Depending on which systematic order of classification is adhered to, the genus Pseudochazara comprises 27–32 species of Graylings (Gross 1978, Lukhtanov 2007, Savela 2015). It has a wide distribution in the Palaearctic region from North Africa to the Himalayas and Mongolia (Tennent 1996, Tshikolovets 2005, Yakovlev 2012). In addition to vague species delimitation, large intraspecific variation has resulted in the description of over 100 subspecific taxa (Lukhtanov 2007) in this intensively studied taxon.

The main reason for the extensive variation in phenotype can be linked with the specific ecological requirements of these butterflies. They are mostly petrophilous and limited to specific rock substrate to which they are perfectly adapted with their camouflaged underside wing pattern and cryptic coloration. Local adaptation to mimic the coloration of the rock substrate is, therefore, one of the main drivers for such large scale diversification (Lorković 1974, Weiss 1980, Hesselbarth et al. 1995, Tennent 1996, but see Anastassiu et al. 2009).

Trying to resolve the systematics of this genus and its species delimitation has been thwarted by the fact that the genitalia of many Pseudochazara species are virtually identical and their wing shape and coloration, both being partially dependant on environmental conditions (Gross 1978, Hesselbarth et al. 1995), is inconsistent. The last comprehensive taxonomic review which was published by Gross (1978) is already outdated. He recognised 24 species, among which P. obscura (Staudinger, 1878) is now considered a subspecies of P. lydia (Staudinger, 1878) (see Eckweiler and Rose 1988), P. aurantiaca (Staudinger, 1878) and P. xerxes Gross & Ebert, 1975 have been reclassified as subspecies of P. beroe (Herrich-Schäffer, 1844) (see Lukhtanov 2007), P. schahrudensis (Staudinger, 1881) is now considered conspecific with P. mamurra (Herrich-Schäffer, 1844) (see Eckweiler 2004) and P. pakistana Gross, 1978 is conspecific with either P. gilgitica (Tytler, 1926) (see Lukhtanov 2007) or P. baldiva (Moore, 1865) (see Wakeham-Dawson et al. 2007). Several members of the Pseudochazara genus from Central Asia that are currently recognised as separate species were considered subspecific taxa in the revision (e.g. P. droshica (Tytler, 1926), P. gilgitica (Tytler, 1926), P. lehana (Moore, 1878)) while P. euxina (Kuznetsov, 1909) from Crimea was entirely neglected. Two additional species were described after the revision, P. kanishka (Aussem 1980a) and P. annieae (Pagès 2007). Following Gross’ revision (1978) the shape of the androconial scales of several Pseudochazara species has proven to be constant, enabling species delimitation (Weiss 1980, Eckweiler and Rose 1989, Wakeham-Dawson and Kudrna 2000, Wakeham-Dawson et al. 2003, Wakeham-Dawson and Kudrna 2005, Wakeham-Dawson 2006, Wakeham-Dawson and Kudrna 2006, Pages 2007, Wakeham-Dawson et al. 2007).

There has been no attempt to reconstruct the phylogeny of the genus or validate species status using molecular markers. Only the taxonomic position within subtribe Satyrina and a sister relationship to Chazara has been established (Peña et al. 2011).

In order to resolve the relationship among Pseudochazara species and re-evaluate their species status, in particular of some European taxa, we employed DNA barcoding – using a standardized gene region (5’ segment of the mitochondrial gene

cytochrome c oxidase subunit I =

No insertions or deletions were observed in the mitochondrial COI gene and therefore the alignment was unambiguous. For the COI dataset 63 unique haplotypes among 108 Pseudochazara sequences were detected. 114 (17.5%) sites were variable and 95 (14.6%) were parsimony informative. The average interspecific genetic distance was 4.9%, but in the case of P. mniszechii the intraspecific diversity ranged from 0 to 6.7% with highly distinct divergent sequences of P. mniszechii tisiphone. No evident barcoding gap was observed separating intraspecific from interspecific pairwise genetic distances (Fig. 1). On the contrary, sharing of identical haplotypes was observed in the following taxa: P. graeca / P. mamurra amymone, P. mamurra mamurra / P. daghestana, and P. beroe aurantiaca / P. alpina. On the other hand, 82% of species comparisons showed high (≥2%) interspecific distances.

The calculated maximum connection for parsimony networks at the default 95% limit was 11 steps, and resulted in 9 separate networks within Pseudochazara. 6 of them contain only single species (P. atlantis, P. turkestana, P. thelephassa, P. lehana, P. kanishka, and P. anthelea), whereas the remaining 3 comprise several closely related species (Figs 2–4). Outgroups were contained in 2 distinct networks (Chazara enervata and Chazara briseis/C. heydenreichi).

The topology of the Bayesian Inference tree of all Pseudochazara samples, including the selected outgroup species (Fig. 5), confirms the monophyly of the genus. High posterior probability values support a basal position of P. atlantis, the only species of the genus present in (and confined to) North Africa. This is somewhat surprising as P. anthelea and P. thelephassa are considered to be morphologically the most distinct and separate species within the genus (Gross 1978). P. atlantis has tentatively been placed into two groups, the ‘mamurra’ species group (Brown 1976), based on androconia shape, and the ‘pelopea’ species group (Wakeham-Dawson and Dennis 2001), on account of the shape of male genitalia. P. atlantis is also distinctive according to the TCS analysis and forms a separate network. In addition, the second basal split within Pseudochazara is well supported, and, apart from some single species clades, three species groups tentatively named as the ‘pelopea’, ‘hippolyte’ and ‘mamurra’ clades received high support. We present the results for these clades separately:

Our study supports the monophyly of the genus Pseudochazara with high posterior probability values of the COI gene tree. Within the genus, however, two conflicting patterns appear with, unexpectedly, deep divergences between presumably conspecific taxa on the one hand and lack of divergence among well-defined species on the other. This is to some extent concordant with similar studies in related genera in the subfamily Satyrinae (Kodandaramaiah and Wahlberg 2009, Nazari et al. 2010, Kreuzinger et al. 2014). The basal position of P. atlantis from North-western Africa as sister group to all remaining Pseudochazara species falls into the first category. Based on distinct male genitalia morphology and wing shape/patterns P. anthelea and P. thelephassa were considered to form the basal split within the genus (Gross 1978, Aussem 1980b, Hesselbarth et al. 1995, Wakeham-Dawson and Dennis 2001). The basal position of P. atlantis is difficult to explain in terms of biogeography, as it indicates a North African origin of the genus, which has its centre of divergence much further eastwards in the Middle East (Hesselbarth et al. 1995, Tshikolovets 2011). P. atlantis is an alpine species distributed only in the Atlas Mountains of Morocco (Tennent 1996), therefore its isolation from the main distribution of the genus could possibly have preceded the last land bridge connections with Europe at the end of the Miocene (Garcia-Castellanos et al. 2009). Hence, its basal position could be an artefact of long-branch attraction (Bergsten 2005) and/or incomplete sampling of the entirely Asiatic species. Therefore, confirmation with additional genetic markers and additional sampling is required.

Another unexpected result is a deep split between P. mniszechii and P. mniszechii tisiphone, species which are very similar in wing patterns/coloration and considered conspecific in current literature (Hesselbarth et al. 1995, Kudrna et al. 2011, Tshikolovets 2011, Eckweiler 2012) and databases (Lukhtanov 2007, Savela 2015, Fauna Europaea 2016). Based on the COI gene tree P. tisiphone Brown, 1980 (stat. n.) is a separate species closely related to two local endemics from the southern part of the Balkan Peninsula, P. orestes and P. cingovskii. Actually P. tisiphone was originally described as a subspecies of P. cingovskii (Brown 1980) and its close relationship was hypothesised also by Wakeham-Dawson and Dennis (2001) based on the similarity of the male genitalia. The low level of genetic differentiation between P. tisiphone, P. orestes, and P. cingovskii indicates a relatively recent speciation, however, we are inclined towards supporting their separate species status based on constant differences in wing patterns/coloration and also their ecological specialization (Pamperis 2009, Verovnik et al. 2013).

A split between P. anthelea anthelea from Asia Minor and P. anthelea amalthea from the Balkan Peninsula has been suggested based on minor differences in male genitalia and consistent differences in female wing coloration between both taxa (Olivier 1996, Wakeham-Dawson and Dennis 2001). They are considered separate morphospecies by Kudrna et al. (2011). We can agree with separate species status as the split between the two taxa is much older compared to almost no differentiation in three morphologically and ecologically well defined species: P. tisiphone, P. orestes, and P. cingovskii. Following this reasoning, P. pelopea tekkensis from NE Iran could also be considered a distinct species, however, inclusion of more samples is needed to confirm this status.

Given the high resolution of the basal clades within the COI gene tree, the lack of differentiation between taxa within the ‘mamurra’ and ‘pelopea’ group was unexpected. In particular, species like P. geyeri and P. daghestana are among the most easily recognisable species in the genus with uniform and very distinct wing patterns/coloration. There are several possible hypotheses to explain this lack of differentiation:

– Incomplete lineage sorting: recent speciation could result in unresolved relationships among these closely related species; however, well-defined species borders in terms of constant wing pattern differentiation coupled with broad overlaps in species ranges challenges this hypothesis.

– Recent gene flow: gene flow between closely related taxa is a known phenomenon (Descimon and Mallet 2009) and masks relationships among species especially with mitochondrial DNA (Gompert et al. 2008). The species involved have broadly overlapping ranges and could sometimes be found syntopic (Aussem 1980c, Hesselbarth et al. 1995), so hybridization is possible. Actually hybridization is documented even among the most distantly related species such as P. anthelea and P. geyeri (Aussem, 1980c). Nuclear markers with higher genetic resolution (e.g. microsatellites, SNPs) would be required to study the contact zones between these taxa to confirm ongoing gene flow. It must be noted that partial exclusion is evident when two or more Pseudochazara species are syntopic, as one is always dominant, while the others appear in very low frequencies (Hesselbarth et al. 1995, Verovnik et al. 2014).

– Pseudogenes or Wolbachia infections: both are common in invertebrates, particularly in arthropods (Bensasson et al. 2011, Gerth et al. 2014, Leite 2012, Ritter et al. 2013). As the vast majority of the haplotypes in the ‘mamurra’ and ‘pelopea’ clades originate from the BOLD database it is impossible to check or correct for this potential error.

The most enigmatic taxon among the 'mamurra' group is P. mamurra amymone from northern Greece and Albania (Eckweiler 2012, Verovnik et al. 2014). Apart from the author’s original description (Brown 1976) little has been published regarding this elusive taxon for a long time. Failed attempts to locate the vaguely described type locality (Cuvelier 2010) have led to several misleading hypotheses, resulting in speculation that it may even be a rare hybrid between P. tisiphone and P. anthelea (Wakeham-Dawson and Dennis 2001, Kudrna et al. 2011). Somewhat surprisingly, the COI gene tree suggests it has a close relationship with P. graeca, another species from the southern Balkan Peninsula. These two taxa have distinct and constant wing patterns and differ in their habitat requirements, with P. mamurra amymone inhabiting steep and hot rocky gorges at lower elevations (Gascoigne-Pees et al. 2014) while P. graeca is predominantly a montane (high elevation) species endemic to Greece (Anastassiu et al. 2009). Thus, despite paraphyly of P. amymone Brown, 1976 (stat. n.) in relation to P. graeca, we believe they both represent valid species within the ‘mamurra’ group. Consequently P. kermana Eckweiler, 2004 (stat. n.), sister species to P. amymone and P. graeca combined, should also be elevated to species rank, although additional populations of P. mamurra in Iran should be examined to confirm this status. Alternatively, all the taxa within the ‘mamurra’ group, including the monophyletic P. schakuhensis, a sister species to the amymone-graeca-kermana clade, should be treated as a single very polymorphic species, a rather more destructive approach given the current taxonomy.

Although we are aware of the pitfalls of using single gene trees in the interpretation of phylogenetic patterns (Nichols 2001), we believe that strongly supported basal branching and splits between taxa, considered conspecific, represent valid insights into speciation in the Pseudochazara genus and together with distinct morphology and ecology allows species delimitation. Hence, we propose separate species status for the following taxa: P. tisiphone, P. amalthea, P. amymone, and P. kermana. This has important conservation implications, as most of these species are local endemics and therefore potentially threatened (Verovnik et al. 2014). Wider taxon sampling and inclusion of nuclear markers would undoubtedly help to a better understanding of the taxonomy of this fascinating butterfly genus.