Microbial Community Composition and Dynamics of Moving Bed Biofilm Reactor Systems Treating Municipal Sewage

Abstract

Moving bed biofilm reactor (MBBR) systems are increasingly used for municipal and industrial wastewater treatment, yet in contrast to activated sludge (AS) systems, little is known about their constituent microbial communities. This study investigated the community composition of two municipal MBBR wastewater treatment plants (WWTPs) in Wellington, New Zealand. Monthly samples comprising biofilm and suspended biomass were collected over a 12-month period. Bacterial and archaeal community composition was determined using a full-cycle community approach, including analysis of 16S rRNA gene libraries, fluorescence in situ hybridization (FISH) and automated ribosomal intergenic spacer analysis (ARISA). Differences in microbial community structure and abundance were observed between the two WWTPs and between biofilm and suspended biomass. Biofilms from both plants were dominated by Clostridia and sulfate-reducing members of the Deltaproteobacteria (SRBs). FISH analyses indicated morphological differences in the Deltaproteobacteria detected at the two plants and also revealed distinctive clustering between SRBs and members of the Methanosarcinales, which were the only Archaea detected and were present in low abundance (<5%). Biovolume estimates of the SRBs were higher in biofilm samples from one of the WWTPs which receives both domestic and industrial waste and is influenced by seawater infiltration. The suspended communities from both plants were diverse and dominated by aerobic members of the Gammaproteobacteria and Betaproteobacteria. This study represents the first detailed analysis of microbial communities in full-scale MBBR systems and indicates that this process selects for distinctive biofilm and planktonic communities, both of which differ from those found in conventional AS systems.

INTRODUCTION

The moving bed biofilm reactor (MBBR) system was developed in the late 1980s for the treatment of domestic and industrial wastewaters. These systems are now operating in more than 22 countries (including New Zealand) and range from large- to small-scale wastewater treatment plants (WWTPs) (35). The MBBR process combines features of both fixed-growth and activated sludge (AS) systems in that the microbial community is largely retained within the reactor as a biofilm on suspended carriers, with a smaller planktonic fraction being present in suspension as free-floating cells or small flocs. MBBR technology offers a number of advantages over conventional technologies for treating waste, including a high effluent quality, no bulking problems, and lower cost. Other advantages of fixed biofilm growth include the decoupling of biomass retention from hydraulic retention time leading to longer sludge ages and low waste sludge volumes (5, 37).

Studies on the microbial community composition of conventional activated sludge systems indicate that the community is typically dominated by aerobic or facultatively anaerobic heterotrophic bacteria belonging to the Betaproteobacteria (41). It is unclear whether similar communities are found in MBBR processes, as there have been no microbiological studies on full-scale wastewater systems. Differences in microbial communities might be expected given that MBBR systems support development of microbial biofilms within which microenvironments support the growth of both anaerobic and aerobic organisms within the same ecosystem (19). Molecular techniques such as FISH and denaturing gel gradient electrophoresis (DGGE) have demonstrated the presence of organisms associated with simultaneous nitrification and denitrification within a lab-scale continuous-flow MBBR system treating wastewater under aerobic conditions (16). The presence of a reduced oxygen gradient within biofilms has the potential to also support the growth of anaerobic ammonium-oxidizing bacteria (anammox) (47). Anammox is a process that is of significant recent interest in the field of wastewater treatment because of the energy efficiencies that can be achieved in the conversion of ammonium and nitrite to nitrogen gas under anaerobic conditions (46).

Biofilm development is a key process in the establishment of an effective MBBR process. To maintain effective gas and nutrient transfer, the ideal biofilm is relatively thin and evenly distributed over the carrier surface (31). This can be influenced by turbulence in the reactors, which also influences substrate and oxygen transfer. Aside from these few key factors, the effects of other operational parameters and influent composition on microbial community structure and function within MBBR systems are poorly understood.

The aim of this study was to investigate the microbial communities in Wellington's Moa Point (MP) and Karori WWTPs and thus to provide the first comprehensive insight into the key microbial groups in full-scale MBBR systems. Approximate monthly samples, consisting of suspended biomass and biofilm scraped from carriers, were collected over a 12-month period from the two plants. A full-cycle community analysis approach, including analysis of 16S rRNA gene libraries, fluorescence in situ hybridization (FISH) and automated ribosomal intergenic spacer analysis (ARISA), was used to determine bacterial and archaeal community composition and dynamics. Total dissolved sulfides and dry/wet weight of biofilm adhering to carriers were also determined. Microbial community analysis was also performed on a sample from a conventional floc-based activated sludge system for comparison.

MATERIALS AND METHODS

Sample sites.

Sampling was carried out at Wellington's MP and Karori WWTPs. The MBBR reactors at both plants contained suspended polyethylene carriers (K1 media; AnoxKaldnes) comprising 30 to 50% of reactor volume. Reactor samples, comprising suspended K1 carriers with adherent biofilm, were collected once a month over a year from three MBBR reactors at MP (designated M1, M2, and M3) and two reactors from Karori (designated K1 and K2) treatment plants. For comparison, mixed liquor from a conventional activated sludge system was collected from a large municipal WWTP in northern New Zealand. Samples were collected in one-liter bottles and transported refrigerated overnight to the laboratory.

Physical and chemical analyses.

Dry and wet weight determinations were made on biofilm from five K1 carriers from each sample. To determine the wet weight, carriers with adherent biofilm were blotted dry on tissue paper for 2 min and then weighed. The samples were then transferred to a desiccator and dried for 1 week and then reweighed. Control samples comprising five unused carriers were also subjected to desiccation and then weighed. Dry and wet weight determinations were made following subtraction of the average weight of the control carriers. The statistical significance of differences between samples was determined using a t test (unequal variance). To measure the suspended biomass, 1 ml of sample was pelleted by centrifugation at 13,000 × g for 10 min in a 1.5-ml microcentrifuge tube. The supernatant was carefully removed, the wet weight determined, and the opened tube subjected to desiccation and then weighed as described above.

Total dissolved sulfide was measured immediately upon receipt of samples by a colorimetric method as outlined previously (13). A standard curve was prepared using various concentrations of sodium sulfide. The absorbance of copper sulfide precipitate for each sample was measured at 460 nm in a UV-Vis spectrophotometer.

DNA extraction.

Total genomic DNA was extracted from biomass using a phosphate, SDS, chloroform-bead beater method as described previously (44). Extracted DNA was dissolved in 50 μl of DNase-free water and stored at −20°C until further analysis.

ARISA.

ARISA was used as a rapid method to profile bacterial community structure and to make comparisons between monthly samples. The intergenic spacer region between the 16S and 23S rRNA genes was amplified using two universal bacterial primers, SDBact and LDBact (33), in a PCR described previously (21). The fluorescently labeled products were purified using a QIAquick PCR purification Kit (Qiagen) and analyzed along with an internal LIZ1200 standard on a 3130XL capillary genetic analyzer using a 50-cm capillary (Applied Biosystems Ltd., New Zealand).

Results from ARISA were analyzed using GeneMapper software (version 3.7) to create bacterial community profiles for each sample. Multidimensional scaling (MDS) plots were constructed from community profile data using Primer 6 software (version 6.1.6). Manhattan distance was chosen as the measure between samples on the MDS plots.

16S rRNA gene analysis. (i) Clone library analysis.

Cloning and restriction fragment length polymorphism (RFLP) analysis of PCR-amplified 16S rRNA genes was performed as generally described previously (7). Forward and reverse primers for PCR amplification of bacterial 16S rRNA genes were 5′-AGRGTTTGATCMTGGCTCAG-3′ and 5′-GKTACCTTGTTACGACTT-3′, respectively (39). Primers used for analysis of archaeal 16S rRNA gene sequences were 5′-TTCCGGTTGATCCYGCCGA-3′ (Arch21F) and 5′-YCCGGCGTTGAMTCCAATT-3′ (Arch958R) (15). Cloned inserts were recovered by PCR amplification using the vector-specific primers PGEM-F (5′-GGCGGTCGCGGGAATTGATT-3′) and PGEM-R (5′-GCCGCGAATTCAACTAGTTGATT-3′) (1).

(ii) RFLP of clones.

Restriction endonuclease HaeIII (Invitrogen) digestion was used to generate RFLP profiles for archaeal clones to investigate diversity prior to selection of clones for sequencing as previously described (7). The resulting products were resolved and visualized by polyacrylamide gel electrophoresis (PAGE) through 6% nondenaturing gels as described previously (38). Gels were run at 120 V for 70 min and then stained with ethidium bromide. Unique RFLP profiles were designated operational taxonomic units (OTUs), and representative clones were selected for sequencing.

(iii) Sequencing and data analysis.

PCR-amplified inserts from representative clones were purified and sequenced using the aforementioned vector-specific primers. Purification and sequencing were performed under contract by Macrogen Inc. (Kumchun-Ku, Seoul, South Korea) using a 3730XL DNA analyzer (Applied Biosystems). A total of 96 clones were sequenced from each bacterial clone library, whereas one clone representing each OTU was sequenced from the Archaea libraries.

Sequence data were processed using the high-throughput pipeline available through the Ribosomal Database Project II (http://rdp.cme.msu.edu) (49) and also compared with the GenBank database sequences (http://www.ncbi.nlm.nih.gov) using nucleotide-nucleotide BLAST (2).

(iv) Phylogenetic analysis.

Sequences were screened for chimeras using Mallard software (version 1.02; School of Biosciences, Cardiff University [http://www.bioinformatics-toolkit.org/Mallard/index.html]) (6), and unreliable sequences were eliminated from further analysis. The remaining partial and full-length 16S rRNA sequences were aligned using the SINA Web alignment tool, SILVA (www.arb-silva.de/aligner) (32). These sequences were imported into the ARB software (25) along with the SSU Ref database (SILVA version 106) to construct phylogenetic trees. The SSU Ref database contains more than half a million curated full-length (>1,200 bp) 16S rRNA sequences that have been previously published or uploaded to public databases (such as GenBank). Phylogenetic trees were constructed using maximum likelihood methods.

(v) FISH.

FISH was used in combination with confocal laser scanning microscopy to examine samples for the presence of methanogens and Deltaproteobacteria (4, 27). Samples were fixed within 12 h of sample collection using 4% paraformaldehyde as described previously (27) and stored at −20°C in a 1:1 phosphate-buffered saline (PBS) and ethanol mixture awaiting further analysis. Slides were prepared from fixed samples by placing 20 μl of a sample into 6-mm diameter wells on Teflon-coated slides (ProSciTech), which were then air dried. Samples were dehydrated by immersion in 50%, 80%, and 98% ethanol, respectively, for 3 min each. Hybridization was carried out in a 50-m Falcon tube chamber at 46°C for 2 h. Oligonucleotide probes were derived from published sequences with a 5′-end fluorescent label (specified in Table 1) and synthesized by Thermo Fisher Scientific (Germany). Each well was hybridized with buffer (0.9 M NaCl, 0.01% SDS, 20 mM Tris-HCl, formamide at optimal concentrations for each probe) and 1 μl of 50 ng/μl probe. Following incubation, slides were immersed in preheated (48°C) wash buffer (20 mM Tris-HCl, 5 mM EDTA, NaCl at optimal concentrations for each probe) and air dried in the dark. Salts were removed from the slides by washing in ice-cold distilled water. DAPI (4′,6-diamidino-2-phenylindole) (10 μg/μl) was then applied as a universal DNA stain. Slides were incubated in the dark for 5 min and then rinsed with distilled water and air dried.

Table 1.

FISH probe sequences

Probe name	Target group	Probe sequence 5′ → 3′	Label	Reference
EUB338	Eubacteria	GCTGCCTCCCGTAGGAGT	CY3	3
MSMX860	Methanosarcinales	GGCTCGCTTCACGGCTTCCCT	CY3	34
DELTA495a	Most Deltaproteobacteria and Gemmatimonadetes	AGTTAGCCGGTGCTTCCT	CY5	24
cDELTA495a	Competitor for DELTA495a	AGTTAGCCGGTGCTTCTT		26

(vi) Confocal microscopy.

FISH slides were viewed using an FV1000 Olympus confocal laser scanning microscope (CLSM) equipped with a 100× oil immersion objective lens. Excitation of fluorescein isothiocyanate (FITC), Cy3, and Cy5 was performed at 488 nm (Ar laser), 543 nm (He-Ne laser), and 635 nm (red diode laser), respectively. Images were viewed by using an FV10-ASW2.0 (Olympus) viewer.

For quantification, at least five z-series of 1-μm depth (6 to 10 sections) were made for each sludge sample with a universal stain (DAPI) and a group-specific probe (MSMX860 or Delta495a). The biovolume percentage was calculated by using DAIME software (version 1.2; http://www.microbial-ecology.net/daime) (14).

RESULTS

Sampling was carried out at Wellington's MP and Karori WWTPs. Both plants are configured to include biological treatment in the form of aerated moving bed bioreactors. Solids are removed by contact stabilization and clarification. MP receives household and industrial waste, including that from an abattoir, and has an average dry weather flow of 822 liters/s with a BOD of 0.23 kg/m³. The Karori WWTP receives only domestic waste with an average dry weather inflow of 20 liters/s with a BOD of 0.37 kg/m³. Dissolved oxygen levels were regulated at similar levels between the two MBBR facilities of interest, with an annual average of 1.56 mg/liter for MP and 1.45 mg/liter for Karori. Elevated conductivity levels (yearly average, 4.072 mS/cm), indicative of seawater infiltration, are consistently detected at the MP plant while levels in influents reaching the Karori plant are low and typical of domestic effluents. A total of 11 approximately monthly samples were collected from each of the two treatment plants over a period of 12 months. Samples were taken from the aeration tank of the MBBRs and comprised K1 carriers with adherent biofilm and reactor fluid.

General characteristics of biofilm and suspended biomass.

The color and odor of biofilms attached to K1 media differed between sites. At MP the biofilms were black in color and had a sulfurous odor. In contrast, biofilm samples from the Karori WWTP were consistently greyish-brown and had no obvious odor. A comparison of biofilm quantity was made by determining the wet and dry weights of biofilms from each MBBR for all time points. No significant differences (P = 0.124) were observed between the dry weight of the biofilm samples from MP (0.029 ± SD 0.033 g, n = 8) and Karori (0.013 ± SD 0.006 g, n = 8). Similarly, no significant differences were observed between the wet weights of biofilm from the two sites (P = 0.379). P values of < 0.05 were considered significant.

To provide an estimate of the amount of suspended biomass, dry and wet weight determinations were also made on pellets prepared from 1 ml of supernatant fluids. Supernatant dry weight values were low (<0.021 g/ml) for all samples. No significant differences (P = 0.427) were found between samples from the two MBBR systems.

Total dissolved sulfides were detected in all samples from the MP site with values ranging between 1.5 to 13 mM. In contrast, sulfide was not detected (<1 mM) in samples collected from the Karori plant. Significant differences (P = 0.00012) were observed in sulfide values between the two MBBR sampling sites.

Bacterial community analysis. (i) ARISA.

ARISA was used as a rapid method to profile and compare the bacterial community structures in both biofilms and suspended biomass. MDS was used to investigate differences between ARISA community profiles over time and between treatment plants. These data indicated differences between the two WWTPs and also between biofilm and suspended communities (Fig. 1).

Fig 1.

MDS plot of ARISA data representing bacterial communities in samples collected over 12 months from MP and Karori WWTPs. Each symbol within the graph represents a bacterial community for one given time point. Minimum stress, 0.1. Bacterial communities from Moa Point (black) and Karori (gray) samples have formed two distinct clusters. Evidence of biofilm communities (enclosed within the black circles) segregating away from the suspended biomass is also displayed.

(ii) 16S rRNA gene analysis.

To determine the bacterial community composition, clone libraries of PCR-amplified 16S rRNA genes were prepared and sequenced from biofilm scraped from carriers and from suspended biomass. This analysis was performed on reactor samples collected at three time points (March 2010, August 2010, and January 2011). Figure 2 presents a summary of phyla detected in a representative sample from each plant. Phylogenetic trees constructed using the entire sequence data set are presented for all phyla (Fig. 3), the Deltaproteobacteria (Fig. 4) and the Epsilonproteobacteria (Fig. 5). A phylogenetic tree of sequences aligning to the Firmicutes is presented in Fig. S1 in the supplemental material.

Fig 2.

Composition of 16S clone libraries in biofilm (A) and suspended biomass (B) from Moa point and Karori reactors. Samples were collected in March 2010. For comparison, a clone library prepared from conventional AS mixed liquor (C) is also shown. Each color on the graph represents the dominant phyla or class found within the community.

Fig 3.

Maximum likelihood tree showing alignment of sequences from clone libraries prepared from biofilm and suspended communities. Numbers within brackets indicate percentages of clones from the MP biofilm, Karori biofilm, MP suspended, and Karori suspended samples, respectively. These numbers differ from those presented in Fig. 2, as they indicate percentages over three time points (March 2010, August 2010, and January 2011). The scale bar indicates 10% sequence divergence.

Fig 4.

Maximum likelihood tree showing alignment of sequences from clone libraries to the sulfate-reducing members of the Deltaproteobacteria. Clone sequences are presented in bold and identified by sample type, date, and clone reference, respectively. Sample type identifiers: M1, Moa Point biofilm; MS1, Moa Point suspended biomass; K1, Karori biofilm; KS1, Karori suspended biomass. Numbers in parentheses after the sequence identifier indicate number of clones within the cluster. Shaded boxes indicate diversity within a group of clones. Dotted lines indicate partial sequences (450 to 750 bp). Open and filled circles indicate clades supported by bootstrap values of ≥75% and ≥ 90%, respectively. Horizontal bars indicate groups within the Syntrophobacterales (Syn.), Desulfobacterales (Dsb.), and Desulfovibrionales (Dsv.). Outgroup consists of sequences from other bacterial phyla.

Fig 5.

Maximum likelihood tree showing alignment of sequences from clone libraries to the Epsilonproteobacteria. Vertical bars with brackets in bold indicate groups within the genus Arcobacter. Details are the same as those provided in Fig. 3 and 4.

Each treatment plant yielded a characteristic community composition that was consistent between reactors and time points (Fig. 2A). Biofilm communities from both plants showed limited bacterial diversity (Shannon-Wiener index of 0.93 to 1.18) and were dominated by Firmicutes (40 to 44% of clones). Phylogenetic alignment of sequences indicated a broad diversity within the Firmicutes (see Fig. S1 in the supplemental material), although the majority of clones represented members of the Clostridia (38% of clones). Strictly anaerobic, sulfate-reducing members of the Deltaproteobacteria (SRBs) were the second most abundant group, comprising 29.4% of the biofilm clone library at MP and 24.7% at Karori. Members of Desulfobacterales (11 to 19%), Syntrophobacterales (8 to 10%), and Desulfovibrionales (0.5 to 1.5%) were the most abundant SRBs found within these samples (Fig. 4). The biofilms from Karori WWTP differed due to an elevated incidence of Betaproteobacteria and Gammaproteobacteria (17.6% of clones), which were both in lower abundance (<7.4%) in the biofilm community from MP. A number of other organisms, including representatives of the phyla Bacteroidetes, Synergistes, Planctomycetes, Verrucomicrobia, and Acidobacteria, as well as various unclassified bacteria, were also detected at low abundance (<5% of clones) in the biofilm samples from both plants. Analysis of the archaeal community from the two MBBR systems indicated a very limited diversity, with only one RFLP pattern detected among the 20 clones screened. Sequence analysis indicated that this pattern represented organisms from the methanogen order Methanosarcinales.

In contrast to the biofilm samples, the suspended biomass from both MBBR systems (Fig. 2B) was dominated by a more diverse group of aerobic organisms from the Alphaproteobacteria (Rhizobiales and Rhodobacterales), Gammaproteobacteria (Pseudomonadales and Aeromonadales), and Betaproteobacteria (Burkholderiales and Rhodocyclales). Members of the Clostridia, representing the majority of the Firmicutes, were also present at MP (14% of clones) and Karori (40% of clones) treatment plants. In addition, MP samples had an elevated level of Epsilonproteobacteria (54% of clones) which were affiliated with the Campylobacteraceae and aligned most closely to Arcobacter spp. (Fig. 5). Though a diversity of sequences were observed, the majority aligned most closely to clones obtained from estuarine and river environments or to Arcobacter nitrofigilis.

The clone library prepared for comparative purposes from a mixed liquor sample from a conventional AS plant (Fig. 2A) was more diverse and dominated by members of the Betaproteobacteria (28%) and Bacteroidetes (25%).

(iii) FISH analysis.

FISH was used as a PCR-independent approach to validate the 16S rRNA gene library results and to investigate the spatial distribution and abundance of Archaea using a Methanosarcinales probe and putative sulfate-reducing bacteria using a Deltaproteobacteria probe. FISH analysis was performed on samples collected between March 2010 and June 2010, September 2010, December 2010, and January 2011.

The distinctive black biofilm from MP revealed colloidal clusters of Deltaproteobacteria buried within the biofilm structures (Fig. 6A). Based on the clone library results, these are likely to be sulfate-reducing bacteria. Samples from Karori also showed the presence of Deltaproteobacteria, but these were fewer in abundance and were filamentous (Fig. 6B). Biovolume analysis confirmed the occurrence of Deltaproteobacteria at both treatment plants, with higher volumes recorded in samples from MP WWTP (Fig. 6). However, there was no significant (P = 0.282) difference observed between the biovolumes of Deltaproteobacteria between the two study sites.

Fig 6.

Biofilm samples scraped from carriers (as seen in column 1) were hybridized with fluorescently labeled probes targeting methanogenic Archaea (MSMX860; red), Deltaproteobacteria (DELTA495a; cyan), and eubacteria (EUB338; green). DAPI (blue) was used as a universal DNA stain. (A) Clusters of Deltaproteobacteria (arrowed “s”) were abundant in black biofilm from MP. (B) Clusters of Deltaproteobacteria from Karori were present as filaments. Methanogenic Archaea (arrowed “m”) are visible in both samples. Biovolume estimates of methanogens and Deltaproteobacteria relative to DAPI-stained material are shown in graphs for both samples.

Methanosarcinales were observed in low numbers (<5% biovolume) in biofilms from the two MBBR systems. However, in samples from the MP plant, these were observed in distinctive clusters around cells hybridizing with the Deltaproteobacteria probe, which represent putative SRBs (Fig. 6A).

DISCUSSION

Since their invention, MBBR systems have been used to treat both municipal and a wide range of industrial wastes, including pulp and paper (43), cheese factory waste (36), and phenolic wastewater (11). Previous studies (35, 51) have reported on the physical and engineering aspects of MBBR systems, while the composition of the microbial communities responsible for the biological activity has received no attention in full-scale systems treating municipal wastewater. This study set out to investigate the prokaryotic community dynamics in a parallel study of two MBBR treatment plants over a 12-month period. The MBBR treatment systems included in this study are situated in the same city and are operated under similar parameters to treat urban municipal wastewater. The major operational difference between the two plants is that the Karori plant receives effluent from a largely residential catchment, while the MP plant services a much larger catchment that includes mixed urban and industrial uses.

Biofilm communities.

Differences in bacterial community profiles between the two WWTPs were expected due to the different sources of waste entering these systems. The results of this study indicated no difference in the amount of biomass retained on carriers, but distinct differences were noted in both coloration and dissolved sulfide concentrations. The observation that MP biofilms were black in color and sulfurous in odor is consistent with the detection of appreciable levels of dissolved sulfides at this site. The detection of sulfides at MP indicates the possible activity of anaerobic SRBs, which are commonly found in anoxic systems (8, 29). Sulfate is reduced to sulfide by SRBs, resulting in a pungent sulfurous odor and upon reaction with metals yields black precipitates.

Despite differences in the color of biofilms from the two sites, both yielded bacterial 16S rRNA gene libraries that were dominated by anaerobes, notably Clostridia and members of the Desulfobacterales and Syntrophobacterales, which are known sulfate-reducing bacteria (29). Both plants also showed similar communities of Archaea, which were dominated by Methanosarcinales. Validation of these results was performed by FISH using probes targeted to the Methanosarcinales and the Deltaproteobacteria. Although the Deltaproteobacteria probe is not specific for sulfate-reducing bacteria, these were the only members of the class detected in the clone libraries. It was therefore presumed that cells hybridizing to the Deltaproteobacteria probe were SRBs. Biovolume analysis confirmed the presence of SRBs in both sample sets but indicated a higher abundance in the MP site and different cell morphology than that seen in the Karori samples. These differences may go some way toward explaining the higher levels of sulfides detected at the MP site. It is also possible that differences in influent composition influence the community structure and function in these biofilms. In addition to domestic wastewater, the MP plant receives industrial waste from an abattoir, expected to carry large amounts of fats, oils, and greases (28). The presence of long-chain fatty acids (LCFA) derived from the degradation of lipids (45) and low water temperatures have been shown to select for SRBs in a wastewater treatment system (9, 22). Enhanced SRB activity would also require access to excess sulfate as an electron acceptor. The source of this in the MP plant has not yet been clearly established, although seawater infiltration is suspected in this system, as indicated by elevated conductivity levels. Seawater is known to contain appreciable concentrations of sulfate, and high concentrations have been found in an MBBR system treating waste from a marine aquarium (20).

FISH and biovolume analysis confirmed the presence of members of the Methanosarcinales in approximately equal abundance at both sites. This analysis also revealed a clear association of these methanogens with Deltaproteobacteria in the MP samples. The majority of SRBs described in the past are putatively free-living (29), but recent studies of the marine environment using FISH have revealed consortia of SRBs and methanogens (10, 18), suggesting a symbiotic bacterial-archaeal interaction. The physiological basis of this symbiosis is still a matter of conjecture. Although methanogens were detected at the Karori site, these were not observed in association with the filamentous SRBs, further supporting the notion that this community differs from that found in the MP samples. Filamentous SRBs belonging to the genus Desulfonema have been detected previously in marine and freshwater sediments (17, 50). Members of this genus exhibit gliding motility, which is presumed to confer the ability to move along gradients within dense mats and avoid predation by grazing protozoa. These features would be equally advantageous in both MBBR biofilms, though filamentous SRBs were detected only in the Karori samples. This further supports the notion that the biofilm communities in these two systems are influenced by external factors, which may include differences in influent composition. Indeed, other studies have confirmed that specialized groups of microorganisms develop over time in response to changes in complex organic matter entering the system (40).

Suspended biomass.

Clone library analysis indicated that the suspended fraction was dominated by aerobic members of the Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, and Epsilonproteobacteria. Members of the Clostridia were also present but in much lower abundance than in biofilm samples, possibly resulting from biofilm detachment processes which are known to occur periodically in such systems (23). A notable difference between the two plants was the high abundance of sequences in the MP clone library that aligned to Arcobacter spp., including Arcobacter nitrofigilis. The genus Arcobacter was proposed in 1991 (48) to describe a group of aerotolerant members of the Campylobacteraceae and has been of increasing interest as an emerging zoonotic pathogen. The genus includes species isolated from a diverse range of aquatic and terrestrial environments as well as the feces and reproductive tracts of domestic livestock, sewage, and abattoir effluents. The genus type species, Arcobacter nitrofigilis, is a nitrogen-fixing organism that was isolated from the root zone of a salt-marsh plant although not all species are salt tolerant (12). The reason for the abundance of Arcobacter spp. in the MP suspension samples remains unclear although the MP plant receives effluent from a local abattoir, and it is possible that they derive from this source.

Consistent differences between the planktonic and biofilm communities are indicated by the MDS analysis of the ARISA results. These differences may be explained by the different conditions that prevail within these structured environments. The relatively short hydraulic retention time within the MBBR system selects for organisms that have a high growth rate in order to withstand washout and to be retained in the reactor. The bulk liquid phase is also aerated, supporting aerobic metabolism. While members of the Clostridia were detected in the suspended phase, the majority of taxa identified in the clone library analysis were from putatively fast-growing aerobic species, including members of the Burkholderiales, Rhodocyclales, Aeromonadales, Pseudomonadales, and Rhodobacterales.

Comparison with activated sludge.

Microbial communities in activated sludge systems have been widely discussed in the literature as the majority of modern wastewater treatment facilities utilize this technology (42). Activated sludge typically comprises free-swimming planktonic cells and aggregated flocs that can be maintained under both aerobic and anoxic conditions. Bacterial communities in AS plants treating industrial and municipal wastewater are typically dominated by Betaproteobacteria, followed by Alphaproteobacteria and Gammaproteobacteria. Other groups of bacteria found in low abundance are Bacteroidetes and Firmicutes (30, 41). The clone library prepared from a conventional AS mixed liquor sample in this study yielded results consistent with studies reported in the literature and suggests that both the attached biofilm and suspended communities in MBBR systems differ substantially from that found in conventional AS.

In summary, this study provides the first detailed investigation of key microbial groups in full-scale MBBR systems treating municipal wastewater. The results indicate that MBBR communities differ substantially from those in conventional AS systems by selecting for two distinct bacterial communities: a biofilm community that is dominated by anaerobes and a suspended community that includes fast-growing aerobic bacteria. The observation, from FISH analyses, of a close association between Methanosarcinales and putative SRBs in the MP biofilms implies a functional relationship between these two groups of microorganisms. Further studies that include consideration of both the prokaryotic and eukaryotic communities are required to elucidate the nature of these microbial relationships and to develop food web models that will underpin manipulation of the system for optimal performance.