Study site
The study was conducted in the Federseemoor (48.092°N, 9.636°E), a peatland of 30 km2 located in the region Upper Swabia in southwest Germany. This peatland has developed via natural terrestrialization from a proglacial lake after the last ice age. As a result, the surface area of the lake declined from 30 to 12 km2. Between 1787 and 1808, the lake was further reduced to a size of 1.4 km2 by drainage activities. The newly gained land of 11 km2 was used as pasture but turned out to be unprofitable due to the recurring high water table. Nowadays it is a nature conservation area, mainly consisting of fen (see van den Berg et al.21 for a vegetation map). The lake Federsee is completely surrounded by reed vegetation (P. australis), with a total area of 2.2 km2 and a density of around 70 living shoots and 75 dead stems per m2. During the measurement period (7–10 June) the Phragmites plants were 1.2 m high. This is half their maximum height, which is reached at the end of July. The high density of Phragmites and lack of other species in the reed belt result from high nutrient concentrations due to wastewater input to the lake since 1951. After 1982, the input of untreated sewage water was stopped, which reduced the nutrient concentrations. Only since 2006 has there been a significant improvement in water quality, and after 2008 the lake water became clear again. The field experiment was installed in the middle of the reed area at around 70 m distance from an eddy covariance (EC) tower, which has been running since March 201321. In a radius of at least 200 m around the EC tower, the vegetation is dominated by Phragmites (see van den Berg et al.21), meaning only reed dominated the measured EC footprint.
Field experiment
Nine plots of 2 m × 2 m were prepared for three treatments with three replicates: (1) clipped reed (CR), to exclude the pressurized flow in the plants; (2) clipped and sealed reed (CSR), to exclude any exchange via plant stems; and (3) control where reed was not manipulated. In the CR and CSR treatments, living and dead reed stems were clipped to about 10 cm above the water table. In the CSR treatment the clipped reed stems was sealed with an acrylic sealant. Since rhizomes connect plants over longer distances, plots were isolated by cutting rhizomes from the reed plants around each plot to a depth of 50 cm, to avoid gas exchange with the surrounding area. The period between preparation of the plots and measurements was minimized (1–2 days) to reduce possible side effects, such as change in substrate availability for methanogens. One day before the first measurement, the water table rose about 20 cm in the whole field, flooding the prepared sealed stems of one plot already prepared for the CSR treatment. Nevertheless, since no gas exchange is expected from the sealed stems, this plot was still included in the experiment. CH4 and CO2 diffusive fluxes from the soil and plant-mediated fluxes were measured with transparent flow through chambers. Pore water was extracted to analyze the effect of the reduced/excluded gas exchange by the plants on soil chemistry. In each plot ebullition was measured as well (see below).
Diffusive and plant mediated CH4 flux
On 7, 9 and 10 June 2016 between 07:00 and 18:00, the gas fluxes of each treatment were alternately measured. Per day, only one of the triplicates per treatment was measured. CH4 fluxes were measured in the middle of the plots with transparent chambers with a diameter of 50 cm. One chamber was 2 m high and was on the control plots. Two chambers were 1 m high and used on the CR and CSR plots. The 1-m chambers were equipped with a small fan of 8 cm × 8 cm that had a flow capacity of 850 l min−1; two fans were installed in the 2-m chamber. Each day one replicate of every treatment was measured, to be able to capture the diurnal cycle for each plot and to minimize disturbance by translocating the chambers. The chambers were connected with 8 m tubing to a multiport inlet unit attached to a fast greenhouse gas analyzer (GGA) with off-axis integrated cavity output spectroscopy (GGA-24EP, Los Gatos Research, USA) measuring the concentration of CH4 and CO2 every second. Every 5 min, the multiport switched between the three chambers, allowing air from each chamber to be alternately pumped through the GGA with a pumping rate of 300 ml min−1 and resulting in four flux measurements per plot per hour (~ 35 measurements per plot per day). The withdrawn air from the chamber was replaced with ambient air through an opening in the chamber. After 1–2 h of continuous measurements, the chambers were ventilated by lifting the chambers to fully replace inside air with ambient air. After 15 min, the chamber was put back and measurements continued. Since it takes a long time before the chamber CH4 gets to equilibrium with the water column, 1–2 h of increasing CH4 concentration in the chamber will have little effect on the measurement accuracy of the CH4 flux (in contrary to the CO2 flux)22. Nevertheless, we used only data from the first 30 min after ventilating to calculate the diffusive flux (five measurements per plot per day), since this is the period where temperature and humidity inside the chamber resemble outside conditions most closely. Only for the comparison between eddy covariance fluxes and chamber fluxes on the control plots we did use data from the whole measurement period.
The concentration for every measurement point was corrected for the change in concentration caused by the inflow of ambient air with known CO2 and CH4 concentrations (measured by the EC station) and outflow of chamber air (both with a flow rate of the pump speed of the Los Gatos). The slope of the corrected chamber concentrations over a 4 min period within the 5 min measurement was used to calculate the flux and was checked for non-linear fluctuations due to e.g. ebullition. Fluxes corresponding to an average chamber concentration of > 100 ppm CH4 were discarded, because of the GGA’s detection limit. In total 11% of the fluxes were discarded.
Ebullition
In each plot ebullition was measured by catching bubbles from a fixed surface with an ebullition trap10, composed of a 20 cm diameter funnel, to which a glass bottle of 300 ml was attached. The bottles were filled with water from the site and the ebullition trap was installed under the water table on 8 June and carefully anchored between reed stems (no open endings of stems were below the trap) on the soil surface around 0.55 m below the water surface. Bubbles were captured in the glass bottle for 18 days, after which the bottles were removed and gas samples were taken in the field. The total volume of ebullition gas was determined and the concentration of CH4, CO2 and N2O were measured by gas chromatography (7890B GC, Agilent Technologies, USA) in the lab.
Environmental variables
In each chamber, temperature and radiation were measured with a temperature/light sensor (HOBO Pendant data logger, Onset Computer Corporation, USA) logging at an interval of 30 s. Every minute soil temperature was measured in each plot in the upper 0–0.05 m with a Soil Water Content Reflectometer (CS655, Campbell Scientific Inc., USA) around 0.56 m below the water table. Air temperature, air relative humidity (HMP155, Vaisala Inc., Finland) and incoming and outgoing shortwave and longwave radiation (CNR4, Kipp & Zonen Inc., The Netherlands) were measured at a height of 6 m close to or at the EC station. Groundwater table was continuously measured with a water level pressure sensor (Mini-Diver datalogger, Eijkelkamp Agrisearch Equipment Inc., The Netherlands) placed at 1.45 m depth in a 2-m long filter pipe that was placed 1.60 m into the soil. Data were recorded at a 30 min interval.
Pore water sampling and analysis
To see if the treatments had any effect on the methane production, pore water samples were analyzed. At two locations in each plot, pore water was extracted anaerobically with ceramic cups (Eijkelkamp Agrisearch Equipment Inc., The Netherlands). Pore water from 10, 20, 30 and 50 cm depth was collected by vacuum suction in syringes and transported to the lab. In the lab, pore water was diluted with a ratio of 1:3. Dissolved organic carbon (DOC) concentration was measured with a Dimatoc 100 DOC/TN-analyzer (Dimatec, Germany). A second pore water sample was taken in vacuumed 13 ml exetainers with 3 g of NaCl. The concentration of CH4 in the headspace of these exetainers, representing the CH4 concentration in pore water, was determined on a HP gas chromatograph (Hewlett Packard, USA). A third pore water sample was fixed with 0.2% 2.2-bipyridin in 10% CH3COOH buffer in the field to determine Fe(II) measuring photometrical absorption at 546 nm in the lab.
Eddy covariance
The EC tower was located at a distance of around 70 m from the prepared plots. The tower was 6 m high and consisted of a LI-7700 open path CH4 gas analyser (LI-COR Inc., USA), a LI-7200 enclosed path CO2/H2O gas analyser (LI-COR Inc., USA) and a WindMaster Pro sonic anemometer (GILL Instruments Limited Inc., UK). Molar mixing ratio/mass density of the gases and wind speed in three directions were measured at a frequency of 10 Hz. Fluxes were calculated for an averaging interval of 15 min with the software EddyPro version 6.1.0. For more detailed information about the set up and calculations of the fluxes, see van den Berg et al.21.
δ13C measurements
CH4 oxidation and transport lead to isotopic fractionation of δ13C of CH423. The difference between δ13C of the CH4 present in the soil and the CH4 emitted to the atmosphere may therefore reveal the importance of both methane oxidation and the different emission pathways.
The δ13C of CH4 tends to be much lower than the natural abundance in organic compounds, because methanotrophic prokaryotes prefer the lighter 12CH4 to 13CH4 thereby increasing the δ13C of CH4. Diffusion rates for 12CH4 are higher than for 13CH414 decreasing the δ13C of the emitted CH423. Although 13C enrichment (compared to produced CH4) has been found in internal spaces of plants due to CH4 oxidation14, the fractionation at the plant-atmosphere surface reduces the δ13C by about 12–18‰ due to the faster transport rate of 12CH4, which makes that emitted CH4 can have a lower fraction of δ13C than the produced CH4. Differences in δ13C between sediment and overall emission are larger for plants with diffusive internal gas transport than for plants with convective gas transport23.
Since fractionation of CH4 emitted through ebullition in shallow waters is negligible, these gas bubbles can be used to know the isotopic composition of CH4 produced in sediment23. We therefore compared the δ13CH4 signature of ebullition gas with the signatures of CH4 from the chambers. Gas samples from the chamber were taken when the CH4 concentration was at least 10 times the ambient concentration, from each plot in the afternoon. The δ13CH4 signature was measured with an isotope-ratio mass spectrometer Delta plus XP (Thermo Finnigan, Germany).
Statistics
Chamber fluxes were measured at different times of the day, which means that environmental variables like temperature and radiation were varying. To be able to compare the different treatments without the variation resulting from environmental conditions, an analysis of covariance (ANCOVA) was conducted with the environmental variables as covariables. For the analysis, the data of the different measurement days were pooled together per treatment. The residuals of the model were normally distributed. With the parameters of the ANCOVA model, average fluxes were calculated with average environmental variables for the period ebullition was measured (8–27 June), to be able to compare the chamber fluxes with ebullition.
To test if the means of the ebullition measurements or pore water concentrations were different between the treatments, an analysis of variance (ANOVA) test was performed with Fishers’s Least Significant Difference (LSD) post hoc test to find the specific differences between the treatments.
Source: Ecology - nature.com
