Use of Lactation Curves for Analysis of Milk Production Data

Abstract

Nonlinear equations were compared with categorical analysis to account for DIM effects on milk production. Five different models for lactation curves were evaluated. Derived from a multiphasic lactation curve, the selected lactation curve appeared to result in random residuals and performed more consistently than the multiphasic curve. Residuals from the fitting of lactation curves were then used for split-plot analysis (continuous model) to estimate treatment effects. Statistical performance of this model was compared with split-plot analysis based on a discrete model with regularly spaced intervals to account for DIM effects (discrete model). The fitting of lactation curves accounted for herd, lactation number, and interaction effects of herd and lactation number and accounted for 34.1 and 44.3% of variance among cows for primiparous and multiparous cows, respectively. The continuous model detected interactions of genetic and management factors with treatment of multiparous cows that were not detected by the discrete model. No statistically significant differences were detected between the two modeling approaches. The continuous model appeared to violate fewer assumptions regarding data distribution than did the discrete model, which reduced the risk of introducing bias during the estimation of treatment effects. The continuous model seemed to be more sensitive to subtle interactions of treatment and other factors.