JOHNSON: PERCOIDEI 



481 



Table 120. Continued. E,\tended. 



bones in relation to the development of the dorsal fin may prove 

 useful in determining the homologies of these additional ele- 

 ments as well as the first three predorsals. 



Caudal skeleton.— The primitive percoid caudal skeleton con- 

 sists of one parhypural with a well-developed hypurapophysis, 

 five hypurals, two pairs of uroneurals. three epurals, one ural 

 centrum, a low neural crest on PU2 and autogenous haemal 

 spines on PU2 and PU3. This configuration is found in at least 

 some members of 54 percoid groups. 



The most common reductions involve fusion of hypurals one 

 and two and hypurals three and four and loss of the posterior 

 uroneural pair. Loss of one epural occurs in only 14 groups, and 

 epurals are completely lacking only in some congrogadids. More 

 extreme reductions, including various combinations of fusions 

 of the hypurals with the parhypural and/or urostyle, loss or 

 fusion of the anterior uroneural pair and fusion of the autoge- 

 nous haemal spines, occur in only a few groups: Acanthoclini- 

 dae, some apogonids, Congrogadidae, Grammatidae, Menidae, 

 Mullidae, Opstognathidae, Plesiopidae, and Pseudochromidae. 



The second preural centrum bears a full neural spine in only 

 two groups, Echneididae and Nandidae, except for occasional 

 anomalous specimens. This full neural spine must be second- 

 arily derived in the echeneidids because these fishes are un- 

 questionably closely related to other percoids that bear the usual 

 reduced neural crest on PU2 (see discussion on utility of larval 

 morphology). Unfortunately, evidence for the origin of this de 

 novo spine in echeneidids is lacking. Although it may represent 

 a captured first epural (there are only two in echeneidids), it is 

 attached and of full length at its earliest appearance in ontogeny. 

 Another possibility is that the second preural centrum of other 

 percoids has been lost in echeneidids, so that the last centrum 

 bearing a full neural spine actually corresponds to PU3. How- 

 ever, presence of the usual autogenous haemal spines on both 

 PU2 and PU3 in echeneidids refutes this hypothesis. The sig- 

 nificance of a full neural spine on PU2 in the Nandidae is un- 

 clear, since the affinities of this family with the Percoidei remain 

 problematic. 



The presence of a procurrent spur and of radial cartilages 

 anterior to the neural and haemal spines of PU3 are probably 

 primitive features (Johnson, 1975, 1983). The procurrent spur 

 is developed to some extent in 50 percoid groups, all but ten of 

 which have a primitive caudal complex. Reductions among these 

 ten groups usually involve only simple hypural fusion. The pro- 

 current spur is never present in groups with fewer than 9 -F 8 

 principal rays. Third preural radial cartilages are found in 45 

 of 66 groups examined for them, about half of which have 

 primitive caudal complexes. 



I'crtchral number. — \ c'rXf:bra.\ number ranges from 23 to about 

 78 in percoids. Gosline (1968, 1971) noted that the "basal 

 number" of vertebrae in percoids is 24-25 (10 -I- 14-15), and 

 this number characterizes 45 of the 91 groups treated in Table 

 1 20; 63 groups have 24-27 vertebrae. Twenty-two groups have 

 vetebral counts greater than thirty, but only five have more than 

 40 vertebrae. Only priacanthids and scalophagids have fewer 

 than 24 (10 + 13). 



G///a/-c/!e5. — Primitively, percoid gill arches contain the follow- 

 ing elements: one basihyal, four basibranchials (the fourth car- 

 tilaginous), three pairs of hypobranchials, five pairs of cerato- 

 branchials, four pairs of epibranchials, four pairs of 



