EPAXIAL MUSCLES AND ACANTHOMORPH RELATIONSHIPS 



129 



lack of parietal spines; Johnson & Patterson (1993: 579) 

 considered and rejected a relationship between dactylop- 

 terids and gasterosteiforms. The monotypic family Bathylu- 

 tichthyidae was recently erected by Balushkin & 

 Voskoboynikova (1990) and placed in the Scorpaeniformes 

 (our Scorpaenoidei) largely on the basis of trend characters 

 variably shared with some cottoid taxa. Although Bathylu- 

 tichthys could have secondarily lost Type 1 epaxial insertions, 

 its position in the Scorpaenoidei should be regarded as 

 provisional. The condition of the parietal and supratemporal 

 canal in Bathylutichthys could be informative, but requires 

 investigation. 



Conversely, Mandrytsa (1991) has recently questioned the 

 inclusion of the Pataecidae in the Scorpaenoidei (his Scor- 

 paeniformes) based on a study of cephalic lateral-line struc- 

 ture. We have examined specimens of two of the three 

 pataecid genera (Aetapcus and Pataecus; Table 1) and found 

 that they have a typical scorpaenoid Type 1 arrangement of 

 their epaxial musculature, corroborating their current posi- 

 tion in the suborder. Ishida's (1994) more detailed analysis of 

 various myological and osteological characters also conclu- 

 sively nests pataecids within the Scorpaenoidei (as the sister 

 group of the Aploactidae). 



Winterbottom (1993) suggested a relationship of gobioids 

 with the scorpaenoid family Hoplichthyidae, but this is not 

 supported by our observations. Gobioids have no association 

 of epaxial muscle with distal portions of the dorsal-fin ptery- 

 giophores, whereas hoplichthyids exhibit a typical scor- 

 paenoid Type 1 pattern. 



The shared Type 1 morphology in a subset of perciforms 

 (blennioids, some cirrhitoids, Apogonidae, some Centropo- 

 midae, Centrogeniidae, Champsodontidae, Cheimarrhich- 

 thyidae, Grammatidae, Haemulidae, Percidae, and 

 Serranidae) implies that closer relationships might exist 

 among these taxa than are presently recognized (cf. Figs 2, 

 4-5, 12-17). The enigmatic family Centrogeniidae is an 

 interesting example because its nomenclatural history reflects 

 the possible relationships suggested by epaxial morphology. 

 Centrogenys vaigiensis, the single included species, and/or its 

 junior synonyms, has variously been classified as a scorpaeni- 

 form (e.g., Day, 1875; Fowler & Bean, 1922), a serranid 

 (e.g., Jordan, 1923; Weber & de Beaufort, 1931; Paxton et 

 al., 1989), or has been suggested to bear a superficial 

 resemblance to cirrhitids (Gosline, 1966; Nelson, 1984). 

 Although Centrogenys does not fit comfortably into any of 

 these taxa as they are currently diagnosed, the similar Type 1 

 epaxial musculature suggests that a detailed anatomical com- 

 parison could provide considerable insight into their interre- 

 lationships. 



In the Centropomidae, we found that extant members of 



the subfamily Latinae (Lates, Psammoperca) have a modified 

 Type 1 epaxial morphology where the muscle insertions to 

 the pterygiophores are separate from the main epaxial body, 

 but are below the spine/pterygiophore articulation (Fig. 12); 

 this arrangement could also be described as a modified Type 

 morphology with a more dorsal position of the normally 

 proximal insertions. The Centropominae (Centropomus) dif- 

 fer in lacking such dorsal epaxial insertions to dorsal-fin 

 pterygiophores (Type 0) (Table 1). Greenwood (1976) 

 hypothesized the monophyly of the Centropomidae, with its 

 two subfamilies as sister taxa, on the basis of two synapomor- 

 phies: pored lateral-line scales extending to posterior margin 

 of caudal fin, and neural spine of second vertebra markedly 

 expanded in an 'anteroposterior direction.' Pored lateral-line 

 scales extend well on to the caudal fin in many acanthomorph 

 fishes, and reach, or nearly reach, the posterior margin of the 

 fin in several families, including sciaenids (Greenwood, 

 1976), moronids (G.D. Johnson, pers. comm.), most pem- 

 pheridids, rhyacichthyids (Springer, 1983) and polynemids. 

 Therefore, this character does not provide convincing evi- 

 dence of relationship, and may be plesiomorphic within 

 perciforms. We also are not convinced that Greenwood's 

 second character (also noted by Gosline, 1966), expansion of 

 the second neural arch, is homologous in centropomines and 

 latines. In adult centropomines (see Fraser, 1968: 455 for 

 discussion of ontogenetic variation), the second neural spine 

 is broadly expanded over most of its length (resulting in a 

 truncated or rounded distal tip to the spine) and closely 

 applied to the first neural spine, which is narrow and sharply 

 pointed (see Fraser, 1968: fig. 14; Greenwood, 1976: fig. 25d; 

 Rosen, 1985: fig. 39B). In contrast, the anterior neural spine 

 morphology of the latines does not differ markedly from the 

 conditions found in various basal perciforms; the second 

 neural spine is only expanded proximally, and is not closely 

 applied to the first neural spine (see Greenwood, 1976: figs 

 25a-c). Given the lack of convincing synapomorphies to unite 

 the subfamilies Latinae and Centropominae, and considering 

 the differences in epaxial morphology (as well as various 

 other anatomical differences listed by Greenwood, 1976), 

 there is no justification for placing them in a single family. 

 Based on their modified Type 1 epaxial morphology, we here 

 remove the African/Indo-Australian genera Lates and Psam- 

 moperca from the Centropomidae to a separate family, 

 Latidae. Hypopterus (Western Australia) and Eolates (Italy 

 [Monte Bolca]), included as latines by Greenwood (1976), 

 presumably also belong to the newly created Latidae. Green- 

 wood (1976) considered Psammoperca macroptera, the type 

 species of Hypopterus, to be a synonym of P. waigiensis, the 

 single species he recognized in Psammoperca; however, 

 recent authors (e.g., Allen & Swainston, 1988: 62; Paxton et 



SCA 



Fig. 7 Type 1 epaxial musculature in the scorpaeniform Pterois radiata (USNM 140493, 63.3 mm SL). Note the insertion of the epaxial 

 muscle on to elements of the second pterygiophore and those posterior to the ninth pterygiophore. Abreviations and other methods of 

 presentation as in Fig. 1. Scale bar = 5 mm. 



