| Abstract: |
Researchers have long appreciated the significant relationship between body size and an animal's overall adaptive strategy and life history. However, much more emphasis has been placed on interpreting body size than on the actual calculation of it. One measure that is especially important for human evolutionary studies is stature. Despite a long history, stature estimation remains plagued by two methodological problems: (1) the choice of statistical estimator, and (2) the choice of reference population upon which to derive the parameters. This work addresses both of these problems. Three reference samples of known stature with maximum humerus and femur lengths recorded are used in this study: a large (N = 2,209) modern human sample from North America, a smaller sample of modern African pygmies (N = 19) from West Africa and a sample of African great apes (N = 85). First, the performance of five commonly used regression models is examined: classical calibration, inverse calibration, major axis, reduced major axis and the zero intercept regression (ratio) model. These five techniques are rigorously tested in both univariate and multivariate categories for their ability to predict stature in situations where there is both extrapolation beyond the mean of the reference sample and differences in limb proportions. The performance of the five estimators is measured using both the root mean squared error and bias. Stature is also estimated for two well-known fossil hominids, A.L. 288-1 ('Lucy') and KNM-WT 15000 ('The Boy') using parameters derived from the reference samples. Results indicate that the choice of regression estimator remains unclear in situations where the allometry differs between the reference population and the sample population, which is often the case with fossil remains. Classical calibration performed best in almost all multivariate analyses of stature. However, results varied in univariate situations. Results of the fossil stature estimations indicate that the A.L. 288-1 fossil hominid has unique limb proportions and was best estimated by the human pygmy reference sample with univariate data and the chimpanzee sample with multivariate data. However, since the limb proportions of WT 15000 are well within the modern human range, the large modern human sample was the best reference sample.
|
