HYBRID SPECIATION IN ANIMALS: EVIDENCE FROM Heliconius BUTTERFLIES

by

Camilo Salazar Clavijo July, 2006

A dissertation submitted to Faculty of Biological Science of Universidad de los Andes in partial fulfillment of the requirements for the degree of Doctor in Philosophy in Biological Science

ACKNOWLEDGEMENTS

Thanks are due first of all to Dr. Mauricio Linares for his initiative and interest in the Heliconius hybridization. He conceived the idea of the recreation of H. heurippa from H. cydno and H. melpomene crosses in his 1989 Ph.D. dissertation. He motivated me to study for Ph.D under his supervision, which turn out to be an interesting and valuable experience. Also, he advised me with his wonderful conversations about Heliconius genetics and with his effective effort in obtaining funds and individuals to carry out the investigation. Special thanks to my home country (Colombia) for providing me with a doctoral and basic science scholarship through COLCIENCIAS and Banco de La Republica. Also, I would like to point out that 90% of the work was carried out in my home country demonstrating that in spite of many troubles good investigation can be conducted in Colombia. Thanks to my Advisor Dr. Eldredge Bermingham for his ideas about the phylogenetical topologies, broad experience in seeing what was important and in providing finance and housing support in the Smithsonian Tropical Research Institute (STRI), where some of the molecular data were gathered. I want to express my thanks for the credibility and support of my Advisor Chris Jiggins who was essential in different periods throughout these four years, helping me with his experimental and theoretical ideas about speciation, the revision of my writings, contributions to the preparation of the manuscript, interest in my academic interaction with other scientists and housing in STRI and in Edinburgh University. Thanks to Dr Jody Hey, who kindly gave me the opportunity to play, understand, and face my interest in biological theory, for his amiability throughout my Rutgers University visit. Dr. Jesse Taylor merits a special mention for his help and interest in developing a coalescence model and spending his time in giving his opinions and suggesting paths to follow in the analysis of migration rates. Dr. Kronforst rendered important help in getting reflectance wing spectra and sharing his molecular data about H. pachinus that shed light about using these markers in H. heurippa. I am grateful for this. Christian Salcedo and Natalia Giraldo collaborated with me in carrying out some of the behavioral experiments. Natalia also, helped with elaboration of some of the figures presented here and in the papers. Thanks for this help. Carlos Arias and Maria Clara Melo deserve a mention for their collaboration in the lab. Also, the former was an important and friendly help in my work in Panama and a good office partner. Enrique Bustos, Emigdio Guiza, Luz Marina 2

Pedraza and Luz Dary Padilla freely helped with stocks and broods care in La Vega and Luz Dary also helped in the lab. Cielo De Oro and Ana Rita Aldana helped with logistics in the elaboration of projects and financial reports. I want to give thanks to all the work partners with whom I shared time in the Genetics Institute and all the students I taught, from whom I learned several things. Thanks to my family for their unconditional and essential support. Finally, I would like to thank Carolina Barriga for the special moments and her help in editing the document as well as to Luz Betancourt for her friendship.

3

CONTRIBUTIONS TO THIS WORK

I hereby state that the work contained in this thesis was entirely carried out by myself, with the exception of the following specific areas where I have included work carried out in collaboration with others:

Chapter 2 was published in Journal of Evolutionary Biology vol. 18: 247-256 with the co-authors Chris D. Jiggins, Carlos F. Arias, Alexandra Tobler, Eldredge Berminghgam and Mauricio Linares. I was responsible for carrying out all of the data collection and analysis and received help from Carlos Arias in the laboratory work, Alexandra Tobler designed ap primers used in the study.

Chapter 3 has been submitted for publication to Genetics. Marcus Kronforst provided primer sequences and aliquots using for amplifying inv, Dll, Sd and w loci. Also, he gave me reference sequences from H. melpomene and H. cydno populations. Jesse Taylor developed the coalescence model using simulation techniques in order to obtain migration rates through different hypotheses, including those that compare hybrid speciation vs. ongoing gene flow. Also, he obtained summary gene flow statistics and likelihood surfaces for each gene.

Chapter 4 describes work that has in large part been published as an article in Nature 441: 247-256, which is included here as an Appendix. In the chapter I summarise the areas of this work that I carried out myself. In addition Christian Salcedo helped with pattern colour models data collection and analysis, and Jesús Mavárez developed and used 12 microsatellite loci to carry out assignment tests for populations of H. melpomene, H. cydno and H. heurippa. Results of this analysis are briefly cited in chapter IV. Also, figures 4.1a and 4.1b were drawn by him. Other contributions of this author were cited in the annex. Chris Jiggins created the two step speciation model depicted in annex figure 8. This model is described in different points throughout the chapters. 4

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ........................................................................................... 2 CONTRIBUTIONS TO THIS WORK......................................................................... 4 TABLE OF CONTENTS ............................................................................................... 5 LIST OF FIGURES........................................................................................................ 7 LIST OF TABLES.......................................................................................................... 8 ABSTRACT .................................................................................................................... 9 PREFACE ..................................................................................................................... 10 CHAPTER I .................................................................................................................. 13 Animal diversification through recombinational speciation ...................................... 13 Theoretical background .......................................................................................... 13 Plausibility of hybrid speciation according to simulated parameters ..................... 16 Hybrid speciation genetics ..................................................................................... 17 Frequency of stabilized recombinant lineages and homoploid hybrid speciation.. 19 Methods and difficulties in reconstructing patterns of reticulate evolution ........... 29 New molecular attempts in hybrid origin detection ............................................... 34 Reproductive isolation and hybrid speciation ........................................................ 36 Conclusion.............................................................................................................. 38 References .............................................................................................................. 39 CHAPTER II ................................................................................................................ 47 Hybrid incompatibility is consistent with a hybrid origin of Heliconius heurippa Hewitson from its close relatives, Heliconius cydno Doubleday and Heliconius melpomene Linnaeus. ................................................................................................. 47 ABSTRACT ........................................................................................................... 47 INTRODUCTION .................................................................................................. 48 RESULTS............................................................................................................... 52 Crosses between H. heurippa and H. melpomene .............................................. 52 Crosses between H. heurippa and H. cydno....................................................... 55 DISCUSSION......................................................................................................... 56 Sterility and the status of H. heurippa................................................................ 56 Recessivity and the genetics of hybrid sterility .................................................. 57 Using the ‘Speciation Clock’ to study relationships .......................................... 58 MATERIALS AND METHODS ........................................................................... 60 Statistical analysis .............................................................................................. 61 Polymerase Chain Reactions methods for Tpi and Apterous ............................. 62 Appendix 2.1 ...................................................................................................... 64 Appendix 2.2 ...................................................................................................... 65 References .............................................................................................................. 66

5

CHAPTER III ............................................................................................................... 76 Resolving introgression and hybrid speciation among Heliconius butterflies ........... 76 ABSTRACT ........................................................................................................... 76 INTRODUCTION .................................................................................................. 77 RESULTS............................................................................................................... 79 Description of gene regions................................................................................ 79 Species relationships and population genetic parameters................................... 80 History of divergence between H. melpomene and H. cydno............................. 85 Isolation-Migration model including H. heurippa ............................................. 86 A hybrid speciation coalescent model ................................................................ 88 DISCUSSION......................................................................................................... 92 Gene flow between H. c. cordula and H. m. melpomene ................................... 92 The history of H. heurippa ................................................................................. 93 CONCLUSIONS .................................................................................................... 98 MATERIALS AND METHODS ........................................................................... 99 Specimen collection............................................................................................ 99 PCR and sequencing methods .......................................................................... 100 Phylogenetic analysis ....................................................................................... 101 Population genetic analysis .............................................................................. 102 SUPPLEMENTARY MATERIAL ...................................................................... 107 Appendix 3.1. ................................................................................................... 107 Appendix 3.2 .................................................................................................... 109 References ............................................................................................................ 110 CHAPTER IV ............................................................................................................. 122 Genetic hybrid trait reconstruction and recombinational speciation in Heliconius butterflies.................................................................................................................. 122 ABSTRACT ......................................................................................................... 122 INTRODUCTION ................................................................................................ 122 RESULTS AND DISCUSSION........................................................................... 124 MATERIALS AND METHODS ......................................................................... 134 Crosses.............................................................................................................. 134 No-choice experiments ..................................................................................... 134 Mate choice experiments: Tetrads.................................................................... 135 Mate choice experiments: Colour pattern models ............................................ 136 References ............................................................................................................ 141 CONCLUSIONS......................................................................................................... 154 ANNEX ........................................................................................................................ 156

6

LIST OF FIGURES Figure 1.1 Schematic representation of chromosomal hybrid speciation model........................ 14 Figure 1.2 Example of phylogenetic network with a single hybrid species (B)......................... 31 Figure 1.3 Speciation in coalescence representation.................................................................. 35 Figure 1.4 Hybrid origin coalescence scenario. ......................................................................... 36 Figure 2.1 Species used in this study. ........................................................................................ 50 Figure 2.2 Geographic distribution of the study species in the eastern Andes........................... 51 Figure 2.3 Segregation of sterility phenotypes in backcross broods. ......................................... 55 Figure 3.1 Phylogenetic relationships of H. heurippa................................................................ 81 Figure 3.2 Allele networks for nuclear genes............................................................................. 83 Figure 3.3 Single locus a, b and multilocus c profile likelihoods for the migration rates......... 91 Figure 3.4 Coalescent model employed in estimating the three species population parameters.94 Figure 4.1 Reconstruction of the H. heurippa wing pattern..................................................... 126 Figure 4.2 Live courtship experiments..................................................................................... 128 Figure 4.3 Relative probabilities of H. heurippa males approaching and courting colour pattern models .............................................................................................................................. 129 Figure 4.4 Colour pattern models for H. melpomene. .............................................................. 130 Figure 4.5 Colour pattern models for H. cydno........................................................................ 131 Figure 4.6 Oviposition preference............................................................................................ 132

7

LIST OF TABLES Table 1.1 Hypothetical example of transgressive segregation. .................................................. 19 Table 1.2 Hypothesized examples of bisexual diploid taxa of hybrid origin. ............................ 20 Table 2.1. Hatch rate of control, F1 and backcross broods. ....................................................... 53 Table 2.2. Collection localities for the Heliconius races used in these crosses. ......................... 60 Table 3.1 Summary of genetic polymorphism data for mtDNA sequences in each population. 82 Table 3.2 Genetic structure (FST) values for comparisons between the three populations. ........ 84 Table 3.3 Summary of genetic polymorphism data for sex-linked and nuclear loci sequences. 85 Table 3.4 Genealogical parameters estimated under the IM model. .......................................... 87 Table 3.5 Multilocus maximum likelihood estimates of the migration rate parameters. .......... 90 Table 3.6 The pattern of segregating sites. ................................................................................. 96 Table 4.1 Relative mating probabilities in no-choice experiments. ......................................... 127 Table 4.2 Number of matings in tetrad mate-choice experiments............................................ 127

8

ABSTRACT

The formation of novel lineages from two or more ancestral species through gene flow, without change in chromosome number, is known as homoploid speciation. Extensive studied cases exist in plants such as Helianthus and Iris in which the genetics, theoretical predictions and ecological circumstances that contribute to the process are well understood. In animals, although plausible, this reticulate pattern is poorly documented. There are several examples suggested by phylogenetic incongruence and intermediate morphological traits. But completely dissected examples are practically absent. The rarity of this kind of speciation is caused principally by two obstacles that any new hybrid lineage faces; that is, the difficulty of achieving: (1) high hybrid fitness and (2) reproductive isolation from parental lineages. H. heurippa, a neotropical butterfly located in the Colombian east Andes, has been proposed as a hybrid entity between H. cydno and H. melpomene. Genomic compatibility experiments with the three species suggest that H. heurippa have less incompatibility with H. melpomene than those observed between H. cydno and H. melpomene, suggesting that this entity is not a differentiated subspecies of H. cydno. Molecular data were superficially concordant with hybrid hypothesis because, of the seven loci evaluated: these taxa were monophyletic and related to H. cydno in mtDNA (CoI-CoII); intermediate in a sexlinked locus (Tpi); and have a mixed alleles pattern with H. melpomene and H. cydno in inv, Dll, w and Sd nuclear markers. Estimates of genetic differentiation showed a pattern consistent with hybrid origin through the loci, but coalescence simulations could only discard an ancient speciation event without recurrent gene flow. Distinguishing between recent hybridization and ongoing gene flow was not possible, partly as consequence of heterogeneity in migration rates per locus. Synthetic hybrid reconstruction suggests that at least two backcrosses to H. cydno and subsequent crosses between heterozygous genotypes are needed to obtain a phenotype similar to wild H. heurippa. This pattern breeds true with wild individuals suggesting that the reconstructed pattern elements are homologous to those found in the wild. H. heurippa has strong reproductive isolation from its parental species through assortative mating, based on colour pattern cues. Hence in this dissertation, all the evidence shows a complete picture of the nature of H. heurippa as a hybrid species in which a hybrid trait, namely pattern colour, produces reproductive isolation overcoming the principal obstacles associated with homoploid speciation. 9

PREFACE

Hybrid speciation asssociated with a change in the level of ploidy has been frequently documented in plants. The relatively high incidence of these kinds of mixed genomes is a product of the spontaneous reproductive isolation between a tetraploid and its diploid parental lineages as a consequence of low fertility triploid bridges and the existence of underdominant rearrangements. A less documented hybrid speciation model is that which proceeds without a change in chromosome number. Some relevant cases are known in plants, specifically in the sunflower genera Helianthus, but both polyploid and homoploid hybrid speciation are poorly known in animals. There are two principal obstacles to hybrid speciation, the achievement of high hybrid fitness and reproductive isolation from parental lineages, but there is no clear consensus in theoretical models about the mechanisms involved in overcoming these barriers. This dissertation focuses on how new stabilized Heliconius species can emerge through introgressive hybridization and the importance of a hybrid trait in its speciation.

Chapter I reviews general theoretical models proposed to explain the plausibility of homoploid speciation, parameters measured by these models and different parameter interaction under simulation that allows production of plausible hybrid species under adverse conditions. High probability of this kind of speciation was obtained when habitat availability and few chromosomal differences were present. But, until now no consideration of other kinds of reproductive isolation (e.g. assortative mating) has been made. Also in this chapter, genetics of hybrid speciation is dealt with and a principal role for additive traits involved in transgressive segregation phenotypes is documented for the majority of the known cases. A summary of the frequency of stabilized recombinant animal lineages is presented with descriptions of the kind of evidence supporting each case. A considerable number of examples fulfil the requisite of having a stable genome, but practically only one (the case study in this dissertation) has relevant data with regard to the importance of the hybrid trait in reproductive isolation.

Then, I consider the methods and difficulties in reconstructing patterns of reticulate evolution and contrast different methodologies. The principal focus was on new algorithms designed to obtain information from sequence data, both to reconstruct 10

reticulations and to estimate demographic parameters as gene flow that could be used to prove homoploid speciation scenarios. Also, morphological and ecological methods were suggested as necessary steps in revealing the history of the hybrid species. Next, the importance of reproductive isolation in hybrid speciation is discussed. Finally future ways to study this kind of speciation and conclusions about its relevance are addressed.

Chapter II describes the results of a series of crossing experiments that investigate hybrid compatibility between three species, Heliconius heurippa, H. cydno and H. melpomene. This is a species group that has rapidly radiated and is of considerable interest with regard to mechanisms of speciation. In particular, studies of colour pattern genetics (see chapter IV) suggest that H. heurippa from the foothills of the Colombian Llanos, may have originated through hybridization between its close relatives, H. cydno and H. melpomene. Because of the rapid radiation and the ongoing introgression within this group of butterflies, traditional phylogenetics methods, based on multiple gene genealogies have given conflicting information. Results describing patterns of genomic compatibility therefore shed new light on the relationships of these species, using accumulation of hybrid breakdown. H. heurippa shows less intrinsic incompatibilities with H. melpomene (asymmetric Haldane’s Rule) than the equivalent cross involving H. cydno and H. melpomene (symmetric Haldane’s Rule). This result is consistent with the idea that H. heurippa is not an H. cydno cognate and also with its possible hybrid origin. Also, simple phylogenetic relationships, using presence or absence of gene colour pattern phenotypes produce incongruent phylogenies in which five colour pattern characters (presence of red band, red line, red dots and absence of brown forceps and iridescence) affiliate Heliconius heurippa with H. melpomene and one character (presence of yellow band) affiliates it with H. cydno. Experimental crosses between H. heurippa with H. cydno and H. melpomene show that H. heurippa has colour traits homologous with colour pattern elements of both putative parental species.

In Chapter III, I attempt to demonstrate that Heliconius heurippa has a molecular pattern consistent with hybrid speciation hypothesis using a multilocus approach. The genome of the putative hybrid is studied, drawing upon both phylogenetic methods and a novel coalescence model. The data was intriguing, showing particular nuclear loci (Dll, inv, Sd and w) that have a mosaic distribution of H. heurippa alleles derived from both parental lineages. Also, H. heurippa forms a monophyletic clade related to H. 11

cydno in mtDNA (CoI-CoII) and has an intermediate position in a sex-linked locus (Tpi). These topological and genetic data superficially seem to strongly support the hybrid speciation hypothesis. However, the novel coalescent model employed here fails to distinguish between alternative hypotheses of high ongoing gene flow and a recent hybrid origin, in part because of between-locus variation in the distribution of molecular polymorphisms. If hybrid speciation is important, it must necessarily occur in taxa with significant rates of introgressive hybridization, such as those in which shared variation is observed, and makes rigorous testing of the alternative hypotheses of hybrid founding versus introgressive hybridization necessary. Since H. heurippa data show patterns of allele sharing similar to those previously used as support for hybrid speciation in other taxa it is necessary to contrast alternative hypotheses as ongoing gene flow before concluding that this kind of speciation occurs in these taxa.

Chapter IV reveals the hybrid speciation nature of Heliconius heurippa. Here a genetic reconstruction of the colour pattern was made through inter-specific crosses between H. melpomene and H. cydno, taking advantage of the colour pattern differences between these species, determined largely by three co-dominant loci. In only two backcross generations to H. cydno, F1 males (H. cydno females times H. melpomene males) times H. cydno females, a virtually identical H. heurippa colour pattern was obtained. Crosses between reconstructed and wild specimens prove that colour elements between parents and H. heurippa are homologous. Once hybrid origin was recreated, reproductive isolation between H. heurippa from H. melpomene and H. cydno was tested. No-choice mating experiments showed a reduced probability of mating in all inter-specific comparisons, with H. heurippa females particularly reluctant to mate with H. cydno and H. melpomene males. In courtship experiments, H. heurippa males were ten times more likely to court their own females. In mating experiments with choice, there is a similar strong assortative mating with low inter-specific matings. Isolation due to assortative mating is considerably higher in comparison with other Heliconius mechanisms involved in species boundaries. Using colour pattern models, a relevant role for hybrid colour pattern as a cue in mate choice was documented. Finally, a model for hybrid origin and speciation of H. heurippa has been proposed, with the evidence collected in this and previous chapters. This study provides the first demonstration that hybrid trait can promote reproductive isolation, solving the principal obstacles for recombinant speciation. 12

CHAPTER I Animal diversification through recombinational speciation Recombinational speciation, the reticulate pattern in which a new species is produced by hybridization of two different lineages with the same ploidy level, is less well understood than the speciation by hybridization involving variation in chromosome number [1,2]. There are two possible reasons for this tendency: homoploid hybrid species could be rare in nature; or its detection is difficult with the morphological, genetic and analytical methods available until now [3].

The rarity of homoploid hybrid speciation is probably due to the fact that hybrid traits are commonly non adaptive and/or are swamped by ongoing sympatric or parapatric gene flow with parental lineages [2,4]. If these difficulties could be overcome readily, then hybrid speciation might be common. However as I mention above, the fact that few documented cases exist could be the product of the limited capacity for detection inherent in the genetic and morphological methodologies. Specifically, confusion between 1) shared polymorphisms retained from hybridization event(s), and 2) recent gene flow between parental species and the possible hybrid entity is common in molecular approaches.

Here, a revision of what it is known about animal hybrid speciation is presented. Ecological, genetic and behavioral conditions that could act in favor of this kind of speciation are discussed. Also, a description of methodological techniques and perspectives in the field are evaluated.

Theoretical background Two models of hybrid speciation, which resolve the isolation from parents, are known [5]. One of these is based on the existence of different chromosomal rearrangements in the parental species. Heterozygous F1 hybrids have a reduction in fertility as a consequence of impaired chromosomal pairing in meiosis or reduction in viable gametes generated by crossing-over between heterologous chromosomal blocks [5]. 13

Although this reduction in fertility in the F1 reduces gene flow between parent species, backcrossing of these individuals may give rise to novel chromosomally balanced individuals with restored fertility [2,5] (Figure 1.1).

Figure 1.1 Schematic representation of chromosomal hybrid speciation model. Parental species (chromosome number 2n=8) with two reciprocal translocations were crossed. Heterozygous F1 hybrids produce 12 unbalanced (unviable) and four balanced (viable) gametes. Parental structures may be recovered in two gametes and the remaining two will be recombinant forms. In the F2 a small fraction of novel fertile and stable homozygotic karyotypes that are partially or completely isolated from the parental species could be produced.

The second model involves the establishment of a novel hybrid linage by external barriers to gene flow [1]. In this case, genetic isolation results as a byproduct of ecological selection and niche differentiation between hybrids and the parent species [6,7]. If by chance a hybrid form is adapted to a different niche from the parental species, then a hybrid species may be established in sympatry as a consequence of ecological isolation. Both models are non exclusive and could contribute synergistically to hybrid speciation. A more general model incorporating the conditions and assumptions mentioned above has recently been proposed [8].

14

In this proposal, chromosomal and genic incompatibility were included as well as strong selection and ecological divergence [8]. Four critical steps are considered:

1) Hybridization is followed by inbreeding and hybrid breakdown due to chromosomal or genic incompatibilities.

2) Hybrids with the highest fertility or viability are favored by selection.

3) A new hybrid genotype may become stabilized if it becomes reproductively isolated from the parental species.

4) Once a hybrid genotype becomes stabilized, it must co-exist with one or both parents or occupy a new ecological niche. Either outcome requires ecological divergence.

Two factors emphasized in this model appear to facilitate hybrid speciation. First, reproductive barriers between a stabilized hybrid genotype and its parents could arise through rapid chromosomal evolution caused by chromosomal breakage [9,10] and high mutation rates [11]. Added to this, higher rates of chromosomal evolution would be enhanced by inbreeding and/or population subdivision because both will reduce effective population sizes and increase the fixation of novel chromosomal rearrangements through drift [6]. The second factor is the availability of suitable habitat. Open geographical space will enhance the establishment and success probability of hybrids. Also, habitat disturbance may contribute to breakdown of premating reproductive barriers between previously isolated parental species, increasing hybridization frequency and providing suitable habitat for hybrid segregates, which often diverge ecologically from both parents [6,7].

15

Plausibility of hybrid speciation according to simulated parameters

The likelihood that hybrid speciation takes place under conditions proposed by models mentioned above has been tested through simulation [5,12]. An early approach assumes a uniform ecological environment, genetic factors affecting viability/fertility, recurrent breeding in the hybrid zone, nonoverlapping generations and genetic differencesvarying through generations [12]. Results obtained indicate that the process is facilitated by higher fertility in the F1 hybrids, a higher selective advantage for the optimal type, and a long hybrid zone interface. Also, high selfing rates accelerate the process, but hybrid speciation does not depend absolutely on this kind of breeding [12]. Interestingly, hybrid genotypes could increase until fixation, supplanting the parental species after a long stasis period (punctuated time). The dynamic of the process is also facilitated by mating between clumped hybrid individuals with fewer chromosomal differences [12]. New insights about the process were obtained from improved simulations that rely on empirical observations in plants, and from the existence of two different chromosomal rearrangements [6]. Geographical isolation was taken as an important factor that may affect the rate of gene flow between parental and hybrid lineages as were habitatspecific ecological performance, high fitness in parental samples from the edge of the group and high fitness in hybrid samples from the center of the group. Cases where the hybrid occupied a habitat in which parents perform poorly were also studied (i.e. the hybrid could colonize empty niche) [5]. Viability and fertility were evaluated independently and genetic isolation was quantified with a set of marked loci [5]. According to this perspective, local extinction of parental species is not a final result, and the chance of hybrid speciation depends on strong habitat-specific ecological selection. Indeed, with weak ecological selection, hybrid speciation almost never occurred, while with strong ecological selection it occurred in approximately 20% of simulation runs [5]. Although the models and conditions in these simulations were obtained from empiric and experimental studies in plants, general statements and requirements could be applied to animals. However, as yet no theoretical or simulated models for this kind of speciation have been proposed for animals.

16

Hybrid speciation genetics A huge amount of theoretical and empirical evidence exists regarding the genetics of hybrid breakdown [2,13-15]. Generally, negative epistasis is involved in hybrid breakdown [14]. The genetic interaction pattern suggested by Bateson [16], Dobzhansky [17] and Muller [18] appears as a principal explanation for hybrid incompatibilities [19]. Dominance is also invoked as an explanation for sex specific sterility or inviability being restricted to the heterogametic sex, a pattern known as Haldane’s rule [20,21]. Also, there are specific cases involving sexual selection and meiotic drift as important contributors to sterility [22,23]. The genetic basis of hybrids of high fitness has not been considered, but is of interest because of its importance in homoploid speciation [24]. Increased hybrid fitness could be reached through different genetic paths. Heterosis, or hybrid vigor, is common in F1 hybrids although their advantage is reduced in later generations. However, new fit hybrids may appear in later generations as a result of favorable epistatic gene interactions or advantageous additive alleles [24]. There are few data supporting the importance of these effects. Burke and Arnold [24] give two reasons for this. Firstly, the occurrence of fit hybrids is a rare event per se and secondly, few studies have dissected specific genotype fitness effects, and many only compare the hybrid statistical performance across filial classes [24]. However, recent experimental and theoretical analysis of hybrid zones sheds light on increased genetic constitution in hybrids [25,26]. There is limited evidence of the epistatic role in plant studies. Favorable heterospecific gene interactions with negative disequilibrium in unlinked markers was produced in synthetic hybrid lineages between Helianthus annus and H. petiolaris [27]. Also, heterospecific cytonuclear interactions that increment hybrid fitness have been observed in crosses between Iris species [28]. Previous examples show that both kinds of interactions have been observed. The importance of epistasis in animal hybrid superiority is until now unexplored. However, genic factors (i.e. Dobzhansky–Muller) are probably important in recombinational speciation. Theoretically, they could act in a different way than chromosomal rearrangements, producing asymmetric effects and high F2 fitness involving complex combinatory scenarios [2].

17

As consequence of the fact that the importance of epistasis in hybrid superiority has not been clearly established, other genetic mechanisms may be involved. Sometimes F2, backcrosses or later-generation hybrid progeny exhibit extreme phenotypes relative to those of either parental forms [29]. The generation of these extreme phenotypes is referred to as transgressive segregation (Table 1.1) [30]. This additive genetic mechanism could be a direct way of diverging from parent species and, if the extreme phenotype also has adaptive value, hybrid speciation might occur [29,31]. Different explanations have been proposed to account for transgressive segregation. Rieseberg et al. [29] summarise them as follows: (1) an elevated mutation rate in hybrids, (2) reduced developmental stability, (3) nonadditivity of allelic effects within a locus or overdominance, (4) nonadditivity of allelic effects between loci or epistasis, (5) the unmasking of rare recessive alleles that are normally heterozygous in the parental taxa, (6) chromosomal number variation and (7) the complementary action of additive alleles that are dispersed between the parental lines. Although each of above may contribute to explain transgressive segregation in particular instances, some seem unlikely to provide a general explanation for this pattern [29]. Genetic studies in tomato and barley provided fairly convincing evidence for expression of rare recessive alleles and complementary gene action in extreme phenotypes [32,33]. Furthermore, Rieseberg’s data compilation [29] gives evidence for transgressive segregation. Opposing effects of different gene alleles on phenotypes within each parental species could have strong positive effects through complementation in hybrid descendence beyond the F1 generation [31]. QTL analysis supports this prediction and gives some interesting insights into the frequency of the phenomenon [34]. Many traits provide evidence that at least one antagonistic locus is needed for transgressive phenotype expression (63.6% from 3252 QTLs evaluated), which supports complementary gene action. Antagonistic QTLs are more common in plant traits than in animal traits. Traits from intraspecific crosses are more likely to have at least one antagonistic QTL than traits from interspecific crosses. Recently, evidence for directional selection as a cause of diversification was observed in QTL analysis acting on interspecific crosses; many cases involved physiological traits with low impact in intraspecific crosses; morphological traits give an understanding of the transgressive action in the last cross type. Although there are several traits proposed as result of hybrid transgression in animals (see Appendix 1 in [29] and supplementary information in [34] for description), sufficient experimental evidence exists for the novel hybrid phenotypes generated by 18

transgressive segregation only in two cases: the cichlid Malawi adaptive radiation and Heliconius butterflies [35-37]. In the first case, skulls of F2 hybrids between two closed related cichlid species showed shape change beyond the range of the parental values, and, in the second, Mendelian pattern colour alleles, with different dominance patterns in the parents, produce a novel combined phenotype in the hybrid [37].

Arguments presented above about the genetics of hybrid speciation suggest that more effort in experimental research on animals is necessary to clarify the importance of the mechanisms that are common in plants, and establish a point of comparison between chromosomal and genic contributions. Theoretical simulations using Fisher’s model of stabilizing selection on multiple traits shows that heterosis arises only when traits are additive, leaving aside dominance and epistasis as forms of achieving hybrid high fitness [26]. These results would act in favor of transgressive segregation as the principal genetic mechanism involved in homoploid speciation.

Table 1.1 Hypothetical example of transgressive segregation. Complementary action of genes with additive effects is shown. Letters in parentheses indicate the species origin of QTL alleles in F2 individuals. (From [34]).

Phenotypic values QTL

Species A (AA genotype)

Species B (BB genotype) Transgressive F2

Transgressive F2

1

+1

-1

+1(AA)

-1(BB)

2

+1

-1

+1(AA)

-1(BB)

3

+1

-1

+1(AA)

-1(BB)

4

-1

+1

+1(BB)

-1(AA)

5

-1

+1

+1(BB)

-1(AA)

T

+1

-1

+5

-5

Frequency of stabilized recombinant lineages and homoploid hybrid speciation Plant hybrid speciation was convincingly documented in ten cases [6,38], most of them with the ecogeographic and postmating isolation hybrid traits established [6,38]. On the other hand, bisexual diploid animals of hybrid origin are less known. Some hypothesized cases exist, where evidence for shared variation with other related species has been documented (Table 1.2) [39]. Nevertheless, nothing is known about the 19

plausibility of the adaptive hybrid traits that produce reproductive isolation with respect to parental lineages (but see below). Also, many of these cases have not been examined with multiple characters nor with experimental approaches for testing reproductive isolation from putative parent species [39]. Both are indispensable conditions for discarding alternative hypotheses and establishing speciation.

Table 1.2 Hypothesized examples of bisexual diploid taxa of hybrid origin. (updated from [39]).

Evidencea

Referenceb

M, A, mt

[40]

Cerion “columna”-like

M, A

[41,42]

Cerion “rubicundum”-like

M

[41]

Andrena montrosensis

M

[43,44]

Papilio joanae

M, mt

[45]

Papilio brevicauda

M, mt

[45]

Heliconius heurippa

M, mt, N, U, C, P

[37,46,47]

Lonicera fly (Rhagoletis sp.)

A, mt, N

[48]

Bosmina coregoni/longispina

M

[49]

Daphnia wankeltae

M, A

[50]

Daphnia cucullata procurva

M, A

[51,52]

Daphnia galeata mendotae

M, A, mt, C, R

[53-55]

Brachymystax sp.

M, A

[56,57]

Catostomus discobolus yarrowi

M, A

[58,59]

Chasmistes brevirostris

M

[60]

Chasmistes liorus mictus

M

[60]

Gila robusta jordani

M, A, mt

[61]

Gila seminuda

M, A, mt

[62]

Luxilus albeolus

M, A, mt

[63,64]

Taxon Mollusks Mercenaria campechiensis texana

Insects

Crustaceans

Fishes

20

Evidencea

Taxon

Referenceb

Mimagoniates microlepis

M

[65]

Pararhynichthys bowersi

M, A

[66-68]

Metriaclima sp. (Makanjila)

M, U

[69]

Neolamprologus marunguensis

mt, N, U

[70]

Xiphophorus clemenciae

mt, N, M, C, E

[71]

M, G, E

[72,73]

M, A, mt

[74]

M, A

[75]

M

[76]

Canis rufus

M, A, mt, N, U

[77,78]

Mus musculus molossinus

M, A, mt

[79]

R

[80]

helminth, trematode Schistosoma sinensium Amphibians Ambystoma tigrinum stebbinsi Reptiles Pseudemoia cryodroma Birds Passer italiae Mammals

Soft corals Alcyonium sp. a

Abbreviations for character sets are: A, allozymes; M, morphology; mt, mtDNA; U, microsatellites; N, nDNA; R, ribosomal DNA (ITS-1); C, genetic crosses; P, morphometrics; R, RFLPs; G, chromosomal differentiation; E, behavioral.

For example, Catostomus discobulus yarrowi was proposed as a hybrid species based on allozymic and morphological variation [59]. However, there is inconclusive evidence about the contribution of C. plebius (a possible parental) to the hybrid genome, and morphological traits could be shared ancestral polymorphisms retained since divergence with C. discobolus and C. plebius [58]. Another case with confusing results is Luxilus albeolus, which was proposed as a hybrid between L. cornutus and L. cerasinus based on the intermediacy of morphological and allozymic characters [64]. Revision of the data added to mtDNA analysis show that the contribution of the allozymic alleles of L. cerasinus to L albeolus is also shared in the allopatric L. cornutus population. This result diminishes the importance of hybrid origin and opens the possibility of allozyme convergence or shared ancestral polymorphisms as explanations for the mixed allozymic pattern [63]. Clearer evidence of introgressive hybridization comes from 21

allozymes, mtDNA and morphology in the case of minnow fishes [62]. Gila seminuda was intermediate to G. elegans and G. robusta in phylogenetic analysis of morphology and allozymic characters, but had mtDNA haplotypes which were only related to G. elegans. Studies with the same markers including other species of the Gila complex suggested that the phylogenetic conflict was a product of past episodic introgression as a consequence of variation in the continuity of drainage in the Colorado basin [61]. This pattern was generated by tectonic activity and aridification cycles that facilitate sympatric introgression and allopatric differentiation [39]. In some cases the introgression is old, yet its effects persist today. Fossil records indicate temporal, stable and distinct populations of possible hybrid origin in the genus Cerion [41]. Invasion of C. dimidiatum 13,000 years ago, in the Bahamas produced hybrid morphologies between this species and the local C. columna. Evaluation of the present population of snails was concordant with a long persistence of hybrid effects beyond C. dimidiatum extinction [42]. However, whether this long process of introgression produces different isolated lineages is an open question [39]. Not all the cases of hybrid origin reveal intermediate morphologies. This precludes intuitive assignment of a phenotype as recombinant [39]. Also, the demographic history and the cycles of reproductive systems (sexual and asexual) may prevent the presence of a clear phylogenetic signal, making the taxa identification controversial [55].

mtDNA and Allozymes suggest that Daphnia mendotae is a stabilized introgressant involving D. rosea and D. galeata [55]. Joint analysis (allozyme, mtDNA, RFLPs and morphology) clearly indicated widespread hybridization between North American D. rosea and D. mendotae [53]. The resulting hybrids are often more abundant than the parent taxa in a lake, providing good conditions for backcrossing. Proposed introgressants are distributed throughout the North American range of D. mendotae, with nuclear gene flow unaccompanied by mtDNA gene flow. Also, lack of mtDNA sequence divergence between D. mendotae and D. galeata together with their great divergence in allozyme loci (resulting in D. mendotae alleles related to D. rosea) supports a nuclear introgression hypothesis and concordance with other allozymic patterns precluding alternate symplesiomorphy or convergence hypotheses [54,55,81,82]. The adaptive significance of these introgressed genes is unknown, but taking into account the fact that D. mendotae is the most morphologically variable and ecologically successful Daphnia species in temperate North American lakes, it is 22

probable that some of the advantages stated above are due to the introgression phenomenon [55]. Information about reproductive isolation in this crustacean group is scarce. Some evidence of postzygotic mechanisms was found in anthropogenic introduction of Eurasian D. galeata into the Laurentian Great Lakes [54]. However, nothing is known about the nature and strength of this isolation. In spite of the intricate ponds and lakes system that could account for allopatry speciation, it is possible that the success of the introgressant D. mendotae may be found to be due to refugial hybridization in Beringia when D. galeata and D. rosea were in contact and the new homoploid lineage took advantage of empty niche space [55,81].

Chromosomal reorganization is fundamental in hybrid speciation models [5,6]. Evidence about its relevance in facilitating this process was observed in the homoploid plant hybrid species [6]. A famous case, in which chromosomal rearrangements were crucial in the hybrid lineage reproductive isolation involves wild sunflowers. Helianthus anomalus was heterozygous for ten chromosomal differences between H. annuus and H. petiolaris. The same ten were responsible for hybrid breakdown in synthetic experiments with both species [10]. In animals, until now the importance of this kind of genome reorganization in recombinant speciation is unexplored. However, Schistosoma sinensium has a mosaic chromosome 2 between the African and Asian classical forms [73]. Although S. sinensium belongs to the Asian group in terms of telomere location, the C-banding pattern of chromosome 2 of this species shares a similarity with the African species [73]. It is important to mention that Asian and African forms of chromosome 2 diverge as a consequence of complex reorganization of segments through pericentric and paracentric inversions that change the chromosome structure from Asian to African variants [73]. Integration of the African type of chromosome 2 into the Asian genome could have been caused through hybrid recombination between African and Asian ancestral species. Interestingly, S. sinensium (W) chromosome had the same structural change [73]. The plausibility of recombinational speciation is complemented by evidence from molecular genetic analysis that shows specific deletions in rDNA spacer region, harbored by S. mansoni and S. haematobium (African species). Other characters including morphology, specificity of snail host and geographical distribution revealed that S. sinensium has a mixed manifestation of S. mansoni (38%) and S. japonicum (62%) [72]. Breeding experiments are necessary in 23

this case to establish the importance of the hybrid rearrangements in reproductive isolation.

Habitat perturbation by humans appears to have increased the incidence of hybridization. Sometimes introgression sufficient to change the genetic constitution of taxa occurs [39]. However, human and natural effects in hybrid formation are difficult to separate [39]. The case of red wolf (Canis rufus) reflects this situation [77,78]. In this century, the red wolf population declined dramatically throughout its geographic range in the southeastern United States, leading to extinction in particular places [77]. Rarity of Canis rufus, led to hybridization of this taxa and C. latrans with phenotypic consequences [83]. Mitochondrial DNA analysis shows a complex pattern of haplotypes: some wolf-like, others coyote-like, and in some cases shared between coyotes and wolves [78]. This result was explained by wild hybridization between the three taxa that shared geographical range in the past [78]. On the other hand, morphological studies have suggested that Canis rufus is a distinct species with an intermediate phenotype and is the ancestor of the gray wolf (C. lupus) and the coyote (C. lantrans) [84]. If this is true, a unique mtDNA and nuclear alleles should exist in red wolves, forming a different basal clade. However, an intermediate phenotype and the complex relationship for mtDNA described above are consistent with an origin by hybridization [77,85]. Probably, the red wolf may represent a phenotype resulting from a several-hundred-year period of hybridization between coyotes and wolves in the south-central United States, which began with habitat changes associated with the arrival of settlers circa 1700 [77]. Subsequently, the extermination of gray wolves, increased the frequency of hybridization of the red wolf with coyote populations, diminishing the traits that are characteristic of its phenotype [77]. Microsatellite analysis is consistent with the historic hybrid origin of the red wolf between coyotes and gray wolves, followed by recent hybridization with coyotes as a consequence of the rarity of gray wolves [77]. Alleles for 10 loci were shared between red wolves and coyotes; diagnostic alleles were observed for coyotes and gray wolves but not for red wolves [77]. The lack of specific diagnostic alleles in red wolves may be a product of their rapid loss in small populations near to extinction [77]. However, high allele diversity and heterozygosis of captive red wolves indicate an opposite tendency, reflecting no variation in effective size [77]. The lack of substructure in the populations of red wolves, the non genetic subdivision observed in coyotes and local divergence of 24

gray wolves are congruent with the hypothesis of divergence and hybridization in the Ice Age Refugia as against a more recent origin directed by human agricultural activity [77].

The two principal problems of hybrid speciation: reproductive isolation and high fitness (see above), may be overcome in the case of parasitic animals [48]. The host-specific life history of these kinds of organisms would facilitate speciation in the face of gene flow and offer a robust solution to ecological problems [86]. Competition between parents and the hybrid entity could be avoided as a result of hybrid lineage shift to a new host thus obtaining an independent resource and instantaneous reproductive isolation, especially if the mating occurs on the host [86]. This kind of recombinational speciation was recently documented in the honeysuckle-Rhagoletis system [48]. Nonnative, brushy Lonicera spp. in the northeastern United States serves as a new niche for a native Ragoletis colonist in its range distribution [87,88]. Unique high-frequency alleles permit the assignation of species of these insects to a specific niche (host) [88]. Notably, the flies that infested Lonicera spp showed a unique mixture of speciesspecific allozyme alleles that indicate a possible hybrid origin between the blueberry maggot (R. mendax) and the snowberry maggot (R. zephyria) [48]. Sequence data from five of the six linkage groups with intermediate frequencies in the R. pomonella and mitochondrial genome confirm this hypothesis [48]. In this particular case other possibilities regarding the origin of the Lonicera fly as a result of incomplete linage sorting, speciation from only one parent, or a hybrid zone maintained by immigration were tested by genetic and simulated probes. In all the comparisons hybridization hypothesis was the most probable explanation for observed mixed genetic constitution [48]. The shift to honeysuckle probably freed the hybrid origin Lonicera fly from parental species competition, facilitating reproductive isolation through epistatic interaction between mate and host choice, as was observed in other Ragoletis species [86]. However, studies on host choice, mate choice and host-specific fitness between the putative parents and the Lonicera fly are needed in order to validate the ecological assumptions and establish that the hybrid trait was causal in speciation [48].

Hybridization may have an important role in adaptive radiations, both in their origin and in the generation of functional diversity after radiation (syngameon hypothesis) [89]. These two different processes may result in species of hybrid origin [89]. 25

Although they do not seem to undergo recombinational speciation, Darwin’s finches provide strong evidence about the influence of hybridization in evolution and diversification [90,91]. Studies of the finches over several years were possible due to geographical isolation in the Galapagos islands, permitting a knowledge of the long term effects of hybridization and other evolutionary forces [39,92]. Interspecific hybrids between Geospiza fortis, G. scandens and G. fuliginosa on Daphne Major, involving F1 and backcross descendants, studied over a four-year period had a higher average fitness than parental types [39,90]. Climatic fluctuations between drought and wet seasons not only account for a selection regime in Daphne Major populations, but also facilitate hybridization [39,90]. Specifically, introduction of genetic variation by hybridization could reduce inbreeding depression during periods of small population size [90]. Hybrid fitness probably undergoes fluctuations with normal variation in climatic conditions, resulting in a balance between hybridization and selection. If novel combinations of high fitness traits occurs in the hybrids, selection could facilitate an adaptive peak shift and speciation [90]. Cases where adaptive radiations culminate in homoploid speciation events were observed in cichlids [69,70]. Malawi, Tanganyika and Victoria lakes are hot spots of cichlids diversity[70]. A lot of endemic cichlid species have arisen in the lakes since less than1million years ago, and despite intensive investigation, the modes and mechanisms producing this diversity are still largely unknown [69,93]. Prezygotic isolation barriers could be important in the initial process of cichlid species diversification [94]. Sexual selection driven by male colour patterns has strong effects in maintain species integrity [69,95,96]. In this scheme low opportunity for hybridization could arise. However, the existence of high rates of polymorphism in mtDNA [97] and microsatellite loci [98], may be evidence of a high incidence of hybridization in these fishes [69]. Events of hybridization due to distortion in the mating signal as consequence of water turbidity [99], fauna translocation [100], lake level fluctuations leading to secondary contact and captivity has been observed [101,102]. Two cases in which speciation by hybridization could contribute to cichlid diversity have been documented [69,70]. In the Tanganyika lamprologines, results from a phylogenetic study show two divergent mitochondrial lineages among Neolamprologus marunguensis individuals [70]. One of these is related to Helianthus-clade and the other to olivaceous-clade. On the other hand all marunguensis individuals were placed in the olivaceous-clade in a TmoM27 nuclear locus [70]. Five microsatellite loci show mosaic pattern alleles in some individuals and unique alleles in others [70]. 26

The genetic divergence in the mtDNA haplotypes, 12.4% in the control region and 5.2% in the cytochrome b, suggest that Neolamprologus marunguensis is a species produced in an ancient hybridization event [70]. This new entity would have arisen in secondary contact between the divergent parental lineages as a consequence of the drop in lake water level and the subsequent increase in level opening access to a new habitat different from that of the parental species, enhancing population hybrid isolation [70]. The hybrid formation in secondary contact is plausible because it is known that environmental fluctuations in the level of the lakes have the potential to reverse reproductive isolation in cichlids [99]. In the Metriaclima zebra species complex in Lake Malawi there are multiple populations that differ in dorsal fin colouration with mutually exclusive distributions [69]. M. zebra habitats have sandy bottoms, deep water and rocky areas, that apparently preclude contact zones between different colour forms, even when they inhabit adjacent areas [103]. No postzygotic isolation between populations was detected in aquarium experiments and prezygotic isolation has not been tested up until the present [69]. The absence of postzygotic barriers suggest that periods of sympatry could potentially lead to hybridization between “red top” (M. thapsinogen) and “blue with bars” (M. zebra) dorsal fin colouration forms. A population of Metriaclima exhibiting mosaic phenotypes intermediate between the two different colour populations mentioned above is located in Makanjila, a place situated midway between Eccles Reef and Chiofu Bay, locations corresponding to the extreme limits of the blue and red fin colours [69]. A simulated test of hybrid F1 population and factorial correspondence analysis on genotypes obtained from four microsatellites suggest a population differentiation between M. thapsinogen and M. zebra with intermediate assignment of Makanjila individuals [69]. These and another genetic inferences indicate that this population (Makanjila) has arisen as result of extensive hybridization between M. thapsinogen and M. zebra or other Metriaclima species [69]. In spite of the persistence of high frequency unique alleles in the Makanjila species and lack of information about reproductive isolation in this taxon [69], this case and the Neolamprologus marunguensis suggest that hybridization among cichlids may be more pervasive than currently appreciated and may be facilitated by the environmental dynamics of the lake as explained before.

27

No evidence about the evolution of reproductive isolation is mentioned in the homoploid hybrid speciation cases proposed above. Until now only two examples deal with this problem, one in fishes and the other in butterflies [37,71]. In the first case, Xiphophorus clemenciae (a swordtail species) is proposed as a hybrid species between X. maculatus and X. hellerii. This is evidenced by the discordance between Xiphophorus clemenciae (swordless species) mtDNA haplotype and swordtail nuclear genome (seven gene regions joining in a unique analysis). Also, X. maculatus times X. hellerii genetic crosses resembled the X. clemenciae median swordtail and geographical distribution is coincident with a probable ancient hybridization event [71]. The authors explain that divergence of the hybrid form is mediated by a preexistent female sensorial bias to mate with swordtail phenotypes so that hybrid speciation could be attained by sexual selection [71]. However, in the mate choice experiments no discrimination between large and median tails was made, with the result that an opportunity is created for gene flow between the swordtail parent and the hybrid, which in turn could dilute the speciation process.

The second case is Heliconius heurippa: a neotropical butterfly of possible hybrid origin between H. cydno and H. melpomene [104]. Evidence about its specific status comes from a high divergence with respect to either of the possible H. melpomene or H. cydno geographical races in 12 microsatellites loci [37]. Heliconius cydno is black with white and yellow marks and mimics H. sapho pattern, while H. melpomene is black with red, yellow and orange marks and mimics H. erato pattern [105]. The hindwing of H. heurippa is indistinguishable from that of sympatric H. m. melpomene, while the yellow band on its forewing is similar to that of parapatric H. cydno cordula [46]. Heliconius cydno and H. melpomene exhibit strong positive assortative mating based on their wing colour patterns [105], and also differ in habitat use [106] and host plant preference [107], but hybrids do occur at low frequency in the wild [108]. H. heurippa is ecologically most similar to H. cydno, which it replaces geographically in the eastern Andes of Colombia [46]. Genetic crosses between H. m. melpomene and H. c. cordula show that a synthetic hybrid pattern virtually identical to wild H. heurippa is obtained from two generations of introgressive hybridization to H. c. cordula [37]. Also, genetic data based in mtDNA, one sex linked locus and four nuclear loci, although not ruling out a considerable incidence of gene flow, shows a phylogenetic pattern consistent with a hybrid origin and introgressive history [47]. Interspecific crosses show intrinsic 28

postzygotic isolation between H. melpomene and H. heurippa but complete fertility between the latter species and H. cydno [46]. Mate choice, courting frequency and pattern colour attraction experiments suggest a strong conspecific preference by both hybrid colour pattern elements (experimental removed red or yellow colour elements, produce a half courtship diminution). Hence, the combination of strong assortative mating, geographic isolation from H. cydno and both intrinsic and ecological postzygotic isolation, has led to the speciation of H. heurippa [37] (See chapters II, III, IV for details).

Methods and difficulties in reconstructing patterns of reticulate evolution

Dowling and Secor proposed three principal reasons why recombinational speciation in animals has been poorly documented [39]. In the first place, when divergence levels between the hybrid and parental lineages is low, horizontal gene transfer will be difficult to discriminate from ancestral polymorphisms. Secondly, if one of the parents is extinct, problems with tracing and recreation of the possible route of hybridization emerge. Finally, the hybrid taxa could develop its own traits with the passing of time, erase the hybrid origin signal, and avoid the detection of the steps of hybridization history. For these reasons the same authors propose that in order to document hybrid speciation a large set of characters (i.e. morphological, molecular, ecological and behavioral) should be evaluated [39]. These factors and the fact that evolutionary history at the level of species is not necessarily a bifurcating tree, but rather a net of alleles and genotypes in which lineages will experience reticulate events as a consequence of homoplasy and recombination (and hybridization itself), could complicate homoploid speciation detection [109,110]. Here, I focus on the capability of molecular methodologies for detecting hybrid speciation, because commonly[110].

At sequence level analysis the reconstruction of hybrid speciation in the form of phylogenetic trees requires special methods that are, as yet, largely unavailable [110]. This is because extensive testing over alternative reconstructions on large simulated phylogenies are necessary to assess the support for particular hypotheses [110]. Another inconvenience is that phylogenetic algorithms do not take into account the effect of population genetic process that would shape species relationships and that could be 29

estimated through coalescent methods [111]. For this reason a more realistic approach for inferring species level history would be one which includes reticulate patterns and in which allele ancestry between species may be traced, genetic process measured and demographic parameters estimated [110].

Linder and Rieseberg [110] mentioned three lines of classical evidence that might be employed to detect and reconstruct hybrid speciation. First, in the absence of other processes that might produce topological incongruence in phylogenetic trees, hybrid speciation could be detected by simply looking at discordant trees through separate data analysis [112]. The second way to detect hybridization relies on combining DNA sequences from multiple independent loci into a single analysis looking for phylogenetic signals that suggest sets of different histories as would be observed using the split decomposition method [113]. A third method, consists of searching the genome for zones with linkage disequilibrium i.e. hybrid species probably having combinations of regions with tightly linked markers coming from each parent displaying gametic correlations [114]. Due to the influence of stochastic, methodological and genetic population history in affecting phylogenetic reconstructions, a powerful test for hybrid speciation would be combining phylogenetic incongruence with linkage disequilibrium. The reason for joining both analyses is that separate trees of individual DNA regions or loci that are part of a tightly linked set of loci should have the topology of only one side of the hybridization. These reconstructions would be topologically incongruent with those obtained from loci belonging to the other parent [110].

Modern biosystematics tools have been developed in order to represent with more accuracy the evolutionary relationships at species level [109]. These methods account for reticulation patterns and together are grouped as part of the network concept. A phylogenetic network is a rooted DAG (directed acyclic graph, see Rieseberg and Linder for details [110]) in which the internal nodes are partitioned into tree nodes and network nodes. A tree node has one ancestral branch and two or more descendant branches. A network node has two ancestral branches and only one descendant branch. In the same way, branches are partitioned in tree branches and network branches in accordance with their node origin. On a flat surface tree branches are vertical or angled and network branches are horizontal (Figure 1.2). 30

Figure 1.2 Example of phylogenetic network with a single hybrid species (B). Depicted as figure 1 in [110].

In this model, DNA sequences are assumed to evolve only on the tree branches, although a small amount of change could theoretically occur in network branches. Constructed in the same way as classical trees, network reconstruction and detection have the follow components: (1) algorithms for reconstructing networks, (2) software capable of generating simulated network patterns for some specific models and with the capacity to evolve DNA sequences over the simulated data, (3) statistical or other supporting methods for the reconstructions [109,110]. In spite of a continuous improvement in the statistical and supporting methodologies, at the moment none of the three classical phylogenetic possibilities nor the network approach for detecting and reconstructing hybrid speciation has been well studied in simulations [110]. For this reason the last part of this chapter is dedicated to a general description of network algorithm reconstructions, the effects of genetic process on these methods and the use of other true simulation tools that would be useful in documenting hybrid speciation.

Performance studies that assess trees and network reconstruction methods have to be able to measure the error (distance) between the phylogeny of a group and the estimate of it. In metric concept, error measures could be symmetric, that is: the counts of false positives (branches in the reconstruction that are not in the model) and false negatives (branches that are in the model but not in the reconstruction) should be equal to zero [110].

31

The use of multiple independent genes together in a combined data analysis in hybrid speciation detection has been low (but see [115-119]), because this method is not completely satisfactory. The reason is the unacceptable number of false positives that are produced in the reconstruction of this kind of analysis. Combining data itself and the lack of biological rationalization in these methods could be the root of the problems encountered [110]. The algorithms in this kind of analysis are grouped in three types: (1) a tree is built and then network branches are added to turn it into a network, optimizing some cost criteria [120-122], (2) a group of trees is built (some times with different data subsets) and the software tries to reconcile them. Conflicts produced by non-reconciliation might be explained by reticulation events. This is the basis on which median networks work [117,123]. Also, the same principle applies to molecularvariance parsimony procedures [124], (3) before doing any reconstruction, incompatibilities in the data are characterized to provide a group of possible resolutions through reticulation. Splits-based methods rely on these last kinds of algorithms [113,116,125]. The result is not a specific network solution, but presents all consistent choices [110].

Promising, but nascent, developmental state phylogenetic incongruence methods could be an alternative way of achieving hybrid speciation evidence. SpNet algorithm, developed by Nakheleh et al. [126] takes into account the phylogenetic signal, independently for each locus network and compares it with a model of reticulation, with congruence measures. An important condition for the correct functioning of the SpNet algorithm is that each evolutionary hybrid event should be independent of whatever other event. Initial simulation studies suggest that this method has lower false positives in comparison with combined data approach [126].

The network algorithms could also be used independently for each locus (as NJ trees are), by doing a specific locus topology and polymorphism interpretation for each genetic marker or evaluating the joint pattern concordance with the hybrid speciation scenario. However, it is important to know that for this and the other methods mentioned above, reticulate patterns could be observed at levels below species (chromosomes and genomes) and in some cases these other kinds of reticulations would be mimetic of real hybrid speciation reticulation patterns. Also, population genetic processes as lineage sorting, independent gene duplication and random loss in multiple 32

genes can produce incongruent tree reconstructions that could be interpreted as hybrid speciation [127].

In particular, in recently diverging species coalescence time of alleles at a single locus is longer than the formation time in real species: partly due to differences in effective population sizes between loci (generally, nuclear loci Ne is quadruple that of organelle genes). As a consequence, stochastic sorting process could be confused with the real species relationships (gene tree/species tree problem [128]). If alleles for different genes assort differently during speciation, then incongruent trees will be reconstructed, reflecting a similar pattern as that expected in a hybrid speciation event [110].

The misinterpretation is severe in the case of duplicated genes, where lineage sorting is likely to occur between alleles of orthologous and paralogous sequences that must be distinguished. Because duplicated genes are subject to random loss in different species as consequence of different molecular factors (for example the random production of pseudogenes), the gene tree/species tree problem is presented in a similar way to the lineage sorting affect in one locus [110]. To avoid the inconvenience of incorrect phylogenetic interpretations the greatest number of orthologous sequences should be included in the analysis. However, genetic drift and natural selection together with other genetic processes may act on the genome, facilitating orthologous gene loss and increasing the possibility of working with paralogous sequences [110]. The duplication nature of a gene is still under active investigation. For this reason, hybrid studies are not safe from incongruence noise as a consequence of paralogous gene duplications [127,129].

Another important force affecting the correct estimation of phylogeny is recombination. This force commonly acts at the population level and combines genome evolutionary history between lineages [130]. Due to the interchange of genomic regions in the meiotic state, recombination might be considered as a type of reticulation [110]. Moreover, sexual reproduction in itself provides an initial possibility of mixing the parents genome before the real meiotic recombination. Both meiotic and sexual recombination can obscure hybrid speciation patterns principally through two effects [110]. Firstly, different alleles at particular loci are inherited in distinct lineages as a 33

consequence of recombination coupled with genetic drift and natural selection. The net effect is the same as lineage sorting, leading to incongruent reconstructions through different loci. Secondly, the assumption that individual sequence represents a single lineage history when running the analyses can produce false results. In fact, in the presence of recombination a single sequence might represent multiple histories. Many investigations have tried to overcome and quantify recombination rates [131]. However, no work that attempts to differentiate between meiotic recombination and hybrid speciation exists [110].

New molecular attempts in hybrid origin detection

Bioinformatics with an analytical view within the evolutionary field can resolve or reduce the problematic effects discussed above. In particular, advances in simulation methods promote the development of divergence population genetics (DPG) a discipline [132] in which the idea is to estimate demographic population history and evaluate the relative importance of gene flow vs. ancestral polymorphisms in the speciation process through summary statistics and interpretation of variation in multilocus data sets [132,133]. These methods rely on coalescence theory and provide the possibility of testing different kinds of results from different evolutionary situations in concordance with parameter estimation [134]. These kinds of analyses were used to reveal the possible processes involved in speciation and the history of different animal groups with the emphasis on recently originated taxa prone to undergo hybridization [135-143]. For a summary of these methods see Table A2 - Appendix 2 in [144] and the references therein. Although, DPG approaches are valuable in addressing most demographic speciation questions, the coalescence framework studied in them is generally the situation of new populations or species that originated backward in time (T1) from a common ancestral population and that may or may not interchange genes (Figure 1.3) and does not incorporate the possibility of hybrid speciation.

34

Figure 1.3 Speciation in coalescence representation. Specifically the diagram corresponds to the Isolation-Migration (IM) model. This model assumes that an ancestral population of constant size and population mutation parameter TA separated in two populations at time T. Each descendant population has its own mutation parameter, T1 and T2, respectively. Also migration occurs between the two populations at rates M1 and M2. This is a general model that includes, as special case, the isolation-without-migration model (by set M1=M2=0; from [133]).

There is therefore a need to establish new coalescent models for the hybrid origin of a third lineage (Figure 1.4). Until now published software deals with interspecies hybridizations and establishes hybrid genomic composition with an assignment test [145], which is not my specific interest here. The real issue is to trace backward the history of shared alleles of possible hybrid species and distinguish this from the signal of recent gene flow. In particular it is possible to get gene flow information through review of the polymorphism pattern, and see the distributions of specific summary statistics in several simulations. It is known that shared polymorphisms increase while fixed differences and within species polymorphisms decrease when gene flow is present [139]. In the case of H. heurippa we have used the ratio of the number of sites segregating in both species 1 and species 2 (possible hybrid species), but not in the species 3, to the number of sites segregating only in species 1 or species 2 [47].

35

Figure 1.4 Hybrid origin coalescence scenario. Pop 1, Pop 2 and Pop 3 are the parental (1,3) and hybrid (2) effective population sizes, A is the effective ancestral population size, T1,T2 represent coalescence times from present until Pop 2 and from Pop 2 until A, respectively. Red arrows show coalescence direction. Taken from [47].

Unfortunately the total likelihood surface was flat throughout different scenarios, probably due to the effect of heterogeneity in migration rates between loci. Nonetheless this hypoethsis testing approach to the study of hybrid speciation, in combination with linkage mapping of the hybrid traits and estimates of linkage disequilibrium will provide a battery of methods to test for hybrid homoploid speciation.

Reproductive isolation and hybrid speciation Beyond detecting possible hybrid genetic species in animal taxa, it is important to establish that the hybrid entities are real species, especially when their origin involves sympatric or parapatric isolation. For this reason it is necessary to establish and quantify the degree of reproductive isolation between the putative parental forms and the hybrid. 36

In cases in which the genetic isolation is not instantaneous (i.e. chromosomal model, see above) general ecological changes are suggested as a main factor involved in reproductive isolation. In particular, Gross and Rieseberg [146] proposed 6 questions to verify the role of ecological divergence in homoploid hybrid speciation. When these questions were answered, results suggest that, at least in plants, ecology could play a major role [146]. The evidence comes from transplant experiments between parents and hybrid habitats, pollinator differences, temporal isolation and extreme phenotypes recreated in laboratory through genetic crosses [146]. Also, evidence for the strength of divergent natural selection in the synthetic hybrids grown in different habitats has been measured, showing that selection could avoid hybrid dilution by gene flow [146]. However, evidence for the importance of divergent ecology in animal hybrid speciation is scarce.

Ecological information is commonly limited to brief descriptions of the habitat that a hybrid species occupies but with no quantitative or experimental evaluation. For example, in the fish genus Gila, the parental species are found sympatrically in the large Colorado River, whereas the hybrid species is found only in the Virgin River, a moderately sized tributary of the Colorado [62]. In the same way Dapnia mendotae occurs in slightly warmer water and shows less vertical migration than one if its sympatric progenitors, D. dentifera [55]. In the soft coral genus Alcyonium, the proposed hybrid species is completely allopatric with its progenitors [80]. But all of these cases do not really show how the different distribution accounts for maintaining the integrity of the hybrid species. Shift to a new resource has been presented as an opportunity to drive hybrid speciation [48]. The idea is that new traits obtained by hybridization could facilitate the colonization of a new host. However, until now only the Lonicera fly met this scenario [48], but there is no evidence about the mating preferences or other reproductive isolation mechanisms between different host races. Until now only two cases have reproductive isolation measures: Heliconius heurippa and Xiphophorus clemenciae (see above and chapter IV) [37,71]. In both, genetic crosses were made to recreate the natural hybrid phenotype. Mating choice experiments between hybrid species and the parents show high discrimination and preference between conspecifics and the homoploid species mediated by the hybrid trait. However, in the Xiphophorus case the preference is for swordtail length and it is not clear if the 37

median size of X. clemenciae maintains this species different with regard to X. hellerii, the possible long swordtail parent.

Methods for establishing species discreteness could help to obtain evidence for reproductive isolation in hybrid taxa. The idea is to use statistical tools to look for clustering. The discrimination power of the tests was proved in sympatric species differentiation [147,148] and evaluation of the gene flow effect in collapsed reproductive barriers [142]. The method makes more sense when various traits: morphological, behavioral, reproductive, and molecular are included [2]. In spite of the advantages of these multivariate techniques, they have been poorly employed in documenting homoploid species. In plants, multivariate methods were used to compare recreated hybrid traits in laboratory crosses and evaluate if these are different from the natural hybrids and if synthetic individuals also have a different genetic cluster with respect to the parents [149]. A large sample of morphological, life-history and physiological traits were included in the comparison of the three Helianthus ancient hybrid species [149]. In animals morphological traits were used in a multivariate space to account for intermediate characters in Gila semimuda, and Metriaclima sp hybrid origin was tested by investigating the spatial distribution of microsatellite frequencies.

Conclusion The difference between the lack of a good understanding about the relevance of hybrid speciation in animals and the well-established importance of this mechanism in plant diversification is in part due to difficult diagnosis and poor knowledge about how the hybrid obtains reproductive isolation from the parents in the former group. In plants QTL techniques, morphometric designs, synthetic hybrid tests and habitat transplants are well developed and provide a broad picture of homoploid speciation. In contrast, the majority of animal hybrid cases rely on mtDNA and nDNA discordant phylogenies along with morphologic mixed traits. The lack of a good number of individuals by brood and clarity in the hybrid trait adaptive value in animals limits the QTL analysis approach broadly used in plants to detect homoploid speciation. On the other hand, chromosomal rearrangements and ecological versatility facilitate reproductive isolation in hybrid plants while in animals RI is more difficult to achieve. The available 38

ecological data on animal hybrid performance is poor, except for those limited to unimodal and bimodal hybrid zones. The same is true for the measurement of reproductive isolation between animal homoploid hybrids and their parents, and the means by which the hybrid trait promotes speciation. As I mention above, only two cases really recreate the hybrid genotype in the laboratory and test for RI, showing an importance of hybrid traits in species mating. Another major problem in testing hybrid origin lies in distinguishing between shared variation due to recent gene flow and that derived from a hybrid origin. Although a clear resolution of this issue is unlikely, new tools using coalescent methods to estimate gene flow show promise. For all the reasons enumerated above, documenting a convincing case for hybrid speciation in animals requires information on multiple markers, synthetic recreation of hybrid traits, measurement of the importance of hybrid traits in ecological or sexual divergence, measurement of the adaptive value of hybrid traits and diagnosis of the nature of the shared variation with the parents. In this view, the role of hybridization as a mechanism that promotes speciation in animals is plausible, but the proposed cases that have poor evidence at present studied in more detail and those that meet most of the criteria indicated above must be corroborated.

References 1. Grant V (1981) Plant Speciation. New York: Columbia Univ. Press. 2. Coyne JA, Orr HA (2004) Speciation. Sunderland, MA: Sinauer Associates. 3. Rieseberg LH (1991) Homoploid reticulate evolution in Helianthus (Asteraceae): evidence from ribosomal genes. Am J Bot 78: 1218-1237. 4. Turelli M, Barton N, Coyne J (2001) Theory and speciation. Trends Ecol Evol 16: 330-343. 5. Buerkle CAM, R. J. Asmussen, M. A. and Rieseberg, L. H. (2000) The likehood of homoploid hybrid speciation. Heredity 84: 441-451. 6. Rieseberg LH (1997) Hybrid origins of plant species. Ann Rev Ecol Syst 28: 359389. 7. Anderson E (1948) Hybridization of the habitat. Evolution 2: 1-9. 8. Templeton AR (1981) Mechanisms of speciation-- a population genetic approach. Ann Rev Ecol Syst 12: 23-48. 9. Shaw DD, Wilkinson P, Coates DJ (1983) Increased chromosomal mutation rate after hybridization between two subspecies of grasshoppers. Science 220: 1165-1167. 10. Rieseberg LH, Van Fossen C, Desrochers AM (1995) Hybrid speciation accompanied by genomic reorganization in wild sunflowers. Nature 375: 313316. 11. Barton NH, Hewitt GM (1985) Analysis of hybrid zones. Ann Rev Ecol Syst 16: 113-148. 39

12. McCarthy EM, Asmussen MA, Anderson wW (1995) A theoretical assessment of recombinational speciation. Heredity 74: 502-509. 13. Johnson N (2000) Speciation: Dobzhanzky-Muller incompatibilities, dominance and gen interactions. Trends Ecol Evol 15: 480-481. 14. Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation. Genetics 154: 1663-1679. 15. Orr HA, Turelli M (2001) The evolution of postzygotic isolation: Accumulating Dobzhansky-Muller incompatibilities. Evolution 55: 1085-1094. 16. Bateson W (1909) Heredity and variation in modern lights.; Seward A, editor. Cambridge: Cambridge Univ. Press. 17. Dobzhansky TH (1936) Studies on hybrid sterility. II. Localisation of sterility factors in Drosophila pseudoobscura hybrids. Genetics 21: 113-135. 18. Muller HJ (1942) Isolating mechanisms, evolution and temperature. Biol Symp 6: 71-125. 19. Orr HA (1996) Dobzhansky, Bateson, and the genetics of speciation. Genetics 144: 1331-1335. 20. Haldane JBS (1922) Sex ratio and unisexual sterility in hybrid animals. J Genet 12: 101-109. 21. Turelli M, Orr HA (1995) The Dominance Theory of Haldanes Rule. Genetics 140: 389-402. 22. Wu C-I, Johnson NA, Palopoli MF (1996) Haldane's rule and its legacy: why are there so many sterile males? Trends Ecol Evol 11: 281-284. 23. Tao Y, Hartl DL (2003) Genetic dissection of hybrid incompatibilities between Drosophila simulans and D. mauritiana.III. Heterogeneous accumulation of hybrid incompatibilities, degree of dominance, and implications for Haldane's rule. Evolution 57: 2580-2598. 24. Burke JM, Arnold ML (2001) Genetics and the Fitness of Hybrids. Ann Rev Genet 35: 31-32. 25. Harrison RG (1990) Hybrid zones: windows on evolutionary process. In: Futuyma D, Antonovics J, editors. Oxford Surveys in Evolutionary Biology. Oxford: Oxford University Press. pp. 69-128. 26. Barton NH (2001) The role of hybridization in evolution. Molec Ecol 10: 551-568. 27. Rieseberg LH, Sinervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene interactions in hybrid speciation: evidence from ancient and experimental hybrids. Science 272: 741-745. 28. Burke JM, Carney SE, Arnold ML (1998) Hybrid fitness in the Louisiana irises: Analysis of parental and F1 performance. Evolution 52: 37-43. 29. Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and speciation. Heredity 83: 363-372. 30. DeVicente MC, Tanksley SD (1993) QTL analysis of transgressive segregation in an interespecific tomato cross. Genetics 134: 585-596. 31. Bell M, Travis MP (2005) Hybridization, transgressive segregation, genetic covariation, and adaptative radiation. Trends Ecol Evol 20: 358-361. 32. Rick CM, Smith PG (1953) Novel variation in tomato species hybrids. Am Nat 87: 359-375. 33. Vega U, Frey KJ (1980) Transgressive segregation in inter- and intraspecific crosses of barley. Euphytica 29: 585-694. 34. Rieseberg LH, Widmer A, Arntz AM, Burke JM (2003) The genetic architecture necessary for transgressive segregation is common in both natural and domesticated populations. Philos Trans Roy Soc Lond B 358: 1141-1147. 40

35. Albertson RC, Kocher TD (2005) Genetic architecture sets limits on transgressive segregation in hybrid cichlid fishes. Evolution 59: 686-690. 36. Albertson RC, Streelman JT, Kocher TD (2003) Directional selection has shaped the oral jaws of Lake Malawi cichlid fishes. Proc Natl Acad Sci, USA 100: 52525257. 37. Mavárez J, Salazar C, Bermingham E, Salcedo C, Jiggins CD, et al. (2006) Speciation by hybridization in Heliconius butterflies. Nature 441:868-871. 38. Howarth DG, Baum DA (2005) Genealogical evidence of homoploid hybrid speciation in an adaptative radiation of Scaevola (Goodeniaceae) in the Hawaiian islands. Evolution 59: 948-961. 39. Dowling TE, Secor CL (1997) The role of hybridization and introgression in the diversification of animals. Ann Rev Ecol Syst 28: 593-619. 40. O'Foighil D, Hilbish TJ, Showman RM (1996) Mitochondrial gene variation in Mercenaria clam sibling species reveals a relict secondary contact zone in the western Gulf of Mexico. Mar Biol 126: 675-683. 41. Goodfriend GA, Gould SJ (1996) Paleontology and chronology of two evolutionary transitions by hybridization in the Bahamian land snail Cerion. Science 274: 1894-1897. 42. Gould SJ, Woodruff DS (1990) History as a cause of area effects: an illustration from Cerion on Great Inagua. Bahamas. Biol J Linn Soc 40: 67-98. 43. Lanham UN (1981) Some Colorado Andrena of the subgenus Scaphandrena of presumed hybrid origin, with special reference to the tarsal claws (Hymenoptera: Apoidea). J Kans Entomol Soc 54: 537-546. 44. Lanham UN (1984) The hybrid swarm of Andrena (Hymenoptera: Apoidea) in Western North America: a possible source for the evolutionary origin of a new species. J Kans Entomol Soc 57: 197-208. 45. Sperling FAH, Harrison RG (1994) Mitochondrial DNA variation within and between species of the Papilio machaon group of swallowtail butterflies. Evolution 48: 408-422. 46. Salazar C, Jiggins CD, Arias CF, Tobler A, Bermingham E, et al. (2005) Hybrid incompatibility is consistent with a hybrid origin of Heliconius heurippa Hewitson from its close relatives, Heliconius cydno Doubleday and Heliconius melpomene Linnaeus. J Evol Biol 18: 247-256. 47. Salazar C., Jiggins CD, Taylor J, Kronforst M, Linares M (2006) Resolving Introgression and Hybrid Speciation among Heliconius Butterflies. Genetics submitted. 48. Schwarz D, Matta BM, Shakir-Botteri NL, McPheron BA (2005) Host shift to an invasive plant triggers rapid animal hybrid speciation. Nature 436: 546-549. 49. Lieder U (1983) introgression as a factor in the evolution pf polytypical plankton cladocera. Int Revue ges Hydrobiol 68: 269-284. 50. Hebert PDN (1985) Interspecific hybridization between cyclic parthenogens. Evolution 39: 216-220. 51. Lieder U (1987) The possible origin of Daphnia cucullata procurva Poppe1887 in the lakes of the Pomeranian Lakeland by hybridization in the past. Hydrobiologia 145: 201-211. 52. Schwenk K, Spaak P (1995) Evolutionary and Ecological consequences of interspecific hybridization in cladocerans. Experientia 51: 465-481. 53. Taylor DJ, Hebert PDN (1992) Daphnia galeata mendotae as a cryptic species complex with interspecific hybrids. Limnol Oceanogr 37: 658-665. 41

54. Taylor DJ, Heber PDN (1993) Cryptic intercontinental hybridization in Daphnia (Crustacea): the ghost of introductions past. Proc Roy Soc Lond B 254: 163-168. 55. Taylor DJ, Hebert, P. D. N. and Colourne, J. K. (1996) Phylogenetics and evolution of the Daphnia longispina group (Crustacea) based on 12S rDNA sequence and allozyme variation. Mol Phylogenet Evol 5: 495-510. 56. Mina MV (1992) An interpretation of diversity in the Salmonid genus Brachymystax (Pisces, Salmonidae) as the possible result of multiple hybrid speciation. J Ichthyol 32: 117-122. 57. Osinov AG (1993) Countercurrent dispersal, secondary contact, and speciation in lenoks of the genus Brachymystax (Salmonidae: Salmoniformes). Genetika 29: 654-669. 58. Crabtree CB, Buth DG (1987) Biochemical systematics of the catostomid genus Catostomus : assessment of C. clarki, C. plebeius and C. discobolus including the Zuni sucker, C. d. yarrowi. Copeia 1987: 843-854. 59. Smith GR, Hall JG, Koehn RK, Innes DJ (1983) Taxonomic relationships of the Zuni mountain sucker Catostomus discobolus yarrowi. Copeia 1983: 37-48. 60. Miller RR, Smith GR (1981) Distribution and evolution of Chasmistes (Pisces: Catostomidae) in western North America. Occ Pap Mus Zool Univ Mich 696: 146. 61. Dowling TE, DeMarais BD (1993) Evolutionary significance of introgressive hybridization in cyprinid fishes. Nature 362: 444-446. 62. DeMarais BD, Dowling TE, Douglas MM, Minckley W. L., Marsh PC (1992) Origin of Gila seminuda (Teleostei: Cyprinidae) through introgressive hybridization : Implications for evolution and conservation. Proc Natl Acad Sci, USA 89: 2747-2751. 63. Meagher S, Dowling TE (1991) Hybridization between the Cyprinid fishes Luxilus albeolus, L. cornutus, and L. cerasinus with comments on the proposed hybrid origin of L. albeolus. Copeia 4: 979-991. 64. Menzel BW (1977) Morphological and electrophoretic identification of a hybrid cyprinid fish, Notropis cerasinus x Notropis c. cornutus, with implications on the evolution of N. albeolus. Comp Biochem Physiol 57B: 215-218. 65. Menezes NA, Weitzman SH (1990) Two new species of Mimagoniates (Teleostei: Characidae: Glandulocaudinae). their phylogeny and biogeography and a key to the glandulocaudin fishes of Brazil and Paraguay. Proc Biol Soc Wash 103: 380426. 66. Goodfellow WN, Hocutt CH, Morgan RP, Stauffer JR (1984) Biochemical assessment of the taxonomic status of "Rhinichthys bowersi" (Pisces: Cyprinidae). Copeia 1984: 652-659. 67. Stauffer JR, Hocutt CH, Denoncourt RF (1978) Status and distribution of the hybrid Nocommis micropogon x Rhinichthys cataractae, with a discussion of hybridization as a viable mode of vertebrate speciation. Am Midl Nat 101: 355365. 68. Stauffer JR, Hocutt CH, Mayden RL (1997) Pararhinichthys, a new monotypic genus of minnows (Teleostei: Cyprinidae) of hybrid origin from eastern North America. Ichthyol Explor Freshwaters 7: 327-336. 69. Smith PF, Konings AD, Kornfield IRV (2003) Hybrid origin of a cichlid population in Lake Malawi: implications for genetic variation and species diversity. Mol Ecol 12: 2497-2504. 70. Salzburger W, Baric S, Sturmbauer C (2002) Speciation via introgressive hybridization in East African cichlids? Mol Ecol 11: 619-625. 42

71. Meyer A, Salzburger W, Schartl M (2006) Hybrid origin of swordtail species (Teleostei: Xiphophorus clemenciae) driven by sexual selection. Mol Ecol 15: 721-731. 72. Greer GJ, Kitikoon V, Lohachit C (1989) Morphological and life cycle of Schistosoma sinensium Pao, 1959, from northwest Thailand. J Parasitol 75: 98101. 73. Hirai H, Taguchi, T., Saitoh, Y., Kawanaka, M., Sugiyama, H., Habe, S., Okamoto, M., Hirata, M., Shimada, M., Tiu, W. U., Lai, K., Upatham, E. S., and Agatsuma, T. (2000) Chromosomal differentiation of the Schistosoma japonicum complex. Int J Parasitol 30: 441-452. 74. Jones TR, Routman EJ, Begun DJ, Collins JP (1995) Ancestry of an isolated subspecies of salamander, Ambystoma tigrinum stebbinsi Lowe: the evolutionary significance of hybridization. Mol Phylog Evol 4: 194-202. 75. Hutchinson MN, Donnellan SC (1992) Taxonomy and genetic variation in the Australian Lizards of the genus Pseudemoia (Scincidae: Lygosominae). J Nat Hist 26: 215-264. 76. Johnston RF (1969) Taxonomy of house sparrows and their allies in the Mediterranean basin. Condor 71: 129-139. 77. Roy MS, Geffen E, Smith D, Ostrander EA, Wayne RK (1994) Patterns of differentiation and hybridization in North American wolflike canids, revealed by analysis of microsatellite loci. Molec Bio Evol 11: 553-570. 78. Wayne RK, Jenks SM (1991) Mitochondrial DNA analysis implying extensive hybridization of the endangered red wolf Canis rufus. Nature 351: 565-568. 79. Yonekawa H, Moriwaki K, Gotoh O, Miyashita N, Matsushima Y, et al. (1988) Hybrid origin of Japanese Mice "Mus musculus molossinus¨: Evidence from Restriction Analysis of Mitochondrial DNA. Molec Bio Evol 5: 63-78. 80. McFadden CS, Hutchinson B (2004) Molecular evidence for the hybrid origin of species in the soft coral genus Alcyonium (Cnidaria: Anthozoa: Octocorallia). Mol Ecol 13: 1495-1505. 81. Taylor DJ, Herbert PDN (1993) Habitat dependent hybrid parentage and differential introgression between neighboringly sympatric Daphnia species. Proc Natl Acad Sci, USA 90: 7079-7083. 82. Taylor DJ, Herbert PDN (1993) A reappraisal of phenotypic variation in Daphnia galeata mendotae : the role of interspecific hybridization. Can J Fish Aquat Sci 50: 2137-2146. 83. Nowak RM (1979) North American Quaternary Canis Monograph 6.; Kansas Uo, editor. Lawrence: Museum of Natural History. 84. Nowak RM (1992) The red wolf is not a hybrid. Conserv Biol 6: 593-595. 85. Wayne RK (1992) On the use of morphologic and molecular genetic characters to investigate species status. Conserv Biol 6: 590-592. 86. Berlocher SH, Feder JL (2002) Sympatric speciation in phytophagous insects: moving beyond controversy? Ann Rev Entomol 47: 773-815. 87. Ramsay M, Williamson R, Estivill X, Wainwright BJ, Ho MF, et al. (1993) Haplotype analysis to determine the position of a mutation among closely linked DNA markers. Hum Mol Genet 2: 1007-1014. 88. Berlocher SH (2000) Radiation and divergence in the Rhagoletis pomonella species group: inferences from allozymes. Evolution 54: 543-557. 89. Seehausen O (2004) Hybridization and adaptive radiation. Trends Ecol Evol 19: 198-207. 90. Grant PR, Grant BR (1992) Hybridization of bird species. Science 256: 193-197. 43

91. Grant PR, Grant BR (1994) Phenotypic and genetic effects of hybridization in Darwin's finches. Evolution 48: 287-296. 92. Grant B, Grant PR (2003) What Darwin's finches can teach us about the evolutionary origin and regulation of biodiversity. BioScience 53: 965-975. 93. Kornfield I, Smith PF (2000) African cichlid fishes: models systems for evolutionary biology. Ann Rev Ecol Syst 31: 163-196. 94. Danley PD, Kocher TD (2001) Speciation in rapidly diverging systems:lessons from Lake Malawi. Molec Ecol 10: 1075-1086. 95. Danley PD, Market JA, Arnegard ME, Kocher TD (2000) Divergence with gene flow in the rock-dwelling cichilds of lake Malawi. Evolution 54: 1725-1737. 96. Knight ME, Turner GF (1999) Reproductive isolation among closely related Lake Malawi cichlids: can males recognize conspecific females by visual clues? Anim Behav 58: 761-768. 97. Parker A, Kornfield I (1997) Evolution of the mitochondrial DNA control region in the mbuna (Cichlidae) species flock of Lake Malawi, East Africa. J Molec Evol 45: 70-83. 98. Markert JA, Danley PD, Arnegard ME (2001) New markers for new species: microsatellite loci and the East African cichlids. Trends Ecol Evol 16: 100-107. 99. Seehausen O, Alphen JJM, Witte F (1997) Effects of water transparency and light regime on sexual selection and the diversity of haplochromine cichlids. Science 277: 1808-1811. 100. Stauffer JR, Bowers NJ, Kocher TD, McKaye K. R. (1996) Evidence of hybridization between Cyanotilapia afra and Psuedotropheus zebra (Teleostei: Cichlidae) following an intralacustrine translocation in Lake Malawi. Copeia: 203-208. 101. Ruber L., Meyer A, Sturmbauer C, Verheyen E (2001) Population structure in two sympatric species of the Lake Tanganyika cichlid tribe Eretmodini: evidence for introgression. Molec Ecol 10: 1207-1225. 102. McElroy DM, Kornfield I (1993) Novel jaw morphology in hybrids between Pseudotropheus zebra and Labeotropheus fullerborni (Teleostei: Cichlidae) from Lake Malawi, Africa. Copeia 1993: 933-945. 103. Markert JA, Arnegard ME, Danley PD, Kocher TD (1999) Biogeography and population genetics of the Lake Malawi cichlid Melanochromis auratus: habitat transience, philopatry and speciation. Molec Ecol 8: 1013-1026. 104. Gilbert LE (2003) Adaptive novelty through introgression in Heliconius wing patterns: Evidence for shared genetic 'tool box' from synthetic hybrid zones and a theory of diversification. In: Boggs CL, Watt WB, Ehrlich PR, editors. Ecology and Evolution Taking Flight: Butterflies as Model Systems. Chicago: Univ. of Chicago Press. 105. Jiggins CD, Naisbit RE, Coe RL, Mallet J (2001) Reproductive isolation caused by colour pattern mimicry. Nature 411: 302-305. 106. Mallet J, Gilbert LE (1995) Why are there so many mimicry rings? Correlations between habitat, behaviour and mimicry in Heliconius butterflies. Biol J Linn Soc 55: 159-180. 107. Smiley JT (1978) Plant chemistry and the evolution of host specificity: new evidence from Heliconius and Passiflora. Science 201: 745-747. 108. Mallet J, McMillan WO, Jiggins CD (1998) Mimicry and warning colour at the boundary between races and species. In: Howard DJ, Berlocher SH, editors. Endless Forms: Species and Speciation. New York: Oxford University Press. pp. 390-403. 44

109. Posada D, Crandall KA (2001) Intraspecific gene genealogies: trees grafting into networks. Trends Ecol Evol 16: 37-45. 110. Linder CR, Rieseberg LH (2004) Reconstructing patterns of reticulate evolution in plants. Am J Bot 9: 1700-1708. 111. Stephens M (2003) Inference under the coalescent. In: Balding DJ, Bishop M, Cannings C, editors. Hanbook of Statistical Genetics. Chichester: John Wiley and Sons Ltd. pp. 636-660. 112. Maddison WP (1997) Gene trees in species trees. Syst Biol 46: 523-536. 113. Bandelt H-J, Dress WM (1992) Split decomposition: a new and useful approach to phylogenetic analysis of distance data. Mol Phylog Evol 1: 242-252. 114. Rieseberg LH, Raymond O, Rosenthal DM, Lai Z, Living-stone K, et al. (2003) Major Ecological Transitions in Wild Sunflowers Facilitated by Hybridization. Science 301: 1211-1216. 115. Sattath S, Tversky A (1977) Additive similarity trees. Psychometrika 42: 319-345. 116. Huson DH (1998) SplitsTree: a program for analyzing and visualizing evolutionary data. Bioinformatics 14: 68-73. 117. Bandelt H-J, Forster P, Rohlf A (1999) Median-Joining Networks for Inferring Intraspecific Phylogenies. Mol Biol Evol 16: 37-48. 118. Xu SZ (2000) Phylogenetic analysis under reticulate evolution. Molec Biol Evol 17: 897-907. 119. Bryant D, Moulton V. NeighborNet: an agglomerative method fro the construction of planar phylogenetic networks.; 2002; WABI, Rome, Italy. Lecture Notes in Computer Science. pp. 375-391. 120. Makarenkov V (2001) T-REX: reconstructing and visualizing phylogenetic trees and ticulation networks. Bioinformatics 17: 664-668. 121. Makarenkov V, Legendre P (2004) From phylogenetic tree to a reticulated network. J Comput Biol 11: 195-212. 122. Clement MD, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 9: 1657-1660. 123. Bandelt HJ, Macaulay V, Richards M (2000) Median networks: speedy construction and greedy reduction, one simulation, and two case studies from human mtDNA. Molec Phylog Evol 16: 8-28. 124. Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491. 125. Huber kT, Watson EE, Hendy MD (2001) An algorithm for constructing local regions in a phylogenetic network. Molec Phylog Evol 19: 1-8. 126. Nakhleh L, Sun J, Warnow T, Linder R, Moret BME, et al. Reconstructing reticulate evolution in species: theory and practice.; 2004; San Diego, California, USA. pp. 337-346. 127. Rokas. A., Williams BL, King N, Carroll SB (2003) Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature 425: 798-804. 128. Nichols R (2001) Gene trees and species trees are not the same. Trends Ecol Evol 16: 358-364. 129. Arvestad L, Berglund AC, Lagergren J, Sennblad B (2003) Bayesian gene/species tree reconciliation and orthology analysis using MCMC. Bioinformatics 19: i7i15. 130. Machado CA, Hey J (2003) The causes of phylogenetic conflict in a classic Drosophila species group. Proc R Soc Lond B Biol Sci 270: 1193-1202. 45

131. Posada D, Crandall KA (2001) Evaluation of methods for detecting recombination from DNA sequences: computer simulatons. Proc Natl Acad Sci, USA 98: 13757-13762. 132. Kliman RM, Andolfatto P, Coyne JA, Depaulis F, Kreitman M, et al. (2000) The population genetics of the origin and divergence of the Drosophila simulans complex species. Genetics 156: 1913-1931. 133. Hey J, Machado CA (2003) The study of structurated populations - new hope for a difficult and divided science. Nat Rev Genet 4: 535-543. 134. Rosenberg NA, Nordborg M (2002) Genealogical trees, coalescent theory and the analysis of genetic polymorphisms. Nat Rev Genet 3: 380-390. 135. Hey J (2005) On the Number of New World Founders: A Population Genetic Portrait of the Peopling of the Americas. PLoS BioL 3: 0965-0975. 136. Won YJ, Hey J (2005) Divergence population genetics of chimpanzees. Molec Biol Evol 22: 297-307. 137. Won YJ, Sivasundar A, Wang Y, Hey J (2005) On the origin of Lake Malawi cichlid species: a population genetic analysis of divergence. Proc Natl Acad Sci, USA 102: 6581-6586. 138. Hey J., Nielsen R. (2004) Multilocus methods for estimating population sizes, migration rates and divergence time, with applications to the divergence of Drosophila pseudoscura and D. persimilis. Genetics 167: 747-760. 139. Machado CA, Kliman RM, Markert JA, Hey J (2002) Inferring the history of speciation from multilocus DNA sequence data: The case of Drosophila pseudoobscura and close relatives. Mol Biol Evol 19: 472-488. 140. Machado CA, Robbins N, Gilbert MTP, Herre EA (2005) Critical review of host specificity and its coevolutionary implications in the fig/fig-wasp mutualism. Proc Natl Acad Sci, USA 102: 6558-6565. 141. Beerli P, Felsenstein J (2001) Maximum likehood estimation of a migration matrix and effective population sizes in n subpopulations by using a coalescent approach. Proc Natl Acad Sci, USA 98: 4563-4568. 142. Gow J, Peichel CL, Taylor EB (2006) Contrasting hybridization rates between sympatric three-spined sticklebacks highlight the fragility of reproductive barriers between evolutionary young species. Molec Ecol 15: xxx-xxx. 143. Rannala B, Yang Z (2003) Bayes Estimation of Species Divergence Times and Ancestral Population Sizes Using DNA Sequences from Multiple Loci. Molec Ecol 164: 1645-1656. 144. Mallet J (2005) Hybridization as an invasion of the genome. Trends Ecol Evol 20: 229-237. 145. Vaha J-P, Primmer CR (2006) Efficiency of model-based Bayesian methods for detecting hybrid individuals under different hybridization scenarios and with different numbers of loci. Molec Ecol 15: 63-72. 146. Gross BL, Rieseberg LH (2004) The Ecological Genetics of Homoploid Hybrid Speciation. J of Heredity 96: 241-252. 147. Neff NA, Smith GR (1978) Multivariate analysis of hybrid fishes. Syst Zool 28: 176-196. 148. Humphries JM, Bookstein FL, Chernoff B, Smith GR, Elder RL, et al. (1981) Multivariate discrimination by shape in realtion to size. Syst Zool 30: 291-308. 149. Rosenthal DM, Rieseberg LH, Donovan LA (2005) Re-creating Ancient Hybrid Species Complex Phenotypes from Early-Generation Synthetic Hybrids: Three Examples Using Wild Sunflowers. Am Nat 166: 26-41. 46

CHAPTER II Published in Journal of Evolutionary Biology 18: 247-256 Hybrid incompatibility is consistent with a hybrid origin of Heliconius heurippa Hewitson from its close relatives, Heliconius cydno Doubleday and Heliconius melpomene Linnaeus.

ABSTRACT Shared ancestral variation and introgression complicates the reconstruction of phylogenetic relationships among closely related taxa. Here I use overall genomic compatibility as an alternative estimate of species relationships in a group where divergence is rapid and genetic exchange is common. Heliconius heurippa, a butterfly species endemic to Colombia, has a colour pattern genetically intermediate between H. cydno and H. melpomene: its hindwing is nearly indistinguishable from that of H. melpomene and its forewing band is an intermediate phenotype between both species. This observation has lead to the suggestion that the pattern of H. heurippa arose through hybridization. I present a genetic analysis of hybrid compatibility in crosses between the three taxa. H. heurippa x H. cydno and female H. melpomene x male H. heurippa yield fertile and viable F1 hybrids, but male H. melpomene x female H. heurippa crosses yield sterile F1 females. In contrast, Haldane’s rule has previously been detected between H. melpomene and H cydno in both directions. Therefore, H. heurippa is most closely related to H. cydno, with some evidence for introgression of genes from H. melpomene. The results are compatible with the hypothesis of a hybrid origin for H. heurippa. In addition, backcrosses using F1 hybrid males provide evidence for a large Z (X)chromosome effect on sterility and for recessive autosomal sterility factors as predicted by Dominance Theory.

47

INTRODUCTION In groups of rapidly radiating species, resolution of phylogenetic relationships based on morphological or DNA sequence characters can be difficult. This can be due to low divergence between closely related species, such that there is not sufficient information in the data to reconstruct a reliable phylogenetic hypothesis. A more difficult problem to resolve is that genetic variation is often shared between species for long periods subsequent to speciation, leading to well-known problem of discordance between gene trees and species trees [128,150]. Even worse, hybridization likely occurs between lineages for considerable periods subsequent to divergence, such that different parts of the genome have distinct historical relationships [139,151].

Hybridization can result in the flow of neutral loci between species [139,152]. Furthermore, introgression might be favoured by natural selection if hybridization results in novel adaptive gene combinations. In plants, hybridization commonly leads to the establishment of novel species with a genome composed of large regions from two or more parental taxa [6,114]. Although the production of such hybrid genomes seems to be rare in animals [153,154], it is likely that occasionally novel combinations of genes derived from hybridization play an important role in adaptive evolution, and perhaps even in speciation [39,155]. Furthermore, it has recently been suggested that colonization of new habitats may increase rates of hybridization and introgression facilitating rapid adaptive radiation in many animal species [89]. Thus, reconstructing relationships among populations and species is complicated in part by 1) shared ancestral variation, 2) gene flow of neutral loci across species boundaries and 3) the possible hybrid origin of novel adaptive gene combinations. Both 2) and 3) clearly violate the assumptions underlying the reconstruction of a bifurcating species tree.

It has been suggested that this problem can be resolved by studying the genealogical relationships of loci that cause sterility [156]. However there are two major problems with this approach. First, despite intensive study, only seven cases of loci causing species incompatibilities have been identified in animals [157-163]. Second, incompatibility is thought to result from the gradual accumulation of certain genes [14,15,21,164], and the genealogical history of any one such locus may not necessarily be representative of the genome as a whole. Instead, where species have recently 48

diverged and hybridization is still common, an estimate of average divergence across the whole genome would be useful to determine species relationships.

I instead propose that an estimate of genomic divergence is possible using traditional crossing techniques. Comparisons of incompatibility with genetic distance show a relatively linear positive relationship in all taxa where sufficient data exists, implying that genomic incompatibility accumulates gradually and at an approximately constant rate between populations [15,165-169]. The accumulation of incompatibility is not expected to be truly linear, and theory predicts a ‘snowball’ effect whereby a linear accumulation of single gene differences would lead to an exponential increase in incompatibility [170]. This is because the latter is caused by epistatic gene interactions rather than single gene effects. Nonetheless this accumulation of incompatibility has been termed a “speciation clock” [166], and while it is clearly not truly clock-like, an estimate of the relative degree of incompatibility between taxa should give an estimate of their relative divergence times. We here aim to determine the relationships of three recently diverged taxa using such data.

In Heliconius butterflies, inter-specific hybridization is known in many taxa[104,108] Wild-caught hybrid specimens are known between several species in the Heliconius melpomene Linnaeus and Heliconius cydno Doubleday (Lepidoptera: Nymphalidae) complex, and there is also shared variation among these taxa at both mtDNA [47] and nuclear genes [171]. One member of this group, Heliconius heurippa Hewitson, is unusual in that its colour pattern appears to share distinct genetic elements derived from two putative parental species. Crossing experiments have shown that elements of the H. heurippa pattern breed true when hybridized with either H. melpomene or H. cydno. Such crosses provide some evidence for genetic homology of pattern elements (Figure 2.1; Linares unpub.), as if similar phenotypes were independently derived in each species then one might expect some disruption of their development in hybrid individuals. Such breakdown does occur in certain crosses in Heliconius, providing evidence for similar yet independently derived phenotypes in other species [172]). In the case of H. heurippa (Figure 2.1b), the hindwing is nearly indistinguishable from that of the subspecies Heliconius melpomene melpomene Linneaus (Figure 2.1c) and breeds true when they are crossed (Figure 2.1e; Linares unpub.), while the yellow forewing band is closer in appearance to the subspecies Heliconius cydno cordula 49

Neustetter (Figure 2.1a) and similarly breeds true (Figure 2.1f; Linares unpub.). Furthermore, in the forewing the pattern of a distal red band and a proximal yellow band is extremely similar to that seen in H. melpomene x H. cydno hybrids, suggesting that the red forewing band could be generated by genes homologous to the red forewing in H. m. melpomene (Figure 2.1e; [104,173]; Linares unpub.). Thus, based strictly on wing colour pattern genetics H. heurippa appears to be intermediate between H. c. cordula and H. m. melpomene and perhaps somewhat closer to the latter (see dotted diagram in Figure 2.1).

Figure 2.1 Species used in this study. a. Heliconius cydno cordula, b. H. heurippa, c. H. melpomene melpomene, d. A hybrid derived from a cross between offspring of two C x (C x H) backcross families, e. F1 (M x H) hybrid and f. F1 (C x H) hybrid, showing a close resemblance to the species H. heurippa. All pictures show dorsal (left) and ventral (right) view. The cladogram shows proposed relationships based on the present study, whereby H. heurippa is most closely related to H. cydno, but may have arisen through introgression of colour pattern gene(s) from H. melpomene (dotted line). However, note that four colour pattern traits, the red forewing band, short forewing red line, red hindwing spots and absence of hindwing brown mark all suggesting H. heurippa is more closely related to H. melpomene.

50

In addition, a colour pattern virtually identical to that of H. heurippa can be created in the laboratory after just a few generations of hybridization between H. c. cordula and H. m. melpomene (Figure 2.1d), subspecies that occur in sympatry near to the current range of H. heurippa, on the eastern slope of the Colombian Andes near Villavicencio (Figure 2.2). This has lead to the hypothesis that H. heurippa has arisen as a result of hybridization between H. melpomene and H. cydno (K. S. Brown & L. Gilbert pers.com; Linares unpub.). Alternatively, the H. heurippa pattern elements may be ancestral to the group, or they may have arisen multiple times using similar developmental pathways. These alternatives are difficult to distinguish, but the hybrid origin hypothesis is clearly supported by the fact that among all the diversity of colour pattern races in both H. melpomene and H. cydno, the two phenotypes that give rise to the H. heurippa pattern are those that occur closest to its current geographic range. Unsurprisingly, given the intermediate nature of H. heurippa, its phylogenetic position is unclear. It has been considered a close relative of H. melpomene based on colour pattern and genitalia [174] and to H. cydno according to mitochondrial DNA [171,175].

Figure 2.2 Geographic distribution of the study species in the eastern Andes. The eastern slopes of the Colombian Andes are shown, with partial distributions of H. melpomene and H. cydno, complete known distribution of H. heurippa and sample localities (black dots). H. melpomene is found throughout the lowlands from Central America to Brazil, while H. cydno is found throughout the Andes. H. melpomene is broadly sympatric with both H. heurippa and H. cydno in the Andean foothills. H. cydno and H. heurippa are probably parapatric around Yopal, but this region remains poorly explored. 51

In order to further unravel the evolutionary origin of H. heurippa, I here describe crosses designed to investigate the hybrid incompatibility of H. heurippa with respect to both H. cydno and H. melpomene. The aim of this study is 1) to examine the degree of hybrid compatibility between these three taxa in order to determine more precisely the relationship of H. heurippa with respect to its putative parental species, and 2) conduct backcross experiments to investigate whether the genetic incompatibilities among these taxa show any evidence for recessive factors and a large Z (X) effect.

RESULTS

Crosses between H. heurippa and H. melpomene Female hybrids between H. heurippa (H) and H. melpomene (M) from the Eastern Andean foothills show asymmetrical sterility. The female offspring of a cross between female H. heurippa and male H. melpomene (H x M) were completely sterile, either failing to lay eggs or laying eggs that never hatched (Table 2.1). Sterile eggs in the F1 generation were consistently smaller in size compared to eggs laid by controls (CS, pers. obs.), similar to phenotypes observed in previous crosses involving H. melpomene from French Guiana [176]. In contrast, the reciprocal cross (M x H) produced female offspring that laid fertile eggs with a hatch rate of 0.56± 0.076, not significantly different to the controls, 0.53±0.059 for M and 0.66±0.06 for H (G2=2.38, n.s.; Table 2.1). Male hybrids were fully fertile in both crosses, with hatch rates of 0.69±0.115 and 0.52±0.09 respectively, following Haldane's Rule (Table 2.1). The average proportion of females was 0.49± 0.05 and did not differ significantly in seven cross classes (G7=8.75, n.s.), four F1 and three backcross broods (a total of 76 broods, 795 adults): 1) (M x M), 2) (H x H), 3) (H x M), 4) (M x H), 5) ((M x H) x H/ F1/M), 6) (H/M x (M x H)) and 7) (H/M x (H x M)). This observation suggests an absence of Haldane's Rule for viability.

52

Table 2.1. Hatch rate of control, F1 and backcross broods. Individuals of hybrid genotype are coded as maternal x paternal genotype where M is H. melpomene melpomene, H is H. heurippa and C is H. cydno. Mean hatch rate, the variance in hatch rate and their standard errors were calculated using the program BetaBino (see Methods). Note that in some brood classes hatch rate may have a bimodal, L or reverse-L shaped distribution.

Cross type

Maternal genotypePaternal genotypeNo. of broodsNo. of eggsMean hatch rate SE Variance SE H. melpomene x H. heurippa crosses

Pure M H

M H

17 22

701 710

0.533 0.664

0.059 0.060

0.051 0.077

0.016 0.019

H M

M H

4 2

128 182

0.695 0.765

0.144 0.22

0.085 0.11

0.05 0.096

HxM HxM

H/F1/M H/F1/M

12 6

127 0

0 0

-

-

-

MxH H/M H/M (M x H) x M

H/F1/M MxH HxM M

17 7 7 25

522 308 302 722

0.564 0.522 0.695 0.504

0.076 0.09 0.115 0.055

0.089 0.044 0.101 0.068

0.021 0.03 0.04 0.015

0.658 0 0 0.602

0.183 0.109

0.114 0.082

0.054 0.031

F1 interspecific

Sterile backcrosses

Fertile backcrosses

F2 interspecific (M x H) x (M x H) (M x H) x (M x H) (M x H) x (M x H) H/M

M/H 4 85 M 4 51 H 1 0 (M x H) x (M x H) 7 220 H. cydno x H. heurippa crosses

Pure C H

C H

27 22

1206 710

0.665 0.664

0.053 0.060

0.071 0.077

0.016 0.019

C H

H C

6 5

616 258

0.461 0.59

0.086 0.135

0.037 0.091

0.020 0.04

CxH HxC H/C (C x H) x (C x H) (C x H) x C (C x H) x H

F1/C/H/F2 H/C/F1 HxC F2/C C/F2/H F2

12 10 3 4 4 2

185 228 351 99 119 24

0.69 0.66 0.50 0.70 0.78 0.876

0.053 0.014 0.067 0.025 0.214 0.098 0.143 0.082 0.144 0.090 0.090 0.0003

0.011 0.015 0.066 0.048 0.065 0.027

F1 interspecific

Backcrosses

Crosses were carried out to distinguish between sterility caused by genomic incompatibility and cytoplasmic factors. Here, the sterile F1 females have cytoplasm and W chromosome from H, but a Z (X) chromosome from M. The female offspring of a fertile F1 female (M x H) backcrossed to an M male, possesses the cytoplasm, Z (X)chromosome and on average 75% autosomes from M. These females should be fully fertile if F1 sterility results from cytoplasmic effects, but in fact this cross shows 53

segregation of sterility phenotypes, ranging from fully fertile to partially sterile, although never with complete sterility (Figure 2.3a). This suggests that interactions between the M Z (X) -chromosome and H autosomal genes are the cause of F1 sterility. The same type of M Z (X) -chromosome-H autosomal interaction could be responsible for the segregation of sterility observed in F2 (M x H) x (M x H) broods: of 9 females tested from two broods, 5 were fertile (0.658±0.183; G2=2.36, n.s.) and 4 were sterile (Table 2.1). In addition, 7 males tested were fully fertile. However, autosomal homozygous interactions (H2 sensu [14]) through homozygous H and/or M autosomes cannot be ruled out as the main cause of sterility in these F2 broods. To verify the importance of the Z (X)-chromosome in causing sterility and to test for autosomal recessive or dominant sterility factors, fertile F1 males were backcrossed to females of the two parental species. In crosses in both directions, female offspring were recovered showing a complete range of fertility, but with pronounced bimodality (Appendix 2.1a,b; Figure 2.3b,c). In the backcross to H. heurippa 26 females were tested from 7 broods, with a ratio of 8 sterile to 18 fertile females, when all classes showing some degree of fertility were combined. Some of these sterile females laid eggs while others did not. There was a highly significant association between sterility and Z (X)-linked Tpi in female offspring from the backcross to H. heurippa (appendix 2.1a; Figure 2.3b). All seven females that were completely sterile lacked the H Tpi insertion and therefore had the M Z (X) chromosome while most females that showed some fertility had the H Z (X) chromosome (G1= 8.43, P<0.05). In the backcross to H. melpomene the H Z (X) -chromosome is being introgressed into a largely H. melpomene autosomal background. The H Z (X) -chromosome is associated with complete fertility in the F1 generation, so we did not expect significant sterility in this cross. Nonetheless, of 34 females tested from three broods, 12 were sterile and 22 were fertile. In all cases sterile females laid eggs that failed to hatch. In this cross I scored two Z(X)-linked markers, Tpi and ap that showed 6 recombinant genotypes in 30 individuals scored, a recombination rate of 0.2. There was no significant association of Z(X)-chromosome genotype with sterility either for each marker individually (Tpi, G1= 0.67, n.s. and Apt, G1= 1.64, n.s.; Fig. 2c; appendix 1b) or both markers together (G3= 2.30, n.s.; Figure 2.2c; Appendix 2.2). Nonetheless the fertility of females was more common when both Tpi and ap alleles belong to H. melpomene: of 21 fertile females, 13 54

had both ap and Tpi alleles from H. melpomene. This result suggests that there may be a weak effect of the H Z (X)-chromosome on sterility too. I also demonstrated segregation of sterility phenotypes among various F2 individuals, although the small sample size means that these crosses are not very informative with respect to the genetics of sterility (Table 2.1).

Figure 2.3 Segregation of sterility phenotypes in backcross broods. The distribution of hatch rates for the female offspring of each cross type is shown. Shading indicates genotypes at Tpi and ap marker loci with superscripts showing the alleles; M: H. melpomene, H: H. heurippa and ?: for individuals that were not analyzed. Female genotypes are shown first.

Crosses between H. heurippa and H. cydno

55

Hybrids between H. heurippa (H) and H. cydno (C) were completely fertile and viable. The female offspring of a cross female H. heurippa x male H. cydno (H x C) laid eggs with a hatch rate of 0.66 ± 0.067; the reciprocal cross (C x H) had a hatch rate of 0.69 ± 0.053. In all F1 and backcross broods tested between H. heurippa and H. cydno, fertility was not significantly different to controls (Table 2.1; G9 = 7.90, n.s.). In general, F1 and backcross broods involving H heurippa x H. cydno and H. heurippa x H melpomene crosses showed similar sex ratios. The average proportion of females was 0.52± 0.05 and did not differ significantly in seven cross classes (G7=11.15, n.s.), four F1 and three backcross broods (a total of 85 broods, 1258 adults): 1) (C x C), 2) (H x H), 3) (C x H), 4) (H x C), 5)((C x H) x F1/C/H/ F2), 6) ((H x C) x H/C/ F1) and 7) (H/C x (H x C)). This observation suggests an absence of Haldane's Rule for viability in these crosses. DISCUSSION

Sterility and the status of H. heurippa H. heurippa is entirely compatible with H. cydno, but shows sex-specific hybrid sterility following Haldane’s Rule, when crossed with H. melpomene. Clearly, H. heurippa is much more closely related to H. cydno than to H. melpomene. This does not rule out a role for hybridization in the origin of H. heurippa, and implies that, if hybridization did occur it must have involved introgression of H. melpomene colour pattern allele(s) into a largely H. cydno genetic background, rather than an even hybrid mixing of the two species. In agreement with this hypothesis, CoI-CoII and Tpi gene sequences group H. heurippa with H. cydno [171,175]. However, given the hybridization and gene flow known to occur between these taxa, estimates of relationships based on single gene loci are likely to be subject to error. Indeed, a third nuclear gene, Mpi, shows almost identical alleles shared between individuals of H. melpomene and H. cydno [177]. The estimate of relationships based on genomic compatibility presented here is a useful complement to previously published gene genealogies and morphological studies, and is perhaps more likely to reflect similarity across the whole genome of the three species.

56

Recessivity and the genetics of hybrid sterility Given the relatively few phylogenetically independent comparisons supporting Haldane’s Rule [178], and the great preponderance of studies in male heterogametic taxa such as Drosophila, there is a clearly a need to study hybrid breakdown in Lepidoptera and birds in order to determine whether there is a common genetic basis to the phenomenon [176,179-181]. With the asymmetry for sterility between H. melpomene and H. heurippa found in this study, there are now 30 cases in Lepidoptera of female sterility and 56 of female inviability, following Haldane`s Rule [166]. In contrast, in Drosophila male sterility is far more frequent than male inviability [180,181] and this pattern has been explained by dominance theory combined with faster male evolution [13,180]. The contrast between Drosophila and Lepidoptera could be caused by the effect of faster-male evolution, but this remains to be demonstrated. Faster-male evolution would act against the faster evolution of sterility in the female heterogametic sex in Lepidoptera [23]. In my crosses, there is evidence for a large Z (X) effect, but as has been discussed, such an effect is largely a result of the backcross design and does not necessarily indicate a greater accumulation of sex-linked versus autosomal sterility factors.

The large X effect is consistent with Dominance Theory, but more convincing support would come from the demonstration of autosomal recessive sterility factors as predicted by the theory [14,21,164]. There is some evidence for such factors in my backcross to H. melpomene. This cross involves introgression of the H Z-chromosome into an H. melpomene genetic background, a combination that is associated with complete fertility in the (M x H) F1. The H autosome by M Z-chromosome interaction that causes sterility in the F1 is also expressed in this brood, but in at most 25% of the offspring. Hence, the observed segregation of sterile and partially sterile phenotypes (<60% hatch rate) in 25 of the 34 offspring (Figure 2.3c) would seem to suggest additional sterility factors. This must result from H2 interactions (sensu [14]) between homozygous H. melpomene autosomal alleles and the H Z-chromosome, or heterozygous H1 interactions between autosomal H. heurippa and homozygous H. melpomene alleles. At present, evidence for a role of the Z (X)-chromosome is inconclusive. Of the 16 individuals from this backcross with both ap and Tpi M alleles, 13 were fertile but only 3 were sterile, clearly in the direction expected if the Z chromosome had an effect on sterility. 57

However, a test for heterogeneity on the data is not significant (Appendix 2.2 ; G3=2.30, n.s.). Therefore, it seems likely that some combination of H1 and H2 interactions are causing sterility in this cross. While my conclusions are clearly limited by the small sample sizes obtainable in Heliconius crossing experiments, the results nevertheless give some indication of the presence of recessive sterility interactions in agreement with Dominance Theory, and suggest means by which this hypothesis could be further tested.

Using the ‘Speciation Clock’ to study relationships My results are perhaps of greater interest in terms of the light that they shed on diversification in Heliconius. In particular I am interested in H. heurippa because of its possible hybrid origin. The crosses show that H. heurippa is completely compatible with H. cydno, but shows asymmetrical hybrid female sterility with H. melpomene. However if H. heurippa resulted from introgression of genes from H. melpomene, I might also expect somewhat less incompatibility in H. heurippa x H. melpomene crosses as compared to H. cydno x H. melpomene. There is evidence that this is indeed the case. In one direction of cross H. melpomene and H. heurippa produce fertile F1 females, but the corresponding H. melpomene x H. cydno cross is known to produce sterile females. The only published analysis of F1 females in an M x C cross was a brood that produced 3 completely sterile females and 4 females with a 20% hatch rate [179]. In the same analysis, evidence from backcrosses to H. melpomene provided further indirect evidence that M x C crosses would produce sterile F1 females, as the Zchromosome was strongly associated with sterility, in contrast to the lack of association found in my broods (Figure 3c). Hence, the available evidence is consistent with the prediction that H. heurippa is genetically intermediate between H. cydno and H. melpomene.

The accumulation of genes that cause genomic incompatibility is presumably a stochastic process [15]. Hence, when the number of genes that differ between taxa is very low, an estimate of divergence based on compatibility is likely to be subject to a large stochastic error. It is therefore important to determine whether the incompatibility observed in my crosses could be explained by just one or very few loci, as this might cast doubt on my conclusions regarding the relationship of these three taxa. First, if a 58

single sex-linked locus were causing sterility I would probably expect a bimodal 1:1 ratio of completely sterile to fertile phenotypes in the backcross to H. heurippa. Hence, the segregation of many distinct sterility phenotypes in this backcross implies that many loci are involved. Second, the presence of sterility in the backcross to H. melpomene shows that there are divergent autosomal loci between H. melpomene and H. heurippa, in addition to the Z and autosomal loci that interact to cause sterility in the F1 generation. Although the asymmetry of sterility in the F1 generation implies that divergence is recent, there is nonetheless evidence that a significant number of sterility factors have accumulated between H. heurippa and H. melpomene, in contrast to the complete fertility observed between H. heurippa and H. cydno.

Given the complete compatibility between H. heurippa and H. cydno, it could be argued that H. heurippa is best considered a sub-species of H. cydno. Nonetheless, it is known that speciation in Heliconius does not always involve hybrid incompatibility. H. erato and H. himera remain distinct in hybrid zones maintained by strong ecological selection against hybrids and assortative mating, with no reduction in viability or fertility of interspecific hybrids [182,183]. In H. cydno and H. melpomene disruptive sexual selection accompanied by mimicry forms a barrier of >99.9%, while genomic incompatibilities lead to isolation of just 70% [179,184]. Unfortunately, good collections are not available from zones of sympatry between H. heurippa and H. cydno in order to test whether the two species do indeed maintain their distinctness in sympatry. However, H. heurippa females show strong assortative mating when tested against both H. melpomene and H. cydno, which likely results in a strong barrier to gene flow in the wild (Salazar & Linares unpub.), suggesting that H. heurippa is correctly considered a good species.

In the case of H. melpomene and H. cydno, it is known that pre-mating isolation results from mate preferences based on colour pattern [105]. In that case, a switch in colour pattern driven by selection for mimicry was the most likely first step in speciation. In contrast, H. heurippa is not mimetic, and instead I have suggested that its colour pattern has arisen through the establishment of a hybrid pattern resulting from introgression of genes from H. melpomene into H. cydno (Figure 2.1; Linares unpub.). If this novel colour pattern also led to strong pre-mating isolation with H. cydno, then H. heurippa 59

may represent a case of speciation initiated by hybridization, a rather unusual phenomena in animals.

MATERIALS AND METHODS

Collection localities are given in Table 2.2 and Figure 2.2. Crosses involving the three species were performed in La Vega, 50 km northwest of Bogotá, Colombia. H. cydno were collected to the west of the Cauca Valley, in the Dagua region and in the foothills of the eastern slope of the Andes. In crosses between these two H. cydno populations no detectable reduction in hybrid fertility or viability was observed ([185]; Salazar in prep.). Stocks of H. m. melpomene and H. heurippa were established from individuals collected near Villavicencio. In all cases, recently emerged virgin females were presented to sexually mature males in order to maximize the probability of obtaining a successful mating. Sexually mature (two or three days old) females show a low mating probability with males that are not from their own species (Salazar & Linares unpub).

Table 2.2. Collection localities for the Heliconius races used in these crosses. Each species has been given a one-letter code used in Tables 2-3. All localities are in Colombia.

Species

Locality

Latitude, Longitude and altitude

H. melpomene melpomene (M)

Chirajara, Cundinamarca

4º 12’48”N, 73º 47’ 70”W, 11501450m.

H. melpomene melpomene (M)

Dele B (RíoCharte), Casanare

5º 25’5”N, 72º 31’ 20”W, 1150m.

H. heurippa (H)

Chirajara, Cundinamarca

4º 12’48”N, 73º 47’ 70”W, 11501450m.

H, heurippa (H)

Buenavista, Cundinamarca

04º 10’30”N, 73º 40’ 41”W, 1270m.

H. cydno (C)

Barro Negro, Casanare

H. cydno (C)

Atuncela, Valle del Cauca

H. cydno (C)

Río Bravo, Valle del Cauca

6º 01’ 6”N, 72º 05’ 47”W, 1050 m. 3º 44’ 3”N, 76º 41’ 53”W, 1400 m. 3º 54’ 13”N, 76º 38’ 18”W, 1000 m.

60

Mated females were kept individually in 1 x 1 x 2 m outdoor insectaries with access to sufficient pollen sources (Lantana and Psiguria) and artificial nectar (10% sugar solution). Plants were provided for oviposition, mainly Passiflora edulis and P. oerstedii. H. heurippa, H. m. melpomene and H. c. cordula from slopes of the Colombian Andes seem to be oligophagous. The three species lay eggs with a similar frequency on all plant species presented to them (including, P. edulis, P. oerstedii, P. maliformis, P. ligularis and P. arborea) in the insectaries. Also the different larval instars grow equally well on all the Passiflora species that were provided to them as food, mainly P. edulis and P. oerstedii (Salazar & Linares unpub). Eggs were collected every 8 days and kept individually in plastic pots with food. Larvae were reared on Passiflora plants grown directly in the insectaries. After pupation they were transferred to baskets for eclosion. The number of eggs laid, hatch rates, and number of eclosing butterflies were recorded. F1 males, F1 females and female offspring of backcrosses were tested for fertility with respect to control broods reared under the same conditions.

Statistical analysis Hatch rate data were compared using likelihood-ratio tests implemented in the BETABINO program (Freeware development by Ziheng Yang, UCL-UK; See [176,179]). Differences between classes could be established with analysis of variance. However, this method can cause heteroscedasticity when the data show non-normal distributions and the sample size is different between classes. This is the case in the hatch rate data collected here. I wish to compare these rates between broods of different types, taking into account real differences in hatch rates between replicates due to genetic or environmental variation. A better way to analyze my data is to use a binomial parameter that changes across replicate broods within each cross type according to a E distribution. The E-binomial distribution can fit the skewed (e.g., Lshaped, reverse L-shaped) or bimodal distributions expected in extreme cases of variation within brood classes (e.g., crosses with sterility segregation). Then comparisons can be done between crosses using the likelihood parameters of mean, variance, and their standard errors. Five alternative models can be fitted with the program when a data set contains replicate broods of several classes: (1) a classical binomial parameter for each class, which assumes a zero brood to brood variance; (2) a 61

single E mean and variance for the entire data set; (3) different means for each class but a single variance; (4) a single mean but different variances; and (5) a different mean and variance for each class. Fitting the models on only part of the data set allows for specific hypotheses to be tested using likelihood ratio tests. For more details about use and implementation see ftp://abacus.gene.ucl.ac.uk/pub/ and the Appendix to [176]. Contingency tables for the segregation of sterility phenotypes and genetic marker loci were tested for heterogeneity using the program ‘Monte Carlo R x C v2.0’ designed by W. Engels, University of Wisconsin. When the expected data in any cell is less than 5, the program uses the Lewontin and Felsenstein (1965) algorithm [186], that takes a large number of random tables with the same marginal sums as the observed data, and simultaneously asks if each table deviates from the expected as much or more than the observed data. The ratio of the number of trials with log likelihood ratio test (LLR) less than or equal to the observed data to the total number of trials can be used as a true probability. All analyses were carried out with 20,000 trials.

Polymerase Chain Reactions methods for Tpi and Apterous Total genomic DNA was extracted from tissue preserved in alcohol with a QIAGEN DNeasy tissue kit (QUIAGEN, Hilden, Germany) and the offspring of broods genotyped for two Z (X)-linked marker loci. Intron 3 of the sex (Z)-linked Triose phosphate isomerase (Tpi) gene was amplified using primers situated in the surrounding exons. Primer sequences, Polymerase Chain Reaction (PCR) conditions and evidence for sex linkage were described previously [171,176]. This intron contains a 33 bp insertion in populations of H. heurippa that is absent in H. melpomene melpomene from the Colombia Andes. This length variation was used to follow segregation of Tpi in backcross broods. Alleles were separated on 2% Metaphor (QIAGEN) agarose gels run for 4h at 125 V and stained with ethidium bromide. All broods proved informative with regard to the segregation of alleles in female offspring, having F1 fathers heterozygous for the Tpi insertion. Primers for the Apterous (ap) gene were developed for Heliconius and shown to be Z (X) -linked in H. melpomene broods (Jiggins unpub.). Sequences for these primers are 5’-TGAATCCTGAATACCTGGAGA-3’ (forward) and 5’GGAACCATACCTGTAAACCC-3’ (reverse), which amplify a 228 bp region of

62

coding sequence in the ap gene, corresponding to position 184 to 412 in a reference sequence of Precis coenia (GENBANK Accession number L42140). Female individuals from Brood 959 were sequenced to search for diagnostic restriction enzyme sites. One site, 31bp from the end of the amplification fragment, was found to be different between alleles derived from H. heurippa and H. melpomene. The difference is due to a single base-pair substitution corresponding to a restriction site recognized by the enzyme Dde1 (5’ to 3’ CTNAG). The ap fragment was amplified using a PCR profile of 94qC for 3 min, followed by 94qC for 30 s, 55qC for 40 s, and 72qC for 1 min for 35 cycles. Amplified DNA was digested with Dde1 (0.025 units/Pl) for 3 h at 37 qC. Digestion products were separated by electrophoresis on 2% Metaphor agarose gel and stained with ethidium bromide. As the grandparents of brood 959 were not available for genotyping, two wild-caught individuals each of H. m. melpomene and H. heurippa were genotyped to confirm that the cut site allele segregating in my cross was derived from the H. melpomene population. This marker was informative in the broods backcrossed to H. melpomene, but not those to H. heurippa, presumably because the restriction site was polymorphic in the H. melpomene population from which parental stocks were derived.

63

SUPPLEMENTARY MATERIAL Appendix 2.1 Hatch rate of eggs laid by the female offspring of backcross broods. Question marks in the Tpi and ap genotype column indicate cases in which the individual was not preserved for genetic analysis. M: H. melpomene allele; H: H. heurippa allele. No. of days in cage indicates the number of days that a female was kept alive after mating.

Brood No. No. No. Hatch Tpi ap of female parent of days in cage of eggs laid rate genotype genotype a. Offspring of backcross to H. heurippa H x (H x M) 146 27 0 M H 146 27 124 23 H H 146 30 0 M H 146 30 10 40 H H 148 34 17 65 M H 148 26 13 100 H H 148 36 5 0 M M 148 30 12 100 H H 148 21 0 ? ? 150 33 14 100 M H 152 40 25 72 H H 152 40 16 81 H H 152 43 8 50 M H 152 25 19 53 H H 152 25 7 14 H H 152 21 0 M H 152 32 0 M H 155 32 28 50 H H 155 30 3 33 M H 257 37 5 60 H H 257 30 23 30 ? ? 257 33 0 M H 258 13 4 100 ? ? 258 30 8 100 ? ? 258 43 7 42 ? ? 258 30 6 0 M H b. Offspring of backcross to H. melpomene M x (H x M) 956 28 6 83 H H 956 20 16 43 H H 956 16 13 0 H H 956 21 12 33 H M 956 20 39 48 M M 956 29 25 52 M M 956 20 10 10 H H 64

956 956 956 956 956 956 959 959 959 959 959 959 959 959 959 959 959 959 959 959 959 959 959 958 958 958 958

26 25 17 20 20 17 20 18 30 16 23 27 30 24 36 36 27 24 15 23 35 17 20 16 23 30 27

4 6 7 10 21 1 8 19 16 3 24 8 11 13 12 16 20 24 20 6 21 9 5 8 10 10 27

50 100 0 100 66 0 0 0 18 0 37 100 54 84 66 0 0 37 40 0 14 0 0 100 0 80 10

? H ? M M H ? H M H M M M H M M M M M ? M M H M M H M

? M ? M M H ? H M M M M M M M H M M M ? M M H M M H H

Appendix 2.2 Segregation of sterility phenotypes and marker genotypes in backcross to H. melpomene broods. The existence of sterility phenotypes in this brood provides evidence for recessive autosomal sterility factors (see text). A test of heterogeneity on this table is not significant (G3= 2.30, n.s.), although there is a tendency towards an association of fertility with the melpomene Tpim apm genotypes.

Genotype Tpih apm Tpih aph Tpim apm Tpim aph

Fertile 3 4 13 1

Sterile 1 4 3 1

65

References 1. Grant V (1981) Plant Speciation. New York: Columbia Univ. Press. 2. Coyne JA, Orr HA (2004) Speciation. Sunderland, MA: Sinauer Associates. 3. Rieseberg LH (1991) Homoploid reticulate evolution in Helianthus (Asteraceae): evidence from ribosomal genes. Am J Bot 78: 1218-1237. 4. Turelli M, Barton N, Coyne J (2001) Theory and speciation. Trends Ecol Evol 16: 330-343. 5. Buerkle CAM, R. J. Asmussen, M. A. and Rieseberg, L. H. (2000) The likehood of homoploid hybrid speciation. Heredity 84: 441-451. 6. Rieseberg LH (1997) Hybrid origins of plant species. Ann Rev Ecol Syst 28: 359389. 7. Anderson E (1948) Hybridization of the habitat. Evolution 2: 1-9. 8. Templeton AR (1981) Mechanisms of speciation-- a population genetic approach. Ann Rev Ecol Syst 12: 23-48. 9. Shaw DD, Wilkinson P, Coates DJ (1983) Increased chromosomal mutation rate after hybridization between two subspecies of grasshoppers. Science 220: 1165-1167. 10. Rieseberg LH, Van Fossen C, Desrochers AM (1995) Hybrid speciation accompanied by genomic reorganization in wild sunflowers. Nature 375: 313316. 11. Barton NH, Hewitt GM (1985) Analysis of hybrid zones. Ann Rev Ecol Syst 16: 113-148. 12. McCarthy EM, Asmussen MA, Anderson wW (1995) A theoretical assessment of recombinational speciation. Heredity 74: 502-509. 13. Johnson N (2000) Speciation: Dobzhanzky-Muller incompatibilities, dominance and gen interactions. Trends Ecol Evol 15: 480-481. 14. Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation. Genetics 154: 1663-1679. 15. Orr HA, Turelli M (2001) The evolution of postzygotic isolation: Accumulating Dobzhansky-Muller incompatibilities. Evolution 55: 1085-1094. 16. Bateson W (1909) Heredity and variation in modern lights.; Seward A, editor. Cambridge: Cambridge Univ. Press. 17. Dobzhansky TH (1936) Studies on hybrid sterility. II. Localisation of sterility factors in Drosophila pseudoobscura hybrids. Genetics 21: 113-135. 18. Muller HJ (1942) Isolating mechanisms, evolution and temperature. Biol Symp 6: 71-125. 19. Orr HA (1996) Dobzhansky, Bateson, and the genetics of speciation. Genetics 144: 1331-1335. 20. Haldane JBS (1922) Sex ratio and unisexual sterility in hybrid animals. J Genet 12: 101-109. 21. Turelli M, Orr HA (1995) The Dominance Theory of Haldanes Rule. Genetics 140: 389-402. 22. Wu C-I, Johnson NA, Palopoli MF (1996) Haldane's rule and its legacy: why are there so many sterile males? Trends Ecol Evol 11: 281-284. 23. Tao Y, Hartl DL (2003) Genetic dissection of hybrid incompatibilities between Drosophila simulans and D. mauritiana.III. Heterogeneous accumulation of hybrid incompatibilities, degree of dominance, and implications for Haldane's rule. Evolution 57: 2580-2598. 24. Burke JM, Arnold ML (2001) Genetics and the Fitness of Hybrids. Ann Rev Genet 35: 31-32. 66

25. Harrison RG (1990) Hybrid zones: windows on evolutionary process. In: Futuyma D, Antonovics J, editors. Oxford Surveys in Evolutionary Biology. Oxford: Oxford University Press. pp. 69-128. 26. Barton NH (2001) The role of hybridization in evolution. Molec Ecol 10: 551-568. 27. Rieseberg LH, Sinervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene interactions in hybrid speciation: evidence from ancient and experimental hybrids. Science 272: 741-745. 28. Burke JM, Carney SE, Arnold ML (1998) Hybrid fitness in the Louisiana irises: Analysis of parental and F1 performance. Evolution 52: 37-43. 29. Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and speciation. Heredity 83: 363-372. 30. DeVicente MC, Tanksley SD (1993) QTL analysis of transgressive segregation in an interespecific tomato cross. Genetics 134: 585-596. 31. Bell M, Travis MP (2005) Hybridization, transgressive segregation, genetic covariation, and adaptative radiation. Trends Ecol Evol 20: 358-361. 32. Rick CM, Smith PG (1953) Novel variation in tomato species hybrids. Am Nat 87: 359-375. 33. Vega U, Frey KJ (1980) Transgressive segregation in inter- and intraspecific crosses of barley. Euphytica 29: 585-694. 34. Rieseberg LH, Widmer A, Arntz AM, Burke JM (2003) The genetic architecture necessary for transgressive segregation is common in both natural and domesticated populations. Philos Trans Roy Soc Lond B 358: 1141-1147. 35. Albertson RC, Kocher TD (2005) Genetic architecture sets limits on transgressive segregation in hybrid cichlid fishes. Evolution 59: 686-690. 36. Albertson RC, Streelman JT, Kocher TD (2003) Directional selection has shaped the oral jaws of Lake Malawi cichlid fishes. Proc Natl Acad Sci, USA 100: 52525257. 37. Mavárez J, Salazar C, Bermingham E, Salcedo C, Jiggins CD, et al. (2006) Speciation by hybridization in Heliconius butterflies. Nature 441:868-871. 38. Howarth DG, Baum DA (2005) Genealogical evidence of homoploid hybrid speciation in an adaptative radiation of Scaevola (Goodeniaceae) in the Hawaiian islands. Evolution 59: 948-961. 39. Dowling TE, Secor CL (1997) The role of hybridization and introgression in the diversification of animals. Ann Rev Ecol Syst 28: 593-619. 40. O'Foighil D, Hilbish TJ, Showman RM (1996) Mitochondrial gene variation in Mercenaria clam sibling species reveals a relict secondary contact zone in the western Gulf of Mexico. Mar Biol 126: 675-683. 41. Goodfriend GA, Gould SJ (1996) Paleontology and chronology of two evolutionary transitions by hybridization in the Bahamian land snail Cerion. Science 274: 1894-1897. 42. Gould SJ, Woodruff DS (1990) History as a cause of area effects: an illustration from Cerion on Great Inagua. Bahamas. Biol J Linn Soc 40: 67-98. 43. Lanham UN (1981) Some Colorado Andrena of the subgenus Scaphandrena of presumed hybrid origin, with special reference to the tarsal claws (Hymenoptera: Apoidea). J Kans Entomol Soc 54: 537-546. 44. Lanham UN (1984) The hybrid swarm of Andrena (Hymenoptera: Apoidea) in Western North America: a possible source for the evolutionary origin of a new species. J Kans Entomol Soc 57: 197-208.

67

45. Sperling FAH, Harrison RG (1994) Mitochondrial DNA variation within and between species of the Papilio machaon group of swallowtail butterflies. Evolution 48: 408-422. 46. Salazar C, Jiggins CD, Arias CF, Tobler A, Bermingham E, et al. (2005) Hybrid incompatibility is consistent with a hybrid origin of Heliconius heurippa Hewitson from its close relatives, Heliconius cydno Doubleday and Heliconius melpomene Linnaeus. J Evol Biol 18: 247-256. 47. Salazar C., Jiggins CD, Taylor J, Kronforst M, Linares M (2006) Resolving Introgression and Hybrid Speciation among Heliconius Butterflies. Genetics submitted. 48. Schwarz D, Matta BM, Shakir-Botteri NL, McPheron BA (2005) Host shift to an invasive plant triggers rapid animal hybrid speciation. Nature 436: 546-549. 49. Lieder U (1983) introgression as a factor in the evolution pf polytypical plankton cladocera. Int Revue ges Hydrobiol 68: 269-284. 50. Hebert PDN (1985) Interspecific hybridization between cyclic parthenogens. Evolution 39: 216-220. 51. Lieder U (1987) The possible origin of Daphnia cucullata procurva Poppe1887 in the lakes of the Pomeranian Lakeland by hybridization in the past. Hydrobiologia 145: 201-211. 52. Schwenk K, Spaak P (1995) Evolutionary and Ecological consequences of interspecific hybridization in cladocerans. Experientia 51: 465-481. 53. Taylor DJ, Hebert PDN (1992) Daphnia galeata mendotae as a cryptic species complex with interspecific hybrids. Limnol Oceanogr 37: 658-665. 54. Taylor DJ, Heber PDN (1993) Cryptic intercontinental hybridization in Daphnia (Crustacea): the ghost of introductions past. Proc Roy Soc Lond B 254: 163-168. 55. Taylor DJ, Hebert, P. D. N. and Colourne, J. K. (1996) Phylogenetics and evolution of the Daphnia longispina group (Crustacea) based on 12S rDNA sequence and allozyme variation. Mol Phylogenet Evol 5: 495-510. 56. Mina MV (1992) An interpretation of diversity in the Salmonid genus Brachymystax (Pisces, Salmonidae) as the possible result of multiple hybrid speciation. J Ichthyol 32: 117-122. 57. Osinov AG (1993) Countercurrent dispersal, secondary contact, and speciation in lenoks of the genus Brachymystax (Salmonidae: Salmoniformes). Genetika 29: 654-669. 58. Crabtree CB, Buth DG (1987) Biochemical systematics of the catostomid genus Catostomus : assessment of C. clarki, C. plebeius and C. discobolus including the Zuni sucker, C. d. yarrowi. Copeia 1987: 843-854. 59. Smith GR, Hall JG, Koehn RK, Innes DJ (1983) Taxonomic relationships of the Zuni mountain sucker Catostomus discobolus yarrowi. Copeia 1983: 37-48. 60. Miller RR, Smith GR (1981) Distribution and evolution of Chasmistes (Pisces: Catostomidae) in western North America. Occ Pap Mus Zool Univ Mich 696: 146. 61. Dowling TE, DeMarais BD (1993) Evolutionary significance of introgressive hybridization in cyprinid fishes. Nature 362: 444-446. 62. DeMarais BD, Dowling TE, Douglas MM, Minckley W. L., Marsh PC (1992) Origin of Gila seminuda (Teleostei: Cyprinidae) through introgressive hybridization : Implications for evolution and conservation. Proc Natl Acad Sci, USA 89: 2747-2751.

68

63. Meagher S, Dowling TE (1991) Hybridization between the Cyprinid fishes Luxilus albeolus, L. cornutus, and L. cerasinus with comments on the proposed hybrid origin of L. albeolus. Copeia 4: 979-991. 64. Menzel BW (1977) Morphological and electrophoretic identification of a hybrid cyprinid fish, Notropis cerasinus x Notropis c. cornutus, with implications on the evolution of N. albeolus. Comp Biochem Physiol 57B: 215-218. 65. Menezes NA, Weitzman SH (1990) Two new species of Mimagoniates (Teleostei: Characidae: Glandulocaudinae). their phylogeny and biogeography and a key to the glandulocaudin fishes of Brazil and Paraguay. Proc Biol Soc Wash 103: 380426. 66. Goodfellow WN, Hocutt CH, Morgan RP, Stauffer JR (1984) Biochemical assessment of the taxonomic status of "Rhinichthys bowersi" (Pisces: Cyprinidae). Copeia 1984: 652-659. 67. Stauffer JR, Hocutt CH, Denoncourt RF (1978) Status and distribution of the hybrid Nocommis micropogon x Rhinichthys cataractae, with a discussion of hybridization as a viable mode of vertebrate speciation. Am Midl Nat 101: 355365. 68. Stauffer JR, Hocutt CH, Mayden RL (1997) Pararhinichthys, a new monotypic genus of minnows (Teleostei: Cyprinidae) of hybrid origin from eastern North America. Ichthyol Explor Freshwaters 7: 327-336. 69. Smith PF, Konings AD, Kornfield IRV (2003) Hybrid origin of a cichlid population in Lake Malawi: implications for genetic variation and species diversity. Mol Ecol 12: 2497-2504. 70. Salzburger W, Baric S, Sturmbauer C (2002) Speciation via introgressive hybridization in East African cichlids? Mol Ecol 11: 619-625. 71. Meyer A, Salzburger W, Schartl M (2006) Hybrid origin of swordtail species (Teleostei: Xiphophorus clemenciae) driven by sexual selection. Mol Ecol 15: 721-731. 72. Greer GJ, Kitikoon V, Lohachit C (1989) Morphological and life cycle of Schistosoma sinensium Pao, 1959, from northwest Thailand. J Parasitol 75: 98101. 73. Hirai H, Taguchi, T., Saitoh, Y., Kawanaka, M., Sugiyama, H., Habe, S., Okamoto, M., Hirata, M., Shimada, M., Tiu, W. U., Lai, K., Upatham, E. S., and Agatsuma, T. (2000) Chromosomal differentiation of the Schistosoma japonicum complex. Int J Parasitol 30: 441-452. 74. Jones TR, Routman EJ, Begun DJ, Collins JP (1995) Ancestry of an isolated subspecies of salamander, Ambystoma tigrinum stebbinsi Lowe: the evolutionary significance of hybridization. Mol Phylog Evol 4: 194-202. 75. Hutchinson MN, Donnellan SC (1992) Taxonomy and genetic variation in the Australian Lizards of the genus Pseudemoia (Scincidae: Lygosominae). J Nat Hist 26: 215-264. 76. Johnston RF (1969) Taxonomy of house sparrows and their allies in the Mediterranean basin. Condor 71: 129-139. 77. Roy MS, Geffen E, Smith D, Ostrander EA, Wayne RK (1994) Patterns of differentiation and hybridization in North American wolflike canids, revealed by analysis of microsatellite loci. Molec Bio Evol 11: 553-570. 78. Wayne RK, Jenks SM (1991) Mitochondrial DNA analysis implying extensive hybridization of the endangered red wolf Canis rufus. Nature 351: 565-568.

69

79. Yonekawa H, Moriwaki K, Gotoh O, Miyashita N, Matsushima Y, et al. (1988) Hybrid origin of Japanese Mice "Mus musculus molossinus¨: Evidence from Restriction Analysis of Mitochondrial DNA. Molec Bio Evol 5: 63-78. 80. McFadden CS, Hutchinson B (2004) Molecular evidence for the hybrid origin of species in the soft coral genus Alcyonium (Cnidaria: Anthozoa: Octocorallia). Mol Ecol 13: 1495-1505. 81. Taylor DJ, Herbert PDN (1993) Habitat dependent hybrid parentage and differential introgression between neighboringly sympatric Daphnia species. Proc Natl Acad Sci, USA 90: 7079-7083. 82. Taylor DJ, Herbert PDN (1993) A reappraisal of phenotypic variation in Daphnia galeata mendotae : the role of interspecific hybridization. Can J Fish Aquat Sci 50: 2137-2146. 83. Nowak RM (1979) North American Quaternary Canis Monograph 6.; Kansas Uo, editor. Lawrence: Museum of Natural History. 84. Nowak RM (1992) The red wolf is not a hybrid. Conserv Biol 6: 593-595. 85. Wayne RK (1992) On the use of morphologic and molecular genetic characters to investigate species status. Conserv Biol 6: 590-592. 86. Berlocher SH, Feder JL (2002) Sympatric speciation in phytophagous insects: moving beyond controversy? Ann Rev Entomol 47: 773-815. 87. Ramsay M, Williamson R, Estivill X, Wainwright BJ, Ho MF, et al. (1993) Haplotype analysis to determine the position of a mutation among closely linked DNA markers. Hum Mol Genet 2: 1007-1014. 88. Berlocher SH (2000) Radiation and divergence in the Rhagoletis pomonella species group: inferences from allozymes. Evolution 54: 543-557. 89. Seehausen O (2004) Hybridization and adaptive radiation. Trends Ecol Evol 19: 198-207. 90. Grant PR, Grant BR (1992) Hybridization of bird species. Science 256: 193-197. 91. Grant PR, Grant BR (1994) Phenotypic and genetic effects of hybridization in Darwin's finches. Evolution 48: 287-296. 92. Grant B, Grant PR (2003) What Darwin's finches can teach us about the evolutionary origin and regulation of biodiversity. BioScience 53: 965-975. 93. Kornfield I, Smith PF (2000) African cichlid fishes: models systems for evolutionary biology. Ann Rev Ecol Syst 31: 163-196. 94. Danley PD, Kocher TD (2001) Speciation in rapidly diverging systems:lessons from Lake Malawi. Molec Ecol 10: 1075-1086. 95. Danley PD, Market JA, Arnegard ME, Kocher TD (2000) Divergence with gene flow in the rock-dwelling cichilds of lake Malawi. Evolution 54: 1725-1737. 96. Knight ME, Turner GF (1999) Reproductive isolation among closely related Lake Malawi cichlids: can males recognize conspecific females by visual clues? Anim Behav 58: 761-768. 97. Parker A, Kornfield I (1997) Evolution of the mitochondrial DNA control region in the mbuna (Cichlidae) species flock of Lake Malawi, East Africa. J Molec Evol 45: 70-83. 98. Markert JA, Danley PD, Arnegard ME (2001) New markers for new species: microsatellite loci and the East African cichlids. Trends Ecol Evol 16: 100-107. 99. Seehausen O, Alphen JJM, Witte F (1997) Effects of water transparency and light regime on sexual selection and the diversity of haplochromine cichlids. Science 277: 1808-1811. 100. Stauffer JR, Bowers NJ, Kocher TD, McKaye K. R. (1996) Evidence of hybridization between Cyanotilapia afra and Psuedotropheus zebra (Teleostei: 70

Cichlidae) following an intralacustrine translocation in Lake Malawi. Copeia: 203-208. 101. Ruber L., Meyer A, Sturmbauer C, Verheyen E (2001) Population structure in two sympatric species of the Lake Tanganyika cichlid tribe Eretmodini: evidence for introgression. Molec Ecol 10: 1207-1225. 102. McElroy DM, Kornfield I (1993) Novel jaw morphology in hybrids between Pseudotropheus zebra and Labeotropheus fullerborni (Teleostei: Cichlidae) from Lake Malawi, Africa. Copeia 1993: 933-945. 103. Markert JA, Arnegard ME, Danley PD, Kocher TD (1999) Biogeography and population genetics of the Lake Malawi cichlid Melanochromis auratus: habitat transience, philopatry and speciation. Molec Ecol 8: 1013-1026. 104. Gilbert LE (2003) Adaptive novelty through introgression in Heliconius wing patterns: Evidence for shared genetic 'tool box' from synthetic hybrid zones and a theory of diversification. In: Boggs CL, Watt WB, Ehrlich PR, editors. Ecology and Evolution Taking Flight: Butterflies as Model Systems. Chicago: Univ. of Chicago Press. 105. Jiggins CD, Naisbit RE, Coe RL, Mallet J (2001) Reproductive isolation caused by colour pattern mimicry. Nature 411: 302-305. 106. Mallet J, Gilbert LE (1995) Why are there so many mimicry rings? Correlations between habitat, behaviour and mimicry in Heliconius butterflies. Biol J Linn Soc 55: 159-180. 107. Smiley JT (1978) Plant chemistry and the evolution of host specificity: new evidence from Heliconius and Passiflora. Science 201: 745-747. 108. Mallet J, McMillan WO, Jiggins CD (1998) Mimicry and warning colour at the boundary between races and species. In: Howard DJ, Berlocher SH, editors. Endless Forms: Species and Speciation. New York: Oxford University Press. pp. 390-403. 109. Posada D, Crandall KA (2001) Intraspecific gene genealogies: trees grafting into networks. Trends Ecol Evol 16: 37-45. 110. Linder CR, Rieseberg LH (2004) Reconstructing patterns of reticulate evolution in plants. Am J Bot 9: 1700-1708. 111. Stephens M (2003) Inference under the coalescent. In: Balding DJ, Bishop M, Cannings C, editors. Hanbook of Statistical Genetics. Chichester: John Wiley and Sons Ltd. pp. 636-660. 112. Maddison WP (1997) Gene trees in species trees. Syst Biol 46: 523-536. 113. Bandelt H-J, Dress WM (1992) Split decomposition: a new and useful approach to phylogenetic analysis of distance data. Mol Phylog Evol 1: 242-252. 114. Rieseberg LH, Raymond O, Rosenthal DM, Lai Z, Living-stone K, et al. (2003) Major Ecological Transitions in Wild Sunflowers Facilitated by Hybridization. Science 301: 1211-1216. 115. Sattath S, Tversky A (1977) Additive similarity trees. Psychometrika 42: 319-345. 116. Huson DH (1998) SplitsTree: a program for analyzing and visualizing evolutionary data. Bioinformatics 14: 68-73. 117. Bandelt H-J, Forster P, Rohlf A (1999) Median-Joining Networks for Inferring Intraspecific Phylogenies. Mol Biol Evol 16: 37-48. 118. Xu SZ (2000) Phylogenetic analysis under reticulate evolution. Molec Biol Evol 17: 897-907. 119. Bryant D, Moulton V. NeighborNet: an agglomerative method fro the construction of planar phylogenetic networks.; 2002; WABI, Rome, Italy. Lecture Notes in Computer Science. pp. 375-391. 71

120. Makarenkov V (2001) T-REX: reconstructing and visualizing phylogenetic trees and ticulation networks. Bioinformatics 17: 664-668. 121. Makarenkov V, Legendre P (2004) From phylogenetic tree to a reticulated network. J Comput Biol 11: 195-212. 122. Clement MD, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 9: 1657-1660. 123. Bandelt HJ, Macaulay V, Richards M (2000) Median networks: speedy construction and greedy reduction, one simulation, and two case studies from human mtDNA. Molec Phylog Evol 16: 8-28. 124. Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491. 125. Huber kT, Watson EE, Hendy MD (2001) An algorithm for constructing local regions in a phylogenetic network. Molec Phylog Evol 19: 1-8. 126. Nakhleh L, Sun J, Warnow T, Linder R, Moret BME, et al. Reconstructing reticulate evolution in species: theory and practice.; 2004; San Diego, California, USA. pp. 337-346. 127. Rokas. A., Williams BL, King N, Carroll SB (2003) Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature 425: 798-804. 128. Nichols R (2001) Gene trees and species trees are not the same. Trends Ecol Evol 16: 358-364. 129. Arvestad L, Berglund AC, Lagergren J, Sennblad B (2003) Bayesian gene/species tree reconciliation and orthology analysis using MCMC. Bioinformatics 19: i7i15. 130. Machado CA, Hey J (2003) The causes of phylogenetic conflict in a classic Drosophila species group. Proc R Soc Lond B Biol Sci 270: 1193-1202. 131. Posada D, Crandall KA (2001) Evaluation of methods for detecting recombination from DNA sequences: computer simulatons. Proc Natl Acad Sci, USA 98: 13757-13762. 132. Kliman RM, Andolfatto P, Coyne JA, Depaulis F, Kreitman M, et al. (2000) The population genetics of the origin and divergence of the Drosophila simulans complex species. Genetics 156: 1913-1931. 133. Hey J, Machado CA (2003) The study of structurated populations - new hope for a difficult and divided science. Nat Rev Genet 4: 535-543. 134. Rosenberg NA, Nordborg M (2002) Genealogical trees, coalescent theory and the analysis of genetic polymorphisms. Nat Rev Genet 3: 380-390. 135. Hey J (2005) On the Number of New World Founders: A Population Genetic Portrait of the Peopling of the Americas. PLoS BioL 3: 0965-0975. 136. Won YJ, Hey J (2005) Divergence population genetics of chimpanzees. Molec Biol Evol 22: 297-307. 137. Won YJ, Sivasundar A, Wang Y, Hey J (2005) On the origin of Lake Malawi cichlid species: a population genetic analysis of divergence. Proc Natl Acad Sci, USA 102: 6581-6586. 138. Hey J., Nielsen R. (2004) Multilocus methods for estimating population sizes, migration rates and divergence time, with applications to the divergence of Drosophila pseudoscura and D. persimilis. Genetics 167: 747-760. 139. Machado CA, Kliman RM, Markert JA, Hey J (2002) Inferring the history of speciation from multilocus DNA sequence data: The case of Drosophila pseudoobscura and close relatives. Mol Biol Evol 19: 472-488. 72

140. Machado CA, Robbins N, Gilbert MTP, Herre EA (2005) Critical review of host specificity and its coevolutionary implications in the fig/fig-wasp mutualism. Proc Natl Acad Sci, USA 102: 6558-6565. 141. Beerli P, Felsenstein J (2001) Maximum likehood estimation of a migration matrix and effective population sizes in n subpopulations by using a coalescent approach. Proc Natl Acad Sci, USA 98: 4563-4568. 142. Gow J, Peichel CL, Taylor EB (2006) Contrasting hybridization rates between sympatric three-spined sticklebacks highlight the fragility of reproductive barriers between evolutionary young species. Molec Ecol 15: xxx-xxx. 143. Rannala B, Yang Z (2003) Bayes Estimation of Species Divergence Times and Ancestral Population Sizes Using DNA Sequences from Multiple Loci. Molec Ecol 164: 1645-1656. 144. Mallet J (2005) Hybridization as an invasion of the genome. Trends Ecol Evol 20: 229-237. 145. Vaha J-P, Primmer CR (2006) Efficiency of model-based Bayesian methods for detecting hybrid individuals under different hybridization scenarios and with different numbers of loci. Molec Ecol 15: 63-72. 146. Gross BL, Rieseberg LH (2004) The Ecological Genetics of Homoploid Hybrid Speciation. J of Heredity 96: 241-252. 147. Neff NA, Smith GR (1978) Multivariate analysis of hybrid fishes. Syst Zool 28: 176-196. 148. Humphries JM, Bookstein FL, Chernoff B, Smith GR, Elder RL, et al. (1981) Multivariate discrimination by shape in realtion to size. Syst Zool 30: 291-308. 149. Rosenthal DM, Rieseberg LH, Donovan LA (2005) Re-creating Ancient Hybrid Species Complex Phenotypes from Early-Generation Synthetic Hybrids: Three Examples Using Wild Sunflowers. Am Nat 166: 26-41. 150. Hudson RR, Coyne JA (2002) Mathematical consequences of the genealogical species concept. Evolution 56: 1557-1565. 151. Pääbo S (2003) The mosaic that is our genome. Nature 421: 409-412. 152. Takahata N, Slatkin M (1984) Mitochondrial gene flow. Proc Natl Acad Sci, USA 81: 1764-1767. 153. Bullini L (1994) Origin and evolution of animal hybrid species. Trends Ecol Evol 9: 422-426. 154. Mayr E (1963) Animal Species and Evolution. Cambridge, Mass.: Harvard University Press. 155. Arnold ML (1997) Natural Hybridization and Evolution. New York: Oxford University Press. 215 p. 156. Ting C, Tsaur S, Wu C (2000) The phylogeny of closely related species as revealed by the genealogy of a speciation gene, Odysseus. Proc Nat Acad, USA 97: 5313-5316. 157. Tao Y, Zeng Z, Li J, Hartl DL, Laurie C (2003) Genetiic dissection of hybrid incompatibilities between Drosophila simulans and Drosophila mauritiana. II. Mapping hybrid male sterility loci on the third chromosome. Genetics 164: 1399-1418. 158. Barbash DA, Awadalla P, Tarone AM (2004) Functional divergence caused by ancient positive selection of a Drosophila hybrid incompatibility locus. PLoS BioL 2: 0839-0848. 159. Presgraves DC, Balagopalan L, Abmayr SM, Orr HA (2003) Adaptative evolution drives divergence of a hybrid inviability gene between two species of Drosophila. Nature 423: 715-719. 73

160. Barbash DA, Siino DF, Tarone AM, Roote JA (2003) A rapidly evolving Mybrelated protein causes species isolation in Drosophila. Proc Nat Acad, USA 100: 5302-5307. 161. Hutter P (2002) X-linked small GTPase and OXPHOS genes are candidates for the genetic basis of hybrid inviability in Drosophila. Dev Genes Evol 212: 504-512. 162. Perez DE, Wu CI, Johnson NA, Wu ML (1993) Genetics of reproductive isolation in the Drosophila simulans clade: DNA marker-assisted mapping and characterization of a hybrid-male sterility gene, Odysseus (Ods). Genetics 134: 261-275. 163. Wittbrodt J, Adam D, Malitschek B, Maueler W, Raulf F, et al. (1989) Novel putative receptor tyrosine kinase encoded by the melanome inducing Tu locus in Xiphophorus. Nature 341: 415-421. 164. Turelli M (1998) The causes of Haldane's rule. Science 282: 889-891. 165. Price T, Bouvier M (2002) The evolution of F1 postzygotic incompatibilities in birds. Evolution 56: 2083-2089. 166. Presgraves DC (2002) Patterns of postzygotic isolation in Lepidoptera. Evolution 56: 1168-1183. 167. Sasa MM, Chippindale PT, Johnson NA (1998) Patterns of postzygotic isolation in frogs. Evolution 52: 1811-1820. 168. Coyne JA, Orr HA (1997) Patterns of speciation in Drosophila revisited. Evolution 51: 295-303. 169. Coyne JA, Orr HA (1989) Patterns of speciation in Drosophila. Evolution 43: 362381. 170. Orr HA (1995) The population genetics of speciation: The evolution of hybrid incompatibilities. Genetics 139: 1805-1813. 171. Beltrán M, Jiggins CD, Bull V, McMillan WO, Bermingham E, et al. (2002) Phylogenetic discordance at the species boundary: gene genealogies in Heliconius butterflies. Mol Biol Evol 19: 2176-2190. 172. Sheppard PM, Turner JRG, Brown KS, Benson WW, Singer MC (1985) Genetics and the evolution of muellerian mimicry in Heliconius butterflies. PhilosTrans R Soc Lond B Biol Sci 308: 433-613. 173. Naisbit RE, Jiggins CD, Mallet J (2003) Mimicry: developmental genes that contribute to speciation. Evol Dev 5: 269-280. 174. Emsley MG (1965) Speciation in Heliconius (Lep., Nymphalidae): morphology and geographic distribution. Zoologica, New York 50: 191-254. 175. Brower AVZ (1994) Phylogeny of Heliconius butterflies inferred from mitochondrial DNA sequences (Lepidoptera: Nymphalidae). Mol Phylogenet Evol 3: 159-174. 176. Jiggins CD, Linares M, Nasbit RE, Salazar C, Yang ZH, et al. (2001) Sex-linked hybrid sterility in a butterfly. Evolution 55: 1631-1638. 177. Bull V, Beltrán M, Jiggins CD, McMillan WO, Bermingham E, et al. (2006) Polyphyly and Gene Flow between Non-sibling Heliconius Species. BMC Biol 4:11. 178. Read A, Nee S (1991) Is Haldane's rule significant? Evolution 45: 1707-1709. 179. Naisbit R, Jiggins CD, Linares M, Salazar C, Mallet J (2002) Hybrid sterility, Haldane’s rule and speciation in Heliconius cydno and H. melpomene. Genetics 161: 1517-1526. 180. Orr HA (1997) Haldane's rule. Ann Rev Ecol Syst 28: 195-218. 181. Laurie CC (1997) The weaker sex is heterogametic: 75 years of Haldane's rule. Genetics 147: 937-951. 74

182. McMillan WO, Jiggins CD, Mallet J (1997) What initiates speciation in passion vine butterflies? Proc Natl Acad Sci, USA 94: 8628-8633. 183. Jiggins CD, McMillan WO, Neukirchen W, Mallet J (1996) What can hybrid zones tell us about speciation? The case of Heliconius erato and H. himera (Lepidoptera: Nymphalidae). Biol J Linn Soc 59: 221-242. 184. Naisbit R, Jiggins CD, Mallet J (2002) Disruptive sexual selection against hybrids contributes to speciation between Heliconius cydno and Heliconius melpomene. Proc R Soc Lond B Biol Sci 268: 1-6. 185. Linares M (1997) Origin of Neotropical mimetic biodiversity from a three-way hybrid zone of Heliconius butterflies. In: Ulrich H, editor. Proceedings of the International Symposium on Biodiversity and Systematics in Tropical Ecosystems, Bonn 1994. Bonn: Zoologishunginstitut und Museum Alexander Koenig. 186. Lewontin RC, Felsenstein J (1965) The robustness of homogenity test in 2xN tables. Biometrics 21: 19-33.

75

CHAPTER III Submitted to Genetics Resolving introgression and hybrid speciation among Heliconius butterflies

ABSTRACT Homoploid hybrid speciation is rare in animals due to the lack of a clear mechanism by which hybrids can generate reproductive isolation from their parents. Heliconius heurippa is a probable hybrid species between Heliconius cydno and Heliconius melpomene. Here I use a multilocus approach drawing upon phylogenetic methods and coalescent model to investigate the origins of the H. heurippa genome. I sequenced a mitochondrial region ( CoI and CoII), a sex-linked locus (Tpi) and four autosomal loci (Dll, inv, w and Sd). H. heurippa was monophyletic at the mtDNA and Tpi loci, but showed a mosaic distribution of alleles derived from both parental lineages at all four autosomal loci. Estimates of genetic differentiation showed H. heurippa to be closer to H. cydno at mtDNA and three autosomal loci, intermediate at Tpi, and closer to H. melpomene at Dll. I attempted to distinguish between a hybrid origin for H. heurippa and ongoing introgressive hybridization as alternative explanations for this shared variation. Although analysis of the data using a neutral coalescent model appears to exclude evolutionary scenarios involving an ancient speciation event with no recurrent gene flow, I was unable to obtain well-resolved estimates of either the migration rates or the numbers of founding individuals. Since the H. heurippa data show patterns of allele sharing similar to those previously used as support for hybrid speciation in other taxa, these analyses should caution against the inference of hybrid speciation based solely on such patterns without explicit testing of alternate hypotheses.

76

INTRODUCTION Hybridization and introgression are important evolutionary forces in plant diversification [6,155]. In particular, polyploid species are well documented in plants and also known in some animal groups [187-189]. In contrast, homoploid hybrid speciation, without a change in chromosome number, is thought to be more difficult and has only been well documented in a handful of plant species [2,6,38,114,190,191] and is considered rare or absent in animals. Nonetheless, in several animal taxa putative examples of hybrid speciation have been proposed. In cichlid fish, transgressive segregation can produce new variants and likely contributes to adaptive radiation [31,89]. Many other cases of apparently hybrid genomes are known, for example, in tephritid flies, Cladocerans, parasites, wolf-like canids, soft corals, Xiphophorus, cyprinid and cichlid fishes [48,55,61,69-71,73,78,80], but direct evidence for a constructive role for hybridization in adaptive divergence is lacking.

One reason why hybrid speciation has been treated with skepticism in animals is the lack of a clear mechanism that would allow a hybrid taxon to become established. In plants, hybrids become isolated from their parental species as a result of several processes, notably uniparental reproduction, distinct flowering time, attracting a different pollinator, occupying different habitat, or spatial isolation due to geographical separation after dispersal [190]. Some of these are unique to plants, while others could occur in animals. However, in none of the proposed examples of hybrid animal speciation has clear evidence been shown of a potential mechanism. One of the best examples is the ‘Lonicera fly’, a species of Rhagoletis with a hybrid genome that has occupied a novel niche feeding on introduced Lonicera plants in North America [48] . It has not been demonstrated, however, whether the inferred steps of hybridization actually generated the Lonicera-adapted phenotype. The major problem, therefore, is that natural hybrids are often unfit as a result of extrinsic or intrinsic hybrid incompatibility, but where they are fertile and viable they must establish reproductive isolation and escape ecological competition with the parental taxa [5].

77

Here I describe a homoploid hybrid animal species in which the role of the hybrid trait in causing reproductive isolation has been well documented [37]. Heliconius colour patterns are aposematic and often mimetic, such that rare colour pattern hybrids are selected against by predators, generating post-mating isolation between divergent forms. Furthermore, patterns play an important role in mate recognition, leading to pre-mating isolation between species and sub-species [192]. The mate preferences that generate this isolation most likely evolve subsequent to the establishment of a novel pattern in a population, making speciation a two-step process. Therefore, if a hybrid pattern can overcome frequency dependent mimetic selection and become established in a population, the novel population then has a high probability of becoming reproductively isolated from its parental species provided the colour pattern change is sufficiently large. Thus, the frequency of hybridization, great diversity of patterns and clear role of pattern shifts in generating reproductive isolation suggests a route for hybrid speciation in these butterflies.

Furthermore, several taxa appear to have hybrid phenotypes and are candidates as potential hybrid species. Most notably H. heurippa, H. timareta and H. pachinus all share elements putatively derived from both H. melpomene and H. cydno [46,104,193]. The latter species are known to hybridize in the wild [104,108], despite strong assortative mating, post-mating isolation and ecological differences [105,107,179,194,195].

The best studied of these is H. heurippa in which crossing experiments have provided evidence for genetic homology of pattern elements shared between H. heurippa, H. cydno and H. melpomene [46]. Furthermore, when H. c. cordula and H. m. melpomene, subspecies that occur near the current range of H. heurippa on the eastern slope of the Colombian Andes, are crossed a virtually identical colour pattern to that of H. heurippa can be created in the laboratory after just a few generations [37]. The phylogenetic position of H. heurippa is unclear, as it has been considered a close relative of H. melpomene based on colour pattern and genitalia [174], but a relative of H. cydno according to mtDNA [175]. Studies of genetic compatibility have shown female hybrid sterility between H. heurippa and H. melpomene, but show complete compatibility between H. heurippa and H. cydno [46]. Finally, behavioral experiments show that the combined hybrid colour pattern of H. heurippa is critical for mate recognition [37]. 78

Removal of either the red element derived from H. melpomene, or the yellow element derived from H. cydno, results in the pattern being less attractive to H. heurippa males. These data therefore provide strong evidence that the Heliconius heurippa colour pattern is a hybrid trait that causes reproductive isolation. Furthermore, microsatellite data show that H. heurippa is a distinct species and not simply a geographic race of H. cydno; H. heurippa is considerably more differentiated than any other geographic populations of H. melpomene or H. cydno sampled in Panama, Colombia and Venezuela [37]. Thus, H. heurippa is arguably one of the most convincing cases of hybrid speciation in animals, but nothing known about the genealogical history of the H. heurippa genome. Here I use multilocus sequence data to investigate the possibility that H. heurippa is a species of hybrid origin.

RESULTS

Description of gene regions I sequenced a mitochondrial region of 1572bp representing 799bp of CoI, 62bp of the tRNALEU and 711bp of CoII from 11 individuals. Including sequences from GenBank, the alignment had 1572 nucleotide sites examined from 69 individuals, of which 155 (10%) were variable. GTR+I+G [196,197] was selected as the most likely model of nucleotide substitution. I obtained 584bp of the Z-linked Triose phosphate isomerase exon 3 (31bp), intron 3 (430 bp) and exon 4 (123 bp), for 11 haplotypes from 11 individuals. Including sequences from GenBank the total alignment had 25 haplotypes from 22 individuals. Identity was confirmed by comparison with reference sequences for H. c. chioneus (AF413788) and H. m. rosina (AF413790).

The four autosomal regions are portions of nuclear loci initially chosen for their potential role in the development of butterfly wing patterns [198]. Three of these, Distal-less, invected and scalloped, are transcription factors [199-201], while the fourth, white, is member of the ommochrome biosynthesis pathway that generates the yellow, orange and red pigments in Heliconius [202]. However there is currently no evidence that any of these genes are involved in pattern divergence in Heliconius, so I consider them as randomly chosen marker loci. I obtained 558bp of Distal-less, corresponding to 79

exon 4/5 (175bp), intron 5 (333bp) and exon 6 (50bp) of Drosophila (NM166689– intron 4 is absent in Heliconius) for 32 alleles from 20 individuals; 439bp of invected exon 2 (67bp), intron 2 (265bp) and exon 3 (107bp) for 42 alleles from 23 individuals; 499bp of white exon 4 (49bp), intron 4 (397bp) and exon 5 (53bp) for 25 alleles from 17 individuals; 483bp of the scalloped gene from exon 7 (13bp), intron 7 (334bp) and exon 8 (136bp) for 22 alleles from 12 individuals.

Species relationships and population genetic parameters None of the three species formed a monophyletic group at all of the genes sampled. In the phylogeny derived from mitochondrial DNA, individuals of the three species fell into three well-supported monophyletic clades (Figure 3.1): 1) an eastern melpomene clade; 2) the cydno clade including all the H. cydno and H. heurippa sampled and seven H. m. melpomene from the eastern Andean foothills; 3) a western melpomene clade. Within the cydno clade, the H. heurippa haplotypes form a monophyletic group with five fixed differences from H. cydno (0.92% net divergence). Genetic diversity in H. heurippa was the lowest of the three species (Table 3.1). Tajima’s D was not significantly different from zero (Table 3.1), suggesting that a small but constant population size rather than a recent bottleneck is more likely to explain the lack of variation in H. heurippa. Interestingly Tajima’s D estimates were significantly negative for two of the three populations of H. melpomene (H. m. rosina and H. m. mocoa) included here for comparison (Table 3.1), possibly reflecting a recent bottleneck or mtDNA selective sweep. As all four autosomal loci showed some evidence for recombination in at least one of the three species (there were insufficient variable sites to obtain a reliable estimate for the sex-linked locus, Tpi Appendix 3.1), I present gene networks rather than bifurcating trees. Among the five nuclear loci, the sex-linked locus Tpi was the only marker that clearly separated all three species. H. c. cordula and H. m. melopomene alleles formed distinct clades separated by five fixed differences and one shared polymorphism, with 1.3% net divergence (Figure 3.2a). H. heurippa alleles also form a distinct cluster (Figure 3.2a) separated by five fixed differences from H. m. melpomene (1.3% net divergence) and by six from H. c. cordula (1.4% net divergence). In concordance with the network groups, FST values showed the species as three distinct populations, with H. heurippa showing greater differentiation from H. c. cordula (FST = 0.791) and H. m. 80

melpomene (FST = 0.719) than that observed between H. c. cordula and H. m. melopomene (FST = 0.498 P<0.05; Table 3.2; Figure 3.2a).

Figure 3.1 Phylogenetic relationships of H. heurippa (Identified with H and the number of individual, see Table S1 for details) with other populations of H. melpomene and H. cydno based on COI and COII sequences. C and M identify H. m. melpomene and H. c. cordula individuals of the putative parental species taken from GenBank (Table S1 and methods). Sequence ID’s beginning with AF and AY indicate GenBank accession numbers for different populations of H. cydno and H. melpomene included for comparison (see methods). H. hecale was used as the outgroup. Branch lengths and probability values (under branches) were estimated using Bayesian analysis (see Methods). Values over the branches indicate bootstrap support derived from a Maximum Parsimony analysis. Countries of origin are identified using the following abbreviations: P = Panama and C = Colombia; Abbreviations of species names are m. = melpomene, c. = cydno. 81

Table 3.1 Summary of genetic polymorphism data for mtDNA sequences in each population. Population

N

S

TW

TW 95% CI

DT

DT |null

H. m. mocoa

22

9

0.0016

0.0014-0.0018

-1.8296

< 0.05*

H. c. chioneus

10

28

0.0064

0.0020-0.0108

0.3504

>0.1

H. heurippa

11

4

0.0009

0.0001-0.0017

-0.83418

>0.1

H. m. melpomene

8

39

0.0099

0.0045-0.0152

-1.5379

>0.05

H. c. cordula

5

9

0.0028

0.0025-0.0031

-0.1974

>0.1

H. m. rosina

10

26

0.0060

0.0037-0.0083

-1.8174

< 0.05*

N, S, TW and DT are the number of individuals, segregating sites, genetic diversity and Tajima’s parameter, respectively. * Indicates statistical significance for departure of DT from neutral expectation at 95%. H. c. weymeri and H. c. chioneus sequences from Colombia were excluded in the polymorphism analysis, because of the low number individuals available in each case (3 and 1, respectively).

Three of the autosomal loci, Dll, inv and w, show a striking pattern in which H. c. cordula and H. m. melpomene are clearly differentiated, but H. heurippa shares variation with both species. At Dll, H. c. cordula and H. m. melpomene (Figure 3.2b) are separated by 10 fixed differences (3% net divergence) and share three polymorphisms. Similarly at inv, H. c. cordula and H. m. melpomene have ten fixed differences (4.6% net divergence) and two shared polymorphisms (Figure 3.2c). At w, one allele of H. m. melpomene was shared with H. c. cordula (M6-1; Figure 3.2d), such that there was only one fixed difference (2% net divergence) and eight shared polymorphisms. Nonetheless, estimates of FST between H. c. cordula and H. m. melpomene were high for all three loci (Dll, 0.621, inv, 0.593 and w, 0.327). In contrast, H. heurippa did not have any fixed differences with either H. c. cordula or H. m. melpomene at any of these three loci. At Dll, H. heurippa was closer to H. m. melpomene (net divergence 0.41%) than H. c. cordula (net divergence 2.28%). At the other two loci, H. heurippa was closer to H. cydno, with inv showing 0.56% and 2.76% divergence with H. c. cordula and H. m. melpomene respectively, while w showed 0.024% and 1% for the same two comparisons. Estimates of FST supported these observations, with Dll showing H. heurippa not significantly differentiated from H. m. melpomene but strongly differentiated from H. c. cordula (Table 3.2), while inv and w showed the opposite pattern with H. heurippa closer to H. c. cordula (Table 3.2).

82

In contrast with the patterns observed in the other five loci, the Sd locus showed no fixed differences among any of the three species (Figure 3.2e). H. c. cordula and H. m. melpomene had seven shared polymorphisms and net divergence of 0.62%. H. heurippa shared 14 and ten polymorphisms with H. c. cordula and H. m. melpomene respectively, representing net divergence of 0.10% and 0.5%. Genetic diversity was generally high in all three species (Table 3.3). Tajima’s D was significantly negative in H. c. cordula (Table 3.3). FST values showed that H. heurippa was more similar to H. c. cordula than to H. m. melpomene (Table 3.2).

Figure 3.2 Allele networks for nuclear genes. Yellow, Red and Blue are H. c. cordula, H. m. melpomene and H. heurippa alleles. Respective alleles are also identified with the letters C, M and H, followed by the individual number and allele number. Black dots are hypothetical ancestors. Sizes of the circles reflect allele frequencies in the population. a. Tpi, b. Dll, c. inv, d. w and e. Sd.

83

Table 3.2 Genetic structure (FST) values for comparisons between the three populations. Population

H. c. cordula

H. c. cordula

(5)

H. heurippa (11) H. m. melpomene (8)

H. heurippa mtDNA

H. m. melpomene

0.905 0.0001* 0.058

0.739

0.44

0.0001*

-

Tpi H. c. cordula

(6)

H. heurippa (11) H. m. melpomene (8)

0.791 0.0001* 0.498 0.0001*

0.719 0.0001*

-

Dll H. c. cordula

(9)

H. heurippa (14) H. m. melpomene (9)

0.479 0.0001* 0.621 0.0001*

0.085 0.066

-

inv H. c. cordula (14) H. heurippa (16) H. m. melpomene(12)

0.107 0.022* 0.593 0.0001*

0.419 0.0001*

-

w H. c. cordula

(7)

H. heurippa

(9)

H. m. melpomene (9)

0.004 0.383 0.327 0.003*

0.176 0.019*

-

Sd H. c. cordula

(7) -

0.037 0.196 0.148 0.119 H. m. melpomene (8) 0.006* 0.020* * Significance at 95% was obtained by bootstrapping (10000 subsamples). H. heurippa

(7)

84

Table 3.3 Summary of genetic polymorphism data for sex-linked and nuclear loci sequences. Data is showed for each population (H. c. cordula, H. heurippa and H. m. melpomene). H. c. cordula locus

n (alleles)

S

TW

TW 95% CI

Tpi

6

9

0.0083

0.0037-0.0129

-0.1548 (-1.4347 to1.5822) 0.4850 (-0.8862 to 0.9428)

0.4150

Dll

9

23

0.0193

0.0101-0.0285

-0.5311 (-1.6696 to1.6159) 0.3300 (-0.7913 to 0.7376)

0.0760

inv

14

51

0.0394

0.0177-0.0571

-1.1859 (-1.6520 to1.6498) 0.1000 (-0.6055 to 0.5822)

0.000001*

w

7

26

0.0281

0.0064-0.0498

-0.3275 (-1.5502 to1.5575) 0.4070 (-0.6332 to 0.6404)

0.1710

sd

7

20

0.0171

0.0128-0.0213

-0.8328 (-1.5533 to1.7872) 0.2440 (-0.7018 to 0.8047)

0.0090*

DT

DT 95% CI

P

DT rec 95% CI

P

H. heurippa locus

n (alleles)

S

TW

TW 95% CI

DT

Tpi

11

2

0.0014

0.0008-0.0036

0.199

(-1.4296 to1.8276) 0.5920 (-1.4296 to 1.6648)

0.5320

Dll

14

29

0.0229

0.0126-0.0331

-0.642

(-1.7616 to1.6671) 0.2950 (-0.7520 to 0.6874)

0.0420

inv

16

42

0.0307

0.0198-0.0416

0.0026

(-1.8007 to1.7247) 0.5590 (-0.6635 to 0.6064)

0.5060

w

9

29

0.0340

0.0246-0.0434

0.1089

(-1.6834 to1.5982) 0.5850 (-0.6231 to 0.6390)

0.6450

sd

7

25

0.0215

0.0127-0.0302

-0.0604 (-1.5947 to1.6325) 0.5260 (-0.6953 to 0.62731)

DT 95% CI

P

DT rec 95% CI

P

0.4530

H. m. melpomene locus

n (alleles)

S

TW

TW 95% CI

DT

Tpi

8

9

0.0078

0.0043-0.0113

1.2504

(-1.6740 to1.7414) 0.9120 (-1.0680 to 0.9984)

0.9890

Dll

9

21

0.0169

0.0048-0.0217

-0.6685 (-1.6864 to1.5708) 0.3010 (-0.7703 to 0.7226)

0.0510

inv

12

29

0.0235

0.0170-0.0300

-1.7198 (-1.7129 to1.6628) 0.024* (-0.6769 to 0.6816)

0.000001*

w

9

31

0.0377

0.0341-0.0413

-1.1847 (-1.6591 to1.6624) 0.1130 (-0.6389 to 0.6184)

0.000001*

sd

8

33

0.0266

0.0161-0.0371

-0.3359 (-1.5973 to1.6506) 0.3790 (-0.6258 to 0.5921)

0.1320

DT 95% CI

P

DT rec 95% CI

P

n, S, TW, DT and DT rec are the number of alleles, segregating sites, genetic diversity and Tajima’s D, calculated without recombination and with recombination, respectively. * Indicates a P=Pr(Dd observed value) statistically significant deviation of Tajima’s D from Hudson 1990 panmixia model [203]. 95% CIs under the model were calculated by coalescent simulations (with fixed S). The two-tailed probability test of significance was derived from simulations (P=0.025)

History of divergence between H. melpomene and H. cydno The Isolation-Migration model, as implemented in [55], was used to infer the population history of H. melpomene melpomene and H. cydno cordula. For this analysis all sampled populations of the two species were included. The IM program uses a coalescent model to estimate population size, time since divergence, and ongoing migration parameters from multilocus sequence data sampled from two sister species. 85

Here, H. melpomene melpomene and H. cydno cordula were estimated to have similar population sizes (§1.3 million individuals), a common ancestral population size t 541, 389 individuals and to have diverged 1.25 million years ago ( Table 3.4). The different loci showed different patterns of gene flow, which was commonly asymmetric. The estimated migration rate from H. c. cordula to H. m. melpomene (m1=1.7x10-5 per generation) was high at mtDNA, as expected from the pattern of shared variation seen in the mtDNA tree. Gene flow estimated in the other direction was considerably lower (m2=0.31x10-6 per generation). Among the remaining loci, Tpi and inv showed zero gene flow in both directions, w and Sd showed moderate gene flow from H. c. cordula to H. m. melpomene but zero in the reverse direction (Table 3.4). Dll had significant gene flow in both directions (Table 3.4).

Isolation-Migration model including H. heurippa The IM model was then used to compare H. heurippa with each of the parental species. The estimated H. heurippa population size was similar in each comparison and considerably smaller than that estimated for the other species (Table 3.4). Unlike the comparison between H. m. melpomene and H. c. cordula, both comparisons involving H. heurippa failed to yield good estimates of the ancestral population size and divergence time. The likelihood surfaces obtained for these parameters were either flat or rising throughout the parameter range investigated. Per-locus migration rate estimates are given in Table 3.4.

86

Table 3.4 Genealogical parameters estimated under the IM model.

T1

Species comparison

H. melpomene-H. cydno

H. cydno - H. heurippa (a)

H. cydno - H. heurippa (b)

H. melpomene- H. heurippa (a)

H. melpomene-H. heurippa (b)

T2

2.25 1.31 4.35 1,310,465

2.39 1.41 7 1,393,588

4.58 2.64 8.70 1,505,195

1.25 0.70 2.32 411,191

****

****

3.85 2.30 8.60 1,556,449

1.69 0.99 3.67 684,094

****

****

mtDNA (x10-6)

t

2.15 1 4.8 1,250,911

4Nm1 1.7 x10-5 1.57 8.78x10-5

Tpi (x10-6)

4Nm2 4Nm1 0.31 0 0.11 1.16x10-4

inv (x10-6)

4Nm2

4Nm1

4Nm2

0

0

0

Dll (x10-6)

w (x10-6)

4Nm1 1.26 0.087 1.03 x10-5*

4Nm2 0.43 0.066 1.65 x10-5*

4Nm1 0.27 0.044 9.61*

~0

~0

4Nm2 ~0

****

0.23 0.59 0,0058 0.21 0 9.48x10-5 2.69x10-5

0

1.01 x10-5 ~0 1.85 2.81 x10-5*

1.05 0.044 2.09*

****

****

0

0

2.67 x10-5 5.76 ~0 1.44 x10-4*

5.61 1.07 0.76 0.42 2.08 x10-4* 4.65*

2.91 x10-5 7.17 ~0 2.57 x10-4

****

5.01 0.60 0.50 0 0.0083 9.20x10-5 3.85x10-5

0

~0

4.74 2.89 0.073 0.083 1.10 x10-5* 6.31*

~0

0

5.48 ~0 0.45 2.50 x10-5*

****

****

****

****

0

0.71 0.060 8.66

~0

0.10 0.010 3.19

0.88 0.044 4.41*

9.16 0.56 3.78x10-5

5.54 x10-5 2.86 3.25 0.057 2,66 x10-4* 1.15x10-5*

Sd (x10-6) 4Nm1 4Nm2 0.40 0.044 ~0 1.03 x10-5*

1.88 3.30 0.13 0.063 1.67 x10-5* 6.17*

****

****

2.26 1.35 0.16 0.17 2.92 x10-5* 1.68 x10-5*

****

****

For comparisons involving H. heurippa two data sets were used, (a) the complete data set and (b) a pared down data set including only those alleles most related to the other species involved in the comparison (see Results). m1 refers to the forward migration rate (per generation) of haplotypes from population 2 into population 1 and m2 refers to migration in the reverse direction. For each cell the first value is the parameter estimate, second is the lower limit and third is upper limit at 95% level. **** indicates that no reliable ML estimate was obtained for a parameter, * unreliable estimate or limit due to flat or incomplete posterior probability distribution sampled and ~0 effectively zero, although the lowest 'bin' does not actually include zero (i.e. low gene flow is probable).

87 88

If H. heurippa did have a hybrid origin, then estimates of ongoing migration might be inflated by the mix of deep and recent coalescence events found in the same population. In other words in a comparison with H. m. melpomene, alleles derived from H. c. cordula might appear to date from an early speciation event while the alleles derived from H. m. melpomene would falsely indicate recent gene flow. In order to control for this possibility, I also ran IM with a pared down data set that included only those H. heurippa sequences which appear to have descended from the parent species included in the pairwise combination. Thus for example, H. c. cordula was compared with only the H. heurippa alleles that fell within the H. cydno clade. Data sets created in this way were used for those loci that clearly group H. c. cordula and H. m. melpomene alleles in independent clusters (Tpi, Dll, inv, w). Unsurprisingly, migration rate parameters were quite different with this pared down data set, but nonetheless significant migration was observed in several comparisons (Table 3.4).

A hybrid speciation coalescent model Since IM cannot model the history of three species or test for the hybrid origin of a species, I developed an alternative method based on a generalization of the structured neutral coalescent implemented in IM. The method explores the distribution of two summary statistics under different historical scenarios for the history of the three species. For these analyses I used a reduced data set that included only the races H. c. cordula and H. m. melpomene that are thought to have given rise to H. heurippa, and also excluded the mtDNA sequences that, probably due to recent introgression, were incompatible with my hypothesis of an origin for H. melpomene and H. cydno that predated that of H. heurippa.

Maximum likelihood estimates (MLEs) for a representative set of simulations in which I assumed symmetric gene flow between the two parental species and the putative hybrid (between Pop 1 and Pop 2 and between Pop 2 and Pop 3) are shown in Table 3.5. Although the numerical values of the MLEs varied substantially depending on the values assigned to the remaining parameters, several features of the likelihood surface were shared both by these calculations and others (not shown) allowing for asymmetric gene flow and different 88

numbers of founders (between 0 and 100). In particular, the log-likelihoods of the multilocus MLE’s of the migration rate parameters consistently fell between –23.3 and – 24.2 for all scenarios involving either ongoing hybridization or hybrid speciation, irrespective of the remaining parameter values and despite the MLEs themselves varying substantially. Only scenarios involving zero gene flow (m12 = m23 = 0) showed significantly lower MLEs, except when the founding event was very recent (T1 = 100,000 generations; Table 3.5).

As examination of the single locus likelihoods reveals (Figure 3.3a,b), the variability of the migration rate estimates appears to be a consequence of between locus differences in the amount of gene flow between H. heurippa and the two putative parent species. Profile likelihood curves for the migration rates for one set of parameter values are shown in Figure 3.3 and are representative of the locus-specific patterns obtained in my other analyses as well (results not shown). Consistent with the IM results, the MLE’s of m12 (x106

) are small for Tpi and Dll (0.0), moderate for inv (0.0-0.6) and Sd (0.0-1.6), and large for

w (1.8-2.0). Likewise, the MLE’s for m23 (x10-6) are small for Tpi and w (0.0), moderate for Dll (0.4-1.2) and inv (0.6-1.6), and large for Sd (0.8-2.0). It is not surprising, therefore, that the multilocus data fail to tell a coherent story about gene flow between these species. Instead, the multilocus likelihood surfaces are relatively flat in the vicinity of the MLE, at least in part because changes in the migration rates lead to increases in some of the single locus likelihoods and decreases in others, the two tending to cancel each other out.

89

Table 3.5 Multilocus maximum likelihood estimates of the migration rate parameters.

N2

T1

f1, f3

m12 (x106)

m23 (x106)

ln(Lk)

4x105

3x105

1,1

1.0

1.2

-23.770

0.0

0.0

-26.687

0.8

0.8

-23.763

0.0

0.0

-25.833

0.4

0.6

-23.500

0.0

0.0

-28.287

0.8

0.6

-23.664

0.0

0.0

-27.555

0.0

0.4

-23.684

0.0

0.0

-24.108

0.0

0.2

-23.749

0.0

0.0

-23.889

0.6

0.6

-23.703

0.0

0.0

-30.730

1.4

1.0

-23.806

0.0

0.0

-30.555

1.2

1.4

-23.394

0.0

0.0

-25.881

0.8

0.8

-23.523

0.0

0.0

-24.741

0.4

0.4

-24.140

0.0

0.0

-30.383

0.4

0.4

-23.988

0.0

0.0

-30.254

10,10

10,0

0,10 4x105

1x105

1,1

10,10

4x10

5

5

6x10

1,1

10,10 7x105

3x105

1, 1

10,10 1x105

3x105

1, 1

10,10

N2 is the population size of population 2 (H. heurippa), while N1 and N3 are treated as fixed parameters (see Methods). T1 refers to the time since the founding of population 2. f1 and f3 refer to the numbers of founders from population 1 and population 2 in a hybrid speciation scenario. Thus 10,0 refers to a non-hybrid speciation involving divergence from population 1, whereas 10,10 refers to a hybrid speciation event involving 10 founders from each parent. m12 and m23 are the migration parameters between population 1 and 2, and 2 and 3 respectively. ln(Lk) is the log likelihood value for this set of parameters given the observed data; the upper value in each cell in this column is the log likelihood at the maximum likelihood estimates of the migration rates, while the lower value is the log likelihood under the assumption of no migration. 90

Figure 3.3 Single locus a, b and multilocus c profile likelihoods for the migration rates m12 and m32, assuming symmetrical migration (m12 = m21; m32 = m23). The profile likelihood for each migration rate was calculated by finding the maximum likelihood when the value of that rate was fixed and the other rate was allowed to vary between 0.0 and 2.0. The remaining parameter values were: f1 = f3 = 1; N2 = 400,000; T1 = 300,000 generations; for the fixed parameters see methods.

91

DISCUSSION Heliconius heurippa was initially identified as a putative hybrid species based on its intermediate colour pattern, which shows a striking similarity to phenotypes produced after just a few generations of hybridization between H. c. cordula and H. m. melpomene [37]. The genealogies presented here provide independent evidence for the hybrid status of the H. heurippa genome. All four autosomal loci showed a pattern in which H. heurippa shares similar alleles with both H. m. melpomene and H. c. cordula. At three of these (inv, w and sd), H. heurippa was most closely related to H. c. cordula, while at the fourth (Dll) it was closer to H. m. melpomene. The remaining loci show H. heurippa as a monophyletic lineage either closest to H. c. cordula (CoI, CoII) or intermediate (Tpi).

Gene flow between H. c. cordula and H. m. melpomene The comparison of H. c. cordula and H. m. melpomene under the IM model indicates ongoing introgressive hybridization at most of the loci studied (except Tpi and inv, m1=m2=0, Table 3.4). In particular there are very closely related mtDNA haplotypes shared between the species (gene flow from H. c. cordula to H. m. melpomene, m1 = 1.7x10-5 per generation; Figure 3.1), moderate gene flow in both directions at one of the autosomal loci, Dll (m1=1.26 x10-6 per generation and m2=0.43 x10-6 per generation), and shared asymmetric gene flow at w and Sd (Table 3.4). These species are known to hybridize in the wild and shared alleles have previously been observed at another autosomal locus, Mpi for which a symmetrical migration rate of 1.54x 10-6 per generation was estimated in Panamanian populations [171,177].

Shared mtDNA variation observed here between H. cydno and H. melpomene in eastern Colombia (Figure 3.1) suggests recent introgression. However given that female hybrids are sterile following Haldane’s rule [179], this is quite a surprising result. The shared mtDNA haplotypes may have been retained as an ancient polymorphism since the speciation of H. melpomene and H. cydno, but this seems unlikely given the evolutionary distance between the species (1.5-2% divergence between the two mtDNA clades). 92

Occasional fertility of F1 females could have provided a route for introgression. Limited fertility has been observed in a cross between a female H. melpomene and a male H. cydno [179]– the wrong direction to explain the introgression observed here, but nonetheless suggestive that F1 female fertility is possible. Indeed, the fact that the H. c. cordula haplotypes are monophyletic relative to H. m. melpomene argues against there being high rates of mitochondrial gene flow between these species. One possibility is that Figure 1 reflects a recent selective sweep by a mtDNA haplotype in both H. m. melpomene and H. c. cydno, probably after the origin of H. heurippa. Such a scenario would be consistent with the negative (albeit statistically insignificant) values of Tajima's D estimated for both populations (Table 3.1) and would explain why the two species proposed to be the parents of the putative hybrid share mitochondrial genomes that are more closely related to one another than either is to H. heurippa. Of course, the picture is further complicated by the mtDNA sequence from individual M5, which - despite having been sampled in an eastern Andean population of H. m. melpomene - is itself most closely related to mtDNA haplotypes from the subspecies H. m. mocoa.

The history of H. heurippa The IM model could not be fitted to the H. heurippa data set, and led to very flat likelihood surfaces for some of the parameter estimates. One possible explanation for this is that H. heurippa is not a sister species of either H. m. melpomene or H c. cordula but instead shares common ancestry with both species following a hybrid speciation event. An important assumption in the IM model is that the two species being compared are sister to one another (Hey 2005 pers. com.), so hybrid speciation would confound the assumptions of the model. Indeed, it seems intuitive that the pattern of allele sharing with both parental species would lead to different alleles having very different coalescence times, and hence to a broad likelihood surface for the estimated time since speciation.

Nonetheless the lack of fit of the data to the IM model, intriguing as it is, does not in itself provide any proof of the hybrid speciation scenario. This led me to develop a three species coalescent model (Figure 3.4) with the aim of discriminating among the following 93

alternative hypotheses: a) shared ancestral variation, b) ongoing levels of introgression and c) hybrid speciation as alternative explanations for the pattern of allele sharing between the study species. Due to the complexity of implementing a full MCMC analysis as used by IM, I instead used a simpler approach whereby I simulated data sets under different parameterizations of the model, calculated summary statistics from the simulated data and used these results to estimate the likelihood of the observed sequence data for each choice of model parameters. The model allowed comparison of the likelihood values under different scenarios involving hybrid speciation or a simple bifurcating origin for H. heurippa with subsequent gene flow.

Figure 3.4 Coalescent model employed in estimating the three species population parameters. For parameter descriptions see methods.

94

Notably, the coalescent model implemented here failed to resolve alternative scenarios for the history of the three species, with the exception that most scenarios involving no gene flow between H. heurippa and the parental species appear to be inconsistent with the sequence data. Although this failure to resolve hypotheses could be a consequence of my use of summary statistics rather than the full sequence data, two observations suggest that the flatness of the multilocus likelihood surfaces (Table 3.5; Figure 3.3c and unpublished data) is not simply an artifact of my statistical methodology. First, these two summary statistics were chosen for my analysis because simulations of the three species coalescent with comparable amounts of sequence data indicated that their distribution should be sensitive to rates of gene flow and founder number. More importantly, examination of the single locus likelihood surfaces revealed that the locations of these MLEs did not depend greatly on my assumptions about the effective population sizes or the timing of the speciation events. However, I did observe that the MLEs differed between the loci analyzed, such that gene flow appears to be negligible for some loci but very large (i.e., 4Nem >> 1) for others (Figure 3.3a,b). Although some of the single locus likelihood profiles are sufficiently flat that the differences in the migration rate MLEs may be attributed to noise, this does not appear to be the case for w in the case of the m12 profiles (Figure 3.3a) or for inv, w, and sd for the m23 profiles (Figure 3.3b), all of which have relatively peaked minima or maxima at zero. Aggregation of these single locus likelihoods produces a flat multilocus likelihood surface around the MLE because changes in the migration rates in this region will increase some of the single locus likelihoods, but decrease the others. Furthermore, this result might have been anticipated from Table 3.6, which shows that the numbers of segregating sites either private to a particular species or shared between pairs of species differs dramatically between loci.

95

Table 3.6 The pattern of segregating sites.

locus

L

samples

s1

s2

s3

s12

s23

s13

sr1

sr3

P

Tpi Dll

313 140

8/11/6 9/14/9

4 3

0 12

4 8

0 1

0 3

0 0

0.00 0.07

0.00 0.15

1.8x10-9 5.4x10-9

inv w sd

202 192 323

12/16/14 9/9/7 8/7/7

9 6 16

9 6 5

18 8 6

2 6 3

7 0 6

0 2 0

0.11 0.50 0.14

0.26 0.00 0.55

7.55x10-9 1.0x10-8 7.7x10-9

L = number of non-coding sites used in the three species coalescent analyses; samples indicates the number of sequences from H. melpomene (Pop 1), H. heurippa (Pop 2), and H. cydno (Pop 3), respectively.; s1, s2, s3 are the numbers of sites segregating only in Pop 1, Pop 2, or Pop 3; s12, s23, and s13 are the numbers of sites segregating in Pop 1 and 2, Pop 2 and 3, or Pop 1 and 3 only; sr1 and sr3 are the two summary statistics (see methods for definition).

There are a number of possible explanations for such interlocus variation. One is that some of the loci sequenced for this study are either themselves subject to natural selection or else are linked to other loci subject to natural selection. Indeed, this seems plausible as I know that there is a strong correlation between the Tpi locus and hybrid sterility in interracial H. melpomene, H. cydno x H. melpomene and H. melpomene x H. heurippa crosses[46,176,179]. If Tpi were associated with genes causing hybrid sterility, this might explain the clear lack of gene flow among the three species at this locus. At the other extreme, Sd shows far more allelic mixing between species than the other loci studied. A similar pattern at another locus, Mpi, has led to suggestions that balancing selection could be maintaining diversity and perhaps promoting introgressive hybridization [171,177] and it is possible that similar processes are occurring at Sd. In summary, although the data clearly imply that introgressive hybridization has occurred between the three species, I am unable to distinguish between the two alternative hypotheses of a) a hybrid founding event for H. heurippa with little subsequent inter-specific gene flow and b) a high ongoing level of introgressive hybridization between the three species. The only scenario that could be clearly rejected is an ancient origin, hybrid or otherwise, with no subsequent gene flow.

Furthermore, the observed pattern of allelic variation is surprising. In the light of previous studies, most notably of Helianthus sunflowers [27,114], the expected pattern following a 96

hybrid speciation event is one of a hybrid genome consisting of genetic “blocks” derived from one or the other of the parental species. These would be seen in my data as a clustering of the hybrid species entirely within one or the other parental species for each locus. However, this is not the general pattern seen in H. heurippa, where I observe allelic variation shared with both parental species at several loci. Intuitively, the latter pattern seems more consistent with high ongoing rates of introgression at many loci, and species differences maintained by selection at the remaining loci. However, to incorporate the effects of selection a more complicated model than that described here would be required.

Overall, the species relationships are consistent with the H. heurippa genome containing a greater contribution from H. cydno than H. melpomene. The H. heurippa mtDNA haplotypes fall within an H. cydno clade, and at three of the five nuclear loci FST values show H. heurippa closer to H. cydno (Table 3.2). This is consistent with what is known about the three species. H. heurippa is a geographic replacement of H. cydno that flies in similar habitats. Where H. cydno flies sympatrically with H. melpomene, the former is associated more with closed canopy forest and tends to be found at higher altitudes [195], both characteristics also observed in H. heurippa populations in eastern Colombia (ML and CJ pers. obs.) Furthermore, laboratory reconstruction of the H. heurippa color pattern from H. c. cordula and H. m. melpomene involves backcrossing F1 male hybrids to H. cydno, with a correspondingly greater contribution from the H. cydno genome. Finally, patterns of hybrid sterility show that H. heurippa is more compatible with H. cydno [37,46].

97

CONCLUSIONS Using sequence data for six gene regions and a combination of phylogenetic and coalescent-based analyses, I have studied the history of H. heurippa and investigated the hypothesis of hybrid speciation. Superficially, the data seem to strongly support the hybrid speciation hypothesis, with Fst analyses showing H. heurippa closer to H. cydno at some loci and closer to H. melpomene at others. This is the pattern that might be expected following a hybrid speciation event, and has been used previously to argue the case for hybrid speciation in other groups [38,71,112]. Furthermore, there are clearly alleles in H. heurippa that could be considered diagnostic for one or other of the parental species at different loci, similar to the ‘private alleles’ found in the ‘Lonicera fly’ that were apparently derived from one or other putative parent. Thus, my data are similar in several aspects to previous examples that have been proposed as good cases for hybrid speciation. Nonetheless, using a neutral coalescence model, I was unable to distinguish the hypothesis of high ongoing gene flow from that of a hybrid origin, or indeed provide good parameter estimates for the population history of H. heurippa.

As far as I am aware, this is the first time that methods based on the coalescent have been applied to the study of a hybrid species, and my results highlight the difficulty of clearly proving the case for hybrid speciation. If hybrid speciation is important, it must necessarily occur in taxa with significant rates of introgressive hybridization, such that where shared variation is observed the alternative hypotheses of hybrid founding versus introgressive hybridization need to be rigorously tested. In part, my failure to distinguish alternative hypotheses is a consequence of statistical methodology: by analyzing summary statistics I have excluded some information that might have better resolved the evolutionary history of these species. This can be seen, for example, in the flatness of the profile likelihoods shown in Figure 3.3. Clearly, further development of these methods, perhaps along the lines already implemented in IM, is needed. However, as my consideration of the single locus likelihoods makes clear, more powerful statistical methods which neglect inter-locus heterogeneity will probably remain inadequate. It may be that my coalescent framework would prove useful with a more extensive genome sampling from which a consistent signal 98

of neutral coalescence might be expected to emerge, although alternatively so much of the genome may be sufficiently linked to sites subject to selection that including more data would merely complicate the picture further.

In summary, the very process of ongoing introgression between H. m. melpomene and H. c.cordula is perhaps what makes the signal of a hybrid origin for a third species so difficult to identify. As in other groups of closely related taxa in which hybridization is frequent, such as Anopheles gambie [204,205] and the Drosophila pseudoobscura group [138,139], discordant patterns are found at different markers such that particular alleles cannot be considered ‘diagnostic’ of a particular species. For this reason, I argue that it is necessary to directly study the history of traits involved in adaptation and reproductive isolation. In the case of H. heurippa, sequence analysis of the genes controlling the different pattern elements would provide a convincing test of their hybrid origin or otherwise. Population genetic data such as that described here could play a role in identifying regions with reduced gene flow that are likely to be associated with adaptation and reproductive isolation.

MATERIALS AND METHODS Specimen collection Butterflies were collected in Colombia and Venezuela (Appendix 3.1), wings removed and stored in glassine envelopes. The bodies were preserved with 100% EtOH in the Universidad de Los Andes (M code) and in DMSO in the Jiggins collection (stri-b code) for H. heurippa (n=11), H. m. melpomene (n=8) and H. c. cordula (n=9) from the eastern Colombian and Venezuelan Andes (Appendix 3.1). Total DNA extractions from a third of a thorax were carried out with the QIAGEN DNeasy tissue kit (QIAGEN, Hilden, Germany).

99

PCR and sequencing methods Cytochrome oxidase subunit 1 (CoI)/tRNALEU/cytochrome oxidase subunit 2 (CoII) region and the sex-linked locus, Triose-phosphate isomerase (Tpi), were amplified from individual genomic DNA using PCR primers and conditions described in Beltrán et. al. [171]. Products were electrophoretically separated on 1.5% low–melting point agarose with ethidium bromide (1 mg/ml). Bands were cut from the gel, dissolved in gelase and 2Pl of this template was used to carry out Dye-terminator sequencing reactions using Big Dye v3.1 kit (ABI Corporation). Cycle sequencing reactions of 10Pl were cleaned over centrisep spin columns with Sephadex® G50 and dried. Amplification products were re-suspended in 4Pl of a 4:0.1 deionized formamide: Crystal violet solution, denatured at 95oC for five minutes and loaded into 5.5% acrylamide gels. Gels were run for 3 hours on an MJ Research BaseStation. Tpi products were cloned before sequencing using pGEMt-T Easy Vector System (Promega). Distal-less (Dll), invected (inv), white (w) and scalloped (sd) also were amplified from individual genomic DNA using PCR primers and conditions as described by Kronforst M [198]. Gel running, cycle sequencing chemistry and clean-up procedure were as above but the samples were resuspended in 12Pl of deionized formamide, denaturated 95oC for five minutes and run on an ABI Prism™ 310 automated DNA sequencer (Applied Biosystems Inc.). These four loci were also cloned before sequencing with the same Promega System. For each individual 3–5 clones were sequenced to identify distinct alleles. Sequences included in the analysis were generally represented by at least two identical clones. Chromatograms were edited and base calls checked using SEQSCAPETM (Applied Biosystems) and SEQUENCHER v4.1 (Gene Codes Corporation, Inc.). Alignments were made using Sequencher, with subsequent manual adjustment and assignment of protein reading-frames carried out in MACCLADE v4.02 [206].

The mitochondrial and Tpi sequences for the same H. c. cordula and H. m. melpomene individuals studied here, were taken from GenBank. (DQ019244-DQ019246, DQ019250,DQ019251, DQ19234-DQ19239 for H. c. cordula and AY548139, DQ019243, DQ019247-DQ019249, DQ019252-DQ019254, DQ019228-DQ019233, DQ019240, 100

AY548151 for H. m. melpomene). In the case of CoI and CoII analysis, I also included H. cydno and H. melpomene sequences from GenBank for comparison: H. c. chioneus (AF413672, AF413707, AF512978, AF512980, AF512985, AF512989, AF512990AF512993, AY548130), H. m. rosina (AF413673, AF413674, AF512971, AF512972, AF512977, AF512982-AF512984, AF512987), H. c. weymeri (AY548114-AY548116), H. m. mocoa (AY548118-AY548129, AY548131-AY548138) and H. hecale (AF413683).

Phylogenetic analysis Mitochondrial (mtDNA), phylogenetic analysis were performed with PAUP* v.4.0b10[207]. A maximum parsimony tree was generated using a heuristic search with TBR branch swapping. Bootstrap values were calculated with 5000 replicates using the same search conditions. MRMODELTEST v2.0 [208] was used to determine the most appropriate model of nucleotide substitution based on hierarchical likelihood ratio tests. The command line for this model was included in a Bayesian analysis with MRBAYES v.3.04 [209] . Four differentially heated Markov chains were initiated from random trees, run for 106 generations and sampled every 100 cycles. Likelihood values were plotted against number of generations to determine the point at which stationarity was reached. All trees sampled before these points were discarded and the remaining tree samples were used to generate a 50% majority rule consensus tree (n =9688). The posterior probability of each clade is provided by the percentage of trees identifying the clade [209].

Recombination complicates interpretation of the species history as different parts of the same loci could have different histories and it is difficult to distinguish homoplasy due to recurrent mutation from that generated by shuffling bases as a consequence of recombination events [130]. Phylogenetic methods such as maximum parsimony (MP) and maximum likelihood (ML) assume an underlying bifurcating tree, an assumption that is violated by intragenic recombination. Therefore phylogenetic trees were reconstructed only for the mtDNA data, as I can reasonably assume that homoplasy is due to recurrent mutation rather than recombination. For nuclear loci I constructed haplotype networks that take into account the presence of persistent ancestral nodes, multifurcations and 101

reticulation. The presence of loops in these networks might reflect recombination events [109]. Networks for nuclear loci were constructed with statistical parsimony in TCS v 1.21[122] (a method that allows the identification of putative recombination events by looking at the spatial distribution in the sequence of the homoplasies defined by the network [210]), considering gaps as missing data and adjusting the parsimony limit to the respective data set. Median-joining networks (a method that does not resolve ties produced by recombination) were also constructed with Network v4.1.1.1[117] (www.fluxusengineering.com) for comparison.

Population genetic analysis Population recombination rates for each nuclear locus were estimated with a coalescent likelihood-based method implemented in the program LDhat [211]. The proportion of permuted data sets with a likelihood equal to or greater than the original data is calculated to establish the statistical significance of the recombination rate [211]. Within and between species genetic variation was assessed for the H. c. cordula, H. m. melpomene and H. heurippa populations as follows. The per site population mutation rate TW [212] with 95% confidence intervals was estimated with DnaSP v4.10.3 [213]. Deviation from a neutral model, and hence the effectiveness of TW in reflecting the effective population size (Ne) was tested by estimating Tajima’s D [214]. Significant departure of Tajima’s D from zero was evaluated both assuming recombination and without recombination. Both tests were carried out because presence of recombinant sites in the nuclear genes leads to a loss of power to reject neutrality [215]. For between species comparisons, the program SITES [216] was used to estimate net divergence between species [217] and the number of shared polymorphisms and fixed differences.

Genetic differentiation between pairs of populations was measured using Wright’s FST [218] adapted for DNA sequence data [219,220] and estimated using DnaSP v4.10.3 [213]. Statistical significance was obtained by bootstrapping i.e. randomly sampling with replacement the values of within population diversity, ʌS, and the values of the between population divergence, ʌB, 10000 times and recalculating FST for each replicate. The P 102

values are based on the proportion of random groupings being equal to or greater than the observed. This bootstrap process was carried out using SEQUENCER v6.1.0 [221]. In order to estimate the role of gene flow in shaping the pattern of shared alleles between species, two different procedures were implemented. First, the Isolation-Migration (IM) model was used for pairwise species comparisons [138] and second a generalization of the IM model was developed here to estimate population parameters for a three species system. The IM model involves a comparison of two species descended from one ancestral population, so is not specifically designed to incorporate a hybrid speciation scenario; however I ran the model on all possible pairs of species to investigate population parameters and in particular estimate rates of gene flow at each locus. In addition, some of the demographic parameters estimated using IM are subsequently used as fixed parameters in my more complex model. In the case of IM, population sizes (T1, T2 and TA), time of divergence (t) and gene flow (m1 and m2) per generation for each locus were estimated for H. m. melpomene and H. c. cordula using the six loci. The same parameter set was estimated for H. melpomene – H. heurippa and H. cydno – H. heurippa pairs. In order to estimate the H. heurippa population size for use in subsequent analyses, an overlap range between 10 runs was estimated (5 melpomene-heurippa comparisons and 5 cydno-heurippa comparisons).

First, gene regions were selected that did not show evidence of recombination, by removing recombinant sections of each alignment after analysis using the Hudson four gamete test [222]. The IM program was then run with a 100,000 step burnin period, 20,218,199 steps for Metropolis coupling algorithm using 10 Markov chains with a geometric increment model over selected gene regions [222]. For each locus two migration rates, m1 and m2 were estimated (–j6 option) and the HKY mutation model was used [223]. Demographic values were obtained through a molecular clock scalar calibration to obtain parameters per base pair and per generation, taking as reference Brower’s 2.3% divergence in mtDNA per million years estimated for insects [224] (i.e. a neutral substitution rate of 1.15% per million years) and with an assumption of four generations per year. This model assumes that the two species arose from a single species some time (t generations) in the past, that 103

the population sizes are constant (but possibly differ between the descendant and ancestral species), and that there has been either bidirectional, unidirectional or no gene flow between the species since population separation.

In the second method, coalescent simulations were used to explore how the distribution of two summary statistics varied with different assumptions about the evolutionary history of the Heliconius m. melpomene, H. heurippa, and H. c. cordula complex. My model is based on a structured neutral coalescent generalizing that implemented in IM [225] which allows the number of populations to vary from three to two to one going from the present into the past (Figure 3). Specifically, I assume that during an initial epoch [0,T1] encompassing the time of sampling (t = 0) there are three populations, labeled Pop 1, Pop 2, and Pop 3 corresponding to H. m. melpomene, H. heurippa, and H. c. cordula, respectively, and which have effective population sizes N1, N2, and N3, and that the forward in time migration rate (per individual per generation) from population i to population j is mij (Figure 3.4). At the outset of the second epoch, spanning [T1, T2], any lineages present in Pop 2 are randomly assigned to f1 founders from Pop 1 and f3 founders from Pop 3, using Wright-Fisher sampling, and these founders are moved to their respective populations of origin (Figure 3.4). Pop 2 then ceases to exist, reducing the process to a coalescent structured by Pop 1 and Pop 3, with the same effective population sizes and pairwise migration rates as before (Figure 3.4). Finally, at the outset of the third epoch, Pop 1 and Pop 3 are merged into a single ancestral population, Pop A, of effective population size NA in which the genealogy is determined by the usual unstructured coalescent (Figure 3.4). This scenario reflects my judgement that H. m. melpomene and H. c. cordula are older than H. heurippa. The model is equally able to accommodate a hybrid origin for heurippa with little or no subsequent gene flow (by taking f1 and f3 both positive and m12, m21, m23, and m32 small) as well as a bifurcating speciation event followed by extensive gene flow involving the other putative parental species.

To determine a genealogy for each gene, I assigned the same number of ancestral lineages to each population as there were sequences of the corresponding species in the sample and then ran the coalescent until the there was only a single lineage (the most recent common 104

ancestor of the entire sample) left; in all cases this individual appeared in the ancestral population preceding the split of Pop 1 and Pop 3. Given this genealogy, a set of sequences was evolved by first assigning a sequence to the root of the tree with nucleotides assigned to each site through independent draws from the stationary distribution of base frequencies. The HKY model [223] was then used to evolve sequences along each branch of the genealogy. Finally, the two summary statistics were calculated for the collection of simulated sequences present at the tips of the tree. These procedures were repeated 10,000 times for each gene and for each combination of parameter values to generate an approximation to the true bivariate distribution of the summary statistics. The likelihood of the observed summary statistics for each locus given the parameter values was estimated by using a kernel density estimator with a bivariate Gaussian kernel with bandwidth h = 0.05 [226]. As the loci were unlinked, I assumed independence among loci and estimated the total likelihood as the product of the five locus-specific likelihoods.

I used independent analyses of the Heliconius sequence data to estimate as many parameters of the above model as possible. Specifically, pairwise IM analyses of the H. m. melpomene and H. c. cordula sequence data were used to estimate N1, N3, NA, T2, m13, and m31 and the following values were used in my simulations: N1 = 7x105, N3 = 1.3x106, NA = 2x105, T2 = 7x105 generations, m13 = 0.0 and m31 = 4.0x10 -6. Note that these estimates, which were obtained through IM analyses of the same non-coding, gap-stripped sequences used in my subsequent coalescent analyses, differ somewhat from those shown in Table 4, which used both coding and non-coding sites. Parameters were converted to absolute units using the mitochondrial substitution rate mentioned above. I used Watterson’s estimator for T plus the preceding estimates of N1 and N3 to estimate the mean substitution rate (per bp per generation) for each of the five loci; separate rate estimates were obtained for the H. m. melpomene and H. c. cydno sequences and then averaged to provide the rates used in my simulations. Finally, the nucleotide substitution model was parameterized by using PAUP* v.4.0b10 [207] to estimate transition/transversion ratios and equilibrium base frequencies for each gene using a three species alignment. The remaining demographic parameters for which I did not have independent values were varied within ranges that I believed to be biologically plausible (see Table 3.5 for a sample). Although ideally I would have liked to 105

have estimated the full multivariate likelihood surfaces for all parameters simultaneously, the computational resources available to me forced me to focus attention on the migration rates, which I varied (independently) in units of 0.2 (x10-6) from 0.0 to 2.0, while exploring relatively few values for the other undetermined parameters.

The two summary statistics used in this study were the ratio (sr1) of the number of sites segregating in both Pop 1 and Pop 2 (but not Pop 3) to the number of sites segregating only in Pop 1 or Pop 2 and the ratio (sr3) of the number of sites segregating in both Pop 2 and Pop 3 (but not in Pop 1) to the number of sites segregating only in Pop 2 or Pop 3. These were chosen from a number of candidate statistics, which were evaluated using coalescent simulations because they were strongly correlated with the migration rates between Pop 2 and Pop 1 or Pop 3. The values of these and related summary statistics for each of the five loci are shown in Table 3.6. These were calculated using gap-stripped, non-coding regions of each gene. The mtDNA data was excluded from this analysis because the mtDNA phylogeny (Figure 3.1) is inconsistent with one of the central assumptions of the coalescent model, namely that H. m. melpomene and H. c. cordula diversified prior to the origin of H. heurippa, and would have dominated the multilocus likelihood calculation had it been included.

Sequence data from this chapter have been deposited with the EMBL/GenBank Data Libraries under accession nos. DQ674383-DQ674451, DQ445385-DQ445415 and DQ445416-DQ445457.

106

SUPPLEMENTARY MATERIAL

Appendix 3.1. Populations sampled with information about the genes sequenced for each individual

Figure code

Identification number

C1

M187

H. c. cordula

C2

M105

H. c. cordula

C3

M104

H. c. cordula

C4

M101

H. c. cordula

C5

M182

H. c. cordula

C6

M110

H. c. cordula

C7

M111

H. c. cordula

C8

M189

H. c. cordula

C9

M199

H. c. cordula

M1

M2 M3

M4

M5

M6

Stri-b-13 M113 M95

Stri-b-9

Stri-b-11

Stri-b-12

species

H. m.melpomene

H. m. melpomene H. m. melpomene

H. m. melpomene

H. m. melpomene

H. m. melpomene

M7

M119

H. m. melpomene

M8

M115

H. m. melpomene

H1

H2

H3

Stri-b-40

Stri-b-51

Stri-b-44

H. heurippa

H. heurippa

H. heurippa

Locality San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W San Cristobal, Merida , Venezuela 7°47’35’’ N, 72°11’44’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Santa Ana, Merida, Venezuela 7° 36’41’’N, 72°18’10’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Santa Ana, Merida, Venezuela 7° 36’41’’N, 72°18’10’’W Santa Ana, Merida, Venezuela 7° 36’41’’N, 72°18’10’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia

CO Tpi

Dll

inv

w

sd

Collection date

*

*

—

—

—

—

16/10/02

*

-

—

—

—

-

13/06/02

*

*

—

-

—

—

13/06/02

*

*

—

—

—

—

13/06/02

*

*

—

-

—

—

16/10/02

-

-

—

—

-

-

9/08/02

-

-

-

—

-

-

9/08/02

-

-

-

—

-

-

9/08/02

-

-

-

—

-

-

9/08/02

*

*

—

—

—

28/02/98

*

*

—

—

—

—

*

*

—

—

—

—

13/06/02

15/07/02 *

*

—

—

—

— 28/02/98

*

*

—

—

—

28/02/98

*

*

—

—

—

— 28/02/98

*

*

-

—

-

-

9/08/02

*

*

-

-

-

-

13/06/02

—

—

—

—

—

— 28/02/98

—

—

—

—

—

28/02/98

—

—

—

—

—

— 28/02/98

107

H4

H5

H6

H7

H8

H9

H10

H11

M17

M12

M145

Stri-b-39

M8

Stri-b-34

M141

M4

H. heurippa

H. heurippa

H. heurippa

H. heurippa

H. heurippa

H. heurippa

H. heurippa

H. heurippa

4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W Chirajara, Cundinamarca, Colombia 4° 12’ 48’’N, 73°47’70’’W

—

—

—

-

—

— 2/08/02

—

—

—

—

—

— 2/08/02

—

-

—

—

-

15/07/02

—

—

—

—

—

28/02/98

—

—

—

—

-

2/08/02

—

—

-

—

-

28/02/98

—

—

-

—

-

15/07/02

—

—

-

-

-

2/08/02

* Mitochondrial and sex-linked sequences for H. c. cordula (C) and H. m. melpomene (M) were taken from GenBank. (DQ019244-DQ019246, DQ019250,DQ019251, DQ19234-DQ19239 for H. c. cordula and AY548139, DQ019243, DQ019247-DQ019249, DQ019252-DQ019254, DQ019228-DQ019233, DQ019240, AY548151 for H. m. melpomene)

108

Appendix 3.2 Recombination rate estimates for five gene regions sampled from populations of H. heurippa and relatives.

locus Tpi Dll inv w sd

TW 0.0083 0.0151 0.0394 0.0281 0.0171

locus Tpi Dll inv w sd

TW 0.0014 0.0229 0.0307 0.0340 0.0215

locus Tpi Dll inv w sd

TW 0.0078 0.0138 0.0235 0.0377 0.0266

H. c. cordula U *** 0.0095 0.0364 0.000001 0.0082 H. heurippa U *** 0.0001 0.0045 0.000001 0.0186 H. m. melpomene U 0.000001 0.0053 0.0091 0.000001 0.000001

U/u *** 0.6291 0.9238 0.0000355 0.4795

P *** <0.05* <0.05* >0.05 <0.05*

U/u *** 0.0043 0.1465 0.0000294 0.8651

P *** >0.05 <0.05* >0.05 <0.05*

U/u 0.00012 0.4 0.3872 0.0000265 0.0000375

P >0.05 >0.05 <0.05* >0.05 >0.05

*Statistical significance to 95% level. TW and U are the genetic diversity and recombination rate parameters, respectively. No values are given for Tpi for H. c. cordula and H. heurippa because there were insufficient segregating sites to provide reliable estimates

109

References 1. Grant V (1981) Plant Speciation. New York: Columbia Univ. Press. 2. Coyne JA, Orr HA (2004) Speciation. Sunderland, MA: Sinauer Associates. 3. Rieseberg LH (1991) Homoploid reticulate evolution in Helianthus (Asteraceae): evidence from ribosomal genes. Am J Bot 78: 1218-1237. 4. Turelli M, Barton N, Coyne J (2001) Theory and speciation. Trends Ecol Evol 16: 330343. 5. Buerkle CAM, R. J. Asmussen, M. A. and Rieseberg, L. H. (2000) The likehood of homoploid hybrid speciation. Heredity 84: 441-451. 6. Rieseberg LH (1997) Hybrid origins of plant species. Ann Rev Ecol Syst 28: 359-389. 7. Anderson E (1948) Hybridization of the habitat. Evolution 2: 1-9. 8. Templeton AR (1981) Mechanisms of speciation-- a population genetic approach. Ann Rev Ecol Syst 12: 23-48. 9. Shaw DD, Wilkinson P, Coates DJ (1983) Increased chromosomal mutation rate after hybridization between two subspecies of grasshoppers. Science 220: 1165-1167. 10. Rieseberg LH, Van Fossen C, Desrochers AM (1995) Hybrid speciation accompanied by genomic reorganization in wild sunflowers. Nature 375: 313-316. 11. Barton NH, Hewitt GM (1985) Analysis of hybrid zones. Ann Rev Ecol Syst 16: 113148. 12. McCarthy EM, Asmussen MA, Anderson wW (1995) A theoretical assessment of recombinational speciation. Heredity 74: 502-509. 13. Johnson N (2000) Speciation: Dobzhanzky-Muller incompatibilities, dominance and gen interactions. Trends Ecol Evol 15: 480-481. 14. Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation. Genetics 154: 1663-1679. 15. Orr HA, Turelli M (2001) The evolution of postzygotic isolation: Accumulating Dobzhansky-Muller incompatibilities. Evolution 55: 1085-1094. 16. Bateson W (1909) Heredity and variation in modern lights.; Seward A, editor. Cambridge: Cambridge Univ. Press. 17. Dobzhansky TH (1936) Studies on hybrid sterility. II. Localisation of sterility factors in Drosophila pseudoobscura hybrids. Genetics 21: 113-135. 18. Muller HJ (1942) Isolating mechanisms, evolution and temperature. Biol Symp 6: 71125. 19. Orr HA (1996) Dobzhansky, Bateson, and the genetics of speciation. Genetics 144: 1331-1335. 20. Haldane JBS (1922) Sex ratio and unisexual sterility in hybrid animals. J Genet 12: 101-109. 21. Turelli M, Orr HA (1995) The Dominance Theory of Haldanes Rule. Genetics 140: 389-402. 22. Wu C-I, Johnson NA, Palopoli MF (1996) Haldane's rule and its legacy: why are there so many sterile males? Trends Ecol Evol 11: 281-284. 23. Tao Y, Hartl DL (2003) Genetic dissection of hybrid incompatibilities between Drosophila simulans and D. mauritiana.III. Heterogeneous accumulation of hybrid 110

incompatibilities, degree of dominance, and implications for Haldane's rule. Evolution 57: 2580-2598. 24. Burke JM, Arnold ML (2001) Genetics and the Fitness of Hybrids. Ann Rev Genet 35: 31-32. 25. Harrison RG (1990) Hybrid zones: windows on evolutionary process. In: Futuyma D, Antonovics J, editors. Oxford Surveys in Evolutionary Biology. Oxford: Oxford University Press. pp. 69-128. 26. Barton NH (2001) The role of hybridization in evolution. Molec Ecol 10: 551-568. 27. Rieseberg LH, Sinervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene interactions in hybrid speciation: evidence from ancient and experimental hybrids. Science 272: 741-745. 28. Burke JM, Carney SE, Arnold ML (1998) Hybrid fitness in the Louisiana irises: Analysis of parental and F1 performance. Evolution 52: 37-43. 29. Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and speciation. Heredity 83: 363-372. 30. DeVicente MC, Tanksley SD (1993) QTL analysis of transgressive segregation in an interespecific tomato cross. Genetics 134: 585-596. 31. Bell M, Travis MP (2005) Hybridization, transgressive segregation, genetic covariation, and adaptative radiation. Trends Ecol Evol 20: 358-361. 32. Rick CM, Smith PG (1953) Novel variation in tomato species hybrids. Am Nat 87: 359375. 33. Vega U, Frey KJ (1980) Transgressive segregation in inter- and intraspecific crosses of barley. Euphytica 29: 585-694. 34. Rieseberg LH, Widmer A, Arntz AM, Burke JM (2003) The genetic architecture necessary for transgressive segregation is common in both natural and domesticated populations. Philos Trans Roy Soc Lond B 358: 1141-1147. 35. Albertson RC, Kocher TD (2005) Genetic architecture sets limits on transgressive segregation in hybrid cichlid fishes. Evolution 59: 686-690. 36. Albertson RC, Streelman JT, Kocher TD (2003) Directional selection has shaped the oral jaws of Lake Malawi cichlid fishes. Proc Natl Acad Sci, USA 100: 5252-5257. 37. Mavárez J, Salazar C, Bermingham E, Salcedo C, Jiggins CD, et al. (2006) Speciation by hybridization in Heliconius butterflies. Nature 441:868-871. 38. Howarth DG, Baum DA (2005) Genealogical evidence of homoploid hybrid speciation in an adaptative radiation of Scaevola (Goodeniaceae) in the Hawaiian islands. Evolution 59: 948-961. 39. Dowling TE, Secor CL (1997) The role of hybridization and introgression in the diversification of animals. Ann Rev Ecol Syst 28: 593-619. 40. O'Foighil D, Hilbish TJ, Showman RM (1996) Mitochondrial gene variation in Mercenaria clam sibling species reveals a relict secondary contact zone in the western Gulf of Mexico. Mar Biol 126: 675-683. 41. Goodfriend GA, Gould SJ (1996) Paleontology and chronology of two evolutionary transitions by hybridization in the Bahamian land snail Cerion. Science 274: 18941897. 42. Gould SJ, Woodruff DS (1990) History as a cause of area effects: an illustration from Cerion on Great Inagua. Bahamas. Biol J Linn Soc 40: 67-98.

111

43. Lanham UN (1981) Some Colorado Andrena of the subgenus Scaphandrena of presumed hybrid origin, with special reference to the tarsal claws (Hymenoptera: Apoidea). J Kans Entomol Soc 54: 537-546. 44. Lanham UN (1984) The hybrid swarm of Andrena (Hymenoptera: Apoidea) in Western North America: a possible source for the evolutionary origin of a new species. J Kans Entomol Soc 57: 197-208. 45. Sperling FAH, Harrison RG (1994) Mitochondrial DNA variation within and between species of the Papilio machaon group of swallowtail butterflies. Evolution 48: 408422. 46. Salazar C, Jiggins CD, Arias CF, Tobler A, Bermingham E, et al. (2005) Hybrid incompatibility is consistent with a hybrid origin of Heliconius heurippa Hewitson from its close relatives, Heliconius cydno Doubleday and Heliconius melpomene Linnaeus. J Evol Biol 18: 247-256. 47. Salazar C., Jiggins CD, Taylor J, Kronforst M, Linares M (2006) Resolving Introgression and Hybrid Speciation among Heliconius Butterflies. Genetics submitted. 48. Schwarz D, Matta BM, Shakir-Botteri NL, McPheron BA (2005) Host shift to an invasive plant triggers rapid animal hybrid speciation. Nature 436: 546-549. 49. Lieder U (1983) introgression as a factor in the evolution pf polytypical plankton cladocera. Int Revue ges Hydrobiol 68: 269-284. 50. Hebert PDN (1985) Interspecific hybridization between cyclic parthenogens. Evolution 39: 216-220. 51. Lieder U (1987) The possible origin of Daphnia cucullata procurva Poppe1887 in the lakes of the Pomeranian Lakeland by hybridization in the past. Hydrobiologia 145: 201-211. 52. Schwenk K, Spaak P (1995) Evolutionary and Ecological consequences of interspecific hybridization in cladocerans. Experientia 51: 465-481. 53. Taylor DJ, Hebert PDN (1992) Daphnia galeata mendotae as a cryptic species complex with interspecific hybrids. Limnol Oceanogr 37: 658-665. 54. Taylor DJ, Heber PDN (1993) Cryptic intercontinental hybridization in Daphnia (Crustacea): the ghost of introductions past. Proc Roy Soc Lond B 254: 163-168. 55. Taylor DJ, Hebert, P. D. N. and Colourne, J. K. (1996) Phylogenetics and evolution of the Daphnia longispina group (Crustacea) based on 12S rDNA sequence and allozyme variation. Mol Phylogenet Evol 5: 495-510. 56. Mina MV (1992) An interpretation of diversity in the Salmonid genus Brachymystax (Pisces, Salmonidae) as the possible result of multiple hybrid speciation. J Ichthyol 32: 117-122. 57. Osinov AG (1993) Countercurrent dispersal, secondary contact, and speciation in lenoks of the genus Brachymystax (Salmonidae: Salmoniformes). Genetika 29: 654669. 58. Crabtree CB, Buth DG (1987) Biochemical systematics of the catostomid genus Catostomus : assessment of C. clarki, C. plebeius and C. discobolus including the Zuni sucker, C. d. yarrowi. Copeia 1987: 843-854. 59. Smith GR, Hall JG, Koehn RK, Innes DJ (1983) Taxonomic relationships of the Zuni mountain sucker Catostomus discobolus yarrowi. Copeia 1983: 37-48. 60. Miller RR, Smith GR (1981) Distribution and evolution of Chasmistes (Pisces: Catostomidae) in western North America. Occ Pap Mus Zool Univ Mich 696: 1-46. 112

61. Dowling TE, DeMarais BD (1993) Evolutionary significance of introgressive hybridization in cyprinid fishes. Nature 362: 444-446. 62. DeMarais BD, Dowling TE, Douglas MM, Minckley W. L., Marsh PC (1992) Origin of Gila seminuda (Teleostei: Cyprinidae) through introgressive hybridization : Implications for evolution and conservation. Proc Natl Acad Sci, USA 89: 27472751. 63. Meagher S, Dowling TE (1991) Hybridization between the Cyprinid fishes Luxilus albeolus, L. cornutus, and L. cerasinus with comments on the proposed hybrid origin of L. albeolus. Copeia 4: 979-991. 64. Menzel BW (1977) Morphological and electrophoretic identification of a hybrid cyprinid fish, Notropis cerasinus x Notropis c. cornutus, with implications on the evolution of N. albeolus. Comp Biochem Physiol 57B: 215-218. 65. Menezes NA, Weitzman SH (1990) Two new species of Mimagoniates (Teleostei: Characidae: Glandulocaudinae). their phylogeny and biogeography and a key to the glandulocaudin fishes of Brazil and Paraguay. Proc Biol Soc Wash 103: 380-426. 66. Goodfellow WN, Hocutt CH, Morgan RP, Stauffer JR (1984) Biochemical assessment of the taxonomic status of "Rhinichthys bowersi" (Pisces: Cyprinidae). Copeia 1984: 652-659. 67. Stauffer JR, Hocutt CH, Denoncourt RF (1978) Status and distribution of the hybrid Nocommis micropogon x Rhinichthys cataractae, with a discussion of hybridization as a viable mode of vertebrate speciation. Am Midl Nat 101: 355-365. 68. Stauffer JR, Hocutt CH, Mayden RL (1997) Pararhinichthys, a new monotypic genus of minnows (Teleostei: Cyprinidae) of hybrid origin from eastern North America. Ichthyol Explor Freshwaters 7: 327-336. 69. Smith PF, Konings AD, Kornfield IRV (2003) Hybrid origin of a cichlid population in Lake Malawi: implications for genetic variation and species diversity. Mol Ecol 12: 2497-2504. 70. Salzburger W, Baric S, Sturmbauer C (2002) Speciation via introgressive hybridization in East African cichlids? Mol Ecol 11: 619-625. 71. Meyer A, Salzburger W, Schartl M (2006) Hybrid origin of swordtail species (Teleostei: Xiphophorus clemenciae) driven by sexual selection. Mol Ecol 15: 721731. 72. Greer GJ, Kitikoon V, Lohachit C (1989) Morphological and life cycle of Schistosoma sinensium Pao, 1959, from northwest Thailand. J Parasitol 75: 98-101. 73. Hirai H, Taguchi, T., Saitoh, Y., Kawanaka, M., Sugiyama, H., Habe, S., Okamoto, M., Hirata, M., Shimada, M., Tiu, W. U., Lai, K., Upatham, E. S., and Agatsuma, T. (2000) Chromosomal differentiation of the Schistosoma japonicum complex. Int J Parasitol 30: 441-452. 74. Jones TR, Routman EJ, Begun DJ, Collins JP (1995) Ancestry of an isolated subspecies of salamander, Ambystoma tigrinum stebbinsi Lowe: the evolutionary significance of hybridization. Mol Phylog Evol 4: 194-202. 75. Hutchinson MN, Donnellan SC (1992) Taxonomy and genetic variation in the Australian Lizards of the genus Pseudemoia (Scincidae: Lygosominae). J Nat Hist 26: 215-264. 76. Johnston RF (1969) Taxonomy of house sparrows and their allies in the Mediterranean basin. Condor 71: 129-139. 113

77. Roy MS, Geffen E, Smith D, Ostrander EA, Wayne RK (1994) Patterns of differentiation and hybridization in North American wolflike canids, revealed by analysis of microsatellite loci. Molec Bio Evol 11: 553-570. 78. Wayne RK, Jenks SM (1991) Mitochondrial DNA analysis implying extensive hybridization of the endangered red wolf Canis rufus. Nature 351: 565-568. 79. Yonekawa H, Moriwaki K, Gotoh O, Miyashita N, Matsushima Y, et al. (1988) Hybrid origin of Japanese Mice "Mus musculus molossinus¨: Evidence from Restriction Analysis of Mitochondrial DNA. Molec Bio Evol 5: 63-78. 80. McFadden CS, Hutchinson B (2004) Molecular evidence for the hybrid origin of species in the soft coral genus Alcyonium (Cnidaria: Anthozoa: Octocorallia). Mol Ecol 13: 1495-1505. 81. Taylor DJ, Herbert PDN (1993) Habitat dependent hybrid parentage and differential introgression between neighboringly sympatric Daphnia species. Proc Natl Acad Sci, USA 90: 7079-7083. 82. Taylor DJ, Herbert PDN (1993) A reappraisal of phenotypic variation in Daphnia galeata mendotae : the role of interspecific hybridization. Can J Fish Aquat Sci 50: 2137-2146. 83. Nowak RM (1979) North American Quaternary Canis Monograph 6.; Kansas Uo, editor. Lawrence: Museum of Natural History. 84. Nowak RM (1992) The red wolf is not a hybrid. Conserv Biol 6: 593-595. 85. Wayne RK (1992) On the use of morphologic and molecular genetic characters to investigate species status. Conserv Biol 6: 590-592. 86. Berlocher SH, Feder JL (2002) Sympatric speciation in phytophagous insects: moving beyond controversy? Ann Rev Entomol 47: 773-815. 87. Ramsay M, Williamson R, Estivill X, Wainwright BJ, Ho MF, et al. (1993) Haplotype analysis to determine the position of a mutation among closely linked DNA markers. Hum Mol Genet 2: 1007-1014. 88. Berlocher SH (2000) Radiation and divergence in the Rhagoletis pomonella species group: inferences from allozymes. Evolution 54: 543-557. 89. Seehausen O (2004) Hybridization and adaptive radiation. Trends Ecol Evol 19: 198207. 90. Grant PR, Grant BR (1992) Hybridization of bird species. Science 256: 193-197. 91. Grant PR, Grant BR (1994) Phenotypic and genetic effects of hybridization in Darwin's finches. Evolution 48: 287-296. 92. Grant B, Grant PR (2003) What Darwin's finches can teach us about the evolutionary origin and regulation of biodiversity. BioScience 53: 965-975. 93. Kornfield I, Smith PF (2000) African cichlid fishes: models systems for evolutionary biology. Ann Rev Ecol Syst 31: 163-196. 94. Danley PD, Kocher TD (2001) Speciation in rapidly diverging systems:lessons from Lake Malawi. Molec Ecol 10: 1075-1086. 95. Danley PD, Market JA, Arnegard ME, Kocher TD (2000) Divergence with gene flow in the rock-dwelling cichilds of lake Malawi. Evolution 54: 1725-1737. 96. Knight ME, Turner GF (1999) Reproductive isolation among closely related Lake Malawi cichlids: can males recognize conspecific females by visual clues? Anim Behav 58: 761-768.

114

97. Parker A, Kornfield I (1997) Evolution of the mitochondrial DNA control region in the mbuna (Cichlidae) species flock of Lake Malawi, East Africa. J Molec Evol 45: 7083. 98. Markert JA, Danley PD, Arnegard ME (2001) New markers for new species: microsatellite loci and the East African cichlids. Trends Ecol Evol 16: 100-107. 99. Seehausen O, Alphen JJM, Witte F (1997) Effects of water transparency and light regime on sexual selection and the diversity of haplochromine cichlids. Science 277: 1808-1811. 100. Stauffer JR, Bowers NJ, Kocher TD, McKaye K. R. (1996) Evidence of hybridization between Cyanotilapia afra and Psuedotropheus zebra (Teleostei: Cichlidae) following an intralacustrine translocation in Lake Malawi. Copeia: 203-208. 101. Ruber L., Meyer A, Sturmbauer C, Verheyen E (2001) Population structure in two sympatric species of the Lake Tanganyika cichlid tribe Eretmodini: evidence for introgression. Molec Ecol 10: 1207-1225. 102. McElroy DM, Kornfield I (1993) Novel jaw morphology in hybrids between Pseudotropheus zebra and Labeotropheus fullerborni (Teleostei: Cichlidae) from Lake Malawi, Africa. Copeia 1993: 933-945. 103. Markert JA, Arnegard ME, Danley PD, Kocher TD (1999) Biogeography and population genetics of the Lake Malawi cichlid Melanochromis auratus: habitat transience, philopatry and speciation. Molec Ecol 8: 1013-1026. 104. Gilbert LE (2003) Adaptive novelty through introgression in Heliconius wing patterns: Evidence for shared genetic 'tool box' from synthetic hybrid zones and a theory of diversification. In: Boggs CL, Watt WB, Ehrlich PR, editors. Ecology and Evolution Taking Flight: Butterflies as Model Systems. Chicago: Univ. of Chicago Press. 105. Jiggins CD, Naisbit RE, Coe RL, Mallet J (2001) Reproductive isolation caused by colour pattern mimicry. Nature 411: 302-305. 106. Mallet J, Gilbert LE (1995) Why are there so many mimicry rings? Correlations between habitat, behaviour and mimicry in Heliconius butterflies. Biol J Linn Soc 55: 159-180. 107. Smiley JT (1978) Plant chemistry and the evolution of host specificity: new evidence from Heliconius and Passiflora. Science 201: 745-747. 108. Mallet J, McMillan WO, Jiggins CD (1998) Mimicry and warning colour at the boundary between races and species. In: Howard DJ, Berlocher SH, editors. Endless Forms: Species and Speciation. New York: Oxford University Press. pp. 390-403. 109. Posada D, Crandall KA (2001) Intraspecific gene genealogies: trees grafting into networks. Trends Ecol Evol 16: 37-45. 110. Linder CR, Rieseberg LH (2004) Reconstructing patterns of reticulate evolution in plants. Am J Bot 9: 1700-1708. 111. Stephens M (2003) Inference under the coalescent. In: Balding DJ, Bishop M, Cannings C, editors. Hanbook of Statistical Genetics. Chichester: John Wiley and Sons Ltd. pp. 636-660. 112. Maddison WP (1997) Gene trees in species trees. Syst Biol 46: 523-536. 113. Bandelt H-J, Dress WM (1992) Split decomposition: a new and useful approach to phylogenetic analysis of distance data. Mol Phylog Evol 1: 242-252.

115

114. Rieseberg LH, Raymond O, Rosenthal DM, Lai Z, Living-stone K, et al. (2003) Major Ecological Transitions in Wild Sunflowers Facilitated by Hybridization. Science 301: 1211-1216. 115. Sattath S, Tversky A (1977) Additive similarity trees. Psychometrika 42: 319-345. 116. Huson DH (1998) SplitsTree: a program for analyzing and visualizing evolutionary data. Bioinformatics 14: 68-73. 117. Bandelt H-J, Forster P, Rohlf A (1999) Median-Joining Networks for Inferring Intraspecific Phylogenies. Mol Biol Evol 16: 37-48. 118. Xu SZ (2000) Phylogenetic analysis under reticulate evolution. Molec Biol Evol 17: 897-907. 119. Bryant D, Moulton V. NeighborNet: an agglomerative method fro the construction of planar phylogenetic networks.; 2002; WABI, Rome, Italy. Lecture Notes in Computer Science. pp. 375-391. 120. Makarenkov V (2001) T-REX: reconstructing and visualizing phylogenetic trees and ticulation networks. Bioinformatics 17: 664-668. 121. Makarenkov V, Legendre P (2004) From phylogenetic tree to a reticulated network. J Comput Biol 11: 195-212. 122. Clement MD, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 9: 1657-1660. 123. Bandelt HJ, Macaulay V, Richards M (2000) Median networks: speedy construction and greedy reduction, one simulation, and two case studies from human mtDNA. Molec Phylog Evol 16: 8-28. 124. Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491. 125. Huber kT, Watson EE, Hendy MD (2001) An algorithm for constructing local regions in a phylogenetic network. Molec Phylog Evol 19: 1-8. 126. Nakhleh L, Sun J, Warnow T, Linder R, Moret BME, et al. Reconstructing reticulate evolution in species: theory and practice.; 2004; San Diego, California, USA. pp. 337-346. 127. Rokas. A., Williams BL, King N, Carroll SB (2003) Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature 425: 798-804. 128. Nichols R (2001) Gene trees and species trees are not the same. Trends Ecol Evol 16: 358-364. 129. Arvestad L, Berglund AC, Lagergren J, Sennblad B (2003) Bayesian gene/species tree reconciliation and orthology analysis using MCMC. Bioinformatics 19: i7-i15. 130. Machado CA, Hey J (2003) The causes of phylogenetic conflict in a classic Drosophila species group. Proc R Soc Lond B Biol Sci 270: 1193-1202. 131. Posada D, Crandall KA (2001) Evaluation of methods for detecting recombination from DNA sequences: computer simulatons. Proc Natl Acad Sci, USA 98: 1375713762. 132. Kliman RM, Andolfatto P, Coyne JA, Depaulis F, Kreitman M, et al. (2000) The population genetics of the origin and divergence of the Drosophila simulans complex species. Genetics 156: 1913-1931. 133. Hey J, Machado CA (2003) The study of structurated populations - new hope for a difficult and divided science. Nat Rev Genet 4: 535-543. 116

134. Rosenberg NA, Nordborg M (2002) Genealogical trees, coalescent theory and the analysis of genetic polymorphisms. Nat Rev Genet 3: 380-390. 135. Hey J (2005) On the Number of New World Founders: A Population Genetic Portrait of the Peopling of the Americas. PLoS BioL 3: 0965-0975. 136. Won YJ, Hey J (2005) Divergence population genetics of chimpanzees. Molec Biol Evol 22: 297-307. 137. Won YJ, Sivasundar A, Wang Y, Hey J (2005) On the origin of Lake Malawi cichlid species: a population genetic analysis of divergence. Proc Natl Acad Sci, USA 102: 6581-6586. 138. Hey J., Nielsen R. (2004) Multilocus methods for estimating population sizes, migration rates and divergence time, with applications to the divergence of Drosophila pseudoscura and D. persimilis. Genetics 167: 747-760. 139. Machado CA, Kliman RM, Markert JA, Hey J (2002) Inferring the history of speciation from multilocus DNA sequence data: The case of Drosophila pseudoobscura and close relatives. Mol Biol Evol 19: 472-488. 140. Machado CA, Robbins N, Gilbert MTP, Herre EA (2005) Critical review of host specificity and its coevolutionary implications in the fig/fig-wasp mutualism. Proc Natl Acad Sci, USA 102: 6558-6565. 141. Beerli P, Felsenstein J (2001) Maximum likehood estimation of a migration matrix and effective population sizes in n subpopulations by using a coalescent approach. Proc Natl Acad Sci, USA 98: 4563-4568. 142. Gow J, Peichel CL, Taylor EB (2006) Contrasting hybridization rates between sympatric three-spined sticklebacks highlight the fragility of reproductive barriers between evolutionary young species. Molec Ecol 15: 739-752. 143. Rannala B, Yang Z (2003) Bayes Estimation of Species Divergence Times and Ancestral Population Sizes Using DNA Sequences from Multiple Loci. Molec Ecol 164: 1645-1656. 144. Mallet J (2005) Hybridization as an invasion of the genome. Trends Ecol Evol 20: 229-237. 145. Vaha J-P, Primmer CR (2006) Efficiency of model-based Bayesian methods for detecting hybrid individuals under different hybridization scenarios and with different numbers of loci. Molec Ecol 15: 63-72. 146. Gross BL, Rieseberg LH (2004) The Ecological Genetics of Homoploid Hybrid Speciation. J of Heredity 96: 241-252. 147. Neff NA, Smith GR (1978) Multivariate analysis of hybrid fishes. Syst Zool 28: 176196. 148. Humphries JM, Bookstein FL, Chernoff B, Smith GR, Elder RL, et al. (1981) Multivariate discrimination by shape in realtion to size. Syst Zool 30: 291-308. 149. Rosenthal DM, Rieseberg LH, Donovan LA (2005) Re-creating Ancient Hybrid Species Complex Phenotypes from Early-Generation Synthetic Hybrids: Three Examples Using Wild Sunflowers. Am Nat 166: 26-41. 150. Hudson RR, Coyne JA (2002) Mathematical consequences of the genealogical species concept. Evolution 56: 1557-1565. 151. Pääbo S (2003) The mosaic that is our genome. Nature 421: 409-412. 152. Takahata N, Slatkin M (1984) Mitochondrial gene flow. Proc Natl Acad Sci, USA 81: 1764-1767. 117

153. Bullini L (1994) Origin and evolution of animal hybrid species. Trends Ecol Evol 9: 422-426. 154. Mayr E (1963) Animal Species and Evolution. Cambridge, Mass.: Harvard University Press. 155. Arnold ML (1997) Natural Hybridization and Evolution. New York: Oxford University Press. 215 p. 156. Ting C, Tsaur S, Wu C (2000) The phylogeny of closely related species as revealed by the genealogy of a speciation gene, Odysseus. Proc Nat Acad, USA 97: 5313-5316. 157. Tao Y, Zeng Z, Li J, Hartl DL, Laurie C (2003) Genetiic dissection of hybrid incompatibilities between Drosophila simulans and Drosophila mauritiana. II. Mapping hybrid male sterility loci on the third chromosome. Genetics 164: 13991418. 158. Barbash DA, Awadalla P, Tarone AM (2004) Functional divergence caused by ancient positive selection of a Drosophila hybrid incompatibility locus. PLoS BioL 2: 08390848. 159. Presgraves DC, Balagopalan L, Abmayr SM, Orr HA (2003) Adaptative evolution drives divergence of a hybrid inviability gene between two species of Drosophila. Nature 423: 715-719. 160. Barbash DA, Siino DF, Tarone AM, Roote JA (2003) A rapidly evolving Myb-related protein causes species isolation in Drosophila. Proc Nat Acad, USA 100: 53025307. 161. Hutter P (2002) X-linked small GTPase and OXPHOS genes are candidates for the genetic basis of hybrid inviability in Drosophila. Dev Genes Evol 212: 504-512. 162. Perez DE, Wu CI, Johnson NA, Wu ML (1993) Genetics of reproductive isolation in the Drosophila simulans clade: DNA marker-assisted mapping and characterization of a hybrid-male sterility gene, Odysseus (Ods). Genetics 134: 261-275. 163. Wittbrodt J, Adam D, Malitschek B, Maueler W, Raulf F, et al. (1989) Novel putative receptor tyrosine kinase encoded by the melanome inducing Tu locus in Xiphophorus. Nature 341: 415-421. 164. Turelli M (1998) The causes of Haldane's rule. Science 282: 889-891. 165. Price T, Bouvier M (2002) The evolution of F1 postzygotic incompatibilities in birds. Evolution 56: 2083-2089. 166. Presgraves DC (2002) Patterns of postzygotic isolation in Lepidoptera. Evolution 56: 1168-1183. 167. Sasa MM, Chippindale PT, Johnson NA (1998) Patterns of postzygotic isolation in frogs. Evolution 52: 1811-1820. 168. Coyne JA, Orr HA (1997) Patterns of speciation in Drosophila revisited. Evolution 51: 295-303. 169. Coyne JA, Orr HA (1989) Patterns of speciation in Drosophila. Evolution 43: 362381. 170. Orr HA (1995) The population genetics of speciation: The evolution of hybrid incompatibilities. Genetics 139: 1805-1813. 171. Beltrán M, Jiggins CD, Bull V, McMillan WO, Bermingham E, et al. (2002) Phylogenetic discordance at the species boundary: gene genealogies in Heliconius butterflies. Mol Biol Evol 19: 2176-2190.

118

172. Sheppard PM, Turner JRG, Brown KS, Benson WW, Singer MC (1985) Genetics and the evolution of muellerian mimicry in Heliconius butterflies. PhilosTrans R Soc Lond B Biol Sci 308: 433-613. 173. Naisbit RE, Jiggins CD, Mallet J (2003) Mimicry: developmental genes that contribute to speciation. Evol Dev 5: 269-280. 174. Emsley MG (1965) Speciation in Heliconius (Lep., Nymphalidae): morphology and geographic distribution. Zoologica, New York 50: 191-254. 175. Brower AVZ (1994) Phylogeny of Heliconius butterflies inferred from mitochondrial DNA sequences (Lepidoptera: Nymphalidae). Mol Phylogenet Evol 3: 159-174. 176. Jiggins CD, Linares M, Nasbit RE, Salazar C, Yang ZH, et al. (2001) Sex-linked hybrid sterility in a butterfly. Evolution 55: 1631-1638. 177. Bull V, Beltrán M, Jiggins CD, McMillan WO, Bermingham E, et al. (2006) Polyphyly and Gene Flow between Non-sibling Heliconius Species. BMC Biol 4:11. 178. Read A, Nee S (1991) Is Haldane's rule significant? Evolution 45: 1707-1709. 179. Naisbit R, Jiggins CD, Linares M, Salazar C, Mallet J (2002) Hybrid sterility, Haldane’s rule and speciation in Heliconius cydno and H. melpomene. Genetics 161: 1517-1526. 180. Orr HA (1997) Haldane's rule. Ann Rev Ecol Syst 28: 195-218. 181. Laurie CC (1997) The weaker sex is heterogametic: 75 years of Haldane's rule. Genetics 147: 937-951. 182. McMillan WO, Jiggins CD, Mallet J (1997) What initiates speciation in passion vine butterflies? Proc Natl Acad Sci, USA 94: 8628-8633. 183. Jiggins CD, McMillan WO, Neukirchen W, Mallet J (1996) What can hybrid zones tell us about speciation? The case of Heliconius erato and H. himera (Lepidoptera: Nymphalidae). Biol J Linn Soc 59: 221-242. 184. Naisbit R, Jiggins CD, Mallet J (2002) Disruptive sexual selection against hybrids contributes to speciation between Heliconius cydno and Heliconius melpomene. Proc R Soc Lond B Biol Sci 268: 1-6. 185. Linares M (1997) Origin of Neotropical mimetic biodiversity from a three-way hybrid zone of Heliconius butterflies. In: Ulrich H, editor. Proceedings of the International Symposium on Biodiversity and Systematics in Tropical Ecosystems, Bonn 1994. Bonn: Zoologishunginstitut und Museum Alexander Koenig. 186. Lewontin RC, Felsenstein J (1965) The robustness of homogenity test in 2xN tables. Biometrics 21: 19-33. 187. Stebbins GL (1950) Variation and Evolution in Plants. New York: Columbia Univ. Press. 188. Levin DA (2002) The Role of Chromosomal Change in Plant Evolution. Oxford: Oxford University Press. 189. Holloway AK, Cannatella D, C.,, Gerhardt HC, Hillis DM (2006) Polyploids with Different Origins and Ancestors From a Single Sexual Polyploid Species. Am Nat 167: E88-E101. 190. Abbott R (2003) Sex, Sunflowers, and Speciation. Science 301: 1189-1190. 191. Arnold ML, Bennett BD (1993) Natural hybridization in Louisiana irises: genetic variation and ecological determinants. In: Harrison RG, editor. Hybrid Zones and the Evolutionary Process. New York: Oxford University Press. pp. 115-139. 119

192. Jiggins CD, Estrada C, Rodrigues A (2004) Mimicry and the evolution of pre-mating isolation in Heliconius melpomene. J Evol Biol 17: 680-691. 193. Mallet J (1999) Causes and consequences of a lack of coevolution in Mullerian mimicry. Evol Ecol 13: 777-806. 194. Mallet J, Joron M (1999) Evolution of diversity in warning colour and mimicry: Polymorphisms, shifting balance and speciation. Ann Rev Ecol Syst 30: 201-233. 195. Estrada C, Jiggins CD (2002) Patterns of pollen feeding and habitat preference among Heliconius species. Ecol Entomol 27: 448-456. 196. Rodríguez FJ, Oliver JL, Marín A, Medina JR (1990) The general stochastic model of nucleotide substitution. J Theor Biol 142: 485-501. 197. Yang Z (1994) Estimating the pattern of the nucleotide sustitution. J mol Evol 39: 105-111. 198. Kronforst M (2005) Primers for the amplification of nuclear introns in Heliconius butterflies. Mol Ecol Notes 5: 158-162. 199. Carroll SB, Gates J, Keys DN, Paddock SW, Panganiban GEF, et al. (1994) Pattern formation and eyespot determination in butterfly wings. Science 265: 109-114. 200. Keys DN, Lewis DL, Selegue JE, Pearson BJ, Goodrich LV, et al. (1999) Recruitment of a hedgehog regulatory circuit in butterfly eyespot evolution. Science 283: 532534. 201. Brunetti CR, Selegue J, Monteiro A, French V, Brakefield PM, et al. (2001) The generation and diversification of butterfly eyespot color patterns. Curr Biol 11: 1578-1585. 202. Nijhout HF (1991) The Development and Evolution of Butterfly Wing Patterns. Washington, DC: Smithsonian Institution Press. 203. Hudson R (1990) Gene genealogies and the coalescent process. In: Futuyma D, Antonovics J, editors. Oxford Surveys in Evolutionary Biology. Oxford: Oxford Univ. Press. pp. 1-44. 204. Turner TL, Hahn MW, Nuzhdin SV (2005) Genomic islands of speciation in Anopheles gambiae. PLoS Biol 3: 1572-1578. 205. Besansky NJ, Krzywinski J, Lehmann T, Simard F, Kern M, et al. (2003) Semipermeable species boundaries between Anopheles gambie and Anopheles arabiensis: evidence from multilocus DNA sequence variation. Proc Natl Acad Sci USA 100: 10818-10823. 206. Maddison WP, Maddison DR (2001) MacClade version 4.02: analysis of phylogeny and character evolution. Sinauer Associates, Sunderland Massachusetts. 207. Swofford DL (2000) PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). 4. ed. Sunderland, Massachusetts: Sinauer Associates. 208. Nylander JAA (2004) MrModeltest 2.0.: Program distributed by the author.Evolutionary Biology Centre, Uppsala University. 209. Huelsenbeck JP, Ronquist FR (2001) MrBayes: Bayesian inference of phylogeny. Bioinformatics 17: 754. 210. Crandall KA, Templeton AR (1999) Statistical methods for detecting recombination. In: Crandall KA, editor. The Evolution of HIV: Johns Hopkins University Press. pp. 153-176.

120

211. McVean G, Awadalla P, Fearnhead P (2002) A Coalescent-Based Method for Detecting and Estimating Recombination From Gene Sequences. Genetics 160: 1231-1241. 212. Watterson GA (1975) On the number of segregating sites in genetical models without recombination. Theor Popul Biol 7: 256-275. 213. Rozas J, Sánchez-DelBarrio JC, Messeguer X, Rozas R (2003) DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 19: 2496-2497. 214. Tajima F (1989) Stadistical method for testing the neutral mutation hypothesis by DNA polymorhism. Genetics 123: 585-595. 215. Wall J (2000) A comparison of Estimators or the Population Recombination Rate. Mol Biol Evol 17: 156-163. 216. Hey J, Wakeley J (1997) Acoalescent estimator of the population recombination rate. Genetics 145: 833-846. 217. Nei M (1987) Molecular Evolutionary Genetics. New York: Columbia University Press. 218. Wright S (1951) The genetical structure of populations. Ann Eug 15: 323-354. 219. Lynch M, Crease TJ (1990) The analysis of population survey data on DNA sequence variation. Mol Biol Evol 7: 377-394. 220. Charlesworth B (1998) Measures of divergence between populations and the effect of forces that reduce variability. Mol Bio Evol 15: 538-543. 221. Kessing B (2000) SEQUENCER, Version 6.1.0. Distributed by the author. In: Naos Marine Laboratories STRI, editor. Panama. 222. Hudson R, Kaplan N (1985) Statistical propieties of the number of recombination events in the history of a sample of DNA sequences. Genetics 111: 147-164. 223. Hasegawa M, Kishino K, Yano T (1985) Dating the human-ape splitting by a molecular clock of mitochondrial DNA. J Mol Evol 22: 160-174. 224. Brower AVZ (1994) Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolution. Proc Natl Acad Sci USA 91: 6491-6495. 225. Nielsen R, Wakeley J (2001) Distinguishing migration from isolation. A Markov chain Monte Carlo approach. Genetics 158: 885-896. 226. Scott DW (1992) Multivariate Density Estimation: Theory, Practice, and Visualization: Wiley-interscience.

121

CHAPTER IV Published in Nature 441: 868-871 Genetic hybrid trait reconstruction and recombinational speciation in Heliconius butterflies

ABSTRACT

Speciation is generally considered to result from genetic differences accumulated from a common ancestor. An alternative is recombinational speciation, considered to be extremely rare, in which a third species emerge having genetic contribution from two ancestors. Here I show that a hybrid trait in an animal species can directly cause reproductive isolation. The butterfly species Heliconius heurippa is known to have an intermediate morphology and a hybrid genome [46], and I have recreated its intermediate wing colour and pattern through laboratory crosses between H. melpomene and H. cydno. Mate preference experiments involved no-choice or choice, and pattern models show that the phenotype of H. heurippa reproductively isolates it from both parental species. There is strong assortative mating between all three species, and in H. heurippa the wing pattern and colour elements derived from H. melpomene and H. cydno are both critical for mate recognition by males.

INTRODUCTION

Recombinational hybrid speciation is considered very rare or poorly documented [2,6,146]. This has been explained by the theoretical prediction that reproductive isolation between hybrids and their parents is difficult to achieve and it is not easy in all cases to recreate the hybrid origin in experimental conditions [2,5,12]. However, if a hybrid phenotype directly causes reproductive isolation from parental taxa, and different kinds of markers can be used to study the route of origin, these difficulties can be overcome. Such a role for a hybrid 122

phenotype has been convincingly demonstrated only in three species of Helianthus sunflowers adapted to different habitats [227]. In animals, the evidence for homoploid hybrid speciation is less convincing, because in the majority of cases the studies are limited to showing phylogenetic discordance as the principal evidence, and reproductive isolation is not measured [48,69,70,89]. In these cases shared genetic variation could also be a result of introgression subsequent to a bifurcating speciation event instead of homoploid speciation [39].

Heliconius cydno and H. melpomene are two closely related species that overlap extensively in Central and South America [228] (Annex Figure 1). Speciation in these taxa involved shifts in colour mimicry rings generating both assortative mating and extrinsic postzygotic isolation due to predator-mediated selection and poor hybrid mating discrimination [105,108,182,192]. Heliconius cydno is black with white and yellow marks, whereas H. melpomene is black with red, yellow and orange marks. Both species exhibit strong positive assortative mating based on their wing colour patterns and also differ in habitat use [106] and host plant preference [107], but inter-specific hybrids do occur at 1/1000 frequency in the wild [108]. Heliconius heurippa has a mixed wing pattern, which has led to the suggestion that this is a hybrid species [46,104]. Its hindwing is indistinguishable from that of sympatric H. m. melpomene, whereas the yellow band on its forewing is similar to that of parapatric H. cydno cordula. Ecologically, H. heurippa is most similar to H. cydno, which it replaces geographically in the eastern Andes of Colombia.

The reality of the specific status of H. heurippa was demonstrated in the Fst values obtained through several loci [47] (Chapter III) in which this taxa occupied different positions between the parents with a different genetic degree of similarity. Also, H. heurippa appears as a structured population different from geographical parental samples in microsatellite analysis [37] (Annex Figure 1).

123

RESULTS AND DISCUSSION

I tested the hybrid origin hypothesis for the H. heurippa colour pattern throughout interspecific crosses between H. cydno cordula and H. m. melpomene, reconstructing a possible route of introgressive hybridization that could have given rise to its mixed pattern elements. This is feasible due to the largely co-dominant loci involved in H. melpomene and H. cydno pattern heredity [173,229]. Three loci controlling the red and yellow bands on the forewing and the brown pincer-shaped mark on the ventral hindwing were studied (Figure 4.1a). Most H. cydno times H. melpomene F1 hybrids seem intermediate to both parents (Figure 4.1a), with both a yellow (cydno) and a red (melpomene) band in the median section of the forewing, whereas the ventral side of the hindwing shows a reduced brown mark intermediate between the parental species. Female F1 hybrids from crosses between H. melpomene and H. cydno show Haldane’s rule [46,179], and thus only male F1 hybrids could be backcrossed to both parents to produce offspring. Backcrosses to H. melpomene produced offspring very similar to pure H. m. melpomene, and further backcross generations never recreate the total H. cydno FW band, necessary for expression of a yellow phenotype similar to H. heurippa (Figure 4.1a). However, after only two generations a phenotype virtually identical to H. heurippa (In Annex: supplementary Figure 3) was produced by backcrossing an F1 male to an H. cydno cordula female and then mating selected heterozygote offspring of this cross (Figure 4.1b). When H. heurippa-like individuals descended from this brood were crossed the pattern bred true, showing that they are homozygous for the red forewing band (BB) and the absence of brown hindwing marks (brbr) characteristic of H. melpomene, and similarly homozygous for the yellow forewing band (NnNn) derived from H. cydno. Also, the pattern of these H. heurippa-like individuals breeds true when crossed to wild H. heurippa (Figure 4.1b), implying that pattern genes segregating in my crosses are homologous with those in wild H. heurippa. Furthermore, these results were concordant with observed fifth-generation wild backcrosses between H. melpomene and H. cydno individuals collected in San Cristóbal, Venezuela by Mavarez et. al. [37], suggesting that the lab hybridization is congruent with the probable natural route for the hybrid H. heurippa origin [37] (Figure 4.1b and Annex: supplementary Figure 4).

124

125

Figure 4.1 Reconstruction of the H. heurippa wing pattern. a. First row, H. cydno cordula, H. heurippa and H. m. melpomene; second row, H. cydno cordulauH. m. melpomene F1 hybrid. Backcrosses to H. cydno cordula and H. m. melpomene are shown in the left and right boxes, respectively. Three major loci regulate the patterns: complete (BB, H. melpomene), intermediate (Bb, heterozygotes) or no (bb, H. cydno) expression of a red band on the dorsal forewing; complete (NnNn, H. cydno), intermediate (NnNb, heterozygotes) or no (NbNb, H. melpomene) expression of a yellow band on the dorsal forewing band; and complete (BrBr, H. cydno), intermediate (Brbr, heterozygotes) or no (brbr, H. melpomene) expression of a brown pincer-shaped mark on the ventral hindwing. Further crosses were performed using individuals with the phenotype marked with asterisk (see b). Forewings are shown in dorsal view, hindwings in ventral view. b. Left box: offspring from crosses between individuals marked with an asterisk in a. The B and Br loci are linked, explaining the absence of the two recombinant genotypes BbNnNnbrbr and bbNnNnbrbr. Right box: offspring of a cross between a laboratory hybrid with genotype BBNnNnbrbr and H. heurippa, showing that the pattern breeds true.

When a new phenotype is generated through hybridization, it needs to gain reproductive isolation to avoid dilution with the parent lineages [146]. I therefore tested the degree to which H. heurippa is isolated from H. melpomene and H. cydno by assortative mating. Nochoice mating experiments showed a reduced probability of mating in all inter-specific comparisons, with H. heurippa females particularly unlikely to mate with either H. cydno or H. melpomene (Table 4.1). When a male of each species was presented with a single female, H. heurippa males were ten times more likely to court their own females than the other species (Figure 4.2). In mating experiments with choice, there was similarly strong 126

assortative mating, although occasional matings between H. cydno females and H. heurippa males were observed (Table 4.2). Isolation due to assortative mating (on average more than 90% isolation between H. heurippa and H. melpomene and more than 75% between H. heurippa and H. cydno) is therefore considerably greater than that caused by intrinsic incompatibilities (about 25% isolation between H. heurippa and H. melpomene, and zero between H. heurippa and H. cydno) [46] or extrinsic postzigotic isolation (about 50%) [230]. The combination of strong assortative mating, geographic isolation from H. cydno and both intrinsic and ecological postzygotic isolation has contributed to the speciation of H. heurippa.

Table 4.1 Relative mating probabilities in no-choice experiments.

Male H. melpomene H. cydno H. heurippa

Female H. melpomene 1 (17) 0.120 (0.048–0.231, 50) 0.100 (0.031–0.022, 40)

H. cydno 0.178 (0.084–0.309, 45) 1 (27) 0.440 (0.255–0.637, 44)

H. heurippa 0.073 (0.023–0.163, 55) 0.022 (0.001–0.096, 45) 1 (22)

For each female type, probabilities were estimated relative to that of intra-specific mating, which was set to 1. Numbers in parenthesis show the 95% maximum-likelihood support limits and the number of females used.

Table 4.2 Number of matings in tetrad mate-choice experiments. Female

Male

H. melpomene H. heurippa 15 0 0 12 H. cydno H. heurippa H. cydno 5 3 H. heurippa 0 5 H. melpomene H. cydno H. melpomene 10.5 0 H. cydno 0 5.5 Mating results of 0.5 are due to simultaneous mating of both pairs during the experiment. H. melpomene H. heurippa

127

Figure 4.2 Live courtship experiments. I placed mature males (> 5 days old) of each genotype in an insectary and introduced a single virgin female (1-5 days old). The number of courtship bouts (sustained hovering by the male over the female) occurring over a period of 15 minutes was recorded. The female genotype was then substituted, with genotype order randomized, such that each panel of males was tested against each female genotype. No mating was permitted in order to not disrupt subsequent behaviour. Males were used only once. In total 30 replicate panels of males were tested against all three female genotypes in more than 450 min of observations. H. melpomene (red dots), H. heurippa (blue dots) and H. cydno (grey dots).

The next step is to investigate whether the recombinational trait is involved in the assortative mating highlighted above. Experiments with dissected wings showed that both elements of the forewing colour pattern of H. heurippa were necessary for the stimulation of approach and courtship (Figure 4.3). H. heurippa males were less than half as likely to approach and court the H. m. melpomene or the H. cydno cordula pattern than their own (Figure 4.3). When either the red or yellow bands were experimentally removed from the H. heurippa pattern, this led to a similar reduction in its attractiveness, demonstrating that both hybrid elements are necessary for mate recognition by male H. heurippa (Figure 4.3).

128

Figure 4.3 Relative probabilities of H. heurippa males approaching and courting colour pattern models. Values are estimated relative to the probability of approach a. and courtship b. of the H. heurippa control pattern shown at the top (probability equal to 1). Model patterns are H. m. melpomene (melp), H. cydno cordula (cord), H. heurippa modified to remove the red band (heuA) and H. heurippa modified to remove the yellow band (heuR). Red, real wings; blue, paper wings. Error bars show maximum-likelihood support limits. (see Methods for an explanation).

Similar results were obtained when these experiments were replicated with printed-paper models (Figure 4.3), showing that the colour pattern itself was the cue rather than pheromones associated with the dissected wings. Additional experiments showed that males of both H. m. melpomene and H. cydno cordula showed a greatly reduced probability of approaching and courting the H. heurippa pattern than their own (Figures 4.4 and 4.5). Given the incomplete postzygotic reproductive isolation between all three species, this

129

pattern-based assortative mating must have a continuing role in generating reproductive isolation between H. heurippa and its relatives.

Figure 4.4 Colour pattern models for H. melpomene. Model experiments were carried out as described in Methods using H. melpomene males. In all comparisons, males of H. melpomene prefer to approach and court the real wings and paper models of their own phenotype.

130

Figure 4.5 Colour pattern models for H. cydno. Model experiments were carried out as described in Methods using H. cydno males. In all comparisons except one, H. cydno prefer to approach and court the real wings and paper models of their own phenotype. The only exception was that the H. cydno males, approach and court the modified H. heurippa pattern with the red band removed at the same frequency as their own pattern.

It is important to establish the relevance of ecology in the homoploid speciation because changes in habitat use could facilitate the process [5,146]. One of the most relevant resources necessary for larval development and adult toxicity in Heliconius is the Passifloraceae plant family [231,232]. As mentioned above, host plant preference could contribute to H. melpomene and H. cydno speciation [233]. In particular H. melpomene is monophagous and H. cydno is oligophagous in Costa Rica [107]. When H. heurippa was tested with respect to oviposition preference an oligophagous pattern similar to H. cydno 131

from the eastern Andes was obtained (Figure 4.6). This result is not a surprise. On a geographic scale H. heurippa replaces in distribution the H. cydno subspecies from north to south in the east Andes [46]. In other words H. heurippa uses the same habitat as H. cydno. As I show in this and other chapters, H. cydno made a major contribution to the hybrid H. heurippa genome with H. melpomene probably contributing fewer genes [47], which is the reason why this homoploid species have a H. cydno like oviposition preference. Furthermore, some period of allopatry with regard to H. cydno probably accounts for the new hybrid pattern colour fixation (see below). In contrast to other cases of homoploid speciation, in which new habitat colonization is important in generating reproductive isolation [48,114], here the fixation and use of colour pattern in mating constitute a sufficient barrier to gene flow.

Intriguingly, H. m. melpomene from the eastern Andes also has an H. cydno oligophagous oviposition pattern (Figure 4.6). Plant preference genes introgressed from H. cydno is a possible explanation, taking into account the fact that this species also shares other genome regions such as mtDNA (see chapter III) [47].

Figure 4.6 Oviposition preference. Five different Passiflora species were given to H. cydno cordula, different races of H. melpomene and H. heurippa mated females in a insectary. Eggs laid at intervals of eight days in each plant for each independent female (approx. 15 females for each species) were counted and the total oviposition frequencies estimated. 132

Geographic diversity in colour pattern observed in Heliconius is probably obtained through a shift balance process in which both genetic drift and frequency-dependent selection are involved [194,234]. Such an event could contribute to isolate the hybrid H. heurippa from H. cydno. Subsequently, the novelty of a new pattern that includes elements from H. cydno and H. melpomene in an additive manner produces a mate discrimination cue that could enhance the divergence in parapatry, facilitating self recognition between the three species (Annex: supplementary Figure 8).

This model composed of two steps could offer an alternative way of overcoming the theoretical difficulty of possible hybrid phenotype dilution in the parents inasmuch as shows rapid increase in reproductive isolation driven by assortative mating. Furthermore, because I am proposing divergence in mate behaviour in a geographically isolated population, reinforcement or some other form of sympatric divergence is not required for speciation to occur.

This study provides the first experimental demonstration of a hybrid trait generating reproductive isolation between animal species, and the first example of a hybrid trait causing pre-mating isolation through assortative mating. None of the theoretical treatments of homoploid hybrid speciation consider the important factors required to facilitate this kind of speciation in animals [5,12]. If variation for mate preference were incorporated, the theoretical conditions favoring hybrid speciation might not be as stringent as has been supposed. Finally, the absence of clear morphological markers as the pattern colour genes studied here could obscure the process avoiding the detection of hybrid speciation in other animal groups that are less studied. Other proposed cases of homoploid hybrid speciation in animals occur in well-studied groups such as African cichlids [69,70] and Rhagoletis flies [48].

133

MATERIALS AND METHODS

Crosses Crosses were performed in La Vega, Colombia, between January 2000 and December 2002 with the use of H. m. melpomene (Virgen de Chirajara, 4.213° N, 73.795° W) and H. cydno cordula (Barro Negro, 6.016° N, 72.091° W), both from the Colombian eastern Andes. I isolated virgin females with older males to produce inter-specific F1 offspring and backcrosses. After mating, females were kept individually in 2u3u2 m3 insectaries and supplied with pollen and nectar from Psiguria and Lantana flowers, and Passiflora vines for oviposition. Eggs and the larval and adult stages were reared as described previously. Colour pattern segregation and designation were studied with the use of nomenclature described previously [173].

No-choice experiments

Description Experiments without choice measure the reluctance of males and females to mate interspecifically; such experiments simulate a natural situation in which males encounter females singly. In the insectaries mentioned above, a virgin female (one to three days old) of each species was presented to ten mature males (more than ten days old) for two days. Males were used only once. Matings were recorded every 30 min from 06:00 until 15:00. Experiments were performed between all combinations of the three species, including control experiments between conspecifics. To detect any unobserved mating, all the females were checked after the experiments for the presence of a spermatophore in their reproductive tract.

134

Statistics: No-choice ML model [235]. A binomial mating probability P ix j was obtained for each combination of i-type female and j-type male, maximizing the expression for loge-likelihood given by: m loge P i x j + n loge (1- P i x j) where m and n are the numbers of trials in which the pair mated or remained unmated, respectively. The loge-likehoods for the P ix j values were maximized using the SOLVER algorithm supplied with Microsoft EXCEL. Support limits for P ix j , which are asymptotically equivalent to 95% intervals, where obtained at the parameter values that led to a decrease in the loge-likelihood of two units [236].

Mate choice experiments: Tetrads

Description Experiments with choice were performed to estimate mating probability in a situation in which males encounter females from different species simultaneously. Pairwise experiments were performed between each combination of the three species, in which a single recently emerged virgin female and a mature male (more than ten days old) of each species were placed in a 2u3u2 m3 insectary over the course of a single day. Thus, each experiment involved four butterflies, for example a female and male H. heurippa and a female and male H. cydno for the comparison between those two species. The first mating only was recorded for each experiment, and individuals were not reused. In cases in which both pairs mated simultaneously they were scored as each having half a mating. At least 15 experiments were performed for each pairwise comparison. Mating probabilities were estimated by likelihood (see below).

135

Tetrads statistics [105,182]

Likelihood was used to estimate the probability P i x j of a mating between female type i and male type j, relative to P m x m of mating within H. melpomene (M). The overall multinomial probability of the results for each experiment were then estimated, e.g. for H. melpomene (H) x H. cydno (C) comparison, I m x m = P m x m/(P m x m + P m x c + P c x m + P c x c), I m x c = P

m x c/(P m x m

+P

mxc

+P

cx m

+P

c x c)

an so on. (6I = 1 for each tetrad). The loge-

likelihood was therefore 6 (X m x m logeI m x m + X m x c logeI m x c and son on) where X m x m is the number of M x M matings and X m x c is the number of M x C matings in that tetrad. Likelihoods were summed over all experiments and maximized by varying P

ix j

values

using the SOLVER algorithm supplied with Microsoft EXCEL. Support limits for P

ixj

,

which are asymptotically equivalent to 95% intervals, where obtained at the parameter values that led to a decrease in the loge-likelihood of two units [236].

Mate choice experiments: Colour pattern models In Heliconius, males use colour patterns to locate females and are choosy in mating, presumably because of the large and costly spermatophore transferred to females [237]. I investigated male preferences for different colour pattern models. Between 10 and 20 males in a 2u3u2 m3 insectary were presented with either dissected natural wings or a printed colour pattern model, fixed to a length of flexible clear nylon. Models were manipulated to simulate Heliconius flight in the centre of a spherical area (60 cm in diameter) demarcated by references in the insectary roof. Randomly ordered pairs of 5-min experiments were performed: first, a control flight with a model of the male’s own colour pattern, and second, an experimental flight with a different colour pattern. Entry to the sphere was recorded as ‘approach’ and sustained fluttering directed at the model as ‘courtship’. At least 25 replicates were performed for each comparison. In addition, models were made in which the H. heurippa pattern was modified to show either the yellow band without red (Heu-A) or the red band without yellow (Heu-R). For the dissected wing models, permanent black marker pen (Pilot ultrafine point no xylene SCA-UF) was used to cover the corresponding band. Paper models were made from digital photographs of wings taken with a Sony 136

Cyber-shot dsc-s85 camera that were printed with a high-performance inkjet printer (Hewlett Packard Deskjet 3820) on special photo-quality calcium paper. Only paper models with reflectance spectra similar to real wings were used (Appendix 4.1). The data were used to estimate the probabilities Qixj that males of type j approached or courted models of type i relative to that of their own type j (which was set to one), using likelihood (see below).

Colour Pattern model statistics [105,192]

I estimated the probabilities Q i x j that males type j approached or courted models type i relative to that of their own type j, using likelihood and setting the model to one, so that any value under one represented preference for the control model. Thus, for M males with M versus C models, the actual probabilities are QA c x m/(QA c x m + 1) that males approach C and 1/(QA c x m + 1) that they approach M. The loge-likelihood for this experiment is therefore ™[XA c x m loge { QA c x m/( QA c x m + 1)} + XA m x m loge{1/(QA c x m + 1)}], where XA c x m is the number of M males approaching C and XA m x m is the number approaching M. Similarly QH i x j parameters were estimated for probability of hovering courtship of the model. The summed loge-likelihood was maximized over all experiments by varying the QH i x j parameters. Confidence intervals for parameters were established using the same likelihood method described for mating experiments.

137

138

139

140

References 1. Grant V (1981) Plant Speciation. New York: Columbia Univ. Press. 2. Coyne JA, Orr HA (2004) Speciation. Sunderland, MA: Sinauer Associates. 3. Rieseberg LH (1991) Homoploid reticulate evolution in Helianthus (Asteraceae): evidence from ribosomal genes. Am J Bot 78: 1218-1237. 4. Turelli M, Barton N, Coyne J (2001) Theory and speciation. Trends Ecol Evol 16: 330343. 5. Buerkle CAM, R. J. Asmussen, M. A. and Rieseberg, L. H. (2000) The likelihood of homoploid hybrid speciation. Heredity 84: 441-451. 6. Rieseberg LH (1997) Hybrid origins of plant species. Annu Rev Ecol Syst 28: 359-389. 7. Anderson E (1948) Hybridization of the habitat. Evolution 2: 1-9. 8. Templeton AR (1981) Mechanisms of speciation-- a population genetic approach. Ann Rev Ecol Syst 12: 23-48. 9. Shaw DD, Wilkinson P, Coates DJ (1983) Increased chromosomal mutation rate after hybridization between two subspecies of grasshoppers. Science 220: 1165-1167. 10. Rieseberg LH, Van Fossen C, Desrochers AM (1995) Hybrid speciation accompanied by genomic reorganization in wild sunflowers. Nature 375: 313-316. 11. Barton NH, Hewitt GM (1985) Analysis of hybrid zones. Ann Rev Ecol Syst 16: 113148. 12. McCarthy EM, Asmussen MA, Anderson wW (1995) A theoretical assessment of recombinational speciation. Heredity 74: 502-509. 13. Johnson N (2000) Speciation: Dobzhanzky-Muller incompatibilities, dominance and gen interactions. Trends Ecol Evol 15: 480-481. 14. Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation. Genetics 154: 1663-1679. 15. Orr HA, Turelli M (2001) The evolution of postzygotic isolation: Accumulating Dobzhansky-Muller incompatibilities. Evolution 55: 1085-1094. 16. Bateson W (1909) Heredity and variation in modern lights.; Seward A, editor. Cambridge: Cambridge Univ. Press. 17. Dobzhansky TH (1936) Studies on hybrid sterility. II. Localisation of sterility factors in Drosophila pseudoobscura hybrids. Genetics 21: 113-135. 18. Muller HJ (1942) Isolating mechanisms, evolution and temperature. Biol Symp 6: 71125. 19. Orr HA (1996) Dobzhansky, Bateson, and the genetics of speciation. Genetics 144: 1331-1335. 20. Haldane JBS (1922) Sex ratio and unisexual sterility in hybrid animals. J Genet 12: 101-109. 21. Turelli M, Orr HA (1995) The Dominance Theory of Haldanes Rule. Genetics 140: 389-402. 22. Wu C-I, Johnson NA, Palopoli MF (1996) Haldane's rule and its legacy: why are there so many sterile males? Trends Ecol Evol 11: 281-284. 23. Tao Y, Hartl DL (2003) Genetic dissection of hybrid incompatibilities between Drosophila simulans and D. mauritiana.III. Heterogeneous accumulation of hybrid

141

incompatibilities, degree of dominance, and implications for Haldane's rule. Evolution 57: 2580-2598. 24. Burke JM, Arnold ML (2001) Genetics and the Fitness of Hybrids. Ann Rev Genet 35: 31-32. 25. Harrison RG (1990) Hybrid zones: windows on evolutionary process. In: Futuyma D, Antonovics J, editors. Oxford Surveys in Evolutionary Biology. Oxford: Oxford University Press. pp. 69-128. 26. Barton NH (2001) The role of hybridisation in evolution. Molec Ecol 10: 551-568. 27. Rieseberg LH, Sinervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene interactions in hybrid speciation: evidence from ancient and experimental hybrids. Science 272: 741-745. 28. Burke JM, Carney SE, Arnold ML (1998) Hybrid fitness in the Louisiana irises: Analysis of parental and F1 performance. Evolution 52: 37-43. 29. Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and speciation. Heredity 83: 363-372. 30. DeVicente MC, Tanksley SD (1993) QTL analysis of transgressive segregation in an interespecific tomato cross. Genetics 134: 585-596. 31. Bell M, Travis MP (2005) Hybridization, transgressive segregation, genetic covariation, and adaptative radiation. Trends Ecol Evol 20: 358-361. 32. Rick CM, Smith PG (1953) Novel variation in tomato species hybrids. Am Nat 87: 359375. 33. Vega U, Frey KJ (1980) Transgressive segregation in inter- and intraspecific crosses of barley. Euphytica 29: 585-694. 34. Rieseberg LH, Widmer A, Arntz AM, Burke JM (2003) The genetic architecture necessary for transgressive segregation is common in both natural and domesticated populations. Philos Trans Roy Soc Lond B 358: 1141-1147. 35. Albertson RC, Kocher TD (2005) Genetic architecture sets limits on transgressive segregation in hybrid cichlid fishes. Evolution 59: 686-690. 36. Albertson RC, Streelman JT, Kocher TD (2003) Directional selection has shaped the oral jaws of Lake Malawi cichlid fishes. Proc Natl Acad Sci, USA 100: 5252-5257. 37. Mavárez J, Salazar C, Bermingham E, Salcedo C, Jiggins CD, et al. (2006) Speciation by hybridization in Heliconius butterflies. Nature 441: 868-871. 38. Howarth DG, Baum DA (2005) Genealogical evidence of homoploid hybrid speciation in an adaptative radiation of Scaevola (Goodeniaceae) in the Hawaiian islands. Evolution 59: 948-961. 39. Dowling TE, Secor CL (1997) The role of hybridization and introgression in the diversification of animals. Ann Rev Ecol Syst 28: 593-619. 40. O'Foighil D, Hilbish TJ, Showman RM (1996) Mitochondrial gene variation in Mercenaria clam sibling species reveals a relict secondary contact zone in the western Gulf of Mexico. Mar Biol 126: 675-683. 41. Goodfriend GA, Gould SJ (1996) Paleontology and chronology of two evolutionary transitions by hybridization in the Bahamian land snail Cerion. Science 274: 18941897. 42. Gould SJ, Woodruff DS (1990) History as a cause of area effects: an illustration from Cerion on Great Inagua. Bahamas. Biol J Linn Soc 40: 67-98.

142

43. Lanham UN (1981) Some Colorado Andrena of the subgenus Scaphandrena of presumed hybrid origin, with special reference to the tarsal claws (Hymenoptera: Apoidea). J Kans Entomol Soc 54: 537-546. 44. Lanham UN (1984) The hybrid swarm of Andrena (Hymenoptera: Apoidea) in Western North America: a possible source for the evolutionary origin of a new species. J Kans Entomol Soc 57: 197-208. 45. Sperling FAH, Harrison RG (1994) Mitochondrial DNA variation within and between species of the Papilio machaon group of swallowtail butterflies. Evolution 48: 408422. 46. Salazar C, Jiggins CD, Arias CF, Tobler A, Bermingham E, et al. (2005) Hybrid incompatibility is consistent with a hybrid origin of Heliconius heurippa Hewitson from its close relatives, Heliconius cydno Doubleday and Heliconius melpomene Linnaeus. J Evol Biol 18: 247-256. 47. Salazar C., Jiggins CD, Taylor J, Kronforst M, Linares M (2006) Resolving Introgression and Hybrid Speciation among Heliconius Butterflies. Genetics submitted. 48. Schwarz D, Matta BM, Shakir-Botteri NL, McPheron BA (2005) Host shift to an invasive plant triggers rapid animal hybrid speciation. Nature 436: 546-549. 49. Lieder U (1983) introgression as a factor in the evolution pf polytypical plankton cladocera. Int Revue ges Hydrobiol 68: 269-284. 50. Hebert PDN (1985) Interspecific hybridization between cyclic parthenogens. Evolution 39: 216-220. 51. Lieder U (1987) The possible origin of Daphnia cucullata procurva Poppe1887 in the lakes of the Pomeranian Lakeland by hybridization in the past. Hydrobiologia 145: 201-211. 52. Schwenk K, Spaak P (1995) Evolutionary and Ecological consequences of interspecific hybridization in cladocerans. Experientia 51: 465-481. 53. Taylor DJ, Hebert PDN (1992) Daphnia galeata mendotae as a cryptic species complex with interspecific hybrids. Limnol Oceanogr 37: 658-665. 54. Taylor DJ, Heber PDN (1993) Cryptic intercontinental hybridization in Daphnia (Crustacea): the ghost of introductions past. Proc Roy Soc Lond B 254: 163-168. 55. Taylor DJ, Hebert, P. D. N. and Colourne, J. K. (1996) Phylogenetics and evolution of the Daphnia longispina group (Crustacea) based on 12S rDNA sequence and allozyme variation. Mol Phylogenet Evol 5: 495-510. 56. Mina MV (1992) An interpretation of diversity in the Salmonid genus Brachymystax (Pisces, Salmonidae) as the possible result of multiple hybrid speciation. J Ichthyol 32: 117-122. 57. Osinov AG (1993) Countercurrent dispersal, secondary contact, and speciation in lenoks of the genus Brachymystax (Salmonidae: Salmoniformes). Genetika 29: 654669. 58. Crabtree CB, Buth DG (1987) Biochemical systematics of the catostomid genus Catostomus : assessment of C. clarki, C. plebeius and C. discobolus including the Zuni sucker, C. d. yarrowi. Copeia 1987: 843-854. 59. Smith GR, Hall JG, Koehn RK, Innes DJ (1983) Taxonomic relationships of the Zuni mountain sucker Catostomus discobolus yarrowi. Copeia 1983: 37-48. 60. Miller RR, Smith GR (1981) Distribution and evolution of Chasmistes (Pisces: Catostomidae) in western North America. Occ Pap Mus Zool Univ Mich 696: 1-46. 143

61. Dowling TE, DeMarais BD (1993) Evolutionary significance of introgressive hybridization in cyprinid fishes. Nature 362: 444-446. 62. DeMarais BD, Dowling TE, Douglas MM, Minckley W. L., Marsh PC (1992) Origin of Gila seminuda (Teleostei: Cyprinidae) through introgressive hybridization : Implications for evolution and conservation. Proc Natl Acad Sci, USA 89: 27472751. 63. Meagher S, Dowling TE (1991) Hybridization between the Cyprinid fishes Luxilus albeolus, L. cornutus, and L. cerasinus with comments on the proposed hybrid origin of L. albeolus. Copeia 4: 979-991. 64. Menzel BW (1977) Morphological and electrophoretic identification of a hybrid cyprinid fish, Notropis cerasinus x Notropis c. cornutus, with implications on the evolution of N. albeolus. Comp Biochem Physiol 57B: 215-218. 65. Menezes NA, Weitzman SH (1990) Two new species of Mimagoniates (Teleostei: Characidae: Glandulocaudinae). their phylogeny and biogeography and a key to the glandulocaudin fishes of Brazil and Paraguay. Proc Biol Soc Wash 103: 380-426. 66. Goodfellow WN, Hocutt CH, Morgan RP, Stauffer JR (1984) Biochemical assessment of the taxonomic status of "Rhinichthys bowersi" (Pisces: Cyprinidae). Copeia 1984: 652-659. 67. Stauffer JR, Hocutt CH, Denoncourt RF (1978) Status and distribution of the hybrid Nocommis micropogon x Rhinichthys cataractae, with a discussion of hybridization as a viable mode of vertebrate speciation. Am Midl Nat 101: 355-365. 68. Stauffer JR, Hocutt CH, Mayden RL (1997) Pararhinichthys, a new monotypic genus of minnows (Teleostei: Cyprinidae) of hybrid origin from eastern North America. Ichthyol Explor Freshwaters 7: 327-336. 69. Smith PF, Konings AD, Kornfield IRV (2003) Hybrid origin of a cichlid population in Lake Malawi: implications for genetic variation and species diversity. Mol Ecol 12: 2497-2504. 70. Salzburger W, Baric S, Sturmbauer C (2002) Speciation via introgressive hybridization in East African cichlids? Mol Ecol 11: 619-625. 71. Meyer A, Salzburger W, Schartl M (2006) Hybrid origin of swordtail species (Teleostei: Xiphophorus clemenciae) driven by sexual selection. Mol Ecol 15: 721731. 72. Greer GJ, Kitikoon V, Lohachit C (1989) Morphological and life cycle of Schistosoma sinensium Pao, 1959, from northwest Thailand. J Parasitol 75: 98-101. 73. Hirai H, Taguchi, T., Saitoh, Y., Kawanaka, M., Sugiyama, H., Habe, S., Okamoto, M., Hirata, M., Shimada, M., Tiu, W. U., Lai, K., Upatham, E. S., and Agatsuma, T. (2000) Chromosomal differentiation of the Schistosoma japonicum complex. Int J Parasitol 30: 441-452. 74. Jones TR, Routman EJ, Begun DJ, Collins JP (1995) Ancestry of an isolated subspecies of salamander, Ambystoma tigrinum stebbinsi Lowe: the evolutionary significance of hybridization. Mol Phylog Evol 4: 194-202. 75. Hutchinson MN, Donnellan SC (1992) Taxonomy and genetic variation in the Australian Lizards of the genus Pseudemoia (Scincidae: Lygosominae). J Nat Hist 26: 215-264. 76. Johnston RF (1969) Taxonomy of house sparrows and their allies in the Mediterranean basin. Condor 71: 129-139. 144

77. Roy MS, Geffen E, Smith D, Ostrander EA, Wayne RK (1994) Patterns of differentiation and hybridization in North American wolflike canids, revealed by analysis of microsatellite loci. Molec Bio Evol 11: 553-570. 78. Wayne RK, Jenks SM (1991) Mitochondrial DNA analysis implying extensive hybridization of the endangered red wolf Canis rufus. Nature 351: 565-568. 79. Yonekawa H, Moriwaki K, Gotoh O, Miyashita N, Matsushima Y, et al. (1988) Hybrid origin of Japanese Mice "Mus musculus molossinus¨: Evidence from Restriction Analysis of Mitochondrial DNA. Molec Bio Evol 5: 63-78. 80. McFadden CS, Hutchinson B (2004) Molecular evidence for the hybrid origin of species in the soft coral genus Alcyonium (Cnidaria: Anthozoa: Octocorallia). Mol Ecol 13: 1495-1505. 81. Taylor DJ, Herbert PDN (1993) Habitat dependent hybrid parentage and differential introgression between neighboringly sympatric Daphnia species. Proc Natl Acad Sci, USA 90: 7079-7083. 82. Taylor DJ, Herbert PDN (1993) A reappraisal of phenotypic variation in Daphnia galeata mendotae : the role of interspecific hybridization. Can J Fish Aquat Sci 50: 2137-2146. 83. Nowak RM (1979) North American Quaternary Canis Monograph 6.; Kansas Uo, editor. Lawrence: Museum of Natural History. 84. Nowak RM (1992) The red wolf is not a hybrid. Conserv Biol 6: 593-595. 85. Wayne RK (1992) On the use of morphologic and molecular genetic characters to investigate species status. Conserv Biol 6: 590-592. 86. Berlocher SH, Feder JL (2002) Sympatric speciation in phytophagous insects: moving beyond controversy? Ann Rev Entomol 47: 773-815. 87. Ramsay M, Williamson R, Estivill X, Wainwright BJ, Ho MF, et al. (1993) Haplotype analysis to determine the position of a mutation among closely linked DNA markers. Hum Mol Genet 2: 1007-1014. 88. Berlocher SH (2000) Radiation and divergence in the Rhagoletis pomonella species group: inferences from allozymes. Evolution 54: 543-557. 89. Seehausen O (2004) Hybridization and adaptive radiation. Trends Ecol Evol 19: 198207. 90. Grant PR, Grant BR (1992) Hybridization of bird species. Science 256: 193-197. 91. Grant PR, Grant BR (1994) Phenotypic and genetic effects of hybridization in Darwin's finches. Evolution 48: 287-296. 92. Grant B, Grant PR (2003) What Darwin's finches can teach us about the evolutionary origin and regulation of biodiversity. BioScience 53: 965-975. 93. Kornfield I, Smith PF (2000) African cichlid fishes: models systems for evolutionary biology. Ann Rev Ecol Syst 31: 163-196. 94. Danley PD, Kocher TD (2001) Speciation in rapidly diverging systems:lessons from Lake Malawi. Molec Ecol 10: 1075-1086. 95. Danley PD, Market JA, Arnegard ME, Kocher TD (2000) Divergence with gene flow in the rock-dwelling cichilds of lake Malawi. Evolution 54: 1725-1737. 96. Knight ME, Turner GF (1999) Reproductive isolation among closely related Lake Malawi cichlids: can males recognize conspecific females by visual clues? Anim Behav 58: 761-768.

145

97. Parker A, Kornfield I (1997) Evolution of the mitochondrial DNA control region in the mbuna (Cichlidae) species flock of Lake Malawi, East Africa. J Molec Evol 45: 7083. 98. Markert JA, Danley PD, Arnegard ME (2001) New markers for new species: microsatellite loci and the East African cichlids. Trends Ecol Evol 16: 100-107. 99. Seehausen O, Alphen JJM, Witte F (1997) Effects of water transparency and light regime on sexual selection and the diversity of haplochromine cichlids. Science 277: 1808-1811. 100. Stauffer JR, Bowers NJ, Kocher TD, McKaye K. R. (1996) Evidence of hybridization between Cyanotilapia afra and Psuedotropheus zebra (Teleostei: Cichlidae) following an intralacustrine translocation in Lake Malawi. Copeia: 203-208. 101. Ruber L., Meyer A, Sturmbauer C, Verheyen E (2001) Population structure in two sympatric species of the Lake Tanganyika cichlid tribe Eretmodini: evidence for introgression. Molec Ecol 10: 1207-1225. 102. McElroy DM, Kornfield I (1993) Novel jaw morphology in hybrids between Pseudotropheus zebra and Labeotropheus fullerborni (Teleostei: Cichlidae) from Lake Malawi, Africa. Copeia 1993: 933-945. 103. Markert JA, Arnegard ME, Danley PD, Kocher TD (1999) Biogeography and population genetics of the Lake Malawi cichlid Melanochromis auratus: habitat transience, philopatry and speciation. Molec Ecol 8: 1013-1026. 104. Gilbert LE (2003) Adaptive novelty through introgression in Heliconius wing patterns: Evidence for shared genetic 'tool box' from synthetic hybrid zones and a theory of diversification. In: Boggs CL, Watt WB, Ehrlich PR, editors. Ecology and Evolution Taking Flight: Butterflies as Model Systems. Chicago: Univ. of Chicago Press. 105. Jiggins CD, Naisbit RE, Coe RL, Mallet J (2001) Reproductive isolation caused by colour pattern mimicry. Nature 411: 302-305. 106. Mallet J, Gilbert LE (1995) Why are there so many mimicry rings? Correlations between habitat, behaviour and mimicry in Heliconius butterflies. Biol J Linn Soc 55: 159-180. 107. Smiley JT (1978) Plant chemistry and the evolution of host specificity: new evidence from Heliconius and Passiflora. Science 201: 745-747. 108. Mallet J, McMillan WO, Jiggins CD (1998) Mimicry and warning colour at the boundary between races and species. In: Howard DJ, Berlocher SH, editors. Endless Forms: Species and Speciation. New York: Oxford University Press. pp. 390-403. 109. Posada D, Crandall KA (2001) Intraspecific gene genealogies: trees grafting into networks. Trends Ecol Evol 16: 37-45. 110. Linder CR, Rieseberg LH (2004) Reconstructing patterns of reticulate evolution in plants. Am J Bot 9: 1700-1708. 111. Stephens M (2003) Inference under the coalescent. In: Balding DJ, Bishop M, Cannings C, editors. Hanbook of Statistical Genetics. Chichester: John Wiley and Sons Ltd. pp. 636-660. 112. Maddison WP (1997) Gene trees in species trees. Syst Biol 46: 523-536. 113. Bandelt H-J, Dress WM (1992) Split decomposition: a new and useful approach to phylogenetic analysis of distance data. Mol Phylog Evol 1: 242-252.

146

114. Rieseberg LH, Raymond O, Rosenthal DM, Lai Z, Living-stone K, et al. (2003) Major Ecological Transitions in Wild Sunflowers Facilitated by Hybridization. Science 301: 1211-1216. 115. Sattath S, Tversky A (1977) Additive similarity trees. Psychometrika 42: 319-345. 116. Huson DH (1998) SplitsTree: a program for analyzing and visualizing evolutionary data. Bioinformatics 14: 68-73. 117. Bandelt H-J, Forster P, Rohlf A (1999) Median-Joining Networks for Inferring Intraspecific Phylogenies. Mol Biol Evol 16: 37-48. 118. Xu SZ (2000) Phylogenetic analysis under reticulate evolution. Molec Biol Evol 17: 897-907. 119. Bryant D, Moulton V. NeighborNet: an agglomerative method fro the construction of planar phylogenetic networks.; 2002; WABI, Rome, Italy. Lecture Notes in Computer Science. pp. 375-391. 120. Makarenkov V (2001) T-REX: reconstructing and visualizing phylogenetic trees and ticulation networks. Bioinformatics 17: 664-668. 121. Makarenkov V, Legendre P (2004) From phylogenetic tree to a reticulated network. J Comput Biol 11: 195-212. 122. Clement MD, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 9: 1657-1660. 123. Bandelt HJ, Macaulay V, Richards M (2000) Median networks: speedy construction and greedy reduction, one simulation, and two case studies from human mtDNA. Molec Phylog Evol 16: 8-28. 124. Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491. 125. Huber kT, Watson EE, Hendy MD (2001) An algorithm for constructing local regions in a phylogenetic network. Molec Phylog Evol 19: 1-8. 126. Nakhleh L, Sun J, Warnow T, Linder R, Moret BME, et al. Reconstructing reticulate evolution in species: theory and practice.; 2004; San Diego, California, USA. pp. 337-346. 127. Rokas. A., Williams BL, King N, Carroll SB (2003) Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature 425: 798-804. 128. Nichols R (2001) Gene trees and species trees are not the same. Trends Ecol Evol 16: 358-364. 129. Arvestad L, Berglund AC, Lagergren J, Sennblad B (2003) Bayesian gene/species tree reconciliation and orthology analysis using MCMC. Bioinformatics 19: i7-i15. 130. Machado CA, Hey J (2003) The causes of phylogenetic conflict in a classic Drosophila species group. Proc R Soc Lond B 270: 1193-1202. 131. Posada D, Crandall KA (2001) Evaluation of methods for detecting recombination from DNA sequences: computer simulatons. Proc Natl Acad Sci, USA 98: 1375713762. 132. Kliman RM, Andolfatto P, Coyne JA, Depaulis F, Kreitman M, et al. (2000) The population genetics of the origin and divergence of the Drosophila simulans complex species. Genetics 156: 1913-1931. 133. Hey J, Machado CA (2003) The study of structurated populations - new hope for a difficult and divided science. Nat Rev Genet 4: 535-543. 147

134. Rosenberg NA, Nordborg M (2002) Genealogical trees, coalescent theory and the analysis of genetic polymorphisms. Nat Rev Genet 3: 380-390. 135. Hey J (2005) On the Number of New World Founders: A Population Genetic Portrait of the Peopling of the Americas. PLoS BioL 3: 0965-0975. 136. Won YJ, Hey J (2005) Divergence population genetics of chimpanzees. Molec Biol Evol 22: 297-307. 137. Won YJ, Sivasundar A, Wang Y, Hey J (2005) On the origin of Lake Malawi cichlid species: a population genetic analysis of divergence. Proc Natl Acad Sci, USA 102: 6581-6586. 138. Hey J., Nielsen R. (2004) Multilocus methods for estimating population sizes, migration rates and divergence time, with applications to the divergence of Drosophila pseudoscura and D. persimilis. Genetics 167: 747-760. 139. Machado CA, Kliman RM, Markert JA, Hey J (2002) Inferring the history of speciation from multilocus DNA sequence data: The case of Drosophila pseudoobscura and close relatives. Mol Biol Evol 19: 472-488. 140. Machado CA, Robbins N, Gilbert MTP, Herre EA (2005) Critical review of host specificity and its coevolutionary implications in the fig/fig-wasp mutualism. Proc Natl Acad Sci, USA 102: 6558-6565. 141. Beerli P, Felsenstein J (2001) Maximum likehood estimation of a migration matrix and effective population sizes in n subpopulations by using a coalescent approach. Proc Natl Acad Sci, USA 98: 4563-4568. 142. Gow J, Peichel CL, Taylor EB (2006) Contrasting hybridization rates between sympatric three-spined sticklebacks highlight the fragility of reproductive barriers between evolutionary young species. Molec Ecol 15: 739-752. 143. Rannala B, Yang Z (2003) Bayes Estimation of Species Divergence Times and Ancestral Population Sizes Using DNA Sequences from Multiple Loci. Molec Ecol 164: 1645-1656. 144. Mallet J (2005) Hybridization as an invasion of the genome. Trends Ecol Evol 20: 229-237. 145. Vaha J-P, Primmer CR (2006) Efficiency of model-based Bayesian methods for detecting hybrid individuals under different hybridization scenarios and with different numbers of loci. Molec Ecol 15: 63-72. 146. Gross BL, Rieseberg LH (2004) The Ecological Genetics of Homoploid Hybrid Speciation. J Hered 96: 241-252. 147. Neff NA, Smith GR (1978) Multivariate analysis of hybrid fishes. Syst Zool 28: 176196. 148. Humphries JM, Bookstein FL, Chernoff B, Smith GR, Elder RL, et al. (1981) Multivariate discrimination by shape in realtion to size. Syst Zool 30: 291-308. 149. Rosenthal DM, Rieseberg LH, Donovan LA (2005) Re-creating Ancient Hybrid Species Complex Phenotypes from Early-Generation Synthetic Hybrids: Three Examples Using Wild Sunflowers. Am Nat 166: 26-41. 150. Hudson RR, Coyne JA (2002) Mathematical consequences of the genealogical species concept. Evolution 56: 1557-1565. 151. Pääbo S (2003) The mosaic that is our genome. Nature 421: 409-412. 152. Takahata N, Slatkin M (1984) Mitochondrial gene flow. Proc Natl Acad Sci, USA 81: 1764-1767. 148

153. Bullini L (1994) Origin and evolution of animal hybrid species. Trends Ecol Evol 9: 422-426. 154. Mayr E (1963) Animal Species and Evolution. Cambridge, Mass.: Harvard University Press. 155. Arnold ML (1997) Natural Hybridization and Evolution. New York: Oxford University Press. 215 p. 156. Ting C, Tsaur S, Wu C (2000) The phylogeny of closely related species as revealed by the genealogy of a speciation gene, Odysseus. Proc Nat Acad, USA 97: 5313-5316. 157. Tao Y, Zeng Z, Li J, Hartl DL, Laurie C (2003) Genetiic dissection of hybrid incompatibilities between Drosophila simulans and Drosophila mauritiana. II. Mapping hybrid male sterility loci on the third chromosome. Genetics 164: 13991418. 158. Barbash DA, Awadalla P, Tarone AM (2004) Functional divergence caused by ancient positive selection of a Drosophila hybrid incompatibility locus. PLoS BioL 2: 08390848. 159. Presgraves DC, Balagopalan L, Abmayr SM, Orr HA (2003) Adaptative evolution drives divergence of a hybrid inviability gene between two species of Drosophila. Nature 423: 715-719. 160. Barbash DA, Siino DF, Tarone AM, Roote JA (2003) A rapidly evolving Myb-related protein causes species isolation in Drosophila. Proc Nat Acad, USA 100: 53025307. 161. Hutter P (2002) X-linked small GTPase and OXPHOS genes are candidates for the genetic basis of hybrid inviability in Drosophila. Dev Genes Evol 212: 504-512. 162. Perez DE, Wu CI, Johnson NA, Wu ML (1993) Genetics of reproductive isolation in the Drosophila simulans clade: DNA marker-assisted mapping and characterization of a hybrid-male sterility gene, Odysseus (Ods). Genetics 134: 261-275. 163. Wittbrodt J, Adam D, Malitschek B, Maueler W, Raulf F, et al. (1989) Novel putative receptor tyrosine kinase encoded by the melanome inducing Tu locus in Xiphophorus. Nature 341: 415-421. 164. Turelli M (1998) The causes of Haldane's rule. Science 282: 889-891. 165. Price T, Bouvier M (2002) The evolution of F1 postzygotic incompatibilities in birds. Evolution 56: 2083-2089. 166. Presgraves DC (2002) Patterns of postzygotic isolation in Lepidoptera. Evolution 56: 1168-1183. 167. Sasa MM, Chippindale PT, Johnson NA (1998) Patterns of postzygotic isolation in frogs. Evolution 52: 1811-1820. 168. Coyne JA, Orr HA (1997) Patterns of speciation in Drosophila revisited. Evolution 51: 295-303. 169. Coyne JA, Orr HA (1989) Patterns of speciation in Drosophila. Evolution 43: 362381. 170. Orr HA (1995) The population genetics of speciation: The evolution of hybrid incompatibilities. Genetics 139: 1805-1813. 171. Beltrán M, Jiggins CD, Bull V, McMillan WO, Bermingham E, et al. (2002) Phylogenetic discordance at the species boundary: gene genealogies in Heliconius butterflies. Mol Biol Evol 19: 2176-2190.

149

172. Sheppard PM, Turner JRG, Brown KS, Benson WW, Singer MC (1985) Genetics and the evolution of muellerian mimicry in Heliconius butterflies. PhilosTrans R Soc Lond B Biol Sci 308: 433-613. 173. Naisbit RE, Jiggins CD, Mallet J (2003) Mimicry: developmental genes that contribute to speciation. Evol Dev 5: 269-280. 174. Emsley MG (1965) Speciation in Heliconius (Lep., Nymphalidae): morphology and geographic distribution. Zoologica, New York 50: 191-254. 175. Brower AVZ (1994) Phylogeny of Heliconius butterflies inferred from mitochondrial DNA sequences (Lepidoptera: Nymphalidae). Mol Phylogenet Evol 3: 159-174. 176. Jiggins CD, Linares M, Nasbit RE, Salazar C, Yang ZH, et al. (2001) Sex-linked hybrid sterility in a butterfly. Evolution 55: 1631-1638. 177. Bull V, Beltrán M, Jiggins CD, McMillan WO, Bermingham E, et al. (2006) Polyphyly and Gene Flow between Non-sibling Heliconius Species. BMC Biol 4: 11. 178. Read A, Nee S (1991) Is Haldane's rule significant? Evolution 45: 1707-1709. 179. Naisbit R, Jiggins CD, Linares M, Salazar C, Mallet J (2002) Hybrid sterility, Haldane’s rule and speciation in Heliconius cydno and H. melpomene. Genetics 161: 1517-1526. 180. Orr HA (1997) Haldane's rule. Ann Rev Ecol Syst 28: 195-218. 181. Laurie CC (1997) The weaker sex is heterogametic: 75 years of Haldane's rule. Genetics 147: 937-951. 182. McMillan WO, Jiggins CD, Mallet J (1997) What initiates speciation in passion vine butterflies? Proc Natl Acad Sci, USA 94: 8628-8633. 183. Jiggins CD, McMillan WO, Neukirchen W, Mallet J (1996) What can hybrid zones tell us about speciation? The case of Heliconius erato and H. himera (Lepidoptera: Nymphalidae). Biol J Linn Soc 59: 221-242. 184. Naisbit R, Jiggins CD, Mallet J (2002) Disruptive sexual selection against hybrids contributes to speciation between Heliconius cydno and Heliconius melpomene. Proc R Soc Lond B Biol Sci 268: 1-6. 185. Linares M (1997) Origin of Neotropical mimetic biodiversity from a three-way hybrid zone of Heliconius butterflies. In: Ulrich H, editor. Proceedings of the International Symposium on Biodiversity and Systematics in Tropical Ecosystems, Bonn 1994. Bonn: Zoologishunginstitut und Museum Alexander Koenig. 186. Lewontin RC, Felsenstein J (1965) The robustness of homogenity test in 2xN tables. Biometrics 21: 19-33. 187. Stebbins GL (1950) Variation and Evolution in Plants. New York: Columbia Univ. Press. 188. Levin DA (2002) The Role of Chromosomal Change in Plant Evolution. Oxford: Oxford University Press. 189. Holloway AK, Cannatella D, C.,, Gerhardt HC, Hillis DM (2006) Polyploids with Different Origins and Ancestors From a Single Sexual Polyploid Species. Am Nat 167: E88-E101. 190. Abbott R (2003) Sex, Sunflowers, and Speciation. Science 301: 1189-1190. 191. Arnold ML, Bennett BD (1993) Natural hybridization in Louisiana irises: genetic variation and ecological determinants. In: Harrison RG, editor. Hybrid Zones and the Evolutionary Process. New York: Oxford University Press. pp. 115-139. 150

192. Jiggins CD, Estrada C, Rodrigues A (2004) Mimicry and the evolution of pre-mating isolation in Heliconius melpomene. J Evol Biol 17: 680-691. 193. Mallet J (1999) Causes and consequences of a lack of coevolution in Mullerian mimicry. Evol Ecol 13: 777-806. 194. Mallet J, Joron M (1999) Evolution of diversity in warning colour and mimicry: Polymorphisms, shifting balance and speciation. Annu Rev Ecol Syst 30: 201-233. 195. Estrada C, Jiggins CD (2002) Patterns of pollen feeding and habitat preference among Heliconius species. Ecol Entomol 27: 448-456. 196. Rodríguez FJ, Oliver JL, Marín A, Medina JR (1990) The general stochastic model of nucleotide substitution. J Theor Biol 142: 485-501. 197. Yang Z (1994) Estimating the pattern of the nucleotide sustitution. J mol Evol 39: 105-111. 198. Kronforst M (2005) Primers for the amplification of nuclear introns in Heliconius butterflies. Mol Ecol Notes 5: 158-162. 199. Carroll SB, Gates J, Keys DN, Paddock SW, Panganiban GEF, et al. (1994) Pattern formation and eyespot determination in butterfly wings. Science 265: 109-114. 200. Keys DN, Lewis DL, Selegue JE, Pearson BJ, Goodrich LV, et al. (1999) Recruitment of a hedgehog regulatory circuit in butterfly eyespot evolution. Science 283: 532534. 201. Brunetti CR, Selegue J, Monteiro A, French V, Brakefield PM, et al. (2001) The generation and diversification of butterfly eyespot color patterns. Curr Biol 11: 1578-1585. 202. Nijhout HF (1991) The Development and Evolution of Butterfly Wing Patterns. Washington, DC: Smithsonian Institution Press. 203. Hudson R (1990) Gene genealogies and the coalescent process. In: Futuyma D, Antonovics J, editors. Oxford Surveys in Evolutionary Biology. Oxford: Oxford Univ. Press. pp. 1-44. 204. Turner TL, Hahn MW, Nuzhdin SV (2005) Genomic islands of speciation in Anopheles gambiae. PLoS Biol 3: 1572-1578. 205. Besansky NJ, Krzywinski J, Lehmann T, Simard F, Kern M, et al. (2003) Semipermeable species boundaries between Anopheles gambie and Anopheles arabiensis: evidence from multilocus DNA sequence variation. Proc Natl Acad Sci USA 100: 10818-10823. 206. Maddison WP, Maddison DR (2001) MacClade version 4.02: analysis of phylogeny and character evolution. Sinauer Associates, Sunderland Massachusetts. 207. Swofford DL (2000) PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). 4. ed. Sunderland, Massachusetts: Sinauer Associates. 208. Nylander JAA (2004) MrModeltest 2.0.: Program distributed by the author.Evolutionary Biology Centre, Uppsala University. 209. Huelsenbeck JP, Ronquist FR (2001) MrBayes: Bayesian inference of phylogeny. Bioinformatics 17: 754. 210. Crandall KA, Templeton AR (1999) Statistical methods for detecting recombination. In: Crandall KA, editor. The Evolution of HIV: Johns Hopkins University Press. pp. 153-176.

151

211. McVean G, Awadalla P, Fearnhead P (2002) A Coalescent-Based Method for Detecting and Estimating Recombination From Gene Sequences. Genetics 160: 1231-1241. 212. Watterson GA (1975) On the number of segregating sites in genetical models without recombination. Theor Popul Biol 7: 256-275. 213. Rozas J, Sánchez-DelBarrio JC, Messeguer X, Rozas R (2003) DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 19: 2496-2497. 214. Tajima F (1989) Stadistical method for testing the neutral mutation hypothesis by DNA polymorhism. Genetics 123: 585-595. 215. Wall J (2000) A comparison of Estimators or the Population Recombination Rate. Mol Biol Evol 17: 156-163. 216. Hey J, Wakeley J (1997) Acoalescent estimator of the population recombination rate. Genetics 145: 833-846. 217. Nei M (1987) Molecular Evolutionary Genetics. New York: Columbia University Press. 218. Wright S (1951) The genetical structure of populations. Ann Eug 15: 323-354. 219. Lynch M, Crease TJ (1990) The analysis of population survey data on DNA sequence variation. Mol Biol Evol 7: 377-394. 220. Charlesworth B (1998) Measures of divergence between populations and the effect of forces that reduce variability. Mol Biol Evol 15: 538-543. 221. Kessing B (2000) SEQUENCER, Version 6.1.0. Distributed by the author. In: Naos Marine Laboratories STRI, editor. Panama. 222. Hudson R, Kaplan N (1985) Statistical propieties of the number of recombination events in the history of a sample of DNA sequences. Genetics 111: 147-164. 223. Hasegawa M, Kishino K, Yano T (1985) Dating the human-ape splitting by a molecular clock of mitochondrial DNA. J Mol Evol 22: 160-174. 224. Brower AVZ (1994) Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolution. Proc Natl Acad Sci USA 91: 6491-6495. 225. Nielsen R, Wakeley J (2001) Distinguishing migration from isolation. A Markov chain Monte Carlo approach. Genetics 158: 885-896. 226. Scott DW (1992) Multivariate Density Estimation: Theory, Practice, and Visualization: Wiley-interscience. 227. Rieseberg LH (2000) Crossing relationships among ancient and experimental sunfower hybrid lineage. Evolution 54: 859-865. 228. Brown KS (1981) The biology of Heliconius and related genera. Ann Rev Entomol 26: 427-456. 229. Linares M (1989) Adaptive microevolution through hybridization and biotic destruction in the neotropics [PhD Dissertation]. Austin: University of Texas. 230. Mallet J, Barton NH (1989) Strong natural selection in a warning color hybrid zone. Evolution 43: 421-431. 231. Spencer KC (1988) Chemical mediation of coevolution in the Passiflora - Heliconius interaction. In: Spencer KC, editor. Chemical Mediation of Coevolution. San Diego: Academic Press. pp. 167-240. 232. Engler HS, Spencer KC, Gilbert LE (2000) Preventing cyanide release from leaves. Nature 406: 144-145. 152

233. Smiley JT (1978) The host plant ecology of Heliconius butterflies in Northeastern Costa Rica [Thesis]: University of Texas, at Austin. 234. Mallet J, Singer MC (1987) Individual selection, kin selection, and the shifting balance in the evolution of warning colours: the evidence from butterflies. Biol J Linn Soc 32: 337-350. 235. Naisbit RE, Jiggins CD, Mallet J (2001) Disruptive sexual selection against hybrids contributes to speciation between Heliconius cydno and Heliconius melpomene. Proc R Soc Lond B Biol Sci 268: 1849-1854. 236. Edwards AWF (1972) Likelihood. Cambridge: Cambridge University Press. 237. Boggs C, Gilbert LE (1979) Male contribution to egg production in butterflies: evidence for transfer of nutrients at mating. Science 206: 83-84.

153

CONCLUSIONS

In this thesis I have demonstrated that H. heurippa is a stable entity with a constant aposematic pattern that has a large distribution range (almost 200 Km) in the Andes foothills. This distinguishes it from hybrids found in a hybrid zone in Venezuela between H. melpomene and H. cydno, where intermediate patterns are less stable and the hybrid distribution is narrow (9.28 to 17.5 Kms). Although H. heurippa is parapatric with H. melpomene in altitudinal range (H. heurippa is found between 1000-1500mts and H. melpomene between 200-1300mts) the first one maintains its integrity partly by microhabitat segregation: H. heurippa inhabits understory forest while H. melpomene inhabits secondary forest. H. heurippa is not an H. cydno or a H. melpomene cognate. Genomic compatibility experiments showed less intrinsic postzigotic isolation between H. heurippa and H. melpomene than between the last species and H. cydno. Also, five colour pattern markers (red band, red dots, red line, both, absence of brown forceps and iridescence) trace a monophyletic H. heurippa’s relation with H. melpomene while only one yellow forewing band related it with H. cydno. When a genetic screen was done with seven molecular markers, incongruent phylogenetic signals were produced. This, in addition with genetic divergence estimates (Fst) showed concordance with a classical pattern proposed for hybrid species. H. heurippa appears more closely related to H. cydno in Co, inv and w, more related to H. melpomene in Dll and intermediate to both species in Tpi. However, when I tried to distinguish between alternative hypotheses of high gene flow and a recent hybrid origin using a coalescence model, only an ancient hybridization event without recent gene flow was discarded. A more rigorous measurement of gene flow is necessary to clarify this point, but at the moment this result does not contradict the principal arguments about the hybrid origin of H. heurippa.

Molecular markers, oviposition and habitat preference suggest that H. heurippa is a hybrid species with a greater genomic contribution from H. cydno than H. melpomene, but with important introgressed colour pattern elements from the last species. H. heurippa colour pattern was reconstructed in the laboratory in basically two generations of F1 males (H. cydno females times H. melpomene males) backcrossed to H. cydno females with posterior 154

crosses between selected pseudo-F2 phenotypes. Furthermore, crosses between reconstructed individuals and between reconstructed and wild H. heurippa, bred true for the colour pattern elements, which suggests genes homology.

H. heurippa is currently isolated from its putative parents: mate choice experiments revealed strong assortative mating between the three species. This gene flow barrier is more effective than other intrinsic and extrinsic postzygotic isolation mechanisms involved in Heliconius speciation. For this reason H. heurippa is a good species in the biological sense. Additionally, mating cues are in part colour based: pattern model experiments show strong H. heurippa male preference for the colour hybrid trait, were both colour pattern elements must be present for courtship stimulation. Being an aposematic and a mating choice signal, this colour pattern solves obstacles proposed for the hybrid speciation theory. Finally, the demographic evidence and geographic distribution of H. heurippa are consistent with a two step scenario, in which the first step involves the formation of a new aposematic signal by hybridization established through genetic drift and dependent natural selection in allopatry with H. cydno, and the second step uses the same signal as a mating clue in parapatry with both parental species.

155

ANNEX

156

Vol 441|15 June 2006|doi:10.1038/nature04738

LETTERS Speciation by hybridization in Heliconius butterflies Jesu´s Mava´rez1*, Camilo A. Salazar2*, Eldredge Bermingham1, Christian Salcedo2, Chris D. Jiggins3 & Mauricio Linares2 Speciation is generally regarded to result from the splitting of a single lineage. An alternative is hybrid speciation, considered to be extremely rare, in which two distinct lineages contribute genes to a daughter species. Here we show that a hybrid trait in an animal species can directly cause reproductive isolation. The butterfly species Heliconius heurippa is known to have an intermediate morphology and a hybrid genome1, and we have recreated its intermediate wing colour and pattern through laboratory crosses between H. melpomene, H. cydno and their F1 hybrids. We then used mate preference experiments to show that the phenotype of H. heurippa reproductively isolates it from both parental species. There is strong assortative mating between all three species, and in H. heurippa the wing pattern and colour elements derived from H. melpomene and H. cydno are both critical for mate recognition by males. Homoploid hybrid speciation—hybridization without change in chromosome number—is considered very rare2–4. This has been explained by the theoretical prediction that reproductive isolation between hybrids and their parents is difficult to achieve3,5,6. However, if a hybrid phenotype directly causes reproductive isolation from parental taxa, this difficulty can be overcome. Such a role for a hybrid phenotype has been convincingly demonstrated only in Helianthus sunflowers7. In animals, the evidence for homoploid hybrid speciation is less convincing. Putative hybrid species are known with mixed genomes8–11, but in these examples shared genetic variation could also be a result of introgression subsequent to a bifurcating speciation event. Heliconius cydno and H. melpomene are two closely related species that overlap extensively in lower Mesoamerica and the Andes12. Speciation in these butterflies has not involved any change in chromosome number13 but is instead associated with shifts in colour patterns that generate both assortative mating and postzygotic isolation due to predator-mediated selection14–17. Heliconius cydno is black with white and yellow marks, whereas H. melpomene is black with red, yellow and orange marks. Both species exhibit strong positive assortative mating based on their wing colour patterns and also differ in habitat use18 and host plant preference19, but interspecific hybrids do occur at low frequency in the wild15. Heliconius heurippa has an intermediate wing pattern, which has led to the suggestion that this is a hybrid species1,20. Its hindwing is indistinguishable from that of sympatric H. m. melpomene, whereas the yellow band on its forewing is similar to that of parapatric H. cydno cordula. Ecologically, H. heurippa is most similar to H. cydno, which it replaces geographically in the eastern Andes of Colombia. Here we first establish that H. heurippa is currently genetically isolated from its putative parents and provide evidence that its genome is of hybrid origin. A bayesian assignment analysis using 12 microsatellite loci scored in populations from Panama, Colombia and Venezuela divides H. cydno (n ¼ 175), H. melpomene (n ¼ 167)

and H. heurippa (n ¼ 46) individuals into three distinct clusters (Fig. 1). Hence, H. heurippa is genetically more differentiated than any geographic race sampled of either species. Moreover, analyses of polymorphism at two nuclear genes (Invected and Distal-less) show no allele sharing between H. cydno and H. melpomene, whereas the H. heurippa genome appears as an admixture, sharing allelic variation with both putative parental species (Supplementary Fig. 2, and C.S., C.D.J. and M.L., unpublished observations). To test the hypothesis of a hybrid origin for the H. heurippa colour pattern, we performed inter-specific crosses between H. cydno

Figure 1 | Geographic distributions and genetic differentiation between H. cydno, H. melpomene and H. heurippa. H. heurippa is sympatric with H. melpomene and parapatric with H. cydno in eastern Colombia. We used the software Structure 2.1 (ref. 29) with the multilocus microsatellite data set to assign individuals to species and detect admixed individuals (namely hybrids). We ran Structure 2.1, varying the burn-in (104 to 105) and run length (105 to 106), number of clusters (one to four), ancestry type (with and without admixture) and allele frequency estimation (correlated and independent) to obtain the highest probability model for the data set. The upper inset shows the results obtained with the best model (three clusters, admixture and independent estimations of allele frequencies). The relative contributions of the three clusters to each individual’s genome are shown in the following colours: blue, H. cydno; red, H. melpomene; green, H. heurippa. Collection site codes: A, Pipeline Road, Panama; B, Parcela 33, Colombia; C, San Cristo´bal, Venezuela; D, La Gira, Venezuela; E, Ocache, Colombia; F, Villavicencio, Colombia. An expanded view of the clusters is shown in Supplementary Fig. 1.

1 Smithsonian Tropical Research Institute, Apartado postal 0843-03092, Panama´, Repu´blica de Panama´. 2Instituto de Gene´tica, Universidad de los Andes, Carrera 1E No 18ª–10, PO Box 4976, Santafe´ de Bogota´ D.C., Colombia. 3Institute of Evolutionary Biology, School of Biological Sciences, University of Edinburgh, West Mains Road, Edinburgh EH9 3JT, UK. *These authors contributed equally to this work.

868

© 2006 Nature Publishing Group

LETTERS

NATURE|Vol 441|15 June 2006

Figure 2 | Reconstruction of the H. heurippa wing pattern. All fore- and hind-wings shown in dorsal and ventral views, respectively. a, First row, H. cydno cordula, H. heurippa and H. m. melpomene; second row, H. cydno cordula £ H. m. melpomene F1 hybrid. Backcrosses to H. cydno cordula and H. m. melpomene are shown in the left and right boxes, respectively. Other individuals are wild hybrids from San Cristo´bal, Venezuela, shown next to their putative genotypes. Three major loci regulate the patterns: complete (BB, H. melpomene), intermediate (Bb, heterozygotes) or no (bb, H. cydno) expression of a red band on the dorsal forewing; complete (NNNN, H. cydno), intermediate (NNNB, heterozygotes) or no (NBNB, H. melpomene) expression of a yellow band on the dorsal forewing; and complete (BrBr, H. cydno), intermediate (Brbr, heterozygotes) or no (brbr, H. melpomene) expression of a brown pincer-shaped mark on the ventral hindwing. Further crosses were performed using individuals with the phenotype marked with an asterisk (see b). b, Left box: offspring from crosses between individuals marked with an asterisk in a. The B and Br loci are linked, explaining the absence of the two recombinant genotypes BbNNNNbrbr and bbNNNNbrbr. Right box: offspring of a cross between a laboratory hybrid with genotype BBNNNNbrbr and H. heurippa, showing that the pattern breeds true. The other individual (M130) is a wild hybrid from San Cristo´bal, Venezuela, with a phenotype very similar to H. heurippa.

cordula and H. m. melpomene to reconstruct the steps of introgressive hybridization that could have given rise to H. heurippa. The colour pattern differences between H. m. melpomene and H. cydno cordula are determined largely by three co-dominant loci controlling the red and yellow bands on the forewing and the brown pincer-shaped mark on the ventral hindwing (see Fig. 2a)21,22. Most H. cydno £ H. melpomene F1 hybrids seem intermediate to both parents (Fig. 2a), with both a yellow (cydno) and a red (melpomene) band in the median section of the forewing, whereas the ventral side of the hindwing shows a reduced brown mark intermediate between the parental species. Female F1 hybrids resulting from crosses between H. melpomene and H. cydno are sterile in accordance with Haldane’s rule1,23, and thus only male F1 hybrids backcrossed to either H. cydno cordula or H. m. melpomene females resulted in offspring. Backcrosses to H. melpomene produced offspring very similar to pure H. m. melpomene, and further backcross generations never produced

individuals with forewing phenotypes similar to H. heurippa (Fig. 2a). However, after only two generations a phenotype virtually identical to H. heurippa (Supplementary Fig. 3) was produced by backcrossing an F1 male to an H. cydno cordula female and then mating selected offspring of this cross (Fig. 2b). In offspring of crosses between these H. heurippa-like individuals the pattern breeds true, showing that they are homozygous for the red forewing band (BB) and the absence of brown hindwing marks (brbr) characteristic of H. melpomene, and similarly homozygous for the yellow forewing band (NNNN) derived from H. cydno. The pattern of these H. heurippa-like individuals also breeds true when crossed to wild H. heurippa (Fig. 2b), implying that pattern genes segregating in our crosses are homologous with those in wild H. heurippa. Furthermore, in a wild population of sympatric H. m. melpomene and H. cydno cordula in San Cristo´bal, Venezuela, we observed natural hybrids at an unusually high frequency (8%), including some individuals very similar to our laboratory-produced H. heurippa-like butterflies (Fig. 2b). Microsatellite data show that these individuals have genotypes indistinguishable from that of H. cydno and must therefore be at least fifth-generation backcrosses (Supplementary Fig. 4). This shows that multiple generations of backcrossing can occur in the wild and that female hybrid sterility is not a complete barrier to introgressive hybridization. The fact that the H. heurippa pattern can be generated by laboratory crosses between H. melpomene and H. cydno, and is also observed in wild hybrids between the two species, establishes a probable natural route for the hybrid origin of H. heurippa. The next step in species formation is reproductive isolation. We therefore tested the degree to which H. heurippa is isolated from H. melpomene and H. cydno by assortative mating. No-choice mating experiments showed a reduced probability of mating in all interspecific comparisons, with H. heurippa females particularly unlikely to mate with either H. cydno or H. melpomene (Table 1). When a male of each species was presented with a single female, H. heurippa males were tenfold more likely to court their own females than the other species (Supplementary Fig. 5). In mating experiments with choice, there was similarly strong assortative mating, although occasional matings between H. cydno and H. heurippa were observed (Table 2). Isolation due to assortative mating, on average more than 90% between H. heurippa and H. melpomene and more than 75% between H. heurippa and H. cydno, is therefore considerably greater than that caused by hybrid sterility (about 25% isolation between H. heurippa and H. melpomene, and zero between H. heurippa and H. cydno)1 or predator selection against hybrids (about 50%)24. Therefore, strong assortative mating, in combination with geographic isolation

© 2006 Nature Publishing Group

869

LETTERS

NATURE|Vol 441|15 June 2006

Table 1 | Relative mating probabilities in no-choice experiments Male

H. melpomene H. cydno H. heurippa

H. melpomene

Female H. cydno

H. heurippa

1 (17) 0.120 (0.048–0.231, 50) 0.100 (0.031–0.022, 40)

0.178 (0.084–0.309, 45) 1 (27) 0.440 (0.255–0.637, 44)

0.073 (0.023–0.163, 55) 0.022 (0.001–0.096, 45) 1 (22)

For each female type, probabilities were estimated relative to that of intra-specific mating, which was set to 1. Numbers in parenthesis show the 95% maximum-likelihood support limits and the number of females used.

from H. cydno and postzygotic isolation from H. melpomene has contributed to the speciation of H. heurippa. We next investigated the role of colour pattern in mate choice. Experiments with dissected wings showed that both elements of the forewing colour pattern of H. heurippa were necessary for the stimulation of courtship (Fig. 3). H. heurippa males were less than half as likely to approach and court the H. m. melpomene or the H. cydno cordula pattern than their own (Fig. 3). When either the red or yellow bands were experimentally removed from the H. heurippa pattern, this led to a similar reduction in its attractiveness, demonstrating that both hybrid elements are necessary for mate recognition by male H. heurippa (Fig. 3). Similar results were obtained when these experiments were replicated with printed-paper models (Fig. 3), showing that the colour pattern itself was the cue rather than pheromones associated with the dissected wings. Additional experiments showed that males of both H. m. melpomene and H. cydno cordula showed a greatly reduced probability of approaching and courting the H. heurippa pattern than their own (Supplementary Figs 6 and 7). Given the incomplete postzygotic reproductive isolation between all three species1, this pattern-based assortative mating must have a continuing role in generating reproductive isolation between H. heurippa and its relatives. Novel patterns in Heliconius probably become established through a combination of genetic drift and subsequent fixation of the novel pattern driven by frequency-dependent selection25. Such an event could have established the hybrid H. heurippa pattern as a geographic isolate of H. cydno. Subsequently, the pattern was sufficiently distinct from both H. melpomene and H. cydno that mate-finding behaviour also diverged in parapatry, generating assortative mating between all three species (Supplementary Fig. 8). This two-stage process indicates a possible route by which the theoretical difficulty of a rapid establishment of reproductive isolation between the hybrid and the parental taxa could have been overcome5,6. Furthermore, because we are proposing divergence in mate behaviour in a geographically isolated population, reinforcement or some other form of sympatric divergence is not required for speciation to occur. Our study provides the first experimental demonstration of a hybrid trait generating reproductive isolation between animal species, and the first example of a hybrid trait causing pre-mating isolation through assortative mating. None of the theoretical treatments of homoploid hybrid speciation have considered the effects of assortative mating5,6. If variation for mate preference were incorporated, the theoretical conditions favouring hybrid speciation might

not be as stringent as has been supposed. Finally, two other species, H. pachinus20 and H. timareta26, have also been proposed as having H. cydno/H. melpomene hybrid patterns, indicating that this process might have occurred more than once. However, whether these cases represent a particularity of Heliconius or a common natural process that has been undetected in other animal groups studied less intensively remains a matter of further study. Suggestively, other proposed cases of homoploid hybrid speciation in animals occur in well-studied groups such as African cichlids8–10 and Rhagoletis flies11. METHODS Crosses. Crosses were performed in La Vega, Colombia, between January 2000 and December 2002 with the use of H. m. melpomene (Virgen de Chirajara, 4.2138 N, 73.7958 W) and H. cydno cordula (Barro Negro, 6.0168 N, 72.0918 W), both from the Colombian Eastern Cordillera. We isolated virgin females with older males to produce inter-specific F1 offspring and backcrosses. After mating, females were kept individually in 2 £ 3 £ 2 m3 insectaries and supplied with pollen and nectar from Psiguria and Lantana flowers, and Passiflora vines for oviposition22. Eggs and the larval and adult stages were reared as described

Table 2 | Number of matings in tetrad mate-choice experiments Female

H. melpomene H. heurippa H. cydno H. heurippa H. melpomene H. cydno

Male

H. melpomene 15 0 H. cydno 5 0 H. melpomene 10.5 0

H. heurippa 0 12 H. heurippa 3 5 H. cydno 0 5.5

Mating results of 0.5 are due to simultaneous mating of both pairs during the experiment.

870

Figure 3 | Relative probabilities of H. heurippa males approaching and courting colour pattern models. Values are estimated relative to the probability of approach (a) and courtship (b) of the H. heurippa control pattern shown at the top (probability equal to 1). Model patterns are H. m. melpomene (melp), H. cydno cordula (cord), H. heurippa modified to remove the red band (heuA) and H. heurippa modified to remove the yellow band (heuR). Red, real wings; blue, paper wings. Error bars show maximum-likelihood support limits (see Supplementary Information for an explanation).

© 2006 Nature Publishing Group

LETTERS

NATURE|Vol 441|15 June 2006

previously23. Colour pattern segregation and designation were studied with the use of nomenclature described previously22. No-choice experiments. Experiments without choice measure the reluctance of males and females to mate inter-specifically; such experiments simulate a natural situation in which males encounter females singly. In the insectaries mentioned above, a virgin female (one to three days old) of each species was presented to ten mature males (more than ten days old) for two days. Males were used only once. Matings were recorded every 30 min from 06:00 until 15:00. Experiments were performed between all combinations of the three species, including control experiments between conspecifics. To detect any unobserved mating, all the females were checked after the experiments for the presence of a spermatophore in their reproductive tract. Tetrads. Experiments with choice were performed to estimate mating probability in a situation in which males encounter females from different species simultaneously. Pairwise experiments were performed between each combination of the three species, in which a single recently emerged virgin female and a mature male (more than ten days old) of each species were placed in a 2 £ 3 £ 2 m3 insectary over the course of a single day. Thus, each experiment involved four butterflies, for example a female and male H. heurippa and a female and male H. cydno for the comparison between those two species. The first mating only was recorded for each experiment, and individuals were not reused. In cases in which both pairs mated simultaneously they were scored as each having half a mating. At least 15 experiments were performed for each pairwise comparison. Mating probabilities were estimated by likelihood (Supplementary Methods). Colour pattern models. In Heliconius, males use colour patterns to locate females and are choosy in mating, presumably because of the large and costly spermatophore transferred to females27. We investigated male preferences for different colour pattern models. Between 10 and 20 males in a 2 £ 3 £ 2 m3 insectary were presented with either dissected natural wings or a printed colour pattern model, fixed to a length of flexible clear nylon. Models were manipulated to simulate Heliconius flight in the centre of a spherical area (60 cm in diameter) demarcated by references in the insectary roof. Randomly ordered pairs of 5-min experiments were performed: first, a control flight with a model of the male’s own colour pattern, and second, an experimental flight with a different colour pattern. Entry to the sphere was recorded as ‘approach’ and sustained fluttering directed at the model as ‘courtship’. At least 25 replicates were performed for each comparison. In addition, models were made in which the H. heurippa pattern was modified to show either the yellow band without red (Heu-A) or the red band without yellow (Heu-R). For the dissected wing models, permanent black marker pen (Pilot ultrafine point no xylene SCA-UF) was used to cover the corresponding band. Paper models were made from digital photographs of wings taken with a Sony Cyber-shot dsc-s85 camera that were printed with a high-performance inkjet printer (Hewlett Packard Deskjet 3820) on special photo-quality calcium paper. Only paper models with a reflectance spectra similar to real wings were used. The data were used to estimate the probabilities Q ij that males of type j approached or courted models of type i relative to that of their own type j (which was set to one), using likelihood. Confidence intervals for parameters were obtained as the values that decreased the difference between two loge likehoods by two units (Supplementary Methods). Microsatellites. Twelve microsatellite loci were genotyped (Hel02, Hel04, Hel05, Hm01, Hm02, Hm03, Hm04, Hm05, Hm06, Hm13, Hm19 and Hm22)28 for 60 individuals of Heliconius heurippa and at least 24 individuals from each of five populations of both H. melpomene and H. cydno. Collection sites were as follows: H. c. chioneus and H. m. rosina from Pipeline Road, Panama (9.1228 N, 79.7158 W); H. c. cordula, H. m. melpomene from San Cristo´bal, Venezuela (7.7678 N, 72.2258 W); H. c. chioneus and H. m. melpomene from Parcela 33, Colombia (5.0668 N, 74.5618 W); H. heurippa and H. m. melpomene from near Villavicencio, Colombia (4.1518 N, 73.6358 W); H. c. weymeri and H. c. cydnides from Ocache, Colombia (3.7038 N, 76.4938 W); H. c. barinasensis and H. m. melpomene from La Gira, Venezuela (9.3348 N, 70.7308 W). The genetic structure of populations was analysed with bayesian assignment tests as implemented in Structure 2.1 (ref. 29). Received 8 November 2005; accepted 22 March 2006. 1.

2. 3. 4.

Salazar, C. A. et al. Hybrid incompatibility is consistent with a hybrid origin of Heliconius heurippa Hewitson from its close relatives, Heliconius cydno Doubleday and Heliconius melpomene Linnaeus. J. Evol. Biol. 18, 247–-256 (2005). Rieseberg, L. H. Hybrid origins of plant species. Annu. Rev. Ecol. Syst. 28, 359–-389 (1997). Coyne, J. A. & Orr, H. A. Speciation (Sinauer, Sunderland, Massachusetts, USA, 2004). Gross, B. L. & Rieseberg, L. H. The ecological genetics of homoploid hybrid speciation. J. Hered. 96, 241–-252 (2005).

5. 6. 7. 8. 9.

10. 11.

12. 13.

14. 15.

16. 17.

18.

19. 20. 21. 22. 23.

24. 25.

26. 27. 28. 29.

McCarthy, E. M., Asmussen, M. A. & Anderson, W. W. A theoretical assessment of recombinational speciation. Heredity 74, 502–-509 (1995). Buerkle, C. A., Morris, R. J., Asmussen, M. A. & Rieseberg, L. H. The likelihood of homoploid hybrid speciation. Heredity 84, 441–-451 (2000). Rieseberg, L. H. Crossing relationships among ancient and experimental sunflower hybrid lineages. Evolution Int. J. Org. Evolution 54, 859–-865 (2000). Salzburger, W., Baric, S. & Sturmbauer, C. Speciation via introgressive hybridization in East African cichlids? Mol. Ecol. 11, 619–-625 (2002). Smith, P. F., Konings, A. & Kornfield, I. Hybrid origin of a cichlid population in Lake Malawi: implications for genetic variation and species diversity. Mol. Ecol. 12, 2497–-2504 (2003). Seehausen, O. Hybridization and adaptive radiation. Trends Ecol. Evol. 19, 198–-207 (2004). Schwarz, D., Matta, B. M., Shakir-Botteri, N. L. & McPheron, B. A. Host shift to an invasive plant triggers rapid animal hybrid speciation. Nature 436, 546–-549 (2005). Brown, K. S. The biology of Heliconius and related genera. Annu. Rev. Entomol. 26, 427–-456 (1981). Brown, K. S., Emmel, T. C., Eliazar, P. J. & Suomalainen, E. Evolutionary patterns in chromosome numbers in neotropical Lepidoptera. I. Chromosomes of the Heliconiini (Family Nymphalidae: Subfamily Nymphalinae). Hereditas 117, 109–-125 (1992). McMillan, W. O., Jiggins, C. D. & Mallet, J. What initiates speciation in passion-vine butterflies? Proc. Natl Acad. Sci. USA 94, 8628–-8633 (1997). Mallet, J., McMillan, W. O. & Jiggins, C. D. in Endless Forms. Species and speciation (eds Howard, D. J. & Berlocher, S. H.) 390–-403 (Oxford Univ. Press, New York, 1998). Jiggins, C. D., Naisbit, R. E., Coe, R. L. & Mallet, J. Reproductive isolation caused by colour pattern mimicry. Nature 411, 302–-305 (2001). Jiggins, C. D., Estrada, C. & Rodrigues, A. Mimicry and the evolution of premating isolation in Heliconius melpomene Linnaeus. J. Evol. Biol. 17, 680–-691 (2004). Mallet, J. & Gilbert, L. E. Why are there so many mimicry rings? Correlations between habitat, behaviour and mimicry in Heliconius butterflies. Biol. J. Linn. Soc. 55, 159–-180 (1995). Smiley, J. T. Plant chemistry and the evolution of host specificity: new evidence from Heliconius and Passiflora. Science 201, 745–-747 (1978). Gilbert, L. E. in Butterflies: Ecology and Evolution Taking Flight (eds Boggs, C. L., Watt, W. B. & Ehrlich, P. R.) 281–-318 (Univ. of Chicago Press, Chicago, 2003). Linares, M. Adaptive Microevolution Through Hybridization and Biotic Destruction in the Neotropics. Thesis, University of Texas, Austin (1989). Naisbit, R. E., Jiggins, C. D. & Mallet, J. Mimicry: developmental genes that contribute to speciation. Evol. Dev. 5, 269–-280 (2003). Naisbit, R. E., Jiggins, C. D., Linares, M., Salazar, C. A. & Mallet, J. Hybrid sterility, Haldane’s rule and speciation in Heliconius cydno and H. melpomene. Genetics 161, 1517–-1526 (2002). Mallet, J. & Barton, N. H. Strong natural selection in a warning-color hybrid zone. Evolution Int. J. Org. Evolution 43, 421–-431 (1989). Mallet, J. & Singer, M. C. Individual selection, kin selection and the shifting balance in the evolution of warning colours: the evidence from butterflies. Biol. J. Linn. Soc. 32, 337–-350 (1987). Mallet, J. Causes and consequences of a lack of coevolution in Mu¨llerian mimicry. Evol. Ecol. 13, 777–-806 (1999). Boggs, C. L. & Gilbert, L. E. Male contribution to egg production in butterflies: evidence for transfer of nutrients at mating. Science 206, 83–-84 (1979). Mava´rez, J. & Gonza´lez, M. A set of microsatellite loci for Heliconius melpomene and close relatives. Mol. Ecol. Notes 6, 20–-23 (2006). Pritchard, J. K., Stephens, M. & Donnelly, P. J. Inference of population structure using multilocus genotype data. Genetics 155, 945–-959 (2000).

Supplementary Information is linked to the online version of the paper at www.nature.com/nature. Acknowledgements We thank E. Garcı´a and the UNET for help at Paramillo Natural Park, San Cristo´bal, Venezuela; R. Castillo, L. Pereira and O. Quintero for butterfly collecting; M. Guerra and L. Gonza´lez for help with the preparation of figures; N. Barton and F. Jiggins for discussion; and L. Gilbert and J. Mallet for inspiring us to study hybridization. This work was funded by the Marie-Curie Fellowships, the Smithsonian Tropical Research Institute, the Fondo Colombiano de Investigaciones Cientı´ficas y Proyectos Especiales Francisco Jose de Caldas COLCIENCIAS, Banco de la Repu´blica, and private donations from Continautos S.A., Proficol El Carmen S.A., Didacol S.A., and F. Arango, Colombia. C.D.J. is supported financially by the Royal Society and by a grant from BBSRC. Author Information The sequences have been deposited in GenBank under accession numbers DQ445384–DQ445414 (Distal-less) and DQ445416– DQ445457 (Invected). Reprints and permissions information is available at npg.nature.com/reprintsandpermissions. The authors declare no competing financial interests. Correspondence and requests for materials should be addressed to M.L. ([email protected]) or J.M. ([email protected]).

© 2006 Nature Publishing Group

871

Supplementary Information Supplementary Methods Mavárez et al 'Speciation by hybridization in Heliconius butterflies'.

Mate choice experiments: No-choice ML model 1. A binomial mating probability P i x j was obtained for each combination of i-type female and j-type male, maximizing the expression for loge-likelihood given by: m loge P i x j + n loge (1- P i x j) where m and n are the numbers of trials in which the pair mated or remained unmated, respectively. The loge-likehoods for the P i x j values were maximized using the SOLVER algorithm supplied with Microsoft EXCEL. Support limits for P i x j , which are asymptotically equivalent to 95% intervals, where obtained at the parameter values that led to a decrease in the loge-likelihood of two units. Mate choice experiments: Colour Pattern Models ML model 2,3. We estimated the probabilities Q i x j that males type j approached or courted models type i relative to that of their own type j, using likelihood and setting the model to one, so that any value under one represented preference for the control model. Thus, for M males with M versus C models, the actual probabilities are QA c x m/(QA c x m + 1) that males approach C and 1/(QA c x m + 1) that they approach M. The loge-likelihood for this experiment is therefore ™[XA c x m loge { QA c x m/(

QA c x m + 1)} + XA m x m loge{1/(QA c x m + 1)}], where XA c x m is the number of M males

approaching C and XA m x m is the number approaching M. Similarly QH i x j parameters were estimated for probability of hovering courtship of the model. The summed loge-likelihood was maximized over all experiments by varying the QH i x j parameters. Confidence intervals for parameters were established using the same likelihood method described for mating experiments.

References 1. Naisbit, R. E., Jiggins, C. D. & Mallet, J. Disruptive sexual selection against hybrids contributes to speciation between Heliconius cydno and Heliconius melpomene. Proc. Roy. Soc. B. 268, 1849-1854 (2001). 2. Jiggins, C. D., Naisbit, R. E., Coe, R. L. & Mallet, J. Reproductive isolation caused by colour pattern mimicry. Nature 411, 302-305 (2001). 3. Jiggins, C. D., Estrada, C. & Rodrigues, A. Mimicry and the evolution of premating isolation in Heliconius melpomene Linnaeus. J. Evol. Biol. 17, 680-691 (2004). 4. Pritchard, J. K., Stephens, M. & Donnelly, P. J. Inference of population structure using multilocus genotype data. Genetics 155, 945-959 (2000). 5. Clement MD, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Mol. Ecol. 9, 1657-1660. 6. Bookstein, F. L. Morphometric Tools for Landmark Data: Geometry and Biology (Cambridge University Press, Cambridge, 1991). 7. Adams, D. C., Rohlf, F. J. & Slice, D. E. Geometric Morphometrics: Ten Years of Progress Following the ‘Revolution’. Ital. J. Zool. 71, 5-16 (2004). 8. Rohlf, F. J. & Marcus, L. F. A revolution in morphometrics. Trends Ecol. Evol. 8, 129-132 (1993). 9. Rohlf, F. J. & Slice, D. E. Extensions of the Procrustes method for the optimal superimposition of landmarks. Syst. Zool. 39, 40-59 (1990)

Supplementary Figure 1. Reproductive isolation between H. cydno, H. melpomene and H. heurippa. We used the software Structure 2.1 4 with the multilocus microsatellite data set to assign individuals to species and detect admixed individuals (i.e. hybrids). We ran Structure 2.1 varying the burnin (104 to 105) and run length (105 to 106), number of clusters (1 to 4), ancestry type (with and without admixture) and allele frequency estimation (correlated and independent), in order to

obtain the highest probability model for the dataset. The figure shows the results obtained with the best model (three clusters, admixture and independent estimations of allele frequencies). The relative contribution of the three clusters to each individual's genome is shown in the Y axis. Blue: H. cydno, Red: H. melpomene and green: H. heurippa. Collection sites: Pipeline road, Panama (1-41 H. cydno, 42-86 H. melpomene); Parcela 33, Colombia (87-110 H. cydno, 111-133 H. melpomene); San Cristóbal, Venezuela (134-163 H. cydno, 164-198 H. melpomene); La Gira, Venezuela (199-244 H. cydno, 245-270 H. melpomene); Ocache, Colombia (271304 H. cydno) and Villavicencio, Colombia (305-342 H. melpomene, 343-388 H. heurippa).

Distal-less

Invected

Supplementary Figure 2. Allele networks for nuclear genes (CS, CDJ and ML unpublished). Yellow, red and blue are H. c. cordula, H. m. melpomene and H. heurippa alleles. Respective alleles are identified with the letters C, M and H, followed by the individual number and allele number. Black dots are hypothetical ancestors. Sizes of the circles reflect allele frequencies in the population. Networks were constructed with statistical parsimony in TCS v 1.21 5 (a method that allows the identification of putative recombination events by looking at the spatial distribution in the sequence of the homoplasies defined by the network), considering gaps as missing data and adjusting the parsimony limit to the respective data set.

Supplementary Figure 3. Morphological comparison of reconstructed hybrids and wild H. heurippa. The shape of the forewing band was studied in 30 H. heurippa females and 37 H. heurippa-like lab hybrid females, using geometric morphometrics methods 6-8. We took X,Y coordinates of four homologous landmarks that describe the general shape of the forewing band in scanned pictures of wings. The variation unexplained by the shape of the forewing band was removed using Generalized Procrustes Superimposition 7,9. The canonical analysis of variance (CAV) used to test for significant differences in the shape of the forewing band between wild H. heurippa and H. heurippa-like lab hybrids found no differences in the wing pattern shapes of the two groups (Wilk's Lambda = 0.9046, p = 0.1768).

Supplementary Figure 4. Introgression analysis of H. cydno and H. melpomene at San Cristóbal, Venezuela. The same microsatellite loci described in Methods were scored in 36 H. melpomene (ind. 1 to 36), 44 H. cydno (ind. 37 to 80) and 9 hybrids (ind. 81 to 89) from Paramillo Natural Park, San Cristóbal, Venezuela. We used the software Structure 2.1 4, in order to assign individuals to species and to detect admixed individuals (i.e. hybrids), based on their multilocus microsatellite genotype. We varied the burnin (104 to 105), run length (105 to 106), number of clusters (1 to 4), ancestry type (with and without admixture) and allele frequency estimation (correlated and independent), in order to obtain the highest probability model for the dataset (which was obtained with two clusters, admixture and independent estimations of allele frequencies). The contribution of each cluster to each individual's genome was used as an introgression index, useful to describe its genotypic class. For example, F1, first-, second-, third- and fourth-generation backcrosses to H. cydno should have, on average, 50%, 25%, 12.5%, 6.25% and 3.13% of their genome introgressed from H. melpomene, respectively. Colours represent the contribution of each species-cluster (blue: H. melpomene, red: H. cydno) to the introgression index. The hybrid individuals cannot be distinguished from other individuals of H. cydno, indicating that multiple generations of backcrossing must have occurred.

Supplementary Figure 5. Live courtship experiments. We placed mature males (> 5 days old) of each genotype in an insectary and introduced a single virgin female (1-5 days old). The number of courtship bouts (sustained hovering by the male over the female) occurring over a period of 15 minutes was recorded. The female genotype was then substituted, with genotype order randomized, such that each panel of males was tested against each female genotype. No mating was permitted in order to not disrupt subsequent behaviour. Males were used only once. In total 30 replicate panels of males were tested against all three female genotypes in more than 450 min of observations. H. melpomene (red dots), H. heurippa (blue dots) and H. cydno (grey dots).

Supplementary Figure 6. Colour pattern models for H. melpomene. Model experiments were carried out as described in Methods using H. melpomene males. In all comparisons, males of H. melpomene prefer to approach and court the real wings and paper models of their own phenotype.

Supplementary Figure 7. Colour pattern models for H. cydno males. Model experiments were carried out as described in Methods using H. cydno males. H. cydno prefer to approach and court the real wings and paper models of their own phenotype and real wings of H. heurippa with the red band removed.

Supplementary Figure 8. A model for the hybrid origin of Heliconius heurippa.