Effects of silage protein degradability and fermentation acids on metabolizable protein concentration: a meta-analysis of dairy cow production experiments.

A meta-analysis was conducted using data from dairy cow production studies to evaluate silage metabolizable protein (MP) concentrations. The data consisted of 397 treatment means in 130 comparisons, in which the effects of silage factors (e.g., date of harvest, wilting, silage additives) were investigated. Within a comparison, a fixed amount of the same concentrate was fed. A prerequisite of data to be included in the analysis was that silage dry matter (DM), crude protein (CP), ammonia N, lactic acid (LA), and total acid (TA) concentrations and digestibility were determined. A smaller data set (n = 248) comprised studies in which silage water-soluble N concentration was also analyzed. The supply of MP was estimated as amino acids absorbed from the small intestine using a model with constant values for ruminal effective protein degradability (EPD) and intestinal digestibility of rumen undegraded protein. Microbial protein was calculated on the basis of digestible carbohydrates and rumen degradable protein (RDP). Alternative models were used to estimate microbial protein formation, assuming the energy values of RDP and TA to be equivalent to 1.00, 0.75, 0.50, 0.25, and 0 times that of digestible carbohydrates. Because EPD values are seldom determined in production trials, they were derived using empirical models that estimate them from other feed components. The goodness of fit of models was compared on the basis of root mean squared error (RMSE) of milk protein yield (MPY) predicted from MP supply (adjusted for random study effect) and Akaike's information criterion. Metabolizable protein supply calculated from basal assumptions predicted MPY precisely within a study (RMSE = 16.2 g/d). Variable contribution of RDP to the energy supply for microbial synthesis influenced the precision of MPY prediction very little, but RMSE for MPY increased markedly when the energy supply of rumen microbes was corrected for TA concentration. Using predicted rather than constant EPD values also increased RMSE of MPY prediction. These observations do not mean that the supply of MP from undegraded feed protein is constant. However, it suggests that our current methods overestimate the range in EPD values and that the techniques have so many inherent technical problems that they can mask the true differences between the feeds. Including new elements in feed protein evaluation models may not improve the precision of production response predictions unless the consequent effects on the supply of other nutrients are taken into account.