Scorpaeniformes: Relationships 

 B. B. Washington, W. N. Eschmeyer and K. M. Howe 



THE order Scorpaeniformes is a large, morphologically di- 

 verse group containing about 20 families (depending on 

 classification used), 250 genera, and over 1,000 species. The 

 order is defined by the presence of a suborbital stay, a posterior 

 extension of the third infraorbital bone which in nearly all species 

 is firmly attached to the preopercle. Infraorbital bones for many 

 scorpaeniform groups were discussed most recently by Poss 

 (1975). Many workers have suggested that the stay may have 

 evolved independently (Matsubara, 1943;Quast, 1965; Green- 

 wood et al., 1966; Poss, 1975; and Nelson, 1976). 



Relationships 



The higher classification of the Scorpaeniformes remains con- 

 troversial and uncertain, both in terms of monophyly and in 

 the definition of families and their relationships. Confusion ex- 

 ists not only at the subordinal levels, but also at lower taxonomic 

 levels. For example, between 1 and 17 families of cottids have 

 been recognized by previous workers. 



Two workers presented hypotheses of relationships within the 

 Scorpaeniformes. Matsubara (1943), in a detailed study of Jap- 

 anese scorpaenoids based on osteological and anatomical char- 

 acters, briefly treated relationships of scorpaenoids to other scor- 

 paeniforms. His graphic presentation of relationships is shown 

 in Figure 240. Several lineages are recognizable: 1) the Hexa- 

 grammidae, Anoplopomatidae, and "generalized" scorpaenids; 

 2) Peristediidae, Triglidae, and Dactylopteridae; 3) "special- 

 ized" scorpaenids, Bembridae, Platycephalidae, and Hoplich- 

 thyidae; 4) Cottidae and Agonidae; and, 5) Cyclopteridae and 

 Liparididae. In 1955, Matsubara refined his hypothesis of re- 

 lationships and presented a classification with categories equiv- 

 alent to three suborders, several superfamiliesand included fam- 

 ilies as follows: 



Cottida 

 Cottina 



Scorpaenicae 



Scorpaenidae, Synanceiidae, Congiopodidae 

 Hexagrammicae 



Anoplopomatidae, Hexagrammidae 

 Plalycephalicae 

 Parabembridae, Bembridae, Platycephalidae, Hop- 

 lichthyidae 

 Cotticae 



Cottidae, Psychrolutidae 

 Agonicae 



Agonidae, Aspidophoridae 

 Triglicae 

 Triglidae, Peristediidae 

 Dactylopterina 



Cephalacanthidae 

 Cyclopteridae 



Cyclopteridae, Liparididae 



Quast (1965) presented a notably different hypothesis of re- 



lationships of the scorpaeniforms. His work was based on char- 

 acters which were useful in comparisons with the hexagrammids 

 and included many characters taken from the earlier works of 

 Gill (1888), Regan (1913a) and Berg (1940). Quast proposed that 

 the Scorpaeniformes included three basic lineages: 1 ) the cottid- 

 hexagrammid (including the Cyclopteridae and Agonidae); 2) 

 the anoplopomatid; and, 3) the scorpaenoid (including all other 

 families). Quast (1965) did not incorporate his recommended 

 revisions in his formal synopsis of scorpaeniforms because he 

 believed that the cottoids and anoplopomatids were still in need 

 of intensive study. 



Several studies of particular character complexes have also 

 contributed to understanding of relationships within the order. 

 Freihofer (1963), in a study of patterns of the ramus lateralis 

 accessorius and associated nerves in teleosts, found three pat- 

 terns of nerves in scorpaeniforms which suggested three groups: 

 1) the Scorpaenidae and Synanceidae; 2) the Hexagrammidae, 

 Cottidae, and Liparididae; and, 3) the Anoplopomatidae. These 

 groupings seem to support Quast's hypothesis of relationships 

 but many families were not examined by Freihofer. Hallacher 

 (1974) provided a summary of gasbladder muscles in the scor- 

 pionfish genus Sebastes and included observations on other 

 scorpaeniforms. Matsubara (1943) treated this feature for Jap- 

 anese scorpaenoids. Hallacher recognized four states of the ex- 

 trinsic muscle in scorpaeniforms. His characters were based on 

 the connections, or lack of connections, of this muscle between 

 the cranium, pectoral girdle, vertebral column, and the gas- 

 bladder. His observations partially supported Matsubara's hy- 

 pothesis of scorpaeniform lineages. 



Scorpaeniform fishes have been considered as pre-perciforms 

 or as perciform derivatives but their relationship to other fishes 

 remains uncertain. Many workers have argued that the Scor- 

 paeniformes evolved from a "generalized" perciform ancestor 

 because of striking similarities in general body form, and ana- 

 tomical and osteological characters of generalized scorpaenids 

 and perciforms (Gill, 1888; Regan, 1913a; Taranets, 1941; Mat- 

 subara, 1943; Gregory, 1959; Quast, 1965; Gosline, 1971; Lau- 

 der and Liem, 1983). Others (Greenwood et al., 1966; Nelson, 

 1976) have tentatively placed the Scorpaeniformes as a distinct 

 pre-perciform group of the Acanthopterygians. 



As previously mentioned, several authors have suggested that 

 the Scorpaeniformes may be polyphyletic and hence, derived 

 from several groups. Greenwood et al. (1966) noted that some 

 scorpaeniforms share similarities of the parietals and cheek 

 muscles with cods, while others share similarities with toad- 

 fishes, and still others with perciformes. Freihofer (1970), on 

 the basis of nerve evidence, suggested that gobiesocids were 

 related to cottoids, especially liparidids. Although several au- 

 thors have suggested that the suborbital stay may have evolved 

 more than once in the Scorpaeniformes, little consideration has 

 been given to the hypothesis that other groups of fishes may 

 have lost the suborbital stay. Within the Scorpaeniformes, sev- 

 eral groups show a reduction or loss of the suborbital stay. 

 Groups which have lost the circumorbital bones, and possibly 

 a suborbital stay (e.g. gobiesocids, callionymids, lophiiforms 



438 



