Causes of inaccurate prediction of volatile fatty acids by simulation models of rumen function in lactating cows

Extant mechanistic models of rumen function are unable to predict the molar proportion of volatile fatty acids (VFA) accurately. In order to make these models useful in investigating theories on nutrient flows that go beyond the rumen, the representations adopted need to be improved. This theoretical study was directed at identifying what parts of a rumen model may be responsible for the inaccurate VFA prediction. For five distinct parts of a model, not involving the description of the microbiological metabolism, the hypothesis was tested that their inappropriate description can be a probable cause of inaccurate VFA prediction. These five parts were: (1) the input functions of feed ingestion; (2) the representation of insoluble, degradable substrate of roughage and concentrate origin; (3) the kinetics of VFA absorption from the rumen; (4) the VFA coefficients that describe the stoichiometry of the conversion of fermented substrate into VFA; and (5) the representation of the rumen particle dynamics. Every hypothesis was tested by modifying the model description and simulating it to steady state. Observations required were derived from digestion trials with lactating dairy cows. Simulation results demonstrated that the predicted molar proportion of rumen VFA concentrations is particularly influenced by VFA absorption kinetics and VFA coefficients. Although the description of particle dynamics also had a large influence with certain choices of its parameterization, it is probably a less important cause of inaccurate prediction when rumen feed degradation (apparent from rumen outflow) is predicted well. In conclusion, to obtain improved predictions of the molar proportions of rumen VFA, further work is required on the representation of VFA absorption kinetics and of VFA coefficients of fermentation stoichiometry. (C) 1997 Academic Press Limited.