Application of Regression Techniques to Studies of Relative Growth in Crustaceans

Publisher Website
Google Scholar
Abstract
Studies of relative growth in crustaceans are often based, in whole or in part, upon the practice of log transformation of all linear morphometric data prior to the calculation of a least squares estimate linear regression for a broad size class of specimens. This and other practices may result in misinterpretation of the regression analysis. We propose that analysis of untransformed data with a Model II regression technique such as the reduced major axis is appropriate for description of relative growth in crustacean linear morphometric features, and that data for broad size-class samples should be examined as iteratively divided subsets, rather than assuming a priori that the entire sample is adequately represented by a single continuous function. A statistical test for evaluation of regression functions on the basis of randomness of residuals is provided.