Predictability of dry matter intake, diet composition and digestibility in beef cattle using fecal near-infrared reflectance spectroscopy

Abstract

The objective of this project was to use F.NIRS to develop calibration equations to predict forage intake, composition of the consumed forage and nutrient digestibility of tropical grasses when the intake is limited by either nutrition or non-nutritional factors. Ten Nelore heifers were housed in individual pens with concrete floors and free access to fresh water during 5 experimental periods divided into 10 sub periods of 7 days each. Animals were blocked in pairs according to age and body weight and, within each block, randomly assigned to a sequence of two treatments: ad libitum or restricted fed. Diet was offered twice daily (at 07:00 and 14:00), composed exclusively of Pennisetum purpureum (Elephant Grass), daily harvested and chopped. Individual feed intake was measured daily by weighing feed offered and orts. During the sampling period (3 out of 7 days per sub period), samples of the feed offered and orts were collected daily and total daily fecal production was measured by collecting all feces in each pen’s floor as soon as the animal excreted them every 6 hours. Dried and ground individual fecal samples from each animal were subjected to NIRS scanning. Spectra were recorded in diffuse reflectance mode using a Fourier transformation NIR spectrometer and stored using OPUS Spectroscopy Software. Fecal spectra were then used to develop equations to predict diet composition, digestibility and dry matter intake. Coefficients of determination for calibration (R2 c) and cross-validation (R2 cv) for prediction of diet composition were greater for % CP (R2 c = 0.99; R2 cv = 0.97), % NDF (R2 c = 0.97; R2 cv = 0.95) and % OM (R2 c = 0.98; R2 cv = 0.97) with an RPD higher than 4.5 for all parameters. For CPd, NDFd and OMd, R2 c and R2 cv were greater than 0.90 and RPD higher than 3. For DMI (R2 c = 0.96; R2 cv = 0.77) and DMd (R2 c = 0.0.93; R2 cv = 0.88) and an RPD lower than 3 for both parameters. F.NIRS showed its potential for prediction of % CP and % NDF of the diet consumed, requiring just one day of sampling, even when intake is limited by non-nutritional factors. Further research is needed to develop more robust and larger calibration equations for the prediction of DMI, however, our results provides good evidence that improved F.NIRS calibrations can be used as an alternative method to monitor diet composition of grazing animals.