{"paper_id":"aff052d6-9c0f-48b8-9626-99e4130ec74f","body_text":"1 \n \nGenomics of Neotropical biodiversity indicators: two butterfly radiations with rampant \nchromosomal rearrangements and hybridisation  \n \nEva SM van der Heijden 1,2, Karin Näsvall 1, Fernando A. Seixas 3, Carlos Eduardo Beserra Nobre 4, Artur \nCampos D Maia4,5, Patricio Salazar-Carrión1,2, Jonah M Walker1,2, Daiane Szczerbowski6, Stefan Schulz6, Ian \nA Warren 2, Kimberly Gabriela Gavilanes C órdova7, María José Sánchez -Carvajal7, Franz Chandi 7, Alex P \nArias-Cruz7, Nicol Rueda -M1,8, Camilo Salazar 8, Kanchon K Dasmaha patra9, Stephen H Montgomery 10, \nMelanie McClure 11, Dominic E Absolon 1, Thomas C Mathers 1, Camilla A Santos 1, Shane McCarthy 1, \nJonathan MD Wood 1, Gerardo Lamas 12, Caroline Bacquet 7,13, André Victor Lucci Freitas 14, Keith R. \nWillmott15, Chris D Jiggins2, Marianne Elias13,16, Joana I Meier1,2 \n \n1. Tree of Life Programme, Wellcome Sanger Institute, UK \n2. Department of Zoology, University of Cambridge, UK \n3. Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA, USA \n4. Graduate Program in Animal Biology, Federal University of Pernambuco, Brazil \n5. Graduate Program in Plant Biology, Federal University of Pernambuco, Brazil \n6. Institute of Organic Chemistry, Technische Universität Braunschweig, Germany \n7. Universidad Regional Amazónica Ikiam, Ecuador \n8. Biology Program, Faculty of Natural Sciences, Universidad del Rosario, Bogotá, Colombia \n9. Department of Biology, University of York, UK \n10. University of Bristol, UK \n11. Laboratoire Écologie, Évolution, Interactions des Systèmes Amazoniens (LEEISA), Université de Guyane, CNRS, IFREMER, Cayenne, France  \n12. Museo de Historia Natural, Universidad Nacional Mayor de San Marcos, Lima, Peru \n13. Smithsonian Tropical Research Institute, Gamboa, Panama \n14. Universidade Estadual de Campinas, Brazil \n15. Florida Museum of Natural History, USA \n16. Institut de Systématique Évolution Biodiversité (ISYEB), CNRS, MNHN, EPHE, Sorbonne Université, Université des Antilles, Paris, France \n \nAuthor contributions: \nPSC, JIM, ESMH, KGGC, MJSC, DS, FC, APAC, CB, NRM, CS, KKD, NN, SHM, MM, KRW, AVLF, CDJ and ME col lected samples. GL, \nKRW and ME oversaw the butterfly identification and taxonomic revision. ESMH, JIM, JMW and IAW performed lab work. ESMH \nmapped and filtered the sequencing data, and ran the phylogenetic and hybridisation analyses. KN did the chromosomal  \nrearrangement analysis. CEBN, ACDM, DS, SS and AVLF analysed the androconial semiochemicals. FAS ran the MSCi analysis. DEA, \nTCM, CAS, SM, KN and JW assembled and curated the new genomes. JIM designed the study with contributions from ESMH, ME \nand CDJ. ESMH, JIM and KN wrote the manuscript with contributions from all authors. \n \nCorresponding author: Joana Isabel Meier, joana.meier@sanger.ac.uk \nKey words: Speciation, biodiversity, butterflies, neotropics, genomics \nClassification: Biological Sciences - Evolution \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n2 \n \nAbstract \nA major question in evolutionary biology is what drives the diversification of lineages. Rapid, recent \nradiations are ideal systems for addressing how new species arise because they still show key \nmorphological and ecological adaptation s associated with speciation. While most studied recent \nradiations have evolved in an insular environment, less research has been carried out on continental \nradiations with complex species interactions. Melinaea and Mechanitis butterflies (Nymphalidae : \nIthomiini) have rapidly radiated in the Neotropics. They are classical models for Amazonian biogeography \nand colour pattern mimicry and have been proposed as biodiversity indicators. We generated reference \ngenomes for five species of each genus, and whol e-genome resequencing data of most species and \nsubspecies covering a wide geographic range to assess phylogeographic relationships, patterns of \nhybridisation and chromosomal rearrangements. Our data help resolve the classification of these \ntaxonomically challenging butterflies and reveal very high diversification rates. We find rampant evidence \nof historical hybridisation and putative hybrid species in both radiations, which may have facilitated their \nrapid diversification. Moreover, dozens of chromosomal f usions and fissions were identified between \ncongeneric species, and even some within species. We conclude that interactions between geography, \nhybridisation and chromosomal rearrangements have contributed to these two rapid radiations in the \nhighly diverse Neotropical region. We suggest that rapid radiations may be spurred by repeated periods \nof geographic isolation during Pleistocene climate oscillations, combined with lineage -specific rapid \naccumulation of incompatibilities during allopatric phases, followed by secondary contact with some gene \nexchange. \n \nSignificance Statement \nUnderstanding factors contributing to rapid speciation is a key aim of evolutionary biology. Here we focus \non two rapid radiations of Neotropical butterflies. Our genomic data with b road taxonomic and \ngeographic coverage reveal rampant hybridisation and chromosomal rearrangements,  each likely \ncontributing to the high diversification rates. Our study highlights the use of genomic data to resolve \ntaxonomically challenging species groups and elucidate drivers of diversification in r apid radiations. We \nshow that for biodiversity hotspots with recent radiations, barcoding is insufficient to characterise species \nrichness due to gene flow and recent speciation. The taxonomic implications of b oth introgression and \nkaryotype diversity for species delimitation are important to consider during monitoring and management \nof biodiversity in these vulnerable habitats. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n3 \n \nIntroduction \nRapid radiations, where a lineage diversifies into many different species over a short time period, are ideal \nsystems for studying how new species evolve (1, 2). They can be driven both by non-adaptive processes, \nsuch as the accumul ation of differences during periods of allopatry leading to incompatibilities upon \nsecondary contact, and by adaptive processes such as adaptation to different ecological niches or sexual \nselection for different traits and preferences (3, 4). Sympatric radiations require some degree of niche \ndifferentiation among the species for stab le coexistence and sufficient reproductive isolation such that \nincipient lineages do not merge (2).  \nWhile most lineages do not readily radiate even in the face of ecological opportunity, some \nlineages are particularly prone to rapid radiations, and do so repeatedly. We are only starting to \nunderstand the factors explaining these lineage-specific differences (5). Most knowledge stems from well-\nstudied radiations that evolved in insular environments with little competition with other species and a \nrelatively simple and geographically limited environment (e.g. Darwin’s finches on the Galapagos Islands, \ncichlid fishes in lakes, Anolis lizards on the Caribbean islands or Hawaiian silverswords) (2, 6). However, \nmany rapid radiations evolved on  the much more complex continents and much less is known about \ndrivers leading to their diversification (7–9). Reduced competition in insular environments allows niche \nspecialisation without being immediately outcompeted when a lineage is not yet well -adapted to its \nenvironment. However, in large continental areas such as the hyperdiverse Neotropics, competition is \nmuch stronger limiting ecological speciation . On the other hand, large and complex environments on \ncontinents may provide more opportunity for allopatric divergence than a smal l island. The speed of \naccumulating incompatibilities in allopatry may thus be more important in rapid radiations on continents \nthan insular environments.  \nHere, we study the drivers of diversification in two rapid continental radiations of the Neotropical \nbutterfly tribe Ithomiini. Ithomiine butterflies (Nymphalidae: Danainae, ca. 400 species in 42 genera) are \nfound across Central and South America (10, 11). They constitute a substantial part of the butterfly species \nassemblage and are regarded as good indicators of spatial patterns of biodiversity in the Neotropics, the \nmost biodiverse areas in the world (10, 12, 13). Sequestration of pyrrolizidine alkaloids from Asteraceae \nand Boraginaceae plants render most Ithomiini unpalatable (14–18), and their colour patterns advertise \nthis unpalatability to predato rs. They form Müllerian mimicry rings, where locally co -occurring species \nhave converged in colour pattern, and thus share the cost of educating predators (10, 19, 20). We focus \non two ithomiine genera, Melinaea and Mechanitis, which have diversified exceptionally fast with most \nspecies younger than a million years (11, 21). Hitherto, the study of these radiations has been hampered \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n4 \n \nby taxonomic challenges. Melinaea and Mechanitis are among the most taxonomically difficult of \nIthomiini, as Fox noted (1967)  (22): ‘these insects [are] so thoroughly confusing and so thoroughly \nconfused by my predecessors’. The species do not differ in genital or other morphological characteristics \nand show substantial intraspecific wing pattern variation and mimicry between taxa. Barcoding does not \nreliably distinguish species either (23, 24). As prior studies have only used few or no genetic markers or \ndid not have broad geographic coverage, the taxonomy is still partially unresolved, despite many \ntaxonomic revisions (22, 24–26, e.g. 27–29). \nWhile the exact causes of their rapid radiations are unknown, different contributing factors have \nbeen proposed. Ecological adaptation may be relevant as species show differences in microhabitats, host \nplants, mimicry rings, and altitude, but on the other hand many species share habitats and host plants, \nand the majority of species  occurs in the lowlands (25, 30–33). Moreover, where species co -exist, they \nconverge in colour patterns and thus assortative mating may rely more strongly on chemical cues. \nIthomiine species differ in male -specific androconial compounds (chemical compounds secreted from \nspecialised wing scales where the fore - and hindwing overlap), which likely act as pheromones (Trigo et \nal. 1996; Schulz et al. 2004; Blow et al. 2023). \nAllopatric accumulation of differences could also have played a role in the rapid diversification of \nithomiini, as this could have occurred in different rainforest refugia during climatic oscillations e.g. in the \nPleistocene (26, 34, but see 35, 36)  or on opposite sides of geographic barriers such as the Andes. Both \nclimatic refugia and the Andes have been proposed as “speciation pumps” in the Neotropics (37), also for \nIthomiini (38), where periods of allopatry followed by secondary contact create favourable circumstances \nfor speciation.  \nAnother factor that might contribute to the diversification of ithomiine butterflies is hybridisation. \nPhylogenetic studies using a limited number of markers have reveal ed mito-nuclear discordances and \nparaphyletic taxa in Ithomiini (23, 36) . This could be due to limited geographic or genetic resolution, \nincomplete lineage sorting in the rapidly speciating lineages, or introgressive hybridisation. While gene \nflow between sister lineages can homogenise their gene pools, opposing speciation, recent studies have \nshown that sometimes introgressive hybridisation from more distant relatives can facilitate rapid \ndiversification by enriching the genetic diversity with novel, potentially adaptive variants or contributing \nto the origin of new hybrid species (39–43). Admixture has been shown to kickstart adaptive radiation \n(e.g. 44, 45) , facilitate parallel adaptation (e.g. 42, 46) , and novel adaptations (e.g. 47), but the role in \nithomiini diversification is hitherto unknown. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n5 \n \nIthomiini butterflies have an unusually high diversity in chromosome numb er (48), which could \nalso contribute to their rapid diversification. Offspring from parents with different karyotypes may suffer \nreduced fitness, due to mismatch in pairing of homologous chromosomes tha t results in aneuploidy, \nmeiotic failure or hybrid sterility (49, 50). Furthermore, chromosomal rearrangements might facilitate \ndivergence in the face of gene flow by accumulation of incompatibilities in low recombining regions (51) \nor if they link together co-adapted variants (52, 53). In Melinaea and Mechanitis butterflies, chromosome \ncounts range from 13 to 30 (48), and chromosomal rearrangements likely contribute to reproductive \nisolation, as a cross between two closely -related Melinaea species with different karyotypes resulted in \nnearly sterile hybrids (54). However, pervasive intraspecific variation in chromosome counts (48, 54)  \nindicates that not all rearrangements reduce fitness and the ir role in speciation thus remains an open \nquestion.  \nHere, we use five reference genomes of each genus and whole -genome resequencing data of \nalmost all species and many subspecies , to resolve taxonomic uncertainties and explore whether \ngeography, introgressive hybridisation or chromosomal rearrangements may have played a role in the \nrapid diversification of Mechanitis and Melinaea butterflies. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n6 \n \nResults \n \nFigure 1 - Rampant cytonuclear discordance in Mechanitis and Melinaea and a need for taxonomic revision \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n7 \n \nCo-phyloplot showing the nuclear and mitochondrial phylogenies of 135 Mechanitis and 109 Melinaea individuals. \nNuclear phylogeny (left) based on 537,500 SNPs for Mechanitis (A) and 784,526 SNPs for Melinaea (B) and full \nmitochondrial genome phylogeny (right). The coloured circles and connecting lines indicate the currently classified \nspecies (25, 29) . Br=Brazil; FG=French Guiana; E -Co=eastern Colombia; W -Co=western Colombia; E -Ec=eastern \nEcuador; W-Ec=western Ecuador; E -Pe=eastern Peru; Pa=Panama, Su=Suriname. The coloured boxes highlight key \nfindings. The node labels show concordance factors, indicating the percentage of trees produced for windows across \nthe genome that contain that clade. \n \nTaxonomic revision \nIn this manuscript, we adhere to the ‘genotypic cluster’ species concept, which defines a species as ‘ a \nmorphologically or genetically distinguishable group of individuals that has few or no intermediates when \nin contact with other such clusters’ (55). Through whole-genome resequencing of 135 Mechanitis and 109 \nMelinaea individuals from across South and Central America, with additional individuals from the \noutgroup genera Forbestra, Eutresis and Olyras, we shed light on the phylogenomic relationships and \ntaxonomy (Fig. 1). Our results, laid out in the following sections (see also Text S1), confirm species versus \nsubspecies status for most known taxa in agreement with (29), support two recent species \nreclassifications (Mel. tarapotensis (54), Mel. mothone (56)) and reveal three additional taxa that need to \nbe elevated to species level (Mel. maeonis stat rest (Hewitson 1869), Mec. nesaea stat rest (Hübner 1820), \nMec. macrinus stat rest (Hewitson 1860), Fig. 1; Text S 1). The placement of Mel. menophilus mediatrix \n(French Guiana) has been uncertain (56); our dataset places it as the most divergent subspecies of Mel. \nmenophilus. A revised, annotated taxonomic list for these two genera can be found in Text S2. \n \nMitonuclear discordance \nIn both Mechanitis and Melinaea, we find rampant mitonuclear discordance, i.e. mismatches between \nmitochondrial and nuclear phylogenies (IQtree2; maximum likelihood) (Fig. 1). For instance, Mec. nesaea \nis sister to Mec. polymnia in the nuclear phylogeny, but to Mec. lysimnia in the mitochondrial phylogeny \n(Fig. 1A - box III; Fig. S1 -2). This result could either be indicative of incomplete lineage sorting (ILS) or \nadmixture (more details below). Mec. messenoides harbours two divergent mitochondrial haplotypes, \nconsistent with barcoding results (23, 25) (Fig. 1A - box II), one clustering with the polymnia-lysimnia clade, \nand the other one with its nuclear sister species, Mec. menapis . The mitochondrially divergent Mec. \nmessenoides individuals do not form separate clades in the nuclear phylogeny, nor do they differ in \ncollection location or subspecies (Table S1; Fig. S1-2).  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n8 \n \nWhile the Melinaea species are clearly differentiated in the nuclear phylogeny, the mitochondrial \nphylogeny shows almost no variation among species, confirming previous barcoding results (56) (Fig. 1B; \nFig. S3 -4). Mel. ludovica, Mel. tarapotensis and Mel. lilis are the only species that form monophyletic \nmitochondrial clades that are no part of the shallow clade without species differentiation. Notably, Mel. \nludovica is the only species that is also an outgroup to the other Melinaea species in the nuclear genome, \nwhereas Mel. lilis and Mel. tarapotensis are sister to Mel. isocomma and Mel. satevis, respectively. \n \nCalibrated phylogenetic tree \nWe approximated the divergence times in the two genera using a Bayesian MCMC method for inferring \ntrees (BEAST2), with one individual representing each lineage to produce a phylogeny calibrated with \ndivergence times from (11) (Fig. 2). The seven Mechanitis species are estimated to have diversified within \nthe past 1.36 million years, resulting in a speciation rate of 1.431 speciation events per lineage/My \n(assuming a pure birth model with a constant speciation rate). The ten species of the core Melinaea clade \n(excluding the most divergent species, Mel. ludovica) have diversified in the past 1.41 million years (Fig. \n2B), giving a speciation rate of at least 1.633. Of the four potential Melinaea species missing in our analysis \n(Mel. ethra, mnasias, mnemopsis and scylax), two likely form part of the core clade (11, 57), which would \nincrease the speciation rate to 1.672.  \nHowever, note that gene flow could affect our estimates , as suggested by discordance between \nthe BEAST2 and IQtree2 topologies (Fig. 2 vs Fig. 1 (note low concordance factors in Fig. 1)): Mec. \nmessenoides is not sister to Mec. menapis anymore, and Mec. nesaea has shifted to be sister to  Mec. \nlysimnia instead of Mec. polymnia. The relationships among lineages within Mec. polymnia also changed. \nWhile ILS could partially explain this, introgression  could also cause  it (see details below). Gene flow \nbetween sister lineages will make their apparent divergence times shorter than the initial split time, and \ngene flow with non-sister lineages will make them appear longer.  \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n9 \n \n \nFigure 2 - Calibrated phylogeny of Mechanitis and Melinaea butterflies with evidence of introgression and \nbiogeographic patterns. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n10 \n \nTime-calibrated BEAST2 phylogenies of Mechanitis (A) and Melinaea (B) with the newly proposed species \nclassification and secondary calibrations from (11) (asterisk at node indicates which node was used to calibrate). One \nindividual was included at the species level, or subspecies -level if they were very divergent. Th e regions in the \noverview map are based on the combined distribution of our subspecies and samples from (10) with region names \nadapted from (11). Arrows between clades indicate potential hybridisation events (based on AIM, Fbranch and BPP). \nThe node labels indicate the age as obtained by the calibration. ‘Co re’ in the Melinaea phylogeny indicates the core \nclade of fast diverging Melinaea, and ‘ingroup’ is a clade referred to in the text. For each clade, cartoon wings based \non representative colour patterns are shown, and as an example, pink stars designate taxa part of the same mimicry \nring. The distribution of our individuals is indicated by coloured dots and coloured rings on a distribution map based \non the subspecies distribution from (10)); map from USGS, Esri, TANA, DeLorme and NPS; see Fig. S6 -7 for larger \nmaps). The chromosome numbers in the right column are based on (48) or our reference genomes (asterisk).  \n \nPhylogeographic patterns \nThrough combining our sampling locations with those compiled by (10) adjusted according to our \ntaxonomic revision (Fig. 2), we assessed the biogeographic distribution of the species. We find four main \nbiogeographic regions (Fig. 2; Fig. S6-7), with most species restricted to one of them. Some sister-species \nare separated by the Andes: Mec. messenoides and mazaeus (Western Amazonia) versus their respective \nsisters Mec. menapis and macrinus (West of the Andes). However, note that the placement of Mec. \nmessenoides differs in the BEAST2-topology and is affected by hybridisation (see next section). Within \nMec. polymnia , Ecuadorian and Colombian individuals from opposite sides of the Andes are highly \ndivergent, indicating little gene flow, though our study includes one putative hybrid (Fig. S8). This pattern \nof species separation by the Andes was not previously as apparent due to species misclassification. \nMost individuals in our Melinaea dataset are from Western Amazonia (east of the Andes) and we \nfind that many sister species are sympatric. Only two species in our dataset occur west of the Andes (Mel. \nlilis and idae) and they form a clade with a third species from east of the Andes ( Mel. isocomma). Our \ndataset lacks a potential third species occurring west of the Andes, Mel. scylax (10), which may represent \na subspecies of Mel. lilis (26, 57). Other potential species missing in our dataset are Mel. mnemopsis from \nWestern Amazonia, Mel. ethra from the Atlantic forest and Mel. mnasias from Western Amazonia and the \nAtlantic Forest (10, 57).  \n \nSignatures of rampant introgression throughout both genera \nPhylogenetic discordance between phylogenies constructed for the nuclear and mitochondrial genome \n(Fig. 1), for different genomic regions inferred with concordance factors (Fi g. 1), and with IQtree2 and \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n11 \n \nBEAST2 (Fig. 1 vs Fig. 2) is indicative of a history of introgression and/or ILS in both genera. We assessed \nhybridisation throughout both radiations with windowed species tree inference with BPP (Fig. S5), excess \nallele sharing between non-sister taxa estimated from Fbranch (Fig. S9) and joint inference of species tree \nwith gene flow using the approximate isolation with m igration (AIM) model (Fig. S10) . The r esulting \nphylogenies and introgression histories are summarised in Fig. 2 (Text S3), but we stress that this is only \none of multiple possible scenarios consistent with our observed patterns of excess allele sharing and gene \ntree discordance.  \nIn Mechanitis, two lineages show uncertain placements. As mentioned, the placement of Mec. \nmessenoides and nesaea varies depending on t he methodology (IQtree2 and BEAST2) and genomic \nregions (Fig. 1A – box II and III). Contrary to the nuclear phylogeny inferred with BEAST2, but in line with \nthe mitochondrial phylogeny, the species phylogeny inferred allowing for gene flow (AIM) places \nmessenoides closer to the polymnia-nesaea-lysimnia clade rather than with menapis-mazaeus-macrinus. \nAIM indicates admixture between these clades (Fig S10A – arrows B-D), and Fbranch also confirms excess \nallele sharing between messenoides and polymnia-nesaea-lysimnia (Fig. S9A - #1). According to the BPP-\nanalysis, Mec. messenoides groups with menapis -mazaeus-macrinus in ~51% of the genome, a nd with \npolymnia-lysimnia-nesaea in ~28% of the genome (sometimes still with Mec. menapis ) (Fig. 5B). The \nbranching order within the polymnia-lysimnia-nesaea clade is also highly variable across the genome: \npolymnia and nesaea are sister in 46.3% of the genome grouping, while 31.7% of trees group  polymnia \nwith lysimnia, and 14.7% group nesaea with lysimnia (Fig. 5C) (see also Fbranch - Fig. S9A #3). In short, \nthese results are consistent with Mec. messenoides being introgressed (Fig 2A - arrows a-d) and gene flow \nbetween polymnia-lysimnia-nesaea.  \nIn Melinaea, IQtree2 and BEAST2 produce the same topology (Fig. 1 vs Fig. 2), with Mel. idae sister \nto Mel. lilis and Mel. isocomma . However, the AIM analysis groups Mel. idae with the  ingroup clade, \nalthough showing extensive gene flow from  the lineage of Mel. lilis and Mel. isocomma (Fig. 2B  - arrow \na; Fig S10B – arrow A,D; see also Fbranch (Fig. S9B - #1)). To resolve the position of Mel. idae, as well as \nputative introgression betw een deeper bra nches of the phylogeny, we ran BPP focusing on the \nrelationships between the lilis-isocomma-idae clade and representatives of the ingroup  clade (mneme, \nmarsaeus and mothone) (Fig. 5E-F). In the majority of the genome (43.4%) lilis, idae and isocomma group \ntogether (like IQTree2 and BEAST2), while in 34.5% of the genome idae groups with mothone-marsaeus-\nmneme (like AIM). Notably, the Z chromosome supports almost exclusively the latter relationship. In 4.3% \nof the genome, isocomma clusters with mothone-marsaeus-mneme. Mel. lilis is almost as often sister to \nidae (30.3%) as to isocomma (38.3%), and less commonly an outgroup to both (4.1%), indicating more \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n12 \n \nrecent shared ancestry between  lilis and idae, which is confirmed by Fbranch excess allele s haring (Fig. \nS10B; Fig S9B - #2). Mel. mothone, marsaeus and mneme also vary in their respective relationships (Fig. \n5E). \nWe further investigated the timing of divergence and introgression for the three species showing \nthe strongest signals of introgression (Mec. messenoides, Mec. nesaea and Mel. idae), using a multispecies \ncoalescent-with-introgression (MSCi) approach. In all three cases, introgression is estimated to be old \n(200-425 kya) (Fig. 3A-B scenario 1; Fig. 3C). For Mec. messenoides and Mel. idae, different replicate runs \nproduce different outcomes reflecting the uncertainty in the placement of these two species in the \nphylogeny. Notably, some replicates indicate the origin of the introgressed lineages closely coincides with \nthe split time from both parents, suggestive of a hybrid origin (Fig. 3A-B scenario 2; Fig. S11; Table S2).  \n \n \nFigure 3 - Three ancestrally introgressed species, with a focus on Mechanitis nesaea \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n13 \n \nA multispecies coalescent-with-introgression model explored the relation and timing of introgression relative to the \ndivergence times, in A) Mel. lilis , Mel. idae  and Mel. marsaeus ; B) Mec. menapis , Mec. messenoides , and Mec. \npolymnia; and C) the Brazilian Mec. lysimnia lysimnia, Mec. nesaea and Mec. polymnia casabranca. D) A closer look \ninto the restored species Mec. nesaea : phylogenetic relationships, chromosome numbers (48), a photo of a \nrepresentative adult and an early fifth instar larva (123, 124) . The photo of M. lysimnia lysimia  is a courtesy of \nAugusto Rosa. E) Overlaid chromatograms of androco nial extracts of representative individuals of Mec. nesaea \n(yellow line), Mec. l. lysimnia  (blue) and Mec. polymnia casabranca  (orange). Peaks: (1) 4 -Hydroxy-3,5,5-\ntrimethylcyclohex-2-enone, (2) Hydroxydanaidal, (3) Methyl hydroxydanaidoate, (4) Methyl far nesoate isomer, (5) \nMethyl (E,E)-farnesoate, (6) m/z 57, 43, 55, 56, 85, (7) Octadecatrienoic acid (cf.), (8) Octadecanoic acid, (9) Ethyl \nlinolenate, (10) (E)-Phytyl acetate (11) Hexacosene, (12) Heptacosene, (13) Nonacosene (not all compounds of Table \nS3 are found in these three individuals). F) NMDS shows the androconial chemical bouquet of Mec. nesaea is clearly \ndistinct from both putative parental lineages, most similar to Mec. polymnia. G) A genome scan of f dM across the \ngenome (in 20kb windows) revea ls that strong signatures of introgression (f dM) between Mec. nesaea  and Mec. \nlysimnia (P1=allopatric polymnia,P2=nesaea,P3=lysimnia,P4=Forbestra) overlaps with regions of high differentiation \n(FST)  between Mec. nesaea and its sister species Mec. polymnia (orange vertical lines - high FST, red dots - high fdM  \nand high F ST). Chromosomal breakpoints between Mec. polymnia and the four other reference genomes are shown \nwith blue bars on top.  \n \nA focus on Mec. nesaea \nAs we propose to re -elevate Mec. nesaea to a species from its previous classification as a subspecies of \nMec. lysimnia, we investigated the relationships and reproductive isolation between Mec. nesaea, Mec. \nlysimnia and Mec. polymnia in more detail. We resequenced 15 Mec. nesaea, 19 Mec. l. lysimnia and 9 \nMec. p. casabranca, of which 24 were sampled from sympatry, and assessed the androconial compounds \nthat might act as pheromones, potentially contributing to assortative mating. Even though Mec. lysimnia \nand Mec. polymnia have been observed to interbreed in nature (58) and putative hybrids have been \nobserved (Fig. S12), we find relatively high F ST throughout the genome between all three species and no \nevidence of ongoing gene flow based on ADMIXTURE analyses (Fig. S8; S13). Furthermore, they have been \nshown to differ in number of chromosomes (Fig. 3D) (48), and we find their androconial bouquets to be \nclearly distinct (Fig. 3E -F, Text S4; Table S3 -4). All these lines of evidence suggest strong reproductive \nisolation between Mec. nesaea and both Mec. polymnia  and Mec. lysimnia. Genome scans for \nintrogression (fdM) between Mec. lysimnia and Mec. nesaea compared to allopatric Mec. polymnia show \nthat this introgression is restricted to few genomic regions  (Fig. 3G; Fig. S14 -S15). The MSCi model for \nMec. nesaea suggests Mec. nesaea diverged from Mec. polymnia prior to Mec. lysimnia introgression. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n14 \n \nHowever, lysimnia-introgression could have contributed key genetic variation to Mec. nesaea  and \nstrengthened reproductive isolation to  Mec. polymnia . Consistent with this hypothesis, we find that \nintrogression peaks coincide with peaks of elevated Dxy and FST between Mec. nesaea and Mec. polymnia \nand mostly show no evidence of excess allele sharing between Mec. nesaea and sympatric Mec. polymnia \ncompared to allopatric Mec. polymnia (Fig. 3G; Fig. S13-S15), but the low levels of ongoing gene flow make \nthese measures poor predictors of reproductive isolation barriers. \n \n \nFigure 4 - Chromosomal rearrangements \nA) Synteny between Melinaea and Mechanitis genomes based on whole genome alignments. Horizontal bars \nrepresent individual chromosomes, with sex -chromsomes (black bar) and chromosomes involved in within -species \npolymorphic fusion -fissions (purple bar) highlighted. The cladogr am is based on Fig. 2 and shows haploid \nchromosome numbers in parentheses. B) Example of differentiation (FST) and breakpoints (blue vertical lines) between \nMec. mazaeus and Mec. menapis along the genome . The red dots indicate windows coinciding with breakpoints. C) \nExamples of HiC-contact maps. Top panel: Sex-chromosomes in Mec. mazaeus. Lower panel: Autosomal fission-fusion \nheterozygote in Mel. mothone. D) Matrix displaying number of fusion -fission rearra ngements between species in \neach genus.  \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n15 \n \nChromosomal rearrangements \nTo study chromosomal rearrangements, we generated haplotype-resolved genomes of five Melinaea and \nfive Mechanitis species using PacBio and HiC-data (Fig. S16), resulting in high quality genomes with >96% \nBUSCO completeness and >94% of the genome assembled to chromosomes (Table S5 -S6). The inferred \nchromosome numbers match the chromosome counts from karyotyping (48). The Mechanitis genomes \nare substantially shorter (291-320 Mb) than the Melinaea genomes (496-661 Mb, Table S5). \nWhereas most Lepidoptera have conserved karyotypes with 31 chromosomes (59), our synteny \nanalysis using BUSCO genes show that Mechanitis and Melinaea genomes are highly rearranged compared \nto the ancestral karyotype (Fig. S17-S19), corroborating earlier findings from two Melinaea genomes (60). \nThe median length of conserved syntenic blocks is 32 -36 genes, compared to 168 genes in the outgroup \nDanaus plexippus (Table S7; Fig. S19). The canonical Z has not undergone any fiss ions and has retained \nthe longest conserved syntenic blocks (225 -227 genes; Fig. 4A; Table S7;  Fig. S19) in our genomes. \nHowever, both genera share a fusion of Z with parts of ancestral autosome 10 and Mechanitis has a further \nfusion with parts of ancestral autosome 6 (Table S8; Fig. S18). In addition, a fusion -fission polymorphism \ninvolving the Z and another autosome was observed between the Zs of the male Mel. isocomma (Table \nS8; Fig. S16). A W -chromosome was ident ified in all females as a segment of varying size depleted of \nBUSCO genes with moderate sequence similarity within genera, but none between (Fig. 4A). We found \nW-autosome fusions in four species  (Table S8). Mec. macrinus  and mazaeus show complex fusions \nbetween the W and multiple autosomes that are partially shared. Four species have two Z chromosomes, \nas by definition, the homologue of the chromosome fused to the W becomes a Z chromosome. Similarly, \nthree species have two or three W chromosomes. Surprising ly, seven individuals were heterozygous for \none or more simple autosomal rearrangement and in Mel. ludovica  we detected a complex chain \nrearrangement involving four autosomes in one haplotype and three autosomes in the other haplotype \n(Table S8; Fig. S16). None of the simple autosomal polymorphisms are shared among taxa. \nWhile chromosome spreads had revealed species differences in chromosome counts (48), we \nhere show that these are not due to a few fissions or fusions, but complex rearrangements. Closely related \nsister species show 3 -47 chromosomal rearrangements (Fig. 4), which could contribute to reproductive \nisolation. To assess if these chromosomal rearrangements confer reproductive is olation barriers, we \nmapped the location of the chromosomal breakpoints between all species pairs to test for an association \nbetween breakpoints and reduced gene flow. We found increased FST in all Mechanitis comparisons, and \nreduced diversity (π) in breakpoint regions in seven of the ten comparisons (Fig. 4B, Fig. S20, Table S9). In \nprinciple, elevated FST could be caused by increased background selection as recombination tends to be \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n16 \n \nreduced at chromosome ends and levels of ongoing gene flow are low (61, 62). However, we also find \nelevated absolute divergence (DXY) in windows coinciding with breakpoints in four of the ten comparisons, \nwhich indicates that background selection alone cannot explain the pattern (63). In Melinaea, we did not \nobserve significantly elevated FST in the breakpoint regions, but we detected an increase in DXY especially \nin the comparisons involving the more distantly related Mel. ludovica (Table S9). \n \nDiscussion \nOur results confirm that the two ithomiine genera Melinaea and Mechanitis represent fast and recent \nradiations. They diverged in the past 1 -2 million years, and speciated much faster than the other well -\nstudied Neotropical butterfly radiation of Heliconius (respectively 1.633 and 1.431 versus 0.324 speciation \nevents per lineage/My (46 species in 11.8 My)) (64). Kawahara et al. (21) previously found a significant \nrate shift towards high diversification rates in ithomiine butterflies (clade L; 0.23 speciation events per \nlineage/My) and we show that among Ithomiini, Mechanitis and Melinaea have an even higher speciation \nrate, consistent with previous results (11). Our results shed light on potential drivers of their rapid \ndiversification, detailed below.  \nGiven the high number of ancient introgression events across both genera, hybridisation may \nhave sped up speciation by boosting gene tic variation as seen in other systems (e.g. 42). We identified \nthree species showing ancient introgression  that might have a hybrid origin. Further research on which \ngenomic regions are more or less likely to introgress could inform us as to what extent introgression \ncontributed to reproductive isolation between those taxa and their parental lineages. For instance, similar \nto Heliconius butterflies, introgression might have contributed to both mimicry ring switches and colour-\nbased assortative mating, which generate reproductive isolation, thereby driving speciation (65).  \nDespite ancient hybridisation, we find little evidence for ongoing gene flow between the species, \nsuggesting strong reproductive isolation. Some of this reproductive isolation is likely attributed to the \nexceptionally high rates of chromosomal rearrangements across both genera, as complex chromosomal \nrearrangements are expected to constitute barriers to gene flow (66, 67) . While Lepidoptera \nchromosomes are holocentric and may be able to tolerate simple fusions and fissions during meiosis, most \nLepidoptera have retained the same highly conserved karyotypes of 31 chromosomes (59, 68) . The \nmassive chromosomal rearrangements we found in ithomiini are thus unusual, but there are other \nlineages that also show high rates of fissions and fusions (e.g. Leptidea (69); and Erebia (70)) and there is \na slight association with variation in chromosome number and speciation rates (71).  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n17 \n \nEven though chromosomal rearrangements are thought to limit gene flow, many of our genomes \nshowed heterozygosity for chromosomal rearrangements. The heterozygote fission -fusions are \nunambiguous in the HiC -data, matching th e intraspecific variation in chromosome counts documented \npreviously (48) and cytogenetical evide nce for pervasive polymorphism in  Mel. satevis cydon (54). This \nmay be akin to Leptidea sinapis, where, despite multiple fusion/fission polymorphisms segregating within \npopulations (69, 72), crosses between chromosomal extremes show low survival of F2 hybrids (73). That \ncomplex chromosomal rearrangements involving multiple chromosomes likely constitute stronger \nbarriers to gene flow compared to simple fusions and fissions has also been found in other taxa such as \nshrews (74) and wallabies (75) and is in line with findings of nearly sterile hybrids in a cross between two \nMelinaea sister species that differ in chromosome count (54). It is thus possible that simple fusions and \nfissions are not strongly s elected against, but if too many sequential f usions accumulate in isolated  \npopulations, they cause reproductive isolation upon secondary contact. \nFurthermore, we find many fusions between sex chromosomes and autosomes. Given the \ndisproportionate role sex chromosomes are proposed to have in speciation (Haldane’s rule and/or large-\nZ effect) (76–78), they may play a strong role in reproductive isolation of ithomiini butterflies. We found \ntwo Z-autosome fusions shared among all congenerics, but no fission in the canonical Z, consistent with \nconserved Z chromosomes in other Lepidoptera lineages (68). The heterozygote Z -autosome fusion in  \nMel. isocomma potentially represents an ongoing fixation of a novel Z -autosome fusion, which might \nrapidly rise in frequency if there are advantageous loci involved (79). The simple W -autosome fusion \nobserved in Mel. menophilus resemble those described in some Heliconius butterflies (80). The complex \nfusion-fissions with multiple Ws  and Zs in Mec. macrinus  and Mec. mazaeus  are similar to Leptidea \nbutterflies (81) and may constitute particularly strong barriers to gene flow (82).  \nSince the divergence times of Mechanitis and Melinaea species match Pleistocene climatic \nfluctuations (2.58 Ma -11.7 ka), and many sister species are separated by the Andes or restricted to \ndifferent geographic regions corresponding to postulated Pleistocene refugia (26, 34, 37), we believe that \nour findings are consistent with the speciation pump hypothesis. This hypothesis postulates that periods \nof allopatry, followed by secondary sympatry, can facilitate diversification (37), as proposed in e.g. fishes \n(45, 83)  and Andean plants (84). The high rates of ch romosomal rearrangements in Mechanitis and \nMelinaea might accelerate the accumulation of chromosomal differences in isolated populations. On \ncoming together in secondary sympatry, these rearrangements likely cause low hybrid fitness, wh ich in \nturn might facilitate adaptation to different niches due to low levels of gene flow. Selection against \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n18 \n \ninterbreeding might also lead to reinforcement of reproductive isolation via assortative mating, e.g. based \non pheromones (18, 85).  \nOur data suggests that hybridisation -derived genetic variation and high rates of chromosomal \nrearrangements both may have played key roles in the fast diversification of Melinaea and Mechanitis, \nand potentially contributed to adaptation and assortative mating facilitating species coexistence. Notably, \nwhile an important role of hybridisation in diversification has been found in both insular environments \n(e.g. silverswords (44), or cichlids (45)) and  continental radiations (e.g. this study, Heliconius (86), or \nRhagoletis flies (87)), rapid radiations with high rates of chromosomal rearrangements seem to be \nrestricted to continents (e.g. this study, shrews (74), or wallabies (75)). Due to more opportunities for \ngeographical isolation on large land masses than in insular environments, factors such as high rates of \nchromosomal rearrangements that speed up the allopatric accumulation of incompatibilities may thus be \nparticularly important in continental radiations.  \n Our study not only shed s light on drivers of continental radiations, but also largely resolves the \ntaxonomy of important biodiversity indicators. Hitherto, the study of these radiations has been hampered \nby taxonomic challenges, whereas our combination of whole-genome resequencing with vast taxonomic \nand geographic coverage, genome assemblies and androconial chemical analysis allowed us to resolve \ntaxonomic issues. Our study confirms that DNA barcoding can be misleading, massively underestimating \nspecies richness, and should only be used with care to assess biodiversity. The taxonomic implications of \nboth introgression and karyotype diversity for species delimitation and designation of conservation units \nare important to consider during monitoring and management of biodiversity in these vulnerable habitats.  \n \nMaterials and methods \nCollecting butterfly specimens \n157 specimens of Mechanitis, 9 specimens of Forbestra, 109 specimens of Melinaea, 1 Eutresis and 1 \nOlyras were collected over the years 2000 to 2023 across Central and South America (Table S1). Adult \nbutterflies were caught with a net, and their bodies were subsequently preserved in ethanol, DMSO or \nflash-frozen and stored at -70°C. Moreover, a few legs from dried museum specimens were used (Florida \nNatural History Museum; Nat ural History Museum London). Wings were photographed and stored in \nenvelopes. The resulting dataset covers almost all species of Melinaea and Mechanitis from a wide \ngeographical range. In addition, 65 individuals of Mec. lysimnia nesaea (status restored to Mec. nesaea in \nthis paper; hereafter called Mec. nesaea), 19 Mec. lysimnia lysimnia and 8 Mec. polymnia casabranca  \nwere collected for the androconial chemical  analysis (Table S3). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n19 \n \n \nDNA extractions & whole genome resequencing \nDNA extractions were done with either the Qiagen MagAttract High Molecular Weight kit (Qiagen ID \n67563), or the Qiagen QiaAmp DNA mini kit (51304), or a PureLink digestion and lysis step followed by a \nmagnetic bead DNA extraction (88). The dried museum specimens were extracted using a Lysis -C buffer \nand a MinElute DNA extraction kit (protocol adapted from (89); Qiagen ID 28006). Library preparations \nwere performed using homemade TN5 -transposase-mediated tagmentation (protocol adapted from \n(90)), or following the manufacturer’s guidelines with the Illumina DNA PCR -free library prep kit and \nsequenced (150 bp paired -end) on Illumina NovaSeq 6000 or NovaSeq X machines at Novogene or the \nWellcome Sanger Institute.  \n \nReference genomes  \nHaplotype-level chromosomally resolved reference genomes were assembled for five species of  \nMechanitis (Mec. messenoides, menapis, mazaeus, macrinus and polymnia) and Melinaea (Mel. ludovica, \nmarsaeus, mothone, isocomma and menophilus). Note that earlier versions of two Melinaea genomes \nwere published previously (60). In short, we combined 12 -57x PacBio HiFi sequencing and 33 -197x \nIllumina sequencing of HiC libraries (haploid coverages, Table S5) and assembled the genomes according \nto the Tree of Life pipelines (https://github.com/sanger-tol/genomeassembly) (Text S5).  \n \nWhole genome mapping \nTo prepare the whole genome data for analysis, read quality was checked with FastQC (v0.11.9) (91). \nSequences below 50 bp were discarded and adapters and PolyG -tails were trimmed with FastP (v0.23.2) \n(92), before they were aligned to Melinaea marsaeus (60) or Mechanitis messenoides using BWA-mem \n(v.0.7.17) (93). Picard was used to remove PCR duplicates (v3.0.0) (94). Samtools (v1.17) (95) and GATK3 \nHaplotypeCaller (v3.8.1.0) (96, 97) were used for variant calling, with a minimum base quality score of 20.  \nVCFtools (v0.1.16) (98) was used for filtering. Based on the distribution of sequencing depth \n(mean Melinaea: 7; Mechanitis: 15), all sites with a mean depth below 3 (Melinaea) or 5 (Mechanitis), and \nabove 15 (Melinaea) or 30 (Mechanitis) were removed. Insertions, deletions, sites with >50% missing data, \nas well as genotypes with a depth below 2 (Melinaea) or 3 (Mechanitis) were removed. The mitochondrial \nDNA was filtered separately, with a maximum depth of 1700 (Melinaea) or 1200 (Mechanitis). \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n20 \n \nPhylogenetic analyses \nFor each genus, we ran a Principal Component Analysis (PCA) using Plink (V1.9) to explore population \nstructure (99) (Fig. S21). Next, we inferred a phylogenetic tree based on a filtered subset of the whole \ngenome sequence data (also including monomorphic sites, thinned to 1 in 500 sites, with a minimum \ngenotype quality of 10). Our filtered VCF -files were converted to phylip with a custom script which calls \nheterozygous sites as ‘ambiguous’ (equal likelihood for both alleles) to generate one sequence per \nindividual (vcf2phylip.py, http://www.github.com/joanam/scripts), and subsequently, IQtree2 (v2.1.2) \n(100) produced phylogenetic trees with ultrafast bootstrap approximation ( -B 1000; UFBoost) (101) and \nthe GTR-model.  \nWe inferred separate phylogenies for mitochondrial and nuclear DNA. The nuclear trees are based \non 537,500 SNPs for Mechanitis and 784,526 SNPs for Melinaea. The mitochondrial phylogenies are based \non the full mitochondrial genome (not thinned) including 11,818 bp for Mechanitis and 11,815 bp for \nMelinaea. For Mechanitis, we included a maximum of six individuals of the same subspecies and country \nin the phylogenetic analyses, thus excluding several Brazilian Mec. polymnia , Mec. nesaea  and Mec. \nlysimnia. These individuals were included in hybridisation -analyses. For Melinaea, all individuals were \nused. \nIn addition, a phylogenetic tree calibrated with divergence times from (11) was produced using \nBEAST2 (102) (following https://beast2-dev.github.io/beast-\ndocs/beast2/DivergenceDating/DivergenceDatingTutorial.html) for the same dataset as the nuclear \nphylogenetic tree, but thinned further to 1 in 5000 sites and with only one individual per lineage. \nDivergence times between Mec. mazaeus and Mec. macrinus (0.39 MYA) and Mel. isocomma and Mel. \nlilis (0.65 MY) were used for calibration. We used the HKY gamma -4 site model with a strict molecular \nclock and the Calibrated Yule model. The model was run for 15.000.000 chains, stored every 5000 trees. \nPhylogenies were visualised using the packages ‘ape’ (v5.7-1) and ‘phytools’ (v2.1-1) in R (Paradis \nand Schliep 2019; Revell 2024) and FigTree  (v1.1.4) (http://tree.bio.ed.ac.uk/software/figtree/). To \ncalculate a constant speciation rate (𝜆) we assumed a pure birth model, with n as the number of species \nand T the time from root to tip: 𝑛 = 𝑒 𝜆∗𝑇 which gives 𝜆 = 𝑙𝑛(𝑛)/𝑇. \n \nDistribution maps \nThe coordinates of our individuals and the individuals from (10) were plotted using the libraries ‘sf’, \n‘tmap’, ‘tmaptools’ and ‘mapview’ in R (103–105). We classified the subspecies of (10) into speci es \nfollowing our taxonomic revision. Our individuals were plotted as dots on top of the overall distribution \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n21 \n \ncombining data with (10). The basemap ‘Esri.WorldTerrain’ was used, pro vided through ‘Leaflet’ (ESRI, \nARCGIS; https://github.com/leaflet-extras/leaflet-providers); Fig. S6-7 has maps including the attribution.   \n \nWindow trees \nIQtree2 was used to produce phylogenetic trees in windows across the genome (one 20kb window every \n200kb) and to calculate gene concordance factors between these window -trees and the whole-genome \nphylogenetic tree (106).  \n \nDSuite analysis \nWe explored excess allele sharing between species or divergent subspecies using DSuite (107). We filtered \nour genomic dataset to be 1 in 100 sites, only including variab le sites. Dtrios calculated D and f4 -ratio \nstatistics for all trios and Fbranch then summarised them as f-branch statistics using a species tree based \non the nuclear phylogeny of Fig. 2 (Table S1). The results of this analysis were visualised with a python  \nscript provided by DSuite.  \n \nAIM with StarBEAST2 \nWe followed the Approximate Isolat ion with Migration (AIM) in StarBEAST2 tutorial to obtain a \nphylogenetic tree with admixture arrows (102, 108). For each genus, we picked one individual per species \n(highest depth), and randomly extracted 150 (Mechanitis) or 200 (Melinaea) 800-1000bp windows across \nthe genome. We used the HKY gamma-4 site model with a strict molecular clock and the Yule model. The \nmigration rate was set to 25 with an initial value of 0.1, and we ran the model for 100.000.000 chains, \nstoring every 5000 trees. We updated the parameters according to the suggestions in the output of the \nfirst run and re-ran the model to improve the results. \n \nSpecies-tree inference \nPhylogenetic relationships across the genome between species of Melinaea and Mechanitis were inferred \nusing the multispecies coalescent (MSC) approach implemented in BPP v.4.6.2 (109), while accounting for \nincomplete lineage sorting. For each species, only the individual with the highest coverage was included. \nTo minimise the effect of linked selection, loci were required to be at least 2 kb from annotated exons. \nBecause the analysis assumes no intra-locus recombination and independence between loci, we selected \nloci of 300 bp and at least 2 kb from neighbouring loci. Sequence alignments were produced for all loci, \nmasking repetitive elements annotated in the refer ence genome using RepeatMasker v4.1.5 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n22 \n \n(http://repeatmasker.org/). For each locus, individuals with more than 80% missing genotype calls were \nexcluded from the alignment and only loci with all individuals passing filters were considered. \nFurthermore, sites with missing genotype calls were removed and loci with less than 30 bp passing filters \nwere excluded. Heterozygous sites were assigned IUPAC codes. Loci were grouped into blocks of 100 loci. \nSpecies-tree estimation was then performed in BPP v.4.6.2 using the A01 analysis (species-tree inference \nassuming no gene flow). Inverse gamma priors (invGs) were applied both to the root age (τ0) and to \neffective population sizes (θ) – invG(3, 0.06) and invG(3, 0.04), respectively. Parameters were scaled \nassuming a mutation rate of 2.9 × 10−9 substitutions per site per generation and 4 generations per year. \nThe MCMC was run for 1,000,000 iterations after 50,000 iterations of burn -in, sampling every 10 \niterations. \n \nDemographic modelling \nFor the three highly introgressed species Mel. idae, Mec. messenoides and Mec. nesaea, we ran a \nmultispecies-coalescent-with-introgression (MSCi) model implemented in BPP v.4.6. 2 (109) to better \nestimate their position in the phylogeny and divergence timing while accounting for admixture. For Mel. \nidae we considered Mel. lilis and Mel. marsaeus as sister/donor species, while for Mec. messenoides, Mec. \nmenapis and Mec. polymnia (West) were chosen. For Mec. nesaea, we used one individual of each of the \nBrazilian populations (Mec. nesaea, Mec. lysimnia lysimnia , Mec. polymnia casabranca ). Loci were \nselected randomly from autosomes, and required to be at least 2 kb from annotated exons and 10 kb from \nthe nearest locus, and a maximum size of 2 kb. For each locus, individuals wit h more than 50% missing \ndata and sites containing missing genotypes or overlapping annotated repetitive elements were removed. \nOnly loci at least 800 bp long after filters and without missing individuals were considered. Heterozygous \nsites were assigned IUPAC codes. Demographic parameters were estimated using a fixed species tree with \nintrogression events (Fig. S11). An inverse gamma prior (invG) was applied for all population size \nparameters θ (α=3; β=0.04) and root age parameter τ (α=3; β=0.06). A beta pr ior was applied to the \nintrogression probability (φ) (α=1; β=1). Three replicate MCMC runs of 1,000,000 iterations, after a burn-\nin period of 50,000 iterations, sampling every 50 iterations were performed for each dataset. Divergence \ntime was calculated based on 4 generations per year, and a mutation rate of 2.90E-09 as in Heliconius (43) \n(T=τ/mutation rate/generations/million years). \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n23 \n \nGenome scans and ADMIXTURE \nFST, DXY and pi (π) were calculated in all species pairs in 20kb windows (including monomorphic sites, with \na minimum of 10,000 sites per window) across the genome with the scripts from Simon Martin \n(https://github.com/simonhmartin/genomics_general). We also ran ADMIXTURE in subsets of Mechanitis \nand Melinaea to see clustering between individuals with k ranging from 2 to 9 (110). For this dataset, we \nused the same filter settings as for the Fbranch analysis (see above). \n \nIntrogression (fdM) scans  \nTo examine gene flow across the genome of Mec. nesaea, we comp uted f dM with the script by Simon \nMartin (ABBABABAwindows.py, https://github.com/simonhmartin/genomics_general) (111) with various \npopulations for P1,2,3,4. If for example P1=Mec. polymnia casabranca (Brazil), P2=Mec. nesaea, P3=Mec. \nlysimnia Brazil, and P4=Forbestra, high fdM values indicate introgression between Mec. lysimnia and Mec. \nnesaea. We also tested other outgroups (Mec. messenoides) and with five single Mec. nesaea individuals, \nto see if the signal was consistent in different individuals (Fig. S14).  \n \nAndroconial Chemistry \nWe obtained samples of the androconial secretions from 92 males (Text S5). To exclude non-androconial \ncompounds from our analyses, for each population the same extraction procedure was adopted with \nwings of conspecific females and two non-androconial wing areas of males. Eight solvent negative controls \nwere also taken for each sampling event. \nThe peak areas o f each chromatogram were integrated with MSD ChemStation E.02.01.1177 \n(Agilent Technologies, USA) to obtain the total ion current signals. A series of linear alkanes (C7–C40) was \nused to determine the linear retention indices (RI) of each compound (112). Compounds were identified \nby comparing their mass spectra and retention indices with those of reference samples available from \npersonal and commercial mass spectral libraries (FFNSC 2, MassFinder 4, NIST17, MACE v.5.1 (113), and \nWiley Registry™ 9th ed.). The peaks exclusive to the androconial samples were used t o determine the \nrelative percentages of each compound per sample. \nDihydropyrrolizines, such as hydroxydanaidal or methyl hydroxydanaidoate, are typically \naccompanied by smaller peaks formed by degradation during GC/MS analyses and/or storage (114). These \ndegradation peaks were excluded from the generated ion chromatograms and statistical analysis. To avoid \nnon-evident contaminants, we only considered compounds present in more than three individuals for any \ngiven taxon. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n24 \n \nThe data were plotted in R (115) using the vegan package \n(https://doi.org/10.32614/CRAN.package.vegan) for nonmetric multidimensional scaling with \n‘monoMDS’, specifying a global model, square root transformation and Wisconsin double standardisation \n(autotransform=TRUE). \n \nChromosome rearrangements \nSynteny and breakpoint analysis was performed using single copy orthologous genes identified with \nBUSCO version 5.7.1 with the lineage da tabase lepidoptera_odb10 and otherwise default options (116). \nTo compare large scale rearrangements in the Mechanitis and Melinaea versus the b utterfly ancestral \nlinkage groups (Merian elements), we used two outgroup genomes, Melitaea cinxia (117) and Danaus \nplexippus (118). The output from BUSCO was filtered with a custom script to contain only complete genes \nlocated in chromosome-sized scaffolds and excluding W. Only single copy genes were included with the \nexception of genes on the neo-Z2, where we also included genes classified as duplicated. The sex linking \nof Z2 appears to be recent and most of the genes had high similarity to the genes on the corresponding \nW’s and were therefore classified as duplicated by BUSCO. Genes that were actually duplicated (occurring \nmore than once or present on other chromosomes than W) on the Z2 were removed. We determined \nsyntenic blocks, excluding single gene translocations, by comparing the position of the BUSCO -genes in \neach genome against the Merian elements (Melitaea cinxia), and visualised the syntenic relationship with \nthe R -package gggenomes (119). Minimum number of fusions were determined by the number of \ndifferent Merian elements located in every query chromosome and fissions were determined as the \nnumber of query chromosomes containing parts of each Merian element for each species. T he \nbreakpoints between all species within Mechanitis and Melinaea was determined by the same principle \nas above using an all against all approach for each genome to compare the BUSCO-gene positions. Synteny \nanalysis of the sex chromosomes and between haplotypes were performed with whole genome alignment \nusing minimap2/2.27 with default settings and -x asm10 (1% sequence divergence) (120) and visualised \nafter removing short alignments (<100kb for multispecies alignment, <500kb for haplotype alignment) \nusing a modified version of the R-package Farre-lab/syntenyPlotteR (121). \nTo investigate the association between chromosomal rearrangements and species divergence we \nmapped the location of the breakpoints between each comparison to the reference genomes Mec. \nmessenoides and Mel. marsaeus. Divergence and diversity was estimated i n 20 kb windows along the \ngenome (detailed above). To determine whether the observed statistics in the breakpoint regions were \ndifferent from the random expectation of regions of the same size and number across the genome, we \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n25 \n \nused permTest with the randomi se function option ‘randomizeRegions’ and evaluate function option \n‘meanInRegions’ for 50,000 permutations, as implemented in the R-package regioneR (122). The analyses \nabove were performed in R v4.4.0 (115). For conservative interpretation multiple correction was \nperformed with Bonferroni adjustment (ɑ/n), where ɑ = 0.05 and n = 10 for the comparisons within each \ngenus, resulting in values considered significantly different if their p -value from the randomisation test \nwas less than 0.005. \n \nFunding \nThis research was funded with the Wellcome  Trust award 220540/Z/20/A, a Branco Weiss Fellowship, a \nRoyal Society University Research Fellowship (URF\\R1\\221041) and a Bateson Research Fellowship by St \nJohn’s College, Cambridge awarded to JIM. ESMH was supported by NERC DTP C -CLEAR, the Zoology \nDepartment of the University of Cambridge, St John’s College, Cambridge, and the Wellcome Sanger \nInstitute PhD programme. CEBN had a PhD Scholarship from Coordenação de Aperfeiçoamento de Pessoal \nde Nível Superior. Sample collection was further supported by Le verhulme trust (KW and ME), a Phyllis \nand Eileen Gibbs Fellowship, ATIP grant, ANR SPECREP and CLEARWING, and HFSP RGP0014/2016 (ME \nand MM), a Research Fellowship from the Royal Commission for the Great Exhibition of 1851 and a Royal \nSociety Research Grant (SHM) and Deutsche Forschungsgemeinschaft, Schu984/12-2 (SS).  \n \nAcknowledgements \nWe thank Ismael Aldas and Raúl Aldaz for assistance with catching butterflies in Ecuador, Mario Tuanama \nand Ronald Mori-Pezo for help with rearing and collecting butterflies in Peru, Augusto Rosa for sampling \nsupport in São Paulo, Erika de Castro for sampling in Brazil, Mathieu Joron for sampling in French Guiana \nand John R. MacDonald for sampling in Panama. We thank the Peruvian, Ecuadorian and Brazilian \nauthorities as well a s the Museo de Historia Natural in Peru and the Museo Ecuatoriano de Ciencias \nNaturales in Ecuador for their support. Many thanks to Dr Blanca Huertas and Robyn Crowther from the \nNatural History Museum in London (NHMUK) for providing butterfly legs of Mel. isocomma and Mel. \nmothone and to Dr Petra Korlevic for assistance with extracting DNA from museum specimens. We are \ngrateful to the editor and reviewers for their helpful comments that have greatly improved our \nmanuscript. \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n26 \n \nData, materials and software availability \nAll code and tables underlying the figures are  found in our GitHub repository: \nhttps://github.com/rapidspeciation/mechanitis_melinaea. \n \nReferences \n1.  J. B. Losos, Adaptive radiation, ecological opportunity, and evolutionary determinism : \nAmerican society of naturalists E. O. Wilson award address. American Naturalist 175, \n623–639 (2010). \n2.  R. G. Gillespie, et al., Comparing Adaptive Radiations Across Space, Time, and Taxa in \nJournal of Heredity, (Oxford University Press, 2020), pp. 1–20. \n3.  E. Gittenberger, Radiation and adaptation, evolutionary biology and semantics. \norganisms, diversity & evolution 4, 135–136 (2004). \n4.  J. E. Czekanski-Moir, R. J. Rundell, The Ecology of Nonecological Speciation and \nNonadaptive Radiations. Trends in Ecology & Evolution 34, 400–415 (2019). \n5.  R. De-Kayne, et al., Why Do Some Lineages Radiate While Others Do Not? Perspectives \nfor Future Research on Adaptive Radiations. Cold Spring Harb Perspect Biol a041448 \n(2024). https://doi.org/10.1101/cshperspect.a041448. \n6.  J. Cerca, et al., Evolutionary genomics of oceanic island radiations. Trends in Ecology & \nEvolution 38, 631–642 (2023). \n7.  S. Claramunt, Discovering exceptional diversifications at continental scales: the case of \nthe endemic families of neotropical suboscine passerines. Evolution (2010). \nhttps://doi.org/10.1111/j.1558-5646.2010.00971.x. \n8.  R. Maestri, et al., The ecology of a continental evolutionary radiation: Is the radiation of \nsigmodontine rodents adaptive? Evolution 71, 610–632 (2017). \n9.  S. Tiatragul, A. Skeels, J. S. Keogh, Morphological evolution and niche conservatism \nacross a continental radiation of Australian blindsnakes. Evolution 78, 1854–1868 (2024). \n10.  M. Doré, et al., Anthropogenic pressures coincide with Neotropical biodiversity hotspots in \na flagship butterfly group. Diversity and Distributions 28, 2912–2930 (2022). \n11.  N. Chazot, et al., Renewed diversification following Miocene landscape turnover in a \nNeotropical butterfly radiation. Global Ecology and Biogeography 28, 1118–1132 (2019). \n12.  G. W. Beccaloni, K. J. Gaston, Predicting the species richness of neotropical forest \nbutterflies: Ithomiinae (Lepidoptera: Nymphalidae) as indicators. Biological Conservation \n71, 77–86 (1995). \n13.  K. S. Brown, “Conservation of Neotropical Environments: Insects as Indicators” in The \nConservation of Insects and Their Habitats, (Elsevier, 1991), pp. 349–404. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n27 \n \n14.  K. S. Brown, Adult-obtained pyrrolizidine alkaloids defend ithomiine butterflies against a \nspider predator. Nature 309, 707–709 (1984). \n15.  P. J. Devries, F. G. Stiles, Attraction of Pyrrolizidine Alkaloid Seeking Lepidoptera to \nEpidendrum paniculatum Orchids. 22, 290–297 (1990). \n16.  J. R. Trigo, K. S. Brown, Variation of pyrrolizidine alkaloids in Ithomiinae: a comparative \nstudy between species feeding on Apocynaceae and Solanaceae. Chemoecology 22, \n22–29 (1990). \n17.  A. V. L. Freitas, et al., Tropane and pyrrolizidine alkaloids in the ithomiinesPlacidula \neuryanassa andMiraleria cymothoe (Lepidoptera: Nymphalidae). Chemoecology 7, 61–67 \n(1996). \n18.  M. McClure, et al., Why has transparency evolved in aposematic butterflies? Insights from \nthe largest radiation of aposematic butterflies, the Ithomiini. Proceedings of the Royal \nSociety B: Biological Sciences 286 (2019). \n19.  F. Müller, Ituna and Thyridia: a remarkable case of Mimicry in Butterflies. Kosmos May, \n100 (1879). \n20.  G. W. Beccaloni, Ecology, natural history and behaviour of ithomiine butterflies and their \nmimics in Ecuador (Lepidoptera: Nymphalidae: Ithomiinae). Tropical Lepidoptera 8, 103–\n124 (1997). \n21.  A. Y. Kawahara, et al., A global phylogeny of butterflies reveals their evolutionary history, \nancestral hosts and biogeographic origins. Nature Ecology and Evolution 7, 903–913 \n(2023). \n22.  R. M. Fox, A monograph of the Ithomiidae (Lepidoptera). Part III. The Tribe Mechanitini. \nMemoirs of the American Entomological Society 22 (1967). \n23.  K. K. Dasmahapatra, M. Elias, R. I. Hill, J. I. Hoffman, J. Mallet, Mitochondrial DNA \nbarcoding detects some species that are real, and some that are not. Molecular Ecology \nResources 10, 264–273 (2010). \n24.  M. Elias, et al., Limited performance of DNA barcoding in a diverse community of tropical \nbutterflies. Proc. R. Soc. B. 274, 2881–2889 (2007). \n25.  R. I. Hill, et al., Ecologically relevant cryptic species in the highly polymorphic Amazonian \nbutterfly Mechanitis mazaeus s.l. (Lepidoptera: Nymphalidae; Ithomiini). Biological \nJournal of the Linnean Society 106, 540–560 (2012). \n26.  K. S. Brown, Geographical patterns of evolution in Neotropical Lepidoptera: differentiation \nof the species of Melinaea and Mechanitis (Nymphalidae, Ithomiinae). Systematic \nEntomology 2, 161–197 (1977). \n27.  W. T. M. Forbes, A second review of Melinaea and Mechanitis (Lepidoptera, Ithomiinae). \nJournal of the New York Entomological Society 56, 1–24 (1948). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n28 \n \n28.  R. M. Fox, A monograph of the Ithomiidae (Lepidoptera). Part II. The Tribe Melinaeini \nClark. Transactions of the American Entomological Society 86, 109–171 (1960). \n29.  G. Lamas, Checklist: Part 4A. Hesperioidea - Papilionoidea. Atlas of Neotropical \nLepidoptera 5 (2004). \n30.  K. R. Willmott, J. Mallet, Correlations between adult mimicry and larval host plants in \nithomiine butterflies. Proceedings of the Royal Society B: Biological Sciences 271 (2004). \n31.  N. Chazot, et al., Mutualistic Mimicry and Filtering by Altitude Shape the Structure of \nAndean Butterfly Communities. The American Naturalist 183, 26–39 (2014). \n32.  M. McClure, M. Elias, Unravelling the role of host plant expansion in the diversification of \na Neotropical butterfly genus. BMC Evolutionary Biology 16 (2016). \n33.  I. Birskis-Barros, A. V. L. Freitas, P. R. Guimarães, Habitat generalist species constrain \nthe diversity of mimicry rings in heterogeneous habitats. Sci Rep 11, 5072 (2021). \n34.  K. S. Brown, Quaternary refugia in tropical America: evidence from race formation in \nHeliconius butterflies. Proc. R. Soc. Lond. B. 187, 369–378 (1974). \n35.  A. Whinnett, et al., Strikingly variable divergence times inferred across an Amazonian \nbutterfly ‘suture zone.’ Proc. R. Soc. B. 272, 2525–2533 (2005). \n36.  K. K. Dasmahapatra, G. Lamas, F. Simpson, J. Mallet, The anatomy of a “suture zone” in \nAmazonian butterflies: A coalescent-based test for vicariant geographic divergence and \nspeciation. Molecular Ecology 19, 4283–4301 (2010). \n37.  J. Haffer, Speciation in Amazonian Forest Birds: Most species probably originated in \nforest refuges during dry climatic periods. Science 165, 131–137 (1969). \n38.  N. Chazot, et al., Contrasting patterns of Andean diversification among three diverse \nclades of Neotropical clearwing butterflies. Ecology and Evolution 8, 3965–3982 (2018). \n39.  J. Mallet, Hybrid speciation. Nature 446, 279–283 (2007). \n40.  The Heliconius Genome Consortium, Butterfly genome reveals promiscuous exchange of \nmimicry adaptations among species. Nature 487, 94–98 (2012). \n41.  R. Abbott, et al., Hybridization and speciation. J of Evolutionary Biology 26, 229–246 \n(2013). \n42.  D. A. Marques, K. Lucek, V. C. Sousa, L. Excoffier, O. Seehausen, Admixture between \nold lineages facilitated contemporary ecological speciation in Lake Constance \nstickleback. Nature Communications 10 (2019). \n43.  N. Rosser, et al., Hybrid speciation driven by multilocus introgression of ecological traits. \nNature 628, 811–817 (2024). \n44.  M. Barrier, B. G. Baldwin, R. H. Robichaux, M. D. Purugganan, Interspecific hybrid \nancestry of a plant adaptive radiation: allopolyploidy of the Hawaiian silversword alliance \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n29 \n \n(Asteraceae) inferred from floral homeotic gene duplications. Molecular Biology and \nEvolution 16, 1105–1113 (1999). \n45.  J. I. Meier, et al., Cycles of fusion and fission enabled rapid parallel adaptive radiations in \nAfrican cichlids. Science 381, eade2833 (2023). \n46.  R. W. R. Wallbank, et al., Evolutionary Novelty in a Butterfly Wing Pattern through \nEnhancer Shuffling. PLoS Biology 14 (2016). \n47.  L. H. Rieseberg, et al., Major Ecological Transitions in Wild Sunflowers Facilitated by \nHybridization. Science 301, 1211–1216 (2003). \n48.  K. S. Brown, B. V. Schoultz, E. Suomalainen, Chromosome evolution in Neotropical \nDanainae and Ithomiinae (Lepidoptera). Hereditas 141, 216–236 (2004). \n49.  C. A. Everett, J. B. Searle, B. M. N. Wallace, A study of meiotic pairing, nondisjunction \nand germ cell death in laboratory mice carrying Robertsonian translocations. Genet. Res. \n67, 239–247 (1996). \n50.  F. J. Ayala, M. Coluzzi, Chromosome speciation: Humans, Drosophila, and mosquitoes. \n(2005). \n51.  K. Yoshida, et al., Chromosome fusions repatterned recombination rate and facilitated \nreproductive isolation during Pristionchus nematode speciation. Nat Ecol Evol (2023). \nhttps://doi.org/10.1038/s41559-022-01980-z. \n52.  M. A. F. Noor, K. L. Grams, L. A. Bertucci, J. Reiland, Chromosomal inversions and the \nreproductive isolation of species. Proc. Natl. Acad. Sci. U.S.A. 98, 12084–12088 (2001). \n53.  L. H. Rieseberg, Chromosomal rearrangements and speciation. (2001). \n54.  M. McClure, B. Dutrillaux, A.-M. Dutrillaux, V. Lukhtanov, M. Elias, Heterozygosity and \nChain Multivalents during Meiosis Illustrate Ongoing Evolution as a Result of Multiple \nHolokinetic Chromosome Fusions in the Genus Melinaea (Lepidoptera, Nymphalidae). \nCytogenet Genome Res 153, 213–222 (2017). \n55.  J. Mallet, A species definition for the Modern Synthesis. TREE 10, 294–299 (1995). \n56.  M. McClure, M. Elias, Ecology, life history, and genetic differentiation in Neotropical \nMelinaea (Nymphalidae: Ithomiini) butterflies from north-eastern Peru. Zoological Journal \nof the Linnean Society 179, 110–124 (2016). \n57.  A. H. B. Rosa, E. P. Barbosa, N. Wahlberg, A. V. L. Freitas, Systematic position and \nconservation aspects of Melinaea mnasias thera (Lepidoptera: Nymphalidae: Danainae). \nNat. Conserv. Res. 9 (2024). \n58.  J. Vasconcellos-Neto, K. S. Brown, Interspecific Hybridization in Mechanitis Butterflies \n(Ithomiinae): a Novel Pathway for the Breakdown of Isolating Mechanisms. Biotropica 14, \n288 (1982). \n59.  R. Robinson, Lepidoptera Genetics (Pergamon, 1971). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n30 \n \n60.  J. Gauthier, et al., First chromosome scale genomes of ithomiine butterflies \n(Nymphalidae: Ithomiini): Comparative models for mimicry genetic studies. Molecular \nEcology Resources 23, 872–885 (2023). \n61.  S. H. Martin, J. W. Davey, C. Salazar, C. D. Jiggins, Recombination rate variation shapes \nbarriers to introgression across butterfly genomes. PLoS Biol 17, e2006288 (2019). \n62.  K. Näsvall, et al., Nascent evolution of recombination rate differences as a consequence \nof chromosomal rearrangements. PLoS Genet 19, e1010717 (2023). \n63.  T. E. Cruickshank, M. W. Hahn, Reanalysis suggests that genomic islands of speciation \nare due to reduced diversity, not reduced gene flow. Molecular Ecology 23, 3133–3157 \n(2014). \n64.  K. M. Kozak, et al., Multilocus species trees show the recent adaptive radiation of the \nmimetic heliconius butterflies. Systematic Biology 64, 505–524 (2015). \n65.  M. Rossi, et al., Adaptive introgression of a visual preference gene. (2024). \n66.  M. Kirkpatrick, N. Barton, Chromosome Inversions, Local Adaptation and Speciation. \nGenetics 173, 419–434 (2006). \n67.  A. Mackintosh, Chromosome Fissions and Fusions Act as Barriers to Gene Flow between \nBrenthis Fritillary Butterflies. Molecular Biology and Evolution 40 (2023). \n68.  C. J. Wright, L. Stevens, A. Mackintosh, M. Lawniczak, M. Blaxter, Comparative \ngenomics reveals the dynamics of chromosome evolution in Lepidoptera. Nat Ecol Evol 8, \n777–790 (2024). \n69.  V. A. Lukhtanov, V. Dincă, G. Talavera, R. Vila, Unprecedented within-species \nchromosome number cline in the Wood White butterfly Leptidea sinapis and its \nsignificance for karyotype evolution and speciation. BMC Evol Biol 11, 109 (2011). \n70.  H. Augustijnen, et al., A macroevolutionary role for chromosomal fusion and fission in \nErebia butterflies. Science Advances (2024). \n71.  J. M. De Vos, H. Augustijnen, L. Bätscher, K. Lucek, Speciation through chromosomal \nfusion and fission in Lepidoptera. Phil. Trans. R. Soc. B 375, 20190539 (2020). \n72.  L. Höök, K. Näsvall, R. Vila, C. Wiklund, N. Backström, High-density linkage maps and \nchromosome level genome assemblies unveil direction and frequency of extensive \nstructural rearrangements in wood white butterflies (Leptidea spp.). Chromosome Res 31, \n2 (2023). \n73.  V. A. Lukhtanov, et al., Versatility of multivalent orientation, inverted meiosis, and rescued \nfitness in holocentric chromosomal hybrids. Proc. Natl. Acad. Sci. U.S.A. 115 (2018). \n74.  S. Garagna, M. Zuccotti, J. B. Searle, C. A. Redi, P. J. Wilkinson, Spermatogenesis in \nheterozygotes for Robertsonian chromosomal rearrangements from natural populations of \nthe common shrew, Sorex araneus. Reproduction 87, 431–438 (1989). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n31 \n \n75.  S. Potter, et al., Limited Introgression between Rock-Wallabies with Extensive \nChromosomal Rearrangements. Molecular Biology and Evolution 39, msab333 (2022). \n76.  J. B. S. Haldane, Sex ratio and unisexual sterility in hybrid animals. Journ. of Gen. 12, \n101–109 (1922). \n77.  L. Z. Carabajal Paladino, et al., Sex Chromosome Turnover in Moths of the Diverse \nSuperfamily Gelechioidea. Genome Biology and Evolution 11, 1307–1319 (2019). \n78.  A. J. Mongue, M. E. Hansen, J. R. Walters, Support for faster and more adaptive Z \nchromosome evolution in two divergent lepidopteran lineages*. Evolution 76, 332–345 \n(2022). \n79.  D. Charlesworth, B. Charlesworth, Sex differences in fitness and selection for centric \nfusions between sex-chromosomes and autosomes. Genet. Res. 35, 205–214 (1980). \n80.  N. Rueda-M, et al., Genomic evidence reveals three W-autosome fusions in Heliconius \nbutterflies. PLoS Genet 20, e1011318 (2024). \n81.  J. Šíchová, et al., Fissions, fusions, and translocations shaped the karyotype and multiple \nsex chromosome constitution of the northeast-Asian wood white butterfly, Leptidea \namurensis. Biol. J. Linn. Soc. 118, 457–471 (2016). \n82.  T. Xiong, et al., A polygenic explanation for Haldane’s rule in butterflies. Proc. Natl. Acad. \nSci. U.S.A. 120, e2300959120 (2023). \n83.  J. April, R. H. Hanner, A.-M. Dion-Côté, L. Bernatchez, Glacial cycles as an allopatric \nspeciation pump in north-eastern American freshwater fishes. Molecular Ecology 22, \n409–422 (2013). \n84.  B. Nevado, N. Contreras‐Ortiz, C. Hughes, D. A. Filatov, Pleistocene glacial cycles drive \nisolation, gene flow and speciation in the high‐elevation Andes. New Phytologist 219, \n779–793 (2018). \n85.  P. M. B. Bacquet, et al., Selection on male sex pheromone composition contributes to \nbutterfly reproductive isolation. Proc. R. Soc. B. 282, 20142734 (2015). \n86.  R. M. Merrill, et al., The diversification of Heliconius butterflies: What have we learned in \n150 years? Journal of Evolutionary Biology 28, 1417–1438 (2015). \n87.  T. H. Q. Powell, et al., The Build-Up of Population Genetic Divergence along the \nSpeciation Continuum during a Recent Adaptive Radiation of Rhagoletis Flies. Genes 13, \n275 (2022). \n88.  M. Kucka, Chan, Yingguang Frank, HMW DNA extraction using magnetic beads. \nprotocols.io (2022). https://dx.doi.org/10.17504/protocols.io.b46bqzan. \n89.  P. Korlević, et al., A Minimally Morphologically Destructive Approach for DNA Retrieval \nand Whole-Genome Shotgun Sequencing of Pinned Historic Dipteran Vector Species. \nGenome Biology and Evolution 13, evab226 (2021). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n32 \n \n90.  S. Picelli, et al., Tn5 transposase and tagmentation procedures for massively scaled \nsequencing projects. Genome Research 24, 2033–2040 (2014). \n91.  Andrews, S, FastQC: A quality control tool for high throughput sequence data. (2010). \n92.  S. Chen, Y. Zhou, Y. Chen, J. Gu, Fastp: An ultra-fast all-in-one FASTQ preprocessor in \nBioinformatics, (Oxford University Press, 2018), pp. i884–i890. \n93.  H. Li, R. Durbin, Fast and accurate short read alignment with Burrows-Wheeler transform. \nBioinformatics 25, 1754–1760 (2009). \n94.  Picard, Picard Toolkit. Broad Institute (2019). https://broadinstitute.github.io/picard/. \n95.  P. Danecek, et al., Twelve years of SAMtools and BCFtools. GigaScience 10 (2021). \n96.  A. McKenna, et al., The genome analysis toolkit: A MapReduce framework for analyzing \nnext-generation DNA sequencing data. Genome Research 20, 1297–1303 (2010). \n97.  V. Ruano-Rubio, et al., Scaling accurate genetic variant discovery to tens of thousands of \nsamples. (2017). https://doi.org/10.1101/201178. \n98.  P. Danecek, et al., The variant call format and VCFtools. Bioinformatics 27, 2156–2158 \n(2011). \n99.  S. Purcell, et al., PLINK: A tool set for whole-genome association and population-based \nlinkage analyses. American Journal of Human Genetics 81, 559–575 (2007). \n100.  B. Q. Minh, et al., IQ-TREE 2: New Models and Efficient Methods for Phylogenetic \nInference in the Genomic Era. Molecular Biology and Evolution 37, 1530–1534 (2020). \n101.  D. T. Hoang, et al., UFBoot2: Improving the Ultrafast Bootstrap Approximation. Mol. Biol. \nEvol 35, 518–522 (2017). \n102.  R. Bouckaert, et al., BEAST 2: A Software Platform for Bayesian Evolutionary Analysis. \nPLoS Comput Biol 10, e1003537 (2014). \n103.  E. Pebesma, Simple Features for R: Standardized Support for Spatial Vector Data. The R \nJournal 10, 439 (2018). \n104.  M. Tennekes, tmap: Thematic Maps in R. J. Stat. Soft. 84 (2018). \n105.  T. Appelhans, F. Detsch, C. Reudenbach, S. Woellauer, Mapview: Interactive viewing of \nspatial data in R. R package version 2.11.2 (2023). \n106.  B. Q. Minh, M. W. Hahn, R. Lanfear, New methods to calculate concordance factors for \nphylogenomic datasets. Molecular Biology and Evolution 37, 2727–2733 (2020). \n107.  M. Malinsky, M. Matschiner, H. Svardal, Dsuite - Fast D-statistics and related admixture \nevidence from VCF files. Molecular Ecology Resources 21, 584–595 (2021). \n108.  N. F. Müller, et al., Joint inference of species histories and gene flow. [Preprint] (2018). \nAvailable at: http://biorxiv.org/lookup/doi/10.1101/348391 [Accessed 7 November 2024]. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n33 \n \n109.  T. Flouri, X. Jiao, B. Rannala, Z. Yang, A Bayesian Implementation of the Multispecies \nCoalescent Model with Introgression for Phylogenomic Analysis. Molecular Biology and \nEvolution 37, 1211–1223 (2020). \n110.  D. H. Alexander, J. Novembre, K. Lange, Fast model-based estimation of ancestry in \nunrelated individuals. Genome Res. 19, 1655–1664 (2009). \n111.  S. H. Martin, J. W. Davey, C. D. Jiggins, Evaluating the Use of ABBA–BABA Statistics to \nLocate Introgressed Loci. Molecular Biology and Evolution 32, 244–257 (2015). \n112.  H. van D. Dool, P. D. Kratz, A generalization of the retention index system including linear \ntemperature programmed gas-liquid partition chromatography. Journal of \nChromatography A 11, 463–471 (1963). \n113.  S. Schulz, A. Möllerke, MACE – An Open Access Data Repository of Mass Spectra for \nChemical Ecology. J Chem Ecol 48, 589–597 (2022). \n114.  S. Schulz, et al., Semiochemicals derived from pyrrolizidine alkaloids in male ithomiine \nbutterflies (Lepidoptera: Nymphalidae: Ithomiinae). Biochemical Systematics and Ecology \n32, 699–713 (2004). \n115.  R Core Team, A language and environment for statistical computing. R foundation for \nstatistical computing (2024). https://doi.org/10.1080/01621459.1972.10481279. \n116.  M. Manni, M. R. Berkeley, M. Seppey, F. A. Simão, E. M. Zdobnov, BUSCO Update: \nNovel and Streamlined Workflows along with Broader and Deeper Phylogenetic \nCoverage for Scoring of Eukaryotic, Prokaryotic, and Viral Genomes. Molecular Biology \nand Evolution 38, 4647–4654 (2021). \n117.  R. Vila, et al., The genome sequence of the Glanville fritillary, Melitaea cinxia (Linnaeus, \n1758). Wellcome Open Res 6, 266 (2021). \n118.  L. Gu, et al., Dichotomy of Dosage Compensation along the Neo Z Chromosome of the \nMonarch Butterfly. Current Biology 29, 4071-4077.e3 (2019). \n119.  T. Hackl, M. J. Ankenbrand, B. van Adrichem, gggenomes: a grammar of graphics for \ncomparative genomics. R package (2024). https://github.com/thackl/gggenomes. \n120.  H. Li, Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics 34, 3094–\n3100 (2018). \n121.  S. Quigley, J. Damas, D. M. Larkin, M. Farré, syntenyPlotteR: a user-friendly R package \nto visualize genome synteny, ideal for both experienced and novice bioinformaticians. \nBioinformatics Advances 3, vbad161 (2023). \n122.  B. Gel, et al., regioneR: an R/Bioconductor package for the association analysis of \ngenomic regions based on permutation tests. Bioinformatics 32, 289–291 (2016). \n123.  M. Carvalho, A. Victor, L. Freitas, E. Barbosa, Immature stages of Mechanitis polymnia \ncasabranca (Nymphalidae, Danainae). (2019). https://doi.org/10.5281/zenodo.2650300. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint \n\n34 \n \n124.  D. H. A. Melo, A. V. L. Freitas, Immature stages of Mechanitis lysimnia nesaea \n(Nymphalidae: Danainae: Ithomiini). Tropical Lepidoptera 33, 79–85 (2023). \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.07.07.602206doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}