66 FISHES OF WESTERN SOUTH AMERICA 
ranges eastward and the creation of large areas having a dry climate at their 
headwaters. 
The existence of terraces showing that Lake Titicaca was once much larger, 
extending as far as Pueard, and that the Desaguadero is paralleled by the shore- 
lines of an ancient strait, were pointed out by Agassiz in 1875 (q. v.).. The present 
shrunken lake is evidence of a progressively drier climate with the upthrust of the 
Bolivian cordillera. That it has been in more or less continuous existence from 
the date of its rise from the sea is indicated by the considerable saline deposits on 
the land, and the brackish character of the waters of Lake Poopé. The terraces 
300-400 feet above the lake are much higher than the present pampa which holds 
back the exit of the lake through the Rio La Paz. This goes to show that when 
they were formed the La Paz area was much higher, and that the cutting down of 
the La Paz gorge through the soft, alluvial formations is of subsequent occurrence. 
The fishes of the lake gained admittance to this closed system from the parent sea 
itself, rather than via an Amazonian tributary, and their lack of kinship with other 
fishes of the continent would lead to the same conclusion. 
In the above paper Eigenmann says that while isolation is a necessary condition 
to species-formation, there is insufficient isolation in Lake Titicaca itself to have 
given rise to so many species. That has been one of the puzzling questions in the 
present study. We find perhaps the most complete, if not the longest, isolation to 
be that of Lake Ascotan, Chile, yet, except for color, its fishes differ little from those 
of the great lake. On the other hand, the nearby Lake Umayo, rather well sepa- 
rated, has been prolific of species not yet found in the greater, parent body. Fur- 
thermore, our collecting records show that there is an assortment of species with 
depth, and between the lake and its tributary streams, the nearest thing to isolation 
I can suggest. But, here as elsewhere, we do not often find more than one or two 
species in any particular locality. 
For a long time I was troubled to answer the question as to the distribution of 
Orestias agassizi, supposedly existing all the way from the Nudo de Pasco in the 
north to Lake Ascotan a thousand miles to the south in the Chilean Andes, con- 
nected by a terrain dissected to a depth of many thousand feet by the Mantaro, 
Apurimac, and Urubamba, an effective barrier except perhaps about the high passes 
of the voleanic western cordillera. How could a species maintain its integrity 
against such odds? There seemed to be a tentative answer in our finding the 
species even at the extreme elevation of 14660 feet in the pass Crucero Alto. But 
now after a careful restudy of all my specimens and comparison with those from 
the California Academy of Sciences, I find that the species are distinct in central 
and southern Peru. The confusion arises from the close resemblance of the imma- 
ture stages, for the adults are quite unlike. We may look to the Crustacea for 
additional evidence of the marine origin of Lake Titicaca, for seven species of 
Orchestiids exist there. 
Resumé: In the total absence of species derived from southern invaders, only 
Astroblepus from the northward, no Pacific slope forms scaling the extreme heights, 
we have only two primary sources of the fishes of the cordilleras and interandine 
valleys. These are the aggressive species from the Amazon and the endemic 
upland forms. 
