VALIDATION OF PROTEIN EVALUATION SYSTEMS BY MEANS OF MILK-PRODUCTION EXPERIMENTS WITH DAIRY-COWS

Protein evaluation systems (CP, DCP, PDI, AAT, AP, MP, AAS and DVE) we re validated using data froM 15 production experiments with dairy cows , carried out in The Netherlands. Only treatments that were deficient in protein according to at least one system were selected. Average pro duction was 31.2 kg fat and protein corrected milk d-1 and 989 g milk protein d-1. The observed milk protein production was compared with mi lk protein production predicted from the protein supply and requiremen ts in each system. The difference between observed and predicted milk protein production expressed as the absolute and relative prediction e rror was smallest for the DVE-system (-2 g d-1; respectively 5.7%) and increased in the order CP (-22 g d-1; 6.7%), PDI (-19 g d-1; 7.8%), D CP (-44 g d-1; 8.8%), AP (-37 g d-1; 9.3%), AAS (100 g d-1; 11.7%), AA T (112 g d-1; 13.4%) and MP-system (204 g d-1; 22.9%). Predictions can be improved when a variable efficiency of milk protein production is used. In the DVE-system the observed efficiency decreased with increas ing protein to energy ratio in the diet and milk production level. It was concluded that under Dutch conditions the prediction of milk prote in production decreased in the order DVE, CP, PDI, DCP, AP, AAS, AAT a nd MP-system.