Predicting Risk from Reducing Nitrogen Fertilization Using Hierarchical Models and On‐Farm Data

Abstract

Current systems for developing N recommendations for corn (Zea mays L.) lack methods to quantify the effects of factors influencing yield responses to N and quantify the uncertainty in N recommendations. We utilized hierarchical modeling and Bayesian analysis to quantify the risk from reducing N to corn using on‐farm observations. Across Iowa, farmers conducted 34 trials in 2006 and 22 trials in 2007. Each trial had a farmer’s normal N rate alternating with a reduced rate (by about 30% less) in three or more replications. Yield losses (YLs) from reduced N were calculated at 35‐m intervals. Posterior distributions were used to identify across‐field and within‐field factors affecting YL and to quantify the risk of economic YL (>0.31 Mg ha−1) from reducing N in unobserved fields. In 2006 (dry May and June), the economic YL for corn after soybean (C‐S) was predicted to be 20% larger than that for corn after corn. Also in 2006, C‐S fields with above‐normal June rainfall had economic YLs 35% larger than those with below‐normal June rainfall, and sidedress applications were about 20% riskier than spring applications. In 2007 for C‐S, N reductions with above‐normal spring rainfall were riskier than with below‐normal spring rainfall. Areas with higher soil organic matter (SOM) had economic YLs about 20% smaller than those with lower SOM. Many on‐farm trials can be conducted across the state and the use of the proposed statistical methodology can improve decisions on where to reduce N applications across and within fields.

Strengths and Limitations of Nitrogen Rate Recommendations for Corn and Opportunities for Improvement

“…A typical adaptive management method to obtain robust evaluations of N management at the field level to fine-tune N practices is to perform replicated strip trials comparing either two or more rates of N or different forms, timing or placement of N. Two treatment replicated strip trials, usually the farmer N rate or an N rate from a recommendation system like the yield goal system or the MRTN system, and a rate that is 45 kg N ha -1 greater or less than the farmer or recommendation system rate, enable farmers to refine N recommendations (Kyveryga et al, 2013). Refinement of N recommendations on an individual field basis is clearly needed for all recommendation systems based on the information provided throughout this paper.…”

Probability of profitable yield response to nitrification inhibitor used with liquid swine manure on corn

“…In our study, the general hierarchical model comprised three levels: data, field, and regional (Kyveryga et al 2013). The data model included a distribution of YR values (YR nj ) within a trial (where n = 1,…,N), for an individual grid cell (where j = 1,…,M) to be normally distributed with a mean l n and a precision k n unique for that trial.…”

“…All calculations were done using the statistical software R (R Development Core Team 2009). More details for using Hierarchical models and Bayesian statistics to analyze data collected from OETs can be found in Kyveryga et al (2013).…”

“…conducted a large-scale on-farm evaluation study to develop procedures for calibrating commonly used N diagnostic tools for measuring lateseason corn N status. Subsequently, Kyveryga et al (2013) identified factors that influence the risk associated with reducing normal N fertilizer application rates by about one-third across Iowa. The objective of this study was to utilize yield monitoring and GPS technologies in OET to identify field and within-field factors for which the use of Instinct (an encapsulated form of nitrapyrin) with LSM might be profitable in spatially variable corn fields in Iowa.…”

Assessing and modeling economic and environmental impact of wheat nitrogen management in Belgium

“…In that way, in Belgium, farmers used to provide N in sufficient and reasonable quantities (60 kg N ha À1 ) at the tiller, stem extension, and flag leaf stages, which relate to Zadok stages 23, 30, and 39, respectively. In recent studies, Dumont et al (2014aDumont et al ( , 2014cDumont et al ( , 2015b successfully transposed the theory of yield distribution analysis (Day, 1965;Kyveryga et al, 2013) onto the study of crop model simulations. They demonstrated that the probability of achieving yields greater than the mean of the corresponding distribution was the highest when Belgian farmers' most typical N rate was applied.…”