03_Mezhzherin_4_2020.indd UDC 595.721:575.2 MORPHOMETRIC VARIATION OF HYBRIDIZING SPECIES AND GYNOGENETIC BIOTYPES OF SPINED LOACHES (COBITIDAE, COBITIS) IN RIVER SYSTEMS OF UKRAINE S. V. Mezhzherin*, L. I. Pavlenko, A. O. Tsyba, T. V. Saliy, M. A. Ghazali Schmalhausen Institute of Zoology of NAS of Ukraine, vul. B. Khmelnytskogo, 15, Kyiv, 01030 Ukraine E-mail: smezhzherin@gmail.com *Corresponding author S. V. Mezhzherin (http://orcid.org/0000-0003-2905-5235) L. I. Pavlenko (http://orcid.org/0000-0002-5295-4258) A. O. Tsyba (https://orcid.org/0000-0001-5838-0948) T. V. Saliy (https://orcid.org/0000-0003-2961-1326) M. A. Ghazali (http://orcid.org/0000-0001-9195-0914) Morphometric Variation of Hybridizing Species and Gynogenetic Biotypes of Spined Loaches (Cobitidae, Cobitis) in River Systems of Ukraine. Mezhzherin, S. V., Pavlenko, L. I., Tsyba, A. O., Saliy, T. V., Ghazali, M. A. — To test the possibility of identifying the species and biotypes of spined loaches of the complex Cobitis elongatoides–taenia–tanaitica using body morphometric characteristics, we considered a pool of specimens from different river systems of Ukraine. The sample included three parental species and seven hybrid biotypes with various genome combinations, and the morphometry was based on 23 body parameters and 26 derivative indices. The variability was analyzed by standard and multivariate statistics. Neither any one measurement, nor their combination allowed identifying the specimens with 95–100 % probability. Pooled, samples had no internal logic of morphological remoteness of hybrid biotypes and parental species in regards to their genetic similarity. Morphometry’s low resolution in case of pooled samples is linked to the specifics of body shape and significant geographic variability which partially evens out the between-group differences, while within samples the resolution was higher without reaching even 95 % diagnostical certainty. All in all, the resolutions obtained for morphometric parameters are within results obtained in similar studies for other hybrid fishes. K e y w o r d s : morphometry, diagnostics, diploid-polyploid complex, spined loaches, Cobitis. Zoodiversity, 54(4): 285–298, 2020 DOI 10.15407/zoo2020.04.285 Morphology 286 S. V. Mezhzherin, L. I. Pavlenko, A. O. Tsyba, T. V. Saliy, M. A. Ghazali True spined loach of the genus Cobitis Linnaeus, 1758 have Trans-Palearctic range with Eastern Asian and Mediterranean-European centers of species diversity. They have for two decades now attracted much attention from scientists working on unsolved problems on the border of evolutional genetics and systematics (Rab et al., 2000; Mezhzherin, Chudakorova, 2002; Janko et al., 2003, 2005, 2007; Culing et al., 2006; Perdices et al., 2016). The special interest towards the group of small fishes is two-fold; firstly, they exhibit an extraordinary taxonomic diversity, often cryptic if only the outer appearance is taken into account. Therefore, when karyo- type diversity (Vasil’ev, 1985; Vasil’ev, Vasil’eva, 1998; Rab et al., 2000; Boroń, 2003; Mezhzherin et al., 2014), allozymes (Šlechtova et al., 2000; Mezhzherin, Pavlenko, 2010) and primary DNA structure (Culing et al., 2006; Perdices et al., 2016) began to be analyzed, within the morphological continuum of the erstwhile Cobitis taenia Linnaeus, 1758, there were revealed several evolutionary and genetically discrete forms. Such variability was the basis for isolation or description of a record, for current systematics, number of species of European vertebrates, many of which can be considered twin species. Whereas seventy years ago in the whole Palearctic there swam only C. taenia (Berg, 1949), the early millennium already sees dozens of species in the area (Perdices et al., 2016; Fish base). Secondly, the loaches are capable of intense interspecific hybridization leading to di-, tri- and tetraploid hybrids as has been known since 1970s (Ueno, Ojima, 1976; Ueno et al., 1980; Vasil’ev, 1985). It has been estab- lished since then that in most European polyploid biotypes’ chromosome sets, there are one or two genomes of C. elongatoides Bacescu, Mayer, 1969 (Bohlen, Rab, 2001; Bohlen et al., 2002; Boroń, 2003; Culing et al., 2006; Mezhzherin, Pavlenko, 2010), which lives in the Danube drainage basin and in the upper reaches of the Rhine and Oder. The genomes of genetically similar C. taenia and C. tanaitica Bacescu, Mayer, 1969 also contribute to polyploids. The range of C. taenia occupies the basins of Dniester, Southern Buh, Dnipro, and Don, and reaches the Volga and also the rivers of the North and Baltic Seas. In the upper reaches of the Rhine and Oder the species hybridizes with C. elongatoides (Boroń, 2003; Culing et al., 2006). The range of C. tanaitica lies more to the south and covers Lower Danube, lower reaches of Southern Bug and Dnipro, rivers of the Azov Sea, in- cluding Don (Vasil’ev, Vasil’eva, 1998). The hybridization between C. taenia and C. tanaitica does not directly result in allotriploids since polyploids with only these species’ genomes are so far unknown (Bohlen, Rab, 2001; Culing et al., 2006; Mezhzherin, Pavlenko, 2009). The wide area of genetic introgressions between these species includes the Lower Dnipro and Seversky Donets river system (Mezhzherin, Pavlenko, 2009), as well as the Southern Bug (Bohlen, Rab, 2001). In triploid loaches, the most frequent genome combinations are as follows: one C. elongatoides and two С. tanaitica (biotype C. elongatoides–2 tanaitica), one C. elongatoides and two C. taenia (biotype C. elongatoides–2 taenia), one C. elongatoides combined with genomes of C. taenia and C. tanaitica (biotype C. elongatoides–tanaitica–taenia) and two genomes of C. elongatoides combined with ge- nome of C. tanaitica (biotype C. 2 elongatoides–tanaitica). Much more seldom, in Ukrainian rivers there occur triploids having genomes of Sabanejewia aurata (De Filippi, 1863) and genomes of C. taenia and C. tanaitica (biotype C. taenia–C. tanaitica–S. aurata) (Mezhzherin et al., 2014). Tetraploids’ genomic structure is much more diverse, yet such biotypes are found significantly more seldom than triploids. Gametogenesis in hybrids is done by abnormal meiosis, and reproduction is mostly through gynogenesis, since in diploids always, and in triploids sometimes, there occurs fertilization which leads to tri- and tetraploids, respectively. Evidently, the presence of specimens with intermediate in regards to the parent species’ phenotypes, creates large problems in practical species diagnostics, moreover in a taxonomical group where interspecies variability of morphometric parameters might be cryptic. An explicit study of various exterior parameters in parental species and hybrid genotypes in Polish waters (Kotusz, 2000, 2008) showed that not only different hybrid biotypes but even the parental species can- not be distinguished with certainty based on such body measurements. Moreover, there was no clear logic of morphological differentiation which should be apparent in a situation where polyploid forms show gene dosage effect. Morphometric parameters, despite being of limited use in systematics of the genus Cobitis, retain a sig- nificant place in taxonomic analysis. For descriptions of new species, body measurements are still an important part of species diagnostics (Vasil’eva, Vasil’ev, 2012; Chen et al., 2015; Mousavi-Sabet et al., 2015) With this in mind, and seeing that the previous research was limited only to Polish populations, we con- sider studying morphometric variability of the parental species and hybrid forms an urgent task, made more relevant by sampling fishes from different river basins, with different ratios of the parental species and hybrid biotypes. Material and methods Our study was based on series of loaches caught in Ukrainian waters: 1 — Lower Danube (45.39, 29.59), 2 — Trans Carpathian rivers, left tributaries of Tisza (48.15, 23.39), 3 — Lower Dnipro (46.60, 32.59), 4 — Sobok River, secondary tributary of the Southern Buh (49.11, 29.03), 5 — lake Babye, Middle Dnipro floodplain (50.47, 30.54), 6 — Irpin River, right tributary of Middle Dnipro (50.52, 30.26), 7 — Navaria, water reservoir on the Shchirets River, left tributary of Dniester (49.74, 23.95), 8 — Styr River, tributary of Pripyat River (50.74, 25.31) (fig. 1). The total number of fishes was 819. 287Morphometric variation of hybridizing species and gynogenetic biotypes of spined loaches… Every fi sh was genetically tested by electrophoretic analysis of three enzymes and structural proteins of its muscle. Th e following loci were studied: Aat-1 (codes the soluble form of aspartatamino transferase), Mdh-1 (codes the soluble form of malate dehydrogenase), Ldh-B (codes a heart subunit of lactate dehydrogenase) and Pt-3 (codes one of the soluble muscle proteins). Electrophoresis was carried out in SDS-PAGE in Tris-Glycine system (Peacock et al., 1965). We also did a cytometric analysis by measuring erythrocyte area (Sezaki et al., 1977). Th e material was then divided into biotypes, and every specimen classifi ed according to the grouping. Sex was determined by the presence or absence of the Canestrini’s organ — the thickening of the fi rst ray of the pectoral fi n, characteristic of males of every species in C. taenia. Th e morphometric analysis was done according to the standard plan adapted to cobitids (Pravdin, 1966). We measured 23 body parameters (fi g. 2): L — total length, SL — standard length, CL — corpus length, ao — snout length, o — eye diameter, c — head length, po — postorbital length, hc — head height, H — depth of body at the deepest point, h– depth of body at the smallest point, aD — antidorsal length, pD — postdorsal length, aV — antiventral length, aA — antianal length, lD — length of dorsal fi n, hD — height of dorsal fi n, lA — length of anal fi n, hA — height of anal fi n, lP — length of pectoral fi n, lV — pelvic fi n length, PV — distance between pectoral and pelvic fi n, VA — distance between dorsal and anal fi n, lcaudv — caudal peduncle length. Variability analysis was done using indices, for which the measurements were taken as % of corpus length. All fi shes were fi xed in 4 % formaldehyde and measured by the same person. Fig. 1. Collection points of spined loaches in the river systems of Ukraine. Th e decoding of the numbering of samples is given in Material and methods. Fig. 2. Body measurements for Cobitis. Th e original fi sh image is from Wilhelm von Wright out of Fries, 1895. 288 S. V. Mezhzherin, L. I. Pavlenko, A. O. Tsyba, T. V. Saliy, M. A. Ghazali In total, 49 variables were taken into account: 23 absolute measurements and 26 indices. Most indices were ratios of various body measurements to corpus length. Besides that, we also used five other indices (ao/c, o/c, io/c, lcaud/aD, lcaud/hc). The number of variables was close to the number of samples in some groups which can cause statistical artifacts of significant separation of the groups (Mitteroecker, Bookstein, 2011). Thus, classification and group separation was conducted with between-group PCA implemented in function group PCA of R (version 3.4.2; R Core Team, 2017) package Morpho (version 2.5.1; Schlager, 2017). It projects data onto orthogonal axes of group means variation and tests pairwise between-group differences with the permutation test (we used 10 000 iterations). Measurements and indices were taken at different scales; some of the variables (L, SL, CL, aA, lcaud/ hc) had high variances. So, log10-transformation was applied in order to stabilize the variance. Differences in biotypes were estimated with permutational ANOVA test implemented in the function aovp of package lmPerm v. 2.1.0 (Wheeler and Torchiano, 2016) and permutational multivariate analysis of variance (PERMANOVA) with Euclidean distance measure and 1000 permutations implemented in the function adonis of package vegan v. 2.5-3 (Oksanen et al., 2018). Mahalanobis distances were estimated as Euclidean distances between group means of the canonical variates (function CVA of the package Morpho). Bootstrap support for the hierarchical clustering was evaluated with package pvclust v. 2.0-0 (Suzuki, Shimodaira, 2015). Results B i o t y p e i d e n t i f i c a t i o n a n d p o p u l a t i o n s t r u c t u r e d e t e r m i n a t i o n . In order to do this, we used a set of four loci and cytometry data which allowed us to assign biotypes and species and separate the loaches into ten groups (table 1). The spined loach C. taenia (TT) in Ukraine is genetically heterogeneous (Mezhzherin, Pavlenko, 2009), which is caused by massive introgressive hybridization with С. tanaitica and by gene flows, in particular, alleles of locus Ldh-B. The Lower Dnipro is an area of in- T a b l e 1 . Electromorph types for four loci in different groups of the hybrid biotypes and species of Cobitis Species and biotypes Loci genotypes Aat-1 Mdh-1 Ldh-B Pt-3 ТТ C. taenia 100–100 100–100 100–100, 100–105, 105–105 100–100 NN C. tanaitica 100–100 100–100, 100–110, 110–110 100–100 90–90 ЕЕ C. elongatoides 105–105, 105–110, 110–110 110–110 100–100 90–90 ЕN C. elongatoides–tanaitica 100–110 100–110, 110–110 100–100 90–90 ENN(N) C. elongatoides–2 (3) tanaitica 100–100–110 100–100–110, 100–110–110, 110–110–110 100–100 90–90–90 ЕNТ(NT) C. elongatoides–2 (3) taenia; C. elongatoides–1 (2) taenia–1 (2) tanaitica 100–100–110 100–100–110 100–100–100, 100–100–105, 100–105–105 90–90–100, 90–100–100 ЕЕ(E)N C. 2 (3) elongatoides–tanaitica 100–105–105, 100–105–110, 100–110–100 100–110–110, 100–105–110, 100–110–100 100–100–100 90–90–90 ЕЕN95 C. 2 (3) elongatoides–tanaitica95 95–100–110–110 100–110–110 100–100–100 90–90–90 ЕЕN95Т C. 2 (3) elongatoides–taenia–tanaitica 95–100–110–110 100–100–110–110 100–100–100–100 90–90–90–100 АNТ C. taenia–C. tanaitica–Sabanejewia aurata 100–100/110 100–100/110 90/100–100 — 289Morphometric variation of hybridizing species and gynogenetic biotypes of spined loaches… trogressive hybridization between C. taenia. Specimens from the region hads lightly more than half of allele genes specific for C. taenia, which lets us to formally assign them to the species. In the Navaria population of diploid C. taenia there was found a unique al- lelomorph Ldh-B105, which shows its evolutionary isolation from other populations. In the populations of the Middle Dnipro basin there were no genes specific for С. tanaitica, and Ldh-B105 was infrequent. This could be viewed to support the species’s relative homogene- ity in the region. Genetically homogeneous С. tanaitica (NN) was found only in the Lower Danube basin. C. elongatoides (EE) was caught in Lower Danube but also in the Trans Carpathian Tisza tributaries. The biotypical structure of the samples is presented in table 2. In the Lower Danube basin the rewere found diploid hybrids С. elongatoides–tanaitica (biotype EN), triploid С. elongatoides–2 tanaitica and very rare tetraploids С. elongatoides–3 tanaitica. The latter two biotypes here are combined (ENN(N)). The most frequent in the Lower Danube basin were triploid hybrids (С. 2 elongatoides–tanaitica), also found in the Tisza basin. Tetraploids (С. 3 elongatoides–tanaitica) were very scarce and so are analyzed together with triploids (С. 2 elongatoides–tanaitica) as the biotype groupe EE(E)N. Triploids in whose chromo- some sets were two genomes of C. elongatoides and one of С. tanaitica, also were the most abundant in the Irpin River basin. However, here they were genetically unstable, which can be seen in the disrupted structure of the electrophoresis of certain enzymes (Mezhzherin, Pavlenko, 2007). They were assigned a distinct biotype (EЕN95). In the Irpin basin we also found tetraploid С. 2 elongatoides–taenia–tanaitica95, separated into biotype EЕNN95. Five biotypes were grouped together as ENT(NT); their chromosome sets included one genome of С. elongatoides and always one or two genomes of C. taenia (C. elongatoides– 2 taenia, С. elongatoides–taenia–tanaitica, С. elongatoides–3 taenia, C. elongatoides–2 tae- nia–tanaitica, С. elongatoides–taenia–2 tanaitica). The reason for their merging was, first of all, insufficient resolution of electrophoresis to distinguish between С. taenia and С. ta- naitica. The only currently known locus which allows distinguishing them, Pt-3, has low ex- pression levels during summer, and when its products are poorly presented on electrophore- grams, it becomes impossible to set apart biotypes with and without С. tanaitica genome. Another triploid biotype, С. taenia–С. tanaitica–Sabanejewia aurata (АТN), has in- stead of the genome of С. elongatoides the chromosomes of Sabanejewia aurata (Mezhzherin et al., 2014). S e x s t r u c t u r e a n d d i m o r p h i s m . As expected, almost all hybrid biotype speci- mens were female (table 3). The scarce males were found for ENN(N) and EEN95, making up 1.4 % and 0.8 % of the biotypes, respectively, and in ANT males were even more frequ- ent — 14.2 %. In diploid species males were also in minority. In C. taenia and C. elonga- T a b l e 2 . Species and biotypes structure of spined loaches’ populations Population ТT NN ЕЕ ЕN ЕNТ(NТ) ENN(N) ЕЕ(E)N ЕЕN95 ЕЕN95Т АNТ Lower Danube 8 21 15 70 237 Tisza Basin 34 6 Lower Dnipro 25 8 Sobok 38 6 Lake Babye 17 19 Irpin 21 37 122 8 Navaria 60 8 Styr 4 45 1 8 N o t e . Biotype designations are explained in table 1. 290 S. V. Mezhzherin, L. I. Pavlenko, A. O. Tsyba, T. V. Saliy, M. A. Ghazali toides they numbered slightly more than a third of the fish, and inС. tanaitica they were not found at all, presumably due to the small sample of the species. Analysis of body measurements and indices for the sexes of the two diploid species shows distinct sex dimorphism; males are smaller and their proportions are different. First of all, their fins are more elongated (table 4). Between-group PCA showed that males and females were different in 68 % cases in С. taenia by measurements and in 76 % by indices; in 56 % cases in C. elongatoides by measurements and in 87 % by indices. In both cases the most contribution belonged to total length and lP/CL, lV/CL, lA/CL, hA/CL (table 4). The dimorphism in females’ and males’ sizes and proportions requires that males be excluded from analysis since they would interfere with variance and discrimination patterns. V a r i a b i l i t y o f m o r p h o l o g i c a l p a r a m e t e r s . Most of the between-biotypes variation of absolute traits can be explained with one between-group principal component (bg-PC1). It described 97.1% of variation and was associated with the unidirectional change of all traits (table 5). The second component explained only 1.3 % of the total variability. The distributions of some biotypes and species in the space of the first two components overlapped to a large extent (fig. 3). Overall classification by absolute traits was 40.7 %. A similar situation is observed for the third and fourth components (fig. 4). Most of the between-group differences in indices were explained with four principal components (table 6), which explained 93.6 % of total variability. The main variation was mostly associated with head indices (bgPC1: ao/CL, hc/CL, c/CL, o/CL) and relative size of fins (bgPC2: lV/CL, hA/CL, lP/CL). All biotypes were overlapping in space of principal components (fig. 5). In total, only 28.2 % of specimens were correctly classified. D i s c r i m i n a t i o n a n d d i a g n o s t i c s o f C o b i t i s u s i n g b o d y m e a s u r e m e n t s a n d i n d i c e s . Analysis of both absolute measurements and indices showed that the grouping factor does have an effect (ANOVA, p < 0.001) on the variability of all traits except lA/CL, VA/CL, lcaud/CL (ANOVA, p > 0.01). However, due to significant transgression not a single trait (index) could be used in practice as a diagnostic tool to distinguish species and biotypes. T a b l e 3 . Sex ratios in spined loaches by species and biotypes Biotypes } { TT 108 57 EE 35 20 NN 8 0 EN 15 0 ENT(NT) 108 0 ENN(N) 69 1 EE(E)N 252 0 EE95 121 1 EEN95T 8 0 ANT 12 2 N o t e . Biotype names are explained in table 1. T a b l e 4 . Total length and body indices showed to be important for group-PCA. Intersex differences were significant (ANOVA, р < 0.01) Parameters С. taenia C. elongatoides males N = 57 females N = 108 males N = 20 females N = 35 M SD M SD M SD M SD L, mm 65.4 6.6 77.1 11.9 65.4 8.9 77.4 17.0 lP/CL 0.198 0.030 0.148 0.021 0.201 0.023 0.149 0.014 lV/CL 0.155 0.023 0.140 0.014 0.170 0.019 0.143 0.012 lA/CL 0.087 0.015 0.079 0.011 0.094 0.018 0.080 0.009 hA/CL 0.169 0.025 0.150 0.015 0.179 0.020 0.157 0.014 N o t e . N — sample size, M — mean, SD — standard deviation. 291Morphometric variation of hybridizing species and gynogenetic biotypes of spined loaches… Multivariate analysis of differences using aggregated traits in samples which had most specimens showed statistically significant heterogeneity (PERMANOVA: Danube, F = 11.3, df = (4, 339), p < 0.001; Irpin, F = 4.0, df = (3, 145), p = 0.003). Yet the factor of assigned biotype explained only a small part of the total variability: Danube partial R2 = 11.8 %, Irpin partial R2 = 7.7 %. Classification in the samples was also unsatisfactory. According to between-group PCA, Danube biotypes were classified correctly in 54 % cases, and the best identified biotypes were EE (9 out of 15, 60 %) and EE(E)N (149 out of 239, 62.3 %). In Irpin the total classification accuracy was 49 %, and the best classified was biotype ENT(NT) (21 out of 30, 70 %). As to the whole aggregate of traits, biotypes also did not differ strongly — overall classification accuracy was 54 %, although between-group diffe ren- ces were statistically significant: PERMANOVA: F = 23.8, df = (9, 725), p < 0.001. Excluding hybrid biotypes did not result in better classification. Between-group PCA allowed true identification of 35 % specimens. Interspecies differences by all traits were not statistically significant, PERMANOVA: F = 1.7, df = (2, 148), p = 0.132. One- dimensional analysis showed statistically significant differences in two traits describing the dorsal fin (lD/ CL, hD/CL; ANOVA, p < 0.001), and the species’ ranges significantly transgressed (table 7). T a b l e 5 . Loadings of the between-group principal components calculated for absolute traits to separate biotypes Variable bgPC1 bgPC2 Explained variance, % 97.09 1.32 L –0.191 –0.083 SL –0.190 –0.132 CL –0.202 –0.178 ao –0.216 0.491 o –0.220 0.213 c –0.213 0.111 po –0.233 0.175 hc –0.227 –0.055 H –0.263 –0.020 h –0.274 0.080 aD –0.190 –0.208 pD –0.208 0.007 aV –0.213 –0.166 aA –0.214 –0.153 lD –0.220 0.038 hD –0.202 0.210 lA –0.194 0.049 hA –0.164 0.173 lP –0.172 0.185 lV –0.147 0.209 PV –0.193 –0.494 VA –0.214 –0.195 lcaud –0.193 –0.271 N o t e . bgPC — between-group principal component. Fig. 3. 95 % confidence interval ellipses of the biotypes in the morphospace of bgPC1 and bgPC2 calculated for log10-transformed absolute traits. Each biotype means are marked with black points and names. The biotypes are explained in table 1. 292 S. V. Mezhzherin, L. I. Pavlenko, A. O. Tsyba, T. V. Saliy, M. A. Ghazali T a b l e 6 . Loadings of the between-group principal components calculated for indices Index bg-PC1 bg-PC2 bg-PC3 bg-PC4 Explained variance, % 41.08 25.87 16.43 10.23 L/CL –0.045 –0.107 –0.006 0.005 SL/CL –0.012 –0.107 –0.014 0.046 ao/CL –0.438 –0.109 0.220 0.030 o/CL –0.371 –0.059 –0.579 –0.076 c/CL –0.231 –0.029 –0.043 0.114 po/CL –0.277 0.110 0.175 0.068 hc/CL –0.146 0.133 –0.062 –0.011 H/CL –0.256 0.340 –0.025 0.243 h/CL –0.303 0.455 0.111 –0.126 aD/CL 0.022 –0.073 –0.102 0.019 pD/CL –0.096 0.039 0.147 –0.302 aV/CL –0.061 0.049 –0.099 0.181 aA/CL –0.057 0.079 –0.060 0.018 lD/CL –0.211 –0.034 0.050 0.626 hD/CL –0.229 –0.116 0.218 0.003 lA/CL –0.159 –0.159 –0.045 0.218 hA/CL –0.125 –0.330 0.171 –0.206 lP/CL –0.189 –0.349 –0.069 0.026 lV/CL –0.144 –0.509 –0.010 –0.072 PV/CL 0.166 –0.017 –0.238 0.349 VA/CL –0.022 0.107 –0.073 –0.150 lcaud/CL 0.077 –0.067 0.009 0.184 ao/c –0.207 –0.079 0.263 –0.084 o/c –0.140 –0.029 –0.536 –0.191 lcaud/aD 0.055 0.006 0.111 0.165 lcaud/hc 0.223 –0.200 0.071 0.195 Fig. 4. 95% confidence interval ellipses of the biotypes in the morphospace of bg PC3 and bgPC4 calculated for log10 transformed absolute traits. Designations the same as on fig. 3. A 293Morphometric variation of hybridizing species and gynogenetic biotypes of spined loaches… If hybrids alone were analyzed, classification was true for 54 % specimens. Intergroup differences using aggregated traits were statistically significant, PERMANOVA: F = 32.9, df = (6, 577), p < 0.001. M o r p h o l o g i c a l r e m o t e n e s s o f b i o t y p e s a n d s p e c i e s According to pair wise comparisons of biotypes (PERMANOVA), both body measurements and indices showed significant differences: 19 out of 45 measurement tests and 23 out of 45 indices (table 8). For 14 pairs the differences were statistically significant in both cases. Yet the link between biotype and trait was not strong; partial R2 ranged from 0.2 to 36.3 % (table 10). Between-group remoteness was estimated by Mahalanobis distances (table 9). For both measurements and indices, the Mahalanobis distances were small: the amplitude was 1.15–3.49 in the terms of standard deviations. Approximately unbiased bootstrap p-values Fig 5. 95 % confidence ellipses of the biotypes in the morphospace of four between-group principal components calculated for indices. Mean groups of each biotype is marked with black point and designation. T a b l e 7 . Mean values, standard deviation and ranges of some body measurements and indices Trait С. elongatoides N = 35 C. taenia N = 108 C. tanaitica N = 8 M SD Min–Max M SD Min–Max M SD Min–Max Ld/CL 0.110 0.010 0.091–0.131 0.101 0.012 0.057–0.130 0.098 0.008 0.086–0.111 hD/CL 0.198 0.014 0.144–0.226 0.183 0.020 0.108–0.235 0.192 0.013 0.171–0.214 pD/CL 0.448 0.038 0.350–0.508 0.451 0.028 0.368–0.517 0.488 0.017 0.468–0.518 Ao/c 0.436 0.042 0.357–0.517 0.410 0.035 0.321–0.500 0.412 0.041 0.348–0.464 pD, mm 25.1 4.4 14–36 24.9 6.9 16–37 29.1 2.2 26–33 CL, mm 55.4 8.8 39–77 55.3 12.7 35–76 59.8 5.4 54–70 N o t e . N — sample size, M — mean, SD — standard deviation, Min — minimum, Max — maximum. 294 S. V. Mezhzherin, L. I. Pavlenko, A. O. Tsyba, T. V. Saliy, M. A. Ghazali T a b l e 8 . Results of the pairwise PERMANOVA Pair of groups df2 Measurements Indices F (1, df2) p Partial R2, % F (1, df2) p Partial R2, % TT vs. EE 141 0.96 0.326 0.7 4.91 0.001 3.4 TT vs. NN 114 1.95 0.139 1.7 1.31 0.222 1.1 TT vs. ENT(NT) 214 2.37 0.103 1.1 1.71 0.110 0.8 TT vs. ENN(N) 175 26.41 0.001 13.1 6.85 0.001 3.8 TT vs. EE(E)N 358 156.97 0.001 30.5 10.06 0.001 2.7 TT vs. EEN95 227 3.21 0.034 1.4 10.3 0.001 4.3 TT vs. ANT 117 0.46 0.590 0.4 1.34 0.231 1.1 TT vs. EEN95T 114 0.44 0.622 0.4 1.36 0.223 1.2 TT vs. EN 121 7.71 0.009 6 2.08 0.058 1.7 EE vs. NN 41 1.06 0.318 2.5 1.75 0.112 4.1 EE vs. ENT(NT) 141 1.39 0.243 1 5.54 0.001 3.8 EE vs. ENN(N) 102 13.2 0.001 11.5 6.85 0.001 6.3 EE vs. EE(E)N 285 65.3 0.001 18.6 10.67 0.001 3.6 EE vs. EEN 154 1.53 0.198 1 7.64 0.001 4.7 EE vs. ANT 44 0.4 0.570 0.9 3.52 0.009 7.4 EE vs. EENT 41 0.35 0.607 0.8 2.48 0.023 5.7 EE vs. EN 48 4.14 0.042 7.9 3.56 0.006 6.9 NN vs. ENT(NT) 114 2 0.151 1.7 0.97 0.408 0.8 NN vs. ENN(N) 75 1.02 0.346 1.3 0.52 0.800 0.7 NN vs. EE(E)N 258 8.46 0.004 3.2 1.48 0.176 0.6 NN vs. EEN 127 3.19 0.055 2.4 2.36 0.029 1.8 NN vs. ANT 17 1.74 0.193 9.3 2.34 0.035 12.1 NN vs. EENT 14 1.3 0.261 8.5 2.17 0.042 13.4 NN vs. EN 21 1.5 0.198 6.7 0.61 0.778 2.8 ENT(NT) vs. ENN(N) 175 25.97 0.001 12.9 4.09 0.002 2.3 ENT(NT) vs. EE(E)N 358 156.55 0.001 30.4 13.28 0.001 3.6 ENT(NT) vs. EEN 227 2.26 0.113 1 11.45 0.001 4.8 ENT(NT) vs. ANT 117 0.23 0.777 0.2 1.42 0.183 1.2 ENT(NT) vs. EENT 114 0.65 0.462 0.6 1.63 0.106 1.4 ENT(NT) vs. EN 121 6.89 0.008 5.4 1.94 0.064 1.6 ENN(N) vs. EE(E)N 319 29.54 0.001 8.5 13.18 0.001 4 ENN(N) vs. EEN 188 39.92 0.001 17.5 16.1 0.001 7.9 ENN(N) vs. ANT 78 12.25 0.001 13.6 4.31 0.003 5.2 ENN(N) vs. EENT 75 6.58 0.004 8.1 2.94 0.013 3.8 ENN(N) vs. EN 82 0.7 0.586 0.9 2.08 0.065 2.5 EE(E)N vs. EEN 371 211.49 0.001 36.3 12.58 0.001 3.3 EE(E)N vs. ANT 261 34.32 0.001 11.6 3.66 0.009 1.4 EE(E)N vs. EENT 258 18.45 0.001 6.7 1.8 0.098 0.7 EE(E)N vs. EN 265 5.75 0.005 2.1 1.01 0.339 0.4 EEN vs. ANT 130 0.48 0.597 0.4 2.65 0.009 2 EEN vs. EENT 127 0.51 0.563 0.4 0.85 0.516 0.7 EEN vs. EN 134 11.2 0.002 7.7 2.6 0.008 1.9 ANT vs. EENT 17 0.34 0.622 2 1.07 0.367 5.9 ANT vs. EN 24 5.6 0.012 18.9 3.02 0.013 11.2 EETN vs. EN 21 4.3 0.029 17 2.06 0.052 9 295Morphometric variation of hybridizing species and gynogenetic biotypes of spined loaches… (AU-values) of the UPGMA clustering were high for EE(E)N and EN, EEN and EENT, TT and ENT(NT) clusters (fig. 6). There was no one-to-one mapping between morphologic distance and genetic remoteness. However, as expected, the most remote proved to be C. elongatoides and the hybrid of golden spined loach (ANT). T a b l e 9 . Mahalanobis distances between biotypes Distances by log10-transformed raw traits TT EE NN ENT(NT) ENN(N) EEN ANT EENT EE 2.00 NN 2.66 3.35 ENT(NT) 1.24 2.25 2.32 ENN(N) 2.29 3.04 1.44 1.94 EE(E)N 1.89 2.72 1.94 2.04 1.51 EEN 2.15 2.25 2.42 1.90 2.33 ANT 2.27 2.20 3.49 2.00 3.14 2.40 EENT 2.38 2.60 2.63 2.36 2.46 1.38 2.95 EN 2.19 2.98 1.87 1.97 1.52 2.22 3.29 2.57 Distances by log10-transformed indices TT EE NN ENT(NT) ENN(N) EE(E)N EEN ANT EENT EE 1.96 NN 2.66 3.32 ENT(NT) 1.24 2.23 2.31 ENN(N) 2.25 2.92 1.39 1.84 EE(E)N 1.73 2.45 1.81 1.82 1.48 EEN 2.12 2.25 2.37 1.89 2.19 1.94 ATN 2.27 2.17 3.49 1.99 3.10 3.01 2.38 EETN 2.37 2.59 2.61 2.36 2.37 2.08 1.37 2.95 EN 2.18 2.90 1.86 1.92 1.51 1.15 2.13 3.28 2.52 Fig. 6. UPGMA clustering of biotypes by Mahalanobis distances calculated for body measurements and indices separately. Rectangles bounds the clusters with more than 90 % AU-support. 296 S. V. Mezhzherin, L. I. Pavlenko, A. O. Tsyba, T. V. Saliy, M. A. Ghazali Discussion The data on variability of plastic traits which reflect body’s absolute parameters and proportions in spined loaches specimens of the complex Сobitis elongatoides–С. taenia– С. tanaitica, show that the analyzed hybrid forms and parental species often do differ sig- nificantly in the traits’ variability patterns in one- or multidimensional space. However, the differences are not significant enough that measurements and proportions could be used to draw a key for species and biotypes with 95–100 % true identifications. Also, the difficulties of identification lie not only in the impossibility to clearly distinguish hybrid forms which should have intermediate morphology relative to parental species, but also, surprisingly, parental species themselves. It pertains first of all to C. elongatoides and С. taenia, which are highly genetically differentiated (Šlechtova et al., 2000; Bohlen et al., 2002), unlike C. taenia and С. tanaitica which have insignificant genetic differences (Mezhzherin, Pavlenko, 2009). On the whole, the data on biotope variability in river systems of Ukraine correspond to results obtained for spined loaches from Poland (Kotusz, 2000, 2008). In the pooled sample of spined loaches, multidimensional analysis not only wiped the borders between species and biotypes; there was no logical structure to their morphological differentiation which a priori should correspond to the degree of genetic difference between the parent species and gene dosage in polyploids. The reasons for this incongruity could be indistinct species differences, absence of additive genetic interactions and the effect of the environment on plastic traits, which is evidently no less if not stronger than genes’ influence on the pheno- type. And thus geographical variability within a species masks interspecies variability. The difficulty with identification using plastic traits does not, theoretically, mean im- possibility to distinctly identify them using qualitative exterior parameters, such as the de- tails of Gambetta zones, shape and number of the melanistic spots at the base of the tail fin, structure of scales and the Canestrini’s organ (Kotusz, 2008). Yet in practice, the traits are not widely used for the C. elongatoides–С. taenia–С. tanaitica species group. Such a trait as the shape of Canestrini’s organ can be used only for males and so only for diploids; scale shape seems to be fairly subjective, and specificity in Gambetta zones is drowned by individual and geographical variability. Thus, distinguishing the species appears practically impossible. A similar situation arose in another diploid-polyploid complex of European freshwater fishes, Carassius (superspecies auratus). In Ukrainian waters it includes one parental species and three hybrid clone biotypes. The discrimination level in the group turned out to be quite low (Mezhzherin, Kokodiy, 2009). When a complex of plastic and meristic traits was used, it reached 86 %, and for plastic traits alone — 80 %, which is similar to resolution for spined loaches. The notion of twin species was designed in 1960s for outwardly indistinguishable but genetically and reproductively isolated taxonomical entities (Mayr, 1966). Recently, it sees more use as a metaphor but not a real scientific description. It turned out that for most of the so-called twin species, it is possible to discriminate them based on some exterior parameters and on distinct genetic specimen series. This means that the problem lies not in the absence of diagnostic traits as such but in the need to interpret them truly; whereas in spined loaches C. elongatoides–С. taenia–С. tanaitica the concept of twin species prob- ably found its pure embodiment. The species of the group do reliably differ in karyotypes, biochemical markers and DNA sequences, yet have no distinct morphological diagnosis. References Berg, L. S. 1949. Freshwater Fish of the USSR and adjacent countries. Part 2. Izd-vo Akad. Nauk SSSR, Leningrad, 470–1382 [In Rassian]. Bohlen, J., Rab, P. 2001. Species and hybrid richness in spined loaches of the genus Cobitis (Teleostei: Cobiti- dae), with a cheklist of European forms and suggestions for conservation. J. Fish. Biol., 59 (1), 75–89. 297Morphometric variation of hybridizing species and gynogenetic biotypes of spined loaches… Bohlen, J., Rab, P., Šlechtova, V., Rabova, M., Ritterbusch, D., Freyhof, J. 2002.Hybridogenous biotypes in spined loaches (genus Cobitis) in Germany withi mplications for conservation. In: Colares-Pereira, M. J., Cowx, I. G., Coelho, M. M., eds. Conservation of f reshwater fishes: options for the future. Wiley-Blackwell, 311–321. Boroń, A. 2003. Karyotypes and cytogenetic diversity of the genus Cobitis (Pisces: Cobitidae) in Poland: a review. Cytogenetic evidence for a hybrid origin of some Cobitis triploids. Folia Biol., 51. (suppl.), 49–54. Chen, Y., Sui, X., Liang, N., , Y. 2015. Two new species of the genus Cobitis Linnaeus from southern China. Chinese J. Oceanol. Limnol, 34 (3), 517–525. Culing, M. A., Janko, K., Boroń, A., Vasil’ev, V. P, Cote, I., Hewitt, G. M. 2006. European colonization by the spined loach (Cobitis taenia) from Ponto-Caspian refugia based on mitochondrial DNA variation. Mol. Ecol., 15 (1), 173–190. Fish base. https://www.fishbase.de/summary/FamilySummary.php?ID=127 Janko, K., Kotlik, P., Rab P. 2003. Evolutionary of asexual hybrid (Teleostei: Cichlidae) inferred from phyloge- netic analysis mitochondrial DNA variation. J. Evol. Biol., 16 (6), 1280–1287. Janko, K., Culing, M. A., Rab, P., Kotlik, P. 2005. Ice age cloning — comparison of the Quaternary evolutionary histories of sexual and clonal forms of spiny loaches (Cobitis, Teleostei) using the analysis of mitochon- drial DNA variation. Mol. Ecol., 14 (10), 2991–3004. Janko, K., Flajšhnanas, M., Choleva, L. et al. 2007. Diversity of European spined loaches (genus Cobitis L.,): an update of the geographic distribution of the Cobitis taenia hybrid complex with a description of new molecular tools for species and hybrid determination. J. Fish. Biol., 71 (2), 387–408. Kotusz, J. 2000. Intra- and interpopulation morphological variability in variability in diploid and varied-ploidy Cobitis from Poland. Fol. Zool., 49 (Suppl. 1), 219–226. Kotusz, J. 2008. Morphological relationships between polyploidy hybrid spined loaches of the genus Cobitis (Teleostei: Cobitidae) and their parental species. Ann. Zool., 58 (4), 891–905. Mayr, E. 1966. Animal species and evolution. Belknap Pres, Harvard Universyty, 1–797. Mezhzherin, S. V., Chudakorova, T. Yu. 2002. Genetic structure of diploid-polyploid spined loaches complex Сobitis taenia (Cypriniformes: Cobitidae) of the Middle Dnieper basin. Rus. J. Genet, 38 (1), 86–92. Mezhzherin, S. V., Kokodiy, S. V. 2009. Morphological variability and differentiation of the goldfish, Caras- sius auratus, and Prussian carp, C. gibelio, (Cypriniformes, Cyprinidae) in reservoirs of Ukraine. Vestnik Zoologii, 48 (1), 39–50 [In Russian]. Mezhzherin, S. V., Pavlenko, L. I. 2007. A case of hybridization in loaches (Osteichthies: Cobitidae: Cobitis) determing genetic instability and expansion. Cytol. Genet., 41 (4), 218–225. Mezhzherin, S. V., Pavlenko, L. I. 2009. Spined loaches(Cypriniformes: Cobitidae: Cobitis) of Ukraine waters, genetic boundaries of the species and natural hybridization. Uzhhorod University Scientific Herald. Ser. Biol., 25, 146–154 [In Russian]. Mezhzherin, S. V., Pavlenko, L. I. 2010. Genetic diversity, оrigin and distribition tendencies of polyploid spined loaches (Cypriniformes, Cobitidae, Cobitis) within Ukraine. Cytol. Genet., 44 (5), 65–77. Mezhzherin, S. V., Pukhtayevych, P. P., Tsyba, А. А. 2014. Absorbing hybridization of Cobitis taenia and Sa- banejewia aurata (Cypriniformes, Cobitidae) in water reservoirs of Northern Ukraine connected with diploid-polyploid complex formation. Vestnik Zoologii, 48 (6), 503–510. Mitteroecker, P., Bookstein, F. 2011. Linear discrimination, ordination, and the visualization of selection gradi- ents in modern morphometrics. Evol. Biol., 38 (1), 100–114. Mousavi-Sabet, H., Vatandoust, S., Esmaeili, H. R., Geiger, M. F. Freyhof, J. 2015. Cobitis avicennae, a new species of spined loach from the Tigris River drainage (Teleostei: Cobitidae). Zootaxa, 3914 (5), 558–568. Oksanen, J., Blanchet, F. G., Kindt, R. et al. 2018. Vegan: community ecology package. R package version 2.5–3. https://cran.r-project.org/web/packages/vegan Peacock, F. C., Bunting, S. L., Queen, K. G. 1965. Serum protein electrophoresis in acrilamyde gel patterns from normal human subjects. Science, 147 (3664), 1451–1455. Perdices, A., Bohlen, J., Šlechtová, V., Doadrio I. 2016. Molecular evidence for multiple origins of the European spined loaches (Teleostei, Cobitidae). PLoS ONE, 11, 1. doi:10.1371/journal. pone.0144628 Pravdin, I. F. 1966. Guide to the study of fish. Food Industry, Moscow, 1–376 [In Russian]. Rab, P., Raboba, M., Bohlen, J., Lusk S. 2000. Genetic differentiation of the two hybrid diploid–polyploid com- plexes of loaches, genus Cobitis (Cobitidae) involving C. taenia, C. elongatoides and C. sp. in the Czech Republic: Karyotypes and cytogenetic diversity. Folia Zool., 49 (Suppl. I), 55–66. R Core Team. A language and environment for statistical computing. 2017. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Schlager, S. 2017. Morpho and Rvcg — Shape Analysis in R. In: Zheng, G., Li, S., Szekely, G., eds. Statistical Shape and Deformation Analysis. Academic Press, 217–256. Sezaki, K., Kobayashi, H., Nakamura, M. 1977. Size of erythrocytes in the diploid and triplod specimens of Carassius auratus langsdorfi. Jap. J. Ichthyol., 24 (2), 135–140. Šlechtova, V., Luskova, V., Šlechta, V. et al. 2000. Genetic differentiation of two diploid–polyploid complexes of spined loach, genus Cobitis (Cobitidae), in the Czech Republic, involving C. taenia, C. elongatoides, and C. spp.: allozyme interpopulation and interspecific differences. Folia Zool., 49 (1), 67–78. 298 S. V. Mezhzherin, L. I. Pavlenko, A. O. Tsyba, T. V. Saliy, M. A. Ghazali Suzuki, R., Shimodaira, H., Suzuki, M. R. 2015. Package ‘pvclust’. R topics documented, 14. https://cran.r- project.org/web/packages/pvclust Ueno, K., Iwai, S., Ojima, Y. 1980. Karyotypes and geographical distribution in the genus Cobitis (Cobitidae). Bull. Japan Soc. Sci. Fish, 46 (1), 9–18. Ueno, K., Ojima, Y. 1976. Diploid–tetraploid complexes in the genus Cobitis (Cobitidae, Cyprinida). Proc. Japan Acad., 52, 446–449. Vasil’eva, E. D., Vasil’ev, V. P. 2012. Cobitis amphilekta sp. nova, a new species of spined loaches (Cobitidae, Cypriniformes) from the Caspian Sea Basin. J. Ichthyol., 52, 3, 200–206. Vasil’ev, V. P. 1985. Evolutionary karyology of fishes. Nauka, Moscow, 1–299 [In Russian]. Vasil’ev, V. P., Vasil’eva, E. D. 1998. Sibling species in the genus Cobitis (Cobitidae) Cobitis rossomeridionalis sp. nova. J. Ichthyol., 38 (8), 580–590 [In Russian]. Wheeler, B., Torchiano, M., & Torchiano, M. M. 2016. Package ‘lmPerm’. R package version, 2.1.0. https:// cran.r-project.org/web/packages/lmPerm Received 22 March 2019 Accepted 25 August 2020