A NEW APPROACH TO STUDYING PHENOTYPIC EVOLUTION IN THE FOSSIL RECORD

How well can paleontological data resolve patterns of phenotypic evolution, given sampling effects and biological and geological confounding factors? I approach this question with a new method for inferring a model of the temporal pattern of phenotypic change in a fossil species, based on morphological data and associated geological information on time and environment (e.g. age model, facies, paleobathymetry). The goal is to characterize a broad suite of models that fit the data at an acceptable level. A key element is thus to define what it means for a given model to fit the observations, which depends on assessing the uncertainty in both data and model. Uncertainties in chronostratigraphic, depositional, paleoecological and morphometric data, together with knowledge on the parameters and their range of values, are combined with the data-model misfit into a probability density reflecting both data fit and model plausibility. Information can then be retrieved on how well the paleontological data are able to resolve parameters of phenotypic evolution, taking into account the stratigraphic and paleoecological context. The method is illustrated using synthetic data.