The use of mid infrared diffuse reflectance spectroscopy to determine quantitatively the chemical composition of tall fescue

Tall fescue samples that were harvested over 3 yr were ground using two methods and oven-or freeze-dried. The samples were then analysed chemically and by mid infrared diffuse reflectance spectroscopy for crude protein, neutral detergent fiber, acid detergent fiber, and in vitro digestible dry matter. Results indicated that the same principles that govern the development of near infrared calibrations apply to the development of calibrations using mid infrared spectra. Overall, results indicated that spectra collected at a resolution of 4 cm(-1) produced better calibrations than did spectra collected at 16 cm(-1) that a wide variety of data treatments needed to de tested, and that oven-and freeze-dried samples calibrated identically. However, unlike previous results using near infrared spectroscopy, samples ground in a Udy cyclone mill (Udy Corp., Fort Collins, CO) were preferred over samples ground in a Christy-Norris mill (Christy and Norris Ltd., Chelmsford, England). A reduction in the size of the calibration set decreased the accuracy of calibrations to the level found for near infrared calibrations. Approximately 56 randomly selected samples out of 223 total were needed for acceptable calibrations. Finally, determinations for samples from a given year using calibrations developed for samples from another year indicated that mid infrared calibrations, at least as developed in this study, may be less robust than near infrared calibrations. In conclusion, the results presented in this study show that mid infrared spectra can be used to develop calibrations to determine the composition of forages and that mid infrared calibrations behave in a manner similar to that of near infrared calibrations.