Darevskia mirabilis Arribas, Ilgaz, Kumlutaş, Durmuş, Avcı & Üzüm, 2013 (Arribas, Ilgaz, Kumlutas, Durmus, Avci & Uzum, 2013)

Arribas, Oscar, Candan, Kamil, Kornilios, Panagiotis, Ayaz, Dinçer, Kumlutaş, Yusuf, Gül, Serkan, Yilmaz, Can, Caynak, Elif Yildirim & Ilgaz, Çetin, 2022, Revising the taxonomy of Darevskia valentini (Boettger, 1892) and Darevskia rudis (Bedriaga, 1886) (Squamata, Lacertidae): a Morpho-Phylogenetic integrated study in a complex Anatolian scenario, Zootaxa 5224 (1), pp. 1-68 : 25-29

Darevskia mirabilis Arribas, Ilgaz, Kumlutaş, Durmuş, Avcı & Üzüm, 2013 . stat. nov.

( Fig. 12b ). Type Locality: Ovit Pass , Kaçkar Mountains, Rize, Turkey .

Distribution: It is known from the southern parts of Rize and Trabzon, especially around Kaçkar region.

Comments: Distinctiveness of it already mentioned in other previous genetic studies ( Rato et al. 2021; Candan et al. 2021), and whose isolated presence in the Kaçkar mountains, without contact with other forms, has made its classification oscillate between rudis and cf. valentini , and that has been genetically revealed in another distant locality (Sarıkamış, Kars, Turkey), a question that will be more deeply studied.

Darevskia rudis rudis seems to be distinct from other forms that have been assigned to rudis s. lat., and that shall be considered now as nominally belonging to another taxon different from D. rudis : D. obscura stat. nov. (see below). Darevskia rudis would have as subspecies Darevskia rudis lantzicyreni ( Darevsky & Eiselt, 1967) comb. nov. ( Fig. 12d ) and D. r. bolkardaghica ( Fig. 12c ).

Darevskia bithynica , together with Darevskia b. tristis, perhaps paraphyletic and harboring more than one taxon within, or perhaps the results (paraphyly) are due to an ancient introgression that obscures its homogeneity.

Darevskia valentini (s. str.) ( Fig. 12f ), monotypical, without any of its former subspecies (latzicyreni or spitzenbergerae ) that belong to other species or are taxa on its own.

Darevskia obscura ( Lantz & Cyrén, 1936) stat. nov., including D. obscura bischoffi comb. nov. and D. obscura macromaculata comb. nov.. The latter seems to be identical in the different analyses done and could be synonymous with obscura s. str. (almost the Turkish populations). Must be mentioned that D. obscura has been postulated as a species on its own by other authors ( Gabelaia et al. 2018 – by geometric morphometrics; Tarkhnishvili et al. 2020b - by head shape morphometrics-; Gabelaia 2019).

It remains to clarify the status of the two forms of the Greater Caucasus ( D. r. chechenica , and D. r. svanetica ): independent from the others or probably closer to D. obscura , but not to the true D. rudis . This point has to be confirmed, however.

DISCUSSION

Phylogenetic reconstruction

The complex structure of the studied group, D. valentini , D. rudis , and their relatives, has been recognized from the first studies to the present ( Lantz & Cyren 1936; Darevsky & Eiselt 1967; Darevsky 1967; 1972; Darevsky & Lukina 1977; Eiselt et al. 1992; Arribas et al. 2013; Rato et al. 2021; Candan et al. 2021). This complexity has always been attractive to researchers who apply both kinds of markers, morphology and/or more recently genetics trying to solve it. Elaborated recent assessments using genetic markers point out that there are more lineages within the D. valentini / D. rudis complexes than the previously suspected ( Candan et al. 2021; Rato et al. 2021). In this study, we aimed to increase the knowledge of the status of the currently recognized genetic lineages by creating the largest datasets, including a remote subspecies not studied so far – D. v. spitzenbergerae – for the first time, to use in both morphological and molecular analyses to clarify the problem. Our phylogenetic results show the presence of several monophyletic clades that reveal themselves as different species ( Fig. 9 ). Of these distinct clades, some had been well-documented for the first time in a recently published study ( Candan et al. 2021), and the authors have accepted that D. valentini s. lat. has more genetic lineages than previously suspected, two of which have been presented there as they should have to be described and named. However, two important shortcomings that we have tried to eliminate here prevented them from their taxonomic description: the lack of morphological study for diagnoses and the absence of samples of one of the up to now two unique subspecies of D. valentini ( D. v. spitzenbergerae ), whose study was unavoidable to make taxonomic decisions.

As seen from the tree topology obtained here, genetically divergent lineages, clades A and B, were detected as monophyletic ( Fig. 9 ). The occurrence of these two highly divergent monophyletic lineages is not only confirmed by the tree topology but also the species delimitation analyses revealed both clades as different species, which is one of the most important factors that paved the way for the here proposed taxonomic revision. Network analyses based on both genetic markers also supported this distinction. In Cyt-b, all clades were placed into their unique positions and they did not share any haplotypes ( Fig. 10A ). In MC1R, which is a nuclear marker and has slower substitution rates, an agreement relatively with a more complex structure was showed. Clade A is represented by a single haplotype (Hap16), while clade B appears to have two haplotypes (Hap16 and Hap26) ( Fig. 10B ). Although these results were suggested by Candan et al. (2021), a definite conclusion could not be made due to the absence of subspecies D. v. spitzenbergerae , a problem now solved. Considering sampling data used in our phylogenetic construction, clade A consists of both D. s. spitzenbergerae from Mergan Plateau (type locality of this relevant and geographically extreme subspecies) and D. s. wernermayeri ssp. nov. from Narlıca Valley as sister taxa ( Fig. 9 ). Although the population located in Narlıca Valley (Van, Turkey) is morphologically included in D. v. lantzicyreni ( Eiselt et al. 1992) , it is genetically more closely related to D. s. spitzenbergerae than to the former. In addition to this, the populations, which were assimilated to D. v. lantzicyreni according to morphology ( Eiselt et al. 1992), represent a completely different lineage (clade B) according to genetics. Such discordant patterns called cryptic speciation are often shown in the lizards ( Ahmadzadeh et al. 2013; Kornilios et al. 2018; Karakasi et al. 2021; Arribas et al. 2022).

Another major point is the status of D. r. mirabilis (clade C). This subspecies was first described by Arribas et al. (2013) from Ovit Pass, a very isolated geographic region in Kaçkar Mountains. Its phylogenetic position is obvious here and reveals that it should be a species as different as clades A and B ( Fig. 9 ). The genetic difference of this taxon was demonstrated by two independent studies. Firstly, Rato et al. (2021) suggested that a clade, called Trabzon-Rize in their study, is genetically distinct and that it should be considered one of the four main lineages of D. rudis . Since they did not distinguish any subspecies, they could not determine that this clade belongs to D. r. mirabilis . The fact that the D. rudis specimens used in their study share the same branch with a specimen we know for certain to be D. r. mirabilis , undoubtedly proves that this clade is a new taxon and the corresponding samples of Rato et al. (2021) belong to it. Secondly, Candan et al. (2021) has also mentioned that it has isolated genetic structure and that its distribution area may be wider than expected because a datum retrieved from GenBank ( Tarkhnishvili et al. 2013), which is located around Sarıkamış (Kars, Turkey), clustered with D. r. mirabilis in the same branch.

Similar to Candan et al. (2021), one of the interesting results obtained within D. valentini / D. rudis complexes is that the specimens belonging to D. r. rudis , D. r. bolkardaghica and D. v. lantzicyreni , cluster together with overlapping. This unexpected pattern makes it difficult to engage the complexity of the group, which unables to apply the current nomenclature and difficulties understanding the main processes underlying genetic variation. Considering the genetic ( Fig. 9 ) and morphological (see results section) evidence together, the most possible scenario seems to accept that D. v. lantzicyreni is really a subspecies of D. rudis , not from D. valentini . Thus, nominal form of D. valentini is only limited to northeast Anatolia (with areas of Georgia, Armenia and Azerbaijan), while the distribution of D. rudis sensu novo, extends from the northeastern Black Sea region to the inner Anatolia and from there to the south up to the Bolkar Mountains.

Finally, the status of some former subspecies of D. rudis also inevitably needs revision. The claim that a member of this group, D. r. obscura , is different has been put forward in a previous study including phenotypic comparison ( Gabelaia et al. 2018; Tarkhnishvili et al. 2020b). The phylogenetic results strongly support these morphological findings (clade G, Fig. 9 ). Above all, D. r. obscura has a phylogenetic position quite closely related to other two former D. rudis subspecies: D. r. bischoffi and D. r. macromaculata . Considering all these results, it seems that accepting the first described form, D. saxicola obscura Lantz & Cyren 1936 , as a species: D. obscura will contribute positively to the clarification of this group.

Morphology derived structure

Considering the studied complex group it seems that there are three large groups, which obviously coincide with the current taxonomy based on morphology (we still use here the old nomenclature to refer them). The most different includes D. bithynica s. str. and D. b. tristis, which had longer heads both concerning its width, and also about their body length, but not in their pilei because other species (especially of the former rudis complex) had smaller (in size and length) but very wide heads. Similarly, the scales that cover the upper part of the crus are small and barely keeled. Also, they had comparatively longer hindlimbs (are the more climbing, based on this characteristic). Osteologically, they have very rarely any B-Type pre-autotomic vertebrae. The sternal fontanelle is frequently reduced or absent in D. b. bithynica . Postorbital and postfrontal are subequal or the postorbital is a bit smaller (different to D. valentini , clade B and D. v. spitzenbergerae ). Squamosal and postorbital overlap commonly in half of the second’s length (as also in D. r. bolkardaghica ), more usually than in other forms of the group. Darevskia b. bithynica and D. b. tristis are identical in ANOSIM. This species is also recovered by genetics. Genetics indicates the possibility that tristis is paraphyletic as presently understood.

The former valentini complex has a broad overlap among the different forms in CDA. These valentini complex samples had comparatively longer limbs, comparatively smaller heads, a greater number of scales in the crus (which in this case corresponds also to smaller scale size, and are no or almost-none keeled), and less markedly, a greater number of ventral and dorsal scales. Anal index, a bit greater (scale comparatively wider) in D. valentini than in D. bithynica or the former rudis complex. One of their supposed taxa, D. v. lantzicyreni , perhaps due to its wide dispersal and the presence of isolated populations, appears somewhat heterogeneous. Darevskia v. lantzicyreni overlaps a few with D. r. bolkardaghica (in males, and even more in females, which would be in agreement with the genetic results and the taxonomic changes proposed above). In turn, D. v. lantzicyreni has the higher dorsalia among the former valentini complex and is the closer of this complex to D. rudis s. str., which would also agree with the genetic analysis (see above) and its relation as conspecific.

Darevskia r. bolkardaghica is characterized by low lamellae (in males and females), and osteologically is characterized because not infrequently shows a weakly ossified rib associated to the third vertebra (an extremely rare character, probably atavistic, associated to small and isolated populations), and the sternal fontanelle adopt singular shapes in sand-clock, irregular cordiform or trilobate in its forepart.Also its squamosal and postorbital bones overlap commonly in half of the second’s length (as in D. bithynica ). Genetically, it is related to D. v. lantzicyreni and D. r. rudis (all three are proposed here as subspecies of D. rudis )

Darevskia v. spitzenbergerae , clade A (here treated as the nominal ssp. of D. spitzenbergerae ), and clade B (here described as a new species) (in males) and in general, as all the former valentini complex (in females) have higher ventralia counts than in D. bithynica or the former rudis complex taxa. Osteologically, this singular form ( spitzenbergerae ) has the interclavicle lateral branches inclined forwards (with only this model in typical D. s. spitzenbergerae ) and in “clade A” (coexisting with some branches backward). Postorbital and postfrontal are subequal or the postorbital is a bit greater (as in D. valentini s. str. or clade B). Nominal taxa spitzenbergerae +clade A, and clade B are recovered as two different species by genetics. Darevskia spitzenbergerae and clade A are primitive forms, among the closely related to D. rudis (sensu novo) and their subspecies (lantzicyreni and bolkardaghica, especially this latter).

Darevskia valentini s. str. seems to be a different taxon (genetics) without its formerly assigned subspecies (is nominotypical). It has Temporalia2 a bit lower than in related taxa. In D. valentini s. str. not infrequently appear some B-Type autotomic vertebrae. Postorbital and postfrontal are subequal or the postorbital is a bit greater (as in clade B or D. spitzenbergerae ).

Darevskia rudis complex is characterized by smaller dorsalia in a great part of the rudis complex (except in D. r. rudis –yet indicated in Arribas et al. 2013, that also is recovered as a different species in genetic analyses), than in the former valentini complex and D. bithynica ssp. (except in D. v. valentini that has lower scores similar to the main former rudis complex). Hindlimb relative length is also comparatively smaller when compared with the former valentini complex and even more with D. bithynica . These differences shall be considered as characteristic of D. obscura and their newly assigned subspecies, which are very few differentiated. Darevskia r. obscura and D. r. macromaculata are near the same by morphology, as suggested in Arribas et al. (2013). They are so similar in all analyses (including non-significant differences in ANOSIM) that they appear to be the same (increased pigmentation in typical Georgian macromaculata, but perhaps not in Turkish specimens, a question to be studied in future). Temporalia1 is somewhat smaller in D. r. macromaculata and D. r. obscura (M, F), and SVL (size) is greater in D. r. bischoffi (M, F).

Paradoxically, D. rudis s. str. is morphologically extreme and a differentiated form within “its” former complex, and is distinguished from the other former rudis complex taxa (now D. obscura sspp.) by its greater values of dorsalia. Also, it is basal to the group in UPGMA. Osteologically, in D. rudis s. str. postorbital and postfrontal are subequal or the first is smaller than the second (as in D. obscura and D. bithynica ).

The two extreme populations displaced towards the south of both classical complexes ( D. rudis bolkardaghica and D. valentini spitzenbergerae ) result in the ones that connect morphologically the former rudis and valentini complexes.This may be because they are the most primitive in both groups, or because of an ecoclimatic convergence in their scalation. Both live on calcareous substrates (siliceous, even volcanic in the other forms), so they have a lighter background color than other forms (darker).

Concordance of genetic and morphological results a) The classic morphological groupings/species ( rudis and valentini complexes) seem to be no longer valid, due to newly discovered “intermediate” taxa, recent speciation, and multiple past and present introgression. The situation is fairly more complex than previously expected.

b) The above-mentioned morphological characteristics of D. bithynica are valid for this species.

c) The above-mentioned morphological characteristics of D. rudis complex are valid for D. obscura (and its sspp. macromaculata and bischoffi).

d) As stated above D. rudis s. str. is distinguished from the other former rudis complex taxa (hereinafter D. obscura sspp.) by its greater values of dorsalia. Also, it is basal to the group in UPGMA.

e) Darevskia r. lantzicyreni comb. nov. and D. r. bolkardaghica are subspecies of D. rudis .

f) Darevskia spitzenbergerae is a different taxon. The fourth axis of females analysis (at the limit of significance) discriminates specially clade A (D. s. wernermayeri nov. ssp.) and in a lesser degree D. v. spitzenbergerae (that genetically cluster together and are very similar in ANOVA), characterized by higher values of lamellae, preanalia and femoralia. Inside this spitzenbergerae genetic clade, preanalia is higher in clade A (D. s. wernermayeri nov. ssp.) and strongly characterizes it (M, F) concerning near all the taxa studied here (and if the Tukey-Kramer multiple comparison test is used instead of the much stricter of Scheffe, is significantly different to all the taxa, including the nominal D. s. spitzenbergerae )

g) Darevskia mirabilis is another taxon, genetically singular, and only moderately differentiated in its morphology within the former rudis complex (morphologically seems more related to D. o. obscura or D. r. bolkardaghica –Anatolian diagonal effect? -), and longtime approached to D. valentini by its pattern, but well isolated genetically. In ANOVA, very low supraciliar granula (M, F) (especially distinctive of this taxon) and gularia (M), and higher circumanalia (M) counts are the most diagnostic characters.

h) The number (tibialia), size and keeling of the crus scales was formerly used to distinguish between the forms assigned to “ rudis ” and “ valentini ” (sensu auctt.) and is distinctive with higher counts (and smaller scales size and keeling) in the former valentini complex (and D. b. bithynica ), appearing, to the contrary, the lower counts (with big size and strong keeling) in part of the former rudis complex ( D. o. obscura , D. o. bischoffi and D. o. macromaculata). However, D. r. rudis , D. b. tristis, D. r. bolkardaghica and D. mirabilis appear in intermediate scores (but in scale size and aspect are more similar to the lower scored ones, the former rudis complex: D. obscura sspp.). This can be seen in Fig. 11 .

There is dissociation between the results based on morphology (more similar to the former taxonomy) and those based on genetics. We believe this difference is due to three factors:

a) They are taxa of very recent differentiation. Speciation in the group is very recent, and although some lineages constitute phylogenetic species and biological species (different niches in sympatry, see for example the excellent work of Tarkhnishvili et al. 2013), their genetic distances are very low and phylogenies can be obscured also by this reason.

b) In many places, they are involved in extensive hybridization and introgression phenomena. The presence of numerous hybrids (even hybrid swarms) between the different forms, intermediate animals and that carry mitochondrial haplotypes that are not related to their nuclear reality, as can be seen in Rato et al. (2021) and Candan et al. (2021).

c) There are new taxa recently described or in the process of description in this study, some of them barely known up to now, that present intermediate, less well-defined, or mosaic characteristics, and that have been forcibly wedged between previously defined groups in former studies.

For these reasons, morphology is useful for the diagnosis of the different taxa and even the study of their intraspecific variability, but not so much for phylogenetic reconstruction.

CONCLUSION

Some geographical regions in the world are very important both in terms of species diversity and the history of the emergence of this biodiversity. Anatolia and Caucasus, which are valuable geographical regions in terms of both these phenomena, host many extraordinary living creatures, from what the genus Darevskia is one of the best examples due to its complex structure, phylogenetic radiation and the parthenogenesis phenomenon. Here, we performed the most comprehensive morphological evaluation as well as a complete molecular phylogenetic evaluation (without any representative taxa excluded) of a group of Darevskia whose taxonomic positions have been discussed for many years. The molecular results showed that the so called rudis group (from Murphy et al. 2000) contains more genetic lineages than anticipated, one of them identified as new species ( D. josefschmidtleri sp. nov.). Morphological comparisons, on the other hand, contributed also to the definition of a new subspecific taxon (as D. s. wernermayeri ssp. nov.).

Based on both morphology and molecular markers, the taxonomic structuring of the studied group is proposed as follows:

Figures

FIGURE 9. Maximum Likelihood (ML) tree (left) and collapsed one for the same tree (right) are given. Numbers on branches indicate the bootstrap and posterior probability (pp) values (ML/BI). Each species delimitation result is shown, and a vertical bar represents each cluster obtained from the analysis. Red circles indicate the internal nodes of each OTUs. The new nomenclature proposed in the text is used.

FIGURE 10. Parsimony networks corresponding to Cyt-b (A) and MC1R (B) represent reconstruction of the studied group. Numbers within parentheses represent a mutational step, black circles missing haplotypes, and colored circles haplotypes. The circle area is proportional to the number of individuals. The new nomenclature proposed in the text is used.

FIGURE 11. Escalation of the calf area in the rudis and valentini complexes. Scale size (great, medium, small) and keeling development (weak keeling, medium keeling, strong keeling). The new nomenclature proposed in the text is used. D. r. lantzicyreni comb. nov. (medium scalesize, medium keeling); D.josefschmidtleri sp.nov. (small scale size, weak keeling);D. s. spitzenbergerae stat. et comb. nov. (small scale size, weak keeling); D. s. wernermayeri ssp. nov. (medium scale size, weak keeling); D. valentini (medium scale size, weak keeling); D. b. bithynica (small scale size, weak keeling); D. b. tristis (medium scale size, weak keeling); D. o. bischoffi comb. nov. (great scale size, strongly keeling); D. o. obscura stat. et comb. nov. (great scale size, strongly keeling); D. o. macromaculata comb. nov. (great scale size, strongly keeling); D. r. bolkardaghica (medium scale size, medium keeling); D. rudis (medium scale size, medium keeling); D. mirabilis stat. nov. (medium scale size, medium keeling.

FIGURE 12. a) Darevskia spitzenbergerae wernermayeri ssp. nov. (Paratype; nº 12, Male; Başeğmez Village, Çaldıran, Turkey); b) Darevskia mirabilis stat. nov. (Paratype; nº 5, Female; Ovit Pass, Kaçkar Mountains, Rize, Turkey); c) Darevskia rudis bolkardaghica (Paratype; nº 1, Male; Karagöl, Ulukışla, Niğde, Central Anatolia, Turkey); d) Darevskia rudis lantzicyreni comb. nov. (nº 23, male; Kümbet Village, Zara, Turkey); e) Darevskia josefschmidtleri sp. nov. (Paratype; nº 20, Male; Yukarınarlıca Village, Çatak, Van, Turkey); f) Darevskia valentini (nº 9, Male; Tepeler Village, Ardahan, Turkey) and temporal area of an Armenian specimen (Karvansaray, Martuni District, Armenia); g) Darevskia spitzenbergerae spitzenbergerae stat. et comb. nov. (nº 1, Male; Cilo Sat Mountains, Hakkari, Turkey)- Also, temporal area of other specimen from the same locality. The new nomenclature proposed in the text is used.