Search

Near infrared (NIR) spectroscopy is in widespread use at grain receival silos and stockfeed mills to determine the protein and moisture contents of cereals used in livestock feeds. These parameters alone, however, are only the first step in the definition of nutritive value, which is best determined by accounting for the inevitable losses that occur following digestion and absorption within animals. When formulating diets for growing pigs, the greatest cost pressure is against the cost per unit of energy and, hence, digestible energy (DE) is a valuable measure of the nutritive value of cereals for this class of livestock. The DE content of cereals for pigs can vary considerably within a cereal type (in excess of 3 MJ kg−1 DM) and a rapid means of assessing this parameter would improve the accuracy of diet formulations for growing pigs and reduce the cost of the diets. To date the use of NIR to predict DE has been limited by the cost associated with obtaining a sufficient number of samples with in vivo measurements over an adequate range necessary to develop a robust NIR calibration. To overcome this problem, a range of cereals (wheat, barley, sorghum, triticale, maize) that had been fed to pigs experimentally, to determine their DE contents were sourced from Australia, Canada, France and New Zealand. NIR spectra of both whole and milled grain were recorded and used to derive calibrations for DE. A disadvantage of this approach is the influence of interlaboratory variation on the reference sample set and the resulting calibration. This study has, for the first time, demonstrated the value of indicator variables in detecting the presence of, and compensating for, interlaboratory variability in a set of animal nutrition data.