POE, Steven; WAKE, Marvalee H.; University of New Mexico; University of California, Berkeley: Quantitative tests of general theories for the evolution of development

Recent interest in integrating development and evolution has focused attention on general theories and models for the ways that development evolves. Three models have been proposed to describe the evolution of ontogeny, primarily for vertebrates. According to the early conservation (EC) model, evolutionary change occurs more frequently in late development whereas early development is relatively conserved. In the hourglass (HG) model, early and late development are relatively labile and an intermediate, “phylotypic” stage is most conserved. A third model is the adaptive penetrance pattern (AP). Verbal and descriptive arguments have been offered in favor of and opposition to each of these models, but only recently have quantitative tests been proposed. The concept of sequence events in heterochrony suggests testable predictions for these models. The EC model predicts changes in timing will be less frequent earlier in ontogeny, the HG model predicts changes will be least frequent at an intermediate stage, and the AP model predicts that changes will be most frequent at an intermediate stage. Simpler models also make predictions for heterochrony. A model we call the adjacency (AJ) model predicts that small changes in ontogenetic position are more likely than large changes. An unconstrained (UC) model allows all types of change to be equally probable such that there is no phylogenetic inertia of ontogenetic position. We present a phylogenetic statistical assessment of these models and show that the more complex EC, HG, and AP models are not required to explain the patterns observed in developmental event data for 14 species of vertebrates. Rather, the simple AJ model is adequate.