On rethinking allometry

Abstract

Analysis of sets of intra-species and inter-species allometric relationships reveals that the inter-species data generally fit an exponential model better than a linear model. The intra-species data seem equally suited to either model. Skewness of the data and the effect of logarithmic transformations on correlation coefficients are examined in the light of these findings. Inter-species data are approximately lognormally distributed and logarithmic transformations are necessary to produce linear relationships. As a consequence, correlation coefficients usually increase after logarithmic transformation of inter-species data.