Effects of substrates on N2O emissions in an anaerobic ammonium oxidation (anammox) reactor

N2O emission in the anaerobic ammonium oxidation (anammox) process is of growing concern. In this study, effects of substrate concentrations on N2O emissions were investigated in an anammox reactor. Extremely high N2O emissions of 1.67 % were led by high NH4-N concentrations. Results showed that N2O emissions have a positive correlation with NH4-N concentrations in the anammox reactor. Reducing NH4-N concentrations by recycling pump resulted in decreasing N2O emissions. In addition, further studies were performed to identify a key biological process that is contributed to N2O emissions from the anammox reactor. Based on the results obtained, Nitrosomonas, which can oxidize ammonia to nitrite, was deemed as the main sources of N2O emissions.

processes are a critical source of atmospheric N 2 O (Kampschreur et al. 2009a;Wunderlin et al. 2013;Shaw and Koh 2012). In addition, research has generally shown that N 2 is the end product of the anammox process (Jetten et al. 2005); however, high N 2 O emission from Anammox processes have also been reported (Kampschreur et al. 2009b). Thus, there is an urgent need to investigate the production of N 2 O in the anammox process and develop methods of controlling and decreasing the greenhouse emissions from the anammox process.
In this study, an anammox reactor was used to study the effects of substrate concentrations on the emission of N 2 O in an anammox process. The relationship between substrate concentrations and N 2 O emissions was studied by changing the influent NH 4 -N concentration. Furthermore, genetic analysis using the 16S rRNA gene was employed to characterize the microbial population of the anammox granules.

Reactor performance
The removal performance of ammonia and nitrite is shown in Fig. 1a, b. Whenever the effluent NO 2 -N concentration fell below 10 mg L −1 , the nitrogen loading rate (NLR) was increased by adjusting the influent nitrogen concentration while maintaining a constant HRT of 8 h. During start-up period, with the influent NH 4 -N and NO 2 -N concentrations set at 73.2 and 88.3 mg L −1 , respectively, effluent NH 4 -N and NO 2 -N concentrations below 7 and 2 mg L −1 were obtained, with the TN removal rate >80 %. Subsequently, at a constant HRT, the influent NH 4 -N and NO 2 -N concentrations were further increased to 100.4 and 124.8 mg L −1 , respectively, and effluent NH 4 -N and NO 2 -N concentrations initially were a little higher, but both soon decreased to below 8 mg L −1 over a 3-day period. These results indicated that the seed anammox sludge could adapt quickly to changes in NLR. On day 24, the influent NH 4 -N and NO 2 -N concentrations were increased to 145.0 and 176.4 mg L −1 , respectively, which were the highest levels used in this study. Under these conditions, the effluent NH 4 -N and NO 2 -N concentrations were 27.0 and 14.8 mg L −1 , respectively. Accordingly, influent NH 4 -N and NO 2 -N concentrations were decreased to 120 and 150 mg L −1 , respectively, and effluent NH 4 -N and NO 2 -N concentrations were then maintained below 18.6 and 9.9 mg L −1 . Overall, the reactor could operate with a stable nitrogen removal rate of over 81 %. Figure 1c shows the ratios of influent NO 2 -N/NH 4 -N, effluent NO 2 -N removal/NH 4 -N removal, and effluent NO 3 -N production/NH 4 -N removal. At the start-up period, influent NO 2 -N/NH 4 -N was set 1.2. Accordingly, effluent NO 2 -N removal/NH 4 -N removal, and Effluent NO 3 -N production/NH 4 -N removal were 1.3 and 0.3, respectively, which were close to values reported by others (Strous et al. 1998). In order to investigate the effect of influent NH 4 -N on N 2 O emission, on day 12 influent NO 2 -N/NH 4 -N was changed to 1.09. As a result, effluent NO 2 -N removal/NH 4 -N removal increased to 1.55. Conversely, effluent NO 3 -N production/NH 4 -N removal decreased to 0.2. The same results were once more affirmed on day 31. Denitrification was considered to be the main reason for the additional NO 2 -N or NO 3 -N removal.  Fig. 2. The conversion ratio of N 2 O was calculated from the removed nitrogen. On the first day, about 0.6-0.64 % N 2 O content was detected in the emission gas. On day 2, the influent pipe became blocked, thus only 0.34 % N 2 O was detected in the emission gas. However, this value increased to 0.54 % over the following 3 days and by day 6 the N 2 O concentration had reached 0.93 %, accompanied with a high effluent NH 4 -N residual. On days 11-13 and 30-32, the effluent NH 4 -N remained at 32-34 and 37-42 mg L −1 , respectively. Under these conditions, the N 2 O emissions were found to be significantly higher than the values associated with low effluent NH 4 -N concentrations. Over the course of the study, N 2 O levels were determined to be 0.6-1.0 % in the off-gas.

Effects of influent NH 4 -N, NO 2 -N, NO 3 -N and nitrogen removal rate on N 2 O emission
Effects of inlet NH 4 -N, NO 2 -N, NO 3 -N and nitrogen removal rate on N 2 O production are shown in Fig. 3. The EGSB reactor used in this study was operated with a high recycle rate. Thus, influent NH 4 -N, NO 2 -N and NO 3 -N were calculated by using the following equation.
where x is the concentration of inlet NH 4 -N, NO 2 -N, NO 3 -N, a is the influent concentration of NH 4 -N, NO 2 -N, NO 3 -N, n is the ratio of recycle rate to influent flow rate, b is the effluent concentration of NH 4 -N, NO 2 -N, NO 3 -N. According to the equation, inlet NH 4 -N, NO 2 -N and NO 3 -N were determined by two factors: changing influent concentrations or different recycle rate. In order to observe the effects of inlet NH 4 -N, NO 2 -N and NO 3 -N, only the recycle rate was changed while the nitrogen loading rate was set with the same value 0.5 kg m −3 day −1 (Fig. 3a-c). Also, the effects of nitrogen removal rate were evaluated with the same influent nitrogen concentrations. As shown in Fig. 3a, average N 2 O content was 0.6 % with an inlet NH 4 -N concentration of 27-28 mg L −1 . Increasing the inlet NH 4 -N concentration from 36 to 57 mg L −1 , N 2 O increased from 0.65 to 1.4 %. Inlet NH 4 -N concentration and N 2 O emission were simulated according to the current data by the following equation with P < 0.03.
where y is the N 2 O emission, x is the inlet NH 4 -N concentration. In a word, increasing inlet NH 4 -N concentration tended to yield a higher N 2 O concentration.
The influences of inlet NO 2 -N and NO 3 -N concentrations were also investigated during the study, though no obvious relationship was found with N 2 O emissions (Fig. 3b, c).

Bacteria community analysis
Hierarchical cluster analysis was used to identify the differences of three bacterial community structures (Fig. 4). The three samples were sampled from the same reactor, showing obvious similarity of community structure. Nitrosomonas, which oxidizes ammonia to nitrite, was detected in all the three samples. In the anammox reactor, it is difficult to keep dissolved oxygen at zero. Thus, the anammox reactor provides the conditions for the growth of Nitrosomonas. However, Nitrosomonas is known to produce N 2 O under low oxygen conditions (7). This was supported by the relationship between Nitrosomonas abundance and N 2 O emission (Fig. 5).
In this study, N 2 O emissions were found to be higher than the reported values. Okabe et al. reported that a N 2 O emission of only 0.05-0.23 % was detected with a nitrogen removal rate of 7.5-15 kg N m −3 day −1 . However, the highest N 2 O concentration of 1.67 % was quantified in this study, which is compared with other the results in Table 1. From Table 1, increasing nitrogen loading rates showed positive effect on decreasing N 2 O concentrations. Longfei et al. (2015) also reported that the increase of nitrogen loading rate could reduce N 2 O emission and they found it is more seasonable if compare the value of N 2 O production per gram N removal (N 2 O emission/nitrogen removal rate). Although higher nitrogen removal rate helps to reduce the footprint of the anammox system, it was difficult to maintain the stable running under high nitrogen removal rate due to floatation of anammox granules and pipe clogging. On the other hand, Kampschreur et al. also found high N 2 O concentrations with 0.6 % in one full-scale anammox reactor. Thus, increasing NLR is effective in reducing N 2 O emission, but N 2 O emissions (2) y = 0.0008x 2 − 0.0387x + 1.039 are inevitable in an anammox reactor. Reducing N 2 O emission is still a concern for anammox applications.
NO 2 -N and NO 3 -N are the substrates for denitrifiers. It is supposed that N 2 O is produced as an intermediate of incomplete heterotrophic denitrification due to low COD/N ratio (Okabe et al. 2011). However, no relationship was found between NO 2 -N, NO 3 -N and N 2 O emission in this study. Thus, it is difficult to explain the increasing N 2 O emission during this study.
Okabe et al. indicated that denitrification by putative heterotrophic denitrifiers present in the inner part of the granule was considered the most probable cause of N 2 O emission from the anammox reactor. In this study, inlet NH 4 -N showed clear relation to N 2 O emission (Fig. 3). Also, Nitrosomonas abundance increased with N 2 O emission (Fig. 5). As shown in Fig. 5, only 0.025 g-N 2 O emitted/100 g-N consumed was observed without Nitrosomonas. And the denitrifires were presumed to contribute the above N 2 O emission. After that, Nitrosomonas abundance increased with N 2 O emission. At last, 0.7 g-N 2 O emitted/100 g-N consumed was observed, which was almost 30 times of the value without Nitrosomonas. The results got in this study showed that Nitrosomonas was the main cause of N 2 O emission. Nitrosomonas competed with anammox bacteria for NH 4 -N. Because anammox bacteria could not oxidize NH 4 -N without NO 2 -N, therefore, supplying enough NH 4 -N is favorable for Nitrosomonas. While oxygen was always insufficient

N 2 O emission (%) References
Granules-based 70,000 7.14 0.6 Kampschreur et al. for NH 4 -N oxidation in one anammox reactor, thus, N 2 O produced due to NH 2 OH oxidation (Wunderlin et al. 2013). The results of this study is partly consistent with the literature showing that NH 2 OH oxidation by AOB was considered the most probable cause of N 2 O production (0.6 % of the nitrogen load) in a full-scale Anammox reactor treating sludge reject water (Kampschreur et al. 2009b). Beyond that, this study could not exclude the possibility of N 2 O emission by denitrifiers. Further study was needed to quantify N 2 O emission contributed by denitrifiers and Nitrosomonas using real wastewater.

Conclusions
One anammox reactor was used to investigate the effect of substrate concentrations on N 2 O emissions. The monitoring N 2 O concentrations were determined as 0.6-1.0 % in the emission gas during this study. Increasing inlet NH 4 -N concentration from 36 to 57 mg L −1 , N 2 O increased from 0.65 to 1.4 %. Reduced inlet NH 4 -N concentrations induced N 2 O emission. The results got in this study suggested that in addition to denitrifiers, Nitrosomonas was also a significant cause of N 2 O emissions.

Anammox reactor and substrate
The reactor had an inner diameter of 14 cm with a total liquid volume of 10 L including a reaction zone of 8 L and a settling zone of 2 L. The reactor was made of acrylic resin and had a water jacket for temperature control. Sampling ports were located at heights of 3, 17, 20 and 25 cm above the reactor bottom. Part of the effluent was collected in a 5-L container (with mixer and heater) for use as recycle (Fig. 6). The pH was adjusted by an online pH controller (TPH/T-10, Tengine, China) using 0.5 mol L −1 H 2 SO 4 . The reactor was enclosed in a black-vinyl sheet to prevent growth of photosynthetic bacteria and algae. The reactor was operated in up-flow mode, with influent introduced at the bottom using a peristaltic pump (BT100-2J, LongerPump, China). A recirculation pump (BT600-2J, LongerPump, China) was used to dilute the influent (Fig. 6) with the treated wastewater in the 5-L recycle container.
The anammox seed sludge used in the reactor was taken from a pilot-scale anammox reactor (unpublished). The seed sludge was granule activated carbon (GAC)-based granules with settling velocity over 150 m h −1 . The initial seeding concentration (mass of mixed liquor suspended solids (MLSS) per liter) was set at 4 g MLSS L −1 .
The reactor was fed with synthetic wastewater with a nitrite to ammonium molar ratio of 1.0-1.2. The detailed composition of the influent is shown in Table 2. The influent storage tank was flushed with nitrogen gas to maintain DO under 0.5 mg L −1 , and Na 2 SO 3 was added to a concentration of 40 mg L −1 (shown to be harmless to Anammox bacteria, Wenjie et al. 2014) to keep the DO level close to zero.

Analytical methods
NO 2 -N and NH 4 -N were measured by the colorimetric method according to Standard Methods (APHA 1995). Total nitrogen (TN) was determined by the persulfate method using the UV spectrophotometric screening method (APHA 1995) for quantification of TN as NO 3 -N (the oxidization product of the persulfate digestion). NO 3 -N (of the original sample) was determined by calculation of the difference of TN and the sum of NO 2 -N and NH 4 -N. The pH was measured by using a pH meter (9010, Jenco, USA), and dissolved oxygen (DO) was measured by using a DO meter (6010, Jenco, USA).

Gas collection and analysis
Gas was collected through the GSS (Fig. 1) and the volume was measured using an inverted cylinder containing tap water with the pH lowered to 3 using 1-N H 2 SO4. Gas analyses were performed by using a GC-112A gas chromatograph (INESA INSTRU-MENT, China).

DNA extraction and high-throughput 16 s rRNA gene pyrosequencing
After 139 days of operation, the particle based granules were taken out from the Anammox reactor. A granular sludge sample was first ground with a pestle under liquid nitrogen. Meta-genomic DNA was extracted using the E.Z.N.A. Soil DNA Kit (OMEGA Biotec. D5625-01, USA) according to the manufacturer's instructions. Amplification of the 16S