Taxonomic Rearrangement of the Erebia tyndarus Species Group (Lepidoptera, Nymphalidae, Satyrinae) Based on an Analysis of COI Barcodes, Morphology, and Geographic Distribution

Despite numerous attempts to reveal the phylogenetic position and taxonomic status of formally described entities, a large number of unresolved taxonomic problems still persist in the E. tyndarus group, mostly due to incomplete species and population sampling, especially in the eastern part of the group’s distribution. Here, we provide a COI barcode study based on essentially improved sampling of the taxa and populations from the Caucasus, including for the first time, data on one of the key taxa in this complex, Erebia iranica, described from the Demavend volcano in Iran. We also analyze the structure of valve in male genitalia and the geographic distribution of the taxa. Our analysis does not confirm the close relatedness and conspecifity of the taxa known in current literature as “E. iranica iranica” (North Iran) and “E. iranica sheljuzhkoi” (Great Caucasus). Instead, the obtained data indicates the need for the taxonomic reorganization of the E. iranica complex and its division into two species: monotypic E. sheljuzhkoi (Great Caucasus) and polytypic E. iranica with subspecies E. iranica iranica (North Iran), E. iranica dromulus (Turkey, Ararat Mt.), E. iranica transcaucasica (Lesser Caucasus), and E. iranica graucasica (Great Caucasus). In addition, our data do not support the recently proposed splitting of E. callias and E. cassioides into multiple allopatric species.

The group of ringlet butterflies similar to Erebia tyndarus (Esper, 1781) can be subdivided into four complexes of closely related cryptic species.
(1) The Euro-Asia Minor complex (= the E. ottomana complex) includes the single species E. ottomana Herrich-Schäffer, 1847 distributed in the mountains of south Europa and in Turkey (HESSELBARTH et al. E. calcarius Lorkoviae, 1953 andE. arvernensis Oberthür, 1908. (3) The taxa of the E. iranica Grum-Grshimailo, 1895 species complex are distributed in the Russian, Georgian, Armenian, and Azerbaijani parts of the Caucasus, and in Turkey and North Iran (HESSELBARTH et al. 1995).
(4) The taxa of the Erebia callias Edwards, 1871 species complex are distributed in Siberia and North America (DE LESSE 1960).
Species delimitation and identification in the Erebia tyndarus group is difficult. The morphology of male genitalia is not uniform within this group; however, the individual variations are very strong and are often comparable with the interspecific differences (e.g. see: DE LESSE 1960). The differences in wing pattern are very subtle or nearly lacking between many species (WARREN 1936, DE LESSE 1960. In spite of morphological similarity, the taxonomic and identification problems within this group can be solved if chromosomal (LORKOVIAE 1949;DE LESSE 1960;ROBINSON 1971;LUKHTANOV 1987) or molecular markers (MARTIN et al. 2002;ALBRE et al. 2008;LUKHTANOV et al. 2009;PEÑA et al. 2015;GRATTON et al. 2016;SCHMITT et al. 2016;NAKATANI et al. 2018) are applied.
An unusual diversity of karyotypes is the most remarkable characteristic of this group. Haploid chromosome numbers (n) range from n=8 in E. calcarius (LORKOVIAE 1949;ROBINSON 1971) to n=51-52 in E. iranica (DE LESSE 1960;LORKOVIAE 1972;LUKHTANOV 1987). These differences provide reliable characters for species delimitation, description, and identification (DE LESSE 1960). However, the parapatric E. tyndarus and E. cassioides (n=10), and the sympatric taxa of the E. iranica species complex (n=51-52) share the same chromosome numbers (ROBINSON 1971) and can be identified on the basis of molecular markers (ALBRE et al. 2008;LUKHTANOV et al. 2009;GRATTON et al. 2016) and subtle differences in morphology (WARREN 1936). The species status of E. tyndarus and E. cassioides (n=10) was also supported by hybridization experiments (LORKOVIAE 1958).
Here we provide an analysis of this group based on new COI barcodes, morphology, and geographic distribution.

Material and Methods
38 specimens (E. iranica iranica, E. iranica transcaucasica, and E. sheljuzhkoi) were processed at the Canadian Centre for DNA Barcoding (CCDB, Biodiversity Institute of Ontario, University of Guelph) using standard high-throughput protocol described in DEWAARD et al. (2008) and resulted in a 658 bp fragment of COI. The BOLD accession numbers of these specimens and their geographic data are presented in Fig. 1. The sequences, pictures, and collection data of these specimens are deposited and can be freely downloaded at the BOLD Public Data Portal (http://www.boldsystems.org/index.php/databases). Erebia iranica transcaucasica was found to be represented by two haplotypes in Turkey (Ispir): h1 (27 specimens, ## LOWAB234-09, LOWAB243-09 to LOWAB254-09, LOWAB256-09 to LOWAB260-09, LOWAB262-09 to LOWAB271-09 and LOWAB293-09) and h2 (2 specimens, ## LOWAB246-09 and LOWAB269-09). The set of voucher specimens of the butterflies is kept at the Zoological Institute of the Russian Academy of Science (St. Petersburg) and at the McGuire Center for Lepidoptera and Biodiversity (University of Florida).
We excluded the GenBank samples LC340508 (Russia, North Caucasus, Dombai) and LC340477 (Armenia, Aragats) from the analysis because of their extremely strong deviation in the nucleotide composition not supported by other sequences from the same localities.

Results and Discussion
Despite numerous attempts to reveal the phylogenetic position and taxonomic status of formally described entities, a large number of unresolved taxonomic problems still persist in the E. tyndarus group, mostly because of incomplete species and population sampling, especially from the eastern part of the group's distribution (MARTIN et al. 2002;PEÑA et al. 2015;GRATTON et al. 2016;SCHMITT et al. 2016). Here we provide a COI-barcode analysis based on essentially improved sampling of the taxa and populations from the Caucasus, including, for the first time, data on one of the key taxa in this complex, Erebia iranica, described from the Demavend volcano in Iran.
The analysis revealed E. ottomana as the most basal lineage (clade) within the group (Fig. 1). Then, the Caucasian (II) and the European-Siberian-American (III) clades were revealed to be highly supported monophyletic groups (Fig. 1).
The Caucasian clade has been shown to include two lineages: the lineage of E. sheljuzhkoi distributed in the Great Caucasus only, and the lineage of E. iranica sensu lato distributed in the Great and Lesser Caucasus as well as in Turkey and North Iran. Thus, these two lineages are sympatric in the Great Caucasus and separated by a significant barcode gap (Table 1) which correlates with a gap in morphology (Fig. 3) and, therefore, represent two different biological species.
Our analysis does not confirm the close relatedness and conspecificity of the taxa known in current literature as "E. iranica iranica" (North Iran) and "E. iranica sheljuzhkoi" (ALBRE et al. 2008, LUKHTANOV et al. 2009). Instead, it demonstrates that these taxa are quite distant with respect to their DNA barcodes. Their conspecificity has been claimed on the basis of identity in chromosome number (n=51-52) (LORKOVIAE 1972;LUKHTANOV 1987). However, molecular studies demonstrate that this character (n=51-52) has a plesiomorphic nature, and, thus, is not evidence for creating any taxonomic or phylogenetic conclusions.
Differentiation in the valve shape between allopatric forms of the E. iranica complex (Figs 3 and 4) is correlated with a relatively low (1.6-2.2%) barcoding gap between them (Table 1). In accordance with the criteria formulated in our work (allopatry + barcoding gap less 3% + no evidence for reproductive isolation, LUKHTANOV et al. 2016) they should be classified as a subspecies, not as a different species.
In general, the analyzed data indicates the need for taxonomic reorganization of the E. iranica complex and its division into two species: monotypic E. sheljuzhkoi ( The European-Siberian-American clade has been shown to include two assemblages: the lineage of E. callas distributed in Siberia and North America (Colorado) (Fig. 1) and the lineage represented by the taxa distributed in the mountain systems of Southern Europe (the Iberian-Alpine-Balkan lineage) (Figs 1 and 2).
The Siberian-American lineage includes the single species E. callias. It has been split recently into several allopatric species such as E. callias sensu stricto, E. sibirica Staudinger, 1881, E. chastilovi Churkin, 2003, andE. przhevalskii Yakovlev, 2012 on the basis of differences in male genitalia (RUBIN & YAKOVLEV 2013). Indeed, the structure of male genitalia is not identical in different populations within this lineage (Fig. 5). The population from the Saur-Tarbagatai mountain system (E. callias sibirica) is especially different, in that it  has a wider shape of valve in male genitalia. Similarly, the wide valve is found in some populations in Mongolia (E. callias przhevalskii). The almost complete absence of differentiation in DNA barcodes (Table 1), along with the presence of transitional forms in male genitalia (Fig. 5), seems to support the subspecies rather than species status of the studied taxa. However, the number of genital preparations and DNA barcodes studied is insufficient to draw definitive taxonomic conclusions in this case. Further research based on more intensive sampling of specimens and multiple molecular markers may shed light on the status and phylogenetic relationships of the Siberian and Mongolian taxa.
The Iberian-Alpine-Balkan lineage consists of the populations that are weakly differentiated with respect to DNA barcodes (Fig. 2), except for the clearly distinct E. hispania and E. rondoui. However, chromosomal analysis, hybridization experiments, and thorough analysis of the distribution in zones of sympatry and parapatry (LORKOVIAE 1958) demonstrated that the rest of the populations represented four different biological species, E. tyndarus, E. cassioides, E. nivalis, and E. calcarius. Based on differences in allozymes, SCHMITT et al (2016), split E. cassioides into three allopatric species; E. cassioides sensu stricto, E. neleus, and E. arvernensis. Although there is a certain logic to this action (SCHMITT et al. 2016) based on the use of the phylogenetic species concept, it contradicts the logic of LORKOVIAE (1958) and DE LESSE (1960), who divided the complex into species based on the data of reproductive isolation, i.e. based on the biological species concept.
Thus, acceptance of the changes proposed by SCHMITT et al. (2016) makes the system of the Erebia tyndarus group eclectic, partly based on the phylogenetic species concept and partly based on the biological species concept. In this situation, adhering to the biological species concept, we consider it more reasonable to use the traditional system (LORKOVIAE 1958) and do not support the splitting of E. cassioides into three species.

Taxonomic conclusion
We propose the following taxonomic arrangement of the E. tyndarus group (haploid chromosome numbers are in parentheses). Recently, a population of the E. tyndarus group was discovered in the Polar Urals (Russia) and was described as a new species, E. churkini Bogdanov, 2008. However, it has not been studied so far in respect to molecular markers and, therefore, has not been included in the species list below.