52 PROCEEDINGS OF THE NATIONAL MUSEUM voi-. ii9 



ing the shape of certain scales and the color pattern similar to those of 

 pyrrhulininans. However, a generic or subgeneric name does not seem 

 advisable for this species. Whether these pyrrhulinian-like characters 

 of espei indicate an archaic natm-e is problematical. I am inclined to 

 believe that they are, especially when one considers that the body 

 shape of this species is more like that of pyrrhulininans than any other 

 nannostominan. On the other hand, it lacks the tube in the second 

 orbital bone that is present in Poecilobrycon and the pyrrhulininans. 

 Presumably the presence of this tube is a primitive character for nan- 

 nostominans. It is possible that the loss of the tube in espei was inde- 

 pendent of its loss in other species of Nannostomus and that my 

 association of espei with species of Nannostomus rather than with 

 Poecilobrycon is artificial. The problem cannot be settled Avith the 

 morphological data at hand. 



Nannostomus digrammus is likewise apart from other species of Nan- 

 nostomus in the highly modified anal fin of the male. It appears, on 

 basis of color pattern and premaxillary teeth, to be more closely 

 related to beckfordi and bijasciatus than to any other known nanno- 

 stominan. 



Both N. beckfordi and A^". bijasciatus appear related. They lack 

 specialized or distinctive characters that set the other species of 

 Nannostomus apart from them. They both lack an adipose fin ; appar- 

 ently both have at least some perforated lateral-line scales and have 

 only moderately modified anal fins in the male. Their color patterns 

 are somewhat different. However, until the color pattern and other 

 characters of the widely distributed beckfordi can be studied adequately, 

 it will remain difficult to evaluate its relationship with bifasciatus. 



Concerning geographical variation within a species, N. espei, N. 

 bifasciatus, and P. harrisoni appear to have restricted ranges, and little 

 or no geographical variation was noted in the specimens utilized in the 

 present study. Nannostomus beckfordi, digrammus, trifasciatii-s, margi- 

 natum, Poecilobrycon unifasciatus, and eques have a wide distribution in 

 South America. Of these, beckfordi and marginatus were noted to 

 have the greatest geographical variation. Indeed beckfordi is so vari- 

 able that future work based on sufficient collections may show good 

 reasons to recognize many subspecies and perhaps even species of this 

 group of related populations. 



Although some morphological, especially live color, variation cor- 

 related with geographical distribution was noted in marginatus, the 

 close relationship of these population samples seems clearer than those 

 of beckfordi. Some degree of difference was found in specimens of 

 trifasciatus from Peru and British Guiana and this, it seemed, could 

 be correlated with color differences in aquarium specimens said to have 



