Tables

Abstract. Temporal drying of upper soil layers of acidic methanogenic peatlands might divert the flow of reductants from CH4 formation to other electron-accepting processes due to a renewal of alternative electron acceptors. In this study, we evaluated the in situ relevance of Fe(III)-reducing microbial activities in peatlands of a forested catchment that differed in their hydrology. Intermittent seeps reduced sequentially nitrate, Fe(III), and sulfate during periods of water saturation. Due to the acidic soil conditions, released Fe(II) was transported with the groundwater flow and accumulated as Fe(III) in upper soil layers of a lowland fen apparently due to oxidation. Microbial Fe(III) reduction in the upper soil layer accounted for 26.7 and 71.6% of the anaerobic organic carbon mineralization in the intermittent seep and the lowland fen, respectively. In an upland fen not receiving exogenous Fe, Fe(III) reduction contributed only to 6.7%. Fe(II) and acetate accumulated in deeper porewater of the lowland fen with maximum concentrations of 7 and 3 mM, respectively. Both supplemental glucose and acetate stimulated the reduction of Fe(III) indicating that fermentative, incomplete, and complete oxidizers were involved in Fe(II) formation in the acidic fen. Amplification of DNA yielded PCR products specific for Acidiphilium-, Geobacter-, and Geothrix-, but not for Shewanella- or Anaeroromyxobacter-related sequences. Porewater biogeochemistry observed during a 3-year-period suggests that increased drought periods and subsequent intensive rainfalls due to global climate change will further favor Fe(III) and sulfate as alternative electron acceptors due to the storage and enhanced re-oxidation of their reduced compounds in the soil.


Introduction
Acidic wetlands (pH<5.0)represent a vast type of northern wetlands in Eurasia and North America (Harriss et al., 1993).Waterlogging, low temperatures, and low nutrient quality of plant litter impair decomposition of plant litter, favoring the accumulation of organic carbon.Emission estimates of the greenhouse gas methane (CH 4 ) from wetlands range from 92 to 232 Tg CH 4 year −1 (Wuebbles and Hayhoe, 2002).Although rates of CH 4 production were shown to be correlated with water-table depth, peat chemistry and vegetation type (Verville et al., 1998), pathways of CH 4 production are still not well understood.Generally, two-thirds of the biogenic CH 4 produced in wetlands originates from acetoclastic methanogenesis (Conrad, 1999).However, H 2 -CO 2 appears to be a significant precursor in northern peatlands (Avery et al., 2003;Horn et al., 2003).Acetate even accumulates in some peatlands as a terminal product of anaerobic decomposition indicating that it is not the primary source of CH 4 that is emitted from such habitats (Hines et al., 2001;Duddleston et al., 2002).Acetate consumption appears to occur in these peatlands after diffusion into oxic environments where it is oxidized to carbon dioxide (CO 2 ).
High-latitude regions are expected to experience a temperature increase as a result of global climate change, and climate models predict a decrease in annual precipitation in most European regions during the next decades (International Panel on Climate Change (IPCC), 2007).Thus, transient drying and oxidation of upper soil layers might divert the flow of reductants from CH 4 formation (Blodau et al., 2004) to other electron-accepting processes due to the renewal of alternative electron acceptors.Atmospheric nitrogen and sulfate depositions might further enhance the activity of sulfate and nitrate reducers under changed climatic conditions in northern peatlands similar to other wetlands (Gauci et al., 2002;Lamers et al., 2002;Vile et al., 2003).However, addition of alternative electron acceptors to ombrotrophic bogs and minerotrophic fens incubated in anoxic jars do not universally divert carbon and electron flow from CH 4 formation (Dettling et al., 2006).
δ 34 S values and 35 S-labeling patterns indicate that the dissimilatory reduction of sulfate is an ongoing process in acidic seeps and fens of a forested catchment in northern Bavaria, Germany (Lehstenbach, Fichtelgebirge) (Alewell and Giesemann, 1996;Alewell and Novak, 2001) like in peat bogs (Wieder and Lang, 1988).In minerotrophic fens which are connected to the groundwater flow, ferric iron [Fe(III)] can be another potential electron acceptor.Fe(II) concentration profiles hint to microbial Fe(III) reduction in pH neutral fens adjacent to agricultural fields (Todorova et al., 2005), and Fe(III) reduction appears to parallel CH 4 formation in northern acidic wetlands (Metje and Frenzel, 2005).Fe(III) reduction might compete with methanogenesis for H 2 or acetate (Roden and Wetzel, 2003).Most Fe(III) reducers can use one or more alternative electron acceptors (Lovley et al., 2004), which might be advantageous in upper peat layers that experience redox fluctuations due to water-table variations or oxygen release by plant roots.While Fe(III) exists predominantly in the solid phase as oxyhydroxide minerals at circumneutral pH, Fe(III) is more soluble under acidic conditions (Lovley et al., 2004).However, most cultured Fe(III) reducers are neutrophiles and have only negligible capacities to reduce Fe(III) under moderately acidic (pH 3-6) conditions.Thus, we have only a marginal understanding of the flow of carbon and reductant in acidic, Fe(III) rich habitats and of their inherent Fe(III) reducing microbial communities (Vile and Wieder, 1993;Cummings et al., 2000Cummings et al., , 2003;;Blodau et al., 2002;Petrie et al., 2003;Adams et al., 2007;Blöthe et al., 2008).
The main objectives of this study were (1) to provide experimental evidence on field based data for diverted flow of reductants from CH 4 formation to other electron accepting processes in upper soil layers of peatlands which differ in their hydrology, and (2) to achieve a better understanding of the flow of carbon in acidic habitats and of their inherent Fe(III)-reducing community.

Field site description
The sites are located in the Lehstenbach catchment (Fichtelgebirge, Northeastern Bavaria, Germany), with a highest elevation of 877 m a.s.l. and an area of 4.2 km 2 .Ninety % is covered by Norway spruce (Picea abies [L.] KARST.) of different age classes, and thirty % of the catchment soils are fens and seeps.Upland soils in the catchment have developed from weathered granitic bedrock and are predominantly Dystric Cambisols and Cambic Podzols (WRB system).The annual precipitation in the catchment varies between 900 and 1160 mm yr −1 and the average annual temperature is 5 • C. The site Schlöppnerbrunnen I (50 • 08 14 N, 11 • 53 07 E) is a fen which is located in the upper part of the catchment (upland fen), dominated by Sphagnum mosses alternately covered with patches of Vaccinium myrtillus (L.), Juncus effusus (L.), Carex nigra ((L.)Reichard)), Carex rostrata (Stokes), and Carex canescens (L.) (Table 1).The yearly mean groundwater table depth approximated 0.2 m.Some lower situated soils in the catchment Lehstenbach are only water saturated if the groundwater level increases during autumn and winter.In these seeps, represented by the site Köhlerloh (intermittent seep), the groundwater level reaches the soil surface during autumn, winter, and spring, whereas it is about 0.5-to-1 m below the surface during summer.Open areas at the seep are vegetated with Sphagnum mosses or are partly covered with dense layer of Vaccinium myrtilus.The fen Schlöppnerbrunnen II (50 • 08 38 N, 11 • 51 41 E) is completely overgrown by Mollinia caerulea (L.Moench), Eriophorum vaginatum (L.), Carex canescens (L.), and Juncus effusus (L.).It is located down slope in the catchment and close to the runoff (lowland fen).The mean groundwater table approximated 0.1 m.
Three intermittent seeps located upstream of the lowland fen were also sampled in October 2003.These seeps were covered with spruce, Vaccinium myrtillus, and some Sphagnum mosses.

Porewater collection
Porewater from the upper 40 cm of the fens site was sampled with dialysis chambers every two months during the time period from the end of July 2001 to July 2004 with an interruption during the winter months due to coverage of the catchment with snow and ice as previously reported (Schmalenberger et al., 2007).The intermittent seep was only sampled during periods of water saturation.The hot summer of 2003 lead to a drying of the topsoil of the lowland and the upland fen down to a depth of 5-to-10 cm.

Anoxic soil microcosms
For determining rates of anaerobic microbial activities, soil samples from three replicate sites and different depths (approximately 0-10, 10-20, and 20-30 cm) were obtained in September 2001 in sterile airtight vessels and transported to the laboratory.Replicates were pooled and processed within 4 h.Fourty g (wet wt) soil was placed into 125-ml infusion flasks (Merck ABS, Dietikon, Switzerland) inside an O 2 -free chamber (100% N 2 gas phase).Bottles were closed with rubber stoppers and screw-cap seals, flushed with sterile argon for 15 min, and incubated in the dark at 15 • C with an initial overpressure of 20-25 kPa argon at room temperature.Headspace gases were taken by sterile, argon flushed syringes from these bottles.Gas values were estimated by Henry's law and included the total amounts in both gas and liquid phases.To facilitate sampling of water soluble parameters, 40 ml anoxic, deionized water with a pH of 5.0 were added to another set of soil microcosms with 40 g (wet wt) soil.Adjustments of pH were performed with sterile solutions of 10 N HCl and 10 N NaOH.All microcosms were done in three replicates.At 8 time points, samples were taken during 15 days of incubation in the dark at 15 • C. Activity rates were calculated by linear regression analysis during the time period of linear increase of reduced compounds or linear decrease of electron acceptors.
For determining the effect of supplemental electron donors on Fe(III)-reducing activities, soil samples (0-10 cm depth) were obtained from the lowland fen in March and October 2002.Thirtyfive g (wet wt) soil was mixed with 70 ml anoxic, deionized water with a pH of 5.0.Glucose (2.5 mM), acetate (2 mM), or lactate (2 mM) was added from sterile anoxic stock solutions; H 2 (10 ml) was added as sterile gas.At 10 time points, samples were taken during 15 days of incubation in the dark at 15 • C. Fe(II) formation rates were calculated by linear regression analysis during the time period of linear increase of Fe(II) (glucose, 4 days; acetate, 8 days; lactate, H 2 and control 7 days).

Enrichments of Fe(III) reducers
Five g (wet wt) soil of the lowland fen (0-10 cm) obtained in October 2003 was mixed with 95 ml of dilution buffer (Küsel et al., 2001) and further diluted in a tenfold dilution series, which were used to inoculate an anoxic, undefined medium with a pH of 5.2 that was supplemented with 40 mM amorphous ferric hydroxide [Fe(OH) 3 ] and either acetate (5 mM) or H 2 (10 ml) as previously described (Blöthe et al., 2008).Bromoethanesulfonate (BESA) was added (15 mM) to inhibit methanogenesis.Tubes were incubated in the dark at 15 • C for 3 months.

Analytical methods
The pH was measured with a U457-S7/110 combination pH electrode (Ingold, Steinbach, Germany).Headspace gases (H 2 , CO 2 and CH 4 ) were measured with Hewlett Packard Co. (Palo Alto, CA, USA) 5890 series II gas chromatographs (Küsel and Drake, 1995).The reduction of Fe(III) was determined after acid extraction by the amount of Fe(II) formed in anoxic incubations.Aliquots (0.2 ml) of the media or of the soil suspension were taken by sterile syringes and transferred to 9.8 ml of 0.5 N HCl and incubated for 1 h at room temperature (Küsel et al., 1999).Fe(II) was measured after the addition of acetate by the phenanthroline method (Tamura et al., 1974).Pedogenic iron (Fe d ) was extracted with dithionite-citrate-bicarbonate solution (Mehra and Jackson, 1960).Poorly crystallized iron oxides, hydroxides, and associated gels (Fe o ) were extracted with acidic ammonium oxalate solution (Schwertmann, 1964).Extracted Fe was measured by atomic absorption spectrometry (Unicam 939 spectrometer, ThermoNicolet GmbH, Offenbach, Germany).Short chain aliphatic acids and alcohols were determined with Hewlett-Packard 1090 series II high-performance liquid chromatographs (Küsel and Drake, 1995).The detection limits for short chain fatty acids approximated 5-10 µM and of alcohols 30 µM at an injection volume of 200 µl.Dissolved organic carbon (DOC) was analyzed with a liquiTOC (Foss-Heraeus, Hanau, Germany) with a detection limit of 50 µM.NH + 4 and NO − 3 /NO − 2 were determined by flow injection analysis with a detection limit of 3 µM (QuickChemAE, Lachat Instruments, Milwaukee, WI).SO 2− 4 was measured by ion chromatography with a detection limit of 3 µM (DX-100 with AS4 A column; Dionex, Sunnyvale, CA).Sulfide was measured by the Cline procedure with a detection limit of 5 µM (Cline, 1969) (Blöthe et al., 2008).Anaeromyxobacter-specific PCR was performed according to the method described by Wu et al. (Ab112F, Ab227R;2006).

Clone library construction
PCR amplicons produced with group-specific 16S rRNA gene primers were cloned using the pGEM®-T vector and Escherichia coli JM109 competent cells according to the manufacturer's instructions (Promega, Madison, WI USA).Clone libraries were screened by restriction fragment length analysis (RFLP) as previously described (Blöthe et al., 2008).All clones screened using RFLP were grouped into phylotypes according to RFLP banding patterns.

Phylogenetic and statistical analyses
Representative clones for each RFLP phylotype were sequenced bidirectionally using a Big-Dye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems, Foster City, CA) on an Applied Biosystems 3100 Genetic Analyzer with Capillary Electrophoresis.Sequences were assembled using Sequencher v4.5 (Gene Codes Corp., Ann Arbor, MI) and prior to phylogenetic analysis, vector sequences flanking the 16S rRNA gene inserts were removed.Previously identified sequences with high sequence similarity to the clones obtained in this study were determined using the BLAST algorithm against the GenBank database available from National Center for Biotechnology Information (NCBI) (Altschul et al., 1990).Clone sequences were checked for chimeras and aligned with reference sequences in the ARB software package as previously described (Blöthe et al., 2008).Dendrograms were constructed in the ARB software package by adding 16S rRNA sequences to the distance-matrix tree using PARSIMONY INTERAKTIV without changing the overall tree topology (Ludwig et al., 2004).The coverage of the clone libraries were calculated (Singleton et al., 2001), and the sampling efficiency within clone libraries was assessed using Analytica Rarefaction 1.3 software (http://www.uga.edu/strata/software/) originally derived by Heck et al. (1975).

Nucleotide sequence accession numbers
The 16S rRNA gene sequences presented in this study have been deposited in the EMBL database under the accession numbers AM941453-AM941457 for Acidiphilium-affiliated 16S rRNA gene sequences, AM941458-AM941489 for Geobacter-affiliated 16S rRNA gene sequences, and AM941490-AM941491 for Geothrix-affiliated 16S rRNA gene sequences.

Fe(III)-reducing activities
The intermittent seep and the lowland fen displayed high Fe(II) formation rates during the first 7 days of incubation and reached high amounts of Fe(II) formed compared to the upland fen (Table 2).The lowland fen was depleted in nitrate, and consumption of sulfate started after Fe(II) formation reached the plateau (Fig. 1).However, CH 4 formation paralleled Fe(III) reduction in all soil depths (Fig. 1, and data not shown).These parallel activities were not observed in the intermittent seep and the upland fen.The upland fen displayed higher sulfate-consuming activities than the lowland fen (Table 2).The intermittent seep showed negligible in situ methane forming activities.
The initial rate of Fe(II) formation in the topsoil of the lowland fen was even slightly higher in March 2002 [1840 nmol g (wet wt soil) −1 d −1 ] than in September 2001 (Table 2).Apparently, the topsoil was more oxidized in March, because the initial Fe(II) concentration approximated only 1.7 µmol g (wet wt soil) −1 compared to 7.3 µmol g (wet wt soil) −1 in September.However, both experiments yielded similar maximum concentrations of Fe(II) at the end of incubation (Figs. 1 and 2) which were equivalent to 70% of the Fe d content of the soil.

Source of iron in the lowland fen
To detect the source of the high Fe d concentrations in the lowland fen (Table 1), 3 intermittent seeps were sampled upstream of the lowland fen, and the upper 5 to 15 cm layer of each soil was incubated under anoxic conditions.These soils formed Fe(II) without delay with rates ranging from 90 to 1050 nmol g (wet wt soil) −1 d −1 .Two of these soils had high initial concentrations of Fe(II) indicating that Fe(III) reduction was an ongoing process in these wetland soils.The amounts of pedogenic iron (Fe d ) and oxalate extractable iron (Fe o in these acidic (pH 3.2) soils ranged from 2.3 to 13.9 g kg −1 and from 1.8 to 12.5 g kg −1 , respectively.

Porewater biogeochemistry
Porewater depth profiles at the intermittent seep showed typical biogeochemical gradients of redox sensitive compounds indicating the sequential utilization of nitrate, Fe(III), and sometimes of sulfate with increasing soil depth during water saturated conditions between autumn (November) and early summer (June) (data not shown; Küsel and Alewell, 2004).Fe(II) reached maximum concentrations of 50 µM in 25 cm depth.Declining sulfate gradients were only detected in early summer samplings.Formate (up to 650 µM), acetate (up to 260 µM), and lactate (up to 85 µM), but not propionate or butyrate were detected in the upper 20 cm of the porewater.In general, the upland and the lowland fen showed stronger reduced conditions than the seep.Both fens showed similar qualitative biogeochemical porewater patterns during the three year sampling period.However, the lowland fen showed up to 35-fold higher concentrations of Fe(II), up to 3fold lower concentrations of sulfate, and up to 2-fold higher concentrations of CH 4 in the porewater than the upland fen.The concentrations of Fe(II) in the lowland fen were highly variable (Figs. 3 and 4).During summer of 2001, 2002, and 2004, the depth integrated (0-to-40 cm) average Fe(II) concentrations approximated 2961, 216, and 92 µM.In general, maximum Fe(II) concentrations occurred below 30 cm depth.
In the lowland fen, the concentrations of nitrate (up to 55 µM), sulfate (up to 140 µM), and negligible concentrations of Fe(II) and ammonia indicated soil oxygenation down to a depth of 25 cm after the snowmelt in 2002, 2003, and 2004 (Fig. 3).Oxygenation might have occurred due to mixing of the porewater with lateral flowing, oxygenated surface water.Drying of the topsoil (upper 5-to-10 cm) occurred during the hot summer in 2003 followed by heavy rain falls prior to sampling in September.Up to 420 µM sulfate (Figs.The concentrations of short chain fatty acids detected in the porewater of the lowland fen were much higher than in the upland fen.In general, concentrations increased with increasing soil depth (Fig. 3).Depth integrated average concentrations in the lowland fen approximated 82, 56, 44, 15, and 4 µM during 2001 to 2004 for acetate, formate, lactate, propionate, and butyrate, respectively.Highest concentrations were detected in November 2001 and December 2002, where concentrations of acetate, propionate, butyrate, and lactate reached up to 3160, 1600, 100, and 95 µM in the porewater (Figs. 3 and 4b, and data not shown).In contrast, maximum concentrations of formate occurred in April 2002 (Fig. 3).Ethanol was never detected.Porewater pH did not decrease in the presence of high concentrations of short chain fatty acids.Acetate concentrations were positively correlated with Fe(II) (R 2 =0.75;Fig. 4).

Effect of supplemental electron donors on Fe(III)reducing activities
Since the high concentrations of acetate observed might be derived from fermentation, glucose and typical fermentative products were supplemented to lowland fen soil.The initial rate of Fe(II) formation was enhanced from 1.84 µmol g (wet wt soil) −1 d −1 to 6.08 µmol g (wet wt soil) −1 d −1 by supplemental glucose, but not with supplemental acetate, lactate, and H 2 .However, a secondary Fe(II)-forming increase occurred after 4 days in acetate microcosms (Fig. 2).Glucose and acetate supplemented microcosms yielded higher maximum Fe(II) concentrations of 24.7 and 26.2 µmol g (wet wt soil) −1 compared to 15.4 µmol g (wet wt soil) −1 of the control.
Glucose was rapidly consumed within 3 days (Fig. 2a) and yielded CO 2 and acetate as end products, and H 2 and ethanol as transient products (Fig. S2, http://www.biogeosciences.net/5/1537/2008/bg-5-1537-2008-supplement.pdf).Supplemental lactate yielded acetate as main product.Both acetate and H 2 were not completely consumed during 15 days of incubation.H 2 but not acetate stimulated the formation of CH 4 (Fig. 2c).Microcosms supplemented with either acetate, H 2 or glucose with soil obtained in October 2002 yielded similar results (data not shown).Formate was detected as a small transient product only in glucose amended microcosms of October 2002.

Molecular detection of Fe(III) reducers
PCR with lowland fen soil yielded products with 16S rRNA primers specific for acidophiles belonging to Acidiphilium and for neutrophiles belonging to Geobacter or Geothrix.No PCR products were obtained with a primer set specific for bioleaching-associated bacteria, Shewanella or Anaeromyxobacter.PCR products with Geobacter specific primers were detected in enrichments obtained from a 10 −3 soil dilution transferred to mineral medium at pH 5.5 supplemented with 40 mM amorphous ferric hydroxide [Fe(OH) 3 ] and 5 mM acetate or H 2 .PCR products with Acidiphilium specific primers were detected up to a 10 −3 soil dilution enrichment amended with acetate but not with H 2 , and PCR products with Geothrix specific primers were detected only in 10 −1 soil dilution enrichments.Screening of 16S rRNA gene clones by RFLP revealed that all phylotypes detected in the enrichments also occurred in the fen soil.
A total of 45 Acidiphilium-16S rRNA gene clones were screened by RFLP, and 13 different phylotypes could be differentiated.Comparative sequence analyses indicated that 5 phylotypes were 95% similar to cultured Acidiphilium or Acidosphaera species.Clones were most closely related (96-98% sequence similarity) to a forest soil or sphagnum peat bog clone (Fig. 5).With the primer pair specific for Geothrix, 20 clones were obtained, but only 2 different phylotypes were obtained with a 96-97% sequence similarity to Geothrix-related sequences.With the primer pair specific for Geobacter, a total of 83 clones were obtained and 40 different phylotypes were obtained.A number of non-Geobacteraceae sequences and chimeras between Geobacteraceae and non-Geobacteraceae were detected.Comparative sequence analysis revealed that 34 of the 40 sequences retrieved showed high sequence identity to Geobacteraceae sequences (Fig. 6); one was related to Geobacter chapellii str.172 (96% sequence similarity), three were related to Geobacter bemidjiensis and Geobacter bremensis (94-96% sequence similarity).Many sequences were similar to Pelobacter spp.(94-97%).Acidiphilium and Geobacter specific 16S rRNA gene clone libraries showed coverages of 57 and 35%, respectively.

Mobilization and oxidation of Fe(II) in the catchment
At the upland fen, porewater concentrations of Fe(II) were low, similar to oligotrophic ombrogenic peatlands that receive most of their iron from atmospheric deposition.The low Fe(III) reduction rates corresponded to the low amounts of oxalate extractable Fe(III) oxides present in the upland fen and indicated that Fe(III) reduction was of minor significance for the oxidation of carbon in this fen similar to other northern peatlands (Blodau et al., 2002).Porewater profiles of Fe(II) and sulfate in the intermittent seep were similar to pH www.biogeosciences.net/5/1537/2008/Biogeosciences, 5, 1537-1549, 2008 neutral fens (Todorova et al., 2005;Dettling et al., 2006) or other wetlands (Roden and Wetzel, 2003).The high Fe(II) formation rate at the seep suggested a mobilization of Fe(II) during waterlogging from autumn to spring.Fe(II) was also spontaneously formed in other waterlogged seeps sampled upstream of the lowland fen further strengthening the suggestion that Fe(III) reduction is an ongoing process in many seeps of this catchment.The lowland fen is connected to a shallow groundwater layer and receives water from these intermittent seeps and fens located in the north-east of the Lehstenbach catchment (Küsel and Alewell, 2004).Due to the acidic (pH 3.1) soil conditions of most seeps, the majority of the reductive dissolved Fe(II) will not adsorb to the solid phase and move with the groundwater flow.Indeed, concentrations of Fe(II) in a nearby groundwater well range from 9-to-143 µM (Küsel and Alewell, 2004).Thus, the lowland fen appears to receive continuously anoxic Fe(II)-rich groundwater.The high accumulation of iron in the upper soil of the lowland fen might have resulted from the oxidation of Fe(II) in oxidized peat surface layers.The high Fe o /Fe d ratios (∼0.8) (Table 1) suggest that Fe(III) precipitated as amorphous oxides or as organic matter complexes.Microbial Fe(II) oxidation might yield colloidal or dissolved forms of Fe(III) readily available for microbial reduction (Roden et al., 2004).The high mean DOC concentration (77 mg L −1 ) in the porewater would further favor Fe(III) reduction, because humic compounds can serve as electron shuttles between Fe(III) reducers and surface-bound Fe(III) sterically not accesible to microorganisms (Lovley et al., 2004).

In situ relevance of Fe(III)-reducing activities
According to the 1:4 ratio of CO 2 production to Fe(III) reduction (Roden and Wetzel, 1996), microbial Fe(III) reduction in the most active upper soil layer (0-10 cm) accounted for 26.7, 6.7, and 71.6% of the anaerobic organic carbon mineralization in the intermittent seep, the upland, and the lowland fen, respectively.In rhizosphere and unvegetated wetland sediments Fe(III) reduction accounts for approximately 65 and 40% of carbon mineralization (Roden and Wetzel, 1996).Thus, Fe(III) reduction can substantially contribute to carbon mineralization also in peatlands and reduce CH 4 emissions.The lowland fen showed sequential Fe(III)-reducing and sulfate-reducing activities but concomitant Fe(III)-reducing and methanogenic activities (Fig. 1).Overlapping or concomitant Fe(III)-reducing and methanogenic activities were not observed in soil samples obtained after the snow melts indicating that prolonged anoxic or reduced conditions in the fen were necessary for the establishment of methanogenic activities.Short-term oxygenation of reduced surface soil during summer that leads to a rapid renewal of the Fe(III) pool might yield a partial shift of the electron flow from CO 2 to Fe(III) by methanogens and help to explain concomitant reduction processes in peatlands (Dettling et al., 2006;Metje and Frenzel, 2005;Paul et al., 2006) or in rice paddy soils after drainage (Krüger et al., 2001).The ability of some methanogens to interact with extracellular quinones, humic acids, and Fe(III) oxides has raised the possibility that methanogens contribute to Fe(III) and humic acid reduction (Bond and Lovley, 2002;van Bodegom et al., 2004).Concomitant activities can be also explained by the use of non-competetive substrates.Concomitant or reversed Fe(III) and sulfate reduction is reported from other fens (Todorova et al., 2005).It was suggested that concomitant reduction processes are due to the presence of more crystalline Fe(III) oxides like hematite or goethite in the soil (Postma and Jakobsen, 1996), which are reduced at lower redox potentials (Straub et al., 2001) than sulfate or CO 2 (Zehnder and Stumm, 1988).However, the high Fe o /Fe d ratios in the top soil hint to amorphous, easily reducible Fe(III) oxides.

Accumulation of acetate
High short chain fatty acid and high Fe(II) concentrations were detected in deeper soil of the lowland fen during autumn to winter in 2001/2002 and 2002/2003 with maximum concentration of 3 and 7 mM for acetate and Fe(II), respectively.The high Fe(II) concentrations in this depth were surprising due to the relatively low amounts of iron in the solid phase.However, porewater biogeochemical gradients in fens are not only based on diffusion like in lake sediments.We can not rule out that a part of the Fe(II) was transported with the groundwater flow.Acetate could have been produced by fermentors using plant polysaccarides from dead root material.Accumulation of acetate up to 1.2 mM is reported from bogs in Michigan or Ontario (Shannon and White, 1996;Blodau et al., 2002) and appears to be due to the absence of acetoclastic methanogenesis (Shannon and White, 1996;Hines et al., 2001).Temporal acetate accumulations might characterize fens and bogs as habitats with specific qualities.The decoupling of methanogenesis from carbon flow might be due to the direct inhibition of accumulated acetate on methanogens, because acetic acid can penetrate through the cell membrane and act as decoupler of the proton motive force under acidic conditions (Williams and Crawford, 1984;Goodwin and Zeikus, 1987) or due to transport and thermodynamic constraints (Beer and Blodau, 2007).
The positive correlation of acetate and Fe(II) in the lowland fen porewater during the 3-year-period (Figs. 3 and 4) suggests either a formation of acetate by incomplete oxidizing Fe(III) reducers or an accumulation of acetate after depletion of the Fe(III) pool by complete acetate-oxidizing Fe(III) reducers.Cultured Geobacter species, which were most closely related to clone sequences retrieved from this fen, like G. chapellii, G. bemidjiensis, and G. bremensis are complete oxidizers (Lovley et al., 2004).In contrast, the also closely related Pelobacter venetianus is an incomplete oxidizer.The low or negligible concentrations of acetate in upper soil suggests oxidation of acetate to CO 2 via aerobic respiration, other oxidative microbial processes or syntrophic oxidation of acetate to CO 2 by the concerted ac-tivity of acetate-oxidizing anaerobes and hydrogenotrophic methanogens (Horn et al., 2003;Metje and Frenzel, 2007).

Fe(III)-reducing microbial communities of acidic habitats
Due to our limited knowledge about Fe(III) reduction in moderate acidic habitats, phylogenetic analyses of Fe(III) reducers based on known 16S rRNA gene sequences are severely limited, and we might miss important genera.Phylotypes related to cultured Acidiphilium or Acidisphaera species were detected in the lowland fen similar to slightly acidic coal mining lake sediments (Blöthe et al., 2008).Most Fe(III)-reducing prokaryotes cultured to date are either neutrophilic or acidophilic and have only minor capacities to reduce Fe(III) at in a pH range from 4 to 6.The acidophilic Acidiphilium cryptum (ATCC 33463) can reduce only small amounts of solid phase Fe(III) at pH 5 (Bilgin et al., 2004).Geobacter sp.CdA-3 that was isolated from a mining-impacted sediment can reduce Fe(III) at a pH range from 5.5 to 8.1 (Cummings et al., 2000), and G. bremensis can reduce Fe(III) down to pH 5 (Straub and Bucholz-Cleven, 2001)  Geobacteraceae dominate only in sediment enrichment cultures incubated under neutral pH conditions (Petrie et al., 2003).Surprisingly, no PCR products of Anaeromyxobacter, or Shewanella related species were obtained from the lowland fen, although microorganisms from these genera are common to various metal-reducing environments.Recently it was shown that Acidobacterium capsulatum is capable of Fe(III) reduction in a pH range of 2 to 5 (Blöthe et al., 2008) and that the potential for dissimilatory Fe(III) reduction is widespread among acidophilic heterotrophic bacteria (Coupland and Johnson, 2008).Acidobacteria are present in peatlands (Dedysh et al., 2006) and have been detected also in the upland fen of this catchment (Schmalenberger et al., 2008).Thus, results from phylogenetic analyses in this investigation provide an incomplete picture.

Conclusions
Our field based experimental results corroborate the hypotheses that a decrease in annual summer precipitation in most northern European regions as a result of global climate change (IPCC 2007) will shift the electron flow away from methanogenesis to alternative anaerobic processes in peatlands.Drying of the top soil down to 10 cm depth during the hot summer 2003 followed by heavy rain falls in September lead to the appearance of sulfate in the porewater and a renewal of Fe(III) which favored the subsequent Fe(III)-and sulfate-reducing activities.Drying and rewetting of the top soil 2003 yielded even higher porewater sulfate concentrations in the lowland fen than in the upland fen despite the average 3-fold lower sulfate concentrations in the lowland fen (Fig. S1).The higher release of sulfate might be due to differences in the reduced sulfur pools, which were subjected to oxidation.The Fe-rich lowland fen contains higher contents of acid volatile sulfur (AVS, i.e. amorphous FeS) and total reduced inorganic sulfur (TRIS, i.e. amorphous FeS, S 0 , and FeS 2 ) (Loy et al., 2004) but lower contents of organic sulfur compared to the upland fen (Paul et al., 2006).Synchrotronbased X-ray spectromicroscopy revealed that the fraction of organic reduced sulfur of these fens is more stable under alternating reduction-oxidation processes than the FeS or FeS 2 pool (Prietzel, personal communication), similar to results obtained with chemical S fractionations in peat bogs (Wieder and Lang, 1988;Wieder et al., 1990).Thus, the presence of Fe(II) seemed to affect both the storage and mobilization of sulphur.An increase of extreme weather conditions like summer droughts and heavy rainfall events during the next decades will amplify the importance of iron for biogeochemical processes in iron-rich peatlands.

Fig. 1 .
Fig. 1.Formation of Fe(II) and CH 4 and consumption of sulfate in anoxic microcosms of soil obtained from the lowland fen (0-10 cm depth) in September 2001.Presented are the averages ± standard deviations of triplicates.

Fig. 2 .
Fig. 2. Effect of the consumption of supplemental electron donors (a) on the formation of Fe(II) (b) and formation of CH 4 (c) in anoxic microcosms of soil obtained from the lowland fen (0-10 cm) in March, 2002.Presented are the averages ± standard deviations of triplicates.

Fig. 3 .
Fig. 3. Porewater concentrations of nitrate, Fe(II), sulfate, pH, DOC, acetate, formate, and buturate in the lowland fen sampled during end of July 2001 to July 2004.Please note that the upper concentrations were combined in a range for better visualization of low concentrations.White triangles below the x-axis flag the snow melt events, the black triangle flags the summer drought followed by a heavy rain fall in September 2003.

Fig. 4 .
Fig. 4. Detailed porewater depth profiles of Fe(II) (a) and acetate (b) in the lowland fen sampled in July, September, and November 2001, and in January 2002.

Fig. 5 .
Fig. 5. Phylogenetic tree showing the relative positions of Acidiphilium-affiliated 16S rRNA gene sequences derived from the lowland fen soil (0-10 cm) obtained in October 2003.Sequences were added to the existing tree without changing the overall tree topology by using the ARB treeing tool PARSIMONY INTERAKTIV.Names and accession numbers (between bracktes) for closest relatives 16S rRNA gene sequences are given.The bar indicates 10% sequence divergence.

Fig. 6 .
Fig. 6.Phylogenetic tree showing the relative positions of the Geobacter-affiliated 16S rRNA gene sequences derived from the lowland fen soil (0-10 cm) obtained in October 2003 as inferred by Parsimony method.Bootstrap values for a total of 100 replicates are shown at the nodes.Names and accession numbers (between bracktes) for all 16S rRNA gene sequences used for comparsion are given.The bar indicates 10% sequence divergence.

Table 1 .
Characteristics of the seep and both fens from a forested catchment a .
a Presented are the averages from duplicate soil samples.b Fe d refers to pedogenic Fe-oxides.c Fe o refers to poorly crystallized Fe-oxides, hydroxides, and associated gels.

Table 2 .
Anaerobic activities of seep and fen soils in anoxic microcosms of September 2001.
a Presented is the average rate observed in triplicate microcosms.b Final Fe(II) concentrations after Fe(III)-reduction was completed.c n.a.not applicable.No CH 4 was formed or no NO −