Carbohydrate quantitative digestion and absorption in ruminants: from feed starch and fibre to nutrients available for tissues.

Carbohydrates are the main source of energy in ruminants. Their site, extent and kinetics of digestion highly impact the amount and profile of nutrients delivered to peripheral tissues, and the responses of the animal, i.e. ingestion, efficiency of production, N and methane excretion, quality of products and welfare. Development of multi-objective feed evaluation systems thus requires a more integrated quantitative knowledge on carbohydrate digestion and yield of terminal products, as well as on their metabolism by splanchnic tissues. The objective of this paper is to review (i) quantitative knowledge on fibre, starch and sugar digestion, volatile fatty acids (VFA) and glucose production and splanchnic metabolism and (ii) modelling approaches which aim at representing and/or predicting nutrient fluxes in the digestive tract, portal and hepatic drainage. It shows that the representation of carbohydrate digestion and VFA yield is relatively homogeneous among models. Although published quantitative comparisons of these models are scarce, they stress that prediction of fibre digestion and VFA yield and composition is still not good enough for use in feed formulation, whereas prediction of microbial N yield and ruminal starch digestion seems to be more satisfactory. Uncertainties on VFA stoichiometric coefficients and absorption rates may partly explain the poor predictions of VFA. Hardly any mechanistic models have been developed on portal-drained viscera (PDV) metabolism whereas a few exist for liver metabolism. A qualitative comparison of these models is presented. Most are focused on dairy cows and their level of aggregation in the representation of nutrient fluxes and metabolism highly differs depending on their objectives. Quantitative comparison of these models is still lacking. However, recent advances have been achieved with the empirical prediction of VFA and glucose production and fluxes through PDV and liver based on the current INRA feed evaluation system. These advances are presented. They illustrate that empirical prediction of ruminal VFA and intestinal glucose production can be evaluated by comparison with measured net portal net fluxes. We also illustrate the potential synergy between empirical and mechanistic modelling. It is concluded that concomitant empirical and mechanistic approach may likely help to progress towards development of multi-objective feed evaluation systems based on nutrient fluxes.