An experiment was conducted in 50-L columns using MFT collected from Mildred Lake Settling Basin at Syncrude Canada Ltd. in Fort McMurray, Alberta, Canada (Figure A1). Detailed information about setting up this experiment has been described by Siddique et al. (2014). Briefly, two 50-L acrylic columns were each filled with 45 L MFT under a curtain of N₂ gas and sealed with an N₂ headspace to maintain anaerobic conditions. One column contained MFT amended with hydrolyzed canola meal as carbon source (final concentration 400 mg C L⁻¹ tailings) to enhance indigenous microbial activity. The canola is not considered a source of viable microbes nor amplifiable DNA because it was hydrolyzed at high pH and 60°C for 24 h. Parallel experiments have not revealed any contributions of microbes from canola amendment to the tailings community (manuscript in preparation). This amendment had been determined previously to promote MFT consolidation at this concentration (Bressler et al., 2010); its preparation is described in Table 2 of a companion paper (Siddique et al., 2014). The other column contained unamended MFT as a baseline control in which only endogenous carbon was available to support microbial activity. The columns were incubated undisturbed in dim light at ~20°C for 213 days. Experimental details and analytical approach for determining biogenic gas production and composition, in situ pH, porewater recovery and consolidation of tailings, soluble cations, and anions, exchangeable cations, and carbonates minerals have been described in a companion paper (Siddique et al., 2014). The same samples were subjected to solid phase biogeochemical analyses including fractionation of iron (Fe) and sulfides, mineralogy of solid phase by Scanning Electron Microscopy (SEM) and synchrotron-based X-ray Absorption Near-Edge Structure (XANES) and characterization of microbial communities by 16S rRNA gene pyrosequencing as described below in this section.

To observe clay architectural patterns, whole (unmanipulated) MFT samples taken from column ports were cryo-fixed immediately using a cryogenic technique (Emitek K1250 cryogenic system) by plunging them into liquid N₂. The frozen samples were transferred under vacuum to the cold-stage of the cryo-preparation chamber for ice sublimation (10 min at pressure 10⁻⁵ Torr). After coating with gold, the samples were transferred into the cryo-SEM chamber, where they remained frozen during SEM imaging.

For conventional SEM and synchrotron XANES spectroscopy, a heavy liquid technique using a hydrophilic sodium polytungstenate solution of Na₆ (H₂W₁₂O₄₀) with density 3 g cm⁻³ (Franke et al., 2007) was used to separate the heavy mineral fraction (density >3 g cm⁻³). Approximately 5 g of MFT solids phase was dispersed in 20 mL Na₆ (H₂W₁₂O₄₀) solution with manual shaking in 50-mL centrifuge tubes. The suspension was centrifuged for 15 min at 3075 g to separate the light and heavy fractions.

For SEM analysis of the heavy fraction to examine mineralogy, samples were coated with carbon using an evaporative carbon coater (Leica EM SCD005). The samples were examined using SEM (JEOL—6301F) with field emission, equipped with a liquid N₂-cooled lithium drifted silicon X-ray energy-dispersive spectrometer (PGT). XANES was performed to determine the Fe-bearing minerals present in the heavy fraction of the MFT solid phase using Fe K-edge (E = 7112 eV). The samples were mounted on Kapton tape (DuPont) and oriented at 45° prior to X-ray beam exposure at the VESPERS beamline in the Canadian Light Source synchrotron facility (Saskatoon, Canada). The scans ranged between 30 eV below and above the absorption edge with a 0.5-eV step size. At least five spectra were collected from each experimental sample and from the reference samples (standards of pure Fe minerals) used in the study. The standard minerals (analytical grade and natural minerals) were purchased from WARD Canada and Sigma-Aldrich (Canada). To avoid oxidation of Fe^II, strictly anoxic conditions were maintained throughout the entire process of sampling, transport, and preparation of the samples for XANES analysis.

The total genomic DNA was extracted from MFT samples using a physical lysis method described previously (Foght et al., 2004). Triplicate 300-μl sub-samples were each extracted twice for a total of six extractions per sample. The precipitated DNA from the six extractions was pooled and used for PCR amplification. A negative control consisting of only extraction reagents was included with every set of samples to ensure contamination was not occurring at any stage of extraction.

After the negative extraction controls had been verified to contain no amplifiable DNA, each genomic DNA extract pool was PCR-amplified in triplicate (3 × 25 μL reactions) using the KAPA2G Robust HotStart DNA polymerase (Kapa Biosystems, Woburn MA). For every triplicate PCR reaction per sample, a negative PCR control was included containing only PCR reagents but no DNA template. Each 25-μL reaction consisted of: 1 μL of genomic DNA, 5 μL of 5X KAPA2G Buffer A, 5 μL of 5X KAPA Enhancer I, 1.25 μL of 100% dimethylsulfoxide, 0.5 μL of a 10 mM dNTP mix, 2 μL of 25 mM MgCl₂, 2.5 μL of a 2.5 μM working solution of forward primer, 2.5 μL of a 2.5 μM working solution of reverse primer, and 0.5 U of the KAPA2G Robust HotStart DNA polymerase; all reagents were molecular biology grade. The universal primers target both Bacteria and Archaea (An et al., 2013). The forward primer 454T_FB has the 926Fw sequence (AAACTYAAAKGAATTGRCGG) at its 3' end and a B-adaptor sequence (CTATGCGCCTTGCCAGCCCGCTCAG) at its 5' end. The reverse primer 454T_RL_X has the 1392R sequence (ACGGGCGGTGTGTRC) at its 3' end, an L-adaptor sequence (CCATCTCATCCCTGCGTGTCTCCGAC) at its 5' end and a sample-specific 10 nucleotide barcode sequence between adapter and primer, which allows for sample identification after multiplexing during pyrosequencing. The thermocycling program consisted of: one initial 5-min denaturation period at 95°C; 10 cycles of (30 s at 95°C, 30 s at 60°C decreasing by 0.5°C/cycle, and 30 s at 72°C); 30 cycles of (30 s at 95°C, 30 s at 55°C, and 30 s at 72°C); and one final 5 min extension period at 72°C.

Triplicate PCR products from each sample were pooled and an aliquot was examined for quality control by electrophoresis on an agarose gel (i.e., single band of expected size). PCR products were then purified using a QIAquick PCR purification kit (Qiagen) according to the manufacturer's instructions. Purified PCR products were quantified in duplicate using the Nanodrop 2000c (Thermo Scientific) and diluted to 25 ng μL⁻¹. Diluted and purified PCR products were again quantified using Nanodrop and visually verified on an agarose gel before being sent to Genome Quebec Innovation Centre, Montreal, QC, Canada for pyrosequencing using a GS FLX Titanium Series XLR70 kit (Roche Diagnostics Corporation). Raw pyrosequencing data were analyzed using Phoenix 2, an in-house developed SSU rRNA data analysis pipeline providing quality control and chimera detection, following the procedure described by Soh et al. (2013). Pyrotag sequences (~7000–10,000 reads per sample) have been submitted to the NCBI Short Read Archive (http://www.ncbi.nlm.nih.gov/sra) under accession numbers SRR621691–SRR621697 representing, respectively: initial, U1, U2, U3, A1, A2, and A3 samples. Quality-verified sequences were compared against the SILVA 102 database (http://www.arb-silva.de) and clustered into Operational Taxonomic Units (OTUs) at 5% distance. OTUs with read abundance <1% in every sample were not considered further in community analysis; hence the total abundance reported is <100%. To compare communities in samples, non-metric multidimensional scaling (NMDS) analysis was implemented in Phoenix 2.

Previous small-scale (2-L) experiments demonstrated that stimulating indigenous microbes in MFT with labile organic carbon sources enhanced clay consolidation and pore water recovery (Bressler et al., 2010). In the current experiment we used larger volumes of MFT (45 L) to reduce “wall effects” in the columns, and low concentrations of water-soluble canola meal hydrolysate as an organic carbon source. A parallel unamended MFT control column showed the effects of microbial metabolism of endogenous carbon (e.g., residual hydrocarbons and their metabolites; Siddique et al., 2007). The stationary acrylic columns containing amended or unamended MFT were incubated at ~20°C (circa in situ temperature in the tailings ponds) under a N₂ headspace to maintain anaerobic conditions. As described in a companion paper (Siddique et al., 2014), the MFT underwent bioconsolidation and de-watering measured, respectively, as a decrease in the height of the mudline (the interface between solids and overlying expressed porewater) and an increase in the volume of “cap water” above the mudline; this bioconsolidation was more pronounced in the amended than the unamended MFT column. Three ports in the sides of each column allowed access for measurements and sample collection; Port 1 in each column accessed the cap water whereas Ports 2 and 3 accessed tailings below the mudline; the latter MFT samples comprised porewater (described in the companion paper) and solid phase, described herein.

Methanogenic conditions prevailed in the amended column during incubation, as inferred from the composition of biogenic gas emitted and trapped: by 75 days, amended MFT had generated 2.8 L of CH₄ and ~0.5 L CO₂ in the column headspace vs. negligible gas emissions from unamended MFT (Siddique et al., 2014). We did not determine the redox potential (E_h) of the tailings during incubation because sampling would have disturbed the MFT structure, but we did measure E_h when samples were withdrawn at 213 days. Samples from the amended column were more reduced, where E_h ranged from −33 mV in cap water to −350 mV at port 3 (Figure 2) vs. the unamended column where suboxic conditions (E_h 209 mV) prevailed in the cap water while the MFT below the mudline experienced anoxic conditions with E_h of −148 at port 3 (Figure 2).

Because the surface properties of clays are of paramount importance for fines consolidation, minerals in the solids phase of amended and unamended MFT samples collected at Ports 2 and 3 were characterized and quantified. Fractionation and quantitation of Fe in the solid phase is important for inferring its role in microbial metabolism under methanogenic conditions and also for observing transformation (dissolution and precipitation) of minerals under reducing conditions. The forms of Fe in the initial MFT before any manipulation are shown in Table 1. Fe fractionation of the solid phase of unamended and amended MFT samples after 213 days of incubation (Figure 3A) was calculated from the data shown in Table 2 using the approaches described in Materials and Methods. In unamended MFT, Fe^III and Fe^II were present in almost equal concentrations and in crystalline form. Oxyhydroxides were major forms of crystalline Fe^III whereas crystalline Fe^II comprised pyrite (FeS₂), siderite (FeCO₃), and vivianite [Fe₃(PO₄)₂.8H₂O].

Synchrotron analyses (XANES; Figure 4) revealed that ferrihydrite (5Fe₂O₃. 9H₂O) and goethite (α-FeOOH) were dominant Fe^III minerals, whereas siderite was the dominant Fe^II mineral in both amended and unamended MFT (Table 3). During incubation and microbial metabolism in amended MFT, Fe^III decreased and Fe^II increased, with amorphous Fe^II as the dominant phase. Chemical analyses showed that amorphous FeS and crystalline vivianite increased in amended MFT whereas the concentration of pyrite remained unchanged (Figure 3B). These analyses indicate that significant transformation of iron minerals occurred in the amended MFT during incubation, concomitant with consolidation, de-watering, and gas production (Siddique et al., 2014).

The greater recovery of porewater from amended vs. unamended MFT (Siddique et al., 2014) led us to use cryo-SEM to examine the solid phase and determine the clay architecture of MFT samples collected at 213 days of incubation (Figure 5). Microscopy revealed aggregation of clay particles in the amended MFT, producing networks of pores between aggregates that presumably would facilitate egress of pore water and biogenic gas from MFT (Figure 5B), compared with the fine pore structure preserved in the unamended MFT (Figure 5A). Newly transformed amorphous Fe minerals coated the charged clay surfaces in the amended MFT and the coatings observed on aggregate surfaces in amended MFT (Figure 5D) were even more apparent using conventional SEM (Figure 5F). Energy dispersive spectroscopy (EDS) analyses performed during cryo-SEM confirmed that the clay aggregates in amended MFT were coated with amorphous Fe minerals (Figure 5H) that, according to mineralogical analyses described above (Figure 3, Table 3), were newly formed by transformation of crystalline iron minerals during incubation. In the unamended MFT, clay particles were observed randomly oriented and closely stacked, without any Fe coating (Figures 5A,C,E,G). These observations imply that transformation of iron minerals has changed the clay surfaces, allowing them to create larger pore structures stabilized by formation of new minerals.

To understand electron flow from labile organic carbon to Fe^III minerals in MFT via indigenous microorganisms, 16S rRNA gene pyrosequencing was performed. NMDS analysis was performed on the total communities in initial MFT (before manipulation) and in unamended and amended MFT samples taken from Port 1 (the water cap) and Ports 2 and 3 (tailings below the mud line) after 213 days incubation. Figure 6A shows that the unamended tailings communities were most similar to the initial MFT, whereas the amended tailings communities and both cap water communities differed from the original tailings communities and from each other.

The microbial community initially present in MFT before manipulation comprised a majority of methanogens (~65% of pyrosequences) dominated by putative hydrogenotrophs (43% Methanolinea/Methanoregula), with fewer acetoclastic methanogen sequences (~20% Methanosaeta) (Figure 6B). The bacterial component included Betaproteobacteria (family Comamonadaceae; 9%) and Chloroflexi (4%) as dominant phyla, with minor contributions by diverse bacterial taxa. After 213 days incubation, the community in unamended tailings below the mudline (Ports 2 and 3) still closely resembled the initial MFT community. However, Port 1 in the unamended column accessed the expressed pore water that collected above the mudline. The community structure of that cap water had shifted radically, being virtually depleted of methanogens but enriched in anaerobic Bacteroidetes (order Sphingobacteriales) and in Alpha- and Gammaproteobacteria, which include numerous facultative anaerobes. The proportion of reads assigned to “rare” taxa (each with abundance <1%) comprised nearly 40% of total reads, with virtually all being bacterial.

The microbial community detected in amended MFT solids also shifted during incubation, compared to the initial and the unamended solids. Fewer hydrogenotrophic methanogen sequences (Methanolinea and Methanoregula) resulted in a lower proportion of total methanogenic sequences, but the proportion of acetoclasts (Methanosaeta) was almost unchanged (Figure 6B). Among bacterial sequences, the diversity and abundance of Clostridiales (particularly Anaerobacter) and Synergistaceae (including Thermanaerovibrio) increased significantly in amended vs. initial and unamended MFT (Table 4), likely reflecting increased fermentation of the hydrolyzed canola meal amendment. In contrast, betaproteobacterial sequences (mainly in the family Comamonadaceae) decreased in amended MFT solids compared to initial MFT. The community in cap water overlying amended MFT diverged substantially from the MFT solids with a precipitous decline of archaeal sequences that were replaced by sequences affiliated with the uncharacterized, uncultivated candidate division W6. Sequences of the order Sphingobacteriales and little-known class Lentisphaeria also increased in abundance. The “rare” microbiome (each <1% abundance) comprised about one-third of the total pyrosequences. Thus, amendment shifted the MFT solids community slightly toward fermenters while retaining a substantial proportion of methanogens (>40%), but expression of pore water radically altered the community structure to become more diverse, no longer resembling the solids-associated community.

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.