Fishes of the Order Cyprinodontes 

 II. An Analysis of the Genera of the Poeciliidae 



Since 1913, when Regan published A Revision of the Cyprinodont Fishes 

 of the Subfamily Poeciliinae/ — an epoch-making- work in the study of this 

 group of tropical American fishes, — a number of new types have been dis- 

 covered, and new light has been thrown on the characters and relationships 

 of forms previously described. It seems advisable, therefore, to bring this 

 revision by Regan up to date and to include at the same time some original 

 observations and deductions. 



As Regan's review demonstrated, by far the most valuable of all the cliar- 

 acters which have been used in the classification of these little fishes are 

 those taken from the structure of the gonopodium^ (this is the intromittent 

 organ of the male, formed by the elongation and singular elaboration of rays 

 3, 4 and 5 of the anal fin; see figures). With the aid of gonopodial charac- 

 ters, Regan and other writers have been able to demonstrate repeatedly that 

 species which previously had been closely associated together in reality should 

 be distributed to distant parts of the series, and that species which had been 

 widely separated in the older classifications were, nevertheless, very closely 

 related. In the older systems the characters chiefly used involved the length 

 of the intestine and the shape and arrangement of the teeth, features corre- 

 lated closely with food, and now known to have been repeatedly and inde- 

 pendently altered on the different lines of evolution within the family (these 

 characters still appear to be of high value in the classification of the Poeci- 

 liidae, however, even now that it has been shown that they cannot safely be 

 employed in the primary division of the group). 



Unfortunately, males are often not taken, even when a considerable series 

 of these fishes are collected, or, if taken, the males may be either young or 

 non-breeding individuals, in either case not showing the characteristic struc- 

 ture of the gonopodium. Unless a species is known and recognized, it can- 

 not be certainly classified in the absence of males with perfect gonopodia ; 

 usually even its approximate position in the series cannot be determined. 

 In fact, species of one cyprinodont family known only from females may be, 

 as they have been at various times, placed in genera belonging to another 

 family of the group. 



Gonopodia may be quickly mounted on a slide, using preferably the 

 glycerine sodium silicate medium recently described by Greaser and Clench,^ 

 and thus readily studied and drawn with the aid of a microscope, or, better, 

 of a projection apparatus. In routine identification most of the gonopodial 

 features may be made out with the aid of a 12X or 18X hand lens. 



1 Regan, Proc. Zool. Soc. London, 1913, pp. 977-1018; figs. 168-173, and pis. 99-101. 

 The Poeciliinae of Regan are equivalent to the Poeciliidae of the present paper. 



- Gonopodia were figured and used as systematic characters as early as 1848 

 (Heckel), 1855 (Poey), and 1907 (Eigenmann), but Regan was the first to use gono- 

 podial characters in a comprehensive revision of the entire group. 



8 Trans. Am. Micr. Soc, 42, 1923, pp. 69-71. 



