Abstract
Simultaneous nitrification and denitrification under low dissolved oxygen conditions is an energy-saving modification of the activated sludge process to achieve efficient nitrogen removal. Geographically distinct full-scale treatment plants are excellent platforms to address the links of microbial community with operating parameters. Mixed liquor samples were collected from a sequencing batch reactor plant, oxidation ditch plant, and step-feed activated sludge plant. Next-Generation Sequencing of the samples showed that the microbial communities were similar at the phylum level among the plants, being dominated by Proteobacteria. Microbial composition of functional groups was similar between the react fill and react phases of the sequencing batch reactors, among four sequencing batch reactors, and among four oxidation ditches. Nitrospira was the only identified genus of autotropic nitrifying bacteria with a relative abundance of 2.2–2.5% in the oxidation ditches and 0.4–0.7% at the other plants. Heterotrophic nitrifying–aerobic denitrifying bacteria were dominated by Dechloromonas with a relative abundance of 0.4–1.0%. Microbial community composition and nitrogen removal mechanisms were related to overall level and local zonation of dissolved oxygen, mixed liquor suspended solids concentration, nitrogen and organic loadings, and solids retention time. Low dissolved oxygen and low organic and nitrogen loadings favored growth of Nitrospira.
INTRODUCTION
The regulations related to nitrogen concentrations in wastewater discharge are becoming more stringent. The wastewater treatment industry constantly searches for more cost-effective nitrogen removal technologies. Biological nitrogen removal is usually achieved via autotrophic nitrification (NH4+ → NO2− → NO3−) under aerobic conditions, and heterotrophic denitrification (NO3− → NO2− → NO → N2O → N2) under anoxic conditions in multiple stages (Metcalf & Eddy/AECOM 2014). Recently, simultaneous nitrification and denitrification (SND) in single reactors is gaining popularity and has been adopted by wastewater treatment plants (WWTPs) to meet discharge limits on nitrogen and biochemical/chemical oxygen demand (BOD/COD) (Daigger & Littleton 2014). SND is a biological nitrogen removal process where nitrification and denitrification take place concurrently in single reactors without defined aerobic and anoxic zones. Activated sludge (AS) systems conventionally set operating dissolved oxygen (DO) concentration at 1–2 mg/L without intentional nitrogen removal and at 3–4 mg/L with enhanced nitrogen removal (Metcalf & Eddy/AECOM 2014). SND is generally achieved in AS under low DO conditions. At a low DO concentration (≤2 mg/L) in the bulk water column, nitrification or nitritation may occur in the outer layer of a microbial floc while denitrification and anammox take place in the inner layer of the floc. In addition to the combination of autotrophic nitrification and heterotrophic denitrification, the mechanisms that are responsible for SND can include the couplings of nitritation–denitritation (NH4+ → NO2− → NO → N2O → N2), nitritation–anammox (NH4+ + NO2− → N2), nitritation–anammox–denitrification, and heterotrophic nitrification–aerobic denitrification (Wen et al. 2013; Metcalf & Eddy/AECOM 2014; Cydzik-Kwiatkowska & Zielińska 2016; Keene et al. 2017; Li & Tao 2017).
In addition to ammonia-oxidizing bacteria, ammonia-oxidizing archaea that also catalyze aerobic autotrophic oxidation of ammonia to nitrite under low DO conditions have been detected in AS (Zhang et al. 2011). Recently, complete ammonia-oxidizing (comammox) bacteria, which can convert ammonia to nitrate in one step, have been discovered (Daims et al. 2015; Van Kessel et al. 2015) and found in low DO AS processes (How et al. 2018; Roots et al. 2019). The diverse groups of microorganisms and treatment mechanisms of low DO SND make it complicated to determine design parameters and optimize operational parameters. Microbial community composition and diversity need to be further explored in full-scale low DO SND systems.
Aeration typically accounts for over 50% of total energy usage in an AS process plant (Metcalf & Eddy/AECOM 2014). Maintaining a lower DO concentration means a larger driving force for oxygen dissolution or saving aeration energy. Extended aeration with low DO is favorable for developing SND (Daigger & Littleton 2014; Keene et al. 2017). Low DO SND has been successfully implemented at full-scale WWTPs without sacrificing treatment performance (Daigger & Littleton 2014; Wen et al. 2015; Keene et al. 2017). Compared with pilot-scale and laboratory bioreactors, a full-scale bioreactor harbors a more diverse microbial community (Xia et al. 2018). WWTPs are fertile testing grounds for a range of fundamental ecological questions (Xia et al. 2018).
Next-Generation Sequencing has been employed to understand microbial community dynamics in various full-scale bioreactors, such as oxidation ditches and conventional AS processes (Zhang et al. 2011; Hu et al. 2012; Wang et al. 2012; Ju et al. 2014; Saunders et al. 2016; Xia et al. 2018). However, earlier studies on microbial community with multiple WWTPs collected samples only from one point each plant, without looking into the variations between treatment trains and time sequences (Zhang et al. 2011; Hu et al. 2012; Wang et al. 2012; Ju et al. 2014; Jo et al. 2016; Saunders et al. 2016; Wang et al. 2016). It remains challenging to find tight links between microbial community and AS functions (Xia et al. 2018). More data on gene amplicon sequencing of AS samples from full-scale WWTPs under different operating conditions are needed to troubleshoot and optimize low DO SND systems.
The objectives of this study were to: (1) investigate the variations in microbial community composition and diversity of AS along a time sequence of sequencing batch reactors (SBRs) and among the parallel-operated SBRs and oxidation ditches as well as the differences among three differently configured low DO SND plants, using Next-Generation Sequencing; and (2) explore the links of microbial community composition and diversity with operating parameters of the bioreactors. The similarity in microbial community composition and assembly of AS decreases as geographic distance between WWTPs increases (Xia et al. 2018). The combination of a SBR plant in Virginia, an oxidation ditch plant in New Jersey, and a step-feed AS plant in Pennsylvania in the USA represents a large spatial scale for reliable identification of microbial community composition and diversity. Meanwhile, sampling along the time sequence of four SBRs and also simultaneous sampling in the four SBRs and four oxidation ditches captured practical, unique environments to identify the relationships of microbial community composition with operating parameters of individual reactors.
MATERIALS AND METHODS
Description of full-scale WWTPs
Three geographically distinct full-scale municipal WWTPs were selected because of their different configurations for biological nitrogen removal (Table 1). Moreover, these plants have different treatment capacities and operating conditions. The Dale City WWTP in Virginia has four SBRs with relatively low mixed liquor suspended solids (MLSS) concentrations. The SBRs are operated in a sequence of mix fill–react fill–react–settle–decant. Each cycle lasts for 6 h. The EDC WWTP in New Jersey has four oxidation ditch trains, each of which contains an anaerobic tank, an oxidation ditch, and a final clarifier. Each oxidation ditch has one surface aerator. The Scranton WWTP in Pennsylvania has a treatment train containing a primary clarifier, an anoxic tank, a step-feed aeration tank, and a final clarifier. The aeration tank is split into ten zones and every four zones received 25% of the inflow.
Operating conditions of three WWTPs in this studya
Parameter . | Dale City . | EDC . | Scranton . |
---|---|---|---|
Design flowb, m3/d | 17,388 | 7,938 | 60,480 |
Average operating flow, m3/d | 11,340 | 4,536 | 34,398 |
Treatment technology | SBR | Oxidation ditch | Step-feed AS |
Operating DO, mg/L | 0.5–1.5 | 0.2–1.3 | 0.5–2.0 |
SRTb, d | 17 | 15 | 12 |
MLSS, mg/Lc | 2,200 | 2,500 | 4,000 |
Influent BOD, mg/L | 214 | 203 | 121 |
Influent BOD Loading, kg/d | 2,429 | 919 | 4,169 |
Influent TSS, mg/L | 196 | 295 | 127 |
Influent TKN, mg/L | 51 | 40b | 26 |
Influent TKN loading, kg/d | 582 | 181 | 901 |
Influent NH3-N, mg/L | 31 | 25b | 23 |
Effluent BOD, mg/L | NDc | 2.4 | 3.7 |
Effluent TSS, mg/L | ND | 2.5 | 11 |
Effluent TKN, mg/L | 0.9 | N/A | 1.8 |
Effluent NH3-N, mg/L | 0.04 | 0.4 | 0.1 |
Effluent NO3−-N, mg/L | 1.4 | 5.0 | 2.5 |
Temperature, °C | 15 | 16 | 10 |
Date of mixed liquor sampling | 2/22/2017 | 11/29/2016 | 1/18/2017 |
Parameter . | Dale City . | EDC . | Scranton . |
---|---|---|---|
Design flowb, m3/d | 17,388 | 7,938 | 60,480 |
Average operating flow, m3/d | 11,340 | 4,536 | 34,398 |
Treatment technology | SBR | Oxidation ditch | Step-feed AS |
Operating DO, mg/L | 0.5–1.5 | 0.2–1.3 | 0.5–2.0 |
SRTb, d | 17 | 15 | 12 |
MLSS, mg/Lc | 2,200 | 2,500 | 4,000 |
Influent BOD, mg/L | 214 | 203 | 121 |
Influent BOD Loading, kg/d | 2,429 | 919 | 4,169 |
Influent TSS, mg/L | 196 | 295 | 127 |
Influent TKN, mg/L | 51 | 40b | 26 |
Influent TKN loading, kg/d | 582 | 181 | 901 |
Influent NH3-N, mg/L | 31 | 25b | 23 |
Effluent BOD, mg/L | NDc | 2.4 | 3.7 |
Effluent TSS, mg/L | ND | 2.5 | 11 |
Effluent TKN, mg/L | 0.9 | N/A | 1.8 |
Effluent NH3-N, mg/L | 0.04 | 0.4 | 0.1 |
Effluent NO3−-N, mg/L | 1.4 | 5.0 | 2.5 |
Temperature, °C | 15 | 16 | 10 |
Date of mixed liquor sampling | 2/22/2017 | 11/29/2016 | 1/18/2017 |
aMonthly averages of influent and effluent for Dale City (September 2015), EDC (July 2016), and Scranton (October 2015). DO: dissolved oxygen; BOD: biochemical oxygen demand; TSS: total suspended solids; TKN: total Kjeldahl nitrogen; ND: not detected.
bOne-time measurements during sampling.
cDesign value.
All the WWTPs were operated at low DO concentrations for SND (Table 1). The SBRs had a DO setpoint of 1.5 mg/L for the react phase and DO reached 0.5 mg/L soon after the beginning of each react fill phase. DO was measured near the aerator, at the middle (0.75 mg/L), and at the end of each oxidation ditch while sampling mixed liquor in the middle. DO concentration varied between 0.5 and 1.0 mg/L from the first to second feeding point and was up to 2.0 mg/L in the last aerated zone of the Scranton aeration tank.
Collection and processing of biomass samples
Table 2 summarizes the samples collected from each plant and the codes used to denote each sample. All the samples were shipped overnight on ice to the American Water laboratory in Delran, USA, and 45 mL of each sample was transferred to a 50-mL sterile tube and centrifuged at 2,500 × g for 5 min at room temperature. After removing the supernatant, the concentrated pellets (approximately 200 mg) were used for DNA extraction.
Samples collected and analyzed from the three WWTPs
WWTP . | Sample collected . | Sample code . |
---|---|---|
Dale City | SBR tank 1 during React Fill phase | DCBio1RF/DCBio1RFdupa |
SBR tank 2 during React Fill phase | DCBio2RF | |
SBR tank 2 during React phase | DCBio2R | |
SBR tank 3 during React Fill phase | DCBio3RF | |
SBR tank 3 during React phase | DCBio3R | |
SBR tank 4 during React Fill phase | DCBio4RF | |
SBR tank 4 during React phase | DCBio4R | |
EDC | Oxidation ditch 1 | EDCBio1/EDCBio1dupa |
Oxidation ditch 2 | EDCBio2 | |
Oxidation ditch 3 | EDCBio3 | |
Oxidation ditch 4 | EDCBio4 | |
Scranton | Aeration tank | Scranton/Scrantondupa |
WWTP . | Sample collected . | Sample code . |
---|---|---|
Dale City | SBR tank 1 during React Fill phase | DCBio1RF/DCBio1RFdupa |
SBR tank 2 during React Fill phase | DCBio2RF | |
SBR tank 2 during React phase | DCBio2R | |
SBR tank 3 during React Fill phase | DCBio3RF | |
SBR tank 3 during React phase | DCBio3R | |
SBR tank 4 during React Fill phase | DCBio4RF | |
SBR tank 4 during React phase | DCBio4R | |
EDC | Oxidation ditch 1 | EDCBio1/EDCBio1dupa |
Oxidation ditch 2 | EDCBio2 | |
Oxidation ditch 3 | EDCBio3 | |
Oxidation ditch 4 | EDCBio4 | |
Scranton | Aeration tank | Scranton/Scrantondupa |
adup indicates that the sample was sequenced in duplicate.
DNA extraction, polymerase chain reaction (PCR) amplification and 16S rRNA gene amplicon sequencing
Upon sample homogenization, DNA was extracted in duplicate with a FastDNA Spin Kit for soil (MP Biomedicals, USA), following the manufacturer's protocol. An optimized bead beating step was incorporated to ensure that all microbial groups were represented. The bead beating was performed using SuperFastPrep-2 (MP Biomedicals) at a speed of 18,000–23,000 rpm for 4 × 40 s. The samples were kept on ice for 1–2 min after each 40-s beating. Extract DNA concentration was analyzed using the Qubit 3.0 dsDNA HS assay kit (Invitrogen, USA) and the duplicate sample with a higher DNA concentration was used for further processing.
A procedure modified from Caporaso et al. (2011) was performed for 16S rRNA gene amplicon sequencing, targeting the V3-4 variable regions. Amplicon library PCR was performed using 2.5 μL of 5 ng/μL extracted DNA as a template and SYBR Green FastMix (Quantabio, USA). The V3-4 primers used in this study were 314F (3′-CCTACGGGNGGCWGCAG-5′) and 805R (3′-GACTACHVGGGTATCTAATCC-5′) (Klindworth et al. 2013). The thermocycler settings for the V3-4 amplicon PCR were: initial denaturation at 95 °C for 3 min, 25 cycles of 95 °C for 30 s, 55 °C for 30 s, 72 °C for 30 s, and final elongation at 72 °C for 5 min. The amplicon libraries were purified using the Agencourt AMPure XP bead protocol (Beckmann Coulter, Ireland). The second PCR was Index PCR, which was conducted to attach dual indices and Illumina sequencing adapters using the Nextera XT index kit (Illumina, USA) in order to pool all samples together. The thermocycler settings for the Index PCR were: initial denaturation at 95 °C for 3 min, eight cycles of 95 °C for 30 s, 55 °C for 30 s, 72 °C for 30 s, and final elongation at 72 °C for 5 min. The amplicon libraries with indices were also purified using the Agencourt AMPure XP bead protocol (Beckmann Coulter, Ireland). After the Amplicon and Index PCR amplification, DNA size was examined with 1% (w/v) agarose gel electrophoresis for quality assurance.
Final library concentration was measured with the Qubit 3.0 dsDNA HS assay kit (Invitrogen, USA). Based on the library concentrations and amplicon sizes, the samples were pooled in equimolar concentrations and diluted to 4 nM. The library pool was then sequenced on an Illumina MiSeq platform (Illumina, USA) using a MiSeq Reagent kit v3 (2 × 300 paired-end) including 45% PhiX control library (Illumina, USA) spike-in and the final library loading concentration of 6 pM. One sample from a reactor at each plant was selected to conduct DNA sequencing in duplicate (Table 2) for quality control.
Bioinformatic analysis
Demultiplexing was performed with the Illumina's MiSeq reporter software to generate FASTQ files for forward and reverse reads after the paired-end sequencing was completed. The sequencing data were analyzed according to the standard Quantitative Insights Into Microbial Ecology (QIIME v1.9.1) protocol (Caporaso et al. 2010). The high-quality sequences were aligned and assigned against the Greengenes database 13_08 using the PyNAST program with 97% identity, and singletons were removed. The lowest number of operational taxonomic unit (OTUs) was 35,547 among the 15 samples, and, therefore, all samples were rarefied to 35,000 sequences for downstream analysis.
The QIIME program was used to calculate Simpson Diversity Index, Shannon Diversity Index, Chao1 richness estimator, phylogenetic diversity, and observed OTUs (Hu et al. 2012; Cydzik-Kwiatkowska & Zielińska 2016). Clustering and principal coordinate analysis (PCoA) were conducted using QIIME on all the 15 samples to show dissimilarity of the microbial communities. Single-factor analysis of variance was conducted using the Excel Data Analysis package to test whether there were significant differences (P ≤ 0.05) in microbial community composition between operating phases, among bioreactors, and among the WWTPs. In addition, canonical correspondence analysis (CCA) was conducted to investigate the relationships of the microbial communities with operating conditions, influent organic characteristics, and influent nitrogen characteristics of the three WWTPs.
RESULTS AND DISCUSSION
The Dale City and Scranton WWTPs had total nitrogen discharge limits and the EDC Plant had an effluent discharge limit for ammonia only. The effluent concentrations of total nitrogen at all three WWTPs (Table 1) were less than 10 mg/L, meeting their discharge limits. All the plants also met their discharge limits on BOD and TSS (Table 1).
Similarity in microbial community composition
Only one phylum of archaea (Euryarchaeota) was detected with an abundance of less than 0.02% in all the samples from the three WWTPs. Between 1.4% and 4.5% of the sequences were not assigned to any phylum in the Greengenes database. There were 31–38 phyla of bacteria identified in all the mixed liquor samples. As shown in Figure 1, Proteobacteria (39–49%) was the most dominant phylum followed by Bacteroidetes (16–25%), and Chloroflexi (6.7–14%) at all the WWTPs. Proteobacteria and Bacteroidetes have been reported as being the two most abundant phyla at other full-scale oxidation ditch and conventional AS WWTPs (Hu et al. 2012; Wang et al. 2012; Cydzik-Kwiatkowska & Zielińska 2016; Xia et al. 2018).
Phylum classification of mixed liquor samples from (a) four SBRs at Dale City Plant, (b) four oxidation ditches at EDC Plant, and (c) and an aeration tank at Scranton Plant. ‘Others’ refer to those with individual abundances below 0.5%.
Duplicate sequencing produced almost the same abundance distribution of the abundant phyla (P ≥ 0.998) for the three samples from the three WWTPs (Figure 1). The react fill and react phases of SBRs 2–4 at the Dale City WWTP were statistically similar in abundance distribution of the 11 abundant phyla (P > 0.993), suggesting similar treatment mechanisms in the react fill and react phases. There were no differences in relative abundance distribution of the abundant phyla among the four SBRs at the Dale City WWTP (P = 1.00) and the four oxidation ditches at the EDC WWTP (P = 1.00).
Dissimilarity among the three WWTPs became remarkable at the class level. α-Proteobacteria and Saprospiraceae were the predominant class and family, respectively, in all the samples from the Dale City SBRs. δ-Proteobacteria (13–16%) and β-Proteobacteria (13–15%) were the top two classes in all the samples from the EDC oxidation ditches. Saprospirae (21%) and Saprospiraceae (17%) were the predominant class and family, respectively, in the Scranton AS.
The three WWTPs had similar numbers of genera detected in individual samples, but the majority were unnamed. There were 435–470 genera detected in the samples from the SBRs, including only 158–187 genera with known names. There were 431–460 genera in the samples from the oxidation ditches, including 162–172 genera with known names. There were 373–383 genera detected in the sample from the step-feed aeration tank, including only 153–154 genera with known names. As shown in Figure 2, only three of the nine to ten abundant genera were named. These unnamed sequences probably represented uncharacterized microorganisms or novel functional gene fragments that were not available in the Greengenes database, which can be further analyzed later as the database is updated.
Abundant genera of mixed liquor samples from (a) SBRs at Dale City Plant, (b) oxidation ditches at EDC Plant, and (c) step-feed aeration tank at Scranton Plant ‘Abundant genera’ include any genus with a relative abundance greater than 2%.
The phylum Chloroflexi contains species of filamentous bacteria (Cydzik-Kwiatkowska & Zielińska 2016; Xia et al. 2018). Filamentous bacteria facilitate formation of microbial flocs, but excessive growth causes sludge bulking. The Dale City SBRs had slightly higher relative abundance of Chloroflexi (7.2–14%) than the other two plants (6.7–11%). In particular, the genus Kouleothrix in Chloroflexi had higher abundances in the SBRs than the step-feed AS and oxidation ditches (Figure 2). It was known that Dale City had filamentous issues that could be due to Kouleothrix.
Similarity in microbial diversity
There were similar Simpson Indexes, Shannon Indexes, Chao1 estimators, phylogenetic diversity values, and OTUs between the two phases of the SBRs, among the bioreactors at each plant, and between the SBR and oxidation ditch plants (Table 3), indicating similar degree of genus concentration, diversity, and richness. The only exception was the relatively lower microbial diversity of the DCBio3R sample, which cannot be explained without further evidence. The Scranton aeration tank had a relatively lower Shannon Diversity Index, Chao1 richness estimator, phylogenetic diversity, and OTUs, which could be attributed to its relatively higher MLSS concentration and shorter SRT (Table 1). Shannon diversity and OTU values of the low DO oxidation ditches and aeration tank in this study were also higher than those normally operated oxidation ditches and conventional AS systems (Hu et al. 2012; Wang et al. 2012). Simpson diversity in the EDC oxidation ditches and Scranton aeration tank was similar to those reported for normally operated oxidation ditches and conventional AS systems (Hu et al. 2012).
Summary of alpha-diversity results for three WWTPs
Sample . | Simpson Index . | Shannon Index . | Chao1 estimator . | Phylogenetic diversity . | Observed OTUs . |
---|---|---|---|---|---|
Dale City | |||||
DCBio1RF | 0.989 | 9.18 | 7,668 | 85.7 | 4,090 |
DCBio2RF | 0.987 | 9.16 | 7,899 | 88.3 | 4,018 |
DCBio2R | 0.986 | 9.11 | 7,826 | 89.1 | 4,153 |
DCBio3RF | 0.992 | 9.25 | 7,907 | 92.6 | 4,111 |
DCBio3R | 0.988 | 8.92 | 6,743 | 85.3 | 3,745 |
DCBio4RF | 0.988 | 9.11 | 7,723 | 88.5 | 4,172 |
DCBio4R | 0.989 | 9.17 | 7,828 | 87.7 | 4,090 |
EDC | |||||
EDCBio1 | 0.993 | 9.61 | 7,056 | 94.1 | 4,000 |
EDCBio2 | 0.995 | 9.63 | 7,092 | 92.4 | 3,935 |
EDCBio3 | 0.995 | 9.71 | 8,026 | 93.2 | 4,210 |
EDCBio4 | 0.995 | 9.65 | 7,585 | 91.2 | 4,123 |
Scranton | 0.988 | 8.97 | 5,839 | 77.8 | 3,454 |
Sample . | Simpson Index . | Shannon Index . | Chao1 estimator . | Phylogenetic diversity . | Observed OTUs . |
---|---|---|---|---|---|
Dale City | |||||
DCBio1RF | 0.989 | 9.18 | 7,668 | 85.7 | 4,090 |
DCBio2RF | 0.987 | 9.16 | 7,899 | 88.3 | 4,018 |
DCBio2R | 0.986 | 9.11 | 7,826 | 89.1 | 4,153 |
DCBio3RF | 0.992 | 9.25 | 7,907 | 92.6 | 4,111 |
DCBio3R | 0.988 | 8.92 | 6,743 | 85.3 | 3,745 |
DCBio4RF | 0.988 | 9.11 | 7,723 | 88.5 | 4,172 |
DCBio4R | 0.989 | 9.17 | 7,828 | 87.7 | 4,090 |
EDC | |||||
EDCBio1 | 0.993 | 9.61 | 7,056 | 94.1 | 4,000 |
EDCBio2 | 0.995 | 9.63 | 7,092 | 92.4 | 3,935 |
EDCBio3 | 0.995 | 9.71 | 8,026 | 93.2 | 4,210 |
EDCBio4 | 0.995 | 9.65 | 7,585 | 91.2 | 4,123 |
Scranton | 0.988 | 8.97 | 5,839 | 77.8 | 3,454 |
Microbial groups responsible for nitrogen removal
SND is often observed when pilot- and full-scale SBRs, oxidation ditches and conventional AS processes are operated at low DO concentrations and relatively long SRTs (Daigger & Littleton 2014; Wen et al. 2015; Keene et al. 2017). In this study, Nitrospira was identified as the only known genus for autotrophic nitrification at all three WWTPs (Figure 3). This genus can be involved in both ammonia oxidation to nitrite and nitrite oxidation to nitrate. No autotrophic bacteria that only perform ammonia oxidation were identified in the samples. However, it is still possible that ammonia-oxidizing bacteria were present at very low abundances compared to Nitrospira because the Next-Generation Sequencing samples were rarefied to 35,000 sequences and subsequently the existence of ammonia-oxidizing bacteria cannot be completely ruled out from the samples. The growth of Nitrospira in WWTPs is favored by low DO (Cydzik-Kwiatkowska & Zielińska 2016). Keene et al. (2017), How et al. (2018), and Roots et al. (2019) reported that Nitrospira were the dominant nitrifying bacteria in low DO (0.2–1.0 mg/L) bioreactors treating municipal wastewater. Nitrospira can include nitrite oxidizing bacteria and comammox bacteria. Nitrite oxidizing Nitrospira outcompeted nitrite oxidizing Nitrobacter under long-term low DO conditions (Liu & Wang 2013).
Relative abundances of nitrifiers (Nitrospira) and denitrifiers (the others) in the three WWTPs.
As shown in Figure 3, the relative abundances of Nitrospira in the oxidation ditches (2.2–2.5%) was much higher than that in the other two plants (0.4–0.7%), presumably because the oxidation ditches, having constant low DO zones, were operated at the lowest DO concentrations (Table 1), which could promote local growth of Nitrospira (How et al. 2018; Roots et al. 2019).
Denitrifying bacteria previously found in municipal wastewater treatment systems (Lu et al. 2014; Xia et al. 2018) were detected in 12 genera with a total relative abundance of 2.0–2.8% in the SBRs, 14 genera with a total abundance of 2.2–2.5% in the oxidation ditches, and 13 genera with a total abundance of 3.1% in the step-feed aeration tank (Figure 3). The relatively higher abundance of denitrifying bacteria in the Scranton aeration tank was probably due to the shortest SRT among the three plants because heterotrophic denitrifying bacteria generally grow faster than autotrophic nitrifying bacteria (Metcalf & Eddy/AECOM 2014).
Dechloromonas was the predominant genus of heterotrophic denitrifying bacteria, followed by Acinetobacter and Hyphomicrobium at all the plants (Figure 3). Heterotrophic nitrifying–denitrifying bacteria, including Acinetobacter, Agrobacterium, Azoarcus, Bacillus, Comamonas, Thauera, Pseudomonas, and Paracoccus (Li et al. 2015; Cydzik-Kwiatkowska & Zielińska 2016; Rout et al. 2017; Chen et al. 2019), were detected at total abundances of 0.6–0.9% in the SBRs, 0.7–0.9% in the oxidation ditches, and 0.4% in the step-feed AS process. The relatively higher abundance of denitrifying bacteria and lower heterotrophic nitrifying–denitrifying bacteria in the Scranton aeration tank could be due to its relatively higher operating DO level and high MLSS (Table 1). None of the anammox genera were detected in any of the samples, indicating the absence of anammox bacteria.
Overall, the Illumina Next-Generation Sequencing results indicated that the oxidation ditches removed nitrogen mainly through the conventional SND process, plus possibly simultaneous heterotrophic nitrification–aerobic denitrification; and that the SBRs and step-feed AS process were probably regulated mainly by simultaneous heterotrophic nitrification–aerobic denitrification.
Dissimilarity in microbial community composition among WWTPs
Beta-diversity expresses similarity between individual communities (Cydzik-Kwiatkowska & Zielińska 2016). It was assessed with PCoA and hierarchical clustering based on the unweighted pair group method with arithmetic mean (UPGMA). The samples from the same plant clustered closely and the duplicate samples clustered together (Figure 4), indicating reliable sequencing results. The three principal coordinates represented more than 92% of total variance (Figure 5). The clustering results were consistent with the PCoA results show that: (1) the microbial communities at the three treatment plants were site-specific and the bioreactors at each plant had similar microbial communities; (2) the largest dissimilarity in principal coordinate 2 (PC2) PC2 separated the Scranton aeration tank at the higher operating DO levels and highest MLSS concentration from the Dale City SBRs and EDC oxidation ditches at lower DO levels and MLSS concentrations; (3) the second dissimilarity in PC1 distinguished the Dale City SBRs that had cyclic aeration from the EDC oxidation ditches and the Scranton step-feed aeration tank that had constant zonation in DO concentration. Therefore, both overall DO level and the pattern of low DO distribution in AS systems can lead to differences in microbial community composition and diversity.
Hierarchical clustering of mixed liquor samples to interpret the distance matrix produced from beta-diversity analysis.
Linking microbial community composition to operating conditions
The CCA analysis (Figure 6) confirmed that DO and MLSS were the primary operating variables that led to the microbial community composition in the AS samples from the Scranton Plant. SRT and influent BOD and TKN concentrations had strong relationships with the microbial community composition at the Dale City Plant, which achieved the highest organic and nitrogen removal efficiencies among the three plants. The lower DO concentrations and lower influent BOD and TKN loadings promoted development of Nitrospira in the oxidation ditches at the EDC Plant. Temperature was not considered an impacting factor in this study because all samples were collected during winter seasons and monthly average data were used in the analysis. Overall, several operating conditions and influent characteristics influenced microbial community composition at the three plants. It should be noted that the technology employed at the three plants were different and the adjustment of operating conditions to achieve enhanced organic and nitrogen removal should consider limitations of specific treatment technology.
CCA plots for linking microbial community compositions to (a) operating conditions, (b) influent organic characteristics, and (c) influent nitrogen characteristics.
CONCLUSIONS
The low DO SBRs had similar composition and diversity of microbial communities during react fill and react phases. Microbial community composition and diversity were almost the same across the bioreactors operating in parallel at the SBR and oxidation ditch WWTPs. The microbial similarity inside and across the WWTPs suggests broad applicability of SND.
Microbial community composition was similar at the phylum level and divisive at the genus level among different process configurations. Statistical analyses suggested that there was more than one factor among operating DO level and low DO zonation, MLSS concentration, nitrogen and organic loadings, and SRT that affected the microbial community composition of the low DO SND systems.
The oxidation ditches had the lowest operating DO levels, constant low DO zones, and low BOD and TKN loadings, resulting in the highest relative abundance of Nitrospira and, consequently, the greater role of the conventional SND process. At relatively higher DO levels, high MLSS, or longer SRT, the SBRs and step-feed aeration tank removed nitrogen using a mixture of mechanisms involving mainly simultaneous heterotrophic nitrification and aerobic denitrification.
ACKNOWLEDGEMENTS
This work was funded by American Water in 2017. We would like to thank operators at the three WWTPs, including Terry Miller at the Dale City Plant, Roger Parr at the EDC Plant, and Christine Wesolowski at the Scranton Plant. Special thanks are extended to Bill Johnson for help on PCR analysis and Jake Metch for help on QIIME analysis.