NEAR INFRARED REFLECTANCE SPECTROSCOPY FOR PREDICTING FORAGE COMPOSITION AND VOLUNTARY CONSUMPTION AND DIGESTIBILITY IN CATTLE AND SHEEP

Near infrared reflectance spectroscopic (NIR) measurements were made on 82 samples of legume (alfalfa and clover), grass (brome, timothy, reed canary grass and meadow foxtail) and legume-grass mixtures using a Neotec model 6100 scanning monochromator. Data on the forages, used for establishing NIR calibrations for predictive relationships and appraising them, were chemical composition and measurements of digestibility and voluntary consumption for cattle and sheep. The primary wavelengths selected by multiple regression techniques were similar to those obtained by other researchers for crude protein, acid and neutral detergent fiber, calcium and phosphorus. Similar primary wavelengths were selected for prediction of digestibility and voluntary intake (g kg −0.75 ) of forages for cattle and sheep, but those selected for voluntary intake on the basis of percentage of body weight differed between animal species. The wavelengths which best predicted animal intake and digestibility in our trials differed from those reported by other researchers. Crude protein, acid detergent fiber, neutral detergent fiber, lignin, acid detergent insoluble nitrogen, calcium, phosphorus and ash concentrations in forage were predicted with standard errors of 1.0, 2.2, 2.9, 1.1, 0.07, 0.15, 0.02, and 1.2%, respectively. The accuracy of predictions for these chemical constituents was similar to that reported by other workers. Digestible energy content, dry matter digestibility, voluntary intake and digestible energy intake of hays by cattle were predicted with standard errors of prediction of 0.59 MJ kg, 2.4%, 7.6 g DM kg −0.75 , and 79 kJ kg −0.75 , respectively. Corresponding values for sheep were 0.96, 4.4, 6.3 and 128. The quantitative importance of variability in animal data in the calibration of the NIR procedure was discussed. This variability accounted for about one-half of the variability of NIR prediction of voluntary DM and digestible energy intake of cattle. This proportion was reduced to approximately one-quarter and one-sixth for digestibility of dry matter and digestible energy content of feed, respectively. Key words: Cattle, sheep, forages, near infrared reflectance spectroscopy, nutritive value