BININDA-EMONDS, O. R. P.*; JEFFERY, J. E.; DONOHOE, A.; RICHARDSON, M. K.; Technical University of Munich; Leiden University; University College London; Leiden University: Reconstructing phylogenetic relationships from developmental timing data

There has been increasing interest recently in analyzing developmental timing data in a phylogenetic framework. Several studies have now shown that changes in such timing data (i.e., sequence heterochrony) map well to �known� phylogenies. Therefore, the obvious question is whether the phylogenetic signal that heterochrony appears to contain can itself be used to infer phylogeny. Because the timing data are continuous and not easily comparable between species, they must be recoded into a form suitable for phylogenetic analysis. We examine three coding methods: event-pairing (EP), which is well established in evolutionary studies of heterochrony, and two methods taken from the analogous problem of inferring phylogeny from changes in chromosomal gene order, breakpoint distances (BP) and junction coding (JC). We examine the entire issue initially from first principles, establishing three desired features a coding method should have. Only JC possesses all three features. Both EP and BP lack one feature each. We then use simulation to examine the performance of the three methods in reconstructing a known model tree of 20 species given a data set of 100 developmental events. Overall, EP consistently outperformed both BP and JC in terms of accuracy and the number of optimal solutions produced (�decisiveness�). Moreover, the performance of EP was largely unaffected by the frequency of heterochrony, whereas both BP and JC suffered increasing decreases in performance as heterochrony became more common. However, we caution that accuracy was never greater than 60%, indicating that current methods of coding developmental timing data are still inadequate for phylogeny inference.