Description of a new species of Alburnus Rafinesque, 1820 (Actinopterygii, Cyprinidae, Leuciscinae) from the Kolpa River in the Sava River system (upper Danube drainage), with remarks on the geographical distribution of shemayas in the Danube

Abstract Alburnus sava, new species, is described from the Kolpa River. The Kolpa is a tributary of the Sava, a major tributary of the Danube River, in the Black Sea basin. Alburnus sava is distinguished from its congeners in the Danube drainage, A. mento and A. sarmaticus, by having 23−27, usually 24−26, gill rakers; the ventral keel usually completely covered by scales (scaleless part maximum 15% of the keel length); 15−16, mode 15, branched pectoral-fin rays; the length of the gill raker at the junction of the arch limbs 65−70% of the length of the opposite outer gill filament; and a relatively long lower jaw (37−40% HL or 112−130% interorbital width). Alburnus sava is a large-sized potamadromous shemaya known to occur in the entire Sava drainage. The taxonomic status of A. mento and A. sarmaticus is confirmed. Alburnus danubicus is discussed and as there are no new arguments, it is kept as a valid species. New details on the distribution of shemayas in the Danube drainage are presented.


Introduction
Shemaya is a Russian vernacular name based on the Persian name shah-mahi, King's fish, used for A. chalcoides Gueldenstaedt, 1772 in the Caspian Sea basin. It is also commonly applied to a number of nominal species and subspecies in the Caspian and the Black seas since Borodin (1904) and Berg (1905Berg ( , 1916. Shemayas were later assigned to a distinct genus Chalcalburnus Berg, 1932 and discriminated from Alburnus by a usually scaled (almost completely or completely covered by scales) ventral keel between the pelvic fins and the anus (vs. usually scaleless in Alburnus) and an increased number of lateral-line scales, usually over 55 (vs. usually less then 55 in Alburnus, with the exception of A. filippii Kessler, 1877 with up to 63) (Berg 1932(Berg , 1933. Recent morphological studies Kottelat 2007a,b, Özuluğ and supported the idea that shemayas form a distinct morphological group of species though included in the genus Alburnus Rafinesque, 1820. Not all shemaya species have been studied using molecular markers but preliminary data support the conclusion that they form a monophyletic group (Perea et al. 2010, Geiger et al. 2014. Earlier authors (e.g., Kessler 1877, Berg 1916, 1923, 1932, 1933, 1949, Abdurakhmanov 1962, Tscherbukha 1965, Elanidze 1983, Movchan and Smirnov 1983) emphasized an extremely high similarity of shemayas in the few characters used in their discrimination − usually only the number of branched anal-fin rays and gill rakers, and some body measurements. For example, a reliable range of the branched anal-fin rays based on large samples examined of the Caspian shemaya is 12−17 (=12½−17½ according to the method of counting by Kottelat and Freyhof (2007) with means 14.6 (Berg 1932) and 14.7 (Abdurakhmanov 1962). The Black Sea forms are characterised by higher numbers of rays, averaging 15.6 in the Don drainage (Berg 1932), 15.9 in the South Bug (Tscherbukha 1965), 15.2 in the Berda River, Sea of Azov basin (Loshakov 1973) with a range of 14−17 (=14½−17½) (i.e. within the range in the Caspian shemaya). However, there are also local forms in the Black Sea basin with a lower number of branched anal-fin rays. For example, the Colchis (or Georgian) shemaya from Western Transcaucasia has a range of branched anal-fin rays of 12−16 (=12½−16½), usually 13−15, and a mode of 14 (Berg 1949, Elanidze 1983. However, later studies Kottelat 2007a,b, Özuluğ and, based on a larger set of characters, discriminated 16 species of shemayas. They consider the following nominal taxa as valid in the Black Sea basin (clockwise from the Danube): A. danubicus Antipa, 1909 (anadromous), A. mento (Heckel, 1836) (landlocked), A. sarmaticus Freyhof & Kottelat, 2007b (anadromous and potamadromous), A. mentoides Kessler, 1859 (resident), A. leobergi Freyhof & Kottelat, 2007b (originally anadromous, also landlocked populations in dammed rivers), A. derjugini Berg, 1923 (resident), A. istanbulensis Battalgil, 1941 (resident), A. schischkovi (Drensky, 1943) (anadromous), and A. mandrensis (Drensky, 1943) (probably, landlocked). Shemayas distributed outside of the Black Sea basin include A. volviticus Kottelat, 2007a andA. vistonicus Freyhof &Kottelat, 2007a, both from lakes in eastern Greece, Aegean Sea basin (landlocked); two species from lakes in the Sea of Marmara basin in Turkey, A. carinatus Battalgil, 1941 (landlocked) and A. nicaeensis Battalgil, 1941 (landlocked, possibly extinct); two species in the Aegean Sea basin in western Turkey, A. attalus Özuluğ & Freyhof, 2007 (resident) and A. battalgilae Özuluğ & Freyhof, 2007 (resident), and A. chalcoides from the Caspian Sea basin (anadromous).
However, this taxonomic scheme still does not answer all remaining questions for the group. For example, it is not clear which species occurs between the ranges of A. derjugini and A. istanbulensis (e.g., in Sakarya, Kızılırmak, and Yesilirmak rivers). It is still to be clarified if there is (was) a single species in the Çoruh as supposed by Berg (1923) or there occurred two species or morphs, resident and anadromous as found by Deryugin (1899). Furthermore, the identification of shemaya in coastal lakes and limans in Bulgaria and Romania, other than Lake Madra, remains unknown.
Current knowledge of the morphology of the Black Sea shemayas is inadequate as the keys offered to distinguish them (Kottelat and Freyhof 2007) are often not helpful since most character ranges overlap. This may be the reason why a taxonomic compromise was offered (Parin et al. 2014) to consider only two shemaya species as valid, the Caspian A. chalcoides and the Black Sea A. mento.
As to the Danube drainage, as indicated above, Freyhof and Kottelat (2007b), based on literature and their own data, assumed an occurrence at least in former times, of two sympatric anadromous shemayas, one lake-dwelling species landlocked in a number of subalpine lakes, and one probable resident river species. The latter was only known from the Kolpa [Kupa] River and only three specimens were available for examination for Freyhof and Kottelat (2007b: 223). These authors found the Kolpa shemaya morphologically similar to A. sarmaticus, but differing in the number of gill rakers (27 vs. 28−34 in 28 individuals from the South Bug and the lower Danube). A morphological analysis of additional material from the Kolpa is important in resolving the taxonomy of the Danubian shemayas. Herein, we examine morphological variation of shemayas and provide a critical review of their historical distributions in the Danube.

Material and methods
The triangular-shaped symphisis of the lower jaws is referred as the chin. The ventral keel is defined as the distance between the base of the anus and the level of the posterior ends of the pelvic-fin bases. The dorsal-fin insertion is the posterior-most point where the last dorsal-fin ray connects with the body. All measurements were made point-topoint with a dial caliper and recorded to the nearest of 0.1 mm. Methods for counting rays, lateral-line scales and scales along the scaleless part of the ventral keel, and for most measurements follow Kottelat and Freyhof (2007). Additional measurements of the cranium, jaws, and operculum as defined in Tables 1 and 3 were made point to point from the anteriormost extremity to the posteriormost extremity (lengths), from the uppermost extremity to the lowermost extremity (depths), and between the lateralmost extremities (widths). Length of the cranial roof was measured from the anterior margin of the supraethmoid to the base of the supraoccipital crest. Vertebral counts are given according to Naseka (1996). Standard length was measured from the tip of the upper jaw to the posterior margin of the hypurals. Head length was measured from the anteriormost extremity of the head (either tip of the upper jaw or the projected symphysis of the lower jaw) to the posterior opercular margin that includes the skin fold. In total, 49 morphometric indices were used for descriptions and statistical analyses as in Tables 1 and 3. These also included the length of the scaleless portion of the ventral keel relative to the keel length (Bogutskaya et al. 2010, fig. 1) and length of a gill raker at the junction of the lower and the upper limbs of the first gill arch (the 6−8th, usually the 7th, gill raker) relative to the opposite outer gill filament (Kottelat and Freyhof 2007, fig. 42). The last two rays in the dorsal and anal fins based on a single pterygiophore were counted as 1½ rays. Numbers of vertebrae and fin rays were counted from radiographs. In total, 16 meristic characters were examined (nine as given in Table 2 and numbers of branched dorsal-fin rays, branched pelvic-fin rays, intermediate vertebrae, total lateral scales, total lateral-line scales, scales above the lateral line, and scales below the lateral line). All characters were obtained from specimens of both sexes and combined in analyses and tables.
Cluster Analysis (CA), Multidimensional Scaling (MDS), Principal Component Analysis (PCA), Discriminant Function Analysis (DFA), and a Kruskal-Wallis test which is helpful for comparison of three and more groups with a presumably nonparametric distribution of variables, were performed using STATISTICA 6.0 and PRIMER v6.1.9 to identify the most important characters that contribute to the differentiation of samples and visualise the degree of morphological separation among the new species, A. mento, A. sarmaticus, and A. leobergi.    Diagnosis. Alburnus sava sp. n. is distinguished from all other species of Alburnus in the Danube drainage by having 23−27, usually 24−26, gill rakers; the ventral keel usually completely scaled (scaleless maximum 15% of the keel length); 15−16 (mode = 15) branched pectoral-fin rays; the length of gill raker 65−70% of the length of the opposite outer gill filament; and a relatively long lower jaw (37−40% HL, 112−130% interorbital width).

Alburnus
Description. The general appearance of Alburnus sava sp. n. can be seen in Figure 1. Relative measurements are provided in Table 1. Variation in ten (of the 16) examined meristic characters is provided in Table 2. The largest specimen, a spent female, is 218 mm SL.   Table 2E.
Scales along scaleless portion of ventral keel 0 1 2 3 4 5 6 7 8 9 10 11 12 13 mean A. sava sp. n., n=13 8* 1 4 0.7 A. sarmaticus, South Bug, n=5 3 2 1.4 A. sarmaticus, Danube, n=15 1 5 5 2 2 2.9 A. leobergi, n=6 1 4 1 2.3 A. mento, n=50 1 5 7 6 9 11 3 3 2 2 1 7.2 As the examined samples are rather small in number of specimens and contain individuals with a wide range in SL, Table 1 also presents the range and mean for the holotype and size groups separately. Body depth at the dorsal-fin origin in the 218 mm-long specimen represented 27% SL and considerably exceeded the range in body depth of smaller specimens (20−23% SL). The same is found for the depth of the caudal peduncle, 10% SL (60% length of caudal peduncle or 1.7 times in its length) vs. 9% (46−54% length of caudal peduncle or 1.9−2.2 times in its length), respectively. However, head length and eye diameter are clearly negatively allometric. The head length mean in the small-size group (63−80 mm SL) is 25.8% SL vs. 22.7% SL in specimens 174 and 218 mm SL; in smaller specimens the head length considerably exceeds the body depth while in the largest specimen it is much smaller than the latter. The eye diameter mean is 7.5% SL (29% HL) in the small-sized group (63−80 mm) vs. 5.1% SL (22.5% HL) in 218 mm long specimen. The mouth is upturned and the mouth cleft is straight. The tip of the mouth is about at a level with the upper margin of the pupil. The lower jaw is long, its length 112−130% interorbital width. The chin is variably developed (Fig. 1). The holotype has a well developed chin while the chin of the paratype is smoothed and slightly projected.
The ventral keel between the pectoral-fin bases and the anus is well pronounced but not sharp and usually completely covered by scales (in 8 specimens, including the holotype) or scaleless (exposed) for 1−2 scales only (Table 2), reaching up to 15% of the keel. Dorsal fin with 3 unbranched and 8½ branched rays. Anal fin with 3 unbranched and 15½ or 16½ branched rays ( Table 2). Origin of anal fin located on (in three specimens) or slightly behind the vertical of the dorsal-fin insertion (Fig. 2).
Total lateral-line scales number (61) No nuptial tubercles in the examined material. Four dissected individuals were females. Overall colouration is silvery with no orange or red pigment at fin bases and no faint dark midlateral stripe in both freshly caught and preserved specimens.
Distribution and habitat. The species is currently known from the Kolpa River drainage, a tributary of the Sava River in the upper Danube drainage, Black Sea basin (Fig. 4). Alburnus sava sp. n. is a potamadromous species, occurring in streams and rivers with moderate to rapid current and a gravel and cobble bottom; in spring, during the spawning season, the species migrates upstream to smaller tributaries to shallow riffles where they spawn.
Etymology. The species name refers to the Sava River. A noun in apposition. Vernacular name. Local names are bucov or velika pliska in Croatian and Serbian, pegunica in Slovene.

Comparisons
The new species belong to the Alburnus mento group of shemayas (former genus Chalcalburnus), which includes A. mento, A. sarmaticus, and A. leobergi, as defined by Frey- hof and Kottelat (2007b) to have 57−62 lateral-line scales to the posterior margin of hypurals and 15−16½ branched anal-fin rays. The maximum size of Alburnus sava sp. n. in examined materials is 218 mm SL (adult female), which is about the body size of mature adults of migratory shemayas in the Black Sea basin, including the Danube − ca. 200−300 mm (vs. 100−130 mm in resident small-sized A. derjugini and A. mentoides) (Drensky 1943, Berg 1949, Tscherbukha 1965, Movchan and Smirnov 1983. Alburnus sava sp. n. does not demonstrate a distinct difference in most morphometric characters from all or either of the other Danubian species and A. leobergi (Tables 1, 3; discussed below regarding statistical analysis). The same is true for numbers of vertebrae (total and in vertebral regions), branched anal-fin rays, and scales in the lateral row, total lateral line and the lateral line to the posterior margin of hypurals) ( Table 2).
However, Alburnus sava sp. n. can be clearly distinguished from A. sarmaticus (Table 2) using two meristic characters. First, the number of gill rakers is 23−27 in A. sava sp. n. vs. 28−34 in A. sarmaticus; this same range for A. sarmaticus was found by Freyhof and Kottelat (2007b: table 3). A wider range in number of gill rakers, 26−33 (mean = 30), was reported by Tscherbukha (1965) for specimens from the South Bug. Second, the ventral keel of A. sava sp. n. is usually completely covered by scales, while in A. sarmaticus the typical state is a scaleless keel along about 10−15% its length (i.e., 1−3 scales exposed (Fig. 5a) found in 16 of 20 specimens). Among specimens of A. sarmaticus examined, neither a completely scaled keel nor a keel exposed along 6 scales (4−6 indicated by Freyhof and Kottelat (2007b) was found.
Though measurements based on the limited material examined cannot be used for taxonomic purposes with regard to being variable depending on season, sex, size and other factors, some relative measurements may have some taxonomic value as reflecting basic morphological differences between these species. Alburnus sava sp. n. differs from A. sarmaticus by a longer upper jaw, 28−31% HL (vs. 24−29% HL); a longer lower jaw, 37−40% HL or 112−130% interorbital width (vs. 34−38% HL or 89−113% interorbital width); and a longer cranial roof, 14−16% SL (vs. 12−15% SL) (Tables 1, 3). Freyhof and Kottelat (2007b) described A. sarmaticus as having 17½ or fewer branched anal-fin rays as an anadromous shemaya in the Danube; they described A. danubicus, a purportedly extinct species from the Danube, as having 17−20½ branched analfin rays; this was based on their opinion of data of Antipa (1909) and Drensky (1943) for an anadromous shemaya from the lower Danube. However, no specimens have been reported since then with 18½ or more branched anal-fin rays and some authors (Halasi-Kovacs andHarka 2012, Stefanov andTrichkova 2015) continue to use danubicus for the Danubian shemaya as the only anadromous species in the Danube. Materials collected by Shishkov [=Chichkoff] and Drensky, still deposited in the National Museum of Natural History in Sofia (Bulgaria), should be examined to clarify the problem.
When compared to A. mento, A. sava sp. n. with 15½ or 16½ branched anal-fin rays do not differ from the former species possessing 13−18½ branched anal-fin rays ( Depth of caudal peduncle (% SL) 9.2 10.3 9.5 0.5 8.8 10.5 9.6 0.4 7.9 9.7 9.0 0.6 9.5 10. Dorsal-fin base length (% SL) 11.7 12.1 11.9 0.2 9.6 12.4 11.1 0.7 9.0 11.6 10.6 0.7 10.4 11.7 11.0 0.5 Snout length (% SL) 6.6 7.2 6.9 0.2 6.0 6.8 6.4 0.2 5.6 7.6 6.5 0.7 6.0 6.9 6.6 0.    Alburnus sava sp. n. and A. mento are distinguished in a much clearer way by the length of the scaleless portion of the ventral keel and the relative length of a gill raker. As shown by Freyhof and Kottelat (2007a, b), these characters have a considerable taxonomic value in the genus Alburnus. In A. sava, the ventral keel is completely scaled or there is a very short scaleless portion of the keel which is only 1 or 2 scales long, maximum 15% of the keel length (the shortest keel among the examined species,   Table 2). In contrast, A. mento has a very long scaleless portion of the keel which is 3 to 13, usually 4−12, scales long (Fig. 5b) that is 25−85% of the keel length (up to 90% according to Freyhof and Kottelat (2007b).
Alburnus sava sp. n. further differs from A. mento by the shape of the symphysial part of the jaws. Neither a pronounced chin nor a knob formed by the symphysis of the lower jaws were found in the new species. In contrast, Alburnus mento of a different size have a very prominent chin and a strong knob on the lower jaw entering a corresponding notch formed at the symphysis of the upper jaws, a feature also typical of the asp Leuciscus aspius.
Note on syntypes of A. mento. A comment should be given clarifying the status of the specimens labelled as syntypes of A. mento in NMW. The original description (Heckel 1836: 225−226) clearly indicates three samples of specimens it was based on. All three samples are still present among specimens labeled as syntypes in NMW. The set of Acquisition Sheets (bound in a number of books at present) is the most reliable source of primary original information which accompanied the samples at the moment of their accession in the collection. The accession information for one more sample labeled as syntypes, NMW 55629, does not match the original description. It contains four specimens registered under the acquisition numbers 1835.VI.1 (2 specimens) and 1835.VI.1a (2 specimens) which say that the specimens were received by exchange from [Dr] Schreibers in 1832 and 1834, respectively, from near Stift Kremsmünster (Kremsmünster Abbey; in the town of Kremsmünster in the district of Kirchdorf an der Krems, Upper Austria, 29 km from Gmunden on Traunsee). However, the source of the acquisition may mean any water body in the surroundings of Kremsmünster where an astronomical and geophysical observatory ("Mathematical tower"), the first in the world, was located which included a natural history museum with botanical and zoological collections (still existing as a weather station). However, the label (from Steindachner's time) says «Traunsee» (probably just based on the original description). As Kremsmünster was not mentioned in the original description by Heckel (1836), even if the specimens are from Lake Traun, they do not belong to the Traunsee sample collected by Heckel in 1824. So, the NMW 55629 specimens do not belong to the type series and the designation of the 150 mm long (SL) specimen from this sample as lectotype by Freyhof and Kottelat (2007b: 217) is invalid.

Statistical analysis
Results of PCA and MDS analyses applied to 16 meristic variables outlined three clusters (Fig. 8). A. sava sp. n. is most distant from A. mento. Alburnus sarmaticus and A. leobergi are comparatively close; specimens of these two species cannot be distinguished based on their meristic characters.
Based on the PCA, the most influential variables are the number of gill rakers and the number of scales along the scaleless portion of the ventral keel−these variables' contributions based on covariances are over 0.24 (Factor 1: 0.668885, Factor 2: 0.316944 and Factor 1: 0.241753, Factor 2: 0.434431, respectively) vs. less than 0.1025 for all other variables. A Kruskal-Wallis test revealed four (including the two mentioned above) characters different on a statistically significant ( To reduce the number of morphometric indices (47 in total, as in Tables 1 and 3) to be used in a DFA, a Kruskal-Wallis test was applied and 10 characters were excluded as not significantly different (p>0.01%) between the samples. Among other 37 characters, a PCA revealed 15 variables whose contribution to either of Factors 1 and 2 or both was the highest (over an arbitrary fixed threshold of 0.01 based on the number of the used variables and their variance); these characters are given in bold in Tables 1 and  3. These 15 indices were used for a DFA (Fig. 9) that revealed three groups corresponding to A. sava sp. n., A. mento, and A. sarmaticus + A. leobergi.  As a further step, the samples of A. sarmaticus and A. leobergi were combined as being indistinguishable by the studied morphometric and meristic characters (Figs 8,9) and a DFA was performed based on 19 most contributing characters (4 meristic and 15 morphometric ones as discussed above) (Fig. 10); predicted classifications for A. sava sp. n., A. mento, and A. sarmaticus + A. leobergi were 100%-correct. DFA statistics values are as follows: Wilks' Lambda 0.00618, approx. F (38, 54) =16.661, p<0.0000, which indicate almost perfect discrimination. Alburnus sava sp. n. is the most distant from A. mento (Squared Mahalanobis Distance equals 103.99); A. sarmaticus+A. leobergi is much closer (54.11) but still significantly distant. The most removed, by the characters considered, are A. mento and A. sarmaticus+A. leobergi (107.02). Partial Lambdas demonstrate the unique contribution of the respective variable to the discriminatory power of the whole model, and the most significant (Partial Lambda <0.9) for discrimination of the samples under consideration were number of scales along scaleless portion of ventral keel (0.481705), cranium width between margins of sphenotics (% cranium roof length) (0.545483), cranium width between margins of pterotics (% cranium roof length) (0.803940), number of gill rakers (0.810206), number of branched pectoral-fin rays (0.830673), number of scales above lateral line (0.845142), length of lower jaw (% interorbital width) (0.878840), and body depth at dorsal-fin origin (% SL) (0.886565).

Discussion on distribution of shemayas in the Danube and adjacent areas
Shemayas are reported in the Danube from the upper reaches down to the delta (Fig. 4).
In the upper Danube, a shemaya, commonly identified as A. mento, was known to occur (or still occurs) in lakes in Germany in systems of the Isar (Starnberger See [Würmsee]) and the Inn (Chiemsee, Simsee, and Waginger See), in Austria in the Traun River system (Traunsee, Attersee, Mondsee, Hallstättersee, Grundsee, Wolfgangsee, probably Fuschlsee and Irrsee) and the Drava [Drau] system (Wörthersee [Vrba]), and in Slovenia in the upper Sava (Bled [Veldeser See, Blejsko Jezero]) (Freyhof and Brunken 2004, Freyhof and Kottelat 2007b, Glowacki 1896, Heckel 1836, Honsig-Erlenburg and Petutschnig 2002, Patzner et al. 1993). Most of these lakes belong to Danubian tributaries on the northern slope of the Alps. Lake Wörthersee is located much further south, on the northern slope of the Karawankas mountain ridge (one of the south-eastern ridges of the Alps) and belongs to the Gurk [Krka] River in the upper Drava system. Alburnus mento was found in the lake, in the inflowing Reifnitzbach and the outflowing Glanfurt River (Honsig-Erlenburg and Petutschnig 2002). Freyhof and Kottelat (2007b) found no morphological difference between A. mento inhabiting Lake Wörthersee and A. mento in the lakes of the Isar and Inn river systems. This means that A. mento has a rather geographically fragmented range and it cannot be excluded that a shemaya which was historically present in Lake Bled (Glowacki 1896, Munda 1926 in the very upper reach of the Sava River belonged to A. mento. No museum samples of the Bled shemaya are known to exist to check this hypothesis. The lake morph (lake phenotype) from different lakes likely have a polyphyletic origin as found for lake Coregonus (e.g., Kottelat 1997, Douglas et al. 1999, Østbye et al. 2006). However, a problem of distinguishing between single (homology) and multiple (homoplasy of ecomorphological traits) origins of the distinct lake phenotype (= A. mento) and its taxonomic implication is beyond the scope of this paper.
In the Danube in Germany, Austria, Hungary, and Croatia downstream to the confluence with the Drava [Drau], one collection specimen (now lost) was known from near Vienna (NMW 55630, a syntype); this specimen may represent the anadromous or potamadromous Danubian species rather than the landlocked A. mento (Berg 1932, Freyhof andKottelat 2007b). We examined the other NMW specimen labeled as A. mento from the Danube at Lobau near Vienna (NMW 54831) but it is a Leuciscus aspius. Shemayas were also known to be found occasionally in the Hungarian section of the Danube downstream of the Austrian section (Hankó 1931) but are extirpated (Halasi-Kovacs and Harka 2012). It was also mentioned as a fish occurring in a lake in the Maros [Mureș] river system (a tributary of the Tisza) (Herman 1887).
Based on records of Glowacki (1896) and Munda (1926), most subsequent authors gave the upper Sava and the upper Drava as areas of historical distribution of shemaya (e.g., Taler 1953, Sabioncello 1967, Povž 2012. Povž and Sket (1990) supposed its occurrence in the Mura [Mur] (a tributary of Drava), the entire Drava, and the lower course of the Sava. Povž et al. (1998) presented a first reliable record (with voucher specimens and a photo) of a shemaya from the Sava system. Several specimens were collected in the Kolpa [Kupa] River, a right tributary of the Sava, between Dol and Sodevci on August 11, 1996, between Dečina and Radenci on September 10, 1996, and up and below the Radenci dam on September 11, 1996.
In Croatia, a shemaya is reported from the Drava, Sava, Kupa [Kolpa] and their tributaries (Bogut et al. 2006, Mrakovčić et al. 2006, Jelić et al. 2016 but no voucher specimens are known except those from the Kolpa. However, it cannot be excluded that the authors based their statements on reliable data from local fishery sources. Further south and eastwards, in Bosnia and Herzegovina, a shemaya is found in the lower Sava and its tributaries Una, Sana, Vrbas, Ukrina, Tinja, Bosna (upper section with tributaries and from Žepče to the confluence with the Sava), and Drina (at Ljubovija and from Loznica to the confluence with the Sava) (Vuković and Ivanović 1971, Vuković 1977, 1982, Kosorić et al. 1980, Hađiselimović and Hamzić 1999, Mikavica and Savić 1999, Drešković et al. 2011, Simonović et al. 2015, Jelić et al. 2016. In Serbia, it is reported from the Danube down to the Iron Gates [Đerdap] at the border with Romania (Janković 1996, Simonović 2001) and from the Zapadna Morava River between the Ovčar Banja and Međuvršje reservoirs (Veljović et al. 1985) and in its tributary Ibar (Šorić 2009).
Formally, the name A. sava sp. n. is applied herein only for the shemaya found in the Kolpa but we suppose that this potamadromous fish inhabits (or inhabited in the past) the entire Sava River system and, probably, the Danube upstream from the Iron Gates being geographically separated from the anadromous shemaya of the lower Danube (A. danubicus and A. sarmaticus).
Right below the Iron Gates (a 134-m-long gorge on the main stream of the Danube dammed in 1964 and1977), shemaya were known in Romania (Bănărescu 1964) and the entire Bulgarian stretch of the Danube with tributaries (e.g., Iskyr, Vit, Osym, Yantra) where it still occurs though rare (numerous publications reviewed by Stefanov and Trichkova (2015). Records of shemaya in the lower section of the Danube (downstream from Cernavodă) in Romania and Ukraine were also numerous (e.g., Drensky 1943, Bănărescu 1961, 1964, Movchan and Smirnov 1983. So, historically, shemaya in the Danube were distributed in almost the entire drainage. The length of the Danube and its major tributaries makes it reasonable to suppose that the drainage was historically populated by an anadromous shemaya (entering the river for spawning from the sea) and a resident potamadromous shemaya. At present, it is difficult to assume how far the anadromous form(s) or species used to migrate upstream in the Danube. All anadromous shemayas in the Black Sea and the Sea of Azov basins spend most of the year foraging in coastal sea areas, estuaries, and limans with salinity up to 10−12‰ and wintering in deep places of lower reaches of rivers. Before damming of rivers, they migrated upstream to reach spawning grounds at a distance from tens to hundreds of kilometers depending on the size of the river and availability of grounds suitable for spawning (riffles with gravel bottom and rapid flow). Shemaya were known to migrate up to Pervomaysk in the South Bug, the Dnieper River Rapids (upstream of the town of Zaporizhia) in the Dnieper, and the Oskol River in the Siversky Donets system of the Don River (Movchan and Smirnov 1983). It looks quite likely that the anadromous shemaya in the main Danube migrated upstream to at least the Iron Gates gorges. Our data support a taxonomic conclusion of previous authors (Bănărescu 1961, 1964, Tscherbukha 1965, Loshakov 1973, Freyhof and Kottelat 2007b) that the South Bug and Dnieper and the Danube are populated by one and the same species.
No materials were available for us to clarify a border between the ranges of A. danubicus, A. sarmaticus, and A. sava sp. n. This question can only be solved when specimens from the mainstream Sava and its tributaries in Croatia, Bosnia and Herzegovina, and Serbia are studied as well as specimens from the Danube below its confluence with the Sava. This task cannot be easily implemented because of the rarity of the fish; though, in some tributaries of the Sava in Bosnia and Herzegovina it is still often recorded (Jelić et al. 2016).