Next-Generation Pyrosequencing Analysis of Microbial Biofilm Communities on Granular Activated Carbon in Treatment of Oil Sands Process-Affected Water

Abstract

The development of biodegradation treatment processes for oil sands process-affected water (OSPW) has been progressing in recent years with the promising potential of biofilm reactors. Previously, the granular activated carbon (GAC) biofilm process was successfully employed for treatment of a large variety of recalcitrant organic compounds in domestic and industrial wastewaters. In this study, GAC biofilm microbial development and degradation efficiency were investigated for OSPW treatment by monitoring the biofilm growth on the GAC surface in raw and ozonated OSPW in batch bioreactors. The GAC biofilm community was characterized using a next-generation 16S rRNA gene pyrosequencing technique that revealed that the phylum Proteobacteria was dominant in both OSPW and biofilms, with further in-depth analysis showing higher abundances of Alpha- and Gammaproteobacteria sequences. Interestingly, many known polyaromatic hydrocarbon degraders, namely, Burkholderiales, Pseudomonadales, Bdellovibrionales, and Sphingomonadales, were observed in the GAC biofilm. Ozonation decreased the microbial diversity in planktonic OSPW but increased the microbial diversity in the GAC biofilms. Quantitative real-time PCR revealed similar bacterial gene copy numbers (>10⁹ gene copies/g of GAC) for both raw and ozonated OSPW GAC biofilms. The observed rates of removal of naphthenic acids (NAs) over the 2-day experiments for the GAC biofilm treatments of raw and ozonated OSPW were 31% and 66%, respectively. Overall, a relatively low ozone dose (30 mg of O₃/liter utilized) combined with GAC biofilm treatment significantly increased NA removal rates. The treatment of OSPW in bioreactors using GAC biofilms is a promising technology for the reduction of recalcitrant OSPW organic compounds.

INTRODUCTION

Open-pit mining of Canada's Athabasca oil sands produces large quantities of bitumen coated sands. The bitumen is extracted from the sands using the Clark hot-water process, in which 1 m³ of bitumen extracted from oil sands generates 4 m³ of oil sands process-affected water (OSPW) (1–3). Over a billion cubic meters of this generated OSPW is currently stored in tailing ponds given the no-release policy for OSPW followed by the oil sands companies (4, 5). This volume of OSPW will continually increase as the oil sands bitumen extraction continues in the Athabasca region until suitable treatment technologies become available to allow for OSPW release (2). This OSPW is contaminated with a large number of inorganic and organic compounds during the extraction process that are known to be toxic to aquatic and terrestrial life (6). Thus, the production of OSPW associated with the extraction of oil sands bitumen is a great environmental issue that needs to be addressed in the near future for the remediation of the oil sands region.

Fresh OSPW has acute, subchronic, and chronic toxicities to aquatic organisms, and the majority of this toxicity has been previously attributed to naphthenic acids (NAs) (2, 7). NAs are a mixture of organic surfactants containing carboxyl groups with a general chemical formula for classical NAs (containing two oxygen molecules) of C_nH_2n+ZO₂, where n is the number of carbon atoms and Z is either zero or a negative even integer representing the number of hydrogen atoms lost due to ring formation (8, 9). Currently, the wetlands used to treat OSPW by the oil sands industry are not effective in eliminating toxicity because many kinds of NAs are recalcitrant to natural biodegradation. Therefore, there is an urgent need for the establishment of adequate OSPW treatment technologies to reduce the continual accumulation and current storage of OSPW in tailing ponds. In addition, extending the recycling capacity of the high-efficiency-treated OSPW may lead to the reduction of freshwater withdrawal from the Athabasca River.

The granular activated carbon (GAC) biofilm technology is very promising for removal of recalcitrant and toxic organic compounds, such as NAs, due to its high adsorptive capacity for organics and high biomass concentration in developed biofilms, which degrades organics in a biofilter configuration (10, 11). It has been reported previously that ozonation can increase the biofilter performance and reduce the operation time by increasing the biological activity and decreasing the organic loading of recalcitrant organics to the biofilter (12). However, the typical operational costs for the production of 1 kg of ozone are in the range of 1.5 to 2.0 U.S. dollars (13, 14); considering these high operational costs, a partial degradation of target compounds in wastewater using lower ozone doses would help to limit costs while providing degraded organics that are more easily degraded in downstream biological treatment. Previously, our research group reported the use of GAC fluidized bed biofilm reactors for the treatment of raw and ozonated OSPW and found that more than 86% and 99.5% NAs were removed from raw and ozonated OSPW, respectively, after the GAC treatment processes (15, 16). Given these positive results, further investigation of the biofilm morphology and microbial community characterization would be beneficial for the improvement of the design and understanding of the operation of biofilm reactors.

Conventional microbial community characterization methods include denatured gradient gel electrophoresis (DGGE), clone library, quantitative PCR (qPCR), terminal restriction fragment length polymorphism (T-RFLP), and fluorescence in situ hybridization (FISH), among others (17, 18). Previously, it has been reported that the conventional molecular biological methods underestimate the overall diversity of the microbial community and are unable to detect rare species in a complicated environmental sample because of a lack of sufficient sequences to capture comprehensive and systematic information on various microbial communities (19). For example, a very limited number of sequences can be generated by the DGGE and clone library methods, and the processes are time-consuming (20, 21). Preferential amplification of rRNA genes with the PCR-based methods may lead to the omission of some microbial species information (21, 22). T-RFLP analysis is PCR based and suffers from the same drawbacks as this technique (23). The FISH technique is fluorescence based, which requires optimization of probe design and hybridization conditions (23). More sensitive technologies are needed to achieve a more precise and complete characterization of microbial communities. Toward this goal, new high-throughput next-generation techniques have been used for environmental matrices, including the characterization of biofilms developed on Athabasca River sediments and soils using ion torrent pyrosequencing (24, 25), and wastewater treatment (18) and raw water distribution (26) using 454 pyrosequencing. For example, Yergeau et al. (24) collected sediments from different locations of the Athabasca River and biofilm samples from rotating annular reactors to perform ion torrent pyrosequencing of biofilm microbial communities. However, few studies have addressed biofilm community analysis for bioreactors aimed at treating OSPW. Among these studies, the DGGE technique has been utilized for the analysis of OSPW biofilm microbial communities on various surfaces, such as polyethylene (PE) (22, 27, 28), polyvinyl chloride (PVC) (22, 28), and GAC (15, 16). However, to our knowledge, no studies have investigated OSPW biofilm formation on GAC using high-throughput pyrosequencing techniques.

Thus, a study on biofilm development on GAC was performed using a batch study with continuous replacement of OSPW into reactors containing GAC. The main objective of the study was to evaluate biofilm growth on the GAC surface and microbial community composition during raw OSPW treatment using next-generation high-throughput 454 pyrosequencing. In addition, it is well known that ozone can be used to degrade the target classical NAs with high cyclization and long chains into lower-molecular-weight oxidized NA compounds, thus increasing their biodegradability (16, 29). Moreover, ozonation can impact the microbial community structure in biofilm reactors and has the potential to increase bioreactor performance (22). Thus, the impact of ozonated OSPW on biofilm growth, biofilm community structure, and overall NA removal was also evaluated to assess the synergistic removal by these individual OSPW treatment processes. The impact of the biofilm bacterial community on OSPW quality was investigated via removal of NAs, chemical oxygen demand (COD), and acid-extractable fractions (AEF).

MATERIALS AND METHODS

Sources of OSPW and GAC.

Process water from an oil sands tailing pond in Fort McMurray, AB, Canada, was sampled and shipped in October 2012. Raw OSPW was received at the University of Alberta in 200-liter barrels and preserved in the dark at 4°C in a temperature-controlled room prior to use in experiments. Selected-grade bituminous-coal-based GAC (SGL 8 × 30), produced by high-temperature steam activation, was purchased from Calgon Carbon Corporation (Pittsburgh, PA). GAC was sterilized at 121°C for 30 min (model 733LS vacuum/gravity system sterilizer; Getinge USA, NY), dried at 104°C for ∼72 h, and allowed to cool in a desiccator before being used in GAC experiments.

Ozonation of OSPW.

Detailed descriptions of the ozonation procedure have been reported previously by Wang et al. (29). Briefly, ozonation of raw OSPW was performed by generating ozone from extradry and high-purity oxygen using an ozone generator (model GSO-40; PCI-WEDECO, Herford, Germany). The gas mixture (ozone and oxygen) containing an ozone concentration of ∼150 mg/liter was sent through a ceramic fine bubble gas diffuser located at the bottom of a 200-liter plastic barrel containing raw OSPW. The ozone concentrations in the feed and off-gas lines were continuously monitored by two identical ozone monitors (model HC-500; PCI-WEDECO) during OSPW ozonation. The residual ozone (unreacted, physically absorbed in the OSPW) was purged using pure nitrogen and the off-gas line ozone concentration was recorded. The Indigo method was used to estimate the residual ozone as per standard methods (30). A utilized ozone dose of 30 mg of O₃/liter was used for ozonation of OSPW to reduce the NA concentrations by 40 to 50% (determined as described in “Water quality parameters” below), thus allowing for residual NAs to be available for biodegradation.

Biofilm growth experimental methods.

The culture of a biofilm on the GAC surface was carried out at room temperature (21 ± 1°C) and 150 rpm on a horizontal shaker with treatments enclosed in 1-liter amber bottles containing 2 g of GAC and 500 ml of raw or ozonated OSPW (Innova 2100 platform shaker; New Brunswick Scientific, USA). For biofilm growth start-up, the reactors treating raw OSPW (n = 4) were used without addition of any bacterial inoculants, whereas the reactors treating ozonated OSPW (n = 4) were inoculated with endogenous raw OSPW bacterial culture (5 ml) prior to starting experiments to promote biofilm growth in the absence of endogenous bacteria, which might have been reduced or eliminated during the ozonation process. Initially, all reactors were run for 4 days in batch mode to allow for the attachment of bacteria on the GAC surface. Subsequently, the reactors were operated in a batch mode with continuous interchange of raw and ozonated OSPW every second day until biofilm growth reached steady state (day 48; see Fig. S1 in the supplemental material). After this steady state of biofilm growth, experiments were carried out with fresh raw and ozonated OSPW for 2 days to observe the removal of organics from the combined adsorption and biodegradation mechanisms (n = 2 for each treatment). To observe only the adsorption mechanism, the biofilm biodegradation on the GAC surface was inhibited using 0.1% (wt/vol) sodium azide, and experiments were carried out with raw (n = 2) and ozonated (n = 2) OSPW. Using these two methods, the impact of the biodegradation mechanism only can be calculated by the difference between measurements in combined and adsorption-only experiments.

Liquid samples from each reactor were collected in triplicate with sterilized pipettes and analyzed for COD, AEF, and NAs. To collect the GAC biofilm samples from the reactors, the GAC was isolated from the treated OSPW by pouring off the OSPW gently, and the GAC was taken from the reactors using a presterilized scoop. The GAC biofilms were then analyzed by heterotrophic plate counting (HPC) (n = 3), confocal laser scanning microscopy (CLSM) (n = 4), and DNA extraction (n = 2).

Bacterial enumeration using HPC.

Bacterial enumeration was performed by HPC using the drop plate method (31) for raw and ozonated OSPW, and their respective biofilms were developed on the GAC surfaces. To extract bacteria from the biofilms, a mass (0.5 to 1.0 g) of GAC biofilm was placed into 2 ml of a sterile phosphate-buffered saline (PBS) solution in a 15-ml sterile tube and mixed using a vortex for 1 min. For OSPW and extracted biofilm samples, a series (7 serial) of 10-fold dilutions was performed and 10 μl of each dilution was plated in triplicate on R2A (Difco) agar culture plates. Plates were incubated at 37°C, with counting of bacterial CFU performed at 24, 48, and 72 h. For biofilm samples, the counts were converted into bacterial CFU per gram of GAC of biofilm. Statistical analysis of HPC was performed using analysis of variance (ANOVA) at a significance level (α) of 0.05.

CLSM imaging.

GAC samples were taken from raw and ozonated OSPW treatments every 6 days for GAC biofilm analysis. The GAC biofilms were stained with SYTO 9 (BacLight Live/Dead bacterial viability kit; Molecular Probes, USA) and concanavalin A (ConA; Molecular Probes, Eugene, OR) lectin conjugated with Texas red for the probing of live cells and extracellular polymeric substances (EPS), respectively (32). Biofilm image observation, acquisition, and biofilm thickness measurements were performed with a CLSM (Zeiss LSM 710; Carl Zeiss Micro Imaging GmbH, Germany). The images were observed and scanned randomly at 4 or 5 positions with a Zeiss Plan-Apochromat lens (×20 magnification, 0.8 numerical aperture). The detailed procedure has been described previously by Hwang et al. (22).

DNA extraction, qPCR, and 454 pyrosequencing.

A PowerSoil DNA isolation kit (MOBIO Laboratories Inc., Carlsbad, CA) was used for isolating DNA from bacterial cells in duplicate. Planktonic bacteria were isolated as a pellet by centrifuging raw and ozonated OSPW at 10,000 rpm for 10 min (Multifuge3S/3S-R; Heraeus, Thermo Scientific, USA). The pellet was collected and added to power soil bead tubes to isolate total genomic DNA. For isolating DNA from the GAC biofilm, a mass of GAC (0.5 to 1.0 g) was added directly to the power soil bead tubes. The manufacturer's protocol was followed for isolating the total genomic DNA. Quantitative real-time PCR (qPCR) was carried out in triplicate using a CFX 96 Touch real-time PCR system (Bio-Rad Laboratories Inc., USA) containing 1× SsoFast EvaGreen supermix, a 0.5 μM concentration of each primer (Integrated DNA Technologies, Coralville, IA), and 5 μl of diluted DNA within a 25-μl total reaction volume. The primer set used for amplification of extracted DNA during qPCR consisted of 907r (5′-CCG TCA ATT CMT TTG AGT TT-3′) and 341f (5′-CCT ACG GGA GGC AGC AG-3′). The qPCR data were analyzed statistically using ANOVA at a significance level (α) of 0.05.

After DNA extraction, the V1 to V3 regions of the 16S rRNA genes were amplified by PCR using 28F and 519R primers (GAGTTTGATCNTGGCTCAG and GTNTTACNGCGGCKGCTG, respectively) and a Qiagen hot-start master mix. The DNA was denatured at 95°C for 5 min, followed by 35 cycles at 94°C for 30 s, 54°C for 45 s, and 72°C for 60 s. Finally, an extension reaction was performed at 72°C for 10 min. The amplified DNA was sequenced and analyzed using a 454/Roche GS-FLX instrument by the Research and Testing Laboratory in Lubbock, TX, as previously described (33). An automated pipeline was used to process raw 454 sequence data. After denoising (USEARCH application) and chimera removal (UCHIIME in de novo mode), the sequences ware clustered into operational taxonomic unit (OTU) clusters with 100% identity (0% divergence) using USEARCH (34). For taxonomic identification, the seed sequences were derived from the NCBI using a distributed .NET algorithm. Based upon the above BLASTn+-derived sequence identity percentage, the sequences were classified at the appropriate taxonomic levels based upon criteria as detailed in the supplemental material.

Trimmed sequences were also processed through the Ribosomal Database Project (RDP) pyrosequencing pipeline (35). The sequences of the samples were further aligned using the RDP aligner tool before the RDP Clustering function was applied. The resulting clusters were submitted to calculate the Chao1 estimator, Shannon-Weaver (H′) index evenness, and the rarefaction curves at the level of 3% dissimilarity, which was considered to be approximately related to species level. The RDP abundance statistics tool was also used to calculate the differences between samples based on the Jaccard method (36) and construct a distance matrix at 3% dissimilarity by using the unweighted pair group method with arithmetic mean (UPGMA).

Water chemistry analysis.

COD in raw and ozonated OSPW treatments was measured according to standard methods (30) after filtering through a 0.45-μm-pore-size nylon filter. AEF were analyzed using the protocol developed and used by the oil sands industry and described by Gamal El-Din et al. (37). Briefly, treatments were filtered through a 0.45-μm-pore-size nylon filter and the pH was adjusted to 2.4 to 2.5 using sulfuric acid. A 50-ml aliquot of this acidified OSPW was extracted with high-pressure liquid chromatography (HPLC)-grade dichloromethane (DCM), dried, and reconstituted in Optima-grade DCM prior to analysis. Fourier transform infrared (FTIR) spectroscopy (PerkinElmer, ON, Canada) was used to measure the AEF for carbonyl stretch equivalents in OSPW using a KBr cell. It should be noted that the AEF determination measures all compounds with functional groups containing carboxylic acids, ketones, and aldehydes; it cannot be used as a direct measure of NA concentrations. However, AEF values are commonly used by the oil sands industries as surrogate measure for NAs in OSPW. One-way ANOVA was applied to determine reliability of triplicate COD and AEF measurements at a significance level (α) of 0.05.

Classical NA concentrations were determined using ultrahigh-pressure liquid chromatography (UHPLC)–high-resolution mass spectrometry (HRMS). The methodology has been detailed previously (15, 29) Briefly, OSPW samples were centrifuged at 11,984 × g for 10 min (5810 R centrifuge; Eppendorf AG, Hamburg, Germany), and then the samples were passed through a Waters Acquity UPLC system (Milford, MA) for the separation of NAs using [¹³C]myristic acid as an internal standard. A high-resolution (full width at half-maximum, 40,000) Synapt G2 HDMS mass spectrometer (m/z from 0 to 600) equipped with an electrospray ionization source operating in negative-ion mode and quadruple time of flight (QTOF) was used for detection, with TargetLynx ver. 4.1 software used to analyze the data for target compounds. The NA concentrations in raw and ozonated OSPW were subcategorized based on carbon and Z numbers to identify the impact of treatment on each class of NAs. The NA concentration based on Z numbers was determined from the summation of NA species for all carbon numbers (Σ[NAs]_n) for each Z number, whereas the NA concentration based on carbon number was determined from the summation of NA species for all Z numbers (Σ[NAs]_Z) for each carbon number.

RESULTS AND DISCUSSION

Characterization of biofilm. (i) Confocal imaging.

Figure S1 in the supplemental material shows confocal images of biofilms developed on the GAC surfaces in raw and ozonated OSPW during the 48-day experimental duration. Bacterial attachment was observed on the GAC surface on day 6 (first sample day) for both treatments, as indicated by the green and red stained areas of the images. The biofilm growth reached steady state after 30 days, and the final biofilm thicknesses on day 48 were around 50 μm for both treatments (see Fig. S1). The biofilm images indicate that the bacteria did not fully cover (i.e., completely green and red images) the GAC surface even after 48 days. The biofilm thicknesses are similar to those found by Hwang et al. (22) for OSPW biofilms produced on polyethylene (PE) and polyvinylchloride (PVC) coupons in continuous bioreactors, where a continuous (full covered) biofilm on the surface was observed on these relatively smooth surfaces. This is expected, as the bacteria form a complex structure of aerobic biofilms formed by discrete aggregates of densely packed cells and interstitial voids on the GAC surfaces as reported previously (38, 39). The ozonated OSPW treatment showed faster initial attachment of bacteria (by day 18), despite having a biofilm thickness similar to that of raw OSPW after 48 days. Ozonation can increase the biodegradability and reduce toxicity of OSPW compounds (40), which might be a plausible reason for faster attachment of bacteria on the GAC surface in ozonated OSPW. The GAC biofilm thicknesses were greater than the previously observed biofilm thickness (34 ± 5 μm) for the treatment of OSPW using fluidized bed biofilm reactors (15, 16). The greater biofilm thickness on the GAC from the current study might be attributed to a reduction in biofilm detachment from the lower shear stress in the batch reactors compared with that in the fluidized bed biofilm reactors (39, 41).

(ii) Quantification of microbial community in the GAC biofilm.

Figure 1 shows the results of bacterial growth from qPCR (Fig. 1A) and HPC (Fig. 1B) analyses of raw and ozonated OSPW over the 48-day experimental duration. The bacterial attachment on the GAC surface was observed on day 6 for both treatments. For both analyses, the bacterial growth increased in the GAC biofilm until reaching a plateau by day 30 for qPCR and day 48 for HPC analyses. The final bacterial concentration calculated using qPCR was ∼10⁹ copies/g of GAC for raw and ozonated OSPW (Fig. 1A). The growth pattern trends were similar over time for both treatments, and abundances were not significantly different from each other (P = 0.008). A previous study by Reaume et al. (42) showed a higher density of cells/gram of GAC (4 × 10¹⁰) in GAC biofilms in a biofilter used for the continuous treatment of municipal wastewater using ATP analysis. However, the use of different methods and wastewaters might be attributed to the observed 10-times-lower bacterial copy number on the GAC surface for the current experiments. The HPC bacterial colony growth increased over time and reached a plateau on day 30 for both treatments (Fig. 1B). The initial numbers (day 0) of cultivable bacteria were 2 × 10³ CFU/ml and 3 × 10³ CFU/ml for raw and ozonated OSPW, respectively. At day 30 and day 48 there were ∼10⁷ CFU/g of GAC in both treatments (Fig. 1B). Although the final bacterial growth metrics cannot be directly compared between qPCR and HPC methods, the ability of each method to detect bacterial growth reliably is important for consistent analysis. The conventional cultivation-dependent microbiological methods typically represent approximately 1 to 5% of bacterial colonies compared to those detected by qPCR. Similar percentages of HPC bacterial growth (1 to 10%) have been reported previously by Naz et al. (43).

FIG 1.

Bacterial growth on the GAC surface over 48 days for raw and ozonated OSPW GAC treatments. (A) qPCR method (bacterial copy number); (B) heterotrophic plate counting method (bacterial CFU). Values are averages (n = 3), and error bars represent standard deviations.

The bacterial growth on the GAC surfaces for raw and ozonated OSPW treatments was accomplished without the addition of an external organic food source to the reactors. Therefore, the microbial community metabolic needs were being fully supplied by the organics from OSPW provided every second day. The ability of the GAC surface to achieve a biofilm containing a high bacterial concentration might be attributed to (i) surface morphology that is irregular, rough, and highly porous, which may enhance bacterial colonization by providing protection from high fluid shear forces (44), and (ii) the high adsorption capacity of GAC (high porosity), which increases the availability of substrates, oxygen, and nutrients, thus attracting bacteria to the GAC surface (45).

Microbial community structure characterization.

The relative abundances of phyla identified by pyrosequencing of planktonic OSPW and GAC biofilm samples are shown in Fig. 2. The sequenced bacterial phyla consisted of Proteobacteria, Nitrospirae, Acidobacteria, Verrucomicrobia, Bacteroidetes, Chloroflexi, and others (Fig. 2). Proteobacteria were most abundant in all samples, with approximately 40, 60, 70, and 90% of overall bacterial abundances for ozonated OSPW, raw OSPW, ozonated GAC biofilm, and raw GAC biofilm (Fig. 2A to D), respectively. The Proteobacteria were also abundant for OSPW biofilms on PE and PVC reported by Hwang et al. (22) and on the Calgary biofilm device reported by Golby et al. (27). Interestingly, the relative Proteobacteria sequence abundances decreased by 10 to 20% in the ozonated treatments versus the analogous raw treatments. In contrast, Acidobacteria and Bacteroidetes sequences were more abundant, at 18% and 13% in the ozonated OSPW samples, compared to 8% and 5% in raw OSPW samples, respectively. The increase in the Acidobacteria for ozonated OSPW samples can be attributed to the formation of biodegradable acidic components from ozonation of raw OSPW leading to preferential selection of this phylum (22). The abundances of Proteobacteria were higher in raw (∼90%) and ozonated (∼70%) OSPW biofilms than in planktonic raw (∼60%) and ozonated (∼40%) OSPW. The increase of Proteobacteria abundances in GAC biofilm samples may have resulted from resistance of this phylum to the toxicity of NAs abundant on the GAC surfaces of this treatment and the broad degradation ability of this phylum. Previously, Yergeau et al. (46) found Proteobacteria abundances to be positively correlated with the concentration of NAs in aerobic biofilms. Thus, as ozonation decreased the OSPW toxicity, the adsorption of organics on the GAC surface in ozonated treatments may have decreased the GAC surface toxicity, resulting in increased abundances of other phyla. Moreover, differences in microbial communities between two biofilms could be due to changes in organic substrate composition, planktonic microbial community diversity, and bacterial surface polymer (e.g., lipopolysaccharides) physicochemical properties (47). The phylum Proteobacteria accounts for more than 40% of the prokaryotic genera which are known to have extreme metabolic diversity (48). Many bacterium of this phylum are ecologically important because they play key roles in the carbon, sulfur, and nitrogen cycles (48). Moreover, the dominant microorganisms of the Athabasca watershed and sediments belong to the Proteobacteria, which are known to degrade recalcitrant bituminous compounds (46). Thus, given this ecological importance and their overall abundances in the current samples, further analysis of the Proteobacteria classes was considered (Fig. 3). The Proteobacteria abundances were mainly composed of the orders Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, and Deltaproteobacteria (Fig. 3A). The ozonated GAC biofilm and raw and ozonated OSPW samples exhibited similar abundances of these groups, other than the Deltaproteobacteria abundances being marginal in the ozonated GAC biofilm sample. The dominant class in raw OSPW GAC biofilm was Gammaproteobacteria, with almost 60% of the total abundance. Our results differed from previously reported Gammaproteobacteria compositions (∼20%) in the aerobically grown biofilm determined using a rotating annular biofilm reactor with Athabasca River sediments as inoculants (24). Ozonation of OSPW led to an increase in the relative abundance of Alphaproteobacteria, whereas the relative abundance of Gamma- and Deltaproteobacteria decreased in the ozonated OSPW samples. It is clear that the Deltaproteobacteria were not able to grow on the GAC biofilm of both raw and ozonated OSPW (Fig. 3A). This inability to grow may be due to increased sensitivity of this group to the higher NA concentrations found on the GAC surface, as noted previously.

FIG 2.

Relative abundances of bacterial community members grouped by phyla. (A) Raw OSPW; (B) ozonated OSPW; (C) raw OSPW biofilm; (D) ozonated OSPW biofilm.

FIG 3.

Relative abundances of different Proteobacteria classes (A) and orders of Proteobacteria (B to D) in raw and ozonated planktonic OSPW and GAC biofilms. (B) Alphaproteobacteria orders; (C) Betaproteobacteria orders; (D) Gammaproteobacteria orders; (E) Deltaproteobacteria orders.

Among the other phyla, the members of Acidobacteria play an important role in the degradation of carbohydrates in boreal forest peat land, as these microbial groups are abundant in RNA-derived sequence libraries and culture studies indicate that they are well adapted to acidic, nutrient-poor conditions (49). Nitrospirae are involved in denitrification, sulfur oxidation, and sulfate reduction, showing their considerable capacity for adaptation to variable geochemical conditions and roles in local biogeochemical cycles (50). It has been previously reported that OSPW contains heteroatomic compounds containing nitrogen and sulfur, which would support the presence of Nitrospirae (51). Chloroflexi play an ecological role in the degradation of carbohydrates and soluble microbial products (52) and thus fulfill an important role in the GAC biofilm treatment process to bioregenerate the GAC from the adsorbed extracellular polymeric substance (EPS) on the GAC surface. Members of the phylum Bacteroidetes are abundant in many marine ecosystems and are known to have a pivotal role in the mineralization of complex organic substrates, such as polysaccharides and proteins, in the marine realm (53). The free-living, epibiotic, and symbiotic lifestyles of Verrucomicrobia are predominantly heterotrophic and can lead to degradation of higher-molecular-weight carbohydrates (polysaccharides), and their wide-ranging occurrence in different freshwater and marine habitats suggests that these bacteria play significant ecophysiological and biogeochemical roles in water columns. Previously, it has been reported that the adsorption capacity of GAC may decrease with bioreactor operating time due to the accumulation of metabolic byproducts and extracellular polymeric substances on GAC surfaces (54). Thus, a combination of these bacteria can degrade the adsorbed EPS from the GAC surface and will help the regeneration of GAC to increase the biofilter lifetime.

Table S1 in the supplemental material shows the overall bacterial diversity statistics of the various OSPW and biofilm samples. The Chao1 (community richness) and Shannon (community diversity) indices confirmed that the richness and diversity of GAC biofilm samples were both less than those of the OSPW samples (see Table S1). This decreased diversity may be attributed to the toxicity of NAs accumulated on the GAC surface, which will allow survival only of bacteria with high tolerances to the high concentrations of adsorbed NAs. Moreover, the higher diversity observed for the ozonated versus raw OSPW GAC biofilm samples (see Table S1) helps to confirm the impact of accumulated NAs on the GAC surface, as the ozonated OSPW has much lower NA concentrations than does the raw OSPW (see “Water quality parameters” below). The bacterial communities of the raw and ozonated OSPW were clustered together, while the two GAC biofilm communities were also grouped. GAC biofilm samples in ozonated OSPW have the largest genetic differences (distances) compared to the other samples, indicating that GAC adsorption significantly impacted the structure of the microbial community.

The phylum Proteobacteria was further identified at the order level of classification for all four samples (Fig. 3A). The Alphaproteobacteria thrive in widely divergent habitats and play a significant role in environmental processes. The most dominant order of Alphaproteobacteria was Rhizobiales in all four samples, ranging from 45 to 75% of total abundances (Fig. 3B). Ozonation decreased the Rhizobiales composition in the planktonic OSPW, while they were still dominant in the biofilm samples. The members of Rhizobiales have diverse functions, including atmospheric nitrogen fixation by living symbiotically in the roots of leguminous plants (55, 56). The other dominant orders of Alphaproteobacteria were Rhodospirillales and Rhodobacterales (Fig. 3B). The order Rhodobacterales includes species of photosynthetic bacteria, and some members of this order are chemoorganotrophs which can metabolize sulfur-containing organic compounds (56). The members of Rhodospirillales are comprised of photosynthetic, flower- and fruit-forming species and chemoorganotrophs that are involved in the partial oxidation of carbohydrates and alcohols (48, 56). The OSPW contains sulfur-nitrogen heteroatomic compounds (51), and the presence of Rhodobacterales and Rhodospirillales indicates the metabolism of sulfur- and nitrogen-containing compounds. Among the other orders, members of Caulobacterales are chemoorganotrophs, whereas Sphingomonadales include phototrophic and chemoorganotrophic species (48, 56). Rickettsiales are pathogenic for humans and animals (57).

The Betaproteobacteria are very heterogeneous with regard to metabolism, morphology, and ecology (48) and have previously been shown to be bitumen degraders found in the sediments of the Athabasca River, its tributaries, and oil sands tailing ponds (46). The major orders observed currently in this class of bacteria include Rhodocyclales, Burkholderiales, and Nitrosomadales (Fig. 3C). The dominant order in raw OSPW (∼40%) was Rhodocyclales, and the composition of this order was decreased from ozonation and in GAC biofilm samples (raw and ozonated OSPW). Burkholderiales was the dominant order (40 to 70%) in ozonated OSPW and biofilm samples. The members of Rhodocyclales are widespread and abundant in wastewater treatment systems, where they can degrade diverse groups of carbons, including aliphatic and aromatic compounds as well as nitrogen and phosphorus (58). Burkholderiales have been reported as the main contributor in the microbial ecology of bioremediation treatments for aromatic decontamination (59). Bacteria in this order can degrade a diverse group of aromatic compounds, including polychloro by-phenyls, naphthalene, and phenanthrene (59). Thus, the increase of Burkholderiales composition in ozonated OSPW and in biofilm samples indicated the higher microbial activity in the biofilm. The bacteria of the order Nitrosomonadales are chemolithotrophs which oxidize ammonia and fix carbon autotrophically from carbon dioxide (60). These bacteria can adapt to low ammonium concentrations and can use carbon dioxide produced from bacterial metabolism (48).

The orders from Gammaproteobacteria were sensitive to ozonation, and a drastic change in compositions was observed among treatments, as shown in Fig. 3D. The order Pseudomonadales (∼50%) was the dominant composition in raw OSPW; its abundance decreased to ∼25% in ozonated OSPW. Alteromonadales was the most abundant order in ozonated OSPW; in raw and ozonated OSPW GAC biofilms, the dominant order was Chromatiales (40 to 48%). The other dominant orders were Xanthomonadales, Oceanospirillales, Legionellales, and Methylococcales (Fig. 3D). The compositions were widely varied in all four samples because of the impacts of ozonation and biofilm formation. Among the orders, Pseudomonadales, Xanthomonadales, Alteromonadales, and Oceanospirillales are polyaromatic hydrocarbon (PAH) degraders, whereas Methylococcales are methane degraders (48, 61). Members of the order Chromatiales are known as phototrophic purple sulfur bacteria, able to perform photosynthesis and store elemental sulfur in cells (62). Legionellales are Gram-negative chemoorganotrophic pathogenic bacteria which survive in protozoan hosts (use amino acids as carbon and energy sources) in the natural environment and are responsible for pneumonia and influenza (48).

Figure 3E shows the Deltaproteobacteria compositions, and the major orders were Desulfuromonadales, Myxococcales, Desulfovibrionales, Desulfurellales, and Bdellovibrionales. The Desulfovibrionales abundances were highest in the raw and ozonated OSPW samples, at 45% and 80%, respectively. The Desulfuromonadales and Myxococcales were most abundant in the raw and ozonated GAC biofilms, respectively. Myxococcales are aerobic and carbon degraders (63–65), whereas the members of Desulfuromonadales, Desulfovibrionales, and Desulfuromonas orders are strictly anaerobic and sulfate/sulfur reducers (65–67). Desulfurellales species are capable of oxidizing saturated fatty acids via sulfur reduction (66, 68). Bdellovibrionales members are Gram-negative, motile, and uniflagellated bacteria characterized by predatory behavior (or an obligatory parasitic life cycle) which can digest the multidrug-resistant pathogenic bacteria associated with human infection (69). Bdellovibrio organisms usually grow in the highly polluted niches, such as industrial effluents, areas in rivers and estuaries where they can degrade organic carbon as prey independently (70).

Water quality parameters.

Concentrations of the individual NAs and total summed NAs for raw and ozonated OSPW both before and after biodegradation treatment are shown in Fig. 4. The relatively low ozone dose (30 mg/liter of utilized ozone) was able to reduce the raw OSPW (16.95 mg/liter) NAs by about 40% after ozonation (10.01 mg/liter). The removal after the various treatments is shown in Table 1 for COD, AEF, and NAs (based on total NAs). Slight improvements in removals of COD and AEF were observed after GAC biofilm treatments for ozonated OSPW (21% and 28%) compared to the removal from raw OSPW (18% and 23%) (Table 1). However, there was a statistically significant reduction in NAs between the raw (31%) and ozonated (66%) OSPW treatments. The removals of COD, AEF, and NAs from the GAC adsorption-only mechanism were evaluated by using 0.1% (wt/vol) sodium azide in raw and ozonated OSPW to inhibit biodegradation (Table 1). The removal from biofilm biodegradation only could then be assessed from the difference between the combined versus GAC adsorption-only experiments for raw and ozonated OSPW (Table 1). For COD, biodegradation accounted for 7.1 to 9.0% of the overall combined treatment removal, with 10.5 to 11.5% due to adsorption. For AEF, biodegradation accounted for 6.0 to 8.4% of the overall combined treatment removal, with 17.4 to 19.6% due to adsorption. Clearly, for the COD and AEF the GAC adsorption plays a greater role in the combined removal; however, the benefit of biodegradation is the complete mineralization of organic compounds to water and carbon dioxide versus their adsorption on the surface, which is a transfer of organics from liquid phase to solid phase.

FIG 4.

NA concentrations in raw and ozonated OSPW before and after 2 days of treatment. (A) Raw OSPW; (B) biodegradation control treatment of raw OSPW using 0.1% (wt/vol) sodium azide; (C) combined GAC adsorption and biofilm treatment of raw OSPW; (D) ozonated OSPW; (E) biodegradation control treatment of ozonated OSPW using 0.1% (wt/vol) sodium azide; (F) combined GAC adsorption and biofilm treatment of ozonated OSPW.

TABLE 1.

Removal of COD, AEF, and NAs in raw and ozonated OSPW after 2 days of treatment

Parameter	% removal from:
	Raw OSPW			Ozonated OSPW
	Combineda	GAC adsorptionb	Biofilm biodegradationc	Combineda	GAC adsorptionb	Biofilm biodegradationc
COD	17.6	10.5	7.1	20.5	11.5	9.0
AEF	23.4	17.4	6.0	28.0	19.6	8.4
Classical NAs	30.7	26.0	4.7	66.7	49.7	17.0

^aCombined: GAC biofilm biodegradation and adsorption.

^bGAC adsorption: GAC biofilm + sodium azide to remove biodegradation potential.

^cBiofilm biodegradation = combined − GAC adsorption.

More interestingly, removal rates of about 4% and 26% of total NAs were attributed to biodegradation and GAC adsorption, respectively, for raw OSPW treatments, compared to 17% and 50% for the respective treatments for ozonated OSPW. Ozonation of OSPW reduced the toxicity and increased the biodegradation of NAs in ozonated OSPW. Thus, the higher removal of NAs from ozonated OSPW treatment than from raw OSPW treatment may be attributed to the increased bioavailability and reduced toxicity of NAs from ozonated OSPW on the GAC surface. Moreover, the higher diversity and richness of microbial communities in the ozonated GAC biofilm may more efficiently degrade the supply of more biodegradable organic compounds produced by OSPW ozonation (see Table S1 in the supplemental material). It has been reported that ozonation produces oxidized NAs (51), which contain more OH groups in their structure and cause less hydrophobicity (71) and thus have less toxicity (72) than do classical NAs. Moreover, it was found that the higher concentrations of bituminous compounds in the sediments of the Athabasca River reduced the biofilm bacterial activity in an annular biofilm reactor because of inhibitory effects (24). Thus, it is postulated that the reduced degradation of NAs on the GAC surface from raw OSPW treatment may be a result of the acute toxicity of the concentrated classical NAs on the GAC surface inhibiting the growth of some microbial orders, such as Bacteroidetes and Acidobacteria, on the GAC.

Interestingly, of the parameters measured, the highest removal rates were observed for NAs in raw and ozonated OSPW treatments rather than COD and AEF. It has been reported that classical NAs include mostly highly branched, cyclic compounds, with high molecular weights and long carbon chains which lead to this higher hydrophobicity (73). Thus, the higher hydrophobicity of the NAs was attributed to the higher adsorption affinity to GAC and biofilms; in addition, higher rates of removal of classical NAs were observed. Moreover, the biodegradation of adsorbed NAs in biofilm and on GAC may produce intermediate biodegradable products such as oxidized NAs, which contribute to COD and AEF. AEF include classical and oxidized NAs and other organic compounds with functional groups containing carboxylic acids, ketones, and aldehydes (74) and have a medium bulk adsorption affinity to GAC. COD represents the oxidation of all organic and inorganic compounds and ions and has the lowest bulk adsorption affinity to GAC.

The removal of individual classical NAs groups in raw and ozonated OSPW based on their carbon and Z numbers is shown in Fig. 5. Higher percentages of NA removal from ozonated OSPW than from raw OSPW were observed for all carbon and Z numbers. Generally, higher rates of removal of NAs were observed at higher carbon numbers (n = 16 to 20) and Z numbers (Z = −10 to −12). The lowest carbon number (n = 11) and Z number (Z = −2) groups exhibited lower rates of removal that may have been an artifact of their lower abundances impacting the quantification. Overall, the biodegradation and adsorption processes provide different impacts on carbon and Z numbers. The lower-molecular-weight (i.e., lower carbon and Z numbers) NAs have been shown to have higher biodegradability in previous studies (4, 75). In contrast, the higher-molecular-weight NAs (i.e., higher carbon and Z numbers) show increased adsorption to GAC due to increases in hydrophobicity and nonpolarity of NAs in these groups (73, 76). These longer molecules can be adsorbed more readily on the GAC surface (73) and can more easily approach the GAC surface to reach GAC pores. Based on this assessment, the combination of biodegradation and adsorption could potentially lead to the removal of all molecular weights of NAs. Thus, in the current study, the higher rates of removal of the species with high carbon and Z numbers indicate a greater contribution of adsorption. However, the impact of biodegradation of these species cannot be discounted, as they may be biodegraded simultaneously with GAC adsorption.

FIG 5.

Relative concentrations of NAs in raw and ozonated OSPW (lower values indicate higher rates of removal) after 2 days of treatment (based on data from Fig. 4). (A and B) Summation of NA concentration based on Z number (Σ[NAs]_z) for each carbon number in the treated OSPW divided by initial concentrations of Σ[NAs]₂₀ in untreated raw (A) and ozonated (B) OSPW; (C and D) summation of NA concentrations based on carbon number (Σ[NAs]_n) for each Z number in the treated OSPW divided by initial concentrations of Σ[NAs]_n0 in the untreated raw (C) and ozonated (D) OSPW.

The 454 pyrosequencing analysis showed that the planktonic OPSW and GAC biofilms contain diverse microbial communities dominated by carbon degraders of simple carbon and polyaromatic carbons (Table 1) which are capable of degrading OSPW organics. Numerous studies have reported that GAC biofilm treatments exhibited simultaneous adsorption and biodegradation mechanisms for the removal of organic compounds from other wastewater matrices (77–79). This combination indicates the ability of GAC to bioregenerate due to the biodegradation activity of microorganisms colonizing the external surface and macropores of the GAC (80). The combined effect of adsorption and biodegradation on contaminant removal rates is dependent on the microbial abundances and compositions, their metabolic rates, and the biofilm retention time. Thus, the current study indicates that a diversity of carbon-degrading bacterial orders (e.g., Burkholderiales) can be considered for the bioaugmentation of biofilm reactors through bacterial seeding for OSPW treatment optimization. The higher rate of removal of NAs from ozonated OSPW (66%) than from raw OSPW (31%) indicated higher microbial activity in the GAC biofilm due to the reduced toxicity and improved biodegradability of ozonated OSPW NAs (81, 82). Given the reductions of the NAs, in addition to the reductions of the COD and AEF, it can be concluded that the GAC biofilm treatment is a promising new technology for the treatment of ozonated OSPW. However, this technology will need further study and optimization to improve the efficiencies of removal of each of these contaminants to become a realistic treatment process for the vast quantities of OSPW.