Upland and in-stream controls on baseflow nutrient dynamics in tile-drained agroecosystem watersheds

Abstract

In landscapes with low residence times (e.g., rivers and reservoirs), baseflow nutrient concentration dynamics during sensitive timeframes can contribute to deleterious environmental conditions downstream. This study assessed upland and in-stream controls on baseflow nutrient concentrations in a low-gradient, tile-drained agroecosystem watershed. We conducted time-series analysis using Empirical mode decomposition of seven decade-long nutrient concentration time-series in the agricultural Upper Big Walnut Creek watershed (Ohio, USA). Four tributaries of varying drainage areas and three main-stem sites were monitored, and nutrient grab samples were collected weekly from 2006 to 2016 and analyzed for dissolved reactive phosphorus (DRP), nitrate-nitrogen (NO3-N), total nitrogen (TN), and total phosphorus (TP). Statistically significant seasonal fluctuations were compared with seasonality of baseflow, watershed characteristics (e.g., tile-drain density), and in-stream water quality parameters (pH, DO, temperature). Findings point to statistically significant seasonality of all parameters with peak P concentrations in summer and peak N in late winter-early spring. Results suggest that upland processes exert strong control on DRP concentrations in the winter and spring months, while coupled upland and in-stream conditions control watershed baseflow DRP concentrations during summer and early fall. Conversely, upland flow sources driving streamflow exert strong control on baseflow NO3-N, and in-stream attenuation through transient and permanent pathways impacts the magnitude of removal. Regarding TN and TP, we found that TN was governed by NO3-N, while TP was governed by DRP in summer and fluvial erosion of P-rich benthic sediments during higher baseflow conditions. Findings of the study highlight the importance of coupled in-stream and upland management for mitigating eutrophic conditions during environmentally sensitive timeframes.