Figure 256.— Carinotetraodon lorteti: lateral 

 view of head, 35.9 nun SL, locality i 



differs externally from Sphoeroides only in having the in- 

 ternal epithelium of the nasal sac deeply pitted rather 

 than smooth or with ridges, and the gill rakers anteriorly 

 on the fourth arch less well developed. The single 

 species, formosa, is often placed in Sphoeroides (as by 

 Fraser-Brunner 1943). The skull structure of formosa is 

 remarkably similar to that of annulatus, being an even 

 more specialized version of it, with the frontals more 

 laterally expanded anteriorly and the sphenotics even 

 more anterodorsally expanded to form at least half of the 

 dorsal margin of the orbit. Only in annulatus among the 

 species of Sphoeroides does the parasphenoid have a dor- 

 sal flange in the interorbital septum which meets the 

 frontal, and formosa also has a similar flange. The color 

 pattern of formosa can be considered a variation on that 

 of annulatus, with more numerous light reticulations and 

 circles and larger dark spots. It is obvious that annulatus 

 is a specialized derivative of a testudineus-Vike stock, and 

 that formosa is a specialized derivative of an annulatus- 

 like stock. Thus, formosa should be included in Sphoe- 

 roides as the most extreme example of frontal widening 

 and sphenotic anterior prolongation, for the pitted 

 epithelium of its nasal sac and smaller gill rakers on the 

 fourth arch certainly are not sufficiently distinctive 

 features to warrant even subgeneric recognition of for- 

 mosa within Sphoeroides. 



Two of the examined species of Sphoeroides in which 

 the skull is only slightly less wide and the snout region 

 only slightly less elongate than in the extreme condition 

 of dorsalis share with dorsalis one external peculiarity 

 not otherwise found among the tetraodontids. In the 

 western Atlantic S. dorsalis and in the eastern Pacific S. 

 lobatus and angusticeps there are a pair of dark dermal 

 flaps which are relatively constant in position on the dor- 



sum (sometimes absent as an intraspecific variation), 

 while Shipp (1974:59) reported that this pair of flaps is 

 also present in the eastern Atlantic S. marmoratus, not 

 studied here. These three species (four including mar- 

 moratus) undoubtedly share a close common ancestry, 

 with the ancestral species becoming divided into two 

 isolated populations with the emergence of the Central 

 American isthmus. The Atlantic population evolved into 

 the highly specialized (in narrowness of skull) dorsalis 

 (and probably marmoratus), while the eastern Pacific 

 population diverged into lobatus and angusticeps with a 

 slightly lesser narrowing of the skull, differing from one 

 another mainly in angusticeps having a plainer colora- 

 tion and having lost all the spines. 



Sphoeroides nephelus, which also has a relatively 

 narrow skull, is probably also related to the line which 

 gave rise to dorsalis, lobatus, and angusticeps, while its 

 closest relative is maculatus (see Shipp and Yerger 

 1969a, b, who show that parvus, not studied here, 

 nephelus, and maculatus constitute a closely related 

 species complex). On the basis of its skull structure, 

 spengleri, with a moderate skull width, is not far remov- 

 ed from that complex, while pachygaster, with a 

 somewhat wider skull, may be derived from a spengleri- 

 like form. 



The deepwater S. pachygaster, with its thick, flabby 

 spineless skin, seems externally to be one of the most dis- 

 tinctive and specialized species of Sphoeroides. However, 

 internally pachygaster is only moderately specialized by 

 a slightly more than moderate width of the skull, a slight 

 anterodorsal prolongation of the sphenotics, a less firm 

 ossification of the skull, and an unusually great amount 

 of sculpturing on the dorsal surface of the skull, especial- 

 ly on the frontals, and by the loss of teeth on the first 



