A gas production technique as a tool to predict organic matter digestibility of grass and maize silage

Seventy-two grass silages and 67 maize silages with known in vivo organic matter digestibility (IVOMD) (0.664 and 0.670, respectively) in sheep were evaluated by an in vitro gas production (GP) technique. Samples of forages (500 mg) were incubated in buffered rumen fluid for 96 h and the gas volumes recorded at 2, 4, 6, 8, 10, 12, 15, 19, 24, 30, 36, 48, 72 and 96 h. Kinetics of gas production were derived from gas volume recordings described by the exponential equation p = A + B(1 − e−ct). By means of a stepwise multiple regression analysis, prediction equations for IVOMD were derived using GP parameters alone and in combination with chemical components (i.e., crude protein (CP), neutral and acid detergent fibres, lignin, and starch for maize silage), and/or silage fermentative indicators (i.e., dry matter, pH and NH3). The predictive equations were subjected to a 'leave-one-out' or 'jack-knifing' cross-validation process. It was necessary to consider silages separately in order to obtain the best IVOMD prediction accuracy. More than 67% of the variation in IVOMD for grass silage was explained by in vitro gas production parameters 'A', 'B' and 'c', but for maize silage only parameter 'c' was related to IVOMD, accounting for 17% of total variance. Inclusion of CP in the multiple regression improved IVOMD prediction, accounting for 77 and 38% of variation in grass and maize silages, and up to 70 and 43% on inclusion of silage fermentative indicators. The highest predictive accuracy occurred when gas production parameters were combined with chemical constituents and silage fermentation parameters (79 and 49%). However, the residual standard deviation of validation of the prediction equation was lower for maize versus grass silage (22.2 versus 34.2), with coefficient of variation of 3.32 and 5.15, respectively. The GP technique can be used to predict IVOMD of grass and maize silages.