Diversification and Coevolution in Brood Pollination

AJB Advance Article published on October 7, 2016, as 10.3732/ajb.1600056. The latest version is at http://www.amjbot.org/cgi/doi/10.3732/ajb.1600056 RESEARCH ARTICLE

AMERICAN JOURNAL OF BOTANY

INVITED PAPER For the Special Section: The Ecology, Genetics, and Coevolution of Intimate Mutualisms

Diversifi cation and coevolution in brood pollination mutualisms: Windows into the role of biotic interactions in generating biological diversity1

David H. Hembry 2,4 and David M. Althoff 3

Brood pollination mutualisms—interactions in which specialized insects are both the pollinators (as adults) and seed predators (as larvae) of their host plants—have been infl uential study systems for coevolutionary biology. These mutualisms include those between fi gs and fi g wasps, yuccas and yucca moths, leaffl owers and leaffl ower moths, globefl owers and globefl ower fl ies, Silene plants and Hadena and Perizoma moths, saxifrages and Greya moths, and senita cacti and senita moths. The high reciprocal diversity and species-specifi city of some of these mutualisms have been cited as evidence that coevolution between plants and pollinators drives their mutual diversifi cation. However, the mechanisms by which these mutualisms diversify have received less attention. In this paper, we review key hypotheses about how these mutualisms diversify and what role coevolution between plants and pollinators may play in this process. We fi nd that most species-rich brood pollination mutualisms show signifi cant phylogenetic congruence at high taxonomic scales, but there is limited evidence for the processes of both cospeciation and duplication, and there are no unambiguous examples known of strict-sense contemporaneous cospeciation. Allopatric speciation appears important across multiple systems, particularly in the insects. Host-shifts appear to be common, and widespread host-shifts by pollinators may displace other pollinator lineages. There is relatively little evidence for a “coevolution through cospeciation” model or that coevolution promotes speciation in these systems. Although we have made great progress in understanding the mechanisms by which brood pollination mutualisms diversify, many opportunities remain to use these intriguing symbioses to understand the role of biotic interactions in generating biological diversity.

KEY WORDS brood pollination; coevolutionary diversifi cation; Ficus ; obligate pollination mutualism; Phyllanthus ; pollinating seed parasite; seed predator; speciation; Yucca

Brood pollination mutualisms—specialized interactions in which those between leaffl ower plants and leaffl ower moths (Kato et al., plants are pollinated by insects which, as larvae, consume seeds of the 2003; Kawakita, 2010), Silene and two genera of moths, Hadena and same plants—have evolved only a handful of times in the entire co- Perizoma ( Kephart et al., 2006 ), globefl owers and globefl ower fl ies evolutionary history of insects and angiosperms. Despite having ( Pellmyr, 1992 ; Després et al., 2002b ), saxifrages and Greya moths arisen infrequently, these mutualisms can be extraordinarily species- (Pellmyr and Th ompson, 1992 ), and senita cacti and senita moths rich and ecologically dominant in some ecosystems and have long (Fleming and Holland, 1998). Th ese interactions have been enor- attracted attention from natural historians, evolutionary biologists, mously important study systems for the evolution of cooperation and and ecologists. Th e most famous of these mutualisms are those be- for coevolutionary biology. In this review, we consider the evidence tween fi gs and fi g wasps (Mayr, 1885; Cook and Rasplus, 2003) and for the mechanisms of diversifi cation of both plants and insects en- yuccas and yucca moths ( Riley, 1892 ; Pellmyr, 2003 ); others include gaged in brood pollination mutualisms, and the role of biotic interactions (and coevolution) in that diversifi cation. 1 Manuscript received 6 February 2016; revision accepted 10 May 2016. In the same way that one might study anoles, cichlids, or Galápa- 2 Department of Ecology and Evolutionary Biology, University of Arizona, P. O. Box 210088, gos fi nches to understand adaptive radiation, brood pollination Tucson, Arizona 85721 USA; and mutualisms (also sometimes referred to as nursery pollination mu- 3 Department of Biology, Syracuse University, 107 College Place, Syracuse, New York, 13244 tualisms, pollinating seed-predation mutualisms, pollinating seed- USA 4 Author for correspondence (e-mail: [email protected]), phone: +1-520-626- parasite mutualisms, and in the case of fi gs, yuccas, and leaffl owers, 9315 obligate pollination mutualisms) have been major study systems for doi:10.3732/ajb.1600056 understanding coevolutionary diversifi cation. We choose the term

brood pollination over other, widely used synonyms in this synthesis In this review, we ask how brood pollination interactions diver- because of its brevity (relative to pollinating seed-predation mutu- sify, and then use this evidence to ask what the role of biotic inter- alism), because fi g wasp larvae technically are infl orescence gallers actions between plants and pollinators (including coevolution) is in rather than seed predators, and also because we wish to compare pro- this diversifi cation process. Th roughout, we will suggest key hy- cesses in otherwise analogous interactions that diff er in whether they potheses that need to be tested to strengthen the link between co- are strictly obligate or include copollinators (below). evolution and diversifi cation. Approaches that focus on coevolution Here we defi ne coevolutionary diversifi cation broadly as diversi- per se have been criticized both for their unworkability (coevolu- fi cation (among populations, species, or clades) in which coevolution is diffi cult to demonstrate) and on empirical and theoretical tion (reciprocal natural selection sensu Janzen, 1980; Th ompson, grounds (there is little, or mixed, evidence for coevolution promot- 2005 ) drives the diversifi cation (sensu Hembry et al., 2014 ; ing diversifi cation, particularly in mutualisms) ( Yoder and Nuismer, note that this is a broader defi nition than that of Althoff et al., 2010 ; Hembry et al., 2014 ). We note therefore that our approach 2014 ). Importantly, coevolutionary diversifi cation may not neces- off ers a more nuanced perspective that is informative whether the sarily create phylogenetic congruence between coevolving clades goal is to emphasize the role of biotic interactions more generally or (Th ompson, 2005 ; Poisot, 2015 ). Th is emphasis on brood pollina- coevolution specifi cally in diversifi cation. Specifi cally, we ask the tion mutualisms is due to four characteristics many of these sys- following questions: how specialized (at the species level) are plants tems share: (1) the highly intertwined life cycles of both plants and and brood pollinators to each other, do they diversify through copollinators, (2) the fact that the pollinators control most (if not all) speciation or some other process that generates phylogenetic con- pollen fl ow in their plant hosts ( Th ompson, 1994 ), (3) the great de- gruence, do they speciate in allopatry, do pollinator clades displace gree of species-level specialization seen between plants and pollina- each other competitively, and is coevolution implicated in either tors in nature (Wiebes, 1979; Weiblen, 2002; Kawakita and Kato, plant or insect speciation? 2006 ; Althoff et al., 2012), and (4) the high species richness in some plant and pollinator clades engaged in these interactions ( Janzen, 1979 ; Kawakita, 2010 ). Although the hypothesis that coevolution WHAT ARE THE BROOD POLLINATION MUTUALISMS? promotes diversifi cation has been invoked for a wide number of organisms and types of interactions ( Ehrlich and Raven, 1964 ; Dietl We defi ne brood pollination mutualisms as mutualisms in which and Kelley, 2002 ; Th ompson, 2005 , 2009 ; Hembry et al., 2014 ), the larvae of pollinators consume seeds or other female infl orescence above attributes of brood pollination mutualisms make them ideal tissue. Brood pollination mutualisms involve diverse taxa from test cases for this hypothesis. many diff erent geographic areas ( Table 1 ). Th ese interactions are Despite the fact that they have been major and infl uential study usually obligate, but in some systems they can be more facultative systems for understanding the role of coevolution (or biotic inter- at least for the plants. For instance, fi gs, yuccas, and leaffl owers are actions more generally) in diversifi cation, we know very little about generally described in the literature as obligately pollinated by their the mechanisms by which the plants and insects engaged in brood respective brood pollinators (obligate pollination mutualisms; e.g., pollination mutualisms diversify. Th is defi ciency is largely due to Pellmyr, 2003 ; Kawakita, 2010 ). In contrast, it is well established the patterns of diversity across the diff erent brood pollination mu- that the senita cactus is not exclusively pollinated by senita moths tualisms. Some of these interactions (e.g., fi gs, leaffl owers) have (Holland and Fleming, 1999) and Silene– Hadena , Silene – Perizoma, hundreds of species, primarily in the tropics (Berg, 1989; Kawakita, and Lithophragma– Greya interactions both exist with copollina- 2010 ), so that taxonomic sampling and indeed, taxonomic defi cits tors and can span the continuum from mutualism to antagonism, in the pollinator insects, become major impediments. Research on even among populations within a pair of interacting species these systems has primarily focused on cophylogenetic patterns at (Th ompson and Pellmyr, 1992 ; Th ompson and Cunningham, 2002 ; macroevolutionary scales and examining patterns of species-speci- Kephart et al., 2006 ). Copollinators also exist in some globefl ower fi city in ecological time (Weiblen and Bush, 2002; Molbo et al., populations ( Suchan et al., 2015 ). Whether strictly obligate or more 2003 ; Machado et al., 2005 ; Kawakita and Kato, 2006 , 2009 ; Cruaud facultative, we will consider all types of brood pollination for two et al., 2012c ; Hembry et al., 2013). Other interactions (e.g., senita, reasons. First, pollination interactions do not need to be exclusively globefl owers, and saxifrages) are very species-poor, and are per- reciprocally obligate for pollinators and plants to exert signifi cant haps thus not particularly suited to studying diversifi cation; how- selection on each other that may fuel speciation (Th ompson and ever, these systems have served as important models in evolutionary Cunningham, 2002 ; Th ompson et al., 2013b ). Second, brood polli- ecology. Some interactions appear to be obligate ( Cook and Rasplus, nation interactions that are not strictly obligate can reveal impor- 2003; Pellmyr, 2003), whereas others vary among species and pop- tant insights about how diversifi cation occurs in such mutualisms ulations along the continuum between mutualism and antagonism (e.g., Th ompson and Rich, 2011 ). (Th ompson and Cunningham, 2002; Kephart et al., 2006). Finally, many of these interactions have only been discovered in the past several decades (Th ompson and Pellmyr, 1992; Fleming and Hol- DIVERSIFICATION IN BROOD POLLINATION MUTUALISMS land, 1998 ; Kato et al., 2003). As a result, despite their conceptual infl uence on the fi eld of coevolution, it has been diffi cult to use Although the patterns of diversifi cation for brood pollination mu- these systems to test the hypothesis that coevolution (or biotic in- tualisms have been elucidated, there has been relatively little re- teractions) promotes diversifi cation. Even so, recent work in many search on the mechanistic basis of diversifi cation. As highlighted systems has begun to reveal common patterns and reinforces the already, part of this lack is due to diffi culties in manipulating the potential for brood pollination mutualisms to serve as important systems and part is due to the relatively recent emergence of some models for the study of biotic interactions and coevolution in systems. Perhaps the best-studied system with regards to patterns diversifi cation. of diversifi cation is the yuccas and yucca moths, that have received OCTOBER 2016 , VOLUME 103 • HEMBRY AND ALTHOFF—DIVERSIFICATION IN BROOD POLLINATION MUTUALISMS • 3

particularly detailed attention at multiple phylogenetic scales. Other groups that have received some research attention include globefl owers and globefl ower fl ies in the Palearctic ( Després et al., 1 ≥ 6

≥ 30

~18 2002a , b , 2003 ; Espíndola et al., 2012a, b , 2014 ; Suchan et al., 2015), ~750 ~500 fi gs and fi g wasps in some parts of their range (particularly islands and the edges of the tropics: Molbo et al., 2003 ; Jackson et al., 2008 ; Number of plant species implicated Renoult et al., 2009 ; Cornille et al., 2011 ; Chen et al., 2012 ; Cruaud et al., 2012b ; Kusumi et al., 2012; Moe and Weiblen, 2010, 2012 ; Darwell et al., 2014; Tian et al., 2015), Lithophragma – Greya interactions in western North America ( Brown et al., 1994 ; Rich et al., 2008; Thompson and Rich, 2011; Thompson et al., 2013b), and leaffl owers in the clade Glochidion in southeastern Polynesia (here- after, Polynesian leafflowers; Hembry et al., 2013 , unpublished manuscript). Taken together, we see a number of common factors warm temperate zones warm temperate

Holarctic that may contribute to diversifi cation in these systems. Below, we propose and provide answers to what we think are the key questions in understanding the role of biotic interactions and coevolution in generating diversifi cation in brood pollination mutualisms. Bear in mind that not all brood pollination mutualisms will diver-

ower moths ower tropics, World Old sify exactly the same way, but we suggest that there are several key Common nameCommon Distribution Senita moth (Sonoran Desert) Nearctic Greya mothsGreya Nearctic western Fig waspsFig subtropics Pantropical, Leaffl Yucca mothsYucca Mexicotropical Nearctic, 35–40 Globefl ower fl ies ower Globefl Palearctic factors that may vary in importance depending on the system. A glossary of key terms is provided in Table 2 .

(1) How specialized are the plants and pollinators to each other?— Knowing how specialized brood pollinated plants and their pollinators are to each other is essential to understanding how they diversify. Accordingly, specialization in brood pollination systems has been a contentious topic in the literature. Early research on fi g

Gracillariidae wasps assumed that each species of fi g has (at least locally) a unique Order, family Order, species of pollinating fi g wasp (e.g., Janzen, 1979 ; Ramírez, 1970 ); this paradigm was later challenged following the advent of molecular phylogenetic methods in delimiting fi g wasp taxa (e.g., Molbo Hymenoptera Lepidoptera: Pyralidae Lepidoptera: Lepidoptera: Prodoxidae Lepidoptera: Anthomyiidae et al., 2003 ; Machado et al., 2005 ; Cornille et al., 2011 ). Recent work examining fi g wasp diversity has demonstrated that genera of fi g wasps and subgenera or sections of Ficus are oft en specialized to each other (e.g., Jackson et al., 2008 ; Cruaud et al., 2012c ; but see

Renoult et al., 2009). Within these reciprocally associated subclades, however, it is not unusual for a single Ficus species to be pollinated by multiple wasp species either sympatrically or allopatrically, although there are still plenty of examples of locally one-to-one inter-

Agaonidae Lepidoptera: Lepidoptera: Epicephala Upiga virescens , Tegeticula Parategeticula Hadena , Prodoxidae Lepidoptera: Perizoma Chiastocheta Noctuidae, Geometridae Lepidoptera: Greya actions (see reviews by Herre et al., 2008; Cook and Segar, 2010; Yang et al., 2015). Th ere are also some good examples of a single fi g wasp species visiting multiple Ficus host species sympatrically (e.g., pollinator sharing; Cornille et al., 2011; Wang et al., 2016 ). owers

owers Yucca has been extensively sampled in the United States, and saxifrages this work has revealed that over much of their range, a single species of Tegeticula is monophagous (although there are a few exceptions; Pellmyr, 2003; Althoff et al., 2012 ). Patterns of specialization in leaffl owers and leaffl ower moths are less clear and may vary among major clades of Phyllanthus s.l. and Epicephala . Th ere are

Family name Common Insect taxon examples of locally monophagous leaffl ower moths ( Kawakita and Moraceae Figs Phyllanthaceae Leaffl Kato, 2006 ), and leaffl owers, which locally only have unique pollinators (e.g., Kawakita and Kato, 2006 ; Kawakita et al., 2015 ; Li and Yang, 2015 ). Th ere are, however, also more complex patterns of sympatric pollinator and host sharing ( Zhang et al., 2012 ; Li et al., 2015 ; Hembry et al., unpublished manuscript). For the less species-rich brood pollination mutualisms, patterns pollination mutualisms. distributions of brood and geographic species diversity, Taxonomy, of specialization are more varied. Within co-occurring assemblages, , Agavaceae Hesperoyucca Yuccas particularly in California, multiple species each of Greya moths and Glochidion, ) Breynia Plant taxon TABLE 1. TABLE Ficus Phyllanthus (includes Yucca Silene Caryophyllaceae Ranunculaceae Trollius Globefl Saxifragaceae Heuchera Lithophragma, stars; schottii Cactaceae Pachycereus Woodland cactus Senita their saxifrage hosts do not appear to have restricted host breadth 4 • AMERICAN JOURNAL OF BOTANY

TABLE 2. Glossary of terms used. Polynesia ( Hembry et al., 2013 , unpublished manuscript). Never- Term Defi nition theless, this interesting hypothesis should be more widely tested in this and other brood pollination mutualisms. Proximately, both Coevolution Reciprocal natural selection between two or more taxa sensu Janzen (1980), Thompson pre- and postoviposition barriers may mediate this specifi city. (1994 , 2005 ): does not require or imply Th ere is evidence not only that host-recognition cues such as fl oral cospeciation or phylogenetic congruence odor may mediate specifi city in leaffl owers( Okamoto et al., 2007 , Cospeciation Process by which host and pollinator speciate 2013 ), but also that larvae may suff er reduced survivorship on the contemporaneously. Diff ers from cases in “wrong” hosts in yuccas and fi gs ( Smith et al., 2009 ; Moe et al., which host and pollinator speciation are 2011 ; Cornille et al., 2011 ; see Althoff , 2016 , in this issue) suggest- decoupled, or phylogenetic tracking. Phylogenetic tracking Process by which pollinator (or host) diversifi es ing that the antagonistic seed-eating part of the interaction may in onto host clade in a manner which matches part be responsible for this specifi city. In the case of fi gs, host spe- the phylogenetic relationships of the host cialization may be proximately due to the ability to successfully ini- taxa, without contemporaneous tiate galls on fi g fl owers ( Ghana et al., 2015 ). More research is cospeciation occurring needed to ascertain the relative roles of the mutualistic and antago- Phylogenetic congruence Pattern in which host and pollinator phylogenies nistic sides of the overall interaction in determining the high levels are topologically more similar than would be expected by chance. This congruence may of specialization seen in brood pollination mutualisms. be observed between species, subclades, or other operational taxonomic units and may (2) Do brood pollination mutualisms show evidence for phyloge- be generated by cospeciation, phylogenetic netic congruence and for cospeciation? — Cophylogenetic analyses tracking, or some other process. have been a major area of focus for the study of the evolution of Host-shift Process in which a pollinator lineage colonizes brood pollination mutualisms, in part because cophylogenetic a new host lineage. patterns can (in principle) be detected with partial extant taxon Duplication Process in which a pollinator (or host) speciates without the other associate speciating. This sampling. At global scales, the most diverse brood pollination term is from the cophylogenetics literature, mutualisms, fi gs (Ficus ) and fi g wasps (Agaonidae) ( Cruaud et al., but may also describe cases in which host 2012c ), Yucca and Tegeticula ( Althoff et al., 2012), and the leaffl ower and pollinator speciation are decoupled. clade Glochidion and its reciprocally associated clade of Epicephala “One-to-one” paradigm Traditional and highly infl uential view of brood ( Hembry et al., 2013 ) show evidence for signifi cant cophylogenetic pollination interactions in which each host structure. For fi gs and gfi wasps, dioecious fi gs show greater evi- species and each pollinator species is (locally or globally) associated with a single unique partner dence for phylogenetic congruence globally with their fi g wasps species. Articulated clearly in early fi g literature than do monoecious fi gs; in particular, copollinators are much such as Ramírez (1970) and Janzen (1979) more likely to be sister species on dioecious than monoecious fi gs (Yang et al., 2015). For leaffl owers and leaffl ower moths in general, the pattern is less clear. Epicephala -pollinated leaffl owers ( Glochid- (Th ompson et al., 2013b; Th ompson, 2013). Th e single species of ion, Breynia , and Phyllanthus ) are globally a polyphyletic assem- European globefl ower (Trollius europaeus) is pollinated by multiple blage that does not show any phylogenetic congruence with species of monophagous globefl ower fl ies in Europe, whereas Asian Epicephala, which is monophyletic (Kawakita and Kato, 2009). globefl ower fl ies use multiple species of globefl owers (Després et al., Within the clade Gomphidium (a subgenus of Phyllanthus ) in New 2002b ; Espíndola et al., 2012a). Many Silene have only one known Caledonia, however, there appear to be two subclades, each of pollinating Hadena , and many reported Hadena are only known for which is reciprocally associated with its own clade of Epicephala 1–2 Silene species; however, generalists using up to nine partner (interestingly, these Gomphidium -associated Epicephala do not species are known in both genera (Kephart et al., 2006). Despite form a monophyletic group; Kawakita and Kato, 2009 ). No studies limited evidence for a traditional “one-to-one” paradigm, these re- have tested whether there is signifi cant cophylogenetic structure cent studies have shown that the most species-rich brood pollina- between any of the other Epicephala -pollinated subclades of Phyl- tion mutualisms are still highly specialized at the species level, and lanthus s.l. (Breynia , Phyllanthus subgenus Gomphidium , Madagas- this specialization is important if we want to ask what role plant– car Phyllanthus, and the P. reticulatus complex) and their Epicephala. pollinator interactions play in their mutual diversifi cation. In addi- None of the above insect–plant systems show exact phylogenetic tion, a common theme seems to be that the insect pollinators are congruence, and at smaller, within-region scales, plants and their usually the more specialized of the interacting partners. pollinators can actually show no signifi cant phylogenetic congru- Why are these interactions so specialized? All else being equal, ence (such as Polynesian leaffl owers and their Epicephala [Hembry intimate mutualisms, such as these, should form ecological net- et al., 2013], or within a co-occurring assemblage of section Ameri- works which are highly reciprocally specialized and modular in cana fi gs and their Pegoscapus wasps [ Jackson et al., 2008 ]). structure ( Guimarães et al., 2007 ; Th ompson et al., 2013a ). All of At even smaller phylogenetic scales such as between pairs of in- these interactions also evolved from parasitic ones and might retain teracting sister-species, there are relatively few examples of phylo- a narrow host breadth from their parasitic ancestors ( Th ompson, genetic congruence. Th e best-studied example is that between 1994 ). Kawakita et al. (2010) proposed that leaffl ower moths have a Joshua trees ( Yucca brevifolia jaegerianaand Y . ) and the sister- more restricted host breadth than nonpollinating herbivorous rela- species pollinators Tegeticula synthetica and T. antithetica in the tives because some aspect of mutualism may promote greater host- Mojave Desert of the United States. In this case, a cophylogenetic specifi city. Whereas this result is intriguing, the fi ndings, based on approach suggests cospeciation, but divergence in Joshua tree Glochidion-associated Epicephala in Japan, appear to diff er from and its yucca moths may not have been contemporaneous (Smith those subsequently reported in China ( Li et al., 2015 ) and French et al., 2008a ). Furthermore, the two races of Joshua tree are not OCTOBER 2016 , VOLUME 103 • HEMBRY AND ALTHOFF—DIVERSIFICATION IN BROOD POLLINATION MUTUALISMS • 5

completely reproductively isolated, which suggests the pollinators some of these systems, speciation has been examined in relatively are not completely host-specifi c ( Yoder et al., 2013 ). few brood pollination mutualisms. However, those groups that have Below the species level, there is an extensive and growing literature been examined closely show evidence for allopatric speciation in on phylogeography and population genetics in brood pollination mu- both plants and pollinators. Comparison of contemporary ranges of tualisms. Many of these studies have examined one taxonomic group sister taxa in a phylogenetic context suggests that allopatric specia- with molecular data and the other group using existing taxonomy tion has occurred in Yucca ( Althoff et al., 2012), Tegeticula ( Althoff (but see Smith et al., 2011 ; Espíndola et al., 2014 ). For instance, widely et al., 2012), the older of two Polynesian Epicephala clades (Hembry distributed fi g species can show substantial spatial turnover and/or et al., 2013), and two species of Greya in California (Rich et al., phylogenetic diff erentiation in their pollinating fi g wasps ( Kobmoo 2008 ; Th ompson and Rich, 2011). Th e well-examined species pairs et al., 2010; Moe and Weiblen, 2010; Chen et al., 2012, Darwell et al., of Y. brevifolia– T . synthetica and Y. jaegeriana – T . antithetica (dis- 2014; Tian et al., 2015), but in some cases distances of >1000 km are cussed in more detail later) are allopatrically distributed (except for required for this turnover or differentiation to become apparent one co-occurring population in Nevada, USA), and this allopatry is ( Kobmoo et al., 2010 ; Tian et al., 2015 ). Trollius europaeus globefl ow- thought to be in part related to a Pliocene marine incursion in the ers and three globefl ower fl y species show phylogeographic structure Mojave Desert (Smith et al., 2008a). Cryptic allopatric lineages also across Europe, but among species the patterns are incongruent and have been demonstrated in Chinese Wiebsia fi g wasps ( Chen et al., idiosyncratic ( Espíndola et al., 2014 ). One Polynesian Epicephala 2012 ) and Southeast Asian and New Guinea Ceratosolen fi g wasps taxon (clade Z of Hembry et al., 2013 ) shows no phylogenetic diff er- (Moe and Weiblen, 2010). Dispersal between Europe and Asia, and entiation across nine diff erent (and in many cases, morphologically concomitant host-shift s, are implicated in the diversifi cation of distinct) species of leaffl ower tree (Glochidion ) in the Society Islands globefl ower fl ies, but there is also much within-European diversifi - (D. Hembry et al., unpublished manuscript). Tegeticula yuccasella ap- cation despite only a single available globefl ower host ( Espíndola pears to be a generalist yucca moth species that pollinates at least et al., 2012a ). Finally, within-species phylogeographic structure based seven diff erent species of Yucca across its range in the central and on geography has been demonstrated in Tegeticula ( Segraves and eastern United States (Pellmyr, 1999). A similar pattern exists with Pellmyr, 2001), Trollius europaeus ( Espíndola et al., 2012b ), and the widely distributed Phyllanthus reticulatus and P. mi c ro car pu s three European globefl ower fl y species (Espíndola et al., 2014). complex in East Asia and their diverse fauna of Epicephala (Luo et al., Most of these examples are from the insects; the hypothesis of allo- 2011 ; Kawakita et al., 2015 ; Li and Yang, 2015 ). patric speciation has not been tested for brood-pollinated plant In many of these cases, it may be that there is cryptic diff erentia- clades other than Yucca . Van Welzen et al. (2015) suggested that tion in the generalist species that is associated with the use of dispersal out of mainland Southeast Asia and perhaps also edaphic diff erent partners; furthermore, phylogeographic patterns are not adaptation to limestone may have promoted diversifi cation both in necessarily evidence for how speciation occurs. In one test of cryp- Breynia and their non- Epicephala -pollinated sister clade. tic diff erentiation of a more generalist species, Leebens-Mack and Th e widespread evidence for allopatric speciation in insects en- Pellmyr (2004) showed only limited host-associated genetic struc- gaged in brood pollination mutualisms is intriguing, especially ture in mtDNA cytochrome oxidase sequences from T. yuccasella given that if there is a system in which allopatry would not be nec- across eastern North America. Such generalist species might repre- essary for speciation, it would probably be brood pollination mutu- sent relatively recent expansions of partner breadth; however, they alisms. Th ompson (1994) suggested that systems in which partner may also indicate that use of diff erent partner species across geo- species control the movement of each other’s gametes would be graphic space does not necessarily lead to genetic diff erentiation. If good candidates for coevolution driving speciation, and it is easy to so, this suggests intriguingly—and in contrast to the traditional co- imagine scenarios in which selection to prevent interspecifi c hy- speciation paradigm—that speciation in plants and pollinators in bridization between plant species might result in restricted move- brood pollination interactions may in many cases actually be de- ment/mating of pollinators across host plant species. Even so, a coupled. Some detailed studies of closely related fi g species and number of studies have argued that hybridization between closely their wasps at the regional scale are not incompatible with such a related sympatric fi gs mediated by shared pollinating fi g wasps may hypothesis (Jackson et al., 2008; Cornille et al., 2011), although to be common (Jackson et al., 2008; Renoult et al., 2009; Cornille et al., the best of our knowledge, broadly “generalist” (rather than oli- 2011 ; Wang et al., 2016 ). Th e role of allopatry in speciation within gophagous) fi g wasps analogous to Tegeticula yuccasella have not pollination brood mutualisms is consistent with examples of poten- been reported. Consequently, some cases of apparent cospeciation tial coevolutionary speciation from other interactions ( Parchman (phylogenetic congruence) may be the result of phylogenetic track- and Benkman, 2002 ; Hoso et al., 2010; Smith and Benkman, 2007; ing of insects onto plants rather than true contemporaneous cospe- Schluter, 2010), in which geographic isolation was always impli- ciation (de Vienne et al., 2013). Alternately, phylogenetic congruence cated in diversifi cation ( Hembry et al., 2014 ). Continued exami- may be generated by strict-sense cospeciation, but perfect congru- nation of the role of geography and biotic selection in brood ence is “erased” over time due to host shift s (e.g., Silvieus et al., pollination mutualisms is crucial for understanding the mecha- 2008 ). Although generalist species are the exceptions in pollination nisms of speciation. brood mutualisms, they provide a way to test the role of the mutualism in leading to population diff erentiation and speciation. More (4) Is there evidence for displacement of one pollinator clade by tests of host-associated divergence among populations of generalist another? —Th at host-shift s are common in brood pollination mutu- species are needed to catch “diversifi cation in action” and comple- alisms seems unquestionable. Only host-shift s or duplication (specia- ment the number of phylogenetic studies above the species level. tion by pollinators on the same host without host speciation; e.g., Page, 2003) followed by diff erential extinction can explain the ob- (3) Do plants and insects in brood pollination mutualisms speciate served phylogenetic incongruence between plants and brood pollina- in allopatry? —Perhaps because of the enormous species richness of tors at fi ne scales (Page, 2003; Jackson et al., 2008; Cruaud et al., 6 • AMERICAN JOURNAL OF BOTANY

2012a ; Hembry et al., 2013 ; Yang et al., 2015 ). Although duplications speciation in brood pollination mutualisms has long attracted at- and diff erential extinction may indeed occur in brood pollination tention (e.g., Ramírez, 1970 ; Kiester et al., 1984 ), but has been diffi cult mutualisms (especially if plant and pollinator speciation is decoupled, to distinguish from the alternative hypotheses that diversifi cation as already described), there are enough examples of shallow incon- in pollinators simply responds to the diversifi cation of their host gruences in plant and pollinator phylogenies to make clear that host- plants (or vice versa). Th e role of coevolution has only been explic- shift s occur. Particularly good examples include multiple shift s itly tested in one case, that of divergence in Joshua trees (Yucca between capsular and fl eshy-fruited yucca hosts in the Tegeticula yuc- brevifolia and Y . jaegeriana ) and speciation in their pollinating casella clade ( Althoff et al., 2012 ) and a series of rapid host shift s in a yucca moths, Tegeticula synthetica and T. antithetica. Th ese species recently colonizing Polynesian leaffl ower moth clade ( Hembry et al., are distributed in the Mojave Desert of the western United States. 2013 ). Host-shift s are also very likely in fi g wasps ( Molbo et al., 2003 ; Yucca brevifolia is pollinated by T . synthetica in the western Mojave Jackson et al., 2008 ; Cook and Segar, 2010 ), and possibly more so in Desert, and Y . jaegeriana is pollinated by T. antithetica in the east- wasps pollinating monoecious fi gs than those pollinating dioecious ern Mojave Desert ( Pellmyr and Segraves, 2003 ). Th ese two inter- fi gs (Yang et al., 2015). Th e eff ects of host-shift s are, however, not well acting plant–pollinator pairs are allopatrically distributed, with the known; they may include stable coexistence of both pollinators on exception of one contact zone (Tikaboo Valley in Nevada) where all the same host, competitive exclusion, or plant speciation ( Cook and four species co-occur ( Smith et al., 2009 ). Segar, 2010; Cornille et al., 2011; McLeish and van Noort, 2012; Extensive work on this interaction has shown a complex and in- Hembry et al., 2013 ). Evidence for competition between co-occurring terestingly nuanced view of the role and mechanisms that coevolu- pollinators has been reported for Elisabethiella fi g wasps on Ficus tion may play in speciation. More divergence is seen in fl oral traits natalensis in South Africa ( Cornille et al., 2011 ) and Epicephala on associated with the interaction than in vegetative traits associated Breynia in China ( Zhang et al., 2012 ). with the interaction ( Godsoe et al., 2008 ), and trait matching (be- One interesting common pattern seen in both Yucca and Glochid- tween fl ower stylar canal and moth ovipositor lengths) is seen at the ion is the possibility of widespread host-shift s by a pollinator clade, species level within each species pair ( Godsoe et al., 2010 ). Th ese potentially resulting in displacement of other pollinator lineages. divergent phenotypes between Y . brevifolia varieties persist in the Althoff et al. (2012) suggest that the pollinating yucca moth genus face of extensive contemporary gene fl ow ( Starr et al., 2013 ; Yoder Tegeticula may have displaced the other pollinating yucca moth genus et al., 2013 ). Furthermore, species distribution modeling suggests Parategeticula over much of its range in North America. Divergence that the abiotic requirements of Y . brevifolia and Y . jaegeriana do time estimates (based on molecular clock rates) suggest that the di- not diff er ( Godsoe et al., 2009 ). In the contact zone in Tikaboo Val- versifi cation of Tegeticula is 2–3 Myr more recent than that of Yucca ley, where all four species co-occur, there are asymmetric diff er- (Pellmyr and Leebens-Mack, 1999; Good-Avila et al., 2006; Smith ences in pollinator specifi city, withT. antithetica visiting both Y . et al., 2008a, b ); furthermore, Parategeticula is primarily restricted to brevifolia and Y. jaegeriana trees but T. synthetica nearly exclu- the southern edge of the range of Yucca today (in Arizona, USA and sively visiting Y . brevifolia trees (Smith et al., 2009). Th is asymme- Mexico). Where they co-occur on the same host, Tegeticula and Pa- try in visitation may explain observed patterns of asymmetric rategeticula oviposition sites diff er although larvae of both feed on introgression between the two Joshua tree varieties (Starr et al., seeds ( Pellmyr et al., 2008 ). Direct tests of competition are needed. 2013 ). However, survivorship of T. antithetica larvae is lower in Y . Similarly, Hembry et al. (2013) report that one Polynesian leaffl ower brevifolia fruits than in Y. jaegeriana fruits (Smith et al., 2009); this moth clade (clade Z) is associated with 12 diff erent host Glochidion diff erence can be explained by trait matching between ovipositor species across three archipelagos. Because this moth clade is restricted and stylar canal length. Furthermore, in the contact zone in the to oceanic islands, none of which (with one exception) have ever been Tikaboo Valley, plant–pollinator specifi city is important in pro- connected to one another, the ancestral moth species must have made moting or reinforcing this coevolution and reproductive isolation a series of host-shift s from an ancestral host onto many new Glochid- between the plants (Smith et al., 2009; Yoder et al., 2013; see Royer ion lineages on each island it colonized. Whether the colonization of et al., 2016 , in this issue). Th e divergence of T. synthetica and T. these Glochidion hosts by Epicephala clade Z resulted in competitive antithetica appears to predate that of Y . brevifolia and Y . jaegeriana , displacement of an earlier pollinator lineage is unclear. Th ere is an and the chloroplast phylogeography of Y . brevifolia and Y . jaegeri- older Epicephala clade in Southeastern Polynesia, and today the de- ana are not congruent with phylogenetic relationships between scendants of these two moth clades co-occur on several islands. On their two pollinators (Smith et al., 2008a). Th ese fi ndings suggest some of these islands the two clades co-occur on the same hosts, that the initial divergence in this system may have been due to a whereas on other islands, they use diff erent host species (D. Hembry vicariant event (such as the Miocene Bouse Embayment in today’s et al., unpublished manuscript). What is unclear for both Yucca or Colorado River Valley) that aff ected Tegeticula , and since this ini- Glochidion , is whether the widespread host-shift s in the pollinators tial event, coevolution between plant and pollinators has contrib- promoted additional diversifi cation. A sobering aspect to the results uted to the trait divergence and speciation that is observed today. from Tegeticula and Epicephala is that these host-shift s were only de- Th ese fi ndings underscore the diffi culty of distinguishing diver- tectable because of extensive extant taxon sampling, suggesting that sifi cation driven by coevolution from that in which either the pol- with additional sampling, similar widespread host-shift s may be de- linator is responding (through coevolution) to divergence in the tectable in leaffl owers elsewhere in the world and in fi gs. Having an plant, or the plants are diversifying in response to their pollinators accurate and complete picture of diversifi cation and host use patterns (as is suggested for other pollination interactions; Hodges and will be absolutely critical for determining the mechanisms of specia- Arnold, 1994; Ramírez et al., 2011). Indeed, the adoption of brood tion in pollination brood mutualisms. pollination did not increase diversifi cation rates inYucca relative to their sister clade ( Smith et al., 2008b ). However, they do suggest a (5) Is there evidence for a role of coevolution in speciation?— Th e number of interesting things about how coevolution may be impli- hypothesis that coevolution accompanies (or even promotes) cated in speciation. First, these results suggest that coevolution acts OCTOBER 2016 , VOLUME 103 • HEMBRY AND ALTHOFF—DIVERSIFICATION IN BROOD POLLINATION MUTUALISMS • 7

in concert with other factors such as geographic isolation (and is tested more broadly. Th ere is relatively little evidence for a “coevo- potentially in confl ict with the hypothesis stated above, that specia- lution through cospeciation” model and relatively limited evidence tion in plants and brood pollinators may be decoupled). Th is fi nd- overall that coevolution promotes speciation in these systems. ing is consistent with those from coevolving interactions other than What little evidence there is suggests that antagonistic coevolution brood pollination mutualisms ( Hembry et al., 2014 ). Second, they rather than mutualistic coevolution might be more important. suggest that although much theoretical and empirical work has fo- Coevolution may accompany or promote speciation as has been cused on whether and how mutualism might promote diversifi ca- demonstrated in Joshua trees, and this hypothesis needs to be tested tion, the antagonistic side of the overall interaction might be more in many additional species pairs of plants and their pollinators for important. Th e traits showing phenotype matching in this system us to truly understand the role of coevolution in pollination brood (moth ovipositor length and plant ovary wall thickness) interact mutualisms. antagonistically, suggesting that to the extent that coevolution is Although there is currently limited evidence of coevolutionary implicated in speciation in brood pollination mutualisms, it may be diversifi cation, there can be no question that the persistence of new through antagonistic coevolution as suggested by theoretical work species and diversifi cation of clades greatly depends on interactions ( Yoder and Nuismer, 2010 ). Furthermore, Althoff (2014) suggested between plants and pollinators. Speciation in brood pollinators and that, in Tegeticula, changes in ovipositor morphology led to corre- their host plants may not involve coevolution, but the interaction lated changes in male reproductive traits that could cause repro- between plants and brood pollinators may play a role in their diver- ductive isolation among pollinator species. Th us, for yucca moths sifi cation at macroevolutionary scales. Coevolution is not restricted changes in antagonistic traits may have a direct link to speciation. to pairwise interactions, and it is likely that whole clades of plants Finally, the Joshua tree–yucca moth example is a case of di- and brood pollinators infl uence each other’s evolution in the same versifi cation through some process which produces phylogenetic manner as has been argued for other mutualistic and antagonistic congruence (although it is not necessarily contemporaneous cospe- interactions (e.g., Ehrlich and Raven, 1964 ; Ricklefs, 2010 ). ciation; Smith et al., 2008a). Th is fi nding is in contrast to the exten- An important consideration is what type of interaction may be sive evidence for host-shift s at various phylogenetic scales in brood structuring the diversifi cation of brood pollinators and their host pollination mutualisms (but see dioecious fi gs in Yang et al., 2015 ). plants. Certainly, these relationships are mutualistic (except of As stated already, the consequences of host-shift s for diversifi cation course in cases of breakdown), but diversifi cation could still be me- in brood pollination mutualisms are not clear; one hypothesis is diated by either antagonistic or competitive interactions (Hembry that the presence of two distantly related (and therefore presumably et al., 2014 ). In antagonistic cases, plants could be under selection reproductively isolated) pollinators, with diff erent functional traits, to escape seed predation as much as they can, while pollinators on a single host plant might act on natural trait variation in the host could be under selection to obtain as much of their host’s seeds as and promote divergence (and speciation) in the host (Cook and possible (or, to switch onto new host species when available). Th is Segar, 2010 ; Hembry et al., 2013 ). Such a process would be a diff er- antagonism could also involve some form of coevolutionary alter- ent mechanism by which coevolution might be implicated in spe- nation (sensu Th ompson, 1994) in which pollinator larval feeding ciation in brood pollination interactions. Because we currently lack causes diff erent selection on plant defenses that leads to short bouts any analogous studies from any other pairs of jointly diverging of coevolution that prompts pollinators to shift to a less-defended plant and pollinator species in any brood pollination mutualisms, plant species. Such a process could be particularly important in testing the fi ndings from the above Joshua tree work with other spe- species-rich regional assemblages in which plants and pollinators cies pairs in similar systems is a high priority. could take advantage of new partners, and new trait combinations that permit this alternation could allow bursts of diversifi cation in a manner similar to that which has been argued for interactions DISCUSSION AND CONCLUSIONS between plants and their herbivores (Ehrlich and Raven, 1964; Futuyma and Agrawal, 2009 ). Although there is a great deal we do not know about how brood Alternately, competitive interactions may govern the diversifi ca- pollination mutualisms diversify, answers to the above questions tion of plants and pollinators, in a manner similar to that which has demonstrate that common patterns are beginning to emerge across been argued for other animal-pollinated plant radiations ( Schluter, systems. Th e most species-rich systems—yuccas, fi gs, and leaffl ow- 2000 ; Armbruster and Muchhala, 2009 ). In such a scenario, plant ers—show signifi cant phylogenetic congruence at large phyloge- species would be competing with each other for pollinators, and netic scales, but this phylogenetic congruence is far from exact. insect species would be competing with one another for host plants. Host-shift s appear to be common in all of these systems, with the Natural selection would thus promote character displacement in possible exception of dioecious fi gs. Th ere is limited evidence for traits that mediate partner-specifi city, and this character displace- both cospeciation and for duplication, and there are no unambigu- ment would promote high specifi city. Any new pollinator or plant ous examples known of strict-sense contemporaneous cospecia- species is therefore a resource for other species, and switches among tion. Much of the observed phylogenetic congruence may simply partners would be analogous to the switches among pollination be due to conservatism in host associations, host tracking, and the syndromes that are seen in plants such as Aquilegia (Hodges and isolation of diff erent subclades in diff erent biogeographic regions. Arnold, 1994 ). Geographic isolation (or more specifi cally, allopatry) appears to In the past two decades, the discovery of several new brood pol- be important in driving speciation in several diff erent brood polli- lination mutualisms, the molecular phylogenetics revolution, and nation mutualisms, particularly in the insects. Some intriguing evi- increased interest in coevolution have led to these mutualisms be- dence suggests that widespread host-shift s by pollinator lineages ing used for rigorous testing of long-standing ideas about the role may displace other pollinator lineages and that speciation in plants of biotic interactions in generating biodiversity. Although we have and insects may be decoupled; these are hypotheses that should be made great progress, there is still much about the diversifi cation of 8 • AMERICAN JOURNAL OF BOTANY

plants and brood pollinators that we do not understand. Many op- Cruaud , A. , J. Cook , Y. Da-Rong , G. Genson , R. Jabbour-Zahab , F. Kjellberg , portunities exist to apply recently developed macroevolutionary, R. A. S. Pereira , et al. 2012a . Fig–fi g wasp mutualism: Th e fall of the strict genomic, and network analytical methods to testing hypotheses cospeciation paradigm? In S. Patiny [ed.], Evolution of plant–pollinator re- about diversifi cation in brood pollination mutualisms. Th ese meth- lationships, 68–102. Cambridge University Press, Cambridge, UK. ods will need to be coupled with focused experimental tests on the Cruaud , A. , R. Jabbour-Zahab , G. Genson , S. Ungricht , and J.-Y. Rasplus . 2012b . Testing the emergence of New Caledonia: Fig wasp mutualism as a role of mutualism and antagonism in generating trait divergence to case study and review of evidence. PLoS One 7 : e30941 . provide a robust picture of coevolution and diversifi cation. Brood Cruaud , A. , N. Rønsted , B. Chantarasuwan , L. S. Chou , W. L. Clement , A. pollination mutualisms have inspired and fascinated evolutionary Couloux , B. Cousins , et al. 2012c . An extreme case of plant–insect codiver- biologists since the late 19th century and will continue to provide a sifi cation: Figs and fi g-pollinating wasps.Systematic Biology 61 : 1029 – 1047 . unique window onto the origins of biological diversity. Darwell , C. T., S. al-Beidh , and J. M. Cook . 2014 . Molecular species delimi- tation of a symbiotic fi g-pollinating wasp species complex reveals extreme deviation from reciprocal partner specifi city. BMC Evolutionary Biology 14 : ACKNOWLEDGEMENTS 189 . We thank Judith Bronstein, Atsushi Kawakita, Tomoko Okamoto, Després , L. , L. Gielly , B. Redoutet , and P. Taberlet . 2003 . Using AFLP to re- Olle Pellmyr, Rafael Raimundo, Kari Segraves, John Th ompson, solve phylogenetic relationships in a morphologically diversifi ed plant species complex when nuclear and chloroplast sequences fail to reveal vari- George Weiblen, and Jeremy Yoder for discussion on this topic over ability. Molecular Phylogenetics and Evolution 27 : 185 – 196 . the past decade and Edward Allen Herre, Jeremy Yoder, and one Després , L. , S. Loriot , and M. Gaudeul . 2002a . Geographic pattern of genetic anonymous reviewer for their many helpful suggestions, particularly variation in the European globefl ower Trollius europaeus L. (Ranunculaceae) with regards to the fi g literature. We also thank Jim Leebens-Mack inferred from amplifi ed fragment length polymorphism markers. Molecular and Jeremy Yoder for organizing the symposium and editing the Ecology 11 : 2337 – 2347 . special issue. David Hembry was supported by the NIH PERT Després , L. , E. Pettex , V. Plaisance , and F. Pompanon . 2002b . Speciation in program at the University of Arizona. AJB acknowledges that DHH the globefl ower fl yChiastocheta spp. (Diptera: Anthomyiidae) in relation to retains the right to provide a copy of the fi nal manuscript to the host plant species, biogeography, and morphology. Molecular Phylogenetics NIH upon acceptance for publication and for public archiving in and Evolution 22 : 258 – 268 . PubMed Central as soon as possible but no later than 12 months de Vienne, D. M., G. Refregier , M. Lopez-Villavicencio , A. Tellier , M. E. Hood , and T. Giraud . 2013 . Cospeciation vs. host-shift speciation: Methods for aft er publication by AJB. testing, evidence from natural associations and relation to coevolution. New Phytologist 198 : 347 – 385 . LITERATURE CITED Dietl , G. P., and P. H. Kelley . 2002 . Th e fossil record of predator–prey arms races: Coevolution and escalation hypotheses. In M. Kowalewski and P. Kelley [eds.], Althoff , D. M. 2014 . Shift in egg-laying strategy to avoid plant defense leads to Th e fossil record of predation. Paleontological Society Papers vol. 8, 353–374. reproductive isolation in mutualistic and cheating yucca moths. Evolution Ehrlich , P. R., and P. H. Raven . 1964 . Butterfl ies and plants: A study in coevo- 68 : 301 – 307 . lution. Evolution 18 : 586 – 608 . Althoff , D. M. 2016 . Specialization in the yucca–yucca moth obligate pollina- Espíndola , A. , S. Buerki , and N. Alvarez . 2012a . Ecological and historical tion mutualism: a role for antagonism? American Journal of Botany 103. drivers of diversifi cation in the fl y genusChiastocheta Pokorny. Molecular DOI: 10.3732/ajb.1600053. Phylogenetics and Evolution 63 : 466 – 474 . Althoff , D. M. , K. A. Segraves , and M. T. J. Johnson . 2014 . Testing forcoevo- Espíndola , A. , B. C. Carstens , and N. Alvarez . 2014 . Comparative phylogeog- lutionary diversifi cation: Linking pattern with process. Trends in Ecology & raphy of mutualists and the eff ect of the host on the genetic structure of its Evolution 29 : 82 – 89 . partners. Biological Journal of the Linnean Society 113 : 1021 – 1035 . Althoff , D. M. , K. A. Segraves , C. I. Smith, J. Leebens-Mack, and O. Pellmyr. Espíndola , A. , L. Pellissier , L. Maiorano , W. Hordijk , A. Guisan , and N. Alvarez . 2012 . Geographic isolation trumps coevolution as a driver of yucca and 2012b . Predicting present and future intra-specifi c genetic structure yucca moth diversifi cation. Molecular Phylogenetics and Evolution 62 : through niche hindcasting across 24 millennia. Ecology Letters 15 : 649 – 657 . 898 – 906 . Fleming , T. H., and J. N. Holland . 1998 . Th e evolution of obligate pollination Armbruster , W. S., and N. Muchhala . 2009 . Associations between fl oral spe- mutualisms: Senita cactus and senita moth. Oecologia 114 : 368 – 375 . cialization and species diversity: Cause, eff ect, or correlation? Evolutionary Futuyma , D. J., and A. A. Agrawal . 2009 . Macroevolution and the biological Ecology 23 : 159 – 179 . diversity of plants and herbivores. Proceedings of the National Academy of Berg , C. C. 1989 . Classifi cation and distribution of Ficus. Experientia 45 : Sciences, USA 106 : 18054 – 18061 . 605 – 611 . Ghana , S. , N. Suleman , and S. G. Compton . 2015 . Ability to gall: Th e ultimate Brown , J. M., O. Pellmyr , J. N. Th ompson , and R. G. Harrison . 1994 . Phylogeny basis of host specifi city in fi g wasps?Ecological Entomology 40 : 280 – 291 . of Greya (Lepidoptera: Prodoxidae) based on nucleotide sequence variation Godsoe , W. , E. Strand , C. I. Smith , J. B. Yoder , T. C. Esque , and O. Pellmyr . in mitochondrial cytochrome oxidase I and II: Congruence with morpho- 2009 . Divergence in an obligate mutualism is not explained by divergent logical data. Molecular Biology and Evolution 11 : 128 – 141 . climatic requirements. New Phytologist 183 : 589 – 599 . Chen , Y. , S. G. Compton , M. Liu , and X.-Y. Chen . 2012 . Fig trees at the north- Godsoe , W. , J. B. Yoder , C. I. Smith , C. S. Drummond , and O. Pellmyr . 2010 . ern limit of their range: the distributions of multiple cryptic pollinators indi- Absence of population-level phenotype matching in an obligate pollination cate multiple glacial refugia. Molecular Ecology 21 : 1687 – 1701 . mutualism. Journal of Evolutionary Biology 23 : 2739 – 2746 . Cook , J. M., and J.-Y. Rasplus . 2003 . Mutualists with attitude: Coevolving fi g Godsoe , W. , J. B. Yoder , C. I. Smith , and O. Pellmyr . 2008 . Coevolution and wasps and fi gs. Trends in Ecology & Evolution 18 : 241 – 248 . divergence in the Joshua tree/yucca moth mutualism. American Naturalist Cook , J. M., and S. T. Segar . 2010 . Speciation in fi g wasps. Ecological Entomology 171 : 816 – 823 . 35 : 54 – 66 . Good-Avila , S. V., V. Souza , B. S. Gaut , and L. E. Eguiarte . 2006 . Timing Cornille , A. , J. G. Underhill , A. Cruaud , M. Hossaert-McKey , S. D. Johnson , K. and rate of speciation in Agave (Agavaceae). Proceedings of the National A. Tolley , F. Kjellberg , et al. 2011 . Floral volatiles, pollinator sharing, and Academy of Sciences, USA 103 : 9124 – 9129 . diversifi cation in the fi g–wasp mutualism: Insights fromFicus natalensis, Guimarães , P. R. Jr ., V. Rico-Gray , P. S. Oliveira , T. J. Izzo , S. F. dos Reis, and and its two wasp pollinators (South Africa) . Proceedings of the Royal Society, J. N. Th ompson . 2007 . Interaction intimacy ects aff structure and coevolu- B, Biological Sciences 279 : 1731 – 1739 . tionary dynamics in mutualistic networks. Current Biology 17 : 1797 – 1803 . OCTOBER 2016 , VOLUME 103 • HEMBRY AND ALTHOFF—DIVERSIFICATION IN BROOD POLLINATION MUTUALISMS • 9

Hembry , D. H., A. Kawakita , N. E. Gurr , M. A. Schmaedick , B. G. Baldwin , and Luo , S.-X. , H.-J. Esser , D. Zhang , and S. S. Renner . 2011 . Nuclear ITS sequences R. G. Gillespie . 2013 . Non-congruent colonizations and diversifi cation in a help disentangle Phyllanthus reticulatus (Phyllanthaceae), an Asian species coevolving pollination mutualism on oceanic islands. Proceedings. Biological not occurring in Africa, but introduced to Jamaica. Systematic Botany 36 : Sciences 280 : 20130361 . 99 – 104 . Hembry , D. H., J. B. Yoder , and K. R. Goodman . 2014 . Coevolution and the Machado , C. A., N. Robbins , M. T. P. Gilbert , and E. A. Herre . 2005 . Critical diversifi cation of life. American Naturalist 184 : 425 – 438 . review of host specifi city and its coevolutionary implications in the g/fifi g- Herre , E. A., K. C. Jandér , and C. A. Machado . 2008 . Evolutionary ecology of wasp mutualism. Proceedings of the National Academy of Sciences, USA 102 fi gs and their associates: Recent progress and outstanding puzzles. Annual (supplement 1): 6558 – 6565 . Review of Ecology, Evolution and Systematics 39 : 439 – 458 . Mayr , G. L. 1885 . Feigeninsecten. Verhandlungen der Zoologisch-Botanischen Hodges , S. A. , and M. L. Arnold . 1994 . Columbines: A geographically wide- Gesellschaft in Wien 35 : 147 – 250 . spread species fl ock. Proceedings of the National Academy of Sciences, USA McLeish , M. J., and S. van Noort. 2012 . Codivergence and multiple host spe- 91 : 5129 – 5132 . cies use by fi g wasp populations of theFicus pollination mutualism. BMC Holland , J. N., and T. H. Fleming . 1999 . Mutualistic interactions between Evolutionary Biology 12 : 1 . Upiga virescens (Pyralidae), a pollinating seed consumer, and Lophocereus Moe , A. M., D. R. Rossi , and G. D. Weiblen . 2011 . Pollinator sharing in dioe- schottii (Cactaceae). Ecology 80 : 2074 – 2084 . cious fi gs (Ficus : Moraceae). Biological Journal of the Linnean Society 103 : Hoso , M. , Y. Kameda , S.-P. Wu , T. Asami , M. Kato , and M. Hori . 2010 . A 546 – 558 . speciation gene for left -right reversal in snails results in anti-predator adap- Moe , A. M., and G. D. Weiblen . 2010 . Molecular divergence in allopatric tation. Nature Communications 1 : 133 . Ceratosolen (Agaonidae) pollinators of geographically widespread Ficus Jackson , A. P., C. A. Machado , N. Robbins , and E. A. Herre . 2008 . Multi-locus (Moraceae) species. Annals of the Entomological Society of America 103 : phylogenetic analysis of neotropical fi gs does not support co-speciation with 1025 – 1037 . the pollinators: Th e importance of systematic scale in fi g/wasp cophyloge- Moe , A. M., and G. D. Weiblen . 2012 . Pollinator-mediated reproductive netic studies. Symbiosis 45 : 57 – 72 . isolation among dioecious fi g species (Ficus , Moraceae). Evolution 6 6: Janzen , D. H. 1979 . How to be a fi g. Annual Review of Ecology and Systematics 3710 – 3721 . 10 : 13 – 51 . Molbo , D. , C. A. Machado , J. G. Sevenster , L. Keller , and E. A. Herre . 2003 . Janzen , D. H. 1980 . When is it coevolution? Evolution 34 : 611 – 612 . Cryptic species of fi g-pollinating wasps: Implications for the evolution of the Kawakita , A. 2010 . Evolution of obligate pollination mutualism in the tribe fi g–wasp mutualism, sex allocation, and precision of adaptation. Proceedings Phyllantheae. Plant Species Biology 25 : 3 – 19 . of the National Academy of Sciences, USA 100 : 5867 – 5872 . Kawakita , A. , and M. Kato . 2006 . Assessment of the diversity and species speci- Okamoto , T. , A. Kawakita , R. Goto , G. P. Svensson , and M. Kato . 2013 . Active fi city of the mutualistic association between Glochidion trees and Epicephala pollination favours sexual dimorphism in fl oral scent. Proceedings of the moths. Molecular Ecology 15 : 3567 – 3581 . Royal Society, B, Biological Sciences 280 : 20132280 . Kawakita , A. , and M. Kato . 2009 . Repeated independent evolution of obli- Okamoto , T. , A. Kawakita , and M. Kato . 2007 . Interspecifi c variation of fl oral gate pollination mutualism in the Phyllantheae–Epicephala association. scent composition in Glochidion and its association with host-specifi c polli- Proceedings of the Royal Society, B, Biological Sciences 276 : 417 – 426 . nating seed parasite ( Epicephala). Journal of Chemical Ecology 33 : 1065 – 1081 . Kawakita , A. , K. Mochizuki , and M. Kato . 2015 . Reversal of mutualism in a Page , R. D. M. 2003 . Introduction . In R. D. M. Page [ed.], Tangled trees: leaffl ower–leaffl ower moth association: Th e possible driving role of a third- Phylogeny, cospeciation, and coevolution, 1–21. University of Chicago party partner. Biological Journal of the Linnean Society 116 : 507 – 518 . Press, Chicago, Illinois, USA. Kawakita , A. , T. Okamoto , R. Goto , and M. Kato . 2010 . Mutualism fa- Parchman , T. L., and C. W. Benkman . 2002 . Diversifying coevolution between vours higher host-specificity than does antagonism in plant–herbivore crossbills and black spruce on Newfoundland. Evolution 56 : 1663 – 1672 . interaction. Proceedings of the Royal Society, B, Biological Sciences 277 : Pellmyr , O. 1992 . Th e phylogeny of a mutualism: evolution and coadaptation 2765 – 2774 . between Trollius and its seed-parasitic pollinators. Biological Journal of the Kato , M. , A. Takimura , and A. Kawakita . 2003 . An obligate pollination mutualism Linnean Society 47 : 337 – 365 . and reciprocal diversifi cation in the tree genusGlochidion (Euphorbiaceae). Pellmyr , O. 1999 . Systematic revision of the yucca moths in the Tegeticula Proceedings of the National Academy of Sciences, USA 100 : 5264 – 5267 . yuccasella complex (Lepidoptera: Prodoxidae) north of Mexico. Systematic Kephart , S. , R. J. Reynolds , M. T. Rutter , C. B. Fenster , and M. R. Dudash . Entomology 24 : 243 – 271 . 2006 . Pollination and seed predation by moths on Silene and allied Pellmyr , O. 2003 . Yuccas, yucca moths, and coevolution: A review. Annals of Caryophyllaceae: Evaluating a model system to study the evolution of mu- the Missouri Botanical Garden 90 : 35 – 55 . tualisms. New Phytologist 169 : 667 – 680 . Pellmyr , O. , M. Balcázar-Lara , K. A. Segraves , D. M. Althoff , and R. J. Littlefi eld . Kiester , A. R. , R. Lande , and D. W. Schemske . 1984 . Models of coevolution 2008 . Phylogeny of the pollinating yucca moths, with revision of the and speciation in plants and their pollinators. American Naturalist 124 : Mexican species (Tegeticula and Parategeticula ; Lepidoptera, Prodoxidae) . 220 – 243 . Zoological Journal of the Linnean Society 152 : 297 – 314 . Kobmoo , N. , M. Hossaert-McKey , J. Y. Rasplus , and F. Kjellberg . 2010 . Ficus Pellmyr , O. , and J. Leebens-Mack . 1999 . Forty million years of mutual- racemosa is pollinated by a single population of a single agaonid wasp spe- ism: Evidence for Eocene origin of the yucca–yucca moth association. cies in continental South-East Asia. Molecular Ecology 19 : 2700 – 2712 . Proceedings of the National Academy of Sciences, USA 96 : 9178 – 9183 . Kusumi , J. , H. Azuma , H.-Y. Tzeng , L.-S. Chou , Y.-Q. Peng , K. Nakamura , and Pellmyr , O. , and K. A. Segraves . 2003 . Pollinator divergence within an obligate Z.-H. Su . 2012 . Phylogenetic analyses suggest a hybrid origin of the fi gs mutualism: two yucca moth species (Lepidoptera; Prodoxidae: Tegeticula ) (Moraceae: Ficus ) that are endemic to the Ogasawara (Bonin) Islands, Japan. on the Joshua tree (Yucca brevifolia; Agavaceae). Annals of the Entomological Molecular Phylogenetics and Evolution 63 : 168 – 179 . Society of America 96 : 716 – 722 . Leebens-Mack , J. , and O. Pellmyr . 2004 . Patterns of genetic structure among Pellmyr , O. , and J. N. Th ompson. 1992 . Multiple occurrences of mutualism populations of an oligophagous pollinating yucca moth (Tegeticula yucca- in the yucca moth lineage. Proceedings of the National Academy of Sciences, sella). Journal of Heredity 95 : 127 – 135 . USA 89 : 2927 – 2929 . Li , H. , Z. Wang , and B. Hu . 2015 . Four new species of Epicephala Meyrick, Poisot , T. 2015 . When is co-phylogeny evidence of coevolution? In S. Morand, 1880 (Lepidoptera, Gracillariidae) associated with two species of Glochidion B. R. Krasnov, and D. T. J. Littlewood [eds.], Parasite diversity and diver- (Phyllanthaceae) from Hainan Island in China. ZooKeys 508 : 53 – 67 . sifi cation: Evolutionary ecology meets phylogenetics, 420–433. Cambridge Li , H. , and X. Yang . 2015 . Th ree new species of Epicephala Meyrick University Press, Cambridge, UK. (Lepidoptera, Gracillariidae) associated with Phyllanthus microcarpus Ramírez B., W. 1970 . Host specifi city of fi g wasps (Agaonidae). Evolution 24 : (Benth.) (Phyllanthaceae). ZooKeys 484 : 71 . 680 – 691 . 10 • AMERICAN JOURNAL OF BOTANY

Ramírez , S. R., T. Eltz , M. K. Fujiwara , G. Gerlach , B. Goldman-Huertas , N. D. Suchan , T. , M. Beauverd , N. Trim , and N. Alvarez . 2015 . Asymmetrical nature Tsutsui , and N. E. Pierce . 2011 . Asynchronous divergence in a specialized of the Trollius – Chiastocheta interaction: Insights into the evolution of nurs- plant–pollinator mutualism. Science 333 : 1742 – 1746 . ery pollination systems. Ecology and Evolution 5 : 4766 – 4777 . Renoult , J. P., F. Kjellberg , C. Grout , S. Santoni , and B. Khadari . 2009 . Cyto- Th ompson, A. R., T. C. Adam , K. M. Hultgren , and C. E. Th acker . 2013a . nuclear discordance in the phylogeny of Ficus section Galoglychia and host- Ecology and evolution aff ect network structure in an intimate marine mutu- shift s in plant–pollinator associations. BMC Evolutionary Biology 9 : 248 . alism. American Naturalist 182 : E58 – E72 . Rich , K. A., J. N. Th ompson , and C. C. Fernandez . 2008 . Diverse historical Th ompson, J. N. 1994 . Th e coevolutionary process. University of Chicago processes shape deep phylogeographical divergence in the pollinating seed Press, Chicago, Illinois, USA. parasite Greya politella. Molecular Ecology 17 : 2430 – 2448 . Th ompson, J. N. 2005 . Th e geographic mosaic of coevolution. University of Ricklefs , R. E. 2010 . Evolutionary diversifi cation, coevolution between popula- Chicago Press, Chicago, Illinois, USA. tions and their antagonists, and the fi lling of niche space. Proceedings of the Th ompson, J. N. 2009 . Th e coevolving web of life. American Naturalist 173 : National Academy of Sciences, USA 107 : 1265 – 1272 . 125 – 140 . Riley , C. V. 1892 . Th e yucca moth and yucca pollination. Annals of the Missouri Th ompson , J. N. 2013 . Relentless evolution. University of Chicago Press, Botanical Garden 3 : 99 – 158 . Chicago, Illinois, USA. Royer , A. M., M. A. Streisfeld , and C. I. Smith . 2016 . Population genomics of diver- Th ompson , J. N. , and B. M. Cunningham . 2002 . Geographic structure and dy- gence within an obligate pollination mutualism: Selection maintains diff erences namics of coevolutionary selection. Nature 417 : 735 – 738 . between Joshua tree species . American Journal of Botany 103. DOI: 10.3732/ Th ompson , J. N. , and O. Pellmyr . 1992 . Mutualism with pollinating seed parasites ajb.1600069. amid co-pollinators: Constraints on specialization. Ecology 73 : 1780 – 1791 . Segraves , K. A., and O. Pellmyr . 2001 . Phylogeography of the yucca moth Th ompson , J. N. , and K. A. Rich . 2011 . Range edges and the molecular diver- Tegeticula maculata: Th e role of historical biogeography in reconcil- gence of Greya moth populations. Journal of Biogeography 38 : 551 – 563 . ing high genetic structure with limited speciation. Molecular Ecology 1 0: Th ompson, J. N., C. Schwind , P. R. Guimarães , and M. Friberg . 2013b . 1247 – 1253 . Diversifi cation through multitrait evolution in a coevolving interaction. Schluter , D. 2000 . Th e ecology of adaptive radiation. Oxford University Press, Proceedings of the National Academy of Sciences, USA 110 : 11487 – 11492 . Oxford, UK. Tian , E. , J. D. Nason , C. A. Machado , L. Zheng , H. Yu , and F. Kjellberg . 2015 . Schluter , D. 2010 . Resource competition and coevolution in sticklebacks. Lack of genetic isolation by distance, similar genetic structuring but diff er- Evolution: Education and Outreach 3 : 54 – 61 . ent demographic histories in a fi g-pollinating wasp mutualism. Molecular Silvieus , S. I., W. L. Clement , and G. D. Weiblen . 2008 . Cophylogeny of fi gs, Ecology 24 : 5976 – 5991 . pollinators, gallers, and parasitoids. In K. J. Tilmon [ed.], Specialization, van Welzen, P. C., K. Pruesapan , I. R. H. Telford , and J. J. Bruhl . 2015 . Historical speciation, and radiation: Th e evolutionary biology of herbivorous insects, biogeography of Breynia (Phyllanthaceae): What caused speciation? Journal 225–239. University of California Press, Berkeley, California, USA. of Biogeography 42 : 1493 – 1502 . Smith , C. I., W. K. W. Godsoe , S. Tank , J. B. Yoder , and O. Pellmyr . 2008a . Wang , G. , C. H. Cannon , and J. Chen . 2016 . Pollinator sharing and gene fl ow Distinguishing coevolution from covicariance in an obligate pollination among closely related sympatric dioecious fi g taxa. Proceedings. Biological mutualism: Asynchronous divergence in Joshua tree and its pollinators. Sciences 283 : 20152963 . Evolution 62 : 2676 – 2687 . Weiblen , G. D. 2002 . How to be a fi g wasp. Annual Review of Entomology 47 : Smith , C. I., O. Pellmyr , D. M. Althoff , M. Balcázar-Lara, J. Leebens-Mack , 299 – 330 . and K. A. Segraves . 2008b . Pattern and timing of diversifi cation inYucca Weiblen , G. D., and G. L. Bush . 2002 . Speciation in fi g pollinators and para- (Agavaceae): Specialized pollination does not elevate rates of diversifi cation. sites. Molecular Ecology 11 : 1573 – 1578 . Proceedings. Biological Sciences 275 : 249 – 258 . Wiebes , J. T. 1979 . Co-evolution of fi gs and their insect pollinators. Annual Smith , C. I., S. Tank , W. Godsoe , J. Levenick , E. Strand , T. Esque , and O. Review of Ecology and Systematics 10 : 1 – 12 . Pellmyr . 2011 . Comparative phylogeography of a coevolved community: Yang , L.-Y. , C. A. Machado , X.-D. Dang , Y.-Q. Peng , D.-R. Yang , D.-Y. Zhang , Concerted population expansions in Joshua trees and four yucca moths. and W.-J. Liao . 2015 . Th e incidence and pattern of copollinator diversifi ca- PLoS One 6 : e25628 . tion in dioecious and monoecious fi gs. Evolution 69 : 294 – 304 . Smith , C. I., J. B. Yoder , W. K. Godsoe , and O. Pellmyr . 2009 . Host specifi c- Yoder , J. B., and S. L. Nuismer . 2010 . When does coevolution promote diversi- ity and reproductive success of yucca moths (Tegeticula spp., Lepidoptera: fi cation? American Naturalist 176 : 802 – 817 . Prodoxidae) mirror patterns of gene fl ow between host plant varieties of Yoder , J. B., C. I. Smith , D. J. Rowley , R. Flatz , W. Godsoe , C. Drummond , and Joshua tree ( Yucca brevifolia: Agavaceae). Molecular Ecology 18 : 5218 – 5229 . O. Pellmyr . 2013 . Eff ects of gene fl ow on phenotype matching between two Smith , J. W., and C. W. Benkman . 2007 . A coevolutionary arms race causes varieties of Joshua tree ( Yucca brevifolia , Agavaceae) and their pollinators. ecological speciation in crossbills. American Naturalist 169 : 455 – 465 . Journal of Evolutionary Biology 26 : 1220 – 1233 . Starr , T. , K. Gadek , J. B. Yoder , R. Flatz , and C. I. Smith . 2013 . Asymmetric hy- Zhang , J. , S. Wang , H. Li , B. Hu , X. Yang , and Z. Wang . 2012 . Diff use coevolu- bridization and gene fl ow between Joshua trees (Agavaceae: Yucca) refl ects tion between two Epicephala species (Gracillariidae) and two Breynia spe- diff erences in pollinator host specifi city. Molecular Ecology 22 : 437 – 449 . cies (Phyllanthaceae). PLoS One 7 : e41657 .