A DSSAT-Based Framework for Site-Specific Nitrogen Management in Winter Wheat: Evaluating Long-Term Weather Effects on Nitrogen Loss Variability Across Yield Zones

Abstract

Nitrogen (N) fertilizers are essential for sustaining the productivity of winter wheat in humid agroecosystems; yet their management poses a significant dual challenge: optimizing farm profitability while reducing environmental impact. In the southeastern U.S., heavy winter rainfall often exceeds evapotranspiration, resulting in substantial N losses through leaching and denitrification, which degrade water quality and reduce N use efficiency (NUE). Although the widespread adoption of yield zones, the fate and flow of N within these contrasting zones remain inadequately measured, particularly under long-term climatic variability. The goal of this study was to systematically quantify N dynamics and environmental losses (Nenvloss) across contrasting high-yielding (HYZ) and low-yielding (LYZ) zones and subsequently develop integrated, sitespecific management strategies to balance conflicting economic and environmental objectives. Due to the difficulty in the direct measurement of subsurface fluxes, this study utilized the DSSAT CSM-CERES-Wheat model to simulate crop-soil-weather interactions and associated N dynamics across contrasting yield zones. The model calibration (2022-23) and evaluation (2023-24) using two distinct seasons of field-observed data validated the model’s reliability in simulating winter wheat growth, yield, and soil N balance. The simulations identified nitrate (NO3) leaching as the dominant loss pathway, accounting for 94-98% of total losses. A distinct yield zone pattern emerged, where HYZ demonstrated higher leaching potential (39-98 kg[N] ha-1) compared to the LYZ (69-70 kg[N] ha-1). This was primarily driven by higher initial soil mineral N (Nmineral) and enhanced soil permeability rather than differences in crop N uptake. Moreover, weather variability significantly influenced these dynamics, with leaching losses increasing by 48% in wet years compared to drought years, highlighting the inadequacy of static nutrient management recommendations in humid regions. This study further explored integrated strategies to address the trade-offs between yield maximization and environmental sustainability. A systems analysis was conducted to optimize both planting dates and N management strategies by integrating the CSM-CERES-Wheat model with a Multi-Criteria Decision Analysis (MCDA) framework over 32 years. The results indicated that optimizing the planting window to mid-October significantly increased yield (21-28%) and reduced Nenvloss (14-19%) compared to late planting. The hybrid Analytical Hierarchy Process-Test for Order of Preference by Similarity to Ideal Solution (AHPTOPSIS) analysis revealed that the farmer’s baseline strategy (T1: 130 kg[N] ha-1 UAN + 5 Mg ha-1 PL-Fall) was economically inefficient despite a higher yield (3.7-3.8 Mg ha-1). The optimal nutrient management strategies varied by yield zone. In the high‑yield zone (HYZ), profitability was maximized with a split application of UAN at 130 kg N ha⁻¹ (T4). In contrast, the low‑yield zone (LYZ) benefited most from a 25% reduction in UAN (to 130 kg N ha⁻¹) combined with fall‑applied poultry litter at 5 Mg ha⁻¹ (T6). This integrated approach increased profitability by 36– 70% (equivalent to $68–$146 ha⁻¹) while simultaneously reducing environmental nitrogen losses (Nenvloss) by 4–8 kg N ha⁻¹. Overall, this comprehensive study emphasized the critical importance of precision agriculture, indicating that Nenvloss in winter wheat was primarily influenced by soil hydrological properties and climatic variability rather than yield potential. These findings support a paradigm shift from uniform, yield-based inputs to weather-responsive, site-specific strategies, providing a robust framework for enhancing NUE, sustaining high productivity, and minimizing the environmental impact of agricultural systems in high-rainfall regions. Artificial Intelligence (AI) Use Disclosure Statement In the preparation of this thesis/dissertation, the following Artificial Intelligence (AI) tools were used: Grammarly and Gemini. These tools were used primarily to assist with grammatical editing and support in coding for data analysis. The author acknowledges full responsibility for the intellectual content of this work and has ensured that all AI-assisted sections have been reviewed and revised for accuracy and appropriate academic style. All AI-generated content was reviewed and validated for relevance, appropriateness, and accuracy before incorporation into the final document to maintain the scholarly integrity of this research.