Using the cornell net carbohydrate and protein system model to evaluate the effects of variation in maize silage quality on a dairy farm.

Abstract

AbstractForage analysis information is used for developing rations, crop rotations and manure nutrient management plans in the Cornell University Nutrient Management Planning System (CuNMPS). The CuNMPS and other similar models use as inputs steady-state conditions on the farm (milk production, herd size, group dynamics, diets, feed composition, farm size and crop production). This steady-state assumption may result in nutrient management plans with large errors, which may introduce risks in income and production variability. One source of error in assuming steady-state conditions is variation in feed composition. The CuNMPS and other nutrient management planning models used to develop nutrient management plans typically rely on either historical forage analysis or infrequently measured compositional values. Methods for incorporating variance have been proposed; however, none have been implemented. The objectives of this study were to determine the amount of variation in a home-raised forage, how this variation affects predicted performance and nutrient excretion by lactating cows and whether this variation can be accounted for in a field-level model, such as the Cornell Net Carbohydrate and Protein System (CNCPS). Each load of maize (Zea mays) harvested for silage was sampled at harvest and analysed for neutral detergent fibre (NDF) and dry matter (DM) on a commercial 500-cow dairy farm. Averages (49.4 and 26.1% for NDF and DM, respectively) and standard deviations (4.52 and 3.59 units for NDF and DM, respectively) along with feeding error (±3% of formulated as fed amounts) were evaluated with the CNCPS (version 4.0). Results indicate that this range in NDF, DM and feeding error caused predicted variation in income over feed costs ($40,000 year-1), feed requirement (73 Mt DM maize silage year-1) and N and P excretion (110 and 37 kg year-1of N and P, respectively) per 100 cows annually. Accounting for this variation in current and future nutrient management models is required to optimize returns over feed costs, while decreasing nutrient excretion.