Percoidei: Development and Relationships 

 G. D. Johnson 



AS the largest and most diverse of the perciform suborders, 

 the Percoidei exemplifies the inadequacies that charac- 

 terize perciform classification. Regan (1913b) defined the Per- 

 coidei "by the absence of the special peculiarities which char- 

 acterize the other suborders of the Percomorphi [=Perciformes]," 

 and seventy years of research in systematic ichthyology have 

 failed to produce a more meaningful definition. In the absence 

 of even a single shared specialization uniting the percoids, the 

 monophyly of this great assemblage of fishes is doubtful. In spite 

 of our inability to adequately define the Percoidei, or because 

 of it, half of the approximately 145 families of perciform fishes 

 are usually referred to this suborder. Greenwood et al. (1966) 

 listed 71 percoid families in their "highly tentative" familial 

 classification of the Perciformes, and Nelson ( 1976) stated that 

 the Percoidei contains 72 families, 595 genera and about 3,935 

 species. 



Percoids are best represented in the nearshore marine envi- 

 ronment and form a significant component of the reef associated 

 fish fauna of tropical and subtropical seas. A few groups are 

 primarily epipelagic or mesopelagic. Association with brackish 

 water occurs in many nearshore marine families, some of which 

 have one or more exclusively freshwater members, but only 

 four families are primarily restricted to freshwaters, the north 

 temperate Percidae and Centrarchidae, the south temperate Per- 

 cichthyidae (with one brackish water species) and the tropical 

 Nandidae. 



In a practical sense, the suborder Percoidei serves the Per- 

 ciformes in much the same capacity as the Serranidae once 

 served the Percoidei itself as a convenient repository for those 

 "generalized" perciform families that cannot obviously be placed 

 elsewhere. I have treated the percoids in a similar sense here, 

 one of practicality and convenience. 1 do not intend to imply 

 or formulate hypotheses about the monophyly of the Percoidei 

 or to consider their intrarelationships as a whole. My major 

 objectives are to provide some preliminary documentation of 

 the variability of a number of character complexes among adults 

 and larvae of those fishes we now call percoids, to suggest what 

 1 believe to be promising avenues of future research and to offer 

 some specific examples illustrating the utility of larval mor- 

 phology in elucidating percoid phylogeny. 



Classification 



As here defined (Table 1 1 9) the Percoidei includes 80 families 

 and 1 2 incertae sedis genera, making it by far the largest and 

 most diverse suborder of teleostean fishes. The overall limits of 

 the suborder are only slightly modified from Greenwood et al. 

 (1966). The Pomacentridae, Embiotocidae and Cichlidae are 

 excluded because of their recent placement in the Labroidei by 

 Kaufman and Liem (1982). The suborder Acanthuroidea is 

 treated separately in this volume, but a recent hypothesis (Mok 

 and Shen, 1983), with which 1 concur, based on additional evi- 

 dence, suggests a close relationship between acanthuroids and 

 the Scatophagidae. The affinities of the questionably monophy- 

 letic Nandidae remain unresolved (Lauder and Liem, 1983), 

 and although the nandids are provisionally included in my list 



of percoid families, they were not considered in the larval and 

 adult tables. The genus Elassoma. formerly a member of the 

 family Centrarchidae, is excluded from the Percoidei, for rea- 

 sons discussed below. The monophyly of the suborder Trachi- 

 noidei, as defined by Greenwood et al. (1966) is suspect, and 

 the affinities of families such as the Mugiloididae, Percophidae, 

 Chiasmodontidae and others may lie with the percoids. How- 

 ever, these families are treated elsewhere in this volume, and 

 of the "trachinoids," only the Opistognathidae are here included 

 as percoids. 



Although the overall limits of the Percoidei are similarly per- 

 ceived in my classification and that of Greenwood et al. (1966), 

 substantive discrepancies result from differences in concepts of 

 family limits. For example, my Serranidae (Johnson, 1983) in- 

 cludes the Pseudogrammidae and Grammistidae of Greenwood 

 et al. (1966). Leptohrama is treated as a monotypic family sep- 

 arate from the Pempherididae (Tominaga, 1965), epigonids are 

 treated as a separate family, etc. The high percentage of mono- 

 typic families that has historically characterized percoid clas- 

 sification is a disturbing but unavoidable problem that can only 

 be remedied with a better understanding of percoid intrarela- 

 tionships. In my classification (Table 1 19), 26 of the 80 families 

 are monotypic and 12 genera, which lack family names, are 

 retained incertae sedis. Families and incertae sedis genera are 

 arranged alphabetically for easy reference and to avoid any in- 

 ference of affinity based on sequence. The classification of 

 Springer (1982) was followed for most families treated by him 

 and otherwise that of Nelson ( 1 976). Below, 1 discuss differences 

 between my classification and that of Spnnger ( 1 982) or that of 

 Nelson (1976), and present some new information about fa- 

 milial relationships. Early life history information contributed 

 substantially to some of these modifications. 



Acropomatidae and Symphysanodon— The "oceanic per- 

 cichthyids" of Gosline (1966) do not share the defining char- 

 acteristics of the Percichthyidae (see below), and are treated here 

 as a separate family, including the following genera— v^cropowa, 

 Apogonops. Doederleinia (=Rhomhoscrranus), Malakichthys, 

 Neoscombrops. Synagrops and V'erilus. I know of no synapo- 

 morphy that unites the acropomatids, and further work will be 

 necessary to test their monophyly. Larvae of four genera are 

 known. Those of Acropoma (Fig. 254C), Doederleinia (Fig. 254D) 

 and Malakichthys are quite similar, but those of Synagrops {Fig. 

 254B) differ in pigmentation, body form, and the presence of 

 more extensive head spination. Although the larvae of Sym- 

 physanodon (Fig. 254A) are unique in their possession of horn- 

 like frontal spines, they are otherwise remarkably similar to 

 those of Synagrops (Fig. 2548), suggesting that these two genera 

 may be closely related. 



Callanthiidae and Grammatidae.— Springer (1982) noted that 

 "there is little evidence to unite" the five genera he included in 

 the family Grammatidae. I concur with this and treat two of 

 these genera, Callanthias and Grammatonotus as a distinct fam- 

 ily, the Callanthiidae (currently under revision in collaboration 



464 



