TRUE AND APPARENT METABOLIZABLE ENERGY VALUES FOR POULTRY OF CANADIAN WHEATS AND OATS MEASURED BY BIOASSAY AND PREDICTED FROM PHYSICAL AND CHEMICAL DATA

Thirty samples of wheat and 28 samples of oats were assayed for true and apparent metabolizable energy (TME, AME). Within grains, the difference TME−AME increased with decreasing AME values; there is evidence that this trend is associated with reduced voluntary consumption of AME assay diets containing low energy grains. The TME and AME data were compared with ME values predicted from physical and chemical data describing the grains. Previously published prediction equations were tested and new equations were derived. Comparisons between predicted and observed data suggested that both the TME and AME values of wheat were predicted with insufficient accuracy and precision for practical use. Similar comparisons using the oat data showed high correlations between observed and predicted values, although the predictions were no more accurate than for wheat; however, when data describing four samples of naked oats were removed, the correlations were reduced substantially. Comparisons involving data for the hulled oats indicated that most equations were able to predict AME better than TME. Multiple regression analysis was used to identify those combinations of variables best able to predict TME data. No combination of variables was best for both wheat and oats. The combinations of variables used in published equations performed quite well. With four variables, the percentage of the TME variation explained was as high as 52 for wheat, 82 for oats and 64 for hulled oats. Predictions based on air-dry data are associated with higher correlations than those based on dry matter data, but the air-dry predictions are the less useful in practice. The reason for this is discussed.