Nine MIW sites were sampled between September and November of 2017 across four mines: Mine 1 (Cu, Au; Baie Verte, Newfoundland), Mine 2 (Zn, Cu; Snow Lake, Manitoba), Mine 3 (Cu, Ni; Sudbury, Ontario), and Mine 4 (Zn, Cu; Flin Flon, Manitoba; Table 1). In September, each mine's main tailings reservoir was sampled from, and when possible water sampling was also carried out in October (Mine 1) and in November (Mine 3). In addition, an input water source to the tailings reservoir at Mine 3, which combines mine treated water and natural tributaries was collected (September 2017), and a receiving environment site at Mine 4 (September 2017). The receiving environment site at Mine 4 was specially targeted as this local creek has experienced legacy acid mine drainage issues from historical mining activities (communication with mining personnel). Mine 3 has been the focus of past pilot research studies examining the relationships between microbial communities and acidity generation in a tailings reservoir (Bernier and Warren, 2005).

Each MIW site underwent (1) geochemical characterization (total sulfur (TotS), sulfate, thiosulfate, sulfite, nitrate, and nitrite concentrations), (2) microbial community structure assessment (16S rRNA, Illumina MiSeq), and (3) sulfur oxidizing microbial enrichment (SoxBac) as outlined in Supplemental Material (Figure S1). On site temperature, pH, dissolved oxygen saturation (%), conductivity, and redox as (ORP) were measured with a YSI 600 XLM (Mine 3) and ProDSS water quality meter (Mine 2 and 4). In situ physiochemical parameters were not available for Mine 1 tailing reservoirs samples, however pH was measured after arrival to the laboratory. Continued sampling from these sites since the 2017 field campaigns reported in this study confirm that circum-neutral pH (as was measured in the samples upon arrival) is typical of Mine 1's reservoir and therefore considered representative of on-site conditions.

Water samples were collected at Mine 1, 2, and 4 sites and shipped for processing from these remote locations to the University of Toronto, while Mine 3 MIW samples were collected and processed directly on site (Figure S1). Mine 1 and 2 tailing reservoirs are significantly shallower (~1–2 m) than Mine 3 and 4 (~40–10 m, respectively). Water samples from the tailings reservoirs (Mine 3 and 4) were collected using a dedicated Van Dorn sampler as the water columns are significantly deeper, while the remainder of samples were collected using surface grab samplers. Both the VanDorn and surface grab sampler were rinsed with 70% ethanol and rinsed with site water three times prior to sample collection. In situ water was collected for geochemical and microbial community assessment from Mine 2 and 4 into 20 liter containers with ethanol-sterilized polyethylene liners (4 mil, polyethylene, 18″ × 24″, Uline S-1379). Liners were fitted into the container and water samples were transferred directly to the liner using either the VanDorn or the surface grab sampler. Once appropriate volumes had been collected (10–20 L), liners were immediately sealed with no headspace. To remain consistent, the same protocol was used to collect water from Mine 3 for microbial community analyses. Mine 1 water samples were collected using ethanol-sterilized plastic containers for sampling and transport of samples. Mine 1 tailing reservoir water was collected by submerging the container until it was overflowing (no headspace).

Due to the remote locations of Mines 1, 2, and 4, water samples were shipped through air transport, which took between 3 and 10 days to arrive at the University of Toronto. Ongoing work is looking at the potential changes to microbial community structure that could occur during that time but all preliminary work to date has supported no significant changes in sulfur speciation occurring during that transport time. This would indirectly support no major activity changes to the sulfur oxidizing portion of the community. Preliminary lab tests have demonstrated the stability of relevant sulfur species (sulfate, thiosulphate, trithionate, and tetrathionate) within these circum-neutral waters for time periods comparable to shipment times (1–10 days) supporting low activity rates of the sulfur oxidizing portions of these microbial communities.

Water samples were collected in triplicate (40 mL volume per replicate) for total sulfur (unfiltered water samples; TotS_UF) and total dissolved sulfur (TotS_0.2um; MIW filtered through Pall Acrodisc® 25 mm 0.2 μm Supor® membrane filters with polypropylene syringes) into 50 mL Falcon^TM tubes pre-spiked with 80 μL of HNO₃ (Optima grade, Fisher Chemical) and then stored at 4°C until further processing. Analysis was carried out at the Commonwealth Scientific and Industrial Research Organization (CSIRO) through inductively coupled plasma atomic emission spectroscopy (ICP-AES) on a Varian730 ES (Mulgrave, Australia). Certified reference stock solutions (AccuStandard New Haven, CT, USA) were used to prepare sulfur calibration standards in 2% v/v HNO₃ (limit of detection (LOD) for sulfur was 0.03 mmol L⁻¹) and concentrations were determined by measuring intensity at the 181.972 nm sulfur emission line. Background/inter-element interferences were corrected for with the Varian Fast Automated Curve-fitting Technique (FACT).

Thiosulfate (S₂O32-) and sulfite (SO32-) analytes were stabilized upon sampling through derivatization with monobromobimane (Rethmeier et al., 1997). Derivatization agents (50 μL of acetonitrile, 50 μL of 50 mmol L⁻¹ HEPES (4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid, ≥99.5%, Sigma), 5 mmol L⁻¹ of EDTA buffer (ethylenediaminetetraacetic acid, 99.4–100.6%, Sigma Aldrich; pH = 8.0, adjusted with NaOH) and 10 μL of 48 mmol/L monobromobimane (>97%, Sigma Aldrich) in acetonitrile were added to 50 μL of water sample in 2 mL glass amber vials. The derivatization reaction was carried out in the dark for 30 min, 100 μL of methanesulfonic acid (~100 mmol) (≥99.5%, Sigma Aldrich) was then added to halt the reaction and samples were subsequently kept frozen until analysis. Thiosulphate and sulphite analyses were carried out on a Shimadzu LC-20AD prominence liquid chromatograpy (LC) system coupled to a fluorescence UV/VIS detector on an Alltima^TM HP C18 reversed phase column (150 mm × 4.6 mm × 5 μm, Grace^TM) held at 35°C using an isochratic mobile phase of 65% of 0.25% acetic acid v/v (pH 3.5 adjusted with NaOH) and 35% methanol at 0.5 mL min⁻¹ for a total run time of 20 min. The excitation wavelength was set to 380 nm and the emission wavelength 478 nm. The sulfite peak was found to elute at ~5 min and thiosulfate eluted at ~7 min. Calibration curves were prepared using sodium sulphite (Sigma Aldrich, ≥98% purity) and sodium thiosulphate (Sigma Aldrich, 99% purity).

Samples for SO42- and nitrogen species (NO2- and NO3-; see Table S1) analyses were collected with no headspace into 1 liter polypropylene sample bottles rinsed three times with site water and stored at 4°C prior to analysis. Dissolved SO42-, NO2-, and NO3- quantification was carried out by spectrophotometry using a HACH DR2800 (HACH Company, Loveland, CO, USA) as described previously in Risacher et al. (2018).

Water samples for enrichments were taken directly on site (Mine 3) or from shipped water (Mine 1, 2, and 4; Figure S1) in 90 mL sterile containers with no headspace and stored at 4°C until enrichment for SoxBac was initiated (within 2–5 days of collection). The indigenous microbial communities from the active mine wastewater systems were enriched for sulfur oxidizing bacteria (SoxBac) in a neutral pH medium (pH~7) that contained thiosulphate (S₂O32-) as sulfur substrate. Thiosulphate was chosen as a suitable sulfur substrate as it is known to be an important intermediate in microbial sulfur cycling and is an abundant and ubiquitous SOI generated during metal extraction from sulfide ores (Warren et al., 2008; Shi et al., 2011; Veith et al., 2012; Miranda-trevino et al., 2013; Liu et al., 2014). Thiosulphate is also monitored most frequently by industrial monitoring programs (Whaley-Martin et al. in preparation). Methodology for growth of sulfur oxidizing bacteria was based on those described elsewhere in detail (Bernier and Warren, 2005). Briefly, three cycles were employed to generate an active SoxBac enrichment from each of the nine parent water samples. Preliminary experiments comparing enrichments grown for 3, 4 and 5 cycles found that rates of acid production were relatively stable in the third cycle, indicating that SoxBac activity was sufficient.

In the first cycle, a 1:1 ratio of mining water to the neutral pH media (described in SI) was prepared under sterile conditions in 150 mL autoclaved flasks in a Biological Safety Cabinet (BSC). SoxBac enrichments samples were then incubated under dark conditions at room temperatures (~comparable to in situ MIW temperatures) until they reached a pH <5. Monitoring of pH changes (used as a proxy for microbial sulfur metabolic activity) was carried out with a sterile probe (immersed in 70% ethanol for a minimum of 2 min and rinsed with sterile (0.2 μm filtered) MilliQ water (Bernier and Warren, 2005). The second and third successive cycles of SoxBac enrichments were created by sterile pipetting 50 mL of cycle one or two SoxBac enrichment water to an autoclave flask with 100 mL of new media (1:2 ratio of parent water to media). The pH was adjusted to ~7 with dilute sterile (0.2 μm filtered) NaOH/HCL solution and enrichments were monitored under the same conditions as cycle 1 and then halted once the pH had decreased <5 (Cycle 3: 3 to 6 days, Table 2). For cycle 3, both initial and final enrichment waters were sampled for geochemistry (pH, S 2O32-, SO32-, SO42-, and TotS_{UF and 0.2μm} determination) and final waters for enrichment consortia characterization (16S rRNA) following the protocols outlined above. Control samples consisting of media and sterilized mine water for each parent sample were also monitored for pH and SO42-, S 2O32-, and SO32- concentrations to determine any abiotic SOI alteration that may have occurred. Control samples showed pH changes only fluctuated around circum neutral conditions (~7–8) and SOI species changes were < LOD.

Five of the six tailings reservoir MIW samples were circum-neutral (pH 6–7.3), while the Mine 3 September tailings reservoir 2 m depth (pH of 4.7), and input water sample (3.7) and Mine 4 receiving environment sample were acidic (4.3; Table 1). Concentrations of TotS_UF ranged from 2.3 (Mine 1) to 17.5 (Mine 4) mmol L⁻¹ with similar ranges seen in the 0.2 μm filtered and TotS fraction of waters (2.3–16.9 mmol L⁻¹) indicating minimal particulate S species within these samples (Table 1). SO42- concentrations ranged from 2.2 mmol L⁻¹ (Mine 1 tailings reservoir) to 11.8 mmol L⁻¹ (Mine 4 tailings reservoir) across all samples (Table 1). Sulfur mass balance identified that a pool of reactive sulfur (available for sulfur oxidation; S_react: [TotS]—[SO42-]) ranged from 1% (at Mine 3 Tailings Pond (0.5 m) in September) to 97% (10 m at Mine 3 in November; Table 1). Thiosulphate concentrations were less than the detection limit for all samples except for the Mine 4 tailings reservoir MIW sample, where a concentration of 0.31 mmol L⁻¹ was detected. Sulfite concentrations were low but detectable, varying between 6 and 18 μmol L⁻¹ across samples (Table 1). Thus, most of these waters contained 1–7 mmol L⁻¹ of intermediate oxidation state sulfur (in other SOI compounds than thiosulphate or sulphite) that can support microbial sulfur oxidative metabolism.

A SoxBac enrichment was successfully generated from the MIW collected at all nine sites consistent with the detectable presence of sulfur at mmol L⁻¹ concentrations within these waters, even at Mine 1, which had the lowest TotS of 2.3 mmol L⁻¹ and the lowest S_react of 100 μmol L⁻¹ (Table 1). Each SoxBac enrichment began with ~20 mmol L⁻¹ S-S₂O32- ending in depleted thiosulphate to concentrations approaching the analytical limits of detection (<0.001 mM), with the exception of those grown from Mine 1 tailings reservoir water (Table 2).

This result was accompanied by high and comparable acidity generation rates (consistent with sulfur oxidation pathways) where pH decreased from circum-neutral conditions (pH ~7) to ~3 in 5–7 days for Mine 2, 3, and 4 MIW enrichments. The SoxBac enrichment grown from September Mine 1 tailings reservoir was only able to decrease pH one unit (6.1) in 5 days but this was the only enrichment experiment where sulfite production was detected (Table 2). Interestingly, the net concentration of sulfate produced was similar between Mine 1 tailing reservoir Sept SoxBac enrichment and Mine 3 Nov 10 m tailing reservoir (3.2 and 3 mmol L⁻¹ respectively, Table 2). However, the Mine 3 enrichment pH dropped to 3.03 by the end of the enrichment cycle, generating 0.93 mmol L⁻¹ more of H⁺ than the Mine 1 SoxBac enrichment (Table 2). Further, this Mine 3 SoxBac enrichment converted 18.6 mmol L⁻¹ of S in S₂O32-, compared to only 2 mmol L⁻¹ S-S₂O32- converted in Mine 1 September tailings reservoir SoxBac enrichment. The ratio of net concentration change of sulfate to thiosulfate (t = initial vs. t = final) and the net production of [H⁺] from initial to final (Figure S2) suggest that Mine 1 SoxBac enrichments should be the most dissimilar to the other enrichment communities.

The biogeochemical research in mining environments to date has been heavily focused on AMD sites (e.g., waste rock containing sulfide minerals) while the organisms governing the geochemistry in circum-neutral tailing reservoirs remains poorly understood. In this study, there was a consistent absence of the extreme acidophilic iron and sulfur organisms typical of AMD environments (Edwards et al., 1999; Baker and Banfield, 2003; Schippers et al., 2010; Kimura et al., 2011; Kuang et al., 2012). These results are not surprising given the circum-neutral pH values of many of the tailings reservoir MIW samples investigated here. However, even in the three acidic MIW samples Acidithiobacillus or Leptospirillum spp. were not detected (Figure 1). Results further show that the parent mine tailing reservoir MIWs (Mine 1, 2, 3, and 4), input water (Mine 3) and receiving environment (Mine 4; Table 1) samples were comprised of diverse and distinct microbial communities (Figure 1), consistent with the wide range of geochemical conditions present across the parent waters (Table 1). Sphingomonadaceae were consistently present in the microbial communities of the four tailings reservoirs examined in this study, with relative abundances ranging from 4 to 72%. Sphingomonadaceae featured prominently in both Mine 3 (September) and Mine 4 (November) tailings reservoir communities, which were both circum-neutral and exhibited similar temperature (~10°C) and oxygen saturation values >50% (Table 1). Sphingmonadaceae have been identified as organic biodegraders (White et al., 1996) with a high metal tolerance (Liu et al., 2017). In the Mine 1 tailings reservoir, also circum-neutral, but exhibiting the lowest Total S concentration of 2.3 mmol L⁻¹ (Table 1) and the highest nitrate concentration (Table S1), the microbial communities were dominated by Rhodobacteriaceae (September) along with Burholderiaceae (October). Both of these families have been associated with sulfur, carbon and nitrogen biogeochemical cycling (i.e., Ziga et al., 2010; Pujalte et al., 2014).

Halothiobacillus spp. emerged as an important sulfur oxidizing bacterium present in three of the four mines, capable of driving circum-neutral pH to acidic conditions, whose presence and relative abundance was the most important factor associated with the observed clustering of parent MIW and enrichment communities (Figures 1–3). Halothiobacillus spp. is a motile chemolithoautotrophic sulfur oxidizing Gammaproteobacterium capable of oxidizing sulfide, elemental sulfur, thiosulfate and tetrathionate to sulfur, sulfite, polythionates, and sulfate (Kelly and Wood, 2000; Sievert et al., 2000; Shi et al., 2011). The acidophilic organisms that dominate AMD environments are likely end-members of a microbial succession where circum-neutral (or alkaline) conditions are biologically transformed through oxidative processes to acidic conditions. In this scenario, these organisms colonize these waters subsequent to mesophillic or moderate acidophilic microbial communities (Korehi et al., 2014). Liu et al. (2014) reported a relationship between microbial communities in mine tailing cores and overall pH, which in that study ranged from circum-neutral to pH 2–3. Specifically, the class Gammaproteobacteria was implicated in possessing a relationship to pH conditions within the mine tailings (Liu et al., 2014). This is consistent with the Gammaproteobacterium, Halothiobacillus dominating SoxBac enrichments from Mine 2–4 and two in situ MIWs where pH was measured to be ~4 observed in this study (Figure 2, Table 2). As mentioned, the genus Halothiobacillus was more prominent in two of the three acidic parent waters [Mine 3 tailings reservoir (September) and Mine 4 receiving environment (Table 1)], while the family Acetobacteraceae were dominant in the acidic input water to Mine 3. Unlike Halothiobacillus, Acetobacteraceae has not been largely associated with sulfur cycling and instead is associated with carbon cycling and propagating acidic environments (White et al., 1996). As the input water at Mine 3 contains a combination of tailings water and natural tributaries, there maybe a flux of organics from natural sources additionally into this sulfur containing water that supports both heterotrophic and acid producing roles of the community.

Although Halothiobacillus has been shown to be a rare group within AMD environments (Schippers et al., 2010; Yang et al., 2014), our results show that this group can account for more than 70% of the microbial community in SoxBac enrichments derived from tailings associated MIWs (Figure 5). The parent waters where Halothiobacillus were detected here (Mines 2–4), contained considerably higher total aqueous sulfur concentrations (7–17 mM) than Mine 1 (~2 mM) and concentrations of nitrate two magnitudes lower than Mine 1 (Table 1; Table S1) discussed further below. While optimal growth conditions for Halothiobacillus occur under circum-neutral conditions, supporting a classification of “moderate acidophile” (Kelly and Wood, 2000; Sievert et al., 2000; Schippers et al., 2010; Marescotti et al., 2012; Yang et al., 2014), Sievert et al. (2000) reported that laboratory Halothiobacillus incubations under neutral conditions were driven as low as pH~3 through sulfur oxidation. Moreover, Shi et al. (2011) isolated an unspecified strain of Halothiobacillus from a lead-contaminated soil and reported a high metal tolerance, consistent with the mining environments assessed here. Thus, Halothiobacillus appears to be highly adapted to metal mine environments due to its ability to tolerate elevated metal conditions (Table S3) and oxidize an array of reduced sulfur compounds and phylogenetic diversity (Figures 3 and 5).

Within these SoxBac enrichments, variation in the ΔH⁺: ΔSO42- ratios appeared to correspond with the presence and abundance of specific members of the Halothiobacillus clade (Figure 3). All enrichments showed a divergence from the expected 1:1 ratio in ΔH⁺:ΔSO42- associated with complete oxidation of thiosulphate to sulfate. However, enrichments that contained higher proportions of Halothiobacillus sequences 10 and 14, exhibited the lowest ratios of net H⁺: SO42- suggesting that relatively more sulfur disproportionation occurred in these enrichments (Warren et al., 2008). Nevertheless, the results of the enrichment experiments found circum-neutral conditions were lowered to acidic pH levels in < ~6 days in enrichments dominated by Halothiobacillus sp. (Table 2) which would be considered particularly fast if these rates were to ensue on an active mine site.

The identification of 16 Halothiobacillus genera (Figure 3) suggests this family is ecologically well adapted to the variable conditions that can occur within tailings reservoirs. The Halothiobacillus genera identified in this study have the capability to drive acidic conditions through thiosulphate metabolism. However, the differential net acid to sulfate generation associated with different compositional make up of Halothiobacillus genera within these SoxBac enrichments suggest they are doing so via different sulfur reactions. This indicates the need to more fully constrain phylogenetic diversity of this clade with respect to sulfur cycling. Further, while all Halothiobacillus genera identified in this study have the capability to drive acidic conditions through thiosulphate metabolism, the differential net acid to sulfate generation associated with different compositional make up of Halothiobacillus genera within these SoxBac enrichments suggest they are doing so via different sulfur reactions. This indicated the need to more fully constrain phylogenetic diversity of this clade with respect to sulfur cycling.

While phylogenetic resolutions were not resolvable to the species level in this study, there was a high sequence similarity of our results to two species within the same genus: Halothiobacillus kellyi and Halothiobacillus neapolitanus (Figure 4). Both H. kellyi and H. neapolitanus have been identified as mesophillic sulfur oxidizers (Kelly and Wood, 2000; Veith et al., 2012; Vikromvarasiri et al., 2015). H. neapolitanus has been specifically proposed as a useful sulfur oxidizer to remove hydrogen sulfide from biogas (Vikromvarasiri and Pisutpaisal, 2014) and to be capable of tolerating high NaCl concentrations (>860 mmol/L; Sievert et al., 2000) which would explain their occurrence in the high salt conditions often associated within tailing reservoirs.

Similar to Halothiobacillus, the closely related genus Thiovirga (present in Mine 1 tailing reservoir SoxBac enrichments) has also been described as a mesophilic, chemolithoautotrophic, sulfur-oxidizing bacterium (Ito et al., 2005; Yang et al., 2011). However, Halothiobacillus or Thiovirga from Mine 1 tailing reservoir was not successfully enriched for despite both being present in the parent waters (n.b. Halothiobacillus was present but at very low concentrations). The geochemical conditions of the Mine 1 tailings reservoir exhibited the lowest total sulfur concentrations (2.3 mM; Table 1), proportionally higher SO42- concentrations (2.2 mM; Table 1), and thus the lowest S_react concentration (100 μmol L⁻¹), i.e., available S substrate for sulfur oxidizers, and two orders of magnitude higher NO32- (Table S1). In these waters it was the Thiovirga genus that was dominant from the Halothiobacillaceae family (Figure 3, Table S4). This result suggests that the pre-existing geochemical conditions of the tailing reservoirs (either those measured in this study or additional unknown parameters), are governing which sulfur oxidizing microbes are dominant and that which genera are present will correspondingly influence the sulfur cycling that occurs and associated acid and sulfate generation.

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.