Introduction

Arsenic (As) is a highly toxic metalloid which is widespread in the environment and causes severe and numerous health problems worldwide. Presence of As in soils, sediments and water is attributed to natural sources, such as weathering of minerals with high As content, and to human activities (use of arsenical fertilizers and pesticides, as well as industrial and mining activities) (Smedley and Kinniburgh 2002). The mobility and toxicity of As strongly depend on its chemical form (Oremland and Stolz 2003). AsIII, which is the predominant form in reduced environments, is more mobile and is considered more toxic than AsV, which is the predominant form in oxic environments. In turn, inorganic As (iAs) is generally recognized as more toxic than the organic forms (Sharma and Sohn 2009), with the exception of the methylated AsIII species (Petrick et al. 2000; Styblo et al. 2000).

In aquatic environments, As may undergo transformations in its chemical form as a consequence of changes in environmental physico-chemical conditions, and interaction with mineral surfaces (oxidation/reduction, surface complexation), but also through biologically mediated reactions (bio-transformation), which can strongly affect its fate, mobility, bioavailability and toxicity. Arsenic speciation and bioavailability in water and sediments are strongly influenced by biological activities (Oremland and Stolz 2005). Studying the effect of algae on As speciation, Hellweger and Lall (2004) proposed a model in which algae absorb AsV (mistaking it for phosphate). Then, inside the cell, an AsV detoxification mechanism takes place, which consists in reducing AsV to AsIII by methylating it to monomethylarsonic acid (MMAV), then MMAV to dimethylarsinic acid (DMAV). Finally, As is excreted as AsIII and/or DMAV, depending on the algal growth rate and the phosphate conditions (Hellweger and Lall 2004). Besides MMAV and DMAV, the products of methylation include trimethylarsine oxide (TMAO) (Duker et al. 2005) and the final product of the methylation pathway, the volatile trimethylarsine (TMAsIII) (Yin et al. 2011a).

At water–sediment interfaces, biofilms consisting of microorganisms (bacteria, fungi, cyanobacteria, algae and protozoa) embedded in extracellular polymeric substances (EPS) mainly composed of polysaccharides are commonly found (Costerton 2007). Consequently, biofilms are the first component to interact with dissolved substances such as nutrients, organic matter, metals and metalloids, as well as other toxicants in aqueous systems (Sabater et al. 2007).

It has been demonstrated that biofilms play a key role in retention of metals and metalloids from overlying water (Nelson et al. 1996, 1999; Friese et al. 1997; Headley et al. 1998; Decho 2000; Dong et al. 2000, 2007; Haack and Warren 2003; Morris et al. 2005; Serra et al. 2009; Beck et al. 2011; Drahota et al. 2014). Pollutants are removed by the biofilm through a variety of mechanisms, including (bio-)sorption, precipitation as sulphides or phosphates and microbial reductive precipitation (van Hullebusch et al. 2003). Biosorption consists of several mechanisms, including ion exchange, chelation, adsorption and diffusion through cell walls and membranes (van Hullebusch et al. 2003).

The interaction between As and biofilms in aquatic systems can be studied from two points of view: As behaviour or its effects on biofilms. Firstly, As biogeochemistry may be modified by the presence of biofilms. The potential for As enrichment in biofilms has been reported by Drewniak et al. (2008) with concentrations of up to 60 mg kg−1 in rock biofilm. Yang et al. (2011) demonstrated that multi-species biofilms, inoculated from a source receiving coal mining effluent, could both sequester and detoxify Se and As. Drahota et al. (2014), studying As adsorption from natural As sources onto natural surface coatings growing on glass slides, confirmed that dissolved As had been sorbed and retained by the biofilm. Tuulaikhuu et al. (2015), who investigated the fate and toxicity of As on periphytic and epipsammic biofilms using a simplified fluvial system including fish, biofilms and sediment, reported that periphytic biofilms also accumulated As, although it was predominantly retained by sediment. Regarding studies into the retention of As by epipsammic biofilms, Prieto et al. (2013) observed that epipsammic biofilms increased AsV sorption with respect to biofilm-devoid river sediments, with a more noticeable effect in the presence of phosphate. Secondly, presence of As in aquatic systems could affect periphyton communities. Thus, arsenate is responsible for inhibiting algal growth and photosynthetic capacity, as well as for decreasing total biofilm biomass, changing community composition (selecting tolerant species, reducing species richness and making biofilms more heterotrophic) and reducing diatom cell sizes and the ability of the community to use phosphorus (Blanck and Wangberg 1988, 1991; Wangberg et al. 1991; Rodríguez Castro et al. 2015; Tuulaikhuu et al. 2015; Barral-Fraga et al. 2016).

Despite this evidence, literature on (bio-)sorption, speciation and detoxification of As by epipsammic biofilms is scarce. Hence, the novelty of the present work is its contribution to understanding the role played by epipsammic biofilms in As biogeochemistry. The main objective is to study the influence of biofilms developed on riverbed sediments on the (bio-)adsorption and/or (bio-)uptake, mobility, (bio-)transformation and detoxification of AsV at microcosm scale using specifically designed systems to control As concentration and speciation over time. The kinetics of the removal process is also evaluated, along with changes in As aqueous speciation, As volatilization, the distribution of As species within biofilms and remobilization of previously retained As. The effect of phosphate as a potential competitor for As sorption and bio-uptake and as a nutrient for biofilm growth is also explored.

Materials and methods

Sediment characterization

Sediment was sampled from Anllóns River (NW Spain), where As contamination is a problem in some sections and where gold mining activities carried out during the Roman Empire, and in the early 20th century, brought about remobilization of associated arsenic and its accumulation in sediments (Devesa-Rey et al. 2008, 2011; Costas et al. 2011). It has been shown that As mobilization in Anllóns riverbed sediments is enhanced under conditions of high salinity, extreme pH or high P concentration, as well as during high-flow resuspension events (Rubinos et al. 2010, 2011).

For this study, sediment was sampled in an uncontaminated area upstream of the Au–As mineralized area known as Ponte de Eguas (43°13′24.46″N 8°45′44.61″W), which is located 8 km downstream from the Town of Carballo (population over 25,000). A complex sample was collected using a small plastic shovel from the top 5 cm at various points at this site and taken to the laboratory in hermetic plastic containers filled with river water to prevent oxidation. Eh and pH in the sediment were measured in situ using a portable device (Hanna Instruments, HI 9025 microcomputer). Eh values obtained with Pt–Ag/AgCl electrodes were corrected to refer them to the standard hydrogen electrode (SHE) by adding 245 mV.

Once in the laboratory, solid sediment samples were freeze-dried and sieved (<2 mm) for characterization. Only some organic debris was eliminated by sieving, so the <2 mm fraction used for the experiment practically represented bulk sediment. The grain size distribution was determined using wet sieving and the pipette method as described by Guitián and Carballas (1976). Total phosphorus (PT) was determined by means of acid digestion (HF:H2SO4:HCl 10:1:10) followed by colorimetric determination (Murphy and Riley 1962). Total organic carbon, nitrogen and sulphur content were analysed using a LECO TruSpec CHNS analyser. Major and trace constituents were determined by X-ray fluorescence spectrometry. The accuracy of the XRF measurements was checked by using certified reference materials. For instance, in the specific case of As, the As concentration (mg/kg) obtained for the reference material BCRCRM-277b was 45.4 ± 4.1 (certified value 47.3 ± 1.6). X-ray powder diffraction was used for semiquantitative mineralogical analysis of the mineral phases present in the sediment.

River water characterization

River water was collected and filtered using 0.45-μm cellulose nitrate membrane filters NCS 045 47 BC (ALBET LabScience, Dassel, Germany) to be used as a biofilm growth medium in the laboratory in order to mimic natural conditions for biofilm development. pH, Eh and electrical conductivity (EC) were measured in situ using portable electrodes (Hanna Instruments, HI 9025 and HI 9033, respectively). Soluble P was measured by means of acid digestion with H2SO4 followed by colorimetric determination with ammonium molybdate (APHA 2005). Total N was determined by segmented flow analysis and colorimetry with a Futura console (AMS Alliance) after filtration through a 0.45-µm membrane Millex-HM (Millipore). Nitrate was measured by ion chromatograph (model Metrohm 850 Professional IC), and ammonia was determined by ion-selective electrode (ISE; Thermo Scientific 9512BNWP). Alkalinity was measured by colorimetric determination using an AquaKem 250 analyser (Thermo Scientific, Waltham). Dissolved organic carbon (DOC) was measured using a total organic carbon analyser (model TOC-5000, Shimadzu, Kyoto). With this equipment, the DOC concentration is obtained by subtracting the inorganic carbon (IC) concentration from the total carbon (TC) concentration. TC was determined by the 680 °C combustion catalytic oxidation method, whereas IC was determined by acidification and sparging. The carbon dioxide generated in both determinations was detected using a non-dispersive infrared (NDIR) gas analyser. Na and K concentrations were measured by atomic emission spectrometry, and Ca and Mg by atomic absorption spectrometry (Spectra AA 220 FS, Varian Inc., Palo Alto). Total Fe, Mn and As concentration was measured by inductively coupled plasma mass spectrometry (ICP-MS, 820MS Varian Inc., Palo Alto), whereas As species were analysed using HPLC-ICP-MS as described below in “Arsenic analyses” section.

Native biofilm growth

The sediment sample (500 g, 31 % water content) was incubated in bioreactors filled with 2 L of filtered As-free water, equipped with systems for air supply, sample collection and volatilized arsenic trapping (Fig. 1). All materials were previously sterilized by autoclaving at 121 °C for 30 min. The flasks were maintained for 3 weeks in an incubation chamber under optimal controlled conditions of light (day/night cycles, 12 h of light, intensity ca. 40 µmol photons m−2 s−1), temperature (20 °C) and air supply (ca. 1 L min−1). The water was replaced weekly to provide the necessary nutrients for biofilm growth. After 3 weeks, an appreciable biofilm layer was clearly perceived on the sediment surface. Epipsammic biofilms, covering sediments in the Anllóns River, are mainly constituted by Bacillariophyceae, which represent >86 % of the total abundance in superficial sediments (Martiñá Prieto et al. 2016). At the Eguas sampling site, Cocconeis, Navicula and Karayevia were identified as the predominant genera. Epipsammic biofilm inocula from this river have been satisfactorily incubated at laboratory scale in experimental fluvial channels, reaching maximum growth at 2–3 weeks, as indicated by chlorophyll-a and soluble carbohydrates contents (Prieto et al. 2016).

Fig. 1
figure 1

Scheme of experimental setup

Full size image

Arsenic retention and speciation

Stock solution of 1000 mg L−1 AsV was prepared by dissolving Na2HAsO4·7H2O (Panreac, Spain) in Milli-Q water. Once the biofilm had developed, an aliquot of AsV solution was spiked, resulting in an initial 500 μg L−1 (6.67 µmol L−1) As exposure concentration. This As concentration, which exceeds the USEPA’s aquatic life criteria maximum concentration (CMC) (acute exposure) for As in freshwater, set at 340 µg L−1 (USEPA 2014), was selected because it was below the effective AsV concentration that produces a 20 % decrease (EC20) in the bioluminiscence of Aliivibrio fischeri model bacterium, which was set at 2.0 ± 0.6 mg L−1 by Rubinos et al. (2014) using the Microtox® acute toxicity screening bioassay and constitutes a measurable threshold of As toxicity, while still enabling detection of quantifiable changes in the As species concentration in the water column. The effect of phosphate on AsV retention and speciation was also studied in identical systems, using an equimolar As:P concentration ratio. P was added to the systems from stock solution of 1000 mg L−1 P, prepared by dissolving Na2HPO4·7H2O (Panreac, Spain) in Milli-Q water. The systems incorporating biofilm were labelled BAS and BASP, for the systems with and without added phosphate, respectively. In parallel, identical systems without biofilm (sediment and river water sterilized by three cycles of autoclaving at 120 °C for 30 min) were developed as reference systems (controls) for experiments, with and without phosphate (named CAS and CASP, respectively). The systems were maintained under the conditions indicated above for two additional weeks, during which aliquots (5 mL) of the water column were sampled daily, using single-use sterile polypropylene (PP) syringes (Braun Inkjet, B Braun AG, Melsungen). The samples were immediately filtered (sterile 0.45-μm Whatman Puradisc 25AS™ syringe filters, GE Healthcare Europe GmbH, Barcelona) and stored frozen (−80 °C), until analysis of total As and its species [AsIII, AsV, MMAV, DMAV, arsenobetaine (AsB)] by ICP-MS and HPLC-ICP-MS, respectively. Relative standard deviations for total As analysis by ICP-MS were <3 %. Additionally, dissolved P, Fe and Mn concentrations were measured (determined by ICP-MS) as well as dissolved organic carbon (DOC) (as previously determined by TOC analyser). At the end of the experiment, the As speciation in the sediment interstitial waters of the systems was also determined. Samples (10 mL) of interstitial water were filtered, frozen at −80 °C and analysed for As species as described above. The pH, Eh and sulphate concentration were also measured in the overlying water of the different systems at the end of the experiment, as described in “Sediment and river water characterization” section. All determinations were carried out in triplicate to ensure the quality of the values obtained.

Arsenic analyses

Analysis of dissolved As species was carried out using high-performance liquid chromatography together with inductively coupled plasma spectrometry (HPLC-ICP-MS). A Varian Prostar 230 HPLC, equipped with a guard column and a Hamilton PRP-X100 anion exchange column (4.1 × 250 mm and 10 µm), was used to separate five primary As species (AsV, AsIII, MMAV, DMAV and AsB) using a 13 min gradient LC method with 12.5 and 30 mM (pH 9) (NH4)2CO3 as mobile phase, flow rate of 1 mL min−1 and injection volume of 50 µL. For quantification, a Varian 820-MS ICP-MS, equipped with collision reaction interface (CRI) technology to reduce polyatomic interferences, was used. Relative standard deviations for total As analysis by ICP-MS were <3 %. The detection limits under the experimental conditions were 2.8, 4.1, 2.9, 4.6 and 2.5 ng L−1 for AsV, AsIII, MMAV, DMAV and AsB, respectively.

Volatilized arsenic

To quantify volatilized As, arsines were trapped using the AgNO3-based chemo-trapping approach described by Mestrot et al. (2009) and Yin et al. (2011a). In this method, arsine (AsH3), monomethylarsine (MeAsH2), dimethylarsine (Me2AsH) and trimethylarsine (TMAs or Me3As) react with AgNO3 and are preserved by oxidation to their pentavalent oxy-species (AsV, MMAV, DMAV and TMAO) (Mestrot et al. 2009).

To prepare the traps, silica gel (2.5–5 mm) was submersed overnight in 5 % HNO3 (w/v) solution and washed with Milli-Q water (18.2 MΩ cm−1 resistivity), then impregnated with 10 % AgNO3 (w/v) solution and placed overnight in an oven at 70 °C (covered with aluminium foil to avoid photodecomposition of AgNO3) (Yin et al. 2011a). Subsequently, the silica gel (~1 g) was loaded in the trap tubes (10-mL sterilized syringe) and held in place by a small quantity of glass wool washed QP (Panreac, Barcelona) at both ends. Trap tubes were again covered with aluminium foil to avoid photodecomposition of AgNO3 and coupled to the flask systems.

At the end of the experiment, 5 mL of 1 % (v/v) hot boiling HNO3 was used to elute the collected As in the trap tubes (Yin et al. 2011b). Eluates were filtered (0.45 µm) and frozen (−80 °C), and total As was measured by ICP-MS.

Remobilization and bioavailability of arsenic

Immediately after the retention, the overlying water was removed and the potential remobilization of As species was assessed by washing the As-loaded sediment–biofilm or sediment, with As-free filtered (0.45 µM) Anllóns River water (1:10 solid:liquid ratio). The aqueous extracts were filtered (0.45 µm), frozen (−80 °C) and analysed for As by HPLC-ICP-MS.

To evaluate As bioavailability, the diffusion gradient in thin films (DGT) technique was used. DGT devices, purchased from DGT Research Ltd. (Lancaster), accumulate metals on a binding agent after they have passed through a well-defined diffusive layer (Davison and Zhang 1994; Zhang and Davison 1995; Zhang et al. 1995). DGT incorporating Fe-oxide gels have been specifically developed for assessing As bioavailability (Stockdale et al. 2008, 2010; Luo et al. 2010).

Arsenic-specific DGT devices were placed onto the sediment surface of all the studied systems for 24 h to afford an operationally defined measure of the “bioavailable” fraction of As. Once removed from the systems, the devices were rinsed with Milli-Q water and opened for removal of the resin gels, which were then eluted with 1 mL of 7.2 M HNO3 for at least 24 h to allow complete extraction of As from the resin. An aliquot from the sample tube was pipetted and diluted 6 times with Milli-Q water prior to analysis by ICP-MS. The mass of As in the resin gel (M), the time-averaged DGT concentrations (CDGT) and the flux (F) of As measured by DGT were calculated according to Zhang and Davison (1995) and DGT® technical documentation.

To determine time-averaged DGT concentrations, the diffusion coefficient of total As was calculated from the diffusion coefficient of individual species using Eq. 1:

$$D_{\text{AsT}} = \mathop \sum \limits_{i = 1}^{5} x_{i} D_{i},$$
(1)

where x i and D i are the fraction of individual As species over the total As and the diffusion coefficient of the individual As species, respectively.

Values for the diffusion coefficient of AsIII and AsV have been reported in the ranges of 5.9–10.1 and 4.9–6.8 × 10−6 cm2 s−1, respectively (Fitz et al. 2003; Panther et al. 2008; Österlund et al. 2010, 2012; Bennett et al. 2010; Luo et al. 2010; Moreno-Jiménez et al. 2013), whereas diffusion coefficient values of 6.10 and 6.30 × 10−6 cm2 s−1 have been reported for MMAV and DMAV, respectively (Österlund et al. 2012).

Extracellular and intracellular arsenic

Extracellular As in the biofilms was determined using the extraction procedure described by Levy et al. (2005) consisting in phosphate extraction, followed by extraction of intracellular As by the method described by Miyashita et al. (2009), consisting of extraction of As in methanol–water from lysed cells. To this end, after exposure to AsV, biofilm samples (ca. 1 g) were gently harvested and rinsed with 10 mL of filtered Anllóns River water (1:10 solid:liquid ratio). Then, the solid phases were submitted to two washing cycles with 10 mL of 0.1 M KH2PO4/K2HPO4 buffer solution (pH 5.95) to extract any extracellular As. The suspensions were shaken for 30 s and allowed to stand for 20 min, then they were centrifuged (3000 rpm, 15 min), the supernatant was filtered (0.45-µm syringe filters), and the extraction cycle was repeated. The eluates of two washes were combined, frozen (−80 °C) and analysed for arsenic species (HPLC-ICP-MS). The remaining solid phases were gently washed with Milli-Q water (18.2 MΩ cm−1 resistivity), centrifuged (3000 rpm, 15 min) and frozen (−20 °C). For determination of intracellular As, solid phases were thawed and intracellular As was extracted with 10 mL of methanol/H2O (1:1, v/v) solution. After standing for 10 min, the suspensions were sonicated for 10 min and centrifuged at 3000 rpm for 15 min. Extraction was repeated twice with 5 mL of methanol/H2O (1:1, v/v) solution. The extracts were combined and evaporated under vacuum to dryness using a rotavapor (Büchi Rotavapor R-200, BÜCHI Labortechnik GmbH, Essen). The dried extracts were redissolved with Milli-Q water (18.2 MΩ cm−1 resistivity), filtered (0.45 µm), frozen (−80 °C) and analysed for As species (HPLC-ICP-MS). At the end of the sequential extractions, solid phases were dried at 105 °C to constant weight to determine the dry weight of the analysed samples. All determinations were carried out in triplicate.

Results and discussion

Sediment and river water characterization

The main physico-chemical properties of the bed sediments and river water are presented in Table 1. The sediment had slightly acidic pH (6.3) and Eh of 235 mV, which is indicative of a suboxic state (ranging between 100 and 400 mV at around neutral pH) of the surface layer, and showed a predominance of the sandy fraction. Semiquantitative mineralogical analysis of the mineral phases showed quartz as the predominant mineral, with about half of the total abundance, followed by microcline and hornblende. The presence of this highly weatherable mineral can be explained because, at this sampling site, the Anllóns River runs over basic rocks, namely peridotite, pyroxenite, amphibolite and serpentinite, and is indicative of a short transport distance of the sediments. The total organic matter content was 1.5 %, whereas P and N concentrations were 370 and 253 mg kg−1, respectively. The C/N ratio was 35.4, which is indicative of organic matter (OM) rich in lignin and cellulose as well as poor in N, attributable to terrestrial origin (Lamb et al. 2006). The total P concentration was lower than the lowest effect level (LEL) established by the Ontario sediment quality guidelines (Persaud et al. 1993), set at 600 mg kg−1. This level of pollution is expected to have no effect on the majority of sediment-dwelling organisms, and the sediment is considered clean to marginally polluted. The total As concentration of the sediment was 15 mg kg−1, which is lower than the values detected in the sampling campaign performed by Devesa-Rey et al. (2008), falling within the range of 33–264 mg kg−1. The As concentration slightly exceeded the “Effects Range-Low” (ERL) fixed by Long et al. (1995) at 8.2 mg kg−1, which represents the upper end of a range of concentrations at which effects would rarely be observed.

Table 1 Physico-chemical properties of river water and bed sediments
Full size table

River water exhibited neutral pH (7.0), low alkalinity (23.3 mg L−1) and low concentrations of P (0.05 mg L−1), nitrate (1.59 mg L−1 as NO −3 -N) and ammonia (0.09 mg L−1 as NH +4 -N). These parameter values classify the river water as having high ecological status because they are below the limits for this status for Spanish rivers in Atlantic and Cantabrian watersheds, fixed at 0.07, 2.26 and 0.16 mg L−1 for P, nitrate (as NO −3 -N) and ammonia (as NH +4 -N), respectively (BOE 2015). The As concentration in the river water was 3.6 µg L−1, mainly in the form of AsV (92 %) with the remainder as AsIII. This concentration is in the range of those previously detected in Anllóns River fresh water by Costas et al. (2011) (0.16–3.96 µg L−1) and lower than the recommended permissible As level in drinking water, set at 10 µg L−1 by the World Health Organization (WHO 1993).

Arsenic retention

The As concentration in the overlying water decreased during the time course of the experiment down to a nearly constant final value. The removal of As was higher in the presence of biofilm (Fig. 2). After 14 days of exposure, the As retention in the BAS system was 91 % ({As}max = 32.5 μmol kg−1) in comparison with the initial concentration in water (500 µg L−1), but it only reached 70 % in CAS ({As}max = 27.8 μmol kg−1). Data were satisfactorily fitted (R 2 > 0.94) to the empirical exponential equation given by Eq. 2.

$$\left[ {\text{As}} \right] = a + b {\text{e}}^{(-t/c)},$$
(2)

where [As] is the As concentration (µg L−1) at time t (days) and a, b and c are adjustment parameters. The equilibrium aqueous As concentration in BAS, defined by the a parameter of the fit, was 46.71 µg L−1, 3.6 times lower than in the system without biofilm (CAS) (167.4 µg L−1). These results are in agreement with those obtained by Prieto et al. (2013), who studied As removal by biofilm in batch experiments and reported an 18 % increase in average As retention for systems incorporating biofilm in comparison with sediments without biofilm.

Fig. 2
figure 2

Evolution of As concentration in overlying water throughout the experiment for BAS and CAS systems

Full size image

In the environmental conditions of this study, a significant reduction of As concentration was achieved in the presence of biofilm, revealing its importance in natural aquatic systems and potential for application in biotechnological systems for water purification. Thus, the final concentrations in the BAS and CAS systems were lower than the USEPA’s aquatic life criteria maximum concentration (CMC), fixed at 340 µg L−1 in surface waters (USEPA 2014), but only the BAS value was lower than the criteria continuous concentration (CCC), fixed at 150 µg L−1. CMC and CCC are the highest concentrations in surface waters to which an aquatic community can be exposed briefly or indefinitely, respectively, without resulting in an unacceptable effect. The BAS final concentration was also lower than the permissible limit for irrigation water (100 µg L−1) (FAO 1985), although slightly higher than the environmental quality standards (EQS) for As in inland surface waters, set at 25 µg L−1 by the Priority Substances Directive in Surface Waters [S.I. No. 272/2009-European Communities Environmental Objectives (Surface Waters) Regulations 2009]. This EQS represents a threshold for annual average concentration of As in surface waters to ensure protection against long-term exposure to pollutants in an aquatic environment.

The epipsammic biofilm not only enhanced the retained As amount but also the retention rate. Thus, BAS exhibited higher initial retention rates (3.84 µmol As kg−1 day−1 at 7 days) than CAS (2.97 µmol As kg−1 day−1 at 7 days) up to 7 days, after which both values were similar.

The presence of the equimolar P concentration in the BASP system slightly enhanced the amount and rate of As removal, reaching 97 % of its initial concentration ({As}max = 36.3 μmol kg−1). The effect was lower in the absence of biofilm (CASP), where As removal reached only 69 % ({As}max = 26.1 μmol kg−1) (Fig. 3). In terms of water quality, the equilibrium As concentration in BASP determined by the exponential equation was 18.85 µg L−1, which is lower than the CMC and CCC values and than the FAO permissible limit for irrigation water but still higher than the EQS. The equilibrium As concentration in CASP was exactly the same as the value mentioned above for CAS (167.4 µg L−1). With respect to retention rates, BASP also exhibited higher values (4.92 µmol As kg−1 day−1) than CASP (3.28 µmol As kg−1 day−1) up to 7 days, after which the values for both systems were similar.

Fig. 3
figure 3

Evolution of As concentration in overlying water throughout the experiment for BASP and CASP systems

Full size image

The soluble P decreased from 50 to 12 µg L−1 in BAS, while in BASP (to which P was added), it decreased from 220 to 23 µg L−1 (Fig. 4). The reduction of the P concentration was attributed to its consumption by the microorganisms composing the biofilm. In CAS, the soluble P concentration remained fairly constant, between 100 and 150 µg L−1, throughout the experiment, whereas in CASP it decreased in the first 3 days from an initial concentration of 312 to 100–150 µg L−1, which can be considered the equilibrium concentration in both systems.

Fig. 4
figure 4

Evolution of P concentration in overlying water throughout the experiment for BAS, CAS, BASP and CASP systems

Full size image

Arsenic speciation

Five soluble As species were detected in the overlying water in the systems incorporating biofilm exposed to As-enriched solution. During the 2 weeks of the experiment, their concentrations followed the order: AsV ≫ AsIII > MMAV ≈ DMAV > AsB. For BAS, aqueous As was mostly (~98 %) in AsV form, whereas AsIII reached only 1.2 % of total As, and MMAV and DMAV represented only 0.6 and 0.7 % of total As, respectively, indicating slight (bio-)methylation by the epipsammic biofilm during the incubation period. In the absence of the biofilm, the speciation of As in the water column changed significantly. In this system, AsIII accounted for up to 39 % (80 μg/L) of the final aqueous As in CAS (Fig. 5a), with the remainder in the form of AsV, with no methylated forms present. The concentration of AsIII for the systems incorporating phosphate (BASP and CASP) coincided with the results observed for the systems without phosphate (Fig. 5b), while methylated aqueous species were again only detected in the presence of biofilm (Fig. 6a, b). The detection of methylated As species in the BAS and BASP systems suggests some occurrence of bio-methylation in the presence of biofilm, albeit insignificant, during the short running time of our study. Similar concentrations of AsIII were found in CASP and in CAS. In both control systems, AsIII was detected in the overlying water after 48 h of exposure. To investigate this fact, physico-chemical water conditions were investigated, as well as the behaviour of compounds susceptible to promote AsV reduction. The pH and Eh values in the overlying water at the end of the experiment showed that both sediment–biofilm and control systems were under near-neutral and oxic conditions (Table 2). There was no evidence of sulphur oxidation, as sulphate concentrations were similar at the end of the experiment for all the systems and similar to the initial concentrations in the river water. Also, Fe concentrations were similar throughout the experiment in all systems (Fig. 7a, b). On the other hand, in the systems without biofilm, Mn concentrations increased up to day 4 and then remained constant at values much higher (up to 1500-fold) than those observed in the systems with biofilm, in which Mn concentrations decreased over time (Fig. 7c, d). Similarly, higher DOC concentrations were measured for CAS (mean value 55.3 ± 4 mg L−1) in comparison with BAS (mean value of 14.0 ± 3.8 mg L−1). These results suggest that oxidation–reduction processes are occurring in the systems without biofilm, as explained in “General discussion” section. Biofilms seem to inhibit reduction of added AsV, by covering the mineral surfaces of the sediments and thus hindering their interaction with aqueous As and maintaining an oxygenated biofilm–sediment interface.

Fig. 5
figure 5

AsIII concentration in overlying water throughout the experiment for BAS and CAS systems (a) and for BASP and CASP systems (b)

Full size image
Fig. 6
figure 6

DMAV concentration in overlying water throughout the experiment for BAS and CAS systems (a) and for BASP and CASP systems (b)

Full size image
Table 2 Values of pH, Eh and sulphates at the end of the experiment
Full size table
Fig. 7
figure 7

Fe and Mn total concentrations in overlying water throughout the experiment for BAS and BASP (a and c, respectively) and for CAS and CASP (b and d, respectively)

Full size image

As in interstitial water

Similarly to what occurred in the overlying water, As concentrations in interstitial water were notably higher (at least 9 times) in the CAS and CASP systems, with a higher percentage of AsIII, than in the BAS and BASP systems, where detectable concentrations of DMAV were also found (Table 3).

Table 3 Total As and As species concentrations in interstitial water at the end of the experiment
Full size table

Volatilized arsenic

Volatile As forms were only detected in systems incorporating biofilm. This fact, together with the detection of methylated As species, may indicate the occurrence of detoxification processes. It has been reported that some microorganisms (bacteria, fungi and algae) form arsine in order to decrease intracellular As, as a (bio-)transformation pathway to cope with As toxicity (Wang et al. 2014). The mean amount of volatilized As after 14 days of exposure was 17.0 ng As, and the bio-volatilization rate was 2.4 ng of volatilized As per mg of added AsV and day. This value is higher than those calculated from reported data (0.28–0.54 ng mg−1 day−1) by Yin et al. (2011a) for Microcystis, Nostoc and Synechocystis cyanobacteria, treated with either 7.5 or 30 mg L−1 AsV for 6 weeks, but falls within the range of those reported (1.3–9.3 ng mg−1 day−1) by Yin et al. (2011b) for As biotransformation by the protozoan Tetrahymena thermophila after 48 and 72 h of exposure to 150–3000 µg L−1 AsV concentrations. With respect to the initial As concentration, the maximum percentage of volatilized As was only 0.002 %. This value is slightly lower than the percentage (<0.1 %) calculated by Yin et al. (2011a) and than the percentages (0.006–1 %) reported by Yin et al. (2011b). In summary, these results may suggest that the epipsammic biofilm carried out detoxification processes with production of volatilized As, although analysing the global As balance, its contribution was of minor relevance.

Remobilization and bioavailability of arsenic

Regarding As remobilization, the release to water of previously retained As was low and varied between 113.0 ± 9.4, 166.8 ± 10.0, 181.1 ± 9.1 and 176.5 ± 7.9 µg kg−1 for BAS, CAS, BASP and CASP, respectively, representing 4.7, 6.1, 8.7 and 9.0 % of the retained As, respectively. In the presence of biofilm, 94 % of the released As was AsV, with only 3 % being AsIII. In contrast, AsIII accounted for 21 % of the released As in the systems without biofilm, where the concentration of remobilized AsIII was 10 times higher than in the systems with biofilm. MMAV was only detected (3.3 % of total As released) in the extracts from the systems with biofilm, while DMAV was not found in any case.

To estimate As bioavailability, the concentrations of As species were determined in the eluates from the DGT devices and are presented in Table 4, jointly with the parameters calculated from these concentrations. Once again, the values obtained were similar for the different systems. It is worth highlighting that the speciation of the DGT extracts revealed that the percentage of AsV was higher than 77 % in all cases, and DMAV was only detected in the BAS samples.

Table 4 Total As and As species concentrations in the extracts of DGT and DGT parameters
Full size table

Extracellular and intracellular arsenic

In biofilm-enriched samples obtained from the BAS and BASP systems, 71.1 ± 1.5 % of the retained As was extractable with 0.1 M KH2PO4/K2HPO4 phosphate buffer, being attributed to extracellular As (Levy et al. 2005), most of it (>99.6 %) in the form of AsV. This fraction may represent an estimation of easily leachable As (Gleyzes et al. 2002). DMAV was identified in phosphate extracts, revealing once again the occurrence of a (bio-)methylation process by the epipsammic biofilm. Regarding intracellular As, extractable with methanol/H2O (1:1, v/v) solutions (Miyashita et al. 2009), the systems with biofilm revealed 28.9 ± 1.5 % of As inside the cells, almost exclusively (>99.8 %) in the form of AsV.

General discussion

In this study, the influence of epipsammic biofilm on the behaviour of AsV in freshwater environments was evaluated. The results reveal that biofilms increase the amount and rate of AsV retention by the sediment. Prieto et al. (2013) also observed, in batch experiments with short-term (24 h) As exposure, that epipsammic biofilm increased AsV sorption by sediments from the Anllóns River and that this effect was more noticeable in the presence of phosphate.

Sediments may retain As due to their content of Fe, Al and Mn (oxy)hydroxides (Oscarson et al. 1981; Jiang et al. 2005), clay mineral (Manning and Goldberg 1997) and organic matter (Thanabalasingam and Pickering 1986). Additionally, the favourable effect of biofilm on As removal may be attributed to the addition of two combined effects: (bio-)sorption and (bio-)accumulation. Firstly, As (bio-)sorption is improved because the biofilm may increase the specific surface area, and the number of sorption sites and consequently As sinks. It has been observed that sorption of metals by biofilms is governed by their interactions with the biofilm matrix, constituted by cells and extracellular polymeric substances (EPS). EPS are mainly constituted by polysaccharides and proteins, as well as nucleic acids, lipids or humic substances, and EPS molecules contain ionizable functional groups which can sequester toxic compounds (van Hullebusch et al. 2003). Moreover, Fe and Mn oxides precipitated as biominerals in biofilms have been identified as responsible for retention of trace elements (Dong et al. 2000; Warren and Haack 2001). Drahota et al. (2014) reported enhancement of As retention by precipitation of poorly crystalline biogenic Mn oxides, in turn induced by growth and accumulation of biofilms. Secondly, AsV (bio-)accumulation takes place when AsV enters cells via phosphate transporters (Páez-Espino et al. 2009). In our study, the extracellular fraction or As biosorption represents the main compartment of the retained As in the systems with biofilm.

Addition of equimolar phosphate concentration had no effect on the As adsorption in the CASP control system. Although phosphate usually competes with arsenate for sorption sites in soils and sediments due to their chemical similarities (Manning and Goldberg 1996; Hongshao and Stanforth 2001), there was no competition between As and P in CASP; the reason may be that phosphate rapidly decreased in the solution (a reduction of 63 % in the first 3 days) while As retention continued up to day 10. The amount of As retained in the control systems was only slightly lower than the maximum adsorption capacity (2.0 µg g−1) determined using a Langmuir model by Prieto et al. (2013) for sediments from the same site (6.6 µg g−1), which could explain the high As concentration that remained in the overlying water of the CAS and CASP control systems.

The presence of phosphate had a slight positive effect on the retention of As by the epipsammic biofilm. This effect could be due to the stimulative effect of phosphate on the growth of the microorganisms, as proved by the almost complete absorption of phosphate by the biofilm, which is supposed to contribute to its growth. Other possible explanations are the increase in the efficiency and/or number of phosphate/arsenate cellular transporters or to the alleviative effect of phosphate against AsV toxicity. This latter effect was observed by Karadjova et al. (2008) for the green microalga Chlorella salina in sea water, by Wang et al. (2013) for two freshwater green algae, and by Rubinos et al. (2014) for A. fischeri bacterium.

The biofilm inhibited the reduction of added AsV, which may be due to the coverage of the mineral surfaces of the sediments, thus preventing their interaction with aqueous As, and/or to the maintenance of a more oxygenated interface due to photosynthesis. Several key factors are reported in literature to control As speciation, such as pH and Eh (Smedley and Kinniburgh 2002), presence of redox pairs such as Fe3+/Fe2+, Mn4+/Mn2+, SO 2−4 /HS− (Cherry et al. 1979), and the effect of natural organic matter (NOM) (Sharma and Sohn 2009). Among all the conditions evaluated, DOC and Mn concentrations were notably higher in the systems devoid of biofilm. Lower DOC concentrations in BAS and BASP could be attributed to DOC release from the sediment, which was inhibited in the presence of biofilm, or to DOC consumption by biofilm heterotrophs. This fact is relevant because an important role in the geochemical behaviour of As is attributed to NOM (Buschmann et al. 2006; Klitzke and Lang 2009). Thus, it has recently been demonstrated that addition of arsenic to DOM solutions results in arsenate reduction as well as arsenite oxidation (Redman et al. 2002). On the other hand, the release of Mn2+ is attributed to the reduction and dissolution of Mn oxyhydroxides of the sediment. Therefore, the results of this study suggest that dissolution and reductive processes are occurring in CAS and CASP, involving AsV reduction and Mn mobilization jointly with NOM, which were inhibited in both the BAS and BASP systems. This inhibition may come about because of the ability of autotrophic microorganisms, which make up the biofilm, to oxygenate the water–sediment interface and hence avoid reductive processes. The inhibition of AsV reduction by the biofilms has relevant geochemical and toxicological implications, since AsIII is usually considered more mobile and toxic than AsV (Sharma and Sohn 2009; Huang 2014).

Arsenic methylation has been demonstrated in different aerobic and anaerobic microorganisms (Kuehnelt and Goessler 2003). The presence of methylated As species in the overlying water, in the interstitial water and within the biofilm in BAS and BASP are indicative of (bio-)methylation processes taking place. Among the methylated As compounds, DMAV was the main species detected in the overlying and interstitial water. MMAV was detected at lower concentrations. This fact could be explained because MMAV is an intermediate in the As methylation process, with a rapid intracellular metabolism and ten times lower permeability to membranes than DMAV (Cullen et al. 1994a, b, c). Higher DMAV concentrations compared with MMAV have also been found in natural environments such as in a German forested catchment by Huang and Matzner (2007), in eutrophic and mesotrophic lakes by Hasegawa et al. (2009) and in marine sediments by Fauser et al. (2013).

In summary, epipsammic biofilms play a key role in the biogeochemistry of As in river environments, by enhancing removal of AsV from water and by strongly affecting As speciation. Biofilms inhibit the occurrence of aqueous AsIII in the water column, which is attributed to their potential ability to oxygenate the biofilm–sediment interface and to carry out detoxification processes inside the cells via production of methylated and volatilized species.

Conclusions

Epipsammic biofilms covering riverbed sediments enhance the removal and retention rate of AsV from the water and also strongly affect the speciation of As in the water column by inhibiting the occurrence of aqueous AsIII, due to their ability to oxygenate aquatic environments and to bring about DOC consumption by biofilm heterotrophs. This fact has noteworthy toxicological and geochemical relevance, for example, for remediation purposes, considering the frequent greater toxicity and mobility of AsIII species. The detoxification of As driven by epipsammic biofilms is supported by the presence of methylated arsenic species such as MMAV and DMAV in the overlying water and within the biofilm, as well by the detection of volatilized arsenic. The enhancement of the retention and toxicity of As in the presence of biofilms highlights their importance in natural aquatic systems and their potential for application in biotechnological water purification systems.