Selective foraging among free-ranging herbivores can make measuring botanical composition of diets challenging. Using near- infrared reflectance spectroscopy (NIRS) on feces for predicting botanical components of individual animal diets is a novel method for studying diet selection. This study was conducted to determine the ability of fecal NIRS to predict the percentage of consumption of Leymus chinensis (Trin.) Tzvel., a dominant species in north China, by sheep (Ovis aries L.). The calibration data set consisted of 47 diets of known L. chinensis composition, paired with corresponding fecal spectra. These pairs were generated in a trial using restricted feeding. Validation pairs (n = 9) were collected in a similar trial that used ad libitum feeding. Derived coefficients of determination (R2) and standard error of calibration were 0.99% and 2.2% for partial least squares (PLS) regression and 0.89% and 7.3% for stepwise regression, respectively. Derived coefficients of determination (r2) and standard error of prediction were 0.78% and 4.8% for PLS regression and 0.90% and 3.2% for stepwise regression, respectively. PLS regression resulted in better calibration than stepwise regression, but when the calibration data set was small, stepwise regression improved the precision and accuracy of predictions compared with the PLS regression. Results of the present study show that a fecal NIRS equation developed from a restricted feeding trial could be used to predict the percentage of L. chinensis in fecal materials collected from voluntary feeding trials. The Rangeland Ecology & Management archives are made available by the Society for Range Management and the University of Arizona Libraries. Contact lbry-journals@email.arizona.edu for further information. Migrated from OJS platform August 2020
Scholarly peer-reviewed articles published by the Society for Range Management. Access articles on a rolling-window basis from vol. 1, 1948 up to 5 years from the current year. Formerly Journal of Range Management (JRM). More recent content is available by subscription from SRM.