Optimizing start-up strategies for the two-inflow nitritation/anammox process: Influence on biofilm microbial community composition

Anammox bacteria have gained increasing usage in the removal of nitrogen from ammonium-rich wastewater, such as industrial wastewater (Daverey et al. 2013), landfill leachate (Wang et al. 2016), and digester supernatant (Gut et al. 2006; Pereira et al. 2019), because of its low-energy consumption and low-sludge generation (Strous et al. 1999; Tsushima et al. 2007). Furthermore, stable operation can be achieved in the two-stage process, wherein nitritation (partial nitrification) and anammox reactions are conducted in separate reactors (Sharon-Anammox) (Hellinga et al. 1998) and a large number of plants are constructed (Van der Star et al. 2007). In contrast, a one-stage nitritation/anammox process, wherein both reactions occur in a single reactor, was developed for treating ammonia-containing wastewater due to reduced investment costs (Sliekers et al. 2003; Lackner et al. 2014). However, providing and maintaining an appropriate environment for both slow-growing ruler microorganisms, the ammonium-oxidizing bacteria (AOB), and anammox bacteria pose a challenge in the one-stage nitritation/anammox process (Strous et al. 1998). Low dissolved oxygen (DO) and intermittent aeration promote simultaneous microbial reactions and enhance total nitrogen removal (Jetten et al. 2001; Yang et al. 2015). Several types of reactors, such as the sequential batch reactor (Joss et al. 2009; Daverey et al. 2013), up-flow anaerobic sludge blanket (UASB) (Li & Sung 2015), and integrated fixed-film activated sludge (Zhang et al. 2015) have been used for one-stage anammox processes. Different types of biofilm carriers, including granular sludge (Pérez et al. 2014), sponge (Zhang et al. 2015), immobilized gel (Isaka et al. 2011), and membranes (Third et al. 2001) have been used to create necessary conditions for coexistence (Pérez et al. 2014). Cho et al. (2011) observed AOB growth on the outer portion of the carrier with anammox bacteria growing internally, whereas Li & Sung (2015) reported that the growth of both bacteria overlapped in the UASB reactor. AOB oxidizes ammonium to nitrite, creating an anoxic niche, which preserves the growth of anammox bacteria; however, each granule must maintain a suitable microbial consortium, which can increase the sensitivity of the long-term operation.

Zulkarnaini et al. (2018) proposed a novel approach, the one-stage partial nitritation/anammox (PNA) process involving two inflows to induce habitat segregation within each microorganism. In this process, wastewater was supplied to a reactor from two lines: Line 1, through the string wound filter placed inside the reactor and Line 2, directly from the bottom of the reactor. The ammonium supplied from the bottom of the reactor was oxidized to nitrite, which subsequently reacted with the ammonium supplied from the filter within. This design facilitated the growth of specific microorganisms at each respective site. A laboratory-scale experiment using a two-inflow PNA process could successfully treat the artificial wastewater; however, the microbial community in the reactor biomass remains unverified. Furthermore, start-up processes are crucial in improving reactor performance. In this study, two reactors were operated with different start-up strategies to determine the optimum start-up method of the two-inflow PNA process and the microbial community in the reactor was analyzed to verify habitat segregation.

Basal medium composition: KHCO₃, 500 mg; KH₂PO₄, 27.2 mg; MgSO₄.7H₂O, 300 mg; CaCl₂.2H₂O, 180 mg; CaCl₂, 136 mg; and 1 mL trace element solutions I and II (Van De Graaf et al. 1996), supplemented with each nitrogen concentration of (NH)₂SO₄ and NaNO₂ (per L ELIX water).

In Run 2, the anammox granules were inoculated and anammox bacteria were cultivated under anaerobic conditions at a low NLR (Period I), which was identical to that of Period I in Run 1. In the initial stage of Period II, the sponges from the nitritation reactor were transferred to the experimental two-inflow reactor and aeration was initiated. The anammox medium, which consisted of ammonium and nitrite, was supplied from Line 1. The concentration of nitrite subsequently reduced from 200 to 100 mg/L and the ammonium concentrations increased periodically from 100 to 300 mg/L. The ammonium medium was also added from Line 2. The DO concentration was maintained at 0.5–1.5 mg/L. During Periods III to V, the ammonium medium was supplied from both lines at the same NLR similar to that of Run 1.

The influent and effluent waters were sampled once weekly and filtered using 0.2 μm pore size membranes (Merck Millipore Ltd, Germany). The concentrations of ammonium, nitrite, and nitrate were analyzed using an ion chromatograph (Shimadzu HIC-SP, Japan). The pH was measured using a pH meter (HORIBA F-71, Japan) and DO was measured using a DO meter (HACH HQ30d, Germany).

Biofilms attached to the filter surface and on the sponge, and granules settled at the bottom were collected on days 265 (Period II) and 425 (Period IV) from both runs. Additionally, inoculated anammox granules and biofilms on the sponge in the nitritation reactor were collected.

DNA was extracted using a PowerSoil DNA Isolation Kit (QIAGEN, Germany) as per manufacturer's instructions. The 16S rRNA genes were amplified via polymerase chain reaction (PCR) using universal forward primer 515F/universal reverse primer 806R (Caporaso et al. 2012). The PCR program consisted of 25 cycles of 10 s at 98 °C, 15 s at 55 °C, and 45 s at 68 °C, and was performed using an Applied Biosystems 2720 thermal cycler (Thermo Fisher Scientific, Japan). The PCR products were sequenced using the Illumina MiSeq method (Illumina, USA). The sequences were processed using the UPARSE line (Edgar 2013) and QIIME software with SILVA_128 as the reference database (Caporaso et al. 2010).

In Run 2 (Figure 3), the operational process in period I was identical to that of Run 1, yielding similar results with an NRR of 0.144 kg N/m³ · d at NLR 0.20 kg N/m³ · d. In Period II, nitritation sponges were added and aeration was initiated. The ammonium medium was supplied from Line 1, whereas the nitrite medium was supplied from Line 2. The nitrite concentration from Line 2 was gradually decreased from 100 mg N/L to zero, whereas the ammonium concentration was increased from 300 to 400 mg N/L. The DO concentration was maintained below 1.5 mg/L via air flow regulation. The anammox reaction efficiently eliminated most of the nitrite and approximately 80% of the ammonium. Both ammonium oxidation and anammox reaction occurred, resulting in an NRR of 0.237 kg N/m³ · d at an NLR of 0.260 kg N/m³ · d. In Period III, nitrite supplementation was discontinued and the ammonium concentration in both lines was maintained at 500 mg N/L. The nitritation/anammox reaction occurred stably. Ammonium conversion efficiency (ACE) was approximately 50% and NRR was 0.277 kg N/m³ · d. Moreover, NLR increased in Period IV and decreased in Period V via varying the flow rate; however, this did not alter the reactor performance. Although the trend of the reactor performance after Period III was similar to that in Run 1, the maximum NRR and nitrogen removal rate (NRE) in Run 2 were higher than those in Run 1. These results suggest that the start-up strategy affects reactor performance and gradual acclimation of bacterial facilitation of nitritation is crucial for enhancing the nitritation/anammox process.

The phylogenetic classification of bacterial sequences at the phylum level is presented in Table 3. Significant differences between the inocula and biomass in the reactors were observed. Dominant phyla in the inoculated anammox granules were Chloroflexi (40.7%) and Planctomycetes (31.9%). Proteobacteria (18.1%), Bacteroidetes (3.6%), and Acidobacteria (2.5%) were also detected. In contrast, Proteobacteria was detected in high abundance (42.0%) in the sponge inoculum. In the two-inflow reactors, Planctomycetes (20.6–52.9%), Chloroflexi (2.4–49.1%), Chlorobi (2.5–38.7%), and Proteobacteria (4.2–12.3%) were detected in all samples.

In the inoculated granules, the most abundant bacterial genus was Candidatus Brocadia (26.9%), which is a common anammox bacterium. Uncultured bacteria envOPS12 (7.3%) and SBR1031 (A4b: 19.5%; SJA101: 11.2%), belonging to Anaerolineae of the Chloroflexi phylum, were also detected. The Chloroflexi phylum plays a beneficial role in providing a filamentous scaffold for floc formation, fermenting carbohydrates, and degrading complex organic compounds to support the growth of other bacterial populations in activated sludge (Speirs et al. 2019). SBR1031 is a heterotrophic bacterium commonly found in anammox reactors without organic carbon compounds. It can utilize extracellular polymeric substances as a carbon source and contribute to the removal of cellular debris and extracellular proteins, which are essential for cell aggregation and anammox sludge granulation (Li et al. 2023; Yu et al. 2023). Denitrifying bacteria Thauera and Dok59, belonging to the family Rhodocyclaceae, also grew (12.3 and 2.5%, respectively) using endogenous organic substances. Additionally, low abundance of Phycisphaerales belonging to the phylum Planctomycetes was detected. In the inoculated sponge in the nitritation reactor fed with ammonium under semi-aerobic conditions, the most abundant bacteria were unclassified Nitrosomonadaceae, an ammonium-oxidizing bacterium. NOB–Nitrospira, were not detected, implying that partial nitrification occurred in the sponge reactor.

In Run 1, the reactor received ammonium via Line 1 and a sufficient amount of nitrite via Line 2 without aeration after seeding anammox granules during Period II. In the granules collected from the bottom of the reactor collected on day 265 (final of Period II), the abundance of Candidatus Brocadia decreased to 5.6% and Phycisphaerales belonging to phyla Planctomycetes increased, because nitrite alone was added from the bottom of the reactor. Furthermore, the abundance of denitrifiers (Thauera and Dok59) decreased. Conversely, the abundance of Candidatus Brocadia was 33.1% in the biofilm attached to the filter surface. Anaerolineae envOPS12 was also detected with high abundance (27.1%); anammox bacteria and Anaerolineae may have detached from the granules and formed a biofilm on the filter surface. On day 425 of Period V in Run 1, in which ammonium was supplied from both lines and aerated after inoculation of the nitritation sponge in Period III, similar communities were observed in all biomasses. Candidatus Brocadia was detected in the range of 26.9–27.1%, and envOPS12 and Ignavibacteriaceae (phyla Chlorobi) were detected in high abundance. Ignavibacteriaceae a is non-phototrophic chemoheterotrophic facultative-anaerobic bacterium family that uses organic matter from other cells for biofilm formation (Connan et al. 2017). Biofilm forming bacteria shifted from anaerobic to facultative-anaerobic via aeration. Although Nitrosomonadaceae in the sponge decreased, they were detected with an abundance of 3.2–3.3% in all biomasses, implying that both nitritation and anammox occurred in the entire reactor.

As mentioned in the section ‘Reactor performances’, the NRR in Run 2 was considerably higher than in Run 1. These results indicate that the growth of nitritation bacteria on the biofilm surface is an important factor in improving NRR. In this study, the sponge helped accelerate the nitritation reaction. The two-inflow reactor facilitated ecological channel maintenance between both bacteria. A sudden change in the environment, such as the start-up in Run 1, was unsuitable for the formation of the biofilm consortium; however, the NRR was relatively low, despite being within the range of previous reports. Beyond its conventional role in nitrogen removal, DO plays an important role in maintaining the balance between AOB and anammox and inhibiting the growth of NOB. Notably, in this study, the DO level was higher than those of previous studies, which typically reported a DO of 0.5 mg/L (Antwi et al. 2019; Li et al. 2019). Despite the elevated DO, nitrate was undetectable and the abundance AOB remained low, indicating effective control over nitrite oxidation. Furthermore, the significantly increased anammox abundance, in comparison to a single inflow reactor, suggests that the two-inflow system created optimal conditions for anammox bacteria.

The filter surface area versus water volume was low due to the presence of a single filter; hence, the biomass amount was insufficient for NRR improvement. The multi-filter process could improve reactor performance. In addition, the sponges attacked the filter surface biomass and prevented its growth on the filter. Sponge removal after AOB acclimation is considered optimal.

The two-inflow PNA process demonstrated a stable performance. The start-up strategy of Run 2, which involved a gradual decrease in nitrite concentration in the influent, resulted in better performance than that of Run 1. The results of microbial community analysis revealed that Candidatus Brocadia and Nitrosomonadaceae were detached from the inoculated granule and sponge biomass, respectively, and formed a thick biofilm on the filter. The two-inflow PNA reactor could induce habitat segregation; however, the NRR did not increase as in previous reports. Further research is required to improve reactor performance.

This research was supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI [Grant Number: JP18K19874], and the Ministry of Higher Education and Research and Technology. The publication was supported by the Faculty of Engineering, Universitas Andalas.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.