ABSTRACT
When the shapes of phylogenetically divergent samples are compared, standard tests of significance (such as Hotelling’s T2) are not applicable. Instead shape differences between taxa should be assessed relative to the amount of time since they last shared a common ancestor. This can be accomplished using the same methods applied to univariate data from an evolutionary time-series. Thirteen samples of viverravid carnivorans (Viverravidae, Carnivora) from the Paleogene of the Bighorn Basin, Wyoming were compared in this way. The amount of divergence time separating samples was determined from their phylogenetic tree and their stratigraphic setting. Branch lengths were first estimated in millions of years, then converted to generations using an allometric equation relating body-mass and generation-time in extant mammals. A Log-Rate-Log-Interval (LRI) distribution was used to estimate the per-generation rate of shape change, the latter of which was used to calculate the expected shape divergence for each pair of taxa. The amount of shape divergence tended to be less than that expected, given the number of generations separating the samples. Shape change was close to a statistical definition of stasis, indicating that long-term rates of molar shape evolution are slow relative to their potential. This may be due to functional constraints on the configuration of crown features and the complexity of the developmental processes controlling their topographical relationships. Phylogenetic comparisons of shape are dependent on accurate estimates of branch lengths, that may be complicated by data and methodological considerations. In particular, many phylogenetic methods are ambiguous when it comes to determining the time interval separating two stratigraphically distinct samples.
