Use of Near Infrared Reflectance Spectroscopy as a Tool for Screening Treated Forages and By-products

¹ Ruminant Nutrition Laboratory, Livestock and Poultry Science Institute, Agricultural Research Service, US Department of Agriculture, Beltsville, MD 20705

The objective of this study was to determine the feasibility of using near infrared reflectance spectroscopy to measure the composition (NDF and CP) of diversely treated forages and byproducts. Samples (325) were selected randomly from 2808 possible sample and treatment combinations generated using 18 feedstuffs, 26 treatments (acids, bases, oxidizing and reducing agents, and heat), and six levels of reagent for each treatment. The NDF and CP were determined and calculated on an OM basis. Stepwise regressions using all 325 samples and 10 nm between wavelengths gave R² and standard error of calibration of .87 and 4.1 for NDF and .93 and 1.5 for CP. Division of the 325 samples into calibration, validation, and test sets consisting of 4 of 9, 2 of 9, and 3 of 9 total samples, respectively, resulted for NDF in average R² and SE of .87, 4.1; 37, 4.3; and .82, 4.9 for the calibration, validation, and test sets, respectively, and for CP .92, 1.6; .94, 1.3; and .86, 2.1. The results for the all calibration set for NDF were similar to those for materials treated with sodium chlorite (less so for CP), indicating the potential for near infrared as a screening tool for treated feedstuffs. However, the poorer results for the test sets indicate that problems may exist with its use because of the wide diversity of samples and treatments present.

Key Words: near infrared • neutral detergent fiber • treated feeds • protein

Submitted on June 25, 1993
Accepted on December 2, 1993