spines in these two species has occurred independently. 

 T. Abe (pers. commun.) has informed me that, addi- 

 tionally, F. pardalis is at least superficially spineless. 

 Lagocephalus and Tetraodon each contain a spineless 

 species, L. inermis and T. mbu. In T. mbu, and, to a far 

 lesser extent, in the two scaleless species of Fugu, there 

 are pits in the skin that appear to be filled with a 

 secreted matter, which is presently under histological 

 investigation by the author. In L. inermis, and, to a lesser 

 extent, in S. pachygaster, the skin of the belly is irregular 

 in comparison to the smooth skin of the rest of the body, 

 but no spines are present. 



With the exception of these spineless species (and 

 there are probably others among the species of tetra- 

 odontids not examined for this work), spines are always 

 present on the belly of tetraodontids, but there is great 

 variation both between and within genera in the degree 

 of covering by prickles dorsally and laterally on the body. 

 In Lagocephalus, for example, several species (lagoceph- 

 alus and laevigatus) have spines only on the belly, while 

 others (lunaris, scleratus, and spadiceus) have prickles 

 variously developed on the dorsum in front of the dorsal 

 fin. In Sphoeroides prickles are present (except in the 

 two naked species) on the belly and dorsum, but are 

 about as frequently absent as present on the side of the 

 body (see Shipp 1974). In Torquigener prickles usually 

 are absent from the sides of the body, but in Ambly- 

 rhynchotes they are present there. In Ephippion the 

 prickles on the sides of the body become enormously 

 enlarged into plates forming a partial carapace (see illus- 

 tration for form of spines at three specimen sizes), while 

 one species of Amblyrhynchotes (piosae) has the pro- 

 jecting portion of the prickle much longer than in any 

 other tetraodontid. In Chonerhinos and Xenopterus the 

 prickles of the belly and sides of the body tend to be 

 larger than in other tetraodontids. 



In a few tetraodontids the scales along the course of the 

 lateral line are modified, being much smaller, more 

 numerous, and with less development of the protruding 

 spinule. For example, in Lagocephalus, the three species 

 (lunaris, scleratus, spadiceus) which have prickles along 

 the back and upper sides in the region of the lateral line 

 show great differentiation of the small scales immedi- 

 ately along the line from those adjacent to them. In the 

 other tetraodontids examined, there is no differentiation 

 of the prickles immediately along the line from those ad- 

 jacent to them, although the course of the lateral line is 

 sometimes clearly indicated in cleared and stained 

 specimens by a wider spacing of the prickles leading to a 

 spineless stripe along the course of the line (especially 

 clearly marked in the two species of Colomesus, and, to a 

 lesser extent, in a few species of Monotreta). 



The internal anatomical diversity of the tetraodontids 

 is much greater than the external. The long-based, 

 many-rayed dorsal fin in Chonerhinos and Xenopterus is 

 reflected in a greater number of vertebrae than in other 

 tetraodontids, Chonerhinos has 26 vertebrae modally 

 (25-27), of which 10 are usually abdominal, while Xenop- 

 terus has 29 or 30 vertebrae, of which 10 or 11 are ab- 

 dominal. All other tetraodontids have modal values 



ranging from 17 to 23, the only species with modal values 

 above 20 being numerous species of Fugu and some of 

 Monotreta and Torquigener. At the opposite extreme, 17 

 vertebrae modally, are all of the species of Canthigaster, 

 that of Carinotetraodon and of Guentheridia, and sever- 

 al species of Lagocephalus, Sphoeroides, and Tetra- 

 odon, the formula always being eight abdominal and 

 nine caudal when the total is modally 17. Since both the 

 Eoplectinae and Triodontidae, ancestral to the tetra- 

 odontids, have 20 vertebrae (9 -(- 11), as do the other 

 basal Triacanthodidae (8 -(- 12), the presence of some- 

 where around 20 vertebrae can be considered general- 

 ized in tetraodontids, and the great increase in number 

 in Chonerhinos and Xenopterus as well as the decrease in 

 number to as low as 17 can both be considered speciali- 

 zations. Moreover, since the Eoplectinae have only a 

 moderately long soft dorsal fin base and a moderate 

 number of rays (14 to 17), with an indication that the 

 number of rays is in the process of being reduced poste- 

 riorly in the series, and the Triodontidae a short-based 

 dorsal fin with a low number of rays (modally 11), the 

 generalized tetraodontid condition can be considered to 

 be a relatively short-based dorsal fin and lowered 

 number of rays (approximately 10 to 12). 



Thus, the great increase in the length of the base and 

 in the number of rays in the dorsal fin of Chonerhinos 

 and Xenopterus must be considered specializations for 

 stronger more sustained swimming, perhaps associated 

 with their invasion of fluviatile fresh waters, although 

 they are also found in coastal marine waters, with the 

 possible exception that the problematical C. africanus 

 may be confined to the Congo. Similarly, the elaborate 

 open cup nasal apparatus and increased number of 

 lateral lines in these two genera are specializations. 

 Chonerhinos and Xenopterus also have specialized skulls 

 (lacking prefrontals in both genera, and with enor- 

 mously enlarged frontals in Xenopterus). 



Chonerhinos and Xenopterus are highly specialized 

 genera. Chonerhinos, which has lengthened the base of 

 the dorsal and anal fins and increased the number of 

 rays to a lesser extent than Xenopterus, and which has a 

 less specialized skull than Xenopterus, obviously is the 

 more generalized of the two. Fraser-Brunner (1943:4) 

 also considered these two closely related genera as highly 

 specialized, with the increased numbers of vertebrae and 

 fin rays "a secondary development." The contention of 

 Le Danois (1959:248) that these two genera are the most 

 primitive of the tetraodontids and ancestral to the 

 diodontids cannot be taken seriously. 



While Chonerhinos and Xenopterus are highly 

 specialized, in my opinion they are not sufficiently differ- 

 entiated anatomically from the other tetraodontids to be 

 recognized as even subfamilially distinct (much less as a 

 separate family). However, one genus of tetraodontids, 

 Canthigaster, is usually recognized as a distinct family 

 both in the general ichthyological literature and in that 

 more specifically devoted to plectognaths, a practice 

 that I have previously followed. Since about two dozen 

 species of Canthigaster are usually recognized as a 

 separate family, and it is being suggested here that this 



