Copda, 2004(2), pp. 260-280 Phylogenetic Analysis of the Genus Gobionellus (Teleostei: Gobiidae) FRANK PEZOLD A cladistic analysis of the gobiid fish genus Gobionellus primarily using characters of the postcranial axial skeleton and the cephalic lateralis system gave evidence that the genus as historically conceived is polyphyletic. Its present recognition relies upon characters common to many species of gobionelline gobies. One group of six spe- cies is most closely related to the genus Gobioides. This group includes Gobionellus oceanicus and retains the name Gobionellus. Gobionellus is diagnosed by an extensive oculoscapular canal running from the snout to above the rear margin of the oper- culum with a unique A'BCDFHKL' pore pattern, a distally flared fourth neural spine that is spatulate in five of the six species, a vertical row of neuromasts on the rear field of the operculum, and elongate gill rakers on the anterior surface and lobes on the posterior surface of the epibranchial of the first gill arch. No unequivocal synapomorphies are offered for the genus excluding Gobioides. Fifteen species pre- viously assigned to Gobionellus are more closely related to species in the genera Oxyurichthys, Oligolepis, and Evorthodus. These species are removed to the resurrect- ed genus Ctenogobius of which Ctenogobius fasciatus is the type species. Ctenogobius is diagnosed by an abbreviated oculoscapular canal that terminates above the preo- perculum with an A'BCDFH' pattern, a simple or triangulate fourth neural spine, a diagonal posterior opercular neuromast row, and a lack of lobes or gill rakers on the anterior surface of the first epibranchial. The lack of lobes or gill rakers on the anterior surface of the first epibranchial is synapomorphic for the genus. One spe- cies originally placed in Gobionellus, Oxyurichthys stigmalophius, exhibits two synapo- morphies diagnostic of Oxyurichthys?a transversely bifid third neural spine and no preopercular canal. It also shares other derived features found in most species of Oxyurichthys?a rounded margin on the tongue, a membranous crest on the nape, a shortened palatine bone, and a single row of teeth in the upper jaw. Putative syna- pomorphies of the gobionelline genera Evorthodus, Gobioides, Oligolepis, and Steno- gobius are discussed. FISHES referred to the genus Gobionellus are caudal fin. As the three other genera were at widespread in subtropical and tropical that time "catchalls," receiving species that coastal waters of the Americas and West Africa. could not be distinguished from a general go- They are common from northern Peru to Baja biid condition (small, fusiform, bottom-dwelling California in the Pacific, from southern Brazil fishes with united ventral fins), Ginsburg was in to North Carolina and Bermuda in the Atlantic fact comparing Gobionellus to a large cross-sec- and throughout the Gulf of Guinea region in tion of gobies (for examples of the taxa repre- the eastern Atlantic. Little is known about the sented in these genera, see Robins and Lachner, biology of most species and some are uncom- 1966; Hoese and Winterbottom, 1979,). Gins- mon across much of their ranges. Most appear burg also separated Gobionellus from the mor- to prefer protected waters with soft sediments phologically similar Oxyurichthys by the number and may be very abundant where discovered. of tooth rows in the upper jaw. Oxyurichthys was Ginsburg (1932) recognized 14 species of Go- noted to have a single row, whereas the 14 spe- bionellus in six subgenera. He distinguished this cies of Gobionellus that Ginsburg (1932) recog- group from the genera Rhinogobius, Ctenogobius, nized had two or more. and Gobius by the relative number of second Two of the 14 nominal species Ginsburg dorsal- and anal-fin rays, and the form of caudal (1932) initially placed in Gobionellus, Biat luzon- fin. Species of Gobionellus were characterized as ica and Biat fontanesii were not seen by him. having one fewer ray in the second dorsal fin These two species differ from the other mem- than in the anal fin and an elongate caudal fin. bers of the genus by a number of features but Ginsburg noted that species in the other three most prominently in having an oculoscapular genera had an equal number of elements for canal with a single median anterior interorbital both fins, or one fewer in the anal fin than in pore (C, see Materials and Methods), a single the second dorsal fin, and a blunt or rounded pair of pores on the snout (B), a posterior otic ? 2004 by the American Society of Ichthyologists and Herpetologists PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 261 pore (E) and a supraotic pore (G); a 3-22110 first dorsal fin pterygiophore pattern; and a broad gill opening (restricted in all the other species). In actuality, these taxa are not closely related to any species of Gobionellus but instead belong to the subfamily Gobiinae (Pezold, 1993). Both nominal species of Biat are appar- ently synonyms of Amblyeleotris fontanesii (Hoese and Steene, 1978; Gilbert and Randall, 19*79). Since Ginsburg's revision, 11 new species have been described or moved to Gobionellus from other genera. The addition of three of these species nullified Ginsburg's (1932) diag- nosis. These were Gobionellus daguae (including Gobionellus panamensis, see Gilbert and Randall, 1979) and Gobionellus liolepis, described by Gins- burg (1953), and Gobionellus stigmalophius, which was described by Mead and Bohlke (1958). Both G. daguae and G. liolepis have the same number of rays in the second dorsal and anal fins. Go- bionellus stigmalophius has a single row of teeth in the upper jaw. Although large individuals of G. liolepis develop two complete rows and large males of G. daguae may also develop a second row, smaller individuals of these species have a single row of teeth in the upper jaw. When Mead and Bohlke (1958) described G. stigmalo- phius, they considered it a "highly modified Go- bionellus" and felt that with its inclusion "the generic limits of Gobionellus and Oxyurichthys closely approach one another." Gilbert and Randall (1979) believed G. stigmalophius and Ox- yurichthys microlepis to be congeneric but did not place the two genera in synonymy pending fur- ther studies of these and several similar genera, including Evorthodus, Paroxyurichthys, Oligolepis, and Waitea. In fact, Paroxurichthys was later dis- covered to be a junior synonym of Gobionellus (Pezold, 1991) based upon a specimen of Gobi- onellus oceanicus. The reference to Waitea (based upon a communication to the authors by D. Hoese) was most likely to Waitea stomias, which is demonstrated later in this work to be a species of Oligolepis. As suggested by Gilbert and Randall (1979), the limits of Oxyurichthys and Gobionellus did not merely approach one another; they could not be distinguished. The only character state for Gobionellus not violated by the additions noted above was the elongate caudal fin?a state found in a number of disparate gobiid genera. In essence, the problem of an ever more inclu- sive Gobionellus was that none of the diagnostic character states was uniquely derived at the level used. Diagnostic characters have previously been offered for Oxyurichthys, Stenogobius and Gobionellus as recognized here (Pezold, 1991) but the species included in the latter genus were not given. In this paper, I offer a hypothesis on the re- lationships of all species presently included in Gobionellus and present diagnoses with postulat- ed synapomorphies for Gobionellus and related genera resulting from this analysis. Gobionellus is restricted to six species. Fifteen species formerly included in Gobionellus are assigned to the genus Ctenogobius (sensu Robins and Lachner, 1966). All references to Ctenogobius and Gobionellus hereafter are used in this restricted sense. One species, G. stigmalophius, is removed to the ge- nus Oxyurichthys and Oxyurichthys occidentalis Boulenger, a West African species, is formally reassigned to Gobionellus (Miller [1981] assigned the species to Gobionellus, but neither he, nor subsequent authors, gave an explanation for the reassignment). The six species of Gobionellus rec- ognized herein are redescribed in Pezold (2004). MATERIALS AND METHODS The study focused on features of the postcra- nial axial osteology and the cephalic lateralis for three reasons. First, these characters have been shown useful in diagnosing supraspecific taxa: postcranial axial osteology by Birdsong (1975) and Birdsong et al. (1988); cephalic free neu- romas! patterns by Iljin (1930), Miller (1973), and Hoese (1983) and cephalic lateralis canal structure by Lachner and McKinney (1979), Takagi (1989), and Pezold (1993). As these character suites exhibit sufficient stability in a group notorious for labile characters, they are often effective in diagnosing genera and occa- sionally higher taxa as well. Second, when this study was initiated in the early 1980s our un- derstanding of gobioid relationships was much more vague than it is today. Although questions of character state polarity within the Gobioidei still remain, the polarization of states was much more tentative then simply because we knew much less about the distribution of character states across the numerous gobioid taxa. A sur- vey of numerous taxa for character suites could be undertaken relatively easily. Finally, focusing on the sensory system and postcranial osteolog- ical characters was appealing in that they of- fered the opportunity for potentially indepen- dent assessments of relationships. The different embryological origins (ectoderm vs mesoderm) of these characters enhance the possibility of in- dependent evolutionary changes where not constrained by selection or linkage. Because Miller et al. (1980) illustrated the conflicting in- formation offered by skeletal and lateralis char- 262 COPEIA, 2004, NO. 2 acters, other characters were included in the analysis that follows if they appeared informa- tive in diagnosing generic limits for the ingroup species. Comparative collections used for the exami- nation of the cephalic lateralis, and free sensory papillae are listed in Pezold (1993). Specimens of Gobionellus examined are listed in Pezold (2004), whereas specimens of Ctenogobius are listed in Materials Examined. Characters exam- ined included extent of oculoscapular canal de- velopment, oculoscapular canal pore patterns, presence/absence of preopercular canal, preo- percular canal pore number, and free neuro- mas! configurations on the cheek and opercle. Canal and canal pore terminology follows Tak- agi (1957) and Akihito et al. (1984). Pore names (sensu Takagi, 1957) are used in discus- sions and notated using the lettering system of Akihito et al. (1984) when reporting configu- rations for a taxon and labeling illustrations. The system of Akihito et al. (1984) has been modified such that, when interorbital pores are single and median in position, they are under- lined (as C and/or D). Pore names given in dis- cussions are followed by the designator letter of the "Akihito system" in parentheses. Free neu- romast patterns are described in reference to topographic location unless otherwise indicat- ed. All lateralis drawings were made with a dis- secting microscope and camera lucida. Features of the postcranial osteology were ex- amined from radiographs and cleared-and- stained specimens. Specimens examined are in- cluded in Birdsong et al. (1988). Characters ex- amined were shape of the basihyal; placement of the first pterygiophore of the spinous dorsal fin; placement of the first pterygiophore of the second dorsal fin; number of precaudal anal fin pterygiophores; number of precaudal verte- brae; number of caudal vertebrae; relative num- ber of segmented rays to vertebrae; form of the third neural spine; form of the fourth neural spine; extent of neural arch completion over the caudal vertebrae; number of epurals; and first dorsal fin pterygiophore insertion pattern. The first dorsal fin pterygiophore insertion pat- tern is given as a formula following Birdsong (1975) and Birdsong et al. (1988). Shape of the basihyal and third neural spine was determined from cleared-and-stained material and by dissec- tion. Twenty characters were coded for the analy- sis. Characters found to be phylogenetically un- informative were not included in the matrix. Much additional information on character state distributions for the postcranial axial skeleton has been previously given in Birdsong et al. (1988) and for the oculoscapular canal system in Pezold (1993). Monophyly of Gobionellus (as historically con- ceived) was tested and phylogenetic relation- ships among its species were described by com- paring them to other gobionelline species (Pe- zold, 1993) with which they have previously been confused or for which close relationships have been suggested (e.g., Mead and Bohlke, 1958; Gilbert and Randall, 1979; Birdsong et al., 1988). The outgroup species are all members of the Stenogobius group of the Gobionellinae rec- ognized by Larson (2001). Gobionellines in the northern Pacific Acanthogobius, Astrabe, and Chasmichthys groups were not included in the analysis, nor were a number of Indo-Pacific gen- era related to Mugilogobius. Larson (2001) pro- posed two species groups within the Gobionel- linae on the basis of four characters. Species of the Stenogobius group, which would include all species historically placed in Gobionellus (with the exception of A. fontanesii), differ from the Mugilogobius group in having anterior nasal pores (A) present, no villi on the head, no in- fraorbital pores (E) and (usually) transverse rows of free neuromasts on the cheek. The lat- ter two features she proposed as derived states within the gobionellines. Larson's Mugilogobius and Stenogobius groups are at this time united to one another and to the northern Pacific group of gobionellines solely by plesiomorphic traits. Phylogenetic analysis was accomplished using PAUP* (D. Swofford, Phylogenetic Analysis Us- ing Parsimony, 4.0 beta 10 vers., Sinauer Press, unpubl.) and NONA (P. Goloboff, NONA vers. 2.0., unpubl.). Heuristic searches were per- formed using TBR branch-swapping algorithms. Trees were collapsed if minimum branch length equaled zero. All characters were unweighted and character states were unordered (nonaddi- tive). Bremer support values (Bremer, 1994) were calculated using SEPAL (B. Salisbury, SE- PAL: strongest evidence and parsimony analyz- er, vers. 1.4. Yale University, New Haven, CT, un- publ.). The data matrix constructed is given in the appendix. Character state polarity among study taxa (ingroup and gobionelline out- groups) was determined by rooting the clado- gram using the outgroup method (Farris, 1982; Nixon and Carpenter, 1993). Cladograms were rooted using a hypothetical taxonomic unit rep- resenting putative ancestral character states (Lundberg, 1972) derived from a survey of states observed in Rhyacichthys aspro (Rhyaci- chthyidae), the sister group to all other gobioid fishes (Miller, 1973; Springer, 1983), odonto- butids (sensu Hoese and Gill, 1993) and butine eleotrids (the Eleotridae are recognized here as PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 263 including the Butinae and Eleotrinae of Hoese and Gill, 1993, but not the Gobiidae). Character conditions for the hypothetical outgroup are denoted by "0" in the matrix and character state descriptions that follow, with four excep- tions. Cheek papillae orientation (character 9), first dorsal fin pterygiophore insertion pattern (15), the relative number of second dorsal fin and anal fin rays (17) and epural number (20) show ambiguous distributions among basal go- bioid taxa. The hypothetical outgroup state was denoted as unknown (?) for those characters. Trees were constructed and edited using WinClada (K. Nixon, BETA vers. 0.9.9, unpubl.) and CorelDRAW (2002 Corel Corporation, un- publ.). RESULTS Both character suites were phylogenetically informative and results applying to higher levels of gobioid relationships have been previously published (Birdsong et al., 1988; Pezold, 1993). Characters useful in assessing generic affinities of species of Gobionellus and Ctenogobius and in- cluded in the analysis are listed below. The cla- distic analyses using PAUP and NONA pro- duced two most parsimonious trees, each with a branch length of 43. A strict consensus tree with a branch length of 45, consistency index of 0.69 and retention index of 0.90 is shown in Figure 1. In the consensus tree, Gobionellus is part of a clade including Gobioides. The Gobioi- des/ Gobionellus clade has an Awaous/ Stenogobius clade as its sister group. Ctenogobius is part of a polytomy (12 branches) including one clade of six species of Ctenogobius, an Oxyurichthys clade and an Evorthodus/Oligolepis clade. The two most parsimonious cladograms from which the con- sensus tree was formed differ in the position of Ctenogobius stigmaturus as a sister group to a Cten- ogobius/Oxyurichthys/Evorthodus/Oligolepis clade or as part of a Ctenogobius clade; and in Cteno- gobius being placed as either the sister group to Oxyurichthys (Fig. 2) or as the sister group (ex- cluding C. stigmaturus) to an Oxyurichthys/Evor- thodus/Oligolepis clade. The consensus tree indicates that Gobionellus and Ctenogobius are more closely related to oth- er taxa than to one another. Character states are listed and described for each character used in the analysis and followed by a discussion of dis- tributions among ingroup taxa and other go- bioid fishes. Character numbers correspond to those given in the data matrix (Appendix 1). The consistency index for each character fol- lows the character state descriptions. 7 20 19 ancestor Gn thompsoni A banana S genivittatus S laterisquamatus G microdon G stomatus G occidental G oceanicus Gdaguae G liolepis Gd broussoneti Gd grahamae C boleosoma C claytoni C fasciatus C sp Brazil C pseudofasciatus C shufeldti C stigmaticus C stigmaturus C thoropsis 20 C1 . ?:? E lyncus 8910 14 Ol acutipennis Ol jaarmani Ol stomias C lepturus C manglicola C phenacus C saepepallens Csagittula C smaragdus Ox keiensis Ox heisei Ox lonchotus Ox microlepis Ox ophthalmonema Ox papuensis Ox stigmalophius Fig. 1. Strict consensus tree of four cladograms generated by parsimony analysis of 20 characters for a hypothetical outgroup and 38 species of the gobi- onelline "Stenogobius group" sensu Larson (2001). The cladogram has a branch length of 45, a consis- tency index of 0.69 and a retention index of 0.90. Bremer support indices are indicated beneath branches. Character state changes are indicated on the branches with number given above. Homoplastic changes are shown by white circles. 264 COPEIA, 2004, NO. 2 ''? E lyricus 01 acutipennis T3- 01 jaarmani o?01 stomias Ox keiensis Ox heisei Ox lonchotus Ox microlepis Ox ophthalmonema Ox papuensis Ox stigmalophius C boleosoma C claytoni C fasciatus C sp Brazil C pseudofasciatus Cshufeldti ^g- C stigmaticus n ? C stigmaturus C thoropsis C lepturus C manglicola C phenacus C saepepallens C sagittula C smaragdus Fig. 2. Cladogram illustrating alternative most parsimonious reconstruction of relationships of spe- cies of Ctenogobius with other gobionellines. In this res- olution, Oxyurichthys is the sister group to a mono phyletic Ctenogobius. Character numbers are noted above the state changes indicated on branches. Ho- moplastic changes are shown by white circles. Number of upper jaw tooth rows (1).?(0) multiple rows; (1) single row. (0.25). Basal gobioids have multiple rows of teeth in the upper jaw. Single rows of teeth appear in the sicydiines, oxuder- cines (Murdy, 1989), and the Mugilogobiusgroup gobionellines Gobiopterus and Calamiana varie- gata (Larson, 2001). Within the Stenogobius group gobionellines, a single row of teeth in the upper jaw has been observed in Evorthodus (Ginsburg, 1931; Daw- son, 1967, 1969), Oxyurichthys (except for Ox- yurichthys keiensis) and several species of Gobioi- des (Murdy, 1998). A single row also occurs in Gobionellus daguae and G. liolepis (Ginsburg, 1953) and Oligolepis. Large males of G. daguae may develop a second row of teeth. Additional specimens of G. liolepis have just recently been collected on a STRI-NMNH cruise off the coast of El Salvador. The several specimens collected are all larger than the types examined in this study. All of these specimens have developed a second complete row of fine teeth in the upper jaw (J. Van Tassell, pers. comm.). I reran the analysis changing the state for G. liolepis to re- flect two rows of teeth in the upper jaw and the results, including consistency and tree length, were unchanged, except for the distribution of states for this character. A single row of teeth appeared as parallel developments in G. daguae and Gobioides africanus instead of a reversal as in the cladogram presented (Fig. 1). One can le- gitimately code this character either way, de- pending upon whether one emphasizes the ap- pearance of a single row through individuals as large as those observed in the type series or the ultimate achievement of two rows. I chose to emphasize the significance of the presence of the single row as did Meek and Hildebrand (1928) and Ginsburg (1953). Tongue form (2).?(0) truncate or emarginate; (1) rounded. (1.00). Emarginate or truncate tongues were observed in the butines Parvipar- ma straminea, Oxyeleotris sp. (blackbanded gud- geon) , Ophiocara porocephala, Butis melanostigma, and Bostrychus africanus and the eleotrines Go- biomorus, Dormitator, Eleotris, Erotelis smaragdus, Hypseleotris compressa, Leptophylipnus, Mogurnda mogurnda, and Guavina guavina. Among the Stenogobius group gobionellines (sensu Larson, 2001), the derived condition was only found in Oxyurichthys, with the exception of Ox. keiensis in which the plesiomorphic condition was ob- served. Palatopterygoid strut (3).?(0) palatine extending midway along ectopterygoid; (1) palatine elon- gate, reaching or nearly reaching quadrate and reinforced with subequal or very reduced ectop- terygoid; (2) palatine short, not reaching along more than the dorsal third of ectopterygoid. (1.00). Harrison (1989) described the plesiom- orphic condition for the palatopterygoid strut in gobioids (as part of a palatopterygoquadrate complex) as consisting of a palatine extending half the length of the ectopterygoid. In the most basal species, an ossified endopterygoid also forms a prominent part of the strut. All but one species of Oxyurichthys has a short- ened palatine in which the posteroventral pro- cess of the palatine is reduced and the ectop- terygoid forms the lower part of the palatopter- ygoid strut (state 2; Harrison, 1989). All other species of the Stenogobius group (sensu Larson, 2001) included in this study have an elongate palatine extending nearly to or meeting the quadrate (state 1). The elongate palatine is typical of gobionel- lines and has been proposed by Harrison (1989) as a derived condition for gobioid fishes. Larson (2001) found the elongate condition present in most Mugilogobius-group gobionel- lines. The elongate palatine described for these taxa forms the major portion of the palatopter- ygoid strut, with the ectopterygoid reduced to a small splint behind the lower tip of the palatine in the most extreme cases. Although a similar palatine structure has been observed in the go- biines Luciogobius grandis and Gobiodon (Harri- son, 1989), the ectopterygoid forms a major part of the palatopterygoid strut and could not PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 265 be considered subequal to the palatine. An elongate palatine also appears in a number of eleotrids (Akihito, 1969; Harrison, 1989), and the odontobutid Odontobutis obscura (Akihito, 1969). The ectopterygoids in these species are long but are subequal to the palatine, splintlike and bordered by the palatine along most of the anteroventral side. All of these species have en- dopterygoids forming a major portion of the strut. Regarding the constitution of the strut as a whole, not just the relative reach of the pala- tine along the ectopterygoid, the strut seen in the gobionellines examined here is distinct. According to Harrison (1989), a shortened palatine as seen in Oxyurichthys also appears in the oxudercines and the amblyopines Taenioides cirratus and Trypauchen vagina. Murdy (1989) and Murdy and Shibukawa (2001) reported this state for the amblyopine genera Trypauchen, Brachyamblyopus, and Odontamblyopus. Although Harrison used the character to associate the ox- udercines, Taenioides and Trypauchen in an Ox- yurichthys lineage, Murdy (1989) discounted the similarity between the oxudercine and ambly- opine forms because of a greater reduction of the palatines in the oxudercines specimens he examined. This is not apparent in Harrison's illustrations nor the illustration in Murdy and Shibukawa (2001). Harrison's illustrations do show a difference in the form of the ectopter- ygoid between the amblyopines T cirratus and T vagina and the oxudercines and Oxyurichthys. The latter group has elongate ectopterygoids forming the lower portion of the strut. Oxyuri- chthys has a splintlike, teardrop-shaped ectopter- ygoid with an attenuated anterior process. The amblyopines have more bluntly formed, rect- angular ectopterygoids (Harrison, 1989; Murdy and Shibukawa, 2001). In a molecular phytog- eny for gobioid fishes derived from sequence data for three protein-coding mitochondrial genes, Thacker (2003) presented representative amblyopines and oxudercines as a sister clade to selected sicydiines and Stenogobius gobionel- lines. Her analysis did not include Oxyurichthys but did include Evorthodus and Ctenogobius. If Oxyurichthys is a sister group to Ctenogobius as proposed here, then the similarities between the palatopterygoquadrate struts of Oxyurichthys and those of the amblyopines and oxudercines are likely caused by convergence. This conclu- sion may be supported by differences in the form of the ectopterygoid among these taxa. Gill rakers on the anterior side of first epibranchial (4).?(0) unmodified gill rakers present; (1) fleshy lobe-like gill rakers present; (2) no rak- ers, tufts of small papillae present; (3) no rak- ers, clumps of large papillae present (Fig. 3). (0.75). The presence of gill rakers on the first gill arch is ancestral for gobioid fishes. Gill rak- ers are present on the anterior surface of the epibranchial of the first gill arch, in varying numbers, in R. aspro (Miller, 1973:407), the bu- tines P. straminea, Oxyeleotris sp. (blackbanded gudgeon), O. porocephala, B. melanostigma, and Bostrychus africanus and the eleotrines Gobiomo- rus, Dormitator, Eleotris, E. smaragdus, H. compres- sa, Leptophilypnus, M. mogurnda, and G. guavina. Rakers in the eleotrids and R. aspro vary from many thin and elongate rakers as seen in Dor- mitator to small prickly knobs as observed in M. mogurnda and B. melanostigma, but they are not fleshy lobes. Species of Ctenogobius have a first gill arch with tufts of tiny papillae on the first epibran- chial but no gill rakers or lobes and four or five broad, triangulate, unconnected rakers on the ceratobranchial. There are no rakers or lobes on the first epibranchial in species of Stenogo- bius, but there are tufts of large papillae present; the rakers on the first ceratobranchial are re- duced and joined by a low membrane. Species of Gobionellus have slender gill rakers on the up- per arch and, with the exception of G. daguae, on the lower portion as well. Gobionellus daguae has a few broad rakers on the first ceratobran- chial similar to those found in Ctenogobius, Ox- yurichthys, Oligolepis, and Evorthodus. Oxyuri- chthys, Oligolepis, and Evorthodus have lobelike structures on the anterior side of the first epi- branchial. Lobes are also found on the epibran- chial of species of Gobionellus, but they are not on the anterior side and, therefore, do not ap- pear to be homologous. Evorthodus may have a single long raker in addition to lobes on the upper arch. Lobes on the anterior side of the first epibranchial are also found in some species assigned to the Gobiinae (e.g., Hoese and Allen, 1977). Fleshy lobes on pectoral fin girdle (5).?(0) none; (1) present. (1.00). The pectoral girdle is smooth in R. aspro, odontobutids and most eleo- trids. A large platelike flap is found on the pec- toral girdle of P. straminea. Fleshy protuberances (lobes) and folds are found in some species of Gobionellinae (Larson, 2001 :fig. 9). Larson (2001) discussed the presence of these struc- tures in a number of Mugilogobius group genera. Fleshy structures on the pelvic girdle are also seen in several northern Pacific endemics. Lobes are present in Chaeturichthys stigmatias and a long fold is found in Acanthogobius flavi- manus, both members of the Acanthogobius group sensu Birdsong et al. (1988). Quietula 266 COPEIA, 2004, NO. 2 B C ,~\ \ '??? \'V \ Fig. 3. An illustration of the first gill arch (right lateral view) as found in (A) Gobionellus oceanicus, (B) Ctenogobius fasciatus, (C) Evorthodus lyricus, (D) Oligolepis acutipennis, (E) Oxyurichthys stigmalophius, and (F) Oxyurichthys keiensis. Examples of observed states for character 4, gill rakers associated with the anterior epi- branchial, are unmodified gill rakers present on anterior epibranchial surface (3A), fleshy lobelike rakers present (C-F) and no rakers present, only small tufts of papillae (B). y-cauda, a member of the Chasmichthys group Anterior nares position relative to cephalic lateralis (Birdsong et al., 1988) has lobes as well. (6).?(0) nares lateral to canals on snout; (1) Among the Stenogobius group gobionellines, nares medial to snout canals. (1.00). The basal only Awaous and Stenogobius have lobes. gobioid condition is for each naris to be lateral PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 267 Fig. 4. Lateral view of the oculoscapular canals, preopercular canals and sensory papillae rows of the cheek and opercle in (A) Oligolepis acutipennis, (B) Ctenogobius Upturns, and (C) Gobionellus occidentalis. Canal pores are labeled according to Akihito et al. (1984) with unpaired pores underlined. Labeled neuromast rows are posterior opercular row (p) and horizontal midcheek row (b). to the cephalic lateralis canals on the snout. In fact, state one was only observed in Stenogobius (Pezold, 1991). Oculoscapular canal (7).?(0) complete laterally (continuous from orbit to rear margin of opercle) with A'BCDEFGHIJKL' pattern; (1) complete lat- erally with A'BCDFHJKL' pattern; (2) complete laterally with ABCDFHKL' pattern (Fig. 4C); (3) complete laterally with A'BDFHKL' pattern; (4) abbreviate laterally, terminating above preoper- cle with A'BCDFH' pattern (Fig. 4B); (5) dis- junct laterally with a separate portion above rear opercle and A'BCDFH' K'L' pattern (Fig. 4A). (0.83). Akihito (1986) reasoned that the primi- tive gobiid oculoscapular canal is probably an interrupted canal with a disjunct temporal por- tion as it is common across many gobiid taxa (for examples, see Pezold, 1993). However, Tak- agi (1989) suggested a general evolutionary trend in gobioid fishes from well developed to reduced cephalic sensory canal formation. This is consistent with the occurrence of complete oculoscapular canals in the Pvhyacichthyidae (Miller, 1973), and a long canal running from near the tip of the snout to the opercular mar- gin is also seen in butine genera Butis, Bostry- chus, Oxyeleotris, and Ophiocara (Pezold, 1993). An extensive lateral oculoscapular canal is con- sidered plesiomorphic in this study. Pores A'BCDEFGHIJKL' are observed in several basal eleotrid genera with extensive canals?the bu- tines Bostrychus, Butis, and Oxyeleotris (Pezold, 1993). There is a tendency toward loss of pores among the gobioids, even within the butines. Pore G is lost in Ophiocara, G, C, and E in some Bostrychus, and G, C, E, and I in Parviparma. A complete oculoscapular canal with pores A'BCDEFGHIJKL' is regarded as the ancestral gobioid state herein. There are at least five different oculoscapular canal structures (Pezold, 1993) occurring among the study taxa (Awaous exhibits more than one; see Watson, 1992, 1996; Pezold, 1993). Of the five, only the disjunct pattern (state 5) spans the two clades containing Cteno- gobius and Gobionellus, occurring in Evorthodus, Gnatholepis, Oligolepis, and Stenogobius. This pat- 268 COPEIA, 2004, NO. 2 tern is also seen in the sicydiine genera Sicyopus, Lentipes, and Stiphodon. In the cladogram, this pattern is the basal state from which the other four states, all unique among gobioid fishes, are derived. The A'BCDFHKL' oculoscapular pore pat- tern is unique to Gobionellus. In the cladogram, this extensive canal structure is derived from the disjunct state and is the precursor to the state described for Gobioides below. An extensive canal (with a different pore pattern) appears independently in Awaous. Similar reversals to an extensive and continuous oculoscapular canal (with different pore patterns) are seen in the gobiines Bathygobius and Gobius (Akihito and Meguro, 1980; Ahnelt, 2001) and the sicydiines Sicydium and Sicyopterus (Parent! and Maciolek, 1993; Pezold, 1993). In Gobioides, the anterior interorbital pores (C) are lost, resulting in an A'BDFHKL' oculo- scapular pore pattern. The oculoscapular canals are separate between the orbits. Anterior inter- orbital pores (C) are absent in oxudercine spec- imens identified as Apocryptes bato and Parapo- cryptes cantonensis, but they also lack much of the oculoscapular canal system (Pezold, 1993). These oxudercine species have lost the anterior portion of the nasal or snout canals leaving them with only a single pair of snout canal pores (B). Some gobionellines of the northern Pacific, such as A. flavimanus, C. stigmatias and Q. y-cauda, have also lost the anterior interor- bital pores. These species have very different ca- nal systems from the form seen in Gobioides (Pe- zold, 1993), showing a reduction in the extent of the snout canals in particular. Mahidolia and some of the gobiosominines lack anterior inter- orbital pores as well; all of these species are be- lieved to have attained this condition from a sin- gle-pored state, and thus the condition would not be homologous. In short, the canal struc- ture and the pore patterns in all of the taxa noted above differ from the condition observed in Gobioides. Ctenogobius and Oxyurichthys have an A'BCDFH' pore pattern. Abbreviated temporal portions of the oculoscapular canal as found in these two gen- era are also seen in Redigobius, Fusiogobius, Gobio- don, Tukugobius bucculentus and the gobiosomi- nines examined but differ in the pores pre- sent?their canal structures are not regarded as homologous. Preopercular canal pores (8).?(0) canal present with three pores, M'NO'(Fig. 4B-C); (1) canal present with two pores, M'O' (Fig. 4A); (2) ca- nal absent. (1.00). Preopercular canals have no more than three pores in gobiids (M'NO') and this condition is regarded as plesiomorphic for the family. The canal is completely lost in Ox- yurichthys and has been reduced independently to a two-pore state in Oligolepis. Outside of the Stenogobius group gobionellines, the canal and its associated pores have been lost or reduced to the two-pore state independently many times (e.g., Takagi, 1989, Larson, 2001). This charac- ter is variable in G. oceanicus?approximately 5% of the specimens examined here had two pores (state 1) in both preopercular canals, nearly 10% had two pores in one canal and three in the other. Cheek papillae orientation (9).?(0) longitudinal (Fig. 4A); (1) transverse (Fig. 4B-C). (1.00). Larson (2001) characterized the scattered, un- organized state found in R. aspro as the plesiom- orphic condition for gobioid fishes. This same approach was taken in initial analyses, but I. J. Harrison (unpubl.) notes that Rhyacichthys "shows several unusual and specialized charac- ters" and suggests that the organization of the cephalic sensory papillae might be another ex- ample of specialization. A longitudinal pattern is observed in the putative rhyacichthyid Proto- gobius attiti (Watson and Pollabauer, 1998; for discussion, see Shibukawa et al., 2001) and odontobutids (Hoese and Gill, 1993), but both states 0 and 1 are found across other eleotrids. The outgroup state was designated as unknown. Within the Gobionellinae, state 0 predominates among species of the Mugilogobius group (Lar- son, 2001) and the northern Pacific species (the Chasmichthys, Acanthogobius, and Astrabe groups of Birdsong et al., 1988; Akihito et al., 1984). Within the Stenogobius group, only Oligolepis ex- hibits state 0. Anteriad extension of horizontal cheek papillae row "b" (10).?(0) not reaching second transverse suborbital row (Fig. 4B); (1) reaching second transverse row (Fig. 4C). (0.33). The plesiom- orphic condition is a midcheek horizontal row ("b" row) (Miller and Wongrat, 1979) that does not reach anteriad as far as the second trans- verse suborbital row. The midcheek horizontal suborbital papillae row extends forward to reach the second trans- verse suborbital row of a series that starts at the anterior eye margin in Gobioides and Gobionellus with the exception of Gobionellus microdon and Gobionellus stomatus. It reaches only the third row in G. microdon and only the third or fourth row in G. stomatus. The derived condition ap- pears independently in both Oligolepis and in most Oxyurichthys (specimens examined for the PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 269 latter genus varied with the row reaching the third or second transverse suborbital rows). There are three transverse suborbital rows an- teriad to the end of row "b" in all specimens of Ctenogobius examined. The same is true for Sten- ogobius genivittatus (Akihito et al., 1984:fig. 156), Awaous banana, Awaous ocellaris (Akihito et al., 1984:fig. 131), and Evorthodus lyricus. Oligolepis acutipennis lacks well-developed transverse rows, whereas Oligolepis stomias has a condition similar to Evorthodus and Ctenogobius (although the transverse rows are still reduced). Position ho- mology for this row with midcheek horizontal rows seen in the sicydiines and Gnatholepis is not clear. Papillae row patterns of the cheek in the latter taxa have poorly developed midcheek horizontal rows at best, and where formed they are on the posterior field. Posterior opercular papillae row (p) (11).?(0) di- agonal (Fig. 4A-B); (1) vertical (Fig. 4C). (1.00). The general gobioid condition is a di- agonal p row crossing the subopercle-opercle symphysis, disjunct from both the anterior oper- cular and subopercular rows (Sanzo, 1911; Akihito et al., 1984; this study). Gobionellus, Gobioides, Awaous, and Stenogobius have a vertical (perpendicular to the long axis of the fish) p row. The p row is not confluent with the anterior opercular row but does reach the subopercular row in most specimens. Several species of Eleotris, both species of Er- otelis, and Leptophilypnus fluviatilis (Akihito et al., 1984; Miller, 1998; Pezold and Cage, 2002) have a p row that usually connects with the anterior opercular and subopercular rows. It is archlike in orientation and is diagonally inclined as it crosses the opercular/subopercular juncture. The upper portion of the row is directed toward the anterior opercular row when not actually connected. In some species of Eleotris the p-row may appear as two intersecting rows, with a di- agonal upper row intersected by a vertical lower row. Shared possession of these row types may be indicative of relationship among Eleotris, Er- otelis, and Leptophilypnus, but the conformations appear to be homoplastic to the rows seen in Gobionellus and the other gobiids noted. Nape crest (12).?(0) without a membranous crest; (1) membranous crest present. (1.00). Most gobioid fishes, including Rhyacichthys, odontobutids and eleotrids, do not have a crest on the nape. A membrane is observed in some gobiines (e.g., Lophogobius, Cristatogobius, Cryp- tocentroides). Among the Stenogobius group gobi- onellines, the derived state is found in Oxyuri- chthys. Fig. 5. Transversely bifid third neural spine of (A) Oxyurichthys keiensis and (B) Oxyurichthys stigmalophius. Arrows indicate the third neural spines which precede the first dorsal fin pterygiophore. Third neural spine (13).?(0) spikelike with one point; (1) transversely broad and bifid (1.00; Fig. 5). The typical gobioid condition, and that observed in Rhyacichthys, odontobutids, and eleotrids, consists of a single spine tip. The bifid condition, not observable in radiographs, has been observed only in species of Oxyurichthys. Fourth neural spine (14).?(0) thin spine; (1) tri- angulate; (2) slightly flared along its length; (3) pikelike, flared along its length with rearward extension just above the base; (4) spatulate with thin base. (0.67; Figs. 6-7). Most gobioids have simple, thin neural spines (Larson, 2001; this study). The fourth neural spine is broadly flared distally, spatulate but constricted at the base in most Gobionellus. The fourth neural spine in G. daguae (Fig. 6D) is not spatulate like that seen in the other species of Gobionellus but is broad- ened distally with a caudad extension above the base. A similar structure to the distally flared fourth neural spine described above has been ob- served in Oxuderces dentatus (Murdy, 1989). Lar- son (2001) described the second through fifth neural spines as variably expanded and bifid or 270 COPEIA, 2004, NO. 2 Fig. 6. The fourth neural spine in (A) Ctenogobius smaragdus, (B) Ctenogobius shufeldti, (C) Gobionellus micro- don, and (D) Gobionellus daguae. Arrows indicate the fourth neural spines which follow the insertion of the first dorsal fin pterygiophore. The arrow in (D) points to the horizontal posterior flange on the fourth neural spine of Gobionellus daguae. split for Mugilogobius and some related gobi- onelline genera, noting the character does not appear in all specimens within a species nor all species within a genus. Evorthodus and most spe- cies of Oligolepis have a fourth neural spine very slightly expanded throughout its length ending in a pointed tip (state 2; Fig. 7). Oligolepis stomias and some species of Ctenogobius have a broad- based triangulate fourth neural spine (state 1). First dorsal fin pterygiophore pattern (15).?(0) 3? 12210; (1) 3-12201. (1.00). Although the an- cestral condition for first dorsal fin pterygio- phore pattern is a matter of debate (see Larson, 2001:32 for a synopsis), the gobionellines gen- erally exhibit a 3-12210 pattern. Gobioides is an exception showing the derived state (Birdsong et al., 1988; Murdy, 1998). The outgroup was coded as unknown. Dorsal fin confluence (16).?(0) separate or barely connected at base; (1) broadly confluent. (1.00). Basal gobioid fishes have separate dorsal fins. Confluent dorsal fins are found in ambly- opine species. Confluent dorsal fins are also found in Ptereleotris monoptera (Randall and Hoese, 1986). Although rarely observed, conflu- ent dorsal fins have appeared independently in gobiids at least twice. Although they were not connected in the specimen of Gobiodon quin- questrigatus examined in this study, the trait is variably expressed in some species of Gobiodon (R. Winterbottom, pers. comm.). Dawson (1967) reported that the dorsal fins were fre- quently connected basally in E. lyricus but not Evorthodus minutus. Gobionellus liolepis and all species of Gobioides have confluent dorsal fins (Murdy, 1998). The presence of the dorsal fin connection in both the amblyopines and the other diverse taxa is regarded as convergent with the condi- tion seen in Gobioides and Gobionellus. Gobioides differs from the amblyopines in jaw structure, including the palatopterygoid strut (Harrison, 1989), in the form of the cephalic lateralis ca- nals and papillae, and does not have the 2:1 ratio of soft rays to underlying vertebrae, a fea- ture synapomorphic for the amblyopines (Pe- zold, 1993; Murdy, 1998). Number of second dorsal and anal fin rays (17).? (0) one more ray in second dorsal fin than anal fin; (1) equal number of rays in second dorsal and anal fins; (2) one more ray in anal fin than PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 271 <&% * 4 + Fig. 7. The fourth neural spine in (A) Evorthodus lyricus and (B) Oligolepis acutipennis. Arrows indicate the fourth neural spine which follow the insertion of the first dorsal fin pterygiophore. second dorsal fin. (0.67). Two conditions have usually been observed in R. aspro, one more sec- ond dorsal-fin ray than anal-fin ray or an equal number of rays in the two fins (Miller, 1973; Larson, 2001), although Akihito et al. (1984) reported two more in the anal fin than the sec- ond dorsal fin for this species. The odontobu- tids generally have at least one more ray in the second dorsal fin than the anal fin (Akihito et al., 1984; Iwata et al., 1985; Larson, 2001). Among eleotrids states 0 and 1 are generally ob- served (Akihito et al., 1984; Larson, 2001; this study), but Akihito et al. (1984) reported state 2 in Butis amboinensis, Ophieleotris sp. and Hypse- leotris cyprinoides. In this study, state 2 was also found characteristic of Dormitator latifrons, Dor- mitator lebretonis, and Dormitator maculatus, and Hypseleotris compressa. The ancestral state was coded as unknown. Among gobionellines, one more dorsal-fin ray or equal numbers of rays were the predom- inant states observed in the Mugilogobius group by Larson (2001). Of the gobionelline species examined in this study, G. liolepis, G. daguae, Go- bioides, Stenogobius, and Awaous had equal num- bers of second dorsal- and anal-fin elements. All others had one more ray in the anal fin than Fig. 8. Drawing of the postcranial axial skeleton Ctenogobius shufeldti (UMMZ 155326) illustrating in- completely formed neural arches over the caudal ver- tebrae. Incomplete arches lack neural foramina. the second dorsal fin. The cladogram hypothe- sizes two separate origins of state one from an- cestors having one more ray in the anal fin than the second dorsal fin. Median/caudal fin confluence (18).?(0) anal fin and second dorsal fin separate from caudal fin; (1) both fins connected to caudal fin. (1.00). Among basal gobioids all median fins are sepa- rate. Within the Gobionellinae, state 1 is ob- served only in Gobioides (with the exception of Gobioides africanus [Murdy, 1998]). Although connected median and caudal fins are obtained in the amblyopines, the condition is not ho- mologous to that observed in Gobioides for rea- sons given under character 16. Thacker (2000) demonstrated that a continuous median fin-fold was synapomorphic for the genus Microdesmus within the monophyletic Microdesmidae. Neural arches of caudal vertebrae (19).?(0) com- plete; (1) incompletely formed (0.25; Fig. 8). Among gobioid fishes, including Rhyacichthys, odontobutids and eleotrids, the neural arches of the last few caudal vertebrae are incompletely formed. The possession of incompletely formed arches in additional vertebrae is regarded as a derived condition. Evorthodus, Oligolepis, Ox. keiensis, most species of Ctenogobius and G sto- matus have incompletely developed neural arch- es over all or most of the caudal vertebrae. The condition of the neural arch reverses to com- plete over the caudal vertebrae in C. stigmaturus and all species of Oxyurichthys except Ox. keiensis. Among other gobiids, the sicydiines and the ox- udercine Boleophthalmus boddarti are also char- acterized by incomplete neural arches over the caudal vertebrae, and Birdsong (1988) reported reduced neural arches over the caudal verte- brae in the gobiine Robinsichthys arrowsmithensis. Epural number (20).?(0) three; (1) two; (2) one. (0.50). Three epurals are found in Rhyacichthys (Miller, 1973) and sporadically in some odon- tobutids (Akihito, 1986). Most eleotrids have two epurals. The hypothetical ancestor was cod- 272 COPEIA, 2004, NO. 2 ed as "?" for this feature because of the lability in basal gobioids. This character is also labile within the gobionellines studied here. A single epural was variably observed in specimens of G. liolepis, Ctenogobius lepturus, Ctenogobius manglico- la, Ctenogobius saepepallens, C. stigmaturus, an un- described species of Ctenogobius from Brazil, Go- bioides sagitta, Ol. stomias and Oxyurichthys loncho- tus. One specimen of Ox. keiensis had three. Lar- son (2001) reported two epurals in all members of the Mugilogobius group except for Brachygo- bius and Chlamydogobius and Calamiana paludo- sus. In her analysis, she used the modal number if there was variation within a species. Using that approach, Awaous and Evorthodus have a single epural; all others studied here have a mode of two. DISCUSSION The data demonstrate that Gobionellus as his- torically conceived is polyphyletic. The evidence indicates that two lineages exist within the ge- nus that are more closely related to other go- bionellines than to one another. One group of six species, paraphyletic in the analysis, includes G. oceanicus and retains the name Gobionellus. The six species recognized are G. daguae, G. lio- lepis, G. microdon, G. occidentalis, G. oceanicus, and G. stomatus. The species are redescribed in an- other work. Although the extensive oculoscapular canal with an A'BCDFHKL' pore pattern is unique to Gobionellus, it is most likely the precursor for the condition found in Gobioides as depicted in the cladogram. The expanded fourth neural spine as seen in Gobionellus, however, is also presented as a synapomorphy for the Gobioides/Gobionellus clade, being subsequently lost in Gobioides. The characters in conflict with the fourth neural spine as a synapomorphy for Gobionellus are the relative number of second dorsal and anal fin rays (17), the confluence of the dorsal fins (16), forward extent of the midcheek suborbital ("b") papillae row (10), and the number of teeth in the upper jaw (1). Four species, G. oceanicus, G. occidentalis, G. daguae, and G. liolepis share the forward extent of the "b" row with Gobioides. Gobionellus daguae and G. liolepis have an equal number of anal- and second dorsal-fin rays as seen in Gobioides. Gobionellus liolepis forms the sister group to Gobioides with which it shares confluent dorsal fins. Gobionellus occidentalis, G. oceanicus, and G. liolepis share pigmentary fea- tures not used in the analysis and approach Go- bioides in elongate body form, but G. daguae seems misplaced as it is reminiscent of Stenogo- bius with a stockier morphology and distinctive pigmentation. The fourth neural spine in G. daguae, although expanded, is not rounded dis- tally, and lacks the spatulate form seen in these other species of Gobionellus. All six species of Gobionellus also have a verti- cal posterior opercular papillae row, and elon- gate gill rakers on the anterior surface and lobes on the posterior surface of the epibran- chial of the first gill arch. All species but G. sto- matus have complete neural arches. It has been suggested by I. J. Harrison (unpubl.) that the presence of lobes on the posterior surface of the first epibranchial could be a synapomorphy for Gobionellus. Whether or not the morphology of the posterior surface of the first epibranchial is a synapomorphy for the genus remains to be tested. It does appear to differ from that ob- served for other Stenogobius group gobionel- lines, but more information is required on the details of gill arch morphology (beyond the an- terior surface of the ceratobranchial and epi- branchial of the first arch) to better define char- acters and assess the homologies and polarities of their associated states. Gobionellus is not combined with Gobioides here because of the clear synapomorphies de- limiting Gobioides, the relatively few phylogenet- ically informative characters available to this study and the likelihood that possibly two char- acters (shape of the fourth neural spine and posterior epibranchial lobes) may be synapo- morphic for Gobionellus. For the sake of nomen- clatural stability, it is best to continue to recog- nize Gobioides and Gobionellus until more sub- stantial support is offered for one alternative or the other. The other group, comprising 15 species, is Ctenogobius; C. fasciatus is the type species. Al- though shown in the consensus tree as part of a polytomy including Oxyurichthys and an Evor- thodus/Oligolepis clade, one of the two most par- simonious trees supports monophyly based upon a derived condition for the first gill arch (Fig. 2). The epibranchial of the first gill arch lacks lobes and rakers but has small tufts of pa- pillae (Fig. 3). The ceratobranchial has four or five broad, triangulate rakers. The relationship between Ctenogobius and Evorthodus is consistent with the molecular phylogeny of gobioid fishes presented by Thacker (2003). In her phylogeny, which did not include Gobionellus, Gobioides, Ox- yurichthys or Oligolepis, Evorthodus, and Ctenogo- bius formed a clade sister to a clade of three Gnatholepis species. The Evorthodus-Ctenogobius- Gnatholepis clade of the molecular phylogeny is in turn sister to Awaous-Stenogobius and sicydi- ines. It is also of note that Thacker's molecular phylogeny supports the recognition of a Steno- PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 273 gobius group of Gobionellinae as proposed by Larson (2001), distinct from Mugilogobius and the northern Pacific species. In fact, by her anal- ysis, the amblyopines and oxudercines form a sister group to the Stenogobius group gobionel- lines and sicydiines. In addition to the synapomorphy of the epi- branchial structure of the first gill arch, Cteno- gobius is characterized by a combination of oth- er features shared with other members of the polytomy: an abbreviated cephalic lateralis ca- nal with an A'BCDFH' pattern, a diagonal pos- terior opercular papillae row and reduced neu- ral arches over the caudal vertebrae (reversed in one species). Most species have a simple fourth neural spine, but it is broad-based and triangulate in six species (Fig. 6). The polarity of this character is ambiguous. Contrary to the cladograms, geographic character-state distri- butions suggest a recent derivation of the sim- ple spine, as a reversal from a basally broad, tri- angulate spine. As noted by Pezold and Gilbert (1987), both eastern Pacific species of Ctenogo- bius, the sole west African species, C. Upturns, and three of the western Atlantic species (C. phenacus, C. saepepallens, and C. smaragdus) have a broad-based triangulate fourth neural spine. The triangulate spine found in Oligolepis stomias parallels the condition found in Ctenogobius. Species assigned to Ctenogobius are Ctenogo- bius boleosoma, Ctenogobius claytoni, Ctenogobius fasciatus, C. lepturus, C manglicola, Ctenogobius phenacus, Ctenogobius pseudofasciatus, C saepepal- lens, Ctenogobius sagittula, Ctenogobius shufeldti, Ctenogobius smaragdus, Ctenogobius stigmaticus, C. stigmaturus, and Ctenogobius thoropsis. An unde- scribed species from Brazil is also included in this genus. Gobionellus atripinnis and Gobionellus comma are regarded as synonyms of C. claytoni and C. saepepallens, respectively. A full descrip- tion of the genus Ctenogobius and a review of and key to its species will be given elsewhere. Gobioides is distinguished by two synapomor- phies: a 3-12201 first dorsal pterygiophore pat- tern (Birdsong et al., 1988; Murdy, 1998) and an oculoscapular canal extending from near the anterior nares to the rear margin of the opercle with an A'BDFHKL' pore pattern. As indicated by the pattern, paired nasal canals are present and separate between the orbits and the ante- rior interorbital pores (C) are absent. Most spe- cies also have confluent median and caudal fins. As noted above, the genus shares several de- rived conditions with Gobionellus. Murdy (1998) has reviewed the species of this genus. Species of Stenogobius are united by two syna- pomorphies in the cladogram?the anterior na- res are medial to the oculoscapular canals (Pe- zold, 1991) and the raker morphology of the epibranchial of the first gill arch. The lateralis canals also terminate high on the snout near the posterior nares and the first gill arch has very small gill rakers which are united by a low membrane on the ceratobranchial (Watson, 1991; this study). Both of the latter features are also considered possible synapomorphies. The genus Oligolepis is delimited by two syn- apomorphies: transverse suborbital papillae rows of the cheek are reduced to form a lon- gitudinal pattern and the preopercular canal has only two terminal pores (M'O'). The epi- branchial of the first gill arch has a single, sim- ple fleshy lobe on the anterior surface (Fig. 3), which may also be synapomorphic in that the conditions observed in Evorthodus and Oxyuri- chthys include either multiple structures on the epibranchial or if a single lobe is present, the lobe has a fingerlike, bifid or trifurcate struc- ture. Species included in this genus are Oligole- pis jaarmani (= Oxyurichthys jaarmani), Oligolepis nijsseni (= Oxyurichthys nijsseni), Oligolepis acuti- pennis and Oligolepis stomias (= W. stomias). The genus Waitea has been distinguished by an ex- tremely elongate jaw and a lanceolate caudal fin (Smith, 1941). These states occur only in males. Similar sexual dimorphism, although not as ex- treme, occurs in some species of Ctenogobius and in Oxyurichthys keiensis. Waitea is placed in syn- onymy here with Oligolepis. The Oligolepis sp. il- lustrated by Akihito et al. (1984) appears to be Ol. stomias. Oligolepis jaarmani and Ol. nijsseni do not possess any of the apomorphies described below for Oxyurichthys. The genus Evorthodus was diagnosed by Gins- burg (1931) and Dawson (1967). Teeth primar- ily form a single row in both jaws, particularly in females and juveniles. Teeth are incisiform and truncate in juveniles, bifid or entire, but incisiform in females and conical or caniniform in adult males (Dawson, 1967, 1969). Although this character was coded as an identical state to the single row of teeth observed in Oxyurichthys, Oligolepis, G africanus and two species of Gobi- onellus, the presence of incisiform teeth in ju- veniles and females is synapomorphic for the genus Evorthodus, and a single epural appears as a synapomorphy for Evorthodus in this analysis. Epural number is a labile character in gobioid fishes showing intraspecific variation in the re- lated gobionelline genera Ctenogobius and Oli- golepis (Birdsong et al., 1988) and was coded in its modal form in this study and by Larson (2001). Murdy (1989) described retractor dorsalis muscle and fifth ceratobranchial structures that may also prove synapomorphic for Evorthodus. The genus Oxyurichthys contains about 16 spe- 274 COPEIA, 2004, NO. 2 cies, all of which share two synapomorphies: a transversely broad and bifid third neural spine (Fig. 5) and no preopercular canal (F. Pezold and H. K. Larson, unpubl. data). All species of Oxyurichthys, except for Ox. keiensis, also share three other synapomorphies: a sharply rounded fleshy tongue, a single row of teeth in the upper jaw (with partial second rows appearing in a few species) and a shortened palatine which does not form part of the lower palatopterygoquad- rate strut. All but Ox. keiensis also show a reversal to complete neural arches over the caudal ver- tebrae. Although all of the representative spe- cies in the cladogram other than Ox. keiensis have a membranous crest on the nape, it is also lacking in other members of the genus (Pezold, 1998). Other diagnostic characters for the ge- nus, excluding Ox. keiensis, are given by Pezold (1991). Gobionellus stigmalophius exhibits all five of the synapomorphies listed above and is also reassigned to Oxyurichthys. Oxyurichthys also has an elongate longitudinal suborbital neuromast row (row b). A review of the species of this ge- nus is being completed with Helen Larson. MATERIAL EXAMINED Comparative materials examined for postcra- nial osteology are listed in Birdsong et al. (1988) and those studied for cephalic lateralis structure and free sensory papillae distribution patterns are listed by Pezold (1993). Species of Rhyacichthyidae, Odontobutidae, Butinae (Eleotridae) and Stenogobius group gobionel- lines (Gobiidae) examined are given below with the collection number followed by the number of individuals in parentheses. Additional infor- mation was obtained from Hoese and Gill (1993), Miller (1973) and Springer (1983). Specimens of Gobionellus examined are given in Pezold (2004), whereas specimens of Ctenogobius are listed below by species, region, and museum catalog number with the number of individuals given in parentheses. Type material is indicated with the original binomen. Institutional abbre- viations are as in Leviton et al. (1985). Rhyacichthyidae. Rhyacichthys aspro: CAS 32758 (1), USNM 247300 (5). Odontobutidae. Micropercops dabryi: USNM 83982(1). M:cropfrcofw sp.: USNM 112474(6), USNM 112475(2), USNM 112508(1). Odontobu- tis obscurus: USNM 86108(1), USNM 71419(5), USNM 84004(1), USNM 86412(5). Percottus glehni: USNM 86108(1), USNM 105188(1). Per- co(At;/,kAgK USNM 77008(1). Butinae. Bostrychus africanus. CAS-SU 40431(1). Bostrychus sinesis: USNM 46802(1), USNM 576093(1), USNM 85907(6). gw6s amAmfrngnaK USNM 51953(1); USNM 272625(3). Ad6 Wis USNM 135895(1); USNM 161618(4), USNM 261350(13), USNM 268461(7), ANSP 63023-9 (17). Butis gymnopomus. USNM 161176(4), USNM 161177(1). Butis koilomatodon: USNM 161233(3). Bostrychus melanopterus. USNM 87928(2). Kribiahri- krwic CAS-SU 63034(1), CAS-SU 63035 (1), USNM 118789(3), USNM 118790(1). Odonteleotris sp.: MCZ 49560(2). C%,A6x3ra/wnx#Wa: CAS-SU 38579(4). OrygW^s &7Wa&c CAS SU 25582(4). Oryg&m&i; marmorafa: USNM 230238(2), ANSP 87352(3), CAS 49455(2). Oa^Wz^ sp.: USNM 103362(1), USNM 119618(1). Parviparma stra- mimac CAS-SU 29701(1). Gobionellinae (Stenogobius group). Awaous gui- neensis: CAS-SU 55635 (2). Awaous stamineusr. ANSP 29510-13(4), CAS 52267(2). Awamw 6a- nama: ANSP 144525 (3), CAS-SU 18573(3), FMNH 93277(3), UF 30510(1), USNM 272622(1), UTMSI 334(3). Awaous transandeanus: CAS 42775(2), TNHC 11519(1), TNHC 11506(1), TNHC 11509(1), TNHC 11511(1). Awaous sp. ANSP 149484(1). Ewnfhx&w fy^cws CAS 57067(1), CAS 52392(3), CAS 52394(3), TCWC 3283.1(3), TNHC 10623(29), UF 100059(12), USNM 144040(1), UNOVC 4306(15). GnafW^aM/gnm- sis. USNM 126530(2). Gnatholepis cauerensis: CAS 51548 (1). Gnatholepis thompsoni: CAS-SU 8364(1), UMMZ 174286(5). Gobioides africanus: BMNH 1939.7.12.33 (1). Gobo6&s amaorgiK BMNH 1909.10.29.110-112(3), BMNH 1968.11.15.77 (1). G0&0665 6rmi?07K6: ANSP 121256(2), CASSU 21381(1), USNM 233612 (11). Gobioidesgrahamae. BMNH 1925.10.28.464(1), BMNH 1925.10.28. 465(1), BMNH 1950.5.15.41(1), BMNH 1959.3. 17.161 (1). Gobioides peruanus: USNM 123616(1). G0&0665 aagd&z BMNH 1862.1.24.27.29(3). 06go- lepis acutipennis: UMMZ 100537(25), USNM 139345(1), RMNH 14325(3). 0&gD&^;aarma7ir USNM 217267(8), USNM 217266(2). 06go&^ nijsseni: ZMA 115270(2). Oligolepis stomias. USNM 51816(1), USNM 257137 (13), USNM 258782(9), USNM 99296(1). Oxyurichthys auchenokpis: RMNH 4506(2). Oryw,6A%; wrmwAw: CA&SU 33137(10). Oxyurichthys heisei: NLU 64915(2). Oxyurichthys keiensis: RUSI 17043(8), RUSI 16786(1). Oxyuri- didi^ kMfAoAw: ANSP 23350(1), ANSP 90998(3), ANSP 28055-56(2), CAS 23328(14), CAS 51062(17), UMMZ 196868(1), USNM 126533(1), USNM 50698(1). Oxyurichthys microlepis: ANSP 88946(1), UMML 14353(4), UMMZ 100268(4), UMMZ 100539(5). Oxyurichthys ophthalmonema: FMNH 91547(10). Oxyurichthys papuensis: BPBM 7354(2), LACM 37382-2(1). OxymiddAysfawJae USNM 346922(2). Oxyurichthys stigmalophius: ANSP 81233(1), ANSP 81855(1), ANSP 144295(1), UMML 3992(1). OxyuT^ifAy; kmfacw- laris. ANSP 100179(1), NLU 71393(9). Oxyuri- PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 275 c%A)? kdbgr CAS 51047(13), CAS 51059(4). Skn- ogD^t; gKM^tza^ ANSP 86151(9); ANSP 28016- 19(4), BPBM 26373(4), BPBM 26380(2), CAS 51056(7), CAS 52011(3), USNM 99878(1). Steno- gobius gymnopomus: RMNH 4552(4). Stenogobius la- terisquamatus: ZMA 116477(2); WAM P27847- 007(3), WAM P28206-002(l), NLU 62766(4). Ctenogobius boleosoma. New Jersey: ANSP 130090(1). Delaware: ANSP 73745(1). Bermuda: MCZ 32987(1). North Carolina: USNM 123295(1); USNM 123301(2); USNM 123298(6); USNM 123299(1). South Carolina: ANSP 142663(24); ANSP 142664(1); USNM 265067(3); USNM 133064(3); USNM 29673(3), syntypes, Go6- ius encaeomus. Georgia: ANSP 71066(1). Florida: UF/FSU 13575(115); ANSP 96801(1); ANSP 113042(1); USNM 123305(4); USNM 265068(1); USNM 123303(3); USNM 30860(40), syntypes, Gobius boleosoma. Alabama: USNM 127465(1). Mis- sissippi: USNM 265008(1). Louisiana: USNM 123319(1). Texas: ANSP 115760(1); ANSP 73643(2); ANSP 115719(1); ANSP 99216(2); ANSP 73849(3). Mexico: GCRL 2879(18); USNM 192269(1). Belize: USNM 265001 (1). Bahamas: ANSP 98654(1). Cuba: USNM 265011(1); USNM 178955(1). Jamaica: USNM 265071(1). Domini- can Republic: USNM 265069(1). Puerto Rico: UMMZ 172793(16); ANSP 144506(1); ANSP 144488(3); USNM 86910(1); USNM 55695(1); USNM 114657(1). Guadeloupe: ANSP 144487(2). Martinique: ANSP 113089(5). Panama: GCRL 4669(66); GCRL 12227(1); ANSP 146903(1); USNM 148716(1); USNM 123344(2); USNM 265000(1); USNM 205204(1); USNM 226379(1); USNM 81825(1). Colombia: GCRL 4787(24); USNM 38655(3). Curacao: ANSP 144507(4). Ve- nezuela: GCRL 15513(9); USNM 123273(1). BRA- ZIL: MAPA 1502(6 [of 35]); AMNH 3836(13); MO-FURG 80-111(3); ANSP 121172(3); ANSP 121173(1); ANSP 121182(2); AMNH 20746(1); AMNH 20705(1). Ctenogobius claytoni. Texas: UMMZ 167639(1), holotype, Gabionellus atripinnis. Tamaulipas: SU 68864(2). Vera Cruz: FMNH 3740(1), holotype, C. claytonii; FMNH 3741, 16900-16902(4), paratypes, Gobius claytonii; FMNH 16903-16906(4), para- types, G. claytonii; FMNH 4572, 16907-16910(5); UMMZ 184472(2); UMMZ 184456(1); UMMZ 184609(1); TNHC 11277(1); TNHC 11287(6); UMMZ 181796(7; one cleared and stained), para- types, G. atripinnis; UMMZ 187725(4), paratypes, G. atripinnis; UMMZ 187703(7); UMMZ 97727(3); UMMZ 187763(1), paratype, G. atripinnis. Ctenogobius fasciatus. Florida: IRCZM 7544 (2). Honduras: UMMZ 199685(1). Costa Rica: TU 24861(2); TU 24877(28); UMMZ 180655(3); UF 11139(10); UF 11176(5); UF 10268(6); UF/FSU 17695(3); UF/FSU 17624(4); UF/FSU 17727(1); UF/FSU 17670(1); UF/FSU 17645(4). Panama: USNM 81874(1); USNM 81875(4); USNM 81876(2); USNM 81819(3); USNM 148715(1); SU 18574(10); FMNH 32186-32188(3); FMNH 32184(1); GCRL 7846(15); GCRL 12773(5); GCRL 10282(2); GCRL 10266(2); GCRL 12777(9); GCRL 3280(1); UF 35956(1); ANSP 122357(1); ANSP 122358(1). Venezuela: UMMZ 147507(31); UMMZ 147536(1); USNM 194104(1). Trinidad: USNM 7549(1), lectotype, C. fasciatus; USNM 198110(1), paralectotype, C. fasciatus; AMNH 26394(2). Barbados: ROM 36366(91); ROM 36216(88); ROM 24325(2); ROM 36363(1). Dominica: USNM 199703(1); USNM 199704(2); USNM 199705(1). Domini- can Republic: UF 30402(1). Haiti: UMMZ 167222(7). Ctenogobius Upturns. Ghana: USNM 264991(2). Congo: MNHN 1967-416(28). Ctenogobius manglicola. Mexico: SU 3095(1), ho- lotype, Go66w mamg&mb; GCRL 4421(23); GCRL 2771(9); GCRL 2658(35). Guatemala: GCRL 5849(57). El Salvador: GCRL 5030(17). Costa Rica: GCRL 3529(180). Panama: AMNH 73935 (60); USNM 123251(2); USNM 81826(2); USNM 101374(1); USNM 119328(2); USNM 123250(6); USNM 123248(15); USNM 123347(17). Colom- bia: GCRL 5142(15); USNM 123252 (3). Peru: GCRL 22308(20). Ctenogobius sp. Piaui: MCZ 46857(2). Espirito Santo: FMNH 93278(2); FMNH 93258(13); FMNH 92356(1); FMNH 93268(4). Rio De Ja- neiro: FMNH 86668(5); UF 19210(2); UF 19209(2); ANSP 121210(2); ANSP 121211(1). Rio Grande Do Sul: UF34127(2); UF 34129(4); MO-FURG 80-150(2); DZUFRGS 0952(1); UF 34128(1); MAPA 1718(3); MAPA 1498(1) DZUFRGS 1023(3); DZUFRGS 0701(1) DZUFRGS 0946(1); MO-FURG 80-34(6 of 35) MAPA 1501(2). Ctenogobius phenacus. French Guiana: UF 34132(1), holotype, Gobionellusphenacus; USNM 244153(3), paratypes; USNM 264990(7), para- types. Surinam: USNM 226247(2), paratypes; USNM 226248(1), paratype. Venezuela: MBUCV- V 14128(2), paratypes. Ctenogobius pseudofasciatus. Trinidad: UF 31201(1), paratype. Venezuela: UMMZ 147535(1). Panama: FMNH 32178(1); USNM 81824(1), paratype; USNM 105109(1), para- type; USNM 123264(1), paratype; USNM 205202(1), paratype. Costa Rica: USNM 201589 (1), paratype; ANSP 109179(1), paratype; UF/ FSU 17696(1), paratype; UF 13517(1); para- type; UF 13518(1), paratype; UF 13519(1), paratype; UF 13520(1), paratype. Honduras UMMZ 199544(1). Belize: FMNH 82076(1) FMNH 86680(1). Florida: TNHC 10859(1) 276 COPEIA, 2004, NO. 2 UF100057(13); IRCZM 5086(4); IRCZM 5102(7); IRCZM 7358(11); IRCZM 7363(5); IRCZM 7543(3). Ctenogobius saepepallens. Brazil (Bahia): GCRL 10917(1). Trinidad: MCZ 58720(1). Venezuela (Isla Cubagua): ANSP 109181(7), holotype, Go- bionellus comma; LACM 20634(1), paratype, G. comma; LACM 20635(1), paratype, G. comma; UF 12793(1), paratype, G. comma. Panama: MCZ 47646(1); GCRL 4668(1); MCZ 58718(14). Co- lombia (Isla Providencia): UF 24450(1); UF 19103(6); UF 25851(1). Belize: FMNH 86618(1); FMNH 77684(5); FMNH 77681(5); FMNH 77685(1); FMNH 77687(3); FMNH 77682(5); FMNH 77686(3); FMNH 77683(5). Mexico (Cozumel): UMML 9458(2). Grand Cayman: UF 13521(25), paratypes, Gobionellus saepepallens; UF 13523(2), paratypes, G. saepepal- lens; UF 13522(5), paratypes, G. saepepallens; ANSP 109177(3), paratypes, G. saepepallens; ANSP 109178(20), paratypes, G. saepepallens. Puerto Rico: UPR 2412(4), paratypes, G saepe- pallens; UF 23017(12); USNM 114656(1). Virgin Islands: UPR 1766(1), paratype, G saepepallens; GCRL 1999(2); UF 13524(1), paratype, G sae- pepallens. Antigua: CAS 37270(1); UF 12759(1), paratype, G. saepepallens; UF 11304(7), para- types, G. saepepallens. Dominica: USNM 199706(1), paratype, G. saepepallens. Bahamas: ANSP 147349(5); UMMZ 186507(2), paratypes, G. saepepallens; ANSP 100519(38), paratypes, G saepepallens; ANSP 109180(1), holotype, G sae- pepallens; ANSP 86135(1), paratype, G saepepal- lens; AMNH 25792(2) paratypes, G saepepallens; AMNH 24936(59); FMNH 73908(2), paratypes, G. saepepallens; USNM 201590(2), paratypes, G. saepepallens. Florida: ANSP 84784(1); ANSP 96802(1); UF 7050(1); USNM 167676(2). North Carolina: MPM 33435(1); MPM uncat. stn. 52 (1); MPM uncat. stn. 87 (1); MPM uncat. stn. 76(2). Ctenogobius sagittula. California: SU 9893(4); SU 12944(1). Mexico: AMNH 5558(6); CAS 51045(4); LACM 1025(35); LACM 34081-2(5); FMNH 57530(3); MCZ 27881(1); SU 169(8); UMMZ 178590(2); UMMZ 184865(27); UMMZ 172256(15); USNM-BOC 2755(2); USNM 39636(3), syntypes, Gobius longicaudus; USNM 59456(7); USNM 46655(1); USNM 123265(6); USNM 214515(1); USNM 30936(7); USNM 43740(3); USNM 265002(1). El Salvador: FMNH 12018(2); FMNH 93706(2); GCRL 16554(4); GCRL 16562(1); USNM 87200(3); USNM 220642(2). Costa Rica: FMNH 86669(3); FMNH 91226(1); LACM 2700(1); LACM 2889(2). Panama: ANSP 151059(1); FMNH 8469(3); MCZ 46472(2); MCZ 46482(1); UF 16208(10); UMMZ 180724(2); USNM 81961(2); USNM 79013(3); USNM 123260(1); USNM 123258(2); USNM 265005(1); USNM 123259(1); USNM 81821(1); USNM 81818(1); USNM 81820(3). Central America: BMNH 1861-8-13- 26(1), holotype, Euctenogobius sagittula. Colombia: CAS 51041(2); CAS 64211(2); FMNH 58481(2); FMNH 86693(10); USNM 257662(7); USNM 257678(235+); USNM 265085(1); USNM 257667(3); USNM 257679(243+). Ecuador: SU 9291(4); USNM 88785(2). Ctenogobius shufeldti. North Carolina: UMMZ 126274(6); USNM 123238(2); USNM 123240(2). South Carolina: USNM 59074(8); USNM 123237(1); USNM 123244(1); ANSP 149878(3); UMMZ 155196(3). Georgia: GCRL 16980(2); UF 25089(1); UMMZ 155219(7); USNM 298633(30); USNM 346221(16); USNM 131223(1). Florida (Adantic Coast): IRCZM 5100(5); IRCZM 5101(2); TNHC 10834(1); UF 7585(4); UF 7741(8); UF 7742(1); UF 7743(1); UF 7744(3); UF 7745(4); UF 7746(1); UF 7747(14); UF 7748(11); UF 7749(3); UF 7750(2); Florida (Gulf Coast): UF 4381 (7); UF/FSU 5334(7); TU 103016(2); TU 120046(1); TU120073(1); UMMZ 163443(3); ANSP 72835(2); ANSP 72981(1); ANSP 73030(13); ANSP 73069(1). Al- abama: UMMZ 163590(28). Mississippi: ANSP 55812-15(4). GCRL 2066(1); GCRL 2067(1) GCRL 2777(1); GCRL 2778(1); GCRL 2779(3) TU 122420(2); TU 122465(5); TU 122501(8) UMMZ 155431(2); UMMZ 163654(4). Louisi- ana: USNM 35202(12), syn types, Gobius shufeldti; USNM 123239(1); USNM 123241(1); USNM 123249(1); ANSP 70796(1); FMNH 51061(2); FSC 10872(47); TU 266(18); UF 33927(3); UMMZ 155326(9); UNOVC 656(708); UNOVC 763(15); UNOVC 769(33); UNOVC 787(223); UNOVC 1232(3); UNOVC 1770(22). Louisiana- Texas (Sabine Lake): ANSP 99078(7); ANSP 115695(2). Texas: TCWC 1619.1(1); USNM 123245(1). Ctenogobius smaragdus. Florida (Atlantic Coast): IRCZM 309(4); IRCZM 2560(4); IRCZM 2606(5); IRCZM 4842(5); TNHC 10860(16); TNHC 10884(1); UF 4807(1); UF 7756(3); UF 11602(14); UF 19336(1); UF 28776(1); UF 28778(1); UMMZ 189754(10). Florida (Gulf Coast): ANSP 71056-65(10); ANSP 71067(1); CAS 51043(7); LACM 1448(6); UF/FSU 24720(7); UF 3433(5); UF 9231(1). Cuba: MNHN 1257(1), holotype, Gobius smaragdus; USNM 4769(2); USNM 37461(1); USNM 264987(1). Virgin Islands: USNM 78150(1). Be- lize: FMNH 86666(1); FMNH 86677(1). Vene- zuela: GCRL 15514(2). Brazil: CAS-SU 52379(1); GCRL 9621(46); MCZ 4624(1); MCZ 13077(1); SU 52386(1); UF 19211(20). Ctenogobius stismaticus. Florida: TNHC PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 277 10703(1); UF 30509(1); UMML 13450(1). Mis- sissippi: GCRL 13824(3). Texas: NLU 69898(4). Cuba: MCZ 13104(1), holotype, Smaragdus stig- maficus MCZ 13122(3); MCZ 13923(2); SU 1936(6). Antigua: USNM 170308(1). Guade- loupe: ANSP 144494(1). Honduras: FMNH 86678(1). Brazil (Rio DeJaneiro): UF 19212(1); UF 19903(1); MCZ 4622(22); UMMZ 201445(2). Ctenogobius stigmaturus. Bermuda: ANSP 148421(1); ANSP 148422(4); ANSP 148423(1); ANSP 148424(14); GCRL 19604(6); GCRL 19605(2); USNM 178903(2); USNM 178904(2). Florida: ANSP 96784(1); CAS 52007(1); IRCZM 1750(4); IRCZM 2564(60); IRCZM 4005(6); UF/FSU 13247(1); UF/FSU 9152(5); UF/FSU 13245(2); UF/FSU13255(3); UE/FSU 11467(1); UF 7081(43); UF 37133(59); UMML 1446(6); UMMZ 189866(10); USNM 35004(1); USNM 57330(31); USNM 57365(4); USNM 57431(2); USNM 57450(8); USNM 65327(1); USNM 73097(3); USNM 89868(1); USNM 89869(2); USNM 89870(1); USNM 89871(1); USNM 89872(2); USNM 89873(1); USNM 89883(1); USNM 89884(1); USNM 264985(1). Cuba: USNM 82512(2). Belize: AMNH 24615(5). Panama: AMNH 73937(16). Ctenogobius thoropsis. Surinam: FMNH 90554(1), holotype, Gobionellus thoropsis; FMNH 94890(2), paratypes. Brazil: USNM 214066(1), paratype; USNM 264992(1), para type. ACKNOWLEDGMENTS This work is based upon a dissertation com- pleted at the University of Texas at Austin. It differs in analysis, but the conclusions are the same. I thank C. Hubbs, J. Rawlins, R. Buskirk, J. Gold, and C. R. Gilbert for assistance during my graduate years. My understanding of go- bioid systematics profited from discussions and correspondence with R. S. Birdsong, C. R. Gil- bert, D. Hoese, E. Lachner, H. Larson, P. J. Mill- er, E. O. Murdy, V. G. Springer, and R. Winter- bottom. R. C. Cashner, D. Hoese, H. Larson, L. Parenti, and R. Winterbottom read earlier ver- sions of this manuscript. The conclusions of this paper are my own. Many individuals at a num- ber of institutions supplied specimens and in- formation on materials in their care: W. Esch- meyer, T Iwamoto, S. Poss, P. Sonoda, and M. Hearne, CAS; B. Chernoff, W. Smith-Vaniz, and E. Bohlke, ANSP; R. R. Miller and A. S. Snyder, UMMZ; C. R. Gilbert, G. H. Burgess, and J. B. Miller, UF; K. E. Hartel, MCZ; A.Wheeler, M. Holloway, and J. Chambers, BMNH; ML. Bauchot, MNHN; V. G. Springer, S. Jewett, and J. T Williams, USNM; R. K.Johnson, D.J. Stew- art, and T Grande, FMNH; C. E. Dawson, GCRL; M. J. P. van Oijen, RMNH; J. Randall, BPBM; R. Winterbottom, ROM; J. Nielson, ZMK; D. Mosier, TNHC; T van den Audenaerde and M. Louette, MRAC; H. Nijssen, ZMA; P. C. Heemstra, RUSI; R. G. Gilmore, Harbor Branch Foundation; and G. Allen, WAM. Figure 8 was illustrated by P. Regan. S. Recoulley and R. Min- ton assisted with figures. Travel support was re- ceived from Sigma Xi, the Smithsonian Institu- tion, the Academy of Natural Sciences of Phil- adelphia and the California Academy of Scienc- es, and by a University Fellowship from the Graduate School, the University of Texas at Aus- tin. LITERATURE CITED AHNELT, H. 2001. Two Mediterranean gobiid fishes with an unusual cephalic lateral line canal system. Cybium 25:261-267. PRINCE AKIHITO. 1969. A systematic examination of the gobiid fishes based on the mesopterygoid, post- cleithra, branchiostegals, pelvic fins, scapula and suborbital. Jpn. J. Ichthyol. 16:93-114. . 1986. Some morphological characters consid- ered to be important in gobiid phytogeny, p. 629- 639. In: T. Uyeno, R. Aral, T. Taniuchi, and K. Mat- suura (eds.). Indo-Pacific fish biology: proceedings of the second international conference on Indo-Pa- cific fishes, Ichthyological Society of Japan, Tokyo. , AND K. MEGURO. 1980. On the six species of the genus Bathygobius found in Japan. Jpn. J. Ich- thyol. 27:215-236. , M. HAYASHI, AND T. YOSHINO. 1984. Suborder Gobioidei, p. 236-289, pi. 235-258. In: The fishes of the Japanese Archipelago. H. Masuda, K. Amao- ka, C. Araga and T. Uyeno (eds.). Tokai Univ. Press, Tokyo. BIRDSONG, R. S. 1975. The osteology of Microgobius sig- natus Poey (Pisces: Gobiidae), with comments on other gobiid fishes. Bull. Fla. St. Mus. Biol. Sci. 19: 135-187. . 1988. Robinsichthys arrowsmithensis, a new ge- nus and species of deep-dwelling gobiid fish from the western Caribbean. Proc. Biol. Soc. Wash. 101: 438-^43. , E. O. MURDY, AND F. L. PEZOLD. 1988. A study of the vertebral column and median fin osteology in gobioid fishes with comments on gobioid rela- tionships. Bull. Mar. Sci. 42:174-214. BREMER, K. 1994. Branch support and tree stability. Cladistics 10:295-304. DAWSON, C. E. 1967. Notes on the species of the goby genus Evorthodus. Copeia 1967:855-857. . 1969. Studies on the gobies of Mississippi Sound and adjacent waters II, an illustrated key to the gobioid fishes. Publ. Gulf Coast Research Lab. Museum, Ocean Springs, MS. FARRIS, J. S. 1982. Outgroups and parsimony. Syst. Zool. 31:328-334. GILBERT, C. R., AND J. E. RANDALL. 1979. Two new 278 COPEIA, 2004, NO. 2 western Atlantic species of the gobiid fish genus Gobionellus, with remarks on characteristics of the genus. Northeast Gulf Sci. 3:27-47. GINSBURG, I. 1931. Juvenile and sex characters of Evor- thodus lyricus (fam. Gobiidae). Bull. U.S. Bur. Fish. 47:117-124. . 1932. A revision of the genus Gobionellus (fam- ily Gobiidae). Bull. Bingham. Oceanogr. Coll. 4:1- 51. . 1953. Ten new American gobiid fishes in the United States National Museum, including addi- tions to a revision of Gobionellus. J. Wash. Acad. Sci. 43:18-26. HARRISON, I. J. 1989. Specialization of the gobioid pal- atopterygoquadrate complex and its relevance to gobioid systematics. J. Nat. Hist. 23:325-353. HOESE, D. F. 1983. Sensory papilla patterns of the cheek lateralis system in the gobiid fishes Acentro- gobius and Glossogobius, and their significance for the classification of gobioid fishes. Rec. Aust. Mus. 35:195-222. , AND G. R. ALLEN. 1977. Signigobius biocellatus, a new genus and species of sand-dwelling coral reef gobiid fish from the western tropical Pacific. Jpn.J. Ichthyol. 23:199-207. , AND A.C. GILL. 1993. Phylogenetic relation- ships of eleotridid fishes (Perciformes: Gobioidei). Bull. Mar. Sci. 52:415-440. -, AND R. STEENE. 1978. Amblyeleotris randalli, a new species of gobiid fish living in association with alphaeid shrimps. Rec. West. Aust. Mus. 6:379-389. , AND R. WINTERBOTTOM. 1979. A new species of Lioteres (Pisces: Gobiidae) from Kwazulu, with a revised checklist of South African gobies and com- ments on the generic relationships and endemism of western Indian Ocean gobioids. R. Ont. Mus., Life Sci. Occ. Pap. 31:1-13. ILJIN, B. S. 1930. Le systeme des gobiides. Trab. Inst. Esp. Oceanogr. 2:1-63. IWATA, A., S.-R. JEON, N. MIZUNO, AND K.-C. CHOI. 1985. A revision of the eleotrid goby genus Odon- tobutis in Japan, Korea and China. Jpn. J. Ichthyol. 31:373-388. LACHNER, E. A, AND J. F MCKINNEY. 1979. Two new gobiid fishes of the genus Gobiopsis and a redescrip- tion of Feia nympha Smith. Smiths. Contrib. Zool. 299:1-18. LARSON, H. K. 2001. A revision of the gobiid fish ge- nus Mugilogobius (Teleostei: Gobioidei), and its sys- tematic placement. Rec. West. Aust. Mus. Suppl. 62: 1-233. LEVITON, A. E., R. H. GIBBS JR., E. HEAL, AND C. E. DAWSON. 1985. Standards in herpetology and ich- thyology. Part I. Standard symbolic codes for insti- tutional resource collections in herpetology and ichthyology. Copeia 1985:802-832. LUNDBERG, J. G 1972. Wagner networks and ances- tors. Syst. Zool. 18:1-32. MEAD, G W., AND J. E. BOHLKE. 1958. Gobionellus stig- malophius, a new goby from the Gulf of Campeche and the Great Bahama Bank. Copeia 1958:285-289. MEEK, S. E., AND S. F. HILDEBRAND. 1928. The marine fishes of Panama. Part III. Field Mus. Nat. Hist. Publ. Zool. Ser. v. 15 (publ. 249), 709-1045. MILLER, P. J. 1973. The osteology and adaptive fea- tures of Rhyacichthys aspro (Teleostei: Gobioidei) and the classification of gobioid fishes. J. Zool., Lond. 171:397-434. . 1981. Gobiidae, p. 1-8. In: Species identifi- cation sheets for fishery purposes, Eastern Central Atlantic; fishing areas 34, 47 (in part). Vol. 2. W. Fischer, G Bianchi and W.B. Scott (eds.). Depart- ment of Fisheries and Oceans Canada and FAO, Rome. . 1998. The West African species of Eleotris and their systematic affinities (Teleostei: Gobioidei). J. Nat. Hist. 32:273-296. , AND P. WONGRAT. 1979. A new goby (Teleos- tei: Gobiidae) from the South China Sea and its significance for gobioid classification. Zool.J. Linn. Soc. 67:239-257 , M. Y EL-TAWIL. R. S. THORPE, AND C.J. WEBB. 1980. Haemoglobins and the systematic problems set by gobioid fishes, p. 195-233. In: Chemosyste- matics: principle and practice. F. A. Bisby, J. G Vaughan, and C. A. Wright (eds.). Syst. Assoc. Spec. Vol. 16. Academic Press, London. MURDY, E. O. 1989. A taxonomic revision and cladistic analysis of the oxudercine gobies (Gobiidae: Oxu- dercinae). Rec. Aust. Mus. Suppl. 11:1-93. . 1998. A review of the gobioid fish genus Go- bioides. Ichthyol. Res. 45:121-133. , AND K. SHIBUKAWA. 2001. A revision of the gobiid fish genus Odontamblyopus (Gobiidae: Ambly- opinae). Ibid. 48:31-43. NIXON, K. C, AND J. M. CARPENTER. 1993. On out- groups. Cladistics 9:413-426. PARENTI, L. R., AND J. A. MACIOLEK. 1993. New sicy- diine gobies from Ponape and Palau, Micronesia, with comments on systematics of the subfamily Si- cydiinae (Teleostei: Gobiidae). Bull. Mar. Sci. 53: 945-972. PEZOLD, F 1991. The status of the gobioid genus Pa- roxyurichthys. Jpn. J. Ichthyol. 37:344-353. . 1993. Evidence for a monophyletic Gobiinae. Copeia 1993:634-643. 1998. Three new species of Oxyurichthys (Te- leostei: Gobiidae) from the Indian and Pacific Oceans. Ibid. 1998:687-695. . 2004. Redescriptions and synonymies of spe- cies of the American-West African genus Gobionellus (Teleostei, Gobiidae) with a key to species. Ibid. 2004:281-297. , AND B. CAGE. 2002. A review of the spinycheek sleepers, genus Eleotris (Teleostei: Eleotridae), of the Western Hemisphere, with comparison to the West African species. Tulane Stud. Zool. Bot. 31:19- 63. -, AND C. R. GILBERT. 1987. Two new species of the gobiid fish genus Gobionellus from the western Atlantic. Copeia 1987:169-175. RANDALL, J. E., AND D. F HOESE. 1986. Revision of Indo-Pacific dartfishes, genus Ptereleotris (Percifor- mes: Gobioidei). Indo-Pacific Fishes 7:1-36. ROBINS, C. R., AND E. A. LACHNER. 1966. The status of Ctenogobius Gill (Pisces: Gobiidae). Copeia 1966: 867-869. SANZO, L. 1911. Distribuzione delle papille cutanee PEZOLD? GOBIONELLUS PHYLOGENETIC ANALYSIS 279 (organ! ciatiformi) e suo valore sistematico nei gobi. Mitt. Zool. Stn. zu Neapel 20:251-328. SHIBUKAWA, K., A. IWATA, AND S. VIRAVONG. 2001. Ter- ateleotris, a new gobioid fish genus from the Laos (Teleostei, Perciformes), with comments on its re- lationships. Bull. Natl. Sci. Mus. (Tokyo), Ser. A 27: 229-257. SMITH, H. M. 1941. The gobies Waitea and Mahidolia. J. Wash. Acad. Sci. 31:409-415. SPRINGER, V. G. 1983. Tyson belos, new genus and spe- cies of western Pacific fish (Gobiidae, Xenisthmi- nae), with discussions of gobioid osteology and clas- sification. Smiths. Contrib. Zool. 390:1-40. TAKAGI, K. 1957. Descriptions of some new gobioid fishes of Japan, with a proposition on the sensory line system as a taxonomic character. J. Tokyo Univ. Fish. 43:96-126. . 1988 (1989). Cephalic sensory canal system of the gobioid fishes of Japan: comparative mor- phology with special reference to phylogenetic sig- nificance. Ibid. 75:499-568. THACKER, C. 2000. Phylogeny of the wormfishes (Te- leostei: Gobioidei: Microdesmidae). Copeia 2000: 940-957. . 2003. Molecular phylogeny of the gobioid fishes (Teleostei: Perciformes: Gobioidei). Mol. Phylo. Evol. 26:354-368. WATSON, R. E. 1991. A provisional review of the genus Stenogobius with descriptions of a new subgenus and thirteen new species. (Pisces: Teleostei: Gobiidae). Rec. West. Aust. Mus. 15:571-654. . 1992. A review of the gobiid fish genus Awaous from insular streams of the Pacific Plate. Ichthyol. Explor. Freshwat. 3:161-176. . 1996. Revision of the subgenus Awaous (Chon- ophorus) (Teleostei: Gobiidae). Ibid. 7:1-18. , AND C. POLLABAUER. 1998. A new genus and species of freshwater goby from New Caledonia with a complete lateral line. Senckenb. Biol. 77: 147-153. MUSEUM OF NATURAL HISTORY, UNIVERSITY OF LOUISIANA AT MONROE, MONROE, LOUISIANA 71209-0504. E-mail: pezold@ulm.edu. Submit- ted: 10 Oct. 2002. Accepted: 21 Jan. 2004. Sec- tion editor: D. G. Buth. 280 COPEIA, 2004, NO. 2 APPENDIX 1. CHARACTER STATE MATRIX FOR Gobionellus, Ctenogobius, RELATED GOBIONELLINES, AND COMPOSITE OUTGROUP. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Outgroup Gobionellus daguae G. liolepis G microdon G. occidentalis G oceanicus G stomatus Ctenogobius boleosoma C. daytoni C. fasciatus C. lepturus C. manglicola C. sp. (Brazil) C. phenacus C. pseudofasciatus C. saepepallens C. sagittula C. shufeldti C. smaragdus C. stigmaticus C. stigmaturus C. thoropsis Oxyurichthys keiensis Ox. heisei Ox. lonchotus Ox. microlepis Ox. ophthalmonema Ox. papuensis Ox. stigmalophius Oligolepis acutipennis Ol. jaarmani Ol. stomias Gobioides broussoneti Gobioides grahamae Evorthodus lyricus Stenogobius genivittatus S. laterisquamatus Awaous banana Gnatholepis thompsoni 0 0 0 0 0 0 0 0 ? 0 0 0 0 0 ? 0 ? 0 0 ? 1 0 1 0 0 0 2 0 0 1 1 0 0 3 0 0 1 0 0 1 1 0 1 0 0 0 2 0 0 1 1 0 0 4 0 1 1 0 0 1 0 0 1 0 0 0 2 0 0 0 1 0 0 4 0 0 2 0 0 1 0 0 1 0 0 0 2 0 0 1 1 0 0 4 0 0 2 0 0 1 0 0 1 0 0 0 2 0 0 1 1 0 0 4 0 0 2 0 0 1 0 0 1 0 0 0 2 0 0 1 1 0 0 4 0 0 2 0 0 1 0 0 1 1 0 0 0 0 0 0 1 0 0 4 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 0 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 0 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 0 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 1 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 1 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 0 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 1 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 0 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 1 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 1 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 0 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 1 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 0 0 0 2 0 1 1 0 0 1 2 0 0 4 0 0 0 0 0 0 0 0 0 2 0 0 1 0 0 1 1 0 0 4 2 0 1 0 0 1 0 0 0 2 0 1 1 1 1 2 1 0 0 4 2 0 1 0 1 1 0 0 0 2 0 0 1 1 1 2 1 0 0 4 2 0 1 0 1 1 0 0 0 2 0 0 1 1 1 2 1 0 0 4 2 0 1 0 1 1 0 0 0 2 0 0 1 1 1 2 1 0 0 4 2 0 1 0 1 1 0 0 0 2 0 0 1 1 1 2 1 0 0 4 2 0 1 0 1 1 0 0 0 2 0 0 1 1 1 2 1 0 0 4 2 0 1 0 1 1 0 0 0 2 0 0 1 1 0 1 1 0 0 5 1 1 1 0 0 0 2 0 0 2 0 1 1 1 0 1 1 0 0 5 1 1 1 0 0 0 2 0 0 2 0 1 1 1 0 1 1 0 0 5 1 1 1 0 0 0 1 0 0 2 0 1 1 0 0 1 0 0 0 3 0 0 1 1 0 0 0 1 1 1 1 0 1 1 0 1 0 0 0 3 0 0 1 1 0 0 0 1 1 1 1 0 1 1 0 1 1 0 0 5 0 0 0 0 0 0 2 0 0 2 0 1 2 0 0 1 3 1 1 5 0 0 0 1 0 0 0 0 0 1 0 0 1 0 0 1 3 1 1 5 0 0 0 1 0 0 0 0 0 1 0 0 1 0 0 1 0 1 0 1 0 0 0 1 0 0 0 0 0 1 0 0 2 0 0 1 0 0 0 5 0 0 0 0 0 0 0 0 0 2 0 0 1