Energy value of cereal grains for broiler chickens. II. NIR prediction calibrations

Context Energy intake by broiler chickens is related to growth rate and time to final weight. Cereal grains are the predominant energy source, but individual samples vary widely in their capacity to provide energy. Accurate assessment of the energy in a grain before incorporation into a diet would improve diet formulation and bird performance. Aims To determine whether near-infrared spectrometry (NIR) technology can provide accurate, real-time prediction of the apparent metabolisable energy (AME) content and intake of cereal grains for broiler chickens. Methods Results from 431 grain measurements in experiments with chickens were used to develop NIR calibrations for grain AME content (MJ/kg as fed) and AME intake (MJ/bird.day) for combined gender, male and female birds offered diets with and without exogenous enzymes. Diets contained 80% wheat, barley, triticale, sorghum, maize, oats or rice varying in genetics, growing conditions and country of origin. Accuracy of NIR calibrations from scans on whole grain or milled grain were compared within grain and across grain species. Key results NIR calibration statistics for combined gender without endogenous enzymes for AME content and relative AME intake (AME intake index) were respectively, as follows: fraction of variance accounted for (RSQcal), 0.88 and 0.85; standard error of cross-validation (SECV), ±0.46 (MJ/kg as fed) and ±3.33 (0–100); and ratio of predicted deviation-calibration robustness (RPD) 2.67 and 2.27. Corresponding values for diets with enzymes were as follows: RSQcal, 0.92 and 0.84; SECV, ±0.31 and ±3.18; and RPD, 3.13 and 2.21. Combined gender calibrations had higher accuracy of prediction than did male or female bird calibrations. NIR statistics based on whole grain scans were slightly superior to milled grain scans and require less labour. Conclusions NIR calibrations developed for combined gender birds fed diets with or without exogenous enzymes for grain AME content, but not for relative AME intake, are suitable for real-time application in broiler diet formulation. Implications NIR calibrations for predicting AME content of cereal grains based on whole grain scans can be used real-time during formulation of broiler diets with or without enzymes and for separating grains during translocation.