Modelling the effect of stocking rate on the lactation profiles of grazing Holstein-Friesian dairy cows using cubic splines

Abstract

The primary purpose of modeling lactation is to predict the dairy cow’s average daily milk yield with minimal error, after adjusting for various environmental factors. While empirical and mechanistic models have been commonly utilised to model the lactation profile of dairy cows, random regression procedures of Legendre polynomials and cubic splines are increasingly being used. The objectives of this study were to compare the lactation profiles and performance of dairy cows on dryland versus irrigated pastures at different stocking rates with or without grain supplementation using cubic splines model. Cubic splines adequately modelled the bi-weekly milk yield data with low residuals and uncorrelated coefficients attributable to the great flexibility of the model. Without supplementation, mean milk yield did not differ, but was slightly higher in cows grazing at 2.5-3.5 cows/ha stocking rate (SR) compared to cows stocked below at 2.0 c/ha and above at 4.0 c/ha (Figure 1). Irrespective of SR, cows on irrigated pasture had higher peaks except those stocked at 4.0 c/ha. Pasture allocation significantly (p<0.05) increased the rise to peak milk yield in cows stocked at lower stocking rates (2.4-2.5 c/ha) compared to those on 2.8-3.5 c/ha but the latter were more persistent and had higher predicted total milk yields. The results demonstrated the accuracy of cubic splines in modeling lactation and that higher stocking rates can improve the efficiency of pasture utilisation when coupled with adequate grain supplementation.