MEASUREMENT OF BIOLOGICAL AND HISTORICAL INFLUENCES ON PLANT CLASSIFICATIONS

Abstract

SummaryPlant classifications have a frequency distribution of taxon sizes which is very skewed (the celebrated “hollow curve” of Willis). In the first part of this paper I suggest that the best general measure of the hollow curve is the variance of the fitted lognormal distribution. By this means we can compare classifications, such as Linnaean classifications with modern ones; the Linnaean classification of the Gramineae is more “lumped” but more “even” than the modern one.In the second part I use phenetic automatic classifications as a model for the classification process to show that (as Walters (1961) has predicted) classifications are heavily influenced by the method used (amount of chaining), but that the biological patterns of character state distribution among organisms also contribute to the hollowness of the curve. Although classifications are influenced by historical factors, underlying biological patterns are still evident in the relative sizes of genera. The model indicates that artificial and “real” components are present in classifications in roughly equal proportions. It seems that Clayton's (1983) phylogenetic “optimism” and Walters' (1986) psycho‐historical “realism” both need modification.Monotypic taxa, in particular, may have a biological meaning. Willis interpreted monotypic genera as nascent and big genera as old. Walters interpreted monotypic genera as young historically and big genera as old historically. I suggest, from relict theory, that monotypic taxa tend to be ancient relicts and large genera tend to represent blooms of recent evolution, and that extinction, not evolution, tends to delimit taxa.