Comparison of nonlinear mathematical models for lamb growth analysis

Abstract

With the objective to determine the best mathematical model that could demonstrate the relationship between body weight and age in crossbred sheep, and to evaluate its applicability in different farms, 9985 live weight records of 837 crossbred lambs of Texel, Ile de France, Suffolk, and Hampshire Down breeds were used. Brody, von Bertalanffy, Richards, Logistic, and Gompertz models were fitted to the weight data, which were compared based on adjusted R 2 , Syx%, chi-square, and Kolmogorov-Smirnov tests, as well as residual analysis. After selecting the model, lambs whose weights at different ages fell within, below, or above ±20% of the values predicted by the model were classified as medium, low, or high performance, respectively. To validate the model, we used a validation dataset comprising 2898 weights of lambs from 6 different farms. Based on statistical analysis and biological characteristics (sigmoid trend of weight over time, weight at birth, and standard deviation), the model that best represented the growth of the lambs was the Richards model. The percentage of lambs classified as below, within, and above the predicted values based on the model was highly similar for the original and validation data, which indicates that this model could be used in different farms. It was concluded that the Richards model best described the live weight growth of crossbred lambs and aided in classifying the lambs as high, medium, or low performance, thereby providing a complementary tool in management decision-making.