Dairy cow milk protein responses to digestible lysine and methionine in the new Inra feeding system

Dairy cow milk yield, protein yield (MPY) and content (MPC), were predicted in response to the supplies of netenergy of lactation (NEL), metabolisable protein (PDI) using curvilinear plateau equations. These responses were combined to marginal responses to Lys and Met Digestible in the Intestine (LysDI and MetDI in % PDI). The responses to NEL and PDI were estimated in a dataset with LysDI on average at 6.7% PDI and MetDI at 1.9% PDI. The curvilinear plateau marginal responses of MPC (dMPC) and MPY (dMPY) to LysDI variations around 6.7% PDI were established on 30 experiments (exp) in which NEL, PDI and MetDI did not vary and on 87 treatments (treat). Similarly, the responses to MetDI variations around 1.9% PDI were established on 21 exp and 63 treat. If LysDI = 7% or MetDI = 2.4% of PDI, 95% of their respective dMPC increase was observed. The best model to predict the dMPC response to combined effects of LysDI and MetDI (RMSE: 0.43 g/kg) was to take the minimum between the linear part of the response to LysDI (dMPCLys = 0.62 + 1.75 × [LysDI – 6.7] with LysDI<6.7; 26 exp and 45 treat; RMSE=0.59 g/kg) and the curvilinear plateau response to MetDI (1.22 + 9.14 × [MetDI – 1.9] – 9.14 × 0.193×ln{1=exp[(MetDI – 1.9) / 0.193)]}; RMSE=0.58 g/kg), in accordance with the barrel concept. Neither Lys nor Met post rumen infusions modified significantly milk yield (29.8±5.7 kg/d). The response of dMPY to both LysDI and MetDI was calculated as dMPC multiplied by milk yield estimated in response to NEL and PDI (RMSE=23 g/d on Lys and Met post rumen infusions database). The RMSEp of dMPC and dMPY obtained using a 2nd dataset for validation (38 exp of rumen protected Met lower and higher than 1.9 and 89 treat) were 0.36 g/kg and 22 g/d, respectively. If the dMPY response is added to the predicted MPY response to NEL and PDI (on a 3rd dataset using 105 exp on rumen protected Met and on dietary Lys supplies and 218 treat): MPYobserved = 8 (±13) + 1 (±0.01) × MPYpredicted with a slope not significantly different from 1. The RMSEp (42 g/d) was twice lower than if MPY was predicted using only NEL and PDI (RMSEp=87 g/d). Combining the effects of LysDI (% PDI) and MetDI (% PDI) to PDI and NEL improves MPY prediction.