A new species of the genus Cottus (Scorpaeniformes, Cottidae) from the Baltic Sea Basin and its phylogenetic placement

. The genus Cottus contains more than 60 species that are common in freshwater bodies of northern Eurasia and North America. Despite the abundance of species, this genus has been insufficiently studied in Eastern European rivers and lakes. The new taxon Cottus cyclophthalmus sp. nov. was found in the Neman/Nemunas and Venta river systems (Baltic Sea Basin). Cottus cyclophthalmus is sister and most morphologically similar to Cottus gobio . The new species has the following diagnostic characters: round, protruding (tubular) eyes near front of head; dermal papillae on top and sides of head, naked body, absence of bony prickles, full trunk canal with 32–36 pores. The description of Cottus cyclophthalmus and a new finding of Cottus microstomus improved understanding of the Eastern European biodiversity. V. & Zhidkov Z. 2022. A new species of genus Cottus (Scorpaeniformes, Cottidae) from the Baltic Sea Basin and its phylogenetic placement. European Journal of Taxonomy 834: 38–57.


Introduction
Freshwater sculpins of the genus Cottus Linnaeus, 1758 distributed in freshwaters of Europe belong to two species groups: the Western European species group "Cottus gobio" and the Eastern European group "Cottus koshewnikowi" (Sideleva et al. 2019).
Molecular genetic study of the mtDNA sequences (control region and cytochrome b) revealed the existence of three phylogenetic lineages: a W lineage occurs to the west of the Baltic Sea, a SE lineage to the south and southeast of it, an E lineage is found to the east and northeast of the Baltic Sea (Kontula & European Journal of Taxonomy 834: 38-57 ISSN 2118-9773 https://doi.org/10.5852/ejt.2022.834.1897 www.europeanjournaloftaxonomy.eu 2022 · Sideleva V. et al. This work is licensed under a Creative Commons Attribution License (CC BY 4.0).

R e s e a r c h a r t i c l e
urn:lsid:zoobank.org:pub:FB37B188-5202-461D-BEA7-B9D23BE346CA Väinölä 2001). In our opinion, the W lineage corresponds to the "Cottus gobio" species group, the E lineage corresponds to the "Cottus koshewnikowi" species group, the SE lineage from South Estonia is Cottus sp. These results support the presence of two species groups ("Cottus gobio" and "Cottus koshewnikowi") identified using morphological characters (length of the trunk canal, number and distribution of spines on the body) (Witkowski 1979).
The species group "Cottus koshewnikowi" is represented by two species: C. koshewnikowi Gratzianov, 1907and C. gratzianowi Sideleva, Naseka & Zhidkov, 2015(Sideleva et al. 2015a, 2015b. This group of species has such morphological features as the presence of bony prickles on the body and a short (incomplete) trunk canal.
The sculpins of each species group are distributed in different parts of Europe (Freyhof et al. 2005;Sideleva et al. 2015a). The representatives of the "Cottus gobio" group are found from the Pyrenees in the west to the Dniester River in the east. The species of the "Cottus koshewnikowi" group are distributed in Eastern Europe from Finland and the Western Dvina (Daugava) River basin in the west to the Urals in the east.
The Dniester River (Black Sea Basin) can be considered a conventional faunistic boundary between the species groups "Cottus gobio" and "Cottus koshewnikowi". However, to the north of the source of the Dniester River, there are large rivers and river systems whose faunas include sculpins with an unclear taxonomic status. These rivers include the Neman/Nemunas and Venta with their tributaries and the Krasnaya River. The mentioned river systems belong to the Baltic Sea Basin.
The fact that these rivers are inhabited by sculpins is known from the publications of Zograf (1907), Zhukov (1958), Alekseev &Probatov (1969), andTylik &Shibaev (2008). The sculpins have been identified as C. gobio. In 2019, specimens of sculpins were caught in the Nemunas/Neman and Venta river systems (including the Krasnaya River) and morphological and molecular genetic studies were carried out. These studies revealed the presence in these rivers of the species C. microstomus and of a new form different from the known species of the genus Cottus.

Study area and sampling
For this study, sculpins were collected in August 2019 in three tributaries of the Neman River: Neris, Žeimena, and Siesartis (Lithuania). In addition, specimens were caught in the Krasnaya River connected to the Neman by an artificial canal (Kaliningrad Region, Russia). In the Venta River system, samples were collected in its small tributary Šerkšnė River (Lithuania) (Fig. 1). The coordinates of each sampling site are presented in the description of the new species.
The sculpins were caught using a hand net and a battery-powered electric fishing gear (HANS GRASSL GmbH, model IG200/2 In total, 237 specimens were included in this study. Topotypes for three species (C. gobio, C. koshewnikowi, and C. metae) have been studied morphologically and genetically. Selecting data from the GenBank NCBI for other species, we took into account the proximity of the collection site to the type habitat. Therefore, we did not take all the data available.
For molecular genetic analysis of mitochondrial DNA, the tissue samples of three species were used: C. koshewnikowi (10 specimens), C. poecilopus Heckel 1837 (4 specimens), and a new species C. cyclophthalmus (31 specimens from all studied localities). For other species of Cottoidei, data from GenBank NCBI (https://www.ncbi.nlm.nih.gov/genbank/) were used (the list of species in Supplementary file 2). Thus, the sequences of the mtDNA control region were taken for five valid species belonging to the "Cottus gobio" group.
The control region is widely used for species identification and differentiation between closely related species of cottoid fishes of the genus Cottus (Kontula & Väinölä 2001;Šlechtová et al. 2004;Yokoyama et al. 2008;Bravničar et al. 2021).

Lateral line, axial skeleton and 3D scan of the lateral, dorsal and ventral views
The sensory canals and pores of the lateral line were studied by injecting methylene blue into the canals with a syringe. The canals were colored blue and photographed (Sideleva 1982).
An X-ray unit (PRDU, manufactured by Eltekhmed) and a Soredex Digora PCT scanner were used to obtain a digital image of the axial skeleton of the sculpins.
A 3D scan of type specimens and the new species Cottus cyclophthalmus sp. nov. was performed at St. Petersburg State University. A RangeVision Spectrum scanner was used. The resulting model was imported into 3D modeling software Autodesk Meshmixer (ver. 3.3.15) which was used to capture an image of the fish in three projections (lateral, dorsal and ventral views).

Statistical analyses
To study morphometric characters, such multivariate statistical methods as principal component analysis (PCA) and discriminant function analysis (DFA) were used. The scheme of measurements (developed by Taliev 1955 andSideleva et al. 2015b) included 26 distances. This scheme is designed specifically for cottids.
For these data, a principal component analysis (PCA) based on the correlation matrix was performed (absolute measurements were used). A reduced set of orthogonal vectors was generated from the original variables. The obtained principal components are considered to be new uncorrelated characters. Most of the original morphometric characters have high and positive factor loadings on the first principal component (PC1). This means that PC1 has the greatest contribution to the total variability and mainly determines the size differences between individuals (Somers 1986). A scatterplot in the space of the second and third components was created to describe the differences between samples not related to size variability. Each sample forms its own cluster (morphospace) on the diagram. The degree of cluster overlap indicates their morphological differentiation.
Morphometric features that were analyzed using PCA were also used for a discriminant analysis. However, relative measurements (% SL) were used for this analysis. The quality of discrimination was assessed based on the Wilks' Lambda and F-test statistics. Wilks' Lambda values close to 0 indicate a strong discrimination. The level of differences between species was determined based on the values of the squared Mahalanobis distance. The contribution of each character to the discriminating power of the model was estimated by the Partial Lambda value. The lower this indicator, the higher the contribution of the variable. Canonical analysis was used to compute orthogonal discriminant functions. The result of this analysis is visualized using a scatterplot of canonical values in the space of the first and second discriminant axes.
Both statistical analyzes were performed using the STATISTICA 10 software (StatSoft).

DNA extraction, PCR amplification and sequencing
DNA was isolated from fin-clip tissue samples (100-200 mg) fixed in 96% ethanol using QIAamp DNA Mini Kit (Qiagen, Germany). The complete CR was amplified using primers L16638 (AACTCCTACCCCTAACTCCCAAAGC) and H1122 (GGAGTGCGGAGACTTGCAT) (Kocher et al. 1989), resulting in 1000 bp amplicons that included fragments of flanking tRNA genes.
Amplification was undertaken in a BioRad C1000 Touch in a 15 μL reaction volume containing 1 × buffer, 1.5 µM MgCl2, 10 µM of each primer, 0.2 µM of each dNTP, 1 μL of template DNA solution, and 1U of HS Taq polymerase (Evrogen, Moscow). The conditions for PCR were as follows: 3 min of initial denaturation at 95°C, followed by 35 cycles of denaturation at 95°C for 20 s, primer annealing at 59.2°C for 60 s, DNA elongation at 72°C for 60 s, and final elongation at 72°C for 10 min. The sequencing of the amplified fragments was performed in a 3500 Genetic Analyzer (Applied Biosystems) using the primers mentioned above.
Isolation and amplification of DNA was carried out using the equipment of the Laboratory of Ichthyology in Zoological Institute RAS (St. Petersburg). Sequencing was performed in Papanin Institute of Biology of Inland Waters RAS (Borok).
The phylogenetic trees were reconstructed using Bayesian analysis in Mr Bayes 3.1.2 (Huelsenbeck & Ronquist 2001) and the Maximum Likelihood (ML) method in IQ-TREE 1.6.12 (Nguyen et al. 2016).
The choice of the best model for nucleotide substitutions was carried out using algorithm implemented in IQ-TREE (HKY+I+G5+F with parameters I = 0.28 and G = 0.51 was selected as best model). The Bayesian inference of phylogeny was done by using the selected model (nst = 2, rates = invgamma). The MCMC process was set for four chains to run simultaneously for 10 7 generations, with sampling trees at every 1000 generations. The first 25% of trees were discarded in the computation of the majority-rule consensus tree. Posterior probabilities were calculated by generating a 50% majority rule consensus tree with the remaining trees. The statistical reliability of the ML tree was assessed by the bootstrap method (2000 pseudoreplications). The phylogenetic tree was visualized using FigTree 1.4.4 software (http://tree.bio.ed.ac.uk/software/figtree).
To assess the genetic diversity of C. cyclophthalmus, the average number of nucleotide substitutions, haplotype diversity (Hd), and nucleotide diversity (π) were estimated. The calculations were performed using the algorithm implemented in DnaSP ver. 6.12.03 (Rozas et al. 2017). Gaps were treated as the fifth state.
Pairwise p-distances between different species and between haplotypes of the same species were calculated using MEGA X (Kumar et al. 2018). The bPTP web server was used for species delimitation (Zhang et al. 2013). The bPTP analysis was performed using 10 5 MCMC generations (thinning interval = 100, burn-in = 0.1).

Diagnosis
Cottus cyclophtalmus sp. nov. has round, protruding (tubular) eyes near front of head; dermal papillae present on top and sides of head, body naked, bony prickles absent; full trunk canal with 32-36 pores.

Etymology
The name of the new species is derived from the Latin word for 'round-eyed' and is associated with the round and convex shape of eyes.

Description
Body shortened, its average maximum depth at origin of first dorsal fin four times SL. Body massive, preanal distance more than half SL. Caudal peduncle short, its length 14-17% SL (14.9% in holotype), average height of caudal peduncle half its length (Table 1). Trunk naked, bony prickles (modified scales) absent.
Head large with smooth dorsal profile from head to back, its length more than 30% SL in type specimens (33% in holotype). Dermal papillae numerous on top and sides of head, sometimes in form of circles.
Postorbital length large, always more than half head length (53% HL in holotype). Snout short, more than half postorbital length (59.4% in holotype). Anterior nostrils small, tubular, highly pigmented; posterior nostrils in form of short tubes. Mouth small, terminal; upper jaw does not reach vertical line of anterior edge of eye. Teeth on jaws and vomer small, numerous, of same shape and size. Upper lip thick, fleshy, twice thicker than lower lip.
Eye round and protruding, near to front of head, average eye diameter 7.3% SL (7.4% SL or 22.4% HL in holotype). Interorbital space narrow, on average 1.5 times less than eye diameter. Preoperculum with three spines; upper spine sharp, directed backwards and slightly curved inward. Second and third spines small, in form of tubercles hidden under the skin. Interbranchial length large, on average 1.5 times less than length of gill slit (16.6% in holotype).
Two dorsal fins follow each other without gap. First dorsal fin low, length of its longest rays 1.5 times as long as rays of second dorsal fin. Narrow light border along edge of first dorsal fin. Second dorsal fin long, its base 2.5 times as long as base of first dorsal fin. Origin of anal fin at short distance (3% SL) from anus, on vertical line of second ray of second dorsal fin; length of longest rays in anal fin 1.2 times that in second dorsal fin. Pectoral fin short, reaching vertical of first ray of second dorsal fin. Pelvic fin long (20% SL), not reaching anus.
The first proximal pterygiophore of the dorsal series is placed between first and second vertebrae. It supports first dorsal fin spine which is in supernumerary position (morphotype A according to Yabe 1985). One interdorsal pterygiophore is placed between first and second dorsal fin. The last proximal pterygiophore of dorsal series supports one or two rays (one in holotype). The last pterygiophore of anal fin supports one or two fin rays (two in holotype).
The caudal skeleton is composed of single hypural-parhypural complex bone and three epurals. The complex bone has deep notch posteromedially and supports principal caudal-fin rays. Medial principal rays of caudal fin are branched.
Number of rays in fins: first dorsal fin with six to eight spines (seven in holotype); second dorsal fin with 16 to 19 rays (17 in holotype); anal fin with 12 to 15 rays (13 in holotype); pectoral fin with 12 to 15 rays (14 in holotype); four rays in pelvic fin; caudal fin with 12 principal rays (eight branched and four unbranched).
Lateral line of Cottus cyclophthalmus is typical of the genus Cottus. All sensory canals (with exception of preopercular-mandibular canal) are interconnected and form a unified system.

Variation of morphometric features and numbers of rays in Cottus cyclophthalmus sp. nov.
For a comparative analysis of morphometric characters of type and non-type specimens of Cottus cyclophthalmus sp. nov. from the rivers Krasnaya, Neris, Šerkšnė, Siesartis and Žeimena, PCA was used (Fig. 4). The variability of 26 external features was analyzed. (Table 2). The principal component analysis revealed a variability of initial data. This data is represented in a scatter plot with uncorrelated second and third principal components. Each component reflects a proportion of a variability of variance-covariance matrix of features. The conducted analysis showed that the first principal component describes 84.6% of total variability of measurements in five studied samples of sculpins. The first component is characterized by close positive values of factor loadings (from 0.758 to 0.982, or 0.918 on average). All other 25 components describe 15.4% of total variability. Figure 4 shows morphospaces of studied samples in the space of second and third components. The morphospaces of samples of C. cyclophthalmus overlap significantly. There are no discrete geographic groupings. This indicates that the five studied samples are not differentiated by morphometric characters and belong to the same species C. cyclophthalmus.
Data on variation in number of rays in dorsal, anal, and pectoral fins in five samples of Cottus cyclophthalmus sp. nov. are presented in Table 3. A comparison of presented data shows that sculpins from different localities have a similar degree of variation in the number of rays in fins. The differences in number of rays in each fin ranged from 2 to 4 values. In the first dorsal fin in the type specimens from Krasnaya River, the number of rays varied only within two ranges (six to seven rays). In non-type individuals from tributaries of the Nemunas/Neman and Venta rivers, the number of rays varied from six to eight. In all samples (except for sculpins from Šerkšnė River), fish with seven rays in the first dorsal fin dominated (76-94% of all specimens). The number of rays in the second dorsal fin ranged from 16 to 19. Type specimens from Krasnaya River had the smallest (16-17) number of rays in the second dorsal fin. The difference between the samples lies in modal values, with the majority of studied fish (69.5%) having 17-18 rays in the second dorsal fin. In the anal fin, the number of rays varied from 12 to 16. Type specimens of Cottus cyclophthalmus had 13-14 rays in anal fin. The same number of rays dominated in 94% of the studied fish. In the pectoral fin, the number of rays varies from 12 to 15. Fish with a modal number of 14 rays dominated in all samples.

Variation in mitochondrial DNA sequences
To identify intraspecific genetic diversity of C. cyclophthalmus sp. nov., we studied nucleotide sequences (858 bp) of mtDNA control region. Table 4 contains the data on haplotypes found in samples of C. cyclophthalmus from four rivers (Krasnaya, Žeimena, Siesartis, and Šerkšnė). Nine haplotypes were identified (CCY1-CCY9). The most common haplotype was CCY6 (52% of specimens of the new species). It was found in sculpins from each of the studied rivers. The sample of a new species from the Šerkšnė River was characterized by the greatest haplotype diversity (6 unique haplotypes). The number of detected polymorphic sites (S) was seven. The nucleotide diversity of mtDNA control region among individuals of C. cyclophthalmus had a low value (π = 0.00139 ± 0.00024). The haplotype diversity (Hd) was high (0.718 ± 0.080). The average number of nucleotide differences was 1.196.
All data indicate a low level of genetic differentiation between the studied specimens. This is the evidence that all individuals belong to species C. cyclophthalmus.

Morphological differentiation between Cottus cyclophthalmus sp. nov. and other species of genus Cottus
The new species C. cyclophtalmus sp. nov. is characterized by morphological features, the states of which are similar to that of C. gobio. The characters that are common to C. cyclophtalmus and C. gobio are as follows: a shortened body with a similar postanal distance (47.2% SL in type specimens vs 48.7); short caudal peduncle (
Cottus cyclophthalmus sp. nov. differs from all European species of the genus Cottus by the following complex of morphological characters: round bulging eyes, short snout, dermal papillae on top and sides of head, thick upper lip, and light coloration of external part of pectoral fin.
The discriminant function analysis (DFA) based on morphometric data revealed significant morphological differences in body shape between three species: C. cyclophthalmus, C. gobio, and C. koshewnikowi (Fig. 5). Statistical analysis (Wilks' Lambda = 0.00852, approx. F (50, 136) = 26.745, p < 0.0000) indicated a significant discrimination of the three species. The new species from the Nemunas/Neman and Venta river systems was the most different from the Western European species C. gobio (squared Mahalanobis distance between these species was 93.82). Mahalanobis distance between samples of C. cyclophthalmus and Eastern European C. koshewnikowi was slightly lower (55.02). Partial Lambdas demonstrating individual contribution of variables to discriminatory power of model are shown in Table 5. The most important characters (Partial Lambda < 0.8) for discriminating samples of species were horizontal diameter of eye and length of base of first dorsal fin. Thus, C. cyclophthalmus is well differentiated from C. gobio and C. koshewnikowi by the complex of morphometric characters.

Phylogenetic placement of Cottus cyclophthalmus sp. nov. in freshwater Cottoidei
An analysis of phylogenetic relationships of freshwater Cottidae was performed based on the mtDNA control region sequences using Bayesian and ML methods. Both analyzes generated trees with similar topology. The Bayesian tree is shown in Fig. 6.
Among the species of the ingroup, the most isolated, basal position in the phylogenetic tree was occupied by Cottus (Rheopresbe) kazika, a catadromous species inhabiting the rivers of Honshu Island (Japan) (Goto et al. 2015). According to the mtDNA control region data, the average p-distance between this species and all other presented lineages was 14.5% (from 10.8 to 18.8%). The average p-distances between the eight major clades ranged from 5.0 to 12.4% (mean value 8.7%). The genetic differences between species varied greatly (0.1-15.6%), and the average interspecific p-distance was 7.3%. The clade of endemic Baikal cottoid fishes was well supported on the phylogenetic tree. In our dataset, it was represented by 23 species from three families (Cottidae, Abyssocottidae, and Comephoridae). All the studied lineages of European representatives of the "Cottus gobio" and "Cottus koshewnikowi" species groups (including the new species C. cyclophthalmus) and the Siberian species Cottus sibiricus from the Ob and Yenisei drainages formed single clade. This result does not confirm the morphological differentiation of the two identified species groups. The morphological similarity of C. gobio and C. sibiricus has previously been known. L. Berg (1949Berg ( : 1148 wrote about the Siberian sculpin: "It is close to C. gobio, which it replaces in Siberia". Nevertheless, these species are well differentiated by osteological features. The ranges of C. sibiricus and C. gobio do not overlap. Cottus koshewnikowi differs from C. gobio by a number of morphological characters: presence of prickles, short trunk canal with smaller number of pores (22-27 vs 30-36). The p-distances between these three pairs of species are rather low (1.1% between C. gobio and C. sibiricus, 1.3% between C. gobio and C. koshewnikowi, and 0.8% between C. koshewnikowi and C. sibiricus).
All of these data suggest complex phylogenetic relationships between these species that require further study. The average interspecific p-distance within this clade was 1.8% (from 0.8 to 2.9%), which could indicate recent divergence between lineages.
The new species C. cyclophthalmus sp. nov. and C. gobio were found to be most closely related taxa. The genetic distance between them was 1.3%. According to the results of species delimitation using the bPTP web server (Fig. 6), the lineage of C. cyclophthalmus has sufficient support (0.498, with a threshold value of 0.331). This result supports its species status.

The first record of the sculpin Cottus microstomus in the Neman/Nemunas River basin
Studying fish from the tributary of the Neman River (Siesartis River), one specimen (ZIN 56723, SL 78.4 mm) was found that did not belong to the new species C. cyclophthalmus sp. nov. The diagnostic features of the found specimen are the following: low depth of caudal peduncle; head smoothly passes into body, leathery wrinkles on upper surface of head; eye oval, not bulging; in male, lobular genital papilla; dorsal, pectoral and caudal fins with dark transverse stripes; bony prickles on trunk absent; trunk canal full (Fig. 7). The number of rays in fins: first dorsal fin with 8 rays, second dorsal fin with 18 rays, anal fin with 14 rays, pectoral fin with 15 rays, and 12 principal caudal-fin rays.
According to these characters, the specimen was identified as C. microstomus. This species is known from the Vistula and Dniester rivers. The type locality of C. microstomus is near the city of Krakow (Poland). There have been no data on the distribution of this species outside the Vistula and Dniester river systems. In the system of the Nemunas/Neman River, representatives of this species have previously not been recorded. Cottus microstomus is rare in the Neman/Nemunas river system. Among 134 individuals caught, only one specimen of C. microstomus was found. The first finding of this species testifies to its wider range and expands the list of fish species of Lithuania.

Conclusion
The new species C. cyclophthalmus sp. nov. differs from other species of the genus Cottus by the set of morphological characters. Taxonomically important features are the shape and location of the eyes, as well as the presence of well-defined dermal papillae above and on the sides of the head. According to other morphological features, this species belongs to the "Cottus gobio" species group.
The shape of the body, the absence of prickles, and the full trunk sensory canal significantly distinguish C. cyclophthalmus from a closely related C. koshewnikowi, which has an adjacent range. Based on morphological and molecular genetic data C. cyclophthalmus is a distinct taxon. Its position within the genus Cottus has been clearly defined.