The Use of Regression Models to Predict Spatial Patterns of Cattle Behavior

Abstract

Predictive models of cattle behavior were developed by applying multiple regression analysis to a body of behavior observations. Seven independent variables were required to describe the spatial patterns of three modes of behavior on an annual basis. Coefficients of determination were 0.50, 0.34, 0.25, and 0.20 for grazing + travel, summer resting, winter resting, and bedding, respectively. Spatial patterns of each mode were predicted for a 125-ha pasture, upon which a separate set of behavior observations had been made. Comparisons of observed and predicted patterns varied from a close fit for grazing to marginal for resting. Validation using the spatial pattern of fecal deposition, however, yielded a close fit. It was concluded that multiple regression models can be useful in predicting spatial patterns of livestock behavior and may have unexploited potential as both management and research tools.