Validation of prediction equations for apparent metabolizable energy of corn distillers dried grains with solubles in broiler chicks

An experiment consisting of 3 nearly identical trials was conducted to determine the AMEₙ content of distillers dried grains with solubles (DDGS) to validate 4 previously published prediction equations for AMEₙ of corn DDGS in broilers. In addition, prior research data were used to generate a best-fit equation for AMEₙ based on proximate analysis. Fifteen samples of DDGS ranging in ether extract (EE) from 4.98 to 14.29% (DM basis) were collected from various dry-grind ethanol plants and were subsequently fed to broiler chicks to determine AMEₙ content. A corn-soybean meal control diet was formulated to contain 15% dextrose and test diets were created by mixing the control diet with 15% DDGS at the expense of dextrose. In each trial, male Ross × Ross 708 chicks were housed in grower battery cages and received a common starter diet until the experimental period. Each cage was randomly assigned to 1 of the dietary treatments (trial 1 and trial 2: control + 6 test diets, 13 replicates per diet; trial 3: control + 3 test diets, 12 replicates per diet). Experimental diets were fed over a 6-d acclimation period, followed by a 48-h total excreta collection period. On a DM basis, AMEₙ of the 15 DDGS samples ranged from 1,975 to 3,634 kcal/kg. Analyses were conducted to determine gross energy, CP, EE, DM, starch, total dietary fiber, neutral detergent fiber, crude fiber (CF), acid detergent fiber, and ash content of the DDGS samples. All results were reported on a DM basis. Application of the 4 equations to the validation data resulted in root mean square error (RMSE) values of 335, 381, 488, and 502 kcal/kg, respectively. Least absolute shrinkage and selection operator technique was applied to proximate analysis data for 30 corn coproducts adapted from prior research and resulted in the following best-fit equation: [AMEₙ (kcal/kg) = 3,673 – (121.35 × CF) + (51.29 × EE) – (121.08 × ash); P < 0.01; R² = 0.70; R²ₐdⱼ = 0.67; RMSE = 270 kcal/kg]. The RMSE values obtained through validation were not consistent with the expectation of predictive performance based on internal measures of fit for each equation. These results indicated that validation is necessary to quantify the expected error associated with practical application of each individual prediction equation to external data.