JOHNSON: PERCOIDEI 



491 



head spination among larval percoids will entail more precise 

 study than has characterized much previous work. Determi- 

 nation of homology will require detailed information about lo- 

 cation, conformation and processes of development of head 

 spines prior to considering the question of compatibility with 

 adult characters. 



Utility of Larval Morphology in 

 Phylogenetic Studies 



The preceding two decades have seen notable advances in our 

 understanding of the evolutionary relationships of teleost fishes; 

 however, as noted above, progress in elucidating the phylogeny 

 of the Percoidei has not kept pace. Many families are poorly 

 delineated and hypotheses about inter- and intrafamilial rela- 

 tionships are few. Lack of progress is chiefly attributable to the 

 size and diversity of the Percoidei. the adaptive malleability and 

 convergence that have characterized percoid evolution and the 

 paucity of conspicuous morphological specializations that can 

 be readily identified as true synapomorphies. With few excep- 

 tions (Burgess, 1974; Dooley, 1978; Kendall, 1979; Johnson, 

 1983), previous studies of percoid phylogeny and classification 

 have failed to consider early life history stages, even though it 

 is obvious that the prodigious variety of larval form and spe- 

 cialization among percoids offers a rich suite of additional char- 

 acters. 



Within many families there is a complexity of larval mor- 

 phology or diversity of larval form that suggests excellent po- 

 tential for the application of larval characters in elucidating 

 generic interrelationships. Particularly promising families in this 

 regard include the Acropomatidae, Apogonidae, Bramidae, Ca- 

 rangidae, Cepolidae, Chaetodontidae, Haemulidae, Lutjanidae, 

 Malacanthidae, Pentacerotidae, Pomacanthidae, Priacanthidae, 

 Sciaenidae, and Serranidae. The intricate bony ornamentation 

 of the larvae of anthiin serranids, for instance, is considerably 

 more complex than that of the adults, and preliminary studies 

 of details of larval head spination and scale development among 

 New World genera indicate that the current generic classifica- 

 tion, based exclusively on adult morphology, should be reex- 

 amined (Carole Baldwin, Abstracts of 1 983 ASIH Annual Meet- 

 ing). Larvae of groups like the apogonids and carangids exhibit 

 a less complex morphology, but the wide range of form and 

 specialization should prove useful in phylogenetic analyses. 



Larval morphology will undoubtedly also prove useful in con- 

 siderations of higher relationships among percoids. At the fam- 

 ily level, a rather simplistic approach is to consider that larvae 

 offer independent tests of hypotheses of monophyly. In other 

 words, do the larvae of each percoid family share one or more 

 derived features that corroborate the monophyly of that family 

 as currently defined on the basis of adult morphology? The 

 answer to this question appears to be yes for many groups, but 

 problems stem from an inadequate understanding of character 

 polarity and the fact that, for most families, larvae of many 

 genera and most species remain undescribed. Nonetheless, this 

 is a useful concept, and the validity and power of such a test 

 will increase as we gain more knowledge of the larvae of various 

 percoid groups. 



Consider, for example, the bearing of larval morphology on 

 several hypotheses of relationship resulting from the recent re- 

 definition of Schultz's (1945) Emmelichthyidae, a polyphyletic 

 assemblage of planktivorous fishes. Heemstra and Randall (1977) 

 transferred Diptcrygonolus to the Caesionidae and Johnson 

 (1980) hypothesized that caesionids are lutjanoid fishes most 



closely related to the lutjanid subfamily Lutjaninae. Caesionids 

 are quite distinctive in body form and upper jaw configuration, 

 but share with the lutjanines a number of osteological features 

 and a specialized adductor mandibulae (similar to that of most 

 carangids) in which a separate division of A, originates on the 

 subocular shelf Subsequent descriptions of larval lutjanines and 

 caesionids (see Table 122) show that they share a distinctive 

 body form, pattern of head spination, precocious first dorsal 

 and pelvic fins with elongate spines and soft rays, and sparse 

 pigmentation (Fig. 256A, B). The hypothesized sister group re- 

 lationship is thus corroborated by larval morphology. 



The Centracanthidae were also removed from the Emme- 

 lichthyidae and hypothesized to be most closely related to the 

 Sparidae (Heemstra and Randall, 1977; Johnson, 1980) based 

 on adult morphology. Although the larvae of these two groups 

 share no obvious specializations, they are quite similar (Fig. 

 2581, J), and are distinguishable from those of the Emme- 

 lichthyidae (Fig. 2591) and the other reassigned groups. Labra- 

 coglossa, placed in a separate family by Heemstra and Randall 

 (1977) is here placed in the family Scorpididae (see section on 

 classification), and the larval form corroborates this placement 

 (Fig. 258A, B). The larvae of inermiids, Inermia and Emme- 

 lichthyops. also removed from the Emmelichthyidae, remain 

 undescribed, but their identification can provide a test of the 

 hypothesis that they are most closely related to the Haemulidae 

 (Johnson, 1980). 



These examples and those that follow demonstrate that early 

 life history stages offer important information that can be used 

 to test previous phylogenetic hypotheses or incorporated with 

 adult characters into new phylogenetic analyses. Additional ex- 

 amples are mentioned in the discussion of familial classification. 

 Where the larvae are known, failure to consider their mor- 

 phology in studies of percoid phylogeny seems hardly justifiable, 

 and may inhibit progress or lead to false conclusions. This point 

 is well-illustrated in the two examples discussed below, in which 

 details of larval morphology provide critical evidence in support 

 of new or previously rejected phylogenetic hypotheses. 



The families Branchiostegidae (=Latilidae) and Malacanthi- 

 dae have been variously united and separated in past classifi- 

 cations. In the most recent revision, Dooley (1978) concluded 

 that "the branchiostegids and malacanthids have few characters 

 in common that might be used to justify their consolidation 

 into a single family" and noted that they "could as easily be 

 aligned with several other percoid families as with each other." 

 He suggested that the malacanthids are possibly "a branch of 

 the labrid-scarid lineage, while the branchiostegids show closer 

 affinities to the serranid-percid line of perciform evolution." In 

 contrast, Robins et al. ( 1 980) recognized a close affinity between 

 the two groups by treating them as subfamilies of the Malacan- 

 thidae. Marino and Dooley (1982) took issue with this classi- 

 fication and stated that there are "several more myological (dif- 

 ferences) why the families are distinct." Actually, Marino and 

 Dooley listed only one myological difference, the absence of 

 adductor mandibulae section A,,,. This difference and the other 

 1 3 listed by Dooley ( 1 978. Table 1 ), including body depth, body 

 shape, and skull contour, have little relevance to the phyloge- 

 netic affinity of these two groups. As for features common to 

 the malacanthids and branchiostegids, Dooley found only three: 

 dorsal and anal fins relatively long and continuous, a single 

 opercular spine, and "grossly similar larval stages." Dooley cor- 

 rectly noted that the first two of these are not particularly mean- 

 ingful because they are fairly common percoid features, but he 



