Shape-adjusted bone mineral density measurements in baboons: other factors explain primate skeletal element representation at Swartkrans

Abstract

Recently, Lam et al. (J. Archaeol. Sci. (2003) in press) reviewed zooarchaeological studies of bulk bone mineral density (bulk BMD), demonstrating the degree to which several studies accurately assessed volumes at regions of interest (ROIs) on bones. It has been suggested that bulk BMD measurements obtained with a clinical computed tomography (CT) technology are preferable to those obtained with a photon densitometry (PD) technology. The basis for this suggestion comes from the unavoidable fact that (non-destructive) PD-based studies have been less precise in estimating area bounded by the periosteal envelope and entirely unable to subtract area bounded by the endosteal envelope. Here we present a new technique for shape-adjusting ROI volumes in the context of re-evaluating baboon bulk BMD data that were published in an earlier study and incorporated no shape-adjustment (J. Archaeol. Sci. 29 (2002) 883). We suggest that this new shape-adjustment method, similar to that of Stahl (J. Archaeol. Sci. 26 (1999) 1347), improves accuracy of volumetric density measurements by producing ROI volumes that are quantified similarly to CT generated ROI volumes (i.e., using a highly automated direct measurement). Correlation coefficients are higher using the shape-adjusted data probably because ROI volumes more accurately reflect the amount of bone by minimizing external air in bulk BMD calculations. While revised bulk BMD values are consistently higher than values without shape-adjustment, once shape-adjustment is incorporated, the rank order of revised baboon bulk BMD values is correlated significantly with the rank order of bovid (i.e., sheep) ROIs. Thus, intertaxonomic differences in the patterns of baboons and similarly sized bovid skeletal element representation are unlikely to be attributable to density-mediated destructive processes, since these processes would be expected to remove skeletal parts of each taxon similarly. For baboon ROIs, rank ordered bulk BMD was not correlated significantly with any of the rank ordered skeletal part frequencies in primate fossil assemblages from Swartkrans Cave (South Africa) investigated previously. Thus, intrataxonomic patterns of primate skeletal part frequency in these fossil assemblages also are not attributable to a density-mediated destructive process. We recommend that bovid bulk BMD values not be used to assess density-mediated processes in a primate assemblage, or vice versa, since correlation coefficients were not 1.0.