Baltimore Ecosystem Study Cary Institute of Ecosystem Studies

2012 BES Annual Meeting Presentation and Poster Abstracts

Effects of integrated stormwater management and stream engineering on nitrogen uptake and denitrification in streams
Newcomer, Tammy
Co-Authors: Tamara A. Newcomer, Sujay S. Kaushal, Paul M. Mayer, Peter M. Groffman, and Melissa M. Grese

Abstract: Restoring urban infrastructure and managing the N cycle represent major challenges for biogeochemistry and society. We investigated how stormwater best management practices (BMPs) integrated into urban stream networks can influence removal of N pollution. We hypothesized that stormwater BMPs are greater “hot spots” for N removal through denitrification than connected floodplain areas because they have ample organic carbon, low dissolved oxygen levels, and high residence time. We tested this hypothesis by examining N cycling in 2 urban stream networks with stormwater BMPs and a forested reference watershed with a pond at the Baltimore Long-Term Ecological Research (LTER) site. At all 3 sites, we used a combination of: (1) 250 stream reach scale mass balances of N conducted monthly for 2 years across streamflow (2) 6 in-stream tracer injection studies to measure seasonal nitrate uptake and groundwater inputs, and (3) 72 15N in situ push-pull tracer experiments to measure seasonal N removal via denitrification in stormwater BMPs and floodplain features. The stormwater BMPs at Spring Branch were inline wetlands installed below a storm drain outfall and at Gwynns Run were a wetland and wet pond configured in an oxbow to receive water during high flow events. As hypothesized, total dissolved nitrogen (TDN) concentrations decreased consistently across sampling dates as water traveled through stormwater BMPs; TDN concentrations decreased from 3.13 ± 0.67 mg/L to 1.87 ± 0.23 mg/L (mean ± SE) at Spring Branch and from 3.15 ± 0.15 mg/L to 1.47 ± 0.22 mg/L at Gwynns Run. Contrary to our hypothesis, mean TDN retention at Spring Branch was higher in a stream reach with connected floodplains, 2.01 ± 0.77 kg/day (mean ± SE), than in the stormwater BMPs, 0.053 ± 0.025 kg/day. Similarly, at Gwynns Run, mean TDN retention (and export) were 3 orders of magnitude higher in the stream reaches, 2.00 ± 1.6 kg/day (mean ± SE), than in the stormwater BMPs, 0.005 ± 0.597 kg/day. Mass balances and in-stream tracer injections produced consistent results for TDN uptake (µg N·m-2·s-1) and both methods demonstrated that groundwater contributed 30-70% of water to the urban streams. Surprisingly, mean 15N in situ denitrification rates were significantly higher in connected floodplain areas of the streams, 133.8 ± 32.4 µg N·kg soil-1·day-1 (mean ± SE), than in stormwater BMPs or the reference pond, 79.2 ± 22.1 µg N·kg soil-1·day-1. Our forest reference site had lower rates of mass uptake, export, and denitrification than urban sites—likely due to low N concentrations. We learned that floodplains can be important “hot spots” for N removal at a watershed and stream network scale; this is likely because these areas receive perennial flow through the groundwater-surface water interface during baseflow and storm events while the BMPs receive intermittent flow associated only with storm events. Future studies of N removal aimed at evaluating effectiveness of engineered features should consider the importance of groundwater-surface water interactions at the watershed scale.