Variation of amino acid content in the small intestine digesta of cattle: development of a prediction model

Abstract

The amino acid (AA) profile of dietary absorbable proteins may not always be the best for optimal milk protein synthesis, as demonstrated by experimental post-ruminal supplementation of methionine and lysine. Nevertheless, the AA profile of absorbable protein can vary widely with the diet. To be able to balance most of diets according to the main limiting AA, it is crucial to predict the AA composition of duodenal digesta. This article describes an attempt to predict the duodenal AA composition, based on the PDI system and literature reports of the AA composition of rumen microbes, rumen undegraded feed proteins and endogenous proteins. Results from this initial model were tested against AA compositions measured in 133 cattle diets. With most AA the tests revealed slight biases (less than 5% differences) in the mean value predictions except for Met and Gly (12 and 14% differences). Moreover, the model tended to underestimate (−0.1 to −18.3%) low AA contents and overestimate (+0.3 to 14.5%) high ones. The initial model was adjusted for these biases by covariance analysis. After correction the data were predicted with a good degree of confidence for Lys Arg, Thr, Val, Leu, Asp, Ser, Glu and Pro (0.65< R 2<0.80). The residual variations were generally low (<6%) except for His, Met, Pro and Gly (7 to 14%). Factors susceptible of inducing bias, such as the estimated proportion and composition of microbial protein AA or AA composition and content in undegraded dietary protein, are discussed.