Rarity and Body Size: Some Cautionary Remarks

Abstract

Listen

Translate article

Dobson and Yu (1993) advocate removal of the statistical interaction of body size with abundance and geographic range size before species are classified with respect to rarity. They argue that the relative rarity of small and large species can be fairly assessed only when any systematic tendencies for different-sized species to have somewhat different natural densities and range sizes have been accounted for. We have some sympathy with this viewpoint, but we urge caution in the general application of such an approach for several reasons. (1) Other variables may be important. Body size is only one of a great many variables for which correlations with abundance and range size have been reported. Other variables include diet, trophic level, reproductive strategy, dispersal ability, and phylogeny (see Brown & Maurer 1987; Gaston & Lawton 1988; Gaston 1988, 1990, 1994a; Nee et al. 1991; Kunin & Gaston 1993; Blackburn et al. 1994). Some of these relationships are doubtless second-order effects resulting from correlations between these additional variables and body size; body size is known to interact with a variety of life-history and ecological parameters (Peters 1983; Calder 1984). Others are likely to be more direct interactions, however, which raises two questions. First, should body size have any logical precedence over other variables for explicit treatment as a covariate of abundance or geographic range size (aside from the fact that body size data are more likely to be available)? Second, shouldn't rarity also be adjusted for these other effects? We suggest that it is arguable whether body size per se should be treated differently from any other variable, particularly given that the cause and effect of many body-size interactions are poorly understood. Following recent trends in comparative biology, one can argue that the basis for removing any effect of body size on abundance or range size is particularly weak unless due account is taken of the consequences of the phylogenetic relatedness of the taxa concerned (see Harvey & Pagel 1991). Failure to recognize this problem could lead to removal of the wrong interactions. (2) Body size is seldom related to abundance or range size in a simple manner. Dobson and Yu's analyses are based on linear relationships of density and range size with body mass. The interactions of either, however, are seldom simple. Descriptions based on correlation coefficients or linear regressions are often rather poor. For example, body size explained more than 20% of the variance in population abundance in only one of 14 animal assemblages examined by Blackburn et al. (1993), and less than 10% in nine of the assemblages. Brown and Maurer (1987) showed that the most abundant landbird species in North America tend-to be medium-sized, not the smallest species. The shape of the relationship between body size and abundance among these species was approximately triangular (see Gaston 1988; Morse et al. 1988; Nee et al. 1991; Blackburn et al. 1993, 1994). Analysis of species rarity in data of this kind assuming a linear relationship between body size and abundance may tend to rank the commonest small species as relatively rare. Clearly, an analysis controlling for the effects of body size on rarity should take account of the shape of the size-abundance relationship. If the relationship between body size and abundance is triangular, however, then analysis is not straightforward. Linear regression cannot be used because the magnitude of residuals of densitybody size plots are themselves likely to be functions of body size: residuals will tend to be constrained to be small at large body sizes in a triangular relationship. Similar problems pertain to the analysis of range sizebody size relationships, which also tend to be triangular in that small-bodied species tend to have a breadth of geographic range sizes and large-bodied species tend to have only large range sizes (Brown & Maurer 1987; Gaston 1994a). One solution may be to fit polynomial regressions to triangular relationships (see Cotgreave et al. 1993), although how the residuals from such relationships would be interpreted is unclear. Dobson and Yu argue that the influence of body size on population density and range size should be removed in assessing rarity when there is a priori evidence of a signiflcant influence of body size. In practice, such evidence may be misleading. Paper submitted December 7, 1993; revised manuscript acceptedAugust 19, 1994.