Using BayesModelS to provide Bayesian- and phylogenetically-informed primate body mass predictions.

Area of Science:

• Paleontology
• Evolutionary Biology
• Biostatistics

Background:

• Accurate body mass prediction is crucial for reconstructing extinct species' ecology.
• Paleontologists often estimate body mass from fossilized dental dimensions.
• Traditional methods used frequentist statistics, with phylogenetic considerations being a recent addition.

Purpose of the Study:

• To apply and evaluate BayesModelS, a phylogenetically informed Bayesian method, for predicting body mass in euarchontan species.
• To compare the predictive accuracy of BayesModelS against ordinary least squares, phylogenetic generalized least squares, and phylogenetic independent contrasts (PICs).

Main Methods:

• Applied BayesModelS to a dataset of 49 euarchontan species.
• Utilized dental and postcranial variables for body mass prediction.
• Compared prediction accuracy with ordinary least squares, phylogenetic generalized least squares, and PICs.

Main Results:

• BayesModelS and PICs demonstrated substantially higher predictive accuracy than ordinary least squares and phylogenetic generalized least squares.
• Improved performance was most notable when using dental proxies or when proxies exhibited high phylogenetic covariance.
• BayesModelS and PIC methods produced less variance in predicted values across different body mass proxies for Notharctus tenebrosus.

Conclusions:

• Phylogenetically informed methods, particularly BayesModelS and PICs, are more effective for predicting body mass in paleontological contexts.
• These methods offer improved accuracy and reduced variance, especially with dental data.
• The study provides scripts to facilitate the application of BayesModelS and PICs for paleontological research.