The essential role for graphs in allometric analysis

The allometric method, which is widely (but somewhat inaccurately) attributed to Julian Huxley, is at the heart of some of the most important theoretical models in the field of evolutionary biology (e.g. the Metabolic Theory of Ecology). The procedure entails fitting a straight line to logarithmic transformations of the original bivariate data and then back‐transforming the resulting equation to form a two‐parameter power function on the arithmetic scale. Although the distribution for logarithms in graphical display may satisfy the requirement for linearity (as well as assumptions of the fitted model for normality and homoscedasticity), this does not guarantee that the power function estimated by back‐transformation will actually describe the distribution for the original data. Situations of this kind arise with some regularity when untransformed observations lack a unitary pattern, thereby rendering them unsuitable for use in allometric research. Quality of the data and sufficiency of the power function can be judged only by examining a graph of observations on the arithmetic scale. Graphical display of untransformed observations needs to be incorporated into the allometric method to ensure that future studies on the influence of body size on the evolution of form and function are not compromised by bad data.
Source: Biological Journal of the Linnean Society - Category: Research Authors: Tags: Comment Source Type: research