346 WHITTAKER 



ural areas defined by species distributions, distributional centers — have no 

 very close relation to the association concept or to Clements' regional vegeta- 

 tion units. 



One direction in the modernization of Clements' association concept has 

 been followed by Braun (1935, 1938, 1947, 1950, 1956) in her work on the 

 eastern forests. Braun's associations are, in manner of definition, derived 

 from Clements'; each is a large-scale, relatively heterogeneous unit which 

 characterizes a geographic region as its climax. Clements' three-part division 

 was shown by Braun to be unrealistic; and Braun recognizes six regional asso- 

 ciations (not including the Lake Forest, the evergreen forest of the Gulf 

 Coastal Plain, and the transitional Western Mesophytic Region). Of impor- 

 tance at least equal to this change of view is the immense complexity of detail 

 in these forests described by Braun (1950). Because of this complexity, 

 Braun's six associations are by no means all that an ecological splitter might 

 choose to recognize. The association is a more or less arbitrary unit (Braun, 

 1947, 1950, p. 525); Braun's associations are no more than Clements' the 

 unique and inescapable solution to division of the eastern forests. Braun's 

 division is not simply true where Clements' was false; it is a better pattern 

 of abstraction from the eastern forests because more realistic, more frankly 

 cognizant of the underlying complexity in seeking that simplification from 

 complexity which is necessary to any abstraction. Thus in Braun's work, 

 retention of some of the essential form of Clementsian ecology is combined 

 with a fundamental change in perspective. 



Very different directions have been followed in Europe. In Braun-Blanquet's 

 (1921, 1932, 1951) system, the one most widely applied by phytosociologists, 

 communities are classified by character species — -species, often rather obscure 

 ones, which are centered in or largely confined to a given community type and 

 which, hence, are more truly characteristic of it than wide-ranging dominants. 

 An association is in general a vegetation unit of the lowest rank which can 

 be defined by character species. For most areas, the approach through domi- 

 nants is thought suitable only for preliminary and superficial work; but an- 

 other vegetation unit, the sociation, is used in areas such as Scandinavia, 

 where communities contain relatively few species. This unit is one of small 

 scale, defined by dominants of the various strata, not of the uppermost stra- 

 tum alone; and it is unrelated to Clements' regional associations. For neither 

 the sociation nor the association is climax stability or regional prevalence re- 

 quired, as for the association of Clements. Although regional vegetation types 

 are recognized there, the American association as such is used by no one 

 in Europe. In contrast to Clements' view of formations and associations as 

 "concrete" entities, most European authors emphasize that the association is 

 an abstraction — the association first comes into being at the phytosociologist's 

 desk (Klapp, 1949, p. 10). Increasingly detailed knowledge of the vegetation 

 of Europe has led to increasing recognition of the limitations inherent in even 



