NO. 3538 CHARACID FISHES — ^WEITZMAN 51 



CAS(IUM) 11690, SL 19.5-23.7 mm., British Guiana, Essequibo River at Gluck 

 Island, Sept. 30, 1908, C. H. Eigenmann.— Two, SU 50234, SL 25.0-27.0 mm., 

 British Guiana, no other data. — Four, SU 50225, SL 29.4-35.4 mm., aquarium 

 specimens, no other data. — Three, SU 50224, SL 25.8-32.5 mm., aquarium 

 specimens, alizarin preparations, no other data. 



Discussion 



In the course of this study three major problems have become ap- 

 parent. The first of these is that of generic allocation; the second and 

 really inseparable one is the relationships among the species; and the 

 third and largest problem, left largely unworked by this study, is geo- 

 graphical variation and subspeciation. To untangle the nomencla- 

 tural mixup and mistaken identifications of past authors has presented 

 some minor difficulties, and the conclusions reached are presented in 

 the synonymies above. 



Generic designation has proved difficult; indeed, I have believed at 

 times that perhaps it would be best to place all generic and subgeneric 

 taxa treated here under Nannostomus, following Sterba and Tucker 

 (1962). However, it seems that to do this would obscure valid differ- 

 ences and relationships. It appears that the only valid difference be- 

 tween Nannostomus and Poecilobrycon is the presence or absence of a 

 sensory tube in the second orbital bone. This difference seems smaU, 

 but it proved to be remarkably constant in all specimens examined. 

 The only other character found to correlate \vith this was a tendency 

 for a longer snout in Poecilobrycon than in Nannostomus. However, 

 measurements show an overlap in snout lengths even though individual 

 bones of the snout region always seemed broader and relatively shorter 

 in Nannostomus than Poecilobrycon. Both Poecilobrycon eques and 

 Poecilobrycon unifasciatus are obviously closely related because of their 

 caudal fin structure and swimming habits. Their tendency toward 

 long snouts and their possession of tubed second infraorbital bones 

 indicate a relationship with Poecilobrycon harrisoni. Indeed, the liv- 

 ing color pattern of unifasciatus suggests relationships both with harri- 

 soni and eques. These facts made it seem desirable to place these 

 fishes under the genus Poecilobrycon with Nannobrycon as a subgenus 

 for unijsaciatus and eques. The color pattern of P. harrisoni suggests 

 relationships with Nannostomus beckjordi, and it is possible that they 

 had a common ancestor. It thus seems that Poecilobrycon is most 

 closely related to that section of the Nannostomus containing beckjordi. 



A few groupings can be determined in Nannostomus itself. Nannos- 

 tomus trijasciatus and marginatum are obviously closely related in color 

 pattern and anal fin structure; indeed, a separate subgenus could be 

 erected for their reception but I do not believe it advisable. Nanno- 

 stomus espei appears apart from all other members of this genus in hav- 



