Prediction of Bermudagrass Composition and Digestibility with a Near‐Infrared Multiple Filter Spectrophotometer

Agronomists, plant breeders, and ruminant nutritionists have long sought a rapid, accurate technique for determining the chemical composition and nutritive value of forages. Recently, near‐infrared (NIR) spectroscopy has been proposed as a way to accomplish this. Although monochrometer‐type instruments have been used to predict forage quality, filter‐type instruments have not been used for this purpose. Compared to monochrometer‐type instruments which can scan the entire NIR region, filter instruments have higher relative optical energies and may be more useful for generating quantitative data for certain forage constituents. Filter instruments can scan selected portions of the NIR in about 20 sec instead of the 1 or 2 min required by monochrometers. The relative low cost of filter instruments also permits their use for routine analytical determinations. This study was conducted to evaluate the applicability of a NIR multiple filter Spectrophotometer for determining, on a dry matter basis, crude protein, neutral detergent fiber, acid detergent fiber, lignin, and in vitro and in vivo dry matter digestibilities for three groups of field‐grown bermudagrass ( Cynodon dactylon (L.) Pers.) cultivars processed as freeze‐dried forage, sun‐cured hay, and drum‐dehydrated pellets. We particularly wanted to determine if the same set of wavelengths could be used for all samples. Primary and secondary wavelengths for predicting each forage constituent were initially selected with a monochrometer type instrument. Corresponding wavelengths were then selected for the filter Spectrophotometer. Samples of known composition and digestibility were run and the resulting reflectance data used to derive a multiple regression equation from which the quality of the unknown samples was determined. The filter spectrophotometer's six “tilting” filters covered selected regions of the NIR spectrum from 1.50 to 2.36 µm. The same filters and wavelengths could be used to predict the composition and digestibility of the freezedried forages and sun‐cured hays. However, new wavelengths had to be selected to determine crude protein in the drum‐dehydrated pellets. Standard errors obtained with the NIR filter instrument were 0.58 to 1.15 for crude protein, 1.39 to 2.77 for neutral detergent fiber, 1.07 to 2.46 for acid detergent fiber, 0.50 to 0.66 for lignin, 1.90 to 3.37 for in vitro dry matter disappearance, and 1.78 to 254 for in vivo digestible dry matter. Generally, these errors compare favorably with those obtained by conventional laboratory methods. Additional research will probably result in standard errors of prediction being lowered even more, which should result in this technique gaining wide acceptance for forage analysis.