MEASUREMENT DEPENDENCY, SCALE DEPENDENCY, AND THE EFFECT OF PHYLOGENETIC AUTOCORRELATION IN GEOMETRIC APPROACHES TO MORPHOLOGICAL DISPARITY ANALYSES

Over the past decade morphological disparity studies have become increasingly popular in paleontology. A variety of approaches have been taken to such studies ranging from the formally morphometric to the semi-quantitative. While comparisons between quantitative and semi-qualitative approaches have been made with respect to their ability to produce broadly consistent results, the issues of measurement dependency and scale dependency have yet to be formally assessed in this context. Moreover, the problem of phylogenetic autocorrelation in the data used to estimate patterns of morphological disparity has been largely ignored. In order to investigate these topics traditional 2D and 3D geometric morphometric approaches were compared and contrasted with a new morphometric approach—eigensurface analysis— that supports direct comparison of 3D surfaces. Preliminary analysis of synoptic morphometric datasets for bivalves, foraminifera, trilobites, and plants suggests disparity measures are strongly influenced by the manner in which measurements are taken (e.g., 2D vs. 3D data) and the effective spatial scale represented by those measurements. Broadly speaking, inter-taxon disparities can be made to appear, grow larger, diminish, and/or disappear depending on how one chooses to measure the form(s) under consideration. While none of these results are incorrect characterizations of the morphologies, all are incomplete if interpreted in isolation. Even more importantly, failure to correct morphometric disparity data for phylogenetic autocorrelation can yield highly misleading estimates both of gross disparity and the patterning of clade-specific disparities over time. These results suggest some of the primary generalizations of the morphological disparity research programme may need to be reassessed, but also provide a much richer and more nuanced analytic vocabulary for undertaking those reassessments.