Examining the Inter-Storm Variability in Nitrogen Removal Efficiency in a Connecticut Constructed Wetland

Suburban stormwater runoff carries large amounts of nitrogen and other pollutants from fertilized lawns, septic systems, and atmospheric deposition, ultimately contributing to hypoxia (low dissolved oxygen levels) in Long Island Sound and severely perturbing aquatic life. Managers of Long Island Sound have set a goal of achieving a 10% reduction in nitrogen from nonpoint sources in order to mitigate hypoxia. More robust stormwater management practices need to be implemented to both attain this goal and keep pace with the increase in suburban sprawl. Wetlands may provide a valuable, but understudied, ecosystem service by removing nitrogen from this runoff. This study examined the nitrogen removal efficacy of a constructed wetland, located in a suburban area of Hamden, Connecticut. Its position directly upstream of the Lake Whitney public drinking water reservoir makes the quality of the runoff discharged from the site especially critical. Our objective was to evaluate the effectiveness of the constructed wetland at removing nitrogen from runoff, and to assess the inter-storm variability in nitrogen removal, in order to determine which variables (water temperature, residence time, and influent nitrogen concentration) might control this efficiency. The input and output of nitrogen, along with the flow of water, were measured at the site for 73 storm events from June 2011 to December 2011 and from April 2012 to December 2012. Stormwater samples were collected using programmable ISCO 3700 autosamplers over the duration of the storm and composited manually into flow-weighted samples for the influent and effluent in order to determine nitrogen event mean concentrations and loads for each storm event. Results show that the constructed wetland exhibited statistically significant nitrogen removal (nitrate: p=0.001, total nitrogen: p<0.001), indicating that the site is an effective tool for filtering nitrogen. However, nitrogen removal was highly variable from storm to storm. According to the general linear model, highest removal rates occurred with higher mean influent nitrogen concentrations for both nitrate (p=0.002) and total nitrogen (p<0.001), as well as with warmer temperatures (p=0.008) and longer residence times (p=0.022) for nitrate. Continued monitoring of Connecticut constructed wetlands can aid in ensuring more efficient reduction in nitrogen loads to Long Island Sound.