Testing the directionality of evolution:: the case of chydorid crustaceans

Although trends are of central interest to evolutionary biology, it is only recently that methodological advances have allowed rigorous statistical tests of putative trends in the evolution of discrete traits. Oligomerization is one such proposed trend that may have profoundly influenced evolutionary pathways in many types of animals, especially arthropods. It is a general hypothesis that repeated structures (such as appendage segments and spines) tend to evolve primarily through loss. Although largely untested, this principle of loss is commonly invoked in morphological studies of crustaceans for drawing conclusions about the systematic placements of taxa and about their phylogeny. We present a statistical evaluation of this hypothesis using a molecular phylogeny and character matrix for a family of crustaceans, the Chydoridae, analysed using maximum likelihood methods. We find that a unidirectional (loss-only) model of character evolution is a very poor fit to the data, but that there is evidence of a trend towards loss, with loss rates of structures being perhaps twice the rates of gain. Thus, our results caution against assuming loss a priori, in the absence of appropriate tests for the characters under consideration. However, oligomerization, considered as a tendency but not a rule, may indeed have had ramifications for the types of functional and ecological shifts that have been more common during evolutionary diversification.