Methanotrophic and Methanogenic Communities in Swiss Alpine Fens Dominated by Carex rostrata and Eriophorum angustifolium

Abstract

Vascular plants play a key role in controlling CH₄ emissions from natural wetlands, because they influence CH₄ production, oxidation, and transport to the atmosphere. Here we investigated differences in the abundance and composition of methanotrophic and methanogenic communities in three Swiss alpine fens dominated by different vascular plant species under natural conditions. The sampling locations either were situated at geographically distinct sites with different physicochemical properties but the same dominant plant species (Carex rostrata) or were located within the same site, showing comparable physicochemical pore water properties, but had different plant species (C. rostrata or Eriophorum angustifolium). All three locations were permanently submerged and showed high levels of CH₄ emissions (80.3 to 184.4 mg CH₄ m⁻² day⁻¹). Soil samples were collected from three different depths with different pore water CH₄ and O₂ concentrations and were analyzed for pmoA and mcrA gene and transcript abundance and community composition, as well as soil structure. The dominant plant species appeared to have a significant influence on the composition of the active methanotrophic communities (transcript level), while the methanogenic communities differed significantly only at the gene level. Yet no plant species-specific microbial taxa were discerned. Moreover, for all communities, differences in composition were more pronounced with the site (i.e., with different physicochemical properties) than with the plant species. Moreover, depth significantly influenced the composition of the active methanotrophic communities. Differences in abundance were generally low, and active methanotrophs and methanogens coexisted at all three locations and depths independently of CH₄ and O₂ concentrations or plant species.

INTRODUCTION

The atmospheric concentration of the highly potent greenhouse gas methane (CH₄) has increased over the past several decades to a current value of ca. 1.8 ppm by volume (ppmv) and is continuing to rise after an apparent stagnation in the early 2000s (1). Natural wetlands are the most important nonanthropogenic CH₄ source, with estimated emissions of 177 to 284 Tg year⁻¹, accounting for 26 to 42% of the global CH₄ budget (2). Wetlands in northern high latitudes (north of 45°N) contribute approximately 44.0 to 53.7 Tg CH₄ year⁻¹ (3). The CH₄ cycle in these environments is driven by microorganisms: methanogenic archaea (methanogens) generate CH₄ in the anoxic zones of the wetland soil as the terminal step of anaerobic degradation of organic matter. These microorganisms represent a monophyletic euryarchaeal lineage and largely utilize hydrogen or carbon dioxide, acetate, or small methylated compounds as substrates (4, 5). On the other hand, a substantial amount of CH₄ generated in wetlands is oxidized before it can reach the atmosphere by aerobic methane-oxidizing bacteria (methanotrophs), which are active mainly at the oxic-anoxic interface (6). These organisms can utilize CH₄ as the sole energy and carbon source (7), and according to current knowledge, they are placed within the phyla Proteobacteria and Verrucomicrobia (8). The proteobacterial methanotrophs are the most diverse group and are further divided into type I (Gammaproteobacteria) and type II (Alphaproteobacteria) methanotrophs (9).

Methane emissions from wetlands are highly variable both spatially and temporally (10–12). Several factors that influence these variations have been identified: temperature, level of the water table, plant cover, pH, or soil characteristics (13–16). In particular, vascular plants play a key role in CH₄ emissions from wetlands, since they affect the individual processes involved in CH₄ cycling, i.e., CH₄ production, oxidation, and transportation (17). Methane production is stimulated in the rhizosphere by root exudates and decaying roots, which ultimately increase the substrate pool available for methanogenesis (18, 19). On the other hand, O₂ transported to the root zone enhances CH₄ oxidation by aerobic methanotrophs (4). Moreover, the plant aerenchyma can act as a gas conduit for transporting substantial amounts of CH₄ directly from the subsurface to the atmosphere, bypassing zones of potential CH₄ oxidation (20–22).

Sedges of the family Cyperaceae (e.g., Carex and Eriophorum spp.) are often the dominant vascular plants in northern wetland systems (23). Differences in CH₄ emissions from different sedges due to species-specific differences in root exudation patterns and gas-transporting mechanisms have been reported (24, 25). For example, Eriophorum angustifolium exhibits a higher CH₄ transport capacity than Carex rostrata (26, 27). Ström et al. (28) reported significantly higher formation of acetate in the rhizosphere of Eriophorum vaginatum monoliths than in that of C. rostrata. Nevertheless, the C. rostrata monoliths showed higher CH₄ emissions than the E. vaginatum monoliths, and this observation was attributed to higher rates of CH₄ oxidation in the latter. Yet the microbial communities driving the CH₄ cycle in the respective soils were not assessed. In general, little is known about potential differences in the methanotrophic and methanogenic communities associated with different wetland plants (29, 30).

Similar plant communities and life zones can be found in alpine regions between ca. 2,000 and 3,600 m above sea level (ASL) and along the Arctic Circle at sea level (31, 32). Nevertheless, these environments differ in several ways. For example, the alpine regions are characterized by diurnal cycles throughout the year, generally higher temperatures, and the lack of extended permafrost, as well as an insulating snow cover and thus ongoing microbial processes in the soil during the winter (see, e.g., reference 33). Nevertheless, the general mechanisms, i.e., microbial CH₄ oxidation and production, and the interplay between methanotrophs, methanogens, and vascular plants, are comparable, making alpine wetlands easily accessible model systems for the enhancement of our knowledge of the CH₄ cycle in the vast arctic and subarctic wetland areas. Studies on CH₄ dynamics in alpine fens have been conducted mainly in the Rocky Mountains and the Tibetan Plateau (34–38). Studies in the European Alps, however, have been limited to an Austrian (16) and several Swiss (32) alpine fens. These studies focused mainly on in situ measurements of CH₄ emissions and concentrations in pore water, reporting spatial and seasonal variability in emissions. In particular, two C. rostrata-dominated, permanently submerged fens in the Swiss Alps, at the Oberaar and the Göschener Alp, have been studied in detail (33, 39, 40). Analyses of physicochemical pore water properties, and of the abundances of methanotrophs and methanogens in the subsurface along the depth profile, were performed. The results showed that variability in the parameters analyzed was most pronounced in the upper 15 to 20 cm below the water table in these particular fens. At the Oberaar site, composition analysis of methanotrophic and methanogenic communities also confirmed this high variability in the upper soil layers (40). Nevertheless, variations in microbial communities with different dominant species of vascular plants have yet to be investigated.

We hypothesize that apparently comparable alpine fens (i.e., permanently submerged, methanogenic, covered by typical wetland sedges, located at similar altitudes, under comparable climatic conditions) show plant-specific differences in their methanotrophic and methanogenic communities. To test this hypothesis, we investigated soils collected from the upper 15 to 20 cm of three Swiss alpine fens dominated by different vascular plants for abundance and composition of the total (gene level of indicative functional genes) and active (transcript level) methanotrophic and methanogenic communities. Two fens were situated at the Oberaar site and were dominated either by C. rostrata or by E. angustifolium, while the third fen, at the Göschener Alp site, was dominated by C. rostrata. This selection allowed for the comparison of (i) fens that differ in plant species but show comparable physicochemical parameters and (ii) fens that are dominated by the same vascular plant but differ in physicochemical characteristics. Molecular biological analyses were supplemented by field-based measurements of CH₄ emissions, physicochemical pore water properties, and soil structural analyses, providing a comprehensive picture of CH₄ dynamics in these environments.

MATERIALS AND METHODS

Study sites and sample collections.

Two geographically distinct sites were selected in central Switzerland, both positioned on siliceous bedrock, i.e., granite: one at the Oberaar (2,320 m ASL; canton of Berne) and one near the Göschener Alp (1,920 m ASL; canton of Uri). The Oberaar site has been described in detail by Franchini et al. (40). It consists of two interconnected minerotrophic fens, fen 1 (ca. 120 by 30 m) and fen 2 (ca. 70 by 15 m), with water flowing from fen 1 to fen 2, from which it is discharged. Both fens are permanently submerged, and the vascular plant vegetation is dominated by species of the Cyperaceae family (namely, Carex spp., Trichophorum caespitosum, and Eriophorum spp.), covering as much as 70% of the surface area. Moreover, the submerged mosses Calliergon sarmentosum and Sphagnum spp. are present in patches (32, 40).

The Göschener Alp wetland complex (ca. 15.9 ha) consists of several nonconnected fens and has been described by Liebner et al. (33). The fen selected for this study (ca. 340 by 40 m) is permanently submerged, and the vascular plant vegetation is dominated by Carex spp.; Eriophorum spp. and Juncus spp. are present in patches.

Within these two sites, three sampling locations were selected for the present study on the basis of their dominant vascular plant species: (i) OA1 in Oberaar fen 1, with a monospecific stand of Carex rostrata; (ii) OA2 in Oberaar fen 2, with a monospecific stand of Eriophorum angustifolium; and (iii) GA in the Göschener Alp fen, with a monospecific stand of C. rostrata (see Fig. S1 in the supplemental material). At OA1 and OA2, the submerged moss C. sarmentosum was also present in patches. The water table levels at OA1 and OA2 were 3 to 4 cm above the soil surface, and in close proximity to OA1, a water flow of ca. 0.5 m min⁻¹ was measured. At GA, the water table level was 0.5 to 1 cm above the soil surface, and the water was stagnant. This selection of study sites and sampling locations allowed us to compare methanotrophic and methanogenic communities associated with different plant species in the same geographical and physicochemical setting and also to compare communities associated with the same plant species but at geographically and physicochemically distinct sites.

Sampling was performed in September and early October 2013. Details on sampling dates, the coordinates of the sampling locations, annual precipitation, and temperatures are listed in Tables 1 and 2. At each location, CH₄ emission into the atmosphere and the pH and electrical conductivity of the pore water were measured in situ, pore water for chemical analyses was sampled, and soil cores for molecular biological analyses were collected in triplicate.

TABLE 1.

Characteristics of study sites

Characteristic	Study site
	Oberaar	Göschener Alp
Elevation (m ASL)	2,320	1,920
Mean precipitation (mm)a	1,673.1	1,682.6
Temp (°C)a
Mean	−0.3	2.6
Max	19.2	22.1
Min	−22.1	−17.4
Site description	2 connected fens with flowing water	Several nonconnected fens with stagnant water

^aData for 2013 are shown. The meteorological data for the Oberaar site were obtained from Grimsel-Hospiz World Meteorological Organization (WMO) station 06744 (46°34′18″N, 08°19′59.5″E, approximately 6 km northeast of the study site at 1,980 m ASL). The meteorological data for the Göschener Alp site were obtained from the nearby forefield of the Damma glacier (46°38′17.95″N, 08°27′40.05″E, at 2,025 m ASL) and were provided by the WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland. For both study sites, the temperature was corrected for the difference in elevation, assuming a temperature change of 0.6°C/100 m.

TABLE 2.

Characteristics of sampling locations

Characteristic	Sampling location
	OA1	OA2	OA3
Sampling date (day.mo.yr)	23.09.2013	04.10.2013	20.09.2013
Coordinates	46°32′51″N, 8°15′41″E	46°32′49″N, 8°15′38″E	46°38′58″N, 8° 29′36″E
Size of sampling area (m)	10 by 3	8 by 3	5 by 5
Water table level (cm)	3–4	3–4	0.5–1
Dominant vascular plant	C. rostrata	E. angustifolium	C. rostrata

Quantification of CH₄ emissions.

Transparent static flux chambers (ca. 30 cm by 30 cm by 30 cm; made of highly transmissive acrylic glass) (see Fig. S1a in the supplemental material) were used to determine CH₄ emissions as described previously (32, 33). In brief, the chambers were carefully placed in standing water, which provided a complete sealing. The average height between the water table and the top of the chamber was measured for each experiment in order to calculate the exact gas volume inside each chamber. The ventilation hole was closed immediately after the placing of the chamber, and a 50-ml gas sample was extracted with a 60-ml syringe every 5 min for a total of 30 min. The gas samples were injected into sealed and preevacuated 20-ml glass vials. Prior to each sample extraction, the gas in the chamber was mixed by pulling and pushing the piston of the syringe five times. Gas concentrations were measured by gas chromatography (GC) as described previously (40), and emission to the atmosphere was calculated based on the linear regression of accumulated CH₄ inside the chamber (25, 41). During each chamber measurement, a clear linear increase in the CH₄ concentration inside the chamber was observed over time (R² = 0.967 to 0.999).

After the removal of the chambers, the biomass of vascular plants positioned within each chamber during measurements was cut at the water table, separated based on physical appearance into green (i.e., photosynthetically active) and brown (i.e., photosynthetically inactive) biomass, dried for 48 h at 60°C, and weighed. The overall weight of both categories was considered 100% for the purpose of determining the fractional dry weight of green and brown biomass.

The diffusive upward flux of CH₄ through the water phase was calculated from CH₄ concentrations in the pore water measured between the depths of 5 and 10 cm (see below), by applying Fick's first law of diffusion according to the method of Hornibrook at al. (42) employed for similar studies (32) and considering a soil porosity of 0.9.

Sampling and physicochemical analyses of pore water.

Pore water sampling was carried out as described in detail elsewhere (32, 33). In brief, a series of brass tubes (inside diameter [i.d.], 0.3 cm; outside diameter [o.d.], 0.4 cm; tip sealed and perforated within the first 10 mm) were inserted into the fens at 1.0, 2.5, 5.0, 7.5, 10.0, 15.0, 20.0, and 25.0 cm below the water table (see Fig. S1b in the supplemental material) (40). The tubes were slowly filled with pore water by using syringes, closed with three-way valves, and equilibrated for at least 1 h. Immediately prior to sampling, the dead volume of the tubes was discarded, and contact of the pore water with air was avoided as much as possible.

To determine concentrations of dissolved CH₄, 5 ml of pore water was transferred to sealed, N₂-flushed 20-ml glass vials containing 100 μl of 1 M HCl and was stored at 4°C prior to processing. Pore water CH₄ concentrations were subsequently calculated from headspace concentrations determined by GC-flame ionization detection (FID) (see above) as described previously (33). For pore water O₂ concentrations, 25 ml of pore water was slowly extracted, 5 ml discarded, and O₂ concentrations determined calorimetrically on site by using a DRr890 colorimeter (Hach Lange, Rheineck, Switzerland) with high- and low-range Dissolved Oxygen AccuVac ampoules (Hach Lange) according to the manufacturer's instructions.

For measurements of dissolved organic carbon (DOC) concentrations, 20 ml of pore water was filtered on site through 0.45-μm nylon filters (prewashed with deionized water) and was immediately transferred to 20-ml glass vials containing 100 μl of 1 M HCl (20). The samples were stored at −20°C until further processing and were subsequently analyzed using a Shimadzu TOC-5000 analyzer (Shimadzu Scientific Instruments, Columbia, MD). The electrical conductivity and pH of the pore water were measured on site using a Multi 350i probe (WTW Laboratory and Field Products; Nova Analytics, Woburn, MA) equipped with an LR 325/01 conductivity cell and a SenTix 51 pH electrode, respectively. In situ water and soil temperatures were recorded using a manual temperature sensor (Testo AG, Lenzkirch, Germany).

To measure acetate concentrations in the pore water, samples were filtered on site (0.45-μm nylon filters), and 10 ml was transferred to 15-ml plastic vials containing 200 μl of 1 M NaOH. Samples were stored at −20°C prior to further processing. Acetate concentrations were subsequently measured by ion chromatography in a system (DX-1000; Dionex, Sunnyvale, CA) equipped with an EG40 OH⁻ eluent gradient generator.

Soil sampling.

Soil samples were collected in triplicate after flux measurements and pore water sampling in close vicinity to the sampling installations. Cubical soil cores were carefully cut out of the fens by hand using a sharp, sterile knife, immediately placed on dry ice for transport, and frozen at −80°C upon arrival at the laboratory. The frozen cores ranged from 18 to 25 cm in length, from 10 to 19 cm in width, and from 4 to 11 cm in height. The cores were sectioned in a climate chamber at −20°C by using a band saw (BS 400 E Electronic; Flott, Remscheid, Germany) (saw blade with 7 teeth per inch, 12 mm wide and 0.4 mm thick). Approximately 1 cm was cut off the rims and was discarded in order to remove soil material disturbed during sampling.

Two vertical slices ca. 1 to 2 cm thick were cut from each cubical core, representing duplicates of the same soil core. The slices were subsequently sectioned horizontally into 1-cm-thick subsamples. The subsamples used for further analyses were selected based on in situ pore water CH₄ and O₂ concentrations of the respective sampling location. Subsample A was positioned ca. 1 to 2 cm below the solid soil surface and was characterized by high O₂ and low CH₄ concentrations (ca. 5 cm below the water table for OA1 and OA2; ca. 2.5 cm below the water table for GA). Subsample B was situated in a transition zone ca. 3.4 to 4.5 cm below the soil surface, with low CH₄ and low O₂ concentrations (ca. 7.5 cm below the water table for OA1 and OA2; ca. 5 cm below the water table for GA), and subsample C was taken from a zone ca. 11 to 12 cm below the soil surface with low O₂ but high CH₄ concentrations (ca. 15 cm below the water table for OA1 and OA2; ca. 12.5 cm below the water table for GA). Photographs of representative soil slices cut from one soil core taken at each location are presented in Fig. S2 in the supplemental material. The subsamples were labeled according to sampling location and relative depth (A, B, or C), e.g., OA1_A. The subsamples were stored at −80°C until further subsampling for structural analysis and nucleic acid extraction.

Structural analysis of soil samples.

To determine the gravimetric water content and structure of the water-saturated soils, ca. 1-cm³ cubes were cut from each subsample described above. The water content was determined as a proxy for soil density by measuring water loss after drying at 60°C for 48 h. Structural analysis, i.e., determination of surface areas of different plant materials, was carried out as described elsewhere (40). In brief, a ca. 1-cm³ frozen cube from each subsample was weighed, thawed in 20 ml distilled water, and disassembled carefully with fine needles. After filtration through a 1-mm sieve, particles smaller than 1 mm (filtrate; referred to below as debris) were filtered through a 2.7-μm glass fiber filter (GF/D; Whatman, Opfikon, Switzerland). After drying at 60°C for 48 h, the dry weight of the debris was determined by subtracting the dry weight of the filter from the total weight. Particles larger than 1 mm were resuspended in 20 ml distilled water and were separated into the following categories based on visible structural differences: roots (fine cylindrical structures with branches), mosses (main stem with small leaves), and unknown material (material with undefined structure, including leaf and stem residues and decayed material). The surface areas of each category were subsequently determined using an Epson Expression 10000XL scanner and WinRHIZO Pro automated image analysis software (Regent Instruments Inc., Quebec, Canada). After filtration through a 2.7-μm glass fiber filter, scanning, and drying, the dry weight of each category was determined.

Extraction of nucleic acids.

Nucleic acids were extracted from ca. 2 g of each subsample, and the sample material was comparable to the material used for structural analysis. Total genomic DNA and total RNA were extracted by using the RNA PowerSoil Total RNA Isolation kit and the RNA PowerSoil DNA Elution Accessory kit (both from MoBio Laboratories, Carlsbad, CA) according to the manufacturer's instructions. Extracts from the duplicate core slices were subsequently pooled for further analyses, and extracted DNA and RNA were quantified using a NanoDrop spectrophotometer (Thermo Scientific, Wilmington, DE).

Residual DNA was removed from the RNA extracts by treating 5 to 10 μl of RNA with 10 U of RQ1 RNase-free DNase (Promega, Fitchburg, WI) and 40 U of RiboLock RNase inhibitor (Thermo Fisher Scientific, St. Leon-Rot, Germany) for 60 min at 37°C, followed by purification with the RNeasy minikit (Qiagen AG, Venlo, The Netherlands). The RNA extracts were subsequently subjected to PCR targeting the bacterial 16S rRNA gene (see Table S1 in the supplemental material) to test for complete removal of DNA. Only RNA extracts that did not result in the amplification of bacterial 16S rRNA genes were processed further. Ten microliters of the DNase-treated RNA sample was subsequently reverse transcribed to cDNA with 1 μl of random hexamer primer using the RevertAid H Minus First Strand cDNA synthesis kit (Thermo Fisher Scientific).

Amplification and quantification of target genes.

Methanotrophic communities were analyzed based on the pmoA gene, which encodes a subunit of the particulate methane monooxygenase, using primer sets A198f–mb661r and A198f–A650r (see Table S1 in the supplemental material). This enzyme catalyzes the first step in the CH₄ oxidation pathway (the transformation of CH₄ to methanol) and can be found in most known methanotrophs (7, 8). Moreover, the mmoX gene, which encodes the soluble methane monooxygenase and is also present in many methanotrophic genomes, was targeted with three different primer sets (see Table S1) to detect potential expression. These primers also allow detection of the genera Methylocella and Methyloferula, which lack pmoA (43, 44). For the analysis of methanogenic communities, primer set mlas–mcrA-rev (see Table S1) was used to target the mcrA gene, which encodes the alpha subunit of the enzyme methyl-coenzyme M reductase (MCR), catalyzing the final step in methanogenesis (45).

PCRs (25-μl reaction mixture) were performed using 1 μl of template DNA or cDNA, 1× PCR buffer, 0.2 μM (final concentration) of each primer, 0.2 mM deoxynucleoside triphosphates (dNTPs), and 0.5 U Taq polymerase (Dream Taq; Thermo Fisher Scientific). Details on primers and amplification conditions are listed in Table S1 in the supplemental material. Various dilutions of the DNA extracts (1:10 to 1:150 in water) were tested in order to avoid inhibition by coextracted substances, and for each DNA extract, the dilution resulting in the highest PCR yield was used for further analyses.

The abundances of genes (DNA) and transcripts (cDNA) of pmoA (primer set A189f–mb661r only) and mcrA per gram of soil (wet weight) were determined using the ABI 7300 real-time PCR system (Applied Biosystems, Foster City, CA), 1× Kapa SYBR Fast quantitative PCR (qPCR) master mix (Kapa Biosystems, Woburn, MA), 0.4 μM (final concentration) of each primer, and 1 μl of environmental DNA (best dilution) or cDNA in a 20-μl final reaction volume (for amplification conditions, see Table S1 in the supplemental material). Duplicate serial dilutions of previously quantified DNA from Methylococcus capsulatus (strain Bath; courtesy of M. M. Svenning, University of Tromsø, Tromsø, Norway) or Methanosarcina barkeri (DSM 800) were used as calibration curves for the quantification of pmoA or mcrA, respectively. Each sample was analyzed in triplicate, and a total of four runs were required for every gene to include all the samples. Amplification efficiencies calculated from the slopes of calibration curves were in the range of 93.4 to 94.8% (R² = 0.9933 to 0.9972) and 92.1 to 93.9% (R² = 0.9988 to 0.9997) for pmoA and mcrA, respectively. The limits of quantification for pmoA and mcrA copy numbers per gram of soil were determined to be 1.2 × 10³ and 0.9 × 10³, respectively.

T-RFLP analyses.

Terminal restriction fragment length polymorphism (T-RFLP) analysis was performed with the pmoA and mcrA genes (DNA) and transcripts (cDNA) to determine the compositions of methanotrophic and methanogenic communities, respectively. Analysis was carried out as described previously (39, 46) using FAM (6-carboxyfluorescein)-labeled primers A189f and mcrA_rev, respectively, and the amplification conditions listed in Table S1 in the supplemental material. Digested products were gel purified using the GeneJET gel extraction kit (Thermo Fisher Scientific). After restriction digestion with MspI (pmoA) or HaeIII (mcrA), the products were purified using the Montage SEQ₉₆ sequencing reaction cleanup kit (Millipore AG, Zug, Switzerland). Digested and purified PCR products (0.5 to 3 μl) were mixed with 10 μl of HIDI formamide and 0.1 μl of MapMarker 1000_ROX (BioVentures, Murfreesboro, TN), denatured for 2 min at 92°C, and placed on ice immediately. Samples were separated on an ABI 3130xl genetic analyzer (Applied Biosystems, Foster City, CA), and the terminal restriction fragments (T-RFs) were determined using the GENEMAPPER software package (version 3.7; Applied Biosystems). Only T-RFs with a length of >20 (mcrA) or >30 (pmoA) bp, a height of >30 fluorescence units (FU), and a relative area of ≥1% of total area were selected for further analyses. Samples showing a total peak area of <15,000 FU were excluded. Binning of T-RFs to operational taxonomic units (OTUs) was based on in silico analysis of clone sequences of this and previous studies (39, 46). After binning, a matrix of relative areas of individual T-RFs was generated for statistical analyses.

Construction and analysis of recombinant clone libraries.

Recombinant clone libraries of pmoA and mcrA genes were constructed and analyzed in order to facilitate the assignment of T-RFs to phylogenetic groups of methanotrophs and methanogens, respectively. Libraries were constructed from one representative sample from each location using the TA Cloning kit (Invitrogen, Carlsbad, CA) and blue-white screening. For each sample, DNA from subsamples A, B, and C were pooled in equivalent amounts prior to PCR amplification, and PCR products were gel purified before ligation. Forty-eight clones from each library were randomly selected and sequenced (sequencing was performed by GATC Biotech, Constance, Germany). The identities of pmoA and mcrA gene sequences were confirmed by database searches using NCBI BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). In silico digestion with MspI (pmoA) and HaeIII was performed using the NEBcutter online tool (New England Biolabs), and the T-RF size was confirmed by T-RFLP analysis of the clones as outlined above. Phylogenetic analysis of the gene and deduced amino acid sequences was carried out using the ARB program package (47). Sequences were placed in existing alignments of representative pmoA and mcrA amino acid sequences by using the neighbor-joining algorithm implemented in ARB.

Statistical analysis.

The means and standard deviations of the individual parameters were calculated by averaging the triplicate values obtained at each location. Spearman's rank correlation of the means of physicochemical pore water properties with depth were calculated using the R software environment for statistical computing and graphics (48).

The compositions of methanotrophic and methanogenic communities based on relative peak areas of individual T-RFs were analyzed separately by using the vegan, version 2.3-0 (49), and BiodiversityR (50) R packages. Variation in community structure among all samples was analyzed by pairwise comparison of the β-diversity based on the Bray-Curtis dissimilarity matrix (51). Canonical-correlation analysis (CCA) was performed using the geographically distinct site (i.e., the Oberaar or the Göschener Alp), the dominant plant species (i.e., C. rostrata or E. angustifolium), the relative depth (i.e., the depth at which subsample A, B, or C was taken), and the target molecule (i.e., gene or transcript) as constraining parameters in the model. These variables are referred to below as site, plant species, depth, and molecule, respectively.

Analysis of variance (ANOVA) of physicochemical pore water properties and multivariate analysis of variance (MANOVA) of T-RFLP profiles were performed by using MATLAB (R2013a; MathWorks) to test for statistical significance of the variables site, plant species, depth, and, for the T-RFLP profiles, molecule. Significance levels for λ (percent of variance) were determined by randomization (bootstrapping) of the observations 10,000 times.

Nucleotide sequence accession numbers.

Selected representative pmoA and mcrA sequences determined in this study have been deposited at the EMBL nucleotide sequence database and are accessible under accession numbers KP071439 to KP071468 and KP071397 to KP071438, respectively.

RESULTS

Plant biomass.

In the Oberaar fen, the two sampling locations differed in their dominant plant species: OA1 was characterized by a monospecific stand of C. rostrata, while E. angustifolium was the only vascular plant present at OA2 (see Fig. S1 in the supplemental material). At both locations, the water table level was 3 to 4 cm, and the submerged moss C. sarmentosum was observed in patches. At sampling location GA in the Göschener Alp fen, the above-ground vegetation consisted largely of C. rostrata, while C. sarmentosum was absent. The water table level was 0.5 to 1 cm above the soil surface, and the water was stagnant.

At these three locations, the vegetation covers appeared to be similarly dense at the time point of sampling (see Fig. S1 in the supplemental material). Nevertheless, analysis of the dry weight of total above-ground biomass revealed that at OA1, 152.6 ± 25.9 g m⁻² of C. rostrata biomass was present, while at GA, the C. rostrata biomass was only 115.7 ± 24.7 g m⁻² (Fig. 1a). Nevertheless, at both locations, ca. 86% (131.6 ± 8.4 and 99.9 ± 25.4 g m⁻², respectively) of the plant material was green, i.e., photosynthetically active. At OA2, the overall biomass of E. angustifolium was 104.8 ± 8.1 g m⁻², but only ca. 51% (53.8 ± 6.9 g m⁻²) was green. However, E. angustifolium leaves tended to turn reddish-brown toward the end of the growing season, and some biomass might have been considered “brown” while it was still photosynthetically active.

FIG 1.

(a) Above-ground green and brown biomass of vascular plants. (b) CH₄ emissions to the atmosphere normalized per square meter per day. Error bars represent standard deviations (n = 3).

Methane emissions.

The CH₄ emissions per square meter and day showed high variation between locations and were significantly lower [F(1, 7) = 14.5; P < 0.01] at OA1 (103.9 ± 20.9 mg CH₄ m⁻² day⁻¹) and OA2 (80.3 ± 10.1 mg CH₄ m⁻² day⁻¹) than at GA (184.4 ± 43.8 mg CH₄ m⁻² day⁻¹) (Fig. 1b). Some variation (as much as 23.7%) was observed between replicates within each location, yet the values were largely similar to those for emissions determined previously at the Oberaar and Göschener Alp fens (32, 33, 40). The diffusive upward fluxes through the water phase were estimated to be 6.4, 3.5, and 2.9 mg CH₄ m⁻² day⁻¹ at OA1, OA2, and GA, respectively, contributing a maximum of 6% to total emissions.

Physicochemical pore water properties.

Pore water CH₄ concentrations were generally lowest at OA1, followed by OA2 and GA (Fig. 2a). At all three locations, concentrations were low just below the water table (0.02 ± 0.01 to 0.09 ± 0.1 mg liter⁻¹) and increased significantly with depth (Table 3). The highest concentrations were measured between depths of 15 and 25 cm, ranging from 1.7 ± 0.3 mg liter⁻¹ (OA1 at a 25-cm depth) to 4.8 ± 2.3 mg liter⁻¹ (GA at a 15-cm depth). Pore water O₂ concentrations were highest within the first 5 cm (OA1 and OA2) or 2.5 cm (GA) below the water table, followed by a steep decrease, with values around or below 0.5 mg liter⁻¹, for the remainder of the depth profile (Fig. 2a).

FIG 2.

Physicochemical parameters of pore water extracted along the depth profile and structural analysis of soil samples. (a) Mean pore water CH₄ (▲) and O₂ (●) concentrations. (b) Mean DOC concentrations (▼) and temperature (◆). (c) Mean pH (○) and electrical conductivity (▶) of the pore water determined in situ along the depth profile. (d) Dry weight of debris (particles of <1 mm) (■) and gravimetric water content (✖) of the soil samples. Dashed lines indicate the approximate position of the soil surface. Error bars indicate standard deviations (n = 3).

TABLE 3.

Spearman's rank coefficients correlating physicochemical pore water properties^a with depth

Sampling site	Spearman's rank coefficient for correlation of the following pore water propertyb with depth:
	CH4	O2	DOC	Temp	pH	Conductivity
OA1	0.9762***	−0.7143	0.6905	−0.9048**	−0.8333*	0.7381*
OA2	0.9286**	−0.6429	0.8810**	−0.6667	−0.9524**	0.9286**
GA	0.9286**	−0.9222**	0.0476	−0.9286**	0.6108	0.9048**

^aDisplayed in Fig. 2.

^bMean values (n = 3) obtained at each depth (n = 8) were used. *, significant at the 0.05 level; **, significant at the 0.01 level; ***, significant at the 0.001 level.

Like CH₄ concentrations, pore water DOC concentrations were generally lower at OA1 and OA2 than at GA (Fig. 2b). At the latter location, the highest DOC concentration, 12.4 ± 0.02 mg liter⁻¹, was measured directly below the water table, but no clear trend with depth could be discerned. On the other hand, at OA1 and OA2, increases with depth were observed (Table 3). Temperatures generally decreased in the first 10 cm of the depth profile at all three locations, with the lowest temperatures at OA2 (Fig. 2b). In the deeper layers, temperatures were rather stable and comparable between the different locations, ranging from 7.7 to 8.7°C.

At all three locations, the pore water was acidic throughout the depth profile (Fig. 2c). At OA1 and particularly OA2, pH decreased significantly with depth (Table 3), reaching values of 5.0 ± 0.3. At GA, values were generally lower, but similar throughout the depth profiles. The electrical conductivity of the pore water showed some variation between replicates but was generally comparable between the different locations. Values ranged from 7.7 μS cm⁻¹ to 44 μS cm⁻¹ (Fig. 2c) and increased significantly with depth at all three locations (Table 3). Acetate concentrations were generally below 2 μM, and no differences between the three locations or along the depth profile could be discerned (data not shown).

Analysis of variance was performed with average values obtained from triplicate samples to identify the effect of the site (i.e., the Oberaar or the Göschener Alp) and plant species (i.e., C. rostrata or E. angustifolium) on the different physicochemical pore water properties. The results showed that the site had a significant effect on the CH₄ [F(1, 70) = 6.34; P < 0.05] and DOC [F(1, 70) = 116.96; P < 0.001] concentrations in pore water, while no significant effect of the plant species was discerned.

Soil water content and structure.

The gravimetric water content and structure of the soil were determined for the relative depths A, B, and C. Overall, at OA1 and OA2, the water content was high, with average values of 90.8% ± 2.8% and 94.0% ± 1.5%, respectively, while GA samples showed a lower water content (average, 82.4% ± 3.5%) (Fig. 2d). However, within each location, the water content differed little between the three different depths. Overall, the high water content suggests low soil density at the sampling locations.

At OA1 and OA2, particles of >1 mm made up >94% and >97%, respectively, of the total dry weight of the soil cubes analyzed (Fig. 2d). This observation indicates generally low levels of decomposition. The samples consisted mainly of plant material and were largely dominated by mosses, which were particularly abundant at OA2, with an apparent increase with depth. In contrast, in the GA samples, the debris content was higher, and particles of >1 mm made up 82 to 91% of the total dry weight, suggesting higher levels of degradation. The samples were characterized by a high relative surface area of unknown material, while mosses were largely absent, with the exception of a single GA_C replicate (see Fig. S3 in the supplemental material). Roots were present in all samples from the three different locations but generally showed higher surface areas in samples from the C. rostrata-dominated locations OA1 and GA than in those from the E. angustifolium-dominated location OA2 (see Fig. S3).

Abundance of methanotrophic communities.

For pmoA, amplification of genes and transcripts was possible from all DNA and most cDNA samples using both primer sets. Amplification of transcripts failed from one GA_C replicate with both primer sets and from two OA1_C replicates with primer set A189f–A650r. mmoX genes could be detected in most samples with primer sets mmoX206f–mmoX886r and mmoX1–mmoX2, but the yield was often low. On the other hand, mmoX transcripts could not be amplified with these primers, with the exception of a single OA2 sample, and thus, no further analyses were carried out. Primer set 534f–1393r resulted in multiple bands for most DNA samples and was not subsequently applied to the cDNA samples.

The abundances of pmoA genes, as determined with primer set A198f–mb661r, were largely similar in all samples independently of sampling location, ranging from (1.33 ± 2.43) × 10⁶ to (7.11 ± 1.48) × 10⁶ copies (g soil)⁻¹ (for OA2_C and OA2_A, respectively) (Fig. 3a). Transcripts of pmoA were generally less abundant than genes [(0.10 ± 0.08) × 10⁵ to (2.68 ± 4.05) × 10⁵ copies (g soil)⁻¹] and were particularly low in GA_A and GA_B. At OA1, a slightly decreasing trend of pmoA transcript abundance with depth was observed, while at OA2 and particularly at GA, the highest transcript abundance was detected in the deepest samples. Statistical analysis showed that pmoA transcript abundance was significantly influenced by the site [F(1, 24) = 4.90; P < 0.05], while pmoA gene abundance showed no significant differences with the site, plant species, or depth.

FIG 3.

Abundances of pmoA (a) and mcrA (b) genes (◆) and transcripts (♢) per gram of soil (wet weight), as determined by qPCR. Error bars indicate standard deviations (n = 3) and are shown only in the positive direction for better readability.

Abundance of methanogenic communities.

Amplification of mcrA genes and transcripts was possible for all samples. The abundance of mcrA genes was generally 1 order of magnitude higher than that of pmoA genes (Fig. 3b), with OA2_B and OA2_C showing particularly high copy numbers [(2.61 ± 0.68) × 10⁸ and (2.14 ± 2.42) × 10⁸ (g soil)⁻¹, respectively]. The abundance of mcrA transcripts was generally 1 to 2 orders of magnitude below the gene abundance (Fig. 3b). Variation between locations was rather low, with copy numbers ranging from (0.59 ± 0.60) × 10⁶ to (1.50 ± 0.97) × 10⁶ (g soil)⁻¹ in OA2_A and OA1_B, respectively.

A statistically significant effect of the plant species on gene abundance was discerned [F(1, 25) = 5.61; P < 0.05], while differences in mcrA transcript abundance between sites, plant species, or depths were not significant.

Composition of methanotrophic communities.

The composition of the methanotrophic communities, as determined by T-RFLP profiles of pmoA genes and transcripts amplified with primer set A198f–mb661r, was quite diverse (Fig. 4a). For the dominant T-RFs, primer set A198f–A650r gave generally comparable results, while overall fewer T-RFs were identified in most samples, particularly at the transcript level (see Fig. S4 in the supplemental material). Moreover, total peak areas were quite low for several transcript profiles. Therefore, further analyses were performed using only the profiles generated with primer set A198f–mb661r.

FIG 4.

Standardized T-RFLP profiles of pmoA genes and transcripts (amplified with primer pair A198f–mb661r) (a) and mcrA genes and transcripts (b) obtained from individual soil samples. The relative peak areas of individual T-RFs are shown. Assignment of T-RFs is based on in silico analysis of sequence data obtained from DNA extracts of selected samples in this study and previous studies (46).

Analysis of the recombinant libraries and comparison with previous studies (39, 46) enabled us to assign a range of T-RFs to type Ia, Ib, and II methanotrophs. T-RF 244, indicative of type II methanotrophs, dominated most samples at the gene level, while T-RF 79 (indicative of types Ia and Ib) was generally dominant at the transcript level (Fig. 4a). In addition, T-RF 241 (indicative of type Ib) was particularly abundant at GA.

Pairwise comparison of all individual samples revealed a grouping according to site, with several GA samples forming a distinct cluster (see Fig. S5a in the supplemental material). Moreover, separate clusters for genes and transcripts were observed, especially for OA1 and OA2. However, grouping according to the plant species or depth was not very strict, likely due to limitations of 1-dimensional comparisons. We therefore applied canonical-correlation analysis (CCA) to further investigate the observed clustering, using site, plant species, depth, and molecule as constraining parameters in the model. Constrained axes explained in sum 42.5% of inertia, strengthening the site-specific differences in methanotrophic communities, as well as separation by molecule (Fig. 5a). In addition, communities in samples from the C. rostrata-dominated locations appeared to be distinct from those in samples from the E. angustifolium-dominated location.

FIG 5.

Canonical-correlation analysis (CCA) of T-RFLP profiles of pmoA (a) and mcrA (b) genes (◆) and transcripts (♢) based on individual T-RFs (not shown). Constraints are the site (i.e., the Oberaar or the Göschener Alp), plant species (i.e., C. rostrata or E. angustifolium), depth (i.e., A, B, or C), and molecule (i.e., gene or transcript). For pmoA, constrained axes CCA 1 to 3 explained 24.0, 14.1, and 3.3% of total inertia, respectively; for mcrA, constrained axes CCA 1 to 3 explained 21.3, 8.8, and 3.0% of total inertia, respectively. Note that only data for CCA 1 and 2 are displayed. Testing CCA results by permutation (number of permutations used for assessing significance of constraints = 500) under a reduced model found all four constraints to be significant (P < 0.01) for both genes.

Further statistical analyses (by MANOVA) of the T-RFLP profiles revealed overall significant differences [F(4, 48) = 20.25; P < 0.001] between the total (genes) and active (transcripts) communities. Considering only pmoA genes, the plant species and site had significant effects on community composition. At the transcript level, the communities differed significantly with the plant species, site, and depth (Table 4; see also Table S2 in the supplemental material).

TABLE 4.

MANOVA analysis of T-RFLP profiles generated from pmoA and mcrA genes and transcripts, to test the effects of site, plant species, and depth on the compositions of methanotrophic and methanogenic communities^a

Variableb	Significancec by MANOVA for:
	Methanotrophs (pmoA)		Methanogens (mcrA)
	Genes	Transcripts	Genes	Transcripts
Site	***	***	***	***
Plant species	*	*	*	NS
Depth	NS	**	*	NS

^aFor values of λ, F, and P see Table S2 in the supplemental material.

^bThe site was the Oberaar or the Göschener Alp; the plant species, C. rostrata or E. angustifolium; the depth, A, B, or C.

^cNS, not significant; *, significant at the 0.05 level; **, significant at the 0.01 level; ***, significant at the 0.001 level.

Composition of methanogenic communities.

Methanogenic communities showed a high diversity of mcrA genes and transcripts. Based on sequence analysis, most T-RFs could be assigned to specific phylogenetic groups. Several T-RFs indicative of Methanosaetaceae and an uncultured fen cluster were abundant in most samples (Fig. 4b), but some differences between the sampling locations were observed. For example, T-RF 191 (Methanosaetaceae) was highly abundant in most OA1 and OA2 samples yet almost absent in GA samples. On the other hand, T-RF 61 (fen cluster) appeared more abundant in the C. rostrata-dominated locations GA and OA1 than in OA2.

Pairwise comparison resulted in a clear clustering according to the site, yet no separation by plant species or molecule was detected (see Fig. S5b in the supplemental material). In fact, transcript profiles generally seemed to be highly similar to the corresponding gene profiles. In addition, some clustering according to depth was observed. Constrained axes of CCA explained in sum 35.4% of inertia, strengthening site- and depth-specific differences, and also indicating some separation of communities according to the plant species (Fig. 5b).

MANOVA of all mcrA T-RFLP profiles revealed significant differences between the total (genes) and active (transcripts) methanogenic communities [F(4, 48) = 9.12; P < 0.001]. In contrast to the findings for methanotrophs, the active methanogenic communities were significantly influenced only by the site, not by the plant species or depth, while for the total communities, all three factors showed significant effects (Table 4; see also Table S2 in the supplemental material).

DISCUSSION

Selection of study sites and sampling locations.

In wetland systems, vascular plants play a crucial role in CH₄ production, oxidation, and emission, and differences in emissions have been reported for different plant species under controlled conditions (see, e.g., references 27 and 28). However, such studies in natural environments are scarce (17, 24, 25). With the Oberaar and Göschener Alp sites, we had the opportunity to analyze monospecific stands of two commonly found wetland plants in situ and to assess CH₄ emissions and methanotrophic and methanogenic communities associated with the different plant species.

The three fens under investigation (OA1, OA2, and GA) were all permanently submerged, located at similar elevations on siliceous bedrock, and characterized by comparable climatic conditions. Nevertheless, locations OA1 and GA differed in physicochemical pore water properties and soil structure, even though both were dominated by C. rostrata. At GA, decay of soil organic matter appeared to be more advanced (likely due to the stagnant water, in contrast with the flowing water at OA1), resulting in higher substrate availability for methanogens and ultimately in higher CH₄ concentrations in pore water and higher CH₄ emissions.

In contrast, locations OA1 and OA2 differed in their dominant plant species but were comparable in environmental conditions. Although the fraction of brown biomass was higher for E. angustifolium at OA2 than for C. rostrata at OA1, these two locations were similar in their physicochemical pore water and soil structural properties. In particular, DOC concentrations in the pore water and the debris contents of the soil were comparable, indicating equivalent levels of soil decomposition and highlighting the comparability of the systems.

Influence of the plant species on methanotrophic and methanogenic communities.

The compositions of the total (gene-based) and active (transcript-based) methanotrophic communities differed significantly with the plant species. Nevertheless, no plant species-specific T-RFs were discerned, and at all three locations, various type Ia and Ib methanotrophs appeared to dominate the active communities. Type I methanotrophs have frequently been reported as key players in habitats with high CH₄ source strength and nutrient levels, although their abundance is lower than that of type II methanotrophs (e.g., 52–54). Type II methanotrophs, on the other hand, are often present in dormant or resting stages but become active under conditions of nutrient limitation or other unfavorable conditions (54–56). In the highly methanogenic fens studied here, type I methanotrophs also seemed to be the key drivers of CH₄ oxidation at the time point of sampling, while type II methanotrophs were present but rather inactive. In fact, the apparent differences between gene and transcript abundances suggest that the majority of the cells were dormant or even nonviable, highlighting the importance of analyzing mRNA in microbial ecology studies. However, the possibility that these differences are partially due to mRNA degradation during the complex procedure of extraction from soil cannot be fully excluded (57). Expression of mmoX, as reported previously for acidic peat ecosystems (58), was not observed in our fens.

Plant species-specific differences in O₂ transportation properties and radial oxygen loss (ROL) in the rhizosphere have been reported for Eriophorum and Carex species. High ROL was observed for E. angustifolium (59), while C. rostrata appeared to lose O₂ only at the distal and lateral root zones (60). Although the steady-state concentrations of O₂ in pore water were comparable at our three study locations, higher O₂ levels along the root systems of E. angustifolium possibly influenced the differences observed in the methanotrophic communities.

The dominant plant species also appeared to have an influence in shaping the methanogenic communities. However, significant differences in abundance and composition between Carex- and Eriophorum-associated communities were observed only at the gene level; the effect of plant species on active communities was less pronounced. Differences with plant species have also been demonstrated for methanogenic communities at the DNA level in a greenhouse-based study (30) and for pmoA and mcrA gene abundances in a constructed wetland system (29). However, these studies have been performed under controlled conditions where the dominant plant species was the only major variable.

Influence of the site on methanotrophic and methanogenic communities.

In the natural field setting of our study, we observed that the site (i.e., the geographically and geochemically distinct site) appeared to have an even stronger influence on the communities than the dominant plant species. The total and active communities of methanotrophs and methanogens showed highly significant differences between the Oberaar and Göschener Alp sites. At location GA, decomposition of the soil organic matter was more advanced than at OA1 and OA2, and higher DOC and pore water CH₄ concentrations, as well as higher CH₄ emissions, were found. Surprisingly, the abundances of total and active methanogens at GA were similar to those at the other locations. The higher CH₄ emissions and pore water CH₄ concentrations at GA are therefore likely influenced by higher DOC concentrations, suggesting a larger substrate pool, and higher rates of methanogenesis mediated by the specific methanogenic communities present at this location. At GA, several fen cluster-specific T-RFs showed higher abundance, while at OA1 and OA2, acetoclastic Methanosaetaceae appeared to be more abundant, even though similar low acetate concentrations were observed in the pore water at all three locations. However, since the fen cluster is composed of uncultured organisms, no information is available on potentially higher turnover rates than those of other methanogens.

Influence of depth on methanotrophic and methanogenic communities.

Significant differences were also discerned between pmoA transcript-based communities present at different depths. Particularly, the C samples were distinct from the upper samples at all three locations. This observation is in agreement with a previous study at the Oberaar site, where we detected pronounced differences in the active methanotrophic communities between the upper 10 cm below the water table and the deeper layers (40). The active methanogenic communities also showed some changes with depth, but the differences were less pronounced, as in the previous study (40). Yet for both communities, depth-specific taxonomic groups could not be identified by the methodologies applied here, suggesting that further analyses at a higher taxonomic resolution are needed.

The convex and concave shapes of the steady-state pore water O₂ and CH₄ concentrations in the upper 10 cm below the water table and the high CH₄ concentrations measured at a depth of 15 cm suggest a dominance of CH₄ consumption in the A and B samples and a dominance of CH₄ production in the C samples. Nevertheless, within the OA1 and OA2 locations, transcripts of pmoA and mcrA were present in similar abundances at all three depths, confirming the coexistence of active methanotrophs and methanogens independently of pore water O₂ and CH₄ concentrations. In contrast, at the GA location, pmoA transcript abundances in the A and B samples were lower than those in the C samples and in all OA1 and OA2 samples. This difference might be influenced by the lack of mosses at GA. Symbiotic relationships between methanotrophic bacteria and mosses have been reported to reduce CH₄ emissions (e.g., 61–63). At the Oberaar site, viable mosses present in the upper 10 cm below the water table (viability demonstrated previously [40]) likely influenced the higher abundances of pmoA transcripts at OA1 and OA2 than at GA. Yet estimations of the diffusive CH₄ upward flux through the water phase suggest that at our study locations, at least 94% of CH₄ emissions occur through the aerenchyma of the vascular plants. The higher CH₄ emissions observed from the GA location cannot, therefore, be explained by a lower abundance of active methanotrophs in the upper soil layers.

Influence of plant species and biomass on CH₄ emissions.

Species-specific differences in CH₄ emission from Carex and Eriophorum monoliths (27, 28) have been attributed to higher rates of rhizospheric CH₄ oxidation by methanotrophs in the Eriophorum monoliths under controlled conditions, likely due to higher O₂ availability (28). Differences in steady-state O₂ concentrations between OA1 and OA2 were not obvious. However, higher pmoA transcript numbers were detected in the OA2_C samples than in the OA1_C samples, suggesting higher methanotrophic activity in the E. angustifolium rhizosphere than in the C. rostrata rhizosphere and, thus, overall lower CH₄ emissions. Nevertheless, the amount of green biomass of E. angustifolium was also less than that of C. rostrata. Therefore, conclusions on species-specific differences in CH₄ emissions from the Oberaar site have to be drawn with caution.

The biomass of the vascular plant generally affects CH₄ emissions, and positive correlations between biomass and emissions have been demonstrated and attributed to increased CH₄ conduction to the atmosphere, as well as to more-extensive root systems and therefore higher substrate availability for methanogens (17, 22, 32, 64). The importance of vascular plants as conduits for direct CH₄ release is highlighted in our study by the low contribution of the diffusive upward flux to overall emissions. We also conducted additional experiments close to OA1 at a later time point, showing that after cutting of the C. rostrata biomass below the water table, CH₄ emissions dropped to ca. 15% of the level at the same location prior to cutting. Nevertheless, the correlation with biomass was not strict in our study; higher emissions from GA than from OA1 were observed, despite a smaller amount of C. rostrata biomass at the former location. As discussed above, these differences can be attributed to differences in subsurface properties.

Conclusions.

The present study allowed comparison of methanotrophic and methanogenic communities associated with different vascular plant species at the field scale. Our results suggest that within the same site (i.e., the same environmental and physicochemical settings), the dominant plant species may affect not only CH₄ emissions but also the composition of methanotrophic and methanogenic communities present in the respective soil. This influence appears to be more pronounced on the active methanotrophs than on methanogens, likely due to differences in O₂ availability in the rhizosphere. Nevertheless, considering the different sites studied here, other environmental and physicochemical parameters appear to have a stronger effect in shaping the communities than the dominant plant species. Our findings highlight the importance of assessing natural conditions when investigating microbial ecology in complex environmental systems.