Continuous morphological data in phylogenetic analyses: An example from the otters (Carnivora: Lutrinae)

In this study, otter (Carnivora: Lutrinae) postcranial osteological measurements were used in a phylogenetic analysis. Previous systematic analyses of the otters (based primarily on discrete cranial characters) yield conflicting results leading to uncertainty regarding the evolutionary history of this group. Continuous postcranial trait data were analyzed using several approaches and techniques. In Chapter 2, continuous data from postcranial morphological measurements were converted into discretely coded characters using several methods. Two types of gap coding (Gap A and Gap B) and two segment coding approaches (12 and 6 segments) were applied and analyzed using a maximum parsimony reconstruction program. The resulting phylogenies appeared to be size related, grouping otters of similar size together. To remove the size related component, the continuous trait data were entered into a maximum likelihood analysis program without coding in Chapter 3. Regression residuals produced from a regression of body size and character measurements were analyzed using the same program. Results from this analysis were mixed. Uncoded data also yielded size related groupings, whereas the regression residuals appeared to have little biological relevance. Results from all the analyses were combined in Chapter 4 to produce a consensus phylogeny. To construct a consensus, data combining, tree combining, and consensus methods were employed. The resulting consensus tree essentially mirrored the majority of results from the source phylogenies. Overall, although the resulting phylogenies were robust and well supported by bootstrap analyses, the essentially size related input data (except regression residuals) resulted in size related phylogenies, and this approach should be undertaken with caution.