Putting Parasemia in its phylogenetic place: a molecular analysis of the subtribe Arctiina (Lepidoptera)

Despite being popular among amateur and professional lepidopterologists and posing great opportunities for evolutionary research, the phylogenetic relationships of tiger moths (Erebidae: Arctiinae) are not well resolved. Here we provide the first phylogenetic hypothesis for the subtribe Arctiina with the basic aim of clarifying the phylogenetic position of the Wood Tiger Moth Parasemia plantaginis Hübner, a model species in evolutionary ecology. We sampled 89 species in 52 genera within Arctiina s.l., 11 species of Callimorphina and two outgroup species. We sequenced up to seven nuclear genes (CAD, GAPDH, IDH, MDH, Ef1α, RpS5, Wingless) and one mitochondrial gene (COI) including the barcode region (a total of 5915 bp). Both maximum likelihood and Bayesian inference resulted in a well‐resolved phylogenetic hypothesis, consisting of four clades within Arctiina s.s. and a clade comprising spilosomine species in addition to Callimorphina and outgroups. Based on our results, we present a new classification, where we consider the Diacrisia clade, Chelis clade, Apantesis clade, Micrarctia Seitz and Arctia clade as valid genera within Arctiina s.s., whereas Rhyparia Hübner syn.n. and Rhyparioides Butler syn.n. are synonymized with Diacrisia Hübner; Neoarctia Neumoegen & Dyar syn.n., Tancrea Püngeler syn.n., Hyperborea Grum‐Grshimailo syn.n., Palearctia Ferguson syn.n., Holoarctia Ferguson syn.n., Sibirarctia Dubatolov syn.n. and Centrarctia Dubatolov syn.n. are synonymized with Chelis Rambur; Grammia Rambur syn.n., Orodemnias Wallengren syn.n., Mimarctia Neumoegen & Dyar syn.n., Notarctia Smith syn.n. and Holarctia Smith syn.n. are synonymized with Apantesis Walker; and Epicallia Hübner syn.n., Eucharia Hübner syn.n., Hyphoraia Hübner syn.n., Parasemia Hübner syn.n., Pericallia Hübner syn.n., Nemeophila Stephens syn.n., Ammobiota Wallengren syn.n., Platarctia Packard syn.n., Chionophila Guenée syn.n., Eupsychoma Grote syn.n., Gonerda Moore syn.n., Platyprepia Dyar syn.n., Preparctia Hampson syn.n., Oroncus Seitz syn.n., Acerbia Sotavalta syn.n., Pararctia Sotavalta syn.n., Borearctia Dubatolov syn.n., Sinoarctia Dubatolov syn.n. and Atlantarctia Dubatolov syn.n. are synonymized with Arctia Schrank, leading to 33 new genus‐level synonymies. Our focal species Arctia plantaginis comb.n. is placed as sister to Arctia festiva comb.n., another widespread aposematic species showing wing pattern variation. Our molecular hypothesis can be used as a basis when adding more species to the tree and tackling interesting evolutionary questions, such as the evolution of warning signalling and mimicry in tiger moths.


Introduction
Tiger moths are a highly diverse group consisting of about 11 000 species worldwide. Of these, approximately 4000 species in 113 genera belong to the subtribe Arctiina (Erebidae: Arctiinae: Arctiini: Arctiina s.l.) (see Weller et al., 2009 and references therein). Their visual appearance and diverse ecology have made them popular among amateur lepidopterists and some species are studied intensively [e.g. the Ornate Moth Utetheisa ornatrix (Linnaeus), Garden Tiger Moth Arctia caja (Linnaeus) and the Wood Tiger Moth Parasemia plantaginis (Linnaeus)], but in general our knowledge of their diversity and phylogenetic relationships is surprisingly limited (Bendib & Minet, 1998;Conner, 2009). New species are still found, perhaps because many are relatively rare, difficult to observe, or may occur in small numbers in remote places (e.g. Micrarctia kautti Saldaitis & Pekarsky, 2015). The present classification of Arctiina s.l. is based mainly on detailed studies based on morphological characters (Dubatolov & de Vos, 2010;Lafontaine & Schmidt, 2010;Fibiger et al., 2011;Vincent & Laguerre, 2014). However, these data have not been subjected to rigorous phylogenetic analyses.
Most tiger moths are chemically defended, advertise their unpalatability with spectacular warning colours and take part in several Müllerian mimicry rings (Conner, 2009). High morphological variability in Arctiinae means that it is difficult to determine unequivocal synapomorphies [shared derived characters that support monophyletic groups (clades)], which makes it challenging to trace the evolutionary relationships within the group (Schmidt, 2007;Weller et al., 2009). As mimicry is very likely to occur within Arctiinae, another phenomenon that can potentially obscure our understanding of the systematics of this group is incomplete lineage sorting. This is likely to be common in many systems, such as mimetic butterflies, resulting from rapid radiation or adaptive introgression facilitated by strong selection on adaptive loci (Kozak et al., 2015). In addition, the tendency of researchers to describe each colourful and uniquely patterned species in its own genus has led to a less informative classification, in which many tiger moth genera are species-poor, monotypic and, in some cases, probably paraphyletic (Weller et al., 2009).
Parasemia plantaginis is the only species in its nominal genus Parasemia Hübner. The species occurs in the Holarctic, forming two distinct clades, one of which corresponds to P. plantaginis ssp. caucasica (Ménétries), with both male and female moths expressing 'interrupted' forewing pattern Honma et al., 2015) and hindwing coloration varying from yellow to red (Fig. 1D). The other clade comprises all other forms of P. plantaginis with various patterns and polymorphic hindwing coloration (Fig. 1A-C;Hegna et al., 2015). The effects of variation in both larval and adult coloration of P. plantaginis on their predation risk and other fitness measures, as well as population genetics, have been intensively studied (e.g. Ojala et al., 2005Ojala et al., , 2007Lindstedt et al., 2011;Nokelainen et al., 2011;Hegna et al., 2013& Galarza et al., 2014 and the species has great potential for becoming a model system in the study of the evolution of warning coloration (Stevens & Ruxton, 2012) and colour polymorphism.
Thus, to further investigate interesting evolutionary questions in this system, such as the evolution of warning signal polymorphism or convergent evolution in mimicry rings, a well-resolved phylogeny of Arctiina is crucially needed (Simmons, 2009;Hegna et al., 2015). With a phylogenetic hypothesis available, it will be possible to determine when colour polymorphisms have evolved in the group and to study the occurrence of mimetic patterns in detail (Simmons, 2009).
The higher classification of tiger moths (Lepidoptera: Erebidae: Arctiinae) was recently studied with molecular methods by Zaspel et al. (2014), but this study had sparse sampling of the species-rich subtribe Arctiina. Zaspel et al. (2014) sampled only Arctia caja from the diverse Arctia group and did not include Parasemia. Parasemia is thought to be closely related to Arctia, with some evidence that it may, in fact, be within the genus (Fibiger et al., 2011). Schmidt's (2007) tree, with combined evidence from barcode and morphology, placed Parasemia in the same clade with Arctia, Platyprepia, Platarctia and Pararctia. With the broadest sampling of related genera so far, Dubatolov (2008) placed Parasemia closest to Hyphoraia, which consists of three species [Hyphoraia aulica (Linnaeus), H. dejeani (Godart) and H. testudinaria (Geoffroy)], and Epicallia (=Arctia) villica (Linnaeus), a monotypic genus, based on morphological characters.
In this study, we infer a molecular hypothesis of the phylogenetic relationships of species in the subtribe Arctiina, aiming to clarify the position of Parasemia within the subtribe. Based on our results, we revise the classification of Arctiina s.s. By doing this we contribute to establishing the relationships among many monotypic genera, stated by Weller et al. (2009) as the next big challenge in arctiine systematics.

Sampling
Many Palearctic Arctiina species are rare and/or occur in areas that are not easily accessible to collectors. However, with the aid of several amateur lepidopterologists and fellow scientists (see the Acknowledgements) we were able to sample many of the species in the subtribe putatively related to Parasemia. The selection of taxa was based on previous studies (Jacobson & Weller, 2002;Schmidt, 2007;Dubatolov, 2008Dubatolov, , 2009Zaspel et al., 2014) and available checklists relevant to our taxon sampling (Dubatolov & de Vos, 2010;Lafontaine & Schmidt, 2010;Fibiger et al., 2011;Vincent & Laguerre, 2014). Within the tribe Arctiini we sampled 11 species representing nine genera of the subtribe Callimorphina and 89 species representing 52 genera of the subtribe Arctiina, but excluded the mostly tropical subtribes Pericopina, Ctenuchina, Euchromiina and Phaegopterina. As outgroups we used Setina sp. (Erebidae: Arctiinae: Lithosiini) and Amata sp. (Arctiinae: Syntomini), which are closely related to Arctiini according to Zaspel et al. (2014).
Our focal study species, P. plantaginis, is placed in Arctiini: Arctiina. To our knowledge, Parasemia together with other genera putatively related to Arctia belong to Arctiina s.s., and, within that, to a lineage that has a Holarctic distribution (Weller et al., 2009). Sampling within Arctiini was thus limited to the Holarctic region, with most species having a Palaearctic distribution, although eight species occurring only in the Nearctic were also included. For species with a wide distribution range we aimed to sample at least two individuals representing different populations to avoid possible bias caused by local adaptive evolution. As we focused our sampling to Arctiina s.s. in the hope of finding the closest relatives of Parasemia, the so-called spilosomine genera and other mainly tropical lineages of Arctiini were left more sparsely sampled. However, including the sequences of Arctiina used by Zaspel et al. (2014) in our analysis broadened our coverage to tropical regions for the spilosomine genera.
We used samples that were as fresh as possible, with the oldest ones sampled successfully being up to 10 years old, stored dry, in alcohol or frozen at −20 ∘ C. For DNA extraction we used either one to two legs of adult specimen or a small piece of tissue (e.g. anal prolegs) from larvae. The barcode (COI) sequences of our samples were cross-checked in the Barcode of Life Data System (Ratnasingham & Hebert, 2007) for those species that already had a reference barcode provided. All our sampled taxa, genes and GenBank accession numbers are provided in Appendix S1.
DNA extraction was conducted using the DNeasy Blood + Tissue extraction kit (Qiagen, Hilden, Germany) both in Turku and Jyväskylä according to the manufacturer's protocols, but assisted by a robot (Kingfisher, Waltham, MA, U.S.A.) in Jyväskylä. Washing and eluting DNA in Jyväskylä was thus done using MagAttract tubes and the KingFisher robot with the programme Qiagen Blood. For polymerase chain reaction (PCR) and primer pairs we followed the laboratory protocols of Wahlberg & Wheat (2008). However, for some older samples processed in Jyväskylä, in cases where we did not obtain enough product to be visualized and purified from agarose gel during the first PCR, we did a second PCR using the first PCR product as a template with the same primers. PCR products were sent to Macrogen Europe in the Netherlands for sequencing, except for part of the barcode region (the 5 ′ half of COI) samples, which were sequenced in Jyväskylä with Big-Dye terminator v3.1, Cycle Sequencing kit (Applied Biosystems, Carlsbad, CA, U.S.A.) and run on an ABI 3130xl Genetic Analyzer (Applied Biosystems). Finally, we aligned DNA sequences manually using mega version 5.2.2 (Tamura et al., 2011) or bioedit (Hall, 1999) and stored them on the web-based voseq database software (Peña & Malm, 2012).

Phylogenetic analysis and checking for errors
To check for erroneous sequences, we performed neighbour-joining and Bayesian analyses on single-gene alignments. These analyses were compared with the combined analysis of all genes, and if the species were placed in a radically different relationship between these two, the original sequence data for the differing gene were examined, and, in cases of possible contamination or low-quality sequence, omitted from further analysis.
We performed both maximum likelihood (ML) and Bayesian inference (BI) analyses on the combined dataset of a minimum of two successfully sequenced gene regions (min. of approximately 1000 bp). The Bayesian information criterion using partition finder v. 1.1.1 (Lanfear et al., 2012) was used to determine the best-fit partitioning scheme and evolutionary model for the dataset, which was partitioned into each codon position for each gene region. For ML analysis we used raxml-hpc2 (Stamatakis, 2014) on XSEDE (Towns et al., 2014) and ran 1000 replicates of bootstrapping to calculate support for ML nodes using the Cipres science gateway (Miller et al., 2010). The BI analyses were carried out using mrbayes v3.2.3 (Ronquist et al., 2012) on the Cipres science gateway. We performed 10 million generations, with sampling every 1000 generations and four chains, one cold and three heated, in two independent runs. The parameters and models of evolution were unlinked across character partitions and the mixed evolutionary model was used. The convergence of the two runs was ascertained by visual inspection of the log-likelihoods stationary distribution, discarding the first 25% of sampled trees, as well as by checking that the final average standard deviation of split frequencies was below 0.05 and that the potential scale reduction factor for each parameter was close to 1. Resulting trees for both ML (Fig. 2) and BI analyses (Appendix S2) were visualized using figtree v.1.4.2. (Rambaut, 2014).

Results
The most optimal partitioning scheme found by partition finder had 16 partitions (out of a total of 24). Most codon positions of each gene were kept in their own partition, except for the following, which were combined: with a non-resolved branching structure. The other subclade of the monophyletic group of 'Arctia' is the 'Mediterranean Arctia', which comprises our focal study species P. plantaginis placed as sister to Eucharia (=Ammobiota/Arctia) festiva (Hufnagel) (BS = 94, BP = 1.0), next to all three Hyphoraia species, which in turn form the sister clade of Atlantarctia ungemachi (Le Cerf), Atlantarctia (=Arctia) tigrina (Villers) and Epicallia (=Arctia) villica (Linnaeus).

A molecular hypothesis of Arctiina phylogenetic relationships
We were able to sample a wide range of Arctiina species throughout their distribution ranges in the Holarctic, while aiming to find all the potential relatives of Parasemia. Our sampling is the most comprehensive to date of the subtribe Arctiina and brings many species that have been difficult to place in a phylogenetic context for the first time. The resolution of our hypothesis could well be further improved by adding samples   Zaspel et al. (2014), is beyond the scope of this study. We find strong support for a large monophyletic grouping of the spilosomine genera as separate from Arctiina s.s. Within Arctiina s.s., four well-supported clades are recovered. We find it most informative, and probably also most stable, to consider these clades to represent the generic level within the subtribe. Each clade and the implications of our results on the taxonomy of Arctiina are discussed further in the following. Formal taxonomic revision of the genera is given in Table 1.
In the broad sense, our molecular hypothesis of the evolutionary history of P. plantaginis and relatives is in concordance with earlier phylogenies by Ferguson (1985), Schmidt (2007) and Dubatolov (2008Dubatolov ( , 2009, which were based on morphological characters, as well as the COI barcode region in Schmidt (2007). Dubatolov (2008Dubatolov ( , 2009) divided the Arctiina s.s. into 'Micrarctiini' and 'Arctiini'. Dubatolov's (2009) 'Micrarctiini' comprises mostly same genera as in our Diacrisia, Chelis and Apantesis clades, but with different hypothesized phylogenetic relationships. All of Dubatolov's (2008) 'Arctiini' are placed in Arctia as delimited below. Dubatolov (2008) divided 'Arctiini' into two clades, one associated with 'northern and mountainous areas of Asia and North America' and the other with 'plains of moderate altitudes', which correspond largely to our subclades 'Northern Arctia' and 'Mediterranean Arctia', but again his tree derived from morphology has a different branching order. Interestingly, Micrarctia is placed as sister to our Arctia.

Spilosomine genera
The Spilosoma group has been considered part of Arctiina s.l. (e.g., Ferguson, 1985) or as a separate tribe or subtribe called Spilosomina (e.g. Schmidt, 2007;Vincent & Laguerre, 2014). Zaspel et al. (2014) did not find Spilosomina separate from Arctiina and discussed whether the division has been made in an attempt to categorize moths by similar appearance. In our tree with a larger sampling of Arctiina, the spilosomine genera come out as a well-supported monophyletic group corroborating the preliminary results of Schmidt (2007) -a hypothesis that is also supported by the light wing coloration shared by many species within the group. However, as the spilosomine genera are highly diverse and globally distributed, with hotspots of diversity in the tropical Asia and Africa (Ferguson, 1985), our sampling does not allow substantive interpretation of the interrelationships within the clade. We agree with Fibiger et al. (2011) that this species group needs more work and a thorough phylogenetic revision. We thus prefer to retain the spilosomine genera in the subtribe Arctiina s.l. for the time being.
Arctiina s.s.: Diacrisia, Chelis and Apantesis clades Diacrisia, Rhyparia and Rhyparioides have been suggested to be closely related in several studies (Ferguson, 1985;Koda, 1987;Dubatolov, 2009). Our analyses corroborate these studies as we also find them to form a monophyletic entity. Species in this clade differ in their adult forewing coloration and pattern from other Arctiina by their bright yellow and red hues. This group has the highest species diversity in Asia. As Diacrisia is the oldest available genus name for these, we synonymize Rhyparia syn.n. and Rhyparioides syn.n. with Diacrisia.
The second clade combines the rather large genera Chelis and Palearctia together with many smaller genera. Ferguson (1985) noted the close relationship of Neoarctia, Holoarctia, Palearctia and Hyperborea. The internal relationships of this clade are not well resolved and would benefit from adding more samples of species and genera than are included in our analysis. Due to the well-supported monophyly of this clade, all genera in the Chelis clade are here combined into Chelis.
The third clade comprises almost solely species assigned to Grammia, but also Notarctia proxima (Guérin-Méneville), Apantesis nais (Drury) and A. vittata (Fabricius). The close relationship of Grammia, Notarctia and Apantesis has previously been suggested based on morphological characters (Ferguson, 1985). Arctia [later in Grammia] obliterata Stretch was placed in its own genus Holarctia by Smith, based on its more variable morphology and wider distribution than other Grammia species. Schmidt (2009) considered the species obliterata to be related and probably basal to Grammia, a view corroborated by our analysis. Contrary to Schmidt (2009), however, we find the clade consisting of Grammia syn.n., Holarctia syn.n., Notarctia syn.n. and Apantesis monophyletic with high support, and therefore place all these genera under Apantesis (see Table 1). Synonymy of Holarctia with Apantesis and Holoarctia syn.n. with Chelis will also clarify the confusion caused by the similar orthography of these two genus names (Ferguson, 1985).

Micrarctia
Micrarctia trigona is an especially interesting case of Arctiinae tiger moths. The tribe Micrarctiini (originally established by Seitz as Micrarctiinae) was used by Dubatolov (1990Dubatolov ( , 2009 to host many superficially similar arctiine genera that could not be placed elsewhere. Later, most of these genera were moved to other (sub)tribes, leaving M. trigona the only genus and species of Micrarctiini. Recently, a second species was described in Micrarctia that is sympatric with M. trigona (Saldaitis & Pekarsky, 2015). This species, M. kautti, is nocturnal, unlike its sister species, and perhaps this is why it had remained unnoticed for so long. It would be intriguing to include M. kautti in an analysis to further elucidate the position of Micrarctia and thus potentially help to resolve the branching order of all four clades within Arctiina s.s. As the position of Micrarctia is not as strongly supported (BS = 86, BP = 0.99) as the other clades (BS = 99-100, BP = 1.0), we prefer to retain it as a valid genus until further work can ascertain its phylogenetic position.

The Arctia clade
The unusually short branching within the Arctia clade and low support values for internal nodes suggest rapid radiation. This type of quick speciation leaves little phylogenetic evidence in the nuclear genes to study the species-level branching. 'Arctia' species (excluding Micrarctia at the base of the clade) form a well-supported clade. The superfluous number of monotypic genera that also causes polyphyly of Arctia is obviously unwarranted. To render the classification more natural, and also simplify it, we combine all these species under Arctia (see Table 1). However, two well-supported subclades can be distinguished -our 'Northern Arctia' and 'Mediterranean Arctia'.
Northern Arctia and A. caja group Many Arctiina species, especially in the 'Northern Arctia' clade, are better adapted to cooler environments than most other noctuoid moths (Ferguson, 1985). Adapting to cold environments could be one mechanism behind the apparently rapid diversification that has occurred in this clade. The subclade has been divided into many monotypic genera containing some of the most rarely encountered species with almost mysterious life histories. For example, there was a gap lasting for decades between the observations of the Menetries's Tiger Moth Borearctia menetriesii in Finland and the next discovered sites are not only separated by hundreds of kilometres but are also in different habitats (Bolotov et al., 2013).
The species in this subclade are very distinctive, with their conspicuous wing patterns, bright colours and large size. The Garden Tiger Moth Arctia caja is no exception, but is in addition very variable in its patterning. Many species, such as A. intercalaris, A. martinhoneyi, A. thibetica, A. brachyptera and A. opulenta, have been split from A. caja based on appearance, but in our molecular hypothesis all these species group together with high support and very little genetic difference. However, as the molecular markers we used in this study are too conservative for inferring interrelationships between very closely related species, other markers should be used to study patterns and levels of differentiation at the species level. We consider the A. caja group to be part of the 'Northern Arctia' clade. Dubatolov (2008) arranged his 'Northern mountainous clade' to (Gonerda + Preparctia) + Sinoarctia + (Borearctia + (Pararctia + Platarctia)) + (Orontobia + (Oroncus + (Acerbia + Platyprepia))). These genera form our 'Northern Arctia' subclade, supplemented with A. caja group, A. flavia, A. rueckbeili and Pericallia matronula. There is also some evidence in our dataset (Appendix S1) indicating that Ebertarctia nordstroemi (Brandt) could belong to the 'Northern Arctia'. According to our hypothesis the Nearctic genus, Platyprepia is closer to the base and not at the tip of the subclade and Sinoarctia sieversi is nested within Preparctia. Based on the short branching, we combine all these genera under Arctia (see Table 1). By so doing, we again move away from the uninformative monotypic genera.
Some other monotypic genera, such as Leptarctia and Palerontobia, that we were not able to sample or to obtain good-quality sequences of, are likely to belong to this subclade, and including them could help to resolve the internal relationships within the subclade. However, we consider it more likely that the low resolution within this subclade results from rapid diversification rather than sparse sampling, as both morphological and molecular data have repeatedly proved indecisive within this subclade (Ferguson, 1985;Dubatolov, 2008Dubatolov, , 2009Weller et al., 2009).

Mediterranean Arctia
This is another subclade consisting of the equally showy and colourful Atlantarctia ungemachi, Arctia (=Epicallia) villica, Arctia (=Atlantarctia) tigrina, Eucharia (=Ammobiota/Arctia) festiva, Hyphoraia spp. and Parasemia. As their distribution ranges meet at the Mediterranean, we call this group 'Mediterranean Arctia'. This monophyletic group includes only a few species, and several of them are already ascribed to Arctia. We combine both this subclade and the 'Northern Arctia' subclade under Arctia (see Table 1). The species in the two subclades are also morphologically quite similar to each other, and these clades lack reliable synapomorphies.

Concluding remarks and future applications of the phylogeny of Arctiina
This study stemmed from the need to find the closest relatives of Arctia plantaginis to be able to further understand the evolutionary origins of its peculiar polymorphic warning coloration and also tiger moths in general. Arctia plantaginis has been suggested to originate in the Caucasus or south-eastern Europe based on COI, ten microsatellite loci haplotypes and species distribution modelling (Hegna et al., 2015). Hegna et al. (2015) hypothesized that, as sexually monomorphic hindwing coloration seems to be ancestral in arctiines, the Caucasian form, A. plantaginis caucasica, of which hindwing coloration varies continuously from yellow to red in both sexes, would be ancestral to all other A. plantaginis. In other populations, female hindwing coloration still varies continuously from yellow to red, but male hindwing coloration is polymorphic and the ground colour can be white, yellow or black (Fig. 1A-D). Based on our results, the closest relatives of A. plantaginis, like Arctia festiva (Fig. 1E), are indeed sexually monomorphic in their hindwing coloration, although many species continuously vary in forewing pattern. This comparison implies that the polymorphism in A. plantaginis male hindwing coloration is a more recent development.
Another obvious application of our phylogenetic hypothesis is in the study of diversification patterns of Arctiina species. Most Arctiina species are diurnal with polyphagous larvae, feeding on, amongst others, dandelion (Taraxacum spp.) and plantain (Plantago spp.), including in the Nearctic, where these plants are naturalized European species (Conner, 2009). Dubatolov (2008Dubatolov ( , 2009 suggests that Arctiina most probably originated in Asia, from where they have spread in multiple occasions to the Western Palearctic and Nearctic. It is also possible, however, that there were some refugia during glaciation periods in the Mediterranean region, which enhanced diversification.
In conclusion, we would like to encourage researchers to study below the surface of these popular, colourful and dazzling species, so as to gain information that escapes our eyes. Our work offers long-awaited clarification of the phylogenetic relationships of Arctiina, especially within Arctiina s.s. -a group of spectacular and popular moths that have been much studied, yet proven difficult to classify with traditional methods. It was beyond our scope to provide a complete systematic revision of Arctiina s.l., with a vast majority of the 4000 species occurring in the tropics, and more work needs to be done to solve the evolutionary relationships between and within clades in this highly diverse and specialized group of moths. We hope that our molecular hypothesis for Arctiina will work as a backbone, where many more tiger moth species can find their relatives. With rigorous phylogenetic hypotheses, it will be possible to tackle many interesting evolutionary questions to come.

Supporting Information
Additional Supporting Information may be found in the online version of this article under the DOI reference: 10.1111/syen.12194 successfully sequenced gene region (min. of approximately 1000 bp) were not included in the final analysis. Samples marked with an asterisk (*) in collection country are from Zaspel et al. (2014).