Search

How frequent is gene flow between species? The pattern of evolution is typically portrayed as a phylogenetic tree, implying that speciation is a series of splits between lineages. Yet gene flow between good species is increasingly recognized as an important mechanism in the diversification of radiations, often spreading adaptive traits and leading to a complex pattern of phylogenetic incongruence. This process has thus far been studied in cases involving few species, or geographically restricted to spaces like islands, but not on the scale of a continental radiation. Previous studies have documented gene flow, adaptive introgression and hybrid speciation in a small subsection of the charismatic Neotropical butterflies Heliconius. Using genome-wide resequencing of 40 out of 45 species in the genus we demonstrate for the first time that admixture has played a role throughout the evolution of Heliconius and the sister genus Eueides. Modelling of phylogenetic networks based on 6848 orthologous autosomal genes (Maximum Pseudo-Likelihood Networks) or 5,483,419 high quality SNPs (Ancestral Recombination Graph) uncovers nine new cases of interspecific gene flow at up to half of the genome. However, f4 statistics of admixture show that the extent of the process has varied between subgenera. Evidence for introgression is found at all five loci controlling the colour and shape of the mimetic wing patterns, including in the putative hybrid species H. hecalesia, characterised by an unusual hindwing. Due to hybridization and incomplete coalescence during rapid speciation, individual gene trees show rampant discordance. Although reduced gene flow and faster coalescence are expected at the Z chromosome, we discover high levels of conflict between the 416 sex-linked loci. Despite this discordant pattern, both concatenation and multispecies coalescent approaches yield surprisingly consistent and fully supported genome-wide phylogenies. We conclude that the imposition of the bifurcating tree model without testing for interspecific gene flow may distort our perception of adaptive radiations and thus the ability to study trait evolution in a comparative framework.

Numerous animal lineages have expanded and diversified the opsin-based photoreceptors in their eyes underlying color vision behavior. However, the selective pressures giving rise to new photoreceptors and their spectral tuning remain mostly obscure. Previously, we identified a violet receptor (UV2) that is the result of a UV opsin gene duplication specific to Heliconius butterflies. At the same time the violet receptor evolved, Heliconius evolved UV-yellow coloration on their wings, due to the pigment 3-hydroxykynurenine (3-OHK) and the nanostructure architecture of the scale cells. In order to better understand the selective pressures giving rise to the violet receptor, we characterized opsin expression patterns using immunostaining (14 species) and RNA-Seq (18 species), and reconstructed evolutionary histories of visual traits in five major lineages within Heliconius and one species from the genus Eueides. Opsin expression patterns are hyperdiverse within Heliconius. We identified six unique retinal mosaics and three distinct forms of sexual dimorphism based on ommatidial types within the genus Heliconius. Additionally, phylogenetic analysis revealed independent losses of opsin expression, pseudogenization events, and relaxation of selection on UVRh2 in one lineage. Despite this diversity, the newly evolved violet receptor is retained across most species and sexes surveyed. Discriminability modeling of behaviorally preferred 3-OHK yellow wing coloration suggests that the violet receptor may facilitate Heliconius color vision in the context of conspecific recognition. Our observations give insights into the selective pressures underlying the origins of new visual receptors.

ToxicHeliconiusbutterflies have yellow hindwing bars that – unlike their closest relatives – reflect ultraviolet (UV) and long wavelength light, and also fluoresce. The pigment in the yellow scales is 3-hydroxy-DL-kynurenine (3-OHK), found also in the hair and scales of a variety of animals. In other butterflies including pierids, which similarly display wing colors that vary in both the UV and the human-visible range, behavioral experiments have indicated that only the UV component is most relevant to mate choice. Whether inHeliconiusbutterflies it is the UV, the human-visible yellow, and/or the fluorescent component of yellow wing coloration that is relevant to mate choice is unknown. In field studies with butterfly paper models we show that both UV and 3-OHK yellow act as signals forH. eratobut attack rates by birds do not differ significantly between the models. Furthermore, measurement of the quantum yield and reflectance spectra of 3-OHK indicates that fluorescence does not contribute to the visual signal under broad-spectrum illumination. Our results suggest that the use of 3-OHK pigmentation instead of ancestral yellow was driven by sexual selection rather than predation.Summary statementHeliconiusbutterflies use a co-opted yellow pigment for communication, while predators are fooled by non-Heliconiusmimics using ancestral yellow pigments.

Analysis of the known and recently discovered differentiation patterns of 58 forest butterfly species (of the total 66) in the tribe Heliconiini, including over 300 races and forms, has led to the drawing of a map of core areas for evolution in the group. The majority of these are not isolated by any obvious modern physiogeographic barriers to dispersal of these butterflies, and probably correspond to forest refugia during the two or three most recent Quaternary dry periods. The map is in good accord with the results of similar studies in other groups of animals (forest vertebrates and invertebrates) and in forest trees, but indicates a larger number of significant operative refugia (38) than other analyses. Bases for the understanding of differentiation patterns in polytypic continental species are developed, in accord with the refuge theory and using intensive and extensive data on local monomorphic populations and secondary contact zones between differentiated forms. The known forms discussed include: Dione juno miraculosa Hering, Eueides libitina Staudinger, Heliconius xanthocles explicata Stichel, H. eleuchia (Hewitson) and H. leucadia Bates together with a supplementary revision of the tribe giving information published since 1972. The following new forms are described: Eueides emsleyi sp. n. and nine new subspecies of Heliconius: H. aoede philipi, H. wallacei araguaia, H. pardalinus orteguaza, H.p. butleri, H. hecale australis, H.h. barcanti, H. clysonymus tabaconas, H. congener aquilionaris, H. sapho candidus. Geographical details of the refugia are given in Appendix I, and in Appendix II the heliconians are assigned to the postulated refugia.

The Heliconius butterflies are a widely studied adaptive radiation of 46 species spread across Central and South America, several of which are known to hybridise in the wild. Here, we present a substantially improved assembly of the Heliconius melpomene genome, developed using novel methods that should be applicable to improving other genome assemblies produced using short read sequencing. Firstly, we whole genome sequenced a pedigree to produce a linkage map incorporating 99% of the genome. Secondly, we incorporated haplotype scaffolds extensively to produce a more complete haploid version of the draft genome. Thirdly, we incorporated ~20x coverage of Pacific Biosciences sequencing and scaffolded the haploid genome using an assembly of this long read sequence. These improvements result in a genome of 795 scaffolds, 275 Mb in length, with an L50 of 2.1 Mb, an N50 of 34 and with 99% of the genome placed and 84% anchored on chromosomes. We use the new genome assembly to confirm that the Heliconius genome underwent 10 chromosome fusions since the split with its sister genus Eueides, over a period of about 6 million years.

The taxonomic identity of the butterfly Heliconius melpomene f. pyritosa var. fumigata Zikán, 1937 (Nymphalidae: Heliconiinae) is discussed based on the discovery of the specimen based on which this name was assigned. The specimen is not a variation of Eueides tales surdus Stichel, 1903, as previously stated, but is in fact a variation of H. melpomene f. pyritosa, which is a synonym of Heliconius erato amalfreda Riffarth, 1901.

The palatability and the ability of neotropical butterflies to escape after being detected, attacked, and captured by wild kingbirds ( Tyrannus melancholicus Vieillot), was investigated by the release of 668 individuals of 98 butterfly species close to the birds, during their usual feeding activities. Most of the butterflies were attacked and eaten. Only the troidine swallowtails ( Parides and Battus ; Papilionidae) were consistently rejected on taste and elicited aversive behaviours in birds. Most other aposematic and/or mimetic species in the genera Danaus and Lycorea (Danainae), Dione, Eueides and Heliconius (Heliconiinae), Hypothyris, Mechanitis and Melinaea (Ithomiinae), Biblis, Callicore and Diaethria (Limenitidinae) were generally eaten. Cryptic and non-mimetic species were always attacked and, if captured, they were also eaten. All Apaturinae, Charaxinae, Nymphalinae, Hesperidae, most Limenitidinae, Heliconiinae ( Agraulis, Dryas, Dryadula and Philaethria ) and Papilionidae ( Eurytides, Heraclides and Protesilaus ) were in this group. Results indicate that the learning process in kingbirds may demand a large mortality in prey populations, even among species generally accepted as unpalatable and aposematic. They also support the assertion that escaping ability and unpalatability evolved in butterflies as alternative strategies to avoid predation by birds. Mimetic relationships among several species are discussed. Evidence for the evolution of aposematism not related to unpalatability, but to escaping ability, was found for two hard-to-catch Morpho species.

A new phylogenetic hypothesis for Heliconius and related genera is presented, based on DNA sequence data from mtDNA combined with a region of the wingless gene. This study also adds eight new taxa to a previous cladistic hypothesis based on the mtDNA alone. Simultaneous phylogenetic analysis of the two gene regions together supports a topology largely in agreement with traditional views of heliconiine relationships based on morphology and suggests that the mtDNA support for the sister relationship between Eueides and H. charithonia is due to convergent evolution of homoplasious mtDNA sites.

The genus Heliconius has been revised more than a dozen times, yet relationships among many of its species groups remain obscure. A reliable phylogenetic hypothesis is desirable, because the genus has been a model system for studies of tropical community ecology, mimicry, and ecological genetics for the last 3 decades. A new cladogram for 35 species of Heliconius and the related genera Eueides, Laparus, and Neruda is presented, based on mitochondrial sequences spanning part of the COI gene and the COII gene. The data support most traditionally recognized species groups and also the monophyly of the above four genera with respect to other heliconiine outgroups. However, Heliconius is paraphyletic with respect to the other three genera. These data will allow a reinvestigation of problematical morphological, behavioral, and cytological traits in the group.

Chromosome counts in meiotic metaphase plates in the gonads of 67 of the probable 68 species of mimetic neotropical heliconian butterflies (Nymphalidae), representing 1524 individuals in 617 subspecies and geographically separate populations from southern Texas to northern Argentina, revealed a consistent haploid number of n = 21 in the genus Heliconius (except for the most advanced species with n = 33, 37, 56, and 60) and n = 31 in the more primitive genera (Eueides, Dryas, Dryadula, Agraulis, and Dione), with a transitional genus (Neruda) showing three species with n = 28-32, 21-22 + 5-10 "microchromosomes", and 20-22 + 1-5 "microchromosomes". The genus Laparus, with a single polymorphic species doris, probably an offshoot of early Heliconius, shows wide karyotypic variation (n = 20-30, 38) sometimes even within a single individual. The two most primitive genera also show much variation: Podotricha has two species with n = 9 and n = 26-29; and Philaethria shows many phenotypically similar species, two with n = 29 and a still uncertain number (at least 3) with n = 88 (most common), 67-72 (most widespread), 62 (very restricted geographically), 52, 21, and 12. Several interspecific hybrids (Heliconius cydno x H. melpomene) showed normal chromosome pairing, while deficient pairing was seen in intersubspecific hybrids in Eueides tales and Heliconius sara. The importance of these results in the evolutionary study of polytypic tropical species is discussed.

The Heliconius butterflies are a widely studied adaptive radiation of 46 species spread across Central and South America, several of which are known to hybridize in the wild. Here, we present a substantially improved assembly of the Heliconius melpomene genome, developed using novel methods that should be applicable to improving other genome assemblies produced using short read sequencing. First, we whole-genome-sequenced a pedigree to produce a linkage map incorporating 99% of the genome. Second, we incorporated haplotype scaffolds extensively to produce a more complete haploid version of the draft genome. Third, we incorporated ∼20x coverage of Pacific Biosciences sequencing, and scaffolded the haploid genome using an assembly of this long-read sequence. These improvements result in a genome of 795 scaffolds, 275 Mb in length, with an N50 length of 2.1 Mb, an N50 number of 34, and with 99% of the genome placed, and 84% anchored on chromosomes. We use the new genome assembly to confirm that the Heliconius genome underwent 10 chromosome fusions since the split with its sister genus Eueides, over a period of about 6 million yr.

The distributional data for the metabolic amino acid 3-hydroxy-L-kynurenine as a wing pigment in nymphalid butterflies would suggest taxonomic groupings essentially in contradiction with those resulting from more broad-scale morphological, breeding and behavioral studies; this indicates that the storage of this pigment is a taxonomically relatively unimportant unit character. The inherent danger of taxonomic misuse of single-compound chemical analysis is discussed, as is the significance of the use of the same chemical compound to achieve a given color by several divergent but mutually mimetic forms. The use of butterfly wing pigments to aid in the resolution of taxonomic problems has been appreciable in the last three decades, with pioneer work especially being done by Ford (1941, 1942, 1944a, 1944b, 1947) on anthoxanthins, pteridines, and other pigment classes. The present study of the distribution in Lepidoptera of the light yellow pigment 3-hydroxy-L-kynurenine (Fig. 2), a known amino acid on the metabolic pathway between tryptophan and nicotinic acid (a vitamin B), was approached with hopes of deriving some taxonomic conclusions from the resulting data (see Brown, 1965; Tokuyama and coworkers, 1967). The fact that this compound was clearly stored by certain well defined groups of butterflies, while clearly not present in other groups (even mimetic forms with visually identical coloration), permitted considerable enthusiasm for the possibility of unravelling some evolutionary problems, particularly in the large family Nymphalidae. The results of the distributional survey, summarized here, will be reported in detail elsewhere. Although they raise interesting questions about the chemical (and ultimately molecular and genetic) basis for mimicry (see below), they are disappointing in terms of taxonomy.' 3-Hydroxykynurenine is a normal metabolite almost certainly existing in small 1 The possible disappointments of chemotaxonomy are also becoming ever more evident in the study of the relationship of secondary plant products to taxonomy; see, for example, Swain (1963). quantities in all Lepidoptera; we have preliminarily confirmed its presence in a wide sampling of butterflies and moths by paper chromatography coupled with fluorescence and the ninhydrin test (Brown, 1965). However, it is stored in large quantities as wing and body pigment (up to 1.5% of total dry weight) only in certain groups of the diverse family Nymphalidae: namely, all species of the subgenus Heliconius (Heliconius),2 most genera of the subfamily Ithomiinae, and females of the genus Catonephele (Nymphalinae). A closely related but still unidentified compound is stored by most subspecies of Anaea (Hypna) clytemnestra of the Charaxinae, and by Colobura dirce and Tigridia (= Callizona) latifascia (Nymphalinae). We have found no evidence for the storage of the compound as pigment outside of the Nymphalidae or in any other neotropical genus within this grouping, comprising about 40% of the known species of butterflies. Specifically, it is absent from the subgenus Heliconius (Eueides), the subfamilies Danainae (including Indo-Australian species) and Acraeinae, and males of Catonephele as well of both sexes of the contiguous genus Myscelia. These latter are the groups of species most clearly allied with those which do store the compound as pigment. Furthermore, all of the groups mentioned (storing and non-storing) are mutually "mimetic" and possess light yellow 2 Albinistic or defective individuals, and even subspecies, containing no light yellow pigment at all are quite frequent in this group.

Background While the ecological factors that drive phenotypic radiations are often well understood, less is known about the generative mechanisms that cause the emergence and subsequent diversification of novel features. Heliconius butterflies display an extraordinary diversity of wing patterns due in part to mimicry and sexual selection. Identifying the genetic drivers of this crucible of evolution is now within reach, as it was recently shown that cis-regulatory variation of the optix transcription factor explains red pattern differences in the adaptive radiations of the Heliconius melpomene and Heliconius erato species groups. Results Here, we compare the developmental expression of the Optix protein across a large phylogenetic sample of butterflies and infer that its color patterning role originated at the base of the neotropical passion-vine butterfly clade (Lepidoptera, Nymphalidae, Tribe: Heliconiini), shortly predating multiple Optix-driven wing pattern radiations in the speciose Heliconius and Eueides genera. We also characterize novel Optix and Doublesex expression in the male-specific pheromone wing scales of the basal heliconiines Dryas and Agraulis, thus illustrating that within the Heliconinii lineage, Optix has been evolutionarily redeployed in multiple contexts in association with diverse wing features. Conclusions Our findings reveal that the repeated co-option of Optix into various aspects of wing scale specification was associated with multiple evolutionary novelties over a relatively short evolutionary time scale. In particular, the recruitment of Optix expression in colored scale cell precursors was a necessary condition to the explosive diversification of passion-vine butterfly wing patterns. The novel deployment of a gene followed by spatial modulation of its expression in a given cell type could be a common mode of developmental innovation for triggering phenotypic radiations.