Prediction of the nutritive value of forages by near infrared reflectance photometry

A filter‐type near infrared photometer equipped with 19 filters was used to predict crude protein, in‐vitro digestibility of organic matter, and in‐vivo and from in‐vitro predicted metabolisable energy in forages. The investigation was performed using pure species of grasses and legumes, grass/legume mixtures and grasses preserved by different methods. The need for specific calibrations is demonstrated. Results obtained using combinations of six filters selected during the calibration procedure were compared with those obtained using the six standard filters recommended for grain analysis. There were indications that selected filter combinations will only give accurate results if the composition of the test samples is very similar to that of the calibration samples, i.e. of the same species or family as the calibration samples. The correlation coefficient between manual methods and n.i.r. reflectance varied from 0.83 to 1.00 for crude protein, 0.48 to 0.91 for rumen digestible organic matter and 0.35 to 0.92 for metabolisable energy. When samples of grass silage were analysed using grass/legume mixtures for calibration of the equipment, the correlations were poor. Although the technique is satisfactory for determinating crude protein in forages utilised in practical farming, the need for specific calibrations when assessing in‐vivo organic matter digestibility or in‐vivo metabolisable energy renders it inferior to chemical/biological methods.