LAUSR

Abstract Stable isotopes can be an important research tool to track C and N in grazing experiments. This study tested different proportions of C4 grass and legume and the correlation between dietary δ13C with fecal δ13C. Forty lambs, male, with average body weight of 20.4 kg, were randomized in a complete block design and fed Tifton-85 (Cynodon spp.) and Alfalfa (Medicago sativa) hays, at different levels of substitution composing five treatments: 1) 100% Tifton-85 hay; 2) 75% Tifton-85 + 25% alfalfa hay; 3) 50% Tifton-85 + 50% alfalfa hay; 4) 75% alfalfa + 25% Tifton-85 hay; 5) 100% alfalfa hay. The trial lasted 27 days, consisting of 22 days for adaptation and 5 days for collection of orts, feed, and feces. Fecal samples were collected directly from the rectum, to avoid contamination. All samples were collected during the 5-d sampling period, being composited at the end. Feed and fecal samples were dried for 72 h in an oven (55oC) and ground in a Willey mill to pass a 2-mm sieve, incubated for 288 h in situ to obtain the indigestibility. All collected samples were subjected to indigestible neutral detergent fiber (iNDF) and carbon stable isotope analysis. Dietary isotopic composition (δ13C) in the iNDF residue processing was −14.98, −18.22, −23.85, −25.99, and −30.64‰ for treatments 1–5, respectively. The δ13C of dietary samples iNDF treated was −17.41, −20.27, −26.06, −27.83, and −31.67‰, respectively for diets 1–5. Fecal δ13C for non-treated samples was −16.23, −20.79, −25.10, -28.8, and −32.31‰, and for iNDF-treated sample −16.65, −21.52, −26.25, 29.20, and −32.06‰, respectively, for diets 1–5. Fecal samples were more depleted than diet samples, and the iNDF changed the fecal δ13C. The models used to back calculate the diets predicted well the diets by fecal δ13C, and the best models had R2 of 0.98 using non-treated samples, with similar results found when using iNDF-treated samples (R2 = 0.97). There is a 13C discrimination between dietary and fecal samples, however, the proportion of C3 and C4 species in the diet can be accurately predicted based on fecal samples using δ13C. The use of iNDF treated samples did not improve the models, as well as the addition of digestibility and indigestibility in the models.