Department of Evolutionary Anthropology, Duke University, Durham, North Carolina.
Department of Molecular Biomedical Sciences, College of Veterinary Medicine, North Carolina State University, Raleigh, North Carolina.
Am J Primatol. 2020 Feb;82(2):e23088. doi: 10.1002/ajp.23088. Epub 2020 Jan 21.
A primate's body mass covaries with numerous ecological, physiological, and behavioral characteristics. This versatility and potential to provide insight into an animal's life has made body mass prediction a frequent and important objective in paleoanthropology. In hominin paleontology, the most commonly employed body mass prediction equations (BMPEs) are "mechanical" and "morphometric": uni- or multivariate linear regressions incorporating dimensions of load-bearing skeletal elements and stature and living bi-iliac breadth as predictor variables, respectively. The precision and accuracy of BMPEs are contingent on multiple factors, however, one of the most notable and pervasive potential sources of error is extrapolation beyond the limits of the reference sample. In this study, we use a test sample requiring extrapolation-56 bonobos (Pan paniscus) from the Lola ya Bonobo sanctuary in Kinshasa, Democratic Republic of the Congo-to evaluate the predictive accuracy of human-based morphometric BMPEs. We first assess systemic differences in stature and bi-iliac breadth between humans and bonobos. Due to significant differences in the scaling relationships of body mass and stature between bonobos and humans, we use panel regression to generate a novel BMPE based on living bi-iliac breadth. We then compare the predictive accuracy of two previously published morphometric equations with the novel equation and find that the novel equation predicts bonobo body mass most accurately overall (41 of 56 bonobos predicted within 20% of their observed body mass). The novel BMPE is particularly accurate between 25 and 45 kg. Given differences in limb proportions, pelvic morphology, and body tissue composition between the human reference and bonobo test samples, we find these results promising and evaluate the novel BMPE's potential application to fossil hominins.
灵长类动物的体重与许多生态、生理和行为特征有关。这种多功能性和提供动物生活洞察力的潜力，使得体重预测成为古人类学中一个常见且重要的目标。在人类古生物学中，最常用的体重预测方程（BMPEs）是“力学”和“形态计量学”：分别使用单一或多元线性回归，将承重骨骼元素的维度以及身高和活体双髂宽作为预测变量。然而，BMPEs 的精度和准确性取决于多个因素，其中最显著和普遍的潜在误差源之一是超出参考样本范围的外推。在这项研究中，我们使用需要外推的测试样本-来自刚果民主共和国金沙萨的 Lola ya Bonobo 保护区的 56 只倭黑猩猩（Pan paniscus）-来评估基于人类形态计量学的 BMPEs 的预测准确性。我们首先评估人类和倭黑猩猩之间身高和双髂宽的系统差异。由于倭黑猩猩和人类的体重和身高之间的比例关系存在显著差异，我们使用面板回归生成一种基于活体双髂宽的新 BMPE。然后，我们比较了两种先前发表的形态计量学方程与新方程的预测准确性，发现新方程总体上最准确地预测了倭黑猩猩的体重（56 只倭黑猩猩中有 41 只预测值在其实际体重的 20%以内）。新的 BMPE 在 25 到 45 公斤之间特别准确。鉴于人类参考样本和倭黑猩猩测试样本在四肢比例、骨盆形态和身体组织组成方面的差异，我们认为这些结果很有希望，并评估了新的 BMPE 对化石人类的潜在应用。
