A 3D Geometric Morphometric Investigation of Relatedness in the Modern Human, Chimpanzee, and Homo naledi Postcranial Skeleton

Studies of skeletal variation form the basis of our understanding of our species’ history and diversity. This most commonly takes the form of comparative, between- species studies aimed at reconstructing phylogenetic relationships. However, studies of within-species variation can provide insights into relatedness at smaller scales, which can shed light on important microevolutionary processes and be used to identify closely related individuals in the absence of DNA. This dissertation assesses the relationship between genetic and skeletal variation within groups of related and unrelated Homo sapiens and Pan troglodytes, with the aim of better understanding the population sampled by the seemingly homogenous Homo naledi assemblage from the Dinaledi Chamber in Rising Star Cave, South Africa. The hypotheses tested were: 1) closely related individuals exhibit less skeletal variation compared to unrelated individuals, and 2) the Dinaledi assemblage exhibits less morphological variation than H. sapiens and P. troglodytes at the species level, more closely resembling a single population.

Skeletal variation within a subset of H. sapiens from the crypt of Christ Church in Spitalfields, London (17-19th centuries A.D) with documented genealogies and the Gombe population of P. t. schweinfurthii was compared to species-wide baselines and the H. naledi assemblage. The data consisted of 3D models of 556 postcranial elements (first metacarpal (Mc1), proximal femur, talus, calcaneus, and navicular) from 187 individuals that were generated from surface scanners or downloaded from MorphoSource.

This dissertation employed a relatively new 3D geometric morphometric workflow that captures the entire shape of an element through the use of an automated landmarking program and feature-aware registration process. Two hundred pseudolandmarks were automatically and optimally placed on each element. Following alignment, multiple univariate and multivariate statistical analyses were used to quantify shape variation within and between the three species, including distributions of Euclidean distances, Procrustes distances to the mean shapes, Principal Components Analyses (PCA), Between-Group PCA, and Discriminant Function Analyses. The coefficient of relationship was used to represent genetic distance between known genetic relatives within modern humans.

The results of the within-species analyses of skeletal variation support Hypothesis 1 in both modern humans and chimpanzees, though the signal of relatedness is differentially expressed within and across elements. In modern humans, the calcaneus can be used to distinguish known close relatives from distantly related and unrelated individuals. The navicular and femur were also found to be relatively good indicators of relatedness. Within chimpanzees, the talus is the most effective at distinguishing the Gombe population from the species-wide chimpanzee sample, followed by the calcaneus and femur. Within H. naledi, the talus varied the least while the navicular varied the most, though the high levels of variation found in the navicular and Mc1 are likely due to the state of preservation of these elements. The results of the interspecies analyses are more ambiguous. When considering the best preserved elements, the H. naledi talus varies the least within the three species, while the femur varies more than those of either H. sapiens or P. troglodytes at the species level. Hypothesis 2 is thus supported for the talus while rejected in the femur, suggesting that it cannot be rejected as a whole and that the patterns of homogeneity previously observed within H. naledi are more nuanced than previously recognized.