Curtis J. Ransom https://orcid.org/0000-0002-1268-7247
Newell R. Kitchen https://orcid.org/0000-0003-2268-9016
Kristen S. Veum https://orcid.org/0000-0002-6492-913X
James J. Camberato https://orcid.org/0000-0002-6677-854X
Fabián G. Fernández https://orcid.org/0000-0002-9539-0050
David W. Franzen https://orcid.org/0000-0003-4862-8086
Carrie A. M. Laboski https://orcid.org/0000-0003-2050-767X
Emerson D. Nafziger https://orcid.org/0000-0002-6578-1624
Robert L. Nielsen https://orcid.org/0000-0002-7673-1681
Date of this Version
Agronomy Journal. 2021;113:5541–5555.
Nitrogen fertilizer recommendations in corn (Zea mays L.) that match the economically optimal nitrogen fertilizer rate (EONR) are imperative for profitability and minimizing environmental degradation. However, the amount of soil N available for the crop depends on soil and weather factors, making it difficult to know the EONR from year-to-year and from field-to-field. Our objective was to explore, within the framework of hydrologic soil groups and drainage classifications (HGDC), which site-specific soil and weather properties best estimated corn N needs (i.e., EONR) for two application timings (at-planting and side-dress). Included in this investigation was a validation step using an independent dataset. Forty-nine N response trials conducted across the U.S. Midwest Corn Belt over three growing seasons (2014–2016) were used for recommendation model development, and 181 independent site-years were used for validation. For HGDC models, soil organic matter (SOM), clay content, and evenness of rainfall distribution before side-dress N application were the properties generally most helpful in predicting EONR. Using the validation data, model recommendations were within 34 kg N ha–1 of EONR for 37 and 42% of the sites with a root mean square error (RMSE) of 70 and 68 kg N ha–1 for at-planting and side-dress applications, respectively. Compared to state-specific recommendations, sites needing ha–1 or no N were better estimated with HGDC models. In contrast, for sites where EONR was >150 kg N ha–1, HGDC models underestimated N needs compared to state specific. These results show HGDC groupings could aid in developing tools for N fertilizer recommendations.