The analysis of consecutive ontogenetic stages, or events, introduces a new class of data to phylogenetic systematics that are distinctly different from traditional morphological characters and molecular sequence data. Ontogenetic event sequences are distinguished by varying degrees of both a collective and linear type of dependence and, therefore, violate the criterion of character independence. We applied different methods of phylogenetic reconstruction to ontogenetic data including maximum parsimony and distance (cluster) analyses. Two different data sets were investigated: (1) four simulated ontogenies with defined phylogenies of six hypothetical taxa, and (2) a set of “real” data comprising sequences of 29 ontogenetic events from 11 vertebrate taxa. We confirm that heterochronic event sequences do contain a phylogenetic signal. However, based on our results we argue that maximum parsimony is a biased method to analyze such developmental sequence data. Ontogenetic events require a special analytical algorithm that would not neglect instances of chronological (horizontal) dependence of this type of data. One coding method, “event-pairing”, appeared to fulfill this requirement in the vertebrate analyses. However, to accurately analyze ontogenetic sequence data, a more sophisticated coding method and algorithm are needed, for example, measuring distances of dependent events.
A Memetic Algorithm for Phylogenetic Reconstruction with Maximum Parsimony