Romanogobio Bănărescu, 1961 is a genus of bottom-dwelling, rheophilic gudgeons with a wide distribution in temperate Eurasia. Six species were reported from the Danube basin (Kottelat and Freyhof 2007, Friedrich et al. 2018). Among them, Romanogobio antipai was described, as Gobio kessleri antipai, by Bănărescu (1953: 300, 318) based on a series of syntypes from the Danube delta at Sulina (12 specimens collected by Grigore Antipa before 1909) and the lower reaches of the Danubian tributary Argeş (one specimen collected by Băcescu). In the same paper, representing a study of morphometric features within the “Gobio kessleri” group of populations distributed in Romania, Bănărescu also recognized a new form, Gobio kessleri kessleri natio banaticus (the name is not available from this publication, but available as Gobio kessleri banaticus from Bănărescu 1960: 121) and compared both new forms with the nominotypical subspecies. As the correct original spelling of the specific name is kesslerii (Kottelat 1997), we use it hereafter.

Bănărescu (1953) distinguished Romanogobio antipai from R. kesslerii and R. k. banaticus on average values of the postorbital distance (10.5% of body length vs 8.5–10%), barbel length (10.5–13% of body length vs 8–11.5%), eye diameter (5.5% of body length vs 5.6–6.4% R. excluding kesslerii (Dybowsky, 1862) from Bulgaria with 4.8–5.6%, and 75.2% of interorbital distance vs 81.9–98.5%), snout length (9.4% of body length vs 9.4–11.7%), and maximum body depth (18.4% of body length vs 1.8–17.7% excluding R. kesslerii from the Dniester with an average of 18.5%). Consequently, no clear differences between the three taxa were presented, and later Bănărescu (1960, 1961) reported the occurrence of specimens morphologically intermediate between R. antipai and R. kesslerii in the Danube tributaries Ialomiţa, Argeș (Dâmboviţa) and Siret (Bizau and Milcov Rivers), recognized as such also by Bănărescu and Nalbant (1973). Bănărescu (1961, 1999: 151) repeated the most typical features of R. antipai from the Danube delta (smaller eye and deeper body), adding smaller body length in adults (“apparently not exceeding 6 cm”), commonly 4 scales (vs commonly 3) between the lateral line and the pelvic-fin origin, caudal-peduncle width at the anal-fin origin commonly about equal to caudal-peduncle depth (vs larger in kesslerii), and short lateral blotches (vs commonly elongated in banaticus). The distribution range was widened (Bănărescu 1961: 344) to include the lower reaches of the Siret River and its tributaries, the Milcov, Putna, and Birlad Rivers, and the lowest reaches of the Argeş and Ialomiţa Rivers. Similar data were later published in the book on the fishes of Romania (Bănărescu 1964: 454–455, fig. 195).

Bănărescu (1992, 1994a, 1999: 150, fig. 21) again restricted the range of R. antipai to the lowest reaches of the Danube, mentioning that before 1959, it was distributed upstream to the Argeş River mouth (some 430 river kilometres). He emphasized that it markedly differed from other members of the R. kesslerii complex by its morphological features, which were related to dwelling in deep water of the main stream of the river.

Bănărescu (1992, 1994a, 1999) treated the Danube delta gudgeon as a subspecies (Gobio kesslerii antipai) but mentioned that it deserved the rank of species. Kottelat and Freyhof (2007) considered it as a valid species and Friedrich et al. (2018) found no arguments to reject this status.

The sample of Romanogobio antipai from the Danube delta at Sulina, described by Bănărescu based on G. Antipa’s collections (before 1909) was not the only one from the Danube delta. Smirnov (1971) provided meristic and morphometric data on a sample of 24 specimens collected by him in April 1961 in the Ukrainian part of the delta (Chilia Arm) near Izmail and identified it as R. kesslerii. This sample was later included in the book on Ukrainian fishes by Movchan and Smirnov (1981: 344, tab. 181). Bănărescu (1999: 158) suggested that Smirnov’s (1971) specimens from Izmail belonged to R. antipai because they had a deeper body and a smaller eye.

The most recent published record of R. antipai from the lower section of the Danube may be that of Marinov (1978) who reported Gobio kesslerii from the main course of the Danube in Bulgaria. The true R. kesslerii only occurs in middle reaches of tributaries and has not been recorded from the deep main stream of the Danube (e.g., Chichkoff 1937, Mihailova 1970, Dikov et al. 1994).

The absence of recent records of R. antipai lead some authors to the conclusion that the species might be extinct (Bănărescu 1994b, Kottelat 1997). Its conservation status was later evaluated as extinct (Kottelat and Freyhof 2007, Freyhof and Kottelat 2008) because it was supposed that all known ichthyological surveys conducted since the 1960s to 2003 in suitable habitats in the Danube delta had failed to find it; however, this statement is not entirely correct as there is a specimen in Natural History Museum ‘Grigore Antipa’ collected in 1992 (described below).

In 2016, TS collected a gudgeon specimen in the Bulgarian sector of the Danube main stream using a bottom drift net at a depth of 8 m, near the village of Vetren (river kilometer 395). This specimen (Figs 1–2) was preliminarily identified by AN as Romanogobio antipai because of its small eye and 8½ branched dorsal-fin rays distinguishing it from both R. kesslerii and R. vladykovi (Fang, 1943). The present note is devoted to a comparative description of the specimen to test this hypothesis.

Methods for counting fin rays and scales, and for measurements, follow Kottelat and Freyhof (2007) except that head length, postorbital length, and interorbital width include the skin fold. In the examined samples, standard length is used for some relative measurements. Body length, which is the length to the posterior margin of the posterior-most scale on the base of the caudal fin (called standard length II by Holčík et al. (1989: fig. 12), is also measured and the data are compared for diagnostic characters taken from the literature. All measurements were made point-to-point with a dial caliper and recorded to the nearest 0.1 mm. Vertebral counts taken from radiographs follow the scheme by Naseka (1996). To avoid probable discrepancy in lateral-line count, we provide not only the number of lateral-line scales to the posterior margin of the hypurals but also numbers of total lateral scales and total lateral-line scales. Statistical analyses were done using Microsoft Excel, Statistica 6.0 (Statistic for Windows. Statsoft; Discriminant Functional Analysis, DFA), and SPSS Statistics V23.0 (IBM SPSS; Cluster Analysis, CA).

General appearence of the NMNHS presumed R. antipai specimen from the Danube at Vetren is shown in Figs 1–2 together with a syntype and a non-type R. antipai specimen in comparison to R. kesslerii and R. vladykovi. Counts, descriptive states of the pectoral-fin length, and measurements are presented in Tables 1, 2. Examined character states in this specimen coincide considerably with those in the three type specimens of R. antipai and five non-type specimens, and demonstrate its differences from the samples of both the R. kesslerii species complex and R. vladykovi. As the standard length averages 96.8% of the BL (calculated in the Romanogobio material examined in this study), the difference between relative measurements (in % SL and in % BL) is slight and the morphometric character states that have been considered as diagnostic for R. antipai vs. R. kesslerii are confirmed. They include maximum body depth, 19–25% SL (17–25.5% of BL; in parentheses, data from Bănărescu (1953, 1961, 1999) and Movchan and Smirnov (1981) are summarized); caudal peduncle depth, 8–9% SL (7–9% of BL) and 35–38.5% of caudal peduncle length; eye diameter, 5–6% SL (5–6% of BL), 20–23% HL (18–24% HL), and 59–68.5% of interorbital width (61–81% of interorbital width). The NMNHS specimen has 6 and 4 scales, respectively, above (to the dorsal-fin origin) and below (to the pelvic-fin origin) the lateral line similar to the three type specimens of R. antipai and the topotypical specimen from Sulina thus confirming the opinion (Kottelat and Freyhof 2007, Friedrich et al. 2018) that this character is one of the most dependable diagnostic characters for the species. All other counts are identical or very close in the NMNHS specimen and R. antipai examined in this study (Table 1). None of the R. antipai specimens had 7½ branched dorsal-fin rays thus confirming its main difference from R. vladykovi characterized by 7½ branched dorsal-fin rays (Naseka et al. 1999; Naseka 2001).

The three examined samples of Romanogobio kesslerii s.l. demonstrate a statistically significant difference in ten morphometric and five meristic characters (Table 3) but the ranges of character values overlap considerably and the number of specimens is small. We combined all specimens in a single sample in order to estimate general ranges of character values without a special analysis of variation within the R. kesslerii complex.

As can be seen from Table 2, examined specimens of R. antipai including the NMNHS specimen from the Danube at Vetren, most clearly (with a gap or ranges only slightly overlapping) differ from both R. kesslerii and R. vladykovi by the caudal peduncle depth (35–38.5% caudal peduncle length vs 25–33 and 39–50, respectively), a wider head (58–64% HL vs 50–59), a smaller eye (5–6% SL and 59–67.5% interorbital width vs 6–8 and 75–130), and a wider interorbital space (32.5–36% HL vs 19.5–31; 8–9% SL vs 5–8) with shallow orbital notches (Fig. 2a–c). Romanogobio antipai can be further distinguished from R. kesslerii, besides the number of scales above and below the lateral line (6 and 4, respectively, in all examined R. antipai vs commonly 5 and 3 in R. kesslerii s.l.), by a deeper body (19–25% SL vs 16–19), a deeper caudal peduncle (8–9% SL vs 6–8), and a deeper head (59–69% HL vs 53–59).

Besides morphometric characters mentioned above, all examined specimens of R. antipai including the NMNHS specimen can be clearly distinguished from R. vladykovi by the number of branched dorsal-fin rays, 8½, in contrast to 7½ found in all specimens of R. vladykovi examined in this study. Naseka (2001: 111) mentioned that 8½ rays can be rarely found in R. vladykovi; a revision of his primary data (radiographs) revealed a single specimen with 8½ branched dorsal-fin rays out of 46 examined. Romanogobio antipai further differs from R. vladykovi by the vertebral structure (Table 1, 2) having abdominal+caudal counts 19+21 or 20+21, which means that the caudal region is longer than the abdominal region vs. commonly (in 52 out of 70 specimens) 20+20 or 21+21 or variants with a caudal region shorter than the abdominal one.

A DFA (Fig. 3) showed differentiation of the three groups of samples identified as R. antipai, R. kesslerii and R. vladykovi (the number of unbranched dorsal-fin rays was excluded from the analysis as demonstrating zero variability within the groups) and the groups were 100% classified as predicted (Table 4). A CA (Fig. 4) supported the grouping.

To conclude, the analysis confirmed previously reported discriminating character states (number of branched dorsal-fin rays, relative size of the eye and the interorbital space, relative depth of the caudal peduncle) and introduces a new character (vertebral counts) for discriminating Romanogobio antipai from R. kesslerii and R. vladykovi. However, relative taxonomic status of these three species still waits for a phylogenetic analysis based on molecular data. It cannot be excluded that R. antipai is a deep-water ecophenotype of either R. kesslerii or R. vladykovi. The new record demonstrates that R. antipai is still extant in the lower Danube but at present can only be found at a greater depth in the main channel and the deltaic branches. Currently classified as Extinct using IUCN criteria, the conservation status of Romanogobio antipai needs revision, in light of the new record from 2016.