Four decades of change: how woody encroachment, fire, grazing, and agricultural legacy shape hydrology, nutrients, and sediment in tallgrass prairie streams

Abstract

Grassland streams are among the most hydrologically variable and least monitored freshwater ecosystems on Earth. This dissertation uses up to four decades of continuous data from Konza Prairie Biological Station (KPBS), a tallgrass prairie Long-Term Ecological Research site in the Kansas Flint Hills, combined with rainfall simulation experiments in the riparian zone, to show that interacting changes in climate, vegetation, and land management are fundamentally transforming hydrology, nutrient biogeochemistry, and sediment dynamics of intermittent prairie streams. I analyzed long-term data on discharge, precipitation, temperature, and woody vegetation expansion to assess influences on hydrology between 1982 and 2020. Kings Creek's daily discharge decreased by over half and lost 55% of its annual flow days without a decline in precipitation. Riparian woody cover expanded two- to sevenfold across one- to four-year fire return intervals. This increase in woody cover, coupled with a 1.2°C temperature increase over the same period, increased evapotranspiration, with the strongest increases during the July to September period. Structural equation modeling confirmed that evapotranspiration is the primary pathway linking woody encroachment to streamflow decline. Stream nutrient concentrations responded to a hierarchy of drivers operating over different timescales. Agricultural legacy exerted the strongest and most persistent influence: stream nitrate below restored croplands was initially high (450 [mu]g N L⁻¹) and declined by 75% over 25 years as restoration progressed, but groundwater nitrate remained near 950 [mu]g N L⁻¹ more than two decades after crop fertilization ceased. Bison grazing doubled stream nitrate relative to ungrazed watersheds, an effect larger than that observed during the first few years after bison were introduced. Cattle elevated both nitrate and total phosphorus through direct riparian disturbance, and these effects were strongest when cattle were actively present on the pasture. Differences in fire frequency had minimal direct effects on nutrients but interacted with grazing to shape decadal trajectories. Suspended sediment dynamics were governed by a dual-pathway framework in which organic and inorganic fractions responded to different controls. Integration of 15 years of watershed monitoring (2009–2023), rainfall simulation experiments with bromide tracer (n = 10), and streambank geomorphic surveys revealed that discharge drove organic sediment (volatile suspended solids, VSS) through dilution, while fire management drove inorganic sediment (inorganic suspended solids, ISS) through mineral soil exposure. An area-normalized erodibility index (Sed/m²) revealed a nonlinear riparian vegetation cover threshold between approximately 60% and 70%, below which areal erodibility increased by approximately two orders of magnitude. Two high-cover sites (Exps 2 and 3) produced no detectable bromide breakthrough despite sub-annual storm depths, while bare bison-grazed soil was 168 times more erodible per unit area than an adjacent fully vegetated surface at matched slope. Generalized additive models (43–46% variance explained) and structural equation modeling confirmed that days since burn exerted the strongest direct effect on the VSS/ISS ratio, establishing prescribed fire frequency as the primary determinant of sediment composition. The VSS/ISS ratio was more sensitive to management than any individual fraction, with biennial burning under bison maintaining the most organic-dominated sediment regime and quadrennial burning producing the most inorganic-dominated. Streambank surveys corroborated the inorganic pathway, with woody cover as the strongest erosion predictor, and the watershed combining the longest fire interval with the lowest canopy cover exhibited the highest bank cut frequency. These findings demonstrate that effective management of grassland streams requires integrated, long-term approaches that account for the coupled dynamics of water, nutrients, and sediment under ongoing climate and vegetation change. Fire and grazing interactions regulate all three domains through nonlinear threshold responses, and temporal scales of years to decades are required to detect directional changes that shorter-term studies systematically underestimate. Prescribed fire frequency emerges as the most broadly effective management lever, while the prevalence of threshold behaviors underscores the importance of monitoring for proximity to critical transitions before they are crossed.