branchiostegal rays 



row, while the similarity in the configuration of the pos- 

 terodorsal region of the skull and of the pelvis clearly 

 indicates that these two genera, forming the hoUardiins, 

 represent a line of triacanthodid evolution distinct from 

 that of the other nine Recent genera, the triacanthodins. 

 The relationships of these two Recent subfamilies to the 

 two fossil subfamilies are discussed in the following sec- 

 tion on subfamilial relationships. 



Among the triacanthodins, as summarized from Tyler 

 (1968:29-31), the deeper bodied genera with well- 

 developed spiny dorsal fins are more generalized than the 

 others, and of these generalized genera, Triacanthodes, 

 which retains a series of teeth internal to the major outer 

 series and which has a deep body, well-developed spiny 

 dorsal fin, and long gill opening, is the most generalized 

 of all. Mephisto and Paratriacanthodes are close deriva- 

 tives of the same ancestral stock which gave rise to 

 Triacanthodes and are simply progressively more 

 modified from it. Johnsonina is also a derivative of a 

 Triacanthodes-like ancestral stock which has indepen- 

 dently further specialized in most ways from it beyond 

 the levels of Mephisto and Paratriacanthodes. 



Atrophacanthus, Bathyphylax, and Tydemania are 

 probably independent derivatives of a stock of Paratria- 



Figure 22.— Hollardia hollardi: lateral view 



of heads of moderate-sized specimen, 62.7 mm SL, (above) 



and extremely large specimen. 174 mm SL 



(the largest Recent triacanthodid fish recorded), 



Caribbean, showing the change in configuration, 



increased suturing in the occipital-otic 



region and decreased amount of cartilage 



visible externally in large specimens. 



canthodes-like level of organization. An Atrophacan- 

 thus-like form, with the last three dorsal spines rudi- 

 mentary, the second pelvic ray absent, and the mouth 

 slightly but distinctly supraterminal, is probably not 

 ancestral to any other living triacanthodid, for the 

 tendencies it shows are for the snout to slightly decrease 

 in length and the conical teeth to increase in number 

 while becoming smaller and more sharply pointed. These 

 tendencies are not compatible to ancestry of any of the 

 other moderately or highly specialized genera. 



However, a Tydemania-Vike form is ancestral to 

 Macrorhamphosodes, while a Bathyphylax-Uke one is 

 ancestral to Halimochirurgus. Thus, the two long- 

 snouted genera which are superficially closely related in 

 actuality have independently evolved long snouts and 

 are not as closely related to one another as to other 

 genera. 



Tydemania and Macrorhamphosodes are the only two 

 genera of triacanthodids in which the teeth are wide and 

 thin and variously pointed to rounded to truncate distal- 

 ly. Tydemania has a snout of moderate length, a distinct- 

 ly supraterminal mouth of moderate width with wide 

 fleshy lips, the truncate teeth well developed in both 

 jaws but slightly fewer in the upper than in the lower jaw, 

 the last three dorsal spines rudimentary, and the second 

 pelvic ray absent. These are all characteristics that one 

 would expect to find in an ancestor of Macrorham- 

 phosodes, in which the elongate snout bears a distinctly 

 supraterminal mouth in a wide fleshy disk, the teeth are 

 more or less truncate and well developed in the lower jaw 

 but few in number or absent in the upper jaw, the last 

 three dorsal spines are rudimentary, and the second 

 pelvic fin ray is absent. 



In Halimochirurgus the distinctly supraterminal 

 mouth has thin lips and feeble teeth in reduced number 

 in both jaws at the end of an extremely long tubular 



