Nymphalidae: Ithomiini) Butterﬂies from North-Eastern Peru

Zoological Journal of the Linnean Society, 2016. With 4 ﬁgures

Ecology, life history, and genetic differentiation in Neotropical Melinaea (Nymphalidae: Ithomiini) butterﬂies from north-eastern Peru

MELANIE MCCLURE* and MARIANNE ELIAS

Institut de Systematique, Evolution, Biodiversite, ISYEB – UMR 7205 – CNRS, MNHN, UPMC, EPHE, Museum National d’Histoire Naturelle, Sorbonne Universites, 57 rue Cuvier, CP50, F-75005 Paris, France

Received 22 December 2015; revised 11 March 2016; accepted for publication 16 March 2016

Butterflies of the genus Melinaea have conspicuous warning colours and are thought to be the prime distasteful models in many cases of mimicry in the Neotropics. Colour pattern variability has made systematics challenging and previous studies have found little to no genetic differentiation. This paper provides detailed descriptions of the immature stages of seven Melinaea taxa from north-eastern Peru, including distribution and host plant use, in addition to measures of genetic differentiation using microsatellite markers and mitochondrial sequences. Development time and immature stages were similar, making it difficult to elucidate taxonomy based on larval morphological characters. All taxa used Juanulloa as a host plant (Solanaceae), except Melinaea ‘marsaeus’ mothone, which occurs at higher elevations and used Trianaea (Solanaceae). The seven taxa show virtually no mitochondrial divergence, suggesting a recent radiation. Microsatellite markers, however, revealed distinct genetic clusters and evidence of admixture, demonstrating a complex diversification history. Ecological and genetic differentiation observed for Mel. ‘marsaeus’ mothone prompts for a taxonomic status revision to Melinaea mothone mothone and the taxonomic status of Melinaea ‘satevis’ tarapotensis remains unclear. Clearly, further work is needed to clarify the systematics and to shed light on the processes driving speciation in this genus.

ADDITIONAL KEYWORDS: aposematic – Danainae – development – distribution – host plants – Lepi- doptera – mimicry – morphology – Solanaceae – taxonomy.

and mimicry (Whinnett et al., 2005; Dasmahapatra INTRODUCTION et al., 2010), and offer an excellent system to study Mullerian€ mimicry, where multiple unpalatable spe- the mechanisms underlying diversification. These cies possess the same warning signal, reduces the butterflies are large, possess conspicuous aposematic negative impact of predation on each species by shar- warning colours, and are distributed across much of ing the cost of educating predators. Mimetic butter- the Neotropics. They are also extensively involved in flies are well suited for studies on speciation, as mimicry rings, including with ‘tiger-patterned’ species often consist of multiple subspecies diverging (black, orange, and yellow) Heliconius, and are for a number of adaptive traits, such as colour pat- thought to drive mimicry in many other Lepidoptera tern, which can then cause reproductive isolation (Brown & Benson, 1974; Beccaloni, 1997). However, through sexual selection and higher hybrid mortality their colour pattern variability and the lack of mor- (Jiggins et al., 2001; Merrill et al., 2012). Butterflies phological differentiation have presented taxonomical in the genus Melinaea (Nymphalidae: Ithomiini) challenges, and consequently the systematics of this have undergone rapid radiation for warning patterns genus remains unclear (Brown, 1977). Previous studies using mitochondrial and nuclear markers have *Corresponding author. E-mail: [email protected] found little to no genetic differences amongst many

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 1 2 M. MCCLURE AND M. ELIAS species of Melinaea, and it is postulated that they general, are extremely difficult to keep and breed in have only recently diverged (Whinnett et al., 2005; cages, owing to specific ecological requirements such Elias et al., 2007; Dasmahapatra et al., 2010). as shaded habitat and plants that provide precursors In other groups of aposematic and mimetic butter- of sexual pheromones. Therefore, mating behaviour flies, detailed descriptions of the life history and of is poorly documented in these species. the immature stages have revealed distinguishing The few scattered records of host-plant use that ecological and morphological characteristics useful exist suggest that Melinaea species are oligophagous for resolving such taxonomic issues (Brown & Fre- on the family Solanaceae, subclade Juanulloeae (see itas, 1994; Hill et al., 2012). Yet, despite Melinaea Table 1). For example, Melinaea lilis has been butterflies being the prime distasteful models in recorded on Merinthopodium neuranthum and Mar- many cases of mimicry and interesting organisms for kea (Dyssochroma) viridiflora in Costa Rica, and on the study of speciation, general biological and life- Juanulloa mexicana in both Costa Rica and Mexico history information for this genus are still lacking. (see Brown & Freitas, 1994; Drummond & Brown, Reasons for the paucity of life-history information 1987). Melinaea menophilus has been recorded on include the difficulty of finding the host-plants of Markea (Hawkesiophyton) ulei in Brazil (Drummond Melinaea species, which are mostly epiphytes & Brown, 1987; Brown & Freitas, 1994), and on Mar- (Knapp, Persson & Blackmore, 1997). In addition, kea sp. (Willmott & Mallet, 2004), J. mexicana Melinaea species, and ithomiine butterflies in (Drummond & Brown, 1987), and Juanulloa

Table 1. Records of host plants for Melinaea, including the country and the altitude at which it was recorded, when available, and the reference from where it was taken.

Melinaea species Host plant Locality Altitude References

Melinaea mneme mneme Markea coccinea Brazil 50–100 m Drummond & Brown (1987) Melinaea mneme mauensis Markea coccinea Brazil 10 m Drummond & Brown (1987) Melinaea ludovica ludovica Markea sp. Brazil 20 m Drummond & Brown (1987) Melinaea ludovica paraiya Markea (Dyssochroma) Brazil Sea level–800 m, Drummond & Brown viridiflora 20–300 m (1987) Melinaea lilis imitata Juanulloa mexicana Mexico, 200–1600 m, sea Drummond & Brown Costa Rica level–1600 m (1987) Melinaea lilis imitata Merinthopodium Costa Rica Sea level–1600 m Drummond & Brown neuranthum (1987) Melinaea lilis imitata Solandra grandiflora Costa Rica Sea level–1600 m Drummond & Brown (1987) Melinaea lilis parallelis Markea (Schultesianthus) Panama Sea level–2000 m Drummond & Brown leucantha (1987) Melinaea marsaeus pothete Markea (Hawkesiophyton) Brazil 200–600 m Drummond & Brown ulei (1987) Melinaea menophilus Juanulloa mexicana Ecuador 280 m Drummond & Brown menophilus (1987) Melinaea menophilus Markea (Hawkesiophyton) Brazil 200–600 m Drummond & Brown ssp. nov. ulei (1987) Melinaea lilis Juanulloa mexicana Mexico Brown & Freitas (1994) Melinaea menophilus Markea (Hawkesiophyton) Brazil Brown & Freitas (1994) ulei Melinaea ludovica paraiya Markea (Dyssochroma) Brazil Brown & Freitas (1994) viridiflora Melinaea ethra Markea (Dyssochroma) Brazil Brown & Freitas (1994) viridiflora Melinaea menophilus Juanulloa ochracea Ecuador K. Willmott, pers. comm. Melinaea mneme mauensis Markea formicarum French 50–100 m M. McClure, pers. observ.; Guiana Fig. S2 Melinaea menophilus zaneka Markea sp. Ecuador Montane Willmott & Mallet (2004)

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 DIFFERENTIATION IN MELINAEA BUTTERFLIES 3 ochracea (K. Willmott, pers. comm.; see Fig. S1E) in collected in proximity to streams and water, confirm- Ecuador. Similarly, Melinaea mneme has been ing Melinaea’s association with humid habitats. Host recorded on Markea coccinea in Brazil (Drummond & plants were also frequently encountered in very Brown, 1987), and M. M. has also successfully reared humid habitats, often near streams and rivers, and Melinaea mneme mauensis on the ant-garden epi- eggs and larvae were found on those plants that were phyte Markea formicarum in French Guiana, despite accessible. Host plants were photographed and botan- it never having been observed to be used in the field ical samples were collected and then dried in silica gel (M. McClure, pers. observ., see Fig. S2). Shifts in or pressed before identification. host-plant usage have been shown to be a major Gravid wild-caught females were placed in cause of diversification in butterflies (Janz, Nylin & 2 9 2 9 2 m outdoor insectaries in Tarapoto, San Wahlberg, 2006; Nylin, Slove & Niklas, 2014). As Martın, where all rearing was carried out. These switches in host plants may be accompanied by cages were in the shade of nearby trees and made of changes in microhabitats and mimicry rings (Will- shade cloth that blocked 50% of sunlight, so as to mott & Mallet, 2004; Willmott & Freitas, 2006), they reflect understorey conditions, and cages were may also cause reproductive isolation and thus watered and sprayed multiple times each day so as potentially lead to speciation. to keep humidity levels high. Butterflies were pro- This paper aims to provide new morphological and vided nourishment in the form of sugar water solu- ecological information that will help clarify the taxo- tion (at approximately 20% sugar concentration) in nomic relationships and the reproductive barriers small suspended cups filled with segments of red that occur in this genus, thereby providing insights straws so as to imitate flowers. Commercially avail- on the mechanisms of speciation. We present able bee pollen reduced into powder and loosely detailed descriptions and illustrations of the imma- packed into small suspended Eppendorf tubes cov- ture stages and information on the life history of ered in red cloth was also provided for nourishment. seven Melinaea taxa (Melinaea satevis cydon, Meli- Juanulloa parasitica collected in the field and then naea ‘satevis’ tarapotensis, Melinaea marsaeus phasi- potted was used for oviposition. Larvae collected in ana, Melinaea marsaeus rileyi, Melinaea ‘marsaeus’ the field and those collected in the cages were reared mothone, Melinaea menophilus ssp. nov. 1, Melinaea individually in transparent plastic containers in the menophilus hicetas) from the departments of San shade behind a nearby building under ambient con- Martın and Loreto, north-eastern Peru, and report ditions. Juanulloa parasitica leaves were offered on the distribution and use of host plants in the dif- ad libitum, and larvae were checked daily for food ferent habitats. We also present an analysis of mito- replacement and cleaning. chondrial sequences and genotyping of 12 All immature stages were measured and pho- polymorphic microsatellite markers to measure dif- tographed. Pictures were taken with a digital camera ferentiation amongst 15 taxa (species and sub- (Olympus OM-D E-M5 with a macro lens 60 mm species); the seven from north-eastern Peru included f2.8), and larval measurements were taken with digi- in this study, in addition to others from Peru, Ecua- tal callipers. All measurements were taken on live dor, French Guiana, and Panama for comparison. larvae a few hours after moulting, and head capsules were collected after moulting and then measured. Head capsule width was the distance between the MATERIAL AND METHODS most external stemmata, and maximum total length Collection localities for the seven reared Melinaea for both larvae and pupae corresponded to the dis- taxa (Mel. satevis cydon, Mel. ‘satevis’ tarapotensis, tance from the head to the posterior margin of the Mel. marsaeus phasiana, Mel. marsaeus rileyi, tenth abdominal segment in dorsal view. Instar Mel. ‘marsaeus’ mothone, Mel. menophilus ssp. nov. duration was also noted. Pupal weight was measured 1, Mel. menophilus hicetas) consisted of forest habi- the morning following pupation using a small porta- tats near Tarapoto [Rio Shilcayo basin: 6°27030″S, ble scale. Newly emerged butterflies of all taxa and 76°21000″W, altitude (alt.) 460 m and the Tunel ridge: of both sexes were kept together in mixed species 6°27011″S, 76°17011″W, alt. 1090 m], Moyobamba outdoor insectaries, with sugar water solution and (6°04034″S, 76°57027″W, alt. 1130 m), Sauce (6°42027″ bee pollen for nourishment, and pyrrolizidine alka- S, 76°13064″W, alt. 619 m), Carachamera (6°25085″S, loid sources in the form of withered Heliotropium sp. 76°15027″W, alt. 280 m), Shapaja (6°36056″S, (Boraginaceae) and Eupatorieae (Asteraceae). Cages 76°09061″W, alt. 195 m), Pongo-Baranquita road were checked frequently every day for mating (6°17053″S, 76°14038″W, alt. 200 m), El Afluente in the events, and the presence of a spermatophore was Alto Mayo (5°39046″S, 77°41045″W, alt. 1320–1733 m), ascertained by palping the female’s abdomen. and Shucushyacu (5°57020″S, 75°53006″W, alt. 183 m), Samples of the seven taxa collected during this Peru, in 2012 and 2013. Many of the samples were study, in addition to samples from the same or

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 4 M. MCCLURE AND M. ELIAS different taxa collected during previous collecting Additionally, we downloaded mitochondrial efforts in Peru, Ecuador, French Guiana, and sequences (the 2297-bp region spanning the genes Panama, and preserved either in ethanol or in salt- cytochrome oxidase subunit I, transfer RNA for leu- saturated 20% dimethylsulphoxide with ethylenedi- cine and cytochrome oxidase subunit II, hereafter aminetetraacetic acid, were genotyped at 12 COI and COII) from GenBank for those same taxa, microsatellite markers developed for Melinaea and except that we omitted Melinaea ‘menophilus’ medi- using the primers and PCR conditions from McClure atrix (Table 2), and Melinaea lilis parallelis was et al. (2014) (Table 2). To assess the extent of genetic used instead of Melinaea lilis messatis. Sequences differentiation and admixture, the number of possible were aligned using CodonCode ALIGNER 6.0. Best genetic clusters (or distinct groups) was investigated partition schemes were investigated using Parti- using STRUCTURE version 2.3.4 (Pritchard, Stevens tionFinder (Lanfear et al., 2012), using the greedy & Donnelly, 2000), which was run with 500 000 algorithm for linked branch lengths. The data set updates of the Markov chain after an initial ‘burn-in’ was partitioned according to the suggested partition of 50 000 updates for one to 17 genetic clusters scheme that contained two sets of nucleotides (the (K = 1–17), with ﬁve replicates at each value of K.No third codon positions of COI and COII, and all the prior information on sampling location was used in other sites), which both evolved under a general the model. To determine the number of clusters that time reversible substitution model plus gamma best describe the data, we used the method described (GTR+G). We used the IQtree web server to per- in Evanno, Regnaut & Goudet (2005), which is based form maximum likelihood phylogenetic analyses of on the second-order rate of change of the log the sequences (Nguyen et al., 2015), and the likelihood. This method was implemented in STRUC- robustness of the topology was assessed using 1000 TURE HARVESTER (Earl & vonHoldt, 2012). bootstraps.

Table 2. Collection locality, number of individuals genotyped using microsatellite markers (N microsats) and mitochondrial sequences (N COI–COII a), and corresponding GenBank accession numbers for each Melinaea species.

N COI– Melinaea species Locality N microsats COII GenBank accession numbers

Melinaea ‘marsaeus’ mothone Peru 26 6 HM051843, HM051847, HM051845, HM051848, HM051844, HM051846 Melinaea ‘marsaeus’ mothone Ecuador 5 5 EU069076, EU068862, EU068863, EU068864, EU069077 Melinaea menophilus ssp. nov. 1 Peru 37 5 EU068871, HM051752, HM051755, HM051757, HM051749 Melinaea menophilus hicetas Peru 18 5 EU068867, HM051736, HM051738, HM051741, HM051742 Melinaea menophilus menophilus Ecuador 11 4 EU069078, EU068869, EU068870, EU069001 Melinaea menophilus zaneka Ecuador 10 1 KU892715 Melinaea marsaeus phasiana Peru 39 6 HM051858, HM051851, HM051865, HM051867, HM051869, HM051870 Melinaea marsaeus rileyi Peru 19 3 DQ078424, DQ078423, DQ078425 (hybrid with Mel. marsaeus phasiana) Melinaea satevis cydon Peru 19 6 EU068872, HM051777, HM051778, HM051779, HM051781, HM051782 Melinaea satevis maeonis Ecuador 11 4 EU068874, EU068967, EU068875, EU069002 Melinaea ‘satevis’ tarapotensis Peru 14 3 EU068876, HM051784, HM051785 Melinaea isocomma simulator Peru 4 4 DQ078404, DQ078402, DQ078403, KF268422 Melinaea lilis messatis Panama 7 1 HM416454 (Mel. lilis parallelis) Melinaea ludovica ludovica Peru 7 2 HM051786, HM051871 Melinaea ludovica ludovica French Guiana 18 None available Melinaea mneme mauensis French Guiana 23 1 JN273731 Melinaea ‘menophilus’ mediatrix French Guiana 17 None available aCytochrome oxidase subunit I, transfer RNA for leucine and cytochrome oxidase subunit II.

RESULTS were most active in the morning (from sunrise to 11:00 h) and in late afternoon (at approximately DISTRIBUTION 15:00–16:00 h), when temperatures were coolest. In There are three widespread species of Melinaea, each the cages, butterflies visited both natural and artifi- consisting of two to three subspecies, in north-east- cial pollen sources (Fig. S4G), and also visited bird ern Peru (departments of San Martın and Loreto) droppings when they were present. Feeding on artifi- (Fig. 1, Table 3). The general pattern of distribution cial pollen sources resembled the feeding behaviour consists of one subspecies occupying transitional observed on bird droppings and on pyrrolizidine forest (e.g. Mel. ‘satevis’ tarapotensis, Mel. menophi- alkaloid containing plants, with the proboscis lus ssp. nov. 1, Mel. marsaeus phasiana, and extended, alternating between periods of actively Mel. ‘marsaeus’ mothone), and the other lowland for- ‘palping’ the source and immobilization, for several est (e.g. Mel. satevis cydon, Mel. menophilus hicetas, minutes to an hour. Feeding would occasionally and Mel. marsaeus rileyi). Melinaea ‘marsaeus’ moth- cease, with the proboscis retracted, only to resume one occurs only in locations of higher elevation, such shortly after. as the Alto Mayo and near Moyobamba (above Both sexes of all studied species were seen visit- 800 m). Although both Mel. ‘marsaeus’ mothone and ing pyrrolizidine alkaloid sources, such as Heliotro- Mel. menophilus ssp. nov. 1 were present in the for- pium sp. (Boraginaceae) and species of Asteraceae est around Tarapoto, their distribution was segre- in the tribe Eupatorieae (Fig. S1M), although males gated by elevation, with Mel. ‘marsaeus’ mothone did so more frequently. Both in Peru and in French occupying only habitats at higher elevations where a Guiana, male butterflies were regularly seen by the different host plant occurs (Trianaea speciosa). Meli- authors feeding on these plants in areas of sec- naea menophilus was the most widely distributed ondary growth. These chemicals are known to con- species in this study, with Mel. menophilus ssp. nov. fer toxicity to the butterflies, and are also thought 1 mainly present in the west (Rio Mayo Valley) near to be used as precursors for pheromones (Brown, Tarapoto, where the transition between lowland and 1984; Boppre, 1986; Schulz et al., 2004; Trigo, Andean forests occurs, and Mel. menophilus hicetas 2011). Like other ithomiines, males have a plume of mainly present in the east, in Amazonian lowland long androconial scales or ‘hair pencils’ on the costa forests from Pongo to Shucushyacu, but with lots of of their hind wings that they display during court- overlap between the two subspecies. ship, most likely to release and disperse the volatile pheromones.

ADULT BIOLOGY MATING All species inhabit primary forests and individuals were observed only in the shady understorey, often Although mating can occur at any time of the day, in close proximity to rivers and streams. Butterﬂies most mating events in the cages were observed early

Figure 1. Relative abundance of the Melinaea taxa in the different localities in north-eastern Peru. The Melinaea taxa are, from top to bottom: Melinaea ‘marsaeus’ mothone, Melinaea menophilus ssp. nov. 1, Melinaea menophilus hicetas, Melinaea marsaeus phasiana, Melinaea marsaeus rileyi, Melinaea satevis cydon, Melinaea ‘satevis’ tarapotensis.

Table 3. Distribution of the host plants in the different locations surveyed in the departments of San Martın and Loreto Peru, the Melinaea species that were collected on them, and the Melinaea species present in that area

Melinaea species for which eggs and/or Melinaea species larvae were collected present in the Locality Host plants present on the host plants habitat

Alto Mayo Markea sp. and Unknown M; S Trianaea speciosa Moyobamba Trianaea speciosa MM;S Tarapoto Trianaea speciosa MM;S;P (Tunel Ridge) Tarapoto (Rio Juanulloa parasitica M; S; H M; S; H; T Shilcayo basin) Sauce Juanulloa parasitica None collected T; S Carachamera Juanulloa parasitica P; S P; S Shapaja Juanulloa parasitica None collected P; S Pongo Juanulloa parasitica C; S; H C; R; P; S; H Shucushyacu Juanulloa parasitica RR;H;P;S;C

M, M. ‘marsaeus’ mothone;S,Melinaea menophilus ssp. nov. 1; H, Melinaea menophilus hicetas;T,Melinaea ‘satevis’ tarapotensis;C,Melinaea satevis cydon;P,Melinaea marsaeus phasiana;R,Melinaea marsaeus rileyi. in the morning, and copula lasted anywhere between Mel. marsaeus rileyi were very rare (N = 1), and like a little over an hour to 24 h. Females were never interspecific crosses, no eggs were ever produced. observed mating more than once, but males mated Mating between Mel. ‘marsaeus’ mothone and with multiple females during their lifespan. Sper- Mel. marsaeus phasiana was never observed. matophores transferred to females during copulation Crosses between Mel. marsaeus phasiana and were absorbed in their entirety after a few days to a Mel. marsaeus rileyi occurred in cages (N = 2) and couple of weeks. Mating appeared to be highly segre- produced offspring, and putative hybrids with inter- gated by colour pattern (e.g. Fig. S1I; M. McClure, mediate phenotypes (Fig. S1K, L) were also collected pers. observ.), and possibly by other species-specific (N = 6) in Shapaja, Pongo-Barranquita road, and cues, such that even though Mel. ‘satevis’ tarapoten- Shucushyacu. The colour pattern of offspring pro- sis and Mel. menophilus ssp. nov. 1 are co-mimics duced by these phenotypic hybrids can resemble that and have overlapping distribution, no mating was of either parent, or be an intermediate variation, observed between the two species. Occasionally, how- with the broods occasionally displaying a range of ever, interspecific matings between species were phenotypes. For example, a ‘hybrid’ female caught in observed, even though the species were not co- Shapaja produced offspring resembling only mimics. The authors captured one mating pair on Mel. marsaeus phasiana, whereas another ‘hybrid’ the Pongo-Barranquita road consisting of a female female caught in Pongo-Barranquita produced both Mel. satevis cydon and a male Mel. marsaeus rileyi. offspring that resembled Mel. marsaeus rileyi and In mixed species insectaries kept at the time of this many intermediate forms. Similarly, a ‘hybrid’ male study, interspecies mating pairs were also observed caught in Shucushyacu that was mated with a cap- in cages, albeit rarely, between Mel. marsaeus phasi- tive-bred virgin Mel. marsaeus rileyi female also pro- ana and Mel. satevis cydon (N=1), Mel. marsaeus duced both ‘pure’ Mel. marsaeus rileyi and phasiana and Mel. menophilus ssp. nov. 1 (N = 2), intermediates. Mel. satevis cydon and Mel. ‘marsaeus’ mothone Melinaea menophilus ssp. nov. 1 and (N = 2), and Mel. satevis cydon and Mel. menophilus Mel. menophilus hicetas readily mated in cages hicetas (N = 1) (see Fig. S1G, H, J), but no eggs were (N = 9; Fig. S1F) and these crosses produced off- ever produced from these crosses despite the pres- spring. Indeed, gravid females of this species caught ence of a spermatophore. in the wild will often produce broods consisting of both Matings between subspecies of the same species subspecies. Resulting offspring consist of both the also occurred, although they varied in propensity Mel. menophilus ssp. nov. 1 and the Mel. menophilus amongst the different species. Even in cages, mat- hicetas phenotypes, albeit with a biased ratio of more ings between Mel. ‘marsaeus’ mothone and Mel. menophilus hicetas being produced.

all, of the eggshell before initiating feeding on the HOST PLANT leaf. Host plants were determined to be J. parasitica in most of the studied areas and T. speciosa at higher Larvae elevations (i.e. above 800 m; see Table 3 and Larvae of all species (Figs S2–S4) have black head Fig. S1A–D), such as the Alto Mayo and Moyobamba, capsules and black thoracic legs in all instars. White but also at higher elevations in the Cordillera Esca- stripes in a characteristic inverted ‘Y’ appear on the lera near Tarapoto. Both plants are epiphytes and black head capsule when larvae moult to the fourth are found in the canopy of primary forest. Larval instar. First-instar larvae have small buds on either feeding damage is circular (see Figs S1B, S2G, S3A.3 side of the mesothorax. By the second instar, these for examples), such that damage can be identified buds have fully developed into the characteristic even on plants high in the canopy. When accessible, mobile filaments that protrude behind the head of eggs and larvae were found on these plants. Some Melinaea caterpillars. First-instar caterpillars are eggs were parasitized by an unidentified species of translucent greenish grey shortly after hatching, but wasp (Fig. S4H), and this was most frequent with reddish brown bands on a creamy background eggs collected from plants along the riverbank and in quickly appear. Larval colour pattern did not differ forest gaps. The only species ever collected from greatly amongst the different species, except for T. speciosa was Mel. ‘marsaeus’ mothone, even Mel. satevis, which resembled Mel. mneme mauensis when Mel. menophilus ssp. nov. 1 was present at low from French Guiana. These have thicker brick red densities. bands and a whiter inverted ‘Y’ on the head capsule, although this difference is less apparent after moulting to the fifth instar. When resting, larvae do so in OVIPOSITION a characteristic ‘J’ position, as described by Brown & Eggs were laid singly on the underside of mature Freitas (1994), and when disturbed, larvae thrash leaves, occasionally with multiple eggs on a single the anterior portion of their bodies, occasionally leaf, and generally on multiple leaves of a given regurgitating gut content. This may be an adaptive plant. Females were observed searching for hosts antipredator behaviour, as seen in caterpillars of and ovipositing at all times of day, except during the other species (e.g. Peterson, Johnson & LeGuyader, warmest hours between 12:00 and 14:00 h, when 1987; McClure & Despland, 2011). most butterflies were inactive. Females spent a few minutes assessing the host plant by hovering over Pupae the leaves and inspecting them with their antennae Pupae (Figs S2–S4) are squat/round in shape and and forelegs before laying their eggs on the under- suspended under a leaf of the host plant. The colour side of the leaves. Some females produced offspring is a shade of yellow with variable black markings, of only the female sex (approximately 17% of all and no differences were observed amongst the differ- wild-caught females). After closer inspection, the ent species. Wing colours start to appear the day authors noticed that in some instances, some of the before butterfly eclosion (e.g. see Figs S2N, S4K). eggs never hatched, and a dead embryo was discov- Pupae of Mel. ‘marsaeus’ mothone are significantly ered when the eggs were dissected (Fig. S4I, J). This heavier than those of the lowland species Mel. may be the result of male-killing caused by infection satevis cydon, Mel. marsaeus phasiana, and by the bacterium Wolbachia and may in turn cause a Mel. marsaeus rileyi (F = 7.045; df = 6, 215; sex-ratio distortion in the population (e.g. Jiggins, P < 0.001; Fig. 2). Melinaea are also sexually Hurst & Majerus, 1998; Charlat, Hurst & Mercot, dimorphic with respect to weight (F = 13.173; df = 1; 2003; Engelstadter€ & Hurst, 2009). 220; P < 0.001), with male pupae weighing 455.60 6.81 mg (mean SEM; N = 90) on average, and females 425.14 5.15 mg (mean SEM; DESCRIPTION OF EARLY STAGES N = 132). Developmental rates were very similar for Data for eggs, instars 1–5, and pupae are shown in all species, taking 26.03 0.12 days (mean SEM; Table 4. N = 105) on average, from the egg being laid to butterfly emergence, under natural conditions (Table 5; Eggs F = 3.03; df = 5, 99;P= 0.058). Eggs (Figs S2F, S3) of all species are creamy white to yellow, with Mel. ‘marsaeus’ mothone having GENETIC DIFFERENTIATION whiter eggs than the other species. The eggs are tal- ler than they are wide and widest in the middle. Mitochondrial sequences show very little to no differ- Newly hatched larvae occasionally consume part, or entiation amongst the Melinaea taxa studied, except

Table 4. Size (mm) and weight (mg) (mean SE) of immature stages of seven Melinaea spp. Egg measurements (N = 15 individuals/ssp.) are width 9 height. Larval measurements (N = 15 individuals/ssp.) are body length 9 width of head capsules, and pupal measurements (N = 15 individuals/ssp.) are body length 9 body width. Pupal weight (mg) was taken the morning after larvae pupated (N = 30 individuals/ssp.).

Melinaea species Egg First instar Second instar Third instar Fourth instar Fifth instar Pupa Weight (mg)

Melinaea 0.87 0.02 9 3.31 0.08 9 6.06 0.21 9 10.11 0.31 9 15.55 0.36 9 22.68 0.54 9 14.65 0.14 9 414.60 8.98 © 06TeLnenSceyo London, of Society Linnean The 2016 satevis cydon 1.36 0.03 0.60 0.02 0.97 0.02 1.36 0.03 2.09 0.03 3.62 0.08 8.13 0.07 Melinaea 0.88 0.01 9 3.91 0.13 9 6.10 0.12 9 9.11 0.23 9 13.31 0.52 9 21.03 0.92 9 14.67 0.15 9 454.29 9.79 menophilus 1.33 0.01 0.62 0.03 0.96 0.03 1.45 0.04 2.18 0.14 3.58 0.08 8.04 0.13 hicetas Melinaea 0.96 0.02 9 4.22 0.18 9 6.14 0.24 9 9.78 0.28 9 15.60 0.45 9 21.75 0.36 9 14.94 0.15 9 453.79 9.65 menophilus 1.35 0.02 0.67 0.01 0.98 0.02 1.50 0.03 2.13 0.04 3.41 0.07 8.36 0.10 ssp. nov. 1 Melinaea 0.92 0.02 9 3.93 0.11 9 6.36 0.25 9 10.21 0.29 9 16.74 0.43 9 23.43 0.47 9 15.16 0.12 9 482.26 9.93 ‘marsaeus’ 1.41 0.01 0.68 0.02 0.99 0.03 1.52 0.03 2.22 0.05 3.57 0.10 8.42 0.09 mothone Melinaea 0.91 0.02 9 4.22 0.12 9 6.34 0.14 9 9.50 0.26 9 13.68 0.46 9 22.52 0.47 9 14.51 0.15 9 410.45 11.83 marsaeus 1.38 0.03 0.69 0.02 1.05 0.02 1.41 0.02 2.03 0.04 3.60 0.08 8.08 0.07 phasiana

olgclJunlo h ina Society Linnean the of Journal Zoological Melinaea 0.92 0.02 9 3.74 0.089 9 5.98 0.11 9 9.06 0.26 9 13.93 0.50 9 19.85 0.73 9 14.35 0.22 9 418.79 10.00 marsaeus 1.33 0.04 0.63 0.02 0.97 0.02 1.45 0.03 2.22 0.03 3.44 0.10 8.21 0.12 rileyi Melinaea 0.93 0.04 9 4.66 0.13 9 6.45 0.41 9 10.54 0.44 9 15.50 0.23 9 24.05 0.93 9 14.68 0.21 9 444.41 12.25 ‘satevis’ 1.31 0.02 0.67 0.02 1.00 0.04 1.45 0.05 2.01 0.04 3.65 0.08 8.23 0.14 tarapotensis Average 0.91 0.01 9 3.79 0.07 9 6.19 0.10 9 9.82 0.14 9 15.48 0.24 9 22.05 0.24 9 14.72 0.07 9 437.49 4.23 1.36 0.01 0.66 0.01 0.98 0.01 1.44 0.01 2.13 0.02 3.55 0.03 8.22 0.04 2016 , DIFFERENTIATION IN MELINAEA BUTTERFLIES 9

600 b a b b

500

400

300

200

100

0 Mel. satevis Mel. menophilus Mel. menophilus Mel. 'marsaeus' Mel. marsaeus Mel. marsaeus Mel. 'satevis' cydon hicetas ssp. nov. 1 mothone phasiana rileyi tarapotensis

Figure 2. Pupal weight (mg) of seven Melinaea taxa (mean SE; N = 30 individuals/taxon). Different lower-case letters indicate significant differences. The dashed line indicates the average pupal weight for all species combined. for Melinaea ludovica and Mel. lilis (Fig. 3). Most spe- with Mel. isocomma simulator from Peru and cies and subspecies represented by several individuals Mel. lilis messatis from Panama. Finally, both popula- are paraphyletic or polyphyletic, with the exception of tions of Mel. ludovica ludovica from Peru and from Mel. ‘satevis’ tarapotensis. However, this taxon is only French Guiana also cluster together. represented by three individuals, and although it is monophyletic, it is very similar to the other Melinaea DISCUSSION taxa (less than 1% divergence; Fig. 3). With the microsatellite data set, STRUCTURE Mimicry is a good example of a trait under strong analyses also detected a relatively low level of struc- ecological selection that can also be used as a mating turing (Fig. 4). The best-supported number of popula- cue. Synchrony between mate choice and disruptive tions in the STRUCTURE analysis was K = 2 (Delta K selection is an important feature of speciation theory peak = 29.8), with the first cluster consisting of because it can trigger rapid speciation (see Jiggins Mel. menophilus and Mel. ‘marsaeus’ mothone from et al., 2001 and references therein). The genus Meli- Peru and Ecuador; a second cluster consisting of naea is involved in ‘tiger’ mimicry complexes with Mel. ludovica, Mel. mneme, Mel. ‘menophilus’ media- other ithomiines and heliconiines in the Neotropics. trix, Mel. isocomma, Mel. lilis, and Mel. ‘satevis’ tara- The taxonomic difficulties of the Melinaea group potensis; and Mel. satevis and Mel. marsaeus stem from the variability of their colour patterns and consisting of an admixture of the two. The second best the lack of differentiation in structural morphological model was found to be K = 6 (Delta K peak = 3.9), and characters such as genitalia and legs (Fox, 1960). It further separated the different species, albeit with lots has therefore been necessary to depend on pattern of admixture. The analyses show that the populations morphology, despite the potential shortcomings of of Mel. ‘marsaeus’ mothone from Peru and Ecuador doing so in a group engaged in Mullerian€ mimicry. cluster together, but separately from the other Meli- Current taxonomy recognizes 12 species, although naea species, including Mel. marsaeus. The second we found little to no genetic differences between well-supported cluster consists of the Mel. menophilus some species and subspecies using mitochondrial from Peru (Mel. menophilus ssp. nov. 1 and markers, confirming previous work on a subset of Mel. menophilus hicetas) and from Ecuador our target species (Whinnett et al., 2005; Dasmahap- (Mel. menophilus menophilus and Mel. menophilus atra et al., 2010). In this study, additional types of zaneka). However, Mel. ‘menophilus’ mediatrix from data enabled us to reassess species boundaries to French Guiana does not cluster with the other some extent, and to shed light on processes driving Mel. menophilus and instead clusters with speciation in the genus Melinaea. Mel. mneme mauensis from French Guiana. Melinaea The close relationship amongst the different Meli- marsaeus phasiana and Mel. marsaeus rileyi also naea species is supported by the similar characteris- cluster together, although all individuals present a tics of the immature stages, and by their similarities high level of admixture. Melinaea satevis cydon from in ecology and host-plant use. In this study, develop- Peru and Mel. satevis maeonis from Ecuador also form mental rates were similar for all species, taking a cluster. However, Mel. ‘satevis’ tarapotensis does not approximately 26 days from the egg being laid to but- cluster with the other Mel. satevis, but rather clusters terfly emergence under natural conditions. The larvae

of all species in this study can use J. parasitica as a host plant, including Mel. ‘marsaeus’ mothone, whose distribution at higher elevations co-occurs with that of the host plant T. speciosa. This study is the first 0.13 0.12 0.17 0.17 0.14 0.12 0.12 0.12 record of the genus Trianaea also serving as a host plant of a Melinaea species, and although no eggs or larvae other than that of Mel. ‘marsaeus’ mothone Mean developmental time (days) were ever recorded on this host in the wild, it is still unclear whether this is a case of differentiation driven 0.07 26.11 0.17 26.06 0.11 25.36 0.12 26.39 0.19 26.15 0.18 26.03 0.10 26.06 0.18 26.33 by a shift in host plants, and/or because of a change in habitat. As changes in host-plant use have been tied to species diversification (Janz et al., 2006; Nylin et al., 2014), this finding warrants further study. Partial geographical isolation of both 0.15 8.87 0.15 8.80 0.10 8.60 0.20 9.00 0.16 8.50 0.15 8.83 0.06 8.84 0.15 9.27 Mel. marsaeus (phasiana and rileyi) and 15 individuals/ssp.) Mel. menophilus (hicetas and ssp. nov. 1) is sug- =

N gested by the observed partition and uneven abundance at the different localities. Each species has a similar distribution, with one subspecies occupying transitional forest, and the other lowland forest. 0.09 4.07 0.13 3.84 0.16 3.89 0.18 4.14 0.11 4.05 0.06 4.13 0.13 4.20 0.11 4.07 Joron et al. (1999) described a similar regional bio- geographical trend, with Mel. marsaeus rileyi and Mel. menophilus hicetas found only in the Lower Huallaga (i.e. lowland forest). Both species also have a transitional zone near Pongo, a known suture and 0.11 2.85 0.07 2.53 0.20 2.80 0.15 2.90 0.10 2.61 0.05 2.72 0.09 2.63 0.16 2.71 hybrid zone (Whinnett et al., 2005; Dasmahapatra et al., 2010). In those areas where Mel. marsaeus

spp. from north-eastern Peru ( phasiana and Mel. marsaeus rileyi meet, putative phenotypic hybrids, albeit rare, have been collected. Those wild-caught male hybrids can be backcrossed

Melinaea with the parental strains, producing offspring of 0.11 2.21 0.11 2.17 0.20 2.20 0.12 2.30 0.11 2.26 0.04 2.21 0.07 2.13 0.13 2.16 either phenotype, as well as various intermediates. Wild-caught females, having already been mated, also produce offspring of either phenotype, and/or intermediates. Females of Mel. menophilus caught in the wild, however, often produce broods consisting 0.10 2.01 0.09 2.02 0.07 2.18 0.06 2.20 0.07 2.26 0.04 2.12 0.09 2.07 0.09 2.13 of both Mel. menophilus ssp. nov. 1 and Mel. menophilus hicetas, and Joron et al. (1999) also noted that these two forms frequently intergrade with one another in the ﬁeld. Indeed, the two subspecies readily mate, apparently showing no assorta- 0.13 2.25 0.12 2.13 0.06 2.20 0.18 2.01 0.12 2.07 0.06 2.13 0.13 2.12 0.19 2.13 tive mating. These crosses produce offspring of

SE) of each larval instar for seven either phenotype in captivity, with the hicetas phe- 3.87 4.07 4.12 4.10 3.91 4.03

notype appearing to be at least partly dominant. Per- haps unsurprisingly, nuclear markers are unable to separate the subspecies of either species. However, because Mel. menophilus appears to form a panmic- tic population rather than two separate allopatric ssp. nov. 1 3.93 mothone

’ populations with assortative mating, Mel. menophi- tarapotensis

’ lus ssp. nov. 1 and Mel. m. hicetas should probably be considered as different colour morphs of the same species, rather than as true subspecies. satevis marsaeus Duration in days (mean ‘ ‘ Present systematists consider Mel. marsaeus as one highly polytypic species (Brown, 1977), whereas earlier work (Fox, 1965) recognized the melanic Melinaea marsaeus rileyi Melinaea speciesMelinaea satevis cydon Egg First instar Second instar Third instar Fourth instar Fifth instar Pupa Melinaea Table 5. Melinaea menophilus hicetas Mean duration (days) 4.01 Melinaea menophilus Melinaea Melinaea marsaeus phasiana Andean foothill taxa, mothone and messenina,as

Figure 3. Maximum likelihood tree for 14 Melinaea taxa, with bootstrap values shown at nodes. The scale bar repre- sents the amount of nucleotide substitutions.

Figure 4. STRUCTURE plot based on 12 polymorphic microsatellite loci for 15 different Melinaea taxa. Bar colours represent posterior probabilities of assignment to inferred genotypic group, where the number of groups (K) was allowed to vary from 2 to 7. Bottom letters correspond to Melinaea taxa as follows: A, Melinaea menophilus ssp. nov. 1 and B, Melinaea menophilus hicetas from Peru; C, Melinaea menophilus menophilus and D, Melinaea menophilus zaneka from Ecuador; E, Melinaea ‘marsaeus’ mothone from Peru and F, from Ecuador; G, Melinaea satevis cydon from Peru; H, Melinaea satevis maeonis from Ecuador; I, Melinaea marsaeus phasiana and J, Melinaea marsaeus rileyi from Peru; K, Melinaea ‘satevis’ tarapotensis from Peru; L, Melinaea lilis messatis from Panama; M, Melinaea isocomma simulator from Peru; N, Melinaea ‘menophilus’ mediatrix from French Guiana; O, Melinaea mneme mauensis from French Guiana; P, Melinaea ludovica ludovica from French Guiana and Q, from Peru. belonging to a distinct species: Mel. mothone mothone The present genetic data contradict the taxonomic and Mel. mothone messenina. Here we show strong status of two other species: Mel. ‘satevis’ tarapotensis differentiation in habitat and host plant preferences from Peru and Mel. ‘menophilus’ mediatrix from and reproductive isolation between Mel. mothone and French Guiana. Although originally considered to be all other Melinaea species, including the part of the Mel. menophilus complex, Fox (1960) con- Mel. marsaeus subspecies. Furthermore, genetic sidered Mel. ‘satevis’ tarapotensis to be part of structure indicates that Mel. marsaeus is composed Mel. maenius as a result of differences in the colour of two distinct taxa, Mel. marsaeus and Mel. moth- pattern, but the taxon was later placed in the one. This is in fact not very surprising, as both Fox Mel. ethra complex by Brown (1977). Melinaea sate- (1965) and Brown (1977) found that taxonomic char- vis cydon and Mel. ‘satevis’ tarapotensis do not co- acters suggested that Mel. mothone was closest to occur in sympatry and no hybrid zone has ever been Mel. menophilus, but chose to keep them taxonomi- documented. There were no ecological differences cally separate (Fox as a separate species, Mel. moth- between these two subspecies, although indeed there one, and Brown as part of Mel. marsaeus) because of were no differences between any of the species minor differences in the male genitalia (Fox, 1965) except for between Mel. mothone mothone and the and because these two species are widely sympatric. remaining species. Larval colour pattern was similar However, in light of the population genetic structure for Mel. satevis cydon and Mel. ‘satevis’ tarapotensis, results, which separate them into two distinct taxa, and differed from the other Melinaea of Peru, resem- and in addition to important ecological differences bling more closely that of Mel. mneme mauensis. and reproductive isolation, Mel. ‘marsaeus’ mothone However, the genetic structure recognizes the two should be considered as a separate species and there- taxa as being fully separate, and indeed, there fore we here revise its systematic status to appears to be reproductive isolation, as no mating Mel. mothone mothone. This species also includes was ever observed. Nuclear markers seem to place Mel. mothone messenina (previously considered as Mel. ‘satevis’ tarapotensis as closely related to Mel. ‘marsaeus’ messenina). Mel. lilis messatis and Mel. isocomma simulator,

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 DIFFERENTIATION IN MELINAEA BUTTERFLIES 13 although this may be an artefact of our low sample may be causing the death of males before they hatch, size for these taxa. Furthermore, no additional eco- and which in turn may cause a sex-ratio distortion logical data were obtained in the present study for in the population (e.g. Jiggins et al., 1998; Charlat these two species. Certainly the taxonomic status of et al., 2003; Engelstadter€ & Hurst, 2009). Charlat Mel. ‘satevis’ tarapotensis should be revised, although et al. (2003) suggested that Wolbachia alteration of which species should contain it remains uncertain. host reproduction has the potential to accelerate Unfortunately, an in-depth ecological study of the divergence, and thus reproductive isolation, between Melinaea species present in French Guiana was populations of different infection status. For exam- beyond the scope of this study, and although ple, Wolbachia has been shown to induce cytoplasmic Mel. ‘menophilus’ mediatrix appears to be genetically incompatibility, which is caused when sperm from distinct from the Mel. menophilus complex, it is not infected males cannot produce viable offspring with possible to confirm ecological differentiation or repro- eggs of females that are not infected by the same ductive isolation. The results from the STRUCTURE strain, and it may play a role in insect speciation by analysis show that Mel. ‘menophilus’ mediatrix clus- generating reproductive isolation in this manner ters with Mel. mneme mauensis. Melinaea ‘menophi- (Rokas, 2000). lus’ mediatrix was originally considered as a form of Finally, all Melinaea taxa were observed feeding Mel. mneme by Fox (1960), but the two species were on bird droppings and pollen, both from natural and later separated by de Less (1970) based on differ- artificial sources. This is not the first account of ences in the ventral hindwing and in chromosome ithomiines feeding on bird droppings (e.g. Ray & numbers. Brown (1977) later tentatively considered Andrews, 1980), but the use of pollen has not been Mel. ‘menophilus’ mediatrix as a north-eastern iso- shown previously in this taxon, unlike, for example, late of the widespread, but now invalid, species in Heliconius, for which pollen is a major source of Mel. maenius. amino acids (e.g. Gilbert, 1972). Although various Mitochondrial sequences of all subspecies of species of butterflies have been observed to carry pol- Mel. menophilus, Mel. marsaeus, Mel. satevis, and len after visiting flowers, pollen-gathering by Helico- Mel. mothone are virtually identical, and the nius differs from involuntary collecting (see Penz & microsatellite STRUCTURE analysis shows a large Krenn, 2000 and references therein). In this group, amount of shared variation amongst all Melinaea spe- pollen is accumulated in the basal third of the pro- cies. This could be because of recent and rapid radia- boscis and saliva is exuded and mixed with the pol- tion, with a large amount of unsorted polymorphism, len through uncoiling and recoiling of the proboscis. or extensive gene flow amongst sympatric lineages, This was not observed in Melinaea, and so pollen col- such as those found in Tarapoto. Extensive gene flow lecting from natural sources, such as Lantana flow- amongst sympatric species has been reported in other ers, may have been accidental. However, Melinaea mimetic butterflies, particularly in the genus Helico- butterflies also collected pollen from artificial nius (Martin et al., 2013), and such gene flow is even a sources, which consisted of bee pollen in a suspended probable cause of mimicry in some cases, where con- Eppendorf tube. However, whether these butterflies vergence in warning signals is attributed to adaptive are able to obtain the amino acids in pollen remains introgression (Heliconius Genome Consortium, 2012). to be tested. In the case of Melinaea, interspecific mating pairs In conclusion, in light of ecological data, reproduc- sometimes occur, both in the wild and in captivity, but tive isolation, and data from nuclear markers, the according to our observations, such mating pairs are status of Mel. ‘marsaeus’ mothone is revised to rare and no eggs or offspring were ever observed. This Mel. mothone mothone. Data from nuclear markers suggests that the species are probably well differenti- show that Mel. ‘satevis’ tarapotensis is separate from ated and that both pre- and postzygotic barriers are Mel. satevis, and reproductive isolation is also pre- in place. By contrast, sympatric species that have sent, suggesting that they may indeed be separate strongly divergent mitochondrial sequences, such as species altogether, although the taxonomic status of Mel. ‘menophilus’ mediatrix and Mel. mneme in Mel. ‘satevis’ tarapotensis remains unclear. Clearly, French Guiana or, to a lesser extent, Mel. ludovica further work is needed to clarify the identity and and Mel. ‘satevis’ tarapotensis in Peru, share a sub- relationship of the different phenotypes and sub- stantial amount of allelic variation at nuclear species of Melinaea. Future work should also evalu- microsatellite loci. Gene flow might therefore occur in ate the role of T. speciosa as a novel host plant, sympatry amongst at least some well-differentiated which may have resulted in the speciation of species, but the extent of such gene flow warrants fur- Mel. mothone mothone, the role of Wolbachia infec- ther studies with more individuals and more markers. tion as a promoter of speciation in Melinaea, and the There also appears to be some infection by the roles of colour pattern and chemical cues such as maternally inherited bacterium Wolbachia, which pheromones in pre-mating reproductive isolation.

Finally, further molecular studies are needed to eval- coalescent-based test for vicariant geographic divergence uate the extent of gene flow amongst sympatric spe- and speciation. Molecular Ecology 19: 4283–4301. cies. This will undoubtedly shed further light on the Drummond BA III, Brown K. 1987. Ithomiinae (Lepi- diversification of the genus Melinaea in particular, doptera: Nymphalidae): summary of known larval food and on patterns of speciation in general. plants. Annals of the Missouri Botanical Garden 74: 341–358. Earl D, vonHoldt B. 2012. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output ACKNOWLEDGMENTS and implementing the Evanno method. Conservation Genet- – This research was funded by le Fonds Quebecois de ics Resources 4: 359 361. Elias M, Hill R, Willmott K, Dasmahapatra K, Brower A, la Recherche sur la Nature et les Technologies Mallet J, Jiggins C. 2007. Limited performance of DNA (FQRNT) as a Postdoctoral Fellowship award to barcoding in a diverse community of tropical butterflies. Pro- M. M., and by a Centre National de la Recherche – ceedings of the Royal Society of London B 274: 2881 2889. Scientifique (CNRS) Actions Thematiques et Incita- Engelstadter€ J, Hurst G. 2009. The ecology and evolution tives sur Programmes (ATIP) grant and Agence of microbes that manipulate host reproduction. Annual Nationale de la Recherche (ANR) grant (SPECREP) Review of Ecology, Evolution, and Systematics 40: 127–149. awarded to M. E. We are grateful to Mathieu Joron, Evanno G, Regnaut S, Goudet J. 2005. Detecting the Barbara Huber, and Nicolas Chazot for access to number of clusters of individuals using the software specimens. We thank Dr Sanda Knapp (Natural His- STRUCTURE: a simulation study. Molecular Ecology 14: tory Museum, UK) for identification of botanical sam- 2611–2620. ples. We also thank the Peruvian authorities and Dr Fox RM. 1960. A monograph of the Ithomiidae (Lepi- Gerardo Lamas (Museo de Historia Natural, Univer- doptera), Part II. The tribe Melinaeini Clark. Transactions sidad Mayor de San Marcos) for research permits of the American Entomological Society 86: 109–171. (236-2012-AG-DGFFS-DGEFFS and 201-2013-MINA- Fox RM. 1965. Additional notes on Melinaea Hubner€ (Lepi- GRI-DGFFS/DGEFFS), Mathieu Chouteau and Ron- doptera: Ithomiidae). Proceedings of the Royal Entomologi- ald Mori-Pezo for their precious help in the field, and cal Society of London B 34: 77–82. Andre Freitas for commenting on earlier versions of Gilbert L. 1972. Pollen feeding and reproductive biology of this manuscript. Molecular work was carried out at Heliconius butterflies. Proceedings of the National Academy the Service de Systematique Moleculaire du Museum of Sciences of the United States of America 69: 1403–1407. National d’Histoire Naturelle (CNRS-UMR 2700). Heliconius Genome Consortium. 2012. Butterfly genome reveals promiscuous exchange of mimicry adaptations among species. Nature 487: 2–6. REFERENCES Hill R, Elias M, Dasmahapatra K, Jiggins C, Koong V, Willmott K, Mallet J. 2012. Ecologically relevant cryptic Beccaloni G. 1997. Ecology, natural history and behaviour species in the highly polymorphic Amazonian butterfly of ithomiine butterflies and their mimics in Ecuador (Lepi- Mechanitis mazaeus sensu lato (Lepidoptera: Nymphalidae: doptera: Nymphalidae: Ithomiinae). Tropical Lepidoptera 8: Ithomiini). Biological Journal of the Linnean Society 106: 103–124. 540–560. Boppre M. 1986. Insects pharmacophagously utilizing defen- Janz N, Nylin S, Wahlberg N. 2006. Diversity begets diver- sive plant chemicals (pyrrolizidine alkaloids). Naturwis- sity: host expansions and the diversification of plant-feeding senschaften 73: 17–26. insects. BMC Evolutionary Biology 6: 4. Brown KJ. 1977. Geographical patterns of evolution in Jiggins F, Hurst G, Majerus M. 1998. Sex ratio distortion Neotropical Lepidoptera: differentiation of the species of in Acraea encedon (Lepidoptera: Nymphalidae) is caused by Melinaea and Mechanitis (Nymphalidae, Ithomiinae). Sys- a male-killing bacterium. Heredity 81: 87–91. tematic Entomology 2: 161–197. Jiggins C, Naisbit R, Coe R, Mallet J. 2001. Reproductive Brown KJ. 1984. Adult-obtained pyrrolizidine alkaloids isolation caused by colour pattern mimicry. Nature 411: defend ithomiine butterflies against a spider predator. Nat- 302–305. ure 309: 707–709. Joron M, Wynn I, Lamas G, Mallet J. 1999. Variable Brown KJ, Benson W. 1974. Adaptive polymorphism asso- selection and the coexistence of multiple mimetic forms of ciated with multiple Mullerian€ mimicry in Heliconius the butterfly Heliconius numata. Evolutionary Ecology 13: numata (Lepid. Nymph.). Biotropica 6: 205–228. 721–754. Brown KJ, Freitas A. 1994. Juvenile stages of Ithomiinae: Knapp S, Persson V, Blackmore S. 1997. A phylogenetic overview and systematics (Lepidoptera: Nymphalidae). conspectus of the tribe Juanulloeae (Solanaceae). Annals of Tropical Lepidoptera 5: 9–20. the Missouri Botanical Garden 84: 67–89. Charlat S, Hurst G, Mercot H. 2003. Evolutionary conse- Lanfear R, Calcott B, Ho S, Guindown S. 2012. Parti- quences of Wolbachia infections. Trends in Genetics 19: tionFinder: combined selection of partitioning schemes and 217–223. substitution models for phylogenetic analyses. Molecular Dasmahapatra K, Lamas G, Simpson F, Mallet J. 2010. Biology and Evolution 29: 1695–1701. The anatomy of a “suture zone” in Amazonian butterflies: a

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 DIFFERENTIATION IN MELINAEA BUTTERFLIES 15 de Less H. 1970. Formules chromosomiques de quelques Peterson S, Johnson N, LeGuyader J. 1987. Defensive Lepidopt eres Rhopaloceres de Guyane. Annales de la Societ e regurgitation of allelochemicals derived from host cyano- Entomologique de France (Nouvelle Serie) 6: 849–855. genesis by eastern tent caterpillars. Ecology 68: 1268–1272. Martin S, Dasmahapatra K, Nadeau N, Salazar P, Wal- Pritchard J, Stevens M, Donnelly P. 2000. Inference of ters J, Simpson F, Blaxter M, Manica A, Mallet J, Jig- population structure using multilocus genotype data. Genet- gins C. 2013. Genome-wide evidence for speciation with ics 155: 945–959. gene flow in Heliconius butterflies. Genome Research 23: Ray T, Andrews C. 1980. Ant butterflies: butterflies that 1817–1828. follow army ants to feed on antbird droppings. Science 210: McClure M, Despland E. 2011. Defensive responses by a 1147–1148. social caterpillar are tailored to different predators and Rokas A. 2000. Wolbachia as a speciation agent. Trends in change with larval instar and group size. Naturwis- Ecology and Evolution 15: 44–45. senschaften 98: 425–434. Schulz S, Beccaloni G, Brown K, Boppre M, Freitas A, McClure M, Chouteau M, Bernard A, Elias M. 2014. The Ockenfels P, Trigo J. 2004. Semiochemicals derived from development and characterization of polymorphic pyrrolizidine alkaloids in male ithomiine butterflies (Lepi- microsatellite loci for the genus Melinaea (Nymphalidae, doptera: Nymphalidae: Ithomiinae). Biochemical Systemat- Ithomiini). Conservation Genetics Resources 6: 891–893. ics and Ecology 32: 699–713. Merrill R, Wallbank R, Bull V, Salazar P, Mallet J, Ste- Trigo J. 2011. Effects of pyrrolizidine alkaloids through dif- vens M, Jiggins C. 2012. Disruptive ecological selection ferent trophic levels. Phytochemistry Reviews 10: 83–98. on a mating cue. Proceedings of the Royal Society of London Whinnett A, Zimmermann M, Willmott K, Herrera N, B 279: 4907–4913. Mallarino R, Simpson F, Joron M, Lamas G, Mallet J. Nguyen L, Schmidt H, von Haeseler A, Minh B. 2015. 2005. Strikingly variable divergence times inferred across IQ-TREE: a fast and effective stochastic algorithm for esti- an Amazonian butterfly ‘suture zone’. Proceedings of the mating maximum-likelihood phylogenies. Molecular Biology Royal Society of London B 272: 2525–2533. and Evolution 32: 268–274. Willmott K, Freitas A. 2006. Higher-level phylogeny of the Nylin S, Slove J, Niklas J. 2014. Host plant utilization, host Ithomiinae (Lepidoptera: Nymphalidae): classification, pat- range oscillations and diversification in Nymphalid butter- terns of larval hostplant colonisation and diversification. flies: a phylogenetic investigation. Evolution 68: 105–124. Cladistics 22: 297–368. Penz C, Krenn H. 2000. Behavioral adaptations to pollen- Willmott K, Mallet J. 2004. Correlations between adult feeding in Heliconius butterflies (Nymphalidae, Heliconi- mimicry and larval host plants in ithomiine butterflies. Pro- inae): an experiment using Lantana flowers. Journal of ceedings of the Royal Society of London B, 271(Suppl.):266– Insect Behavior 13: 865–880. 269.

SUPPORTING INFORMATION Additional supporting information may be found online in the supporting information tab for this article: Figure S1. A, female Melinaea menophilus ssp. nov. 1 laying an egg on a leaf of Juanulloa parasitica;B, J. parasitica with flowers; C, Trianaea speciosa with flowers in its habitat; D, close-up of a T. speciosa flower; E, Juanulloa ochracea Cuatr from Ecuador, with flowers; F, mating pair of Melinaea menophilus hicetas and Mel. menophilus ssp. nov. 1; G, mating pair of Melinaea marsaeus phasiana and Mel. menophilus ssp. nov. 1; H, mating pair of Melinaea satevis cydon and Mel. menophilus hicetas; I, mating pair of Melinaea ‘marsaeus’ mothone; J, mating pair of Mel. ‘marsaeus’ mothone and Mel. satevis cydon;K,L,Melinaea marsaeus phasiana/rileyi phenotypic hybrid; M, Melinaea butterflies visiting flowers of Eupatorieae. Figure S2. Life stages of Melinaea mneme mauensis in French Guiana. A, newly emerged butterfly; B, C, female laying eggs on leaves of Markea formicarum; D, E, host plant Ma. formicarum in its natural habitat (ant-garden); F, single egg; G, first instar; H, second instar; I, third instar; J, fourth instar; K, fifth instar; L, pre-pupae; M, pupae; N, pupae a few hours before butterfly emergence, showing wing coloration. Figure S3. A, Melinaea satevis cydon;B,Melinaea menophilus hicetas and Melinaea menophilus ssp. nov. 1; C, Melinaea ‘marsaeus’ mothone;D,Melinaea marsaeus phasiana;E,Melinaea marsaeus rileyi;F,Melinaea ‘satevis’ tarapotensis. 1, butterfly; 2, egg; 3, first-instar larva; 4, second-instar larva; 5, third-instar larva. Figure S4. A, Melinaea satevis cydon;B,Melinaea menophilus hicetas and Melinaea menophilus ssp. nov. 1; C, Melinaea ‘marsaeus’ mothone;D,Melinaea marsaeus phasiana;E,Melinaea marsaeus rileyi;F,Melinaea ‘satevis’ tarapotensis;G,Melinaea butterfly feeding on bee pollen from a feeder in the cages; H, unidentified parasitic wasp collected from Melinaea eggs; I, unhatched egg containing a dead embryo; J, dead embryo from dissected unhatched egg; K, Melinaea pupae showing wing colour patterns shortly before eclosion. 1, butterfly; 2, fourth-instar larva; 3, fifth-instar larva; 4, prepupae; 5, pupae.