Taxonomic revision of Phoxinus minnows (Leuciscidae) from Caucasus, with description of a new narrow-ranged endemic species

Taxonomic revision of Phoxinus from the Caucasus revealed two distinct species. One species, P. colchicus , was known from eastern drainage of Black Sea, but was recorded also in the middle reach of the Kuban (Sea of Azov basin), for the first time. The Kuban population represents a genetically unique sub-lineage of P. colchicus . Its ancestors might have colonized the Kuban system through the event of ancient river capture. Another species inhabits only the Adagum River basin in the lower Kuban and represents a new narrow-ranged endemic species – Phoxinus adagumicus sp. nov. According to mtDNA phylogeny (COI and cyt b ), P. adagumicus sp. nov. represents deeply divergent and one of the two early branched lineages of the genus Phoxinus being distant to other species (min. p -distance = 0.074) including geographical neighbors – P. chrysoprasius from Crimean Peninsula and P. colchicus from the Caucasus. The new species differs from most Phoxinus species by frequently occurring single-row pharyngeal teeth (modal formula 5–4). The narrow geographic range (ca. 55 km in length and 15–20 km in width) and high anthropogenic load on local water systems suggests the new species is under threat and needs protection.


Introduction
Minnows of the genus Phoxinus Rafinesque, 1820 are small freshwater fish in the family Leuciscidae Bonaparte, 1835, which prefer rheophilic environment and are widespread in Eurasia from northern Spain eastward to the Anadyr and Amur drainages in Russia and China.Before the implementation of genetic methods to the taxonomy, the genus Phoxinus was represented by a few species despite its wide geographic range (Kottelat 1997).Although detailed morphological studies still contribute to identification of new species in Phoxinus (Kottelat 2007;Bianco and De Bonis 2015), introduction of genetic methods to Phoxinus taxonomy has resulted in a significant increase in the numbers of newly described species, and taxonomic and geographic range revisions (Palandačić et al. 2015(Palandačić et al. , 2017(Palandačić et al. , 2020;;Vucić et al. 2018;Denys et al. 2020;Turan et al. 2023).In particular, recent genetic studies have revealed: i) deep genetic divergences within the genus, and ii) presence of numerous yet undescribed species.On the other hand, discovered patterns of genetic diversity do not always correspond to the morphology-based classification (Palandačić et al. 2017).
The Caucasus is a mountainous transcontinental region between the Black and Caspian seas.It is considered a significant Pleistocene refugium and a hotspot (i.e., the Caucasus Biodiversity Hotspot) of endemism for both plants and animals (Gadzhiev et al. 1979;Kolakovskiy 1980;

Sampling
Fishes were collected by the authors from different localities using frame and seine nets (mesh size 6-8 mm) (Fig. 1; Suppl.material 1).Fishes were euthanized in a solution of clove oil and photographed in an aquarium with artificial lighting using a Nikon D5300 camera (Nikon Corporation, Tokyo, Japan) equipped with a Nikkor 60 mm f/2.8G lens (Nikon Corporation, Tokyo, Japan).Fin clips (pectoral or pelvic) were collected from some specimens (DNA-vouchers) and fixed in 96% ethanol for subsequent DNA extraction in the laboratory.Most of the collected specimens were preserved in 10% formalin (form.),whereas some smaller specimens were preserved in 96% ethanol for molecular analysis.Subsequently, formalin-fixed specimens were washed out in running water and transferred to 70% ethanol for longterm storage.
The types (holotype, part of paratypes), additional and comparative material are deposited in the Fish collection of the Papanin Institute for Biology of Inland Waters of Russian Academy of Sciences, Borok, Russia (IBIW_ FS); the rest of the paratypes are kept in the Zoological Institute of the Russian Academy of Sciences, Saint Petersburg, Russia (ZISP) and the Zoological Museum of the Moscow University, Moscow, Russia (ZMMU).
Sex was determined by the size of the pectoral fins.External meristics were counted on the left side.The total number of the pectoral and pelvic-fin rays was counted on the left fins.Scales above lateral line were counted between lateral line and base of first unbranched ray of dorsal fin; scales below lateral line were counted between lateral line and base of first unbranched ray of anal fin.In both cases, lateral line scales were not taken into account.The number of anterior gill rakers of the first gill arch was counted on the left and right sides of the specimens.The number of the scale rows was counted on the left and right breast patches and average value was taken according to Bogutskaya et al. (2019).Counts of meristic characters (except for the axial skeleton) and assessment of qualitative characters were done using a MC-2-ZOOM stereomicroscope (Micromed, Saint Petersburg, Russia).Counts of vertebrae and pterygiophores followed Naseka (1996) and Bogutskaya et al. (2019) based on radiographs made by a PRDU (v.II) X-Ray equipment (ELTECH-Med, St. Petersburg, Russia).Images of pharyngeal teeth were taken using a JEOL JSM-6510LV scanning electron microscope (Jeol, Tokyo, Japan).
Measurement indexes were statistically processed in Microsoft Excel.Comparison of multiple samples was carried out using the Kruskal-Wallis test followed by the Dunn's post hoc test with Bonferroni correction [rstatix (Kassambara 2020) and tidyverse (Wickham et al. 2019) packages in R version 4.3.1 (Ihaka and Gentleman 1996)].Differences between sexes were tested using the Mann-Whitney U test in Past 4.12b (Hammer and Harper 2001).Principal component analysis (PCA) was performed using the ggfortify (Tang et al. 2016) package in R.
DNA chromatograms were checked for errors in FinchTV 1.4.0 (Rothgänger et al. 2006), and the DNA sequences were aligned using the ClustalW algorithm in MEGA7 (Kumar et al. 2016).Phylogenetic analysis was performed on COI (567 bp) and cytb (1089 bp) concatenated sequences.In addition to the 29 newly determined COI and cytb sequences in this study, 294 concatenated sequences of all available Phoxinus spp.were mined from the GenBank (derived from the studies of Imoto et al. 2013;Xu et al. 2014;Palandačić et al. 2015Palandačić et al. , 2017Palandačić et al. , 2020;;Ramler et al. 2016;Schönhuth et al. 2018;and unpublished works).Three outgroups representing genera Pseudaspius, Rhynchocypris, and Oreoleuciscus were selected according to the previous phylogenetic studies (Palandačić et al. 2015(Palandačić et al. , 2020)).(Suppl.material 5).Only unique haplotypes were used in downstream phylogenetic analyses.
The average intra-group as well as the average pairwise intergroup p-distances using concatenated COI+cytb sequences data set were calculated using the MEGA7 program (Kumar et al. 2016) with 1000 bootstrap replicas.

Phylogenetic placement and genetic distance
Phylogenetic Bayesian tree of the genus Phoxinus shows that Phoxinus adagumicus sp.nov.has its own cluster representing one of the earliest branches with a position between the earliest branch of Phoxinus (P.tumensis and Phoxinus sp. from Far East) and large clade represented the other species from Europe (Fig. 2).Although ML-tree shows some differences in topology, the early branching of Phoxinus adagumicus sp.nov.compared to the most of European species is retained (Suppl.material 7).Phoxinus adagumicus sp.nov.shows the lowest genetic distance to P. cf. morella (Leske, 1774) from basins of the North and Baltic seas and adjacent upper reaches of the Danube (p-distance = 0.074±0.006)(Table 1), although it is almost equally close to P. marsilii Heckel,  Comparative material.Suppl.material 1. Material used in the genetic analysis.Suppl.material 5. Etymology.The new species is named after the Adagum River, left tributary of the lower reach of the Kuban River, where the species occurs; adagumicus -an adjective.
Diagnosis.Phoxinus adagumicus sp.nov. is distinguished from geographically close species (P.chrysoprasius and P. colchicus) by the presence and predominance of specimens with single-row pharyngeal teeth on one or both sides and a combination of characters, none of which is unique, as follows: head depth at nape 54.1-64.8%HL (mean 60.5), and head depth through eye 45.3-51.6%HL (mean 48.0); head length 2.7-3.7 (mean 3.1) times caudal peduncle depth in females and 2.7-3.2(mean 2.9) times in males; body width at dorsal-fin origin 1.4-2 (mean 1.6) times caudal peduncle depth in females and 1.3-1.6 (mean 1.4) times in males; mean number of scale rows on left and right breast patches 3-9 (mean 6.1); scales below lateral line 8-14 (mean 11.8); number of circumpeduncular scales 37-50 (mean 41.5); 3 rd -6 th type of breast scalation (mode 4 th type).
Description.The general appearance of P. adagumicus sp.nov. is shown in Figs 3-4.Morphometric measurements for the holotype and type series with level of significance of sex-related differences are given in Table 2; meristic and qualitative characters for specimens from the type locality are given in Table 3.Primary morphological data for specimens from the type locality (holotype, paratypes and additional material) are given in Suppl.material 2; meristic and qualitative characters of P. adagumicus sp.nov.and other Phoxinus spp.are given in Suppl.material 3; morphometric measurements of P. adagumicus sp.nov., P. chrysoprasius and P. colchicus and their comparison are given in Suppl.material 4.
The most common pharyngeal teeth formulae are 5-4 (n = 11), 5-4.1 (n = 6), 1.5-4 (n = 5) and 1.5-4.1 (n = 7) (Fig. 5, Table 4).Total number of vertebrae in the holotype 41, 39-41 in the paratypes, and 39-42 in the add.Qualitative characters.Pectoral fins do not reach beginning of pelvic fins, except for a few males (ca.7% in total).In most specimens (ca.70%), the tip of upper lip above the horizontal level of lowest point of the eye, in some specimens (ca.25%) at the level, and in 5% of the specimens below the level.Origin of the anal fin is mainly behind the vertical of the posterior insertion of the dorsal fin.(ca.53%), often at the vertical (39%), rarely ahead (8%).Free margin of the dorsal fin mainly straight or slightly convex, anal fin slightly convex or rarely slightly concave.3 rd -6 th type of breast scalation (mode 4 th type).
Coloration.Live coloration of females outside the spawning period is brown, gray or light golden hues (see Fig. 4).In males, the color is similar, but with a greater dominance of golden or greenish hues.In spawning coloration of females, golden hue increases significantly and the coloration becomes more contrasting, in general.The spawning coloration of males is also more contrasting, with dominance of green color with variations towards yellow-green or black-green.The operculum stains are blue and the suboperculum is yellow in both sexes, but this is much more pronounced in males.Red hues appear in spawning coloration and are concentrated at the base of the pelvic, pectoral and anal fins, as well as around the mouth.The specimens preserved in formalin had a yellowish color, which is somewhat darker with a brown tint in the upper parts.Sexual dimorphism.Significant differences were observed in 18 out of 41 morphometric characters (Table 2).In addition to some classical sex characteristics in Phoxinus minnows (e.g., narrower pectoral fins and less bright colors in females), females of the new species generally have shorter anal and pelvic fins, smaller eye diameter, and a higher ratio of predorsal length to head length.
Taxonomic remarks.The presence of Phoxinus minnow in the left lower tributaries of the Kuban has been documented since the first half of the 20 th century (Malyatsky 1930;Berg 1949;Sukhanova and Troitsky 1949), and all previous researchers have attributed this population to P. phoxinus.Emtyl and Ivanenko (2002) used the name 'Phoxinus phoxinus kubanicum sp.nov.' for the minnows from Trans-Kuban rivers (Aphips, Il, Ubin, Abin, and Adagum) and Lake Abrau (Black Sea basin).Even though the name is accompanied by a comparative description, it cannot be considered as valid because it does not comply with the criteria stipulated in art.16.4 of the International Code of Zoological Nomenclature (Ride et al. 1999) for species-group names proposed after 1999, as it is not accompanied by an explicit preservation of a holotype or syntypes for the nominal taxon (art.16.4.1.)and a statement of deposit in a collection (art.16.4.2.) (Bogutskaya et al. 2023).The names P. phoxinus kubanicum, P. phoxinus kubanicus [sic], and P. kubanicus [sic], after an attempt to describe this species (Emtyl and Ivanenko 2002), were subsequently used several times (Otrishko and Emtyl 2013a, 2013b, 2013c;Karnaukhov 2020); however, according to Bogutskaya et al. (2023), an ICZN commissioner, Nikita Kluge (pers.comm.), taxonomists Pyotr Petrov (pers.comm.) and Boris Kataev (pers.comm.), they cannot be considered as available names.
It is worth noting that the original description of 'P.phoxinus kubanicum' does not correspond to the

A B
morphological diagnosis of minnows from the Adagum basin rivers obtained in this research.For example, the two-row formula (2.5-4.2) of the pharyngeal teeth is indicated (Emtyl and Ivanenko 2002), but here, none out of 33 examined specimens had such a formula (Table 4).Also, it is necessary to emphasize that the given image of 'P.phoxinus kubanicum ' (fig. 69 in Emtyl and Ivanenko 2002) does not relate to this species but is borrowed from the monograph of L.S. Berg (1949, fig. 447) and refers to the minnows inhabiting Lovozero (Northern Russia), which most probably belong to Phoxinus sp.clade 17 of Palandačić et al. (2020).Thus, 'P.phoxinus kubanicum' should be considered as an unavailable name.Type locality.Pryamaya Shchel River (44.8538°N, 37.8417°E) upstream of Nizhnebakanskaya, Krasnodar Krai, Russia.A tributary of the Bakanka River → Adagum River → Kuban River → Sea of Azov.
Distribution and habitats.An endemic species living in the northwestern Caucasus in the Adagum River basin, a tributary of the Kuban (Fig. 1).The species has a rather limited range with only 55 km between most distant known occurrences.The species has been found only in small streams located in mountainous and foothill areas -in a zone of a width about 15-20 km along the northern slope of the western part of the Main Caucasian Range.Example of habitat for this species (Abin River) is shown in Fig. 6.Habitat of P. adagumicus sp.nov. in other parts of the Kuban basin and on the Black Sea coast of the Caucasus (Lake Abrau) indicated according to literature data (Berg 1949;Sukhanova and Troitskiy 1949;Emtyl et al. 1994;Emtyl and Ivanenko 2002;Karnaukhov 2020) was not confirmed by our research and may require additional study (see Discussion).
Morphological comparisons.PCA of morphometric characters shows that P. adagumicus sp.nov. is more overlapping with P. chrysoprasius than with P. colchicus (Fig. 7).The highest loadings for P. adagumicus sp.nov.are: head length/caudal peduncle depth, body width at dorsal-fin origin/caudal peduncle depth and eye horizontal diameter (% interorbital width).Sex differences in all three species are divergent in the second component.
The occurrence of single-row pharyngeal teeth, frequent in P. adagumicus sp.nov.(Fig. 5, Table 4), is rare in the genus Phoxinus.In comparison with the geographically neighboring species, P. colchicus and P. chrysoprasius, P. adagumicus sp.nov.has unique formulas of pharyngeal teeth: 5-4, 1.5-4, 5-4.1, 5-5.1 and 1.5-5.1.According to our data, P. colchicus and P. chrysoprasius do not show single-row formulae even on one side, while in P. adagumicus sp.nov.formula 5-4 is found in 1/3 of all studied specimens.In P. colchicus and P. chrysoprasius, formula 2.5-4.2 is modal, but this is absent among individuals of P. adagumicus sp.nov.Single-row pharyngeal teeth were indicated as a unique feature for P. apollonicus Bianco et De Bonis, 2015 from the basin of Lake Skadar in Montenegro (Bianco and De Bonis 2015).

P. chrysoprasius P. colchicus
Female Male in Italy, Slovenia, and Croatia (Bogutskaya et al. 2019), P. adagumicus sp.nov.has a smaller number of anal-fin pterygiophores in front of the first caudal vertebrae (3-7, mean 4.4 vs. 3-8, mean 5.5 and 3 rd -6 th types of scalation pattern of the breast and anterior belly with predominance of 4 th type (vs. 2 nd -7 th types with predominance of 3 rd type) (Suppl.material 3).
Compared to P. strandjae from the rivers of the Black Sea basin, Bulgaria and the rivers of the Marmara Sea, Turkey (Bogutskaya et al. 2019(Bogutskaya et al. , 2023)), P. adagumicus sp.nov.has a smaller number of anal-fin pterygiophores in front of the first caudal vertebrae (3-7, mean 4.4 vs. 4-8, mean 5.6); and 3 rd -6 th types of scalation pattern of the breast and anterior belly with predominance of 4 th type (vs. 3 rd -12 th types with predominance of 6 th , 7 th , 9 th and 11 th types) (Suppl.material 3).

Discussion
This study clarified the taxonomy, morphology, genetics, and distribution of the Phoxinus minnows inhabiting the Caucasus including the Kuban basin, a large riverine system in the Northern Caucasus that is the richest in endemic species compared to fish fauna in remaining European Russia (Abell et al. 2008).Two distinct species, P. adagumicus sp.nov.and P. colchicus, were identified.Geographic ranges of P. adagumicus sp.nov.and P. colchicus in the Kuban basin are separated from each other by at least 145 km in a straight line, and also differ in altitude that is ranging within ca.50-130 m above sea level for P. adagumicus sp.nov.and within ca.280-580 m for P. colchicus.Phoxinus adagumicus sp.nov.was found only in the Adagum basin, while another species, inhabiting the middle reach of the Belaya River, refers to P. colchicus -species widely distributed in the rivers draining the east coast of the Black Sea.Phoxinus adagumicus sp.nov.represents deeply divergent lineage that is earlier branched than other European species.Therefore, P. adagumicus sp.nov.may represent a relic lineage of European minnows, the ancestors of which were among first colonizers of Europe from the East.
The population of P. colchicus in the Belaya River system is apparently a result of the past river capture event.The Kuban population of P. colchicus is the single population recorded outside the Black Sea basin.The uniqueness of haplotypes from the Kuban basin may indicate rather long isolation of this population or be a result of founder effect.The upper reaches of the Belaya River system share watershed with upper reaches of the rivers Shakhe, Sochi and Achipse belonging to the Black Sea basin.River captures with naturally translocated fish individuals between the Kuban and Black Sea tributaries might be a common phenomenon.For example, the recent discovery of the Black Sea populations of Barbus tauricus Kessler, 1877 in some tributaries of the Lower Kuban might be a result of past colonization through main channel of the Kuban River or be an event of river captures (Levin et al. 2019b).
Large morphological variation and overlap between species are a challenge for Phoxinus taxonomy.However, Phoxinus adagumicus sp.nov.differs from other closely-related or geographically neighboring species in pharyngeal teeth formula, having reduced number of teeth and rows.The Phoxinus spp.usually have two-rowed pharyngeal teeth (Berg 1949) while P. adagumicus sp.nov.represents one-rowed formula as strictly dominating over two-rowed.Remarkably, one-rowed teeth formula was found in only one another species -P.apollonicus from the basin of Lake Skadar in Montenegro (Bianco and De Bonis 2015), which was suggested to be synonymized with P. karsticus Bianco et De Bonis, 2015(Palandačić et al. 2017).A decrease in the number of teeth (4.2 and 4.1) is also observed in another Western Balkan species -P.ketmaieri Bianco et De Bonis, 2015, synonymized subsequently with P. lumaireul (Palandačić et al. 2017).Traditionally, the phylogenetic value of pharyngeal tooth numbers and their organization (formula) was high (Chu 1935;Tao et al. 2019) but some recent studies showed great phenotypic plasticity of this character from one side (Shkil et al. 2010;Bolotovskiy and Levin 2011) as well as inheritance of the teeth row numbers from another side (Shkil' and Levin 2008).Nevertheless, in our opinion, future studies should pay more attention to the pharyngeal teeth formula since it is often neglected in the current taxonomical studies.
The geographic distribution of P. adagumicus sp.nov.might be wider than its known distribution range in the Adagum River basin.Emtyl et al. (1994) stated that minnows from the rivers Dyurso (Black Sea basin) and Ayuk (tributary of the Psekups River, Kuban basin) are different from "common minnow" (authors did not use Latinian name of species using common name instead, we assume that this is some Phoxinus from the European part of the USSR) and "Colchis minnow" (which refers to P. colchicus).The habitat of minnows in the Ayuk basin was noted earlier (Sukhanova and Troitskiy 1949), where it was identified as 'P.phoxinus'.This species was previously reported for the Dyurso River (Luzhnyak 2003).Our examination of the Phoxinus material from this river and nearby (Ozereyka River), only confirmed the presence of P. colchicus.Emtyl and Ivanenko (2002), when described 'P.phoxinus kubanicum', indicated Lake Abrau in the Black Sea basin as the habitat for this species in addition to the rivers of the Kuban basin.The Phoxinus was recorded in the lake previously (Malyatsky 1930;Berg 1949), but current studies could not confirm its presence in both Lake (Luzhnyak 2003) and its tributary, the Abrau River (our field survey in 2023).What Phoxinus species lived there is unknown, but given the presence of P. colchicus in adjacent rivers, one may suggest that this is the same species.In addition, Emtyl and Ivanenko (2002) noted the following locations of the Kuban Phoxinus outside the Adagum and Psekups basins: the Aphips and Ubin Rivers.It is worth noting that the Aphips River basin including its left tributary, the Ubin River, are located between the Adagum and Psekups basins (Fig. 1).At the same time, Phoxinus was never recorded in the Aphips basin by our numerous field surveys during 2015-2023, or by earlier researchers (Sukhanova and Troitskiy 1949), which also pointed to a large gap in the distributional range between the Adagum and Psekups basins.From all of the above, we can conclude that the entire or the main range of P. adagumicus sp.nov. is located within the Kuban River in its sub-basin -the Adagum River system.The taxonomic status of Phoxinus from the Ayuk River in the Psekups basin, should be clarified in future research.
We have to consider other peculiarities of the distribution of P. adagumicus sp.nov. in the Kuban basin in the light of recent report on its finding (clade 19 by Palandačić et al. 2020) in the Belaya River (tributary of the Kuban).Clade 19 was also reported by Bogutskaya et al. (2023) for "Adagum River at Krymsk, Kuban drainage".Habitat of the Phoxinus in the Belaya basin is confirmed by our studies but according to genetic results (Fig. 2) and morphological analysis (Fig. 7), this population belongs to P. colchicus.The latter is common in the Belaya River and its tributaries at least from Khamyshki vil.down to Tulsky town along ca.60 km (Artaev et. al 2021).Thus, we have to conclude that either two species of the genus Phoxinus coexist in the Belaya basin, or the coordinates in Palandačić et al. (2020) are indicated incorrectly.We are inclined to the second opinion.

Figure 1 .
Figure 1.Map of localities of Phoxinus spp.sampled for this study.Localities are designated in Suppl.material 1.

PFigure 2 .
Figure2.BI consensus tree of concatenated COI and cytb mtDNA sequences representing all available Phoxinus species in Genbank combined with our data set.Number of some clades is given according to the study ofPalandačić et al. (2020).Species from the Kuban basin are highlighted with color.Bootstrap values/posterior probabilities above 80/0.8are shown; asterisks represent of 100/1 bootstrap/posterior probabilities values.Scale bar is in expected substitutions per site.The nodes with multiple specimens were collapsed to a triangle, with the horizontal depth indicating the level of divergence within the node.
material, commonly 40 or 41.Number of abdominal vertebrae in the holotype 22, 22-23 in the paratypes, and 21-24 in the add.material, commonly 22 or 23.Number of caudal vertebrae in the holotype 19, 17-19 in the paratypes, and 16-19 in add.material, commonly 18. Number of predorsal abdominal vertebrae in the holotype 15, 14-15 in the paratypes, and 14-16 in the add.material, commonly 15.Number of anal-fin pterygiophores in front of the first caudal vertebrae in the holotype 3, 3-6 in the paratypes,

Figure 4 .
Figure 4. Live coloration of Phoxinus adagumicus sp.nov.from the Zybza River.Side and ventral views are of the same specimens (IBIW_FS_339).Fish of spawning or post-spawning coloration were caught on 8 th May 2022.

Figure 7 .
Figure 7. PCA of morphometric characters for Phoxinus spp.under comparison and loading plot showing how strongly each character influences principal components.Specimens of P. colchicus from the Kuban River basin (Khamyshinka River, a tributary of Belaya River) are encircled by black color.

Table 1 .
Genetic p-distances between species or groups of Phoxinus spp.for concatenated COI and cytb mtDNA sequences.The averages of interspecies distances are given below diagonal, the standard errors are given above diagonal; the intraspecies divergence is given in a diagonal in bold. P

Table 3 .
Meristics and scalation pattern of Phoxinus adagumicus sp.nov.from type series and additional material from the type locality (primary data see in Suppl.material 2).Additional material from non-type localities is given in Suppl.material 3.