Distinct Microbial Community Performing Dissimilatory Nitrate Reduction to Ammonium (DNRA) in a High C/NO3− Reactor

A dissimilatory nitrate reduction to ammonium (DNRA) microbial community was developed under a high organic carbon to nitrate (C/NO3−) ratio in an anoxic semi-continuous sequencing batch reactor (SBR) fed with glucose as the source of carbon and NO3− as the electron acceptor. Activated sludge collected from a municipal wastewater treatment plant with good denitrification efficiency was used as the inoculum to start the system. The aim of this study was to examine the microbial populations in a high C/NO3− ecosystem for potential DNRA microorganisms, which are the microbial group with the ability to reduce NO3− to ammonium (NH4+). A low C/NO3− reactor was operated in parallel for direct comparisons of the microbial communities that developed under different C/NO3− values. The occurrence of DNRA in the high C/NO3− SBR was evidenced by stable isotope-labeled nitrate and nitrite (15NO3− and 15NO2−), which proved the formation of NH4+ from dissimilatory NO3−/NO2− reduction, in which both nitrogen oxides induced DNRA activity in a similar manner. An analysis of sludge samples with Illumina MiSeq 16S rRNA sequencing showed that the predominant microorganisms in the high C/NO3− SBR were related to Sulfurospirillum and the family Lachnospiraceae, which were barely present in the low C/NO3− system. A comparison of the populations and activities of the two reactors indicated that these major taxa play important roles as DNRA microorganisms under the high C/NO3− condition. Additionally, a beta-diversity analysis revealed distinct microbial compositions between the low and high C/NO3− SBRs, which reflected the activities observed in the two systems.

Dissimilatory nitrate (NO 3 -) reduction is part of the nitrogen cycle driven by microorganisms with the ability to use NO 3as their electron acceptor. Known dissimilatory NO 3 --reducing pathways include denitrification, which generates nitrogenous gases as its products, and dissimilatory nitrate reduction to ammonium (DNRA), which produces ammonium (NH 4 + ). Both of these processes occur under anoxic environments and utilize the same types of electron donors, such as organic carbon, sulfur, and iron (6,12,28). Therefore, denitrifiers and DNRA microorganisms are competitors for electron donors and electron acceptors (NO 3and nitrite [NO 2 -]), as well as for the habitats in which they grow. Although the microorganisms responsible for denitrification have been extensively examined, limited information is currently available on the DNRA microbial group.
Although different nomenclatures have been termed for this dissimilatory NH 4 + -forming process, including NO 3 -or NO 2ammonification, it is mainly known as DNRA. This microbial group may be further classified into respiratory and fermentative types (7,22), which use different metabolic pathways to catalyze the formation of NH 4 + . DNRA microorganisms have been hypothesized and observed to occur in a high electron donor, limited NO 3environment, which is normally referred to as a high carbon to NO 3 -(C/NO 3 -) condition, whereas denitrifiers appear to prefer a low C/NO 3condition (7,34). Thermodynamically, the energy gained per mole electron donor and per mole NO 3differs for denitrification and DNRA, with the former obtaining a higher amount of energy per glucose molecule and the latter gaining more energy per NO 3molecule (32). These findings suggest how DNRA microorganisms compete with denitrifiers in a NO 3 -limiting ecosystem. The influence of the C/NO 3ratio on NO 3 --reducing communities or a pure culture was recently reported (23,(36)(37)(38)44), with selective pressure of the ratio on the NO 3reduction pathways being observed.
In agriculture, DNRA is considered to be a beneficial process because it preserves nitrogen fertilizers within soil (29). However, this pathway is undesirable in a biological wastewater treatment system due to its production of NH 4 + , a waste product that is generally removed by nitrification-denitrification. The growth and activity of DNRA microorganisms decrease denitrification efficiency, and increase the waste load on nitrification as well as total nitrogen discharged with the effluent. DNRA activity was previously detected in anaerobic digester systems (1,2,13,21), a pilot plant treating sulfate (SO 4 2-)-and NO 3 --containing wastewater (14), a lab-scale denitrifying reactor (4), and aquaculture nitrogen removal systems (10). The extent of DNRA measured among these studies suggests its competitive potential in high organic-loaded wastewaters; however, the microorganisms responsible for the process have rarely been identified in these environments.
In order to control the occurrence of microorganisms with the capacity for DNRA within a wastewater treatment system, it is essential to recognize their identity and diversity as the first step towards maneuvering these populations to the required process. Therefore, in order to identify potential DNRA microorganisms in wastewater ecosystems, an anoxic semicontinuous sequencing batch reactor (SBR) maintained at a high C/NO 3ratio was applied to monitor subsequent microbial community adaptation when using activated sludge as the inoculum. In a direct comparison, a low C/NO 3reactor was operated under otherwise the same environmental conditions and using the same inoculum. The objectives of the present study were (i) to investigate the presence and activity of the DNRA pathway under low and high C/NO 3anoxic sludge ecosystems and (ii) to identify the composition of the resulting microbial communities using an Illumina MiSeq 16S rRNA sequencing analysis. In order to trace the occurrence of DNRA, the stable-isotope tracers 15 NO 3and 15 NO 2were utilized to track the formation of 15 NH 4 + in low and high C/NO 3 -SBR sludge samples.

Operation of the reactors
The two reactors were maintained under low and high C/NO 3ratios, respectively, and operated in the semi-continuous sequencing batch mode, which provided environments for the growth of microorganisms favored under each condition. During each cycle, the systems were set to the continuous feeding batch mode in order to maintain stable concentrations of substrates in the reactors, while also retaining a large amount of biomass sludge in each SBR. Both SBRs were started with an inoculum (1,393±42 mg mix liquor suspended solids [MLSS] L -1 ) from a municipal wastewater treatment plant operated under an anaerobic/oxic/anoxic/oxic (AOAO) process exhibiting good denitrification efficiency with a nitrogen removal rate of approximately 60%. SBRs were run in 1-L working volume reactors with six-blade turbine stirrers, and were maintained at 20°C in a control temperature room. Mixing was performed at 150 rpm in order to maintain a homogenous state inside the systems. During days 1-15 of the operation, the reactor cycle was set to 6 h, which comprised 320 min of continuous feeding and mixing, 30 min of sludge settling, and 10 min of effluent withdrawal. Only unsettled sludge was removed with the discharged effluent during the end of the cycle. The volume exchanged per cycle was set at 1/4 of the reactor working volume, whereas the flow rate was set at 47 mL h -1 (equal to a dilution rate of 0.047 h -1 ), resulting in a hydraulic retention time (HRT) of 24 h. After day 15, the cycle was changed to 12 h (680 min of continuous feeding and mixing) while the exchanged volume remained the same; therefore, the flow rate was adjusted to 22 mL h -1 during this period (equal to a dilution rate of 0.022 h -1 ) with 48 h of HRT. The flow rates of the influent and effluent were controlled by peristaltic pumps (Masterflex; Cole-Parmer, Chicago, IL, USA), whereas the reactor cycle was set using timers. The solid retention times (SRT) of each reactor during the operation were 3.40±0.45 and 2.52±0.97 d for the low and high C/NO 3systems, respectively.
Glucose was selected as the carbon and energy source for microbial growth, and its concentration was varied in order to achieve COD/ NO 3 --N ratios of 4/1 and 8/1 for the low and high C/NO 3 -SBRs, respectively (for details, see Table S1). NO 3was supplied at the same concentration in both systems. The medium fed to the reactors during the 6-h cycle period was composed of the following (L -  (40). The concentrations of all nutrients were doubled during the 12-h cycle in order to obtain the same loading as previously. Media were autoclaved and later supplemented with filter-sterilized glucose solution in order to meet the specified COD/NO 3 --N ratio for each reactor. Media were then flushed with argon (Ar) gas for 30-60 min (depending on the volume prepared) before being added to the system.

Chemical analysis
Water samples were collected daily from both SBRs in order to monitor changes in inorganic nitrogen and organic carbon in the systems. Samples were initially filtered through a glass microfiber filter with a pore size of 0.7 μm (Grade GF/F; GE Healthcare, Little Chalfont, UK) before the chemical analysis. The concentrations of NO 3 -, NO 2 -, and SO 4 2were analyzed by ion chromatography (IC) equipped with an anion column (861 Advanced Compact IC; Metrohm, Herisau, Switzerland), whereas NH 4 + was measured by IC with a cation column (761 Advanced Compact IC; Metrohm). Organic carbon was monitored in the form of dissolved organic carbon (DOC) using a total carbon analyzer (TOC-V; Shimadzu, Kyoto, Japan). Changes in biomass were quantified in terms of MLSS as described by the Standard Methods for the Examination of Water and Wastewater (3). The pH of each withdrawn sample was measured using a pH meter.

Stable isotope incubations
The stable isotope tracers, 15 NO 3and 15 NO 2 -, were applied to verify the occurrence and extent of DNRA in the low and high C/NO 3 -SBR samples. This was performed by monitoring the amount of 15 NH 4 + produced from the dissimilatory 15 NO 3or 15 NO 2reduction. Sludge for the incubations was taken from the reactors at the end of the operation (on days 60 and 54 for the low and high C/NO 3systems, respectively). Experiments were conducted in the batch mode and samples were collected in time series. 15 NO 3and 15 NO 2were used in the form of Na 15 NO 3 and Na 15 NO 2 (NLM-157-PK and NLM-658-PK, respectively, both with 98%+ 15 N atom; Cambridge Isotope Laboratories). Before the incubations, sludge taken from the reactors was centrifuged to remove the original medium, washed once with fresh medium (same composition as that prepared for SBRs, excluding NH 4 Cl), and then placed into 100-mL serum bottles along with new medium. Serum bottles were closed with butyl rubber stoppers and sealed with aluminum caps before flushing with Ar gas for 10 min. The incubations were then started by injecting either 15 NO 3or 15 NO 2into the serum bottles to a final concentration of 20 mg N L -1 , along with 14 NH 4 Cl to a final concentration of 100 mg N L -1 and glucose as appropriate to make the same COD/NO 3 --N ratio as in the reactors. All stock solutions were flushed with Ar gas before use. Samples were placed on a rotary shaker set at 150 rpm and temperature was maintained at 20°C. All incubations were performed in triplicate for the sludge of each reactor. The measurement of 15 NH 4 + produced from the dissimilatory 15 NO 3or 15 NO 2reduction was conducted as previously described (19). Since the original medium from the reactor had been washed off the sludge before the incubations, the amounts of 14 NO 3and 14 NO 2in the samples were considered to be negligible and changes in 15 NO 3and 15 NO 2were measured using the IC method (861 Advanced Compact IC; Metrohm).

Illumina MiSeq 16S rRNA sequencing analysis of microbial communities
The microbial communities of the low and high C/NO 3 -SBRs were examined using the Illumina MiSeq 16S rRNA sequencing method. Samples analyzed included sludge collected on days 27, 30, 38, 42, and 48 from both reactors as well as the inoculum used to start the systems. Total microbial DNA was extracted from samples using the FastDNA TM SPIN Kit for Soil (MP Biomedicals, Santa Ana, CA, USA) as described by the manufacturer's protocol. Extracted DNA was then used as a template for PCR amplification with the 341F and 805R primer pair (targeting the V3 and V4 regions of the 16S rRNA gene). The PCR reaction mixture (25 μL) was prepared from a TaKaRa Ex Taq TM kit (Takara, Otsu, Japan) with 12 ng of the DNA template, 0.2 μM of each dNTP, 2 mM MgCl 2 , 0.2 μM of each primer, and 1.25 U of TaKaRa Ex Taq TM . The PCR thermal steps were as follows: 94°C for 3 min; 25 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s, and then a final 72°C for 5 min. Four PCR reactions were conducted per sample to reduce bias during PCR amplification. PCR products were then checked by 1.5% (w/v) agarose gel electrophoresis in order to ensure that the correct-sized products were amplified (approximately 460 bp).
Four replicates of each sample were subsequently combined and purified with a NucleoSpin ® Gel and PCR Clean-up kit (Macherey-Nagel, Bethlehem, PA, USA) according to the manufacturer's protocol. The concentrations of the PCR products were examined with a spectrophotometer (NanoDrop TM 2000c Spectrophotometer; Thermo Fisher Scientific, Waltham, MA, USA). The 16S rRNA amplicons obtained were then prepared in a 50-μL final volume reaction containing 5 μL of each Illumina Nextera XT index primer (Illumina, San Diego, CA, USA) and 2X KAPA HotStart ReadyMix (Kapa Biosystems, Wilmington, MA, USA) to index each sample. The PCR thermal steps used were as follows: 94°C for 3 min, followed by 8-10 cycles of 98°C for 20 s, 55°C for 30 s, and 72°C for 30 s, and then a final 72°C for 5 min. The indexed PCR amplicons were then purified with Agencourt AMPure XP beads (Agencourt Bioscience, La Jolla, CA, USA), pooled, and diluted to a final loading concentration of 4 pM. DNA paired-end sequencing was conducted on the Illumina MiSeq platform with MiSeq Reagent Kits v3 (600 cycles; Illumina) at the Omics Sciences and Bioinformatics Center (Chulalongkorn University, Bangkok, Thailand). Sequences were assembled by merging forward and reverse reads, and the operational taxonomic unit (OTU) picking process was performed using QIIME software (version 1.9.0) with an open reference method. Taxonomic assignment from the OTUs to related taxa was performed using UCLUST against the Greengenes database (version 13_8, with a 97% sequence identity threshold). Sequence reads in this study may be accessed at the NCBI Short Read Archive (SRA) under accession numbers SRX3440460-SRX3440470. OTUs with ≥1% and ≥2% maximum relative abundance were selected as representatives for class-level and genus-level analyses, respectively. The beta-diversity of each sample was evaluated in QIIME, and the results obtained were used for the principal coordinate analysis (PCoA) using the weighted UniFrac distance metric (24). Statistical analyses of the results obtained were performed using a two-sample t-test on SPSS Statistic ver. 22 (IBM, Armonk, NY, USA).

Competition between denitrification and DNRA in low and high C/NO 3 -SBRs
In the operation of the reactors, semi-continuous SBRs enriched with NO 3 --reducing communities exhibited the ability to maintain a stable C/NO 3ratio during the feeding period (see Fig. S1) as well as the biomass within the systems.
Apart from the C/NO 3ratio, the applied dilution rate is another factor known to affect competition for NO 3in a reactor setting, in which DNRA microorganisms appear to require a low dilution rate for growth (27). This feature may be related to the kinetic aspect of these microorganisms because the requirement for a low dilution rate indicates a long generation time for growth (23), the need for a low NO 3concentration to kinetically compete with denitrifiers, or both. Therefore, the dilution rate used in the present study was selected from a range that was expected to allow DNRA to occur. All other operating parameters, besides the glucose concentration, were the same in the low and high C/NO 3 -SBRs in order to allow the C/NO 3ratio to be the determining factor for the successful NO 3 --reducing pathway in each ecosystem.
Competition between the two NO 3reduction pathways was initially verified by observing changes in the inorganic nitrogen profiles of the low and high C/NO 3 -SBRs (Fig. 1). During the initial phase of the low C/NO 3reactor operation, NO 3was detected in the system in the range of 17.7-52.5 mg N L -1 (47.8-82.3% NO 3conversion) and was consumed to less than 1.0 mg N L -1 (>99% NO 3conversion) after day 17, which suggested that NO 3had become limited for heterotrophic growth or that it was used by autotrophic NO 3reducers. NO 2was only detected on day 4 (1.7 mg N L -1 ) and then remained below the detection limit (see Fig. S2). The level of NH 4 + (23.2-95.1 mg N L -1 ) was always lower than the added NH 4 + concentration, and, thus, there was no obvious indication of DNRA activity, or at least not at a level that exceeded the process of NH 4 + assimilation. Therefore, the major NO 3 -reducing pathway under the COD/NO 3 --N ratio of 4/1 was assumed to be denitrification because depleted inorganic nitrogen was most likely converted to nitrogenous gases. The existence of DNRA in this system was subsequently verified with stable-isotope tracers (see below).
In the high C/NO 3 -SBR, the COD/NO 3 --N ratio was started at 8/1 and the system was observed in order to establish whether it has the ability to sustain the growth of DNRA microorganisms. After 4 d of operation, the NO 3level was depleted (<0.2 mg N L -1 ). This rapid consumption of NO 3was likely to be due to the higher glucose concentration in the high C/NO 3reactor than in the low C/NO 3reactor, which, in turn, required electron acceptors (NO 3and NO 2 -) to complete the reactions catalyzed by the microorganisms. Therefore, NO 3became a limiting substrate in this system, whereas NO 2was undetectable throughout the experiment, suggesting the rapid conversion of NO 2to other end products. The amount of NH 4 + in the high C/NO 3 -SBR was increased to a markedly higher level than the supplied concentration, possibly accounting for 30.1-64.4% and 68.6-100% of the NO 3conversion during days 20-37 and 38-54, respectively. The active formation of NH 4 + most likely indicates the occurrence of DNRA in this high C/NO 3system. Judging from the portion of NO 3converted to NH 4 + during days 20-37, DNRA microorganisms and denitrifiers appeared to have a relatively similar share in NO 3reduction. However, after day 37, DNRA microorganisms appeared to be the principal NO 3reducers in the reactor, with a minor fraction of reduced NO 3being released as gases (<14.5% during peak NH 4 + -forming activity). The diverged nitrogen profiles observed in the two SBRs indicated that the difference in the ratios selected for the low and high C/NO 3reactors induced partitioning in the pathways of microbial NO 3reduction. Since the COD/NO 3 --N of 8/1 already supported the growth and activity of DNRA microorganisms, the ratio was not increased from this value. COD/NO 3 --N of 7.7 was successfully used to enrich a DNRA community in an acetate-fed continuous culture (36). This range of COD/NO 3 --N values appears to be sufficiently high to promote the reduction of NO 3to NH 4 + . However, other parameters, such as the type of organic carbon and its fermentability, also need to be considered as additional factors that affect the COD/NO 3 --N threshold partitioning the pathways of NO 3reduction.
With respect to the cell biomass grown in the two SBRs, the MLSS in the high C/NO 3system was greater (3,008±996 mg L -1 ) than that observed in the low C/NO 3system (2,740±272 mg L -1 ). Since the same amount of NO 3was supplied to both reactors, the higher cell mass in the high C/NO 3system may be due to the higher mole of organic carbon required per mole of NO 3for DNRA than for denitrification (32), thereby providing a greater carbon and energy source for microbial growth. It may also have been caused by the growth of other microorganisms apart from NO 3reducers because an organic carbon concentration above the stoichiometric requirement for NO 3 -reduction is expected to support other anaerobic heterotrophs, such as fermentative bacteria and SO 4 2reducers. Sulfide (S 2-) produced from SO 4 2reduction may sustain DNRA microorganisms by acting as an electron donor (6). However, the effects of S 2in the high C/NO 3reactor may be small considering the amount of glucose added, with SO 4 2being supplied in the media at 24.6±0.66 mg S L -1 . Therefore, if all the SO 4 2supplied was reduced to S 2-, 1.35% (2.69±0.07 mg N L -1 ) of the added NO 3would be used by S 2--driven DNRA (see Supplementary for details), and, hence, its effects on this process were expected to be negligible relative to those on the organic carbon-driven reaction.

Potential activity of DNRA in sludge from low and high C/NO 3 -SBRs
Batch incubations with low C/NO 3sludge revealed that 15 NO 3and 15 NO 2were mostly reduced to end products that did not remain in the liquid phase ( Fig. 2A, B), and, thus, they were assumed to be converted into nitrogenous gases. This was mostly attributed to the activity of denitrifiers, which produce nitric oxide (NO), nitrous oxide (N 2 O), and dinitrogen (N 2 ) gases from the reduction of NO 3and NO 2 -. However, a slight increase in 15 NH 4 + was measured with the 15 NO 3and 15 NO 2tracers, although at small concentrations only (0.19 and 0.23 mg N L -1 , respectively). This weak activity indicated the presence of a small amount of DNRA microorganisms in the low C/NO 3community. 15 NH 4 + and 15 NO 3 -/ 15 NO 2assimilation in all the incubations was expected to be suppressed due to the presence of a high concentration of 14 NH 4 + (100 mg N L -1 ). Therefore, this process was not taken into account for the balance of nitrogen conversion, and all of the 15 NH 4 + produced was assumed to be left in the media.
Regarding high C/NO 3sludge, a marked increase in the level of 15 NH 4 + was observed when 15 NO 3or 15 NO 2was added (Fig. 2C, D). Therefore, DNRA activity was confirmed by the evidence of NH 4 + production from the reduction of NO 3and NO 2 -. The portion of consumed 15 NO 3 -/ 15 NO 2not detected in the liquid phase was assumed to be converted to gaseous products. The addition of 15 NO 2instead of 15 NO 3did not significantly change the activity of DNRA microorganisms in the low and high C/NO 3sludge incubations. Hence, NO 2does not appear to have had a different effect on the nitrogen conversion pathways from that observed for NO 3 -, at least for the samples and conditions tested in this experiment.
Net changes in 15 N-nitrogen for all incubations (calculated based on the initial and final concentrations of 15 NO 3 -, 15 NO 2 -, and 15 NH 4 + ) are shown in Fig. 3. The main products of the reduction of 15 NO 3and 15 NO 2in low C/NO 3sludge were in gaseous forms (Fig. 3A), and, thus, denitrifiers were assumed to be the main contributors to nitrogen conversion in these samples. The 15 NH 4 + produced from DNRA microorganisms inhabiting the low C/NO 3 -SBR was 1.4 and 1.3% of the amount of 15 NO 3and 15 NO 2consumed, respectively, indicating a small role for these microorganisms in the denitrifierdominated community. In contrast, the main product in high C/NO 3sludge was 15 NH 4 + in the 15 NO 3and 15 NO 2incubations (Fig. 3B). The proportions of nitrogenous gases produced in the high C/NO 3incubations (21.9 and 34.5% of consumed 15 NO 3and 15 NO 2 -, respectively) were lower than those converted to 15 NH 4 + . However, these gases may not come exclusively from the activity of denitrification. Previous studies revealed that DNRA also released NO and N 2 O (30, 39, 42); however, the exact production pathways remain unclear. Since high C/NO 3sludge was predominated by DNRA activity, the nitrogenous gases produced may come from denitrification, DNRA, or both.
At the end of the incubations, the net 15 NH 4 + generated in the high C/NO 3sludge was 46.6 and 65.5% of the consumed 15 NO 3and 15 NO 2 -, respectively. The extent of 15 NH 4 + formed here suggested the competitive potential of DNRA microorganisms in activated sludge when under a continuous supply of high C/NO 3loading. Additionally, NO 3and NO 2had the ability to induce DNRA in a similar manner; however, these two nitrogen species were previously reported to exert different stimulating effects on the pathway in Escherichia coli (26,41). In conclusion, the results from stable isotope tracers confirmed that NH 4 + detected in the high C/NO 3 -SBR was mainly the product of DNRA, and that the addition of glucose at a COD/NO 3 --N ratio of 8/1 may be selected for the growth and activity of DNRA microorganisms; however, denitrifiers may still co-exist to a certain level.

Microbial communities of low and high C/NO 3 -SBRs
An Illumina MiSeq 16S rRNA analysis revealed that from all the samples collected, 98.4-99.8% of sequences were identified as Bacteria and ≤0.01% as Archaea, while it was not possible to designate 0.2-1.6% of the reads. The OTUs assigned from the sequence reads were classified into 17 main classes (Fig. 4) belonging to seven phyla, including Proteobacteria, Firmicutes, Bacteroidetes, Chloroflexi, and the candidate divisions GN02, OD1, and TM7. Microbial  richness, as estimated by the Chao1 index, and microbial diversity, as calculated by the Shannon index, revealed a significant difference (P<0.05) between the low and high C/ NO 3microbial populations (see Table S2). Lower richness and biodiversity were observed in high C/NO 3samples, whereas low C/NO 3populations were more diverse with higher community richness. Therefore, a higher C/NO 3ratio appeared to allow certain bacteria to thrive with elevated proportions, while the lower C/NO 3ratio nurtured diverse bacteria that, among the dominating ones, had relatively fair shares of the microbial composition, as indicated by the relative abundance of the major OTUs in the system (see Fig.  S3). The relationships between the inoculum and the low and high C/NO 3microbial communities were revealed in the PCoA plot (Fig. 5) using a beta-diversity measure of each sample. The PCoA plot showed distinct clustering within the low and high C/NO 3samples, with the distance between the beta-diversities of the two reactors' populations conveying the differences in microbial compositions among these two ecosystems. The average relative abundance of the major taxa identified (Fig. 6) indicated that the microbial communities of the low and high C/NO 3 -SBRs were distinct from each other with rarely shared OTUs among them, which reflects partitioning in the NO 3 --reducing activities observed in the two systems.
Among the top OTUs identified in the high C/NO 3reactor, those with the potential of being DNRA microorganisms included those closely related to the genus Sulfurospirillum, the family Lachnospiraceae, the proposed genus PSB-M-3, and the species Geobacter lovleyi. These taxa were present at an average abundance of ≤0.4% (from the five sampling days) in the low C/NO 3system, which emphasized their roles in shaping the NH 4 + -producing community in the high C/NO 3 -SBR. The OTU affiliated with Sulfurospirillum (OTU105876) was one of the OTUs detected with the highest average abundance in the high C/NO 3reactor (16.5%). The genus Sulfurospirillum (belonging to the class Epsilonproteobacteria) is known to exhibit versatile metabolism ranging from SO 4 2reduction, S 2oxidation, and DNRA (18,25,31). However, the amounts of NO 3and organic carbon available in the high C/NO 3 -SBR were expected to promote NO 3reduction to NH 4 + in these microorganisms, while the influence of sulfur compounds was expected to be minimal because the level of sulfur-driven NO 3reduction in the reactor was small. Three OTUs related to Sulfurospirillum were found at a high abundance in the high C/NO 3system, but were present at ≤0.02% of the total community in the low C/NO 3reactor, which indicated that the high C/NO 3condition was highly selective for the growth of these populations. The family Lachnospiraceae (in the class Clostridia of the phylum Firmicutes), the other highest average abundant OTU (16.5%) (OTU567875), is not known to contain DNRA microorganisms, but was previously found in anoxic microcosms performing DNRA (23). Apart from the highest OTU, another top OTU in the high C/ NO 3 -SBR was related to PSB-M-3 (OTU355578) in the family Erysipelotrichaceae of the phylum Firmicutes, with an average abundance of 8.9%. The high proportions of the taxa affiliated with Lachnospiraceae and PSB-M-3 suggest their essential role in the high C/NO 3ecosystem; however, further investigations are needed in order to elucidate their actual ecological function in NO 3 --reducing communities.  The OTU assigned to the species G. lovleyi (OTU586655), a member of the class Deltaproteobacteria, was detected with an average abundance of 7.2% in the high C/NO 3reactor. G. lovleyi is known to conduct DNRA (33) and taxa related to this species have also been found to dominate DNRA enrichment cultures fed with acetate (36)(37)(38). Previous findings and the present results indicate the common occurrence of these DNRA microorganisms when the conditions favor NO 3reduction to NH 4 + . Beside those with DNRA potential, the remaining major OTUs were found to be related to bacteria with fermentative abilities, including the OTUs assigned to the class BD1-5 (OTU65013), genus Tolumonas (OTU4439030), and genus Paludibacter (OTU72348). There is no direct evidence to show that these taxa perform NO 3reduction (9,16,35), and, thus, they most likely acted as contributors supplying simpler fermented products to the rest of the community.
In the low C/NO 3 -SBR, the OTU detected with the highest average abundance (6.9%) was assigned to the family Comamonadaceae (OTU143252) in the class Betaproteobacteria. Several members of this family have been recognized as denitrifying bacteria, such as Comamonas and Acidovorax (43). Other OTUs affiliated with known denitrifiers include those related to Rhodobacter (OTU321409), Dechloromonas (OTU536847), and Flavobacterium (OTU188193) (5,8,11). OTUs related to the families Chitinophagaceae (OTU181810 and OTU4083690) and Oxalobacteraceae (OTU782472) were also identified in the low C/NO 3community, in which some members are known to reduce NO 3or perform complete denitrification (20). The detection of these major OTUs in the low C/NO 3reactor was in accordance with the denitrifying activity observed in the system. Additionally, certain denitrifiers, such as Shewanella loihica and several nirKcontaining bacteria, have been found to contain the DNRA pathway in their genomes (17,44). Therefore, the weak activity of DNRA detected in the low C/NO 3 -SBR may be attributed to one of these denitrifiers because no other known DNRA microorganism was identified. Between the two NO 3 --reducing ecosystems, smaller changes in the microbial community were observed for the low C/NO 3reactor (in which the enriched populations were still denitrifiers), while the high C/NO 3condition was enriched for a markedly different microbial composition from the inoculum (see Fig. S4). This result demonstrated that the selected C/NO 3ratios applied to the two SBRs were already critical for the partition of microbial populations to those with different functional abilities.

Conclusion
In the high C/NO 3 -SBR, DNRA was observed by the marked increase in the NH 4 + concentration. Nitrogen conversion, monitored via 15 NO 3and 15 NO 2tracers, revealed that DNRA microorganisms were the major contributors to NO 3 -/ NO 2reduction under a high COD/NO 3 --N ratio of 8/1, at which NO 3and NO 2induce DNRA in a similar manner. With a low COD/NO 3 --N ratio of 4/1, denitrifiers were the major NO 3 -/NO 2reducers. Additionally, the high C/NO 3conditions enriched for microbial populations markedly differed from those in the low C/NO 3reactor and the inoculum. These populations were comprised of OTUs closely related to known DNRA microorganisms (Sulfurospirillum and G. lovleyi) as well as fermentative bacteria and those that may be capable of both functions. These bacteria were rarely present in the low C/NO 3 -system, which harbored a community of denitrifiers. Therefore, the C/NO 3ratios applied to each SBR provided an environment that partitioned the pathways of NO 3reduction to either nitrogenous gases or NH 4 + as the end product, which was the result of the distinctive microbial composition found in each SBR system.