COLBERT, M.W.; The University of Texas at Austin: Ontogenetic Sequence Analysis: a parsimony-based method for characterizing developmental hierarchies

Most animals lack longitudinal data documenting sequential changes during ontogeny. Accordingly, sequences must be estimated using cross-sectional samples that have been ordered into series based on age or maturity criteria. Size has been the primary criterion for ordering samples, but size variability is prevalent, and its use in ordering can distort patterns of sequence variation. To better assess ontogenetic variation, a parsimony method is presented that utilizes a matrix of scored ontogenetic characters to unravel hierarchical developmental patterns. This method, the ‘Ontogenetic Sequence Analysis’ (OSA), involves PAUP analysis in two treatments of this matrix of irreversible ontogenetic character transformations. The results of OSA include ‘sequence map’ diagrams that depict reticulating networks of ontogenetic sequences leading from the least to the most mature observed phenotypes. From these maps all most parsimonious sequences can be derived, allowing comparison of both intra- and interspecific sequence variation. Consensus-based methods can also be used to derive a single ‘most-likely’ sequence. Placed in a phylogenetic context, these sequence data provide a framework for interpreting heterochronic patterns of developmental evolution. The OSA method is explored using three examples of mammalian postnatal skeletal ossification, including longitudinal human samples, age-calibrated cross-sectional human data from the literature, and original, non-calibrated data on tooth eruption and cranial suture closure in extant species of Tapirus. These examples illustrate the validity of the OSA method for discovering sequence patterns, and emphasize the potential of sequence maps and consensus sequences for evolutionary, ontogenetic, and demographic studies.