xxviii INTEODUCTION. 



a cutting back of the western rivers has resulted in their capture, with tlie result 

 that the eastern rivers with which they were formerly connected have diminished in 

 size and fail to reach the main stream. 



The Eio Grande Province differs from the Mississippi Province principally in the 

 paucity of types. Not one of the endemic uearctic families is present ; the Esocid* are 

 absent, and the Percidiv and Centrarchidse are represented by a few species only. A 

 considerable proportion of the true fresh-water fishes belong to the Cyprinidce, whilst 

 the presence of Cichlida^ and Characinidie is a positive feature which distinguishes this 

 area from other parts of the Nearctic llegiou. 



The E,io Lerma System (excluding the Rio Grande de Santiago below the falls, and 

 properly including, as Dr. Meek has shown, the Rio San Juan, a tributary of the Panuco, 

 as well as the isolated lakes in the States of Miclioacair and Mexico) has so peculiar 

 a fish-fauna that it may be regarded as a separate sub-region of the Nearctic Region. 



The viviparous Cyprinodontidas of the sub-family Characondontina? are characteristic 

 of and nearly peculiar to this, the Lerma, Sub-region, in which the Atlierinid genus 

 CIdrostoma is represented by a number of species which show a remarkable diversity. 

 Both these groups are probably derived from marine ancestors which entered the river 

 at a remote epoch ; none of the marine types (Gobiidas, Mugilid;E, &c.) which are 

 found at the p>rescnt day in neighbouring rivers, sucli as the Balsas and Panuco, have 

 been able to make their way into the Lerma System, from which neotropical fishes are 

 also absent. 



The Cyprinida3 of this sub-region differ considerably from those of the Rio Grande, 

 as five of the seven genera are endemic. With the exception of a Cat-fish {Amiitrus) 

 and a Lamprey (Lampctra), all the fishes of the Lerma Sub-region belong to the three 

 groups already mentioned. 



Below, in comparing some of the shore-fishes of the Pacific and Atlantic coasts of 

 Central America, it is shown that in many cases D. S. Jordan's generalization — to the 

 effect that a form occurring in a certain area has as its nearest relative a form 

 inhabiting a neii^'hbouring area separated by some sort of barrier from the first — holds 

 good. Li such cases isolation appears to have been a factor in determining specific 

 differentiaticjn. 



Jordan's generalization also holds good for many groups of fresli-watei' fishes, but 

 f(n' many others it does not. As examples of the latter we may instance the Cichlid 

 fauna of Lakes Managua and Nicaragua and the greater part of the fisli-fauna of the 

 Lerma System. 



