The Taxa of the Hyponephele lycaon – H. lupina Species Complex (Lepidoptera, Nymphalidae, Satyrinae): Deep DNA Barcode Divergence despite Morphological Similarity

The genus Hyponephele includes about 40 species distributed throughout the southern part of the Palaearctic area. Within this genus, the taxa of the H. lycaon – H. lupina species complex are similar with respect to the wing pattern and genitalia structure. Here we revise this group using analysis of butterfly morphology, DNA barcodes, and study of the type material. We show that, with a few exceptions, the species in this group are allopatric in distribution. Allopatry in combination with phenotypic similarity may be theoretically interpreted as evidence for the conspecifity of these taxa. Here we falsify this hypothesis by using DNA barcode analysis. We show that the species of this complex are genetically very distant and cannot be combined together as a polytypic species. We also demonstrate that H. lupina consists of two deeply diverged allopatric clades, H. lupina s.s. and H. mauritanica comb. & stat. nov. The barcode p-distance between these taxa (3.4-4.9%) is significantly higher than the generally accepted ‘standard’ minimum interspecific divergence (2.0-3.0% ) threshold. These two clades can also be distinguished by the color of the upperside of the wing in males (brown with conspicuous golden reflection in H. lupina; dark brown without golden reflection in H. mauritanica) and by details in male genitalia and male androconia structures. Syntypes of Hyponephele sifanica, H. cheena cheena, H. cheena iskander, and H. cheena kashmirica are studied and figured.

The species-rich genus Hyponephele Muschamp, 1915 comprises about 40 species distributed in the Palaearctic region, from N. Africa and Portugal to Far Eastern Russia and China, with the center of the species diversity in Central Asia (ECKWEILER & BOZANO 2011). Morphologically, this is a very homogenous group. Interspecific differences in male genitalia structure are not pronounced or absent (HIG-GINS 1975;SAMODUROW et al. 1995SAMODUROW et al. , 1996SAMODUROW et al. , 1997SAMODUROW et al. , 1999aSAMODUROW et al. , 1999bSAMODUROW et al. , 2000SAMODUROW et al. , 2001. The male wing pattern is similar in many species, and the female wing pattern can be characterized as extremely similar in many species (see the figures in ECKWEILER & BOZANO 2011). Therefore, it is not surprising that there are numerous unresolved taxonomic problems within the genus. Several partially overlapping and partially alternative taxonomic hypotheses have been suggested to describe the species and subspecies diversity in the group of species close to H. lycaon (Rottemburg, 1775) and H. lupina (Costa, 1863) (SAMODUROW et al. 1995;ECKWEILER & BOZANO 2011). (Costa, 1863) (SAMODUROW et al. 1995;ECKWEILER & BOZANO 2011).
We are not aware of any attempts to solve taxonomic problems in the genus Hyponephele using molecular data, although GenBank has scattered information on the mitochondrial DNA barcodes of some species (WAHLBERG et al. 2003;LUKHTANOV et al. 2009;DINCÃ et al. 2015;YANG & ZHANG 2015;FENG & NING 2012;LUKHTANOV & NOVIKOVA 2015;DAPPORTO et al. 2019). Although DNA barcode data alone has a limited value in taxonomy (DASMAHAPATRA et al. 2010;PAZHENKOVA & LUKHTANOV 2019), combining the morphological and DNA barcode data is an efficient method for creating new and testing old species-level taxonomic hypotheses (LUKHTANOV et al. 2016).
In this work, we attempt (1) to obtain and analyze a more complete and systematized set of DNA barcodes for representatives of the group of species close to H. lycaon and H. lupina, and then (2) to use these DNA barcodes in combination with data on morphology and geographic distribution to test the previously formulated hypotheses on taxonomic relationships in the H. lycaon -H. lupina species complex.
We pay special attention to the taxa H. maroccana, H. sifanica, H. cheena, H. lupina, and H. interposita and argue that H. mauritanica (Oberthür, 1881) comb. & stat. nov. is a distinct species (although it can also be interpreted as an extremely diverged subspecies of H. lupina). Here we do not analyze the taxonomic structure of the species H. lycaon, H. lycaonoides, and H. galtscha in depth which will be considered in a later publication.

Museum work
The Natural History Museum (London, UK) (BMNH) was visited in order to study the type material of Hyponephele sifanica, H. cheena cheena, H. cheena iskander Hemming, 1941, andH. cheena kashmirica (Moore, 1893). The collections of BMNH, of the Zoological Institute of the Russian Academy of Sciences (St. Petersburg, Russia) (ZIN) and the McGuire Center for Lepidoptera and Biodiversity (University of Florida, Gainesville, Florida, USA) (MGCL) were used for analysis of butterfly morphology and geographic distribution.

Morphological analysis
For genitalia preparation, adult abdomens were soaked in hot (90°C) 10% KOH for 3-10 min. They were then transferred to water, and the genitalia were carefully extracted and macerated under a stereomicroscope with the help of a pair of preparation needles or with the help of a needle and watchmaker's tweezers. Once cleansed of all unwanted elements, they were transferred and stored in glycerin. The cleansed genital armatures were handled, studied, and photographed while immersed in glycerin, free from the pressure they would have been subjected to if mounted and, therefore, free from the ensuing distortion. Photographs of the genitalia were taken with a Leica M205C binocular microscope equipped with a Leica DFC495 digital camera, and processed using the software Leica Application Suite, v.4.5.0.

DNA barcode analysis
Standard COI barcodes (658-bp 5' segment of mitochondrial cytochrome oxidase subunit I) were studied. COI sequences were obtained from 30 specimens representing the H. lycaon -H. lupina species complex (Table 1). Legs were sampled from these specimens, and sequence data from the DNA barcode region of the COI were obtained at the Canadian Centre for DNA Barcoding (CCDB, Biodiversity Institute of Ontario, University of Guelph) using protocols described in HAJIBABAEI et al. (2005), IVANOVA et al. (2006 and DEWAARD et al. (2008). The examined specimens were deposited at ZIN and at MGCL. Photographs of these specimens, as well as their collecting data are available in the Barcode of Life Data System (BOLD), projects Butterflies of Palearctic (BPAL) and Butterflies of Palearctic Part B (BPALB) at http://www.boldsystems.org/.
Sequences were aligned using the software BioEdit (HALL 1999) and edited manually. Phylogenetic hypotheses were inferred using the Bayesian approach (BI) as described previously (PRZYBY£OWICZ et al.
On the periphery of the distribution of the complex, the species are found in allopatry: H. lycaon in Central Europe, S. Siberia, and the Russian Far East; H. sifanica in Central China and H. cheena in N. Pakistan and N. India (Fig. 1).

Wing pattern and male androconial patch
The shape of the male androconial patch is the most important morphological trait to distinguish between the species of Hyponephele (SAMODUROV et al. 1995). Although it is variable, sometimes in an individual or at the subspecies level, it is mostly species-specific. We selected four types of androconial patches in the H. lycaon -H. lupina group: (1) the patch is long and narrow (H. lycaon type) (Fig. 3a), found in H. lycaon, H. maroccana, and H. sifanica; (2) the patch is long and broad (H. lupina type) ( Fig. 3b-d), found in H. lupina ( Fig. 3c and H. mauritanica (Fig. 3c,d); (3) the patch is very broad, trapezoidal (H. interposita type) (Fig. 3e), found in H. interposita; (4) the patch is toothed; in width, it is intermediate between those of H. lycaon and H. lupina (H. cheena type) (Fig. 3f), found in H. cheena.
The three species with the H. lycaon type of male androconial patch (Fig. 3a) are also similar in respect to their wing pattern (ECKWEILER & BOZANO 2011) (Fig. 4).
The androconial patch is similar, but not identical, in H. lupina and H. mauritanica. In H. mauritanica (Fig. 3c, d), it is slightly wider and more convex than   in H. lupina (Fig. 3b). These two taxa are also differentiated in respect to other characteristics of wing color and pattern (Fig. 5). The wing color of male H. mauritanica is dark brown without any golden reflection (Fig. 5a). The wing color of male H. lupina is brown with a conspicuous golden reflection (Fig. 5d); this reflection is especially strong in H. lupina rhamnusia (Freyer, 1845) (Fig. 5g). In males of H. mauritanica, the hindwing underside has uniform pattern and color, with no postdiscal light band (Fig. 5b), while in H. lupina this light band is usually better developed (Fig. 5e). In females of H. mauritanica, the wing upperside (Fig. 5c) is much darker than in H. lupina (Fig. 5f). With respect to the wing pattern and androconial shape, H. cheena is intermediate between H. lycaon and H. lupina (Figs 3-6).
In the H. lycaon -H. lupina species complex, Hyponephele interposita has the most derived wing pattern with a much broader androconial patch in males (Fig. 3e) and without a second postdiscal black ocellus on the forewing in females (see ECKWEILER & BOZANO 2011).

Male genitalia
The male genitalia of the H. lycaon -H. lupina complex were studied by HIGGINS (1975)   ( Fig. 7). Our data confirm that the valve of H. mauritanica is slightly wider at the base. In addition, the point of inflection of the dorsal margin of the valve from the basal part, which is directed horizontally, to the apical part, which is directed slightly upward, has a different position in these two taxa. In H. lupina, it is shifted to the middle of the valve, while in H. mauritanica, it is shifted to the base of the valve.

DNA barcode analysis
Allopatry in combination with phenotypic similarity may be theoretically interpreted as evidence for the conspecifity of the taxa H. lycaon, H. maroccana, H. sifanica, and H. cheena. In our research, we tested this hypothesis by using DNA barcode analysis. A phylogenetic analysis of this marker revealed seven strongly supported monophyletic groups within the studied species complex (Fig. 8). These groups correspond to the known taxa: H. sifanica, H. interposita, H. lupina, H. mauritanica, H. maroccana, H. lycaon, and H. cheena. The constructed tree is mostly unresolved at the level of basal clades, and the hierarchy of these discovered clades remains unclear. The exceptions are taxa H. lupina and H. mauritanica, which form a clade with support of 1. Hyponephele interposita appears to be a sister to this clade with intermediate support (0.91) while Hyponephele lycaon and H. maroccana show as sister groups, but with low support (0.7).
The calculation of the uncorrected DNA barcode p-distances showed that the species in the complex are genetically very distant and cannot be combined together as a polytypic species. In the group of allopatric taxa H. lycaon -H. maroccana -H. sifanica -H. cheena We also demonstrated that the divergence between allopatric clades, H. lupina (S. Europe, Anatolia, Levant, the Caucasus, Iran, Central Asia) and H. mauritanica (Iberian Peninsula) (Fig. 8) is deep, confirming the conclusion of HINOJOSA et al. (2018). The barcode p-distance between these taxa ranges from 3.4% (minimum divergence) to 4.9% (maximum divergence). Thus, the minimum divergence (3.4%) is higher than the generally accepted 'standard' minimum interspecific divergence (2.0-3.0%) threshold (HEBERT at al. 2013;LUKHTANOV et al. 2016).

Taxonomic conclusion
The distinct DNA barcode gap between H. lupina and H. mauritanica that we reported is correlated with a morphological hiatus (see above). Previously, we  argued that two or more allopatric clusters of individuals can be classified as different species if the presence of a distinct DNA barcode gap between them is correlated with a morphological hiatus, and if the COI genetic distance between them is deeper than the 'standard' DNA-barcode species threshold (2.0-3.0%) (LUKHTANOV et al. 2016). In accordance with these criteria, H. lupina and H. mauritanica can be considered as two distinct species (although they can be also interpreted as two extremely diverged subspecies of H. lupina).
It should be noted that, with regard to the DNA barcodes, H. lupina is very uniform throughout the entire range, with the exception of southern Italy (Sicily, Calabria). The latter population forms a wellsupported cluster on the tree. Thus, DNA barcodes do not support the subspecies H. lupina intermedia (Staudinger, 1886) from Central Asia (see SAMODU-ROW et al. 2001), but they do support H. lupina rhamnusia (described from Sicily and larger than other subspecies, with a more intense light golden reflection on the wing upperside). There are no molecular data for H. lupina cypriaca Riley, 1921, but it is morphologically distinct, with a darker wing color in both males and females and characteristic dark, dusty yellow markings on the wings of females (ECKWEILER & BOZANO 2011). Therefore, the subspecies status of H. lupina cypriaca Riley, 1921 should be preserved.
We propose the following taxonomic arrangement of the H. lycaon -H. lupina species group.