Contextualizing Wetlands Within a River Network to Assess Nitrate Removal and Inform Watershed Management

Abstract

Aquatic nitrate removal depends on interactions throughout an interconnected network of lakes, wetlands, and river channels. Herein, we present a network‐based model that quantifies nitrate‐nitrogen and organic carbon concentrations through a wetland‐river network and estimates nitrate export from the watershed. This model dynamically accounts for multiple competing limitations on nitrate removal, explicitly incorporates wetlands in the network, and captures hierarchical network effects and spatial interactions. We apply the model to the Le Sueur Basin, a data‐rich 2,880 km2 agricultural landscape in southern Minnesota and validate the model using synoptic field measurements during June for years 2013–2015. Using the model, we show that the overall limits to nitrate removal rate via denitrification shift between nitrate concentration, organic carbon availability, and residence time depending on discharge, characteristics of the waterbody, and location in the network. Our model results show that the spatial context of wetland restorations is an important but often overlooked factor because nonlinearities in the system, e.g., deriving from switching of resource limitation on denitrification rate, can lead to unexpected changes in downstream biogeochemistry. Our results demonstrate that reduction of watershed‐scale nitrate concentrations and downstream loads in the Le Sueur Basin can be most effectively achieved by increasing water residence time (by slowing the flow) rather than by increasing organic carbon concentrations (which may limit denitrification). This framework can be used toward assessing where and how to restore wetlands for reducing nitrate concentrations and loads from agricultural watersheds.

High‐Frequency Sensor Data Reveal Across‐Scale Nitrate Dynamics in Response to Hydrology and Biogeochemistry in Intensively Managed Agricultural Basins

“…Watersheds where stored nitrogen was already in the form of NO 3 − , for example, in groundwater as opposed to soil organic matter, would presumably have less lag between peak NO 3 − and peak Q although the effect of such a system on hydrology is not known. A more slowly changing hydrograph, such as that expected for a landscape with significant water storage or river network branching, would result in an increased Q at higher NO 3 − , as the movement of water is slowed down, although this could be offset by an overall decreased NO 3 − due to increased water residence time and in‐stream NO 3 − removal (Czuba et al, ).…”

The Power of Environmental Observatories for Advancing Multidisciplinary Research, Outreach, and Decision Support: The Case of the Minnesota River Basin

“…In the Le Sueur, for instance, we were able to combine the earlier work on sediment sources with intensive biogeochemical data collection to look at the role of sediment on phosphorus sources and transformations (Baker, ). We also focused our most intense modeling efforts in the Le Sueur (Cho et al, , ; Czuba et al, , ). Lastly, partnering with the IML‐CZO increases both the ability to compare results from the MRB with other intensively managed agricultural watersheds and the potential to link discoveries in critical zone research from the IML‐CZO field sites with the most intensive monitoring (Upper Sangamon River in Illinois and Clear Creek in Iowa) back to the MRB.…”

“…The opportunity to collect and analyze a wide array of environmental data in a large, intensively managed watershed has already led to scientific insights ranging from quantifying changes in hydrologic connectivity and system response (Foufoula‐Georgiou et al, ), to identifying the importance of hydrologic change on erosion of near‐channel sediment sources and associated nutrients (Baker, ; Kelly & Belmont, ; Vaughan et al, ), and to highlighting the ways in which biogeochemical processes can alter pollutant export behavior (e.g., Czuba et al, ; Hansen et al, ). Our research has direct implications in large agricultural watersheds where the source of water quality impairments (e.g., fine sediment, nitrate, or phosphorus) is spatially complex and may involve the interference of multiple stressors on the landscape (i.e., intensive agriculture, climate change, and loss of wetlands).…”

“…The wide spatial and temporal coverage of these data sets enabled testing of novel hypotheses about the watershed‐scale influences of wetlands on downstream nitrate concentrations (i.e., Figures c and d) under different flow conditions (Hansen, Dolph, & Finlay, ; Hansen et al, ) with results contrasting from other empirical watershed‐scale studies in part because of the wide variation in wetland and shallow lake cover in the MRB (Hansen et al, ; Strayer et al, ). The data sets were used in the development of a process‐based river network framework for predicting nitrate and dissolved organic carbon concentrations as a function of the location and specific attributes of wetlands (Czuba et al, ). The data set also allowed for quantification of multiple varied biophysical processes important to nutrient spiraling and transport in streams and rivers of our study region, including phosphorus sorption‐desorption and the role of sediment on modulating phosphorus form and bioavailability (Baker, ), and algal assimilation (Dolph, Hansen & Finlay, ).…”

“…Many of these modeling efforts represent the culmination of efforts to merge interdisciplinary data sets detailed above to gain new insights into watershed‐scale processes. As channel network structure impacts the distribution and structure of ecosystem processes and functions (Benda et al, ; Campbell Grant et al, ; Carrara et al, ), these models allow exploration of critical hot spots where fluxes accumulate or transform (Czuba et al, ; Czuba & Foufoula‐Georgiou, ) and help provide insight into ecosystem processes in heterogeneous dendritic networks in IML (McCluney et al, ). We describe four examples to highlight the wide range of scientific questions that can be addressed with data‐driven models, from bedload transport in networks to mussel population dynamics, and to enable other researchers to find all of the different models that were guided and validated by the collected data sets.…”