Overcoming problems with the use of ratios as continuous characters for phylogenetic analyses

Abstract

The use of quantitative morphometric information for phylogenetic inference has been an intensely debated topic for most of the history of phylogenetic systematics. Despite several drawbacks, the most common strategy to include this sort of data into phylogenetic studies is the use of ratios, that is quotients between morphometric variables. Here, we discuss one particular problem associated with such methodology: the fact that the often arbitrary election of which variable serves as numerator and which as denominator affects the phylogenetic outcome of the analysis. We describe the cause for such an effect, and study its implications with the use of several published data matrices. Alternative coding schemes for ratio characters result in very different phylogenetic hypotheses, an effect that may even be strong enough to affect studies that combine continuous and discrete morphological information. Some of the resulting incongruence is produced by the differences in magnitude of the continuous characters involved, although different rescaling techniques are shown to decrease, but not eliminate, the confounding effect. To eliminate such problematic effect, ratios should be either log‐transformed before their use or replaced by more effective ways to capture morphometric information.