It is generally accepted that cases of species’ polyphyly in COI trees arising as a result of deep intraspecific divergence are negligible, and the detected cases reflect misidentifications or/and methodological errors. Here we studied the problem of species’ non-monophyly through chromosomal and molecular analysis of butterfly taxa close to Melitaea didyma (Esper, 1779) (Lepidoptera, Nymphalidae). We found absence or low interspecific chromosome number variation and presence of intraspecific variation, therefore we conclude that in this group, chromosome numbers have relatively low value as taxonomic markers. Despite low karyotype variability, the group was found to have unexpectedly high mitochondrial haplotype diversity. These haplotypes were clustered in 23 highly diverged haplogroups. Twelve of these haplogroups are associated with nine traditionally recognized and morphologically distinct species M. chitralensis Moore, 1901, M. deserticola Oberthür, 1909, M. didymoides Eversmann, 1847, M. gina Higgins, 1941, M. interrupta Colenati, 1846, M. latonigena Eversmann, 1847, M. mixta Evans, 1912, M. saxatilis Christoph, 1873 and M. sutschana Staudinger, 1892. The rest of the haplogroups (11 lineages) belong to a well-known west-palaearctic species M. didyma. The last species is particularly unusual in the haplotypes we obtained. First, it is clearly polyphyletic with respect to COI gene. Second, the differentiation in COI gene between these mostly allopatric (but in few cases sympatric) eleven lineages is extremely high (up to 7.4%), i.e. much deeper than the “standard” DNA barcode species threshold (2.7–3%). This level of divergence normally could correspond not even to different species, but to different genera. Despite this divergence, the bearers of these haplogroups were found to be morphologically indistinguishable and, most importantly, to share absolutely the same ecological niches, i.e. demonstrating the pattern which is hardly compatible with hypothesis of multiple cryptic species. Most likely such a profound irregularity in barcodes is caused by reasons other than speciation and represents an extraordinary example of intra-species barcode variability. Given the deep level of genetic differentiation between the lineages, we assume that there was a long period (up to 5.0 My) of allopatric differentiation when the lineages were separated by geographic or/and ecological barriers and evolved in late Pliocene and Pleistocene refugia of north Africa, the Iberian and Balkan Peninsulas, the Middle East and Central Asia. We discuss the refugia-within-refugia concept as a mechanism explaining the presence of additional diverged minor haplogroups within the areas of the major haplogroups. We also provide the first record of M. gina in Azerbaijan and the record of M. didyma turkestanica as a new taxon for Russia and Europe.
