Comparison of Nitrogen Removal Eciencies and Microbial Communities of Full-Scale Anaerobic, Anoxic and Aerobic Processes in Municipal Wastewater Treatment Plants having Low and High COD/TN Ratios

Full-scale anaerobic, anoxic and aerobic (A 2 O) process is used worldwide for biological nutrient removal (BNR). However, operation parameters for nitrogen removals and information of microbial communities related to nitrogen removal in full-scale A 2 O wastewater treatment plants (WWTPs) having low and high COD/TN ratios are not available. Based on the analysis of four full-scale A 2 O WWTPs, it is suggested that maintaining longer SRT of ≥ 30 day and DO of ≥ 0.9±0.2 mg-O 2 L -1 is needed to improve nitrogen removal eciency under low COD/TN ratio ( ≤ 3.7). On other hand, at high COD/TN ratio ( ≥ 4.2), DO level of ≥ 2.6 mg-O 2 /L and typical SRT of 19 ‒ 25 days would be suggested. It was conrmed that phosphorus removal eciency signicantly improved under BOD/TP ratio of > 20 for A 2 O process in these full-scale WWTP. Microbial distribution analysis showed that ammonia-oxidizing archaea (AOA) was abundant under conditions of low DO level, longer SRT, high temperature and low COD/TN ratio ( ≤ 3.7). Nitrosomonas sp. are mostly found in aerobic tank of full-scale A 2 O WWTPs. However, abundances of Nitrosomonas sp. are proportional to DO and NH 4+ concentrations for WWTPs with high COD/TN ratio. Nitrosospira sp. are only found under operating condition of longer SRT for WWTPs with low COD/TN ratio. Abundances of Nitrobacter sp. are proportional to DO concentration and temperature rather than abundance of Nitrospira sp. Predominance of nosZ-type denitriers were found at low COD/TN ratio. Abundance of denitriers by using nirS genes was over abundance of denitriers by using nirK genes at high COD/TN ratios WWTPs. (4±0.5 mg-O 2 L -1 ), shorter SRT of 17 day and HRT (3.6 hr.) were found at SK1. At TH2, quite high DO level (2.6±0.2 mg-O 2 L -1 ), SRT of 19 day, and longer HRT (15.4 hr.) were found at TH2 and quite high DO level (3±0.3 mg-O 2 L -1 ), quite longer SRT of 26 day, and longer HRT (10.1 hr.) were found at SK2. Longer SRT of ≥ 26 day with DO level ( ≥ 0.9±0.2 mg-O 2 L -1 ) could be suggested to use as operation parameters when treating low COD/N ratio wastewaters at TH1 or COD/N ratio of 4.2 at SK2.


Introduction
Organic matters and inorganic nutrients are the main contaminants to be treated in municipal wastewaters. Discharge of inorganic nutrients such as nitrogen (N) and phosphorus (P) compounds with to environment is responsible for eutrophication or algal blooms and toxic effects to aquatic life. For this reason, organic matter and inorganic nutrients have to be removed before discharging wastewaters to environment. Biological treatment processes are often recommended because of high removal e ciency and inexpensive operation cost compared to physical and chemical treatment processes. Modi ed Ludzack-Ettinger (MLE) consisting of anoxic and aerobic processes is speci cally designed for N removal. A 2 O consists of anaerobic, anoxic and aerobic processes and is used worldwide to remove both N and P biologically in which abundance of microorganisms responsible for various nutrient removal is achieved under designed environmental conditions. With alkalinity provided for nitri cation step, A 2 O processes can remove both nitrogen and phosphorus e ciently, producing good settling sludge. The energy cost is low and operation is relatively simple. Internal nitrate recycle through proper control of recycling return activated sludge (RAS) from aerobic zone to the anoxic zone is the key to operate A 2 O process successfully [1]. When designing A 2 O processes, wastewater characteristics and operation parameters including contact time of anaerobic tank, solids retention time (SRT), hydraulic retention time (HRT), DO concentration, etc. must be taken into consideration. Wastewater with proper carbon and total nitrogen ratio, i.e., COD/TN ratio, results in good process performance. Typical COD/TN ratio of ≥ 4 for denitri cation process (anoxic tank) and BOD/TP ratio of 20 for biological phosphorus removal process (anaerobic tank) are observed for an A 2 O processes with excellent N and P removal [1]. Insu cient carbon source for denitri cation process occurs in municipal wastewater with low COD/TN ratio, resulting in low N removal performance [2]. External carbon source addition [3] or maintaining longer SRT [4] are the most effective approaches to improve biological N removal performance for wastewater with low COD/TN ratio. Investigating the effects of SRT on N removal e ciency, Liu et al. [5] reported that system with SRT at 40 day outperformed those with shorter SRTs (5, 10, and 20 day). Phanwilai et al. [4] achieve excellent N removal with a step feed treatment process operated at SRT of 60 day. Biological N removal can be improved by optimizing DO levels. In case of keeping very low DO level of 0-0.5 mg-O 2 L − 1 , ammonia-oxidizing archaea (AOA) would be the dominant microorganisms group responsible for N removal [6].
Increasing abundance of ammonia-oxidizing bacteria (AOB) was reported with a high DO level of 1.9-3.5 mg-O 2 L − 1 [6]. The domination of Nitrospira was observed at DO below 1.0 mg-O 2 L − 1 [7].
Temperature and free ammonia (FA) are also important factors affecting the microbial community. Temperature at 10-20ºC was reported to be the optimum range for Nitrospira [8] and the temperature at 24-25ºC is favorable for Nitrobacter [7]. FA was an inhibitor of NOB activity [9]. Furthermore, Nitrobacter is more sensitive to FA than Nitrospira [10].
This research work focused on the nitrogen removal performance, identi cation and quanti cation of microbes from anaerobic, anoxic, and aerobic tanks of A 2 O processes in four full-scale municipal WWTPs having low (≤ 3.7) to high (≥ 4.2) COD/TN ratios. In this work, only nitrogen removal e ciencies were compared and discussed by using results from microbial quantity and communities of bacteria related to nitrogen removal, such as AOA, AOB, nitrite-oxidizing bacteria (NOB), and denitrifying bacteria (DNB) in these four full-scale A 2 O WWTPs. In addition, the results from this work could be applied to increase N removal e ciencies of these four full-scale WWTPs.

Materials And Methods
Description of the study sites A 2 O processes from four full-scale WWTPs with wide ranges of COD/TN ratios were selected as the study sites. Two WWTPs were located in Ding Daeng, Bangkok and Samut Prakan Province, Thailand, and are denoted as WWTP TH1 and TH2, respectively. The other two full-scale WWTPs, Dalseocheon and Sincheon WWTPs, were located in Daegu, South Korea, and are denoted as WWTP SK1 and SK2, respectively. A 2 O processes in these full-scale WWTPs were mainly designed for BNR, especially for biological both nitrogen and phosphorus removals. The in uent COD/TN mass ratios at TH1, TH2, SK1, and SK2 were 3.7, 8.4, 4.2, and 5.6, respectively. TH1 is considered to have low COD/TN ratio (≤ 3.7), while TH2 has relatively high COD/TN ratio (> 4) due to the wastewater being mainly generated from aircrafts, business and commercial buildings, such as hotels and airlines' o ces, in the surrounding of Suvarnabhumi airport.
The schematic layout of the full-scale A 2 O processes in the TH1 and TH2 is shown in Fig. 1 (A). The schematic layout of full-scale A 2 O processes in the SK1 and SK2 (with 1st and 2nd clari ers) are shown in Fig. 1 (B). No primary clari er was designed for TH1 and TH2. Two internal recycles are designed in these plants. The rst one is from aerobic zone to anoxic zone. The second one is for the return activated sludge (RAS) which was recycled from the 2nd clari er back to the anaerobic zone.
All wastewater samples from these four full-scale WWTPs were collected from each sampling points (anaerobic, anoxic, and aerobic zones) twice (between 2018 and 2019).

Wastewater quality analysis
All in uent and e uent wastewater samples were analyzed for BOD, COD, NH 4 + -N, NO 2 --N, NO 3 --N, organic-N, TKN, TN, TP, TSS, and SS following the Standard Method [11]. Selective e uent samples were measured for E. coli.

Microbiological analysis
Molecular analysis of microbes was conducted for selected sludge samples from anaerobic, anoxic, and aerobic zones. Before DNA extraction step, the sludge of each tank was harvested and kept on ice. One mL of sludge was used for DNA extraction according to the procedures of Zhou et al. [12].
Polymerase chain reaction (qPCR) Using quantitative polymerase chain reaction (qPCR) analysis focused to the microbial abundance.  Table S1. Oligonucleotide primers for PCR ampli cation via qPCR are shown in Table S2.
Denaturing gradient gel electrophoresis (DGGE) Using denaturing gradient gel electrophoresis (DGGE) analysis focused to the microbial communities responsible for N removal. Total bacteria were identi ed via 16S-rRNA EUB gene. Ammonium-oxidizing bacteria and archaeal were identi ed through AOB-and AOA-amoA genes. Nitrite-oxidizing bacteria (NOB) were identi ed Nitrospira and Nitrobacter via 16S rRNA NSR and Nitro genes, respectively. Denitrifying bacteria (DNB) were identi ed via nirS, nirK and nosZ genes.
Oligonucleotide primers for DGGE are shown in Table S3. Each 25-µL reaction mixture was added with 1 µ L -1 of template DNA with concentration of 10-20 ng µ L -1 , 10X Ex Taq™ buffer, 5 units/µL TaKaRa Ex Taq™, 2.5 mM dNTP Mixture, and 10 pmol of each primer, and the mixture was nally diluted with nuclease-free water. All PCR reactions were performed by using a T100™ Thermal cycler (BioRad Laboratories, CA and USA). (Bio Rad Laboratories, Inc.). The PCR product of 15 µL was loaded into individual lanes on 8% (W/V) acrylamide gel with 35-55% gradient for EUB target and with 35-50% gradient for AOB target. And electrophoresis step was performed in 1x TAE buffer at 58 ºC with a constant voltage at 80 V for 16 h. The shaped DNA band on acrylamide gel was excised by a scalpel. The DNA fragments were eluted by milli-Q water and set aside in a refrigerator for overnight, and then ampli ed PCR with the same primer without attached CG-camp. Sequencing bases were aligned by using database of the National Center for Biotechnology Information (NCBI).

Loading rate and removal rate calculation
The removal e ciencies (%) of nutrients and contaminants were calculated as Eq. (1).
where C inf and C out are concentrations (mg L -1 ) of water quality parameters in in uent and e uent of a treatment process, respectively.
Free ammonia (FA) was calculated by Eq. (5) according to Anthonisen et al. [13]. where NH 4 + inf is ammonium concentration of in uent (mg-N L -1 ) and T is the temperature of e uent (°C). respectively. Flow rates from high to low are SK2 > SK2 ~ TH1 > TH1. Compared with TH1 (30 mg L -1 ), TH2 (198 mg L -1 ), and SK1 (75 mg L -1 ), high BOD values (260 mg L -1 ) were found at SK2 because the WWTP received industrial wastewaters as well (approximately 25% of total volume). TH1 received wastewater with very low BOD. It is because that TH1 treated wastewater collected from a combined sewer system with domestic sewage being diluted by stormwater. In ltration and in ow are able to enter this combined sewer system. Meanwhile, the high temperature inside the sewer lines could promote the degradation of BOD. Finally, the wide practice of septic tanks installation in the residential houses could remove BOD before wastewater entering the sewer lines.
The average nitrogen concentration and removal e ciencies in each month are shown in Figure 2. A 2 O processes are designed for biological N and P removals and suitable for municipal WWTP [1]. However, total nitrogen (TN) removal e ciency at TH1 was quite low (only 49%) compared to other plants. The low TN removal performance could be explained by the very low COD/TN ratio (3.7) in the wastewater received at TH1. The low nitrogen removal e ciencies were also reported by Liu et al. [14] for WWTPs treating wastewater of relatively low COD/TN ratios. It was reported that denitri cation process could not signi cantly occur due to the insu cient carbon source for denitri cation process with wastewater having relatively low COD/TN ratios of lower than 4. On the contrary, e cient TN removal (86%) was reported for TH2 which received wastewater with COD/TN ratio of 8.4.
For WWTPs treating low COD/TN ratio (≤ 4) wastewater, maintaining longer SRT (≥ 60 day) is recommended to overcome the low TN removal e ciency [4] as the longer SRT would increase nitrifying bacteria abundancy. Meanwhile, long SRT could also enhance NH 4 + removal by increasing nitri cation activity [1]. The effects of SRT on NH 4 + removal were reported by [5]. The e uent NH 4 + concentrations in activated sludge process were reported for SRTs at 5 day ( The results of TP removal con rm that the A 2 O processes at three WWTPs (BOD/TP ratio of > 20) was able to remove P concentration well (more than 96% of TP removal e ciencies) except TH1 which has lowest BOD/TP ratio of 13 and achieved only 35% of TP removal. To increase phosphorus removal e ciency at TH1 (in case there is low BOD:TP ratio), the chemical treatment by using alum as coagulant would be recommended.
Nitrogen-cycling microbial abundances and predominated existing would be relative with various environmental factors such as dissolved oxygen level, SRT, temperature, pH, ammonium loading rates (ALRs), etc., and will be discussed in the following section.

Nitrogen-cycling microbial abundances and communities
Abundances and communities of ammonia-oxidizing bacteria (AOB) and ammonia-oxidizing archaeal (AOA) Autotrophic nitrifying bacteria responsible for ammonia oxidation process are detected at WWTP TH1 and belonged to two orders: Nitrosomonadales (a liated to Nitrosomonas sp. Nitrosospira sp., Nitrosococcus sp., and Thiobacillus sp.) and Rhodocyclales (a liated to Azospira sp., Thauera sp., and Zoogloea sp.) as shown in Table 3. Zhang et al. [15] reported that in full-scale municipal WWTPs, the most important genera of AOB were Nitrosomonas and Nitrosospira. Furthermore, they mentioned that Nitrosomonas were the most dominated ones. Consistently, in the full-scale A 2 O WWTPs, Nitrosomonas sp. are the most dominant AOB, especially in the WWTPs TH2, SK1, and SK2 which were operated at high DO levels. By contrast, the microbial community of Nitrosospira sp. was found at the TH1 because this WWTP was operated under long SRT, a favor condition for the growth of Nitrosospira sp. (see Table 3). Although the abundance of Nitrosospira sp. is less than that of Nitrosomonas sp., the existence of Nitrosospira sp. might be suitable a factor for satisfying good nitri cation process when the conditions are not optimal for growth of nitrifying bacteria [16]. Figure 3A shows the abundance of AOA-amoA genes at WWTP TH1, which is the highest among the WWTPs full-scale A 2 O process investigated, and the signi cance of each tank at TH1 shown the high mean difference of letter grouping in blue clustered column, (p < 0.05), see Table S4. The lower DO level, high temperature, and longer SRT (> 30 day) operated in this plant would signi cantly promote the growth of AOA [17]. Gao Figure 3A). Other factors such as the high NH 4 + loading rate could also increase AOB abundance. The predominated AOB-amoA gene over AOA-amoA gene at TH2, SK1, and SK2 compared to TH1 could be attributed to the higher NH 4 + loading rates in these plants (see Table 2), and the signi cance of the gene (p < 0.05) shown difference of letter grouping in orange clustered bar chart (see Table S4). The typical design DO level for a nitrogen-removal process of around 2 mg-O 2 L -1 was recommended by [1].
Although abundance of AOA was not found three WWTPs TH2, SK2, and SK3, AOA and AOB would collaborate and offer possible advantage in ammonia oxidation rates at lower ammonia concentration at TH1. It is postulated that in the practical operation, it is desire to maintain low DO level in an aerobic tank to reduce energy and sustaining SRT range base on characteristics of each full-scale WWTP, the abundance of AOA might be possible group of microorganisms to collaborate with AOB for nitri cation process. However, in further research on suitable DO level and SRT range would be investigated to nd the optimum conditions of growth AOA that could collaborate with AOB.
Abundances and communities of nitrite-oxidizing bacteria (NOB) Figure 3B shows that Nitrobacter sp. was more abundance than Nitrospira sp. at WWTP TH1. The DO levels (0.7 to 1. The optimal temperature ranges for Nitrobacter and Nitrospira growth are still ambiguous. Huang et al. [7] concluded that Nitrobacter was favorable species under the temperature ranges of 24-25ºC while Nitrospira dominated at relatively high temperature range of 29-30ºC. On the contrary, Alawi et al. [20] indicated that lower temperature range of 10-20ºC was the optimum condition for Nitrospira growth. Roots et al. [21] mentioned that Nitrospira Meanwhile, Nitrobacter is more sensitive to the free ammonia (FA) concentration compared to Nitrospira [10]. Mehrani et al. [9] reported that FA was a major inhibitor on NOB activity. FA concentrations at these three WWTPs (0.25 mg-N L -1 for TH2, 0.32 mg-N L -1 for SK1, and 0.29 mg-N L -1 for SK2) were higher than that at TH1 (0.17 mg-N L -1 ). It could be postulated that FA concentration was inhibitor to decrease abundance of Nitrobacter in these three WWTPs.
In this work, only qPCR technique was used to identify both Nitrobacter and Nitrosipra. The speci c primers to detect nitrifying bacteria population via the DGGE technique were not used. As Nitrosipra are able to complete oxidation of NH 4 + direct to NO 3 without into NO 2 -(complete ammonia oxidizer, comammox process), the speci c primers to detect nitrifying bacteria population for Nitrobacter and Nitrosipra are recommended in the further research. In practical, if the information of Nitrosipra in full-scale WWTP is reliable, a new approach of comammox process would be applied for increasing BNR in the future.
Abundances and communities of denitrifying bacteria (DNB) Three coding genes of nitrite (nirK or nirS) and nitrous oxide (nosZ) reductases were evaluated for the abundance of denitrifying bacteria from these four fullscale WWTPs. As indicated in Figure 3C, higher abundance of nosZ-type denitri ers was found at TH1 among WWTPs investigated due to the low COD/TN ratio of ≤ 3.7 in TH1 (see Table S4). The effects of COD/TN ratio on the abundance of nosZ-type denitri ers were consistent with the results reported by Yuan et al. [24] who reported that the nosZ-type denitri ers was two orders of magnitude more at the in uent COD/TN ratio of 4.6 (1.29 × 10 8 copies) compared to that at COD/TN ratio of 8.4 (1.31 × 10 6 copies) at the Beijing municipal WWTP in China.
The average number of DNB copy presenting at TH1 shows that nosZ-type denitri ers in anoxic and anaerobic tanks were most dominated. Wang et al. [25] found that the abundance of nosZ was a good indicator for rechecking anoxic and anaerobic conditions, having more oxygen concentration those conditions. Base on this result, it can be concluded that the DO level in the anoxic and anaerobic tanks of TH1 were quite high, and denitrifying bacteria could not use NO 3 electron acceptor for denitri cation process, resulting in poor denitri cation e ciency at TH1 in anoxic condition. As shown in Table 2, DO level in anoxic tank at TH1 was 0.3±0.1 mg-O 2 L -1 . It should be noted that the low denitri cation e ciency at TH1 could also attributed to the low COD/TN ratio in the receiving water.
Tallec et al.
[26] and Jia et al. [27] indicated that low DO concentration in WWTPs favors nitrous oxide (N 2 O) production during nitri cation/denitri cation process. High abundance of nosZ gene in denitri es was also found in aerobic tank of WWTP TH1. Henry et al. [28] indicated that nosZ-type denitri ers could be responsible in N 2 O production. It could be postulated that the BNR process at TH1 could produce higher N 2 O gas among WWTPs investigated due to the low DO level of this plant (0.9±0.2 mg-O 2 L -1 ).
On the other hand, the high abundance of nirS-type denitri ers and less abundance of nosZ-type denitri ers were found in anaerobic and anoxic tanks at TH2, SK1, and SK2 due to high DO centration (2.6-4.5 mg-O 2 L -1 ) operated at the A 2 O process. Meanwhile, nirS-type denitri ers were higher than the nirK-type denitri ers at all full-scale WWTPs. Complete denitri cation is possible with nirS-type denitri ers [25]. Che et al. [29] found a predominance of nirS-type level over nirK-type of all eight full-scale municipal WWTPs in different cities of China. Based regression analysis, Zhang et al. [30] suggested that the abundance of nirK-type denitri ers was correlated with temperature and nirS-type denitri ers was linearly correlated with temperature and ammonium concentration.
Both heterotrophic and autotrophic communities of denitrifying bacteria were found as indicated in Table 3. Heterotrophic denitrifying bacteria (Ilumatobacter sp., Comamonas sp., Rhodoferax sp., Terrimonas sp., Niabella sp., Sediminibacterium sp., Tistrella sp., Oryzobacter sp.) are normally found in WWTPs [31,32] Autotrophic denitrifying bacteria belonging to Arcobacter (a liated Arcobacter suis) relate to pathogenic bacteria that were found in high abundance in the municipal full-scale biological N and P removals processes [33]. Kristensen et al. [34] reported that pathogenic Arcobacter bacteria was not found in WWTPs with longer SRT (25-35 day) because they could be able to pass through both anoxic and aerobic tanks. In this work, Arcobacter suis was only found at SK1. Other lamentous autotrophic denitrifying bacteria were commonly found in wastewater worldwide and were presented. Chloro exi plays a role in sludge occulation and is more commonly found in WWTPs designed to remove nutrients, and most appearance with a long SRT operation and expose the biomass to anaerobic conditions [35]. Haliscomenobacter sp. were lamentous bacteria and satis ed being in phosphorus concentrations [36]. Their lamentous bacteria were found and achieved to remove phosphorus in A 2 O processes.

Conclusions
High nitrogen removal performances of full-scale A 2 O process (TH2 and SK2 that high COD/TN ratios were 8.4 and 5.6, respectively) are successful with operation parameters, such as DO level (≥ 2.6 mg-O 2 L -1 ) and SRT (19-26 day). However, to improve nitrogen removal e ciency at WWTPs TH1 (COD/TN ratio of ≤ 3.7) and SK1 (COD/TN ratio of ≤ 4.2) would be possible if longer SRT (> 30 day) is operated. The result from this work is con rm that the A 2 O processes are be able to remove P concentration well in case there are BOD/TP ratios of ≥ 19.
Low DO level (0.9 ± 0.2 mg-O 2 L -1 ) and long SRT under low COD/TN ratio (≤ 3.7) at TH1 is responsible for the high abundances of AOA over AOB.
Nitrosospira could be an appearance at the long SRT maintaining. In contrary, high COD/TN ratio (> 4.2) contributed the abundance of AOB over AOA. The predominated Nitrosomonas were the most existence and others AOB population as Azospira, Nitrosococcus, Thiobacillu, Thauera, and Zoogloea were found.
The nirS outnumbered nirK-type denitri ers. Nitrospira sp. are more competitive than Nitrobacter sp. at the low operational DO concentration. However, abundance of Nitrobacter could be higher than abundance of Nitrospira under lower temperature conditions. Chloro exi and Haliscomenobacter representative for the autotrophic denitrifying bacterium and Ilumatobacter, Comamonas, Rhodoferax, Terrimonas, Niabella, Sediminibacterium, Tistrella, and Oryzobacter species played on the heterotrophic denitrifying bacteria. In case of high abundance of gene-type denitri ers (nosZ) could be found in both anaerobic and anoxic tanks. It could be indicated that there were quite high DO concentrations in both tanks. Maintaining low DO level as operation condition in WWTP full-scale A 2 O process for saving energy, it could be postulated to produce high amount of N 2 O gas rather than operating high DO level. The future research in this area should be recommended in full-scale WWTP.

Declarations
Availability of data and materials The data used to support the ndings of this study are available from the corresponding author upon request.

Competing interests
The authors declare they have no competing interests.
Funding Tables   Table 1 Operational parameters of the full-scale A2O processes Parameter anaerobic, anoxic, and aerobic processes Remark: + is DGGE band presenting on acrylamide gel, Ana is an anaerobic tank, Anx is an anoxic tank, and Aer is an aerobic tank. Figure 1 The schematic layout of the full-scale WWTPs at TH1 and TH2 WWTPs (A), and SK1 and SK2 WWTPs (B)