The protocols for obtaining synthetic arsenate-doped 2-line ferrihydrite by adsorption (initial AD samples) or co precipitation (initial CP samples) were adapted from standard methods 1617. Chlorinated salts were preferred over ferric nitrates in order to avoid any use of nitrates as a growth electron acceptor by bacteria. The conventional steps of centrifugation and washing with MilliQ water to remove chlorides were simplified. A first 0.36 M FeCl₃ solution (10 L at pH < 2) was prepared in a continuously stirred reactor and labeled "Solution A", and a second 0.22 M Na₂HAsO₄ solution (600 ml at pH 8.7) was prepared and labeled "Solution B". The initial AD samples were prepared by adding an appropriate amount of 10 M NaOH solution to 5 L of the continuously stirred Solution A so as to obtain a pH of ~6 and precipitate the hydrous ferric oxide (2LFh). The pH was then maintained constant by automatic titration (Mettler DL21) with a 1 M NaOH solution, and equilibrated for 1 hour. The next step was to add 300 mL of Solution B, regulated to pH ~6 by the addition of a 1 M HCl solution, and continuously stir the mixture for 2 hours. The initial CP samples were prepared by adding 300 mL of Solution B to 5 L of continuously stirred Solution A (at pH 2). The co-precipitation of As-2LFh was achieved by neutralizing this mixture to pH 6, using an appropriate amount of 10 M NaOH. The pH was then maintained constant by automatic titration with a 1 M NaOH solution and equilibrated for 2 hours.

In both preparations, the solids were washed twice with 1 L of deionized water and recovered by Büchner filtration. The initial AD and CP solids had a moisture content of about 85–90 wt.% and were stored at 4°C until used. The As/Fe molar ratio in the freeze-dried solids was about 5.5% (Table 1) as measured after hot HCl extraction 18.

The AD and CP experiments consisted in incubating various amounts of As-2LFh in a growth medium both with and without FR. The experiments were conducted under sterile conditions (three 100°C cycles of one hour at 24-hour intervals) in 600 mL plasma flasks under a CO₂ atmosphere at 20 °C with continuous stirring for about two months. For the biotic experiments, 50 mL of FR inoculum were added to 450 mL of a selective liquid medium called CAsR1; this growth medium had been used previously for growing arsenic-reducing bacteria 19 and the composition is described in Additional File 1. All biotic experiments were performed in duplicate. The same experiments with 500 mL of culture medium CAsR1, but without bacteria, served as control.

Experiments AD1 and CP1 were essentially identical and used 3.1 mM Fe and 0.17 mM As (respectively adsorbed or co-precipitated). Experiments AD5 and CP5 used 15.3 mM Fe and 0.84 mM As (respectively adsorbed or co-precipitated; see Table 1).

The supernatant in all the experiments was sampled (10 mL) weekly with a syringe under anaerobic conditions. After filtration at 0.1 μm and acidification with HCl, total iron was analyzed by Atomic Absorption Spectrometry (AAS) (Varian SpectrAA 300 Zeeman) with a detection limit of 0.06 mg L^-1. The Fe(II) concentration was measured colorimetrically with the ferrozine reagent using an Agilent 8453E UV-Visible 20. An aliquot of the filtered sample was immediately passed through an anionic exchange column that retains As(V) but not As(III) 21 and, after acidification, analyzed by AAS with a detection limit of 5 μg L^-1.

The solid concentrations of contained Fe and As were measured after hot HCl extraction 18 and are listed in Table 1. The solid phases were recovered by centrifugation (Jouan CR412, 15 minutes at 6000 rpm) and freeze-dried. The dry solids were maintained under anaerobic conditions until mineralogical characterization.

Microscopic observations and qualitative analyses were performed at the University of Orléans using a Philips CM20 Transmission Electron Microscope (TEM) with a CCD Gatan camera at 200 kV. The TEM samples were prepared by dispersing the powdered samples in alcohol by ultrasonic treatment, dropping them onto a porous carbon film supported on a copper grid, and then drying them in air. Crystalline phases were determined by X-ray powder diffraction (XRD), the XRD data being analyzed using a diffractometer with a Bragg Brentano geometry (Siemens D5000) equipped with a monochromator and based on cobalt K_α1 radiation (λ = 1.78897 Å). Acquisition time for the XRD patterns in the 4–84°2θ interval was set at one second per 0.02°2θ step (= 1 hour) for rotating samples. The crystalline phases were then identified using Diffrac-AT software in conjunction with the Joint Committee on Powder Diffraction Standards (JCPDS) database. Samples were loaded on aluminum plates.

The arsenic oxidation state was measured by X-ray Absorption Near Edge Spectroscopy (XANES), using the procedure described by Morin et al. 22. The data were recorded at the As K-edge (11859 eV) at room temperature in transmission mode on the D44 bending-magnet beamline at the LURE (Orsay, France). A Si(511) double-crystal monochromator yielded an energy resolution of approximately 0.5 eV, the energy being calibrated by a double-transmission setup using an Au foil as reference. Summing 6 scans for each sample yielded reliable signal/noise ratios. Proportions of As(III) and As(V) were determined by linear least squares fitting of the XANES spectra using the spectra of two relevant model compounds, cpp3 and cpp5, consisting of Fe(III)-As(III) and Fe(III)-As(V) amorphous hydroxides, respectively. Absolute accuracy on the As(III)/As(total) ratio was ± 3% 22. According to this calibration procedure, components lower than 5% are not significant. The absence of As(III) oxidation due to the X-ray beam was checked by replicating the spectra at the same points. Taking into account what it is known about the photo-oxidation of As(III) onto iron oxides, only a very weak oxidation was expected in these samples because of the high As/Fe molar ratio (about 5.5%). Moreover, in contrast to the new generation of beamlines, the intensity of the LURE (Orsay) beam was quite low.

Total bacteria were counted under an optical light microscope (Nikon, with magnification ×400) using a Thoma cell. The isolation and identification of pure strains from the FR community with phylogenetic characterization are described in Additional File 1.

When redox reactions proceed under thermodynamic equilibrium, the solubility of an Fe(III) oxyhydroxide can be obtained by measuring the redox potential Eh of the suspension. Taking, for example, pure Fe(III) oxyhydroxide

Fe(OH)₃(s) + 3H⁺(aq) = Fe³⁺(aq) + 3H₂O(l)

and combining this with the redox half-reaction

Fe³⁺(aq) + e^- = Fe²⁺(aq)

we obtain the redox potential Eh (in mV):

E H ( F e ) = 59 × ( log ⁡ ( K s p K r e d ) − 3 p H − log ⁡ a F e 2 + ) MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGacaGaaiaabeqaaeqabiWaaaGcbaGaemyrau0aaSbaaSqaaiabdIeaibqabaGccqGGOaakcqWGgbGrcqWGLbqzcqGGPaqkcqGH9aqpcqaI1aqncqaI5aqocqGHxdaTdaqadaqaaiGbcYgaSjabc+gaVjabcEgaNjabcIcaOiabdUealnaaBaaaleaacqWGZbWCcqWGWbaCaeqaaOGaem4saS0aaSbaaSqaaiabdkhaYjabdwgaLjabdsgaKbqabaGccqGGPaqkcqGHsislcqaIZaWmcqWGWbaCcqWGibascqGHsislcyGGSbaBcqGGVbWBcqGGNbWzcqWGHbqydaWgaaWcbaGaemOrayKaemyzau2aaWbaaWqabeaacqaIYaGmcqGHRaWkaaaaleqaaaGccaGLOaGaayzkaaaaaa@5802@

where K_sp is the solubility product for Eq. (1), K_red (= 10^13.05 at 25 °C) the equilibrium constant for Eq. (2), and a_Fe2+ the activity of aqueous ferrous iron.

When the crystal grain is infinitely large, the solubility product K_sp can approximate the crystal solubility K_so. In the case of small particles with a large surface, the solubility K_so of the particles is highly dependent on the surface energy 23:

log ⁡ K s o = log ⁡ K s p + 2 3 α V 1 ln ⁡ ( 10 ) R T γ S 2 r MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGacaGaaiaabeqaaeqabiWaaaGcbaGagiiBaWMaei4Ba8Maei4zaCMaem4saS0aaSbaaSqaaiabdohaZjabd+gaVbqabaGccqGH9aqpcyGGSbaBcqGGVbWBcqGGNbWzcqWGlbWsdaWgaaWcbaGaem4CamNaemiCaahabeaakiabgUcaRKqbaoaalaaabaGaeGOmaidabaGaeG4mamdaaGGacOGae8xSdeMaemOvayvcfa4aaSaaaeaacqaIXaqmaeaacyGGSbaBcqGGUbGBcqGGOaakcqaIXaqmcqaIWaamcqGGPaqkcqWGsbGucqWGubavaaWaaSaaaeaacqWFZoWzdaWgaaqaaiabdofatbqabaaabaGaeGOmaiJaemOCaihaaaaa@53A9@

where α is a geometric factor of the nucleus (e.g. α = 3 for a sphere), r the radius of a spherical particle, V the molar volume (m³/mol), R the ideal gas constant (J/K/mol), T the temperature (K), and γ_S the mean surface free energy (surface tension in J/m²).

The surface tension γ_S can be regarded as the total surface energy divided by the total surface area of the crystal grain. Although this relationship has only been established for colloidal particles, it is assumed that it also holds for nano-particles. The radius of the particles and the associated size-dependent term of the solubility depend in turn on the procedure followed to synthesize the solid (e.g. on aging time, impurities, etc.). In a suspension containing As-rich ferrihydrite, the redox potential (in mV) imposed by the Fe(III)/Fe(II) couple between pH 5 and 9 is given by:

E H ( F e ) = 59 × ( log ⁡ ( * K s o K r e d ) − 3 p H − log ⁡ a F e 2 + ) MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGacaGaaiaabeqaaeqabiWaaaGcbaGaemyrau0aaSbaaSqaaiabdIeaibqabaGccqGGOaakcqWGgbGrcqWGLbqzcqGGPaqkcqGH9aqpcqaI1aqncqaI5aqocqGHxdaTdaqadaqaaiGbcYgaSjabc+gaVjabcEgaNjabcIcaOiabcQcaQiabdUealnaaBaaaleaacqWGZbWCcqWGVbWBaeqaaOGaem4saS0aaSbaaSqaaiabdkhaYjabdwgaLjabdsgaKbqabaGccqGGPaqkcqGHsislcqaIZaWmcqWGWbaCcqWGibascqGHsislcyGGSbaBcqGGVbWBcqGGNbWzcqWGHbqydaWgaaWcbaGaemOrayKaemyzau2aaWbaaWqabeaacqaIYaGmcqGHRaWkaaaaleqaaaGccaGLOaGaayzkaaaaaa@58DC@

where *K_so is the solubility of As-rich ferrihydrite.

The redox potential imposed by the As(V)/As(III) couple between pH 5 and 6.96 is similarly given by the equilibrium between H₂AsO₄^- and H₃AsO₃:

E H ( A s ) = 59 × ( 1 2 log ⁡ K 1 − 3 2 p H + 1 2 log ⁡ ( a H 2 A s O 4 − a H 3 A s O 3 ) ) if pH < 6.96 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGacaGaaiaabeqaaeqabiWaaaGcbaqbaeqabeGaaaqaaiabdweafnaaBaaaleaacqWGibasaeqaaOGaeiikaGIaemyqaeKaem4CamNaeiykaKIaeyypa0JaeGynauJaeGyoaKJaey41aq7aaeWaaeaajuaGdaWcaaqaaiabigdaXaqaaiabikdaYaaakiGbcYgaSjabc+gaVjabcEgaNjabdUealnaaBaaaleaacqaIXaqmaeqaaOGaeyOeI0scfa4aaSaaaeaacqaIZaWmaeaacqaIYaGmaaGccqWGWbaCcqWGibascqGHRaWkjuaGdaWcaaqaaiabigdaXaqaaiabikdaYaaakiGbcYgaSjabc+gaVjabcEgaNjabcIcaOKqbaoaalaaabaGaemyyae2aaSbaaeaacqWGibasdaWgaaqaaiabikdaYaqabaGaemyqaeKaem4CamNaem4ta80aa0baaeaacqaI0aanaeaacqGHsislaaaabeaaaeaacqWGHbqydaWgaaqaaiabdIeainaaBaaabaGaeG4mamdabeaacqWGbbqqcqWGZbWCcqWGpbWtdaWgaaqaaiabiodaZaqabaaabeaaaaGccqGGPaqkaiaawIcacaGLPaaaaeaacqqGPbqAcqqGMbGzcqqGGaaicqqGWbaCcqqGibascqGH8aapcqaI2aGncqGGUaGlcqaI5aqocqaI2aGnaaaaaa@6E83@

where K₁ (= 10^21.76 at 25°C) is the reduction reaction constant of H₂AsO₄^- to H₃AsO₃.

At pH above 6.96, the same redox potential is given by the equilibrium between HAsO₄^2- and H₃AsO₃:

E H ( A s ) = 59 × ( 1 2 log ⁡ K 2 − 2 p H + 1 2 log ⁡ ( a H A s O 4 2 − a H 3 A s O 3 ) ) if pH > 6.96 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=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@6CCC@

where K₂ (= 10^28.7 at 25°C) is the reduction reaction constant of HAsO₄^2- into H₃AsO₃.

The threshold Fe²⁺ activity (called Fe_cr) for which E_H(Fe) and E_H(As) are equal, is given between pH 5 and pH 9 by:

log ⁡ F e c r = log ⁡ K r e d ∗ K s o K 1 0.5 − 3 2 p H − 1 2 log ⁡ ( a H 2 A s O 4 − a H 3 A s O 3 ) if pH < 6.96 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=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@770A@

log ⁡ F e c r = log ⁡ K r e d ∗ K s o K 2 0.5 − p H − 1 2 log ⁡ ( a H A s O 4 2 − a H 3 A s O 3 ) if pH > 6.96 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=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@7461@

Assuming that the redox reactions proceed under kinetic constraints due to microbial activities rather than thermodynamic equilibrium, the successive use of terminal electron acceptors can be rationalized in terms of the redox reaction energy yields 24. When initial E_H(As) is lower than initial E_H(Fe) then As-rich ferrihydrite is initially more susceptible to bacterial reduction than aqueous As(V) 12. The bacterial reductive dissolution of ferrihydrite could first occur according to the reaction:

0.125CH₃COO^- + Fe(OH)₃(s) + 1.875H⁺ → 0.25HCO₃^- + Fe⁺⁺ + 2.5H₂O

at a rate R₁ given by the law:

R 1 = ν max ⁡ N [ F e ( I I I ) ] K m + [ F e ( I I I ) ] R P F R P F = 1 − a F e 2 + F e c r MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=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@6314@

where ν_max is the maximum specific Fe(III) reduction rate per cell, N the cell density for the biomass 1 in cells mL^-1, K_m the half saturation constant in mM, [Fe(III)] the substrate concentration in mM, and RPF the Redox Potential Factor with Fe_cr the maximum value of Fe²⁺(aq) activity calculated by Eq. (8) or (9).

The first term in the equation (11) predicts that DIRB will continue to metabolize until the concentration of the Fe(III) substrate vanishes (Monod empirical law). The second term of this rate law (RPF) gives an energetic explanation for a threshold phenomenon 2526: when the RPF is lower than a threshold value (e.g. RPF = 0.1), the bacterial reduction of As(V) in solution could also occur according to the redox reaction:

0. 25CH 3 COO − + H 2 AsO 4 − + 0.75 H + → 0.5 HCO 3 − + H 3 AsO 3 ° if pH < 6.96 0.25 CH 3 COO − + HAsO 4 2 − + 1.75 H + → 0.5 HCO 3 − + H 3 AsO 3 ° if pH > 6.96 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=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@9FC3@

at a rate R₂ which is equal to zero if RPF > 0.1 and which otherwise is given by:

R 2 = ν max ⁡ ∗ N ∗ ( H 2 A s O 4 − ) K m ∗ + ( H 2 A s O 4 − ) R P F ∗ if pH < 6.96 R 2 = ν max ⁡ ∗ N ∗ ( H A s O 4 2 − ) K m ∗ + ( H A s O 4 2 − ) R P F ∗ if pH < 6.96 R P F ∗ = ( 1 − a H 3 A s O 3 A s max ⁡ ) MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xI8qiVKYPFjYdHaVhbbf9v8qqaqFr0xc9vqFj0dXdbba91qpepeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGacaGaaiaabeqaaeqabiWaaaGcbaqbaeqabeGaaaqaauaabeqaciaaaeaacqWGsbGudaWgaaWcbaGaeGOmaidabeaakiabg2da9GGaciab=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@AD23@

where ν*_max is the maximum specific As(V) reduction rate per cell, N* is the cell density for the biomass 2 in cells mL^-1, K*_m is the half saturation in mM, (H₂AsO₄^-) and (HasO₄^2-) are the concentrations of As(V) in mM, and RPF* is the Redox Potential Factor with As_max a maximum activity of H₃AsO₃.

It should be noted here that bacterial reaction in the model is assumed to proceed only in the forward direction; the case in which aqueous Fe(II) decreases after an initial increase, which should stop As(V) reduction (RPF > 0.1), is not be considered.

Interactions between aqueous species and the solid are described by a two-layer surface complexation model 27 with two types of adsorption site being considered, ≡(s)FeOH and ≡(w)FeOH for strong and weak sites, respectively. The two sites are assumed to have equal acid-base intrinsic constants and different affinities for sorbate ions (Table 2). Recommended site densities used for HFO, which has a surface area of about 600 m²g^-1, are 0.005 mol/mol Fe for ≡(s)FeOH and 0.2 mol/mol Fe for ≡(w)FeOH 27. These densities are based on maximum sorption densities, although there are substantial discrepancies in estimates of the maximum sorption density on HFO for arsenic. Adsorption maxima of 0.31, 0.4 and 0.6 mol As (mol Fe)^-1 have been achieved for As(III), and 0.25 mol As (mol Fe)^-1 for As(V) 2829. It should be noted that Dixit and Hering 30 recently observed an increase of 0.4 mol As(III) per mol Fe(II) adsorbed. Here we used a value of 0.4 mol/mol Fe for weak sites concentration in the modeling of our high As/Fe molar ratio experiments and 0.2 mol/mol Fe in the modeling of Bengali sediments experiments with a lower As/Fe molar ratio. This sorption site concentration decreases linearly with the solid Fe concentration.

Modeling geochemical speciation, sorption and microbial reactions was done with the program REACT included in the Geochemist's Workbench^® Release 4.0.3 31. Activity coefficients in REACT were calculated using the Debye Hückel law (at an ionic strength of about 0.04 M). Aqueous reaction constants used for thermodynamic calculations are the values included in the LLNL version 8, release 6 dataset (called "thermo.com.V8.R6+.dat" by REACT), except for some arsenate and arsenite complexes listed in Table 2. The protonation and ferrous complexation constants of arsenate were recommended by Whiting (1992) as a result of a literature research 32. The protonation constants of arsenite, being consistent with the sorption data calculated by Dixit and Hering 28, were extracted from the MINEQL V4.5 database 33. The equilibrium sorption constants used in REACT are based on data reported by Dzombak and Morel (1990) 27, apart from more recent data for arsenate As(V) and arsenite As(III) 28 and for ferrous iron Fe(II) and carbonate species 34, which are listed in Table 2.

The XRD diagrams show that As-2LFh synthesized in the laboratory corresponds to 2-line ferrihydrite, clearly characterized by two diffuse bands at 2.55 and 1.45 Å. The As-2LFh sterilization step in the culture medium at 100 °C had no measurable effect on the crystallinity of the solid. Moreover, most of the solids after two months of incubation were still 2-line ferrihydrite. Vivianite Fe(II)₃(PO₄)₂.8H₂O was also clearly identified by X-ray diffraction analysis, but to a lesser extent and only after the second month of experiments AD1 and CP1 with FR (Table 1). A close association between vivianite and 2-line ferrihydrite in experiment CP1 with FR was noted from Scanning Transmission Electron Microscopy observations (Fig. 1). No green-rust or magnetite was identified, but bobierrite Mg₃(PO₄)₂.8H₂O was characterized in experiment AD1.

Iron concentrations in the aqueous phase were monitored during the incubations under strict anaerobiosis at initial pH ~6, both with and without inoculation of FR. Because FR cell suspensions were maintained with Fe(III) as growth electron acceptor (See Additional File 1), the inoculum was not free of Fe(II); consequently there was an initial concentration of aqueous Fe(II) in the incubation experiments with FR bacteria (Fig. 2). In addition, ferrous iron Fe(II) was the predominant valence state of iron at the beginning of incubation in the experiments with FR.

Fe release was observed only in samples inoculated with FR, and not in the control experiments (Fig. 3). Moreover, the kinetics of Fe(III) reductive dissolution occurred in two phases during the two-month period of incubation: a first phase of increasing dissolved iron concentration Fe(II) during the first month (Figs. 2 and 3), associated with bacterial growth (Fig. 2) and a decrease of Eh (Fig. 4); a second phase, during the second month, marked by a stabilization of aqueous Fe concentrations in experiments AD5 and CP5 and even by a decrease in experiments AD1 and CP1. The maximum Fe release was about 3% in experiments CP5 and AD5 and 8–10% in experiments CP1 and AD1 (Fig. 3). The initial reduction rate was higher in the CP than the AD experiments (Table 3), but there was no difference in the percentage of released Fe at the end of the experiments (Fig. 3).

Analytical results indicate that about the same concentration of As was solubilized in pentavalent form within few hours after each incubation with and without FR. This initial solubilization of about 3 mg L^-1 (~40 μM in Table 4), which represents about 5% of the total As quantity for experiments AD5 and CP5 and 25% for AD1 and CP1, seems not to depend on the initial solid concentration of As-2LFh (Fig. 2). So as to distinguish between abiotic and biotic release of As, and also to facilitate comparison between As and Fe release, the proportions of As and Fe release have been calculated in Figure 3 by subtracting the As and Fe concentrations after initial equilibrium from the concentrations measured over time, and normalizing the differences to the total quantity of As and Fe (Table 1).