Morphometric analysis of old world Talpidae (Mammalia, insectivora) using partial - warps scores

Abstract.—We illustrate one approach to bridging the gap between the statistical methodology of multivariate morphometrics and the graphically oriented methods of geometric morphometrics for landmark data. Partial-warp scores (elements of the weight matrix from relative-warp analysis) are used as variables to describe nonaffine shape variation. The results of various multivariate analyses can then be visualized by reconstruction of images as weighted linear combinations of thin-plate splines. The method was applied to data on seven species of Old World moles of the family Talpidae. There was no evidence for sexual dimorphism or any interaction between sex differences and species differences. There were large statistically significant differences among species with respect to nonaffine shape differences (local deformations). Small but statistically significant differences were found among species with respect to the uniform component of shape variation. [Landmark data; geometric morphometrics; thin-plate spline; relative warps; Talpidae; Insectivora.] Multivariate statistical methods can be combined with the graphically oriented methods of geometric morphometrics t