Use of segmented polynomials to study the lactation curve of Murrah and their crossed breeds of buffaloes under an extensive production system in the State of Sao Paulo
The objective of this study was to evaluate the use of segmented polynomials, described by the two models quadratic-quadratic (PSQQ) and that with three quadratic segments (PSQQQ), for adjustment of the mean lactation curve of Murrah females and their crosses under a system of extensive production, with once daily milking. Studies of this nature in buffaloes are practically non-existent. Milk production data from twelve herds, including 1280 lactations and 8839 test-day milk weights were analyzed. Test days were spaced at 30-day intervals, during the period from 1995 to 2001. Lactations included in the analysis were based on a minimum of four and a maximum of nine test days. For the PSQQ model, the mean curve reached its “Knot” (junction point between the segments) in the second month of lactation, while for the PSQQ model the “Knots” were estimated to be in the second and fourth months. The parameters of each function were estimated by interactive procedures, using PROC REG of SAS. Criteria used to verify the goodness of fit for each function were: adjusted coefficient of determination (R2A); lack of fit test (LOF); and graphic distribution of residuals and of residual lag. The two models closely described variations in milk production throughout the lactation with the lack of fit tests not significant. The coefficients of determination were practically identical for both models under study (R2A = 91.3%). (Text in Portuguese).