DNA barcoding and species delimitation in two species groups of Delias from South East Asia and the Sino-Himalayan zone (Insecta: Lepidoptera: Pieridae)

In this study, we test the proficiency of DNA barcoding in taxon delimitation of a sample of Delias species. With more than 250 species, Delias is the most speciose and diverse genus of butterflies. Most Delias species show striking colours on the ventral wing surfaces, likely exhibited as aposematic signals to predators. The systematic arrangement of Delias has greatly improved recently, since the genus has been the subject of some molecular phylogenetic accounts. Most of these studies have focussed on the lineages from New Guinea where the peak of species diversity occurs. Here we study the D. belladonna and the D. pasithoe species groups, mainly distributed across the Sino-Himalayan region and Southeast Asia, respectively. The wing colour pattern is often important for separating species. However, this is not the case with the D. belladonna species group where the great similarity in the wing pattern between different taxa could result from character convergence or parallelism promoted by mimicry. In these taxa, only male genitalia allow a correct species assignment. The multispecies coalescent model provided the framework for delimiting species according to a phylogenetic species concept. We applied this method in comparison to the detection of a barcode gap based on the genetic distance. Full-length DNA barcodes (658 bp, COI), were obtained and analyzed from 53 Delias specimens, representing eight species attributable to the D. belladonna species group (D. belladonna, D. berinda, D. lativitta, D. patrua, D. sanaca, D. subnubila, D. surprisa, D. wilemani) and seven species attributable to the D. pasithoe species group (D. acalis, D. benasu, D. crithoe, D. hidecoae, D. ninus, D. pasithoe, D. woodi). The results show that the DNA barcodes in the studied Delias species groups yielded satisfactory results, proved useful in disentangling cryptic species, and resulted in species delimitation that, with a few exceptions, broadly agreed with the taxonomic arrangement established through morphology. However, there was no automatic way to get a reliable inference on the species studied. Indeed, both PTP and GYMC methods based on the multispecies coalescent model (MSC), do not statistically distinguish structure associated with population isolation vs. species boundaries. They proved to be useful tools for advancing hypotheses, that still require validation of delimited species using other data sets, such as morphological, ecological, and geographical information.