Biogeochemical factors affecting mercury methylation rate in two contaminated floodplain soils

An automated biogeochemical microcosm system allowing controlled variation of redox potential (E H) in soil suspensions was used to assess the effect of various factors on the mobility of mercury (Hg) as well as on the methylation of Hg in two contaminated floodplain soils with different Hg concentrations (approximately 5 mg Hg kg −1 and>30 mg Hg kg−1). The experiment was conducted under stepwise variation from reducing (approximately −350 mV at pH 5) to oxidizing conditions (approximately 600 mV at pH 5). Results of phospholipid fatty acids (PLFA) analysis indicate the occurrence of sulfate reducing bacteria (SRB) such asDesulfobacter species (10Me16:0, cy17:0, 10Me18:0, cy19:0) orDesulfovibrio species (18:2 ω6,9), which are considered to promote Hg methylation. The products of the methylation process are lipophilic, highly toxic methyl mercury species such as the monomethyl mercury ion [MeHg], which is named as MeHg here. The ln(MeHg/Hgt) ratio is assumed to reflect the net production of monomethyl mercury normalized to total dissolved Hg (Hgt) concentration. This ratio increases with rising dissolved organic carbon (DOC) to Hg t ratio (ln(DOC/Hgt) ratio) (R2 = 0.39,p < 0.0001,n = 63) whereas the relation between ln(MeHg/Hgt) ratio and lnDOC is weaker ( R2 = 0.09; p < 0.05; n = 63). In conclusion, the DOC/Hg t ratio might be a more important factor for the Hg net methylation than DOC alone in the current study. Redox variations seem to affect the biogeochemical behavior of dissolved inorganic Hg species and MeHg indirectly through related changes in DOC, sulfur cycle, and microbial community structure whereas EH and pH values, as well as concentration of dissolved Fe3+/Fe2+ and Cl − seem to play subordinate roles in Hg mobilization and methylation under our experimental conditions.


Introduction
Mercury (Hg) is one of the most hazardous heavy metals, posing a risk to humans and environment (e.g.Wolfe et al., 1998;Gibicar et al., 2006;Bergeron et al., 2011).It is distributed widespread all over the world and can be found in various environmental compartments such as floodplains (e.g.Devai et al., 2005;Overesch et al., 2007;Rinklebe et al., 2009).Many floodplain soils have accumulated large amounts of Hg as a result of atmospheric deposition or through transport from the watershed (e.g.Boening, 2000;During et al., 2009;Rinklebe et al., 2010).Large floodplain areas along the Wupper River (Germany) are heavily polluted with Hg due to the discharge of waste originating from textile industry, particularly from dye factories, and metal industry during the last centuries.
The mobility, bioavailability, ecological and toxicological effects of Hg are strongly dependent on its chemical speciation (Ullrich et al., 2001).Methylation of inorganic Hg is an important process, which can fundamentally change its mobility, bioavailability, and toxicity (Boening, 2000).The

T. Frohne et al.: Biogeochemical factors affecting mercury methylation rate
products of this methylation process are lipophilic, highly toxic methyl mercury species such as dimethyl mercury [Me 2 Hg] or the monomethyl mercury ion [MeHg + ], which is named as MeHg in the following.Both methyl mercury species exhibit a significant risk to humans and wildlife due to its neurotoxicity and tendency to accumulate in the food chain (Wolfe et al., 1998;Boening, 2000;King et al., 2002;Li et al., 2010).
Generally, the mobility and methylation of Hg in frequently flooded soils is determined by a range of factors, such as redox potential (E H ), pH, dissolved organic carbon (DOC), sulphur (S), chloride (Cl − ), iron (Fe), and total dissolved Hg (Hg t ) content (e.g.Skyllberg et al., 2003;De-Laune et al., 2004;Sunderland et al., 2006).Dissolved organic carbon interacts strongly with Hg by the formation of Hg-DOC complexes (e.g.Ravichandran, 2004;Khwaja et al., 2006;Feyte et al., 2010).The high affinity of Hg to DOC can partly be attributed to the binding of Hg with reactive sulfur groups in the hydrophobic acid fraction of DOC (Karlsson and Skyllberg, 2003;Shanley et al., 2008).
Mercury immobilization can be induced under anoxic conditions due to the formation of hardly soluble Hg sulfides (e.g.Skyllberg et al., 2003;Du Laing et al., 2009).Sulfides (S 2− ) are generated through sulfate (SO 2− 4 ) reduction, which is mainly catalyzed by microorganisms.Sulfate reducing bacteria (SRB) (e.g.Desulfovibrio desulfuricans, Desulfobulbus proprionicus) mediate the formation of S 2− as a result of respiration processes that require SO 2− 4 as a terminal electron acceptor (King et al., 2002).During Hg methylation, microorganisms increase their resistance to Hg by rendering the Hg 2+ ion ineffective in disturbing the normal biochemical processes within the cell (Boening, 2000).This methylation process has been found to be mainly conducted by SRB and Fe reducing bacteria (Compeau and Bartha, 1985;Macalady et al., 2000;Fleming et al., 2006;Merritt and Amirbahman, 2009).Furthermore, it might be possible that other organisms such as aerobic bacteria, fungi, and seaweed may play a role in Hg methylation as suggested for tropical environments and mangrove wetlands (Coelho-Souza et al., 2006;Wu et al., 2011).The microbial community composition in soils and sediments can be characterized by the analysis of phospholipid fatty acids (PLFA) (e.g.Macalady et al., 2000;Rinklebe and Langer, 2006;Langer and Rinklebe, 2009).This method allows identifying the presence of SRB (Taylor and Parkes, 1985;Coleman et al., 1993;Macalady et al., 2000).
Iron can influence the dynamics of Hg in soils.For instance, Fe(hydr)oxides are able to adsorb Hg; thus, they can act as important Hg sinks (Fernandez-Martinez et al., 2006;Harris-Hellal et al., 2011).Moreover, the mobility and methylation of Hg can be influenced by Cl − , for example through the formation of Hg chloride or MeHg-Cl complexes (Davis et al., 1997;Skyllberg et al., 2003).
Although the presence of Hg and MeHg in the environment has been frequently documented (e.g.van Faassen, 1975;Boening, 2000;Agusa et al., 2005;Devai et al., 2005;Gibicar et al., 2006), mechanistic experiments aimed to study the redox-induced mobilization and immobilization of Hg and MeHg as well as information on biogeochemical factors affecting the methylation rate of Hg in floodplain soils are very scarce up to date.
Thus, our aim was to assess the impact of E H , pH, DOC, SO 2− 4 , Fe, and Cl − on the mobility and methylation of Hg in two floodplain soils with different Hg contamination levels (approximately 5 and >30 mg Hg kg −1 , respectively) under acidic to neutral pH conditions.Therefore, we used an automatic biogeochemical microcosm system allowing establishing definite, computer-controlled redox conditions in soil suspensions.

Study site
Soil samples were collected from two floodplain soils (Wupper 1 = W1; Wupper 2 = W2) at the lower course of the Wupper River (Germany) close to the confluence into the Rhine River (Fig. 1).The study sites are located about 15-20 km to the north of Cologne, Germany, near the town Leverkusen, (W1: 51 • 4 0.48 N, 6 • 59 0.77 E; W2: 51 • 5 4.1 N, 7 • 0 12.61 E).The distance between the two study sites is about 2 km.The long term average annual precipitation is approximately 800 mm and the long term average annual air temperature is 10.8 • C (DWD, 2009).The geological parent material consists of sediments of the Rhine River ("Niederrheinische Bucht"), which is predominantly shale from Devonian origin ("Rheinisches Schiefergebirge").The study sites are used as grassland and are flooded seasonally by the Wupper River, usually in springtime (Wupperverband, 2009).The Wupper River is approximately 115 km in longitude with an average gradient of 0.4 %.The discharge averages 15.4 m 3 s −1 .The catchment area of the Wupper River comprises 814 km 2 .Both soils are classified as Eutric Fluvisols (IUSS-ISRIC-FAO, 2006).

Sampling, pre-treatment, and analysis of bulk soil
Soil samples were collected from the genetic A-horizons (0-10 cm for W1; 0-32 cm for W2).Soil sampling was performed in four replicates of about 1 kg each which were merged to one sample.For chemical analysis, soil material was homogenized, air-dried, and sieved to <2 mm.Subsamples were ground in an agate disc mill.Physico-chemical soil properties were determined according to standard methods (Schlichting et al., 1995).Total C (C t ) and total N (N t ) were determined via dry combustion and thermal conductivity detection using a C/N/S-Analyzer (Vario EL Heraeus, Analytik Jena, Germany).A C-MAT 550 (Stroehlein, Germany) was used to measure inorganic C by dry combustion and IRdetection.Soil organic C was calculated as the difference between C t and inorganic C. Particle-size distribution was measured using the pipette sampling technique by wet sieving and sedimentation (Blume et al., 2000).Total metal concentrations of the soil were quantified after digestion using aqua regia (37 % HCl + 65 % HNO 3 , volume ratio 3:1) ignoring the immobile silica-bound fraction.Total Hg was analyzed by a cold vapor atomic absorption spectrometer (FIMS 400, Perkin Elmer, USA).The calibration range was 0.5-20 µg l −1 .The dissolved samples were appropriate diluted for this calibration range using 0.01 M nitric acid.Analytical accuracy was achieved by the use of certified reference material (IAEA 405, IAEA 433, NIST 2709, and LGC-6139).Inductively coupled plasma atomic emission detection (Ciros CCD, Spectro, Germany) was used for determination of Fe.For the determination of total S, the soil was compressed to pellets and S was measured by energy dispersive X-ray spectroscopy (XLAB 2000, Spectro, Germany).The calibration (using wax pellets) was done in the concentration range 0.0150-15 g kg −1 .The following 15 reference materials had been used: GBW 7309-7312, GBW 7409-7411, LKSD 1, LKSD 4, NIST 2704, andNIST 2710.For the soil microbial analysis (PLFA), fresh soil samples were sieved to <2 mm and thereafter frozen at −20 • C.After storage, samples were allowed to thaw at 4 • C for one day and 4 h at 20 • C before analysis.Phospholipid extraction and PLFA analysis were performed following the standard procedure described by White et al. (1979) and Frostegård et al. (1991).Lipids were extracted with a modified singlephase mixture chloroform-methanol-citrate buffer (1:2:0.8v/v/v) (Bligh and Dyer, 1959).The resulting lipid material was fractionated into neutral lipids, glycolipids, and polar lipids by a silica-bonded phase column.The polar lipids were transesterified to the fatty acid methyl esters by a mild alkaline methanolysis (Guckert et al., 1985).Samples were analyzed by gas chromatography/mass spectroscopy using a Hewlett-Packard 6890 series gas chromatograph with a HP-5MS column (60.0 m length, 0.25 mm internal diameter, coated with a cross-linked 5 % phenyl methyl rubber phase with a film thickness of 0.25 µm) interfaced to an Agilent 5973 mass selective detector.The resulting chromatograms were evaluated by mass spectra, retention times, and nonadecanoic acid methyl ester (19:0) as the internal standard (N-5377, Sigma Chemical, Inc.).The analytical quality was confirmed by the repeated analysis of a standard bacterial acid methyl ester mix and a 37-component FAME mix (47080-U and 47885-U, Supelco, USA).PLFA were designated using the nomenclature described by Feng et al. (2003).More details regarding the method of PLFA analyses can be found in Rinklebe and Langer (2006) and Langer and Rinklebe (2009).

Biogeochemical microcosm system
Flooding events were simulated using an automated biogeochemical microcosm system in the laboratory (Fig. 2).This system allows establishing pre-defined redox conditions in soil suspensions by flushing them with nitrogen (N 2 ) or oxygen (O 2 ).Thus, it is possible to study the effect of E H almost independent from other parameters.Recently this system was described in detail by Yu and Rinklebe (2011) and successfully used in previous studies for the investigation of trace gases (Yu et al., 2007), for the quantification of mercury emissions (Rinklebe et al., 2010), and for the determination of the dynamics of trace metals (Rupp et al., 2010;Frohne et al., 2011).The current study was conducted in four independent trials for each soil.The microcosms (MCs) were filled with 200 g air-dried soil and 1600 ml deionized water.Homogenous conditions were reached by stirring the slurry continuously.Redox potential, pH, and temperature in each MC were monitored every ten minutes by electrodes (Meinsberger Elektroden, Germany) and stored by a data logger.The pH values of MC 8 and the E H values of MC 1 after approximately 800 h incubation could not be monitored due to an error of the electrodes.The measured redox potential values were normalized to pH 5, because the mean pH during the experiment was around 5 for both soils.Thus, the corrected values will be referred to as "E HatpH5 " in the following.
Straw and glucose were added to each MC to provide an additional source of organic matter for microorganisms.As a result, levels of E HatpH5 decreased (Fig. 4).This process was accelerated by continuously flushing the MCs with N 2 for several days.When lowest E H values were reached, the first sample was taken from each MC.Thereafter, E Hvalues were increased in steps of approximately 100 mV by adding O 2 .Thereby, E H was kept within the set E H -windows ±10-20 mV around the aimed value by supplying O 2 or N 2 automatically when the outer limits of the E H -windows were exceeded.Redox potential was maintained for approximately 24 h within each window and afterwards set to the next window.Sampling was conducted approximately 24 h after reaching each new E H -window (Fig. 4).The soil/water ratio remained the same during the experiment.After achieving the highest E H levels, N 2 was added to lower E H again.

Sample preparation, sub-sampling, and storage during the redox experiment
The slurry in the MCs was sampled using a plastic syringe connected with a PTFE tube.The slurry samples were immediately centrifuged for 15 min at 3000 rpm.Afterwards, the supernatants were filtered through a 0.45 µm Millipore membrane (Whatman Inc., Maidstone, UK) under N 2 -atmosphere.Thereafter, the filtrate (defined as the soluble fraction) was aliquoted to subsamples for subsequent analysis.For measuring Hg t , the first 10 ml subsample was preserved with 200 µl 0.2 M bromine monochloride solution (BrCl) and stored at 8 • C in bottles of acid rinsed borosilicate glass with PTFE-lined caps.A second subsample (8 ml) was stored in acid rinsed glass bottles at 8 • C for the analysis of MeHg.Another 10 ml subsample was stabilized by addition of 400 µl 65 % HNO 3 for analysis of total Fe and total S. Another subsample (10 ml) was stored at −20 • C and gradually thawed for the determination of DOC, Cl − , and SO 2− 4 .

Chemical analyses of the redox experiment samples
Total Hg was measured with cold vapor atomic fluorescence spectrometry (CV-AFS) (mercur duo plus, Analytik Jena, Germany).Mercury standard solutions were prepared by diluting mercury standard solution 1000 mg l −1 Hg (CertiPur, Merck) with deionized water.A 7-point calibration curve The soil microcosm system: ( including a blank sample was used for sample analysis.An intern reference sample was analyzed after every 10 samples to check the instrument drift.The drift was satisfying for all measurements.The detection limit was 10 ng l −1 .The relative standard deviation of repeated measurements was below 3 % for all samples.The analyses of MeHg in the subsample were conducted by gas chromatography with atomic emission detection (GC-AED).An amount of 2 ml of the sample was spiked with 4 ml buffer solution (pH 4.5) and 20 µl Na-propylborat solution (2 % in THF).The solution was stirred for 10 min.The Hg species were enriched from the aqueous phase by solid phase microextraction (SPME) in the headspace mode.Analytes were enriched onto a 100 µm polydimethylsiloxane (PDMS) -fibre for 30 min at a temperature of 30 • C. The prepared samples were stored at 15 • C until measurement.Samples were processed automatically by a multipurpose sampler (MPS2, Gerstel, Mülheim, Germany) combined with a gas chromatograph Hewlett-Packard 6890 (Agilent, Waldbronn, Germany) and a microwave-induced plasma atomic emission detector jas 2350 (jas GmbH, Moers, Germany).Thermal desorption was carried out directly in the injector of the gas chromatograph for 1 min at 200 • C. The analyses were carried out using an HP1 column (25 m × 0.32 mm × 0.17 µm) and He as carrier gas.Injection was performed in splitless mode and the oven was programmed from 40 • C (2 min) to 280 • C at 25 • C min −1 .Reagent gases for the AED were O 2 and H 2 , the make-up gas flow (He) was 130 ml min −1 .The Hg emission line 254 nm was monitored.Only monomethyl mercury (MeHg + ) -and no dimetyhl mercury (Me 2 Hg)could be detected.The detection limit for MeHg + was 0.8 ng Hg l −1 .The calibration range was 1-100 ng Hg l −1 .Quality control was carried out using following reference materials: IAEA 405, IAEA 433, and CRM 462.Total Fe and S were quantified by inductively coupled plasma optical emission spectrometry (ICP-OES) (Ultima 2, Horiba Scientific, Unterhaching, Germany).A four-point calibration was conducted by diluting single standard and multi element solutions (CertiPur, Merck) with deionized water.Analysis was conducted in three replications.The relative standard deviation of replicate analysis was below 5 %.
Dissolved organic carbon was measured after 2-point calibration with a TOC-analyzer (TOC-V E , Shimadzu, Kyoto, Japan).Measurement was performed in two replications for each sample.The detection limit was 1 mg l −1 .Sulfate and Cl − were determined using an ion chromatograph (Personal IC 790, Metrohm, Filderstadt, Germany) with a Metrosep A Supp 4 -column.The detection limit was 0.03 mg l −1 for both ions.

Calculations and statistical analysis
Mean values of E H and pH data measured every 10 min originating from 3, 6, 12, and 24 h periods prior to sampling were calculated.Values below the detection limit were excluded for the statistical analyses.Thereafter correlation and regression analyses between Hg t and MeHg on the one hand and E H , pH, DOC, SO 2− 4 , Cl − , Fe, and S on the other hand were conducted.Relations between MeHg/Hg t and DOC/Hg t on one hand and DOC and Hg t on the other hand were also calculated.Mean E H and pH values of the 6 h period prior to sampling were used because they revealed the highest regression coefficients in most cases.Correlation analysis was conducted by SPSS 19.ORIGIN 6.0 was used to calculate regressions and descriptive statistics.For regression analyses, the naturally logarithmised values (ln) of Hg t , MeHg, DOC, Cl − , SO 2− 4 , and Fe were taken, because the range of the values differed for several orders of magnitude.According to Fowler et al. (2006), the strength of the correlations was categorized in our study as follows: r < 0.20 (corresponds to the coefficient of determination R 2 < 0.04) represent very weak correlations; r between 0.20 and 0.39 (R 2 between 0.04 and 0.15) weak correlations; r between 0.4 and 0.69 (R 2 between 0.16 and 0.48) modest correlations; and r > 0.69 (R 2 > 0.48) strong correlations.

Properties of the bulk soils
Selected properties of the studied bulk soils are provided in Table 1.The soils of W1 and W2 mainly consist of sand and silt.The content of organic carbon is relatively high and the contents of inorganic carbon are 0.01 % (W1) and 0.005 % (W2).The pH is slightly acidic to neutral.Soil W1 is contaminated with approximately 5 mg kg −1 Hg, whereas soil W2 is higher contaminated and contains >30 mg Hg kg −1 (Table 1).For both soils, concentrations of Hg exceed the action value of 2 mg Hg kg −1 set by the German Federal Soil Protection Ordinance (BBodSchV, 1999).

Redox experiment
The variations (mean, median, and range) of the measured parameters during the E H experiment are provided in Table 2.The range was 0.09-8.27µg l −1 for Hg t and 1.3-101 ng 1 −1 .for MeHg.The mean values were 0.99 µg l −1 for Hg t and 14 ng l −1 for MeHg.The E H ranges from −335-601 mV (all data).The pH ranges from 4.1-7.2(all data) with mean values around 5 for both soils.During the experiment, E H and pH reveal a very weak significant negative correlation (R 2 = 0.02; p < 0.01; n = 47,941).The development of E HatpH5 measured in the slurry every 10 min during the experiment and concentrations of Hg t in the soluble fraction at the sampling time are given in Fig. 4. The lowest E HatpH5 levels were around −150 and 0 mV in MCs 1-4 (W1) and between −100 and −350 mV in MCs 5-8 (W2).The highest E HatpH5 levels were around 500 mV for W1 and 500-600 mV for W2.The development of pH measured in the slurry every 10 min and the values of MeHg in the soluble fraction at the sampling points during the experiment are given in Fig. 5.The mean initial pH was 7.0 ± 0.2 for W1 (MCs 1-4) and 5.7 ± 0.3 for W2 (MCs 5-8).The pH dropped rapidly in all MCs to values between 4 and 5. Afterwards, the pH slightly increased in all MCs when increasing the E H stepwise.A relationship between Hg t and E H, or between pH and MeHg in the course of the experiment is not obvious (Figs. 4 and 5).
The (Fig. 6).There was a weaker negative relationship between ln(MeHg/Hg t ) and lnHg t (R 2 = 0.18; p < 0.001; n = 63) and a weak positive relation between ln(MeHg/Hg t ) and lnDOC (R 2 = 0.09; p < 0.05; n = 63).LnHg t values were correlated with lnMeHg values and regression analysis showed a modest positive relationship (R 2 = 0.16; p < 0.005; n = 63).Results of the regression analysis between lnDOC, lnCl − , lnSO 2− 4 , E H, lnFe, pH and lnHg t on one hand and lnMeHg in the soluble fraction on the other hand are provided in Table 3.With the increase of Hg t and MeHg, Cl − , SO 2− 4 , and  3).Values of E H (Table 3) and S (data not given) are not related to Hg t or MeHg.
Values of Cl − and SO 2− 4 in the soluble fraction were weakly associated with E HatpH5 (linear relationship with R 2 = 0.15; p < 0.01; n = 67 for Cl − and curved relationship R 2 = 0.13; p < 0.05; n = 67 for SO 2− 4 respectively).Iron contents in the soluble fraction revealed a modest negative relationship to E HatpH5 (R 2 = 0.33; p < 0.001; n = 67).Contents of DOC did not have a significant relationship with E HatpH5 (data not shown).

Direct impact of E H and pH on the mobility and methylation of Hg
The biogeochemical behavior and the dynamics of Hg and MeHg under changing redox conditions are affected by various factors.Our original hypothesis was that systematic changes of E H from anaerobic to aerobic conditions should have a considerable impact on the methylation of Hg since it has been reported that MeHg increases with decreasing E H (DeLaune et al., 2004;Sunderland et al., 2006).On the other hand, Ullrich et al. ( 2001) stated that anaerobic conditions might favor the reduction from Hg 2+ to hardly soluble Hg 0 , which in turn may reduce Hg mobility and Hg methylation because of reduced bioavailability.However, in the current study, a direct impact of E H on Hg t or MeHg concentrations could not be detected (Figs. 4 and 5; Table 3).The variations of Hg t values during the experiment seem to be almost independent from E H variations (Fig. 4).Wallschläger et al. (1998) found consistent with our results, that the mobility of Hg is less influenced by changing redox conditions.Hintelmann and Wilken (1995) also reported that absolute E H might not be the most important factor regulating Hg methylation activity in anoxic sediments.The results of the current study appear to confirm those assumptions.Redox potential has also an effect on the pH.Generally it is well established that pH increases during reduction because reduction processes consume protons (e.g.Yu et al., 2007).Accordingly, a similar behavior was generally observed in our study.Literature data on effects of pH on the mobility and methylation of Hg are contradictory.Some authors found enhanced mobility and methylation of Hg at low pH (Boening, 2000;Ullrich et al., 2001;Wu et al., 2011).This was attributed to the fact that DOC is more positively charged at low pH and therefore has weaker tendencies to form complexes with Hg, enhancing its availability for methylating bacteria (Ravichandran, 2004).This process could have occurred in the current study as well and might contribute to explain the modest negative relationship between MeHg and pH (Table 3).In contrast, low pH can decrease Hg methylation in anoxic sediments, maybe due to the suppression of bacterial activity at low pH (Gilmour and Henry, 1991).However, the results presented here show a less clear effect of pH on Hg t and on MeHg.Although the relationship between MeHg and pH is modest (Table 3), a mutual development between MeHg and pH is not obvious (Fig. 5) suggesting that additional factors are needed to explain MeHg variations.Accordingly, Wallschläger et al. (1996) have shown that the influence of pH on the solubility of Hg is relatively low compared to other metals (e.g.Cd, Ni, Co, Zn, Cu, Pb).The direct impact of E H on the behavior of Hg and MeHg seems to be very weak in our study.Instead, indirect effects of E H and pH on the mobility and methylation of Hg through E H or pH related changes of other determining factors such as concentrations of DOC, Fe, Cl − , and SO 2− 4 should be more important under our experimental conditions.

Impacts of DOC, Fe, Cl − , and SO 2− 4 on the mobility and methylation of Hg
Total mercury and MeHg concentrations were positively related to DOC concentrations in the current experiment whereas this relationship is stronger between MeHg and DOC (Table 3).Similar results have been obtained by other authors (Covelli et al., 2009;Obrist et al., 2009;Feyte et al., 2010) since Hg and MeHg tend to form complexes with organic carbon.The interaction between DOC and Hg t respectively MeHg can partly be attributed to the binding of Hg with reactive S groups in the dissolved organic molecules, especially in the hydrophobic acid fraction of DOC (Karlsson       and Skyllberg, 2003;Ravichandran, 2004;Khwaja et al., 2006;Shanley et al., 2008).
In addition to the impact of DOC due to a complexation of Hg t and MeHg, a positive relationship between ln(MeHg/Hg t ) and ln(DOC/Hg t ) was found in our study (Fig. 6).The MeHg/Hg t ratio is assumed to reflect the net production of MeHg normalized to the Hg concentration or the methylation efficiency respectively (Shanley et al., 2005;Skyllberg et al., 2007).Low MeHg/Hg t ratios can be due to low Hg methylation or to high MeHg demethylation rates (Remy et al., 2006).Thus, increasing DOC/Hg t ratio might have favored Hg net methylation or decreased demethylation in the current study.One reason for rising DOC/Hg t ratio might be increasing DOC concentrations.In this case, increasing DOC could have promoted Hg net methylation or depressed demethylation in our study.Other studies have shown that high DOC contents can promote Hg methylation by enhanced SRB activity (see below), since DOC can serve as an important carbon source for bacteria (Davis et al., 1997;Ullrich et al., 2001;Lambertsson and Nilsson, 2006).Furthermore, DOC can contribute to abiotic methylation of Hg by donating methyl groups (Weber, 1993).However, abiotic methylation seems to be a process of minor importance compared to biological methylation (Avramescu et al., 2011).The positive relationship between ln(MeHg/Hg t ) and ln(DOC/Hg t ) seems to be important in our experiment.In contrast, the relationship between ln(MeHg/Hg t ) and lnDOC is weak, indicating that DOC alone might be a weaker factor in determining the Hg net methylation as also found by Skyllberg et al. (2003).Instead, the DOC/Hg t ratio seems to play a more important role for Hg net methylation.
We also observed a negative relationship between ln(MeHg/Hg t ) and lnHg t .A positive relationship was found between lnHg t and lnMeHg as also reported by Sunderland et al. (2006) and Ouddane et al. (2008).Thus, increasing Hg t concentrations seem to have an inhibitory effect on Hg net methylation but may lead to higher total MeHg concentrations.High Hg concentrations can generally affect soil microorganisms in different ways.First, it is reported that high Hg t contents can exhibit toxic effects on methylating bacteria resulting in a depression of MeHg production (Ullrich et al., 2001).Secondly, microorganisms in Hg contaminated soils can be well adapted to Hg stress.This adaption can favor the selection of Hg tolerant bacteria in soils (Oliveira et al., 2010;Ruggiero et al., 2011).As a consequence, demethylation can be stimulated at high inorganic Hg concentrations by Hg tolerant bacteria which lead to reductive demethylation of MeHg (Schäfer et al., 2004).Bacterial Hg resistance is encoded by the mercury resistance (mer) operon encoding proteins that act amongst other factors in mercury detoxification.Most Hg resistant isolates contain merB (organomercury lyase), merA (mercuric reductase), merP, and merR genes (Lapanje et al., 2010;Ruggiero et al., 2011).Both toxic effects of Hg t on methylating bacteria and the occurrence of Hg tolerant bacteria might have occurred in the current study in parallel.However, the correlations between Hg t , MeHg, and MeHg/Hg t are relatively low indicating that Hg t appears to have limited utility as a predictor of Hg net methylation and MeHg concentrations.This is in good agreement with Ouddane et al. (2008) who indicates that the production of MeHg is dependent on other parameters such as SRB in addition to total Hg concentrations in sediments with high Hg methylation activity.
Concentrations of MeHg and Hg t in the soluble fraction can also be influenced by the redox cycling of S, which is abundant in both soils (Table 1).The relationships between Hg t and SO 2− 4 , and between MeHg and SO 2− 4 are moderate (Table 3) what might indicate that both Hg t and MeHg may be linked to the sulfur cycle.Brümmer (1974) mentioned that sulfides are generated from sulfates below E H −50 mV at pH 7 (corresponds to E H 68 mV at pH 5).As the E HatpH5 values fell below 68 mV in all MCs during the incubation (Fig. 4), the formation of sulfides is most likely in our experiment.The soil slurries turned black and developed a typical odor with decreasing E H , which also points towards the formation of sulfides.We did observe a correlation between E HatpH5 and SO 2− 4 .However, this relationship is weak and not linear.Relationships between S t and E HatpH5 , Hg t , or MeHg could not be found.One reason for that can be the rapid internal cycling of S, which makes SO 2− 4 concentrations a poor indicator for SO 2− 4 reduction rates (Koretsky et al., 2007;Yu et al., 2007).When the concentration of reduced inorganic S reaches a certain value the solubility and speciation of Hg 2+ may be controlled by the precipitation of insoluble HgS or the formation of charged polysulfide Hg-complexes as previously reported by several authors (e.g.Davis, 1997;Benoit et al., 2001;Du Laing et al., 2009).This may result in decreasing Hg concentrations in the dissolved phase when sulfates are removed from the dissolved phase upon reduction to sulfides.On the other hand, this may also result in the solubilisation of mercury upon oxidation of sulfides to sulfates under oxic conditions.Both processes may explain the positive correlations we observed between dissolved SO 2− 4 and Hg t and MeHg concentrations.The formation of HgS at low E H can also decrease the availability of Hg 2+ for methylation, consequently reducing MeHg production (Ullrich et al., 2001;Han et al., 2008).In contrast, reducing conditions can promote microbial mediated S reduction, which in turn can increase Hg methylation (Duran et al., 2008).In addition, high sulfide concentrations in marine environments containing organic matter seem to promote the uptake of Hg 2+ by methylating bacteria such as sulfate reducing bacteria (SRB) maybe due to enhanced Hg bioavailability in mixed DOM-Hg-S complexes (Benoit et al., 2001;Sunderland et al., 2006).Sulfate reducing bacteria have been identified to be the principal methylators of inorganic Hg in sediments (Compeau and Bartha, 1985;King et al., 2002).The range of bacterial activity is large due to the variation in quantity and quality of organic matter, abundance of SRB, temperature, and SO 2− 4 availability (Pallud and van Capellen, 2006).Various PLFA have been frequently used as biomarkers (e.g.SRB) to describe the microbial community structure in different environments (Taylor and Parkes, 1985;Coleman et al., 1993;Macalady et al., 2000;Wegener et al., 2008).The PLFA which might indicate the presence of Desulfobacter are 10Me16:0, cy17:0, 10Me18:0, and cy19:0 (Kohring et al., 1994).Those PLFA were abundant in both studied soils (Fig. 3).The fatty acids 10Me16:0 and 10Me18:0 might serve as indicators for Desulfobacter and additionally for actinomycetes (Taylor and Parkes, 1983;Frostegård, 1993).The polyunsaturated fatty acid 18:2ω6,9 might indicate the occurrence of Desulfovibrio (Macalady et al., 2000) or fungi (Frostegård, 1993).Desulfobulbus species are characterized by unbranched fatty acids such as 15:0 (Taylor and Parkes, 1983), which is widely distributed among different bacterial taxa (Macalady et al., 2000).Iron reducing bacteria such as Geobacter species are also able to methylate mercury in pure cultures at rates comparable to Desulfobulbus (Fleming et al, 2006;Kerin et al., 2006;Windham-Myers et al., 2009).Additionally, Avramescu et al. (2011) Kohring et al. (1994), Geobacter metallireducens mainly consists of the PLFA 16:1ω7c and 16:0, which are abundant in our soils (Fig. 3).These PLFA can also be found in many other organisms (Zelles, 1997).However, we might speculate that Fe reducing bacteria could be present in our soils and therefore might also contribute to Hg methylation in the current study.The soil W1 contains a higher quantity of PLFA compared to W2, both in numbers (W1 = 26; W2 = 20) and in total PLFA biomass (W1 ≈ 27 000 pmol g −1 ; W2 ≈ 17 000 pmol g −1 ).The number of PLFA which are related to SRB is 6 for both soils and the total biomass of these PLFA is approximately 5000 pmol g −1 for both soils.As measured by the total PLFA biomass, W1 contains 19 % PLFA related to SRB whereas W2 contains 29 % PLFA related to SRB.This indicates that SRB are more dominant at site W2.These values lead to the assumption that the relatively high Hg concentrations at the W2 site seem to have no direct toxic effect on SRB.
The statistical relationships between Hg t and Cl − as well as between MeHg and Cl − are modest (Table 3) indicating that interrelations of these parameters might exist.Chloride can influence Hg speciation due to the competition of Cl − with Hg t and MeHg for binding sites of soil particles, which can reduce Hg adsorption onto soil particles and promote the release of Hg into the aquatic phase (Yin et al., 1997;Liu et al., 2009).This process probably occurred in our study.Moreover, the behavior of Hg in the soluble fraction can partly be affected by the formation of Hg-Cl-complexes, which is relevant at E HatpH5 >500 mV (Davis, 1997).Under the mostly acidic pH conditions which occurred in the current study, the formation of partly water soluble HgCl 2 as well as nearly water insoluble Hg 2 Cl 2 is possible (Davis, 1997;Ullrich et al., 2001).Generally, in solution [MeHg]Cl is formed in the presence of Cl − (Skyllberg et al., 2003).Thus, rising Hg t and MeHg concentrations with increasing Cl − concentrations in our study might indicate the formation of Hg-Cl compounds probably mostly under aerobic conditions (Davis et al., 1997;Takeno, 2005).The weak relationship between E HatpH5 and Cl − reveals that Cl − concentrations are not decisively influenced by E H .
In general, Fe(hydr)oxides are able to adsorb Hg to a certain extent (Fernandez-Martinez et al., 2006;Liu et al., 2009;Harris-Hellal et al., 2011).Moreover, Mehrotra and Sedlak (2005) explained decreased mercury methylation in anoxic wetland slurries upon amendment of Fe(II) by reduced availability of Hg for methylation due to the formation of FeS which subsequently decreased the pool of bioavailable neutral mercury-sulfide species.Hollweg et al. (2009) indicate that Hg interacts with inorganic sulfur ligands in FeS complexes decreasing the bioavailability of Hg.In the current study, both soils contain considerable amounts of Fe (Table 1) and Fe(hydr)oxides should precipitate at high E H which is confirmed by the negative relationship between Fe in the soluble fraction and E HatpH5 .However, no evidence was found in this study that Hg t was linked to the Fe cycle whereas MeHg showed a weak relationship to Fe (Table 3).This may be attributed to the fact that DOC competes with Fe(hydr)oxides for binding Hg and MeHg (Feyte et al., 2010) and the DOC contents were high in our study.Additionally, pH might be an important factor in this context, because Hg is only preferentially sorbed to Fe(hydr)oxides in the neutralalkaline pH-range (Ullrich et al., 2001).In summary, redox variations seem to affect the concentrations of dissolved Hg t and MeHg indirectly through related changes in DOC, sulfur cycle, and microbial interaction and community structure whereas E H and pH values, as well as concentration of dissolved Fe and Cl − seem to play subordinate roles in Hg mobilization and methylation under our experimental conditions.

Conclusions
In our E H experiment the ln(DOC/Hg t ) ratio is positively correlated to net MeHg production.This indicates that the ln(DOC/Hg t ) ratio seems to play an important role for the Hg net methylation.Dissolved organic carbon itself can mobilize Hg and MeHg due to the formation of soluble complexes.Mercury methylation also seems to be linked to the S chemistry while the influence of Fe and Cl − on Hg methylation and speciation seems to be weak in our study.However, the methylation of Hg seems to be affected by the soil microbial community.On the one hand, Hg methylation might be favored by reducing conditions through enhanced microbial activity such as SRB bacteria, as indicated by the presence of the respective PLFA biomarkers.On the other hand, reducing conditions might lead to the formation of hardly available HgS what might contribute to a decrease of MeHg production.In conclusion, future studies on the fate of mercury in wetland soils should include silty and clayey soil material and should focus on the specific role of the soil microbial community structure.

Table 1 .
Selected properties of the studied soils (C org : organic carbon, N t : total nitrogen) total metal concentrations (aqua regia soluble), and total sulfur (S) of the studied bulk soils Wupper 1 (W1) and Wupper 2 (W2).

Table 2 .
Variations (mean, median, range)of concentrations of elements and compounds in the soluble fraction as well as pH, and redox potential (E H ) in the slurry.
DOC in the soluble fraction increased.These correlations were strong for MeHg and DOC and modest for the other parameters.Iron was positively related to MeHg but not to Hg t .Values of pH revealed a modest negative relation to MeHg and no correlation to Hg t (Table

reported that T. Frohne et al.: Biogeochemical factors affecting mercury methylation rate
Fe reduction through Fe reducing bacteria might decrease demethylation.According to