Evaluation and Prediction of Site and Extent of Protein Digestion of Feedstuffs Commonly Fed in Beef Cattle Finishing Diets

Abstract Meeting metabolizable protein requirements of finishing cattle without overfeeding protein requires reliable estimates of site and extent of digestion. Current estimates of rumen degradable (RDP) and undegradable protein (RUP), as well as intestinal digestibility of RUP (dRUP) may be inadequate. Thus, the objectives of this experiment were to quantify RDP, RUP, and dRUP of common feedstuffs and evaluate nutrient composition as a predictor of RDP, RUP, and dRUP. Ruminally cannulated steers (n = 6) consuming a finishing diet were enrolled in replicated 3 × 3 Latin squares. Within each Latin square period, steers were assigned a feedstuff classification for in situ ruminal incubation. Single-source ingredients were classified as grain (steam-flaked and whole-shelled corn, wheat, and milo), protein (wet and dried corn distillers grains, Sweet Bran, cottonseed meal, canola meal, and soybean meal), and roughage (cotton burrs, alfalfa hay, wheat hay, sorghum × sudangrass hay, wheat silage, and corn stalks). Nutrient composition was determined for all feedstuffs using wet chemistry procedures. Feedstuffs were lyophilized, ground through a 2-mm screen, aliquoted into in situ bags, and ruminally incubated for 0, 4, 8, 12, 24, or 48 h. Samples were composited within incubation time point, analyzed for crude protein (CP), corrected for microbial contamination, and fit to non-linear models to estimate RDP and RUP using calculated passage rate. Subsamples of in situ digested feedstuffs underwent in vitro pepsin-pancreatin digestion and CP analysis to estimate dRUP. Statistical analyses were conducted using JMP Pro 16.0 and R (v4.1.0). On average, RUP estimates were 37.0% decreased (P < 0.05) for protein ingredients, yet grains and roughages were not different (P > 0.73) compared with NASEM (2016) empirical values, respectively. However, average dRUP was 11.9% and 33.1% reduced (P < 0.01) for protein ingredients and roughages, yet grain dRUP was not different (P = 0.16) compared with NASEM (2016) values, respectively. Neutral detergent fiber (NDF) and CP explained a substantial portion of the variation in RDP (P < 0.01, R2 = 0.94, RMSE = 2.26). Acid detergent fiber, NDF, and CP explained considerable variation in dRUP (P < 0.01, R2 = 0.91, RMSE = 5.43). Models were predictive when applied to independent data from the literature for estimates of RDP (R2 = 0.93; RMSE = 3.72) and dRUP (R2 = 0.83; RMSE = 6.06). There was slope (-0.09; P < 0.05) and mean bias (-1.30; P < 0.01) for RDP prediction validation but only mean bias (4.90; P < 0.01) in dRUP prediction validation models. Collectively, these data suggest that dRUP of protein ingredients and roughages are less than values reported by the NASEM (2016). Furthermore, CP and NDF may be useful for predicting RDP, RUP, and dRUP.