Data Package Metadata   View Summary

We conducted a continental-scale gradient study of N2 fixation and denitrification, sampling 30 study streams across 13 ecoregions in 2017-2019. Twenty-four of the 30 study streams were part of the National Ecological Observatory Network (NEON) or the StreamPULSE network (NSF Macrosystems), where we could leverage continuously-collected data on key environmental factors (water temperature, light availability, background nutrient concentrations, dissolved O2 and temperature at adequate temporal resolution to model stream metabolism) as covariates to analyze the patterns in N process rates observed across biomes. The remaining six streams were sites of long-term study by our lab or our collaborators and had similar long-term data available. Four of the 30 streams (Arikaree River CO, Kings Creek KS, New Hope Creek NC, Pilgrim River MI) were visited on more than one sampling occasion to provide some measure of year-to-year variation in the rates and co-occurrence of N2 fixation and denitrification. Watershed characteristics were determined using GIS; delineated watersheds for NEON sites were downloaded from the NEON Aquatic Watershed tool (https://hub.arcgis.com/datasets/neon::neon-aquatic-watershed-2/about ), while watersheds for the other 18 sites were delineated using 10-m resolution Digital Elevation Models (DEMs; USGS 2022 https://apps.nationalmap.gov/downloader/#/). Watersheds were used to determine watershed area, slope, elevation (min, mean, max) and aspect. Land cover for each watershed was determined using 30-m resolution data from the National Land Cover Database (NLCD; ESRI 2022, https://www.arcgis.com/apps/instant/media/index.html?appid=fc92d38533d440078f17678ebc20e8e2).

Rates were measured on stream substrates, then scaled to a whole-reach rate by accounting for the relative benthic cover of each substrate type. On each sampling day, stream reaches were first mapped for habitat characteristics and substrate abundance. The most common substrates observed were rocks, sediment, macrophyte and wood, although additional rare substrates were included where possible (additional substrates sampled were algae, bedrock, and coarse particulate organic matter). We tabulated the percent coverage of each substrate type from habitat maps, and we grouped similar substrates into overarching categories (e.g. coverage of clay, silt, sand, and fine particulate organic matter was summed to “sediment” category). During mapping we also measured physical characteristics of the stream depth at 4-7 transects equally spaced along the reach, including wetted width and water, GPS location, and canopy cover in 4 directions using a densiometer. Water samples (both unfiltered and filtered through 0.45 um membrane filters) were collected for later analyses of water chemistry (dissolved organic carbon, DOC; total dissolved nitrogen, TDN; ammonium; nitrate+nitrite; soluble reactive phosphorus, SRP; total dissolved phosphorus, TDP; total phosphorus, TP).

Samples of each mapped substrate were then collected and placed into chambers. Polycarbonate food storage containers (2L) were used for rock and larger macrophyte substrates (Dodds et al. 2017; Eberhard et al. 2018). The chamber lids were sealed airtight with a Viton o-ring and were fitted with a 13x20 mm septa for sample collection. For sediment, wood, and smaller macrophyte substrate, chambers were made 0.475L glass mason jars and lids were similarly fit with an airtight sampling septum. For rock and wood, enough substrate was collected to approximately cover the bottom of the chamber and substrate surface area was measured after incubation by tracing. For all other substrates, a known surface area was collected either by coring (sediment) or by collecting all substrates in an area determined by the surface area of the sampling chamber lid. Volume of substrates placed in each chamber was determined either through direct measurement (sediment cores), or via displacement. Between 6 and 16 replicates of each substrate type were set up for each stream depending on the relative cover of each substrate type.

Blank chambers were also set up to account for changes in gas concentrations caused by container effects and/or changes in physical conditions like temperature. Blank chambers contained either rocks from outside the stream (to account for changes in sample chambers with stream rocks) or only stream water (to account for changes in sample chambers with all other substrates). Two to four replicates of each substrate and/or chamber type were used, and blanks were treated the same as all sample chambers. Additionally, air blank chambers were included for all substrates to test for natural generation of N2O or ethylene; these chambers were set up the same as sample chambers (contained real stream substrates and received nutrient amendments for denitrification assays; note that there were no unamended air blanks for denitrification) but had an air headspace introduced instead of an acetylene headspace.

Once filled, all chambers were sealed and N2 flux was measured. Chambers were divided into three different types: initial to quantify conditions at the start of the incubation (typically 2 reps for each substrate), light chambers, and dark chambers (wrapped in aluminum foil to inhibit autotrophic activity). Initial chambers were sampled immediately, then light and dark chambers were incubated in the stream for ~2 hrs. To sample dissolved gas concentrations from all chambers, triplicate water samples were siphoned into 12mL Exetainers with minimal air contact and zinc chloride was added as a preservative to a final concentration of 0.26% W/V. Barometric pressure and temperature were measured at the time of each sample collection for calculation of N2 gas saturation. A membrane-inlet mass spectrometer (MIMS) with 2-temperature calibration was used to analyze N2 and argon (Ar) concentrations in the samples, and data were converted from MIMS signals to gas concentrations using the R package mimsy (Kelly et al 2020). Because of concerns about bubble formation during incubation and sediment contamination of collected samples, we do not report or interpret the N2:Ar data here.

Following the N2 flux measurements, the foil was removed from all chambers, and the chambers were divided into 3 groups for measurements of denitrification and N2 fixation using acetylene block and reduction assays. For denitrification, we measured rates with and without amendments of C and N (0.62 g/L of both glucose and NaNO3), because the acetylene block method inhibits nitrification and because previous studies have shown that both C and N can become limiting in closed chamber incubations like these, so measuring without nutrient amendments can underestimate denitrification rates. We also added chloramphenicol (2 g L-1) to suppress additional protein synthesis during the incubation in both nutrient amended and unamended denitrification chambers. However, chambers were not sparged with nitrogen or helium to create anoxic conditions.

Following any nutrient amendments, an acetylene-filled balloon was added to each chamber, sized to equal 20% of the total chamber volume. Chambers were filled with stream water and sealed underwater, then balloons were popped with a needle through the sampling septum to introduce an acetylene headspace. Chambers were then shaken for approximately 20 seconds to equilibrate the gas dissolved in the water with that in the headspace. Initial headspace samples were collected within 5 minutes of sealing the chambers. Chambers were placed in the stream for a 2-hour incubation to maintain ambient stream temperatures, then shaken again to equilibrate before final headspace samples were collected and water temperature in the chamber was measured. All gas samples were placed into evacuated 9-mL serum vials and kept in the dark until analyzed.

Initial and final headspace samples were measured using a SRI 8610C gas chromatograph (GC). For denitrification, nitrous oxide (N2O) concentrations were measured with GC column oven set to 40 °C, Hayesep D column, He carrier gas, and an electron capture detector. For nitrogen fixation, ethylene concentrations were measured with GC column oven set to 40 °C ramping to 110 °C after 2.5 min and equipped with a Hayesep T column, He carrier gas, and a flame ionization detector. Gas standards (1000 ppm N2O, 100 ppm ethylene; Matheson Tri Gas) were run every 10-20 samples, and the standard peak areas were used to calculate concentrations in the headspace samples.

After the incubations, subsamples of substrate and/or associated biofilms were collected from each chamber for biomass determination and 16S rRNA Illumina sequencing. Rocks were taken out of the chamber and scrubbed using a small brush. Subsamples of the scrubbate were filtered through 0.7um GFF filters for chlorophyll a and ash-free dry mass (AFDM) analyses, and poured into sterile 15 mL falcon tubes. 12-mL sediment cores were taken from sediment chambers using a 10 mL syringe and placed into the falcon tubes for 16S sequencing, while the rest of the sediments were collected and frozen for later determination of AFDM. Wood substrates were sampled using a pocketknife to cut off approximately 4 surface shavings from each stick within a chamber, which were then placed into falcon tubes with chamber water for 16S sequencing. Macrophytes, algae, CBOM and other organic substrates were sampled by tearing off a small part of the collected biomass and placing it in a falcon tube with the chamber water for 16S sequencing. All samples were placed in a mobile -20 degree C freezer immediately after collection and stored in a -10 degree C freezer upon return to the lab.

We compiled environmental data that was available for the the study sites for use in predictive modeling. We obtained water temperature (WaterTemp_C_mean, WaterTemp_C_max), discharge (Discharge_m3s_mean), and light (Light_PAR_mean) from NEON (https://data.neonscience.org/), StreamPulse (https://data.streampulse.org/), and iUtah via CUAHSI HydroShare (https://www.hydroshare.org/group/6) data repositories. For NEON sites (identification codes in Organization_siteID column), we used the neonUtilities R package to request water temperature (DP1.20053.001), light (DP1.00024.001), and discharge (DP4.00130.001) from both the upstream and downstream sensor locations at each site on the day of sampling. All readings collected on the day of sampling at each sensor locations were either averaged ("_mean") or used to select the daily maximum value (WaterTemp_C_max). For streamPULSE sites, we used the StreamPULSE R package (Vlah and Berdanier 2019) to obtain water temperature, light, and discharge from the repository.

Citations:

Dodds, W. K., Burgin, A. J., Marcarelli, A. M., & Strauss, E. A. (2017). Chapter 32 - Nitrogen Transformations. In G. A. Lamberti & F. R. Hauer (Eds.), Methods in Stream Ecology (Third Edition) (pp. 173-196). Academic Press. https://doi.org/https://doi.org/10.1016/B978-0-12-813047-6.00010-3

Eberhard, E. K., Marcarelli, A. M., & Baxter, C. V. (2018). Co-occurrence of in-stream nitrogen fixation and denitrification across a nitrogen gradient in a western U.S. watershed [journal article]. Biogeochemistry, 139(2), 179-195. https://doi.org/10.1007/s10533-018-0461-y

Vlah, M., and A. Berdanier. 2019. StreamPULSE: Run stream metabolism models on StreamPULSE data.