species and five in one (quadricornis) , i.e., the 15th 

 vertebra helps to support the last anal fin basal pteryg- 

 iophore in Lactophrys and Rhlnesomus but not in Acan- 

 thostracion. In Lactophrys and Rhlnesomus suturing be- 

 tween adjacent vertebrae extends from the 5th and 6th 

 back to between the 12th and 13th, while in Acantho- 

 stracion it extends back to between the 13th and 14th 

 (erroneously stated by Tyler 1963a:171, to be between 

 the 12th and 13th in A. quadricornis). Lateral flanges 

 across the centra are slightly better developed in Lacto- 

 phrys and Rhlnesomus than in Acanthostracion. 



On the basis of the above, only two genera should be 

 recognized in the Lactophrysinae, with Lactophrys = 

 Rhlnesomus being more generalized in all of the charac- 

 teristics listed above than Acanthostracion. Whether the 

 condition of having the carapace slightly open behind the 

 dorsal fin in L. trlgonus is a hold over from the ancestral 

 aracanids that gave rise to the ostraciids or is a case of 

 the subsequent reduction of the carapace in this region 

 from a more immediate ancestor with a complete bridge 

 across the back behind the dorsal fin is impossible to say 

 with assurance. In the only fossil ostraciid, the Eocene 

 Eolactorla, the carapace is complete behind the dorsal 

 fin. I suspect, especially in light of the fact that the 

 carapace is complete behind the dorsal fin at least oc- 

 casionally in large specimens of L. trlgonus just as it is in 

 Eolactorla and all other ostraciids, that the condition in 

 L. trlgonus is a secondary reduction of the posterior end 

 of the carapace and not the retention of a primitive con- 

 dition. It seems possible that L. trlgonus and R. 

 blcaudalis are slightly more closely related to one 

 another than to R. trlqueter, despite the behind the dor- 

 sal fin carapace similarity between blcaudalis and trl- 

 queter. In addition to blcaudalis and trlgonus both hav- 

 ing postanal carapace spines, both have the anterior 

 nostril bulbous, more so in blcaudalis and at all sizes, 

 and less so and only in large specimens of trlgonus. It 

 would be intriguing to know whether the anterior nostril 

 in trlqueter, which is a smaller species than the other 

 two, would become bulbous were it to reach sizes com- 

 parable to trlgonus. 



Acanthostracion is probably derived from a Lacto- 

 phrys = Rhlnesomus-Vike stock, undoubtedly from a 

 form with the carapace closed behind the dorsal fin and 

 with postanal spines. Acanthostracion is specialized in- 

 ternally in the ways mentioned above, while the presence 

 of preorbital spines could indicate either that they are a 

 new development from the Lactophrys = Rhlnesomus- 

 like line which give rise to it or that the immediate 

 ancestry of the Lactophrys = Rhlnesomus line had preor- 

 bital spines which were lost by the Recent species of that 

 group but retained by the line leading to Acantho- 

 stracion. Within Acanthostracion, quadricornis is ob- 

 viously the most specialized form (the addition of 1 

 vertebra to the postanal series for a total of 5 in the series 

 and 19 for the entire column, both unique in the family, 

 and the total number of vertebrae unique in the super- 

 family; the modal reduction by one of the number of pec- 

 toral fin rays, unique in the subfamily; the frequent 

 presence of isolated scale plates above and below on the 



caudal peduncle, unique in the family), and notacan- 

 thus, polygonlus, and gulneensls are more closely related 

 to one another than to quadricornis, with the rela- 

 tionship between notacanthus and polygonlus being es- 

 pecially close on the basis of the coloration, carapace, 

 and caudal fin shape characteristics discussed by Tyler 

 (1965b:275). 



In the Ostraciinae the generic relationships, and which 

 genera are generalized versus specialized, are less clear 

 than in the lactophrysins. The rectangular carapace of 

 Ostraclon and Rhynchostraclon rhlnorhynchus can be 

 considered a specialization, being further removed from 

 the laterally compressed oval of aracanids and the high 

 crested triangle of lactophrysins and of Tetrosomus, with 

 the high crested but otherwise rectangular form of R. 

 nasus being a variant of the Ostraclon and R. rhlno- 

 rhynchus form and the pentangular carapace of Lactorla 

 a low crested variant of the triangular form. 



In the caudal skeleton the presence of a haemal canal 

 piercing the more posterior haemal spines and entering 

 the caudal plate, as found only in Lactorla cornuta 

 among the ostraciins, is clearly a relatively generalized 

 condition in comparison to the highly specialized solid 

 postanal haemal spines, usually reduced in size and 

 never pierced by the haemal canal, which also does not 

 enter the caudal plate, found in L. fornaslnll, Tetro- 

 somus, Ostraclon, and Rhynchostraclon. The reduction 

 in size of several of the neural spines below the last dor- 

 sal fin basal pterygiophore, so that the pterygiophore is 

 supported only basally by a neural spine, as found in L. 

 fornaslnll, and, especially, Tetrosomus, seems special- 

 ized in comparison to the more normal neural spines in 

 this position found in L. cornuta, Ostraclon, and Rhyn- 

 chostraclon. However, the relatively small size and poor 

 development of shelves supporting the carapace on this 

 pterygiophore in L. fornaslnll and Tetrosomus are much 

 closer to the condition found in the lactophrysins and 

 aracanids, and must be considered either more general- 

 ized than or a secondary reduction in size from the far 

 more massive last dorsal fin basal pterygiophore with 

 well-developed specialized shelves found in L. cornuta, 

 Ostraclon, and Rhynchostraclon. 



The only moderately developed lateral flanges across 

 the centra in Tetrosomus and Rhynchostraclon would 

 seem slightly less specialized than the larger flanges of 

 Lactorla and Ostraclon. The relatively small lateral shelf 

 for carapace support on the last anal fin basal pterygio- 

 phore in L. cornuta seems the least removed from the 

 condition in lactophrysins and aracanids and thus less 

 specialized than the much larger anterolateral shelf 

 found in L. fornaslnll, Tetrosomus, Ostraclon, and 

 Rhynchostraclon. 



The number of postanal vertebrae, three in Lactorla 

 and Tetrosomus and four in Ostraclon and Rhyncho- 

 straclon, can be compared only to the condition in lacto- 

 phrysins. In aracanids there is a far less close association 

 of the last (and other) anal fin basal pterygiophore with 

 the haemal spines of the caudal vertebrae (with only the 

 proximal end of the element at all associated with a par- 

 ticular haemal spine instead of most of its posterodorsal 



