Submitted to: International Grasslands Congress
Publication Type: Proceedings
Publication Acceptance Date: 7/4/2005
Publication Date: 7/15/2005
Citation: Coleman, S.W., Johnson, C.E., Reiling, B.A., Mislevy, P. 2005. Prediction of n fractions of warm-season grasses with near-infrared reflectance spectroscopy. June 24, 2005 - July 2, 2005. Dublin, Ireland. XX International Grasslands Congress: Offered Papers. p.260.
Interpretive Summary: Warm season grass are the predominant forages for livestock feed in the southern U.S.A. While they grow rapidly and produce abundant forage during the growing season, they often are lacking in quality or the ability to support animal growth and milk production. The protein concentration and availability is often deficient. Modern Nutrient requirements from the National Research Council (NRC)divides feed protein into ruminally degradable and undegradable, signifying those that may be broken down in the rumen by bacteria and those that escape degradation and are passed intact into the lower tract for absorption. However, the methods for determining these forms is very laborious and expensive and is impractical for most forage sources. Furthermore, the relative amounts of different proteins change with forage species, maturity, and other environmental factors. For this reason we tried to determine the relative proportions of soluble, representing degradable, and insoluble, representing undegradable, protein by near-infrared reflectance spectroscopy (NIRS), a quick method of analysis. The forages were Bermudagrass, Bahiagrass, and stargrass. Our evaluations suggested that determined fractions such as total soluble, acid detergent insoluble, and tri-chloro-acetid acid insoluble protein fractions were readily predicted with NIRS. Those fractions that were calculated from intermediates had more inherent error and were not related to NIRS. This suggests that NIRS can be used to classify forage proteins.
Technical Abstract: The common forages for beef production in much of the U.S.A. are warm-season (C4) grasses which are very productive, but often lower in forage quality than temperate forages. Current feeding standards from the National Research Council have adapted the Cornell Net Carbohydrate and Protein System to more accurately characterize forage quality. As these procedures are tedious, data are limited on genetic and management factors influencing the fractions in C4 species. The objective of this research was to determine if near-infrared reflectance spectroscopy (NIRS) could be used to predict the various N fractions in thee C4 grasses, bahiagrass (Paspalum notatum), bermudagrass (Cynadon dactylon), and stargrass (Cynodom nlemfuensis) over five N fertilizer levels (0,40,70,110,147 kg ha-1) and five harvest dates per year on over a two-year period. Spectal data were collected on each dried, ground sample from 400-2500 nm with a NIRS Systems 6500 spectrophotometer. calibration equations were developed using partial least squares regression. A robust method was used to evaluate the equations by deleting one harvest date from the calibration data and using samples from that date for validation, and repeating for each harvest date. Calibration statistics indicate a good relationship with the laboratory determinations (TCA insoluble N, Soluble protein N, NDIN, ADIN).Fractions calculated by difference (e.g., b1 and b2) were more difficult to predict, especially when calculated as a fraction of N. When the robust evaluation was used, small but significant bias existed among the interactions of species, harvest date and occasionally fertilizer level. Part of the bias was due to bahiagrass having a higher proportion of insoluble N fractions (e.g., ADIN). Though average bias was small, it increased with increasing ADIN levels indication non-linearity of predicted values when compared to laboratory values. The results of this study demonstrate that NIRS can be used to predict fractions of N related to the solubility and degradability in the rumem. However, prediction of calculated fractions were rather poor, but could be calculated from predicted primary fractions such as total, NDIN and ADIN.