New equations for body mass estimation in bovids: Testing some procedures when constructing regression functions

Abstract

Body mass is an important attribute of organisms. For extinct species, body mass is usually predicted from allometric equations relating body mass with some variable(s) from the skeleton in extant species. However, even for extant species, individual associated body mass data are scarce, and mean body mass values taken from literature are commonly used when developing regressions. In this paper we evaluate the correlation between three external body measurements and the body mass of bovids with associated body mass, in order to find those variables adequate for body mass estimation in this family. The importance of using individuals with associated body weight to develop regressions was analyzed, as well as the use of Phylogenetic Generalized Least Squares (PGLS) and weighting procedures. Six equations were obtained with those variables suitable for body mass estimation (head-body length and thorax circumference), three by means of OLS and three by PGLS. Both phylogenetic and non-phylogenetic procedures render the same mean prediction error. The use of weighting procedures does not improve the predictive capacity of the equations. The percentage of prediction error (%PE) was lower than in previous works in ungulates (10–14%). The three equations were repeated with literature body mass data, and in all cases the mean %PE was around 50% higher than with associated body mass data. The multiple regression is strongly recommended for body mass estimation, when both variables are available. Body mass was estimated for three bovid species: Bos primigenius, Myotragus balearicus and Pseudoryx nghetinhensis.