Modeling sow precision feeding based on farm-specific body composition reduces feed costs and environmental load

Modern sows are heavier than those used 30 years ago to establish the equations linking body energy, protein and fat to body weight (BW) and backfat thickness (BT). These equations are among others used in nutritional models to assess individual daily requirements to implement a precision feeding strategy (PF). Using calibrations of equations and a simulation approach, the objective of the current study was to evaluate the impact of four feeding strategies over six successive gestations on sows’ long-term performances. A database of 3098 gestations from 1121 sows containing sows’ parity, BW, BT, feed intake, and litter performances was used to calibrate predictive equations of sows’ body chemical composition in a nutritional model. During gestation, a conventional feeding strategy (CF, same amount for all sows) was compared to a standard feeding strategy (SF) adjusted on individual energy requirements (variable amount per sow), and to two PF strategies adjusted on sow energy and amino acids (AA) requirements (PF AA , variable amount of feed per sow and variable amount of lysine content over days and sows) and with additional phosphorus (P) requirements (PF AA-P , variable amount of feed per sow and variable amounts of lysine and P contents over days and sows). The database was split into a training set (to fit body chemical composition equations from literature) and a testing set (to evaluate the differences between observed and predicted BW and BT at farrowing with calibrated equations). For the calibrated equations, the root mean squared error (RMSE) of sows’ BW and BT at farrowing were 10 kg and 1.7 mm, respectively. Compared to the parameters before calibration, RMSE of sows’ BW increased by 3 kg and RMSE of sows’ BT decreased by −5.2 mm. At the sixth farrowing, BT was 2 mm lower for CF compared to BT target and BT of other feeding strategies (P < 0.001). The BT variability in the herd was also 19% greater for CF than SF, PF AA, and PF AA-P (P < 0.001). Over six successive gestations, feed costs were reduced by 17 € and 26 €, while nitrogen (N) efficiency and P efficiency increased by 30 and 5% and by 15 and 30%, respectively for PF AA and PF AA-P strategies compared to CF (P < 0.001). In conclusion, based on in silico results, feeding the gestating sows individually according to their energy requirements enabled to better reach the target BT at farrowing with the succession of cycles. Adjusting the nutrient ies to individual sow requirements, AA considered alone or in combination with P, reduces feed costs and enhances nutrient efficiencies in the long-term. • Effects of four feeding strategies were simulated over six successive gestations. • Calibrating body composition equations at farm level better met body observations. • Feed supply to sow energy need reduced backfat thickness variability at farrowing. • Adjusting nitrogen and phosphorus supplies improved efficiencies of use by 3%. • Adjusting lysine supply for six gestations decreased feed costs by 17 €.