Methods of energy evaluation of feed ingredients for rabbits and their accuracy

Abstract

The influence of the methodology used in the evaluation of feed ingredients for rabbits on the precision of determination is reviewed. The procedures for calculation of the energy values and their standard errors are shown and discussed for each method. The direct method used for evaluating some relatively balanced feedstuffs such as alfalfa hay or wheat bran gives consistent results, but over-or underestimation and low repeatability is obtained with most imbalanced ingredients. For the substitution method, two calculation procedures of the energy values are compared, showing greater accuracy with the equation in which the gross energy of the ingredient is determined directly instead of estimated by difference. The basal diet and the rate/s of substitution must be designed to avoid a great nutrient imbalance in all the experimental diets. The substitution rate is the main factor in the precision of estimation of feedstuff evaluation, being directly related to the accuracy. Thus, the standard error of the ingredient energy value estimated by difference is 13.4, 6.4 and 2.9 times higher than the standard error of diets for 10%, 20% and 40% substitution rates, respectively. The number of replicates is of limited importance in comparison with the substitution rate; thus, the numbers of rabbits required to estimate the mean energy value within ±5% with a 95% confidence are 37 and 10 for 20% and 40% substitution rates, respectively. Therefore, the energy value of ingredients that are usually included in diets in high proportions (20-30%) can be obtained with good precision by difference using only one rate of substitution. Several levels of inclusion (four or more) must be used when interactions between ingredients are expected or very low rates of inclusion are used. In this case, the energy value is estimated by extrapolation of the regression between the digestible energy of diets and the substitution rates and a greater precision than by difference is obtained. The multiregression method is rarely used because of the great number of ingredients included in the diets and because of the existence of interactions among them, preventing the correct estimation of their energy values.