Science of the Total Environment 466–467 (2014) 490–502

Contents lists available at SciVerse ScienceDirect

Science of the Total Environment

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

Natural attenuation process via microbial oxidation of arsenic in a highAndean watershed

Eduardo D. Leiva a, Consuelo d.P. Rámila a, Ignacio T. Vargas a, Cristian R. Escauriaza a, Carlos A. Bonilla a,Gonzalo E. Pizarro a, John M. Regan b, Pablo A. Pasten a,⁎a Department of Hydraulic and Environmental Engineering, Pontificia Universidad Católica de Chile, Santiago, Chileb Department of Civil and Environmental Engineering, The Pennsylvania State University, University Park, PA, USA

H I G H L I G H T S

• Dissolved As decreases in a stream from a hydrothermal source in Chilean Altiplano.• As attenuation is governed by As(III) oxidation and a pH decrease.• The oxidation of As(III) is mediated by As(III)-oxidizing microorganisms.• Dissolved As attenuation is correlated with As immobilization on Fe-rich sediments.• Biological oxidation coupled to sorption/coprecipitation limits the flux of As from hydrothermal waters.

⁎ Corresponding author at: Departamento de Ingenieríade Ingeniería, Pontificia Universidad Católica de Chile, VSantiago, Chile. Tel.: +56 2 2354 4872; fax: +56 2 2354

E-mail address: ppasten@ing.puc.cl (P.A. Pasten).

0048-9697/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.scitotenv.2013.07.009

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 January 2013Received in revised form 29 June 2013Accepted 2 July 2013Available online xxxx

Editor: F.M. Tack

Keywords:Fluvial arsenicBiological arsenic oxidationAroA-likeChemical arsenic speciationAltiplano

Rivers in northern Chile have arsenic (As) concentrations at levels that are toxic for humans and other organ-isms. Microorganism-mediated redox reactions have a crucial role in the As cycle; the microbial oxidation ofAs (As(III) to As(V)) is a critical transformation because it favors the immobilization of As in the solid phase.We studied the role ofmicrobial As oxidation for controlling themobility of As in the extreme environment foundin the Chilean Altiplano (i.e., N4000 meters above sea level (masl) and b310 mmannual rainfall), which are con-ditions that have rarely been studied. Ourmodel systemwas the upper Azufre River sub-basin, where the naturalattenuation of As from hydrothermal discharge (pH 4–6) was observed. As(III) was actively oxidized by a micro-bial consortium, leading to a significant decrease in the dissolved As concentrations and a corresponding increasein the sediment's As concentration downstream of the hydrothermal source. In-situ oxidation experiments dem-onstrated that the As oxidation required biological activity, andmicrobiological molecular analysis confirmed thepresence of As(III)-oxidizing groups (aroA-like genes) in the system. In addition, the pHmeasurements and solidphase analysis strongly suggested that the As removal mechanism involved adsorption or coprecipitation withFe-oxyhydroxides. Taken together, these results indicate that the microorganism-mediated As oxidation contrib-uted to the attenuation of As concentrations and the stabilization of As in the solid phase, therefore controlling theamount of As transported downstream. This study is the first to demonstrate the microbial oxidation of As inAltiplano basins and its relevance in the immobilization of As.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Arsenic (As) is a toxic element widely distributed in natural envi-ronments (Cullen and Reimer, 1989; Smedley and Kinniburgh, 2002).The contamination of surface and groundwater by As is a seriousenvironmental concern because such contamination limits the useof the water and has adverse effects on human health (Berg et al.,

Hidráulica y Ambiental, Escuelaicuña Mackenna 4860, Macul,5876.

rights reserved.

2001; Bhattacharya et al., 2002; Smedley and Kinniburgh, 2002).Although the behavior of As in nature has beenwidely studied in recentyears (Berg et al., 2001; Manning et al., 1998; Nickson et al., 1998;Nordstrom, 2002; Oremland and Stolz, 2003; Roussel et al., 2000;Stollenwerk et al., 2007), the biogeochemical controls governing themobilization, stabilization, and release of As into river systems are stillunclear.

The mechanisms of As attenuation in natural environments includea combination of physical, chemical, and microbial factors (Wang andMulligan, 2006). The sorption onto iron (Fe) oxyhydroxides is themajor natural attenuation process for the removal and sequestration(stabilization) of As species (Drahota et al., 2012; Leblanc et al., 1996).

491E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

Co-precipitation of As and Fe also contribute to the stabilization ofAs in solid phase (Fuller et al., 1993). However, changes in the As speci-ation, particularly the oxidation from As(III) to As(V), have a crucialrole in the release and mobilization of As because As species differ intheir solubility, toxicity, transport, and bioavailability (Aposhian andAposhian, 2006; Casiot et al., 2005; Drahota et al., 2009; Masscheleynet al., 1991; Smedley andKinniburgh, 2002). Therefore,microorganismscan oxidize As and therefore increase its sorption onto mineral phasesbecause As(V) binds more strongly to Fe oxyhydroxides than As(III)(Oremland and Stolz, 2003). The impact of As oxidation reactions cata-lyzed by microorganisms on As attenuation may vary according to theclimatic and geochemical conditions of a particular system (Lin andPuls, 2003; Smedley and Kinniburgh, 2002).

The occurrence of As in arid environments is widespread dueto geogenic or anthropogenic sources (Baker et al., 1998; Del Razoet al., 1990; Guo et al., 2003; Smedley et al., 2002). Northern Chile isseriously affected by high concentrations (N1 mg l−1) of As in rivers(Alsina et al., in press; Caceres et al., 1992; Landrum et al., 2009;Pizarro et al., 2010; Queirolo et al., 2000; Romero et al., 2003; Sancha,1999), which generates serious problems for the population reliantupon these waters in this arid region. The biogeochemical processesthat control the fate of As in river systems are quite specific becausethe extreme conditions, which are high altitude (N4000 masl) andhigh aridity (b310 mm year−1 of precipitation), impact the environ-ment. In arid and semi-arid systems, metal enrichment processesmay occur via capillary transport (Dold and Fontboté, 2001). The mobi-lized elements (e.g., As) are transported toward the surface of thesoils and sediments where the presence of oxygenmay facilitate oxida-tion (Bechtel et al., 2001; Dold and Fontboté, 2001; Smedley andKinniburgh, 2002). Therefore, the As(III) oxidation and the As stabi-lization (especially as As(V) species) by adsorption processes occur pre-dominantly under oxidizing conditions (Smedley and Kinniburgh,2002). The role of biological reactions, particularly the reactionsmediat-ed by As-oxidizing microorganisms, in the biogeochemical cycling ofAs still is unknown for these types of systems.

The Chilean Altiplano is characterized by a scarcity of water andrich mining activity. Particularly, the Azufre River sub-basin containsa complex mixture of natural and anthropogenic contaminants fromdifferent sources, seriously affecting the quality of the surface waterresources. We observed that in the upper section of this sub-basin,the As concentrations are high (1.0–3.5 mg l−1); additionally, thehydrothermal waters and sediments exhibit naturally elevated con-centrations of As (0.8 mg l−1 and N4 g kg−1 respectively). Interest-ingly, the total dissolved As (AsD) concentrations are attenuated(Δ − 70% AsD), and a noticeable amount of As oxidation (Δ + 90%As(V)/AsD) occurs in the hydrothermal springs (Leiva et al., 2011). Ef-fective risk management and remediation efforts for such As-enrichedfluvial systems as Azufre River sub-basin requires the understandingand quantification of the changes in chemical speciation, as well asthe biogeochemical control of the mobilization/stabilization of As inthe solid phase. In this regard, the Altiplano in northern Chile is anideal site to study the processes that control As mobilization underextreme conditions.

2. General setting

2.1. Climatic and environmental characteristics of the model site

The Altiplano is an area in the central Andes (15°–34°S) coveringthe western part of Bolivia, northern Chile, southern Peru, and north-ern Argentina, with an average altitude of 3600 masl (Allmendingeret al., 1997). The Chilean Altiplano is an elevated plateau 4000 maslcovered by numerous andesitic stratovolcanoes that can reach up6500 masl (Muñoz and Charrier, 1996; Stern et al., 2007). Particularly,the geological formations include conglomerates of the Upper Pliocene–Pleistocene, sandstones, shales, and rhyolitic ignimbrites (Salas et al.,

1966). This area has average temperatures of 0 °C, concentrated rainfallbetween 50 and 300 mm year−1, and potential evaporation of approxi-mately 600–1200 mm year−1 (DGA, 1987). The climatic conditions onall timescales are closely related to changes in the upper-air circulationand by the amount of near-surface water vapor, impacting wet and dryconditions and inducing strong rainfall fluctuations (Garreaud et al.,2003).

The Azufre River sub-basin is located in the XV Region of Arica andParinacota, in northern Chile (Fig. 1). This sub-basin is part of theLluta River Watershed (LRW) and extends between 18° 00′–18° 30′S and 70° 20′–69° 22′ W. The Azufre River originates in the uppersection of the LRW in the Altiplano and has an average altitude of4250 masl (Leiva et al., 2011). This area has a semi-arid climate: rain-fall rarely exceeds an average of 310 mm annually and occurs asintense thunderstorms from late December to late February. Thisarea is characterized by water scarcity caused by the limited rainfall,low humidity, strong solar radiation and high evaporation rates(4.9 ± 0.5 mm d−1) overall. The Azufre River sub-basin is stronglyinfluenced by the Tacora Volcano, which is the northernmostvolcano in Chile and located at 17°43′S, 69°46′W with an altitudeof 5980 masl. The Tacora volcano is a stratovolcano in the CentralVolcanic Zone (CVZ) of the Andes (15°–28°S) (Stern et al., 2007;Clavero et al., 2005). The CVZ is a chain of quaternary stratovolcanoesand andesitic–rhyodacitic domes (Allmendinger et al., 1997 and Kayet al., 1999). The Tacora volcano has been active since at least theMiddle Pleistocene and displays permanent fumarolic (CO2 and SO2

degassing) and hydrothermal features in the southwest flank of the vol-canowithin a drainage area contributing to the Azufre River (Capaccioniet al., 2011; Clavero et al., 2005, 2006). Additionally, the legacy ofsulfur mining is observed on the Tacora crater (~5500 masl) and thewest flank, where tailings and waste rock deposits may be observed.Acid mine drainage (pH b 2.0; As b 0.4 mg l−1, Fe b 10 mg l−1) fromuncontrolled exposure to water and atmospheric O2 is observed atthe foot of mine tailings. The mining legacy and the dry climate limitthe growth of vegetation, but in the upper section of the watershed,there is awetlandwhere different biogeochemical processes can controlthe environmental fate of As and other contaminants.

The study site is located in this wetland, which runs alongthe middle reach of the Azufre River and receives a flow of severalhydrothermal features; these hydrothermal features finally drain intothe Azufre River as described in Fig. 1. Hydrothermal springs are charac-terized by elevated As content, causing themobilization of high concen-trations of dissolved As (N1 mg l−1) and Fe (N1 mg l−1) to the riverflow (Leiva et al., 2011).

2.2. Microbiological context

Microbial processes have a significant impact on the mobility andspeciation of As (Oremland and Stolz, 2003). The transformationsmediated bymicroorganisms play an important role in their modula-tion of As behavior in soils, wetlands, groundwater, and surfacewaters, as well as in the toxicological characteristics of environmen-tal As (Oremland and Stolz, 2003; Mukhopadhyay et al., 2002; Tsai etal., 2009). As(III) is more toxic and mobile than As(V) and its pres-ence usually entails a comparatively larger environmental healthrisk and concern (Mondal et al., 2006; Pierce and Moore, 1982). Bac-teria capable of oxidizing As(III) have been found in sediments, soils,mine tailings, hydrothermal sites, and natural waters (Donahoe-Christiansen et al., 2004; Gihring and Banfield, 2001; Gihring et al.,2001; Hamamura et al., 2009; Inskeep et al., 2007; Oremland et al.,2002; Oremland and Stolz, 2003; Santini et al., 2002). Both hetero-trophic and autotrophic bacteria can oxidize As. Heterotrophic bac-teria oxidize As(III) through a detoxification mechanism and do notobtain energy from the reaction (Anderson et al., 1992; Silver andPhung, 2005). In contrast, autotrophic bacteria oxidize As(III) as achemolithoautotrophic growth strategy, where carbon fixation is

Fig. 1. (a) Map of Chile highlighting the location of the upper section of the Azufre River sub-basin in the Lluta River watershed of northern Chile; (b) transects selected for thestudy. Each transect drains into the Azufre River.*Distance after the origin of the Azufre River (panel a).

492 E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

coupled with the oxidation of inorganic As(III) under aerobic, nitrate-reducing, or denitrifying conditions (Oremland et al., 2002; Rhineet al., 2006; Santini et al., 2002). Microbial oxidation of As is mediatedby enzymes called arsenite oxidases, such as AroA/B, AsoA/B, andAoxA/B (Inskeep et al., 2007; Kashyap et al., 2006; Lebrun et al., 2003;Santini and vanden Hoven, 2004). In hydrothermal systems, the As(III)oxidation activity correlates to the presence of arsenite-oxidizingmicro-organisms (Hamamura et al., 2009; Inskeep et al., 2007; Macur et al.,2004). However, there is no information available concerning the im-pact of arsenite-oxidizing microorganisms on hydrothermal systems at

extreme altitude and in arid lands with high As concentrations in theground and surface waters.

2.3. Environmental and water resources significance

High concentrations of AsD from the Azufre River (1.0–3.5 mg l−1)strongly determine the influx of As into the Lluta River Valley. The LlutaRiver is a precious water resource for a population of ~213,800 locatedin the city of Arica, which is the capital of the XV Region of Arica andParinacota, as well as in many surrounding villages. The Azufre River

493E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

mixes downstream with more alkaline waters (pH increase), conse-quently forming Al hydroxides and Fe oxyhydroxides (Guerra et al.,2012). These solid phases are well-known sorbents of As and occur assuspended solids or coatings on other particles, which are transporteddownstream or deposited depending on the flow velocity. Intensestorms during the wet season trigger events that mobilize As-enrichedsediments deposited along the river bed. The imminent constructionof a dam to increase water availability throughout the year for agricul-tural purposes causes additional concern because the new structurewill most likely become an As repository. The microbial reactions thatcontrol the influx of As at the head waters play a significant role in theoverall As budget and the flux of As into the Lluta River Valley waterresources. Specifically, this research was performed to investigate theinfluence of the As oxidation processes mediated by arsenite oxidizerson the changes in As speciation and attenuation within a hydrothermalspring; these findings will aid the development of future remediationstrategies. Furthermore, we also expect that this case will serve as amodel for other high-altitude hydrothermal systems containing elevat-ed AsD concentrations of reduced As in the CVZ.

3. Methods

3.1. Experimental design in the study site, sampling and on sitemeasurements

The hydrothermal site chosen for detailed study in this work islocated on the west flank of the Tacora volcano (Fig. 1b). Two hot sur-face runoff channels that drain from the same hot spring were chosenfor a transect sampling, each containing four sampling points forhydrogeochemical and microbiological analyses. The average flowvelocities were 25.4 cm s−1 in transect 1 and 12.9 cm s−1 in transect2, based on suspended particle travel times within a distance of 2 min the surface water flow (procedure modified from Langner et al.,2001). The transects were selected based on preliminary data analy-ses from July, 2010 that indicated a decrease in the AsD concen-trations downstream from the discharge of the hydrothermal springsource. In addition, the sampling points were chosen based on theirproximity to high concentrations of Fe-oxyhydroxides, allowingthe simultaneous evaluation of the effects caused by the changesin As speciation and the natural attenuation of AsD concentrations(adsorption onto Fe-oxyhydroxides).

As-rich waters, sediments, and microbial samples were collectedfrom five field campaigns, occurring in October 2010, February 2011,July 2011, November 2011, and May 2012. Water samples from theAzufre River were collected in 300 ml High Density Polyethylene(HDPE) bottles and were immediately filtered in the field with 0.22-μmnylon filters. Filtered subsamples were acidified with HNO3 (1:100 v/v,Merck) for the analysis of the AsD concentration (McCleskey et al.,2004). Unfiltered subsampleswere also acidified to a pHof approximate-ly 2 in the field with HNO3 immediately after sampling (1:100 v/v,Merck) for total As (AsT) (includes particulate As) concentration analysisalong the river.

Water samples from the hydrothermal transects were collectedin 300 ml HDPE bottles and immediately filtered in the field with0.22-μm nylon filters. Filtered subsamples were acidified to pH 2with HNO3 for later laboratory measurements of AsD, total dissolvedFe (FeD) concentrations, and major dissolved cation analysis (sodium,potassium, calcium). For the As speciation analysis (As(III) andAs(V)), the filtered subsamples were acidified in the field to a pH ofapproximately 3 with HCl (1:100 v/v, Merck) (McCleskey et al.,2004).

Additional filtered and non-acidified water samples were col-lected along the Azufre River and in sampling points of the hydrother-mal transects for the later determination of dissolved anions (sulfate,nitrate, and chloride) analysis. All water samples (filtered/acidified,

unfiltered/acidified, and filtered/non-acidified) were stored in dark-ness at 4 °C until processing and analysis occurred within 1 week ofsampling.

The on-site water analyses included temperature (PHC301, HACH),pH (PHC301, HACH), dissolved oxygen (DO) (LDO101, HACH), andelectrical conductivity (CDC401, HACH) (Hq40dMulti, HACH, Loveland,CO, USA). Undisturbed surface sediment samples (depth 6 cm) at eachsampling point of the hydrothermal transects were collected with50 ml polypropylene tubes (BD Biosciences, Mountain View, CA) and150 mlHDPE bottles, andwere kept in darkness at 4 °C until processingand analysis.

3.2. Hydrogeochemical and solid-phase analyses

AsT (unfiltered, acid-soluble), AsD (filtered, acid-soluble), and FeD(filtered, acid-soluble) analyses in water samples were carried outby Total X-ray Reflection Fluorescence (TXRF) (S2 Picofox, TotalX-ray reflection fluorescence spectrometer, Bruker AXS, Billerica,MA, USA) (Borgese et al., 2009). The AsT, AsD, and FeD concentrationsin the water samples were confirmed by Inductively Coupled PlasmaOptical Emission Spectrometry (ICP-OES) (PerkinElmer Optima 7300DV ICP-OES, Shelton, CT, USA) (data not shown). The dissolved inor-ganic As species [As(III) and As(V)] were measured in the laboratoryusing filtered/HCl-acidified subsamples. The dissolved As(III) concen-trations were determined by hydride generation with cysteine andsodium borohydride reductants, as well as Inductively Coupled Plas-ma Mass Spectrometry (ICP-MS) detection (PerkinElmer ELAN 6100,Shelton, CT, USA) (detection limit b0.002 mg l−1) (Garbarino et al.,2002). The As(V) concentration was determined by subtracting theAs(III) content from the AsD concentration measured with TXRF.The contents of the dissolved inorganic Fe species [Fe(II) and Fe(III)]were determined by spectrophotometry. The ferrous (Fe(II)) iron con-centration was determined in the field immediately after sampling infiltered/non-acidified subsamples with a HACH spectrophotometerDR/2010 (1–10 Phenanthroline method 8146, adapted from StandardMethods for the examination of water and wastewater 15th ed 201(1980)) (HATCH, Loveland, CO, USA). Ferric (Fe(III)) iron concentrationwas determined by subtracting the Fe(II) content from the FeD concen-tration measured with TXRF. Sodium, potassium, and calcium wereanalyzed via ICP-OES from the filtered/acidified subsamples. Sulfate,nitrate, and chloride were analyzed using Ion Chromatography (IC)(882 compact IC plus, Metrohm, Herisau, Switzerland) with filtered/non-acidified subsamples.

The sediment samples were dried at 40 °C and the analysis of thetotal As in the solid phase was performed with the prior microwavedigestion (Microwave Accelerated Reaction Systems, model MARS(CEM corporation, Mattheus; North Carolina 28106)) of ~100 mg ofdry sample (b2 mm) in 9 ml of concentrated HNO3 and 3 ml of con-centrated HCl (EPA method 3051A). The As content of the digests wasdetermined by TXRF. The results of the As analyses in the solid phasewere confirmed by ICP-OES (PerkinElmer Optima 7300 DV ICP-OES,Shelton, CT, USA) (data not shown).

3.3. Analytical quality control

Continuous quality assurance/quality control (QA/QC) procedureswere performed for both the field measurements and laboratoryanalyses by TXRF, ICP-OES and IC. All field and laboratory instrumentswere calibrated using certified reference standards or analytical gradesynthetic standards. The acceptable standard deviation for results ofmetals, cations, and anions was always b10%, and all analyses wereinterspersed with and checked against standard reference materials.

The AsT, AsD and FeDmeasurements of the water samples and theAs content of acid digest samples were measured by TXRF in tripli-cate. The QC of the measurement was performed by inserting a syn-thetic standard with a known concentration of As and Fe into the

494 E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

analytical run every 20 samples. Similar to the acid digest samples,the measured concentrations of trace elements (As, Cu, Pb, Fe, Mn)in certified reference standards (Nist 2781, Nist 2702 and SRM 2709aSan Joaquin Soil) were measured with an error range of 5 ± 1% oftheir certified values. The quantitative TXRF analysis was conductedby adding an internal Gallium (Ga) standard of known concentration.In addition, the metal concentrations in the blanks were alwaysbelow the TXRF lower limits of detection (LLD) calculated for thewater samples (b0.01 mg l−1 for As, and b0.03 mg l−1 for Fe). TheQC of the equipment was achieved by monitoring the operationalparameters. A gain correction (adjustment of the spectroscopic ampli-fication) was performed daily and the spectroscopic resolution, sensi-tivity and accuracy of quantification were checked monthly.

The measurements of metals (As and Fe) and cations (sodium,potassium, and calcium) made via ICP-OES were performed in dupli-cate, and a QC sample composed of certified concentrations of metalsand cations was incorporated into the analytical run every 10 sam-ples. The blanks for the metals (As, Fe) were always under the detec-tion limit of the equipment (b0.006 mg l−1 for As, and b0.01 mg l−1

for Fe). The same was observed for the cation analysis; the values ofblanks were under the limit of detection for calcium (b0.01 mg l−1),sodium (b0.01 mg l−1) and potassium (b0.01 mg l−1). QC of equip-ment was accomplished by the periodic monitoring of the calibrationcurve slope. Furthermore, the accuracy and precision of the measure-mentswere evaluated by checking the results against certified standardsamples.

The anion analyses were performed by IC in duplicate and includedan internal quality control sample every 20measurements composed ofsynthetic standards of known anion concentrations to corroborate theresults. For the TXRF and ICP-OES analyses, the IC analysis also reportedblank values under the detection limit for chloride (b0.34 mg l−1), sul-fate (b0.061 mg l−1) and nitrate (b0.076 mg l−1). The QC of equip-ment included a calibration curve of the anions, which was checkedmonthly against synthetic analytical grade standards. After each mea-surement, the columnwaswashed continuouslywith themobile phase.

3.4. In-situ arsenite-oxidation assays

The biotic and abiotic rates of As(III) oxidation were evaluated inthe presence of both alive and dead microbial communities, respec-tively. For the abiotic oxidation assays, sediment samples (64 cm3)upstream from the points of each transect [transect 1 (2.7 m),transect 2 (4.2 m)] were transferred to 500 ml bottles. The sedimentsamples were covered with 10 ml of 37% (v/v) formaldehyde, beforethe addition of 300 ml of spring water and 2 mg l−1 of As(III). Thebiotic oxidation assays were performed following the same protocol,except that formaldehyde was not added to the sediments. Thesample bottles were incubated in a neighboring hot spring of similartemperature [transect 1 (~30 °C), transect 2 (~22 °C)]. Water sub-samples (25 ml) were taken at 0, 3, 5, and 12 min, filtered in thefield using 0.22-μm nylon filters, and analyzed to determine the Asspeciation and AsD. This protocol was modified from that ofLangner et al., 2001.

3.5. Microbiological analyses

3.5.1. Sampling and DNA/RNA extractionThe microbiological samples were collected in sterile 50 ml

polypropylene tubes (BD Biosciences, Mountain View, CA) and pre-served for DNA (4 °C) or RNA extraction (RNASafer Stabilizer Reagent,MoBio, Carlsbad, CA, USA). Metagenomic DNA isolation was performedusing the Power Soil DNA Isolation Kit (MoBio, Carlsbad, CA, USA).The integrity of the DNA was verified via 1% (w/v) agarose gel electro-phoresis prepared with SYBR Green DNA binding dye (1×) (Invitrogen,Carlsbad, CA, USA). The DNA quantity and purity were measured with

spectrophotometry based on the A260 and A260/A280 ratio. The DNApreparations were stored at −20 °C until analysis.

For the RNA extraction, environmental samples were preservedin the field with RNASafer Stabilizer Reagent (MoBio, Carlsbad, CA,USA), before being stored at 4 °C in the laboratory. The total RNAwas extracted from 0.5–1.0 g of wet sample using the Power SoilTotal RNA isolation kit (MoBio, Carlsbad, CA, USA). The extracted RNAwas treated with RNase-free DNase (Promega, Madison, Wisconsin,USA), recovered with a phenol-chloroform solution (1:1), and pre-cipitated with ethanol. Finally, the RNA was resuspended in 50 μl ofnuclease-free water. The RNA quantification was measured at A230.RT-PCR was performed using a SuperScript RT-PCR System (Invitrogen,Carlsbad, CA, USA). To verify the absence of DNA contamination, RT-PCRcontrols were performed without the addition of reverse transcriptase.Finally, the cDNA preparations were stored at−20 °C until analysis.

3.5.2. Polymerase chain reaction (PCR) amplificationPCR amplification of the aroA-like geneswas performedwith degen-

erate primers [aroA95f (5′-TGYCABTWCTGCAIYGYIGG) and aroA599r(5′-TCDGARTTGTASGCIGGICKRT T)] (Hamamura et al., 2009) from en-vironmental DNA and cDNA. The reactions (50 μl) were preparedusing similar masses of the DNA and cDNA templates, as well as 1 μMof each primer; the program used was previously described elsewhere(Hamamura et al., 2009). The amplified products (500 bp) were visual-ized via agarose gel electrophoresis, and the aroA-like gene ofAgrobacterium tumefaciens 5A (aoxAB) was used as the positive control.Subsequently, 16S rRNA gene amplification from the environmentalDNA was performed using the universal primers 8F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1392R (5′-ACGGGCGGTGTGTAC-3′) (Amannet al., 1995).

3.5.3. Terminal restriction fragment length polymorphism (T-RFLP)analysis

The T-RFLP analyses were carried out using DNA samples fromthe hydrothermal source (0 m), because it presented the highest con-centrations of As(III) and fed both transects. T-RFLP analysis wasperformed using the previously described universal primers for the16S rRNA gene (Amann et al., 1995); the 8F primer was labeled withthe fluorophore 6-FAM and the 1392R primer was left unchanged. Thelabeled products were digested separately with 20 U of HhaI and MspIendonucleases and precipitated as described elsewhere (Moran et al.,2008). The DNA fragments were separated via capillary electrophoresis(ABIPrism 310, Applied Biosystems), and the fragment sizes wereestimated using the ROX 1000 internal standard (Applied Biosystems,Foster City, CA, USA). The T-RFLP data were plotted as the area (stan-dardized as relative abundance using the sum of all peak areas) ofeach restriction fragment fluorescence peak against the fragmentsizes. The terminal restriction fragments (T-RFs) with lengths b50and N700 bp and those representing b0.5% of the total area were notincluded in the analyses (Schutte et al., 2008).

The in silico DNA restriction analysis of the 16S rRNA gene se-quences from arsenite-oxidizing microorganisms (Hamamura et al.,2009; Inskeep et al., 2007) was enacted to create a new database ofpredicted T-RFs that included aroA-like positive microorganisms.The database was generated using the Virtual Digest tool, which isa component of the Microbial Community Analysis (MiCA) onlineresource (Shyu et al., 2007); the new predicted T-RFs were added tothe Ribosomal Database Project (RDP) (Cole et al., 2009). Subsequently,the phylogenetic analysis was performed by importing the generateddatabase into the PAT program (Kent et al., 2003) to allow a greatertolerance range for the sizes of the T-RFs; a sizing error of 1 bp wasallowed for the smallest T-RFs (50–200 bp) and a 3 bp error wasallowed with the longest T-RFs (200–700 bp). The database wasmatched to the T-RF lengths generated from the T-RFLP profiles ofthe hydrothermal spring. Two datasets were generated for the possibleassignments of the T-RFs obtained from the digestion with each

495E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

restriction enzyme (HhaI and MspI), and these datasets were crossedto refine the assignment of candidate species at the sampling point.

4. Results

4.1. Arsenic concentrations in the Azufre River

The AsT and AsD concentrations along the Azufre River wereobserved between 3.4 and 1.0 mg l−1 (Fig. 2a), widely exceeding themaximum level for drinking water (10 μg l−1) (WHO, 1993). The AsTand AsD were very high at the origin of the Azufre River (distance0 m) (3.4 ± 0.3 mg l−1 and 3.1 ± 0.3 mg l−1, respectively), possiblydue to hydrothermal discharges with high As concentrations. Therewere no significant differences between the AsT and AsD concentra-tions because the pH along the river was very low b2.5, preventingthe sequestration of As by particulate solids.

Downstream of the Azufre River, in the wetland area (400–1500 m)the AsT and AsD concentrations decreased. This decrease was explainedby a dilution effect caused byflowingwaterwith lower As concentrationsandby the active natural As attenuation processes in this area. Theflow inthe Azufre River is ~14 l s−1 upstream of the wetland and ~30 l s−1

downstream of the wetland. The dilution effect is made evident bychanges in chloride concentrations (Fig. 2b), which decreased by approx-imately 60% (from 536 to 205 mg l−1). The AsD concentrations exhibit a

Fig. 2. Decreases in the AsT and AsD concentrations (a); and dissolved chloride(b) along the Azufre River in the wetland area. The graphs are presented relative to thedistance from the origin point of the Azufre River (P1). Altitude (masl) is shown for refer-ence. The wetland area extends between 400 and 1500 m after the origin of the AzufreRiver. The data are presented as the mean ± SEM (error bars).

greater reduction (from 3.4 to less than 1.0 mg l−1) along this samelength of the river that cannot be explained solely by this phenomenon(the stream flow increases only 2.1 times). Based on these data, we stud-ied the attenuation of As in hydrothermal transects in this area of thewetland.

4.2. Arsenic geochemistry in the transects

The measured AsD (Fig. 3) exhibited a marked decrease in thesampling points farthest from the source (Δ − 72% for transect 1and Δ − 76% for transect 2), demonstrating a clear attenuation ofthe aqueous phase As concentration. The AsD concentration at thesource was 0.8 ± 0.2 mg l−1. In addition, we also observed an increasein the concentration of As in the solid phase at the last 3 points of eachtransect (Fig. 4), which correlates with the attenuation observed inaqueous phase. Interestingly, the source presented the highest con-centration of As in the solid phase (6.4 ± 1.7 mg kg−1 soil) and XRDanalysis revealed that the lyophilized sediments of the transects areadid not contain XRD-crystalline minerals and were therefore assignedas unidentified amorphous Fe-oxyhydroxides.

The results of the dissolved As speciation analysis demonstrateda marked oxidation of As in both transects (Fig. 7). Similarly, the Fespeciation analysis illustrated that Fe(III) predominated in bothtransects (Fig. 8). Additionally, the Fe concentrations in the solidphase were constant along both transects: 285.1 ± 46.9 g kg−1 soilfor transect 1 and 292.6 ± 63.2 g kg−1 for transect 2 (consideringall sampling points). There were no significant concentrations ofsuspended solids (AsD/AsT 0.91–0.95) or turbidity (4.0 ± 2.2 NTU)measured in the bulk water.

The parameters analyzed along the transects revealed that thespecific electrical conductivity was similar at the different samplingpoints of each transect (~4 mS cm−1), and the dissolved oxygenmeasurements revealed the water oxygenation from the sourceuntil the discharge of the runoff in the Azufre River (2 mg l−1 to4.5 mg l−1). There was a marked decrease in temperature (−11 °Cin transect 1 and −13.5 °C in transect 2), and the pH decreaseddownstream from the source (Fig. 5). The decrease in pH was moreevident in transect 2 than in transect 1. While in transect 1, the pHdeclined slightly, and in transect 2 the decrease was more noticeable,falling rapidly to reach values near 4.

The sulfate, sodium, and chloride concentrations remained nearlyconstant downstream from the hydrothermal source (Fig. 6) in bothtransects. The same was observed for nitrate, calcium, and potassium(data not shown). The AsD/SO4

2−, AsD/Na+, and AsD/Cl− ratiosexhibited a marked reduction in both transects, which was likelyinfluenced by the AsD attenuation along the two transects.

4.3. In-situ oxidation experiments

As(III) oxidation was not observed when the biological activitywas eliminated via formaldehyde addition to the samples from eithertransect (Fig. 9). Therefore, the As(III)/As(V) ratio remained constantwhen the biological activity was eliminated. The in-situ oxidationexperiments performed in the presence or absence of sunlight revealedno significant differences in As speciation (data not shown).

The water residence times in both transects were relatively short at1.04 ± 0.12 min for transect 1 and 1.0 ± 0.04 min for transect 2, andthe extent of the As oxidation cannot be explained by slow abiotic oxi-dation reactions. However, in the solid–liquid interface, the residencetimesmay be higher. Additionally, theAs(III) oxidation can proceed abi-otically via Mn oxides, which assist in the oxidation of As(III) and areconsequently reduced to Mn(II) (Manning et al., 2002; Scott andMorgan, 1995). However, the Mn concentrations in the solid phase ofthe transects were low; concentrations of 0.37 ±0.06 g kg−1 soil fortransect 1 and 0.36 ± 0.09 g kg−1 for transect 2 were observed, con-sidering all sampling points.

Fig. 3. As concentrations in the sampling points of transect 1 (a) and transect 2 (b). The AsD concentrations are expressed as the relative concentration (Ci/C0) down gradient of thedischarge from the hydrothermal source (Ci = concentration at distance i, C0 = concentration at source). C0 in aqueous phase was 0.8 ± 0.2 mg l−1. The data are presented as themean ± SEM (error bars).

496 E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

4.4. Molecular analyses

The biological As oxidation is an enzymatic reaction mediatedby arsenite oxidase (AroA). We searched for aroA-like genes in themetagenomic DNA extracted from the sampling points. In all sam-pling points, the putative aroA-like gene was amplified (Fig. 10a),indicating that there was the potential to oxidize As biologically.Additionally, the aroA-like gene expression analysis (from cDNA)confirmed that the aroA transcript was produced in the first threesampling points of both transects (Fig. 10b, upper gel) and can beinduced in the last point (two transects) via exogenous addition of1 ppm of As(III) [as NaAsO2] (Fig. 10b, lower gel).

The in silico analysis of the T-RFLP results supported the presenceof several operational taxonomic units with species assigned toarsenite-oxidizing bacteria (Table 1), including Hydrogenophaga sp.,Thiobacillus sp. and Thermus thermophilus. The other microorganismsidentified were all uncultured bacteria, which were also found inenvironmental samples and sites with similar characteristics to ourstudy site (e.g., Uncultured eubacterium env. OPS 6; YNP). Two ofthese bacteria presumably correspond to members of the Chloroflexiand Alphaproteobacteria classes.

5. Discussion

5.1. Arsenic natural attenuation occurs in the transects

The difference between the reduction of chloride concentrations(~60%) and the reduction in the As concentrations (~71%) along theAzufre River indicates than an active natural As attenuation process

Fig. 4. Solid phase As concentrations in the sampling points of transect 1 (a) and transect 2 ((Ci/C0) down gradient of the discharge from the hydrothermal source (Ci = concentration aThe data are presented as the mean ± SEM (error bars).

was active in this area. In particular, the reduction in the As concen-trations in the surface runoff of hydrothermal waters reveals thepresence of active As stabilization processes. Additionally, therewere no significant changes in the ionic matrix of this runoff to sug-gest any local dilution processes were occurring.

The stability of the conductivity (approximately 4 mS cm−1) alongboth transects suggests that there is no mixture of waters to explainthe results of As attenuation; although the evidence is compelling, it isnot conclusive. However, the conservative behavior of the anions andcations observed in both transects precludes the possibility of dilution.The obvious reduction in the AsD/SO4

2–, AsD/Na+, and AsD/Cl− ratiossupports the hypothesis that the As attenuation has a different geo-chemical behavior than the tracer ions, such as SO4

2−, Na+, and Cl−.Therefore, the changes in the AsD concentrations along the transectsare not caused by dilution; if dilution occurred, then the As/ion ratioswould remain stable throughout both transects. While a dilution effectwith waters from the same ionic matrix but lower As concentrationswould give this same outcome, there are no known hydrothermalwater sources with lower As concentrations able to explain the ob-served phenomenon, and the stream flow along the two transectsexhibited no substantial change.

5.2. Arsenic stabilization occurs onto Fe minerals

The decreasing AsD and increasing amount of solid phase Asacross the last three points of the two transects might be explainedby an increase in the adsorption of As onto the Fe-oxyhydroxidespresent in this area. As adsorption onto Fe-oxyhydroxides and the

b). The As concentrations in the solid phase are expressed as the relative concentrationst distance i, C0= concentration at source). C0 in solid phase was 6.4 ± 1.7 mg kg−1 soil.

Fig. 5. Geochemical parameters in the water samples from the sampling points of transect 1 (a) and transect 2 (b). DO (○) and temperature (△) reveal the expected behavior forhydrothermal waters in contact with the atmosphere. The decrease in pH (□) is important for the adsorption processes of As, and stability of the conductivity (⋄) suggests thatthere is no dilution. The data are presented as the mean ± SEM (error bars).

497E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

formation of ferric arsenate precipitates are the main natural attenu-ation processes for the removal of As (Drahota et al., 2012; Haffertand Craw, 2008; Leblanc et al., 1996), supporting the hypothesis ofthe relevance of these minerals in the attenuation of As in the studiedtransects (Fig. 3). Incidentally, amorphous Fe-oxyhydroxides have ahigh adsorption capacity for As because they have a large surfacearea and therefore provide more active sorption sites (Wang andMulligan, 2006). Consequently, the presence of unidentified amor-phous Fe-oxyhydroxides lends credence to the idea that these min-erals are crucial for the attenuation of As observed in both transects.The obvious As(III) oxidation to As(V) in both transects couldenhance the adsorption of As onto Fe-oxyhydroxides under varying pHconditions. As(V) interacts with Fe-oxyhydroxides more strongly thanAs(III) (Oremland and Stolz, 2003) because the adsorption mechanismis strongly influenced by pH changes. Therefore, the As(III) oxidationto As(V) and the subsequent As(V) adsorption onto Fe-oxyhydroxidesmight explain the As attenuation observed downstream from thehydrothermal streams.

The presence of similar Fe concentrations in the solid phase ofboth transects downstream of the hydrothermal source suggeststhat changes in the concentrations of As in the aqueous phase weredue to an active process at the solid–liquid interface and not by in-creasing As sorbent concentrations (Fe-oxyhydroxides). pH decreasealong the transects supports this because at low pH it does notfavor the formation of these minerals. However, the circumneutralpH in the hydrothermal source and the presence of Fe(III) in theaqueous phase may favor the formation of Fe-oxyhydroxides in thislocation and therefore increase the availability of sediments withfree surface sites for As(III) adsorption. This phenomenon would ex-plain the higher As concentrations in solid phase at this location(source) (Fig. 4). Additionally, the co-precipitation of As and Fe canalso occur during the formation of Fe-oxyhydroxides (e.g., Ferrihydrite),which may help to mitigate the As concentration in aqueous phase(Fuller et al., 1993). Additionally, the co-precipitation of As(V) andFe(III) may be favored by the As(III) oxidation (Jia and Demopoulos,2008; Richmond et al., 2004; Twidwell et al., 2005). However, the low

concentrations of suspended solids (turbidity b 6 NTU and AsD/AsT0.91–0.95) invalidate the supposition that the As attenuation in aque-ous phase occurred because of the As adsorbed on the suspended par-ticulate phases. These results support the idea that the mineral phasespresent in the runoff beds of both transects were responsible for theobserved attenuation of As.

5.3. pH control on arsenic immobilization

The correlation between the pH reduction and AsD attenuation inthe transects supports the hypothesis that the oxidation of As(III) toAs(V) favors As stabilization on the solid phase by increasing theAs(V) adsorption onto solid phase. As(III) and As(V) are stronglychemisorbed on Fe-oxyhydroxides, but this process is pH dependent.While As(V) is adsorbed onto Fe-oxyhydroxides between pH 4 and 7,with an optimal adsorption occurring around pH 4, As(III) is morestrongly adsorbed between pH 6 and 9; and this pattern is mostnoticeable with the amorphous Fe-oxyhydroxides (De Vitre et al.,1991; Dixit and Hering, 2003; Manning et al., 1998; Pierce andMoore, 1982).

Therefore, the decreasing pH along the transects (Fig. 5) revealspossible changes in the differential adsorption of As(III) and As(V)because, at circumneutral pH, the adsorption of As(III) is maximizedand favored over As(V) adsorption (Dixit and Hering, 2003; Goldbergand Johnston, 2001). In the sampling points downstream from thehydrothermal source, As(V) predominates in aqueous phase and thepH becomes more acidic (~pH 4). As the As(III) is oxidized to As(V)along the transects, and the decrease in the pH increases the As(V)sorption; therefore, decreasing pHs partially promotes the As attenua-tion observed in the transects. However, the high As concentration inthe solid phase found at the source can be explained by the pH presentat this point, which is approximately 6 (Fig. 5); at this pH, the adsorp-tion of As(III) is promoted, which has a higher concentration at thispoint (Fig. 7). Additionally, this sampling point is the hydrothermalsource with naturally emerging high AsD concentrations. However,the specific behavior of adsorption onto the Fe-oxyhydroxides is

Fig. 6. Sulfate, sodium, and chloride concentrations in the sampling points of study transect 1 (a) and transect 2 (b). AsD/SO42−, AsD/Na+, and AsD/Cl− ratios in study transect 1 (c) and

transect 2 (b). Both transects reveal the stability of ion concentrations between the different sampling points. The AsD/ion ratios indicate amarked reduction in both transects. The ratioswere calculated from the As and ion concentrations (mg l−1) and were plotted by multiplying the values by 104. The data are presented as the mean ± SEM (error bars).

498 E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

dependent on the acid–base characteristics of the As(III) and As(V)solutions, the affinity of surface functional groups of the solid, andthe cooperative or competitive effects of the anions and cations in theaqueous phase (Davis and Kent, 1990; Kent and Fox, 2004; Grafeet al., 2004).

Fig. 7. Arsenite oxidation in the water samples from transect 1 (a) and transect 2 (b). ConcenThere is an obvious oxidation of As in both transects from the hydrothermal source (0 m).

5.4. Arsenite oxidation and natural attenuation is dependent onbiological activity

The results of the in-situ experiments demonstrate that microbialactivity is necessary for the oxidation of As(III). Additionally, these

trations of As(III) and As(V) are presented relative to the concentration of the total AsD.

Fig. 8. Fe speciation (Fe(II) and Fe(III)) in the water samples from transect 1 (a) and transect 2 (b). Concentrations of Fe(II) and Fe(III) are presented relative to the concentration ofthe total dissolved Fe. In both transects, Fe(III) predominates over Fe(II) and is stable along the transects.

499E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

results confirmed that abiotic As(III) oxidation caused by DO oranother electron acceptor was not significant compared to the oxida-tion activity in the presence of microbial activity at similar time scales.The kinetics of the oxygen-mediated As oxidation has a reportedhalf-life of one year (Eary and Schramke, 1990). Additionally, abioticAs(III) oxidation may occur in association with Fe-oxyhydroxides,where As(III) is adsorbed, oxidized, and finally desorbed through apH-dependent process (Belzile and Tessier, 1990; Horneman et al.,2004; Smedley and Kinniburgh, 2002). However, the most importantabiotic pathway of As oxidation in natural systems is the transfer ofelectrons to Mn oxides (Manning et al., 2002; Oscarson et al., 1981).As(III) is adsorbed on the Mn oxide surfaces and undergoes severalredox reactions on the mineral surface before being released to theaqueous phase (Parikh et al., 2010; Scott and Morgan, 1995). The Asoxidation rate promoted by the Mn oxides depends on the availablesurface area, surface charge, and pH (the oxidation rate is slightly higher

Fig. 9. In-situ As(III) oxidation assays with sediments from transect 1 (a) and transect 2 (b)

at low pH) (Chiu and Hering, 2000; Oscarson et al., 1983). AlthoughMnoxides may form on the surface of Fe-oxyhydroxides and As(III) can beoxidized more rapidly at high temperatures (N48 °C) through a photo-catalytic process (Schwenzer et al., 2001), the low concentrations ofMnin solid phase (b0.5 g kg−1) and the results of in-situ experimentsdiscount this mechanism. Additionally, the formation of Mn oxides isfavored at a higher pH than those observed in this system (pH 4–6)(Morgan, 1967).

The photochemical oxidation of As(III) was also studied by Huget al. (2001), who observed a 90% removal of As(III) within 2 to 3 hin solutions containing Fe(II, III) and citrate. At 25 °C under naturalsunlight, the half-life of As(III) is 0.7 h at pH 5 in a solution of18 μM Fe(III) and 100 mg l−1 DOC (Kocar and Inskeep, 2003).Accordingly the photocatalytic As oxidation has faster kinetics(hours) than a strictly oxygen-dependent abiotic As oxidation. How-ever, the lack of significant differences between the in situ oxidation

. As oxidation is evident when the biological activity is not inhibited by formaldehyde.

Fig. 10. Agarose gel displaying the amplification of an aroA-like gene (a) and aroA-like expression (b) in the microbial samples from the study transects. All samples were alsosupplemented with 1 ppm As(III) to confirm the aroA-like gene's functional role (lower panel of (b)). The 16S rRNA gene was used as a control for DNA quality, and Agrobacteriumtumefaciens 5A was a positive control for arsenite oxidase activity.

500 E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

experiments in the presence and absence of light demonstratesthat the solar radiation does not cause the observed As oxidation inthis site at the measured time scale (Fig. 7). Consequently, in othernatural systems with pHs ranging from 3 to 6, the abiotic oxidationof As(III) is negligible relative to the biologically mediated oxidation(Eary and Schramke, 1990; Langner et al., 2001; Taylor, 2007). There-fore, in natural environments, the available evidence indicates thatthe rates of abiotic As oxidation by photochemical, photocatalytic, orFe or Mn oxides-mediated reactions are significantly lower than theoxidation mediated by microorganisms (Cherry et al., 1979; Eary andSchramke, 1990; Ying et al., 2011). Therefore, the absence of As oxida-tion observed when microbial processes were neutralized (Fig. 9)supports the idea of microbial activity performing the majority of theAs(III) oxidation.

Similarly, the microbiological analyses suggest that the functionalbiological activity of the As oxidation promotes the geochemicalchanges observed in the As(III), As(V), and AsD concentrations.Among the microorganisms found by T-RFLP analyses,Hydrogenophaga

Table 1Closest matching species from the 16S rRNA gene T-RFs assignment in the 16S rRNA datab

Closest match species 16S rRNAacc. no.(GenBank)

AroA-likeacc. no.(GenBank)

AroA-like microorganismsHydrogenophaga sp. str. CL3 DQ986320 EF015462Thiobacillus sp. str. S1 DQ986319 EF015459Thermus thermophilus X07998, NC_006461 BAD71923

Other microorganismsUncultured eubacterium env. OPS 6 AF018191 –

Uncultured eubacterium clone BBACe10 GU357466 –

Uncultured Chloroflexi bacterium HQ397171HQ397189HQ397194

–

Uncultured bacterium DA067 –

Uncultured bacterium GQ921453GQ921460GQ921465

–

a CAO: chemoautotrophic arsenite oxidizer.b HAO: heterotrophic arsenite oxidizer.

sp. and Thiobacillus sp. are chemoautotrophic arsenite oxidizers(CAOs); both of these organisms belong to the class Betaproteobacteria,are common in soils, and are frequently associated with metal-contaminated sites (He et al., 2007; Heinzel et al., 2009). We alsofound T. thermophilus, which is a heterotrophic arsenite oxidizer(HAO), revealing the variety of bacterialmetabolisms thatmight impactthe As cycle in this area. Recent studies have revealed that these organ-isms are present in hydrothermal systems with high As concentrations(Hamamura et al., 2009; Inskeep et al., 2007) and might contributeto the oxidation of As(III) in natural systems (Hamamura et al., 2009).For the other uncultured bacteria found in the study site, there is noinformation available regarding their As oxidation activity. How-ever, the phylogenetic analyses conducted by other authors have iden-tified microorganisms belonging to both classes (Chloroflexi andAlphaproteobacteria) commonly found in arsenic-impacted soils andgeothermal environments (Hamamura et al., 2009; Inskeep et al.,2007). The presence of As-oxidizing bacteria in this area is consistentwith the aroA-like presence and expression, suggesting that these

ases. The microbial characterization corresponds to the hydrothermal source (0 m).

Phylogenetic affiliation Arsenic metabolism AroA-like gene

β-Proteobacteria CAOa aroABβ-Proteobacteria CAOa aroABDeinococcus–Thermus HAOb aroAB

– – –

– – –

Chloroflexi – –

– – –

α-Proteobacteria – –

501E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

populations are responsible for the As oxidation observed in thestudied transects. Therefore, the results of the in situ As oxidation exper-iments, the aroA-like gene presence/expression, and the T-RFLP analysisstrongly support the hypothesis that arsenite-oxidizingmicroorganismscontribute to the As oxidation and consequently to the natural As atten-uation observed in the study site.

6. Conclusions

Our results indicate that there was a natural attenuation of As in themodel system. The AsD concentrations decreased along the transects,while the conservative ion concentrations along the transects discountthe possible effects of dilution or mixing of waters. The rapid oxidationof As(III), which occurred within a few meters of the hydrothermalsource (As(V)/AsD increases of+Δ95% [transect 1] and + Δ100% [tran-sect 2]), suggests the possible involvement of As oxidation in the naturalattenuation process. The increase in the As concentrations in the solidphase after the source correlates with the As oxidation in transects 1and 2, suggesting that the stabilization is mediated by the adsorptionof As onto Fe-oxyhydroxides and controlled by the environmental pH.The results of the in-situ oxidation experiments reveal that livemicroor-ganisms were required to catalyze the As(III) oxidation at the studiedtime scale. Complementary microbiological molecular analyses supportthese findings and suggest that both the oxidation of As(III) and the at-tenuation of AsD is mediated directly by microorganisms via enzymaticactivity (AroA-like). The results of the aroA-like gene expression sug-gests that changes in the expression of this gene can depend uponAs(III) concentration. The T-RFLP analysis supported the presence ofarsenite-oxidizing bacteria in the studied area and reveals the autotro-phic and heterotrophic metabolisms that are most likely to be active inthe As oxidation.

Our findings contribute to a better understanding of biogeochem-ical processes of As oxidation, as well as the biogeochemical controlsof As in fluvial systems impacted by As-rich hydrothermal discharges;these As-rich hydrothermal discharges occur in upper section of theAzufre River sub-basin. The description of the microbial As oxidationreactions involved in these processes is crucial for improving our un-derstanding of the As cycle in extreme, arid environments. Together,these results illustrate the complexity of the interactions involved inthe mobilization and stabilization of As due to the interplay betweenredox and sorption reactions at the mineral-liquid interface involvingbiotic and abiotic reactions at different scales.

Future research should include detailed spectroscopic analysis of thesolid-phases to determine the speciation of As and Fe on the surface ofthe Feminerals and to confirm the specific mechanisms involved in thestabilization of As at the molecular level. Additional efforts are neces-sary to describe the environmental conditions that might inhibit or en-hance this attenuation process. This knowledge is crucial for the designof effective riskmanagement and remediation efforts forfluvial systemsenriched with As from hydrothermal discharges.

Acknowledgments

This research was supported by a FONDECYT 1100943/2010, aFONDECYT 1130936/2013, and a CONICYT grant 24121233/2012, aswell as Fulbright Scholar Award 9561 for J.M.R. This study was also par-tially supported by a CORFO09CN14-5709 grant and a CONICYT/FONDAP15110020 grant. Thanks are extended to the reviewers for the correc-tions and suggestions, which significantly improved the manuscript.

References

Allmendinger RW, Jordan TE, Kay SM, Isacks BL. The evolution of the Altiplano–Punaplateau of the Central Andes. Annu Rev Earth Planet Sci 1997;25(1):139–74.

Alsina MA, Zanella L, Hoel C, Pizarro GE, Gaillard J-Fo, Pasten PA. Arsenic speciation insinter mineralization from a hydrothermal channel of El Tatio geothermal field,Chile. J Hydrol 2013. (in press).

Amann RI, Ludwig W, Schleifer KH. Phylogenetic identification and in-situ detection ofindividual microbial-cells without cultivation. Microbiol Rev 1995;59:143–69.

Anderson GL, Williams J, Hille RL. The purification and characterization of arseniteoxidase from Alcaligenes faecalis, a molybdenum-containing hydroxylase. J BiolChem 1992;267:23674–82.

Aposhian HV, Aposhian MM. Arsenic toxicology: five questions. Chem Res Toxicol2006;19:1–15.

Baker LA, Qureshi TM, Wyman MM. Sources and mobility of arsenic in the Salt Riverwatershed, Arizona. Water Resour Res 1998;34(6):1543–52.

Bechtel A, Sun Y, Püttmann W, Hoernes S, Hoefs J. Isotopic evidence for multi-stagebasemetal enrichment in the Kupferschiefer from the Sangerhausen Basin, Germany.Chem Geol 2001;176(1):31–49.

Belzile N, Tessier A. Interactions between arsenic and iron oxyhydroxides in lacustrinesediments. Geochim Cosmochim Acta 1990;54:103–9.

Berg M, Tran HC, Nguyen TC, Pham HV, Schertenleib R, Giger W. Arsenic contaminationof groundwater and drinking water in Vietnam: a human health threat. Environ SciTechnol 2001;35:2621–6.

Bhattacharya P, Jacks G, Ahmed KM, Routh J, Khan AA. Arsenic in groundwater of theBengal delta plain aquifers in Bangladesh. Bull Environ Contam Toxicol 2002;69:538–45.

Borgese L, Zacco A, Bontempi E, Colombi P, Bertuzzi R, Ferretti E, et al. Total reflectionof X-ray fluorescence (TXRF): a mature technique for environmental chemicalnanoscale metrology. Meas Sci Technol 2009;20:084027.

Caceres L, Gruttner E, Contreras R. Water recycling in arid regions — Chilean case.Ambio 1992;21:138–44.

Capaccioni B, Aguilera F, Tassi F, Darrah T, Poreda RJ, Vaselli O. Geochemical and isotopicevidences of magmatic inputs in the hydrothermal reservoir feeding the fumarolicdischarges of Tacora volcano (northern Chile). J Volcanol Geoth Res 2011;208(3):77–85.

Casiot C, Lebrun S, Morin G, Bruneel O, Personné JC, Elbaz-Poulichet F. Sorption andredox processes controlling arsenic fate and transport in a stream impacted byacid mine drainage. Sci Total Environ 2005;347:122–30.

Cherry JA, Shaikh AU, Tallman DE, Nicholson RV. Hydrogeochemistry: arsenic speciesas an indicator of redox conditions in groundwater. J Hydrol 1979;43:373–92.

Chiu VQ, Hering JG. Arsenic adsorption and oxidation at manganite surfaces. 1. Methodfor simultaneous determination of adsorbed and dissolved arsenic species. EnvironSci Technol 2000;34:2029–34.

Clavero JE, Solar E, Polanco V, Amigo A. Preliminary seismic and diffuese CO2 flux char-acterization of active volcanoes from the Central Andes of Northern Chile. Proceed-ings IASPEI 2005; General Assembly, Santiago, Chile; 2005.

Clavero, J.E., Gracia, M., Gardeweg, M., 2006. Mapa Geológico del área de Villa IndustrialVisvirí, escala 1:100.000. Servicio Nacional de Geología y Minería, Chile.

Cole JR, Wang Q, Cardenas E, Fish J, Chai B, Farris RJ, et al. The Ribosomal DatabaseProject: improved alignments and new tools for rRNA analysis. Nucleic Acids Res2009;37:D141–5.

Cullen WR, Reimer KJ. Arsenic speciation in the environment. Chem Rev 1989;89(4):713–64.

Davis JA, Kent DB. Surface complexation modeling in aqueous geochemistry. Reviewsin mineralogy, mineral–water interface geochemistry, 23. Mineralogical Societyof America; 1990. p. 177–260.

De Vitre R, Belzile N, Tessier A. Speciation and adsorption of arsenic on diagenetic ironoxyhydroxides. Limnol Oceanogr 1991;36:1480–5.

Del Razo LM, ArellanoMA, Cebrian ME. The oxidation states of arsenic in well-water froma chronic arsenicism area of northern Mexico. Environ Pollut 1990;64:143–53.

Dirección General de Aguas (DGA). Balance Hídrico de Chile. Santiago, Chile: Ministeriode Obras Públicas, Dirección General de Aguas; 1987.

Dixit S, Hering JG. Comparison of arsenic(V) and arsenic(III) sorption onto iron oxideminerals: implications for arsenic mobility. Environ Sci Technol 2003;37:4182–9.

Dold B, Fontboté L. Element cycling and secondary mineralogy in porphyry coppertailings as a function of climate, primary mineralogy, and mineral processing.J Geochem Explor 2001;74(1):3–55.

Donahoe-Christiansen J, D'Imperio S, Jackson CR, Inskeep WP, McDermott TR.Arsenite-oxidizing hydrogenobaculum strain isolated from an acid-sulfate-chloride geothermal spring in Yellowstone National Park. Appl Environ Microbiol2004;70:1865–8.

Drahota P, Rohovec J, Filippi M, Mihaljevic M, Rychlovský P, Cervený V, et al. Mineral-ogical and geochemical controls of arsenic speciation and mobility under differentredox conditions in soil, sediment and water at the Mokrsko-West gold deposit,Czech Republic. Sci Total Environ 2009;407:3372–84.

Drahota P, Filippi M, Ettler V, Rohovec J, Mihaljevic M, Sebek O. Natural attenuation ofarsenic in soils near a highly contaminated historical mine waste dump. Sci TotalEnviron 2012;414:546–55.

Eary LE, Schramke JA. Rates of inorganic oxidation reactions involving dissolved oxygen.In: Melchior DC, Bassett RL, editors. Chemical modeling of aqueous systems II, Wash-ington: ACS symposium series, 30. American Chemical Society; 1990. p. 379–96.

Fuller CC, Davis JA, Waychunas GA. Surface-chemistry of ferrihydrite. 2. Kinetics ofarsenate adsorption and coprecipitation. Geochim Cosmochim Acta 1993;57:2271–82.

Garbarino JR, Bednar AJ, Burkhardt MR. Methods of analysis by the US geologicalsurvey national water quality laboratory—arsenic speciation in natural-watersamples using laboratory and field methods. US Geological Survey Water-ResourcesInvestigations Report 02-4144; 2002.

Garreaud R, Vuille M, Clement AC. The climate of the Altiplano: observed current con-ditions and mechanisms of past changes. Palaeogeogr Palaeocl 2003;194(1):5–22.

502 E.D. Leiva et al. / Science of the Total Environment 466–467 (2014) 490–502

Gihring TM, Banfield JF. Arsenite oxidation and arsenate respiration by a new Thermusisolate. FEMS Microbiol Lett 2001;204:335–40.

Gihring TM, Druschel GK, McCleskey RB, Hamers RJ, Banfield JF. Rapid arsenite oxida-tion by Thermus aquaticus and Thermus thermophilus: field and laboratory investi-gations. Environ Sci Technol 2001;35(19):3857–62.

Goldberg S, Johnston CT. Mechanisms of arsenic adsorption on amorphous oxidesevaluated using macroscopic measurements, vibrational spectroscopy, and surfacecomplexation modeling. J Colloid Interf Sci 2001;234:204–16.

Grafe M, Nachtegaal M, Sparks DL. Formation of metal-arsenate precipitates at thegoethite–water interface. Environ Sci Technol 2004;38:6561–70.

Guerra P, Gonzalez C, Escauriaza C, Bonilla C, Pizarro G, Pasten P. Chemical–hydrodynamiccontrol of arsenic mobility at a river confluence. G-Goldschmidt Abstracts 2012.Mineral Mag 2012;76(6):1788.

Guo H, Wang Y, Shpeizer GM, Yan S. Natural occurrence of arsenic in shallow ground-water, Shanyin, Datong Basin, China. J Environ Sci Heal A 2003;38(11):2565–80.

Haffert L, Craw D. Processes of attenuation of dissolved arsenic downstream fromhistoric gold mine sites, New Zealand. Sci Total Environ 2008;405:286–300.

Hamamura N, Macur RE, Korf S, Ackerman G, Taylor WP, Kozubal M, et al. Linking micro-bial oxidation of arsenic with detection and phylogenetic analysis of arsenite oxidasegenes in diverse geothermal environments. Environ Microbiol 2009;11:421–31.

He ZG, Xiao SM, Xie XH, Zhong H, Hu YH, Li QH, et al. Molecular diversity of microbialcommunity in acid mine drainages of Yunfu sulfide mine. Extremophiles 2007;11:305–14.

Heinzel E, Hedrich S, Janneck E, Glombitza F, Seifert J, Schlomann M. Bacterial diversityin a mine water treatment plant. Appl Environ Microbiol 2009;75:858–61.

Horneman A, van Geen A, Kent DV, Mathe PE, Zheng Y, Dhar RK, et al. Decoupling ofAs and Fe release to Bangladesh groundwater under reducing conditions. Part I:evidence from sediment profiles. Geochim Cosmochim Ac 2004;68:3459–73.

Hug SJ, Canonica L, Wegelin M, Gechter D, Von Gunten U. Solar oxidation and removalof arsenic at circumneutral pH in iron containing waters. Environ Sci Technol2001;35:2114–21.

InskeepWP,MacurRE,HamamuraN,WarelowTP,Ward SA, Santini JM. Detection, diversityand expression of aerobic bacterial arsenite oxidase genes. Environ Microbiol 2007;9:934–43.

Jia Y, Demopoulos GP. Coprecipitation of arsenate with iron (III) in aqueous sulfatemedia: effect of time, lime as base and co-ions on arsenic retention. Water Res2008;42(3):661–8.

Kashyap DR, Botero LM, Franck WL, Hassett DJ, McDermott TR. Complex regulation ofarsenite oxidation in Agrobacterium tumefaciens. J Bacteriol 2006;188:1081–8.

Kay SM, Mpodozis C, Coira B. Magmatism, tectonism, and mineral deposits of the CentralAndes (22°–33°S latitude). In: Skinner BJ, editor. Geology and ore deposits of theCentral Andes, 7. Society of Economic Geology Special Publication; 1999. p. 27–59.

Kent DB, Fox PM. The influence of groundwater chemistry on arsenic concentrationsand speciation in a quartz sand and gravel aquifer. Geochem Trans 2004;5:1–12.

Kent AD, Smith DJ, Benson BJ, Triplett EW. Web-based phylogenetic assignmenttool for analysis of terminal restriction fragment length polymorphism profiles ofmicrobial communities. Appl Environ Microbiol 2003;69:6768–76.

Kocar BD, Inskeep WP. Photochemical oxidation of As(III) in ferrioxalate solutions.Environ Sci Technol 2003;37:1581–8.

Landrum JT, Bennett PC, Engel AS, Alsina MA, Pasten PA, Milliken K. Partitioning geo-chemistry of arsenic and antimony, El Tatio Geyser Field. Chile. Appl Geochem2009;24(4):664–76.

Langner HW, Jackson CR, McDermott TR, InskeepWP. Rapid oxidation of arsenite in a hotspring ecosystem, Yellowstone National Park. Environ Sci Technol 2001;35:3302–9.

Leblanc M, Achard B, Ben Othman D, Luck JM, Bertrand-Sarfati J, Personne JC. Accumu-lation of arsenic from acidic mine waters by ferruginous bacterial accretions (stro-matolites). Appl Geochem 1996;11:541–54.

Lebrun E, Brugna M, Baymann F, Muller D, Lievremont D, Lett MC, et al. Arseniteoxidase, an ancient bioenergetic enzyme. Mol Biol Evol 2003;20:686–93.

Leiva ED, Rios PL, Escauriaza CR, Bonilla CA, Pizarro GE, Pasten PA. Arsenic mobilizationin a high Andean watershed impacted by legacy mining. L-Goldschmidt Abstracts2011. Mineral Mag 2011;75(3):1295.

Lin Z, Puls RW. Potential indicators for the assessment of arsenic natural attenuation inthe subsurface. Adv Environ Res 2003;7(4):825–34.

Macur RE, Langner HW, Kocar BD, InskeepWP. Linking geochemical processes with micro-bial community analysis: successional dynamics in an arsenic-rich, acid-sulphate-chloride geothermal spring. Geobiology 2004;2:163–77.

Manning BA, Fendorf SE, Goldberg S. Surface structures and stability of arsenic(III) ongoethite: spectroscopic evidence for inner-sphere complexes. Environ Sci Technol1998;32:2383–8.

Manning BA, Fendorf SE, Bostick B, Suarez DL. Arsenic(III) oxidation and arsenic(V)adsorption reactions on synthetic birnessite. Environ Sci Technol 2002;36:976–81.

Masscheleyn PH, Delaune RD, PatrickWH. Effect of redox potential and ph on arsenic spe-ciation and solubility in a contaminated soil. Environ Sci Technol 1991;25:1414–9.

McCleskey RB, Nordstrom DK, Maest AS. Preservation of water samples for arsenic(III/V) determinations: an evaluation of the literature and new analytical results.Appl Geochem 2004;19(7):995–1009.

Mondal P, Majumder CB, Mohanty B. Laboratory based approaches for arsenic remedi-ation from contaminated water: recent developments. J Hazard Mater 2006;137:464–79.

Moran AC, Hengst MB, De la Iglesia R, Andrade S, Correa JA, Gonzalez B. Changes inbacterial community structure associated with coastal copper enrichment. EnvironToxicol Chem 2008;27:2239–45.

Morgan J. Chemical equilibria and kinetic properties of manganese in natural water. In:Faust SJ, Hunter JV, editors. Principles and applications of water chemistry. NewYork: John Wiley and Sons Inc.; 1967. p. 561–624.

Mukhopadhyay R, Rosen BP, Phung LT, Silver S. Microbial arsenic: from geocycles togenes and enzymes. FEMS Microbiol Rev 2002;26(3):311–25.

Muñoz N, Charrier R. Uplift of the western border of the Altiplano on a west-vergentthrust system, northern Chile. J S Am Earth Sci 1996;9(3):171–81.

Nickson R, McArthur J, Burgess W, Ahmed KM, Ravenscroft P, Rahman M. Arsenicpoisoning of Bangladesh groundwater. Nature 1998;395:338.

Nordstrom DK. Public health: enhanced: worldwide occurrences of arsenic in groundwater. Science 2002;296:2143–5.

Oremland RS, Stolz JF. The ecology of arsenic. Science 2003;300:939–44.Oremland RS, Hoeft SE, Santini JM, Bano N, Hollibaugh RA, Hollibaugh JT. Anaerobic

oxidation of arsenite in Mono Lake water and by a facultative, arsenite-oxidizingchemoautotroph, strain MLHE-1. Appl Environ Microbiol 2002;68:4795–802.

Oscarson DW, Huang PM, Defosse C, Herbillon A. Oxidative power of Mn(IV) and Fe(III)oxides with respect to As(III) in terrestrial and aquatic environments. Nature1981;291:50–1.

Oscarson DW, Huang PM, Liaw WK, Hammer UT. Kinetics of oxidation of arsenite byvarious manganese dioxides. Soil Sci Soc Am J 1983;47:644–8.

Parikh SJ, Lafeerty BJ, Meade TG, Sparks DL. Evaluating environmental influences onAs-III oxidation kinetics by a poorly crystalline Mn-oxide. Environ Sci Technol2010;44:3772–8.

Pierce ML, Moore CB. Adsorption of arsenite and arsenate on amorphous iron hydroxide.Water Res 1982;16:1247–53.

Pizarro J, Vergara PM, Rodriguez JA, Valenzuela AM. Heavy metals in northern Chileanrivers: spatial variation and temporal trends. J Hazard Mater 2010;181:747–54.

Queirolo F, Stegen S, Mondaca J, Cortes R, Rojas R, Contreras C, et al. Total arsenic, lead,cadmium, copper, and zinc in some salt rivers in the northern Andes of Antofagasta,Chile. Sci Total Environ 2000;255:85–95.

Rhine ED, Phelps CD, Young LY. Anaerobic arsenite oxidation by novel denitrifyingisolates. Environ Microbiol 2006;8(5):899–908.

Richmond WR, Loan M, Morton J, Parkinson GM. Arsenic removal from aqueous solu-tion via ferrihydrite crystallization control. Environ Sci Technol 2004;38:2368–72.

Romero L, Alonso H, Campano P, Fanfani L, Cidu R, Dadea C, et al. Arsenic enrichment inwaters and sediments of the Rio Loa (Second Region, Chile). Appl Geochem 2003;18:1399–416.

Roussel C, Bril H, Fernandez A. Arsenic speciation: involvement in evaluation of envi-ronmental impact caused by mine wastes. J Environ Qual 2000;29:182–8.

Salas R, Kast R, Montecinos F, Salas I. Geología y recursos minerales del Departamento deArica. Provincia de Tarapacá. Boletín, No. 21. Santiago: Instituto de InvestigacionesGeológicas; 1966. [119 pp.].

Sancha A. Full-scale application of coagulation processes for arsenic removal in Chile: asuccessful case study. In: Chappell W, Abernathy C, Calderon R, editors. Arsenicexposure and health effects. Amsterdam: Elsevier; 1999. p. 373–8.

Santini JM, vanden Hoven JM. Molybdenum-containing arsenite oxidase of thechemolithoautotrophic arsenite oxidizer NT-26. J Bacteriol 2004;186:1614–9.

Santini JM, Sly LI, Wen A, Comrie D, De Wulf-Durand P, Macy JM. New arsenite-oxidizing bacteria isolated from Australian gold mining environments—phylogeneticrelationships. Geomicrobiol J 2002;19:67–76.

Schutte UME, Abdo Z, Bent SJ, Shyu C, Williams CJ, Pierson JD, et al. Advances in the useof terminal restriction fragment length polymorphism (T-RFLP) analysis of 16SrRNA genes to characterize microbial communities. Appl Microbiol Biotechnol2008;80:365–80.

Schwenzer SP, Tommaseo CE, Kersten M, Kirnbauer T. Speciation and oxidation kinet-ics of arsenic in the thermal springs of Wiesbaden spa, Germany. Fresenius J AnalChem 2001;371:927–33.

Scott MJ, Morgan JJ. Reactions at oxide surfaces. 1. Oxidation of As(III) by syntheticbirnessite. Environ Sci Technol 1995;29:1898–905.

Shyu C, Soule T, Bent SJ, Foster JA, Forney LJ. MiCA: a web-based tool for the analysis ofmicrobial communities based on terminal-restriction fragment length polymor-phisms of 16S and 18S rRNA genes. Microb Ecol 2007;53:562–70.

Silver S, Phung LT. Genes and enzymes involved in bacterial oxidation and reduction ofinorganic arsenic. Appl Environ Microbiol 2005;71(2):599–608.

Smedley PL, Kinniburgh DG. A review of the source, behaviour and distribution ofarsenic in natural waters. Appl Geochem 2002;17:517–68.

Smedley PL, Nicolli HB, Macdonald DMJ, Barros AJ, Tullio JO. Hydrogeochemistry ofarsenic and other inorganic constituents in groundwaters from La Pampa, Argentina.Appl Gochem 2002;17(3):259–84.

Stern CR, Moreno H, Lopez-Escobar L, Clavero JE, Lara LE, Naranjo JA, et al. Chileanvolcanoes. In: Moreno T, Gibbons W, editors. The geology of Chile. London: TheGeological Society; 2007. p. 147–78.

Stollenwerk KG, Breit GN, Welch AH, Yount JC, Whitney JW, Foster AL, et al. Arsenic attenu-ation by oxidized aquifer sediments in Bangladesh. Sci Total Environ 2007;379:133–50.

Taylor WP. Relationship among geochemical processes and microbial community struc-ture in a unique high-arsenic, sulfidic geothermal spring in Yellowstone NationalPark. [MSc Thesis]Bozeman, MT, USA: Montana State University; 2007 [112 pp.].

Tsai SL, Singh S, Chen W. Arsenic metabolism by microbes in nature and the impact onarsenic remediation. Curr Opin Biotech 2009;20(6):659–67.

Twidwell LG, Robins RG, Hohn JW. The removal of arsenic from aqueous solution bycoprecipitation with iron (III). In: Reddy RG, Ramachandran V, editors. ArsenicMetallurgy. Warrendale, PA: TMS; 2005. p. 3–24.

Wang SL, Mulligan CN. Natural attenuation processes for remediation of arsenic con-taminated soils and groundwater. J Hazard Mater 2006;138:459–70.

WHO. Guidelines for drinking water quality, recommendations. vol. 1. World HealthOrganization; 1993.

Ying SC, Kocar BD, Griffis SD, Fendorf S. Competitive microbially and Mn oxide mediatedredox processes controlling arsenic speciation and partitioning. Environ Sci Technol2011;45:5572–9.