Prediction of growth curves in Nigerian sheep using non-linear mathematical models

Abstract

Information on growth pattern is required for selection and management plans in sheep production. In this study, weights of 156 sheep comprising 36 Balami, 37 Uda, 41 Yankasa and 42 WAD were measured monthly from birth to the 12th month and used to model the relationship between weigh and age data by fitting the Logistic, Gompertz, Von Bertalanffy, Verhulst and Brody mathematical growth models. The goodness of fits of the models were assessed using the coefficient of determination, R2, the mean square error, MSE and Akaike's information criterion, AIC. Over all, the growth model parameter A ranged from 31.9 (Verhulst) to 115 (Brody, which was consistently higher than in other growth models). Parameter B was highest in Verhulst (7.60) and lowest in Logistics (-0.277). Parameter K was consistently lowest (-0.005) in the Logistics model and generally highest in the Verhulst model. All the models had high R2 ranging from 0.81 to 0.99, while MSE ranged from 0.20 to 14.5. As in other measures of fitness, similar order of ranking of models was exhibited by the AIC which values ranged from 42.1582 to 16.8644. All the models presented similar predictions of body weight with generally low residuals between them and the observed weights at the same stages of growth. This indicated that all the mathematical models compared and fitted the body weight data well and could be used to predict sheep weights, but Brody and von Bertalanffy were best.