Predicting milk responses to cereal-based supplements in grazing dairy cows

Abstract

The feeding of cereal-based supplements is common in the Australian dairy industry, as it allows cows to increase intakes of total dry matter (DM) and metabolisable energy (ME), while achieving greater stocking rates, greater pasture utilisation and greater milk production per hectare than occurs when cows are fed pasture-only diets. However, for this practice to be profitable, it is important to know how much extra milk, milk protein and milk fat are produced for each kilogram DM consumed. This is difficult to determine in such a complex biological system. We combined information from 24 concentrate-feeding experiments using meta-analysis techniques, so as to develop improved prediction models of the milk, milk protein and milk fat produced when cereal-based concentrates are fed to grazing, lactating dairy cows. Model terms, consistent with biological processes, linear, quadratic and factorial, were selected according to statistical significance. The models were then tested in two ways, namely, their goodness of fit to the data, and their ability to predict novel production data from a further six, unrelated, experiments. A sensitivity analysis was also undertaken to determine how sensitive these predictions are to changes in key inputs. The predictive model for milk yield was shown to very closely reflect milk yield (kg/cow.day) measured under the experimental conditions in unrelated experiments (r = 0.96), with very little bias (Lin’s bias correction factor = 0.98) and high concordance (Lin’s concordance coefficient = 0.95). Predictions generated by multiplying predicted milk protein concentration by predicted milk yield closely matched observed milk protein yield (kg/cow.day) (r = 0.96, Lin’s bias correction factor = 0.98, Lin’s concordance coefficient = 0.95), and predictions found by multiplying predicted milk fat concentration by predicted milk yield closely matched observed milk fat yield (kg/cow.day) (r = 0.94, Lin’s bias correction factor = 0.99, Lin’s concordance coefficient = 0.93). Factors included in the new models for milk, milk protein and milk fat yield reported here have been identified previously as elements that can influence milk production. The value to the dairy industry from being able to predict profitable amounts of concentrates to feed at various stages throughout lactation is considerable. For farmers and their advisers, being able to apply these models to estimate the immediate marginal milk protein and milk fat responses to supplementary feeds should lead to more robust, efficient and profitable milk production systems.