Multiphasic growth models for cattle

There are several ways of generalizing classical growth models to describe the complex nature of animal growth. One possibility is to construct a model based on a sum of several classical growth functions. In this paper, such multiphasic growth models for breeding bulls of the Czech Pied cattle based on the sum of two logistic functions are studied. The logistic function was chosen as a base for the models due to the relatively low degree of nonlinearity for the growth data. The paper describes three steps of constructing such a multiphasic growth model: in the first step a model with four unknown parameters is considered, in the second step the number of model parameters which are to be estimated is increased to five and in the third step a general model with six parameters is used. In each step, statistical properties of the considered model are checked. The residual variability of the best fitting model is on average approx. 8 times lower than the residual variability of classical Gompertz model which is often used by breeders to model cattle growth.