Impact of Nitrate on the Structure and Function of Bacterial Biofilm Communities in Pipelines Used for Injection of Seawater into Oil Fields

Abstract

We studied the impact of NO₃⁻ on the bacterial community composition, diversity, and function in in situ industrial, anaerobic biofilms by combining microsensor profiling, ¹⁵N and ³⁵S labeling, and 16S rRNA gene-based fingerprinting. Biofilms were grown on carbon steel coupons within a system designed to treat seawater for injection into an oil field for pressurized oil recovery. NO₃⁻ was added to the seawater in an attempt to prevent bacterial H₂S generation and microbially influenced corrosion in the field. Microprofiling of nitrogen compounds and redox potential inside the biofilms showed that the zone of highest metabolic activity was located close to the metal surface, correlating with a high bacterial abundance in this zone. Upon addition, NO₃⁻ was mainly reduced to NO₂⁻. In biofilms grown in the absence of NO₃⁻, redox potentials of <−450 mV at the metal surface suggested the release of Fe²⁺. NO₃⁻ addition to previously untreated biofilms induced a decline (65%) in bacterial species richness, with Methylophaga- and Colwellia-related sequences having the highest number of obtained clones in the clone library. In contrast, no changes in community composition and potential NO₃⁻ reduction occurred upon subsequent withdrawal of NO₃⁻. Active sulfate reduction was below detection levels in all biofilms, but S isotope fractionation analysis of sulfide deposits suggested that it must have occurred either at low rates or episodically. Scanning electron microscopy revealed that pitting corrosion occurred on all coupons, independent of the treatment. However, uniform corrosion was clearly mitigated by NO₃⁻ addition.

Biofilms are matrix-embedded communities of microorganisms attached to an interface (usually solid-liquid). They represent an important mode for bacterial growth in natural, medical, and industrial ecosystems (18, 24, 57). Their appearance can be either beneficial or detrimental. In engineered systems, such as wastewater treatment, cooling water systems, and food processing, biofilms cause problems for the particular operation (20). In the oil industry, deoxygenated seawater is often injected into reservoirs for secondary oil recovery and pressure maintenance. Biofilms adherent to surfaces within these injection systems and in the reservoir harbor diverse communities of bacteria. They cause significant operational problems due to fouling, corrosion of iron and steel alloys, reduction of flow, souring (contamination of oil and gas by bacterially generated sulfide), and reservoir plugging (20, 61). Biofilms may provide niches for sulfate-reducing bacteria (SRB) that form hydrogen sulfide (H₂S) from sulfate-containing seawater. H₂S is toxic, corrodes surfaces in pipelines and other equipment, and reduces the quality of oil and gas, therefore causing great financial losses worldwide.

Biocides are frequently used to control the growth of biofilms in such systems, particularly to inhibit SRB and other corrosive microorganisms. However, the effect of biocides is limited, and due to environmental risks, their use is restricted. A more environmentally sound and potentially cost-efficient approach based on the addition of NO₃⁻ or NO₂⁻ to injection waters (i.e., waters used for injection into oil and gas fields) to control souring has been applied in the field with variable results (16, 33, 60, 65, 66). Several mechanisms have been proposed for NO₃⁻/NO₂⁻-mediated souring control: (i) outcompetition of SRB by NO₃⁻/NO₂⁻-reducing bacteria (NRB) for available electron donors; (ii) inhibition of SRB by NO₃⁻ reduction products (i.e., NO₂⁻ and NO) from heterotrophic NRB and/or autotrophic NR sulfide-oxidizing bacteria; (iii) switch of some SRB from SO₄²⁻ reduction to dissimilatory NO₃⁻ reduction to ammonium (DNRA); (iv) removal of H₂S through H₂S oxidation with NO₃⁻ as electron acceptor by NR sulfide-oxidizing bacteria (15, 19, 27, 41, 44, 50).

NO₃⁻/NO₂⁻ treatment of oil field reservoirs may also prevent or mitigate microbially influenced corrosion (MIC) in the production facilities, including pipelines (64). The impact of this treatment and the long-term consequences for MIC are poorly understood. Field investigations on NO₃⁻/NO₂⁻ efficacy for souring and corrosion control have previously focused on planktonic communities, cell suspensions, or debris instead of substrate-attached intact biofilms (34, 64). However, maintaining the structural integrity of biofilms is necessary for understanding mass transfer in the biofilms and cell-cell and cell-environment interactions, as well as the processes and conditions causing corrosion problems owing to MIC. Local concentrations of NO₃⁻ and its reduction products inside the biofilms are unknown but are essential for evaluating the effectiveness of the treatment, as is the in situ impact of NO₃⁻ on a complex bacterial community.

We investigated biofilms derived from an injection system that prepared anoxic seawater. NO₃⁻ was added to the water that was transported via a 9-km-long subsea pipeline to the injection well, where it was pumped into the field. Until injection commenced 4 years prior to the study, no significant amounts of H₂S had been detected in the injection water. However, since SRB had previously been found in biofilm debris obtained from the injection pipeline (34), it was important to ascertain whether those or other bacteria continued to grow even in the presence of NO₃⁻ and if bacteria or the treatment itself induced corrosion in the injector and pipelines. Microsensor profiling was combined with ¹⁵N and ³⁵S labeling, 16S rRNA gene-based fingerprinting, and sulfur isotope fractionation in order to (i) link structure and function in the injection water biofilms; (ii) identify dominant bacteria therein; (iii) assess the impact of short- and long-term NO₃⁻ exposure on the bacterial community composition, diversity, and function; and (iv) identify potential corrosion damage and deposits underlying the biofilms.

MATERIALS AND METHODS

Biofilm establishment and system conditions.

Biofilms were grown inside an offshore water injection system being treated with NO₃⁻ at an oil field in the North Sea. Removable carbon steel coupons (0.2-cm² surface area) made from the same structural pipeline material were mounted in Robbins devices (RD) to facilitate in situ sampling of biofilms grown on the coupons. The steel was marine grade, low-alloy mild steel (AISI 1018) typically containing 0.15 to 0.2% carbon, 0.6 to 0.9% manganese, 0.05% sulfur, and 0.04% phosphor (by weight). RD were installed as side-stream attachments, in parallel with the pipes of the system, under continuous flow at 2 to 3 liters/min and 18 to 20°C. They were positioned downstream and upstream of the NO₃⁻ dosing point (Fig. 1) to allow direct comparison between the NO₃⁻-amended (NA) biofilms and the biofilms grown without NO₃⁻ (control). The pressure levels in the up- and downstream RD were 11 × 10⁵ and 248 × 10⁵ Pa, respectively. The NA and control biofilms were grown over 4 months prior to on-site measurements with microsensors, incubations with [¹⁵N]NO₃⁻ and [³⁵S]SO₄²⁻, and sampling for DNA extraction. After initial sampling, the RD were exchanged; thus NA biofilms no longer received NO₃⁻ [hereafter, NA(−) biofilms], whereas control biofilms then received NO₃⁻ [hereafter, control(+) biofilms]. During the 4 days following this exchange, these biofilms [NA(−) and control(+)] were monitored daily to study the biological and biogeochemical responses under changed conditions. Although the target NO₃⁻ dosage in the injection water for field treatment was 1 mM, dosage irregularities caused NO₃⁻ levels to vary between 0 and 1.3 mM during the entire trial period. The injection water was filtered (40-μm pore size) and deoxygenated in a Minox system (Minox Technology AS, Norway) (Fig. 1a). Briefly, O₂ was stripped out of the water with N₂ gas, which was subsequently recovered by reacting O₂ with methanol in presence of a catalyst to H₂O, CO₂, and CO. The diffusion of CO₂ into the water phase lowered the pH from pH 8.0 to pH 6.8 to 7.3. Treated water had a salinity of 36‰ (as measured on a refractometer) and contained 27 mM SO₄²⁻.

FIG. 1.

(a) Positions of RD in the seawater deoxygenation (DO) system. O₂ was removed from the water in a two-stage gas-stripping process after filtration. (b) Schematic diagram of the experimental treatments. Black and white bars indicate exposure to NO₃⁻-rich (1 mM; NA biofilms) and NO₃⁻-free (control biofilms) seawater, respectively. RD were exchanged to opposite conditions on day 123. ¹⁵N and ³⁵S tracer incubations were done on days 123 and 128 (hatched bars).

Microprofiling.

On each of the 4 days after the preincubation period, one biofilm-covered coupon was collected from the RD and placed in a continuous flow cell. Fresh injection water was flushed with argon to remove dissolved oxygen and kept at 18 to 20°C and pH 8.0 (to buffer against acid production); the water was circulated from a medium tank (6 liters) through the flow cells with a velocity of 1 cm s⁻¹. Depth profiles were measured with microsensors for NO_x (sum of NO₃⁻, NO₂⁻, and N₂O), NO₂⁻, N₂O, redox potential (against an Ag/AgCl reference), pH, H₂ (detection limit of 0.1 μM), and H₂S (detection limit of 1 μM) in biofilms amended with 1 mM NaNO₃⁻ (1, 28, 35, 46, 53). Profiles of H₂S, redox potential, pH, and H₂ were measured in biofilms receiving no NO₃⁻. The profiles were recorded using automated profiling software (μ-Profiler; L. Polerecky, Max Planck Institute for Marine Microbiology, Bremen, Germany) in 10- to 25-μm intervals at a vertical angle of 20°. Areal fluxes of NO₃⁻ and NO₂⁻ through the diffusive boundary layer were calculated using Fick's first law as described previously (21). Diffusion coefficients used for NO₃⁻ and NO₂⁻ in the seawater were 16.1 × 10⁻⁶ and 15.3 × 10⁻⁶ cm² s⁻¹, respectively (37).

¹⁵N incubations and analysis.

The rates of NO₃⁻ reduction were measured by ¹⁵N stable isotope incubations using [¹⁵N]NO₃⁻ as a tracer, and the ¹⁵N products (N₂O, N₂, and NH₄⁺) were determined by mass spectrometry. Biofilm coupons were transferred to 6-ml Exetainers (Labco, High Wycombe, Buckinghamshire, United Kingdom), which were then completely filled with NO₃⁻-free anoxic injection water amended with 485 μM Na¹⁵NO₃ (98 atom%; Sigma Aldrich, Germany) and 485 μM Na¹⁴NO₃ in order to obtain maximum sensitivity for the isotope pairing method (45). The Exetainers were sealed, excluding air bubbles. The first set of samples, NA and control(+) biofilms, were incubated for 2.5, 4, 6, 15, 27, and 41 h (Fig. 1). In the second set of incubations, NA(−) and control biofilms were incubated for 7 h after being spiked with NO₃⁻ label (Fig. 1). Negative controls for each series were prepared and processed in a similar way as the samples, except that they received fresh coupons without biomass. The incubation was performed at 21 to 23°C in the dark, and bacterial activity was stopped using 1.5 ml of zinc acetate (20%, wt/vol). Rate calculations were based on the initial 7 h of the experiments for the comparison of the two incubation series. Liquid phase ¹⁴⁺¹⁵NH₄⁺ was quantified by converting NH₄⁺ with hypobromite iodine (69) to N₂ and measured in the same way as N₂ and N₂O. External ¹⁴⁺¹⁵NH₄⁺ standards were used for calibration. Gas samples were extracted from a 1-ml helium headspace and injected into a gas chromatograph mass spectrometer (VG Optima, Micromass, United Kingdom) for the N₂ and N₂O isotope composition analysis.

³⁵S tracer incubation and analysis.

Exetainers containing single biofilm coupons were filled with in situ water enriched with [³⁵S]SO₄²⁻ tracer (Amersham Bioscience) (final total activity of 100 kBq) and incubated in the dark at room temperature. The incubation was stopped after 2, 4, 6, 12, 24, and 41 h by adding 2 ml of zinc acetate (20%, wt/vol). SO₄²⁻ reduction rate determinations were done using a cold chromium distillation procedure (29), and SO₄²⁻ was quantified by nonsuppressed anion chromatography (Waters 510 pump, Waters IC-Pak anion exchange column, and Waters 430 conductivity detector) (eluent, 1 mM isophthalic acid; flow, 1.0 ml min⁻¹).

Chemical analyses.

NO₃⁻ and NO₂⁻ from liquid samples were quantified on a chemiluminescence detector (Thermo Environmental Instruments, Franklin, MA) as described previously (6). The dissolved organic carbon (DOC) and organic acid contents of in situ water were quantified by high-temperature combustion using a DOC clone (49) and ion chromatography, respectively.

DAPI staining.

After being embedded in 22-oxyacalcitriol cryo-compound (Leica Microsystems GmbH, Wetzlar, Germany), biofilms were removed from the metal surface as a whole block of frozen 22-oxyacalcitriol. Sections 5 μm thick were made using a cryo-microtome (Microm HM 505 E), immobilized on glass slides, and dehydrated in an ethanol series (50, 80, and 96%; 3 min each). DAPI (4′,6′-diamidino-2-phenylindole) was added to each section, and samples were incubated for 10 min in the dark and then examined on a Zeiss Axioplan microscope (Carl Zeiss, Jena, Germany).

Corrosion analyses and S isotope composition.

A paraffin-impregnated graphite electrode was used to identify solid iron sulfide compounds in biofilm samples by applying voltammetry of microparticles (48). Following removal of the biofilm (see above), the steel surfaces of coupons from the NA and control biofilms were examined by scanning electron microscopy ([SEM] Hitachi SEM 4004) equipped with energy dispersive X-ray analysis (EDAX PV 9900 UTW). The quantity and stable sulfur (S) isotope composition of the reduced S fraction in the steel coupons (duplicate samples comprised of steel shavings) and from pipeline debris obtained from areas adjacent to the RD receiving no NO₃⁻ (in triplicates) were determined. The acid-volatile sulfide (AVS) and chromium-reducible sulfur (CRS) fractions were determined by two-step HCl-hot chromous chloride distillations (8). H₂S was trapped as zinc sulfide in 5% zinc acetate solution, and an aliquot of the zinc sulfide suspension was used for sulfide determination according to the method of Cline (9). The remaining zinc sulfide was converted to AgS for stable S isotope analysis. S isotope ratios (³⁴S/³²S) on the AVS and CRS fractions were analyzed by combustion isotope ratio monitoring mass spectrometry (5). AgS samples were combusted to SO₂ in an elemental analyzer (EuroVector) and introduced into a ThermoElectron Finnigan MAT Delta⁺ gas mass spectrometer via a ThermoElectron Finnigan MAT Conflo II interface. Stable isotope ratios of ³⁴S/³²S are given in the δ-notation with respect to the SF₆-based Vienna-Canyon Diablo troilite (V-CDT) standard (11): δ³⁴S(‰) = (R_sample/R_V-CDT − 1) × 10³, where R is ³⁴S/³²S.

Molecular analysis.

Biofilm material for denaturing gradient gel electrophoresis (DGGE) and 16S rRNA gene clone library construction was obtained either from scraping pipelines adjacent to the RD or by flushing the whole RD with sterilized in situ water. Sampling took place after 4 months of growth and again after 4 days of exposure to opposite conditions. All biofilms (biomass of 300 to 500 mg each) were subjected to nucleic acid extraction (bead beating) as described earlier (40). PCR was performed using 50 ng of the DNA extracts in the presence of the bacterial domain-specific primers GM5F (with GC clamp) and 907RC (42). The PCR was run using a hot-start touchdown program to minimize nonspecific amplification. The annealing temperature was decreased from 65 to 55°C every two PCR cycles, after which 12 PCR cycles were performed at 55°C. Equal amounts (1,000 ng) of triplicate PCR products were loaded on the DGGE gel. DGGE was carried out using a Bio-Rad D-Code system run at 60°C at a constant voltage of 200 V for 3.5 h. For cloning, 16S rRNA genes were PCR amplified using the primers GM3F and GM4R (42) at an annealing temperature of 42°C for 30 cycles. PCR products were purified using a QIAquick PCR purification kit (Qiagen, Hilden, Germany), and 200 ng was cloned using a TOPO TA cloning kit (version E; Invitrogen, Karlsruhe, Germany) according to the manufacturer's instructions. Obtained clones were screened by checking their PCR products by gel electrophoresis. The PCR products of the inserts were purified on Sephadex G-50f columns (Amersham Bioscience AB, Uppsala, Sweden) and then sequenced. Sequences were analyzed using ARB software (39) and were inserted into the reconstructed tree by applying parsimony criteria without allowing changes in topology. The number of operational taxonomic units was determined by calculating similarity matrices among the sequences in the ARB database. Sequences with >97% similarity were defined as operational taxonomic units. Rarefaction curves were calculated using the program Analytic Rarefaction (version 1.3, Stratigraphy Lab, University of Georgia; [http://www.uga.edu/∼strata/software]). The coverage of the clone libraries, species richness, species evenness, and Shannon-Weaver index of diversity were calculated for each of the four clone libraries, as reported elsewhere (2, 22, 62).

RESULTS

Biofilm morphology and bacterial distribution.

The NA and control biofilms grown under in situ conditions for 4 months were visibly different. NA biofilms were fluffy, brownish-yellow, and 150 to 200 μm thick (Fig. 2a). On microscopic examination, the biofilms could be divided into five structurally distinct zones (Fig. 2d and e). DAPI-stained biofilm sections showed that most bacterial cells in the NA biofilms were clustered in a 50-μm-thick horizon near the bottom. This zone was characterized by a sponge-like structure containing voids and channels (Fig. 2d and e, zones 2 and 3). Cells were more sporadically located in a 10-μm-thick crust-like layer directly attached to the metal surface (Fig. 2e, zone 1), while no cells were found in the upper part of the biofilm (Fig. 2e, zones 4 and 5). The control biofilms were ash gray to black, compact, and 50 to 80 μm thick (Fig. 2g), with very low biomass in the 10-μm-thick crust on the metal surface (Fig. 2j and k, zone 1). The biofilms increased in thickness (two to fourfold) during fixation and/or storage (Fig. 2), but the overall structural features were preserved.

FIG. 2.

NA (a to f) and control (g to l) biofilms visualized by different approaches. Fresh coupons are shown in panels m and n. Binocular images show complete carbon steel coupons before (a and g) and after (b and h) removal of biofilm cover and after exposure of biofilms to sulfide (c and i). The phase-contrast microscopy images of 5-μm-thick biofilm cross-sections show structurally distinct zones, indicated by the numbers (d and j). Epifluorescence microscopy was applied to the cross-sections from the phase-contrast microscopy (e and k). Positive DAPI signals appear in white. SEM (backscattered electron) images show vertical coupon cross-sections. The arrows indicate MnS inclusions associated with pits (l) and surrounded by organic material inside pits (f).

Biofilms were then exposed to the opposite growth conditions for 4 days. NA(−) biofilms were found to be visibly patchier than NA biofilms. Zones 4 and 5 disappeared, and cells were evenly distributed across zones 1 to 3. In the control(+) biofilms cells were more abundant in zone 1 than before the NO₃⁻ addition. Some additional cells also appeared in zone 2 (data not shown).

Spatial distribution of microenvironments and metabolic activity.

Microsensor profiles showed clear functional differences among NA and control biofilms (Fig. 3). In the NA biofilms, NO₃⁻ (0.96 mM in water phase) penetrated throughout the entire biofilm and decreased to 0.86 mM close to the metal surface (Fig. 3a). The nearly linear curve of the NO₃⁻ profiles inside the biofilm indicated that most NO₃⁻ consumption occurred in a ∼50-μm-thick zone close to the metal surface (at a depth of 150 to 200 μm), i.e., below the maximum depth reached by the microsensor. In this zone, up to 50 μM NO₂⁻ was produced from the NO₃⁻. No clear redox gradients were observed in the NA biofilms, nor was H₂S, H₂, or N₂O detected. In contrast, the control biofilms exhibited strong redox gradients, ranging from 0 mV in the water to −450 mV on the metal surface (Fig. 3b). Nevertheless, neither H₂S, H₂, nor N₂O was detected in these samples either.

FIG. 3.

(a and b) Microprofiles of NO₃⁻, NO₂⁻, redox potential (RP), and pH in NA and control biofilms grown in situ on carbon steel coupons for 4 months. (c and d) The situation 4 days after biofilm exchange to opposite NO₃⁻ conditions, designated NA(−) and control(+) biofilms. H₂S, H₂, and N₂O were not detected in any biofilm. Zero depth indicates the biofilm-water boundary; dashed lines represent the estimated metal coupon surface position. Averages and standard deviations were calculated from three profiles measured across the biofilm.

After the RD were exchanged, both biofilm types showed metabolic responses within 4 days (Fig. 3c and d). In the NA(−) biofilms, the redox potential initially decreased from 0 to −330 mV near the metal surface within the first 24 h and then stabilized at −215 mV (Fig. 4). In the control(+) biofilms, NO₃⁻ uptake started 24 h after the NO₃⁻ addition. Most of the NO₃⁻ was converted to NO₂⁻, again, in a ∼30-μm-thick layer close to the metal surface (Fig. 3d and Table 1). In the control(+) biofilm the redox potential on the metal surface increased from −450 to −270 mV over 96 h (Fig. 3d and 4). Net NO₃⁻ fluxes in NA and control(+) biofilms were almost equal, whereas the NO₂⁻ flux in control(+) biofilms was three times higher than in NA biofilms (Table 1). All biofilms were anoxic during the measurements, as assessed by an oxygen microelectrode (52).

FIG. 4.

Redox potential dynamics on the coupon surface in NA (○) and control (▪) biofilms during the 4-day tests. Arrows indicate the time of NO₃⁻ removal from NA biofilms and the addition of NO₃⁻ to control biofilms. Averages and standard deviations from three values measured at spots across the metal surface are shown.

TABLE 1.

N fluxes calculated from microprofiles and ¹⁵N isotope measurements

Condition and biofilm typea	Microsensor flux (n)b		Conversion rate with 15N incubation (n = 1)b
	NO3−	NO2−	14+15NO3−	14+15NO2−	29+30N2	29+30N2O	15NH4+	Recovery (%)
Before exchange (4 mo)
NA	−16.6 ± 2.4 (3)	5.4 (1)	−6.02	4.97	0.44	0.00	0.35	96
Controlspike	—	—	−0.35	0.08	0.00	0.00	0.26	98
After exchange (4 days)
NA(−)spike	—	—	−4.68	4.17	0.25	0.01	0.20	99
Control(+)	−22.5 ± 2.6 (3)	14.0 ± 2.3 (2)	−2.12	1.24	0.01	0.00	0.83	98

^aSpike, biofilms previously incubated without NO₃⁻ (4 months or 4 days, as indicated) and then dosed with ¹⁴⁺¹⁵NO₃⁻.

^bValues are μmol of N cm⁻² day⁻¹. n, number of replicate measurements; —, not measured.

During the 4-month growth period, the total DOC content in the water column ranged between 2 and 50 mg liter⁻¹, and concentrations of the volatile fatty acids formate, acetate, propionate, and butyrate were below 0.1 mg liter⁻¹. The potential for denitrification under such oligotrophic conditions was assessed on an NO₃⁻-amended (1 mM) NA biofilm using an N₂O microsensor and the acetylene block method (54). While no N₂O was formed in the absence of methanol, the N₂O level near the metal surface increased within the following 37 min after the addition of methanol as a carbon source (1 mM) (Fig. 5). This indicated that the electron donor supply ordinarily limits the denitrification in the system.

FIG. 5.

Test of carbon limitation as measured by N₂O production in an NA biofilm using an N₂O microsensor and applying the acetylene block technique. Data represent N₂O formation before (0 min) and after (12, 18, and 37 min) the addition of 1 mM methanol in the presence of 1 mM NO₃⁻.

N and S conversion rates.

The NO₃⁻ reduction products and N conversion rates in the biofilms were determined from ¹⁵N incubations (Table 1). The highest NO₃⁻ reduction rate among all incubations was measured in NA biofilms. Consistent with the microprofiling results, NO₂⁻ was the main intermediate product from NO₃⁻ reduction in NA biofilms, with N₂ and NH₄⁺ as end products. NO₃⁻ added to control biofilms (necessary for the determination of potential NO₃⁻reduction rates) was solely converted to NH₄⁺, suggesting DNRA.

During the opposite experiment, NA(−) biofilms receiving NO₃⁻ after a 4-day exposure to NO₃⁻-free conditions still reduced NO₃⁻ to NO₂⁻, N₂, and NH₄⁺ at rates comparable to NA biofilms (Table 1). In the control(+) biofilms, the same nitrogen products as in NA biofilms were formed but at lower rates. As the level of N₂ constantly increased after 24 h of ¹⁵N incubation in control(+) biofilms (data not shown), this suggests denitrification. Net fluxes of NO₃⁻ and NO₂⁻ in NA and control(+) biofilms calculated from microprofiles were generally higher than rates determined by ¹⁵N measurements, presumably due to mass transfer limitations during the isotope incubations and to heterogeneity between the biofilms.

The SO₄²⁻ reduction rates in all biofilms were below the detection limit of the radiotracer method of 38 pmol of SO₄²⁻ cm⁻³ day⁻¹, as determined according to the method of Kallmeyer et al. (29).

Effect of NO₃⁻ on bacterial diversity.

DGGE and 16S rRNA gene cloning showed distinct differences in microbial communities within the two biofilms, with higher species richness (number of bands) in the control than in the NA biofilms (Fig. 6, left panels). After the conditions were changed, within 4 days both techniques showed insignificant community changes in the NA(−) biofilms, but the diversity in control(+) biofilms had declined (Fig. 6, right panels). In NA biofilms under NO₃⁻ and NO₃⁻-free conditions, all sequences belonged to the Gammaproteobacteria, and the highest number of obtained sequences were closely related to Methylophaga spp., uncultured environmental clones obtained from Arctic pack ice (7), and Colwellia spp. (see Fig. S2 in the supplemental material). The community structure of the control biofilms was comprised of different proteobacterial sequences (alpha, gamma, delta, and epsilon subgroups) as well as those related to the Cytophaga-Flavobacteria-Bacteriodetes group. The vast majority (80%) of obtained sequences was related to the gammaproteobacterial Arctic pack ice clones (7) (see Fig. S3 in the supplemental material). Following NO₃⁻ addition, the percentage of sequences within the clone library compiled from control(+) biofilms shifted in favor of Methylophaga- and Colwellia-related sequences (together, 76%), while the percentage of sequences related to the Arctic pack ice clones was reduced from 80 to 22% (Fig. 6, right). The coverage within the four clone libraries was 81 to 93% (see Fig. S1 and Table S1 in the supplemental material).

FIG. 6.

(Top) DGGE fingerprints (in triplicate) of PCR-amplified 16S rRNA gene fragments obtained from NA and control biofilms before (4 months) and after (4 days) exchange using bacterial primers GM5F and 907RC. Arrows indicate the visible dominant bands. (Bottom) 16S rRNA gene cloning showing differences in the bacterial community composition in NA and control biofilms before and after exchange to opposite NO₃⁻ conditions. The relative abundance (percent) of sequences in each clone library (i.e., treatment) with the closest relative and the taxonomic affiliation is given. α, Alphaproteobacteria; γ, Gammaproteobacteria; δ, Deltabateria; ɛ, Epsilonbacteria; Alterom, Alteromonas sp.; Arcob, Arcobacter sp.; ARK, environmental Artic pack ice clones; CFB, Cytophaga/Flavobacteria/Bacteriodes; Colwell, Colwellia sp.; Methyloph, Methylophaga sp.; Roseob, Roseobacter sp.; Oleispir, Oleispira sp.; uncult, uncultured; mar, marine.

Corrosion analysis.

The matrix of control biofilms contained greigite (Fe₃S₄) while no known iron corrosion product was detected in the matrix of NA biofilms. The surface of newly manufactured coupons showed grinding marks (Fig. 2n), which disappeared from the surface under the control biofilms, indicating uniform corrosion (Fig. 2h). Conversely, the grinding marks were still intact under NA biofilms (Fig. 2b), indicating mitigation of uniform corrosion by NO₃⁻. However, SEM imaging of vertical coupon cross-sections revealed the presence of irregularly shaped, deep (up to 200 μm) pits underneath both types of biofilms (Fig. 2f and l). These pits were not present in newly manufactured coupons. Pits were often narrow and deep troughs, but laterally running subsurface pits were also found. SEM-energy dispersive X-ray analysis showed that biofilm residues on all coupon surfaces were enriched in iron ions and oxides, but only control biofilms contained iron sulfides. Embedded in the steel matrix orthogonal to the surface, needle-shaped manganese sulfide (MnS) inclusions were often associated with pits (one example is shown in Fig. 2f and l). The MnS inclusions were surrounded by an amorphous matrix, rich in the elements carbon, iron, oxygen, phosphor, and sulfur (Fig. 2f), which is characteristic for organic matter. Sulfur as AVS from the MnS particles was determined to account for 0.27 ± 0.00 weight percent of the steel and had a δ³⁴S_AVS of +1.54 ‰ ± 0.04‰. In the pipeline debris, AVS and CRS made up ca. 0.1 and 0.8 weight percent of the total sulfur (S_total) with δ³⁴S_AVS of −9.73 ‰ ± 0.05‰ and δ³⁴S_CRS of −15.28 ‰ ± 0.21‰, respectively.

DISCUSSION

Spatial distribution of microenvironments and metabolic activity.

For the first time, the spatial distribution of bacteria and their metabolic activities within biofilms from a water injection system for oil recovery have been described in situ with minimal disturbance of the biofilm structure. The spatial distributions of metabolic processes and cell densities within the biofilms correlate well (Fig. 2 and 3). The biofilms were heterogeneous in structure, and most of the N conversions occurred in a narrow ∼30- to 50-μm-thick zone at the biofilm base. Upon addition to the biofilms, NO₃⁻ penetrated the biofilm through to the metal surface, where it was mainly converted to NO₂⁻. The NO₂⁻ was partly further reduced to N₂ and NH₄⁺, as revealed by ¹⁵N labeling (Table 1) and partly diffused out of the biofilm (Fig. 3a and d). The open structure of the biofilms may facilitate mass transfer of nutrients and electron donors from the overlaying water down to the metal surface, where most bacterial cells were found. Here, alternative sources of electron donors for NO₃⁻ reduction and for other reductive biotransformations might have been anodic iron and cathodic hydrogen from the corroding steel (12, 68).

Redox potential measurements clearly showed that all NO₃⁻-treated biofilms exhibited fewer negative values than those biofilms without NO₃⁻. It is evident that the NO₃⁻ treatment drastically affected the dominant metabolic processes in the biofilms. It should be mentioned that the platinum microsensor used to measure the oxidation reduction potential status in the biofilms does not respond to all redox couples but primarily to Fe²⁺, sulfide, and H₂. Notably, the sensor does not respond to NO₃⁻ or NO₂⁻ and responds very slowly to oxygen. Furthermore, it is important to note that the different redox couples in a solution are rarely in equilibrium (63). For example, dissolved NO₃⁻ does not chemically react with H₂S; thus, a mixture of both is stable in the absence of a catalyst. In the absence of electroactive species, the redox signal will not stabilize, as it does in the anoxic water phase and the boundary layer. Near the metal surface, however, stable, reproducible readings indicated the presence of electroactive compounds (Fig. 3 and 4). The signal near the metal surface increased upon the NO₃⁻ addition to the control biofilms, but since the sensor does not respond directly to NO₃⁻, this increase was caused by an NO₃⁻-induced decrease in the concentrations of electroactive compounds such as Fe²⁺, S²⁻, or H₂ (36). As free H₂S and H₂ were not detected, dissolved Fe²⁺ is probably the main determinant for the redox signal and is suggestive of Fe corrosion. The decrease in redox potential after the removal of NO₃⁻ in NA biofilms can be explained by both enhanced Fe²⁺ release from the steel surface and decreased oxidation of Fe²⁺ (Fig. 4). The former process must have been significant in the case of the control coupons that showed substantial surface loss. Conclusive evidence on corrosion can be obtained from the redox potential and SEM data. Interpreted with care and in combination with other methods, the redox microsensor is a useful tool to detect and localize both the effect of treatment and/or MIC processes under in situ anoxic conditions.

Impact of NO₃⁻ on biofilm communities.

The presence of NO₃⁻ led to greatly increased rates of biomass growth and respiratory activity (NO₃⁻ reduction). Rates of NO₃⁻ reduction and turnover were several orders of magnitude higher than SO₄²⁻ reduction in the control samples, the latter being below the detection limit of the radiotracer method (i.e., 38 pmol of SO₄²⁻ cm⁻³ day⁻¹). Indicators of dissimilatory SO₄²⁻ reduction in the control samples were nearly nonexistent: there was no free H₂S, SO₄²⁻ reduction rates were undetectable with sensitive radiotracer techniques, and known SRB were not found (except for a single gammaproteobaterial clone) (Fig. 6), although the latter were previously detected in biofilm debris obtained at the end of the same 9-km-long injection pipeline (34). The finding of Fe₃S₄ together with the background SO₄²⁻ reduction rates measured here and the low abundance of SRB in our control biofilm samples, as opposed to their previous detection (34), are indicative of variable temporal abundance and/or activity of SRB within the system. With regard to MIC, it should be noted that the number of bacteria, either attached or planktonic, does not necessarily correlate with the extent of corrosion; the latter is considered to be linked to metabolic cell activity (reference 3 and references cited therein).

The addition of NO₃⁻ led to a shift from a low-biomass but species-rich community to a high-biomass NO₃⁻-reducing community with low diversity. This effect of NO₃⁻ on the bacterial community structure and diversity can most plausibly be explained by interspecies competition and direct inhibition by NO₂⁻, as was previously suggested (25, 27, 50, 60). In the presence of NO₃⁻, domination by NRB of indigenous control biofilm populations occurred due to the competition for limited electron donors under the prevailing oligotrophic conditions although total DOC concentrations could occasionally peak (up to 50 mg/liter), e.g., during algal blooms in the sea. Possible electron donors in the system were methanol, formaldehyde, formate, CO₂, and CO that derive from the complete and incomplete deoxygenation reaction with methanol and can be carried over from the deoxygenation system into the injection water (Fig. 1a). Other possible electron donors were organic carbon from decaying algal biomass, sulfide, Fe⁰, and H₂. NO₃⁻ reduction is thermodynamically more favorable than SO₄²⁻ reduction and has a higher biomass yield (72). Populations able to denitrify can therefore rapidly become dominant and efficiently consume the electron donor. The inhibition by NO₂⁻ can well lead to a community shift. It is a strong inhibitor of the sulfite reductase and thus inhibits SRB activity, as was previously shown in field and laboratory applications (26, 32, 43, 44, 50). NO₂⁻ levels of 0.05 mM (Fig. 3) at the base of the biofilms may have contributed to the change in community composition. It cannot be concluded here which specific bacterial group that appeared after the addition of NO₃⁻ to control biofilms was responsible for the NO₂⁻ production. Strains of Methylophaga (strictly aerobic methylotrophs), Colwellia (facultative anaerobic chemoheterotrophs), and Oleispira (facultative anaerobic chemoheterotrophs) have all been reported to reduce NO₃⁻ to NO₂⁻ (13, 47, 70, 71). Colwellia-related bacteria may have reduced the NO₂⁻ partly to N₂, as was reported for cultures of Colwellia maris (47).

NO₃⁻ added to control biofilms was converted to NH₄⁺ via NO₂⁻ as an intermediate. NH₄⁺ was potentially formed either by the indigenous populations through DNRA or abiotically, i.e., by the reaction of NO₃⁻/NO₂⁻ with metallic iron (32). However, the increased production of NH₄⁺ during the ¹⁵N incubation relative to the negative control indicated that NH₄⁺ formation was biogenic. Candidates capable of DNRA in the control biofilms were fermentative Vibrio spp. (31, 67). DNRA is energetically favorable under conditions of electron acceptor limitation (low levels of NO₃⁻/NO₂⁻ relative to the electron donor) and yields energy (ΔG′⁰ of −679.61 kJ mol⁻¹). It is also considered a detoxification pathway for NO₂⁻ in organisms containing an NO₂⁻ reductase (23, 67). DNRA might have protected indigenous populations from immediate NO₂⁻ inhibition in the control biofilms. However, the prolonged addition (4 days) of NO₃⁻ to these biofilms led to a rapid decrease in diversity, probably because NRB populations (r strategists) (30) grew faster than indigenous populations (K strategists). Furthermore, the relative decrease of the indigenous population could be due to the enhanced formation by NRB of NO₂⁻ that cannot be detoxified anymore via DNRA.

Conversely, the removal of NO₃⁻ from the NA biofilms resulted neither in a community shift nor in a decrease in potential NO₃⁻ reduction. Even after 4 days without NO₃⁻, upon the readdition of NO₃⁻, it was instantly reduced. The period of absence of NO₃⁻ was likely too short for the NRB populations to disappear and the typical control biofilm community to establish. This is favorable for the operation of the injection water system as fluctuations in the NO₃⁻ concentrations caused by shutdowns in NO₃⁻ dosing or maintenance work, for example, will not lead into instant growth of SRB. However, this stability in the community composition is not coupled to a functional stability; i.e., the redox potential does decrease within days upon NO₃⁻ removal (Fig. 4). Therefore, a continuous addition of NO₃⁻ might be preferred to prevent corrosion.

Sources and sinks of H₂S.

Microsensor, molecular, and tracer methods all suggest that the dissimilatory SO₄²⁻ reduction occurred at low rates in the control biofilms. However, metal sulfides made up part of the crust found both on the coupons and in debris collected from a pipeline adjunct to the RD. The H₂S produced might have reacted quickly with iron ions embedded in the biofilm matrix to form black iron sulfide precipitates, thereby preventing its detection. Indeed, ∼140 μM sulfide experimentally added to the water phase above coupons instantly resulted in FeS precipitation (Fig. 2c and i), and the sulfide concentrations inside the biofilms remained below the detection level (data not shown). The finding of Fe₃S₄ exclusively in the control biofilms also provides indirect evidence for the formation of H₂S. This H₂S must have derived from either (i) dissimilatory reduction of SO₄²⁻ by SRB, occurring at very low rates or episodically; (ii) bacterial degradation of S-containing amino acids (e.g., from algae, seawater organic particles, or the biofilm itself [14]); or (iii) biotic or abiotic dissolution of sulfide from S-containing inclusions embedded in the steel matrix (details below). The stable S isotope composition (³⁴S/³²S) of the sulfides in pipeline debris indicates that bacterial SO₄²⁻reduction has occurred. Dissimilatory SO₄²⁻ reduction and the associated isotope fractionation are the only processes that could lead to the depleted values observed in the debris, i.e., δ³⁴S_AVS of −9.73 ‰ ± 0.05‰ and δ³⁴S_CRS of −15.28 ‰ ± 0.21‰. The sulfide derived from the MnS inclusions or biosynthetic sulfur alone would have an S isotope composition of close to +1.5‰ or +18‰, respectively. Conversely, the sulfide in the debris was depleted in ³⁴S relative to both seawater SO₄²⁻ (+20‰) and the MnS (+1.5‰). While some of the sulfide may derive from MnS dissolution of steel inclusions, the depleted S in the sulfide crust is consistent with the low rates of background SO₄²⁻ reduction and consequent sulfide production in the injector.

Corrosion damage.

NO₃⁻ injection into reservoirs for souring control may prevent or mitigate the risks from MIC within the injection pipelines (38, 66). However, corrosion occurred on the coupon surfaces underneath both NA and control biofilms and so was not entirely prevented by the NO₃⁻ treatment, although its addition did change the type and extent of corrosion. NA biofilm coupons were less affected by uniform corrosion than the control coupons due to the inhibition of sulfide production. The variation in corrosion patterns between NA and control biofilms can be explained by the nature of the biofilms and their distribution over the metal surface. Biofilms are well known to affect the electrochemistry of the near-surface milieu as they modify the physicochemical environment within which the corrosion reactions occur (58). Matrix-bound Fe²⁺ and Fe³⁺ are thought to shuttle electrons from the steel to a suitable electron acceptor such as NO₃⁻, NO₂⁻, and O₂, leading to differential corrosion cells that accelerate corrosion (4). It was estimated for control biofilm coupons that 2% of the S_total from the steel matrix (which contained 535 μmol of S_total cm⁻³) could dissolve during the pitting process. This released sulfide could contribute up to 11% of the S_total in the biofilm crust.

The NO₃⁻ treatment did not protect against pitting corrosion and may have even accelerated it via the chemical oxidation of metallic iron with NO₃⁻/NO₂⁻ with consequent reduction to NH₄⁺ (32, 51, 68). Indeed, NO₂⁻ concentrations of 30 and 50 μM at the metal surface in the biofilms (Fig. 3) are in the range where NO₂⁻ is reported to be corrosive (<3.5 mM) (reference 26 and references cited therein). Nevertheless, NO₃⁻ is used to inhibit SRB activity in seawater oil recovery systems, with one of the intentions being to reduce MIC pitting. We cannot deduce from the data to what extent either of these processes contributed to the pitting on NO₃⁻-amended coupons; however, it is evident that the NO₃⁻ addition strongly reduced the uniform surface corrosion.

Many pits on the corroded coupons were closely associated with needle-shaped MnS inclusions embedded in the steel matrix (Fig. 2f and l). Notably, this was independent of the NO₃⁻ treatment. MnS inclusions are ordinarily formed from manganese and sulfur additions to the steel during its manufacturing. Dissolving inclusions are known to increase the susceptibility of steel to electrochemical pit formation (reference 59 and references cited therein). In addition, bacteria seem involved in the initiation and propagation of pitting at these particles (10). The association of organic matter and MnS inclusions in the coupons implies that the pit formation was indeed microbially influenced. MnS inclusions dissolve into Mn²⁺ and H₂S at a pH and redox potential below 4.8 and −100 mV, respectively (17). The pH in the injection water was 6.8 to 7.3; thus, a pH gradient of at least two units over a distance of up to 400 μm is needed to obtain a pH at the bottom of the pits of below 4.8. Such steep pH gradients were earlier found and were attributed to the formation of protons from the hydrolysis by Fe²⁺ and the formation of hydroxides (32, 36). We suggest that released Mn²⁺ and H₂S from MnS inclusions may serve as precursors for biochemical manganese and sulfur cycles in the biofilms. In the absence of NO₃⁻, the dissolved H₂S might further react with iron to form greigite (Fe₃S₄) inside the biofilm at close to neutral pH values. The overall reaction of MnS dissolution and Fe₃S₄ formation from metal iron is thermodynamically favorable under standard conditions and pH 7 and is proposed to proceed as follows: 4MnS(s) + 3Fe⁰(s) + 8H⁺(aq) → Fe₃S₄(s) + 4Mn²⁺(aq) + 4H₂(aq), where ΔG′⁰ is −32.86 kJ mol⁻¹.

The removal of cathodic H₂ by bacteria and the oxidation of Mn²⁺ with tracers of oxygen from the reaction accelerate the consumption of metal iron and the production of Fe₃S₄, thus enhancing corrosion. Under in situ conditions, the energy gain from this reaction is supposed to be even higher. Greigite is a metastable iron sulfur compound formed at low pH from preexisting mackinawite (FeS) by biogenic aldehydic carbonyls (55, 56). As it is only stable at pH values between 5.5 and 7.0 in a range of redox potential between −200 and 400 mV (56), the exclusive detection of Fe₃S₄ inside the control biofilms is indicative of its formation inside the biofilm. Whether the MnS dissolution and Fe₃S₄ formation are mediated by microbes and coupled to iron oxidation to produce energy is interesting and needs further investigation.

Summary and conclusions.

Our combined results provide the first insights into complex metal-microbe interactions under in situ conditions and show that NO₃⁻ amendments strongly induced structural and functional changes in the marine biofilms. NO₃⁻ addition led to a shift from a low, but species-rich, biomass to a high-biomass, NO₃⁻-reducing community with low diversity. Rates of NO₃⁻ reduction and turnover were several orders of magnitude higher than dissimilatory SO₄²⁻ reduction. Greigite, exclusively detected in the control biofilms, provides evidence that SO₄²⁻ reduction occurred either at low rates or episodically. Corrosion analyses revealed that uniform corrosion was microbially influenced, while this was less clear for pitting corrosion. Whereas pitting corrosion was not affected by the treatment, uniform corrosion was mitigated by the addition of NO₃⁻. MnS inclusions embedded in the steel matrix may play an important role in the pitting process. It is proposed that bacteria might be involved in their dissolution and combine this with the oxidation of metal iron to gain energy in the absence of NO₃⁻.