Comments on taxonomical status of three gobiid nominal species (genus Gobius ) described from Brazilian coast (Teleostei: Gobiiformes: Gobiidae)

. Three species described and currently included in the genus Gobius from the Brazilian coast are herein discussed. Gobius uranoscopus Sauvage is regarded as junior synonym of Bathygobius soporator based on information on holotype deposited in Muséum National d’Histoire Naturelle, Paris; G. silveiraemartinsi Ihering, known only in Rio Grande do Sul was erected based on a short description, and type material is not available, but I considered that it is junior synonym of Ctenogobius shufeldti; Gobius boekeri Ahl, 1931, described from the coast of Pará, Brazil, is a junior synonym of Gobionellus oceanicus, as both taxa are virtually identical in meristic characters and presumable morphometric differences that Ahl used to substantiate the validity of his new species actually fall into the morphometric range observed in G. oceanicus specimens and may be attributable to individual variation and to the image quality by Ahl’s original description.


INTRODUCTION
Family Gobiidae (gobies) is the most diverse group of acanthomorph fishes, with at least 189 genera and more than 1,700 species (Nelson et al., 2016), and has a pivotal role in coral reef, marine and brackish environments. Besides its extraordinary diversity, the taxonomy of gobies is not yet satisfactorily known, with several new species being discovered each year and many others awaiting description. The Brazilian EEZ has reported 28 genera and 51 species to date (Moura et al., 2003).
Gobius was erected by Linnaeus (1758) to include seven species he had diagnosed based on fused pelvic fin rays, forming a disk (Pinnae ventrales unitae). In prior studies on gobiids, most described taxa were put into this genus (including taxa described from the Brazilian coast), which then became a large wastebasket of poorly defined groups, although many of such taxa have been allocated to other genera (for a historical perspective of gobiid taxonomy, see Koumans, 1931). Today the genus Gobius refers to taxa circumscribed to Baltic, Black, Mediterranean and North Seas, and eastern Atlantic as well (for more information, see Pinchuk, 1976Pinchuk, , 1977Miller, 1986), and thus other species were removed and an immense wealth of new genera have been created. On the other hand, three nominal species of Gobius were described from the Brazilian coast: Gobius uranoscopus, Sauvage, 1882;G. silveiraemartinsi Ihering, 1893;and G. boekeri Ahl, 1931, but they have not been treated since then, so these names continued valid despite their real taxonomic status remained unknown to date. This study aims to discuss the taxonomy of the abovementioned nominal species.

MATERIALS AND METHODS
Descriptions of morphological data, whenever not mentioning the original description, are provided according to Murdy & Hoese (2003) and Tornabene et al. (2010). Institutional abbreviations are in accordance with Eschmeyer and Fricke (2015). Abbreviations in the text are as follows: SL = standard length; HL = head length; D. = dorsal fin elements (spines and rays); A. = anal fin elements (spines and rays); P. = pectoral fin rays; ll = scales on longitudinal row; GR = gill rakers (upper + lower). Unknown data from the original description or for lack of type material for each nominal species is indicated along with the text by question marks ("???").
Morphometric data on species were taken mainly from photographs and images. Comparative material of Gobionellus oceanicus deposited in Museum of Zoology, University of São Paulo, Brazil was also examined. The only information on type locality of this species is the term "Brésil", written in the original description. The MNHN's website informs that holotype would have been taken in 10°S-30°W, but this is probably wrong since such coordinates are located outside the Brazilian EEZ.
The original description of Gobius uranoscopus is not much detailed, but counts of scales in longitudinal row (98 scales) are clearly wrong since only Sicydium plumieri, among western Atlantic gobiids, possesses so many scales on flanks (Watson, 2000). Also, G. uranoscopus does not have long, comb-like teeth on upper jaw and five curved teeth on each side of lower jaw as in Sicydium. Further examination on images of the holotype revealed this specimen has about 42 scales on longitudinal row over flanks.
In MNHN's website, Gobius uranoscopus was recognized as an unidentified Bathygobius, which is confirmed herein based on its head morphology and the presence of upper pectoral fin rays ( Fig. 1) as detached silk-like filaments (Miller & Smith, 1989;Murdy & Hoese, 2003 This species has been described based on two specimens taken in southern Brazilian coast, (Camaquã, Rio Grande do Sul state). The type material of this species is unknown and the original description is poorly detailed and without any image. Ihering stated G. silveiraemartinsi specimens present seven spines in first dorsal fin, 2 spines and 11 rays in second dorsal fin, two spines and 11 rays in first anal fin, and 15 pectoral fin rays. The presence of seven spines are typical of gobiid subfamily Gobiosomatinae (Birdsong, 1975), but no member of this subfamily encountered in Brazil possesses only 15 pectoral fin rays, fin counts provided by Ihering are possibly erroneous, and, evidently, Gobius silveiraemertinsi probably is not a gobiosomatinae, since all gobiosomatines reported from southern Brazilian coast present different values of dorsal, anal and pectoral fins, such as Microgobius meeki, Evermann & Marsh (D. VII+11-12; A. 9-10; P. 17-21; Birdsong, 1981), and Gobiosoma hemygymnum (Eigenmann & Eigenmann, 1888) Fischer et al., 2011;Malabarba et al., 2013). Gobius silveiraemartinsi differs from Awaous tajasica by having larger eye (25% of HL vs. less than 20% of HL), more than 11 anal fin rays, and pigmentation without black oblique lines behind the pectoral fin base (present in Awaous; Watson, 1996); from Evorthodus lyricus by the absence of two round black blotches on the caudal fin base (present in E. lyricus; Murdy & Hoese, 2003); from Gobioides broussonnetii by having large eye (about 25% HL vs. 5% or less; Murdy, 1998); from Gobionellus oceanicus by possessing about 15 pectoral rays (vs. 16-20) and presumably less than 14 dorsal fin rays and 15 anal fin rays (vs., respectively, 14 and 15 rays); from G. stomatus by the absence of vertical dark markings on flanks (Pezold, 2004), and from Ctenogobius boleosoma by not presenting humeral black blotch (Ginsburg, 1932;Gilbert & Randall, 1979).
Gobius silveiraemartinsi is mostly similar to Ctenogobius shufeldti in pigmentation, particularly in the presence of "Von Augen zu Munde laufen zwei kurze braunen Streifen, über den Backen ein dritter längerer (p. 34)"; the presence of two dark brown stripes from eye to upper jaw, the largest one on cheek, is typical of C. shufeldti (Fig. 2) and it is shared only with C. sagitulla from eastern Pacific (Ginsburg, 1932). Regardless of the disagreement on counts of first dorsal fin spines between these taxa, G. silveiraemartinsi is herein regarded as junior synonym of Ctenogobius shufeldti.
Gobius boekeri Ahl, 1931 syn. nov. (Fig. 3 Holotype of this species was taken in Soure, Marajó, in the state of Pará, Brazil, but this type specimen, which would have been sent to Zoologisches Museum, Berlin (ZMB), is presumably lost. On the other hand, the description Ahl provided for this species is very straightforward and included a decent illustration of the type specimen. The author considered his new species quite similar to Gobionellus oceanicus, but argued his taxa differ from that species by having a more elongated body and body depth larger on dorsal fin origin. However, further examination of the original description and image compared with the material of Gobionellus oceanicus (see also Pezold, 2004) revealed that body depth and other morphometric values fall into the range, which is observed for G. oceanicus (Table 1). Apparent differences between these taxa may be attributable to individual variation. Another possibility is that the original image, although already stated herein as of reasonable quality, might be somewhat distorted and it is not so detailed, thus making measurements difficult and rendering differences in some body proportions such as caudal peduncle length.
Gobius boekeri is remarkably similar to Gobionellus oceanicus in all meristic characters (e.g. dorsal and anal fin counts, scales counts), and also in many features as presence of ctenoid scales on trunk, black round umeral blotch below first dorsal fin ray, and dark vertical stripe below eye, thus G. boekeri is herein regarded as a junior synonym of Gobionellus oceanicus.