carbon and sulfur cycling by microbial communities in a gypsum-treated oil sands tailings pond

Carbon and Sulfur Cycling byMicrobial Communities in aGypsum-Treated Oil Sands TailingsPondE S T H E R R A M O S - P A D R O N , †

S Y L V A I N B O R D E N A V E , † S H I P I N G L I N , †

I Y S W A R Y A M A N I B H A S K A R , †

X I A O L I D O N G , ‡ C H R I S T O P H W . S E N S E N , ‡

J O S E P H F O U R N I E R , §

G E R R I T V O O R D O U W , † A N DL I S A M . G I E G * , †

Petroleum Microbiology Research Group, Department ofBiological Sciences, University of Calgary, 2500 UniversityDrive NW, Calgary, Alberta, Canada T2N 1N4, Sun Center ofExcellence for Visual Genomics, Faculty of Medicine, 3330Hospital Drive NW, Calgary, Alberta, Canada, T2N 4N1, andSuncor Energy Inc., Ft. McMurray, Alberta, Canada

Received August 18, 2010. Revised manuscript receivedNovember 11, 2010. Accepted November 15, 2010.

Oil sands tailings ponds receive and store the solid andliquid waste from bitumen extraction and are managed topromote solids densification and water recycling. The pondsare highly stratified due to increasing solids content as a functionof depth but can be impacted by tailings addition andremoval and by convection due to microbial gas production.Wecharacterizedthemicrobialcommunities inrelationtomicrobialactivities as a function of depth in an active tailings pondroutinely treated with gypsum (CaSO4 ·2H2O) to acceleratedensification. Pyrosequencing of 16S rDNA gene sequencesindicated that the aerobic surface layer, where the highest levelof sulfate (6 mM) but no sulfide was detected, had a verydifferent community profile than the rest of the pond. Deeperanaerobic layers were dominated by syntrophs (Pelotomaculum,Syntrophus, and Smithella spp.), sulfate- and sulfur-reducing bacteria (SRB, Desulfocapsa and Desulfurivibriospp.), acetate- and H2-using methanogens, and a variety of otheranaerobes that have been implicated in hydrocarbon utilizationor iron and sulfur cycling. The SRB were most abundantfrom 10 to 14 mbs, bracketing the zone where the sulfatereduction rate was highest. Similarly, the most abundantmethanogens and syntrophs identified as a function of depthclosely mirrored the fluctuating methanogenesis rates.Methanogenesis was inhibited in laboratory incubations bynearly50%whensulfatewassuppliedatpond-levelconcentrationssuggesting that in situ sulfate reduction can substantiallyminimize methane emissions. Based on our data, we hypothesizethat the emission of sulfide due to SRB activity in the gypsumtreated pond is also limited due to its high solubility andoxidation in surface waters.

Introduction

The Athabasca oil sands region in northeastern Alberta,Canada harbors an enormous fossil fuel resource that is ofglobal economic importance. The resource is estimated tocontain between 1.7 to 2.5 trillion barrels of heavily biode-graded oil, or bitumen, in an area spanning 75,000 km2 (1).Approximately 20% of the resource is within 75 m of thesurface allowing recovery by mining (1). The mined bitumenis separated from associated sands and clays using an alkalinehot water extraction process. The resulting solid and liquidwaste is stored on-site in large settling areas referred to astailings ponds. The ponds typically consist of 20 to 30 wt %solids (sands and clays), ∼1 to 3 wt % of residual bitumen,and slightly alkaline water (2). The ponds also containnaphtha (low molecular weight hydrocarbons used as abitumen diluent) and naphthenic acids (3), known to be toxicto a variety of aquatic organisms (4). Once the extractionwaste enters the tailings ponds, the solids are allowed tosettle, and the resulting clear water phase is reused in thebitumen extraction process. To produce 1 barrel (159 L) ofbitumen, 1 m3 of oil sands is extracted with 3 m3 of water,producing 4 m3 of tailings (5). Hence, typical daily productionof 105 barrels of bitumen per day releases approximately400,000 m3 of tailings per day in the region (6). Tailings pondmanagement is thus a key component of oil sands operationswith the ultimate goal of reclamation. Government mandateshave recently established aggressive timelines for tailingspond use and reclamation (7) so an understanding of allfactors that impact the operation and life cycle of tailingsponds is critical.

In current practice, approximately 80% of the water thatenters into tailings ponds is recycled back into the bitumenextraction process. The presence of fines (particles <44 µm)confounds the recycling of the remaining water by forminga stable suspended network that densifies very slowly (2). Oilsands operators use different approaches to increase the rateof densification. One approach is through the use of gypsum(CaSO4 ·2H2O, typically added at 1 kg per m3 of tailings), whereCa2+ serves as the densification agent (2). More rapiddensification allows for more extensive water recycling aswell as for the eventual formation of a trafficable surface tosupport reclamation efforts.

Previous work has shown that tailings ponds harbordiverse microorganisms (8, 9). Sulfate reduction and metha-nogenesis are key anaerobic processes, and such microbialactivity has been implicated in releases of hydrogen sulfideand methane from some tailings ponds (9-12). It has alsobeen noted that the presence of sulfate from gypsum additioninhibits methanogenesis (12). Low molecular weight hydro-carbons derived from diluent were shown to be the primarycarbon and energy sources under methanogenic conditions(5). Molecular analysis of samples from Mildred Lake SettlingBasin (MLSB), the largest tailings pond in the region operatedby Syncrude, showed a prevalence of fermenters and sulfate-reducing bacteria (SRB), and both H2- and acetate-usingmethanogens (13). The concentrations of sulfate and ofdissolved and entrapped methane as well as the mostprobable numbers of SRB and methanogens were reportedas a function of depth for the MLSB (9). Depth-dependentin situ microbial activity and the corresponding microbialpopulations remain poorly understood. Such information,together with a thorough knowledge of pond microbiota asa function of depth, is critical to evaluate the benefits andrisks (such as potential H2S and CH4 emissions) of microbiallycatalyzed processes in tailings ponds.

* Corresponding author e-mail: [email protected].† University of Calgary.‡ Sun Center of Excellence for Visual Genomics, Faculty of

Medicine.§ Suncor Energy Inc.

Environ. Sci. Technol. 2011, 45, 439–446

10.1021/es1028487 2011 American Chemical Society VOL. 45, NO. 2, 2011 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 439

Published on Web 12/03/2010

Thus in the present work, we assessed the microbialcommunities responsible for carbon and sulfur cycling inlight of microbial activity and biogeochemical measurements(14) in an active oil sands tailings pond as a function of depth.The results indicate that an understanding of pond microbialprocesses can yield substantial insights and benefits fortailings pond management.

Experimental SectionSampling, Pond Chemistry, and Anaerobic Rate Assays.Samples were collected from tailings pond 6, an active pondmanaged by Suncor Energy Inc., in October 2008. Tailingswere collected from a location near the center of the pond(UTM 466358E - 6319838N) at ∼1.5 m intervals from thesurface down to 18.3 m below surface (mbs). The surfacewater at the sampling location was poorly clarified, but asolids-free sample could be obtained. Tailings containingsuspended solids reached just below the surface water.Although tailings density was not measured, the samplesobtained were considered to be from the ‘preconsolidated’zone, and a subsequent sampling event showed that the watercontent ranged from approximately 60% at 3 mbs to below30% by 12 mbs (E. Ramos-Padron, unpublished results).Approximately 1 L of tailings from each depth was collectedinto sterile plastic Nalgene bottles filled to the top to avoidoxygen contamination. During sampling, pH, temperature,and dissolved oxygen levels were recorded (14). The sampleswere shipped the same day to the laboratory and immediatelyplaced into an anaerobic glovebag containing 90% N2 and10% CO2 for storage. In order to minimize microbialcommunity and chemical changes in the samples, sub-samples for biodiversity studies were removed and stored at-70 °C and soluble sulfate and sulfide concentrations weredetermined (15, 16) within 24 h of sampling. Sulfate reductionrates (SRR) were determined using a radiotracer technique(17), and methanogenesis rates were assessed by addingtailings to anoxic, sterile serum bottles and monitoringmethane production for 36 d (14). Concentrations of acetate,propionate, and butyrate were determined by HPLC aspreviously described (15).

Microcosm Experiments. All incubations were preparedin the anaerobic glovebag. For determining the effect ofsulfate addition on methanogenesis, 10 mL of tailings (from6.1 mbs) were combined with 20 mL of a sterile, anoxicmineral salts medium ((18), but with no added rumen fluid).Sulfate from an anoxic stock solution (100 mM Na2SO4) wasadded to achieve final concentrations ranging from 0.2 to 12mM. Na2SO4 was used instead of CaSO4 (gypsum) as thesulfate source, as the latter has low water solubility and isdifficult to add reproducibly to incubations (11). Triplicateincubations were prepared for each sulfate concentration.Methane formation was monitored over time using a Hewlett-Packard Model 5890 gas chromatograph equipped with athermal conductivity detector using helium as the carriergas and a packed stainless steel column (18′′ long × 1/8′′diameter, Poropak R, 80/100, Supelco). Sulfate concentrationswere monitored using the barium chloride turbidimetricmethod (19). Duplicate autoclaved controls were preparedfor each test condition.

For sulfide oxidation tests, 3 mL of tailings from eachdepth were added to serum bottles containing 60 mL of asterile, anoxic mineral salts medium (20) and amended with3 mM sodium sulfide. Experiments were initiated by adding45 mL of air (containing 20% O2, enough to consume all ofthe added sulfide), and dissolved sulfide concentrations weremeasured every 20 min for up to 100 min using the coppersulfate method (19). Sulfate concentrations were measuredthroughout the experiments by ion chromatography (15).Controls were prepared with autoclaved tailings or in mediumonly.

DNA Extraction, Amplification of 16S Genes, and Py-rosequencing. Clay-rich tailings strongly adsorb DNA, mak-ing DNA extraction difficult (21). Because initial attemptsdid not yield DNA, a DNA adsorption competitor was used.We combined a bead beating DNA extraction kit (FastDNASpin Kit for Soil; MP Biomedicals) and skim milk powder(Fluka Analytical) (40 mg ·g-1 of tailings) as the DNAcompetitor (21) to successfully obtain DNA. This procedurewas used for all samples from 1.5 to 18.3 mbs, yielding anaverage of 0.17 µg DNA per g of tailings. No skim milk powderwas needed for DNA extraction from the surface watersample, because clays were absent.

Isolated community DNA (2 ng µL-1) was subjected topolymerase chain reaction (PCR) amplification of 16S rDNAgenes using 12.5 µL of 2xPCR Master Mix (Fermentas), 10.5µL of nuclease-free water (Fermentas), 1 µL of genomic DNA(2 ng), and 0.5 µL of FLX Titanium amplicon primers 454T-RA and 454T-FB (20 pmol µL-1) for a 25 µL PCR reaction.These have the sequences for 16S primers 926f (aaa ctY aaaKga att gac gg) and 1392r (acg ggc ggt gtg tRc) as their 3′-ends. Primer 454T-RA has a 25 nt A-adaptor (CGTATCGC-CTCCCTCGCGCCATCAG), whereas primer 454T-FB has a25 nt B-adaptor sequence (CTATGCGCCTTGCCAGCCCGCT-CAG). Following PCR amplification (95 °C, 3 min; 25 cyclesof 95 °C 30 s, 55 °C 45 s, 72 °C 90 s; 72 °C 10 min; final holdat 4 °C) with primers 454T-RA and 454T-FB, the PCR productwas verified on a 0.7% agarose gel and purified with aQIAquick PCR Purification Kit (Qiagen). Its concentrationwas determined on a Qubit Fluorometer (Invitrogen), usinga Quant-iT dsDNA HS Assay Kit (Invitrogen). PCR products(typically 20 µL of 5 ng µL-1) were sent to the Genome Quebecand McGill University Innovation Centre, where they weresubjected to 10 PCR cycles with primers 454T-RA-X and 454T-FB, where X is a 10 nt multiplex identifier barcode. Thebarcoded PCR products were analyzed by pyrosequencing,using a GS FLX Titanium Series Kit XLR70 (Roche DiagnosticsCorporation). DNA from skim milk powder was extractedand amplified with the same protocol as used for the tailingspond samples.

Analysis of Pyrosequence Data. Pyrosequencing dataanalysis was performed using a locally developed Perl scriptpackage (available from the authors upon request). Everyraw pyrosequence had to pass a multistage quality inspectionto remove low quality reads and minimize sequencing errorsthat can be introduced during the pyrosequencing process(22, 23). Sequences were rejected if they: (i) did not haveperfect match with the pyrosequence primers, (ii) containedany ambiguities, (iii) had a typical length of 1 standarddeviation from the mean length after removing primersequences, (iv) had average quality scores were below 25(recommended for the Roche Genome Sequencer FLX System22, 23), and (v) had a homopolymer sequence longer thaneight base pairs. The remaining high quality sequences werecompared against the nonredundant SSU ref data set ofSILVA102 (24) using the Tera-BLAST algorithm on a Time-Logic Decypher system (Active Motif, Inc.) consisting of 12boards. Tera-BLAST results were used to screen for prob-lematic, chimerical, and eukaryotic sequences that were thenremoved. Any sequence having a best alignment coveringless than 70% or having a best BLAST search hit e-value ofgreater than e-50 was excluded as a problematic sequence.Putative chimeras were identified using a two-stage approach.First, sequences having a best alignment covering less than90% of the trimmed read length but with greater than 90%sequence identity to the best BLAST match were identifiedas potential chimeras. Second, potential chimeras wereexcluded from further analysis if they were also identified aschimeras at minimum 80% bootstrap support in chime-ra.slayer in the Mothur software package (25). In total, weeliminated 83,777 of 156,749 sequences using the described

440 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 45, NO. 2, 2011

quality control approach. The sequences passing all qualitycontrols were then clustered into OTUs (Operational Taxo-nomic Units) at a 3% distance using the complete linkagealgorithm in Mothur. A taxonomic consensus of representa-tive sequences from each OTU was derived from the recurringspecies within 5% of the best bitscore from a BLAST searchagainst the SILVA database.

Since the skim milk powder used as a clay-bindingcompetitor in the DNA extraction process contained its ownDNA sequences (Table S2), these had to be subtracted fromthe sequences in the tailings samples in order to more closelyreflect microbial diversity associated only with tailings.Following the analysis described above, data were rankedaccording to the most abundant sequences in the skim milkpowder in a descending manner (Table S2). Followinginspection of the data, the top 11 reads were considered tobe associated primarily with skim milk powder and wereremoved from the entire data set. These deleted sequencescomprised 99% of the sequences in the skim milk powder.The remaining sequences in the sampling set were recal-culated to total 100%. The correction step resulted in a finalnumber of 62,511 high quality reads. Of these, 93% could betaxonomically identified at the phylum level, 91% at the classlevel, and 59% at the genus level (Table S3). The entire setof high quality reads were assigned Genbank accessionnumbers HQ026812 to HQ104921.

Fast UniFrac (26) was used to quantitatively compare thephylogenetic compositions of each sampling depth usingthe corrected sample set. Samples were clustered togetherusing the hierarchical clustering method (UPGMA) based onweighted and normalized UniFrac distances (27). The newickformat of the sample relation tree generated from FastUniFrac was visualized using Dendroscope (28).

ResultsPond Chemistry and Microbial Activity. At the time ofsampling, the temperature of pond 6 was 7 °C at the surface,increasing to 19 °C at 18.3 mbs, the pH ranged from 7.0 to7.8 (Table S1), and the dissolved oxygen was highest at thesurface with a rapid drop below the surface water (14). Ofthe organic acids measured, only acetate was detected andwas only present at some depths (Table S1). We previouslyreported the depth-dependent sulfate and sulfide concen-

trations and sulfate reduction and methanogenesis rates forpond 6 (14). In summary, we found that sulfate concentrationswere highest at the surface where no sulfide was detected,but then the levels of these components dropped to nearzero with increasing depth. In deeper zones, sulfide andsulfate concentrations increased again ((14), Table S1),correlating with the highest measured SRR. Methanogenesisrates fluctuated as a function of depth and were among thelowest when the SRR was highest ((14), Figure 1). Theseobservations suggested an inhibition of methanogenesiswhen sulfate was present. To test this, laboratory incubationscontaining anaerobic tailings collected from a depth wherethe methanogenesis rate was relatively high (6.1 mbs) werechallenged with increasing concentrations of sulfate. Indeed,we observed that increasing sulfate concentrations resultedin correspondingly lower methane production (Figure 2). Inincubations containing 2 mM sulfate, which approximatesthe sulfate concentration in some depths of the pond,methane production was reduced by almost 50%. Interest-ingly, some methanogenesis was still observed in the presenceof 12 mM sulfate, a concentration far greater than what wasmeasured in pond samples ((14), Table S1).

We hypothesized that the absence of sulfide from thesurface layers is due to its chemical or microbial oxidation.When sulfide was incubated with tailings in the presence ofair (20% v/v O2), complete oxidation was observed within 80to 100 min (Figure S1A). The sulfide oxidation rate was onlymarginally faster in nonautoclaved tailings, indicating thatsulfide oxidation is mainly a chemical process. Sulfate wasnot detected as an end-product in these incubations (FigureS1B), supporting chemical oxidation as a primary process(that typically does not proceed beyond sulfur). Hence, themeasured high sulfate concentration at the pond surface((14), Table S1) may be caused by chemical and microbialoxidation of sulfide to sulfur, followed by microbial oxidationof sulfur to sulfate.

Microbial Community Composition as a Function ofDepth. A phylum-level survey of the pond microbial com-munity composition is shown in Figure 3. In general, samplescollected from 1.5 to 18.3 mbs exhibited a similar compositionwith the exception of the sample collected at 16.8 m showinga higher abundance of Actinobacteria than the other depths.In contrast, the surface water sample was distinctly different

FIGURE 1. Depth-dependent profiles of (A) total SRB (9) and sulfate reduction rates (4) and (B) total syntrophs (2), methanogens (9),and methanogenesis rates (O). Microbial group abundances are calculated as the % of total corrected reads from Table S3. Sulfatereduction and methanogenesis rates are replotted from ref 14 for comparative purposes. Total SRB include the genera Desulfocapsa,Desulfurivibrio, Desulfobacterium, Desulfuromonas, Desulfotomaculum, Desulfobulbus, Desulfomicrobium, Desulfobacca,Desulfofustis, Desulfosarcina, Desulfobacter, and Desulfovibrio; total syntrophs include the genera Pelotomaculum, Syntrophus, andSmithella; total methanogens include the genera Methanosaeta, Methanoregula, Methanolinea, Methanomethylovorans,Methanobacterium, and Methanoculleus.

VOL. 45, NO. 2, 2011 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 441

at the phylum level, which was confirmed by a cluster analysis(Figure S2). This was also evident at the genus level, wheremembers of the methylotrophic genus Methyloversatilis, theN2-fixing genus Azospirillum, and the planctomycete genusGemmata were most prominent and where the majority of theorganisms identified at greater depths were absent (Table S3).

Despite phylum-level similarities between the nonsurfacesamples, differences between communities were apparentat the genus level. Absent from the surface but found in thedeeper anaerobic zones were sequences affiliating with threesyntrophic genera that included Pelotomaculum, Smithella,and Syntrophus (Figure 4A). The most prominent metha-nogens identified were members of the obligate acetate-using genus Methanosaeta, along with Methanolinea andMethanoregula, two H2-using genera (29, 30; Figure 4B). Themost abundant sequences that affiliated phylogeneticallywith SRB were Desulfocapsa and Desulfurivibrio spp. (Figure4C), followed by Desulfobacterium and Desulfuromonas spp.(Table S3). Other genera that were abundant only in thenonsurface samples included Leptolinea, Thauera, Rhodof-erax, and Acidovorax (Figure 4D). Some sequences were onlyabundantly found in the deepest regions sampled, includingmembers of the genera Brachymonas, Thiobacillus, andCellulomonas (Figure 4E).

DiscussionIntegration of Microbial Activity and Community Profiles.The management of bitumen extraction waste in tailingsponds is an integral part of the oil sands mining operations

in northern Alberta, which provide a substantial portion ofthe global fossil fuel inventory. The optimization of tailingsmanagement practices is ongoing. Thus understanding theunderlying processes that influence tailings behavior andultimate reclamation is an important endeavor. Microbialcommunities in the tailings ponds can impact pond man-agement as their activities influence densification, may leadto gas emissions, and promote hydrocarbon biodegradation(5, 9, 12, 31). Understanding how microbial activity is affectedas layers of consolidated material form and fill in the pondover time is also needed.

We therefore assessed the microbial community structurein relation to measured microbial activities as a function ofdepth in an active tailings pond treated with gypsum in orderto understand the impact on carbon and sulfur cycling andtailings management. Holowenko et al. (9) previously re-ported methane and sulfate concentrations in a differenttailings pond (MLSB) as a function of depth from 1 to 20 mbsat two sampling events a year apart. Methane concentrationsvaried with depth from 1 to 77 mL CH4 L-1 of tailings, whereassulfate concentrations were highest at 1 mbs (150 to 175 mgL-1 or 1.6-1.8 mM) and dropped with increasing depth (9).We also found the highest sulfate concentrations at the pondsurface (6 mM), which also dropped to near zero withincreasing depth but then rebounded at even greater depths((14), Table S1). This deeper sulfate-containing zone (10 to14 mbs, Table S1) presumably provided a source of sulfatefor SRB, as the highest SRR was measured within these depths(at 12.2 mbs, Figure 1A). These observations, coupled with

FIGURE 2. Effect of sulfate concentrations on methane production in laboratory incubations containing tailings pond materialsampled from Suncor pond 6.

FIGURE 3. Microbial community profile at the phylum level as a function of depth in Suncor tailings pond 6.


the measured fluctuating methanogenesis rates ((14), Figure1B), attest to the highly stratified nature and heterogeneityof oil sands tailings ponds (9, 13).

The microbial community changes as a function of depth,assessed using state-of-the-art sequencing technologies,confirm the stratified nature of the pond and indicate whereoil-associated substrates or metabolites are presumablymetabolized coupled with sulfide and/or methane produc-tion. It should be pointed out that since our survey was PCR-based, it likely suffered from PCR biases even though theprimers used were designed to amplify a wide range ofBacteria and Archaea (32). For example, groups like theEpsilonproteobacteria, appeared to be underrepresented(data not shown). Hence the survey presented here has itslimitations and should not be overinterpreted.

However, the sequences identified in the 16S surveyaligned closely with the measured microbial activities.Notably, the abundance profile of SRB mirrored the measuredSRR (Figure 1A) and the syntroph and methanogen abun-dance profiles closely mimicked the fluctuating methano-genesis rates (Figure 1B). Further, many of the identifiedgenera have been reported as community members in otheranaerobic and/or hydrocarbon-associated environments. Akey finding of the tailings pond 16S survey was that a varietyof syntrophs were prominent, with Pelotomaculum as themost abundant genus (Figure 4A). Cultured members of this

genus are syntrophs that utilize propionic acid in metha-nogenic environments (33, 34). Members of Syntrophus canmetabolize a variety of organic compounds including fattyacids and alcohols and are also prevalent in anaerobicconsortia that metabolize oil to methane (35-37). Smithellaspp. have been detected in methanogenic environments andin oil-degrading methanogenic consortia (38). Such syntrophicactivity is usually coupled to that of H2- or acetate-usingmethanogens, and indeed both kinds of methanogens wereprevalent throughout the tailings pond, peaking in abundanceat 7.6 mbs where Syntrophus and Smithella spp. also peakedand where the methanogenesis rate was among the highest(Figures 1B, 4A and 4B). However, consortia of syntrophsand SRB in which syntrophs catalyze the hydrocarbon attackand SRB use the resulting H2, acetate, and other low molecularweight electron donors for sulfate reduction can also beenvisioned when sulfate is available (39). For example, themost abundant SRB identified are known to utilize H2 and/or low molecular weight fatty acids, although Desulfocapsaspecies can also utilize hydrocarbons (40, 41). Acetate wasdetected in some of the depths sampled (Table S1), showingthat a readily available carbon source is present or isgenerated (e.g., through hydrocarbon metabolism) that cansupport many members of the tailings pond community.Such observations suggest a complex interplay among SRB,syntrophs, and methanogens and in fact all three groups of

FIGURE 4. Distribution of selected most abundant microbial genera (as % of total corrected reads) in Suncor tailings pond 6 as afunction of depth. (A) Syntrophs: Pelotomaculum (0), Smithella (2), and Syntrophus (O); (B) Methanogens: Methanosaeta (O);Methanoregula (9); Methanolinea (4); (C) Sulfate- and sulfur-utilizers: Desulfocapsa (2); Desulfurivibrio (4); (D) Other abundantgenera, Leptolinea (2); Thauera (O); Rhodoferax (9); Acidovorax (4); (E) Genera most prevalent in the deepest samples taken:Thiobacillus (4); Brachymonas (9); Cellulomonas (O).


organisms are relatively abundant throughout the pond(Figures 4A, 4B, 4C). The availability of sulfate likely dictatesthe dominant syntrophic processes. Cultivated members ofother major genera identified also have known hydrocarbon-degrading ability including Brachymonas, Thauera, Rhod-oferax, and Cellulomonas (42-45). The nature of the sub-strates used by the identified organisms is not currentlyknown, but bitumen diluent components (low molecularweight hydrocarbons) that enter tailings ponds can bebiodegraded under methanogenic conditions (5). Naphthenicacids may also serve as carbon and energy sources in tailingsbut can be recalcitrant (3, 4, 8, 9). Determining the effectsof such carbon sources on tailings community compositionwill help to clarify which of the detected anaerobes use thepotential substrates and will be the focus of a futurecontribution. The finding that a high abundance of SRBsequences correlates with high SRR is also significant andwill allow rapid gene probe-based evaluation of the abilityof pond communities to catalyze sulfate reduction.

Carbon and Sulfur Cycling and Implications for TailingsPond Management. On average, the sum of soluble sulfideand sulfate concentrations throughout pond 6 was 2.5 mM,less than the 7.3 mM that would be expected based on thereported average dose of 1 kg of gypsum per m3 of tailings(2). We hypothesize that this difference is due primarily tosulfide precipitation or oxidation, not H2S emission. Theconcentration of acid-soluble metal sulfides (e.g., FeS) intailings ponds can be up to 10-fold higher than the dissolvedsulfide concentration (E. Ramos-Padron, unpublished), thussome of the missing sulfide may be retained in the tailingsponds in precipitated form. Some of the identified communitymembers, including Rhodoferax, Acidovorax, and Thioba-cillus spp., are known to play roles in the cycling of sulfuror iron species in anoxic environments (44, 46, 47) and maybe contributing to such precipitation. Moreover, hydrogensulfide (H2S) has high water solubility, which is furtherincreased by dissociation to HS-. However, H2S formed indeep pond layers may partition in bubbles of methane(solubility in water only 25 ppm at 20 °C) also formedthroughout the pond (e.g., Figure 1B), which move upwardallowing H2S to migrate to upper layers. Repartitioning ofsuch “hitchhiking” H2S into water and its oxidation to sulfuror sulfate near the surface (through chemical and microbialprocesses, Figure S1) may contribute to the higher sulfateconcentrations measured at the surface. Members of theabundant Desulfocapsa and Desulfurivibrio genera may alsobe contributing to this process through sulfur and thiosulfatedisproportionation to sulfide and sulfate (48) or sulfuroxidation (49), respectively. Hence, although a certain levelof H2S emission can potentially occur, based on our data wehypothesize that sulfide produced via microbial sulfatereduction largely remains in the pond through precipitationand oxidation. Further work examining the nature of thesulfur species formed at the pond surface and the measure-ment of actual H2S gas emissions will be required to test thishypothesis.

The measured sulfate reduction activity can also affectpotential methane emissions. Our data showed that thepresence of sulfate impedes methane production, even atthe concentrations that are found in the tailings pond (e.g.,2 mM, Figure 2). Such an effect would be predicted basedon the known energetic and thus competitive advantage ofsulfate-reducing versus methanogenic processes. Becausethe reduction of CO2 to methane and the reduction of sulfateto sulfide both require 8 electrons, sulfate reduction by SRBwill prevent the formation of an equivalent amount ofmethane, irrespective of which oil electron donor ultimatelyserved to reduce sulfate or CO2. We previously calculatedthat the average SRR of 10 mmol m-3 d-1 could represent amethane emission saving of 2 million L of methane per day

for a tailings pond occupying a volume of 107 m3 (14), althoughsuch a calculation remains to be validated by comparingwith real-time pond emissions.

However, methane formation in tailings may also helpwith pond management by increasing the rate of densifi-cation. In a long-term study where mesocosms were preparedwith mature fine tailings (MFT) from the MLSB, Fedorak etal. (12) found that methanogenic MFT showed a higher rateof densification than nonmethanogenic MFT. These resultswere consistent with observations made by Syncrude opera-tors who noticed that an increase in the densification of MFTcoincided with the onset of methane emissions from thistailings basin. It is currently not known whether the methaneitself, or another factor such as the microbial biomass, causesincreased tailings densification. An open question in thisregard is whether other microbes contribute to tailingsdensification. Sulfate-reducing activity will almost certainlyprevent mobilization of heavy metal ions in tailings byprecipitating these as metal sulfides. However, it is not knownwhether SRB or other genera like the filamentous, anaerobicLeptolinea (50) also promote tailings densification, whichlike methanogens are abundant throughout pond 6 (Figure4D). Thauera, also prevalent throughout the pond layers,was recently found to contribute to tailings densification inlaboratory experiments (31). In addition to SRB, syntrophs,and methanogens, the deepest zones of the tailings pondsharbor bacteria not found elsewhere (Figure 4E), includingBrachymonas, Thiobacillus, and Cellulomonas. We do notknow what specifically attracts them to such an environment,but the possibility that such organisms actively promotedensification should be considered.

AcknowledgmentsL.M.G. and G.V. were supported by funding from GenomeCanada, Genome Alberta, the Government of Alberta, andGenome BC. C.W.S. is the Principal Investigator of theGenome Canada/Genome Alberta-funded “Distributed andIntegrated Bioinformatics Platform for Genome Canada”,used for the work described here. This work was alsosupported by an NSERC Industrial Research Chair Award toG.V., which was also supported by Baker Hughes Inc.,Commercial Microbiology Ltd. (Intertek), Computer Mod-elling Group Ltd., ConocoPhillips Co., YPF SA, AramcoServices, Shell Canada Ltd., Suncor Energy DevelopmentsInc., Yara International ASA, and by the Alberta EnergyResearch Institute. The authors are grateful for administrativesupport by Drs. Rhonda Clark and Sean Caffrey.

Supporting Information AvailableTwo figures (S1 and S2) and three tables (S1-S3) showadditional geochemical data, sulfide oxidation results, andmolecular ecology findings. The material is available free ofcharge via the Internet at http://pubs.acs.org.

Literature Cited(1) Zhou, S. Q.; Huang, H. P.; Liu, Y. M. Biodegradation and origin

of oil sands in the Western Canada Sedimentary Basin. Petrol.Sci. 2008, 5, 87–94.

(2) Chalaturnyk, R.; Scott, J. D.; Ozum, B. Management of oil sandstailings. Petrol. Sci. Technol. 2002, 1025–1046.

(3) Scott, A. C.; MacKinnon, M. D.; Fedorak, P. M. Naphthenic acidsin Athabasca oil sands tailings waters are less biodegradablethan commercial naphthenic acids. Environ. Sci. Technol. 2005,39, 8388–8394.

(4) Clemente, J. S.; Fedorak, P. M. A review of the occurrence,analyses, toxicity, and biodegradation of naphthenic acids.Chemosphere 2005, 60, 585–600.

(5) Siddique, T.; Fedorak, P. M.; Foght, J. M. Biodegradation ofshort-chain n-alkanes in oil sands tailings under methanogenicconditions. Environ. Sci. Technol. 2006, 40, 5459–5464.

(6) Government of Alberta. www.environment.alberta.ca/02011.ht-ml (accessed June 2010).


(7) Energy Resources Conservation Board. 2010. www.ercb.ca/docs/documents/directives/Directive074.pdf(accessed March 30 2010).

(8) Herman, D. C.; Fedorak, P. M.; Mackinnon, M. D.; Costerton,J. W. Biodegradation of naphthenic acids by microbial popula-tions indigenous to oil sands tailings. Can. J. Microbiol. 1994,40, 467–477.

(9) Holowenko, F. M.; MacKinnon, M. D.; Fedorak, P. M. Metha-nogens and sulfate-reducing bacteria in oil sands fine tailingswaste. Can. J. Microbiol. 2000, 46, 927–937.

(10) Fedorak, P. M.; Coy, D. L.; Salloum, M. J.; Dudas, M. J.Methanogenic potential of tailings samples from oil sandsextraction plants. Can. J. Microbiol. 2002, 48, 21–33.

(11) Salloum, M. J.; Dudas, M. J.; Fedorak, P. M. Microbial reductionof amended sulfate in anaerobic mature fine tailings from oilsand. Waste Manage Res. 2002, 20, 162–171.

(12) Fedorak, P. M.; Coy, D. L.; Dudas, M. J.; Simpson, M. J.;Renneberg, A. J.; MacKinnon, M. D. Microbially-mediatedfugitive gas production from oil sands tailings and increasedtailings densification rates. J. Environ. Eng. Sci. 2003, 2, 199–211.

(13) Penner, T. J.; Foght, J. M. Mature fine tailings from oil sandsprocessing harbour diverse methanogenic communities. Can.J. Microbiol. 2010, 56, 459–470.

(14) Bordenave, S.; Ramos, E.; Lin, S.; Voordouw, G.; Gieg, L.; Guo,C.; Wells, S. Use of calcium sulfate to accelerate densificationwhile reducing greenhouse gas emissions from oil sands tailingsponds. Proceedings of the Canadian International PetroleumConference, 2009; Paper 2009-081 (accessible through thedatabase One Petro).

(15) Lambo, A. J.; Noke, K.; Larter, S. R.; Voordouw, G. Competitive,microbially-mediated reduction of nitrate with sulfide andaromatic oil components in a low-temperature, WesternCanadian oil reservoir. Environ. Sci. Technol. 2008, 42, 8941–8946.

(16) Cline, J. D. Spectrophotometric determination of hydrogensulfide in natural waters. Limnol. Oceanogr. 1969, 14, 454–458.

(17) Ulrich, G. A.; Krumholz, L. R.; Sulfita, J. M. A rapid and simplemethod for estimating sulfate reduction activity and quantifyinginorganic sulfides. Appl. Environ. Microbiol. 1997, 63, 1672–1630.

(18) McInerney, M. J.; Bryant, M. P.; Pfennig, N. Anaerobic bacteriumthat degrades fatty acids in syntrophic association with metha-nogens. Arch. Microbiol. 1979, 122, 129–135.

(19) Kaster, K. M.; Grigoriyan, A.; Jenneman, G.; Voordouw, G. Effectof nitrate and nitrite on sulfide production by two thermophilic,sulfate-reducing enrichments from an oil field in the North Sea.Appl. Microbiol. Biotechnol. 2007, 75, 195–203.

(20) Hubert, C.; Nemati, M.; Jenneman, G.; Voordouw, G. Contain-ment of biogenic sulfide production in continuous up-flowpacked-bed bioreactors with nitrate or nitrite. Biotechnol. Prog.2003, 19, 338–345.

(21) Takada-Hoshino, Y.; Matsumoto, N. An improved DNA extrac-tion method using skim milk for soils that strongly adsorb DNA.Microbes Environ. 2004, 19, 13–19.

(22) Huse, S. M.; Huber, J.; Morrison, H. G.; Sogin, M. L.; Welch,D. M. Accuracy and quality of massively parallel DNA pyrose-quencing. Genome Biol. 2007, 8, R143.

(23) Huse, S. M.; Welch, D. M.; Morrison, H. G.; Sogin, M. L. Ironingout the wrinkles in the rare biosphere through improved OTUclustering. Environ. Microbiol. 2010, 12, 1889–1898.

(24) Pruesse, E.; Quast, C.; Knittel, K.; Fuchs, B. M.; Ludwig, W. G.;Peplies, J.; Glockner, F. O. SILVA: a comprehensive onlineresource for quality checked and aligned ribosomal RNAsequence data compatible with ARB. Nucl. Acid Res. 2007, 35,7188–7196.

(25) Schloss, P. D.; Westcott, S. L.; Ryabin, T.; Hall, J. R.; Hartmann,M.; Hollister, E. B.; Lesniewski, R. A.; Oakley, B. B.; Parks, D. H.;Robinson, C. J.; Sahl, J. W.; Stres, B.; Thallinger, G. G.; Van Horn,D. J.; Weber, C. F. Introducing mothur: open-source, platform-independent, community-supported software for describing andcomparing microbial communities. Appl. Environ. Microbiol.2009, 75, 7537–7541.

(26) Hamady, M.; Lozupone, C.; Knight, R. Fast UniFrac: facilitatinghigh-throughput phylogenetic analyses of microbial communi-ties including analysis of pyrosequencing and PhyloChip data.ISME J. 2010, 4, 17–27.

(27) Lozupone, C.; Hamady, M.; Knight, R. UniFrac-an online toolfor comparing microbial community diversity in a phylogeneticcontext. BMC Bioinf. 2006, 7, 371.

(28) Huson, D.; Richter, D.; Rausch, C.; Dezulian, T.; Franz, M.; Rupp,R. Dendroscope: An interactive viewer for large phylogenetictrees. BMC Bioinf. 2007, 8, 460.

(29) Imachi, H.; Sakai, S.; Sekiguchi, Y.; Hanada, S.; Kamagata, Y.;Ohashi, A.; Harada, H. Methanolinea tarda gen. nov., sp. nov.,a methane-producing archaeon isolated from a methanogenicdigester sludge. Int. J. Syst. Evol. Microbiol. 2008, 58, 294–301.

(30) Yashiro, Y.; Sakai, S.; Ehara, M.; Miyazaki, M.; Yamaguchi, T.;Imachi, H. Methanoregula formicica sp. nov., a novel methane-producing archaeon isolated from methanogenic sludge. IJSEMPapers in Press. Published August 10, 2009 as doi:10.1099/ijs.0.014811-0.

(31) Bordenave, S.; Kostenko, V.; Dutkoski, M.; Grigoryan, A.;Martinuzzi, R. J.; Voordouw, G. Relation between the activityof anaerobic microbial populations in oil sands tailings pondsand the sedimentation of tailings. Chemosphere 2010, 81, 663–668.

(32) Engelbrektson, A.; Kunin, V.; Wrighton, K. C.; Zvenigorodsky,N.; Chen, F.; Ochman, H.; Hugenholtz, P. Experimental factorsaffecting PCR-based estimates of microbial species richnessand evenness. ISME J. 2010, 4, 642–647.

(33) Lueders, T.; Pommerenke, B.; Friedrich, M. W. Stable-isotopeprobing of microorganisms thriving at thermodynamic limits:Syntrophic propionate oxidation in a flooded soil. Appl. Environ.Microbiol. 2004, 70, 5778–5786.

(34) Imachi, H.; Sakai, S.; Ohashi, A.; Harada, H.; Hanada, S.;Kamagata, Y.; Sekiguchi, Y. Pelotomaculum propionicum sp.nov., an anaerobic, mesophilic, obligately syntrophic propi-onate-oxidizing bacterium. Int. J. Syst. Evol. Microbiol. 2007,57, 1487–1492.

(35) Zengler, K.; Richnow, H. H.; Rossello-Mora, R.; Michaelis, W.;Widdel, F. Methane formation from long-chain alkanes byanaerobic microorganisms. Nature 1999, 401, 266–269.

(36) Jones, D. M.; Head, I. A.; Gray, N. D.; Adams, J. J.; Rowan, A. K.;Aitken, C. M.; Bennett, B.; Huang, H.; Brown, A.; Bowler, B. F. J.;Oldenburg, T.; Erdmann, M.; Larter, S. R. Crude oil biodegrada-tion via methanogenesis in subsurface petroleum reservoirs.Nature 2008, 451, 176–180.

(37) McInerney, M. J.; Struchtemeyer, C. G.; Sieber, J.; Mouttaki, H.;Stams, A. J. M.; Schink, B.; Rohlin, L.; Gunsalus, R. P. Physiology,ecology, phylogeny, and genomics of microorganisms capableof syntrophic metabolism. Incredible Anaerobes: from Physiologyto Genomics to Fuels 2008, 1125, 58–72.

(38) Gray, N. D.; Sherry, A.; Hubert, C.; Dolfing, J.; Head, I. M.Methanogenic degradation of petroleum hydrocarbons insubsurface environments: remediation, heavy oil formation, andenergy recovery. Adv. Appl. Microbiol. 2010, 72, 137–161.

(39) Grigoryan, A.; Voordouw, G. Microbiology to help solve ourenergy needs: methanogenesis from oil and the impact of nitrateon the oil field sulfur cycle. Ann. N.Y. Acad. Sci. 2008, 1125,345–352.

(40) Winderl, C.; Anneser, B.; Griebler, C.; Meckenstock, R. U.;Lueders, T. Depth-resolved quantification of anaerobic toluenedegraders and aquifer microbial community patterns in distinctredox zones of a tar oil contaminant plume. Appl. Environ.Microbiol. 2008, 74, 792–801.

(41) Bombach, P.; Chatzinotas, A.; Neu, T. R.; Kastner, M.; Lueders,T.; Vogt, C. Enrichment and characterization of a sulfate-reducing toluene-degrading microbial consortium by combiningin situ microcosms and stable isotope probing techniques. FEMSMicrobiol. Ecol. 2010, 71, 237–246.

(42) Rouviere, P. E.; Chen, M. W. Isolation of Brachymonas petro-leovorans CHX, a novel cyclohexane-degrading beta-proteo-bacterium. FEMS Microbiol Lett. 2003, 227, 101–106.

(43) Biegert, T.; G. Fuchs, G.; Heider, J. Evidence that anaerobicoxidation of toluene in the denitrifying bacterium Thaueraaromatica is initiated by formation of benzylsuccinate fromtoluene and fumarate. Eur. J. Biochem. 1996, 238, 661–668.

(44) Eriksson, S.; Ankner, T.; Abrahamsson, K.; Hallbeck, L. Propy-lphenols are metabolites in the anaerobic biodegradation ofpropylbenzene under iron-reducing conditions. Biodegradation2005, 16, 253–263.

(45) Brito, E. M. S.; Guyoneaud, R.; Goni-Urriza, M.; Ranchou-Peyruse, A.; Verbaere, A.; Crapez, M. A. C.; Wasserman, J. C. A.;Duran, R. Characterization of hydrocarbonoclastic bacterialcommunities from mangrove sediments in Guanabara Bay,Brazil. Res. Microbiol. 2006, 157, 752–762.

(46) Straub, K. L.; Schonhuber, W. A.; Buchholz-Cleven, B. E. E.;Schink, B. Diversity of ferrous iron-oxidizing, nitrate-reducingbacteria and their involvement in oxygen-independent ironcycling. Geomicrobiol. J. 2004, 21, 371–378.

(47) Kellermann, C.; Griebler, C. Thiobacillus thiophilus sp. nov., achemolithoautotrophic, thiosulfate-oxidizing bacterium isolatedfrom contaminated aquifer sediments. Int. J. Syst. Evol. Mi-crobiol. 2009, 59, 583–588.


(48) Finster, K.; Liesack, W.; Thamdrup, B. Elemental sulfur andthiosulfate disproportionation by Desulfocapsa sulfoexigenssp. nov., a new anaerobic bacterium isolated from marinesurface sediment. Appl. Environ. Microbiol. 1998, 64, 119–125.

(49) Sorokin, D. Y.; Tourova, T. P.; Mussmann, M. Dethiobacteralkaliphilus gen. nov. sp. nov., and Desulfurivibrio alkaliphilusgen. nov. sp. nov.: two novel representatives of reductive sulfurcycle from soda lakes. Extremophiles 2008, 12, 431–439.

(50) Yamada, T.; Sekiguchi, Y.; Hanada, S.; Imachi, H.; Ohashi, A.;Harada, H.; Kamagata, Y. Anaerolinea thermolimosa sp. nov.,Lilinea saccharolytica gen. nov., sp. nov. and Leptolinea tar-divitalis gen. nov., sp. nov., novel filamentous anaerobes, anddescription of the new classes Anaerolineae classis nov. andCaldilineae classis nov. in the bacterial phylum Chloroflexi. Int.J. Syst. Evol. Microbiol. 2006, 56, 1331–1340.

ES1028487


carbon and sulfur cycling by microbial communities in a gypsum-treated oil sands tailings pond

Documents