Prediction of diet quality parameters

Abstract

Research was conducted to determine the feasibility of predicting percent dietary crude protein (CP), phosphorus (P), and digestible organic matter (DOM) of white-tailed deer (Odocoileus virginianus) through analysis of feces with near infrared reflectance spectroscopy (NIRS). Feeding trials were conducted with stall-fed, white-tailed deer. Seventy-six rations were formulated with combinations of over 50 forage species representing a broad range of values and composition. Fecal samples were collected on days 6 and 7 after feeding the rations. Micro-Kjeldahl procedure was used to determine CP and P of each ration. Digestible organic matter was determined by in vitro analysis corrected for known in vivo digestibility. Fecal NIRS scans were paired with dietary CP, P and DOM values to establish a reference data set and resulting equations. A 7-wavelength equation for CP, 8-wavelength equation for P, and 4-wavelength equation for DOM, allowed good predictions of the data set: le=0.94, 0,94, and 0.89, SEC=0.70@ 0.02, and 2.64, n--l 1 1, 114, and 117, respectively. Validation of selected equations with an independent sample set yielded acceptable results: SEP=0.87, 0.02, 2.89, R2=0.94, 0.93, and 0.85, for CP, P and DOM, respectively. Geographic spectral stability of predictive equations was assessed for four different ecological communities. Performance was encouraging given the limited data set and vegetational regions forage was harvested from. A hierarchical equation system formulated with six separate equations for CP, P and DOM enhanced prediction capabilities. Results of this study indicate prediction of diet quality for white-tailed deer with NIRS fecal scans can provide managers with critical information necessary for proper nutritional management of free-ranging deer populations. Further work is necessary to increase geographical robustness of the predictive equations.