Methane and sulfate profiles within the subsurface of a tidal flat are reflected by the distribution of sulfate-reducing bacteria and methanogenic archaea

Abstract

The anoxic layers of marine sediments are dominated by sulfate reduction and methanogenesis as the main terminal oxidation processes. The aim of this study was to analyze the vertical succession of microbial populations involved in these processes along the first 4.5 m of a tidal-flat sediment. Therefore, a quantitative PCR approach was applied using primers targeting the domains of Bacteria and Archaea, and key functional genes for sulfate reduction (dsrA) and methanogenesis (mcrA). The sampling site was characterized by an unusual sulfate peak at 250 cm depth resulting in separate sulfate–methane transition zones. Methane and sulfate profiles were diametrically opposed, with a methane maximum in the sulfate-depleted zone showing high numbers of archaea and methanogens. The methane–sulfate interfaces harbored elevated numbers of sulfate reducers, and revealed a slight increase in mcrA and archaeal 16S rRNA genes, suggesting sulfate-dependent anaerobic oxidation of methane. A diversity analysis of both functional genes by PCR-denaturing gradient gel electrophoresis revealed a vertical succession of subpopulations that were governed by geochemical and sedimentologic conditions. Along the upper 200 cm, sulfate-reducing populations appeared quite uniform and were dominated by the Deltaproteobacteria. In the layers beneath, an apparent increase in diversity and a shift to the Firmicutes as the predominant group was observed.

denaturing gradient gel electrophoresis, functional genes, marine subsurface, real-time PCR, tidal flats

Introduction

Coastal marine environments such as estuaries and tidal areas represent an important link between land and the open sea. They receive nutrient input from both, and as a result are characterized by intense primary production and heterotrophic activity (Dittman, 1999; Poremba et al., 1999; Kim et al., 2005). As a consequence, pronounced microbial activities in the upper sediment layers generate steep chemical gradients. Oxygen is depleted within the uppermost few millimeters, and anoxic conditions prevail beneath this layer (Böttcher et al., 2000). Sulfate reduction is considered to be the most important process in organic matter remineralization (Jorgensen, 1982; Llobet-Brossa et al., 2002). The dissimilatory reduction of sulfate can be linked to the oxidation of substrates that are difficult to degrade under anoxic conditions, such as alkanes and aromatic compounds (Hansen, 1994), or even to the anaerobic oxidation of methane at sulfate–methane transition zones (Iversen & Jorgensen, 1985; Thomsen et al., 2001; Ishii et al., 2004). Generally, methanogenesis becomes the dominant terminal oxidation process when sulfate is depleted. In most sediments, the sulfate-depleted zone is located at a depth of tens of centimeters to several meters. These subsurface sediment layers became a focus of microbiological investigations only recently (Thomsen et al., 2001). The community composition of subsurface tidal-flat sediments was studied by a cultivation-based approach (Köpke et al., 2005) and by a phylogenetic survey using domain-specific primers (Wilms et al., 2006a, b).

A more directed strategy to study the distribution of physiologic groups in the environment is the analysis of functional genes such as those that encode for the dissimilatory sulfate reductase (dsr) and the methyl-coenzyme M reductase (mcr) (Lueders et al., 2001; Dhillon et al., 2003; Earl et al., 2003; Kondo et al., 2004; Leloup et al., 2004, 2006).

In the present work, we provide a detailed analysis of the sulfate-reducing and methanogenic communities from the surface to a depth of 5 m in a tidal-flat sediment. Two complementary approaches were applied. Quantitative PCR (qPCR) (Higuchi et al., 1993) was used to estimate the abundance of the Bacteria and Archaea, as well as of dsrA and mcrA genes. The resulting depth profiles were correlated with those of porewater sulfate and methane. The quantitative approach was complemented by a qualitative diversity assessment of both metabolic groups, using functional primers for PCR-denaturing gradient gel electrophoresis (DGGE).

Materials and methods

Sample collection

The sampling site Neuharlingersieler Nacken (53°43′270N and 07°43′718E) is located in the back barrier tidal area of the island of Spiekeroog, which is part of the German Wadden Sea. The biogeochemical settings of this site have been described recently (Köpke et al., 2005; Wilms et al., 2006a). Sediment cores up to 5 m in length were collected by vibrocoring using aluminum tubes (diameter 8 cm) in June 2002 (core A), October 2003 (core B), February 2004 (core C) and September 2005 (core D). The liners of cores A–C were cut longitudinally, whereas core D was cut directly across to avoid degassing of methane during sampling. Subcores for further analyses were taken using cut-off 5-mL syringes from the innermost part of the cores every 20 cm, beginning at a depth of 40 cm. Additionally, shorter sediment cores were taken by hand at the same position to obtain undisturbed subsamples from the sediment surface and a depth of 20 cm. After sampling, all subcores taken for molecular analysis were stored at −20°C until further processing.

Sulfate and methane measurements

Porewater was gained from sediment samples by centrifugation and filtration through 0.2-μm membrane filters. Porewater sulfate concentrations were measured by ion chromatography with conductivity detection (Sykam, Gilching, Germany) as described previously (Sass et al., 2001).

For measuring methane concentrations, 2 cm³ of sediment was added immediately after subsampling to 20 mL of sodium hydroxide solution (2.5%) in gas-tight tubes. Headspace samples (20 μL) were analyzed on a Varian CX 3400 gas chromatograph (Varian, Darmstadt, Germany) equipped with a Plot Fused Silica column (No. 7517; 25 m by 0.53 mm, Al₂O₃/KCl coated; Chromopack, Middleburg, the Netherlands) and a flame ionization detector.

Total cell counts

Sediment samples were fixed by the addition of glutaric dialdehyde (0.2-μm filtered, 2% final concentration) and stored at 4°C in the dark. Prior to analysis, Tween-80 (0.2-μm filtered, 0.01% final concentration) was added to the fixed sediment slurries, and the samples were ultrasonicated (3 × 10 s). An aliquot (5–10 μL) was diluted 1000-fold in particle-free phosphate-buffered saline (0.9 g of NaCl, sodium phosphate buffer 15 mM, pH 7.4, 0.2-μm filtered), thoroughly shaken, and filtered through a white polycarbonate membrane (0.2-μm pore size, 25 mm in diameter, Anodisc 25, Whatman, Maidstone, UK). Cell staining with 4,6-diamidino-2-phenylindol (DAPI), and counting was performed according to Süß. (2004).

DNA extraction and quantification

Total genomic DNA was extracted from 0.5 g of sediment of each subcore using the FastDNA SpinKit (Q-BIOgene, Carlsbad, Canada). DNA concentrations were quantified fluorometrically in a microtiterplate reader (FLUOstar Optima, BMG Labtechnologies, Offenburg, Germany) using a 1 : 200 diluted PicoGreen reagent according to a modified manufacturer's protocol (Molecular Probes, Eugene, OR). In contrast to the original instructions, only the tenth part of each volume and 1 μL of the extracted DNA and lambda-DNA in different concentrations from 100 ng μL⁻¹ to 1 ng μL⁻¹ were used.

Quantitative PCR

Primer sets specific for different phylogenetic domains and functional genes were used for the quantitative (real-time) PCR (qPCR) assay (Table 1) (Lane, 1991; Muyzer et al., 1995; Hales et al., 1996; Wagner et al., 1998; Vetriani et al., 1999; Dhillon et al., 2003). Before quantification, serial dilutions of standard DNA had to be prepared. The standard organism for the bacterial 16S rRNA gene and dsrA gene approach was Desulfovibrio vulgaris^T (DSM 644). A culture of Methanosarcina barkeri^T (DSM 800) was used as a standard for archaeal 16S rRNA gene and mcrA gene quantifications. Genomic DNA was extracted from liquid cultures using the FastDNA SpinKit (Q-BIOgene, Carlsbad, Canada). These DNA extracts were amplified using the bacterial primer pair 8f and 1492r or the archaeal primer pair S-D-Arch-0025-a-S-17 and S-^*-Univ-1517-a-A-21 (Table 1), respectively. Conditions for the Eubacteria-specific PCR were described by Süß. (2004), and those for the Archaea-specific approach by Vertriani et al. (1999). The PCR amplicons were purified and adjusted to a volume of 50 μL using the PCR purification kit, in accordance with the manufacturer's instructions (Qiagen, Hilden, Germany), and served as a target for the qPCR standard curves. For this, they were diluted from 1 : 10³ to 1 : 10¹⁰. As a standard for the functional genes, the total extracted genomic DNA of Desulfovibrio vulgaris^T and Methanosarcina barkeri^T were used in dilutions of 1 : 10¹–1 : 10⁷.

1

Oligonucleotide sequences used in the denaturing gradient gel electrophoresis (DGGE) and quantitative PCR (qPCR) approaches

Target	Primer (Reference)	DNA sequence (bp)	Product size	Annealing-temperature/ fluorescence measured (°C)	Approach
dsrA gene	dsr-1F Wagner et al. (1998)	5′-ACSCACTGGAAGCACG-3′	c. 1900	58	DGGE
	dsr-4R Wagner et al. (1998)	5′-GTGTAGCAGTTACCGCA-3′
dsrA gene	dsr-1F Wagner et al. (1998)	5′-ACSCACTGGAAGCACG-3′	c. 500	60	DGGE
	GC-dsr-500r Dhillon et al. (2003)*	5′-GC-clamp-CGGTGMAGYTCRTCCTG-3′
mcrA gene	mcrA1f Luton et al. (2002)	5′-GGTGGTGTMGGATTCACA	c. 500	58	DGGE
		CARTAYGCWACAGC-3′
	GC-mcrA500r Luton et al. (2002)	5′-GC-clamp-TTCATTGCRTAG
		TTWGGRTAGTT-5′
Bacteria	8f Lane (1991)	5′-AGAGTTTGATCCTGGCTCAG-3′	c. 1500	54	qPCR
	1492r Lane (1991)	5′-GGTTACCTTGTTACGACTT-3′			Standard
Archaea	S-D-Arch-0025-a-S-17 Vetriani et al. (1999)	5′-CTGGTTGATCCTGCCAG-3′	c. 1500	48	qPCR
	S-*-Univ-1517-a-A-21 Vetriani et al. (1999)	5′-ACGGCTACCTTGTTACGACTT-3′			Standard
Bacteria	519f Lane (1991)	5′-GCCAGCAGCCGCGGTAAT-3′	c. 390	50/82	qPCR
	907r Muyzer et al. (1995)	5′-CCGTCAATTCCTTTGAGTTT-3′
Archaea	S-D-Arch-0025-a-S-17 Vetriani et al. (1999)	5′-CTGGTTGATCCTGCCAG-3′	c. 320	48/80	qPCR
	S-D-Arch-0344-a-S-20 Vetriani et al. (1999)	5′-ACGGGGCGCAGCAGGCGCGA-3′
dsrA gene	dsr-1F Wagner et al. (1998)	5′-ACSCACTGGAAGCACG-3′	c. 450	58/82	qPCR
	dsr-500r Dhillon et al. (2003)*	5′-CGGTGMAGYTCRTCCTG-3′
mcrA gene	ME1f Hales et al. (1996)	5′-GCMATGCARATHGGWATGTC-3′	c. 300	54/82	qPCR
	ME3r Hales et al. (1996)	5′-TGTGTGAASCCKACDCCACC-3′

*

Reverse and complementary.

qPCR amplification was performed in a volume of 25 μL containing: 12.5 μL of the DyNAmo HS SYBR Green qPCR Kit (Finnzymes Oy, Espoo, Finland), 0.2 mM each primer, 0.6 ng μL⁻¹ bovine serum albumin (BSA) (only 0.2 ng μL⁻¹ BSA was used for the bacterial approach), and 10 μL of the 1 : 10 diluted DNA templates. Thermal cycling was performed using a Rotor-Gene, RG-3000 four-channel multiplexing system (Corbett Research, Sydney, Australia) with the following parameters: 95°C initial hold for 15 min to activate the Taq polymerase, followed by 50 cycles of amplification, with each cycle consisting of denaturation at 94°C for 10 s, followed by 20 s of annealing at primer-specific temperatures (Table 1), and an extension step of 30 s at 72°C. Fluorescence was measured at the end of each amplification cycle for 20 s at temperatures given in Table 1.

To verify the results, every quantification was repeated three times at the same concentrations of all chemicals and templates. To convert the detected gene targets into cell numbers, averages of 3.8 and 2 copies of the 16S rRNA gene were estimated for bacteria and archaea, respectively (Fogel et al., 1999).

PCR-DGGE analysis

Amplification of dsrA gene fragments was performed via nested PCR. For this, a part of the dsrAB operon (c. 1.9 kb) was amplified by a first step using the modified primers dsr1F and dsr4R (Table 1). For the second PCR, primers GC-dsr500r and dsr1F were used to amplify dsrA gene fragments of suitable lengths for DGGE analysis (c. 500 bp). The GC-clamp-containing primer GC-dsr500r was originally published as the forward primer 1F1 by Dhillon et al. (2003). In our approach, it was used as a reverse primer. The partial mcrA gene (c. 500 bp) was amplified directly by the use of primer mcrA1f and the GC-clamp-containing primer GC-mcrA500r (Luton et al., 2002). PCR mixtures (50 μL) were composed of 0.2 mM dNTPs, 1.5 mM MgCl₂, 0.2 mM each primer, 1 × Red Taq Buffer (Sigma, Munich, Germany), 0.6 ng μL⁻¹ BSA, 1 U of Red Taq DNA polymerase (Sigma, Munich, Germany) and 1 μL of target DNA. Templates for the second step of the nested PCR were 1 μL of 1 : 500 dilutions of amplicons obtained by the first PCR. For both amplification steps for the dsrA gene, PCR was set to 30 cycles, whereas for the mcrA gene, 40 cycles were applied. PCR reactions were carried out in a thermal cycler (Mastercycler, Eppendorf, Hamburg, Germany) under the following conditions: a denaturation temperature of 96°C for 30 s, the respective primer-specific annealing temperature (Table 1) for 45 s, and an elongation temperature of 72°C for 1 min, followed by a final elongation step at 72°C for 10 min. PCR products were checked by agarose electrophoresis, purified, and adjusted to a volume of 10 μL using the MinElute PCR purification Kit (Qiagen, Hilden, Germany). The amplicons were separated by DGGE on an INGENYphorU-2 system (Ingeny, Goes, the Netherlands) using a denaturing gradient from 40% to 70% (where 100% denaturant contains 7 M urea and 40% formamide). Electrophoresis and DGGE pattern analysis was performed as described previously (Wilms et al., 2006a).

Sequencing and phylogenetic analysis

DGGE bands obtained by the dsrA gene approach were excised, treated and sequenced as described previously (Del Panno et al., 2005). The PCR protocol was modified for reamplification: 96°C for 30 s, 60°C for 45 s and 72°C for 1 min (25 cycles), and a final elongation at 72°C for 10 min. The phylogenetic calculations were conducted by applying the phylogenetic analysis software arb (Ludwig et al., 2004) using reference sequences. All partial dsrA gene sequences obtained in this study have been deposited in the EMBL database under accession numbers AM234729–AM234743.

Results

Sulfate and methane profiles

Previous investigations of the sampling site (cores A–C) had revealed relatively uniform lithologic profiles, with sand-dominated upper layers, mud at the bottom of the cores, and an intermediate shell layer (c. 170–240 cm in depth) (Wilms et al., 2006b). This shell layer corresponded to a subsurface sulfate maximum, indicating a possible influx of sulfate-rich water from a nearby tidal creek. The porewater sulfate profiles taken in four sampling campaigns were also highly similar, with minor variations at the sediment surface and in the magnitude of the deep sulfate peak. Therefore, the methane profile that was measured in September 2005 (core D) can be assumed to also reflect the situation in cores A–C.

Sulfate concentrations at the surface were c. 28 mM, and the concentrations rapidly decreased to values below 1 mM beneath a depth of 50 cm (Fig. 1a). Methane concentrations at the sediment surface did not exceed 0.5 nmol cm⁻³, whereas in the sulfate minimum zone, concentrations of up to 125 nmol cm⁻³ were measured. In the zone of the subsurface, sulfate maximum between 200 and 380 cm depth methane concentrations declined again to values c. 5 nmol cm⁻³. Beneath a depth of 400 cm, sulfate was depleted to concentrations below 0.2 mM, whereas methane increased again to values of 100 nmol cm⁻³. These unusual profiles resulted in the presence of three sulfate–methane transition zones at depths around 100, 220 and 370 cm. Standard deviations between the investigated cores were c. 0.04 mM for low sulfate concentrations and up to 4.26 mM for high values.

1

Depth profiles of site Neuharlingersieler Nacken. Higher values of microorganisms and functional genes were determined at the sulfate–methane transition zones (100 and 200 cm). Squares, values for core A; triangles, values for core B; circles, values for core C; and diamonds, values for core D. The curves show average values. (a) Sulfate (black symbols) and methane (gray symbols) profiles along the sediment core. (b) Calculated numbers of bacteria (black symbols) and archaea (gray symbols) per gram of sediment. (c) Calculated numbers of dsrA genes (black symbols) and mcrA genes (gray symbols) per gram of sediment.

Quantification of prokaryotes and functional genes

Similar to the relatively minor variations in geochemical gradients, total cell counts in the different cores differed only slightly, as revealed by epifluorescence microscopy. In surface sediment generally, 3–5 × 10⁸ cells cm⁻³ were found. Numbers declined to 10⁸–10⁷ cells cm⁻³ at a depth of 300 cm (data not shown).

The ratio of archaeal to bacterial 16S rRNA gene copies as determined by qPCR generally increased from 1 : 100 at the surface to 1 : 3 at a depth of 450 cm. The calculated standard deviations for replicate quantifications of one core were mainly between 10 and 50%. Differences between corresponding layers of all cores were slightly higher. Estimated bacterial numbers were highest at the sediment surface, whereas archaeal numbers peaked at a depth of 140 cm (Fig. 1b). Beneath a depth of 200 cm, the estimated cumulative prokaryotic cell numbers decreased to 10⁶ cells g⁻¹ of sediment. In the sulfate-depleted zone (140–180 cm) and the shell layer, bacterial numbers were consistently low, but they strongly decreased within the mud layers beneath. The highest numbers of archaeal targets were detected in the sulfate-depleted layers, and consistently declined within the shell and mud layers.

The depth profiles of the functional genes representative of sulfate-reducing prokaryotes and methanogenic archaea generally correlated well with those of the bacterial and archaeal 16S rRNA genes (Fig. 1c). High values for the dsrA and mcrA genes were obtained from near-surface sediments. The decrease with depth in the numbers of dsrA genes followed the general bacterial trend. The peak of archaeal 16S rRNA gene copies at a depth of c. 140 cm was well reflected by the profile obtained for the mcrA gene. At the upper two sulfate–methane transition zones, elevated numbers of dsrA gene copies were detected, suggesting anaerobic methane oxidation coupled to sulfate reduction in these layers. These local maxima at depths of c. 100 and 200 cm coincided with a faint local maximum of mcrA as well as of bacterial and archaeal 16S rRNA gene copies. In the deep mud layers, the numbers detected were generally at least one order of magnitude lower than in the shell and sand layers, so no clear maxima could be recognized.

The ratio of dsrA genes and bacterial 16S rRNA gene numbers ranged from 1 : 50 to 1 : 20 throughout the entire sediment column. Surprisingly, the ratio of mcrA genes and archaeal 16S rRNA targets decreased from 1 : 3 to 1 : 7 in the sandy sediment layers to less than 1 : 100 at a depth of 300 cm.

DGGE analysis of mcrA and dsrA genes

The diversity analysis of mcrA and dsrA genes by DGGE resulted in distinct clusters reflecting different compartments of the sediment column (Figs 2 and 3). These are characterized by geochemical settings. Samples obtained from corresponding depths of the different cores (A–C) revealed very similar patterns, and generally showed less overlap with samples from other depths.

2

Cluster analysis of denaturing gradient gel electrophoresis (DGGE) band patterns with specific primers for the mcrA gene. The zone from 220 to 280 cm had a different community composition. The other depths were characterized by low sulfate concentrations, reflecting the competition between sulfate reducers and methanogens. The dendrogram was calculated by Pearson correlation and upgma. Spatial variations in community composition (marked by an asterisk) were reflected by the affiliation of single sediment layers of a given core to different layers of other cores.

3

Cluster analysis of denaturing gradient gel electrophoresis (DGGE) band patterns with specific primers for the dsrA gene and positions of bands sequenced. A different sulfate-reducing community composition was found at depths of 280 and 320 cm than in other layers, reflecting decreasing sulfate and low methane concentrations. Bands revealing sequences that are closely related to the same dsrA gene are marked by the same letter. Phylogenetic affiliation, sequence similarities and depth distribution are given in Table 2.

One cluster of mcrA gene patterns was detected at depths of 220 and 280 cm, characterized by the deep sulfate peak and decreasing methane concentrations. The other patterns formed a second cluster. Because of outliers (marked by an asterisk in Fig. 2), subclusters of other compartments are not clearly visible. However, layers containing high and low methane concentrations seem to possess similar mcrA genes. Whereas band patterns of dsrA amplicons yielded different clusters, they correlated with the sedimentologic conditions rather than with geochemical gradients (Fig. 3). The sand- and shell-dominated upper layers (down to a depth of 220 cm) revealed relatively uniform patterns, with a single dominant and a few faint additional bands. A major shift was observed for the mud layers from 220 cm downwards. They were characterized by a higher variability and apparently increasing band numbers.

Sequencing of the main dsrA amplicons along the upper 220 cm (band A, Fig. 3) revealed a similarity of 75% to the dsrA gene from the deltaproteobacterium Desulforhopalus singaporensis^T (Table 2) and of 90% to the Wadden Sea clone DSR_SpoogII_043_C07 (Mussmann, personal communication). Most sequences originating from the layers beneath a depth of 220 cm (band C) were closely related to the dsrA gene of the firmicute Desulfotomaculum thermosapovorans^T (77% similarity). Three other sequences with low similarities among each other were affiliated to Desulfobacca acetoxidans^T as closest described phylotype (67–71% similarity).

2

Overview of dsrA gene phylotypes from sulfate-reducing prokaryotes detected by PCR-denaturing gradient gel electrophoresis (DGGE). The letters and numbers correspond to the appearance and position of the different detected phylotypes given in Fig. 3. DsrA sequences belonging to the same phylotype are labeled with the same letter. Percentages in brackets are indicative for similarities between multiple detected dsrA sequences. For each phylotype, the most closely related sequence, the closest cultivated organism and its appearance with depth are given.

Appearance of dsrA sequences	Closest described relative	Similarity (%)	Closest phylotype	Similarity (%)	Distribution with depth (cm)
					0	160–180	220–360
A (99%)	Unc.* sulfate-reducing bacterium mCR119	84	Desulforhopalus singaporensisT	75	+	+
B (98%)	Unc. sulfate-reducing bacterium VHS057	82	Desulfonatronum lacustreT	73	+	+
C (91–100%)	Unc. sulfate-reducing bacterium UI-DSR43	77	Desulfotomaculum thermosapovoransT	77	+	+	+
D	Unc. sulfate-reducing bacterium KA064	85	Desulfobacca acetoxidansT	71		+
E	Unc. sulfate-reducing bacterium KS03	78	Desulfobacca acetoxidansT	67		+
F	Unc. sulfate-reducing bacterium xCR031	77	Desulfonema limicolaT	73		+
G	Unc. sulfate-reducing bacterium Sed2-DSR11	86	Desulfobacca acetoxidansT	67		+	+

*

Unc., uncultured.

Discussion

In the present study, the population sizes of bacteria, archaea, sulfate reducers and methanogens along the geochemical gradients of tidal-flat sediment were estimated by applying qPCR. It turned out that the population sizes of the different groups correlated well with the respective geochemical gradients, and that anaerobic methane oxidation coupled to sulfate reduction might occur in these sediments.

Prokaryote quantification in sediments

In our investigation, total cell counts in deeper layers showed higher values than calculated after qPCR analysis. It is possible that total cell counts obtained by DAPI staining overestimate the actual in situ numbers by including cell-like particles. These false-positive signals increase with decreasing cell numbers, due to the need to concentrate sample material. A novel method with neglectable background signals using SYBR green as fluorescence dye was developed by Lunau et al. (2005). This technique yielded the same cell numbers for Wadden Sea surface sediments (Lunau and Hilker, personal communications) as our direct counts by DAPI, whereas for deeper layers, a 10-fold lower cell number was counted. These findings would correspond much better with the numbers of prokaryotes calculated by qPCR. However, there are other potential reasons for differences between direct counts and those obtained by qPCR, such as varying DNA extraction efficiencies, coextraction of PCR-interfering humic-like substances, or varying 16S rRNA gene copy numbers (Fogel et al., 1999).

Abundance of methanogens within the subsurface

Functional genes indicative for methanogens and sulfate reducers were detected along the entire sediment column from the surface down to a depth of 450 cm. This is surprising, because both physiologic groups generally compete for the same substrates and hence are thought to not co-occur. However, methylothrophic methanogens might escape direct competition with sulfate reducers by utilization of certain methylated substrates (e.g. methylamines, dimethylsulfide) not utilized by the latter (Madigan et al., 2003). On the other hand, the relatively high numbers of mcrA targets detected just beneath the oxidized surface sediments could also be a result of occasionally occurring high levels of organic matter supplied by sedimentation. This short-term substrate surplus may rid the methanogens of direct competition and offer a short-term niche (Oremland & Polcin, 1982; Fitzsimons et al., 2005). The high numbers of archaea and mcrA gene targets between depths of 100 and 200 cm indicate that these organisms are responsible for the methane peak within the sulfate-depleted zone.

At first sight, it seems surprising that the high methane values at a depth of 450 cm were not accompanied by strongly increasing numbers of methanogens or mcrA genes. However, this can be partly explained by sedimentologic parameters. Compared to the coarse sand and shell layers, the fine-grained mud has a very low hydraulic conductivity, diminishing diffusive fluxes and leading to relatively steep chemical gradients. On the other hand, cell counts obtained by epifluorescence microscopy were already more than one order of magnitude lower than those for the sand layers. This trend was supported by the results of the qPCR, which indicated an even stronger decrease. A potential bias may also be caused by insufficient coverage of the target phyla by the PCR primers used. In a previous study, we detected a new branch within the Euryarchaeota present in the deep mud layers by applying primers targeting eukaryotic rRNA genes (Wilms et al., 2006b). These archaea were not detected in the assays using archaeal primers, suggesting a potential underestimation of the actual in situ numbers of some groups of the Archaea.

Relative abundance of sulfate-reducing bacteria

The ratio of dsr and bacterial 16S rRNA targets in deeper sediment layers of site Neuharlingersieler Nacken was similar to that at the sediment surface (c. 5%). However, the described qPCR bias for the quantification of prokaryotes and functional genes should be roughly the same. Therefore, the calculated percentage of sulfate reducers does not change significantly. Even though they only account for 5–10% of the whole community, they contribute greatly to biochemical processes within subsurface sediments. However, the calculated number was two to five times lower than described for other marine sediments (Sahm et al., 1999; Mußmann et al., 2005). The discrepancy might be explained by the different methodologic approaches used in these studies. Applying fluorescence in situ hybridization combined with catalyzed reporter deposition (CARD-FISH), Mußmann et al. (2005) investigated the uppermost layers of an adjacent tidal flat. Sulfate reducers were estimated to represent c. 10% of the total cell count, which is in the same order of magnitude as our results. The deviations might be due to the different sedimentologic conditions of the investigated sampling sites. Sahm et al. (1999) used slot-blot hybridization to calculate a proportion of sulfate-reducing bacteria (SRB) on the total prokaryotic rRNA of 18–25% in sediments of Aarhus Bay, Denmark. However, this calculation is based on the assumption that all microorganisms contain the same amount of ribosomal RNA. This favors the quantification of microorganisms with higher numbers of ribosomes. In general, the method of choice depends on the question to be answered. DNA-based qPCR and CARD-FISH show comparable results when certain abundancies are determined. The big advantage of analyzing RNA rather than DNA is that conclusions can specifically be drawn concerning active members of the microbial communities.

Diversity of the sulfate-reducing populations

At first sight the findings obtained by DGGE analysis suggest an increasing diversity of sulfate reducers with depth. A vertically structured community of sulfate-reducing bacteria was found that was governed by more than a single environmental factor. Cluster analysis reflected the sedimentologic conditions to a certain extent. The sand and shell layers grouped together, and the mud layers formed a separate cluster. However, similar patterns were found for sulfate-rich surface layers and the sulfate-depleted zone (160–180 cm). In this respect, our results differ from previous reports, which described changing communities of sulfate-reducing bacteria in surface sediments along an estuarine salinity gradient and concomitantly increasing sulfate concentrations (Nedwell et al., 2004; Leloup et al., 2006).

In general, the detected shift from sulfate-reducing bacteria affiliated to the Deltaproteobacteria to those of the Firmicutes is an agreement with previous analyses of the same sampling site. With a molecular approach, a shift from the Proteobacteria to the Chloroflexi with depth indicated similar properties as found in deep biosphere habitats (Wilms et al., 2006a). With a cultivation-based approach, a shift to the Firmicutes with depth was found, indicating the presence of inactive spores (Köpke et al., 2005). Thus, the detection of dsrA genes derived from the Firmicutes might also be due to the accumulation of spores in deeper layers. The assessment of active microorganisms can only be achieved by RNA-based methods. However, the analysis of sulfate-reducing communities via dsrA sequence analysis is prone to two more shortcomings: the limited sequence length (c. 350 bp), and – despite extensive recent work (Wagner et al., 1998; Loy et al., 2002) – still relatively low numbers of dsr sequences in the databases. This hinders the correct affiliation of a detected sequence to a certain phylogenetic group. On the other hand, the dsrA gene sequences detected by the DGGE approach can be expected to be present in significant numbers. In general, as dsrA gene sequences must represent 1% or more to be detected as DGGE bands (Murray et al., 1996). Therefore, the results presented here point in the same direction as those obtained by CARD-FISH (Mußmann et al., 2005). Mußmann et al.. found that the Desulfobulbaceae (including Desulforhopalus spp.) represented up to 4% of the bacterial community, whereas the most abundant SRB belonged to the Desulfosarcinales. The latter were present in the DGGE analysis as well, but produced only a relatively faint band (Fig. 3, band F; related to Desulfonema limicola).

Heterogeneity of the dsrA gene sequences

Although the total number of DGGE bands increased with depth, only two relatively homogeneous groups of sequences were found. The sequences of nine dsrA gene fragments were affiliated to the dsrA gene of Desulfotomaculum thermosapovorans^T, but showed different migration behavior in the DGGE gel. This genomic diversity was detected because the DGGE approach, in principle, can separate bands differing only by a single nucleotide (Fischer & Lerman, 1983). According to the wobble hypothesis (Crick, 1966), sequence mutations of functional genes occur more frequently than mutations in the highly conserved 16S rRNA gene. Sequence variations of the Desulfotomaculum thermosapovorans^T-affiliated dsrA genes were, in 75% of all cases, caused by a mutation on the third nucleotide of the triplet code, and obviously had less detrimental impact. At present, we can only speculate about the functional meaning of the remaining microheterogeneity of the dsrA sequences; it might reflect adaptation to environmental conditions and occupation of different ecologic niches (Jaspers & Overmann, 2004).

Anaerobic methane oxidation

In several investigated sediments, elevated numbers of sulfate reducers and methanogens within sulfate–methane transition zones suggest anaerobic methane oxidation coupled to sulfate reduction (Iversen & Jorgensen, 1985; Thomsen et al., 2001; D'Hondt et al., 2004; Parkes et al., 2005). This process might also occur at site Neuharlingersieler Nacken. This is not only indicated by the elevated numbers of mcrA and dsrA gene copies at the sulfate–methane transition zones; the steep slope of the methane profile at the upper edge of the subsurface sulfate peak at a depth of 170–220 cm suggests a consumptive process.

Highly specialized consortia of sulfate reducers (Desulfosarcina or Desulfococcus) and members of the archaeal ANME groups have been supposed to conduct this reaction in sediments overlying methane hydrates (Boetius et al., 2000; Orphan et al., 2002). Their molecular signatures have also been found in coastal sediments from Aarhus Bay (Thomsen et al., 2001). Although the consortia were just recently detected by CARD-FISH in a nearby sand flat (Ishii et al., 2004), at present we do not know whether other phylogenetic groups might be involved in this process at site Neuharlingersieler Nacken. At the sulfate–methane interface of a deep sub-seafloor habitat, Parkes et al. (2005) did not find members of the ANME groups as well. Here, the archaeal community was dominated by members of the Methanobacteriales and Methanosarcinales group, and this was also found for the subsurface of site Neuharlingersieler Nacken (Wilms et al., 2006b).

Ecologic relevance and outlook

Our results indicate that methane production within sulfate-depleted layers of the subsurface of tidal-flat sediments is mediated by methanogenic archaea. Even though elevated methane concentrations were determined in the water column above these sediments (Grunwald, personal communication), the methane profile suggests a consumptive process within the sediment. One explanation is microbial methane oxidation that controls the methane emission. Without this process, the concentration of this climatically active gas might be even higher. A detailed investigation of the methane flux and the cultivation of organisms that are involved in the methane cycle is currently being performed in a project of the research unit ‘BioGeoChemistry of Tidal Flats.’

Acknowledgements

We thank the Senckenberg Institute, Department for Marine Research, and especially the crew of the RV Senckenberg, for their help in recovering the sediment cores. We also thank Katja Ziegelmüller, Melanie Beck and Yvonne Hilker for assistance in sequencing DGGE bands, and analyzing sulfate and methane concentrations. This work is part of the research project ‘BioGeoChemistry of Tidal Flats’ funded by the Deutsche Forschungsgemeinschaft (DFG).

Author notes

Present address: Henrik Sass, School of Earth, Ocean and Planetary Sciences, Cardiff University, Main Building, Park Place, Cardiff, CF10 3YE, Wales, UK

Editor: Gary King