Análise de curvas de lactação utilizando modelos lineares e não lineares

Abstract

Milk is one of the main foods consumed in the world, due to its nutritional and organoleptic attributes. Its composition changes according to the breed, food, gestation time and some other factors characteristic of the lactating cow, and is still influenced by the place of production, such as climate and pasture. Milk production can be represented graphically during lactation of a cow, this representation is called a lactation curve. In this sense, the objective of this work was to classify the main models used to describe the lactation curves in linear and non-linear, indicating the most appropriate model of each group, besides evaluating the non-linearity of the non-linear models in the description of the lactation curves of the Girolando race. Data from 165 lactations of 89 3/4 Holstein + 1/4 Gir cows reared in a pasture milk production system with supplementation were used, the average production was calculated every 20 days up to 310 days of lactation. Seventeen models of lactation curves available in the literature were used. The estimation of the parameters of the models was made using the least squares method, whereas the non-linear ones required the iterative Gauss-Newton method, also called linearization method, and the initial values of the iterative methods were defined based on the literature. on lactation curves of other breeds. Among the main results, it can be highlighted that the best linear model to describe the lactation curves of the Girolando breed was the Cobuci et al. (2000), the most suitable non-linear model to describe the Girolando lactation curve was Wood’s model (1967). The Cobuci et al. (2000) is an interesting alternative for modeling growth curves. And the Wood model (1967), a non-linear model, has parameters with practical interpretation, making it possible to estimate important aspects of the curve.