Comparison of the German and the Dutch protein evaluation system in dairy nutrition when applied to different rations with graded protein levels.

The Dutch (DVE) and the German (nXP) protein evaluation system were compared using data from four specially designed dairy feeding trials. The criterion for comparison was the divergence between predicted and observed milk protein yield with the respective protein evaluation system. Different forages were fed in each series of experiments, namely red clover/meadow grass, meadow grass, grass silage and maize silage, respectively. To achieve a diversified database, protein was supplied at 2 or 3 different levels in each experiment. Across all measured data, the two protein evaluation systems resulted sometimes in considerable differences between predicted and observed protein yield. The observed yield was overestimated by 223 g per cow and day with the DVE system and by 144 g per cow and day with the nXP system. The magnitude of the bias varied depending on the type of ration fed. In experiments with fresh forage the difference between predicted and observed milk protein yield was far greater with the DVE system than that calculated with the nXP system (difference 322 g vs. 162 g), whereas after feeding grass silage the differences between the two protein evaluation systems were small (194 g vs. 182 g). In experiments with maize silage the predicted value was close to the observed value both with the DVE and the nXP system (difference 48 g vs. 80 g).