at SciVerse ScienceDirect

Environmental Pollution 178 (2013) 254e263

Contents lists available

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

The impact of oscillating redox conditions: Arsenic immobilisation incontaminated calcareous floodplain soils

Christopher T. Parsons a,b,*, Raoul-Marie Couture b,3, Enoma O. Omoregie c,1,4,Fabrizio Bardelli a,5, Jean-Marc Greneche d,6, Gabriela Roman-Ross e,2,7, Laurent Charlet a,8

a Environmental Geochemistry Group, ISTerre, University of Grenoble I, B. P. 53, 38041 Grenoble, FrancebDepartment of Earth and Environmental Sciences, University of Waterloo, 200 University Ave. W, Waterloo, ON N2L 3G1, Canadac School of Earth, Atmospheric and Environmental Sciences, and Williamson Research Centre for Molecular Environmental Science,The University of Manchester, Manchester UKM13 9PL, UKd Laboratoire de Physique de L’Etat Condensé, UMR CNRS 6087, Institut de Recherche en Ingénierie Moléculaire et Matériaux Fonctionnels IRIM2F,FR CNRS 2575, Université du Maine 72085, Le Mans Cedex 9, FranceeDepartment of Chemistry, Faculty of Sciences, University of Girona, Campus de Montilivi, 17071 Girona, Spain

a r t i c l e i n f o

Article history:Received 14 August 2012Received in revised form29 January 2013Accepted 27 February 2013

Keywords:Redox cyclingRedox oscillationFloodingArsenic mobilityFloodplainsWater fluctuating zone (WFZ)

* Corresponding author.E-mail addresses: chris.parsons85@gmail.com (C

uwaterloo.ca (R.-M. Couture), omoregie@cab.intfabrizio.bardelli@gmail.com (F. Bardelli), jean-m(J.-M. Greneche), gabriela.roman@amphos21.com (Ggmail.com (L. Charlet).

1 Current address: Centro de Astrobiología, Instituespacial, Carretera de Ajalvir Km 4, Torrejón de Ardo

2 Current address: Amphos 21 consulting S.L. Passe1a, E08019 Barcelona, Spain.

3 Tel.: þ1 5198884567x31321.4 Tel.: þ34 915206461.5 Tel.: þ33 476635198.6 Tel.: þ33 243833301.7 Tel.: þ34 935830500.8 Tel.: þ33 476635198.

0269-7491/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.envpol.2013.02.028

a b s t r a c t

Arsenic contamination of floodplain soils is extensive and additional fresh arsenic inputs to the pedo-sphere from human activities are ongoing.

We investigate the cumulative effects of repetitive soil redox cycles, which occur naturally duringflooding and draining, on a calcareous fluvisol, the native microbial community and arsenic mobilityfollowing a simulated contamination event.

We show through bioreactor experiments, spectroscopic techniques and modelling that repetitiveredox cycling can decrease arsenic mobility during reducing conditions by up to 45%. Phylogenetic andfunctional analyses of the microbial community indicate that iron cycling is a key driver of observedchanges to solution chemistry. We discuss probable mechanisms responsible for the arsenic immobili-sation observed in-situ. The proposed mechanisms include, decreased heterotrophic iron reduction dueto the depletion of labile particulate organic matter (POM), increases to the proportion of co-precipitatedvs. aqueous or sorbed arsenic with a-FeOOH/Fe(OH)3 and potential precipitation of amorphous ferricarsenate.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Arsenic is an infamous carcinogen (WHO IARC, 2004), ubiqui-tous in the environment and subject to a variety of mobility altering

.T. Parsons), raoul.couture@a-csci.es (E.O. Omoregie),arc.greneche@univ-lemans.fr. Roman-Ross), charlet38@

to Nacional de Técnica Aero-z, 28850 Madrid, Spain.ig de Garcia i Faria, 49-51, 1�-

All rights reserved.

processes induced by redox changes which occur in temporallyflooded soils. The toxicity and mobility of arsenic are dependent onoxidation state and chemical speciation (Bissen and Frimmel,2003), therefore understanding the effects of repetitive floodingis essential. This is emphasised by recent work showing that hy-drological management can impact arsenic mobility in shallowalluvial aquifers (Benner, 2010; Neumann et al., 2010).

Floodplains are used extensively for agriculture (Verhoeven andSetter, 2010), in Europe 79% of riparian area is intensively cultivated(Tockner and Stanford, 2002) despite frequently hosting elevatedconcentrations of arsenic (Du Laing et al., 2009a; Overesch et al.,2007). Common sources of arsenic include mine-effluent, pesti-cides and poultry waste (Smedley and Kinniburgh, 2002), althoughdisperse contamination may be geogenic (Winkel et al., 2008) ordue to atmospheric deposition (Couture et al., 2008). Tracingarsenic origin on floodplains is problematic due to its mono-isotopic nature, lack of durable chemical source signatures anddiversewatershed land-use. Soils in riparian zones frequently act assinks for river-borne contaminants due to their fine particle size

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263 255

and hence high surface area (Lair et al., 2009) but can also act ascontaminant sources due to remobilisation (Roberts et al., 2010).Re-mobilised contaminants threaten human health through accu-mulation in crops (Meharg and Rahman, 2003) and contaminationof shallow alluvial aquifers (Ahmed et al., 2004). The currentmaximum recommended concentration for As in drinking water is10 mg/L (WHO IARC, 2004) which may be exceeded by remobili-sation from sediments (Smedley and Kinniburgh, 2002).

Significant attention has been paid to mechanisms responsiblefor arsenic mobility in soils during reducing conditions (Islam et al.,2004; McGeehan and Naylor, 1994) and to hydrological transportprocesses determining arsenic fluxes from sediments (Mukherjeeet al., 2008; Nath et al., 2009). Additionally recent field studieshave advanced our understandingof the combinedhydrological andbiogeochemical processes affecting arsenic mobility in nature (DuLaing et al., 2009a; Neumann et al., 2010). However, there is ascarcity of experimental studies accurately isolating biogeochemicalprocesses during the oscillating redox conditions experienced infloodplains, paddy fields, shallow aquifers and other water fluctu-ating zones (WFZ) (Kögel-Knabner et al., 2010; Stucki, 2011).Experimental studies until nowalso focused on acidic soils and havenot considered carbonate buffered systems (Frohne et al., 2011;Thompson et al., 2006b). It is unclear as to whether cumulative,inter-cycle changes occur over time due to oscillating redox condi-tions orwhether such conditions result in a systemwhich alternatesbetween oxic and reduced end-points. The objectives of this studyare to determine the mechanisms controlling arsenic mobility in acarbonate-buffered soil following arsenic contamination, and toimprove our understanding of redox-oscillating environments. Tosimulate redox cycles experienced by floodplain soils, bioreactorexperiments were conducted on the top-horizon (0e15 cm) of anarsenic-doped calcic fluvisol prone to phreatic and fluvial inunda-tion. Changes within the bioreactors were monitored throughoutthe experiment using an array of biogeochemical tools. The sam-pling location, flooding modes and extent are illustrated in Fig. 1.

Aqueous chemistry (cations, anions, dissolved organic and inor-ganic carbon (DOC and DIC)), mineralogy (powder X-ray diffraction(XRD)), solid arsenic and iron speciation (X-ray absorption spec-troscopy (XAS) and 57Fe Mössbauer spectrometry) were monitoredin addition to changes in the bacterial community (16S rRNA).Thermodynamic andkinetic geochemicalmodelling implemented inPHREEQC (Parkhurst et al., 1999) aided interpretation. The model isused as a diagnostic tool to interpret themeasured temporal changesand as a prognostic tool to determine the feasibility of potentialmechanisms controlling observed changes in As mobility.

2. Materials and methods

2.1. Field site characterisation

2.1.1. Soil samplingThe soil used in this study was sampled from the eastern floodplain of the Saône

River between Macon and Pont-de-Vaux in Ain, France. The land is used for pastureand corn production. Soil was sampled from the top-horizon (0e15 cm) of a mollic-fluvisol in a minor irrigation channel (46.373107�N, 4.879856�E). Established soilsampling protocols were observed (U.S. EPA, 2000).

2.1.2. Soil characterisationPrior to preliminary and bioreactor experiments the soil was characterised.

Elemental composition was determined by total acidic dissolution(HNO3 þ HF þ H2O2, H3BO3 þ HF)(U.S. EPA, 1996) and ICP-MS analysis (Agilent7500ce, Agilent Technologies, France). The particle size distributionwas analysed bylaser granulometry (Malvern Mastersizer 2000, Malvern instruments, France) andthe bulk mineralogy by XRD (full methodology described in 3.5).

2.1.3. Preliminary experiment: determination of natural redox oscillations within thestudy area

To aid experimental design of bioreactor experiments, the extent of redoxfluctuation occurring naturally in the soil during flooding was determined in a

preliminary flooding experiment. A passive diffusion pore-water sampler wasdeployed (Hesslein, 1976) in the soil, within a polycarbonate box which was sub-sequently flooded for 30 days. This setup has been previously described by Guedronet al. (2011). The pH, Eh, anions, cations and Fe speciation (Stookey,1970) in resultingpore water were analysed. A description of the sampling device and analyses isprovided in the Supplementary Information (Section 1.2.1).

2.2. Main experimental design and redox oscillation procedure

A bioreactor system, based on designs by Thompson et al. (2006b) was filledwith 1 L of soil suspension (<1mm fraction,100 g L�1) equilibrated for 1monthwith12 mM of arsenic (sodium-arsenate) in order to simulate a severe point sourcecontamination event. The suspension was subjected to multiple cycles of reductionand oxidation to determine the cumulative effects of redox-cycling on arsenicmobility. Eh variation was induced by modulation between sparging of N2 (7 days)and air (7 days). A total of 5.5, 14 day cycles over a period of 77 days were conductedat constant temperature (30 �C) with sampling on days 1, 4 and 7 of each half-cycle.Each 14 day cycle effectively served a reproducible replicate of the previous cycleand therefore full replicate experiments were not conducted.

2.3. Aqueous chemistry analyses

All chemicals were analytical grade from Fluka, SigmaeAldrich or Merck.Standards and reagents were prepared with 18 MU cm�1 water (Millipore). Syringe-sampled soil suspensions were centrifuged and the supernatant filtered to 0.22 mmprior to all aqueous analysis. Analysis of total Na, K, Ca, Mn, Fe and As concentrationsin the aqueous phase was performed with ICP-OES after dilution and acidification,using a Perkin Elmer OPTIMA 300 DV (Perkin Elmer, France). Matrix-matchedstandards were used for all calibrations and NIST validated multi-elemental solu-tions were used as internal controls. DOC/DIC concentrations were determinedusing a Shimadzu TOC-5000 (Shimadzu, France), all glassware was burned at 400 �Cfor 4 h before use. Chloride, Nitrate and Sulphate were analysed by ion chroma-tography using a Metrohm 761 Compact IC. Eh and pH were recorded every 30 swithin the reactors using Xerolyt Solid polymer open-junction electrodes. Allaqueous analyses were conducted in triplicate and the error for all techniques was<5%. A more detailed description of the sampling and analytical procedures isprovided in Supplementary Information.

2.4. Microbial community analysis

To monitor changes in the composition of the bacterial community during theexperiment additional suspension samples were taken on days 7 (reducing), 67(oxic) and 77 (reducing). Two-gram suspension sub-samples, concentrated bycentrifugation, were used for DNA and RNA extractions. Bacterial 16S rRNA sequencelibraries were generated from extracted DNA and RNA in order to characterise themicrobial community and identify organisms which were metabolically active. Thefull extraction, amplification and sequencing procedures are described inSupplementary Information (Section 1.3.1). Subsequent phylogenetic analysis wasconducted using the ARB software packagewith the Silva 98 database (Ludwig et al.,2004; Pruesse et al., 2007). Sequences generated have been deposited in the Gen-bank database (accession numbers: JQ976475-JQ976602) (Fig. 2).

2.5. Powder X-ray diffraction analysis

Powder XRD analysis of mineralogy was conducted on days 1 and 77. Less than2 mm, 2 mm and 0.2 mm fractions were analysed using a Bruker D5000 equippedwith a Kevex Si(Li) solid detector and a Cu Ka1 þ 2 radiation source. Larger fractionswerewet ground in a centrifugal mill. Intensities were recorded at 25 �C over a rangeof 2e80� 2qwith a step interval of 0.02� 2q and a counting time of 3 s per step. Full-widths at half-maximum intensity (fwhm) were determined for diffraction maximausing the EVA program (Bruker).

2.6. 57Fe Mössbauer spectroscopy

Mössbauer spectrometry was performed on solid fractions (obtained bycentrifugation) sampled on days 4, 14, 70 and 77. Solids were deposited on poly-carbonate holders which were capped and sealed with epoxy resin. Samples wereplaced in an anoxic container (N2 atmosphere) before transport for Mössbaueranalysis to a cryostat where the sample was maintained under He atmosphere.

The Mössbauer spectra were recorded at 77 K using a constant accelerationspectrometer and a 57Co source diffused into a rhodiummatrix. Velocity calibrationswere conducted using a-Fe foil at room temperature (RT, 295 K). Mössbauer spectrawere described using the Mosfit model (MOSFIT: Teillet and Varret unpublishedprogram) using a combination quadrupolar doublets and magnetic sextets. Theheight, shift and hyperfine field parameters of these components were thencompared to reference values in order to determine structural and speciation in-formation (Fig. 5). The proportion of each iron phase may also be determined fromthe relative sorption area.

Fig. 1. Characterisation of flooding TOP: Map of the extent of the Saône floodplain showing the sampling location (yellow star), towns (red circles), topography (METI/ERSDACet al., 2009) and the hydrological network (Institut Geographique National de France, 2010) (in blue). Coordinates are in decimal degrees based on the WGS84 geoid. BOTTOM: Anidealised cross-section of the floodplain at the sampling location illustrating the flooding modes and extent predicted to occur yearly (long dashed line) and every 100 years (shortdashed line). INSET: Bar chart comparison of pore water chemistry before and after 30 days of laboratory flooding of soils from the sampling location. (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version of this article.)

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263256

2.7. X-ray absorption spectroscopy

XAS measurements were performed at the As K-edge (11,867 eV) at the GILDA(BM-08) beamline (Pascarelli et al., 1996) of the European Synchrotron RadiationFacility (Grenoble, France).

Beam-line optics are described in Supplementary Information. In order toreduce the thermal contribution to the Debye-Waller factors and to preventbeam-induced redox reactions, all samples were measured in high vacuum

(w10�5 mbars) at 77 K. Local structure around As was obtained by quantitativerefinements of the EXAFS signal. Fits were performed in the back-transformedreciprocal space (k) in the range 4e12 �1, using software (Monesi et al., 2005)based on MINUIT routines from CERN libraries (James, 1975). The ATOMS (Ravel,2001) package was used to generate the atomic clusters centred on the absorberatom, which were used as reference structures for calculating theoreticalamplitude and phase back-scattering functions with the FEFF8 package(Ankudinov et al., 1998).

Fig. 2. Function of the microbial community LEFT: Table showing putative roles for bacteria identified by 16S rRNA sequence analysis at different times during this study based onclosest cultivated isolate. RIGHT: Stack plot comparing the relative dominance of identified clones in reducing and oxidising half-cycles.

Fig. 3. Aqueous chemistry and results of modelling measured (black) and modelled (red) DOC, Eh, pH, Fe(s) (grams of reducible goethite and ferrihydrite), Fe(aq) and As(aq) datawith time during reactor experiments. Sampling points for XANES, 57Fe Mössbauer and microbial community analysis are shown on the Eh curve (XANES/Mössbauer ¼ open bluesquares, 16S rRNA ¼ open green circles). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263 257

A

B

Fig. 5. 57Fe Mössbauer spectra and interpretation. A: An example Mössbauer spectraobtained on day 77 at T ¼ 77 K shown together with modelled hyperfine contributions.Green ¼ averaged sextuplet contribution corresponding to Fe(III) present in ferrichydroxides, Red ¼ structural Fe(III) present in clays, Blue ¼ averaged doublet contri-bution corresponding to structural Fe(II) present in clays. B: Hyperfine parameters ofmodelled contributions to soil 57Fe Mössbauer spectra compared to literature valuesfor various iron mineral components. Green squares correspond to modelled sextupletcomponents, Red circles and blue triangles correspond to modelled doublet compo-nents. Grey diamonds correspond to literature values of a-FeOOH, grey trianglescorrespond to structural Fe(III) in clays and grey circles correspond to structural Fe(II)in clays. (For interpretation of the references to colour in this figure legend, the readeris referred to the web version of this article.)

Fig. 4. Thermodynamic predictions of arsenic mineralogy: Pourbaix diagram of theFe, As, CO2 and H2O system in the reactor suspension with arsenic as the principalspecies.

PAs and

PFe ¼ 10�5 mol kg�1,

PC ¼ 10�4 mol kg�1, 303 K, 105 Pa. The ferric

arsenate stability fields are constructed with log Ksp values of 24.86 (dashed) and 26.62(solid) which correspond to the range of values provided in Langmuir et al. (2006) forferric arsenates in the presence of ferric oxides in natural sediments. Point data aremeasured Eh and pH couples from one full cycle.

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263258

3. Results and discussion

3.1. Soil characterisation and preliminary flooding experiments

Total acidic dissolution and ICP-MS analysis showed that the soilcontained amoderate arsenic background concentration and a highiron concentration (29.9ppm and 8.9% respectively), which is oftenobserved in mollic fluvisols (Du Laing et al., 2009b). Laser gran-ulometric analysis revealed that the soil was dominated by silt andclay (24% coarse silt (31e63 mm), 25% fine silt (3.9e31 mm) and 34%clay (1e3.9 mm)). XRD analysis showed that the main crystallinephases were quartz, calcite, illite and chlorite. Total organic mattercontent was 33.4 g kg�1 from loss of ignition. These characteristicsare indicative of a low energy deposition environment consistentwith flooding at low flow rate. Soil pH was 7.8 and was buffered bycalcite (total CaCO3 was 148 g kg�1). FromMössbauer spectra it wasshown that approximately 30% of total iron is present as FeOOH,50% as structural Fe(III) in clays and 20% as structural Fe(II) in clays(Fig. 5). This demonstrates the presence of a large pool of labileFe(III).

During the preliminary flooding experiment, described in 2.1.3,an Eh change from þ500 to �310 mV occurred, crossing thetheoretical MnO2/Mn2þ, NO3

�/NH4þ, HAsO4

2�/H3AsO3, FeOOH/Fe2þ

and SO42�/HS� thermodynamic equilibria (James and Bartlett, 1999),

indicative of major changes to aqueous chemistry (Fig. 1). A con-current drop in pH from 7.8 to 6.9 suggests that in this soil accu-mulation of CO2 in pore-water due to respiration exerts a greaterinfluence on pH than the production of hydroxide by reductiveprocesses. An increase in Fe2þ(aq), Mn(aq) and DOC concentrationswas also recorded, in addition to elimination of aqueous sulphate,indicative of heterotrophic iron, manganese and sulphate reduc-tion. The high final Fe2þ(aq) concentration (0.43 mM), despite sig-nificant probable sorption to clay minerals, indicates the dominantrole of iron reduction in this soil during flooding.

3.2. Active microbial community during redox-cycling

A broad genetic diversity was present in all investigated samplesfrom bioreactor oscillation experiments, due to the variety of en-ergy generating processes possible in such a dynamic environment.Community structure did not appear have been appreciably altereddue to arsenic addition prior to equilibration of the soil suspension.

Similarity between samples based on phylotype compositionwas low (<28%), regardless of the redox cycle. Distinct patterns in16S rRNA phylotypes based on presence or absence; as result ofalternating redox conditions were not observed. This is unsurpris-ing given that many bacteria are capable of operating under bothaerobic and anaerobic conditions (DeAngelis et al., 2010). Themetabolisms of many phylotypes detected in this study could notbe inferred, due to their low identity to properly described bacterialisolates. However, phylotypes related to known heterotrophicbacteria (93 of 127 sequences) were the largest group of bacteriadetected. Presumably these bacteria were responsible for oxidationof organic matter (OM). Phylotypes related to Fe-cycling bacteria(IRB), including Geobacter and Leptothrix were detected, whichdemonstrates the importance of ironeredox processes during theexperiment (Fig. 2). Geobacter sp. are known IRB (Lovley et al., 1993)and likely contributed to Fe(III)-reduction during reducing half-cycles. However, these phylotypes were also detected during oxi-dising half-cycles. Some Geobacter species have been shown to useoxygen as a terminal electron acceptor (TEA) (Lin et al., 2004). It ispossible that these organisms also contributed to OM oxidationduring the oxidising half-cycles demonstrating adaptation a redoxoscillating environment. Phylotypes related to Fe-oxidising bacte-rium Leptothrix sp. (Van Veen et al., 1978) were only detected

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263 259

during the oxidising half-cycles, indicating that microbial oxidationof Fe(II) only occurred during these periods. Pseudomonas relatedphylotypes were detected under both oxidising and reducing con-ditions. This group of bacteria contains metabolically diversegroups of heterotrophic bacteria (Moore et al., 2006) includingbacteria capable of the oxidation of iron (Straub et al., 1996) andsulfide (Mahmood et al., 2009) and reduction of arsenate(Freikowski et al., 2010). However, given the presence of manyheterotrophic bacterial species in both oxidising and reducingphases, it is likely that these organisms were responsible for theoxidation of OM during both phases.

3.3. Aqueous chemistry cycling and cumulative effects

As suggested by the microbial community composition, exten-sive intra-cycle changes to aqueous chemistry occurred within thesuspension during cycling. Measured geochemical parametersillustrate these changes (Fig. 3). The reductive processes, and henceEh decrease, were driven by the consumption/oxidation of DOCcoupled to the reduction of successive TEAs by the bacterial com-munity. This resulted in simultaneous OH� production and hencepH rise. The reintroduction of air simulated the initiation of oxi-dising conditions after a flooding event, whereby reduced speciesare oxidised by biotic and abiotic processes. Although there is ev-idence of Mn cycling in the investigated soil (Supplementary Figs. 3and 4), total Fe solid concentrations are approximately 30 timesgreater than total Mn concentrations. We propose that the micro-bial reduction of ferric iron, shown to be present as poorly crys-talline goethite and Fe(III)-rich clay minerals via 57Fe Mössbauerspectrometry (Fig. 5), is the dominant reductive process duringanoxic cycles and is likely to exert a greater control over arsenicmobility than other reduction processes.

Intra-cycle Eh changes are similar to those exhibited in fieldbased Eh-monitoring studies of flooded soils (Ponnamperuma,1972; Vorenhout et al., 2004) and to those observed in previousredox oscillating experiments (Thompson et al., 2006a, 2006b),crossing various redox boundaries (Figs. 3 and 4). DOC concentra-tion increases during each reducing half-cycle indicating that ratesof microbially mediated solubilisation of particulate organic matter(POM) and desorption exceeded rates of DOC consumption byheterotrophs. In previous redox-oscillation experiments, DOC wasmanually replenished by adding sucrose (Thompson et al., 2006a)at the beginning of each reducing half-cycle to provide constantinitial DOC. Here, the system is left to oscillate in order to observethe attenuation of labile organic carbon without external replen-ishment. The mechanisms responsible for the replenishment ofDOC are likely the hydrolysis of POM by bacterial enzymes (Vetteret al., 1998) and desorption of DOC frommineral surfaces due to pHincrease (Grybos et al., 2009). During subsequent oxidising half-cycles, bacterial consumption of DOC occurs at a greater rate thanit is replenished by hydrolysis, resulting in a decrease in DOCconcentration. The pH of this soil is buffered by CaCO3 and does notdecrease during oxic cycles as would be expected from the reactionstoichiometry. The amplitude of the oscillations in DOC concen-tration decreases during successive cycles as labile POM isexhausted by bacterial hydrolysis and the labile DOC pool is min-eralised. This cumulative decrease is representative of OM con-sumption in many managed floodplain soils which are subjected tosuccessive cycles of phreatic but not fluvial flooding, which limitsinput of organic rich sediment (Kirk, 2004). On floodplains the al-luvial aquifer and the river are strongly inter-dependent and oftenact as one hydrological unit. In these circumstances, as is the caseover much of the Saône floodplain, flood barriers are unable toprevent a raise in water levels on the floodplain and hence anoxiaduring sustained high river levels. Barriers do however prevent

overbank flooding and the deposition of fresh sediment includinglabile OM.

In such situations, where deposition of fresh sediment isrestricted, as the most labile OM is gradually depleted, itsdecreasing availability may eventually limit the rate of heterotro-phic metabolism (Neumann et al., 2010), leading to a decrease inreductive processes shown to be responsible for the mobilisation ofcontaminants such as iron-bound arsenic (Islam et al., 2004).Following depletion of labile OM heterotrophic reductive processeswill likely continue at a slower rate using less labile organic frac-tions as they are degraded by other parts of the microbialcommunity.

Pore-water Fe2þ concentrations, present following iron reduc-tion during the equilibration period decreased during the firstredox cycle and remained low throughout the experiment(<70 mM) despite microbial evidence indicative of iron reduction.Thermodynamic predictions suggest that low Fe2þ(aq) duringreducing half-cycles may be due to the formation of secondary ironminerals such as magnetite, which remains supersaturated withrespect to the reactor suspension at all times (SI� 9.23 calculated inPHREEQC). Although this prediction is supported by previousstudies of iron reducing bacteria where nano-magnetite is the finalproduct of reduction (Coker et al., 2008) neither Mössbauer spectranor XRD results provided evidence for the presence of magnetite asit is likely a minor component of the solid iron pool (seeSupplementary Information Section 1.4). Additionally complexa-tion of Fe2þ with organic matter has been shown inhibit precipi-tation altering solubility (Rose and Waite, 2003) and sorption ofFe2þ to a variety of mineral and solid organic surfaces has beenshown to be kinetically favourable compared to precipitation offerrous minerals at circumneutral to high pH (Charlet et al., 1998).Metal-hydr(oxides) and phyllosilicateminerals have been shown tooffer substantial sorbent surfaces for Fe2þ(aq) with phyllosilicatesbecoming increasingly important at high pH (Jaisi et al., 2008). TheFe-rich clays (illite and chlorite) identified by XRD analysis consti-tute a high proportion of the soil mineral content in this and manyother floodplain soils. Precipitation of ferrous minerals such asFeCO3(s) has been shown to limit arsenic mobility by providing asuitable surface for sorption (Charlet et al., 2011). Therefore fastsorption of Fe2þ to redox stable minerals such as recalcitrant metaloxides and phyllosilicates and Fe complexation with organics mayincrease arsenic mobility during reducing conditions by restrictingprecipitation of ferrous carbonate or sulfide minerals.

Intra-cycle mobilisation of arsenic was observed during eachreducing half-cycle (up to 335 mM), however, upon re-establishment of oxidising conditions As(aq) was re-immobilised,returning to a base-level of approximately 36 mM. Similar varia-tions in arsenic concentration have previously been documented inthe field in response to variable groundwater level resulting invariable redox conditions (Du Laing et al., 2009a). As can be seen inFig. 4, during reducing half-cycles the most thermodynamicallyfavourable arsenic species is arsenite (H3AsO3 and H2AsO3

�)whereas the conditions established during oxidising half cyclesfavour the formation of arsenate (HAsO4

2�) and potentially precip-itation of amorphous ferric arsenate. Arsenite is often considered tobe more mobile than arsenate (Borch et al., 2010) due its neutralcharge (first pKa at 9.2). This is supported by redox oscillating ex-periments at low pH demonstrating increased mobility of arsenite(Frohne et al., 2011) However, at circumneutral to high pH, soilsorption capacities for arsenite are often greater than for arsenate(Dixit and Hering, 2003). Transition from arsenate to arseniteduring reducing half-cycles is therefore unlikely to be the mainmechanism accounting for mobilisation of arsenic in calcareoussoils at pH w8. Alternative mechanisms shown to release arsenicduring reducing conditions include reductive dissolution of As-

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263260

bearing iron (hydr)oxide (Du Laing et al., 2009a; Erbs et al., 2010;Frohne et al., 2011) and competitive sorption with DOC (Bauer andBlodau, 2006), and it is these mechanisms which we consider mostinfluence measured changes in intra-cycle arsenic mobility. Therelatively high arsenic concentrations persisting during oxic cyclesmay be hypothetically explained by surface site saturation due tohigh arsenic concentrations.

In addition to intra-cycle changes, cumulative effects of redoxcycling on arsenic mobility in-situ are present, and have not to ourknowledge been previously described. Whilst As concentrationsduring oxidising half-cycles remained constant, successive redox-cycles resulted in a 45% decrease in mobility between the firstand fifth reducing half-cycles.

In order to evaluate the feasibility of mechanisms for inter andintra-cycle arsenic immobilisation a thermodynamic and kineticmodel of redox cycling in the bioreactor was developed.

3.4. Thermodynamic and kinetic modelling

The model, implemented in PHREEQC (Parkhurst et al., 1999),simulates multiple cycles of biologically mediated oxidation andreduction in a homogeneous soil system containing OM, an activemicrobial community and representative mineralogy. The code al-lows for both thermodynamic equilibria and kinetic reactions. To ourknowledge this is the first such model to incorporate redox oscil-lating conditions andheterotrophic respiration rather than one-timereduction/oxidation events. Although it is probable that multipleTEAs including Mn(VI), nitrate, arsenate and sulphate were used forenergy generation in the bioreactors only ferric iron and O2 areimplemented in the model due to evidence from 16S rRNA, aqueouschemistry and Mössbauer analyses indicating that they were themost abundant and utilised TEAs in the system. As a first step ininvestigating the mechanisms responsible for As immobilisation wecalculated the saturation index (SI ¼ log IAP/Ksp) of the aqueousphase at each sampling point with respect to solid arsenates. Thiscalculation indicates that the aqueous phase was undersaturatedwith respect to major arsenate phases throughout the experiment(SI��5) with the exception of ferric arsenate minerals (amorphousferric arsenate and scorodite) which oscillated between supersatu-ration and undersaturation across cycles. Ferric arsenates arecommonly associated with low pH environments but recent studiesdemonstrate that precipitation and dissolution of amorphous ferricarsenate may occur relatively rapidly (days) even at neutral pH(Fujita et al., 2012;Harveyet al., 2006; Langmuir et al., 2006; Paktuncand Bruggeman, 2010). Ferric-arsenate phases are also often re-ported to be present in contaminated soils (Langmuir et al., 2006).Precipitation and dissolution of amorphous ferric arsenate is there-fore allowed in the model during the equilibration period prior tocycling and during redox cycles.

Table 1Rate controlled reactions, formulations and constants implemented in the PHREEQC cod

Description Reaction Kinet

Hydrolysis of particulateorganic matter

ðCH2OÞpart /RHydroðCH2OÞdis

RHydr

Respiration CH2Oþ O2/CO2 þ H2ORRes

Reduction of ferrihydrite 4FeOOHþ CH2Oþ 8Hþ/4Fe2þ þ CO2 þ 7H2O RFerri

Reduction of goethite 4FeðOHÞ3 þ CH2Oþ 8Hþ/4Fe2þ þ CO2 þ 11H2O RGoet

Slow sorption of arsenic AsðmobÞ/AsðimobÞ RASImRelease of slow sorbed

arsenicAsðimobÞ/AsðmobÞ ð½AS�

Oxidation of Fe2þ Fe2þ þ 0:25O2 þ Hþ/Fe3þ þ 0:5H2O RFeOx

The reactor suspension is modelled as a homogeneous solutionin thermodynamic equilibrium with calcite, illite, ferric-arsenate,goethite and the reactor head-space gas phase. Redox potential iscontrolled primarily by the Fe2þ/Fe3þ redox couple using a log kvalue of 13.6. In the model, pH is used as an input parameter andwas fixed at the beginning of each half cycle. This was implementedin order to overcome problems in estimating fluctuating CO2(aq)outgassing due to a combination of microbial metabolism, slowdiffusion of gas to the suspension’s headspace and the effect of thecarbonate buffering system. Further efforts are needed to fullycharacterise and accurately mode the dynamic of carbonates alongredox oscillations in high pH, calcite buffered substrates (Jourabchiet al., 2008).

Selected reactions controlling pore-water chemistry are rate-controlled and are managed in kinetic blocks each correspondingto 7 day half-cycles. Pore-water and solid phase chemistry arecarried onto each successive cycle using PHREEQC’s SAVE and USEcommands, while gas headspace composition is reset with alter-nating N2(g) or O2(g) at the beginning of each cycle. These reactionsand kinetic formulations are summarised in Table 1. TEA con-sumption is implemented via Monod kinetic formulation (Canavanet al., 2006) and degradation of solid organics by hydrolysis usingthe 1G model (Van Cappellen and Wang, 1995). For detailed de-scriptions of all reactions and for a full table of constants seeSupplementary Information (Section 1.5 and SupplementaryTable 5).

Intra-cycle oscillation of aqueous arsenic concentration iseffectively reproduced by the combination of kineticallycontrolled As sorption/desorption and the precipitation/dissolu-tion of amorphous ferric-arsenate following thermodynamicpredictions (Fig. 4). It has been shown that IRB are capable ofreducing iron within ferric arsenates leading to arsenate releaseinto solution (Cummings et al., 1999), therefore we accept thatthis process is also probably microbially mediated. With succes-sive cycles, and hence a decrease in DOC availability, iron reduc-tion rate decreases and hence less ferric-arsenate dissolves witheach cycle.

To fully capture the slow inter-cycle attenuation of As (R5) arate-controlled arsenic sorption parameter, analogous to co-precipitation (Couture et al., 2010) was implemented. Irreversibleincorporation of arsenic due to uptake during sorption has beenfrequently reported (e.g. Zhang and Selim, 2008). In addition, As co-precipitated with ferrihydrite is significantly less mobile thanadsorbed arsenic upon development of reducing conditions (Erbset al., 2010). Co-precipitated arsenic is released to solution whenthe precipitate is dissolved.

Whilst the model is a simplification of the many processesoccurring in natural soils during flooding, particularly with respectto organic matter degradation, it was able to reproduce with good

e.

ic formulation & constants ID

oðtÞ ¼ R0eð�atÞ R1

¼ Rakom½CH2O�

½O2�KmðO2Þ þ ½O2�

!R2

¼ kom½CH2O�

½FeðOHÞ3�KmðFerrÞ þ ½FeðOHÞ3�

!�Kin

Kin þ ½O2�� R3

hite ¼ kom½CH2O�

½FeOOH�KmðGoethiteÞ þ ½FeOOH�

! KinFeðOHÞ3

KinFeðOHÞ3 þ ½O2�

!�KinO2

Kino2 þ ½O2�� R4

¼ kimob ½As�ð½FeðOHÞ3� þ ½FeOOH�Þ R5: ½FeðOH3Þ� þ ½AS� : ½FeOOH�Þ � ð4� R3þ 4� R4Þ R6

¼ kFeOx½O2�½Fe2þ� R7

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263 261

accuracy all of the measured chemical parameters. This demon-strates that the arsenic attenuation measured during redox cyclingin batch experiments may be successfully described simply by acombination of co-precipitation with ferric-(hydr)oxides, decreasein ferric arsenate dissolution due to depletion of DOC and labilePOM and the kinetic uptake of As by ferric-oxide phases. This resultemphasises the importance of labile POM and DOC in arseniccontaminated soils supporting the recent findings of Neumannet al. (2010) and Benner (2010) and indicates the importance ofhydrological management on floodplains as a control on arsenicmobility. However, it should be noted that the adequate solutionoffered by combination of processes included in the model shouldnot be considered unique. Other processes absent from the model,particularly interactions with organic matter are likely to playimportant roles controlling arsenic mobility in the bioreactor sys-tem and in the natural environment.

3.5. Solid arsenic speciation

To investigate the effect of redox cycling on solid arsenicspeciation, X-ray absorption spectrawere recorded at the As K-edgeon samples taken at the end of the first and last oxidising andreducing cycles (Fig. 6). Linear Combination Fitting (LCF) of thesamples’ XANES spectra using As(III) and As(V) reference spectra,shows that all spectra demonstrate the presence of a mix of tri-valent and penta-valent arsenic species, with an increase in theproportion of As(III) to As(V) during reducing cycles. This isconsistent with the dissolution of As(V)-bearingminerals predicted

Fig. 6. XANES at the arsenic K-edge Arsenic K-edge (11,867 eV) XANES spectrarecorded at 77 K. Spectra from suspension solids are coloured according to time ofsampling: green ¼ end of reducing cycle, red ¼ end of oxidising cycle. As(III) and As(V)adsorbed on pure calcite were used as references (black and blue curves, respectively).An estimation of the As(III) and As(V) contributions to natural samples derived fromLinear combination fitting (LCF) is shown above each spectrum (the error is about 5%).Dashed lines are the linear combination fits. Detailed information on XAS measure-ments and structural parameters is reported in the Supplementary Information (Sec-tion 1.10.2). (For interpretation of the references to colour in this figure legend, thereader is referred to the web version of this article.)

by the model and of reduction of As(V) to As(III) either directly byarsenic reducing bacteria or indirectly due to chemical reduction.

EXAFS refinements revealed the average local structure aroundAs. In all samples two AseO bonds of differing length (w1.7 andw1.8 �A) were observed, compatible with a mixture of As(V) andAs(III) oxyanions. Features beyond the first coordination shell werenegligibly weak, indicative of a disordered structure. This is prob-ably due to incorporation of arsenic into poorly crystalline phasessuch as amorphous ferric oxyhydroxides and ferric arsenate. Thefull results of LCF, coordination numbers, bond lengths and fittedEXAFS spectra are provided in the Supplementary Information(Section 1.10.2).

Following successive cycles, the ratio of As(III)/As(V) remainedconstant during oxidising half-cycles, however, decreased consid-erably during reducing half-cycles. This is consistent with themodel’s prediction of arsenate accumulation in a poorly orderedferric arsenate phase as well as the slow uptake of As(V) duringsorption resulting in persistent arsenate in the solid phase. Itshould be noted that the persistence of arsenite under oxidisingconditions also indicates that oxidation processes were probablystrongly kinetically controlled.

4. Conclusions

In addition to confirming often observed changes to arsenicmobility under variable redox conditions (Frohne et al., 2011) weshow through wet chemistry experiments that repetitive cycling ofredox conditions in arsenic contaminated soils, results in cumula-tive changes to arsenic speciation and mobility. These resultsindicate that redox oscillation in natural environments due toflooding and draining has the potential to stabilise contaminants inthe solid phase and limit aqueous concentrations. Previous studieshave demonstrated that decreasing As concentrations in pore-water during cycles of flooding and draining are due to physicalremoval of arsenic in receding floodwater (Roberts et al., 2010). Inthis study, evidence from thermodynamic and kinetic modelling ofexperimental data, complemented by XAS and 57Fe Mössbauerspectrometry suggest that decreased mobility of arsenic by 45%during reducing conditions can be entirely attributed to changes tomineralogy, OM and microbial activity. The proposed mechanismsfor As mobility reduction are increased co-precipitation of surfacesorbed arsenic and the depletion of labile POM and DOC due to lackof solid organic matter recharge, resulting in decreased rates ofheterotrophic iron reduction and ferric arsenate dissolution. In thisstudy numerous parameters were allowed to vary concomitantly inorder to best represent changes occurring during natural redoxoscillations induced by flooding. Therefore we are unable todetermine from experimental evidence the relative importance ofeach of the proposed processes and mechanisms to reduce arsenicmobility. We accept that other processes, notably competition forsorption sites between aqueous arsenic species and dissolved or-ganics, probably also contributed to the measured reduction inarsenic mobility.

Since redox oscillating conditions are prevalent in a variety ofnear-surface arsenic contaminated environments, critical for foodproduction and groundwater abstraction, further research into thelong term effect of redox oscillations on the soil matrix andcontaminant mobility under redox oscillations is necessary.Although only arsenic was investigated in this study cumulativeeffects may also be present for other inorganic or organic con-taminants. We predict that flooding length and periodicity arelikely to play important roles in determining contaminant mobilityand cumulative trends in redox oscillating systems where manyprocesses are kinetically controlled. Periodicity and length of soilflooding inmanywatersheds is, to a certain extent, controllable due

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263262

to ubiquitous damming. We propose that a greater appreciation ofthe biogeochemical response of soils to varying flooding periodicitymay improve hydrological management of contaminated soilsallowing greater control over contaminant mobility.

Acknowledgements

We would like to thank Philippe Van Cappellen for his help andadvice during the construction of the conceptual model and duringthe writing and review process. We also gratefully acknowledgefinancial support from the AquaTRAIN Marie Curie ResearchTraining Network (Contract no. MRTN-CT-2006-035420) funded bythe European Commission Sixth Framework Programme (2002e2006).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.envpol.2013.02.028.

References

Ahmed, K.M., Bhattacharya, P., Hasan, M.A., Akhter, S.H., Alam, S.M.M.,Bhuyian, M.A.H., Imam, M.B., Khan, A.A., Sracek, O., 2004. Arsenic enrichment ingroundwater of the alluvial aquifers in Bangladesh: an overview. AppliedGeochemistry 19, 181e200.

Ankudinov, A.L., Rehr, J.J., Conradson, S.D., 1998. Real-space multiple-scatteringcalculation and interpretation of X-ray-absorption near-edge structure. Phys-ical Review B 58, 7565e7576.

Bauer, M., Blodau, C., 2006. Mobilization of arsenic by dissolved organic matter fromiron oxides, soils and sediments. Science of the Total Environment 354, 179e190.

Benner, S., 2010. Hydrology: anthropogenic arsenic. Nature Geoscience 3, 5e6.Bissen, M., Frimmel, F.H., 2003. Arsenic d a review. Part I: occurrence, toxicity,

speciation, mobility. Acta Hydrochimica et Hydrobiologica 31, 9e18.Borch, T., Kretzschmar, R., Kappler, A., Cappellen, P.V., Ginder-Vogel, M., Voegelin, A.,

Campbell, K., 2010. Biogeochemical redox processes and their impact oncontaminant dynamics. Environmental Science & Technology 44, 15e23.

Canavan, R.W., Slomp, C.P., Jourabchi, P., Van Cappellen, P., Laverman, A.M., Van denBerg, G.A., 2006. Organic matter mineralization in sediment of a coastalfreshwater lake and response to salinization. Geochimica et Cosmochimica Acta70, 2836e2855.

Charlet, L., Morin, G., Rose, J., Wang, Y., Auffan, M., Burnol, A., Fernandez-Martinez, A., 2011. Reactivity at (nano)particle-water interfaces, redox pro-cesses, and arsenic transport in the environment. Comptes Rendus Geoscience343, 123e139.

Charlet, L., Silvester, E., Liger, E., 1998. N-compound reduction and actinideimmobilisation in surficial fluids by Fe(II): the surface [6-point triple bond;length half of m-dash]FeIIIOFeIIOHo species, as major reductant. ChemicalGeology 151, 85e93.

Coker, V.S., Bell, A.M.T., Pearce, C.I., Pattrick, R.A.D., Van der Laan, G., Lloyd, J.R.,2008. Time-resolved synchrotron powder X-ray diffraction study of magnetiteformation by the Fe(III)-reducing bacterium geobacter sulfurreducens. Amer-ican Mineralogist 93, 540e547.

Couture, R.-M., Gobeil, C., Tessier, A., 2008. Chronology of atmospheric deposition ofarsenic inferred from reconstructed sedimentary records. Environmental Sci-ence & Technology 42, 6508e6513.

Couture, R.-M., Gobeil, C., Tessier, A., 2010. Arsenic, iron and sulfur co-diagenesis inlake sediments. Geochimica et Cosmochimica Acta 74, 1238e1255.

Cummings, D.E., Caccavo, Fendorf, S., Rosenzweig, R.F., 1999. Arsenic mobilizationby the dissimilatory Fe(III)-reducing bacterium Shewanella alga BrY. Environ-mental Science & Technology 33, 723e729.

DeAngelis, K.M., Silver, W.L., Thompson, A.W., Firestone, M.K., 2010. Microbialcommunities acclimate to recurring changes in soil redox potential status.Environmental Microbiology 12, 3137e3149.

Dixit, S., Hering, J.G., 2003. Comparison of arsenic(V) and arsenic(III) sorption ontoiron oxide minerals: implications for arsenic mobility. Environmental Science &Technology 37, 4182e4189.

Du Laing, G., Chapagain, S.K., Dewispelaere, M., Meers, E., Kazama, F., Tack, F.M.G.,Rinklebe, J., Verloo, M.G., 2009a. Presence and mobility of arsenic in estuarinewetland soils of the Scheldt estuary (Belgium). Journal of EnvironmentalMonitoring 11, 873.

Du Laing, G., Rinklebe, J., Vandecasteele, B., Meers, E., Tack, F.M.G., 2009b. Tracemetal behaviour in estuarine and riverine floodplain soils and sediments: areview. Science of the Total Environment 407, 3972e3985.

Erbs, J.J., Berquo, T.S., Reinsch, B.C., Lowry, G.V., Banerjee, S.K., Penn, R.L., 2010.Reductive dissolution of arsenic-bearing ferrihydrite. Geochimica et Cosmo-chimica Acta 74, 3382e3395.

Freikowski, D., Winter, J., Gallert, C., 2010. Hydrogen formation by an arsenate-reducing Pseudomonas putida, isolated from arsenic-contaminated groundwaterin West Bengal, India. Applied Microbiology and Biotechnology 88, 1363e1371.

Frohne, T., Rinklebe, J., Diaz-Bone, R.A., Du Laing, G., 2011. Controlled variation ofredox conditions in a floodplain soil: impact on metal mobilization and bio-methylation of arsenic and antimony. Geoderma 160, 414e424.

Fujita, T., Fujieda, S., Shinoda, K., Suzuki, S., 2012. Environmental leaching charac-teristics of scorodite synthesized with Fe(II) ions. Hydrometallurgy 111-112,87e102.

Grybos, M., Davranche, M., Gruau, G., Petitjean, P., Pedrot, M., 2009. Increasing pHdrives organic matter solubilization from wetland soils under reducing condi-tions. Geoderma 154, 13e19.

Guedron, S., Cossa, D., Grimaldi, M., Charlet, L., 2011. Methylmercury in tailingsponds of Amazonian gold mines (French Guiana): field observations and anexperimental flocculation method for in situ remediation. Applied Geochem-istry 26, 222e229.

Harvey, M.C., Schreiber, M.E., Rimstidt, J.D., Griffith, M.M., 2006. Scorodite disso-lution kinetics: implications for arsenic release. Environmental Science &Technology 40, 6709e6714.

Hesslein, R.H., 1976. An in situ sampler for close interval pore water studies.Limnology and Oceanography 21, 912e914.

Institut Geographique National de France, 2010. BD Carthage e Réseau hydro-graphique français.

Islam, F.S., Gault, A.G., Boothman, C., Polya, D.A., Charnock, J.M., Chatterjee, D.,Lloyd, J.R., 2004. Role of metal-reducing bacteria in arsenic release from Bengaldelta sediments. Nature 430, 68e71.

Jaisi, D.P., Liu, C., Dong, H., Blake, R.E., Fein, J.B., 2008. Fe2þ sorption onto nontronite(NAu-2). Geochimica et Cosmochimica Acta 72, 5361e5371.

James, B., Bartlett, R., 1999. Redox phenomena. In: Sumner, M. (Ed.), Handbook ofSoil Science. CRC Press, Boca Raton, FL, pp. B169eB194.

James, F., 1975. Minuit e a system for function minimization and analysis of theparameter errors and correlations. Computer Physics Communications 10, 343e367.

Jourabchi, P., Meile, C., Pasion, L.R., Van Cappellen, P., 2008. Quantitative interpre-tation of pore water O2 and pH distributions in deep-sea sediments. Geo-chimica et Cosmochimica Acta 72, 1350e1364.

Kirk, G., 2004. The Biogeochemistry of Submerged Soils. Wiley, Chichester; Hobo-ken N.J.

Kögel-Knabner, I., Amelung, W., Cao, Z., Fiedler, S., Frenzel, P., Jahn, R., Kalbitz, K.,Kölbl, A., Schloter, M., 2010. Biogeochemistry of paddy soils. Geoderma 157,1e14.

Lair, G.J., Zehetner, F., Fiebig, M., Gerzabek, M.H., Van Gestel, C.A.M., Hein, T.,Hohensinner, S., Hsu, P., Jones, K.C., Jordan, G., Koelmans, A.A., Poot, A.,Slijkerman, D.M.E., Totsche, K.U., Bondar-Kunze, E., Barth, J.A.C., 2009. How dolong-term development and periodical changes of riverefloodplain systemsaffect the fate of contaminants? Results from European rivers. EnvironmentalPollution 157, 3336e3346.

Langmuir, D., Mahoney, J., Rowson, J., 2006. Solubility products of amorphous ferricarsenate and crystalline scorodite (FeAsO4$2H2O) and their application toarsenic behavior in buried mine tailings. Geochimica et Cosmochimica Acta 70,2942e2956.

Lin, W.C., Coppi, M.V., Lovley, D.R., 2004. Geobacter sulfurreducens can grow withoxygen as a terminal electron acceptor. Applied and Environmental Microbi-ology 70, 2525e2528.

Lovley, D.R., Giovannoni, S.J., White, D.C., Champine, J.E., Phillips, E.J.P., Gorby, Y.A.,Goodwin, S., 1993. Geobacter metallireducens gen. nov. sp. nov., a microor-ganism capable of coupling the complete oxidation of organic compounds tothe reduction of iron and other metals. Archives of Microbiology 159, 336e344.

Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H., Yadhukumar, Buchner, A.,Lai, T., Steppi, S., Jobb, G., Förster, W., Brettske, I., Gerber, S., Ginhart, A.W.,Gross, O., Grumann, S., Hermann, S., Jost, R., König, A., Liss, T., Lüßmann, R.,May, M., Nonhoff, B., Reichel, B., Strehlow, R., Stamatakis, A., Stuckmann, N.,Vilbig, A., Lenke, M., Ludwig, T., Bode, A., Schleifer, K., 2004. ARB: a softwareenvironment for sequence data. Nucleic Acids Research 32, 1363e1371.

Mahmood, Q., Zheng, P., Hu, B., Jilani, G., Azim, M.R., Wu, D., Liu, D., 2009. Isolationand characterization of Pseudomonas stutzeri QZ1 from an anoxic sulfide-oxidizing bioreactor. Anaerobe 15, 108e115.

McGeehan, S.L., Naylor, D.V.,1994. Sorption and redox transformation of arsenite andarsenate in twoflooded soils. Soil Science SocietyofAmerica Journal 58, 337e342.

Meharg, A.A., Rahman, M.M., 2003. Arsenic contamination of Bangladesh paddyfield soils: implications for rice contribution to arsenic consumption. Envi-ronmental Science & Technology 37, 229e234.

METI/ERSDAC, NASA/LPDAAC, USGS/EROS, 2009. ASTER Global DEM Validation(Summary Report for the 30 m World-coverage Dataset).

Monesi, C., Meneghini, C., Bardelli, F., Benfatto, M., Mobilio, S., Manju, U., Sarma, D.,2005. Local structure in LaMnO3 and CaMnO3 perovskites: a quantitativestructural refinement of Mn K-edge XANES data. Physical Review B 72.

Moore, E., Tindall, B., Santos, V., Pieper, D., Ramos, J.-L., Palleroni, N., 2006. Nonmed-ical: pseudomonas. In: The Prokaryotes. Springer, New York, pp. 646e703.

Mukherjee, A., Bhattacharya, P., Savage, K., Foster, A., Bundschuh, J., 2008. Distri-bution of geogenic arsenic in hydrologic systems: controls and challenges.Journal of Contaminant Hydrology 99, 1e7.

Nath, B., Chakraborty, S., Burnol, A., Stuben, D., Chatterjee, D., Charlet, L., 2009.Mobility of arsenic in the sub-surface environment: an integrated hydro-geochemical study and sorption model of the sandy aquifer materials. Journal ofHydrology 364, 236e248.

C.T. Parsons et al. / Environmental Pollution 178 (2013) 254e263 263

Neumann, R.B., Ashfaque, K.N., Badruzzaman, A.B.M., Ashraf Ali, M., Shoemaker, J.K.,Harvey, C.F., 2010. Anthropogenic influences on groundwater arsenic concen-trations in Bangladesh. Nature Geoscience 3, 46e52.

Overesch, M., Rinklebe, J., Broll, G., Neue, H.-U., 2007. Metals and arsenic in soils andcorresponding vegetation at Central Elbe river floodplains (Germany). Envi-ronmental Pollution 145, 800e812.

Paktunc, D., Bruggeman, K., 2010. Solubility of nanocrystalline scorodite andamorphous ferric arsenate: implications for stabilization of arsenic in minewastes. Applied Geochemistry 25, 674e683.

Parkhurst, D.L., Appelo, C.A.J., (US), G.S., 1999. User’s Guide to PHREEQC (Version 2):a Computer Program for Speciation, Batch-reaction, One-dimensional Trans-port, and Inverse Geochemical Calculations. US Geological Survey, Reston, VA.

Pascarelli, S., Boscherini, F., D’Acapito, F., Hrdy, J., Meneghini, C., Mobilio, S., 1996. X-ray optics of a dynamical sagittal-focusing monochromator on the GILDAbeamline at the ESRF. Journal of Synchrotron Radiation 3, 147e155.

Ponnamperuma, F.N., 1972. The Chemistry of Submerged Soils. Academic Press,pp. 29e96.

Pruesse, E., Quast, C., Knittel, K., Fuchs, B.M., Ludwig, W., Peplies, J., Glöckner, F.O.,2007. SILVA: a comprehensive online resource for quality checked and alignedribosomal RNA sequence data compatible with ARB. Nucleic Acids Research 35,7188e7196.

Ravel, B., 2001. ATOMS: crystallography for the X-ray absorption spectroscopist.Journal of Synchrotron Radiation 8, 314e316.

Roberts, L.C., Hug, S.J., Dittmar, J., Voegelin, A., Kretzschmar, R.,Wehrli, B., Cirpka, O.A.,Saha, G.C., Ashraf Ali, M., Badruzzaman, A.B.M., 2010. Arsenic release from paddysoils during monsoon flooding. Nature Geoscience 3, 53e59.

Rose, A.L., Waite, D., 2003. Effect of dissolved natural organic matter on the kineticsof ferrous iron oxygenation in seawater. Environmental Science & Technology37, 4877e4886.

Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source, behaviour and dis-tribution of arsenic in natural waters. Applied Geochemistry 17, 517e568.

Stookey, L.L., 1970. Ferrozine e a new spectrophotometric reagent for iron.Analytical Chemistry 42, 779e781.

Straub, K., Benz, M., Schink, B., Widdel, F., 1996. Anaerobic, nitrate-dependent mi-crobial oxidation of ferrous iron. Applied and Environmental Microbiology 62,1458e1460.

Stucki, J.W., 2011. A review of the effects of iron redox cycles on smectite properties.Comptes Rendus Geoscience 343, 199e209.

Thompson, A., Chadwick, O.A., Boman, S., Chorover, J., 2006a. Colloid mobilizationduring soil iron redox oscillations. Environmental Science & Technology 40,5743e5749.

Thompson, A., Chadwick, O.A., Rancourt, D.G., Chorover, J., 2006b. Iron-oxidecrystallinity increases during soil redox oscillations. Geochimica et Cosmochi-mica Acta 70, 1710e1727.

Tockner, K., Stanford, J.A., 2002. Riverine flood plains: present state and futuretrends. Environmental Conservation 29.

U.S. EPA, 1996. Method 3052 Rev 0 -Microwave Assisted Acid Digestion of Siliceousand Organically Based Matrices.

U.S. EPA, 2000. Soil Sampling (Standard Operating Procedures No. SOP 2012 Rev0.0). U.S Environmental Protection Agency.

Van Cappellen, P., Wang, Y., 1995. Metal cycling in surface sediments: modeling theinterplay of transport and reaction. In: Metal Contaminated Aquatic Sediments.Ann Arbor Press, Chelsea.

Van Veen, W.L., Mulder, E.G., Deinema, M.H., 1978. The Sphaerotilus-Leptothrixgroup of bacteria. Microbiological Reviews 42, 329e356.

Verhoeven, J.T.A., Setter, T.L., 2010. Agricultural use of wetlands: opportunities andlimitations. Annals of Botany 105, 155e163.

Vetter, Y.A., Deming, J.W., Jumars, P.A., Krieger-Brockett, B.B., 1998. A predictivemodel of bacterial foraging by means of freely released extracellular enzymes.Microbial Ecology 36, 75e92.

Vorenhout, M., Van der Geest, H.G., Van Marum, D., Wattel, K., Eijsackers, H.J.P.,2004. Automated and continuous redox potential measurements in soil. Journalof Environmental Quality 33, 1562e1567.

WHO IARC, 2004. Some Drinking-water Disinfectants and Contaminants, IncludingArsenic: Summary of Data Reported and Evaluation. In: IARC Monographs onthe Evaluation of Carinogenic Risks to Humans, vol. 84. World Health Organi-sation International Agency for Research on Cancer.

Winkel, L., Berg, M., Amini, M., Hug, S.J., Annette Johnson, C., 2008. Predictinggroundwater arsenic contamination in southeast Asia from surface parameters.Nature Geoscience 1, 536e542.

Zhang, H., Selim, H., 2008. Reaction and transport of arsenic in soils: equilibriumand kinetic modeling. In: Advances in Agronomy. Elsevier, pp. 45e115.