Pathways of N removal and N2O emission from a one-stage autotrophic N removal process under anaerobic conditions

To investigate the pathways of nitrogen (N) removal and N2O emission in a one-stage autotrophic N removal process during the non-aeration phase, biofilm from an intermittent aeration sequencing batch biofilm reactor (SBBR) and organic carbon-free synthetic wastewater were applied to two groups of lab-scale batch experiments in anaerobic conditions using a 15N isotopic tracer and specific inhibitors, respectively. Then, the microbial composition of the biofilm was analysed using high-throughput sequencing. The results of the 15N isotopic experiments showed that anaerobic ammonium oxidation (Anammox) was the main pathway of N transformation under anaerobic conditions and was responsible for 83–92% of N2 production within 24 h. Furthermore, experiments using specific inhibitors revealed that when nitrite was the main N source under anaerobic conditions, N2O emissions from heterotrophic denitrification (HD) and ammonia-oxidizing bacteria (AOB) denitrification were 64% and 36%, respectively. Finally, analysing the microbial composition demonstrated that Proteobacteria, Planctomycetes, and Nitrospirae were the dominant microbes, corresponding to 21%, 13%, and 7% of the microbial community, respectively, and were probably responsible for HD, Anammox, and AOB denitrification, respectively.

Nitrous oxide (N 2 O), a powerful greenhouse and ozone-depleting gas, has a lifetime of approximately 118 to 131 years and is 300-fold more potent than carbon dioxide (CO2) 1,2 . N 2 O contributes 6 to 8% of the anthropogenic greenhouse effect worldwide 3 . Moreover, the atmospheric concentration of N 2 O has increased at an annual rate of 0.2 to 0.3% over the past decade 4 . N 2 O can be produced in biological wastewater treatments, especially treatments involving biological nitrogen (N) removal 5,6 . Recently, wastewater treatment plants (WWTPs) were found to exhibit gradually rising N 2 O emissions due to increases in population density and industrial activity 7 . Therefore, studying the N 2 O emissions of biological N removal systems is beneficial for controlling the greenhouse effect and protecting the ozone layer.
The one-stage autotrophic N removal process is especially well suited for treating wastewater containing high ammonia but low organics, such as landfill leachate, livestock wastewater and agricultural effluent 8 , because it has several advantages: a low demand for aeration, no consumption of organic carbon and low sludge production 9,10 . In a spatial model of biofilm from a one-stage completely autotrophic N removal process, ammonia-oxidizing bacteria (AOB) and anaerobic ammonium-oxidizing bacteria (AnAOB) grew in different regions according to the concentration of dissolved oxygen (DO) 11 . In this case, ammonia was initially oxidized to nitrite by AOB located in an area of higher DO, i.e., the surface of the biofilm. Then, the nitrite and remaining ammonia are converted to N 2 by AnAOB in anaerobic zones 8 . Kartal et al. 12 presented Eq. 1 to describe the Anammox process. Ammonium is the main N source during one-stage autotrophic N removal. Meanwhile, nitrite is produced by the oxidation of ammonia, and N 2 forms through the pairing of one N atom from ammonium and another N atom from nitrite 13 . Although the Anammox process is not fully understood, it is generally thought to produce no N 2 O gas 14,15 . Thus, improving the Anammox activity would be beneficial for reducing N 2 O emissions. However, the Anammox activity and its contribution to the removal of total N (TN) have not been measured in one-stage autotrophic N removal, making reducing the N 2 O generated in this process difficult.
In addition, heterotrophic denitrifying bacteria were also found in the systems used to treat wastewater containing high levels of ammonia-N without organics 16 , which suggests that heterotrophic denitrification (HD) is likely an additional pathway for N removal in the one-stage autotrophic N removal process. Traditionally, AOB denitrification and HD have been considered the two main pathways responsible for N 2 O emissions from biological N removal processes when DO is limited 17,18 . The presence of AOB and HD bacteria in the system indicates that the one-stage autotrophic N removal process might be a potential source of N 2 O emissions. In HD, N 2 O is believed to be an intermediate produced during denitrification that can be converted into N 2 by nitrous oxide reductase (N 2 OR) 19 . In contrast, AOB denitrification is thought to contribute the same level of N 2 O emissions as HD, or perhaps more, in terrestrial and marine ecosystems because of the lack of genes encoding traditional N 2 OR 20,21 . Typically, AOB denitrification can be influenced by the concentration of DO or elevated nitrite 22,23 , where as HD is closely related to nitrite accumulation, oxygen inhibition and the presence of biodegradable organic compounds [24][25][26] .
However, the contributions of AOB denitrification and HD to N 2 O emissions when the one-stage autotrophic N removal processis used to treat high-ammonia-N, organic-free wastewater remains unclear, especially under anaerobic conditions, such as non-aeration during the application of intermittent aeration or the inner space of the micro-biofilm environment when limited oxygen is supplied to the bulk liquid. Clearly, the emission of N 2 O under anaerobic conditions is an important contribution of the total N 2 O emissions of this system. Therefore, better understanding these mechanisms is essential for formulating operating strategies to minimize N 2 O.
This study was conducted to investigate the pathways of N removal and N 2 O emission from a one-stage autotrophic Nitrogen removal process under anaerobic conditions. Biofilm from a sequencing batch biofilm reactor (SBBR) was used for two groups of batch experiments, and the microbial composition was analysed. First, an 15 N isotope tracer technique was applied to investigate the contributions of Anammox and denitrification to TN removal via a one-stage autotrophic N removal process (batch test 1). Then, the N 2 O emissions corresponding to AOB denitrification and HD were quantified using specific inhibitors in this system (batch test 2). Finally, the microbial diversity and functional microorganisms associated with N 2 O emissions were analysed via high-throughput sequencing technology.

Results and Discussion
Performance of N transformation in the SBBR. The SBBR operated for more than one year with a stable effluent nutrient level and TN removal efficiency exceeding 80%. Figure 1 presents the N transformation performance of the SBBR in the final month of operation. The effluent TN remained in the range of 37.9-40.4 mg N L −1 , and the TN removal efficiency was 80.6 ± 0.6% ( Fig. 1(A)). The N compounds involved in the cycle are also shown in Fig. 1(B). The NH 4 + -N concentration gradually decreased from 89.3 mg N L −1 to 0 mg N L −1 as NO 3 − -N production increased from 11.2 mg N L −1 to 31.2 mg N L −1 , whereas the NO 2 − -N concentration did not exceed 5 mg N L −1 during this whole phase. In particular, NH 4 + -N exhibited a higher disappearance rate during aeration phases than during non-aeration followed by the increase of NO 2 − -N. This behaviour suggests that nitrosation occurred during the aeration phase, whereas during the non-aeration phase, NH 4 + -N and NO 2 − -N simultaneously disappeared via the Anammox process. These results indicate that nitrosation-Anammox is the main pathway for N removal in this system. However, during 22 to 24 h of NO 2 − -N degradation, the NH 4 + -N phase was completely removed, suggesting that denitrification occurred.
The N 2 O emissions corresponding to a single cycle of the SBBR are shown in Fig. 1(C). According to Eq. 6, the N 2 O-N emission factor throughout the process (EF (total) ) was 3.3% in the SBBR, which is similar to the result reported by Liu et al. 27 , and 2.7% of the TN input was converted to N 2 O-N in the simultaneous nitrification-denitrification (SND) process with intermittent aeration (aeration DO:1.5-2.0 mg/L). Jia et al. 28 , who used a lower DO (0.35-0.80 mg/L) during the aerobic phase, found that EF (total) was 7.7%. These results indicated that at the one-stage, completely autotrophic N removal and SND processes likely had similar sources of N 2 O emission, mainly during phases of low DO. However, the rates of N 2 O emission during the aeration intervals were much higher than those during the non-aerated intervals, probably because the later are associated with lower gas/liquid transfer coefficients. As a result, N 2 O emission occurs in both production processes, and stripping from the liquid arises during aerated intervals. Furthermore, the dissolved N 2 O increased during the non-aeration phase, suggesting that this phase is an important stage in N 2 O generation and may generate more N 2 O than the aeration phase. Specifically, the maximum rate of N 2 O emission was observed between 4 and 6 h, when the increase in nitrite was maximized. This finding indicates that N 2 O emission was affected by nitrite accumulation. Table 1  -N to rNH 4 + -N was 1.34, which is similar to the Anammox stoichiometry (1.32) for this ratio according to Strous et al. 29 and van der Heijden et al. 30 . This finding indicates that Anammox plays the main role in the N removal process. During the test, the ratio of rNH 4 + -N to rNO 2 − -N decreased gradually from 91% to 60%, indicating a gradual increase in the relative contribution of denitrification to N removal.

Pathways of N removal.
The value of R 30/29 (Fig. 2) was determined by IRMS, and the relative contributions of Anammox and denitrification were calculated (Fig. 2(B)) via Eqs 2 and 3. The results showed that R 30/29 gradually increased from 0.09  to 0.19; thus, 83-91% of all N 2 was produced by Anammox, and 9-17% was generated via denitrification. These results suggested that Anammox plays the primary role in N removal, consistent with the conclusion drawn above. In addition, the relative contribution of denitrification was found to gradually increase during the operation. Previous studies have shown that autotrophs supply heterotrophs with soluble microbial products (SMPs) for use as electron donors and carbon sources 31,32 ; subsequently, in turn, autotrophs receive inorganic carbon from heterotrophs metabolizing SMPs 33 . Therefore, the increased denitrification was probably attributable to the synthesis of SMPs, which can act as a potential electron donor for denitrification, by AOB.  (Fig. 2(B)), and the EF (total) was 1.6%, as calculated using Eq. 6. The EF (total) of batch test 1 was significantly lower than that of the SBBR (3.3%) because of the absence of nitrification, which is another source of N 2 O emission under aerobic conditions 31,32 . Furthermore, DO exerts an important influence on N 2 O emission from denitrification via HD bacteria and AOB 33,34 , and the DO concentration of batch test 1 differed substantially from that of the SBBR, which may also affect N 2 O emission. The isotopic composition of N 2 O from batch test 1was determined by IRMS (Fig. 2(C) -N and inhibitors (Fig. 3). As NO 2 − -N was added to system (group II), AOB denitrification and HD occurred simultaneously, and the average N 2 O-N release rate was 11.6 μ g (g·MLSS·h) −1 . Meanwhile, with the addition of inhibitors (group III), AOB denitrification was inhibited, and the average release rate of N 2 O-N was 7.5 μ g (g·MLSS·h) −1 . Thus, the release rate reduction of 4.1 μ g (g·MLSS·h) −1 reflects the activity of AOB denitrification. Calculations based on the N 2 O emissions results showed that 36% and 64% of N 2 O emissions were from AOB denitrification and HD, respectively, during the denitrification process, implying that HD is the main pathway of N 2 O emission under anaerobic conditions. Microbial distributions. Figure 4 presents the microbial composition of the biofilm based on the 16S rDNA amplicon pyrosequencing. These results suggest that the dominant microorganisms in the biofilm were Candidatus brocadia, Anaerolineaceae, Gemmatimonadaceae, Ardenticatenia, Nitrospira, Xanthomonadales, Nitrosomonas and Denitratisoma, with relative abundances of 11.2%, 10.4%, 10.1%, 8.7%, 7.0%, 4.2%, 4.1%, and 3.3%, respectively ( Fig. 4(A)). C. brocadia, Nitrosomonas and Denitratisoma have been reported to be Anammox, AOB denitrification and HD bacteria, respectively 35 . In addition, Nitrospira has been shown to be distributed in the outer layers of biofilms and to possess the ability to convert nitrite into nitrate 36 , whereas Xanthomonadales was classified as a member of Gamma proteobacteria, which are regarded as a type of HD bacteria. However, the roles of some species in N removal remain unknown. Thus, each phylum was classified based on 16S rDNA to investigate the biological bases for N removal and N 2 O emissions ( Fig. 4(B)). Chloroflexi, Proteobacteria, Acidobacteria, Planctomycetes, Gemmatimonadetes, Nitrospirae and Bacteroidetes were the main phyla. Most of the Anammox bacteria, HD bacteria and AOB for wastewater treatment could be classified as Proteobacteria, Planctomycetes and Nitrospirae, respectively [37][38][39] , which corresponded to 21%, 13%, and 7% of the total bacteria in the biofilm of this system. Thus, these bacteria might be the main sources of N 2 O emissions under anaerobic conditions.

Conclusions
The relative contributions of denitrification and Anammoxto N 2 production were calculated to investigate the N removal pathways in a one-stage autotrophic N removal system under anaerobic conditions. Anammox played the most important role in N removal, and denitrification emitted the most N 2 O, despite contributing little to N removal. Furthermore, HD created more N 2 O emissions than AOB denitrification under anaerobic conditions, although AOB denitrification was expected to be the more worrisome source of these emissions. Therefore, improving Anammox and decreasing denitrification contributed to reducing the N 2 O emissions of the system.

SBBR operation and synthetic wastewater.
The SBBR consisted of a rigid Plexiglas ® cylinder with an effective volume of 30 L, including approximately 9 L (30%, V/V) of flexible medium for biofilm growth. The bioreactor was operated at 30 ± 2 °C with intermittent aeration (aeration:non-aeration = 2 h:2 h) and a cycle time of 24 h (i.e., 4 min of feeding, 23 h of reaction, 30 min of settling and 26 min of decanting). The DO concentration in the aeration phase was controlled at 1.5 to 2.0 mg L −1 . In each cycle, approximately 10.5 L of wastewater was fed into the bioreactor, and the same amount of supernatant was with drawn after settling, resulting in a hydraulic retention time (HRT) of 48 h. The synthetic wastewater fed into the parent SBBR contained 1.13-g L −1 NH 4 HCO 3 (200-mg L −1 NH 4 + -N), 583.61-mg L −1 NaHCO 3 and 20-mg L −1 KH 2 PO 4 . NH 4 HCO 3 and KH 2 PO 4 were added as N and phosphorus sources, and NaHCO 3 was used to regulate the pH between 7.8 and 8.2. In addition, an appropriate amount of trace elementswere added to support microorganism growth, as described by Jia et al. 40 .

Isotopic tracer experiment.
To distinguish the contributions of Anammox and denitrification to N removal in the one-stage autotrophic N removal process, a 15 N-NaNO 2 isotopic tracer was added to a sealed bottle with an active volume of 100 ml that contained 10 g wet weight of biofilm from the SBBR and 90 ml of synthetic wastewater ( Table 1). The biofilm had been previously incubated for 5 h to remove nitrate from the biofilm. Next, helium gas was introduced to eliminate DO from the sealed Erlenmeyer flask containing the biofilm and pure water, and the temperature was controlled at 30 ± 2 °C. Then, the pure water was replaced with synthetic wastewater that was continuously sparged with helium gas; all other conditions remained constant. The wastewater con-  − -N. Finally, 100 μ l of 50% ZnCl 2 was added to the liquid samples to inhibit microbial activity. The isotope composition of N 2 was analysed to quantify the contributions of Anammox and denitrification to N 2 production. In incubations with 15 NO 2 − and NH 4 + , N 2 production via Anammox consisted of one N atom from NO 2 − and another from NH 4 + , leading to the production of 29 N 2 , whereas the denitrification of two N atoms from NO 2 − was assumed to produce 30 N 2 . However, because the F of 15 NO 2 − was not 100%, 28 N 2 and 29 N 2 were produced via Anammox, and 28 N 2 , 29 N 2 , and 30 N 2 were generated via denitrification. Therefore, the N 2 production mass of Anammox and denitrification could be respectively calculated according to Thamdrup and Dalsgaard 41 , The calculations were described as Eqs 2 and 3: where D N 2 and A N 2 represented the mass of N 2 produced by denitrification and Anammox, respectively; P 29 and P 30 represent the production amount of 29 N 2 and 30 N 2 , respectively, and F represents the fraction of 15 N in the NO 2 − pool. In this system, Anammox and denitrification were the only two pathways of N removal, the relative contributions of denitrification (Cd) and Anammox (Ca) to N 2 production can be described as the ratio of D N 2 to D N 2 plus A N 2 and that of A N 2 to D N 2 plus A N 2 , respectively. Therefore, Cd and Ca can be described by Eqs. 4 and 5, respectively: In which R 30/29 represents the ratio of 30 N 2 production to 29 N 2 production The isotopic composition of N 2 O was also investigated. N 2 O was generated as an intermediatein both nitrification and denitrification during the process of biological N removal 42 . Therefore, denitrification should be the only pathway of N 2 O emission under anaerobic conditions, and N 2 O should possess an isotopic composition similar to that of the N 2 produced by denitrification; that is, the ratio of 46 N 2 O to 45  Experiments involving specific inhibitors. The use of inhibitors can facilitate investigating the magnitudes of the various processes at the source of N 2 O production under anaerobic conditions. Allylthiourea (ATU) was used as the inhibitor of the nitrification of ammonia to nitrite, whereas NaClO 3 was used to inhibit the conversion of nitrite to nitrate catalysed by nitrite oxido-reductase 28 . The co-use of ATU and NaClO 3 can effectively inhibit the production of N 2 O via AOB denitrification 37 , whereas N 2 O emissions by heterotrophic bacteria are not significantly affected by the presence of ATU and NaClO 3 37 . Therefore, the emission of N 2 O produced by HD alone and by both AOB denitrification and HD can be quantified by batch experiments with or without the inhibitors.
Thus, three batch experiments were conducted: (I) no addition of nitrite or inhibitor, (II) the addition of nitrite, and (III) the addition of both nitrite and nitrification inhibitors (ATU and NaClO 3 ). Three devices were assembled for these the batch experiments using an isotopic tracer; then, a 1-L mixture containing 100 mg wet weight of biofilm and 900 ml of wastewater (NH 4 + -N: 9.7 mg L −1 ; NO 2 − -N: 1.8 mg L −1 ; and NO 3 − -N: 23.6 mg L −1 ) from the SBBR were introduced into a sealed Erlenmeyer flask, and then, NaNO 2 , ATU, and NaClO 3 were added to the effluent at concentrations of 100.0 mgN L −1 , 10.0 mg L −1 , and 1.0 g L −1 , respectively. Helium gas was introduced into the wastewater to ensure anaerobic conditions. The solution and off-gas in the devices were sampled every 6 h for 24 h, and the concentrations of NH 4 + -N, NO 2 − -N, NO 3 − -N and TN in the wastewater were measured to investigate the characteristics of N transformation. The N 2 O emissions were also detected to identify the contributions of AOB denitrification and HD. The amount of N 2 O emissions can be described as follows: II-I, the sum of AOB denitrification and HD; III-I, HD; and (II-I)-(III-I), AOB denitrification (Fig. 5).
Physicochemical analysis. The concentrations of TN, NH 4 + -N, NO 2 − -N, and NO 3 − -N were measured using a flow injection analyser (HachQuickchem 8500S2, Hach Inc., Loveland, CO, USA). Alkalinityand biomass dry weight (mixed liquid suspended solids, MLSS) were measured according to standard methods for water and wastewater 43 . The concentration of N 2 O was determined with an Agilent 7820A gas chromatograph (Agilent Technology Inc., Santa Clara, CA, USA) according to Jia et al. 40 . The dissolved N 2 O in wastewater was determined using the head space gas method described by Tsuneda et al. 44 . The values of R 30/29 for N 2 and R 46/45 for N 2 O were measured by isotope-ratio mass spectrometry (IRMS;MAT253, Thermo Finnigan LLC, San Jose, CA, USA) according to the method described by Cao et al. 45 . The N 2 O-N emission factors per TN converted during the interval i-i + 1 (h) and the whole process were calculated using Eqs 7 and 8, respectively:

Microbial composition.
To analyse the microbial composition in the one-stage autotrophic N removal process, biofilm from the SBBR was collected and centrifuged for to extract the DNA. The total genomic DNA was extracted using an E.Z.N.A. ® Soil DNA Kit (OMEGA Bio-Tek, Inc., Norcross, GA, USA), and the bacterial 16S rDNA genes of the biofilm were sequenced using Illumina MiSeq technology at the Shanghai Majorbio Bio-pharm Technology Co., Ltd. (Shanghai, China). Ultra-fast sequence analysis (USEARCH) was used to cluster the operational taxonomic units (OTUs) of a 16S DNA gene based on 97% similarity, and the statistical abundances of different OTUs in the samples reflect those of different microbial species. Then, the microbial composition was analysed according to sequencing information and data from the National Center of Biotechnology Information (NCBI) reference genome. Finally, microorganisms were classified as Anammox bacteria, AOB and HD bacteria based on the pathway of N metabolism. Simultaneously, the relative proportions of these microorganisms were calculated based on the OTU abundances.