Field samples were acquired from the utility water system at a SAGD site in northern Alberta. Water samples (2 L) were shipped in 1 L Nalgene (VWR International, Edmonton, AB) bottles filled to the brim to exclude oxygen. Sample PW7 was a freshwater sample from near the well-bore region with a temperature of 11°C. Sample P0866, was immediately upstream of the water treatment facility, where SBS was injected and was between 20 and 25°C. Samples P0866S and P0848S were taken downstream from this facility (Figure 1). All samples were stored in a Coy (Grass Lake, MI) anaerobic hood with an atmosphere of 90% N₂ and 10% CO₂ (N₂-CO₂, Praxair, Calgary, AB). A pipe segment representing a corrosion failure was also obtained, along with water sample PW8 from this site (Figure 1). The pipe segment was shipped in a bucket, together with pipe-associated water (PW8-PAW). Pipeline associated solids (PW8-PAS) were obtained by scraping the pipe segment in the anaerobic hood and suspending the scrapings in 0.2 μm filtered PW8-PAW, representing 10% of the internal volume of the pipe segment, as described elsewhere (Park et al., 2011). Upon arrival, 1 L of each water sample was filtered using a 0.2 μm Millipore nylon membrane (VWR International, Edmonton, AB) to collect biomass. This was immediately frozen at −80°C for use in DNA extraction.

While some water chemistry data were provided with the samples (bicarbonate, sodium), their chemical composition was characterized further. Sulfide concentrations were assessed colorimetrically using N, N-dimethyl-p-phenylenediamine (Trüper and Schlegel, 1964). Aqueous ferrous iron concentrations were assessed using a colorimetric assay using ferrozine (Park et al., 2011). Ammonium concentrations were assayed using the indophenol method and the pH of the samples was measured using a Thermo Scientific Inc. Orion 370 model pH meter (VWR International, Mississauga, ON). To measure acetate concentrations, a high-performance liquid chromatograph (HPLC) from Waters (Mississauga, ON), model 515, equipped with a Waters 2487 model UV detector set at 220 nm and an organic acid column (250 × 4.6 mm, Alltech Prevail, Guelph, ON) eluted with 25 mM KH₂PO₄ of pH of 2.5 was used. Samples (1 mL) were centrifuged at 13,300 rpm for 5 min, 300 μ L of the resulting supernatant was acidified using 20 μ L of 1 M H₃PO₄, and eluted at a flow rate of 0.8 mL/min. Sulfate concentrations were monitored using a Waters 600 model HPLC equipped with a Waters 432 conductivity detector and an IC-PAK anion column (150 × 4.6 mm, Waters) eluted with 24% (vol/vol) acetonitrile, 2% butanol, and 2% borate-gluconate concentrate. Following centrifugation 100 μ L of the sample supernatant was added to 400 μ L of the acetonitrile solution and eluted at 2 mL/min.

Two carbon steel coupons (American Society for Testing and Materials, ASTM a366, containing 0.015% carbon, 5 × 1 × 0.1 cm) were incubated with 50 mL of each water sample in a 120 mL serum bottle. The coupons were cleaned according to National Association of Corrosion Engineers (NACE) protocol RP0775-2005. The coupons were polished with 400 grit sandpaper, washed in a dibutylthiourea solution for 2 min, followed by a saturated bicarbonate solution for 2 min. The coupons were briefly rinsed in deionized water and then acetone and quickly dried in a stream of air. The coupons were weighed three times using an analytical balance and the average weight was recorded as the starting weight. The serum bottles were sealed with a headspace containing N₂-CO₂. During incubation, headspace methane and aqueous acetate concentrations were monitored. For methane measurements, 0.2 mL of the headspace was sampled periodically using a syringe that had been flushed with N₂-CO₂ and injected into a Hewlett-Packard (Mississauga, ON) model 5890 gas chromatograph at 150°C, equipped with a flame-ionizing detector set to 200°C and a stainless steel column (0.049 × 5.49 cm, Porapak R 80/100, Supelco, Oakville, ON). Following incubation, surface deposits were removed from each coupon using a non-scratching wipe. The coupons were cleaned according to the NACE protocol and again weighed three times. The weight loss (Δ W) was converted into a general corrosion rate (CR, mm/year) by the equation: CR = 87600 × Δ W/A × T × D where A is the coupon surface area (cm²), T is the incubation time (hours), and D is steel density (7.85 g/cm³), as described previously (Park et al., 2011).

Two carbon steel coupons (ASTM a366) were incubated in 50 mL of anaerobic Coleville synthetic brine (CSB-K) medium (Callbeck et al., 2013) in sealed 120 mL serum bottle microcosms with N₂-CO₂. Each microcosm was anaerobically inoculated with 2.5 mL of each field sample (5% inoculum of field sample). As a control, two corrosion coupons were incubated anaerobically with 50 mL of medium only. The incubation periods lasted between 8 and 16 weeks at 32°C, under constant shaking at 100 rpm. During this time, concentrations of methane were measured using gas chromatography and concentrations of acetate were measured using HPLC. For incubations with PW8, PW8-PAW, or PW8-PAS only corrosion rates were determined from metal weight loss. For incubations involving samples PW7, P0866, P0866S, and P0848S, biomass was also collected from each microcosm. The metal coupons were removed from the microcosm and washed with CSB-K medium to remove any unattached material. The coupons were gently scraped with a blade and the resulting biomass was considered representative of the biofilm population. The coupons were then cleaned according to the NACE protocol, and general corrosion rates were determined.

Biomass, collected from each field sample and from the biofilm fractions, was used for DNA extraction. DNA was extracted, using a bead-beating procedure outlined by the manufacturer of the FastDNA® Spin Kit for Soil (MP Biomedicals, Santa Ana, CA). DNA was eluted using 10 mM Tris-Cl pH 8.5 (Buffer EB from the QIAquick kit, Qiagen, Toronto, ON). The concentration of DNA extracted from each sample was quantified using the Qubit Fluorometer, and Quant-iT™ dsDNA HS Assay Kit (Invitrogen, Burlington, ON).

PCR amplification of 16S rRNA genes from isolated community DNA was performed using primers 926Fw (AAACTYAAAKGAATTGRCGG) and 1392R (ACGGGCGGTGTGTRC) using the Taq Polymerase I Kit (Qiagen, Toronto, ON). PCR was for 5 min at 95°C, followed by 20 cycles of 30 s at 95°C, 45 s at 55°C, and 90 s at 72°C and finally 7 min at 72°C. A second PCR step used FLX titanium amplicon primers 454_RA_X and 454T_FwB for an additional 10 cycles. These have the sequences for primers 926Fw and 1392R at their 3′ ends. Primer 454T_RA_X has a 25 nucleotide A-adaptor (CGTATCGCCTCCCTCGCGCCATCAG) and a 10 nucleotide multiplex identifier barcode sequence X. Primer 454T_FwB has a 25 nucleotide B-adaptor sequence (CTATGCGCCTTGCCAGCCCGCTCAG). The resulting 16S rRNA PCR amplicons were purified and sent for pyrosequencing to the Genome Quebec and McGill University Innovation Centre (Montreal, QC). Pyrosequencing was done using a Genome Sequencer FLX Instrument, using a GS FLX Titanium Series Kit XLR70 (Roche Diagnostics Corporation, Laval, QC). Pyrosequencing and bioinformatic analysis of the sequences obtained have all been described elsewhere (Park et al., 2011; Soh et al., 2013).

Desulfovibrio vulgaris Hildenborough and Acetobacterium woodii (DSM 1030) were used as model SRB and acetogen strains, respectively and were grown in Widdel-Pfennig (WP) medium with 10 mM sulfate and 30 mM NaHCO₃ (Caffrey et al., 2007). Microcosms were inoculated with a culture of A. woodii, a culture of D. vulgaris, a co-culture of both strains, or no inoculum. A fifth microcosm containing a culture of D. vulgaris was amended with 3 mM acetate. Incubations were in 160 mL serum bottles containing 40 mL of medium and a headspace of 80% H₂ and 20% CO₂ Growth was measured by determining the optical density at 600 nm using a Shimadzu model 1800 UV spectrophotometer (Mandel, Guelph, ON) and by monitoring concentrations of acetate, sulfate, and sulfide.

To test corrosivities of these cultures, two carbon steel corrosion coupons (ASTM a366) were added to each of the microcosms. The headspace of these microcosms was either 80% H₂ and 20% CO₂, 16% H₂ and 84% CO₂ or N₂-CO₂, as indicated. Concentrations of acetate and sulfate were measured during incubation. Following incubation, the coupons were cleaned as per NACE protocol, metal weight loss was determined and this was converted into general corrosion rates.

The samples obtained from the water gathering system for a SAGD facility are indicated in Figure 1. Relative to samples obtained previously from the GM formations (Park et al., 2011), these had lower concentrations of NaCl (0–200 ppm, compared to 4000–6000 ppm for GM) and higher concentrations of sulfate (0–64 ppm, as compared to 0–2 ppm for GM). Both sets of samples had high bicarbonate concentrations (520 ppm, as compared to 400–1600 ppm for GM). Sulfide (0–1.5 ppm), ammonium (not shown) and acetate (0–19 ppm) were present in low concentrations (Table 1). Ferrous iron was low, except in the pipe-associated solids sample (Table 1: PW8-PAS), which had 1546 ppm, indicative of corrosion (Park et al., 2011).

Incubation of samples P0866, P0866S, and P0848S with carbon steel coupons and a headspace of N₂-CO₂ for 70 days gave formation of 2.3, 1.0, or 2.6 mM headspace methane, respectively (Figure 2A). No significant methane was formed in incubations with sample PW7 or the water-only control. Weight loss corrosion rates were higher (p < 0.00032) for incubations with samples P0866, P0866S, and P0848S (Figure 2B: average 0.018 ± 0.0040 mm/yr, n = 6) than with sample PW7 or the water-only control (Figure 2B: average 0.0060 ± 0.00091 mm/yr, n = 4). The formation of acetate was evaluated in a separate experiment indicating formation of 0.45 to 3.2 mM acetate in the aqueous phase of these incubations, whereas no acetate was formed in the water-only control (Figure 2C). Neither methane nor acetate were formed in the absence of steel coupons (results not shown).

In a variation of this experiment serum bottles with 70 mL CSB-K medium, containing carbon steel coupons, 30 mM of bicarbonate and a headspace of N₂-CO₂ were inoculated with 3.5 ml of field sample. Acetate and methane were monitored as a function of time, and weight loss corrosion was determined at the end of the experiment. Production of up to 0.6 mM acetate was seen for all samples. This appeared to precede production of methane and go through a maximum (Figure 3A). Production of 1.9 to 3.0 mM headspace methane was observed with all samples, except P0866 (Figure 3B). The biofilms present on the coupons were isolated prior to processing of the coupons for measuring weight loss corrosion. The corrosion rates observed for incubations with samples PW7, P0866S, and P0848S (average 0.0068 ± 0.0012 mm/year, n = 6) were higher (p < 0.00076) than those with P0866 and the medium control (average 0.0035 ± 0.00031 mm/year, n = 4), as shown in Figure 3C. Hence, although both acetate and methane were formed as part of the corrosion process, the formation of headspace methane appeared to be correlated with the corrosion rate, i.e., P0866S, the sample with the highest corrosion rate (0.0083 mm/yr) also had the highest headspace methane concentration (3.0 mM).

Similar results were obtained with samples PW8, PW8-PAW, and PW8-PAS, the latter representing pipe-associated solids suspended in microfiltered pipe-associated water (PW8-PAW). Incubation of metal coupons in these samples gave up to 0.38 mM acetate in the aqueous phase and 1 to 2 mM headspace methane (Figures 4A,B). The highest concentrations of acetate and methane were produced in incubations with PW8-PAS, which also had the highest corrosion rate (p < 0.0026) at 0.020 mm/yr (Figure 4C). When these field samples were inoculated in CSB-K medium, production of up to 1.2 mM acetate was seen to precede production of up to 2 mM of headspace methane (Figures 5A,B). Acetate and methane production were observed in incubations with PW8, PW8-PAW, and PW8-PAS, which had higher corrosion rates (p < 0.12) (average 0.0099 ± 0.0049 mm/yr, n = 12) than the uninoculated control (0.0058 ± 0.00042 mm/yr, n = 4) (Figure 5C).

Pyrosequencing of 16S rRNA amplicons for all seven field samples and for four coupon enrichment biofilms, gave 1357 to 6417 quality controlled reads for each sample (Table 2). The compositions of the microbial communities derived from these results are compared in Figure 6. Although the water samples were all from a single pipeline system, it is evident that the microbial communities in water samples retrieved along the flow path were distinct. The community of PW7 near the fresh water producing well-differed from those of PW8 and P0866, both further downstream and all of these were distinct from the communities at P0866S and P0848S downstream from the SBS injection point (Figures 1, 6). Microbial communities of PW8-PAS and PW8-PAW formed a separate branch in the dendrogram together with that of PW8. Communities in coupon enrichment biofilms differed significantly from those in field samples (Figure 6). Microbial communities formed following incubation with P0866, P0866S, and P0848S were very similar, whereas PW7 (Biofilm) was distinct from this cluster (Figure 6).

The microbial community of water sample PW7 was dominated by Betaproteobacteria of the genus Acidovorax (Table 2: 61.70%). This taxon had decreased to 2.49% of the community in water sample PW8, which also harbored significant fractions of methanogens of the class Methanobacteriaceae (16.44%), the genera Methanosaeta (14.46%) and Methanoculleus (3.36%), as well as of Deltaproteobacteria of the genera Desulfocapsa (7.07%) and Desulfobulbus (4.81%). The pipeline solids at this location, PW8-PAS, also harbored high fractions of Deltaproteobacteria/Desulfovibrio (9.69%), of Firmicutes/Acetobacterium (5.58%), and of Firmicutes/Desulfosporosinus (13.45%). Many of these taxa use hydrogen (and possibly Fe⁰) as electron donor for the reduction of CO₂ to methane or to acetate (Acetobacterium) or for the reduction of sulfate to sulfide (Desulfobulbus, Desulfosporosinus, Desulfovibrio). The water associated with the pipe sample (PW8_PAW) had a high fraction of Dietzia (25.91%), which has been associated with hydrocarbon degradation (Bihari et al., 2011). The community composition of water sample P0866, taken further downstream, had high fractions of Betaproteobacteria/Rhodocyclaceae (48.23%), capable of nitrate reduction and of Deltaproteobacteria of the genera Desulfurivibrio (23.52%), Desulfovibrio (4.41%), and Desulfomicrobium (5.43%). Treatment with SBS gave rise to communities dominated by Desulfocapsa in water samples P0866S (95.76%) and P0848S (95.28%), which grows by disproportionation of bisulfite to sulfate and sulfide, as discussed elsewhere (Park et al., 2011). Hence, a succession of communities is present along the water flow path, with a very strong shift being observed by the addition of SBS. It is interesting that the microbial communities in the water samples, obtained from the water-transporting pipeline studied here, appeared to change so rapidly as a function of distance (Figure 6). This would seem to be at odds with the fact that the transport time of water from PW7 to P0848S is less than 1 day. These large changes in community composition may indicate that most biomass grows on the pipe walls as biofilms, is sloughed off at a high rate and appears to dominate the microbial community compositions seen. This explains why Desulfocapsa becomes a major community component immediately downstream from the SBS injection point (Table 2, Figure 1). The planktonic population would not be able to adjust in such a short time, but if the Desulfocapsa is biofilm-grown, it may continuously be swept into the flowing water.

Incubation of field samples with CSB-K medium, which contains CO₂ and bicarbonate but lacks sulfate, in the presence of metal coupons led to their colonization by Methanobacteriaceae, which were present at 71.75, 89.31, and 79.86% in incubations with P0866, P0866S, and P0848. In the incubation PW7, only 4.49% of the reads were of the class Methanobacteriaceae. The latter biofilm had a high fraction of the acetotrophic methanogen Methanosaeta (17.14%), as well as of Pseudomonas (52.82%). Compared to the field samples, the biofilm communities also showed an increased presence of the acetogen Acetobacterium (Table 2). The colonization of carbon steel coupons by methanogens and acetogens (Table 2) can explain the production of acetate and methane observed in these incubations (Figure 3).

SRB of the genus Desulfovibrio or Desulfomicrobium need acetate as a carbon source when deriving energy for growth from reduction of sulfate with H₂ or Fe⁰ (Badziong et al., 1979; Dinh et al., 2004; Caffrey et al., 2007; Enning et al., 2012). When acetate is absent or limiting, this can in principle be provided by acetogens from H₂ (or Fe⁰) and CO₂. In order to test this hypothesis we studied the growth of the model SRB Desulfovibrio vulgaris Hildenborough and the model acetogen Acetobacterium woodii in WP medium, containing bicarbonate and sulfate and a headspace of 80% H₂ and 20% CO₂. A monoculture of A. woodii grew poorly in this medium as judged by turbidity (Figure 7A) and produced up to 1.6 mM acetate after 10 days (Figure 7B). A monoculture of D. vulgaris did not grow (Figure 7A), did not have detectable acetate (Figure 7B) and reduced only 2 mM sulfate to sulfide (Figures 7C,D). Addition of 3 mM acetate to the monoculture of D. vulgaris gave strong growth, sulfate reduction, and sulfide production, while using 1.5 mM of added acetate. Following a lag phase of 2 days needed for the production of 1.3 mM acetate (Figure 7B), the co-culture of D. vulgaris and A. woodii exhibited similar growth, sulfate reduction and sulfide production as the monoculture of D. vulgaris with added acetate. No sulfate reduction or acetate production was seen in the inoculum-free control. Hence, A. woodii can provide the acetate needed by D. vulgaris for growth under chemolithotrophic conditions.

When carbon steel coupons were added to serum bottles, containing WP medium with sulfate and bicarbonate and a headspace of N₂-CO₂ and inoculated with D. vulgaris, D. vulgaris and 3 mM acetate, A. woodii, D. vulgaris, and A. woodii, or no inoculum, similar corrosion rates were observed of 0.0039 to 0.0055 mm/yr in all incubations (Figure 8C). This indicates that Fe⁰ alone is not a good electron donor for sulfate or carbon dioxide reduction by these two type cultures. In order to determine whether Fe⁰ can be used as an electron donor co-metabolically with H₂, the experiment was repeated with a headspace of 80% (vol/vol) H₂ and 20% CO₂ (excess H₂ for reduction of 10 mM sulfate) or 16% H₂ and 84% CO₂ (insufficient H₂ for reduction of 10 mM sulfate). The highest corrosion rates were observed with limiting H₂ for incubations with D. vulgaris and 3 mM acetate (0.0113 mm/yr) or with the coculture of D. vulgaris and A. woodii (0.0104 mm/yr), as shown in Figure 8B. These were higher (p < 0.000068) than the corrosion rates observed for incubations with D. vulgaris alone without added acetate, with A. woodii alone or with the uninoculated control (average 0.0074 ± 0.0015 mm/yr, n = 12). In the presence of excess H₂ lower corrosion rates, between 0.0054 and 0.0059 mm/yr, were observed for all incubations except for the co-culture, which was somewhat higher (0.0076 mm/yr), as shown in Figure 8A. Overall, these results indicate that the co-culture of D. vulgaris and A. woodii can co-metabolically corrode carbon steel at a rate comparable to that when acetate is provided directly to the D. vulgaris culture.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.