Estimation of Stoichiometric Parameters for Rumen Fermentation of Roughage and Concentrate Diets

Abstract

A biochemical model coupled with an iterative nonlinear least-squares program was used to estimate stoichiometric parameters for fermentation of soluble carbohydrate, starch, hemicellulose, cellulose and protein in the rumen. Fermentation parameters were deduced for two sets of literature data, on (1) a roughage group of 60 diets fed to 137 animals and (2) a concentrate group of 48 diets fed to 374 animals. Soluble carbohydrate and starch fermentation parameters were much different for the roughage and concentrate groups. Differences were also noted in fermentation parameters for hemicellulose and cellulose. Model sensitivity to imposed changes of iteratively deduced parameters supported the conclusion that the estimated values are unique for all substrates except protein. In vitro estimates of isolated substrate fermentation from the literature agreed favorably with those indicated for hemicellulose. Cellulose fermentation was estimated to yield an acetate to propionate ratio of 7.8 to 13.1:1. This was higher than estimates in the literature of 1.3 to 3.6:1. Predicted fermentation patterns were within the range of normal biological variation noted for actual patterns. Systematic error components were identified as lignin plus ash as a percentage of feed dry matter and percentage roughage in the diet for the roughage and concentrate groups, respectively. Significant bias remained only in the roughage group for acetate and butyrate. Predicted vs observed acetate, propionate and butyrate production, for diets where comparison data were available, had slopes (correlation coefficients) of .84 (.98), 1.03 (.97) and .94 (.98), respectively. By this method, products of fermentation were estimated with nearly the same accuracy with which they can be measured experimentally.