Before processing butterflies were put in glassine envelopes and kept alive for less than one hour. The abdomen was removed and fixed in a fresh fixative (3:1, 96% ethanol: glacial acetic acid) for further chromosomal analysis. The butterflies were kept dry before DNA-extraction processing.

To distinguish between species in new material we used standard mitochondrial DNA barcode (657 bp length) which was obtained for 85 samples. The DNA was extracted from new butterflies as described in Vishnevskaya et al. (2016). Chromosome preparation techniques used for the new material are also described in Vishnevskaya et al. (2016). All the preparations were held in “Chromas” Core Facility, Research Park, Saint Petersburg University. Sequencing was held in Research Resource Centre for Molecular and Cell Technologies, Research Park, Saint Petersburg University. The obtained sequences were aligned and compared using BIOEDIT v. 7.2.5. Information on sequences (GenBank ID) is present in Suppl. material 1.

In our research we paid special attention to studying the wing coloration in the UV spectrum invisible to human eye. For the analysis we used wings separated from the body as well as several pinned butterfly specimens.

To analyze the wing coloration in the UV spectrum, we used methods based on the ones described in the articles by Meyer-Rochow (1991) and Pecháček et al. (2019), with modifications for specific objects and tasks of the current study: the diaphragm closed; the exposition of 15 seconds; the ISO set at 400. In order to detect UV patterns on the wings of butterflies (male and female), photographs of the wings were taken from the upperside and from the underside. As a source of ultraviolet we used UV lamps, emitting light at a wavelength of 320–380 nm (which matches with the maximum sensitivity of photoreceptors sensitive to ultraviolet radiation in the eyes of insects; Meyer-Rochow 1991). Photographs were taken with a Canon camera (EOS 1100 D) with a modified sensor. The camera was supplied with a lens that transmits UV light (Kyoei Kuribayashi, 35mm F/ 3.5), supplemented with a filter that cuts off visible light from 380 nm for shooting in UV (Baader U-Filter, 320–380 nm, Baader Planetarium, Germany). Before the analysis images were processed in PHOTOSHOP v. 23.4.1. to make the color balance equal in all images. Then the images were converted to grayscale.

Images of upper and underside of the wings in the visible spectrum were obtained using the scanner Epson Perfection 4870 Photo. The images of several pinned butterflies were taken with the Canon EOS 5D Mark IV camera, a Laowa 100mm f/2.8 2X Ultra Macro APO lens and a Canon Macro Twin Lite MT-26EX-RT flash.

We analyzed 294 specimens (Table 1) from 16 species of the monomorphic complex of the subgenus Agrodiaetus (Suppl. material 1). The taxa representing admetus clade included 232 samples, while of the dolus clade included 62 samples. All the photographs of wing coloration of the species in visible and UV light are present in Suppl. materials 2–28.

Samples of monomorphic species complex showed no difference in the upperside of the wings in the UV light. They did not reflect the UV light resulting in black color in the photographs. Therefore, in the analysis we used only photographs of UV patterns of the underside of the wings.

The underside of the Agrodiaetus butterfly wing can reflect and partially or absolutely absorb the UV light with different degree. To make it more clear what an absolute reflectance and absolute absorption is, we provide a photograph of the blue butterfly male with the blue upperside color (Polyommatus hamadanensis de Lesse, 1959) and the butterfly from the brown-species complex (P. orphicus orphicus Kolev, 2005) (Fig. 1).

The underside of the wings varied greatly between species and within species showing different variations of the pattern of UV light reflectance and absorption. We use the word “pattern” since within a single wing one can see reflecting and absorbing areas.

We distinguished 13 different types of UV patterns (Fig. 2), which we illustrate in schemes in Fig. 3. The patterns were distinguished based on the characters described in Table 2.

Different combinations of the character states described in Table 2 appear in different types of UV patterns.

Type A. Background color is almost white. The background color is uniform.

Type B. Background color is light grey. The background color is uniform.

Type C. Background color is light grey. There is a long white band along the inner margin of the forewing. The band is limited above by the second cubital vein (Cu2). The apical area is lighter than the wing area above the Cu2.

Type D. Background color is grey. There is a long white band along the inner margin of the forewing. The band is limited above by the second cubital vein (Cu2). The apical area is lighter than the wing area above the Cu2. On the whole the pattern is similar to the Type C, but differs in the color of the background.

Type E. Background color is grey. The eyespots with white contouring are vivid. The submarginal marking on both wings and the white streak on the hindwing are vivid.

Type F. Background color is light grey or grey. The submarginal area is darker then the background of the forewing.

Type G. The costal area is lighter than the forewing area limited by R4+5, M3 and the Discal cell (or space 6, space 5, space 4 according to the scheme of the wing provided by Eckweiler and Bozano 2016, Lycaenidae part IV). The white band along the inner margin can be wide and extend above the Cu2.

Type H. Background color is grey. The background color is uniform.

Type I. Background color is dark grey (almost black). A white band along the inner margin of the forewing is short (it ends in the middle of the wing). The band is limited above by the Cu2.

Type J. Background color is dark grey (almost black). A white band along the inner margin of the forewing is long (it ends by the margin of the wing). The band is limited above by the Cu2.

Type K. Background color is dark grey (almost black). A white band along the inner margin of the forewing is wide: it extends above the Cu2.

Type L. Background color is dark grey (almost black). A band along the inner margin of the forewing is grey and long (it ends by the margin of the wing). The band is limited above by the Cu2. There are white (lighter) markings within the band on the margin of the forewing.

Type M. Background color is dark grey (almost black). A band along the inner margin of the forewing is grey and short (it ends in the middle of the wing).

The UV pattern distribution within the monomorphic complex species is demonstrated in the Table 3. It shows high variability of UV pattern within the species and between the species. The numbers in the table cells show the number of butterflies with a specific ultraviolet pattern. Though the sample size differed among species, darker types of UV pattern (ultraviolet absorption) were found in a larger proportion of butterflies. For example, the lightest pattern type A was found in four species. Interestingly, type E pattern, which was present in P. pseudorjabovi Lukhtanov, Dantchenko, Vishnevskaya & Saifitdinova, 2015, P. rjabovianus Koçak, 1980 and P. valiabadi Rose & Schurian, 1977, was found only in the samples, collected in 2003 and 2014. In 2024 during the expedition to the same locality (Talysh, Azerbaijan), we collected a huge sample of P. pseudorjabovi (63 specimens), none of which had the type E pattern.

When we compared the patterns of the underside of the wings in visible light with those in UV light, we found that in many cases some elements of the visible pattern were not visible in the UV images (Fig. 4).

For example, Fig. 4 shows differences in the absence and presence of white streak, submarginal marking and even some black eye-spots on the underside of the wings between visible and UV spectra. In some cases, the discal spot was absent in UV light while in others the white streak was present in UV image as well as the submarginal marking and the eye-spots. As the presence or absence of pattern elements in the UV images appeared to be random between the species and within the species, we did not take them into account when we described the types of UV patterns.

Polyommatus admetus lineage includes two sister clades: admetus clade and dolus clade, which were revealed by the analysis of two mitochondrial markers, COI and COII (see Kandul et al. 2004). Within the admetus group (Fig. 5) all types of UV pattern were present. For example, absolutely all types were found in P. pseudorjabovi. On one hand in this study P. pseudorjabovi was represented by the largest sample. On the other hand, the samples of P. admetus and P. ripartii (Freyer, 1830) were significantly smaller, however, they still exhibited a fairly large number of UV pattern types.

Within the dolus group, the distribution of UV pattern appeared to be quite uneven (Fig. 6). No species could demonstrate all types of UV patterns. Pattern types C and D were not present at all within the species of this group. Probably a lower sample size (in comparison to admetus group) was the reason for such distribution of UV pattern in the dolus group.

To identify patterns in the distribution of UV pattern types within the monomorphic species complex, we compared their occurrence in different species habitats, in allopatric populations of a single species, and in sympatric populations of different species.

Some species samples contain both sexes. UV pattern distribution between sexes is represented in Tables 4, 5. All females demonstrated dark UV pattern types but only females of P. pseudorjabovi, also demonstrated light pattern types D and F. Excluding females from the whole sample had little effect on the overall distribution of UV pattern types (Table 5). In P. admetus, pattern M was demonstrated only by females, as well as pattern D in P. pseudorjabovi and pattern I in P. rjabovianus.

Within Armenian species populations we observed a shift in UV pattern to darker types with types A-H absent (Suppl. material 29a). In contrast, Azerbaijani species populations demonstrated all the variants of the UV pattern types (as it was mentioned above all patterns were documented for P. pseudorjabovi). Type E was found only in species P. pseudorjabovi, P. rjabovianus, and P. valiabadi (Suppl. material 29b). Balkan Agrodiaetus populations showed quite high variety of UV pattern, lacking types E and I (Suppl. material 30a). Populations of the species inhabiting Turkey did not demonstrate pattern types A-E (Suppl. material 30b). Within Iranian populations the UV pattern appeared variable, with pattern types C and I absent (Suppl. material 31a). Agrodiaetus butterflies in Russia were represented by five exemplars of P. ripartii collected in the Altai Mountains (Suppl. material 31b). These specimens demonstrated patterns H, J and K.

When we analyzed the distribution of patterns in allopatric populations of individual species we observed the following: in almost all cases, one common pattern was present in all allopatric populations of a species, while some patterns occurred only in populations inhabiting specific territories (see Suppl. materials 32–34). For example, in P. demavendi patterns J and K were common for all populations, but patterns D, F, and G were present only in Iran, and pattern L only in Armenia. In P. orphicus, common patterns included H, L, and M, while P. orphicus orphicus from Bulgaria also demonstrated patterns G and K. In P. admetus all populations demonstrated pattern J. In addition to a J-type pattern there were also types F and K in a population from Greece; a population from Turkey demonstrated F, G, I, K, and L; a population from Armenia type K, and a population from Azerbaijan demonstrated almost all UV pattern types except A, E, H, and L. Polyommatus admetus from Bulgaria and Iran displayed only the common pattern type J. In P. rjabovianus we observed three common types, E, H and M, while a population from Iran also demonstrated types A and L. Polyommatus ripartii was the only species without common pattern among its allopatric populations. However, three patterns were shared among three groups of territories: pattern J in Russia, Turkey, and Bulgaria, pattern L in Turkey, Armenia, and Greece, and pattern M in Turkey, Greece, and Bulgaria. Moreover, each of these patterns was represented by different subspecies inhabiting the respective territories.

While the sample of Agrodiaetus species on the whole did not seem to be small, being divided into sympatric populations the number of specimens in each sample fell dramatically. Probably, this could influence the results of the analysis of UV pattern distribution in sympatric species. Nonetheless we would like to comment on the facts we observed while comparing UV pattern in sympatric populations. The distribution pattern appeared to resemble the one in allopatric populations. We observed some UV patterns which were common for several species; at the same time, we observed UV patterns which were unique to one of the species. For example, in Armenian sympatric populations in Gnyshic (Suppl. material 35a) UV patterns K and L were common for several species. At the same time species P. demavendi demonstrated an extra pattern J, P. emmeli Lukhtanov & Dantchenko, 2024 an extra pattern I, and P. eriwanensis Forster, 1960 an extra pattern M. In Sevan, populations of P. emmeli and P. demavendi antonius Lukhtanov & Dantchenko, 2024 did not demonstrate any common UV pattern (Suppl. material 35b).

The sample from Azerbaijan as well as the sample from Bulgaria contained females. Within Azerbaijani syntopic populations in Mistan (Suppl. material 36a) we did not observe any common pattern for P. pseudorjabovi, P. admetus malievi Dantchenko & Lukhtanov, 2004, and P. rjabovianus, but we observed unique patterns for populations in this area: C and D for P. admetus malievi, and A, B, F, and L for P. pseudorjabovi. In Pirasorah (Suppl. material 36b) only P. pseudorjabovi demonstrated unique patterns, while in Hoveri (Suppl. material 36c) each species population had specific UV pattern. In Monidigah (Suppl. material 36d), where P. pseudorjabovi and P. rjabovianus were collected, both species demonstrated common and unique patterns. When we analyzed the distribution of pattern in females, we observed a small shift to each pattern depending on a species. For example, in Mistan, P. admetus malievi and P. pseudorjabovi shared common pattern I and J; however, in P. pseudorjabovi pattern I was found in six individuals and pattern J in one individual, and in P. admetus malievi pattern J was found in six individuals and pattern I in one individual. Moreover, we also observed unique patterns: M for P. admetus malievi and F, K, and L for P. pseudorjabovi.

In Pirasorah the sample of P. rjabovianus was represented by the only female. At the same time P. pseudorjabovi demonstrated patterns which were only represented by females in this area (Suppl. material 36b).

In Monidigah both species were represented by just one exemplar of a female, but each exemplar demonstrated different UV pattern type.

In Bulgaria females of P. ripartii pelopi Brown, 1976 were represented by just one exemplar, while P. orphicus by seven exemplars. Females of both species demonstrated UV pattern type M.

On the whole in Balkan region sympatric species demonstrated more vivid differences. Thus in Greece, Timfrist (Suppl. material 37a) P. ripartii pelopi and P. timfristos Lukhtanov, Vishnevskaya & Shapoval, 2016 shared common pattern L, at the same time each population demonstrated unique patterns. In Kalavrita, three sympatric species, P. admetus, P. ripartii pelopi, and P. aroaniensis, demonstrated only unique patterns (Suppl. material 37b). In Granitis (Suppl. material 37c) P. nephohiptamenos Brown & Coutsis, 1978 and P. orphicus eleniae Coutsis & De Prins, 2005 demonstrated a common pattern L, as well as two unique patterns each. Finally, in Bulgaria, Hvoina, P. orphicus orphicus and P. ripartii pelopi males exhibited only unique patterns (including females both species shared common pattern M) (Suppl. material 37d).