Wastewater management has experienced different stages, including pollutant removal, resource recovery, and water nexus. Within these stages, anaerobic ammonia oxidation-based biotechnology can be incorporated for nitrogen removal, which can help achieve sustainable wastewater management, such as reclamation and ecologization of wastewater. Here, the physiology, metabolism, reaction kinetics and microbial interactions of anammox bacteria are discussed, and strategies to start-up the anammox system are presented. Anammox bacteria are slow growers with a high doubling time and a low reaction rate. Although most anammox bacteria grow autotrophically, some types can grow mixotrophically. The reaction stoichiometric coefficients can be affected by loading rates and other biological reactions. Microbial interactions also contribute to enhanced biological nitrogen removal and promote activities of anammox bacteria. The start-up of the anammox process is the key aspect for its practical application, which can be realized through seed selection, system stimulation, and biomass concentration enhancement.
Sustainable wastewater management was proposed.
Anammox process is the key for nitrogen removal from wastewater.
High growth rate anammox bacteria could be selectively enriched.
Microbial ecology of anammox bacteria was comprehensively summarized.
Strategies for starting-up of the anammox process were concluded.
Proper management of wastewater is vital for protecting our living conditions, which can support the sustainable development of our society and environment. Previously, three stages of wastewater management have been implemented: pollutant removal from wastewater, resource recovery from wastewater and water nexus (Figure 1) (Mo & Zhang 2013; Leck et al. 2015; Grasso 2019). Conventionally, pollutant removal from wastewater is the focus of wastewater treatment to meet the discharging standard, which can benefit the receiving water bodies. Wastewater contains pure water, organic carbon, nitrogen, phosphorus, and other elements. Currently, the concept of resource recovery, energy recovery, as well as water reclamation/ecologization has become the new trend in wastewater management. Through the application of this new concept, organic carbon in wastewater can be concentrated indirectly or converted directly by the anaerobic methanogenesis process to recover methane as renewable energy (Yin & Wu 2019). The phosphorus in wastewater can be recovered through precipitation as fertilizer, while nitrogen is mainly removed as a pollutant. Following compound recovery or removal, the water quality is significantly improved, and can be reclaimed for industrial and environmental applications. Furthermore, the wasted water and the treated water can be incorporated into the concept of nexus with the environment, ecology, energy, food, and other areas (Mo & Zhang 2013; Leck et al. 2015). To achieve the proposed purposes, green technology should be applied for nutrient removal or recovery, which can contribute to the sustainable management of wasted water.
For nitrogen management in wastewater, conventional biological nitrogen removal is mainly achieved through full nitrification and denitrification processes (Figure 2) (Daims et al. 2006). Generally, for full nitrification, ammonia nitrogen (NH4-N) in wastewater is oxidized to nitrite nitrogen (NO2-N) by ammonia-oxidizing bacteria (AOB) and then to nitrate nitrogen (NO3-N) by nitrite oxidizing bacteria (NOB) under aerobic conditions, with oxygen as the electron acceptor. Subsequently, NO3-N is denitrified to nitrogen gas (N2) through activities of denitrifiers with organic carbon as the electron donor. However, with the new concept of energy recovery, after organic carbon recovery from wastewater such as through anaerobic digestion, there is no adequate organic carbon for conventional denitrification. Therefore, the autotrophic nitrogen removal process is becoming a promising technology for energy neutral or even energy positive results in wastewater treatment (Wang et al. 2015). Autotrophic nitrogen removal is mainly achieved through partial nitrification and anaerobic ammonia oxidation (anammox) processes. For partial nitrification (or nitritation), only the first step carried out by AOB would occur, and the second step by NOB would be inhibited. After nitritation, in the following anammox process, NH4-N reacts with the produced NO2-N, and both are converted to N2.
The anammox process is carried out by anammox bacteria (Kartal et al.2007a, 2007b), which are slow growers. Therefore, the key aspect for the anammox process is to enrich anammox bacteria. Lots of reviews have been conducted to discuss the physiology of anammox bacteria, the history, and technologies to start the process (i.e., Oshiki et al. 2011). However, till now, no reviews have been focusing on the systematic summary of the microbial characteristics of physiology, metabolism, kinetics and interactions, and strategies for start-up of the anammox process based on the microbial characteristics. In this review, the key clue for enhancing the growth of anammox bacteria based on microbiological characteristics and also strategies for enhancing system start-up will be focused on, which could provide a framework for practical application of the anammox process with the better management of wastewater resources.
ANAMMOX BACTERIA AND THEIR PHYSIOLOGY
Initially, the existence of anammox bacteria for removing ammonia with nitrite or nitrate as the electron acceptor was proposed by Broda (1977) based on the thermodynamic analysis. Later, the anammox phenomenon was discovered in a denitrifying bioreactor by Mulder et al. (1995). Thereafter, much research was conducted and practical applications were tested. After 3.5 years of project application, the first full-scale project was established for treating reject waters of anaerobic sludge digestion in Rotterdam (van der Star et al. 2007). Currently, more than 200 anammox facilities have been put into operation worldwide (Cao et al. 2017).
Six genera of anammox bacteria within the phylum Planctomycetes have been confirmed, including Candidatus Kuenenia, Ca. Brocadia, Ca. Anammoxoglobus, Ca. Anammoximicrobium, Ca. Jettenia, and Ca. Scalindua. Among these, the first five types are commonly found in wastewater treatment and freshwater systems, while the last one is commonly found in saline environments such as sea water and sediments (Kartal et al. 2013; Ali et al. 2015a; Guo et al. 2016; Lawson et al. 2017). Physiology and environmental factors affect the distribution of types of anammox bacteria significantly. For example, under low nitrogen loading rate (NLR) conditions, Ca. Brocadia anammoxidans, Ca. Jettenia, Ca. Anammoxoglobus, and Ca. Kuenenia are dominant (Li et al. 2017; Reino et al. 2018; Zhu et al. 2018), while under high NLR conditions, Ca. Brocadia sinica and Ca. Kuenenia stuttgartiensis dominate (Cho et al. 2018; Yang et al. 2018b). Furthermore, Ca. Brocadia and Ca. Brocadia fulgida are mainly observed at 6–15 °C (Hendrickx et al. 2014; Awata et al. 2015; Laureni et al. 2015; Lotti et al. 2015b), while K. stuttgartiensis was found at 25–45 °C (Isaka et al. 2008).
Although few anammox bacteria have been successfully isolated, they usually grow in the form of highly compact spheres, with diameters ranging from 0.6 to 1.0 μm (van Niftrik et al. 2004; Duan et al. 2012). So far, most studies have focused on the enriched anammox bacteria within biological reactors. Various proportions of anammox bacteria have been reported from different studies with varied reactor configurations, operational conditions and growth conditions (Ni et al. 2010). The purities of enriched anammox bacteria were 64 and 74% in the fluidized bed reactor (van de Graaf et al. 1997) and the sequencing batch reactor (SBR) (Strous et al. 1998), respectively. Anammox enrichment purity of 97.6% was achieved in a membrane bioreactor (MBR) inoculated with 60–80% purity granular anammox from the first full-scale anammox reactor (van der Star et al. 2008). Using cultured activated sludge with less than 10% anammox purity as the seed, the purity of enriched anammox bacteria could be up to 97.7% (van der Star et al. 2008).
NUTRIENT METABOLISM OF ANAMMOX BACTERIA
Generally, there are two types of nitrogen metabolic pathways for anammox bacteria. In anammox pathway I, NO2-N is reduced to hydroxylamine and then combines with NH4-N to form hydrazine, which is found in Ca. Brocadia anammoxidans (van de Graaf et al. 1997), while in anammox pathway II, NO2-N is reduced to nitric oxide (NO) and then combines with NH4-N to form hydrazine, which is found in K. stuttgartiensis (Strous et al. 2006). For the clarification of the metabolic pathways, the production of NO can be confirmed by using the NO scavenger PTIO (2-phenyl-4,4,5,5,-tetramethylimidazoline-1-oxyl-3-oxide). For genes responsible for the nitrite reduction, neither nitrite reductases nirS nor nirK were detected in the B. fulgida and B. sinica; nirS encoding cytochrome cd1-type NO-forming nitrite reductase was observed in K. stuttgartiensis and Ca. Scalindua profunda, and nirK encoding copper-containing NO-forming nitrite reductase was found in Ca. Jettenia caeni (Gori et al. 2011; Oshiki et al. 2015).
Anammox bacteria are autotrophic microorganisms, which use inorganic carbon such as CO2 as the carbon source. There are various identified CO2 fixation pathways, such as the pentose phosphate cycle (Calvin cycle), reductive acetyl coenzyme A pathway (acetyl-CoA), reductive tricarboxylic acid cycle, 3-hydroxypropionate bicycle, and 4-hydroxybutyrate cycles. Most anammox species, including K. stuttgartiensis, Ca. Jettenia asiatica, B. fulgida, and Ca. Scalindua profunda, possess the Wood-Ljungdahl pathway (also the reductive acetyl-CoA pathway) (Gori et al. 2011; Hu et al. 2012). In the anammox process, it is proposed that some nitrite is oxidized to nitrate to generate the electrons for CO2 fixation, and the nitrite oxidation might be catalyzed by a nitrate oxidoreductase (NarGH) in K. stuttgartiensis (Strous et al. 2006). For the acetyl-CoA pathway, it should occur by electron transfer at very low redox potentials for NAD+ reduction (−0.32 V), CO2 reduction to formate (−0.44 V), and acetyl-CoA synthesis (−0.5 V). Therefore, the Wood-Ljungdahl pathway requires high energy input. Strous et al. (2006) proposed that the high reducing power electrons derived from the hydrazine oxidation to N2 (E0′ = −0.75 V) could be channeled towards NAD+ and CO2 reduction to sustain the carbon fixation.
Organic carbon utilization
Because anammox bacteria are autotrophic, the presence of organic carbon can inhibit their growth. However, anammox species Ca. Ananimoxoglobus propionicus, K. stuttgartiensis, and B. fulgida are all shown to be able to co-metabolize fatty acids (Guven et al. 2005; Strous et al. 2006; Kartal et al. 2007b, 2008). Huang et al. (2014) found that Ca. Jettenia asiatica could grow at low acetate (≤120 mg/L)/propionate (≤200 mg/L) concentrations, while the acetate concentration of no more than 240 mg/L caused the decrease in ammonium consumption rate by 33% and by 29% for propionate with <400 mg/L. In addition, Kangwannarakul et al. (2018) found that the short term addition of 0.25 and 0.5 mM acetate did not affect the anammox activity, while the long term addition of 0.25 mM acetate could decrease the anammox activity.
Acetate could be activated by an acetyl-CoA synthetase-like protein (kustc1128) in a heterologous host, as well as whole cells of K. stuttgartiensis (Russ et al. 2012), which might lead to direct incorporation of acetate into cell biomass by anammox bacteria. However, based on δ13C values of lipids and substrates, it was proposed that acetate might not be directly incorporated into the biomass, but would first degrade into CO2 and then be fixed via the acetyl-CoA pathway (Kartal et al. 2008). By nanometer-scale secondary ion mass spectrometry (NanoSIMS) scanning, Tao et al. (2019) revealed that the enriched J. asiatica could utilize acetate and propionate at a >10 times higher efficiency than bicarbonate incorporation, and acetate and propionate were likely not assimilated directly, but were first oxidized to CO2 for the follow-up autotrophy. Both B. sinica and J. caeni possess AMP-Acs and ADP-Acs, which can both catalyze acetate into acetyl-CoA via different mechanisms. ADP-Acs catalyzes the synthesis of acetyl-CoA from acetate in a single step, while AMP-Acs synthesizes it in two steps (Starai & Escalante-Semerena 2004). The AMP-Acs route is a high affinity pathway, and the reaction occurred at a low acetate concentration (Krivoruchko et al. 2015). Russ et al. (2012) found that an AMP forming acetyl-CoA synthetase gene (acs) of K. stuttgartiensis could be functionally expressed to convert acetate to acetyl-CoA in Escherichia coli.
The utilization of organic carbon can benefit the growth of some anammox bacteria. For instance, B. fulgida cells had a high metabolic capability to oxidize acetate (Kartal et al. 2008), and they could outcompete other coexisting anammox bacteria with the addition of acetate, NH4, NO2, and NO3 (Winkler et al. 2012; Jenni et al. 2014). The acetate addition could trigger the conversion of adenosine triphosphate (ATP) to adenosine monophosphate (AMP) in Ca. Brocadia (Feng et al. 2018). In addition, the biomass of B. fulgida showed an increase under certain organic carbon to nitrogen (C/N) ratios (Jenni et al. 2014). On the contrary, Ca. Jettenia asiatica showed no superiority in growth under mixotrophic conditions (Huang et al. 2014).
Furthermore, organic carbon can be utilized as the electron donor for nitrate reduction in anammox bacteria, which is called the dissimilatory nitrate reduction to ammonium (DNRA) (Winkler et al. 2012). A. propionicus and B. fulgida were confirmed to oxidize propionate and acetate in NO3-N reduction, and dominated in ecosystems where propionate or acetate are constantly available (Kartal et al. 2007b, 2008). Under low C/N ratios, anammox bacteria, such as B. sinica, could be promoted or even co-cultured with heterotrophs, which was due to the high rate of partial DNRA (Shu et al. 2016; Castro-Barros et al. 2017).
MICROBIAL KINETICS AND INTERACTIONS
Theoretically, under steady-state conditions, approximately 11% of the nitrogen load is converted to nitrate. Kowalski et al. (2019b) found that as much as 60% of the nitrogen load could be oxidized to nitrate in a non-aerated and lid-covered reactor, indicating that some nitrate would be produced by a shift in the anammox metabolism towards NO3. The constitutive expression of nitrite oxidoreductase (NXR) in anammox bacteria might enable them to use NO2 immediately when the energy source becomes available, causing the increased production of nitrate under anoxic conditions. In addition, it should be mentioned that NXR is not limited to obligate NOB (sNOB), but is also found in physiologically diverse bacteria (dNOB), and sNOB and dNOB might both contribute to the excess nitrate production in anammox systems (Sorokin et al. 2012; Daims et al. 2016; Li et al. 2020).
In addition, production of microbial composition such as extracellular polymeric substances (EPS) and soluble microbial products (SMP) requires investment of energy derived from the substrate utilization (Oshiki et al. 2011). The planktonic B. sinica cells could produce 109 μg extracellular protein/mg volatile suspended solids (VSS) and 29 μg extracellular carbohydrate/mg VSS at the stationary growth phase, which were higher than the production found during the exponential growth (Zhang et al. 2017b). Zhang et al. (2017b) found that the allocation of energy to the EPS production may hinder the planktonic anammox bacteria from growing at the maximum rate, whereas immobilized cells could use most of the energy for cellular growth.
The doubling time of anammox bacteria directly affects their growth and the start-up of the anammox system (Ali & Okabe 2015). Anammox bacteria grow slowly, with typical doubling times ranging from one to several weeks (van der Star et al. 2008; Lotti et al. 2014a). Zhang et al. (2017a) obtained the result showing that when free-living planktonic B. sinica and J. caeni cells were immobilized in polyvinyl alcohol and sodium alginate gel beads and cultivated in an up-flow column reactor with high substrate loading rates at 37 °C, the μmax was determined to be 0.33 1/d and 0.18 1/d (corresponding the doubling time of 2.1 d and 3.9 d), respectively. For specific types of anammox bacteria, the doubling times of genus Brocadia varied from 1.6 to 7 d (Isaka et al. 2006; Tsushima et al. 2007a; Bae et al. 2010; Oshiki et al. 2011; Lotti et al. 2014b; Chi et al. 2018), depending on growing conditions (temperature and ammonium levels). However, other anammox bacteria might have high doubling times, such as Ca. Jettenia at 14.4 d, Ca. Scalindua at 14.4 d, and Ca. Anammoximicrobium at 32.1 d (Awata et al. 2013; Khramenkov et al. 2013; Ali et al. 2015a). Therefore, enrichment of Ca. Brocadia may result in a quick start-up of the anammox process. On the one hand, by applying a low sludge retention time (SRT), anammox bacteria with a low doubling time may be enriched. However, due to the relatively slow growth rates of the anammox bacteria and low biomass yield, maintaining a high SRT is important for system performance (Trigo et al. 2006; Chamchoi & Nitisoravut 2007). Therefore, cascade SRTs could be adopted, with initial low SRTs for selecting anammox bacteria with high growth rates and then high SRTs for biomass concentration enhancement (Miao et al. 2018).
The half-saturation constant (Ks) value for nitrite was 0.48 ± 0.29 mg N/L for B. sinica, 0.50 ± 0.013 mg N/L for J. caeni, and 0.0063 mg N/L for Ca. Scalindua sp. (Oshiki et al. 2016). Additionally, the inhibition constant (Ki) value for nitrite was <224 mg N/L for Ca. B. sinica, 154 mg N/L for J. caeni, and 105 mg N/L for Scalindua sp. (Oshiki et al. 2016). Based on the above kinetic coefficients, to obtain exponential growth of anammox bacteria, the nitrite concentration must be maintained at least in the range between two times its Ks value and half of the Ki value; for example, 0.96–112 mg N/L for B. sinica, 1.0–78 mg N/L for J. caeni and 0.0032–53 mg N/L for Ca. Scalindua sp. (Zhang et al. 2017b). This should be considered for enriching anammox bacteria or initiation of the anammox system.
Many environmental factors can inhibit the anammox reaction. Yang et al. (2018a) found that the anammox reactor could operate well with free ammonia (FA) at 13.65 ± 2.69 mg/L and free nitrous acid (FNA) at 39.49 ± 10.95 μg/L, while it was inhibited when FA and FNA concentrations reached 29.65 mg/L and 77.02 μg/L, respectively. After high substrate shocking, the abundance of Ca. Brocadia decreased while that of Ca. Jettenia increased (Yang et al. 2018a). In addition, Yang et al. (2018a) observed that overdoses of calcium or magnesium had adverse effects on the operation of anammox reactors by inhibiting anammox activity.
Microbial loading rates or activities
The applied loading rates or microbial activities affect the reactor size and the process footprint. Generally, low loading rates have been applied in the anammox process. For example, Zhang et al. (2016) reported specific anammox activity (SAA) values between 0.3 and 0.5 g N/g VSS·d. However, high loading rates or activities have been also obtained. For example, Xu et al. (2019) found that the SAA of anammox granules could reach up to 5.6 g N/g VSS·d, while the NLR and NRR (nitrogen removal rate) could be 76.7 kg N/m3·d and 70.0 kg N/m3·d, respectively. Tang et al. (2010) found that with dominant Ca. Brocadia, NRR of 11.7 kg N/m3·d, SAA up to 0.7 g N/g VSS·d, and high biomass concentration of 28.4 g VSS/L were achieved with the efficient anammox sludge granulation. Jetten et al. (2009) proposed that the large membrane surface area serves to accommodate more respiratory proteins and leads to a higher maximum growth rate. Furthermore, anammox bacteria might adjust their activated membrane surface area to the substrate availability to cope with a large range in growth rates. By this means, when ample substrate is available, all membrane surface area would be activated, while when substrate is limited, only part of the internal membranes would be activated and the bacterium would still be capable of energy conservation (Jetten et al. 2009). In addition, Ni et al. (2019) found that in the expanded granular sludge bed reactor (EGSB) and the parent SBR, the NRR was 0.61 vs. 0.99 kg N/m3·d, which caused the dominant anammox bacterial genus to shift from Ca. Kuenenia to Ca. Brocadia.
The temperature tolerance of anammox activity has been reported to be dependent on the species of the anammox bacteria (Magrí et al. 2013). For example, the marine anammox bacteria Ca. Scalindua favor lower temperatures for their growth compared to the wastewater anammox species (van de Vossenberg et al. 2008; Awata et al. 2013). Additionally, anammox bacteria have been observed to be able to alter their lipid membrane to adapt to temperature (Rattray et al. 2010). The maximum activity of the anammox reaction was observed from 35 to 40 °C; when the temperature was raised gradually, the anammox activity showed an irreversible decrease at 45 °C due to biomass lysis (Dosta et al. 2008). In addition, at a very low temperature (15 °C), the anammox system became unstable due to nitrite accumulation, thereby affecting the anammox activity (Dosta et al. 2008). In addition, in a moving bed biofilm reactor (MBBR), the inhibition of anammox activity at 10 °C was observed (Persson et al. 2014). On the contrary, it was also reported that B. fulgida enriched from wastewater treatment plant sludge could survive at 10 °C (Hendrickx et al. 2014). Reino et al. (2018) successfully operated a B. anammoxidans anammox reactor at 11 °C, treating low-strength synthetic influent or nitrite-amended pre-treated real urban wastewater, and the achieved NLR was 1.8 ± 0.1 g N/L·d and 1.2 ± 0.5 g N/L·d, respectively. In the study of Lotti et al. (2014b), pre-treated municipal wastewater was treated by the anammox process with a dosage of nitrite; with the enrichment of B. fulgida, volumetric NLR higher than 0.4 g N/L·d and biomass specific NRR of 50 ± 7 mg N/g VSS·d were obtained at 10 °C (the doubling time was around 35, 77 and 132 d at 20, 15, and 10 °C). In particular, the up-flow anaerobic sludge blanket (UASB) reactor with plug-flow hydrodynamics and high sludge retention capabilities could improve the anammox activity at low temperatures considerably despite the mass transfer limitations (Reino et al. 2018).
Fernández et al. (2012) reported 50% inhibition in the SAA at the FA concentration of 38 mg NH3-N/L. The optimal concentration of FA to maintain stable operation of a granular reactor was found to be less than 20–25 mg NH3-N/L. Tang et al. (2014) found that in biofilm reactors, only concentrations as high as 57–187 mg NH3-N/L caused inhibitory effects, suggesting that biofilm reactors are more resilient to FA inhibition. However, Jaroszynski et al. (2011) reported much higher anammox performance at a pH of 6.5 with the average FA concentration of 0.4 ± 0.3 mg NH3-N/L than at a pH of 7.8 ± 0.2 with the bulk FA averaging 4 ± 3 mg N/L.
Microbial interactions and their contribution to enhanced nitrogen removal
In the anammox-based nitrogen removal systems, diverse microbial interactions exist. Generally, in oxygenated environments, AOB may provide nitrite for anammox bacteria, while NOB may compete for nitrite. Cooperation between anammox bacteria and AOB has been confirmed (Third et al. 2001; Schmidt et al.2002a, 2002b; Sliekers et al. 2002; Vlaeminck et al. 2007). In anoxic environments, nitrate-reducing bacteria may produce nitrite for anammox bacteria under electron donor limitation conditions, while denitrifying microbes will also compete for nitrite when sufficient electron donors are available. The DNRA process could supply anammox bacteria with the necessary ammonium (Kartal et al. 2007a). Furthermore, Tao et al. (2013) found that some heterotrophs could compete with denitrifiers by mineralizing organic compounds faster than denitrifiers, which played a critical role in sponsoring anammox bacteria.
Ye et al. (2018) found that AOB could survive in the anammox reactor, but had extremely slow growth rates. For example, the anaerobic activity of Nitrosomonas eutropha was approximately 50-fold slower than the dedicated anaerobic ammonium oxidizer Ca. Brocadia anammoxidans, and more than 200 times slower than the aerobic activity of Nitrosomonas eutropha itself (Jetten et al. 2001). Since 2015, when Daims et al. (2015) demonstrated that Nitrospira sp. could perform both nitrification stages, this phylum has been observed to be relatively abundant in anaerobic ammonia oxidizing systems (Pinto et al. 2015; Ciesielski et al. 2018; Ziembińska-Buczyńska et al. 2019). All complete nitrifiers identified to date belong to sublineage II of the genus Nitrospira (Daims et al. 2001; Lebedeva et al. 2011). Under certain conditions, Nitrospira may not compete with the anammox bacteria, but supports anammox bacteria by providing nitrite through canonical ammonia oxidation (van Kessel et al. 2015; Ciesielski et al. 2018).
In the autotrophic nitrogen removal systems, washout of NOB is the key aspect for maintaining the anammox activity. Distribution between flocs and biofilm could be a good strategy to achieve the washout of NOB from systems. Laureni et al. (2019) found that floc removal is an effective operational strategy to achieve selective washout of NOB, and hybrid systems rather than solely biofilm systems would be more flexible in controlling NOB in mainstream nitritation and anammox (PN/A) applications. By separating NOB and anammox bacteria in flocs and biofilm, the direct competition for NO2 between NOB and anammox bacteria was identified as key mechanism leading to a difference in the actual growth rates of AOB and NOB (mNOB < mAOB in flocs) and allowing the selective NOB washout over a broad range of simulated SRTs (6.8–24.5 d) (Laureni et al. 2019). In addition, by separating microbial communities between biofilm and flocs in the anammox system, denitrification activity present in flocs can protect anammox activity in biofilm, resulting in high ammonia removal efficiency and resistance to high organic loadings (Yang et al. 2019). In the study of Lotti et al. (2015a), the evaluation of the process in a plug-flow granular sludge-based pilot-scale reactor (4 m3) was conducted at 19 ± 1 °C, which was continuously fed with the actual effluent of the A-stage of the wastewater treatment plant (WWTP), and anammox bacteria were able to grow under mainstream WWTP conditions and new granules were formed and efficiently retained in the system, while heterotrophic biomass grew preferentially in flocs and was efficiently washed out of the system.
Microbial interactions could be incorporated into different processes to enhance nitrogen removal. For example, Xie et al. (2018) applied a novel technology integrating the anammox and denitrifying anaerobic methane oxidation (DAMO) reactions in a membrane biofilm reactor (MBfR), and an effluent total nitrogen (TN) concentration below 3.0 mg N/L and the TN removal rate of 0.28 kg N/m3·d could be achieved. In this process, 30–60% of the nitrate produced during anammox was reduced back to nitrite by DAMO archaea, and then the produced nitrite was removed by the anammox and DAMO bacteria with contributions of >90% and <10%, respectively (Xie et al. 2018).
START-UP OF THE ANAMMOX SYSTEM
The start-up of the anammox system is the bottleneck for its practical application, especially under situations without adequate seeding biomass. The first full-scale granular anammox reactor was started up successfully after 3.5 years' operation, which was longer than the planned 2 years, and the possible reasons were concluded (Abma et al. 2007). First, operational stability could have been affected by incidental biomass loss and freezing problems; second, microbial toxicity could have been caused by high nitrite concentrations, possible methanol slip from the Sharon reactor, and incidental discharges from chemical toilets to the digester; finally, mixing problems might have caused dead zones inside the reactor and inhibition from the formed sulfide (Abma et al. 2007). After start-up, the reactor operated stably even at the loading rate of 10 kg N/m3·d, which could be due to the formation of dense anammox granules with settling velocities higher than 100 m/h (Abma et al. 2007).
Because anammox bacteria are slow growers with a doubling time of 11 days (Strous et al. 1998), the successful start-up and operation of an anammox system must maintain a high concentration of anammox bacteria and enhance their activities (stimulation or avoiding inhibition). During anammox reactor start-up, four consecutive phases of cell lysis, lag phase, activity elevation, and stationary occur (Tang et al. 2013). Many studies have been conducted to enhance the start-up of the anammox system through controlling the four phases. For example, signal substances can be dosed to enhance the microbial activity and also reduce the lag phase (Zhao et al. 2018). Generally, several types of strategies can be summarized, including microbial source and kinetic selection, stimulation with the addition of specific substances, and biomass retention with biofilm or membrane-based systems (Figure 3).
Start-up based on microbial selection and kinetic control
The slow growth rate of anammox bacteria in combination with inhibition and operational problems makes the start-up of the anammox process difficult. Therefore, suitable inoculum should be selected. In addition, suitable nutrient concentrations should be provided to maintain or enhance the microbial reaction activity, while simultaneously avoiding microbial inhibition. These can be achieved through choosing suitable microbial sources and selectively acclimating anammox bacteria with different microbial kinetics (such as the half saturation coefficients, inhibition coefficients and the maximum growth rate).
The inoculum can affect not only the amount of the anammox bacteria, but also the types of anammox bacteria. If an adequate amount of anammox biomass is available, it can be used as the seed directly. On the contrary, without adequate anammox biomass sources, fast start-up of full-scale anammox process involves the stepwise cultivation of mature anammox biomass in laboratory- and pilot-scale reactors, which are subsequently switched to inoculate the full-scale reactor (Wett 2006; van der Star et al. 2007). Tang et al. (2013) proposed that the mixed inoculation with nitritation sludge as the main inoculum could be adopted as an efficient approach for full-scale anammox process start-up, where the obvious anammox activity appeared on 22 d when the reactor was seeded with nitritation sludge, which was only 1/3 of that inoculated with anaerobic granular sludge. Usually, Ca. Brocadia was the most dominant genus present in the enrichments transited from nitrifying sludge, denitrifying sludge, and anaerobic granular sludge, while Ca. Scalindua existed in the enrichment seeded with marine sediments and Ca. Jettenia from river sediments (Tang et al. 2010). By combining the anaerobic baffled reactor and MBR, starting up the cold-anammox process (13 °C) was achieved after 75 d, 45 d, and 90 d through inoculating flocculent nitrification sludge, anaerobic granular sludge and flocculent denitrification sludge, and the anammox species (Ca. Brocadia caroliniensis, B. sinica and Ca. Jettenia asiatica) with large maximum growth rates contributing to the rapid start-up of the cold-anammox process (Wu et al. 2018).
Anammox bacteria were partially inhibited at 50 mg NO2-N/L and totally inactivated at 100 mg NO2-N/L (Jetten et al. 2001). It is important to maintain the system NO2-N concentration below the toxic level (Pyanert et al. 2004), especially during the start-up stage. To alleviate inhibition, mixing in the anammox reactor is a strategy to avoid inhibition induced by high nitrite levels or sulfide formed in dead zones (Strous et al. 1999b). The rapid increase in nitrite could inhibit anammox activity, and gradually increasing nitrite should be adopted as another strategy for the operation of full-scale reactors (Ni et al. 2011). The feeding strategy can also affect the selection of certain types of anammox bacteria. For example, the feeding strategy with a relatively low nitrite concentration to prevent nitrite inhibition is also a selective approach for Ca. Brocadia, which could not be enriched in high nitrite fed reactors (Gaul et al. 2005; van der Star et al. 2008). However, to increase the removal loading rate, increasing the influent nitrite concentration combined with the stepwise reduction in hydraulic retention time (HRT) could be adopted (Sliekers et al. 2003). Besides nitrite, high ammonium and nitrate concentrations of 25 and 50 mM could also inhibit anammox bacteria activity (Dapena-Mora et al. 2007), which should be considered during anammox system start-up, especially for treating ammonia-rich wastewater.
Start-up with stimulation
Microbial activities can be enhanced by increasing the activity of the electron transport chain, including the enzyme activity and electron transfer activity. As discussed below, this can be achieved through nutrient ensurance, intermediate enhancement, and electron transfer enhancement.
In addition to nitrogen macronutrient, trace element nutrients are very important for microbial activities. In particular, some trace elements are key co-factors of functional enzymes. Bi et al. (2014) found that the start-up time of the anammox process could be shortened from 70 to 58 d with 0.06 mM Fe2+ and 50 d with 0.09 mM Fe2+. For anammox bacteria, insertion of a ferrous iron atom into the porphyrin macrocycle by the enzyme ferrochelatase creates heme, which can provide catalytic and electron transfer operations, and serve a critical role in the assembly of major enzyme complexes, including HDH (hydrazine dehydrogenase), HZS (hydrazine synthase), NirS (nitrite reductase), hydroxylamine oxidoreductase (HAO), and cytochrome bc1 complex, among others. By dosing ferrous iron in the medium, the assembly of such enzymes could be enhanced for improving microbial activities directly.
During anammox, many intermediates are produced, and some intermediates can provide more energy for anammox bacteria. High energy supply is very important for the anammox bacteria due to the low energy production from anaerobic reactions. Ganesan & Vadivelu (2019) found that with the addition of 10 mg/L hydrazine, only 7 weeks were needed to stabilize and successfully operate the anammox process, whereas 12 weeks were necessary without the addition of hydrazine. Externally added hydrazine provides a greater energy source once metabolized, which can be used by the anammox bacteria (Yao et al. 2015).
Reduced graphene oxide (RGO) was reported to have a greater ability of electron transfer than graphene oxide by approximately three orders of magnitude, with the TN removal rate and enzyme activity increasing by 10.2% and 1.5–2 fold, respectively (Wang et al. 2013). By dosing RGO to the up-flow column reactor, the start-up period of the anammox process could be shortened from 67 to 49 d, which also enhanced anammox activity and stability even against the high NLR impacts (Yin et al. 2016). Based on the RGO biotransformation analysis, the applied stimulation may be strongly associated with the excellent electron transferability of RGO and its performance as a redox mediator (Yin et al. 2015, 2016). RGO could participate in the electron transfer from hydrazine dehydrogenase to cytochrome bc1 complex due to the faster electron transfer ability. Thus, with the addition of RGO, the enhancement of enzyme activity directly accelerated ATP synthesis and catabolism of anammox biomass. In addition, RGO could act as a scaffold for bacteria attachment (Ruiz et al. 2011), which can help anammox bacteria form macroflocs for enhancing the cell density, as anammox bacteria were not active until the cell concentration was higher than 1010–1011 cells/mL (Strous et al. 1999a). In addition, by the dosage of 0.1 g/L of graphene oxide, greater EPS production was obtained accompanied with a maximum increase of 10.26% in anammox activity (Wang et al. 2013).
The observed microbial activity is the cooperation among all system microorganisms rather than just individual activity. Quorum sensing is a mechanism responsible for the microbial interactions within the microbial ecology, which can enhance microbial activity and gene expression. Quorum sensing is realized through the exchange of signal substances. By adding C12-HSL-containing supernatant (signal substance) into the continuously stirred tank reactors, the start-up time of the anammox process was reduced from 80 to 66 days, and the NLR was also enhanced to 1.6 times that of the control reactor (Zhao et al. 2018). The possible reason is that signal substances could increase the secretion of EPS, resulting in better enrichment of anammox bacteria (Zhao et al. 2018).
Start-up with biomass retention
The limiting factor in fast start-up of biofilm reactors is not only the activity but also the ability to retain anammox biomass in the system and increase the attachment of anammox to the carrier material (Kowalski et al. 2019a). Many factors can affect the sludge settlement properties. For example, Wett (2007) found that nitrite concentration above 10 mg/L could deteriorate the settlement of suspended sludge, with the sludge volume index above 170 mL/g. In conventional reactors, strategies can also be adopted to enhance biomass settlement. For example, by seeding mixed activated sludge to start-up an anammox SBR with settling option (gradually reducing the setting time to 10 min), successful start-up of the anammox process was achieved, and the NLR was up to 506 g N/m3·d and total NRR reached 433 g N/m3·d by enriched Ca. Brocadia (Ye et al. 2018). In addition, MBR can also be adopted to retain biomass.
In biofilm systems, biofilm carriers and strategies to enhance microbial attachment should be developed to enhance biofilm growth. By comparing biofilm carriers of sponge, volcanic rock, and charcoal, Lu et al. (2018) found that using porous material as a carrier for biofilm development is an effective strategy for practical application of the anammox reactor, which can enhance biomass attachment and also create better anaerobic conditions for anammox bacteria. By using MBBR with a novel composite carrier, where the zeolites and floating materials were combined in the spherical shell and distributed evenly by the spherical polyhedron, the PN/A process could be realized in 53 days, and the TN removal efficiency reached around approximately 85% at an influent ammonium concentration of 50 mg/L (Lv et al. 2019). Kowalski et al. (2019b) discovered that rapid attachment of the anammox biomass was achieved in a reactor with media that had a predeveloped layer of a heterotrophic biofilm, and the SAA increased by almost 400% as compared to seed values. For the integrated PN/A process, anammox bacteria growth on biofilm carriers can be achieved first and then enrich nitrifiers, which can shorten the start-up duration (Zekker et al. 2013; Feng et al. 2019). In addition, the control strategy of the anammox process should be different for ammonia-rich wastewater and municipal wastewater. For example, FA could be applied for inhibiting NOB when treating ammonia-rich wastewater (Zekker et al. 2013). However, cascade oxygen supply can be applied when treating municipal wastewater with low ammonia concentrations (Feng et al. 2019).
For anammox granule formation, UASB reactors could be used. Even in UASB reactors, the problem of anammox sludge washout may occur due to the higher HRTs than the designed hydraulic loading (ca. HRT 4 h), which can be overcome by collecting and returning the washed out biomass (Li et al. 2012). Abma et al. (2007) found that fluctuations in the up-flow velocity caused incidental biomass losses, which should be removed efficiently through careful increments of the up-flow velocity to promote granule formation. In addition, inside the UASB reactor, floating granule accumulation and continuous adhesion on the edge of the three-phase separator are frequently encountered and block the gas vent and the effluent pipe, leading to severe sludge decay and loss, and even deterioration of reactor operation (Ni et al. 2019). To solve the biomass loss in conventional EGSB, a novel three-phase separator configuration was incorporated with an anammox granule circulating EGSB, and achieved stable operation of anammox processes by promoted granules circulation, retention, and reaction (Ni et al. 2019).
For membrane-based systems, suspended or biofilm-based biomass can be incorporated. Ni et al. (2010) used the anammox nonwoven membrane reactor to form aggregates in the reactor and biofilm on the interior surface of the non-woven membrane, and the NLR and NRR reached 1,263 mg N/L·d and 1,047.5 mg N/L·d, respectively, with a maximum specific ammonium consumption of 51 nmol/mg protein·min after eight months of operation. In an up-flow fixed-bed biofilm column reactor with nonwoven fabric sheets as the biomass carrier, the anammox reaction (Ca. Brocadia anammoxidans was observed within 50 days, and a total NRR of 26.0 kg N/m3·d (specific NRR of 1.6 kg N/m3·d) obtained was attributed to the high anammox bacteria density (ca. 16 g VSS/L, more than 70% of total bacteria were anammox bacteria) (Tsushima et al. 2007b).
In biofilm-based anammox systems, nutrient diffusion is the key aspect for system performance, and the high NLR could prevent substrate transport limitation inside the biomass (Nicolella et al. 2000). For example, high anammox activity in up-flow column reactors was evenly observed throughout the immobilized gel beads due to faster and deeper substrate transport (Ali et al. 2015b). Due to the high relatively effective diffusivity, anammox reactors using artificially immobilized biomass tend to exhibit higher nitrogen removal performance than ones using naturally aggregated granules (Ali et al. 2015b). A reactor containing gel beads with biomass concentration of 0.33 g VSS/L achieved a NRR of 10.8 kg N/m3·d and SAA of 278.5 ± 30.9 μmol-29N2/g VSS·h in just 35 days, whereas the reactor containing granular biomass of 2.5 g VSS/L could achieve only a NRR of 3.5 kg N/m3·d and SAA of 184.7 ± 30.9 μmol-29N2/g VSS·h (Ali et al. 2015b). In biofilm systems, biomass concentration is important, and specific surface area is more important due to substrate transport limitation. This was also confirmed by Zhu et al. (2018), where the optimal granule sludge size was 0.5–0.9 mm for enhanced anammox bacteria abundance, activity and specific reaction rate.
Anammox-based biological processes play an important role in the sustainable management of wastewater, especially for nitrogen control. Anammox bacteria are slow growers with diverse phyla, which can be enriched from different environmental and operational conditions. The isolation of anammox bacteria should consider the relationship of microbial interaction from material metabolism, electron transfer and also information exchange. Suitable organic carbon contents may enhance activities of certain types of anammox bacteria, which should be carefully considered when treating real wastewater. The reaction stoichiometric coefficient can be affected by loading rates and other biological reactions. Microbial interactions also contribute to the enhanced biological nitrogen removal and promote activities of anammox bacteria, and functional microorganisms should be balanced for system stability. The start-up of the anammox process is the key aspect for its practical application, which can be realized through seed selection, system stimulation, and biomass concentration enhancement. Incorporation of the anammox process could achieve the sustainable management of wastewater, especially for the reclamation of water, recovery of energy and fertilizer, and also the control of pollutants.
This research was supported by the National Natural Science Foundation of China (No. 51738005).
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.