58 ANNALS NEW YORK ACADEMY OF SCIENCES 



the Amazon, and yet there is as great, if not greater, identity of fishes 

 there than in the case of the Paraguay. 



3. There are connections between the Orinoco and the Amazon and 

 the Sao Francisco, and yet there is less similarity between their faunas 

 than between these of the Paraguay, Guiana and the Amazon. 



The key to the explanation of the distribution of the characinids lies in 

 the 50 genera which are not only widely distributed, more or less cos- 

 mopolitan, but the more generalized members of their families. The 

 same is also strongly indicated by the fact that when these widely dis- 

 tributed generalized genera arrived in the different environmental com- 

 plexes, there resulted less conformity between the species than between 

 the genera and a greater similarity between the species which lived in 

 similar environments, even though they are not connected, than between 

 the species in dissimilar environments which are connected. 



I have already shown that the Piano Alto separates the La Plata basin 

 from the Amazon. This being the case, we have to look for another ex- 

 planation of the distribution of the fishes other than river connections. 

 The most natural way to seek the explanation is first to consider the 

 fishes which live on the Piano Alto. 



I have found the following genera on the Brazilian highlands (a max- 

 imum list) : 



Siluridae : 



Callichthys, Pydiclium, Rhamdia, Pimelodus, Pimeladella, Plecostomus, 

 Doras, Trachycorystes, Auclienipterus, Heleogenes, Loricaria, Hoplos- 

 temum and Corydoras. 



Characinidse : 



Erythrinus, Hoplias, Holerytlirinus, Characidium, Creatochanes, Pazcllu- 

 riclithys, Acestorhynchus, Curimatus, Moenkliausia, Astynax, Tetrago- 

 nopterus, PhyrrTiulina, Poecilocharax, Serrasalmo and Chalcinus. 



Cichlidse : 



Geophagus, CrenicicMa, JEquidens, Cichlasoma and in two places Hereto- 



gramma. 

 Rivulus, Symoranchus marmoratus, Hypopomus brevirostris, Gynmotus 



carapo, Eigenmannia virescens and Sternopygus macrurus. 



All of the above genera are not only widely distributed but also the 

 most generalized types of their sub-families. They are represented in 

 the highlands proper by only a few species. For example, I found only 

 twenty-five species in the highlands of Parana above the big Iguassu 

 Falls. The following is the list which I collected in the highlands of 

 northern Goyaz : 



