ESTIMATES OF GENETIC AND ENVIRONMENTAL PARAMETERS FOR REPEATED MEASURES OF MILK PRODUCTION IN GOATS

Abstract The objective of this study was to compare mathematical functions in adjusting the mean and individual lactation curve of goats and to estimate genetic and environmental parameters for repeated measures of milk production. 183 lactations were used, 121 of Saanen goats and 62 of Alpine goats, of animals belonging to the Laboratory of goat and Sheep of UFPB/CCHSA. The milk control was performed at intervals of seven days. The adjustment was made to the mean curve using six mathematical functions: inverse polynomial, linear hyperbolic, incomplete gamma, quadratic logarithmic, linear and quadratic, and adjusted by interactive processes through non-linear regression. The criteria used to verify the fit quality for each function were adjusted coefficient of determination (R2a), percentages of deviation between observed and estimated total yields, percentages of typical curves, mean absolute deviation and mean square of residues. It was verified that any of the models tested can be used for mean curve estimates, but for the study of the individual curves, the incomplete gamma model should be preferred because it presents better estimates of the components of the lactation curve.

摘要：本研究旨在比较可用于拟合山羊平均泌乳曲线与个体泌乳曲线的数学函数，并针对重复测量的产奶量数据估计遗传与环境参数。本研究共使用183条泌乳记录，其中萨能山羊（Saanen goats）121条、阿尔卑斯山羊（Alpine goats）62条，实验动物均来自UFPB/CCHSA绵羊与山羊实验室。产奶量测定间隔为7天。本研究采用6种数学函数对平均泌乳曲线进行拟合，具体包括逆多项式函数、线性双曲函数、不完全伽马函数（incomplete gamma）、二次对数函数、线性函数与二次函数，并通过非线性回归的交互迭代流程完成拟合。用于验证各函数拟合优度的评价指标包括校正决定系数（adjusted coefficient of determination, R²a）、观测与估计总产奶量间的偏差百分比、典型曲线偏差百分比、平均绝对偏差以及残差均方。研究结果表明，所有供试模型均可用于平均泌乳曲线的估计，但针对个体泌乳曲线的相关研究，应优先选用不完全伽马函数模型，因其对泌乳曲线组分的估计效果更优。