88 F- E. Clements 



by an absence of it, with low rainfall and high temperature as 

 causal factors. Though the Arctic tundra is sometimes regarded 

 as a desert, this is clearly a return to the primitive meaning of the 

 word and is warranted neither by vegetation nor by climatic 

 conditions. As a consequence, an accurate and practicable defi- 

 nition of the term apparendy must be based primarily upon 

 vegetation and rainfall, with some consideration of the animals 

 and of high temperature during the growing season at least. 

 In brief, a desert climax is one marked by the absence of forest 

 or grassland, a critical deficit in rainfall, and a high potential 

 evaporation caused by excessive heat and often by high winds 



also. 



In applying this test to the vegetation of North America, it 

 becomes clear that the term desert is often loosely employed by 

 ecologist and geographer alike. Thus, the so-called sagebrush 

 desert does not meet the requirements over all or almost all of 

 its extent, and even the more xeric Larrea community is desert 

 in only a part of the area concerned. The most effective test is 

 afforded by tracing the boundary of the grassland climax in the 

 Southwest and, when this is correlated with rainfall or, better 

 still, rainfall/evaporation, where this is possible, it provides the 

 best delimitation of desert available today. In general, the 

 isohyet of 5 inches marks the disappearance of grass dominants 

 on the climax level, and thus is the readiest means of setting off 

 the desert climax. Much attention has been given to following 

 this boundary on the ground, but it will suffice here to state that 

 the desert, as characterized by Larrea, Franseria, and their typi- 

 cal associates, is confined to the Death Valley, Mohave, and 

 Colorado regions and to a larger but less known area in Mexico. 

 Though small districts of similarly low rainfall occur to the 

 north, there temperature and evaporation are less telling, and 



