The 82nd Annual Meeting of the American Association of Physical Anthropologists (2013)

What is the role of geometric morphometrics in testing functional hypotheses? A case study using 3D pelvic shape


Human Evolutionary Biology, Harvard University

Thursday Morning, 301D Add to calendar

The pelvic girdle is a complex structure with critical roles in locomotion and obstetrics, and efforts to model these mechanical effects on its shape remain difficult because it is not amenable to structural simplification. While geometric morphometric (GM) analysis yields important information regarding changes in shape among groups, its utility in testing biomechanical hypotheses has been questioned. This study evaluates the contributions of both univariate and GM methods to testing hypotheses relating pelvic form to locomotor function.

Three-dimensional landmarks were collected on a phylogenetically-broad pelvic sample of 787 individuals from 40 primate taxa. Linear interlandmark distances were calculated to facilitate testing of biomechanical hypotheses, and a principal components (PC) analysis was performed on Procrustes coordinates. Both linear dimensions and PC scores were subjected to phylogenetic ANOVA by locomotor group. Many of the hypotheses relating linear dimensions to locomotor loading mechanics were rejected, but lower ilium cross-sectional area varies as predicted among groups (F=17.28, p<0.001), demonstrating an adaptive signal. Shape analyses support the univariate results, with significant differences evident along the first five PCs separating vertical clingers and leapers from arboreal and terrestrial quadrupeds (all p<0.05). While both analytical approaches suggest that ilium dimensions differ among locomotor groups, the GM analysis also suggests that ischiopubic shape differentiates groups. Although GM provides additional quantitative results beyond the univariate analyses, it cannot replace all tests of specific, directional hypotheses of pelvic biomechanics and adaptation; it would be prudent to use both approaches concurrently until more targeted GM hypothesis-testing methods are developed.

comments powered by Disqus