Energy value of cereal grains for broiler chickens. III. Suitability of NIR for predicting AME of wheat for diet formulation

Context Cereal grains vary widely in energy content, which can result in poor performance of chickens when wrongly assumed apparent metabolisable energy (AME) values are used in diet formulation. Aims To evaluate the suitability of using near-infrared spectrometry (NIR) calibrations to predict AME content of grains when formulating diets for chickens. Methods Measured AME contents for 111 cereal grains were used to determine the accuracy of their prediction by (i) calculations based on physiochemical measurements, (ii) near-infrared spectrometry (NIR) calibration, and (iii) two published values for each grain type. Suitability of using NIR calibrations to predict AME content of wheat was evaluated in a chicken growth experiment. Wheat with NIR-predicted AME content of 13.59 MJ/kg as fed was formulated into four diets from starter to withdrawal or substituted by another wheat with NIR-predicted AME content of 12.45 MJ/kg as fed. Key results The NIR calibration was the most accurate method for predicting grain AME content for broiler chickens of the 111 grains from different grain species. Chickens fed the NIR-predicted higher-AME wheat diet had a lower feed conversion ratio (FCR g feed/g gain), resulting from non-significant faster growth rate and lower feed intake, than chickens fed the diet based on wheat with a predicted lower AME content when measured over a 42-day period. Conclusions Diets based on wheat with higher AME content predicted by NIR resulted in more efficient use of feed by chickens for growth than did diets based on wheat with a predicted lower AME content. Implications NIR is superior to other methods for predicting AME content of cereal grains for formulation of diets for commercial production of chicken meat.