Science of the Total Environment 472 (2014) 185–193

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

Iron oxides stimulate microbial monochlorobenzene in situtransformation in constructed wetlands and laboratory systems

Marie Schmidt a, Diana Wolfram a, Jan Birkigt a, Jörg Ahlheim c, Heidrun Paschke b,Hans-Hermann Richnow a, Ivonne Nijenhuis a,⁎a Department of Isotope Biogeochemistry, Helmholtz Centre for Environmental Research — UFZ, Permoserstrasse 15, 04318 Leipzig, Germanyb Department of Analytical Chemistry, Helmholtz Centre for Environmental Research — UFZ, Permoserstrasse 15, 04318 Leipzig, Germanyc Department of Groundwater Remediation, Helmholtz Centre for Environmental Research — UFZ, Permoserstrasse 15, 04318 Leipzig, Germany

H I G H L I G H T S

• MCB removal in anoxic gravel bed of a planted and an unplanted constructed wetland was accompanied by iron(II) mobilisation.• Higher MCB removal related to Phragmatis plants and summer season• MCB mineralisation was stimulated with nitrate and iron(III) as electron acceptor in anoxic laboratory microcosm.• MCB removal appears mainly linked to iron reduction; benzene removal may be linked to both sulphate and iron reduction in situ.

⁎ Corresponding author. Tel.: +49 341 235 1356; fax: +E-mail address: ivonne.nijenhuis@ufz.de (I. Nijenhuis)

0048-9697/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.scitotenv.2013.10.116

a b s t r a c t

a r t i c l e i n f o

Article history:Received 11 September 2013Received in revised form 30 October 2013Accepted 30 October 2013Available online xxxx

Keywords:Constructed wetlandChlorobenzeneIn-situ biodegradationTransition/gradient zonesIron and nitrate

Natural wetlands are transition zones between anoxic ground and oxic surface water which may enhance the(bio)transformation potential for recalcitrant chloro-organic contaminants due to the unique geochemical con-ditions and gradients. Monochlorobenzene (MCB) is a frequently detected groundwater contaminant which istoxic and was thought to be persistent under anoxic conditions. Furthermore, to date, no degradation pathwaysfor anoxic MCB removal have been proven in the field. Hence, it is important to investigate MCB biodegradationin the environment, as groundwater is an important drinking water source in many European countries.Therefore, two pilot-scale horizontal subsurface-flow constructed wetlands, planted and unplanted, wereused to investigate the processes in situ contributing to the biotransformation of MCB in these gradientsystems. The wetlands were fed with anoxic MCB-contaminated groundwater from a nearby aquifer inBitterfeld, Germany. An overall MCB removal was observed in both wetlands, whereas just 10% of the originalMCB inflow concentration was detected in the ponds. In particular in the gravel bed of the planted wetland,MCB removal was highest in summer season with 73 ± 9% compared to the unplanted one with 40 ± 5%.Whereas theMCB concentrations rapidly decreased in the transition zone of unplanted gravel to the pond, a sig-nificant MCB removal was already determined in the anoxic gravel bed of the planted system. The investigationof hydro-geochemical parameters revealed that iron and sulphate reduction were relevant redox processes inbothwetlands. In parallel, the addition of ferric iron or nitrate stimulated themineralisation ofMCB in laboratorymicrocosmswith anoxic groundwater from the same source, indicating that the potential for anaerobicmicrobialdegradation of MCB is present at the field site.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Wetlands, river and lake sediments are ecosystems which are char-acterized by steep gradients of numerous physicochemical parameterslike redox potential and oxygen concentrations. They provide highlyreactive environments creating hot spots for microbial activity. The

49 341 235 450822..

ghts reserved.

interaction of these chemical, physical and (micro)-biological processesis crucial for biochemical gradients and element cycling of e.g. iron andsulphur (Borch et al., 2010) but also for turn-over of organic substancesincluding organic contaminants (Bauer et al., 2008; Imfeld et al., 2009).

In wetlands, these microbial metabolic hot spots include e.g. therhizosphere or the interface between the groundwater and surfacewater providing a microbial habitat, high turnover rates of organiccarbon (Stern et al., 2007) and redox processes of relevant elements(O2, N, P, Fe,Mnand S) (Borch et al., 2010), which are affected by annualvariation depending on the season and vegetation dynamics (Kadlec

186 M. Schmidt et al. / Science of the Total Environment 472 (2014) 185–193

et al., 2012). Several natural processes e.g.microbial degradation, oxida-tion, volatilisation, and/or plant uptake contribute to the organic carbonturnover in wetlands (Imfeld et al., 2009; Kadlec et al., 2012).

Contaminant removal in natural and constructed wetlands or re-latedwater treatment systems (horizontal and vertical sediment col-umns) has been reported for compounds including ammonium(Bauer et al., 2008), pesticides e.g. hexachlorocyclohexanes (HCH)(Bhatt et al., 2009; Brunke and Gonser, 1997), pharmaceuticals suchas ibuprofen and clofibric acid (Matamoros et al., 2009) as well as met-alloids e.g. arsenic (Bauer et al., 2008) or for typically encounteredgroundwater contaminants such as the chloroethenes (Imfeld et al.,2008; Kassenga and Pardue, 2002; Lorah and Voytek, 2004; Novaket al., 2002), benzene (Rakoczy et al., 2011) and monochlorobenzene(Braeckevelt et al., 2007a,b).

Monochlorobenzene (MCB), a recalcitrant and ubiquitous ground-water contaminant (EPA, 2009) accumulates under anoxic conditionsas for example found in the contaminated aquifer (Bitterfeld/Wolfen re-gion, Germany) of a former production site of lindane (γ-HCH) as resultfrom microbial conversion of HCHs. The aerobic microbial degradationof MCB is well characterised (Haigler et al., 1992; Müller et al., 1996;van Agteren et al., 1998).Moreover, MCB biodegradation under hypoxicconditions was proven for laboratory cultures of Acidovorax sp. andPseudomonas sp. (Balcke et al., 2008). However, to date only few reportsare available for anaerobicMCBbiodegradation. Evidences for the in situbiodegradation were obtained using compound specific stable isotopeanalysis in combination with stable isotope tracer experiments(Braeckevelt et al., 2007a; Kaschl et al., 2005; Nijenhuis et al., 2007;Stelzer et al., 2009). The mineralisation of 13C-labelled MCB andincorporation of MCB-derived carbon into biomass was proven inlaboratory microcosms (Nijenhuis et al., 2007; Stelzer et al., 2009).Martinez-Lavanchy et al. (2011) suggested the involvement ofProteobacteria, Fibrobacteres andmicrobial members of the candidatedivision OD1 in the anaerobic MCB degradation, though; a single MCBdegrading strain was not identified. Additionally to a completemineralisation, the dechlorination of dichlorobenzene (DCB) and MCBto benzene by Dehalobacter sp. was reported (Fung et al., 2009; Lianget al., 2011).

To date, it is not clear how and via which pathwaysMCB is removedfrom contaminated anoxic aquifers. MCB may undergo different degra-dation reactions in situ (Fig. 1). From laboratory experiments it isknown that under strictly anoxic conditions MCB can be i) reductivelydechlorinated to benzene (Fung et al., 2009) which can be subject tomineralisation, shown for nitrate, iron and sulphate reducing as wellas methanogenic conditions (Burland and Edwards, 1999; Liang et al.,

Fig. 1. Proposed anaerobic degradation pathways, suggested (dashed line) and published (solidzene. Reductive dechlorination of DCB via MCB to benzene (open arrows) reported by Fung etunder nitrate (Burland and Edwards, 1999), iron (Lovley, 2000) and sulphate reducing conditiacceptor was suggested as metabolic pathway at corresponding reducing conditions (this stud

2013; Lovley, 2000; Vogt et al., 2007). Alternatively, ii) we postulatethat MCB can be directly mineralised with nitrate, iron, sulphate orCO2 as terminal electron acceptor. Therefore we propose that thesetwomain pathways, in whichMCB functions as either i) electron accep-tor or ii) electron donor, may take place sequentially or simultaneouslyin the environment leading to MCB mass reduction.

To investigate the MCB removal under anoxic conditions and atanoxic/oxic interfaces and to elucidate the contributing pathwaysin situ, a model subsurface flow constructed wetland, representinga highly reactive gradient system, was used. Previously, MCB in situbiodegradation was demonstrated in the model wetland and ap-peared to correlate with ferrous iron mobilisation. However, theuse of aquifer material with high organic carbon content (browncoal) restricted the analysis of processes in the previous study(Braeckevelt et al., 2007a,b).

Therefore, the aquifer material was replaced by characterisedgravel as sediment to provide a more defined system for theinvestigations. Additionally, reference experiments with anoxicgroundwater from the field site were conducted using laboratorymicrocosms.

In our study we aimed at understanding the processes contribut-ing to anaerobic MCB degradation with the following questions: 1) isMCB degraded in the model system, 2) which potential for microbialMCB degradation provides the contaminated groundwater and 3)which conditions support the anaerobic microbial MCB degradationin situ?

2. Material and methods

2.1. System description

Two horizontal subsurface flow wetlands, planted and unplanted,respectively, were set up at the SAFIRAproject area in Bitterfeld/Wolfen,Germany (Wycisk et al., 2003) and consisted of a stainless steel basinwith the dimension of 6 m (length) × 1 m (width) × 0.6 m (depth),each (Fig. 2/Supporting Information (SI) Fig. S1). Both wetlandscontained a gravel bed (5 m × 1 m × 0.5 m) and an open water pond(1 m × 1 m × 0.4 m) at the outflow side. The gravel with a grain sizeof 0.6 to 6 mm (commercially obtained from the “KiesgrubeWallendorf”, quaternary gravel of the “Saale”-terrace, Germany) had aporosity of 41% and contained 252 ± 67 mg kg−1 iron, based on ex-traction with 0.5 N HCl (analysis with ICP-AES, SI, Table S1). To assessthe vegetation effect on MCB removal and wetland geochemistry, onewetland segment was planted with common reed (Phragmites australis

line), for the degradation of dichlorobenzene (DCB),monochlorobenzene (MCB) and ben-al. (2009) with hydrogen as electron donor. Anaerobic oxidation of benzene was reportedons (Vogt et al., 2007). Direct mineralisation of MCB with NO3

−, Fe3+ or SO42− as electron

y).

Fig. 2. Schematic drawing of the planted horizontal subsurface flowmodel wetland (HSSF-CW), constructed as steel tub, with 3 inflow ports at the left site, one outflow tube at the waterpond (1 × 1 × 0.4 m) at the right site installed in a depth of 0.5 m. The HSSF-CW was filled with gravel which closed with a lake bank and the open water pond. Dimensions in m.

187M. Schmidt et al. / Science of the Total Environment 472 (2014) 185–193

(Cav.) Trin ex. Steud.). MCB contaminated anoxic groundwater wasdirectly pumped from the contaminated aquifer (Bitterfeld-Wolfen,Saxony-Anhalt, Germany; (Heidrich et al., 2004)) via a groundwaterwell into each wetland simultaneously via 3 inlet ports (input flowrate of 1 L h−1). The theoretical retention time, assuming a homoge-neous flow through the complete system, was estimated to be30 days.

2.2. Monitoring and sampling procedure

The hydrochemistry of the groundwater from well HB 5 (16 to 22 mdeep) for the experiment period (2010–2011) is presented in theSupporting Information (SI, Table S2). The hydrochemistry of porewater samples in both wetlands was monitored during 11 samplingcampaigns from May 2010 to November 2011, whereas sampling wasstopped in the winter months. Samples were taken axial at 0.5, 1, 2, 3,4 m (gravel bed) and 5.8 m (water pond) distance from the inflow in0.3, 0.4 and 0.5 m depth from the gravel surface (Fig. 2). The inflowwas sampled at the inflow operating pump. Samples from the pondwere obtained exclusively in 0.4 m depth in 2010 and in 2011 in allthree depths (0.3, 0.4, 0.5 m) to investigate an assumed depth depen-dent distribution. Seventeen (year 2010) and 19 (year 2011) locationswere sampled (Fig. 2) in each wetland, respectively. Water sampleswere taken successively using stainless steel lances (inner diameter:3.5 mm) via pumping (flow rate 60 mL min−1) with a peristalticpump (Reglo-Analog, ISMATEC, Wertheim-Mondfeld, Germany)equipped with elastic tubes (inner diameter: 4 mm; ISO-VERSINIC,LIQUID-scan (Friedrichshafen, Germany)). On average 140 mL porewater was taken per sampling location each, representing a bulk volumewith extraction radius of 4.4 cm. Sample aliquots were prepared, indi-vidually preserved depending on the analysis method (SI, Table S2),closed airtight and stored at 8 °C until analysis as described below.

2.3. Chemicals

Chemicals were purchased from Sigma-Aldrich Chemie (includingFluka, Supelco (Bellefonte, PA, USA) and Riedel de Haën (Seelze,Germany)), Merck (Darmstadt, Germany) Chemotrade (Leipzig,Germany), or Linde Gas AG (Pullach, Germany) at the highest purityavailable. Gases were purchased from Airproducts (Hattingen,Germany). Benzene and 1,3-dichlorobenzene (1,3-DCB) were obtainedfrom Merck, monochlorobenzene (MCB) from Fluka; 1,2-dichloroben-zene (1,2-DCB), and 1,4-dichlorobenzene (1,4-DCB) from Supelco.Benzene-[13C6] was purchased from Sigma-Aldrich (St. Louis, USA).MCB-[13C6] and 1,2-DCB-[13C6] were purchased from Chemotradewith N98% purity.

2.4. Analytical methods

2.4.1. Physico- and geochemical parameters of the pore water samplesConcentrations of the following inorganic parameters were

determined: ammonium (NH4+), nitrate (NO3

−) and nitrite (NO2−),

phosphate (PO43−), iron(II) (Fe(II)) and iron total (Fetot), chloride (Cl−),

sulphate (SO42−) and sulphide (HS-−/S2−) and manganese total

(Mntot). Triplicate measurements of single water samples were done,whereas the standard deviation was always below 10%. Redox potential(Eh), dissolved oxygen concentration (DO), pH value, electrical conduc-tivity (EC) and the water temperature (T) were measured during sam-pling in the field. Parameters analysed are given in the SupportingInformation, Table S1 including sample volume, preservation of sample,analytical method and the limit of detection of the respective method.

2.4.2. Analysis of benzene, MCB and DCBConcentrations of benzene, MCB, DCB and metabolites were

analysed by gas chromatography using an HP 6890A gas chromato-graph with flame ionisation detector (Agilent Technologies, Palo Alto,USA) equipped with the fused silica capillary column Rtx®-VMS(30 m × 0.25 mm × 1.4 μm, Restek GmbH, Bad Homburg Germany).Four mL pore water aliquots were prepared in 10 mL headspace (HS)vials (Supelco Analytical, USA), 500 µL saturated sodium sulphate(pH 1) was added and vials were closed with polytetrafluoroethylenecoated septa (VWR™ International, Radnor, PA, USA). Vials were incu-bated for 5 min at 60 °C whilst shaking to enhance compound transferfrom liquid to gas phase prior to automatic injection (GERSTEL MPS2Multi-Purpose Sampler; GERSTEL GmbH & Co. KG, Germany) of 500 μLof headspace sample. Injector temperature was 250 °C, the split ratiowas 1:1 and helium was used as carrier gas at a flow rate of 45 mL min−1. The temperature programme was: initial temperature 35 °C(2 min), 4 °C min−1 to 60 °C, 20 °C min−1 to 225 °C (1 min). Detectortemperature was set to 280 °C. Per sampling location in the wetland du-plicate samples were taken and analysed applying an external threepoint calibration using commercial standards from Supelco (Bellefonte,PA, USA) containing benzene, MCB, 1,2- and 1,4-DCB. Concentration of1,3-DCB was always below the detection limit.

Stable carbon isotope composition of benzene, MCB, 1,2-DCB and 1,4-DCB was determined by gas chromatography combustion isotope ratiomass spectrometry (GC-C-IRMS). Compounds were gas chromatographi-cally separated using the column Zebron-ZB1 (60 m × 0.32 mm × 1 μm,Phenomenex, Inc.; Torrance, USA). Before automatic (CTC Analytics,CombiPAL) injection of 500 μL headspace sample, vials were incubatedfor 10 min at 60 °C. The temperature programme of themethodwas: ini-tial temperature 40 °C (5 min), 4 °C min−1 to 120 °C, 3 °C min−1 to150 °C, 20 °C min−1 to 300 °C (5 min). Helium was used as carrier gas.Injector temperature was 250 °C. Detector temperature was 280 °C. Sta-ble carbon isotope composition of the reaction products CO2 and CH4

Fig. 3.MCB concentration loss (%) calculated for gravel bed (=difference of sample at 4 m(averaged for 0.3, 0.4 and 0.5 mdepth) to inflow concentration) and for pond (=differenceof water pond sample (2010: in 0.4 m and 2011 averaged for 0.3; 0.4; 0.5 m depth) to in-flow concentration) for the unplanted (A) and the planted (B)wetland, respectively, duringthe sampling in 2010 and 2011. Error bars indicate standard deviation of themean concen-tration loss (n = 3 depth). No standard deviation is given for pond single samples (depth0.4 m (n = 1) during 05–10 to 11–10).

188 M. Schmidt et al. / Science of the Total Environment 472 (2014) 185–193

were determined at an isotherm of 40 °C using a CP-PoraBOND column(50 m × 0.32 mm × 5 μm, J&W Scientific, Waldbronn, Germany) forseparation. The carbon isotope composition is reported in δ-notation(‰) relative to the Vienna Pee Dee Belemnite standard (V-PDB, IAEA —

Vienna) (Coplen et al., 2006).

2.5. Laboratory microcosms

Anoxic laboratory microcosms with contaminated anoxic ground-water from the aquifer were set up to investigate the MCB, DCB orbenzene biodegradation potential of the present microbial commu-nity. Microcosms were prepared in an anaerobic glovebox (98–96%N2/2–4% H2) (CoyLab, Grass Lake, MI, USA). Serum bottles (120 mL)were filled with 50 mL groundwater, amended with 1 mg L−1 ofresazurin as a redox indicator, closedwith Teflon-coated grey butyl rub-ber stoppers and crimped with aluminum crimps. The 38 microcosmsets consisted of two autoclaved abiotic controls and three biotic micro-cosms, each. Vitamin solution (0.5 mL) and corresponding electronacceptors (NO3

− [5 mM], SO42− [5 mM], α-FeO(OH) [1 g/50 mL])

were added under sterile and anoxic conditions. Microcosms wereamended with non-labelled or 13C-labelled benzene, MCB and 1,2-DCB (100 μmol L−1), respectively. In microcosms with 13C-labelledsubstrate, 25% benzene were substituted by [13C6]-benzene, 22% MCBwere substituted by [13C6]-MCB and 23% 1,2-DCB were substituted by[13C6]-1,2-DCB. Due to the presence of benzene, MCB and DCB in theused groundwater (for groundwater composition see SI, Table S2), theactual substrate concentration in each microcosm was higher. Micro-cosms with 13C-labelled/non-labelled substrate and without amend-ment of additional electron acceptors were set up as biotic control.The potential for reductive dechlorination ofMCB and DCB, respectivelywas analysed with the addition of lactate or hydrogen/acetate as elec-tron donor/carbon source. Detailed composition of each microcosmset is presented in the Supporting Information in Table S3. Microcosmswere incubated statically in the dark at 20 °C. Concentrations (singleanalysis of 0.5 mL sample in 10 mL HS vial) and carbon stable iso-tope ratios of benzene, MCB, DCB and products (triplicate analysisfor 13C/12C [‰] of substrates and single analysis of 13C/12C [‰] ofthe product; e.g. CO2 and CH4) were monitored for triplicate bottles(GC-FID and GC-C-IRMS; see Section 2.4.2). Bottles were regularlysampled (after 35, 96, 188, 372, 492, 687 and 992 days) to analysethe concentration of the aromatic hydrocarbons (benzene, MCBand DCB) and to determine the carbon isotope signatures of CO2

and CH4 as potential degradation products within the anoxic oxida-tion as well as MCB and benzene as products of reductive dechlorina-tion (after 992 days). The changes of isotope values and the 13C/12Catom [‰] of CO2 were used to estimate the MCB and benzene degra-dation in these microcosms (Richnow et al., 1999).

2.6. Statistical methods

Principal component analysis (PCA)was performed using the R soft-ware (RCoreTeam, 2013) in order to evaluate trends of hydro-chemicalparameters of the planted and unplanted wetland. The following pa-rameters were included in the correlation analysis: Fe(II), Fetot, MCB,1,2-DCB, 1,4-DCB, benzene (B), CH4, SO4

2−, HS−/S2−, Mntot, NH4+, Cl−,

EC, Eh, DO and pH. Parameters, with concentrations at or below thelimit of detection (LOD) (NO3

−, LOD: 0.15 mg L−1; NO2−, LOD:

0.05 mg L−1) were omitted from themulti-correlation analysis. Param-eters only determined in the 2nd year, such as redox potential and DO,were just included in the PCA of the 2nd year data set. The redox poten-tial was excluded from the PCA of the 1st year data set due to discontin-uous measurement. Correlation analysis was done using the Spearmanrank sum coefficient. PCA calculationswere based on a dissimilarity dis-tance matrix (Euclidean/Ward method). The PCA was used to analysethe relationship between the different sampling points: inflow, 0.5 m,1 m, 2 m, 3 m and 4 m, in upper (0.3 m depth), middle (0.4 m depth)

and bottom (0.5 m depth) layer and in the pond (0.4 m depth); seealso Fig. 2, with reference to their respective pore water parameters.The sampling location corresponds to the object and the chemical pa-rameters to the descriptors (represented by the vectors) of the multi-variate analysis. All data points were computed for each sample takenin X/Y coordinates (X = horizontal = distance and Y = vertical =depth in the wetland) in 2010 and 2011, separately.

3. Results and discussion

3.1. Monochlorobenzene removal in the wetlands

MCB removal was determined for every sampling campaign in bothwetlands during the two years of investigation and the extent of MCBconcentration loss was found to vary in the unplanted and plantedwet-land as well as during different seasons (Fig. 3). The MCB concentrationloss was calculated for the sampling points at different distances fromthe inflow at 4 m in the gravel bed (mean of 0.3, 0.4 and 0.5 m depth)and at 5.8 m in the pond (0.4 m depth for 2010 and mean of threedepths for 2011), and compared to the inflow concentration (Fig. 3).

MCB removal at 4 mdistance (gravel bed) from the inflowwas abouttwice higher (two-sample T-test; significance: p 0.01) in the plantedwetland with 59 ± 14% on average for two years (N = 11 samplingmonths) compared to the unplantedwetland (28 ± 12% removal), indi-cating a positive effect of the plants. Including the MCB decrease in the

189M. Schmidt et al. / Science of the Total Environment 472 (2014) 185–193

pond, the overall removal was not significantly different with 88% (±14%) for the planted and 73% (±28%) for the unplanted system (averageof 11 sampling months). In both wetlands MCB concentration loss washigher in the summer season (June, July, August; Fig. 3) whereas highestMCB removal (73 ± 9%, average of sampling points at 4 m) was identi-fied in the gravel bed of the planted wetland.

Increased removal in summer compared to spring/autumn may bedue to higher air temperature and lower precipitation in July andAugust (SI, Table S4). These weather conditions could have increasedevaporation and phytovolatilisation of volatile organic compounds(VOCs) like MCB and benzene and potentially contributed to the MCBoverall removal. However no significant increase of temperature inthe unplanted gravel bed was determined (data not shown). Hence,emission of MCB most likely was not a dominant elimination processin the unplanted as well as in the planted system, as reported for a sim-ilar planted wetland (Braeckevelt et al., 2011).

In another study with planted constructed wetlands, plant uptakeand phytovolatilisation of benzene as contribution to benzene removalwas found to beminor. In contrast, over 70% of benzene removalwas re-lated to microbial degradation (Seeger et al., 2011).

A dilution effect of theMCB concentration and other parameters e.g.chloride and sulphate in the gravel bedwas observed for bothwetlands,caused by strong rain events inMay and June 2010 (SI, Table S4). Higherwater volume in thewetland led to a typical stratification pattern of thesubstances, with increasing concentrations (MCB and Cl−) from the topto the bottom zone in thewetland. This patternwas present after strongrain and absent after or during droughts in e.g. June and August 2010(SI, Fig. S2).

In the planted wetland, oxygen is thought to be actively introducedfrom roots to the rhizosphere in the waterlogged gravel, which en-hances the uptake of water and nutrients for plants and provides condi-tions for aerobic microbial oxidation of substances (Armstrong, 1964)e.g. MCB and benzene. Oxygen intrusion in the gravel would facilitatesulphide and ferrous iron oxidation with potentially visible orange-brownish coloration of roots; however sampled wetland plants didnot support this assumption (SI, Fig. S3). Oxygen concentrations in theplanted wetland were lower than in the unplanted one suggesting aquick consumption if introduced by plants.

Even under hypoxic conditions, transformation of MCB takesplace (Balcke et al., 2008). However, if oxygen was entering the sys-tem, it would be rapidly consumed by respiration processes e.g. oforganic substances such as root exudates as well as due to abiotic re-actions with e.g. reduced iron or sulphur species. Indeed, dissolvedoxygen analysed in the pore water samples was generally below1 ppm in the gravel (SI, Table S5) and reduced iron and sulphur spe-cies were observed in the gravel, assuming MCB degradation undermainly anoxic conditions. However, in pond and unsaturated zonegravel/pond of both wetlands, aerobic degradation under oxic orhypoxic conditions may have taken place. Indeed the pond and theupper (0.3 m depth) gravel at 4 m (DO: 5.6 ± 3.3 mg L−1 and2.3 ± 1.9 mg L−1) were oxic, whereas the main gravel part from0.5 till 3 m was anoxic (DO: 0.23 ± 0.26 mg L−1) (averaged dataare shown in SI, Table S6). Furthermore the presence of reducediron in the gravel bed supports the observed anoxic conditions.

3.2. Hydro-geochemical and physico-chemical conditions in the wetlands

The evolution of selected hydro-geochemical parameters during thefirst year of wetland monitoring was investigated using the principalcomponent analysis (PCA) as a multivariate statistical approach. Meanvalues of each parameter for inflow, gravel bed and pond, including or-ganic and inorganic substances are presented in the Supporting Infor-mation for the first (Table S5) and second (Table S6) monitoring year.Datawere plotted for the unplanted (Fig. 4A) and planted (Fig. 4B) wet-land separately in order to test the difference in the evolution of theparameters.

In general, samples formed two and eight clusters (based on Spear-man correlation matrix) in the unplanted and planted wetland, respec-tively (SI, Fig. S4A/B). One small group, containingmainly samples takenfrom the pond and at 4 m distance from the inflow and a larger groupwith samples from the inflow and the first 3 m flow path of gravelbed were identified in the unplanted wetland. In contrast, analysis ofthe planted wetland resulted in eight clusters, where two could beclearly distinguished, accounting for inflow and pond samples, whereasfive clusterswere composed of gravel samples, which indicated a highervariability of conditions in the planted in comparison to the unplantedgravel.

In the unplantedwetland, benzene, MCB, 1,2-DCB, 1,4-DCB, SO42−,

Cl−, NH4+ and EC correlated positive with the sampling depth and

negative with the distance. Thus, a removal of chloro-aromaticsand benzene along the flow path and in shallow gravel zones was in-dicated. Moreover, the concentration loss of the contaminant seemsto be linked to sulphate reduction. However, sulphate and sulphidewere not correlated. Sulphide was often found at the limit of detec-tion and at maximum with 0.5 mg L−1 likely as a result of sulphideremoval by precipitation with Fe(II). In contrast pH, Mntot and Fetot,mainly consisting of Fe(II) (see SI, Table S2), correlated positivewith the distance, which indicates microbial iron reduction and aniron mobilisation along the flow path. Moreover, sulphide correlatednegative with iron, which fits to the formation of iron sulphide species(most likely Fe2S). Methane production with increasing depth impliesmethanogenic activity of archaea in hyporheic zones as expected.

In the planted wetlandMCB, 1,4-DCB, benzene, sulphate, Cl− and ECcorrelated negative with the distance, as observed in the unplantedwetland, indicating mass loss of chloro-aromatics and benzene alongthe flow path. Additionally, removal of MCB and benzene with increas-ing depth in the gravel bed was indicated. Furthermore Fe(II), Fetot,Mntot and pH correlated positive with depth and distance indicatingan iron reduction in deeper zones and along the flow path in the gravel.Axes 1 and 2 together explained 50.9% and 47.7% of the variation of thedata in the unplanted and planted wetland, respectively (cluster, SI,Fig. S4A/B). Thus the hydro-geochemical data explained approximately50% of the distribution of the data.

The hydro-geochemical evolution during the second year ofunplanted and planted wetland was investigated (SI, Fig. S5) byperforming a principal component analysis that revealed a similarpattern compared to the first year.

3.3. MCB removal in the wetlands along the flow path with iron andsulphate reduction

The presence of the reduced species Fe(II) and sulphide in thegravel bed and the principal component analysis suggested iron re-duction and sulphate reduction as important redox processes in thewetlands potentially linked to MCB removal (Fig. 4). Therefore, ferrousiron and sulphate were investigated as indicators for these processeswith respect to MCB removal in both wetlands along the flow path.The deepest sampling zone at 0.5 m with dissolved oxygen concentra-tions below 1 mg L−1 (SI, Table S6), reduced redox conditions andiron and sulphur present in the reduced form (SI, Tables S5 & S6) indi-cating suboxic to anoxic conditions (McMahon and Chapelle, 2008),was chosen to investigate anoxic processes. Four sampling campaigns(months: May, July, August and September 2010) were compared toidentify trends influenced by seasons and vegetation, which are pre-sented in Fig. 5 for the unplanted (A) and planted wetland (B).

Importantly, an increase of the ferrous iron concentration along theflow path was observed in both wetlands. As ferrous iron in theinflowing groundwater was b0.5 mg L−1, it is most likely that iron re-ducing bacteria were present in the water and gravel bed, that utilisediron oxides, covering the gravel particles as electron acceptor (gravelcontained 252 ± 67 mg kg−1 iron). The iron mobilisation was morepronounced and earlier along the flow path in the planted wetland

Fig. 4. Principal component analysis (PCA) ordination plot of the hydrogeochemical variables sampled in the unplanted (A) and planted (B) wetland. Represented are the cor-relations between the geochemical parameters of the 1st monitoring year 2010 (N = 6). The vectors represent Fe2+, ferrous iron; Fetot, total iron; MCB, monochlorobenzene;1,2-DCB, 1,2-dichlorobenzene; 1,4-DCB, 1,4-dichlorobenzene; B, benzene; CH4, methane; SO4

2−, sulphate; HS−/S2−, sulphide; Cl−, chloride; EC, electric conductivity; Mntot, totalmanganese; NH4, ammonium; pH; distance, horizontal sampling point position from inflow to pond (inflow, 0.5 m, 1 m, 2 m, 3 m, 4 m, pond) and depth, vertical sampling pointposition (upper, middle and bottom sampling layer). Single data points are not shown and represented by the vector time.

190 M. Schmidt et al. / Science of the Total Environment 472 (2014) 185–193

compared to the unplanted one with an increase in ferrous iron corre-sponding to the main decrease in MCB in the planted wetland (Fig. 5).

A steep decrease in the sulphate concentration (from 1029 to492 mg L−1) and corresponding increase in the sulphide concentration(from 0.08 to 0.3 mg L−1) were determined in the transition zone grav-el/pond of the unplanted wetland, which suggested sulphate reduction.However, sulphide concentrations measured in both wetlands werebelow 1 mg L−1 (SI, Tables S5 & S6) and at the limit of detection.Most probably, sulphide and ferrous iron, formed by microbial activity,

Fig. 5. The evolution of sulphate (◊), ferrous iron (■) and monochlorobenzene (MCB) (▲) conc(inflow, IN = 0, sampling points in bottom layer (0.5 m depth) and pond, P in 0.4 m depth) inwere done (SO4

2− and Fe2+), whereas the standard deviation was always below 10%. Mean vb10%).

build Fe2S-species, which adsorb to the sediment matrix (Billon et al.,2001) and led to underestimations of sulphide and ferrous iron.

In the unplanted wetland, significant MCB removal was determinedat the transition zone between gravel bed/pond (between 4 m andpond) in all sampling months, e.g. in August (2010) a decrease from5.8 to 0.01 mg L−1 (Fig. 5A). Mineralisation of MCB with molecularoxygen as terminal electron acceptor can be assumed in this transitionzone, as oxic conditions were observed in the pond (SI, Table S6). How-ever, the ferrous iron concentrations increased continuously along the

entration is exemplarily presented in different seasons (1st year) axial along the flow pathunplanted (A) and planted wetland (B). Triplicate measurements of single water samplesalues of duplicate analysis are shown for MCB (standard deviation of external calibration

191M. Schmidt et al. / Science of the Total Environment 472 (2014) 185–193

flow path up to a maximum of 4.9 mg L−1 (at 3 m in July) and furtherdecreased in the pond (0.03 mg L−1). The steep decline of ferrousiron in the oxic water pond suggests a rapid oxidation of dissolvedFe(II) to insoluble Fe(III) oxides and/or the formation of Fe-sulphidesat the anoxic/oxic interface of the transition zone from gravel to thepond.

In the planted wetland, the MCB concentration decreased alreadystrongly in the first meters of the gravel bed (Fig. 5). Higher reductionin the planted gravel was observed in the summer months (July andAugust) and was less pronounced in spring and autumn. Ironmobilisation was more pronounced compared to the unplanted system(up to 9 mg L−1). Interestingly, the decrease in MCB concentration wasaccompanied by a steep ferrous iron increase in July. The course of thesulphate concentration along the flow path was similar to the one ofthe unplanted wetland, with the exception of July and August. In Julyit even increased in the pond, indicating accumulation of sulphate dueto water evaporation or oxidation of sulphide and FeS2. Similar to theunplantedwetland,MCB transformation appears to be linked to iron re-duction in the planted wetland. MCB transformation linked to sulphatereduction cannot be excluded, however, was most likely not present.Additional contribution of aerobic degradation to the MCB removal isprobably relevant in the pond, at the transition zone and in the upperlayers of the wetland.

3.4. MCB degradation in laboratory microcosms

To test the capacity of indigenousmicroorganisms of the groundwa-ter for biodegradation ofMCB, benzene andDCB, laboratorymicrocosmswere prepared with the MCB contaminated groundwater from theBitterfeld aquifer, with amendment of different electron donor/acceptorpairs (SI, Table S3). No reductive dechlorination of DCB andMCB to ben-zene in microcosms amended with lactate or hydrogen/acetate as elec-tron donor/carbon source was observed over two years of incubation(data not shown). However, a significant decrease in MCB and benzeneconcentration (up to 90%) was observed in all microcosms, includingabiotic controls (SI, Fig. S6), likely as a result of sorption to e.g. browncoal particles in the groundwater (Wycisk et al., 2003) and as previouslyobserved (Nijenhuis et al., 2007).

To discriminate biotic from abiotic removal reactions we applied astable isotope tracer approach with 13C-benzene, 13C-MCB and 13C-

Fig. 6. Carbon stable isotope ratio of CO2 during 13C-MCB (A) and 13C-benzene(B) transformation to 13CO2 in microcosms with contaminated groundwater amendedwith NO3

− (◊, black dotted line), α-FeO(OH) (□, red dotted line), SO42− (Δ, green dotted

line). Biotic control (○, blue dotted line) was set up to investigate mineralisation underinitial groundwater conditions over the course of incubation time of about 1000 days.The dashed line represents the mean of all abiotic (autoclaved) controls (bold blackdashed line), with −23 ± 1‰ for 13C-MCB (A) and −24 ± 1‰ for 13C-benzene(B). Mean values and standard deviation of 13C/12C-ratio (‰) of CO2 were calculatedfrom triplicate biotic microcosm bottles per set. (For interpretation of the referencesto colour in this figure legend, the reader is referred to the web version of this article.)

1,2-DCB. The evolution of 13C-enriched CO2 was used as indicator forthe mineralisation of the 13C-labelled substrate. The isotope composi-tion of CO2 did not change significantly in any of the microcosm setsamended with 13C-1,2-DCB (data not shown). All microcosm sets for13C-MCB and 13C-benzene showed an increasing 13C/12C ratio for CO2

over time with the exception of the abiotic controls (Fig. 6). In 13C-MCB microcosms the NO3

− amendment resulted with 598 ± 55‰ inthe highest δ13C-CO2 isotope values, followed by goethite (α-FeO(OH)) with 471 ± 111‰ at day 992, indicating that mineralisationof MCB to CO2 was coupled to nitrate and iron reduction, respectively.The increase in 13CO2 corresponded to an estimated degradation of12% and 9% of added MCB with NO3

− and α-FeO(OH), respectively. Forthe first time, a stimulation of MCBmineralisation by addition of nitrateand iron(III) could therefore be demonstrated.

Contrary, amendment of SO42− did not stimulate the MCB

mineralisation with an estimated mineralisation of 1%, compared to2% in the biotic controls where groundwater derived sulphate wasalso present (approx. 8 mmol L−1). This confirms that mineralisationof MCB with sulphate may be very slow and might be inhibited bySO4

2− or by its product sulphide. However, reduced iron was presentin sufficient amount allowing formation of FeS reducing the concentra-tion of free sulphide (see SI, Table S3).

For benzene, at day 188, mineralisation was highest with nitrate(δ13C-CO2 = 249 ± 25‰) and goethite (84 ± 37‰), while minimalmineralisation activity was observed with sulphate and in the bioticcontrol. However, at day 372 the trend changed and the highest en-richment in 13C-CO2 was observed in the sulphate amended bottles(δ13C-CO2 = 1000 ± 90‰). Finally, after 992 days 13C-benzenemineralisation was highest for nitrate (δ13C = 1241 ± 312‰), butsimilar to sulphate (1054 ± 96‰) and goethite (989 ± 215‰)amended microcosms and the biotic control (914 ± 176‰)(Fig. 6), corresponding to an estimated MCB degradation of 21, 17,18 and 16%, respectively (SI, Table S8).

Approximate degradation rates were maximal with nitrate and ironfor both MCB and benzene at 16.8 and 30.4 nmol L−1 day−1 for nitrateand 13.8 and 21.2 nmol L−1 day−1 for iron, respectively (SI, Table S8).

Complete mineralisation of benzene (initial concentration:111 ± 2 μM) requires 467 μM SO4

2− (SI, Tables S7 & S8); however,after 992 days 10.3 mM sulphate was consumed, suggesting that sul-phate reduction was coupled to oxidation of other substances in thegroundwater. Sulphide could not be detected in themicrocosm as resultof binding to reduced iron, added in the form of FeSO4 (SI, Table S2). Innitrate amended microcosms, nitrate concentrations decreased by1.5 mM and 1.8 mM for the benzene and MCB sets, respectively, suffi-cient for the observed mineralisation of 30.1 and 16.6 μmol L (SI,Table S8). In the case of the goethite amended microcosms, only4.3 μmol L−1 (MCB) and 5.1 μmol L−1 (benzene) of ferrous iron wasobserved after 992 days which would not be sufficient for the observedmineralisation. Formation of Fe2S with sulphide in themicrocosms like-ly led to an underestimation of reduced iron. With a background con-centration of more than 8 mmol L−1 sulphate in the microcosms,enough sulphide can be formed by sulphate reducing bacteria.

Methanogenic activity with respect to MCB and benzene degrada-tion was not observed in any of the microcosms. Isotope ratios for CH4

were analysed for all active and biotic control microcosms amendedwith 13C-labelled MCB and were stable with −43 ± 1‰ and methanedid not increase.

3.5. Processes contributing to the in situ microbial degradation of MCB inthe wetland

MCB removal was determined in both wetlands along the flowpath, whereas it was higher in the planted one, in particular in thesummer months, July and August (Fig. 3). Several processes couldhave contributed to the overall removal of MCB including sorption,evapotranspiration or degradation.

192 M. Schmidt et al. / Science of the Total Environment 472 (2014) 185–193

In both segments, sorptionmay occur to e.g. the brown coal particlespresent in the groundwater and in the planted segment, to the plantmaterial. Additionally, evaporation in the unplanted one or evapotrans-piration by plants in the planted system,may contribute toMCB remov-al (Braeckevelt et al., 2011).

However, only microbial biodegradation leads to a sustainablebreakdown of MCB in the ecosystem. Indeed, in our microcosm study,stimulation ofMCBmineralisationwas observed by amendment of goe-thite or nitrate. Moreover, benzene mineralisation was stimulated withgoethite, nitrate or sulphate as reducing agents.

In thewetland gravel, only Fe(III) and sulphatewere available as po-tential electron acceptorswhile in the upper sediment layer, at the plantroots and in the pond, oxygen may have been the main terminal elec-tron acceptor used.

Theoretically 9.9 mg L−1 oxygen are consumed for the complete ox-idation of an inflow MCB concentration of 5 mg L−1 (C6H5Cl + 7O2 + 4 H2O → 6 HCO3

− + 7 H+ + Cl−); however oxygen concentra-tions in the gravel bed would not be sufficient for significant aerobicdegradation (≤0.1 mg L−1). At the anoxic/oxic transition zone in thesediments and in the pond aerobic degradation may further have con-tributed to a major extend to MCB removal. In the gravel bed, twomain degradation pathways could have contributed to MCB degrada-tion: reductive dechlorination of MCB to benzene with subsequent oxi-dation of benzene or the direct oxidation ofMCBwith Fe(III) or sulphateas electron acceptor (Fig. 1).

Reductive dechlorination of MCB to benzene may have taken placein the anoxic deep gravel zones with subsequent anaerobic degradationunder sulphate reducing condition as suggested by the microcosmexperiment. Indeed, benzene was present; however, it was initiallypresent in the groundwater as contaminant and its concentration de-creased similarly to MCB. Additionally, no significant change in the car-bon stable isotope composition of MCB and of benzenewas observed asit would have been expected during dehalogenation (Fung et al., 2009;Liang et al., 2011). Further, benzene degradation was taking place inthe microcosms with MCB, as benzene did not accumulate in themicrocosms.

Sulphate and iron reduction/mobilisation processes were indicat-ed in the wetland. In the wetland gravel maximal concentrations forferrous iron were determined with 3 mg L−1 (unplanted) and25 mg L−1 (planted) in 2011 accounting for a theoretical degrada-tion of 0.2 mg L−1 and 1.8 mg L−1 MCB in the unplanted andplanted wetland assuming an exclusive consumption of electronsfor MCB oxidation (SI, Table S7). However, it is likely that ferrousiron concentrations were underestimated due to formation of Fe2Sspecies in the gravel bed and continuous (re-)oxidation to Fe(III) asresult of oxygen intrusion.

Oxygen may penetrate anoxic zones of the wetland or may bereleased by roots (Armstrong, 1964) where it reacts spontaneouslywith Fe(II) and supports the growth of Fe(II)-oxidizing bacteria(Weiss et al., 2003), generally stimulating iron cycling. Further,plants release organic and inorganic exudates in the rhizospherewhich increase the microbial response and mobilisation of ferrousiron. The strong sulphide smell and the visible black precipitationin the pond of both wetlands, especially at the last sampling cam-paign (SI, Fig. S5), indicated Fe2S formation, supporting the reduc-tion of Fe(III) to Fe(II) and would account for observed decreasedtotal iron concentration (SI, Table S5). Very low mineralisation ofMCB (1 to 2%) was observed with sulphate in microcosms.Additionally, the increase in sulphide in the gravel bed was notsignificant. The mineralisation of 5 mg L−1 MCB would consume14.9 mg L−1 sulphate but this amount is, however, not distinguish-able from the variation of the high sulphate background concentra-tion (~1 g L−1) in the groundwater. Most likely, benzenedegraders used the sulphate as terminal electron acceptor as ob-served in the microcosms and as has been previously described(Vogt et al., 2007).

4. Conclusion

In the presented study the processes contributing to the anaerobicMCB degradation were investigated in a planted and unplanted con-structed horizontal subsurface-flow wetland system operated withgroundwater from a contaminated aquifer in Bitterfeld (Germany).We aimed at understanding the processes contributing to anaerobicMCB degradation with the following questions: 1) is MCB degraded inthe model system, 2) which potential for microbial MCB degradationprovides the contaminated groundwater and 3) which conditions sup-port the anaerobic microbial MCB degradation in situ.

In summary, MCB removal was demonstrated in the planted andunplanted wetland system, which was potentially linked to iron re-duction in the anoxic gravel filter. In the unplanted wetland highestdecrease in MCB concentration was determined in the transitionzone gravel/pond likely due to a given enhanced microbial contami-nant transformation as a result from steep gradients in e.g. oxygenand stimulation of e.g. iron cycling. In laboratory microcosms, stimula-tion of MCB mineralisation with nitrate and goethite as possible termi-nal electron acceptor were proven, revealing the potential of theindigenous microbial community to mineralise MCB under nitrate andiron reducing conditions. In the groundwater, nitrate was detectedonly in small concentrations and iron wasmostly present in its reducedform. Stimulation of iron cycling due to oxygen intrusion into the wet-land can be assumed leading to availability of Fe(III) species suggestingthat iron reduction may be the main redox process in MCB degradationin anoxic zones of these systems. Reductive dechlorination of MCB tobenzene (as well as DCB to MCB/benzene) could neither be detectedin the microcosms nor in the wetlands, which might be due to compe-tition of dehalogenating and sulphate/iron reducing bacteria as well asthe absence of the respective microorganisms or lack of appropriateconditions for organohalide-respiration in the groundwater.

In the planted and unplanted wetland, processes may take place se-quentially or simultaneously. In the unplanted system MCB removalmostly took place in the transition zone gravel/pond, where oxidationwith molecular oxygen is assumed, whereas in the planted wetlandthe main removal was in the gravel bed. The plants have an importantimpact on the biodegradation of MCB in the wetland, providing habitatfor the microbial community and enhancing substance turnover for el-ement cycles, creating specific redox conditions. To date it is not knownwhichmicroorganismsmay be directly involved inMCB removal in situunder anoxic conditions. Therefore, the role of specific microorganismsaswell as the composition and dynamics ofmicrobial community at dif-ferent redox zones and their involvement in theMCB degradation needto be elucidated to provide a full understanding of the process. Detailedunderstanding of MCB degradation under in situ conditions shouldallow design of treatment and management options of contaminatedgroundwater.

Acknowledgements

The SAFIRA project of the UFZ provided the pilot-scale constructedwetland for this study. We are grateful to Dr. Matthias Gehre, UrsulaGünther and Falk Bratfisch, for their scientific, technical and analyticalsupport. We thank the Analytical Chemistry Department of the UFZfor measurements of the inorganic parameters and especially OliverThiel and Sabine Täglich of the Department of Groundwater Remedia-tion for their assistance in the field and laboratory work. We thankGwenaël Imfeld for discussion and help during performing the statisti-cal analysis with R and Kevin Kuntze for discussion. Without the sup-port of all the sample takers, the collection of all the data sets wouldnot have been possible, especially Sascha Lege, Conrad Dorer, BarbaraKlein and many others. This study was funded by the EU FP7 projectGenesis (contract number: 226536) and SAFIRA II Project “Compart-ment Transfer” (CoTra) of the Helmholtz Centre for EnvironmentalResearch — UFZ.

193M. Schmidt et al. / Science of the Total Environment 472 (2014) 185–193

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.scitotenv.2013.10.116.

References

Armstrong W. Oxygen diffusion from roots of some British bog plants. Nature 1964;204:801.

Balcke GU, Wegener S, Kiesel B, Benndorf D, SchlömannM, Vogt C. Kinetics of chloroben-zene biodegradation under reduced oxygen levels. Biodegradation 2008;19:507–18.

Bauer RD, Maloszewski P, Zhang YC, Meckenstock RU, Griebler C. Mixing-controlledbiodegradation in a toluene plume — results from two-dimensional laboratoryexperiments. J Contam Hydrol 2008;96:150–68.

Bhatt P, Kumar MS, Chakrabarti T. Fate and degradation of POP-hexachlorocyclohexane.Crit Rev Environ Sci Technol 2009;39:655–95.

Billon G, Ouddane B, Boughriet A. Chemical speciation of sulfur compounds in surfacesediments from three bays (Fresnaye, Seine and Authie) in northern France, andidentification of some factors controlling their generation. Talanta 2001;53:971–81.

Borch T, Kretzschmar R, Kappler A, Cappellen PV, Ginder-Vogel M, Voegelin A, et al. Bio-geochemical redox processes and their impact on contaminant dynamics. Environ SciTechnol 2010;44:15–23.

Braeckevelt M, Reiche N, Trapp S, Wiessner A, Paschke H, Kuschk P, et al. Chlorobenzeneremoval efficiencies and removal processes in a pilot-scale constructed wetlandtreating contaminated groundwater. Ecol Eng 2011;37:903–13.

Braeckevelt M, Rokadia H, Imfeld G, Stelzer N, Paschke H, Kuschk P, et al. Assessment of insitu biodegradation of monochlorobenzene in contaminated groundwater treated ina constructed wetland. Environ Pollut 2007a;148:428–37.

Braeckevelt M, Rokadia H, Mirschel G, Weber S, Imfeld G, Stelzer N, et al. Biodegradationof chlorobenzene in a constructed wetland treating contaminated groundwater.Water Sci Technol 2007b;56:57–62.

BrunkeM, Gonser T. The ecological significance of exchange processes between rivers andgroundwater. Freshw Biol 1997;37:1–33.

Burland SM, Edwards EA. Anaerobic benzene biodegradation linked to nitrate reduction.Appl Environ Microbiol 1999;65:529–33.

Coplen TB, BrandWA, GehreM, GroningM,Meijer HA, Toman B, et al. After two decades asecond anchor for the VPDB delta13C scale. Rapid CommunMass Spectrom 2006;20:3165–6.

EPA. Chlorobenzene fact sheet. EPA; 2009.Fung JM, Weisenstein BP, Mack EE, Vidumsky JE, Ei TA, Zinder SH. Reductive

dehalogenation of dichlorobenzenes and monochlorobenzene to benzene in micro-cosms. Environ Sci Technol 2009;43:2302–7.

Haigler BE, Pettigrew CA, Spain JC. Biodegradation of mixtures of substituted benzenes byPseudomonas sp strain-Js150. Appl Environ Microbiol 1992;58:2237–44.

Heidrich S, Schirmer M, Weiss H, Wycisk P, Grossmann J, Kaschl A. Regionally contami-nated aquifers — toxicological relevance and remediation options (Bitterfeld casestudy). Toxicology 2004;205:143–55.

Imfeld G, Aragones CE, Zeiger S, von Eckstadt CV, Paschke H, Trabitzsch R, et al. Trackingin situ biodegradation of 1,2-dichloroethenes in a model wetland. Environ SciTechnol 2008;42:7924–30.

Imfeld G, Braeckevelt M, Kuschk P, Richnow HH. Monitoring and assessing processes oforganic chemicals removal in constructed wetlands. Chemosphere 2009;74:349–62.

Kadlec RH, Martin DC, Tsao D. Constructedmarshes for control of chlorinated ethenes: an11-year study. Ecol Eng 2012;46:11–23.

Kaschl A, Vogt C, Uhlig S, Nijenhuis I, Weiss H, Kastner M, et al. Isotopic fractionationindicates anaerobic monochlorobenzene biodegradation. Environ Toxicol Chem2005;24:1315–24.

Kassenga GR, Pardue JH. Fate of TCE in constructed and natural wetland mesocosms.Wetlands and remediation II; 2002. p. 25–32.

Liang X, Howlett MR, Nelson JL, Grant G, Dworatzek S, Lacrampe-Couloume G, et al.Pathway-dependent isotope fractionation during aerobic and anaerobic degradationof monochlorobenzene and 1,2,4-trichlorobenzene. Environ Sci Technol 2011;45:8321–7.

Liang X, Devine CE, Nelson J, Sherwood Lollar B, Zinder S, Edwards EA. Anaerobic conver-sion of chlorobenzene and benzene to CH(4) and CO(2) in bioaugmentedmicrocosms.Environ Sci Technol 2013. http://dx.doi.org/10.1021/es3043092. [in press].

Lorah MM, Voytek MA. Degradation of 1,1,2,2-tetrachloro ethane and accumulation ofvinyl chloride in wetland sediment microcosms and in situ porewater: biogeochem-ical controls and associations with microbial communities. J Contam Hydrol 2004;70:117–45.

Lovley DR. Anaerobic benzene degradation. Biodegradation 2000;11:107–16.Martinez-Lavanchy PM, Dohrmann AB, Imfeld G, Trescher K, Tebbe CC, Richnow HH, et al.

Detection of monochlorobenzene metabolizing bacteria under anoxic conditions byDNA-stable isotope probing. Biodegradation 2011;22:973–82.

Matamoros V, Hijosa M, Bayona JM. Assessment of the pharmaceutical active compoundsremoval in wastewater treatment systems at enantiomeric level. Ibuprofen andnaproxen. Chemosphere 2009;75:200–5.

McMahon PB, Chapelle FH. Redox processes and water quality of selected principal aqui-fer systems. Ground Water 2008;46:259–71.

Müller D, Schlömann M, Reineke W. Maleylacetate reductases in chloroaromatic-degrading bacteria using the modified ortho pathway: comparison of catalytic prop-erties. J Bacteriol 1996;178:298–300.

Nijenhuis I, Stelzer N, Kästner M, Richnow HH. Sensitive detection of anaerobicmonochlorobenzene degradation using stable isotope tracers. Environ Sci Technol2007;41:3836–42.

Novak D, Plomb D, Allen W. Treating landfill-contaminated groundwater using surface-flow wetlands. Wetlands and remediation II; 2002. p. 161–8.

Rakoczy J, Remy B, Vogt C, Richnow HH. A bench-scale constructed wetland as a model tocharacterize benzene biodegradation processes in freshwater wetlands. Environ SciTechnol 2011;45:10036–44.

RCoreTeam. R: a language and environment for statistical computing. R Foundation forStatistical Computing; 2013.

Richnow HH, Eschenbach A, Mahro B, Kästner M, Annweiler E, Seifert R, et al. Formationof nonextractable soil residues: a stable isotope approach. Environ Sci Technol1999;33:3761–7.

Seeger EM, Reiche N, Kuschk P, Borsdorf H, Kaestner M. Performance evaluation using athree compartment mass balance for the removal of volatile organic compounds inpilot scale constructed wetlands. Environ Sci Technol 2011;45:8467–74.

Stelzer N, Imfeld G, Thullner M, Lehmann J, Poser A, Richnow HH, et al. Integrativeapproach to delineate natural attenuation of chlorinated benzenes in anoxic aquifers.Environ Pollut 2009;157:1800–6.

Stern J, Wang Y, Gu B, Newman J. Distribution and turnover of carbon in natural andconstructed wetlands in the Florida Everglades. Appl Geochem 2007;22:1936–48.

van Agteren MH, Keuning S, Janssen DB. Handbook on biodegradation and biologicaltreatment of hazardous organic compounds. Dordrecht: Kluwer Academic Publisher;1998.

Vogt C, Gödeke S, Treutler HC, Weiss H, Schirmer M, Richnow HH. Benzene oxidationunder sulfate-reducing conditions in columns simulating in situ conditions. Biodegra-dation 2007;18:625–36.

Weiss JV, Emerson D, Backer SM, Megonigal JP. Enumeration of Fe(II)-oxidizing andFe(III)-reducing bacteria in the root zone of wetland plants: implications for a rhizo-sphere iron cycle. Biogeochemistry 2003;64:77–96.

Wycisk P, Weiss H, Kaschl A, Heidrich S, Sommerwerk K. Groundwater pollution andremediation options for multi-source contaminated aquifers (Bitterfeld/Wolfen,Germany). Toxicol Lett 2003;140–141:343–51.