JOHNSON: PERCOIDEI 



469 



but I have observed a similar condition in the gobiid Dormi- 

 tator.) 



Elassoma seems to exiiibit a confusing mosaic of character 

 states variously shared with atherinomorphs, cyprinodontoids, 

 percopsiforms, perciforms, and gobioids. Resolution of the evo- 

 lutionary affinities of this genus could be important to our un- 

 derstanding of acanthomorph interrelationships, and I intend 

 to examine this problem more fully. 



Epigonidae. — Eraser ( 1 972a) treated Epigonns. Florencwlla and 

 Rosenhlattia as a subfamily (Epigoninae) of the Apogonidae, 

 but I find no evidence to suggest that these genera are closely 

 related to other apogonids. They are primitive with respect to 

 apogonids in possessing two pairs of uroneurals and a procurrent 

 spur (Johnson, 1975), but specialized in several features listed 

 below. Moreover, the two anal spines of epigonines and apo- 

 gonids, usually cited as evidence of their close relationship, are 

 not homologous (see discussion on median fins). The Epigonidae 

 are here recognized as a distinct family, including Brinkmanella. 

 Sphyraenops and Eraser's epigonines. These five genera share 

 the following specializations: rostral cartilage greatly enlarged, 

 ascending processes of premaxillaries reduced or absent; pre- 

 maxillary articular cartilages enlarged; endopterygoids large, 

 metapterygoids notably reduced; infraorbitals more than six. 

 The larvae of Sphyraenops (Pig. 257A) resemble those of Epigo- 

 nus (Fig. 257B) but differ in possessing well-developed head 

 spination. 



Girellidae, Kyphosidae, Microcanthidae, Neoscorpis. Parascor- 

 pididae and Scorpididae. — Springer (1982; pers. comm.), fol- 

 lowing Jordan (1923) and Golvan (1965), included microcan- 

 thids, Neoscorpis. Parascorpis and scorpidids in the family 

 Scorpididae, but no convincing evidence for uniting them has 

 been presented, and they are treated separately here. The Scor- 

 pididae is here restricted to Scorpis. Medialuna. Lahracoglossa 

 and Bathystethus. The latter two genera were treated as a sep- 

 arate family, Labracoglossidae, by Springer. Scorpidids share 

 similar meristic and osteological features (not derived) and com- 

 parable scale morphology. An unusual small slip of muscle ex- 

 tends from the basioccipital to the first vertebra in Scorpis and 

 Lahracoglossa. but its presence has not been confirmed in the 

 other two genera. The larvae of Scorpis and Bathystethus are 

 undescribed hut those of Lahracoglossa (Fig. 258A) and Me- 

 dialuna (Fig. 258B) share a similar body form, generalized head 

 spination, late fin development and pigment pattern with larvae 

 of the Girellidae (Fig. 258C). Girellids are specialized in several 

 osteological features with respect to the Scorpididae (see Table 

 1 20) and have a unique adductor mandibulae in which A, inserts 

 on the lateral surface of the dentary (Johnson and Fritzsche, 

 in prep.). The distinctive larval form shared by scorpidids and 

 girellids suggests that they may be sister groups. Convincing 

 evidence supporting a close relationship between the Scorpi- 

 didae and the Microcanthidae (Microcanthus. Atypichthys and 

 Neatypus) or the Kyphosidae (Kyphosus, Seclator and Her- 

 nwsilla) is lacking. Furthermore, the larvae of the latter two 

 families (Figs. 259G, J) do not possess the salient features of 

 scorpidid or girellid larvae, but more closely resemble those of 

 the Teraponidae (Fig. 259H). The larvae of Neoscorpis and Par- 

 ascorpis are unknown, and available anatomical information is 

 insufficient to clarify the systematic position of these two genera. 



Malacanthidae. — See discussion on utility of larval morphology. 



Moronidae {Morone and Dicentrarchus), Lateolahrax and 5/>j- 

 /perca.— Gosline (1966) included the Moronidae (using the name 

 Roccus). Lateolahrax and Siniperca (=Coreoperca) in his "es- 

 tuarine and freshwater percichthyids." I treat these separately, 

 because I lack evidence of their affinities with the Fercichthyi- 

 dae, with one another, or with any other percoid group. It is 

 interesting to note that the Moronidae share one of the two 

 synapomophies of the Centropomidae described by Greenwood 

 (1976)— the lateral line extends almost to the posterior margin 

 of the caudal fin. In addition, moronids have an auxilliary row 

 of lateral line scales on the caudal fin above and below the main 

 row, as does the centropomid Lates. Although both of these 

 conditions occur elsewhere in generalized percoids (e.g., Neo- 

 scorpis. some species of Lutjanus. and the percid subfamily 

 Luciopercinae) and may actually be primitive for the Percoidei 

 (Springer, 1983), the possibility of a moronid-centropomid 

 relationship seems plausible and should probably be investi- 

 gated further. Unfortunately, as is typical of most fresh or brack- 

 ish water spawners, the larvae of these groups (Fig. 260) exhibit 

 relatively direct development and consequently offer little phy- 

 logenetic information. 



Percichthyidae. — The Percichthyidae of Gosline (1966) repre- 

 sents a polyphyletic assemblage defined on the basis of shared 

 primitive features. I am unable to find synapomorphies that 

 support recognition of the assemblage as a monophyletic group. 

 I restrict the Percichthyidae to the following genera, which occur 

 only in freshwaters of Australia and South America: Percolates 

 (brackish water), Plectroplites. Macquaria. Maccullochella. Per- 

 cichthys, Percilia. Bostockia. Gadopsis. Nannoperca. Edelia. and 

 Nannatherina. The monophyly of the family is supported by a 

 series of nested synapomorphies, only a few of which are men- 

 tioned here. The scales of most of these genera are similar and 

 unlike those of the excluded genera in having the posterior field 

 filled with simple, only slightly amputated (see McCully, 1970), 

 needle-like ctenii (those of Bostockia. Gadopsis and Nannath- 

 erina are secondarily cycloid). The three most generalized gen- 

 era. Percolates, Plectroplites. and Macquaria are very similar 

 biochemically [MacDonald (1978) synonymized them on this 

 basis], and the latter two share two morphological specializa- 

 tions with Macidlochella. Percichthys. Percilia. Bostockia and 

 Gadopsis: enlarged sensory pores on the dentary and a separate 

 inner division of adductor mandibulae section A,. The three 

 most derived genera, Nannoperca. Edelia and Nannatherina 

 (heretofore treated as kuhliids) share with Bostockia a similar 

 vertebral number (29-33), a distinctive asymmetrical nasal ro- 

 sette, and a number of reductive specializations (absences of the 

 subocular shelf, procurrent spur, and supracleithral sensory ca- 

 nal, reduced numbers of procurrent caudal rays, dorsal spines, 

 branchiostegals and trisegmental pterygiophores, and an inter- 

 rupted or absent lateral line). Systematic placement of the enig- 

 matic Gadopsis has proved problematic, even in recent years. 

 It has generally been treated as a monotypic family and variously 

 assigned to the Percoidei (Greenwood et al., 1 966), Ophidioidei 

 (Gosline, 1 968), Perciformes with proposed affinities to the Tra- 

 chinoidei and Blennioidei (Rosen and Patterson, 1969) or a 

 separate order Gadopsiformes (Scott, 1962). The percoid affin- 

 ities of Gadopsis are manifest in the anatomy of the dorsal gill 

 arches, caudal skeleton and median fin supports. Its affinities 

 with the Percichthyidae are indicated by a number of features 

 shared with some percichthyid genera, including the configu- 

 ration of the adductor mandibulae noted above. Gadopsis shares 



