Building No-Tillage Maize Nitrogen Recommendation Algorithm that Considers Improvements in Soil Health

Author

Dwarika Bhattarai, South Dakota State UniversityFollow

Document Type

Dissertation - Open Access

Award Date

2023

Degree Name

Doctor of Philosophy (PhD)

Department / School

Agronomy, Horticulture, and Plant Science

First Advisor

David E. Clay

Abstract

In long-term no-till fields, farmers have reported that less N is required to optimize maize (Zea mays L.) yields in long term no-tillage fields than conventional tillage fields. These reductions may be attributed to improved soil health resulting from increasing soil organic matter, higher soil microbial activities, and improved water and nutrient use efficiency. However, the impact of soil health measurements on fertilizer-N requirement has not been determined. The objective of this dissertation was to compare different regional N recommendation models to measured values and develop a maize fertilizer-N recommendation model, using machine learning approaches, that includes adjustments based on soil health measurements. The research was conducted for three years at 16-dryland sites that were under no-tillage practice for at least 6-years. The effect of six N rates (0, 45, 90, 135, 180, and 224 kg N ha^-1 ) on maize grain yield was evaluated. Soil samples for nitrate-N (NO₃-N), ammonium-N (NH₄-N), pH, EC, and phospholipid fatty acid (PLFA) were collected from various depths before planting and after harvest. Climate variations influenced the maize yield across experimental sites. Comparison of error rates and bias showed that at lower cost/value ratios the current South Dakota and North Dakota N models had lower error rates and biases than models used in Nebraska, Iowa, and Minnesota. Further, using soil health measurements the support vector machine (SVM) algorithm outperformed several other machine learning algorithms for forecasting the soil yield potential. The top five predictor variables were total N, total C, growing degree days (GDD), soil microbial biomass, and bacterial biomass. The overall findings from this study suggested that soil organic C, total N, inorganic N, soil microbial biomass in addition to the climate variables, rainfall, and temperature, can be used to predict the soil yield potential.

Library of Congress Subject Headings

Corn -- Fertilizers.
Nitrogen fertilizers.
No-tillage.
Soils -- Quality.

Recommended Citation

Bhattarai, Dwarika, "Building No-Tillage Maize Nitrogen Recommendation Algorithm that Considers Improvements in Soil Health" (2023). Electronic Theses and Dissertations. 607.
https://openprairie.sdstate.edu/etd2/607