Improving estimates of body mass in American black bears using morphometrics and non-linear models

Abstract

Measuring the body mass of American black bears (Ursus americanus) can be challenging because of their large size, and if equipment to weigh individuals is undersupplied. Our purpose was to estimate body mass of Florida black bears (U. a. floridanus) by developing models (linear and non-linear) that use morphometrics that can be reasonably easy to obtain (e.g., chest girth and body length). We compared our models with a previously published model for Florida black bears to determine whether prediction of body mass could be improved. Our models were built with current data (2012–2018; n = 532) collected across Florida, USA, by the Florida Fish and Wildlife Conservation Commission. We partitioned the data into training and test subsets using 10-fold cross-validation with 100 iterations. Model fit was assessed by comparing root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2) of observed and predicted values. Based on RMSE, MAE, and R2, our optimal regression model for predicting mass (M) of both female and male bears used both chest girth (G) and total body length (L) as predictors in the non-linear form M = aGb × Lc. Our optimal model was a better fit than the previously published model when both were applied to the full data sets from the current and previous study and to an independent data set. We applied our optimal non-linear regression models built from live bear data to morphological data collected from bear carcasses (n = 544), mainly road mortalities. We found that the live-bear models acceptably estimated mass of dead bears for both sexes. Estimating the mass of live and dead bears can expedite handling time of individuals, fill in data gaps, and provide valuable information on the Florida black bear; our approach may be applicable to American black bears range-wide.