Morphological correlations in evolution: Consequences for phylogenetic analysis

Methods of inferring phylogenetic relationships have received an enormous amount of attention in recent years. One missing component in much of this work concerns the analysis of the data being used. Most techniques of phylogenetic inference, including parsimony, assume the independence of characters, and the assignment of equal weights to characters assumes that they have an equal probability of change. Although systematists using molecular data have begun to deal with issues of character correlation and weighting, they are generally avoided by those using morphological data. Systematists have paid relatively little attention to morphological characters that may be correlated for reasons other than commonality of descent. We review some of the reasons that the assumption of independence of characters may be violated for morphological features, and suggest ways in which detailed analysis of the phenotype may lead to both a priori and a posteriori rationales for weighting of morphological characters. Just as it seems prudent to explore phylogenetic hypotheses that are less than the most parsimonious, systematics would be well served if researchers more fully explored the causes and consequences of character correlations that may exist in morphological data sets. Our two examples, the evolution of sexual dimorphism in voiceless frogs and the role of heterochrony in the evolution of chaenopsid fishes, illustrate that character weighting provides potential insights into both systematic relationships and the evolution of character complexes.