Assessment of the Canadian model predicting daily milk yield and milk fat percentage using single-milking dairy herd improvement samples

The use of adjustment factors with alternate morning (AM) and evening (PM) milk tests to predict daily milk yield and fat percentage from single milking could lead to erroneous daily data. The aims of this study were to evaluate the relationship between predicted daily milk yield or milk fat percentage, calculated using single milking samples and Canadian adjustment factors, and the actual daily milk yield or milk fat percentage, as well as to explore feeding and management variables, that could improve daily predictions. A total of 4277 Holstein cows in 100 dairy herds were enrolled. Separate PM and AM milk samples were collected for each cow using in-line milk meters. Daily milk yield and milk fat percentage predictions were computed from single-milking samples using adjustment factors taking into account milking interval and milking time. Concordance correlation coefficients between actual daily milk yields and daily milk yield predictions from PM (0.970) and AM (0.974) milkings were higher than those between actual daily milk fat percentages and daily milk fat percentage predictions from PM (0.897) and AM (0.917) milkings. There were only slight prediction improvements when days in milk, parity, and some feeding management variables were entered in models aiming to explain residuals.