490 



ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 



somewhat larger, equally heterogeneous assemblage of percoid 

 groups (including the Ambassidae, Centracanthidae, Centro- 

 pomidae. Cirrhitidae, Moronidae, Percidae, Pomatomidae and 

 Pseudochromidae) has minimal head spination, consisting of 

 only a few small spines along the posterior, and usually lateral, 

 margins of the preopercle. In most instances, these spines are 

 so small and isolated that it is difficult to imagine that they serve 

 any useful function. 



The most common pattern of head spination among larval 

 percoids is one in which, in addition to small to moderate pre- 

 opercular spines, small spines may also occur on other bones 

 of the opercular series (interopercle, subopercle and opercle) and 

 on various bones of the pectoral series (cleithrum, postcleith- 

 rum, supracleithrum, posttemporal and tabulars). This pattern 

 occurs in many of the more generalized families that have usu- 

 ally been considered "basal" percoids, including the Acropo- 

 matidae, Gerreidae, Girellidae, Haemulidae, Kyphosidae, 

 Sciaenidae, Scorpididae, Sparidae and Teraponidae, and it must 

 be primitive for at least some large subgroup of percoid families. 



Two additional levels of complexity in this artificial hierarchy 

 involve modifications of cranial bones (frontal and supraoccip- 

 ital) in addition to opercular and pectoral series spination. Mod- 

 ifications of the frontal bones occur only in those larvae with 

 opercular and pectoral series spination and encompass several 

 types of ornamentation. Frontal surface rugosity is found in a 

 few apogonids, bramids and serranids as well as in Acantho- 

 cepola, Lobotes, Hapalogenys, Pseudopenaceros and Sphyrae- 

 nops. Johnson and Keener (1984) noted this condition in larval 

 Alphestes. but it was not previously considered in descriptions 

 of percoid larvae. With closer examination, cranial rugosity will 

 undoubtedly be detected in larvae of other percoid and non- 

 percoid groups. It probably offers an efficient way to strengthen 

 the neurocranium during early development. Frontal spines or 

 serrations are most frequently borne along the supraorbital ridge. 

 Coryphaena, Rachycentron, Lobotes. and some carangids have 

 one large, broad-based supraorbital spine, but the more com- 

 mon condition is a series of supraorbital spines or serrations. 

 These are found in lutjanids, malacanthids, monodactylids, po- 

 macanthids, Stereolepis. some acropomatids, carangids, hae- 

 mulids, sciaenids, and serranids as well as in most groups with 

 supraoccipital modifications. More elaborate ornamentation, 

 consisting of a series of parallel serrated ridges on the dorsal 

 surface of the frontals, characterizes larval malacanthids, pria- 

 canthids, Synagrops and some anthiin serranids. 



The most extreme example of frontal spination is seen in 

 Symphysanodon (Fig. 254A). A longitudinal serrated crest above 

 the supraorbital ridge on each frontal bone continues posteriorly 

 as a long, spike-like serrated spine extending to about the middle 

 of the spinous dorsal fin. The only other example of large paired 

 cranial spines among larval perciforms is found in istiophorids, 

 where the spines originate from the pterotics. This "homed" 

 effect occurs elsewhere in larvae of many scorpaeniform groups 

 (e.g., Scorpaenidae and Triglidae) and in the beryciforms, Di- 

 retmus and Anoplogaster, but in these groups the large paired 

 spines are parietal in origin. With the exception of occasional 

 minute spines or small ridges, larvae of perciform fishes never 

 develop parietal ornamentation, and it is tempting to speculate 

 that the presence of variously developed parietal spines among 

 larvae of many scorpaeniform groups offers support for the often 

 questioned monophyly of the Scorpaeniformes. In any case, this 

 uncommon feature should be examined in future considerations 

 of higher relationships among acanthopterygian fishes. The 



monophyly of the Beryciformes has recently been questioned 

 (Zehren, 1979), and it is interesting to note that although Di- 

 relmus. Anoplogaster and at least some trachicthyoids share 

 larval parietal spines with scorpaeniforms, holocentrids lack 

 them, instead possessing frontal, supraoccipital and preoper- 

 cular spination similar to that seen in more elaborately orna- 

 mented larval percoids. 



Modifications of the supraoccipital, representing the last cat- 

 egory of complexity in head spination, occur in those larvae 

 which also have opercular series, pectoral series and frontal 

 ornamentation. Simple forms of supraoccipital ornamentation 

 include a small peak-like median crest (Chaetodipterus, Pagrus, 

 Polyprion, Sphyraenops, and some acropomatids, apogonids, 

 carangids and anthiin serranids) or a serrated, ridge-like crest 

 (Synagrops. some sciaenids and anthiin serranids). The more 

 extreme form is a large, vaulted, variously serrate spine-like 

 crest that projects beyond the posterior margin of the cranium 

 and is well-developed in preflexion larvae soon after hatching. 

 This type of crest characterizes larval cepolids, Hapalogenys, 

 leiognathids, lethrinids, (lobotids?), pentacerotids, priacanthids 

 and Scoinbrops. To my knowledge, it occurs elsewhere only in 

 the larvae of holocentrid beryciforms and the caproid Antigonia. 



The so called "tholichthys" larvae of the Chaetodontidae and 

 Scatophagidae (Fig. 262A-D) perhaps represent the ultimate in 

 head bone modification among larval percoids. The cranial bones 

 and many of the other exposed bones of the head are thickened 

 and rugose, effecting an armor-like protective covering. In chae- 

 todontids the posttemporal and supracleithrum are rugose and 

 expanded posteriorly as large laminar plates. The preopercle is 

 similarly expanded anteriorly and posteriorly and at its angle 

 bears a broad, flattened or serrated, terete spine. In scatophagids 

 the preopercle is rugose and expanded, but, unlike chaetodon- 

 tids, the supracleithrum is unmodified. The posttemporal is 

 rugose, its dorsal portion is somewhat expanded, and its ventral 

 half extends posteriorly as a very blunt, thick, spine-like pro- 

 jection. Also notable is a large, thick, rugose protuberance cov- 

 ering the pterotic. Although not identical, the larvae of chae- 

 todontids and scatophagids share a unique physiognomy, the 

 details of which should be investigated in relation to possible 

 close affinity of these two families. 



Spination on circumorbital, nasal, premaxillary and maxillary 

 bones is generally found only in those larval percoids with cra- 

 nial ornamentation, and it is almost exclusively in these larvae 

 that other specializations, such as elongate serrate fin spines and 

 spinous scales occur. In addition, opercular and pectoral series 

 spination is usually more extensive and almost always includes 

 an elongate and/or serrate spine at the angle of the preopercle. 



In summary, there seem to have been some common evo- 

 lutionary constraints on the order in which morphological com- 

 plexity and specialization of larval percoids has progressed, but 

 a simple direct relationship between this ordered progression 

 and phylogenetic affinity among families is not apparent. In fact, 

 the assemblages of taxa that characterize the various levels of 

 complexity discussed above are quite diverse and not compat- 

 ible with what little we do understand about percoid affinities 

 based on adult morphology. Furthermore, it is clear that elab- 

 orately ornamented larvae have arisen independently several 

 times within monophyletic groups otherwise characterized by 

 larvae with only generalized opercular and preopercular spi- 

 nation. Examples include the haemulid Conodon, the sparoid 

 family Lethrinidae and the serranid subfamily Anthiinae. Res- 

 olution of the phylogenetic significance of intricate patterns of 



