Variation in energy value and digestibility of sorghum in sows and its rapid prediction using near-infrared spectroscopy

Three experiments were conducted to evaluate the energy value and apparent total tract digestibility (ATTD) of nutrients in 11 sorghum samples (S1–S11) fed to sows at different physiological stages, and to assess the feasibility of predicting energy value and nutrient digestibility using fecal and ingredient near-infrared reflectance spectroscopy (NIRS). In Exp. 1, 36 gestating sows (Landrace × Yorkshire; parity 2) at day 50 of gestation were allotted to 12 dietary treatments, including one basal diet and 11 sorghum test diets, in a replicated 12 × 2 Youden square design. Test diets contained 30% sorghum and digestible energy (DE), metabolizable energy (ME), and ATTD of nutrients were determined using the total collection method. In Exp. 2, 36 lactating sows (Landrace × Yorkshire; parity 2) were assigned to the same 12 dietary treatments in a replicated 12 × 2 Youden square design. Digestible energy and ATTD of nutrients were determined using the indicator method. In Exp. 3, fecal and sorghum ingredient samples were used to develop partial least squares regression (PLSR) models based on NIRS. Results showed that sorghum source significantly affected energy value and nutrient digestibility in both stages (P < 0.001). In gestating sows, DE and ME of sorghum ranged from 15.84 to 17.29 MJ/kg and from 15.50 to 17.04 MJ/kg, respectively, whereas DE ranged from 14.98 to 17.18 MJ/kg in lactating sows. Tannin concentration was negatively correlated with DE in both gestating and lactating sows (r = −0.93 and −0.90, respectively; P < 0.01).  For NIRS prediction, fecal chemical composition was predicted with good accuracy, whereas DE and nutrient digestibility showed moderate predictive performance. Ingredient spectra also showed moderate potential for predicting DE and nutrient digestibility, with performance similar to that obtained from fecal spectra. These results demonstrate that tannin is a major contributor to the variation in energy and nutrient utilization of sorghum fed to sows, and that NIRS combined with PLSR has potential as a rapid and practical tool for evaluating sorghum energy utilization in sows.