Core Ideas The geographic scope, scale, and unique collaborative arrangement warrant documenting details of this work. The purpose of this article is to describe how the research was undertaken, reasons for the research methods, and the project's potential value. The project generated a valuable dataset across a wide array of weather and soils that allows evaluation of N decision tools. Due to economic and environmental consequences of N lost from fertilizer applications in corn (Zea mays L.), considerable public and industry attention has been devoted to the development of N decision tools. Needed are research and databases and associated metadata, at numerous locations and years to represent a wide geographic range of soil and weather scenarios, for evaluating tool performance. The goals of this research were to conduct standardized corn N rate response field studies to evaluate the performance of multiple public‐domain N decision tools across diverse soils and environmental conditions, develop and publish new agronomic science for improved crop N management, and train new scientists. The geographic scope, scale, and unique collaborative arrangement warrant documenting details of this research. The objectives of this paper are to describe how the research was undertaken, reasons for the methods, and the project's anticipated value. The project was initiated in a partnership between eight U.S. Midwest land‐grant universities, USDA‐ARS, and DuPont Pioneer. Research using a standardized protocol was conducted over the 2014 through 2016 growing seasons, yielding a total of 49 sites. Preliminary observations of soil and crop variables measured from each site revealed a magnitude of differences in soil properties (e.g., texture and organic matter) as well as differences in agronomic and economic responses to applied N. The project has generated a valuable dataset across a wide array of weather and soils that allows investigators to perform robust evaluation of N use in corn and N decision tools.

Determining which corn (Zea mays L.) N fertilizer rate recommendation tools best predict crop N need would be valuable for maximizing profits and minimizing environmental consequences. Simultaneous comparisons of multiple tools across various environmental conditions have been limited. The objectives of this research were to evaluate the performance of publicly‐available N fertilizer recommendation tools across diverse soil and weather conditions for: (i) prescribing N rates for planting and split‐fertilizer applications, and (ii) economic and environmental effects. Corn N‐response trials using standardized methods were conducted at 49 sites, spanning eight US Midwest states and three growing seasons. Nitrogen applications included eight rates in 45 kg N ha−1 increments all at‐planting and matching rates with 45 kg N ha−1 at‐planting plus at the V9 development stage. Tool performances were compared to the economically optimal N rate (EONR). Over this large geographic region, only 10 of 31 recommendation tools (mainly soil nitrate tests) produced N rate recommendations that weakly correlated to EONR (P ≤ .10; r2 ≤ .20). With other metrics of performance, the Maximum Return to N (MRTN) soil nitrate tests, and canopy reflectance sensing came close to matching EONR. Economically, all tools but the Maize‐N crop growth model had similar returns compared to EONR. Environmentally, yield goal based tools resulted in the highest environmental costs. Results show that no tool was universally reliable over this study's diverse growing environments, suggesting that additional tool development is needed to better represent N inputs and crop utilization at a larger regional level.

Determination of in-season N requirement for corn (Zea mays L.) is challenging due to interactions of genotype, environment, and management. Machine learning (ML), with its predictive power to tackle complex systems, may solve this barrier in the development of locally based N recommendations. The objective of this study was to explore application of ML methodologies to predict economic optimum nitrogen rate (EONR) for corn using data from 47 experiments across the US Corn Belt. Two features, a water table adjusted available water capacity (AWC wt ) and a ratio of in-season rainfall to AWC wt (RAWC wt ), were created to capture the impact of soil hydrology on N dynamics. Four ML models-linear regression (LR), ridge regression (RR), least absolute shrinkage and selection operator (LASSO) regression, and gradient boost regression trees (GBRT)-were assessed and validated using "leave-one-location-out" (LOLO) and "leave-one-year-out" (LOYO) approaches. Generally, RR outperformed other models in predicting both at planting and split EONR times. Among the 47 tested sites, for 33 sites the predicted split EONR using RR fell within the 95% confidence interval, suggesting the chance of using the RR model to make an acceptable prediction of split EONR is ~70%. When RR was used to test split EONR prediction with input weather features surrogated with 10 yr of historical weather data, the model demonstrated robustness (MAE, 33.6 kg ha -1 ; R 2 = 0.46). Incorporating mechanistically derived hydrological features significantly enhanced the ability of the ML procedures to model EONR. Improvement in estimating in-season soil hydrological status seems essential for success in modeling N demand.

Nitrogen (N) fertilizer recommendation tools could be improved for estimating corn (Zea mays L.) N needs by incorporating site-specific soil and weather information. However, an evaluation of analytical methods is needed to determine the success of incorporating this information. The objectives of this research were to evaluate statistical and machine learning (ML) algorithms for utilizing soil and weather information for improving corn N recommendation tools. Eight algorithms [stepwise, ridge regression, least absolute shrinkage and selection operator (Lasso), elastic net regression, principal component regression (PCR), partial least squares regression (PLSR), decision tree, and random forest] were evaluated using a dataset containing measured soil and weather variables from a regional database. The performance was evaluated based on how well these algorithms predicted corn economically optimal N rates (EONR) from 49 sites in the U.S. Midwest. Multiple algorithm modeling scenarios were examined with and without adjustment for multicollinearity and inclusion of two-way interaction terms to identify the soil and weather variables that could improve three dissimilar N recommendation tools. Results showed the out-of-sample root-mean-square error (RMSE) for the decision tree and some random forest modeling scenarios were better than the stepwise or ridge regression, but not significantly different than any other algorithm. The best ML algorithm for adjusting N recommendation tools was the random forest approach (r 2 increased between 0.72 and 0.84 and the RMSE decreased between 41 and 94 kg N ha −1). However, the ML algorithm that best adjusted tools while using a minimal amount of variables was the decision tree. This method was simple, needing only one or two variables (regardless of modeling scenario) and provided moderate improvement as r 2 values increased between 0.15 and 0.51 and RMSE decreased between 16 and 66 kg N ha −1. Using ML algorithms to adjust N recommendation tools with soil and weather information shows promising results for better N management in the U.S. Midwest.

Controlled-release and slow-release fertilizers can effectively supply nitrogen (N) while mitigating N loss. To determine the suitability of these fertilizers for plants in semi-arid environments, these fertilizers need to be evaluated under varying placement and temperature conditions. Several urea fertilizers were evaluated, including: uncoated, sulfur-coated (SCU), polymer-coated-sulfur-coated (PCSCU), and polymer-coated (PCU) with projected release timings between 45 and 180 d. Nitrogen release was measured under daily fluctuating or static temperatures applied either to the surface or buried in the soil. A second experiment consisted of two PCU sources and added a hanging bag placement comparison and low and high soil moisture treatments. For the first Experiment, the N in uncoated urea released shortly after application. The SCU and PCSCU treatments released > 80% of the N before the first sampling date. With fluctuating temperatures, the PCU 45, 75, 120, and 180 incorporated into the soil released N within +9, +9,-22, and-68 d of their expected timing. However, they released their N within 35 d when surface applied. Conversely, with static temperatures, PCU products released slowly, releasing under 80% for the entire study. The second experiment verified these results and showed no difference between low and high moisture and minimal release with fertilizer not in contact with soil. Each coated fertilizer in these studies exhibited slow/control release properties, but the PCU (surface applied) and SCU/PCSCU (surface applied or incorporated in soil) release was much more rapid than expected. Our research suggests that, although the SCU and PCSCU showed minimal slow-release properties (regardless of placement), the PCU fertilizers incorporated in the soil do have a controlled release approximate to what is expected, but have a much more rapid release when surface applied.