Department of Geology and Geophysics, Yale University, New Haven, Connecticut, United States of America.
PLoS One. 2013 Nov 29;8(11):e82000. doi: 10.1371/journal.pone.0082000. eCollection 2013.
Scaling relationships between skeletal dimensions and body mass in extant birds are often used to estimate body mass in fossil crown-group birds, as well as in stem-group avialans. However, useful statistical measurements for constraining the precision and accuracy of fossil mass estimates are rarely provided, which prevents the quantification of robust upper and lower bound body mass estimates for fossils. Here, we generate thirteen body mass correlations and associated measures of statistical robustness using a sample of 863 extant flying birds. By providing robust body mass regressions with upper- and lower-bound prediction intervals for individual skeletal elements, we address the longstanding problem of body mass estimation for highly fragmentary fossil birds. We demonstrate that the most precise proxy for estimating body mass in the overall dataset, measured both as coefficient determination of ordinary least squares regression and percent prediction error, is the maximum diameter of the coracoid's humeral articulation facet (the glenoid). We further demonstrate that this result is consistent among the majority of investigated avian orders (10 out of 18). As a result, we suggest that, in the majority of cases, this proxy may provide the most accurate estimates of body mass for volant fossil birds. Additionally, by presenting statistical measurements of body mass prediction error for thirteen different body mass regressions, this study provides a much-needed quantitative framework for the accurate estimation of body mass and associated ecological correlates in fossil birds. The application of these regressions will enhance the precision and robustness of many mass-based inferences in future paleornithological studies.
现生鸟类的骨骼尺寸与体重之间的比例关系常用于估算化石鸟类以及祖鸟类的体重。然而，很少有用于限制化石质量估计精度和准确性的有用统计测量值，这使得无法对化石的可靠上下限体重估计进行量化。在这里，我们使用 863 种现生飞鸟样本生成了 13 种体重相关性及其统计稳健性的度量。通过为单个骨骼元素提供具有上下限预测区间的稳健体重回归，我们解决了高度破碎化的化石鸟类体重估计的长期问题。我们证明，在整个数据集的测量中，无论是普通最小二乘回归的系数确定还是预测误差的百分比，最精确的用于估计体重的代理变量是肩甲骨肱骨关节面的最大直径（关节盂）。我们进一步证明，这一结果在大多数调查的鸟类目中都是一致的（18 个目中有 10 个）。因此，我们建议，在大多数情况下，这个代理变量可能为飞行化石鸟类提供最准确的体重估计。此外，通过为 13 种不同的体重回归提供体重预测误差的统计测量值，本研究为化石鸟类中准确估计体重和相关生态特征提供了急需的定量框架。这些回归的应用将增强未来古鸟类学研究中许多基于质量的推断的精度和稳健性。
