Abstract There is increased interest in precision-feeding systems that account for individual-animal variation and temporal dairy herd dynamics to increase feed-efficiency and reduce environmental footprints. The Ruminant Farm Systems (RuFaS; Hansen et al., 2021) model simulates individual animals on a daily timestep using a stochastic, Monte Carlo framework to represent the phenotypic diversity and population dynamics expected in a dairy herd. Within RuFaS, individual animal nutrient requirements are used to inform pen-level diet formulations and feed delivery. However, nutritional requirements for feeding dairy cattle were recently updated via consideration of both animal and diet-related factors. Our aim was to calculate individual animal predictions for dry matter intake (DMI), as well as energy and metabolizable protein requirements, and compare the methods described in the 7th (NRC, 2001) and 8th (NASEM, 2021) editions of the Nutrient Requirements for Dairy Cattle. For that purpose, we simulated dairy herds (ca. 6,000 individual Holstein animals per simulation, mean adult body weight = 671 kg, standard deviation = 79, mean daily milk production = 39.8 kg·day-1·cow-1 , standard deviation = 7.5) for 10+ year time periods with an updated version of RuFaS. The simulated herds provided a wide range of values with which to test the nutrient requirement models, and their implications. The NASEM DMI estimates were less across the board but began to converge towards NRC calculated values at the larger simulated animal body weights. The principal differences in energy requirements between the two models (NRC vs. NASEM) were: 1) predicted requirements for pregnancy were significantly greater for NASEM than NRC across all life history stages; 2) NASEM displayed slightly greater energy requirements for growth in first and second lactation cows, and; 3) NASEM predicted decreased maintenance requirements for all heifer stages but the opposite was true for mature cows. Metabolizable protein requirements were greater following NASEM calculations in heifers over 300 kg, but strikingly similar to NRC-computed values for lactating cows. Overall, results from this modeling study align with our expectations and demonstrate successful implementation of the two nutrition models within the RuFaS model. These outputs confirm the usefulness of data obtained to explore applications of the two nutrient requirement models under different precision feeding management practices and, to some extent, to quantitatively evaluate their viability of the models in edge cases that can occur within the expected variation of a herd’s population. Future work will explore downstream consequences of the nutrient requirement models within the automated, least-cost ration formulation framework using nonlinear programming and metrics related to manure output and storage. These data will guide further developments of RuFaS as whole farm simulation model and provide insights into expected variation in nutritional requirements within a dynamic set of animals.
Read full abstract