The use of near infrared reflectance spectroscopy (NIRS) for prediction of the nutritive value of barley for growing pigs

Abstract

There is a need in the feed industry for a rapid means of evaluating the nutritive value of feeds and feed ingredients. Chemical analysis provides only basic information and most of the laboratory techniques take too long for this information to be of use in feed formulation at the feed mill. Near infrared reflectance spectroscopy (NIRS) has been proposed as an alternative means of predicting nutritive value. In this study, NIRS was used to predict the digestible energy (DE) concentration and in vitro ileal digestibility of crude protein (CP) and total-tract digestibility of energy of locally produced barley. The calibration and validation statistics were developed using modified partial least squares (MPLS). Derivatisation and scatter correction procedures were carried out to reduce interference from external effects. The correlations between actual and predicted DE values, based on both calibration (R2 0.93) and validation (R2 0.69), were strong with corresponding low standard errors of calibration (SEC) and cross validation (SECV) (SEC 0.128, SECV 0.279). Strong correlations were also observed between predicted and actual in vitro digestibility values for both calibration and validation exercises. It was noted that validation weakened the correlations (R2 0.73 vs. 0.50 for in vitro ileal digestibility of CP and 0.80 vs. 0.68 for in vitro total tract digestibility of energy) and fractionally increased the standard errors (0.016 vs. 0.020 for in vitro ileal digestibility of CP and 0.018 vs. 0.024 for in vitro total tract digestibility of energy). The correlations obtained by cross validation of the lowest SECV equations were not significantly different to those obtained by the scatter correction treatments. The strong relationships and low standard errors obtained between the actual and predicted values indicates that NIRS may be of use in predicting the nutritive value of barley for growing pigs, although more research is required to include larger sample sets.