Whole-bone shape of hominoid manual proximal phalanges

Functional morphologists have long noted that skeletal adaptations in primate phalanges reflect locomotor behavior. While most studies have successfully used two-dimensional measurements to quantify general features of phalanx shape, a whole-bone three-dimensional analysis may better capture more subtle aspects of phalanx morphology that have not been quantifiable but are functionally meaningful. Here, we compare linear measurement (LM) and weighted spherical harmonic/sliding semilandmark (SPHARM-sliding) analyses of the manual third proximal phalanx (PP3) in extant hominoids (Homo, Pan, Gorilla, Pongo, Symphalangus, Hylobates; n = 292) and specimens attributed to Australopithecus afarensis (n = 2) and Homo neanderthalensis (n = 2). Morphological variation was summarized using principal component (PC) analysis. Differences between extant taxa were tested for using non-parametric MANOVAs (LM) and Procrustes distance resampling (SPHARM-sliding). Linear discriminant analyses (LDA) were performed using PC scores to assess whether the SPHARM-sliding or LM analysis better predicts group memberships of extant and fossil specimens. In both analyses, PC1 separates taxa along a locomotor gradient, and all extant genera are significantly different from one another (p ≤ 0.01) aside from Pongo versus Symphalangus in the LM analysis (p = 0.053). Only the SPHARM-sliding analysis found significant differences between taxa within each genus (p ≤ 0.04), and differences were even significant among Gorilla subspecies (p < 0.001). LDAs indicated that accuracy, separation effectiveness, and confidence were greater for the SPHARM-sliding analysis in predicting group membership among extant specimens, as well as fossil memberships to an extant group. Overall, results demonstrate that whole-bone, high-density landmark analyses can highlight nuanced features of PP3 morphology and may serve better for making inferences about fossils.