Gnatholepis Bleeker, 1874

Larson, Helen K. & Buckle, Duncan J., 2012, A revision of the goby genus Gnatholepis Bleeker (Teleostei, Gobiidae, Gobionellinae), with description of a new species, Zootaxa 3529 (1), pp. 1-69 : 7-12

Gnatholepis Bleeker, 1874

Stenogobius (Gnatholepis) Bleeker, 1874: 318 , type species Gobius anjerensis Bleeker, 1851 , by original designation and monotypy.

Gnatholepis — Koumans 1931: 86–87; Goren 1979: 42; Yoshino in Masuda et al. 1984: 251; Hoese 1986: 790; Maugé 1986: 369; Birdsong et a l. 1988: 188; Parenti and Thomas 1998: 270; Brito and Miller 2001: 257; Larson 2001: 42; Randall and Greenfield 2001: 2; Larson and Murdy 2001: 3592; Thacker and Cole 2002: 840; Murdy and Hoese 2003: 1791; Larson and Wright 2003: 127; Thacker 2004a: 573 ; Larson and Buckle 2005: 67.

Diagnosis. First dorsal fin spines VI; second dorsal fin rays I,10–12; anal fin rays I,11–12; pectoral fin rays 14–19; lateral scale count 26–31; transverse scale count backward (TRB) 9–11. Head and body somewhat compressed, snout short, steep and may be rounded, mouth rather small, subterminal to inferior, with snout often partly overhanging tips of jaws; lower lip with thin triangular flap or fold ventroposteriorly (greatly reduced in two species). Eyes set high on head, orbit partly above dorsal profile. Opercle and preopercle covered with scales in all but two species (no scales present on side of head).

Lateral canal pores on head with anterior nasal pore just anterior to anterior naris, posterior nasal pore beside each posterior naris, pair of anterior interorbital pores, single posterior interorbital pore, a postorbital pore, a terminal pore over opercle and anterior and posterior temporal pore in separate posterior portion of oculoscapular canal over opercle; three preopercular pores present ( Fig. 1 ). Transverse sensory papillae pattern on head, papilla rows short; with four short, often broken, vertical rows on cheek below eye, row d in two sections, row b may be absent or reduced; ot row on opercle usually in two sections and os and oi rows short; in some specimens ( Fig. 1 ); e rows along lower jaw forming two or three short transverse rows near chin.

Anterior eighth to one-third of first gill slit closed by membrane. Inner face of upper limb of first gill arch (to lesser extent in other arches), covered with low dense papillae which may form clumps or groups; dorsal portion of arch may have short fleshy protuberances ending in one or several papillae; outer faces of gill filaments covered with fine papillae, especially along axis of filament (papillae most pronounced on inner face filaments of first arch and on both faces of filaments of remaining arches). Tongue short, fleshy, bilobed. Anterior naris in short tube, may have small flap posteriorly; posterior naris low, rounded. Pectoral girdle (edge of cleithrum) smooth, with no fleshy knobs or lobes. Pelvic fins fused, oval; margin of fraenum fimbriate, fringed with many fine pointed lobes or thin lappets. Genital papilla in males small, narrow and slender, narrowing to pointed tip; papilla in females short, fleshy and rounded.

Characteristic colour pattern with black mark atop eyeball, dusky to black line from eye crossing cheek to rictus, dark and/or light spot or blotch above pectoral base; body pale with variable pattern of rows of dark lateral blotches, spots and fine lines.

Dorsal pterygiophore pattern 3-12210; 26 vertebrae (10 precaudal and 16 caudal), rarely 10+15 or 10+17; usually two epurals (usually one in G. ophthalmotaenia ), two anal fin pterygiophores present before first caudal haemal spine; neural spines of first few vertebrae slender, pointed; fifth ceratobranchials narrowly triangular, interdigitated posteriorly along median line and with high flange ventrally, dorsally covered with long slender teeth; metapterygoid short, relatively small, with short triangular dorsal process, not contacting or forming bridge to quadrate; palatine stout, broadened anteriorly, not reaching quadrate; pterygoid small, relatively slender, less than half the length of palatine, in contact with quadrate; quadrate somewhat forked; three or four ossified gillrakers; scapula unossified; first epineural inserting on parapophysis of first vertebra.

Remarks. Having the fifth ceratobranchials medially interdigitated (especially posteriorly) is a character that Gnatholepis shares to some degree with Stenogobius (in which the fifth ceratobranchials are partly interdigitated in a few species; character not widely reviewed) (Larson 2001). In Evorthodus lyricus (an Atlantic gobionelline) these bones are joined but not interdigitated ( Parenti & Thomas 1998; HKL pers. observ.). Most gobioids have these bones separate. Oxyurichthys , Oligolepis and Gobionellus (Pezold & Larson in prep.). In Stenogobius the frenum is smooth to finely crenulate (HKL pers. observ.).

Watson (1992) illustrated the small fleshy papillae along the axes of each gill filament in Awaous and considered that the feature was unique to that genus. The papillae covering most of the gill filaments and undersides of gill arches in Gnatholepis greatly resemble those present in species of Awaous . Stenogobius laterisquamatus also possesses fine papillae along the gill filaments, though papillae are more sparse and absent from the gill arches (HKL pers. observ.). We found that these fleshy papillae are present along the gill arches and gill filaments of all species of Gnatholepis . This feature bears further observation.

All but two species of Gnatholepis possess a small triangular flap or fold in the end of the lower lip near the rictus; its function is unknown. The two small species Gnatholepis argus and G. gymnocara lack this structure, as do most gobioids.

Gnatholepis has the dorsal portion of the gill arch with one or more short fleshy protuberances extending into the oral cavity and ending in one or several papillae. They greatly resemble those fleshy lobes present in sand-dwelling gobiines such as Valenciennea and Amblygobius ( Hoese & Allen 1977: fig. 4). The condition in Awaous is similar, but it has a broad fleshy pad covered with small papillae (similar to those covering the gill filaments).

Gnatholepis species have characteristic variably developed black blotches or lines over the dorsal part of the eye. The shape and arrangement of these markings often seem to be species-specific (see species accounts). Additionally, the upper half (at least) of the eyeball is known to fluoresce under red light in at least one species of Gnatholepis ( Michiels et al. 2008) ; this feature has not yet been widely surveyed. The black marking on the dorsal surface of the eye may be part of inter-species communications.

Results of CAP analysis of Gnatholepis . The canonical analysis of principal co-ordinates (CAP) on seven Gnatholepis species ( G. anjerensis , G. caudimaculata n. sp., G. cauerensis , G. knighti , G. opthalmotaenia , G. pascuensis and G. thompsoni ; Table 3) is illustrated in Figure 2 . CAP axes 1 and 2 explain 62% of the total variation within the 13 (m = 13) canonical axes required to best discriminate between the species. In Figure 2 , G. anjerensis occupies the uppermost region of the CAP plot separating from other species along axis 2 by differences, primarily, in the proportion of: head length to standard length, preorbital width in head length and jaw length in head length. Gnatholepis thompsoni , G. cauerensis and G. pascuensis occupy the right-hand side separating most along axis 1 primarily by increased pectoral ray counts and lateral scale counts. Gnatholepis caudimaculata , G. ophthalmotaenia and G. knighti occupy the left-hand side separating along axis 1 primarily influenced by greater proportions of head and eye widths in the head length. Although the seven species cluster into different areas in ordination space, there are no distinct separations illustrated by the CAP plot, although separation of some species could occur along other canonical axes given that 38% of the canonical variation is not illustrated in Figure 2 .

The leave-one-out allocation success is based on the entire 13 canonical axes and showed 64% (183/285) overall correct allocation. However, the allocation success of individual species varies considerably (Table 3). The only distinct species shown is G. pascuensis which achieved 100% correct allocation of the five samples. Gnatholepis thompsoni had a good classification rate of 81.6% whilst G. opthalmotaenia and G. anjerensis had reasonable classification success (73.8% and 71.2%, respectively). Gnatholepis knighti , G. cauerensis and G. caudimaculata n. sp. had lower classification success rates (61.9%, 49.4% and 48% respectively) (Table 3). These results support the CAP illustration ( Figure 2 ) that Gnatholepis species have subtle differences in their morphometric counts and proportions and that the effectiveness of the selected variables to discriminate species varies. Most importantly, meristic counts and morphometric proportions do not provide clear taxonomic identification, with the exception of G. pascuensis (a very small sample size).

Several Gnatholepis species have distinctive features and/or geographic distributions that can be used to distinguish, or isolate, that species from others. These species include G. caudimaculata n. sp., which has a distinctive vertical oval blackish to dark grey blotch across the caudal fin base and is known from the Red Sea and Persian Gulf; G. opthalmotaenia has distinctive black spots on the anal and dorsal fins and breast scales that reach forward to the rear margin of the preopercle (with an Indo-West Pacific distribution); and G. pascuensis has high pectoral fin ray counts (18–19) with a distinctive pattern of two rows of five elongate dense black blotches on the side of the body and is known only from Rapa. Removal of these three distinguishable species from the CAP analysis reduces unnecessary misclassification of the other species which do not share these distinctive features (for example 18, 13 and 6 specimens were incorrectly assigned as G. caudimaculata n. sp., G. opthalmotaenia and G. pascuensis respectively (see Table 3)).

Re-analysis of the remaining four species ( G. anjerensis , G. thompsoni , G. cauerensis and G. knighti ) using CAP results in a small improvement to the overall misclassification success (69% or 144/209, Table 4; Fig. 3 ). Gnatholepis cauerensis and G. knighti have the lowest classification success (57.5% and 66.7% respectively) (Table 4). Gnatholepis cauerensis is most commonly misclassified with G. thompsoni (Table 4: 18 samples; G. thompsoni is restricted to the Atlantic) and G. knighti (12 samples; restricted to Hawaii), while G. knighti is most

Analysis of meristics and proportions using the CAP analyses procedure above shows that the seven Gnatholepis species analysed differed subtly in body shape and/or scale or fin ray counts, with only one species, G. pascuensis , distinctly different. Characteristic body/fin pigment and geographic location were identified by the authors to distinguish three species: G. caudimaculata n. sp., G. opthalmotaenia and G. pascuensis . Four species ( G. anjerensis , G. thompsoni , G. cauerensis and G. knighti ) remain problematic for reliable identification using meristic counts, morphometric proportions, body and fin pigment and geographic location. The authors have used all the features above (as well as the 'gestalt' of the specimens as an identification tool) when sorting specimens of Gnatholepis to species, as a number of the museum lots examined contained more than one species. While current information does not reliably and easily distinguish all species, it is possible that a deeper understanding of live specimen colour and genetic differences in combination with taxonomic information may provide better identification of species within Gnatholepis .

Figures

FIGURE 1. Sensory pores and papillae in Gnatholepis anjerensis, SAIAB 2112, Mauritius, 46 mm SL male. Scales omitted.

FIGURE 2. Canonical analysis of principal co-ordinates plot for the seven Gnatholepis species with vectors overlaid. CAP axes 1 and 2 explain 62% of the variation from 13 canonical axes.

FIGURE 3. Canonical analysis of principal co-ordinates plot for the four Gnatholepis species without reliably distinctive colour pattern, with vectors overlaid. CAP axes 1 and 2 explain 64% of the variation from 14 canonical axes.