Comparaison de différentes méthodes d'évaluation du coefficient d'utilisation digestive des aliments concentrés par le ruminant

Publisher Website
Google Scholar
Abstract
Digestibility of a ruminant compound feed cannot be directly measured. Various methods have been proposed to evaluate the digestibility coefficients in such a situation and those methods are compared. The data used were derived from measurement obtained with 4 mixed diets (roughage + concentrate) and took into account 16 components. The classical method, or difference method, gives unreliable results when the percentage of the component provided by the concentrate in the diet is < 20%. The use of ponderated values did not modify statistically the digestibility coefficient values, but allowed reduction of the safety ranges. Similar conclusions were obtained when individual variations of the roughage digestibility were taken into account instead of a sole mean value (variable hay difference method). With the equation-method, there is a minimal value of the incertitude which is a function of the number of eliminated data and the value of the minimal difference allowed between 2 levels of concentrate incorporation. This mean level was .apprx. 10%. A stepwise multiple regressive model including the level of concentrate was also used to test phenomena of associative digestibilities, the influence of the feeding level and various interactions. Results obtained with the equation and regression methods were rather similar. They were not so highly correlated with those provided by the difference method. Comparison of these various methods indicated that the reliability is largely influenced by the mean concentrate percentage, but also by the variations of these percentages. The regression method seemed to be the most suited because of its possibility to test simultaneously the influence of different exogenous parameters. The importance of knowing the specific reliability and the method used to determine each published digestibility coefficient are emphasized.