LACTATION CURVES IN ANGUS, BROWN SWISS AND THEIR CROSSBRED COWS: I. COMPARISON OF PARAMETER ESTIMATION PROCEDURES

Abstract

A good prediction of cow lactation curves permits the improvement of cow/calf production efficiency, by defining management and breeding strategies. The objective of this study was to compare two procedures, linear (L) and nonlinear (N), to estimate parameters of lactation curves in cows of Angus (A), Brown Swiss (P) and their reciprocal crosses AP (A x P, and P x A). The possibility of using a weighted least squares estimation procedure was also studied. Milk production data (297 lactations) taken weekly for 180 days postpartum were recorded using the weigh-suckleweigh technique. The two procedures were evaluated for each lactation using the following equations: 1) Yt = atbe-ct; 2) Yt = ae-ct; 3) Yt = t/aect; and 4) Yt = ae(bt+ct²). For L, parameters of the four equations were estimated by least squares regression using logarithmic transformations; and for N, parameters were estimated using nonlinear least squares regression. For both, L and N, the patterns of residuals were analyzed. The residual mean squares (CMR) were used as a criterion to evaluate estimation of the curves. The CMR were analyzed, within genotype and lactation number, by a model that includes the effects of estimation procedure and cow. There were no differences (p>0.10) in CMR between L and N, except for Equation 3 where six out of 16 comparisons were significant (p