| Summary: | In the late 1970s Derek Roberts presented the first systematic analyses of the impact of
climate on human morphological variability. One of his key findings was that humans
follow Allen's Rule, which holds that there is a positive correlation between peripheral
body part size and temperature in homeothermic species.
Roberts' conclusions regarding the applicability of Allen's Rule to humans have been
widely accepted. However, three features of his analyses are potentially problematic.
First, he used a sample that is strongly biased towards warm climates. Second, he
maximized sample size of each limb segment at the expense of among-segment
comparability. Third, he ignored the confounding effects of phylogeny.
In this study the reliability of Roberts' conclusions were evaluated in relation to the
aforementioned problems. In the first set of analyses, Roberts' analyses were replicated to
ensure the dataset is comparable to his. In the second, the impact of sampling was
investigated by examining the range of variation present especially among the warm
climate populations. In the third, the impact of over-representation of warm temperature
populations was evaluated by stratifying random samples by temperature. In the fourth,
stratified analyses were conducted using populations with data for all limb segments. In
the last set of analyses, the influence of phylogeny was investigated through phylogenetically-controlled correlation analyses.
The first set of analyses was consistent with Roberts' conclusions indicating that the dataset is appropriate. In the second, the slope of the regression line was variable such that both positive and negative correlations were obtained for many segments, which emphasizes the impact of sample selection. The third set supported the idea that overrepresentation of warm climates is problematic as only two segments yielded statistically significant correlations. In the fourth, the pattern of limb proportion variation did not support previous analyses, suggesting that Roberts' decision to maximize sample size at the expense of among-segment comparability is problematic. The fifth set of analyses indicated that phylogeny has a significant impact on the results of the correlation analyses.
Together, the results of the analyses reported here cast doubt on the claim that humans
follow Allen's Rule. === Arts, Faculty of === Anthropology, Department of === Graduate
|
|---|
