Antunes de Almeida, Luiz Felipe2025-08-082025https://hdl.handle.net/2097/45224Improving the nitrogen (N) economy in soybeans (Glycine max L.) production is essential for developing more efficient, profitable, and environmentally sustainable agricultural systems. While soybeans can meet much of their N demand through biological N₂ fixation, uncertainty remains around when and where fertilizer inputs are needed and how environmental conditions influence N uptake, seed yield, and quality. This dissertation combines field data, crop modeling, and machine learning to provide data-driven insights into soybean N dynamics across diverse U.S. environments. Chapter 1 provides a summary and presentation of the background and objectives of this dissertation. Chapter 2 uses a standardized protocol to evaluate soybean yield responses to N and sulfur (S) fertilization across 26 field trials located across the Midwest region of the United States (US). Results showed that N fertilization rarely increased yield and often introduced uncertainty. In contrast, S fertilization improved N uptake, yield, and N status in some environments. An apparent N dilution curve was developed to help identify in-season plant N limitations, improve nutrient diagnosis, and provide more precise recommendations. Chapter 3 explores the effect of weather and soil factors that influence seed yield, N₂ fixation, and their uncertainty using data from 35 sites. Precipitation, vapor pressure deficit, and soil texture emerged as key factors explaining variability in both yield and N₂ fixation. Small additions of S helped improve N uptake and yield stability in environments with limited organic matter or water availability. Chapter 4 focuses on characterizing seasonal patterns of N₂ fixation using plant samples collected at reproductive stages and analyzed through Bayesian modeling. Peak N₂ fixation typically occurred between full pod and seed fill stages but varied widely depending on water availability and atmospheric conditions. These results highlight the limitations of relying on single-timepoint measurements and the value of time-series data for modeling complex biological processes. Together, the findings of this dissertation support more targeted, data-informed recommendations for improving yield and N use efficiency while delivering practical tools to help farmers, advisors, and researchers advance soybean productivity and long-term sustainability.en-USSoybean nitrogen fixationNitrogen balanceBayesian modelingNutrient management in soybean systemsSoil and weather interactionsQuantitative agronomyDissertation