Predicting Risk of Reducing Nitrogen Fertilization Using Hierarchical Models and On-Farm Evaluation Data.

Current nitrogen (N) fertilizer recommendation systems for corn (Zea mays L.) do not consider factors that influence yield responses to N at different levels and the uncertainty in N recommendations. We used hierarchical modeling (HM) and Bayesian statistics to quantify the risk from reducing N applications in corn based on observations from on-farm evaluations trials. Across Iowa, farmers conducted 34 trials in 2006 and 22 trials in 2007. Each trial had a normal N rate alternating with a reduced N rate (about one third less) in >3 replications, and yield losses (YL) from reduced N were calculated at 35-m intervals. Posterior distributions were used to identify field and within-field level factors affecting YL, and to quantify the risk of economic YL from reduced N in unobserved fields. In 2006 (dry May and June), the economic YL for corn after soybean (C-S) was predicted >20% than for corn after corn. Also in 2006, C-S fields with above normal June rainfall had economic YL by 35% less than those with below normal June rainfall, and sidedress N applications were about 20% more riskier than spring applications. In 2007 for C-S, N reductions with above normal spring rainfall were more riskier than with below normal, and within-fields areas with higher SOM had economic YL by about 20% less than those with lower SOM. Many on-farm evaluation trials can conducted across the state to quantify the risk from reducing N and identifying management, weather, and soil factors that affect this risk at different levels.