edge of the fourth arch and in having nine caudal fin 

 rays, while being more generalized than either of the 

 other two subgroups by always retaining a prefrontal, 

 and more generalized than the C. affinis, atinga, reticu- 

 latus. and tigrinus subgroup by the constant presence of 

 two hypohyals and a pharyngobranchial on the third 

 arch, the constant absence of teeth on the fifth 

 ceratobranchial, and the normal presence of a nasal tube 

 with two nostrils. The only specialized feature that the 

 exclusively Atlantic subgroup shares with C. orbicularis 

 is the presence of nine caudal fin rays, but otherwise the 

 Atlantic subgroup is far more generalized than C. orbicu- 

 laris. 



Whether the three subgroups within Chilomycterus 

 should be recognized subgenerically is a matter of opin- 

 ion, but on a practical level I doubt that such categories 

 would be generally used and none are here proposed. 



In summary, C. orbicularis is most closely related to 

 the C. affinis, atinga, reticulatus, and tigrinus subgroup 

 and in its own way is about as specialized as them, 

 probably having diverged from a common ancestral line 

 at a time when the nasal apparatus was still a tube with 

 two nostrils. The C. antennatus, antillarum, maure- 

 tanicus, schoepfi, and spinosus subgroup is more distant- 

 ly related to the other two and in most ways is more 

 generalized than them, probably having diverged from 

 the line leading to the other two subgroups at a time 

 when there were still 10 caudal fin rays and gill rakers on 

 the anterior edge of the fourth arch, with the subse- 

 quent loss of one of the caudal rays and of those gill 

 rakers while in general becoming less differentiated from 

 the basal stock than did the other two subgroups. 



The three species of Diodon studied, two of which 

 (holocanthus and hystrix) have all of the spines erectile 

 and the other (jaculiferus) with only a few behind the 

 pectoral fin erectile, scarcely differ osteologically, the 

 prefrontal in jaculiferus being slightly smaller than in the 

 other two. The two species oi Dicotylichthys studied, one 

 (nicthemerus) with all the spines erectile and the other 

 (punctulatus) with only those on the head erectile, are 

 equally similar osteologically and can be distinguished 

 from Diodon internally only by the presence of a spe- 

 cialized well-developed band of teeth on the fifth cera- 

 tobranchial, this being entirely toothless in Diodon, 

 while externally they differ only by Diodon having a 



nasal tube with two nostrils versus the specialized bi- 

 lobed tentacle of Dicotylichthys. I do not think that this 

 combination of two characters is of sufficient magnitude 

 to warrant generic recognition, it being no greater, and in 

 actuality less, than that between the three unnamed sub- 

 groups of Chilomycterus. It is suggested that Dico- 

 tylichthys be placed in the synonymy of Diodon. 



Moreover, the skeletal structure of Diodon and Dicoty- 

 lichthys is nearly indistinguishable from that of the en- 

 tirely Atlantic subgroup of Chilomycterus. Both Diodon 

 and Dicotylichthys differ from it only by the presence of 

 at least a few erectile spines and of gill rakers on the 

 anterior edge of the fourth arch, and Dicotylichthys ad- 

 ditionally by the presence of teeth on the fifth cerato- 

 branchial, a rather unimpressive array of differences. 



Diodon, Dicotylichthys, and the entirely Atlantic sub- 

 group of Chilomycterus have apparently diverged little 

 from the generalized ancestral stock. The presence of a 

 full covering of erectile spines without massive radiating 

 bases would seem to be the most generalized condition in 

 diodontids, as it is the closest to the normal condition in 

 tetraodontids. In this view the development of a huge 

 triradiate base to the scale, which makes it nonerectile, is 

 a specialization, and the species of Diodon and Dicoty- 

 lichthys with all of the spines erectile are more generaliz- 

 ed than those with only a minority of them erectile, while 

 by the same token Chilomycterus, all of whose species 

 have all of the spines with huge triradiate bases, is the 

 most specialized of all. 



In short, Diodon and Dicotylichthys do not seem 

 anatomically different enough to warrant separate 

 generic recognition, and while they are more generalized 

 and easily distinguishable osteologically from the two 



Chilomycterus : orbicularis 



Chilomycterus: affinis, atinga, 

 reticulatus, tigrinus 



Figure 303.— Hypothesized 

 phylogenetic relationships of 

 the genera of Diodontidae. 



Dicotylichthys 



Chilomycterus : antennatus, antillarum, 

 mauretanicus, schoepfi, spinosus 



Prodiodon 



