Prediction of the metabolizable energy value of whole-crop wheat from laboratory-based measurements

The accuracy with which several laboratory-based measurements predict the metabolizable energy (ME) value of whole-crop wheat (WCW) was determined. Twenty-six WCW forages differing in variety (cv. Slepjner, Hussar and Cadenza), maturity at harvest (milk, cheese and dough stages) and treatment applied (urea-treated, untreated or acid-based additive treated) were harvested in 2 years and conserved anaerobically in 200-l barrels. The forages were then scanned using near infrared reflectance spectroscopy (NIRS) and analysed for chemical composition, in vitro rumen fluid-pepsin digestibility, in vitro neutral detergent-cellulase plus gamannase digestibility, in vitro fermentation gas production and in situ rumen degradability ME was calculated using measured energy losses in faeces and urine and predicted energy losses as methane. The relationships between ME and the laboratory-based measurements were determined by regression. Guess energy was consistently the best predictor of ME (R-2 = 0.53 and 0.86 in years 1 and 2 respectively). However the autocorrelation involved, militates against the prediction of ME from gross energy. None of the chemical constituents or biological techniques gave a good, robust prediction of ME. However, an NIRS calibration developed using the WCW samples was highly correlated (R-2 = 0.68) with ME. This work therefore suggests that traditional laboratory-based, food evaluation techniques are unsuitable far predicting the ME content of WCW but that NIRS holds promise.