a large number of small, at least semi-isolated, scale plates 

 in a long patch both dorsally and ventrally but not 

 mediolaterally. The carapace is high crested both dor- 

 sally and ventrally so that in cross-sectional outline it 

 would be in the form of a laterally compressed oval. The 

 carapace does not extend posteriorly beyond the level of 

 the rear of the bases of the dorsal and anal fins. This can 

 probably be taken as the generalized aracanid carapace 

 condition, and the one on which the Recent species have 

 modified mainly only in a greater specialization of the 

 caudal peduncular scale plates. Unfortunately, the in- 

 ternal anatomy of P. dubia is mostly unknown, being 

 largely hidden from view by the carapace in the few 

 specimens presently available. However, the more pos- 

 terior portion of the vertebral column is relatively easily 

 seen in some of the specimens, and the indications are. 

 that the caudal plate is as fully fused together as in Re- 

 cent species and that the anterior anal fin basal pteryg- 

 iophores diverged from the midline. The teeth were dis- 

 crete and low in number, similar to those of Recent 

 species. Proaracana has only one known specialized 

 feature relative to the Recent species of the family, this 

 being that it has only 10 caudal fin rays rather than the 

 11 found in all Recent species. 



In triacanthids and balistids the ventral edge of the 

 parasphenoid anterior to the orbit is narrow, and the 

 pharyngobranchials consist of a toothless suspensory ele- 

 ment followed by three (triacanthids) or two (balistids) 

 elements bearing large teeth. One would expect that the 

 more generalized of the aracanids would have been 

 derived from these conditions. Strophiurichthys and 

 Kentrocapros have the ventral edge of the parasphenoid 

 the least expanded in the family, and Strophiurichthys 

 has the best developed dentition on the two pharyngo- 

 branchials (third and fourth) bearing large teeth, as well 

 as minute teeth on the second pharyngobranchial. Ken- 

 trocapros has minute teeth on the second pharyngo- 

 branchial but fewer large teeth on the third and fourth 

 pharyngobranchials than in Strophiurichthys. 

 Strophiurichthys has a high dorsal crest, similar in mag- 

 nitude to that of Proaracana if the high dorsal spiny 

 process in the latter is ignored. 



In short, of the genera examined, Strophiurichthys ap- 

 pears to be the most generalized of the Recent forms. A 

 form like it probably gave rise on the one hand to a line 

 leading to Kentrocapros and hence Aracana and on the 

 other hand to a line leading to Anoplocapros (not studied 

 internally) and Capropygia-Caprichthys, the latter two 

 monotypic genera being so similar that they scarcely 

 merit distinction. In the line leading to Kentrocapros the 

 high crested dorsal ridge was lost, with the back becom- 

 ing flattened or only very gently convex, while the ven- 

 tral carapace ridge was also reduced in size. The pharyn- 

 geal dentition was slightly decreased and the degree of 

 development of large spiny processes increased. Aracana 

 is obviously closely related to Kentrocapros, sharing with 

 it a flattened back and a more or less comparable 

 pharyngeal dentition, and differing from it externally 

 only by a continuing slight increase in the number of 

 large spiny carapace processes and the probable reten- 



tion from a slightly more generalized ancestor than the 

 Recent Kentrocapros of a deeper ventral carapace ridge. 

 Internally Aracana is further specialized by the ap- 

 parently usual loss of one of the two anterior branchios- 

 tegal rays, and, more importantly, by the development of 

 a wide lateral expansion of the ventral edge of the para- 

 sphenoid just in front of its articulation with the antero- 

 medial edges of the prootic shelves. In one respect Ken- 

 trocapros and Aracana have remained slightly more 

 generalized than the ancestral Strophiurichthys-like 

 form in that the individual scale plates of the two an- 

 terior caudal peduncular saddles are less fully con- 

 solidated into functionally single pieces. 



Probably a close derivative of the ancestral Strophi- 

 urichthys-like form is Anoplocapros, whose internal 

 structure has not been studied but which is very similar 

 to Strophiurichthys externally in the features of its 

 relatively spineless high crested carapace. Anoplocapros 

 differs from Strophiurichthys mainly in having a com- 

 plete ring of scales around the posterior region of the 

 caudal peduncle instead of two saddles. However, this is 

 surely a minor distinction, and McCulloch and Waite 

 (1915:479) pointed out that the ring "may be incomplete 

 in the young." In short, the two only narrowly separated 

 posterior peduncular saddles as found in Strophiurich- 

 thys have simply met mediolaterally in adult Anoplo- 

 capros. The Strophiurichthys-Anoplocapros line 

 probably also gave rise to Capropygia and Caprichthys, 

 both of which are specialized by the loss of even minute 

 teeth on the second pharyngobranchial and the reduc- 

 tion in number of large teeth on the third and fourth 

 pharyngobranchials, while the carapace extends back 

 further posteriorly than in any of the other genera. 



The most distinctive internal feature linking Capro- 

 pygia and Caprichthys is the development of a moderate 

 lateral expansion of the ventral edge of the para- 

 sphenoid just behind its articulation with the vomer, in 

 contrast to the more posteriorly placed lateral expansion 



Aracana 

 Kentocapros 



Caprichthys 



Capropygia 



Anoplocapros 



Strophiurichthys 



Figure 146.— Hypothesized phylogenetic 

 relationshipg of the genera of Aracanidae. 



205 



