Hydrological and Water Quality Dynamics under Land Use Change in Intensively Managed Agricultural Watersheds: Integrated Approaches for Sustainable Water and Land Management

Abstract

Land use and land cover changes in agriculturally intensive regions are increasingly recognized as critical drivers of hydrological and water quality dynamics, driven by rising global food demand. In South America, the rapid expansion of soybean cultivation has reshaped many basins into intensively managed agricultural systems, posing significant challenges for reconciling agricultural production with water resource sustainability, yet the complexity of these responses under interacting climatic and management pressures remains insufficiently understood. This dissertation develops an integrated modeling framework that synthesizes long-term hydrological and, in particular, water quality observations with crop-specific land use data, nitrogen (N) budget assessments, and ensemble machine learning to disentangle these coupled impacts and establish a transferable basis for sustainable land and water management. Rapid agricultural intensification in the past four decades has substantially altered land and water dynamics in southern Brazil, with long-term monitoring records showing a coupled pattern of declining flows and increasing pollutant concentrations. Water quality deterioration indicated an early-warning indicator of land use impacts. Land use-climate interactions, particularly soybean fraction with precipitation, strongly influence these trajectories. Assessments of N balances in the La Plata Basin demonstrated sustained N surpluses within soybean-dominated watersheds, with the large inputs derived from biological N fixation and fertilizer applications. Fertilizer inputs have risen faster than non-fertilizer sources, increasing N surplus with high loss potential. About one-third of soybean areas were identified as critical zones for N export risk. Scenario simulations indicate that targeted N surplus reductions in these hotspots can lower riverine N concentrations without compromising yields, offering a feasible pathway for nutrient mitigation. Nutrient source attribution identified fertilizer type as a major determinant of stream N concentrations across 285 watersheds in southern Brazil. Ensemble machine learning analysis attributed approximately half of the observed N to synthetic fertilizer inputs and about one-third to two-fifths to manure inputs, with crop composition further modulating these contributions. These findings underscore the importance of integrating land use planning with nutrient source management to achieve sustained water quality improvements. By linking crop-specific land use patterns, climate variability, and fertilizer sources, this work advances mechanistic understanding of watershed-scale nutrient dynamics. The framework offers a robust, transferable tool for diagnosing ecohydrological risks and guiding targeted, evidence-based strategies to mitigate nutrient pollution while sustaining agricultural productivity under concurrent land use and climate pressures.