Predicting Probability of Corn Response to Nitrogen Rates Using Bayesian Hierarchical Models.

Estimating the probability of profitable yield response to N rates in different N management practices for corn (Zea mays L.) reduces uncertainties in N management. We conducted 110 replicated on-farm strip trials in Ohio and Indiana in 2009, 2010, 2012, and 2013. The strips were completely randomized in each trial with four N rates (112, 168, 224, and 280 kg N ha^-1) and three replications. Yield changes between two rates (112 vs 168 kg N ha^-1, 168 vs 224 kg N ha^-1, and 224 vs 280 kg N ha^-1), were calculated for areas within two strips that were 25 m in length.  Bayesian hierarchical models will be used to estimate profitable yield changes in different management practices such as rotation, tillage, N form and timing. The models will also evaluate how spring rainfall, soil drainage, and soil organic matter content affect the probability of profitable yield change. The number of trials in this dataset is not sufficient to provide reliable estimates in the future for all fields in the agroecosystem where the data were collected. However, when sufficient trials are available the methodology used to estimate probabilities of a profitable yield change could identify best management practices in this agroecosystem and improve decisions on N management practices.