Measurement and modeling of denitrification in sand-bed streams of varying land use

Abstract

Processes that govern transport and transformation of aquatic nitrogen are of growing importance due to increases in anthropogenic nitrogen input from fertilizer application and fossil fuel combustion. Denitrification, the incremental reduction of soluble nitrate to gaseous end products, is the main pathway in which nitrogen is biologically removed from aquatic ecosystems. In this study denitrification is measured from sediment cores in five streams in central Minnesota, USA, using denitrification enzyme activity (DEA) assays as well as microbiological techniques including the amplification of nirS gene fragments through qPCR. Hydraulic and environmental variables are measured in the vicinity of the sediment cores to determine a possible mediating influence of fluid flow and chemical variables on denitrification activity. Denitrification rates measured using DEA analysis with amended nutrients ranged from 0.02-10.1 mg-N m-2 hr-1. Denitrification rates measured without amended nutrients were a factor of 5.35 less on average and ranged from 0.03-0.98 mg-N m-2 hr-1. The abundance of the denitrifier gene nirS was positively correlated with denitrification potential measurements (R2 = 0.79, P < 0.001) for most of the streams studied. NirS distribution in one of the sites, a field scale experimental stream called the Outdoor StreamLab, followed the spatial distribution of benthic organic matter closely along the sediment bed and through the sediment column. Predictive models to determine nitrate uptake via denitrification were derived from hydraulic, morphologic and water quality variables. The first used hydraulic data collected over three summers in the Outdoor StreamLab. A Gaussian-type function was fit to these data and was dependent on fluid flow and channel characteristics within the stream system. The second model was derived following dimensional analysis on data from the Outdoor StreamLab and four other natural streams of varying watershed and in-stream conditions. This predictive model integrated not only stream hydraulic data but also environmental, morphological and DEA measurements for nutrient-amended and unamended samples. The proposed model explained 75% and 60% of the variability in amended and unamended DEA rates, respectively. Results from this study verify that denitrification is ubiquitous across varying stream systems but is most dependent on the distribution of sediment organic matter and interstitial pore space as well as stream hydraulic characteristics.