Four sampling locations were established across a salinity gradient for static chamber sampling. Two sampling locations were established at a site with natural, unrestricted tidal exchange at Sage Lot Pond in a high (Unrestricted-High; mean 2020 salinity of ~25) and low (Unrestricted-Low; mean 2020 salinity of ~12) salinity location relative to the site and two were established at the impounded Herring River in a high (Impounded-High; mean 2020 salinity of ~9) and low (Impounded-Low; mean 2020 salinity of ~4) salinity location relative to the site. Each sampling location was equipped with three replicate collars for static chamber measurements and one pair of air/soil temperature sensors. In addition, data from a 3m tall mast at Sage Lot Pond equipped with sensors to measure relative humidity, temperature, and photosynthetically active radiation (PAR) are included.
To facilitate static chamber CH4 and CO2 flux measurements, round PVC collars (55 cm inner diameter, 24 cm height, 3 replicates per sampling location (n = 4)) were inserted into the sediment surrounding intact vegetation to a depth of at least 10 cm more than four weeks prior to initial sampling. On each sample collection date, we used a modular system of clear, round acrylic chambers to enclose the vegetation and sediment surface for flux measurement. Chamber sections were 55 cm diameter and 0.91 m tall. Depending on the height of the vegetation at the sampling location we used either two (unrestricted sites at Sage Lot Pond) or three (impounded sites at Herring River) stacked chambers for a total chamber height of 1.83 or 2.74 m and empty chamber volume of 0.43 or 0.64 m3. The topmost chamber always had a transparent lid permanently affixed with silicone caulking and a vent tube to allow for pressure equilibration. Two battery-powered fans (10 cm in diameter) were attached to the inside of the chamber to homogenize air and temperature sensors (HOBO TidbiT v2, Onset Computer Corporation) were mounted to the inside of the top and bottom chambers to record chamber air temperature. All samples were collected during mid-day hours between 10:00-14:00 local standard time. Chamber air was considered equilibrated based on visual inspection of the stability of a linear concentration change in the gases of interest, this was typically less than 60 seconds.
Field data required to calculate fluxes (i.e., chamber head space, chamber deployment time) were recorded on field datasheets and later entered into electronic spreadsheets. Concentrations of CH4 and CO2 during 3-7 minutes incubations were measured and recorded using a portable optical feedback-cavity enhanced absorption spectrometer (LI-7810 CH4/CO2/H2O Trace Gas Analyzer, LI-COR, Inc.) sampling at 1 Hz. Chamber flux incubations were conducted both in the light with transparent chambers to quantify net ecosystem exchange, and in the dark with opaque covering of the chambers to quantify ecosystem respiration in the absence of photosynthesis.
We used a sliding regression (“rollRegres” package in R, Christoffersen, 2019) to identify the 150-second window during the incubation where a simple linear regression of CH4 or CO2 concentration by time had a maximum R2. We verified that all linear models used to calculate flux had a significant slope (p less than 0.001) and an RMSE ≤ 1 ppm for CO2 and ≤ 10 ppb for CH4. One measurement did not have a slope significantly different than 0, and that flux was therefore set to 0.
We also assessed each incubation for non-linear changes in CH4 concentration over time as a signal of potential ebullition. When potential ebullition was observed, in addition to applying the sliding regression described above, we estimated CH4 concentration change based on the initial and final concentration of CH4 gas (VCS, 2015).
The rate of concentration change of CO2 and CH4 (slope or the difference between final and initial when ebullition was suspected) was converted to an area-based flux using the Ideal Gas Law and the volume and footprint of the chamber, along with measured chamber air temperature (pooled across the two temperature sensors). Calculations were executed with the “conc_to_flux” function in the R package “ecoflux” (Shannon, 2018).
This process step and all subsequent process steps (unless otherwise noted) were performed by the same person, Rebecca Sanders-Demott.
Christoffersen, B., 2019, rollRegres: Fast Rolling and Expanding Window Linear Regression. R package version 0.1.3. https://CRAN.R-project.org/package=rollRegres
Shannon, J., 2018, ecoFlux: Functions for ecological flux studies including sap flux and soil/stem C efflux (0.3.1). https://rdrr.io/github/jpshanno/ecoflux/
Verified Carbon Standard (VCS), 2015, Approved VCS Methodology VM 0033: Methodology for Tidal Wetland and Seagrass Restoration. Version 1.0, 20 November 2015, Sectoral Scope 14. Developed by Verra with Restore America’s Estuaries and Silvestrum. 1–115. https://verra.org/wp-content/uploads/2018/03/VM0033-Tidal-Wetland-and-Seagrass-Restoration-v1.0.pdf