Prediction of legume silage digestibility from various laboratory methods

The potential of different laboratory methods to predict legume silage organic matter digestibility (OMD) in vivo was evaluated by using data from thirty‐three pure legume silages in seven experiments. The samples were analysed for crude protein concentration, cell wall composition, in vitro digestibility by the methods of Tilley and Terry [Journal of the British Grassland Society, 18 (1963) , 104–111; OMD_T&T], pepsin‐cellulase solubility (OMS) and gas production (OMD_GAS), and for indigestible neutral‐detergent fibre concentration in situ (INDF). The relationships between the results obtained by the laboratory methods and in vivo OMD, all expressed as ratios, were studied using linear univariate regression models with experiment as a random variable (mixed model). Legume silage digestibility could be estimated with acceptable accuracy by different in vitro methods, but not from the chemical composition of the samples. The highest accuracy in OMD prediction was found with OMS (RMSE 0·0113; R² = 0·965) followed by OMD_GAS (RMSE 0·0149; R² = 0·944), OMD_{T& T} (RMSE 0·0149; R² = 0·940) and INDF (RMSE 0·0168; R² = 0·925). The relationships between the in vitro methods and in vivo digestibility are not universal, and should be determined separately for each laboratory and type of forage. Part of the error in OMD prediction can be attributed to errors in in vivo OMD determination.