How Dairy Cattle Diet Formulation Can Deal with Uncertainty

Diet formulation models for dairy cattlerequire estimates of feed composition provided as table values or from feed analysis. In addition to feed composition, models use predicted milk production and body weight for when the ration will be offered, and internal constants such as digestibility coefficients for specific nutrients. Current models do not account for uncertainty of feed analysis, animal performance, or other internal constants; they simply overestimate requirements by applying “safety” margins or adjustments above estimated requirements to compensate for the risk of underfeeding. Optimal safety margins can be calculated by balancing the increased ration cost against the potential loss in milk income from the risk of underfeeding due to uncertainty and variation. For the previous 5-yr average milk and feed prices, the optimal safety margin for diet CP was 0.95 times the SE in predicted requirements and supply. The overall safety margin for multiple sources of uncertainty can be quantified as the sum of squared SE terms times the safety factor (e.g., 0.95) to account for uncertainty in feed analysis, animal production, intrinsic model uncertainty, and variation among animals (assuming no covariance among factors). Possible pitfalls in use of safety margins include failure to understand that variance of ration nutrient composition is less than the variance for individual feeds, and failure to square SE before adding. These mistakes result in overfeeding of nutrients beyond the economic optimum. Explicitly understanding the sources of uncertainty in diet formulation and feeding would enable more accurate compensation for uncertainty.