by ecological causes, that is, ecologically de- frequently prezygotic isolation is strengthened pendent selection against hybrids. In ecotypes as a byproduct of ecological character dis- of Gasterosteus aculeautus stickleback fishes and placement. The problem could be serious, as Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 167

Figure 4. Multiple types of interactions between species might contribute to phenotypic divergence and speciation within a single system, in this case speciation between sympatric limnetic and benthic forms of Gasterosteus aculeatus stickleback fishes. (A) Illustrations of the two forms (courtesy of L. Nagel). (B) Reproduction character displacement (RCD) due to reinforcement. Modified from Rundle and Schluter (1998) and reprinted with permission of Wiley-Blackwell. (C) Ecological character displacement (ECD) in trophic traits due to interspecific competition. Modified from Schluter and McPhail (1993) and reprinted with permission of Elsevier. (D) Character displacement (CD) in defensive armor due to predation. Modified from Vamosi and Schluter (2004) and reprinted with permission of Wiley-Blackwell. exemplified by studies of Gasterosteus aculeatus dation was implicated in character displace- stickleback fishes, which show that multiple ment of armor (Vamosi & Schluter 2004) processes can simultaneously contribute to (Fig. 4). How much the latter two forms of dis- character displacement: reinforcement was im- placement affect the first is unknown, although plicated in reproductive character displace- the tests of reinforcement in this system did ment of mating preferences (Rundle & Schluter control for ecological character displacement 1998; Albert & Schluter 2004); competition (Rundle & Schluter 1998). was implicated in ecological character displace- In summary, because different processes can ment of trophic traits (Schluter & McPhail yield similar patterns, they can confound each 1992; Schluter 1994; Schluter 2003); and pre- other’s effects. Alternatively and on the upside, 168 Annals of the New York Academy of Sciences the different processes might also complement fitness than their sympatric counterparts. How- one another. For example, ecological charac- ever, these females are better at distinguish- ter displacement due to competition could re- ing conspecific males from heterospecific males sult in reduced hybrid fitness (e.g., Pfennig & and thus do not incur the hybridization costs Rice 2007), which then drives the process of that allopatric migrants (to regions of sympatry reinforcement. Future studies of reinforcement with the other species) do (Pfennig & Simovich should consider the different effects of ecolog- 2002; Pfennig 2003; Pfennig & Pfennig 2005; ical divergence on prezygotic and postzygotic Pfennig 2007). In contrast, females of allopatric isolation, as well as the effects of different types Drosophila serrata reject conspecifics from the of ecological interactions. area of sympatry with D. birchii, possibly be- cause both male mating cues and female pref- Connections between the Evolution erences are tuned optimally (Higgie & Blows of Prezygotic Isolation between 2007). Species and Sexual Selection Whether this kind of conspecific rejection within Species can lead to the formation of new species has rarely been explored. An exception concerns Although species recognition and sexual se- different species of frogs from the genus Lito- lection within species are inherently related ria. These species appear to have speciated (Ryan & Rand 1993 for a review), how they by reinforcement, but, because of increases in interact during speciation has received little at- prezygotic isolation due to reinforcement, they tention (Howard 1993; Higgie & Blows 2007; have also evolved divergent mate preferences Lemmon 2009). This is a gross oversight, as between populations within species (Hoskin the evolution of prezygotic isolation in response et al. 2005). A similar pattern has been reported to maladaptive hybridization between incipi- in Timema cristinae walking stick insects. In this ent species might clearly affect mating patterns, system, although females are selected to be and even the evolution of prezygotic isolation, more discriminating against males from a single within species. For instance, if female mate adjacent population that is adapted to feeding recognition systems are under strong stabilizing on a different host plant species, this selection selection in sympatry, then choosy females that has indirectly resulted in increased mating dis- preferentially mate with conspecifics in sympa- crimination against foreign males from multiple try may delay their mating with highly suitable other populations, including even males from conspecifics in allopatry (Ortiz-Barrientos pers. other populations that use the same host (Nosil observ. and Nosil 2007). Thus, sympatric fe- et al. 2003; Nosil 2007). males might incur a cost in allopatry for being We propose that this mechanism—by which “choosier.” Overall, the net gain in fitness of the effects of reinforcement within a particular being choosy is high in sympatry because mal- taxon pair (e.g., a species pair) cascade to inci- adaptive hybrids are not produced, but low (or dentally result in reproductive isolation among even negative) in allopatry. other taxon pairs (e.g., between sympatric and A major trend across studies examining links allopatric populations within the species in the between reinforcement and sexual isolation aforementioned pair)—be termed the “cascade within species is that mating discrimination reinforcement hypothesis” (Fig. 1). We further evolving in response to maladaptive hybridiza- postulate that this hypothesis will often in- tion induces widespread effects on mate choice volve sexual selection within species. Such cas- within species (Table 3, Fig. 1). For instance, cade effects of reinforcement on the evolution in frogs from the genus Spea, sympatric females of reproductive isolation within species may of S. multiplicata reject allopatric males of the be due to females recognizing and preferring same species despite such males having higher males from their own population based on a Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 169 population-specific trait, instead of an ecology- lution of prezygotic isolation among conspecific specific or species-specific trait (Zouros & populations. Presumably, this newly evolved D’Entremont 1980; Kelly & Noor 1996; Higgie postzygotic isolation would feed back on et al. 2000; Hoskin et al. 2005; Lemmon 2009). strengthening prezygotic isolation. These types If such cascade effects are common, then re- of feedback loops between prezygotic and post- inforcement could contribute to speciation be- zygotic isolation have been noted before (see tween ecologically similar pairs of populations, Servedio 2009 for a review), but they have not between populations that are geographically been extended to the geography of speciation. separated from one another, and between con- The effects of reinforcement on mating be- specific populations (Pfennig & Ryan 2006). havior within species also provide insights into The cascade reinforcement hypothesis is our ability to detect reinforcement. For exam- likely to operate when recognition systems in ple, in species experiencing high levels of gene females act in similar genetic ways between flow between sympatric and allopatric popu- and within species (for example, stickleback lations of the same species, we should expect fish tend to use body size as cue for choos- alleles causing increased prezygotic isolation ing mates both between and within species between species (such as female mate dis- [Schluter 2001]). However, changes in other crimination against males from other species) fitness components between species (e.g., due to spread over the entire species range rela- to adaptation to the environment) could also tively quickly, erasing the pattern of reproduc- incidentally drive the evolution of reproduc- tive character displacement that is traditionally tive isolation among conspecific populations. used to detect reinforcement. This is particu- Therefore, a tendency for different traits and larly likely to occur if mating discrimination, genes to be involved in between- versus within- which evolved in sympatry, has little or no cost species mate choice (Arbuthnott 2009), would in areas of allopatry (for example, if prezygotic not preclude the possibility of cascading rein- isolation between species is neutral with respect forcement. However, these differences in mate to sexual selection within species). It is in such choice genetics make a simple prediction: when scenarios that cases of reinforcement may go females use the same cues between and within unrecognized (Walker 1974). Why then, have species, conspecific reproductive isolation will researchers typically found signatures of re- tend to also be behavioral. In contrast, when inforcement in many instances of divergence interspecific cues are different from those used (Fig. 1)? within species, reproductive isolation among One answer is that alleles conferring prezy- conspecific populations might have been based gotic isolation between species fail to spread to on traits involved in divergent ecological adap- the rest of the species range (i.e., to the allopatric tation (such as morphological or physiological range). Such a failure of the spread of alle- traits). les conferring discrimination could occur for The cascade reinforcement hypothesis pre- numerous reasons including strong asymmet- dicts that the evolution of prezygotic isola- rical gene flow going from allopatry into sym- tion between sympatric species can lead to patry, physical linkage of genes affecting mate the evolution of reproductive isolation within preference and local adaptation to sympatric species. Interestingly,because reinforcement af- ecological conditions, and assortative mating fects only prezygotic isolation, the process of alleles that are advantageous in sympatry but cascade speciation among conspecific popula- disadvantageous in allopatry (such as rejection tions could proceed in the absence of postzy- of suitable mates in allopatry (Moore 1957; gotic isolation among these conspecific popu- Barton & Hewitt 1985; Butlin 1993; Howard lations, or it could allow for the evolution of 1993). In particular, mating discrimination postzygotic isolation as a byproduct of the evo- alleles can be disadvantageous in allopatry 170 Annals of the New York Academy of Sciences Continued Pfennig 2005; Pfennig 2007; Pfennig 2008; Reyer 2008) 2005; Smadja & Butlin 2006) 2000; Higgie &Blows 2007; Higgie &Blows 2008) (Pfennig & haplotypes Shared Low (Hoskin et al. High (Higgie et al. leads to sexual selection within species creates new allopatric species sexual selection optimum in allopatry Reinforcement Reinforcement There exists a females reject allopatric males despite high fitness females reject allopatric males females reject sympatric males Sympatric Sympatric Allopatric selection on body size selection on body size hydrocarbons Call, indirect Call, indirect is is S. L. S. Intrinsic Preferred if multiplicata the mother; facultative overdomi- nance if bombifrons the mother when genimaculata form S is the mother, and normal hybrid development if the N form is the mother Reduced fitness Hybrid lethality found found Commonly Commonly Unknown Absent Cuticular agent in the isolation between and preference between Selective Hybrids postzygotic male trait Female Gene flow inviability inviability attempting to mate het- erospecifically Hybrid Hybrid Wasted time Mating Discrimination Effects on Mate Choice Within Species S. D. birchii and bombifrons genimaculata forms N and S and Spea multiplicata System studied wild effects within species effects Implications populations References Litoria Drosophila serrata TABLE 3. Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 171 ) ∗ Noor et al. 2001a, 2001b; Ortiz- Barrientos et al. 2004; Ortiz- Barrientos & Noor 2005 unpublished results 2005; Jaenike et al. 2006) 2002; Nosil et al. 2003; Nosil 2007) High (Noor 1995; High (Telschow et al. High (Nosil et al. leads to sexual selection within species leads to sexual selection within species leads to sexual selection within species Reinforcement Reinforcement Reinforcement ∗ females reject allopatric males females reject foreign males regardless of host association females court slowly allopatric males Sympatric but unknown within species Unknown Sympatric Courtship song, is is D. recens D. D. Intrinsic Preferred when is the father, and hybrid male sterility when subquinaria the father sterility is slightly stronger when pseudoobscura the mother Hybrid lethality Absent Body size Sympatric hybridization in the past found Evidence of Commonly Rarely found Hybrid male agent in the isolation between and preference between Selective Hybrids postzygotic male trait Female Gene flow induced hybrid inviability inferiority sterility Wolbachia Extrinsic hybrid F1 hybrid male

Continued D. recens D. and host plant ecotypes and persimilis D. subquinaria System studied wild effects within species effects Implications populations References Timema cristinae D. pseudoobcura TABLE 3. 172 Annals of the New York Academy of Sciences because: (1) they cause increased mating dis- sympatric species. However, selection might crimination against suitable conspecific males favor the evolution of other forms of prezy- in allopatry, (2) allopatric females could reject gotic isolation, such as various migration mod- sympatric males that have changed their mor- ification behaviors (Mayr 1942; Fisher 1958; phology or behavior as a correlated response to Balkau & Feldman 1973; Yukilevich & True the evolution of female prezygotic isolation, (3) 2006). An example of such a behavior is habi- females may have traded their ability to choose tat preference, which reduces movement and the fittest conspecific mates for the ability to thus gene flow between habitats (Coyne & Orr discriminate against heterospecifics so that in 2004). Like mate preference, migration mod- allopatry they choose less fit males (which is ification can evolve to avoid maladaptive hy- disadvantageous), or (4) they traded their own bridization. In addition to selection against fecundity/survival (or the fecundity/survival of hybrids, selection might often act on immi- their offspring) for the ability to choose conspe- grant parental genotypes that migrate to an cific mates, which would make them less fit in alternative habitat (Hendry 2004; Nosil et al. allopatry (so the alleles that confer discrimina- 2005). Whether scenarios in which selection tion are selected against in allopatry). Finally, acts against immigrants should be considered even if alleles for increased mating discrimina- examples of reinforcement in the traditional tion are neutral outside of sympatric regions, sense is debatable (Butlin 1995). However, se- it may take a long time for them to spread lection against hybrids and immigrants can out of sympatry and across the species range, clearly drive the evolution of both types of particularly when effective population sizes are prezygotic isolation: mate preference and mi- large and hybrid zones narrow.In all such cases, gration modification (such as habitat prefer- we expect the pattern of reproductive charac- ence). ter displacement to linger for longer periods The relative importance of these two dif- of time, although these conditions might actu- ferent mechanisms of reinforcement, and how ally interfere with the completion of speciation they interact, remains unclear. A recent theo- (that is, the spread of between-species mating retical model examined relative fixation prob- discrimination throughout the species range, abilities for mate preference (their assortative including allopatry). mating) and migration modification when the In conclusion, the consequences of geo- two evolved simultaneously (Yukilevich & True graphically localized reproductive isolation are 2006). An extension of this model examined vast, and they include the rapid formation of allele frequencies in polymorphic populations new subspecies, the evolution of strong sexual that had yet to fix either form of prezygotic isolation within species, and the evolution of isolation (that is, the early stages of the speci- cascading discrimination. Similarly, changes in ation process) (Nosil & Yukilevich 2008). Both sexual selection as a byproduct of the evolution models found that the relative likelihood of re- of prezygotic isolation can explain why some- inforcement via mate preference versus migra- times the signature of reproductive character tion modification was highly dependent on se- displacement persists despite extensive gene lection strength. flow between populations. Under weak selection, reinforcement by ei- ther mechanism is unlikely.Under intermediate The Role of Habitat versus Mate selection, the conditions favoring the rise and Preference in Generating fixation of one mechanism favored the rise and Prezygotic Isolation fixation of the other. However, mate preference evolved somewhat more readily than migration Research on reinforcement has primarily fo- modification. Populations of T. cristinae walking cused on divergent mating preferences between sticks, that exhibit polymorphism in habitat and Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 173 migration preferences and experience such in- termediate selection, supported these predic- tions (Nosil & Yukilevich 2008). Specifically, both mate and habitat preference are subject to reinforcement in the walking stick system, but mate preference shows more strongly ac- centuated divergence in the face of gene flow (see also Nosil et al. 2006). Under strong selec- tion, the results were very different. The evo- lution of migration modification generally in- terfered with the evolution of mate preference, by decreasing migration between populations and thereby reducing reinforcing selection for assortative mating. Thus, fixation probabilities and allele frequencies for mate preference ver- sus migration modification were negatively cor- Figure 5. Correlation between the probabilities related when selection was strong (Fig. 5). In of fixation for assortative mating (divergent mate essence, divergence in habitat preference can preferences) versus migration modification (divergent habitat preferences) when selection is strong (s in result in a lack of selection to avoid maladaptive the graphs represents selection strength against im- hybridization in geographically sympatric pop- migrant and hybrid matings in both niches). The sce- ulations, because such hybridization no longer nario depicted represents the simultaneous evolution occurs. of each form of reproductive isolation under a niche- A final point is that congruence of the re- based isolation model where selection against im- migrants and hybrids is ecological in nature. Data sults for allele frequencies versus fixation prob- are from (Nosil et al. 2005). Under strong selection, abilities suggests that similar patterns of rein- as depicted here, the evolution of migration modifi- forcement are expected during different stages cation generally interfered with the evolution of as- of the speciation process. These studies as- sortative mating by decreasing migration between sumed a one-allele scenario, and further work populations thereby reducing selection for assortative with a two-allele scenario is required, as are mating. Such patterns were observed for both allele frequencies in polymorphic populations and for fix- analytical solutions to verify the simulation ation probabilities, suggesting that similar patterns results. are expected for different stages of the speciation process. Modified from Nosil & Yukilevich 2008 and reprinted with permission of the Linnean Society. Comparative Studies: The Effects of Geography and Gene Flow forcement (see Coyne & Orr 2004 for a review), few other comparative studies have been con- Comparative studies of reinforcement, ducted. There are some notable exceptions, of which examine numerous population or species which we consider three. pairs, are extremely rare, but are necessary The first example stems from a compara- for inferring broad generalities about reinforce- tive analysis of song structure in 163 species of ment. The most well known example concerns antbird (Thamnophilidae). In this study, habi- Drosophila, and revealed that prezygotic isola- tat differences and reinforcement were both tion evolves faster than postzygotic isolation, found to contribute to divergent song evo- but that this pattern is driven almost entirely by lution (Seddon 2005). Specifically, songs dif- accentuated prezygotic isolation in sympatric fered between habitats and were more diver- taxa (Coyne & Orr 1988). Although this im- gent between sympatric than allopatric species portant finding rekindled enthusiasm for rein- pairs. Although mate preference itself was not 174 Annals of the New York Academy of Sciences measured, if song affects mate choice then dif- ferences between habitats and reinforcement could combine to promote speciation. Sim- ilarly, Lukhtanov and colleagues (Lukhtanov et al. 2005) showed that the sympatric distri- bution of 15 relatively young sister taxa of Agro- diaetus butterflies strongly correlates with differ- ence in male wing color, and that this pattern most likely results from reinforcement. Again though, mate preference itself was not exam- ined. The final example considers multiple con- specific populations of Timema cristinae walking stick insects and addresses theoretical issues re- lating to how relative population sizes and lev- els of gene flow can affect reinforcement. Nu- Figure 6. Among twelve populations of Timema merous theoretical models have demonstrated cristinae walking stick insects, female mating dis- that high levels of gene flow between diverging crimination against males from other populations is populations can erode the effects of reinforc- strongest when gene flow into the population is inter- ing selection, and thus prevent reinforcement mediate. Shown here is the size the host patch occu- (Sanderson 1989; Servedio & Kirkpatrick 1997; pied by a population that is adjacent to a study pop- ulation, relative to the study population itself (gray Cain et al. 1999; Servedio 2004). However, boxes, study population; black boxes, population ad- gene flow also generates the opportunity for se- jacent to study population). X-axis values represent lection against hybridization to occur. Thus mi- the proportion of total area, study population plus ad- gration can exert a dual effect during reinforce- jacent population, that is occupied by the adjacent ment. As noted by Coyne and Orr (2004, pp. population (actual values denoted by black circles on 371) “reinforcement requires some gene flow, the axis). Host plant patch size and insect population size are known to be strongly, positively correlated but not too much.” (r > 0.50, Sandoval 1994). In turn, molecular, mor- A potential prediction is that the effects of re- phological, and behavioral data all indicate that the inforcement are maximized when gene flow is relative size of the population adjacent to a study intermediate. Indeed, several theoretical mod- population is significantly positively correlated with els have observed this to be the case, and in par- the level of gene flow into the study population (Nosil et al. 2003; Nosil 2007, 2008 for details). The y- ticular Kirkpatrick (2000) gave the dual effects axis is female mating discrimination against foreign of migration on reinforcement explicit theoret- males from other populations that use the alternative ical consideration. Nosil and colleagues (2003) host (absolute value of mean copulation frequency examined the effects of migration on the out- with foreign males minus mean copulation frequency come of reinforcement in natural populations of with males from the female’s own population). Boxes walking stick insects. The results demonstrate illustrate the different geographical scenarios. The curve was estimated using the nonparametric cubic the dual effects of gene flow: the magnitude spline (dashed lines show standard errors from 1,000 of female mating discrimination against males bootstrap replicates) (Schluter 1988). Modified from from other populations was greatest when gene Nosil et al. (2003) and reprinted with permission of flow between populations adapted to alternate the Royal Society of London. host plants was intermediate (Fig. 6). In essence, reinforcement was maximized when gene flow choice (i.e. when the sizes of coexisting popula- was high enough to allow the evolution of re- tions are similar). inforcement, but low enough to prevent gene The generality of the walking stick results is flow from eroding adaptive divergence in mate unknown, although at least three other studies Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 175 have documented stronger effects of reinforce- in plants. For instance, flowering plants have ment on the less abundant species within a hy- the potential to evolve prezygotic isolation in bridizing pair, which undergoes more frequent similar ways to marine organisms with open encounters with heterospecifics (Waage 1975; fertilization, in which gametic interactions are Noor 1995; Peterson et al. 2005). Further stud- key to mate choice (Swanson & Vacquier 2002; ies are required to generate insight into whether Palumbi 2008). Plants also exhibit great vari- migration will tend to facilitate divergence via ation in flowering time, pollinator behavior, increased opportunity for reinforcing selection and mating systems (including the evolution or to act as a homogenizing force due to gene of selfing), and as such, these traits also could flow. In particular, some recent models have ex- potentially be recruited when evolving repro- amined the dynamics of reinforcement or char- ductive isolation in sympatry in response to acter displacement along clines or gradients reductions in hybrid fitness. Because many of (Lemmon et al. 2004; Goldberg & Lande 2006) these traits occur at different stages of their life rather than between discrete patches, and em- cycle, plants that shift from one reproductive pirical reinforcement studies considering gene stage to another could possibly escape the costs flow along clines are lacking. of maladaptive hybridization. Consider two hybridizing plant species that produce hybrid offspring with reduced fitness, Reproductive Character with one of them incurring cost to heterospe- Displacement, Maternal cific pollination in mixed pollen loads (for ex- Investments, and Reinforcement ample, heterospecific pollen grains consistently in Plants outcompete or swamp conspecific pollen grains (e.g., Fishman & Wyatt 1999). If there is vari- Most data on reinforcement comes from an- ation in the population for flowering time or imal studies (Coyne & Orr 2004). The few mating system, then alleles responsible for poor carried out in plants have provided mixed re- competitive ability might become associated sults. Moyle and co-workers (2004) found that with alleles contributing to early flowering or in three genera of angiosperms, prezygotic and selfing (Vallejo-Marin & Uyenoyama 2004). postzygotic isolation evolved at similar rates in As a result, poor competitors increase fitness. both allopatry and sympatry (that is, sympa- These associations might be enough to trigger try did not accentuate the rate of evolution). or complete speciation by reinforcement. This In contrast, Kay and Schemske (2008) found mechanism predicts that variation for mating that in a genus of tropical gingers, prezygotic system or RCD for flowering time exists in isolation evolved faster in sympatric pairs than sympatric areas of hybridizing species. Further, in allopatric pairs, possibly in response to poor it predicts that in heterospecific matings be- growth of F1 individuals. As in the case of gin- tween selfers and outcrossers, the selfers should gers, where style length seems to explain most be poorer competitors in heterospecific pollen of the variation in prezygotic isolation, several competition experiments. If selfers happen to other studies have shown that the position of be strong pollen competitors, then most likely anthers (a male trait) may evolve to specify selfing occurs at an appreciable frequency in precise pollen placement on pollinators (Levin sympatry for reasons other than unpredictable 1978). Modification of female and male organs pollination environments (Kalisz et al. 2004). in plants might suggest that behavioral prefer- As mentioned in a previous section, the ge- ences could evolve in plants and perhaps be netic architecture underlying postzygotic and selected during reinforcement. prezygotic isolation, and their correlations, will Understanding how plant traits affect “mate determine the likelihood of reinforcement. A choice” is crucial for studying reinforcement recent study in morning glory species found 176 Annals of the New York Academy of Sciences selection for reproductive character displace- ment in areas of overlap between Ipomea hed- eracea and I. purpurea. Specifically, in sympatric conditions I. hederacea plants with anthers clus- tered around the stigma are less susceptible to heterospecifics pollination from I. purpurea than plants with anthers of different heights (Smith & Rausher 2008). However, floral traits in I. heder- acea constrain character displacement: positive genetic correlations between anther height and floral traits constrain the response to selection for anther clustering (selection for the tallest anther to decrease in size, and selection for the shortest anther to increase in size). This study demonstrates how correlations among Figure 7. Patterns of reproductive character dis- traits under selection, and maternal invest- placement in Australian angiosperm genera. In four ment in producing hybrids, can adversely affect out of seven genera, a sympatric sister species pro- duces larger seeds than the other sympatric pair and the likelihood of reinforcement despite favor- the allopatric sister taxa. Species comparisons in- able genetic architectures, particularly, when volve three sister taxa of which one is allopatric to the such correlations constrain selection trajecto- other two; in this case, taxon C is allopatric to taxa A ries (note that this applies to both animals and and B. Each symbol in the graph represents the mean plants). value for seed mass for one of the species used in a comparison. Boxes show the 95% quantile distribu- Quantitative measures of costs to hybridiza- tion when considering all species simultaneously. tion are usually lacking, and therefore it is difficult to evaluate empirically the effects of such costs on the success of reinforcement (for a notable exception see Pfennig & Simovich Two major predictions follow from the pre- 2002). Maternal investment in hybrid produc- vious discussion. First, sister plant species that tion and its added cost to hybridization re- produce large seeds are more likely to be found mains largely unexplored in reinforcement. All in sympatry than in allopatry, whereas sister things being equal, if two pairs of hybridiz- plants species with smaller seeds are equally ing species incur similar costs to hybridizing likely to be found in sympatry or in allopatry.In (males are sterile, for example), then species a more general sense, as maternal investment in whose females invest relatively more in pro- hybrids increases, the tendency for sister species ducing such hybrids would experience greater to be found in sympatry should increase. Sec- selection against choosing heterospecifics. This ond, we expect to detect reproductive char- difference could increase the chances of evolv- acter displacement for seed mass. Preliminary ing prezygotic isolation in sympatry. Although data derived from seven genera of Australian this is applicable to both animals and plants, flowering plants show that mean seed mass an obvious trait for increased plant maternal is greater for sympatric sister species than for investment is seed mass. Many plants produce the sister allopatric species (Fig. 7), consistent few, but large seeds, whereas others produce with reproductive character displacement for many, but small seeds. Most likely, plants pro- maternal investment. However, maternal in- ducing few, large seeds incur greater costs as- vestment differences could have evolved before sociated with maladaptive hybridization and sympatry. In this case, differences in seed mass therefore might be, on average, more prone to would set an asymmetry in costs associated undergo reinforcement (Coyne & Orr 2004). with maladaptive hybridization and lead to the Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 177 evolution of prezygotic isolation only in one of in nature. Examples include one-allele mod- the two hybridizing species. els, genes for reproductive isolation residing Asymmetries in prezygotic and postzygotic in genomic regions exhibiting reduced recom- isolationarecommoninbothplantsandani- bination, genetic correlations between pref- mals (Turelli & Moyle 2007), and perhaps the erences and alleles causing genetic incom- evolution of maternal investment differences is patibilities, and widespread ecologically based at the heart of such phenomena. In its simplest selection against hybrids. form, asymmetric postzygotic isolation during It has only recently become apparent that reinforcement predicts also asymmetric evolu- speciation by reinforcement brings unexpected tion of prezygotic isolation in sympatry. This consequences. For example, an important ob- is seen in some systems, as in plants from the servation is that evolving prezygotic isolation genus Anigozanthos, where crossing pairs with between species can lead to strong effects on greater pollen sterility in one direction of the sexual selection within species, perhaps to the cross produce fewer seeds than the reciprocal point of cascading into speciation events be- cross that enjoys higher pollen fertility (Steve tween sympatric and allopatric populations of Hopper pers. comm.). Alternatively, seed mass a single diverging species (we have called this differences could have evolved in response to the cascade reinforcement hypothesis). There- maladaptive hybridization. Counterintuitively, fore, we suggest that studies examining prezy- females that already incur hybridization costs gotic isolation between species should make would possibly reduce their chances of hy- additional efforts to consistently measure its in- bridizing by increasing the actual costs of pro- cidental effects on mating behavior between ducing a hybrid; although this appears to be conspecific populations. This phenomenon selection for postzygotic isolation, we should also might explain why we detect reinforce- note that this would be a case for increased ma- ment or reproductive character displacement ternal investment, which, as mentioned above, despite some gene flow between sympatric and increases the overall selective cost associated allopatric portions of the species range. These with hybridization. Subtle similarities can be speciation dynamics suggest that female pref- found in insects, for instance, where some fe- erences often act in similar ways genetically males spend more energy delaying copulation between and within species (or between and possibly at the expense of their own foraging within sympatric versus allopatric populations). for food or mating with conspecifics. In gen- However, this need not be the case for the eral, maternal investment could modulate hy- cascade hypothesis to operate, and the sup- bridization costs and possibly lead to the faster position actually runs counter to reports that evolution of prezygotic isolation in sympatry. traits involved in between- versus within-species mate choice are often not one and the same (Arbuthnott 2009 for review). This discrepancy Conclusions requires further attention. In general, to bet- ter understand the causes and consequences Divergence in the face of gene flow can be of reinforcement, more data are required on unlikely (Felsenstein 1981), yet many cases of the trade off between how enhanced mating sympatric speciation and reinforcement have discrimination (the evolution of prezygotic iso- been uncovered. Although we still remain lation) increases fitness in sympatry versus how largely ignorant about the frequency of these such discrimination might reduce fitness in al- processes, it is evident that both the genetic lopatric areas. architecture of postzygotic and prezygotic iso- Finally,speciation by reinforcement in plants lation and the demographic and ecological remains quite controversial, despite the ob- conditions for speciation with gene flow exist vious favorable conditions for its evolution: 178 Annals of the New York Academy of Sciences multiple paternity, formation of hybrid zones, Albert, A. Y.K., & Schluter, D. (2004). Reproductive char- variable maternal investment in hybrid off- acter displacement of male stickleback mate prefer- spring, unpredictable pollination environ- ence: Reinforcement or direct selection? Evolution, 58(5), 1099–1107. ments, and fertilization assurance mecha- Arbuthnott, D. (2009). The genetic architecture of insect nisms. Genetic correlations between selected courtship behaviour and premating isolation. Hered- traits showing reproductive character displace- ity,inpress. ment and other traits under different selection Balkau, B. J., & Feldman, M. W. (1973). Selection for regimes could heavily constrain the evolution of migration modification. Genetics, 74, 171. prezygotic isolation in sympatry. In conclusion, Barbash, D. A., Roote, J., et al. (2000). The Drosophila melanogaster Hybrid male rescue gene causes inviability although reinforcement remains an active area in male and female species hybrids. Genetics, 154(4), of research, much work remains to be done be- 1747–1771. fore even single cases are understood well, and Barton, N. H., & Hewitt, G. M. (1985). Analysis of hybrid even then it is only through comparative studies zones. Ann. Rev. Ecol. Syst., 16, 113–148. that we will be able to disentangle the anecdo- Boughman, J. W. (2001). Divergent sexual selection en- Nature tal from the recurrent. The good news is that hances reproductive isolation in sticklebacks. , 411(6840), 944–948. the conceptual framework and tools to address Boughman, J. W. (2002). How sensory drive can promote gaps in our understanding exist. speciation. Trends Ecol. Evol., 17(12), 571–577. Butlin, R. K. (1993). Barriers to gene flow. Nature, 366, Acknowledgments 27. Butlin, R. K. (1995). Reinforcement: an idea evolving. D. Ortiz-Barrientos is funded by the Aus- Trends in Ecology & Evolution, 10(11), 432–434. tralian Research Council and The University Butlin, R. K. (2005). Recombination and speciation. Mol. Ecol., 14(9), 2621–2635. of Queensland. During the writing of this arti- Cain, M. L., Andreasen, V., et al. (1999). Reinforcing cle, P. Nosil was funded and hosted by the In- selection is effective under a relatively broad set of stitute of Advanced Study, Wissenschaftskolleg, conditions in a mosaic hybrid zone. Evolution, 53, Berlin, as part of a working group on specia- 1343. tion. We thank the members of that group, A. Chen, J., Ding, J., et al. (2008). A triallelic system of S5 is a major regulator of the reproductive barrier and Meyer, J. Mallet, and J. Feder, and members compatibility of indica japonica hybrids in rice. Proc. of the OB Lab, especially Peter Prentis, Luke Natl. Acad. Sci., 105(32), 11436–11441. Ambrose, and Antonia Posada, for continuing Coyne, J., & Orr, H. A. (2004). Speciation. Sunderland, discussions about speciation, and in particular Massachusetts: Sinauer Associates, Inc. speciation in the face of gene flow. We thank Coyne, J. A., & Orr, H. A. (1988). Patterns of speciation Carl Schlichting and anonymous reviewers for in Drosophila. Evolution, 43(2), 362–381. Dettman, J. R., Sirjusingh, C., et al. (2007). Incipi- their insights and editorial suggestions. ent speciation by divergent adaptation and antag- onistic epistasis in yeast. Nature, 447(7144), 585– Conflicts of Interest 588. Dobzhansky, T. (1937a). Genetic nature of species differ- The authors declare no conflicts of interest. ences. Am. Nat., 71(735), 404–420. Dobzhansky, T.(1937b). Genetics and the origin of species.New York: Columbia University Press. References Dobzhansky, T.(1940). Speciation as a stage in evolution- ary divergence. Am. Nat., 74(753), 312–321. Adam, D., Dimitrijevic, N., et al. (1993). Tumor suppres- Doi, M., Matsuda, M., et al. (2001). A locus for female sion in Xiphophorus by an accidentally acquired pro- discrimination behavior causing sexual isolation in moter. Science, 259, 359–376. Drosophila. Proc. Natl. Acad. Sci. USA, 98(12), 6714– Albert, A. Y. K. (2005). Mate choice, sexual imprinting, 6719. and speciation: A test of a one-allele isolating mecha- Endler, J. A., & Basolo, A. L. (1998). Sensory ecology, nism in sympatric sticklebacks. Evolution, 59(4), 927– receiver biases and sexual selection. Trends Ecol. Evol., 931. 13(10), 415–420. Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 179

Etges, W. J., & Noor, M. A. F. (Eds.) (2002). Genetics of Howard, D. J. (1993). Reinforcement: Origin, dynamics, mate choice: From sexual selection to sexual isolation edited and fate of an evolutionary hypothesis. In R. G. Har- volume. Dordrecht, The Netherlands: Kluwer Aca- rison (Ed.), Hybrid zones and the evolutionary process (pp. demic Publishers. 46–69). New York: Oxford University Press. Felsenstein, J. (1981). Skepticism towards Santa Rosalia, Jaenike, J., Dyer, K. A., et al. (2006). Asymmetrical rein- or why are there so few kinds of animals? Evolution, forcement and Wolbachia infection in Drosophila. Plos 35, 124. Biol., 4(10), 1852–1862. Fisher, R. A. (1958). The genetical theory of natural selection. Kalisz, S., Vogler, D. W., et al. (2004). Context-dependent New York: Dover. autonomous self-fertilization yields reproductive as- Fishman, L., & Wyatt, R. (1999). Pollinator-mediated surance and mixed mating. Nature, 430(7002), 884– competition, reproductive character displacement, 887. and the evolution of selfing in Arenaria uniflora Kay, K. M., & Schemske, D. W. (2008). Natural selection Caryophyllaceae. Evolution, 53, 1723. reinforces speciation in a radiation of neotropical Gavrilets, S. (2004). Fitness landscapes and the origin of species. rainforest plants. Evolution, 62(10), 2628–2642. Princeton, NJ: Princeton University Press. Kelly, J. K., & Noor, M. A. (1996). Speciation by rein- Goldberg, E. E., & Lande, R. (2006). Ecological and re- forcement: a model derived from studies of Drosophila. productive character displacement on an environ- Genetics, 143(3), 1485–1497. mental gradient. Evolution, 60(7), 1344–1357. Kirkpatrick, M. (2000). Reinforcement and divergence Good, J. M., Dean, M. D., et al. (2008a). A complex ge- under assortative mating. Proc. R. Soc. London Ser. B, netic basis to X-linked hybrid male sterility between 267, 1649. two species of house mice. Genetics, 179(4), 2213– Kirkpatrick, M., & Barton, N. (2006). Chromosome in- 2228. versions, local adaptation and speciation. Genetics, Good, J. M., Dean, M. D., et al. (2008b). Asymmetry and 173(1), 419–434. polymorphism of hybrid male sterility during the Kronforst, M. R., Young, L. G., et al. (2007). Reinforce- early stages of speciation in house mice. Evolution, ment of mate preference among hybridizing Helico- 62(1), 50–65. nius butterflies. J. Evol. Biol., 20(1), 278–285. Greenberg, A. J., Moran, J. R., et al. (2003). Ecological Lemmon, A. R., & Kirkpatrick, M. (2006). Reinforcement adaptation during incipient speciation revealed by and the genetics of hybrid incompatibilities. Genetics, precise gene replacement. Science, 302(5651), 1754– 173(2), 1145–1155. 1757. Lemmon, A. R., Smadja, C., et al. (2004). Reproductive Greig, D. (2007). A screen for recessive speciation genes character displacement is not the only possible out- expressed in the gametes of F1 hybrid yeast. PLoS come of reinforcement. J. Evol. Biol., 17(1), 177–183. Genet., 3(2), e21. Lemmon, E. M. (2009). Diversification of conspecific sig- Greig, D. (2008). Reproductive isolation in Saccharomyces. nals in sympatry: geographic overlap drives multi- Heredity. dimensional reproductive character displacement in Hatfield, T., & Schluter, D. (1999). Ecological speciation in frogs. Evolution, 63(5), 1155–1170. sticklebacks: Environment-dependent hybrid fitness. Levin, D. A. (1978). The origin of isolating mechanisms Evolution, 53(3), 866–873. in flowering plants. Evol. Biol., 11, 185–317. Hendry, A. P. (2004). Selection against migrants con- Lukhtanov, V. A., Kandul, N. P., et al. (2005). Reinforce- tributes to the rapid evolution of ecologically de- ment of pre-zygotic isolation and karyotype evolu- pendent reproductive isolation. Evol. Ecol. Res., 6(8), tion in Agrodiaetus butterflies. Nature, 436(7049), 385– 1219–1236. 389. Higgie, M., & Blows, M. W. (2007). Are traits that experi- Machado, C. A., Matzkin, L. M., et al. (2007). Mul- ence reinforcement also under sexual selection? Am. tilocus nuclear sequences reveal intra- and inter- Nat., 170(3), 409–420. specific relationships among chromosomally poly- Higgie, M., & Blows, M. W. (2008). The evolution of morphic species of cactophilic Drosophila. Mol. Ecol., reproductive character displacement conflicts with 16(14), 3009–3024. how sexual selection operates within a species. Evo- Masly, J. P., Jones, C. D., et al. (2006). Gene transposition lution, 62(5), 1192–203. as a cause of hybrid sterility in Drosophila. Science, Higgie, M., Chenoweth, S., et al. (2000). Natural selection 313(5792), 1448–1450. and the reinforcement of mate recognition. Science, Masly, J. P., & Presgraves, D. C. (2007). High-resolution 290(5491), 519–521. genome-wide dissection of the two rules of speciation Hoskin, C. J., Higgie, M., et al. (2005). Reinforcement in Drosophila. PLoS Biol., 5(9), e243. drives rapid allopatric speciation. Nature, 437(7063), Mayr, E. (1942). Systematics and the origin of species.New 1353–1356. York: Columbia University Press. 180 Annals of the New York Academy of Sciences

Mayr, E. (1963). Animal species and evolution. Cambridge, Noor, M. A. F. (1997). How often does sympatry affect MA: Belknap Press. sexual isolation in Drosophila? Am. Nat., 149, 1156. Michalak, P., & Malone, J. H. (2008). Testis-derived mi- Noor, M. A. F., Grams, K. L., et al. (2001a). The ge- croRNA profiles of African clawed frogs Xenopus netics of reproductive isolation and the potential for and their sterile hybrids. Genomics, 91(2), 158–164. gene exchange between Drosophila pseudoobscura and Moehring, A. J., Llopart, A., et al. (2006a). The genetic D. persimilis via backcross hybrid males. Evolution, 55, basis of postzygotic Reproductive isolation between 512. Drosophila santomea and D. yakuba due to hybrid male Noor, M. A. F., Grams, K. L., et al. (2001b). Chromo- sterility. Genetics, 173(1), 225–233. somal inversions and the reproductive isolation of Moehring, A. J., Llopart, A., et al. (2006b). The genetic species. Proc. Natl. Acad. Sci. USA, 98, 12084. basis of prezygotic Reproductive isolation between Nosil, P. (2007). Divergent host plant adaptation and Drosophila santomea and D. yakuba due to mating pref- reproductive isolation between ecotypes of Timema erence. Genetics, 173(1), 215–223. cristinae walking sticks. Am. Nat., 169(2), 151–162. Moehring, A. J., Teeter, K. C., et al. (2007). Genome-wide Nosil, P. (2008). Speciation with gene flow could be com- patterns of expression in Drosophila pure species and mon. Mol. Ecol. 17(9), 2103–2106. hybrid males. II. Examination of multiple-species Nosil, P., Crespi, B. J., et al. (2002). Host-plant adaptation hybridizations, platforms, and life cycle stages. Mol. drives the parallel evolution of reproductive isolation. Biol. Evol., 24(1), 137–145. Nature, 417(6887), 440–443. Moore, J. A. (1957). An embryologist’s view of the species Nosil, P., Crespi, B. J., et al. (2003). Reproductive isolation concept. In E. Mayr (Ed.), The species problem (pp. driven by the combined effects of ecological adap- 325–338). Washington, DC: American Association tation and reinforcement. Proc. Biol. Sci., 270(1527), for the Advancement of Science. 1911–1918. Moyle, L. C. (2005). Assessing the origin of species in the Nosil, P., Crespi, B. J., et al. (2007). Natural selection genomic era. Gen. Biol., 6(4), 4. and divergence in mate preference during speciation. Moyle, L. C. (2007). Comparative genetics of potential Genetica, 129(3), 309–327. prezygotic and postzygotic isolating barriers in a Ly- Nosil, P., Sandoval, C. P., et al. (2006). The evolution of copersicon species cross. J. Hered., 98(2), 123–135. host preference in allopatric vs. parapatric popula- Moyle, L. C., & Graham, E. B. (2005). Genetics of hybrid tions of Timema cristinae walking-sticks. J. Evol. Biol., incompatibility between Lycopersicon esculentum and L. 19(3), 929–942. hirsutum. Genetics, 169(1), 355–373. Nosil, P., Vines, T. H., et al. (2005). Perspective: Repro- Moyle, L. C., & Nakazato, T. (2008). Comparative genet- ductive isolation caused by natural selection against ics of hybrid incompatibility: sterility in two Solanum immigrants from divergent habitats. Evolution, 59(4), species crosses. Genetics, 179(3), 1437–1453. 705–719. Moyle, L. C., & Nakazato, T. (2009). Complex epista- Nosil, P., & Yukilevich, R. (2008). Mechanisms of re- sis for Dobzhansky-Muller hybrid incompatibility in inforcement in natural and simulated polymorphic Solanum. Genetics, 181(1), 347–351. populations. Biol. J. Linn. Soc., 95(2), 305–319. Moyle, L. C., Olson, M. S., et al. (2004). Patterns of Orr, H. A., & Irving, S. (2005). Segregation distortion in reproductive isolation in three angiosperm genera. hybrids between the Bogota and USA subspecies Evolution, 58(6), 1195–1208. of Drosophila pseudoobscura. Genetics, 169(2), 671– Nakazato, T., Jung, M.-K., et al. (2007). A genomewide 682. study of reproductive barriers between allopatric Orr, H. A., Masly, J. P., et al. (2007). Speciation in populations of a Homosporous fern, Ceratopteris Drosophila: from phenotypes to molecules. J. Hered., richardii. Genetics, 177(2), 1141–1150. 98(2), 103–110. Navarro, A., & Barton, N. H. (2003). Accumulating Ortiz-Barrientos, D., Counterman, B. A., et al. (2004). postzygotic isolation in parapatry: A new twist The genetics of speciation by reinforcement. Plos on chromosomal speciation. Evolution, 57(3), 447– Biol., 2(12), 2256–2263. 459. Ortiz-Barrientos, D., & Noor, M. A. (2005). Evidence Noor, M. A. (1995). Speciation driven by natural selection for a one-allele assortative mating locus. Science, in Drosophila. Nature, 375(6533), 674–675. 310(5753), 1467. Noor, M. A. (1999). Reinforcement and other conse- Ortiz-Barrientos, D., Reiland, J., et al. (2002). Recom- quences of sympatry. Heredity, 83(Pt 5), 503–508. bination and the divergence of hybridizing species. Noor, M. A., Garfield, D. A., et al. (2007). Divergence Genetica, 116(2–3), 167–178. between the Drosophila pseudoobscura and D. persimilis Palumbi, S. R. (2008). Speciation and the evolution of ga- genome sequences in relation to chromosomal inver- mete recognition genes: pattern and process. Heredity, sions. Genetics, 177(3), 1417–1428. 102(1), 66–76. Ortiz-Barrientos et al.: Evolving Prezygotic Isolation in Sympatry 181

Peterson, M. A., Honchak, B. M., et al. (2005). Rela- tion of reproductive isolation in flycatchers. Science, tive abundance and the species-specific reinforce- 318(5847), 95–97. ment of male mating preference in the Chrysochus Saetre, G. P., Kral, M., et al. (1999). Dynamics of a clinal (Coleoptera : Chrysomelidae) hybrid zone. Evolution, hybrid zone and a comparison with island hybrid 59(12), 2639–2655. zones of flycatchers Ficedula hypoleuca and F. albicollis. Pfennig, D. W., & Rice, A. M. (2007). An experimen- J. Zool., 247, 53–64. tal test of character displacement’s role in promot- Saetre, G. P., Moum, T., et al. (1997). A sexually selected ing postmating isolation between conspecific popu- character displacement in flycatchers reinforces pre- lations in contrasting competitive environments. Evo- mating isolation. Nature, 387(6633), 589–592. lution, 61(10), 2433–2443. Sanderson, N. (1989). Can gene flow prevent reinforce- Pfennig, K. S. (2003). A test of alternative hypotheses ment? Evolution, 43, 1223. for the evolution of reproductive isolation between Sandoval, C. P. (1994). The effects of the relative geo- spadefoot toads: Support for the reinforcement hy- graphic scales of geneflow and selection on morph pothesis. Evolution, 57(12), 2842–2851. frequencies in the walking-stick Timema cristinae. Pfennig, K. S. (2007). Facultative mate choice drives adap- Evolution, 48(6), 1886–1879. tive hybridization. Science, 318(5852), 965–967. Sawamura, K., Tomimura, Y., et al. (2006). Estab- Pfennig, K. S. (2008). Population differences in condition- lishing interspecific mosaic genome lines between dependent sexual selection may promote divergence Drosophila ananassae and Drosophila pallidosa by in non-sexual traits. Evolutionary Ecol. Res., 10(5), 763– means of parthenogenesis. Genet. Res., 88(1), 1–11. 773. Sawamura, K., Zhi, H., et al. (2008). Genetic analysis Pfennig, K. S., & Pfennig, D. W. (2005). Character dis- of female mating recognition between Drosophila placement as the best of a bad situation: fitness trade- ananassae and Drosophila pallidosa: application of offs resulting from selection to minimize resource and interspecific mosaic genome lines. Genetica, 133(2), mate competition. Evolution, 59(10), 2200–2208. 179–185. Pfennig, K. S., & Ryan, M. J. (2006). Reproductive char- Schluter, D. (1988). Estimating the form of natural selec- acter displacement generates reproductive isolation tion on a quantitative trait. Evolution, 42, 849–861. among conspecific populations: an artificial neural Schluter, D. (1994). Experimental evidence that competi- network study. Proc. Roy. Soc. Biol., 273(1592), 1361– tion promotes divergence in adaptive radiation. Sci- 1368. ence, 266(5186), 798–801. Pfennig, K. S., & Simovich, M. A. (2002). Differential Schluter, D. (2000). The ecology of adaptive radiation. Oxford: selection to avoid hybridization in two toad species. Oxford University Press. Evolution, 56, 1840. Schluter, D. (2001). Ecology and the origin of species. Presgraves, D., Balagopalan, L., et al. (2003). Adaptive Trends Ecol. Evol., 16, 372. evolution drives divergence of a hybrid inviabil- Schluter, D. (2003). Frequency dependent natural selec- ity gene between two species of Drosophila. Nature, tion during character displacement in sticklebacks. 423(6941), 715–719. Evolution, 57(5), 1142–1150. Presgraves, D. C., & Stephan, W. (2007). Pervasive adap- Schluter, D., & McPhail, J. D. (1992). Ecological Char- tive evolution among interactors of the Drosophila hy- acter Displacement and Speciation in Sticklebacks. brid inviability gene, Nup96. Mol. Biol. Evol., 24(1), The Am. Nat., 140(1), 85. 306–314. Schluter, D. & McPhail, J. D. (1993). Character displace- Reyer, H. U. (2008). Mating with the wrong species can ment and replicate adaptive radiation. Trends in Ecol- be right. Trends Ecol. Evol., 23(6), 289–292. ogy & Evolution, 8, 197–200. Rieseberg, L. H. (2001). Chromosomal rearrangements Schug, M. D., Smith, S. G., et al. (2007). The genetic and speciation. Trends. Ecol. Evol., 16(7), 351–358. structure of Drosophila ananassae populations from Rundle, H. D., & Schluter, D. (1998). Reinforcement of Asia, Australia and Samoa. Genetics, 175(3), 1429– stickleback mate preferences: Sympatry breeds con- 1440. tempt. Evolution, 52(1), 200–208. Seddon, N. (2005). Ecological adaptation and species Rundle, H. D., & Whitlock, M. C. (2001). A genetic inter- recognition drives vocal evolution in neotropical su- pretation of ecologically dependent isolation. Evolu- boscine birds. Evolution, 59(1), 200–215. tion, 55(1), 198–201. Servedio, M. R. (2004). The what and why of re- Ryan, M. J., & Rand, A. S. (1993). Species recognition search on reinforcement. PLoS Biol., 2(12), 2032– and sexual selection as a unitary problem in animal 2035. communication. Evolution, 47(2), 647–657. Servedio, M. R. (2009). The role of linkage disequilib- Saether, S. A., Saetre, G. P., et al. (2007). Sex rium in the evolution of premating isolation. Heredity, chromosome-linked species recognition and evolu- 102(1), 51–56. 182 Annals of the New York Academy of Sciences

Servedio, M. R., & Kirkpatrick, M. (1997). The effects of of Wolbachia versus genetic incompatibilities on re- gene flow on reinforcement. Evolution, 51(6), 1764– inforcement and speciation. Evolution, 59(8), 1607– 1772. 1619. Servedio, M. R., & Noor, M. A. F. (2003). The role of Ting, C. T., Tsaur, S. C., et al. (1998). A rapidly evolving reinforcement in speciation: Theory and Data. Ann. homeobox at the site of a hybrid sterility gene. Science, Rev. Ecol. Evol. Syst., 34(1), 339–364. 282(5393), 1501–1504. Silvertown, J., Servaes, C., et al. (2005). Reinforcement of Turelli, M., & Moyle, L. C. (2007). Asymmetric postmat- reproductive isolation between adjacent populations ing isolation: Darwin’s corollary to Haldane’s rule. in the Park Grass Experiment. Heredity, 95(3), 198– Genetics, 176(2), 1059–1088. 205. Urbanelli, S., & Porretta, D. (2008). Evidence of reinforce- Smadja, C., & Butlin, R. (2006). Speciation – A new role ment of premating isolation between two species of for reinforcement. Heredity, 96(6), 422–423. the genus Ochthebius Coleoptera: Hydraenidae. Evo- Smith, R. A., & Rausher, M. D. (2008). Experimental lution, 62(6), 1520–1527. evidence that selection favors character displacement Vallejo-Marin, M., & Uyenoyama, M. K. (2004). On the in the ivyleaf morning glory. Am. Nat., 171(1), 1–9. evolutionary costs of self-incompatibility: incomplete Sun, S., Ting, C. T., et al. (2004). The normal function reproductive compensation due to pollen limitation. of a speciation gene, Odysseus, and its hybrid sterility Evolution, 58(9), 1924–1935. effect. Science, 305(5680), 81–83. Vamosi, S. M., & Schluter, D. (2004). Character shifts in Swanson, W. J., & Vacquier, V. D. (2002). The rapid evo- the defensive armor of sympatric sticklebacks. Evolu- lution of reproductive proteins. Nat. Rev. Genet., 3, tion, 58(2), 376–385. 137. Via, S., & West, J. (2007). The genetic mosaic: A key phase Sweigart, A. L., Fishman, L., et al. (2006). A simple ge- in ecological speciation. Mol. Ecol., 57, 138–145. netic incompatibility causes hybrid male sterility in Waage, J. K. (1975). Reproductive isolation and the mimulus. Genetics, 172(4), 2465–2479. potential for character displacement in the dam- Takahashi, A., Tsaur, S. C., et al. (2001). The nu- selflies, Calopteryx maculata and C. aequabilis Odonata: cleotide changes governing cuticular hydrocar- Calopterygidae. Syst. Zool., 24(1), 24–36. bon variation and their evolution in Drosophila Walker, T.J. (1974). Character displacement and acoustic melanogaster. Proc. Natl. Acad. Sci. USA, 98(7), 3920– insects. Amer. Zool., 14, 1137. 3925. Yatabe, Y., Kane, N. C., et al. (2007). Rampant gene ex- Tao, Y., Araripe, L., et al. (2007). A sex-ratio meiotic change across a strong reproductive barrier between drive system in Drosophila simulans. II: An X-linked the annual sunflowers, Helianthus annuus and H. peti- Distorter. PLoS Biol., 5(11), e293. olaris. Genetics, 175(4), 1883–1893. Tao, Y., Hartl, D. L., et al. (2001). Sex-ratio segregation Yukilevich, R., & True, J. R. (2006). Divergent outcomes of distortion associated with reproductive isolation in reinforcement speciation: The relative importance of Drosophila. Proc. Natl. Acad. Sci. USA, 98(23), 13183– assortative mating and migration modification. Am. 13188. Nat., 167(5), 638–654. Tao, Y., Masly, J. P., et al. (2007). A sex-ratio meiotic Zouros, E., & D’Entremont, C. J. (1980). Sexual isolation drive system in Drosophila simulans. I: An autosomal among populations of Drosophila mojavensis: Response suppressor. PLoS Biol., 5(11), e292. to pressure from a related species. Evolution, 34(3), Telschow, A., Hammerstein, P., et al. (2005). The effect 421–430.