A comparison of catarrhine genetic distances with pelvic and cranial morphology: implications for determining hominin phylogeny.

Inferring the evolutionary history of the hominins is necessarily reliant on comparative analyses of fossilized skeletal anatomy. However, the reliability of different primate skeletal regions for recovering phylogenetic relationships is currently poorly understood. Historically, postcranial variation has largely been conceived of as reflecting locomotory and postural adaptation. The shape of the os coxae is central to such discussions given the divergent morphology displayed by the bipedal hominin pelvis relative to other primate taxa. While previous cladistic studies have suggested that postcranial and cranial datasets do not differ in terms of their propensity for homoplasy, methodological issues such as the numbers of characters and their quantification make it difficult to evaluate these findings. Here, we circumvent these problems by constructing morphological distance matrices based on cranial, mandibular and os coxae three-dimensional shape. Statistical comparisons of these morphological distance matrices against a single genetic distance matrix for 11 catarrhine taxa show that cranial and os coxae shape reflect genetic relationships better than the mandible when humans are included, and that the cranium and os coxae do not differ statistically in terms of their genetic correlations. When humans were excluded from the analyses, all three anatomical regions were equally strongly correlated with genetic distance. Moreover, a second analysis focusing solely on os coxae variation of 16 taxa demonstrated that os coxae shape correctly recovers catarrhine taxonomic relationships at the sub-family level, even when humans are included. Taken together, our results suggest that there is no a priori reason to favor cranial shape data over os coxae morphology when inferring the genetic relationships of extant or extinct primate taxa. Morphological similarities between humans and other primates differ depending on the skeletal element, suggesting that combining skeletal elements into a single analysis may provide more accurate reconstructions of genetic relationships.