22 FRESH-WATER FISHES OF SOUTH AMERICA EIGENMANN. 



found elsewhere only in the Amazons and northward. These are very 

 probably late immigrants from the Amazon. This sj-stem may provi- 

 sionally be set apart as the San Francisco province. 



South of the Eio San Francisco is a province well marked both posi- 

 tively and negatively. Very many genera found to the north and to 

 the south (see List iii) have no representatives here, while a large num- 

 ber of genera are peculiar to the region. Its northern and southern 

 limits can not yet be defined ; roughly speaking, it includes all the 

 Atlantic slopes of Minas Geraes, Bahia, and Eio de Janeiro. The chief 

 river, and the one most thoroughly explored, is the Parahyba with its 

 tributaries Muriahc and Kio Preto. Other rivers are the Itabapuana, 

 Doce, Mucuri, Jequitinhonha, Pardo, Paraguassu. 



The isolation of this j)rovince proves in a very decided manner that 

 the large number of genera of the La Plata which are also found in the 

 Amazons have not reached the La Plata by way of the sea. The region 

 may be termed the Atlantic province. 



The mountain streams of Colombia, Ecuador, Peru, and Bolivia are 

 inhabited by a number of peculiar genera (xv) and a large number of 

 peculiar species, especially of Fygidium and Cluvtostomus. The pecul- 

 iarities are such that these mountain regions may readily be distinctly 

 separated as the Andean province. The genera l^rcmopli'dus and Astro- 

 hlepus ascribed to the Magdaleua may belong to this province. Its 

 boundaries are necessarily very irregular and as yet not well defined. 

 Species of Pygidium, which are here especially abundant, are also found 

 in the coast rivers of Peru and southward to Chili, thus forming an 

 important portion of the Fuegian fauna. The most important body of 

 water is Lake Titicaca and the headwaters of both the eastern and 

 western slopes are included. 



Of the La Plata province little need be said at this time. A very 

 large part of it has not yet been explored. At present the province 

 must be distinguished by its negative characters. The genus Gochliodon 

 is so nearly related to Amazonian genera that it is of no great impor- 

 tance. The way in which Amazonian genera may enter the La Plata 

 system has been pointed out above. 



With the present data the Brazilian ^ubregion may provisionally be 

 divided into the following provinces: (1) Pacific, (2) Andean, (3) JViag- 

 dalena, (4) Orinoco, (5) Guiana, (G) Amazonian, (7) San Franciscan, (8) 

 Atlantic, (9) La Plata. 



This account would not be complete without a few words in regard 

 to Central America and Mexico. The latter may be dismissed with the 

 statement that its northern half contains North American forms chiefiy 

 while its southern half has a large proportion of Central American 

 forms. The Central American fauna consists of very few northern 

 types, the great majority being modified representatives of South Amer- 

 ican forms. There does not exist at present a sufiicient barrier to pre- 



