Integrative Taxonomic Analysis Doubles Number of Species in the Central Asian Butterfly Genus Lyela (Lepidoptera, Nymphalidae, Satyrinae)

Simple Summary

The Central Asian genus Lyela has never been the subject of taxonomic revisions due to the fact that, with the exception of the widespread L. myops myops, the other described taxa are very local and extremely rare in entomological collections. In my research, I managed to collect material which represents all described species and the main populations of Lyela. Analyses of the pattern and coloration of butterfly wings, male genitalia and mitochondrial DNA barcodes show that the genus Lyela includes not three species, as previously thought, but six species.

Abstract

Lyela Swinhoe, 1908 is a small Central Asian butterfly genus, in which three species were previously recognized based on comparison of wing patterns. The present study, based on an extensive population sample across the entire range of Lyela and using integrative taxonomy methods, confirmed the monophyly of the genus and revealed the paraphyly of the most widespread species, Lyela myops sensu auct. The genus is shown to include six species, L. myops (Staudinger, 1881) (Kazakhstan, northern Kyrgyzstan, northwestern China, and southwestern Mongolia), L. tashkumirica Lukhtanov, 2024, stat. nov. (Fergana Valley in Kyrgyzstan), L. babatagi Tshikolovets, 1998, stat. nov. (southern Uzbekistan and eastern Turkmenistan), L. tekkensis (Staudinger, 1886), stat. nov. (southwestern Turkmenistan and northeastern Iran), L. macmahoni Swinhoe, 1908 (Pakistan and Afghanistan), and L. amirica Wyatt, 1961 (Afghanistan). Each of these species represents a monophyletic unity with respect to the COI gene and is separated from the other species by a distinct barcoding gap and structural differences in the male genitalia.

1. Introduction

Lyela Swinhoe, 1908 is a small Central Asian butterfly genus known to include three species, L. myops (Staudinger, 1881), L. macmahoni Swinhoe, 1908 and L. amirica Wyatt, 1961 [1]. The range of the genus extends from western Kazakhstan and northeastern Iran in the west to Baluchistan (Pakistan) in the south, the Altai Mountains and Mongolia in the east, and central Kazakhstan in the north [2,3,4,5,6,7,8,9,10,11,12,13]. The species of this genus superficially resemble ringlet butterflies of the genus Erebia Dalman, 1816. It is therefore not surprising that the first Lyela species was originally described as Erebia myops [14], i.e., as a species of the genus currently included in the subtribe Erebiina [15]. However, already in 1908, Seitz transferred this species to the genus Coenonympha Hübner, 1819 [16], which is currently considered part of the subtribe Coenonymphina [15]. Taking into account the opinion of Chapman [17], who found that Erebia myops is distant from the true Erebia, Swinhoe [18] described the genus Lyela to include the species L. macmahoni (the type-species of the genus) and L. myops. For the same reasons, and probably unaware of Swinhoe′s work, Muschamp [19] described the genus Dubierebia Muschamp, 1915 (the type species: Erebia myops), noting that this genus occupies an intermediate position between Erebia and Coenonympha. In 1990, D’Abrera formally treated Coenonympha, Lyela and Dubierebia as three distinct genera, noting that Lyela and Dubierebia should possibly be regarded as congeneric with Coenonympha for structural reasons [20].

The closeness of Lyela to Coenonympha has been confirmed by modern molecular studies [21,22,23,24,25]. However, despite the abundance of molecular works devoted to the phylogenetic analysis of the subtribe Coenonymphina, the status of Lyela (genus, subgenus or synonym) and the position of this group in the system remain a subject of debate [4,25]. Even the analysis of whole genome data did not resolve this issue [26]. In the latter work, depending on the method of phylogenetic analysis, the position of the genus Lyela on the phylogenetic tree changes: either it appears as a sister lineage to all other Holarctic Coenonymphina, or as a sister group to Coenonympha nolckeni Erschoff, 1874, with the clade (Lyela + C. nolckeni) being a sister to all other Coenonympha species, including Triphysa Zeller, 1850.

The instability of the position of Lyela on the phylogenetic reconstructions led to the formulation of two extravagant taxonomic hypotheses. According to one of them, Lyela is a synonym of Coenonympha [21]. According to the other hypothesis, the genus Lyela includes the distantly similar and clearly not very closely related species C. nolckeni [25]. Both of the latter hypotheses are not reliably supported by molecular data and contradict the more conservative and generally accepted morphologically based concepts of Coenonympha and Lyela as two distinct genera [3,4,6,7,8,9,10,11,12]. Therefore, in my research I adhere to the traditional concept of Lyela as an oligotypic Central Asian genus of the subtribe Coenonymphina.

In the late 20th century and in the 21st century, several taxa of subspecific rank were described in the genus Lyela [27,28,29]. However, the genus Lyela has never been the subject of taxonomic revisions. This may be due to the fact that, with the exception of the widespread L. myops myops, the other described taxa are very local and extremely rare in entomological collections. In my research, I managed to collect material (including suitable for obtaining DNA barcodes), which represents all described species and the main populations of Lyela. Based on this material and using integrative approach, including analyses of the pattern and coloration of butterfly wings, male genitalia and mitochondrial DNA barcodes, in this article I propose a taxonomic revision of the genus Lyela.

2. Materials and Methods

2.1. DNA Barcoding

Standard mitochondrial DNA barcodes (658 bp fragments of the cytochrome c oxidase subunit I gene) were obtained for 31 samples of the genus Lyela at the Department of Karyosystematics of the Zoological Institute of the Russian Academy of Sciences. DNA was extracted from single legs removed from dried voucher specimens. For the majority of samples, the target 658-bp fragment of COI was amplified using the primers LepF1 and LepR1 [30]. For the rest of the samples, most of which were 20 or more years old, shorter overlapping fragments were amplified using the pair primer combination LepF1-MH-MR1 (311-bp amplicon) and MH-MH1-LepR1 (407-bp amplicon) followed by their concatenation [31]. Sequences were obtained using ABI 3730XL sequencer (Applied Biosystems, Waltham, MA, USA). Sequences were edited to remove ambiguous base calls and primer sequences, and assembled using SEQUENCHER (Gene Codes, Ann Arbor, MI, USA) (https://www.genecodes.com/sequencher-features, accessed on 15 September 2025). All new sequences were submitted to GenBank (accession numbers PX391132–PX391162).

2.2. Selection of Target Group for Phylogenetic Analysis

The phylogenetic relationships between the genera and subgenera of the subtribe Coenonymphina are not well understood [26], making it difficult to choose an outgroup for the genus Lyela. Therefore, I decided to include in the phylogenetic analysis, in addition to all species of the genus Lyela, all available species of the genus Coenonympha, including the phylogenetically most distant species C. oedippus Fabricius, 1787 and C. nolckeni, as well as representatives of the genera Triphysa and Sinonympha Lee, 1974. According to [24], Oressinoma sorata Godman & Salvin, 1868 and O. typhla Doubleday, 1849 were taken as the closest outgroup for all Holarctic Coenonymphina, and Argyronympha rubianensis Grose-Smith, 1889 was taken as a distant outgroup to root the tree.

2.3. DNA Matrix and Samples

A 4816 bp DNA matrix was analyzed, which was obtained by concatenating the mitochondrial COI gene (positions 1–1475) and five nuclear genes: EF-1a (positions 1476–2700), wg (positions 2701–3100), RpS5 (positions 3101–3717), GAPDH (positions 3718–4408), and MDH (positions 4409–4816). This matrix was created according to the methodology of Talavera et al. [32], who proposes to combine a monolocus data set consisting of multiple DNA barcodes with a multilocus backbone data set consisting of a ‘guided’ selection of taxa with multigene sampling [32]. Matrices of this kind result in high phylogenetic resolution provided that the backbone data set includes at least one individual with multigene data for each genus and the DNA barcode set represents all or nearly all species-level taxa [32].

In the studied matrix, the backbone data set consisted of 30 specimens representing 24 Coenonympha species, including the most phylogenetically isolated species C. oedippus and C. nolckeni, as well as Triphysa phryne (Pallas, 1771), Lyela tekkensis (Staudinger, 1886), Sinonympha avinoffi (Schaus, 1927) (=amoena Lee, 1974, see [33] for the rationale of this synonomy), Oressinoma sorata, O. typhla, and Argyronympha rubianensis. In creating the backbone data set, the published nuclear and mitochondrial genes available for the subtribe Coenonymphina were used [21,22,23,24]. The DNA barcode dataset was represented by 39 COI fragments, 37 of which belong to the genus Lyela and two sequences belong to the species C. nolckeni.

In total, the analyzed matrix (backbone + barcode datasets) included 69 samples. Information about this matrix, including nucleotide sequences, taxon names, collection localities, field/laboratory specimen numbers, and GenBank accession numbers for the COI gene, is presented in full in Table S1 in Supplementary Materials. The GenBank accession numbers for the nuclear genes EF-1a, wg, RpS5, GAPDH, and MDH in this matrix are contained in publications [21,22,23,24].

2.4. Phylogenetic Analysis

The Bayesian analysis was performed using the program MrBayes3.2 [34]. Three schemes of the matrix partition were used: (1) analyzing the entire matrix as a single partition, (2) using partition by gene, and (3) using partition by gene for COI, EF1a, wg, RpS5, GAPDH, and MDH, and partition by codone position for COI. In all schemes, the Bayesian analysis of the concatenation COI + EF1a + wg + RpS5 + GAPDH + MDH (Table S1 in Supplementary Materials) was performed using nst = mixed, which allows MrBayes to sample all potential substitution models according to their posterior distribution. The parameter rates (gamma or invgamma) was selected separately for each partition using the program jModelTest [35]. All program settings were specified before running the program for 30,000,000 generations. The first 7500 trees (out of 30,000) were discarded prior to computing a consensus phylogeny and posterior probabilities. The consensus of the obtained trees was visualized using FigTree 1.4.4 (http://tree.bio.ed.ac.uk/software/), accessed on 15 September 202). All three partition schemes resulted in the same topology of the reconstructed consensus trees and almost identical values of posterior probabilities. Therefore, in the following section, the results are presented when the entire matrix was analyzed as a single partition.

The Maximum likelihood analysis was performed in MEGA 11 program [36] using GTR + G + I model, with 10,000 bootstrap replicates.

The COI p-distances (%) within and between the taxa were calculated using the MEGA 11 program [36].

2.5. Morphological Analysis

Morphological analysis was performed as described previously [37,38]. Briefly, to obtain preparations of male genitalia, the butterfly abdomen was placed in a hot (95 °C) 10% KOH solution for 6 min. Then, the abdomen was transferred to distilled water, and the genitalia were removed from the abdomen using a pair of dissecting needles and thin tweezers. After cleaning the genitalia from chitin residues and soft abdominal tissues, they were washed first in 45% and then in 96% ethanol and transferred to a tube with glycerol for a long-term storage. The genitalia were examined and photographed by immersing them in a container with glycerol, without pressing them with a cover glass, therefore, without distorting their native structure and.

2.6. Digital Documentation

Photographs of the genitalia were taken using a Leica M205C binocular microscope (Leica Microsystems, Wetzlar, Germany) equipped with a Leica DFC495 digital camera and processed using Leica Application Suite software version 4.5.0. The butterfly photographs were taken with a Nikon D810 digital camera (Nikon Corporation, Minato City, Tokyo, Japan) equipped with a Nikon AF-S Micro Nikkor 105 mm lens, using the ring full-spectrum lamp as a lighting source.

3. Results

3.1. Phylogenetic Analysis

The Bayesian analysis revealed Lyela as a distinct monophyletic group (Figure 1). A sister group to Lyela was not identified due to low support for relationships between the Lyela, (Coenonympha + Triphysa) and (C. nolckeni + Synonympha avinoffi) clades. Two well-supported subclades were found within Lyela. One of them (which can be called the northern subclade) includes three well-supported clusters. These three clusters correspond to three traditionally recognized taxa known as L. myops myops, L. m. tashkumirica Lukhtanov, 2024 and L. m. babatagi Tshikolovets, [1998]. Since it will be argued below in the Discussion section that they represent distinct species, on the tree (Figure 1) they are designated as L. myops, L. tashkumirica and L. babatagi. The relationships between these three clusters were unresolved. The support for the L. myops + L. tashkumirica clade is extremely low (BPP = 0.55) and in fact it can be said that the three lineages of the northern subclade form a polytomy on the phylogenetic tree.

Specimens from the southern, Tien Shan part of the range of Lyela myops myops (Kyrgyzstan: Issyk-Kul, Chu River valley, Talas Range, Kazakhstan: Karatau Mts), differ from other samples from Kazakhstan by one fixed substitution C⇔T at position 133 of the studied DNA barcoding fragment and may probably represent a separate subspecies, the description of which will become possible in the future as more material accumulates.

DNA barcodes of the previously described subspecies L. myops mangystavica Lukhtanov, 1994 from western Kazakhstan were found to be identical to DNA barcodes of L. myops myops specimens from eastern Kazakhstan, where the taxon was originally described. This confirms the synonymization of L. myops mangystavica Lukhtanov, 1994 with the nominotypical subspecies [3,10].

The second subclade (which can be called the southern subclade) includes three well-supported lineages that correspond to the taxa L. myops tekkensis, L. macmahoni and L. amirica. Thus, the traditionally recognized species L. myops is revealed as a paraphyletic group, partly falling into the northern subclade (L. myops myops + L. myops tashkumirica + L. myops babatagi) and partly falling into the southern subclade (L. myops tekkensis). Within the second subclade, L. macmahoni and L. amirica are revealed as sister species, together forming a clade sister to L. m. tekkensis.

Maximum likelihood analysis resulted in a topology similar to that revealed by the Bayesian analysis (Figure S1). Within the genus Lyela, the same supported six-species clusters and the same relationships between species were identified. Coenonympha nolckeni was found as the sister taxon to the remaining Holarctic Coenonymphina species. A weakly supported clade (Triphysa phryne + Synonympha avinoffi + C. oedippus) was identified as the sister taxon to the remaining Coenonympha species. However, the relationships between these major clades outside the genus Lyela were low in support. Therefore, phylogenetic relationships outside the genus Lyela remain unclear.

MEGA 11 revealed relatively low COI p-distances within the studied taxa (0.00–0.79%) and relatively high COI p-distances between the studied taxa (2.36–6.28%) (

Table 1. Intra-(upper diagonal) and interspecific (all other cells) DNA-barcode p-distances (%) in the genus Lyela.

	Myops	Tashkumirica	Babatagi	Tekkensis	Macmahoni	Amirica
myops	0.00–0.70
tashkumirica	2.89–3.97	0.00
babatagi	2.52–4.27	2.75–2.91	0.15–0.61
tekkensis	3.96–5.62	5.81–6.12	5.02–5.49	0.30–0.79
macmahoni	4.67–6.45	6.11–6.13	5.62–5.79	4.10–4.41	0.00
amirica	5.40–6.39	6.28–6.43	5.63–5.94	4.57–5.02	2.36–2.67	0.00–0.15

).

3.2. Comparison of Male Genitalia

Comparison of the male genitalia (Figure 2) revealed two groups of individuals. The first group includes individuals whose valvae have a simple structure, with the apex of the valva covered only by tiny microdenticles. Within the first group, individuals of L. myops myops are characterized by narrow valvae with a thinned apex. Individuals of L. myops tashkumirica have valvae that are wider in their middle part, without thickening in the distal part. Lyela myops babatagi is characterized by the presence of wide, massive valvae.

The second group includes individuals whose valvae have a complex structure, with the apex of the valva covered not only with tiny microdenticles, but also with large teeth. A probable synapomorphy of L. macmahoni and L. amirica is the expanded apical part of the valva. At the same time, the valvae of L. macmahoni and L. amirica are clearly distinguished from each other: the valvae of L. macmahoni are massive, and the valvae of L. amirica are narrow with a thinned middle and apical parts.

3.3. Wing Pattern

Butterflies of the four traditionally recognized subspecies of L. myops (Figure 3 and Figure 4A,B) are characterized by the presence of relatively large ocelli (eye spots) in the apical part of the forewings, as well as well-defined median bands and a number of small submarginal ocelli on the underside of the hindwings. These characters are part of the ground plan of the wing pattern of nymphalids [39] and are probably plesiomorphies. Lyela macmahoni (Figure 4C,D) and L. amirica (Figure 4E,F) are characterized by the reduction in several basic elements of the wing pattern. Lyela macmahoni lacks the red-brown field on the forewings and the median band on the underside of the hindwings. Lyela amirica lacks the apical ocellus on the forewing and the median band on the underside of the hindwings.

Among the commonly recognized subspecies of L. myops, L. myops tekkensis (Figure 4A,B) is the most differentiated with respect to wing pattern. It is markedly distinguished by the reduced size of the red-brown area on the forewings in both sexes. In L. myops tekkensis, this area ends near the apical ocellus. In other subspecies, this area widens toward the outer edge of the wings.

Lyela myops tashkumirica (Figure 3C,D) differs from L. myops myops (Figure 3A,B) and L. myops mangystavica in its noticeably larger size. In addition, in L. m. tashkumirica, the costal margin of the forewing is more convex, and the apex of the forewing is more rounded. In L. m. tashkumirica, the red-brown area on the forewings is not as wide and has a more diffuse border with the dark edge of the wing. The dark marginal stripe on the forewing is wider, and the underside of the hindwing is less contrasting, with a wider median band. Lyela myops tashkumirica has a noticeable external resemblance to L. myops babatagi (Figure 3E,F) due to the same large size, but differs in a more rounded apex of the forewing, bright whitish hairs on the costal margin of the upperside of the forewing in males, and the apical part of the underside of the forewing, which is dusty with whitish scales. In L. myops tashkumirica, this dusty area extends to the apical ocellus, whereas in L. myops babatagi, the apical ocellus is completely surrounded by the main ochre-brown background.

Lyela myops tashkumirica has a more contrasting underside of the hindwings with a light field between the outer edge of the median band and a row of submarginal ocelli. In L. myops tashkumirica this light area is covered with small grey dots. In L. myops babatagi this part of the wing is covered with dark strokes. In L. myops tashkumirica the outer edge of the median band on the underside of the hind wing has an uneven, irregular contour, and in L. myops babatagi this edge is wavy.

4. Discussion

4.1. Taxonomic Interpretation of the Discovered Lineages

Traditionally, the genus Lyela is considered to contain three species, L. myops, L. macmahoni and L. amirica [1]. If I consider the last two species of the above-mentioned, there is no doubt about their identity. Lyela macmahoni and L. amirica have striking differences in the wing pattern (Figure 4C–F). They live parapatrically in the vicinity of Kabul in Afghanistan [12], but they do not mix. They occupy different ecological niches. While L. macmahoni is a lower montane semi-arid species, like most other species of the genus, L. amirica is found in the upper montane belt [40]. As our study showed, each of them represents a monophyletic unity with respect to the COI gene; they are separated by a clear barcoding gap and have serious structural differences in the structure of the male genitalia. Thus, these two species have properties, which are considered essential criteria of species in virtually all species concepts [41].

The species Lyela macmahoni and L. amirica are relatively young. Taking into account previously proposed calibrations based on the mitochondrial DNA divergence in insects [42,43,44], it can be assumed that the age of their divergence is in the range of 0.65–1.5 million years. They have the lowest level of interspecific differentiation of DNA barcodes in the genus Lyela (2.36%), which is combined with the highest level of morphological differentiation. This is probably reflected by the fact that they live in parapatry, that is, in conditions where morphological differences can be enhanced due to selection against interspecific hybrids with reduced fertility, as is suggested for butterflies of the subgenus Agrodiaetus [45] and the genus Cupido [46].

As for the species L. myops in its traditional sense, when it includes all populations from Iran to Mongolia, it is characterized by a plesimorphic wing pattern type, going back to the ground plan of the nymphaloid wing pattern [39]. Our data show that it consists of four discrete clusters (myops, tashkumirica, babatagi, and tekkensis). These clusters have a level of DNA barcode differentiation exceeding that between the obvious species L. amirica and L. macmahoni. These genetic differences are correlated with differences in the structure of the male genitalia and less noticeable, but still existing, differences in wing pattern. It is also interesting that L. myops in its traditional sense is not monophyletic, and populations from western Turkmenistan and Iran (tekkensis) fall into a southern Lyela subclade together with L. amirica and L. macmahoni, and this subclade is contrasted by the northern subclade. Therefore, based on the criteria that were detailed earlier [47], these four clusters should be interpreted as allopatric species, L. myops (Staudinger, 1881), L. tashkumirica Lukhtanov, 2024, stat. nov. L. babatagi Tshikolovets, 1998, stat. nov. and L. tekkensis (Staudinger, 1886), stat. nov., and not subspecies of a single species.

4.2. Biogeographic Scenario

The conducted multilocus phylogenetic analysis, as well as previous studies [26], did not lead to a completely resolved phylogeny of the Holarctic genera and subgenera of the tribe Coenonymphina. Therefore, the sister group for the genus Lyela remains unknown, and this complicates phylogeographic reconstructions. Nevertheless, based on the topology of the obtained tree (Figure 1) and modern species ranges (Figure 5), it can be assumed that the origin and diversification of the genus occurred within Central Asia [25]. It can also be assumed that the first key event in the divergence of the genus was the division of the original range into two parts—northern (north and east of the Amu Darya River) and southern (south and west of the Amu Darya River). Based on the calibrations [42,43,44] and the identified level of differentiation of the COI gene (Table 1), this occurred 1.1–2.65 million years ago, and this dating is most likely conservative (underestimated) since it is based on the analysis of uncorrected genetic distances and does not take into account hidden nucleotide substitutions.

In the northern part of the range, the northern subclade split into three lineages. One of them is currently found in the vast semi-arid spaces of northern Central Asia (myops). The second is endemic to the semi-arid foothills of the Fergana basin (tashkumirica). The third is endemic to the semi-arid lowlands of southern Uzbekistan and eastern Turkmenistan (Kugitang Range) (babatagi). Species of this northern subclade are not found in either the highlands or the hyper-arid plains of southern Central Asia, providing effective geographic isolation between them. The fact that these lineages do not show even signs of a hierarchical genetic structure suggests that all three lineages originated from a common ancestor at approximately the same time (0.7–1.7 million years ago)

The species of the southern subclade form a clear hierarchical structure on the phylogeny. Based on this structure, it can be assumed that the lineage leading to L. macmahini and L. amirica arose from a common myops-like ancestor. This ancestor penetrated the territory of the Baluchistan Plateau and the adjacent mountains of Afghanistan, where it diverged into two species.

Thus, the genus Lyela is characterized by a significantly greater species diversity than previously assumed and includes six species, not three. Five of these species are local endemics with very limited ranges. This type of geographic distribution and endemism is interesting, but not unique. For example, it is found in another Central Asian butterfly genus Phoenicurusia Verity, 1943 (Lycaenidae) [47]. In the genus Phoenicurusia, Ph. alexandra (Püngeler, 1901) is found in southern and central Kazakhstan (analogous to L. myops), Ph. dilutior (Staudinger, 1881) is endemic to the Fergana Basin (analogous to L. tashkumirica), Ph. sogdiana (Zhdanko, 1990) is found in the lowlands of southern Uzbekistan (analogous to L. babatagi), Ph. phoenicurus (Lederer, 1870) is found in the Kopetdag in western Turkmenistan and northeastern Iran (analogous to L. tekkensis), and Ph. balucha (Howarth, 1976) is found in Pakistan and Afghanistan (analogous to L. macmahoni). Species of these two genera are members of the same local biota, and therefore most likely they have common history, resulting in similar species ranges.

5. Taxonomic Conclusions

The following taxonomic arrangement of the genus Lyela is proposed:

Genus Lyela Swinhoe, 1908 (TS: Lyela macmahoni Swinhoe, 1908)

Lineage macmahoni (=subgenus Lyela)

L. macmahoni Swinhoe, 1908 (=macmahoni shigekoae Sakai, 2015; =macmahoni rekai Sakai, 2015)

L. amirica Wyatt, 1961

L. tekkensis (Staudinger, 1886), stat. nov.

Lineage myops (=subgenus Dubierebia Muschamp, 1915; TS: Erebia myops Staudinger, 1881)

L. myops (Staudinger, 1881) (=myops mangystavica Lukhtanov, 1994)

L. tashkumirica Lukhtanov, 2024, stat. nov.

L. babatagi Tshikolovets, [1998], stat. nov.