A resampling approach and implications for estimating the phalangeal index from unassociated hand bones in fossil primates.

Primate fossil assemblages often have metacarpals and phalanges from which functional/behavioral interpretations may be inferred. For example, intrinsic hand proportions can indicate hand function and substrate use. But, estimates of intrinsic hand proportions from unassociated hand elements can be imperfect due to digit misattribution. Although isolated metacarpals can be identified to a specific digit, phalanges are difficult to assign to a specific ray. We used a resampling approach to evaluate how estimates of intrinsic hand proportions are affected by such uncertainty. First, the phalangeal index-intermediate phalanx length plus proximal phalanx length divided by metacarpal length-for the third digit was calculated for associated specimens of terrestrial, semiterrestrial, and arboreal taxa. We then used resampling procedures to generate distributions of "composite digits" based on resampled ratios in which phalanges from the second, fourth, and fifth rays, and from different individuals, were chosen randomly. Results confirm that the phalangeal index for associated third digits significantly discriminates groups. We also found that resampled ratios had significantly lower means, indicating that using composite digits is prone to systematic underestimation. Resampled ratios also generated distributions with greater variance around the means that obscured distinctions between groups, although significant differences between the most arboreal and terrestrial taxa are maintained. We conclude that using unassociated phalanges to calculate a phalangeal index is prone to sampling bias. Nevertheless, a resampling approach has the potential to inform estimates of hand proportions for fossil taxa, provided that the comparative sample is constrained to mimic the fossil composition.