Abstract
Two experiments were conducted to validate the sensitivity and accuracy of in vitro digestible energy (IVDE) determined with a computer-controlled simulated digestion system (CCSDS) to predict metabolizable energy (ME) of diets for roosters. In experiment 1, soybean hulls were added to a basal diet (calibration diet 1) at 2.06, 4.12, 6.17, 8.23, 10.28, 12.32, or 14.37% of the diets (calibration diets 2-8) to produce an interval of approximately 80 kcal ME/kg. The sensitivity was measured by comparing the determined and actual IVDE of the diets. With these data, a linear model was developed to predict ME from IVDE. In experiment 2, validation diets were identical except they were composed of different cereal ingredients. For each diet, the correlations and ratios between IVDE and ME were analyzed to test the sensitivity of IVDE to predict ME across different ingredients. In experiment 1, a slope of 0.9899 was calculated in a linear regression of determined IVDE on actual IVDE (R2 = 0.9998; P < 0.01). The ratio of determined IVDE to actual IVDE was 0.9878. The ratio of IVDE to apparent metabolizable energy (AME) and to nitrogen-corrected AME (AMEn) was 1.03 and 1.05, respectively. The linear models to predict ME from IVDE were AME = 0.8449 × IVDE + 451 (R2 = 0.9812, residual standard deviation [RSD] = 28 kcal/kg; P < 0.01) and AMEn = 0.8357 × IVDE + 436 (R2 = 0.9821, RSD = 27 kcal/kg; P < 0.01). In experiment 2, a significant simple correlation was observed between the IVDE and AME or AMEn of validation diets (r > 0.97; P < 0.01). The ratio of IVDE to AME and to AMEn was 1.04 and 1.05, respectively. Predicted and determined AME or AMEn of 8 validation diets differed by less than 100 kcal/kg. The regression of determined AME or AMEn against predicted AME or AMEn (R2 ≥ 0.9466; P < 0.01) resulted in an overlapped line where Y = X. These results suggest the IVDE determined with CCSDS is highly sensitive and can be used to accurately predict the ME of diets for roosters across a wide range of cereal grains.