On the Description of Equine Somatic Growth Using Nonlinear Functions.

A better knowledge of the growth of horses can facilitate sounder intervention in their health control, nutrition and management. This study compared five growth functions with either diminishing returns (monomolecular and Michaelis-Menten) or asymptotic sigmoidal (Gompertz, Richards and Lopez) behavior for describing the growth curves of horses, based on the relationship between a given somatic growth trait (body weight, withers height, or body weight to withers height ratio) and age. Data used in this study were those available in the literature for quarter horse males and thoroughbred foals and fillies from birth to maturity. Estimates of the model parameters were obtained using nonlinear regression procedures and the Marquardt-Levenberg algorithm. Functions were compared using goodness-of-fit statistics such as adjusted coefficient of determination and root mean square error. The results obtained with the Richards, Lopez and Gompertz equations suggest that curves are not sigmoidal, and that the diminishing returns models are adequate for describing the growth curves in horses.