Chemical Engineering Journal 431 (2022) 134087 Contents lists available at ScienceDirect Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej Coupling anammox and heterotrophic denitrification activity at mainstream conditions in a single reactor unit Xènia Juan-Díaz, Lluc Olmo, Julio Pérez *, Julián Carrera GENOCOV Research Group, Department of Chemical, Biological and Environmental Engineering, School of Engineering, Universitat Autònoma de Barcelona, Ed. Q- Campus UAB, Spain A R T I C L E I N F O A B S T R A C T Keywords: Mainstream partial nitritation/anammox (PN/AMX) has attracted large attention in the last decade. Two-stage Mainstream anammox configurations have been pointed out as an appealing technology for the application of the PN/AMX process Heterotrophic denitrification at mainstream conditions. However, the process requires of an efficiency improvement by designing a new Organic carbon technology to remove the nitrate produced by anammox bacteria. A new reactor configuration was developed by Cooperation network Single reactor unit coupling anammox to heterotrophic denitrification in a single reactor unit while avoiding competition between Effluent quality both processes. The addition of acetate as an external C–source allowed the removal of nitrate by heterotrophic microorganisms without hampering the anammox process by maintaining high nitrogen removal rates (0.16 ± 0.03 g N L–1 d–1) and nitrogen removal efficiencies (91 ± 8 %) (on average from acetate addition periods). Further, a proper organic load management showed to be effective to face the drawbacks derived from previous mainstream treatment stages (i.e. undesired nitrate production), without compromising effluent COD concen­ trations. By withdrawing samples along the sludge bed, nitrogen compounds and COD concentrations were determined at different reactor heights. This set of experimental data demonstrated that the combination of anammox and heterotrophic denitrification in a single reactor unit was possible as anammox activity dominated within bottom sludge bed sections, while heterotrophic denitrification occurred within middle and upper sludge sections. Microbial diversity results of 16S rRNA gene-targeted sequencing analyses confirmed that anammox and heterotrophic denitrifiers communities occupied two differentiated sludge bed sections along the reactor being dominated by Candidatus Brocadia (30 %) and Thauera (33 to 46 %), respectively. When heterotrophic denitrification occurred, the average N2O emissions with COD addition was ca. 50 % lower than that in periods without COD addition. 1. Introduction captured within the A-stage via a high-rate activated sludge (HRAS) reactor in which a short SRT is imposed to enhance the removal of COD Anammox bacteria oxidize ammonium by using nitrite as electron through adsorption and to enhance higher yields of biogas production acceptor to produce nitrogen gas [1]. The implementation of the from the digestion of the obtained sludge [2,3,7]. The remaining ni­ anammox process at mainstream conditions offers the possibility to trogen is autotrophically removed in the B–stage by the partial nitrita­ move from energy consuming to neutral (or even positive) wastewater tion/anammox (PN/AMX) process. To date, the PN/AMX process treatment plants (WWTPs) [2]. Compared to conventional nitrification/ stability has been reported by using one-stage and two-stage configu­ denitrification, nitrogen removal via the anammox process allows to rations at temperatures as low as 11 ◦ C with biofilms developed on reduce oxygen consumption by a 60 % and does not requires any organic plastic carriers [8,9] and by using granular sludge [10,11], respectively. C–source [3]. The treatment of municipal sewage via anammox-based However, the obtained nitrogen removal rates (NRR) of one-stage con­ processes allows the uncoupling of organic matter (i.e. chemical oxy­ figurations were not competitive compared to conventional treatments. gen demand, COD) and nitrogen removal. A widely accepted configu­ This is mainly due to the fact that this type of biofilm reactors with ration for the implementation of the anammox process at mainstream ammonium oxidizing bacteria (AOB) and anammox growing on inner conditions is the A/B process [4–6]. The incoming organic matter will be surfaces of plastic carriers are strongly rate limited by the external mass * Corresponding author. E-mail address: julio.perez@uab.es (J. Pérez). https://doi.org/10.1016/j.cej.2021.134087 Received 2 July 2021; Received in revised form 4 December 2021; Accepted 7 December 2021 Available online 12 December 2021 1385-8947/© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 transfer rate [12]. For this reason, the use of granular sludge reactors for 2. Materials and methods two-stage PN/AMX configurations might be an appealing alternative to grant process stability and allow for high nitrogen loading rates 2.1. Reactor operation and configuration [10,11,13–15]. The maximum nitrogen removal efficiency of the PN/ AMX process is limited by the nitrate produced by anammox bacteria An UAnSB reactor with an effective working volume of 13 L, (up to 10 % of the influent nitrogen). The increasing stringent discharge excluding the gas–liquid–solid (GLS) separator phase (12 L), was con­ limitations requires of further increase in the process efficiency by verted into a CANDLE reactor for the combination of the anammox and designing a process that includes the removal of the nitrate produced by heterotrophic denitrifying activity in a single reaction unit. The inner anammox bacteria [16]. The conventional process upgrade would be to diameter of the column was 120 mm and the total-reactor-height to enhance heterotrophic denitrification by adding an external C-source (e. column-diameter ratio was 12:1. The CANDLE reactor was previously g., acetate). An efficient design of this last treatment unit might also operated as an UAnSB reactor for 350 days at temperatures ranging from contribute to mitigate the negative effects of (already identified) process 20 to 10 ◦ C treating a real mainstream wastewater by maintaining a challenges such as the instability of the previous PN reactor during short rather constant NRR (0.10 ± 0.01 g N L–1 d–1) [13]. The hydraulic periods of time (for instance, nitrate production at the lowest temper­ residence time (HRT) was calculated by considering the liquid fraction atures in winter [17]); potential nitrate production in the A-stage during within the sludge bed. The liquid fraction volume corresponded to the warm seasons [18,19]; non-ideal nitrite to ammonium feeding ratios to voidage of the sludge bed (i.e., the fraction of the sludge bed volume that the anammox reactor [13]; or the daily nitrogen loads oscillations of the is occupied by the voids) and was experimentally determined. The main water line [15]. average HRT was 1.0 ± 0.2 h. The reactor had five different sampling An appealing alternative would be to add the external C-source into points at heights of 0 m (S1), 0.05 m (S2), 0.16 m (S3), 0.25 m (S4) and the anammox reactor. The combination of anammox and heterotrophic 0.36 m (S5) (Fig. 2). The main difference (compared to the UAnSB denitrification in a single reactor unit would allow to enhance the reactor) is that the CANDLE reactor presents two inlets with the corre­ robustness of the PN/AMX process at mainstream conditions. Never­ sponding diffusors located (i) in the bottom section to feed a real theless, heterotrophic denitrifying bacteria might deteriorate the mainstream wastewater and (ii) at 0.16 m from the reactor bottom (i.e. anammox process, as they present higher growth rates and could at the height of sampling S3) to feed an external C-source (i.e. sodium outcompete anammox bacteria utilizing nitrite as electron acceptor acetate) (Fig. 2). Reactor temperature was continuously monitored and [20]. Also, high COD/N ratios and long–term exposure to organic sub­ recorded on-line using a Pt1000 sensor (Axiomatic, S.L, Spain) located in strates might result into a loss of anammox activity [21,22]. the core of the sludge bed. CANDLE reactor operation was divided in An ad hoc reactor design has been developed to convert an Up-flow four periods. Anammox Sludge Bed (UAnSB) reactor into a reactor unit with minimal nitrate production (i.e. CANDLE, Coupling ANammox and Denitrifica­ • Period I. To evaluate reactor performance without the addition of an tion in a singLE unit). This reactor therefore would be integrated in the external C-source, i.e. only mainstream wastewater addition at 20 ◦ C. treatment of the main waterline as depicted in Fig. 1. The proposed This period will be used as base line in terms of reactor performance configuration enabled the use of anammox and heterotrophic deni­ for the assessment of the effects of adding an external C-source. trifying bacteria in a single unit by preventing competition for nitrite During this period disturbances were applied in the inflow with between both groups of bacteria. The CANDLE reactor was operated for regards to: (i) the nitrite to ammonium ratio compared to the theo­ 200 days at temperatures ranging from 20 to 14 ◦ C by treating a real retically desired for the anammox process; (ii) the inflow nitrate mainstream wastewater. Substrate distribution along the sludge bed concentrations, mimicking a low performance of the previous sections was investigated to determine the distribution of the anammox treatment stages (e.g., nitrification during carbon removal in the A- and heterotrophic denitrifying activities in the reactor. Nitrous oxide stage and nitrate produced in the PN reactor at low temperatures). (N2O) emissions were monitored, whereas process reactor performance • Period II. Reactor performance was evaluated by adding acetate as was complemented by 16S rRNA gene-targeted sequencing analyses for an external C-source at 20 ◦ C. Additionally, the same disturbances the characterization of the microbial community. described in period I were applied. Further, N2O emissions were Fig. 1. Two potential treatment schemes for the integration of anammox in the main waterline of urban WWTPs. The treat­ ment scheme is based on the A/B configuration, where the A- stage is in charge of the COD removal and the B-stage is in charge of the N removal. The A-stage is a high rate activated sludge (HRAS) reactor to enhance both the COD removal through adsorption and the energy recovery from the excess sludge produced. The B-stage removes the nitrogen with anammox, through two different approaches: A) Two-stage PN/AMX process, splitting PN and anammox in two reactor units, being the second one the focus of research in the present study and named Coupling ANammox and Denitrification in a singLE unit (CANDLE). B) The partial denitritation/anammox (PD/A) process (with either one or two stages) fed with the effluent of a nitrification reactor. 2 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 Fig. 2. (A) Picture of the lab-scale CANDLE reactor (sludge bed section). (B) Schematic diagram of the reactor set-up with the cor­ responding peripheral instrumentation, feeding and sampling points. The fraction of the CANDLE reactor vessel highlighted in solid orange corresponds to the sludge bed. Sampling points are denoted as S1, S2, S3, S4 and S5 and the height from the reactor base is provided for each one of them. (For interpretation of the references to colour, the reader is referred to the web version of this article). assessed under optimal and limited COD conditions for heterotrophic treatment stages. Consequently, variable nitrate concentrations were denitrification. measured in the inflow real mainstream wastewater throughout reactor • Period III. To re-evaluate reactor performance with only mainstream operation, as well as variations in the feeding ammonium/nitrite con­ wastewater addition at 20 ◦ C. This experimental period is designed centration ratio, diverging from the theoretical one desired for the to analyse the potential negative effects of the external COD addition anammox process (see Supporting Information Figs. A.3 and A.2, on the anammox process after stopping the previous acetate addition respectively). (i.e. period II). Additionally, it was also used to assess the N2O emissions without external COD addition, for enabling comparison 2.3. Profile concentrations along the sludge bed with periods in which the external organic source was fed to the reactor. Ammonium, nitrite, nitrate, COD and acetate concentrations were • Period IV. Reactor performance was assessed when adding acetate measured along the different sampling points of the CANDLE reactor at whereas a temperature decrease from 20 to 14 ◦ C was applied, to steady state conditions to assess the distribution and contribution of explore the influence of temperature on both process (anammox and anammox and of heterotrophic denitrification activities. Further, ace­ heterotrophic denitrification) taking place in a single reaction unit. tate concentrations were measured along the sludge bed on day 132. N2O emissions were also evaluated at 17 ◦ C under over-dimensioned Distribution of sampling points is depicted in Fig. 2. COD conditions. 2.4. Heterotrophic denitrifying ex-situ batch activity tests 2.2. Wastewater characteristics Ex-situ batch tests were used to assess heterotrophic denitrifying The CANDLE reactor was fed with real mainstream wastewater activity before and during acetate addition on days 43 (period I) and 98 coming from the effluent of an urban WWTP located in an industrial area (period II), respectively. Tests were carried out in duplicates using of Catalonia, NE Spain, therefore the wastewater contained almost no septum-closed bottles of 125 mL with 2.5 ± 0.3 g VS L–1 and 0.3 ± 0.1 g biodegradable COD. As an effluent coming from a PN reactor at main­ VS L-1 for tests of days 43 and 98, respectively. For tests of day 43, stream conditions was not available at the time of this study, nitrite or biomass from sampling S3 was employed as it corresponded to the ammonium were added according to the needs of anammox stoichi­ sludge bed section where acetate was added. For tests of day 98, the ometry [1]. Therefore, this study is not aiming to validate the perfor­ employed biomass corresponded to the sludge located within the GLS mance of the PN reactor. Ammonium and nitrite were added as separator phase. The distinct sample point explains the difference of (NH4)2SO4 and NaNO2, respectively to maintain a ratio between the biomass concentrations among the performed tests. The biomass was nitrite and ammonium concentrations of 1.2 ± 0.2. Further character­ washed and re–suspended in a medium containing the corresponding ization of the real mainstream wastewater after the ammonium and nitrite and/or nitrate and C–source (i.e. acetate) concentrations together nitrite addition is provided in Supporting Information (Table A.1). The with the macro and microelements to avoid nutrient limitation [23]. real mainstream wastewater was stored in a tank within lab-facilities. All septum bottles were flushed with N2 to maintain anoxic condi­ The storing tank containing the influent was flushed with N2 every tions. Tests were conducted at 180 rpm and at 20 ◦ C. Two sets of ex­ time the tank was refilled (3 – 5 weeks) to keep DO concentration in the periments were performed for each sampling period (days 43 and 98): range of 0.2 to 2 mg O2 L–1 and maintained at 20 ± 1 ◦ C. The DO con­ (1) the consumption of nitrite or nitrate in the presence of acetate as an centration measured in the influent did not negatively affect the per­ easily biodegradable C-source and (2) the consumption of nitrite and formance of the anoxic processes occurring in the reactor (i.e., anammox nitrate in the presence of acetate to evaluate if the biomass presented a and heterotrophic denitrification). These potential negative effects preference for any of these two electron acceptors. Sampling time could be ruled out because the performance of the reactor was not depended on the consumption rate of substrates in each experiment. altered when the DO concentration in the tank, where the inflow Substrate consumption rates were calculated by linear regression of wastewater was stored, changed from 0.2 to 2 mg O2 L-1. The stored nitrite or nitrate concentrations of three to six bulk liquid-phase grab wastewater experienced some degree of nitrification, which allowed to samples. Specific heterotrophic denitrifying activities (mg N g VS–1 d-1) simulate a potential nitrate production of the previous mainstream were calculated by dividing the substrate consumption rates (mg N L-1 d- 3 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 1 ) by the biomass concentration (g VS L–1) of each performed test. The emission rates (Eq. (3)) to the relative nitrogen removed by the anam­ employed COD/N ratio (4.2 ± 0.4 g COD g-1N-NO3-) was slightly higher mox process or by both the anammox and heterotrophic denitrifying than the theoretical stoichiometric relationship for the biological het­ processes during a defined period. erotrophic denitrification process using acetate as a C–source (3.9 g COD rN2 O = QN2 ⋅CN2 O (3) g–1N-NO3-) (Eq. (1), [24]). 2.3NO−3 + 2.1CH3 COO− + 4.4H+ →CH1.8 O0.5 N0.2 + 3.1CO2 + 4.4H2 O where CN2 O corresponds to the measured N2O off-gas concentrations (mg N-N2O L-1) and QN2 corresponds to the nitrogen gas flow (13 L d-1) + 1.1N2 added within the GLS separator phase (see Fig. 2) and controlled by a Further information regarding COD, nitrite and nitrate concentra­ means of flowmeter (Bronkhorst, The Nethelands). tions for each performed ex-situ batch test can be found in Supplemen­ Nitrogen gas production by the anammox and heterotrophic deni­ tary Information (Table A.2). trification processes were not considered in QN2 as they were 7-to-10 times lower than the external nitrogen gas supply. 2.5. Analytical methods 2.7. Microbiological quantification and characterization Liquid samples from influent and effluent of the CANDLE reactor were analysed 4 to 5 times per week. Samples were filtered (0.22 μm) Microbiological community composition was identified by using before analysis. Nitrite (influent) and nitrate (influent and effluent) next-generation sequencing analysis. Two biomass samplings were car­ concentrations were analysed off-line with ionic chromatography using ried out: (i) before acetate addition (day 42) from samplings S1 and S3 ICS–2000 Integrated Reagent-Free IC system (DIONEX Corporation, and (ii) during acetate addition (day 97) from sampling points S1, S3, S4 USA). For accurate results, effluent nitrite concentrations were and from the GLS separator phase. DNA was extracted by using the Soil measured by using colorimetric Hach Lange kits (LCK342, Hach Lange, DNA Isolation Plus KitTM (Norgen Biotek Corp, Canada) following the Germany). Ammonium concentrations (influent and effluent) were manufacturer protocol. The quantity and quality of the extracted DNA analysed off-line by means of a gas selective electrode (GSE) (AMTAX sc, was measured by using the NanoDrop 1000 Spectrophotometer (Thermo Hach Lange, Germany). Ammonium and nitrite samples from profile Fischer Scientific, USA). A 260/280 nm ratio of 1.9 ± 0.1 was used as concentrations and from ex-situ batch activity tests were determined by quality cut-off and a minimum of 17 ng μL− 1 of extracted DNA was colorimetric Hach Lange kits (LCK303 and LCK342, respectively, Hach guaranteed to perform sequencing. Paired-end sequencing of the Lange, Germany). Total and soluble COD concentrations were analysed extracted DNA was performed on an Illumina MiSeq platform by the by using colorimetric Hach Lange kits (LCK314 Hach Lange, Germany). Research and Testing Laboratory (Lubbock, Texas, USA). Bacterial 16S Acetate concentrations were measured by gas chromatography (7820A, rRNA variable regions V2–V4 were targeted using the primer pair 515F- Agilent Technologies, USA). Total solids (TS) and volatile solids (VS) 806R for general bacteria and the specific primer pair 368F-820R for the concentration were analysed according to Standard Methods [25]. anammox population. Additional information of the bioinformatics Biomass characteristics were monitored using a stereomicroscope. protocol can be found in Supporting Information (A1.3). Average particle size was measured by a laser diffraction analysis system (Malvern Mastersizer Series 2600, Malvern instruments Ltd., UK). N2O 3. Results and discussion off-gas concentrations were analysed within the GLS phase of the CANDLE reactor with a Clark-type N2O microsensor calibrated accord­ 3.1. Combining anammox and heterotrophic denitrifying activity in a ing to manufacturer’s instructions (UNISENSE, Denmark). single reactor unit: Process performance The CANDLE reactor configuration was proposed to couple anam­ 2.6. Calculations mox and heterotrophic denitrification processes in a single reactor unit. To prevent the competition among anammox and heterotrophic deni­ Nitrogen loading rates (NLR), NRR and organic loading rates (OLR) trifying bacteria, an external C-source was fed (i.e. acetate) to the middle were calculated by using the effective CANDLE reactor volume of 13 L, sludge bed section (i.e. at the height of sampling S3), whereas the real excluding the GLS separator phase. NLR was calculated considering the mainstream wastewater was fed to the bottom sludge section (see ammonium, nitrite and nitrate concentrations of the real mainstream Fig. 2). The obtained results showed that acetate addition (period II and wastewater. Likewise, NRR was calculated considering the removed IV, respectively) allowed the removal of the nitrate produced by ammonium, nitrite and nitrate concentrations. The OLR was calculated anammox bacteria, without hampering the anammox process (Fig. 3, considering the necessary acetate to reduce (i) the nitrate produced by Table 1). This, in turn, significantly improved effluent quality while the anammox process and (ii) the inflow nitrate according to the stoi­ maintaining rather high and constant NRR (0.16 ± 0.03 g N L–1 d–1) and chiometric relation of 3.9 g COD g-1N-NO3- (see Eq. 1) considering a high NRE (91 ± 8 %) (on average from acetate addition periods, Fig. 3). ratio of 0.78 g COD g− 1 sodium acetate. The variations of the acetate Reactor performance was successfully maintained when temperature flow rate to meet the required OLR were manually and periodically decreased from 20 to 14 ◦ C (Fig. 3). In case of increasing inflow nitrate implemented. As a result, the requirements just described were not al­ concentrations (i.e. eventual nitrate production in previous treatment ways fully achieved due to the fluctuating inflow nitrate concentrations stages), the OLR was adjusted to maintain a good effluent quality, throughout periods II and IV. In case of a full-scale application, this without hampering the autotrophic nitrogen removal by anammox could be easily managed by a proper automatic control strategy. Ni­ bacteria (Fig. 3). Even when acetate was added over the stoichiometric trogen removal efficiency (NRE) was calculated based on Eq. (2): requirements (day 93, from days 111 to 114, day 132, from days 172 to (N− NH+ − − + − − 4 +N− NO2 +N− NO3 )inf − (N− NH4 +N− NO2 +N− NO3 )eff 182 and 184 to 192), COD effluent concentrations remained in the range NRE= (N− NH+4 +N− NO − 2 +N− NO − ) 3 inf of COD inflow concentrations (on average for acetate addition periods (2) 40 ± 5 and 47 ± 10 mg COD L–1 for influent and effluent, respectively), showing that mainly the non–biodegradable COD entering the reactor where N − NH+ 4 , N − NO2 and N − NO3 are the ammonium, nitrite and − − was the only organic source discharged, so that the external acetate was nitrate concentrations in the influent (inf) and in the effluent (eff), either directly used for heterotrophic denitrifiers or may have been respectively. stored as polymers by the cells (Fig. 4 and Table A.3, Supporting In­ The emission factor (EF) for N2O was calculated based on the N2O formation). Furthermore, despite acetate was added as an external C- 4 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 Fig. 3. Long-term operation of the CANDLE reactor at mainstream conditions. (A) Nitrogen loading and removal rates (NLR and NRR, respectively) and nitrogen removal efficiency (NRE). (B) Effluent concentrations of ammonium, nitrite and nitrate. Total nitrogen concentrations in the effluent refer to the sum of ammonium, nitrite and nitrate. Table 1 Operational parameters for the different periods without acetate addition (I and III) and during acetate addition (II and IV). NLR: Nitrogen Loading Rate; NRR: Nitrogen Removal Rate; NRE: Nitrogen Removal Efficiency. Period Time NLR (g NRR (g NRE Total Ammonium Nitrite Nitrate Nitrite/ Nitrate Nitrite/ lapse N L-1 N L-1 d- (%) effluent effluent (mg N effluent Effluent ammonium produced/ ammonium (days) d-1) 1 ) nitrogen (mg L-1) (mg N L-1) (mg N L-1) consumed ammonium influent N L-1) consumed Period I 43 0.15 0.10 ± 83 ± 18 ± 11 4±4 1±2 17 ± 9 1.3 ± 0.1 0.24 ± 0.04 1.2 ± 0.3 pre- ± 0.03 0.03 4 acetate addition Period II 97 0.18 0.17 ± 92 ± 5±4 2±2 1±2 3±4 1.2 ± 0.2 – 1.2 ± 0.2 acetate ± 0.02 0.03 7 addition Period III 31 0.19 0.15 ± 79 ± 13 ± 4 4±2 3±2 7±2 1.3 ± 0.1 0.24 ± 0.10 1.3 ± 0.1 no- ± 0.01 0.02 8 acetate addition Period IV 21 0.16 0.14 ± 88 ± 8±5 5±4 2±2 2±2 1.4 ± 0.2 – 1.2 ± 0.2 acetate ± 0.02 0.01 8 addition source to enhance heterotrophic denitrification, this did not affect addition, from days 140 to 149, the nitrate production to ammonium effluent COD concentrations which remained under 125 mg COD L-1 consumption ratio was lower than the stoichiometrically proposed for throughout the operation (see Fig. 4), below the discharging limit ac­ the anammox process (i.e. 0.17 ± 0.05), still detecting a small degree of cording to the European legislation. heterotrophic denitrifying activity. However, reactor performance Reactor operation period was divided into four different phases decreased compared to that of period II, resulting into lower NRE (79 ± (Table 1). The high total nitrogen concentrations in the effluent of 8 %) and NRR (0.15 ± 0.02 g N L-1 d–1) (Table 1). The lower reactor period I (i.e. without acetate addition) was caused by the high incoming performance was attributed to the lack of an external C-source, highly nitrate concentrations and by the oscillating nitrite to ammonium reducing the heterotrophic denitrifying activity and, thus, into high feeding ratios (1.2 ± 0.3) (see Figs. A.3 and A.2, respectively in Sup­ effluent nitrate concentrations (7 ± 2 mg N L–1, Table 1). Also, the porting Information). As soon as heterotrophic denitrifying activity was non–adjusted nitrite to ammonium feeding ratio (Fig. A.2, Supporting promoted by adding acetate (period II), higher NRE (92 ± 7 %) and NRR Information) of this period resulted into an excess of ammonium in the (0.17 ± 0.03 g N L-1 d-1) and a better effluent quality (5 ± 4 mg N L–1) effluent (4 ± 2 mg N L–1, Table 1) and, thus, worsened effluent quality. were achieved by maintaining high NLR (0.18 ± 0.02 g N L-1 d–1) During period IV, the performance of the CANDLE reactor at low (Table 1). temperatures was assessed. For this purpose, reactor temperature was From day 140 to 171 (period III), reactor performance was reassessed decreased from 20 to 14 ◦ C while acetate addition was reactivated by to rule out the possible negative effects of the previous external C-source maintaining the NLR in the range of previous periods (0.16 ± 0.02 g N addition on the anammox process (i.e. only mainstream wastewater L–1 d-1, Table 1). As the OLR was mainly over-dimensioned during feeding) by maintaining a similar NLR than that of previous periods period IV, nitrate concentrations remained very low in the effluent (0.19 ± 0.01 g N L-1 d–1). During the first 10 days after stopping acetate (Table 1). Simultaneously, high effluent ammonium concentrations 5 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 withdrawn samples of the sludge bed. Anammox activity was detected within bottom sludge sections, as observed by the reduction of ammonium and nitrite compounds from S1 to S3 sludge bed sections in both conditions, with and without acetate addition (Fig. 5A). The biomass located within this sludge section pre­ sented a brown-to-reddish colour, with a dense granule structure (see Fig. A.5, Supporting Information). 3.3. Effect of heterotrophic denitrifying activity on N2O emissions Nitrate concentrations increased along the bottom sludge bed sec­ tions, according to the nitrate produced by anammox bacteria (Fig. 5A). When acetate was supplied, nitrate concentrations decreased after reaching the sludge bed section corresponding to the acetate feeding point (i.e. sampling S3) (Fig. 5A). The highest COD concentrations corresponded to the acetate bed section addition and decreased along the upper sludge bed sections, in agreement with the reduction of nitrate concentrations by heterotrophic denitrifiers (Fig. 5B). The biomass located beyond the acetate feeding point presented fewer compact ag­ Fig. 4. Soluble chemical oxygen demand (COD) concentrations of the influent gregates with a brownish colour and fluffy structure (see Fig. A.5, real mainstream wastewater (white dots) and of the effluent (black squares) Supporting Information), although the sludge presented similar average during CANDLE reactor operation. The external C-source (i.e. acetate) is diameter sizes with those of the bottom bed sections (Fig. A.4, Sup­ considered within effluent concentrations. porting Information). The obtained results indicate that heterotrophic denitrifying activity were detected (5 ± 4 mg N L-1, Table 1) caused by a misbalanced nitrite was located within middle and upper sludge sections, while anammox to ammonium feeding ratio to the anammox process (1.2 ± 0.1, see activity was relegated to the bottom sludge section. The reason was Fig. A.2 in Supporting Information for information of the nitrite to attributed to the suitable distribution of acetate addition beyond the ammonium feeding ratio). Overall, a better effluent quality and NRE middle sludge section of the CANDLE reactor. This was proved when were achieved compared to periods without acetate addition (Table 1). measuring acetate concentrations along the sludge bed sections, as they Nevertheless, the N-NO2–/N-NH4+ influent ratio needs to be better were only detected within middle sludge sections, while no acetate controlled to avoid excessive discharging nitrogen compounds, whereas diffusion was observed to the bottom sludge bed sections, ensuring the the COD load should be properly managed to completely reduce nitrate performance of the anammox process (Fig. 5B). Further, the lower concentrations. Despite the temperature decrease, reactor performance granule size and fluffy structure of heterotrophic denitrifying bacteria of period IV was only slightly lower compared to the previous acetate could have also favored heterotrophic denitrifying community to remain addition period at 20 ◦ C (period II), as NRR and NRE only decreased within the middle and upper sludge bed section. from 0.17 ± 0.03 g N L-1 d-1 and 92 ± 7 % (period II) to 0.14 ± 0.01 g N When acetate was added above the stoichiometric requirements (day L-1 d-1 and 88 ± 8 % (period IV) (see Table 1). The capability of an 132), the COD effluent concentration remained in the range of the COD UAnSB reactor of dampening a temperature drop at mainstream con­ inflow concentration, whereas a residual nitrate concentration was ditions was previously demonstrated to be a result of the so-called detected (ca. 0.1 mg N L-1) (Fig. 5A). Additionally, mass balances indi­ reactor overcapacity [13]. This is the capacity of the middle and cated that COD consumption was higher than the stoichiometrically upper part of the sludge bed to counterbalance the activity decrease of required for nitrate reduction by using acetate as an organic source the bottom sludge bed section when temperature drops. According to (Fig. 5B). Acetate can be intracellularly accumulated as poly­ this, the CANDLE reactor also seemed to show overcapacity to damp hydroxybutyrate (PHB) [26–28]. This is consistent with results from temperature drops. Thus, the main reason of the lower NRR and NRE in period III, as heterotrophic denitrification still occurred after stopping period IV can be attributed to the non–adjusted nitrite to ammonium acetate addition (days 140 to 149, see section 3.1). This suggests that feeding ratio, rather than to the effects of temperature decrease. heterotrophic denitrifying bacteria could have stored organic com­ The coexistence of anammox and heterotrophic denitrifying activ­ pounds when acetate addition was over-dimensioned and that they were ities in a single reactor was demonstrated to be a promising technology capable of further consume them under starvation conditions of an to overcome the associated drawbacks of the two-stage autotrophic external organic feeding. However, the effect of heterotrophic de­ biological nitrogen removal process at mainstream conditions; that is, nitrifiers of consuming COD from lysis or endogenous material under potential nitrate production or a misbalanced nitrite to ammonium ratio organic starvation conditions cannot be neglected. Further studies in the effluent of the PN reactor. Further studies should be performed to should be performed to evaluate if acetate was accumulated as PHB and evaluate the CANDLE reactor performance under fluctuating nitrogen how this affected heterotrophic denitrifiers when they were exposed to and organic loading rates as well as under different organic carbon substrate limiting conditions. sources at pilot scale. Different studies reported that anammox activity became deterio­ rated when influent with high COD concentrations or high COD/N ratios 3.2. Location of anammox and heterotrophic denitrifiers along the bed was fed in different reactor configurations [21,29]. However, the results prevents competition of nitrite of this study not only provided evidence that anammox activity remained unaffected after the continuous external C-source supply, but Substrate concentration measurements (nitrogen compounds and that the coexistence of anammox and heterotrophic denitrification COD) were performed along the different sampling points of the sludge processes was possible regardless the COD/N ratio and the incoming bed (see Fig. 2) for three different operational days: during acetate acetate concentrations thanks to the employed reactor configuration (i. supply (period II, on days 125 and 132, when acetate addition was under e., two different feeding points along the sludge bed) which avoided and over-dimensioned, respectively) and without acetate addition (day acetate diffusion to the bottom sludge bed section, mainly colonized by 168, period III). Additionally, biomass characteristics were examined anammox. using a stereomicroscope and particle size was measured for different N2O off-gas concentrations were monitored within different 6 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 Fig. 5. Ammonium, nitrite, nitrate concentrations (A) and soluble chemical oxygen demand (CODS) and acetate (measured as COD) concentrations (B) corre­ sponding to different sludge bed heights sections 0 m (S1), 0.05 m (S2), 0.16 m (S3), 0.25 m (S4) and 0.36 (S5) of the CANDLE reactor during (period II, days 125 and 132, when acetate addition was under and over-dimensioned, respectively) and without acetate (period III, day 168) addition. *u.d, under detection limit. (For interpretation of the references to colour, the reader is referred to the web version of this article). operational periods of the CANDLE reactor with and without acetate reactor were lower when acetate addition occurred (Fig. 6). This agrees addition at 20 ◦ C (period II and III, respectively) and with acetate with the fact that heterotrophic denitrifiers can act as an N2O sink when addition at 17 ◦ C (period IV) (Fig. 6). The measured N2O emission factor COD is not limiting denitrification [33]. From days 125 to 130 (period (EF) (0.002 %, on average from all available measurements) was II), acetate addition was deliberately under–dimensioned, resulting into maintained low during the different operational periods, with and a peak of N2O off-gas concentrations (Fig. 6). This confirmed that the without acetate addition. This is consistent for a reactor where the main amount of N2O that the denitrifiers can consume is smaller under COD biological activity is the anammox process since previous studies indi­ limiting conditions [31]. To demonstrate the positive effects of COD for cated that anammox lacks of genes producing N2O [30]. As anammox do mitigation of N2O emissions, acetate addition was completely stopped not produce N2O, the measured N2O emissions may account on other on day 140 (period III, see Fig. 3). This resulted in a gradual increase of microbial pathways. They can be attributed to incomplete heterotrophic the N2O off-gas concentrations (Fig. 6). However, this slow and pro­ denitrification activity [31] due to the limited organic matter contained gressive increase was not expected. During the first days after stopping in the mainstream wastewater or because of the imposed external C- acetate addition (days 140 to 149), nitrate reduction was still active (i.e. source limiting conditions. They can also be related to the AOB of the the nitrate production to ammonium consumption ratio was 0.17 ± incoming wastewater transported to the CANDLE reactor and located at 0.05, see Fig A.1, Supporting Information). A plausible possibility of the the bottom of the sludge bed “i.e. just in the entry”, where they can gradual N2O off-gas concentrations increase is related to the capability benefit from the residual dissolved oxygen (DO) of the inflow waste­ of heterotrophic denitrifying bacteria to reduce a fraction of the nitrate water. Actually, N2O production by AOB is enhanced under low oxygen produced by the anammox process by using the stored organic com­ concentrations or anoxic environments, prevailing conditions within pounds during the previous acetate addition period (i.e. exposed to anammox reactors [32]. organic limiting conditions). When the nitrate produced to the ammo­ The measured average N2O off-gas concentrations of the CANDLE nium consumption ratio agreed with the stoichiometric produced by the Fig. 6. N2O off-gas concentrations during period II (acetate addition) and period III (no acetate) at 20 ◦ C and during period IV (acetate addition) at 17 ◦ C. 7 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 anammox process (0.29 ± 0.03; Fig A.1), i.e. heterotrophic denitrifying and removal of low–quality sequences can be found in Supplementary activity was severely limited by the absence of organic matter, N2O off- Information (Fig. A.9). All libraries presented an average length of 290 gas concentrations increased and became rather stable at a maximum bps per sequence. value (Fig. 6). Finally, when acetate addition was re–established and During period I (before acetate addition), the microbial community over–dimensioned (period IV, from days 175 to 177), N2O emissions presented a high abundance of Candidatus Brocadia in both S1 (28 %) decreased to minimal values, confirming that heterotrophic denitrifiers and S3 (12 %) (Fig. 8). Also, different abundances of heterotrophic acted as an N2O sink (Fig. 6). denitrifying bacteria were detected. Chloroflexi, Acidobacteria and Chlorobi (specifically Ignavibacterium genus) phylum were observed 3.4. Ex-situ assessment of heterotrophic denitrifying activity with abundances ranging from 8 to 21 % (Fig. 8). These phyla represent a large fraction of the microbial diversity within anammox reactors Ex-situ heterotrophic denitrifying batch tests aimed to assess het­ without the addition of any external C–source [35–37]. Under these erotrophic denitrifying activity of biomass located above the external C- conditions, heterotrophic denitrifying bacteria can degrade the extra­ source feeding before (day 43, period I) and during (day 98, period II) cellular polymeric substances (EPS) produced by anammox bacteria to acetate addition to the CANDLE reactor by using nitrite and/or nitrate as use it as an organic carbon source while encoding nitrite and/or nitrate different electron acceptors. Before acetate addition (period I), nitrite respiration [37,38]. Overall, no significant differences were detected on and nitrate consumption rates were in the same range (Fig. 7). When the microbial community within the bottom and middle sections of the both electron acceptors were added together, they were simultaneously sludge bed (see similitude results from Jaccard Index, Table A.4 in reduced although nitrate consumption rates were higher (see Table A.2 Supporting Information). Further, anammox bacteria were present in in Supporting Information). When assessing heterotrophic denitrifying both sludge bed sections in a significant amount. activity after a period of acetate supply to the CANDLE reactor (i.e. day After ca. 2 months of acetate addition, microbial composition 98), nitrite and nitrate consumption rates clearly increased compared to changed within the middle sludge section S3, while it remained almost previous tests (Fig. 7). Nitrate consumption rates were ca. four times unaffected within the sludge bed section S1 (see Table A.4 in Supporting higher than those of nitrite, indicating a preference of the denitrifying Information). Candidatus Brocadia still presented a high abundance biomass towards nitrate as electron acceptor (see Table A.2 in Sup­ within the bottom sludge bed section (30 %), while its abundance porting Information). This was confirmed when both electron acceptors became significantly reduced within upper and middle sludge sections were added together, as nitrate was firstly reduced. Under this situation, (<3 %) (Fig. 8 and Fig. A.7 in Supporting Information). However, the nitrite consumption started right after nitrate depletion (ca. 22 h after small anammox fraction of S3 was still dominated by the genus Candi­ starting batch activity tests). Nitrite consumption rates were ca. 5 times datus Brocadia, according to the obtained results from the specific higher than the achieved when only nitrite was used as electron acceptor primer (Fig. A.8 in Supporting Information). Within middle and upper (Fig. 7). This might be attributed as nitrite reduction occurred during the sludge layers, the microbial composition was strongly dominated by the exponential growth phase of heterotrophic denitrifying bacteria, which genus Thauera (33 to 46 %), resulting into a decrease of the previously present doubling times of ca. 9 h at 20 ◦ C [34]. The ex-situ batch tests detected heterotrophic denitrifiers before acetate supply (Fig. 8 and performed during the period of acetate addition showed that hetero­ Fig. A.7 in Supporting Information). Thauera was reported to provide trophic denitrifiers preferred to use nitrate over nitrite as electron nitrite for anammox bacteria through reducing nitrate to nitrite in acceptor, although in the absence of nitrate, nitrite reduction was also denitritation/anammox processes [38–41]. Also, a high abundance of possible. Thauera was found in a CANON reactor [42] and when coupling anammox and denitritation in a single reactor fed with acetate as a C- 3.5. Impact of acetate addition on microbial community source [39]. Thauera prefers to use nitrate over nitrite as electron acceptor [39]. Nevertheless, nitrite can be also reduced by Thauera de­ The total number of sequences for each library after quality analysis nitrifiers after nitrate depletion if a sufficient amount of organic matter is still available [39]. These results are in accordance with the ex-situ batch activity tests performed during acetate addition (i.e. day 98) when both electron acceptors were added, as nitrite reduction occurred right after nitrate consumption. Microbial diversity results confirmed that anammox activity domi­ nated the bottom sludge bed section of the CANDLE reactor. Results from profile concentrations showed that nitrate reduction took place beyond the acetate feeding point (i.e. sampling S3), in accordance with the developed heterotrophic denitrifying community (i.e. Thauera). The obtained microbial community data confirmed that the development of a significant denitrifying community in the middle and upper sections of the CANDLE reactor did not hinder the stability of the anammox com­ munity located within the bottom sludge bed section. 3.6. Implication of the findings This research proposed a new reactor configuration to couple anammox and heterotrophic denitrifying processes in a single reactor unit by treating a real mainstream wastewater (containing almost no Fig. 7. Specific heterotrophic denitrifying activities from ex-situ batch tests biodegradable COD and amended with ammonium and nitrite) at tem­ before (black) and during acetate addition (white) on days 43 and 98, respec­ peratures from 20 to 14 ◦ C. A high and stable nitrogen removal per­ tively. Activities before acetate addition (in black) have been plotted at a smaller Y axis scale in the Supporting Information (Fig. A.6). The applied COD/ formance was obtained, even under extra incoming nitrate N-NOx ratios and nitrate and nitrite concentrations of each ex-situ batch test can concentrations (i.e., when simulating a poor performance of previous be found in Supplementary Information (Table A.2). *Corresponds to the nitrite mainstream stages). Full-scale implementation of anammox in the main consumption rate after nitrate depletion within the ex-situ batch tests where waterline is known to currently present as main bottleneck the pro­ both electron acceptors were added. duction of nitrite (i.e., circumventing the production of nitrate by nitrite 8 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 Fig. 8. Microbial diversity using the 515F-806R primer for libraries before and after ca. two months of acetate addition (day 42 and 97, respectively) at different CANDLE reactor heights: 0 m (S1) and 0.16 m (S3). Relative abundance was calculated by only considering those microorganisms in which the number of 16S copies was higher than 3 % of the total copies. (For interpretation of the references to colour, the reader is referred to the web version of this article). oxidizing bacteria (NOB)). Despite previous reports have highlighted the nitrogen removal rates for one-stage PD/AMX systems at 27 ◦ C [39] and stability of the partial nitritation in granular sludge reactors (when the at temperatures ranging from 30 to 20 ◦ C [45] were reported. PN/AMX is split in two different units, as in Fig. 1A) [10,11], this was For a qualitative comparison of the PN/AMX process proposed in this never demonstrated feasible with real wastewater with large reactor study with the PD/AMX configuration, two possible main water lines of volumes. If this technology would demonstrate its feasibility, still there an urban WWTP are defined in Fig. 1. Considering the main objective of is the risk of potential episodes of nitrate production, particularly during both configurations is to enable maximum energy recovery as biogas, the colder days in the winter season. The CANDLE reactor offers a new both processes (PD/AMX and PN/AMX) should be integrated in the approach for mainstream anammox as it allowed to overcome the mainstream with a previous organic matter recovery step, such a HRAS associated drawbacks of previous mainstream stages (e.g., NOB prolif­ system, to redirect the organic matter to anaerobic digestion. After this eration at low temperatures within the PN step) as demonstrated by the first step, both configurations will receive an ammonia nitrogen inflow low total nitrogen effluent concentrations (≤8 mg N L-1; see Table 1). A with a low C/N ratio. The main advantage of PD/AMX over PN/AMX better reactor performance could have been obtained by optimizing the systems is the removal of the nitritation step, which requires a strict inflow nitrite to ammonium ratio for the anammox process and by control of NOB proliferation. Despite this, a proper control strategy to constantly adapting the organic load to the nitrate concentrations. For obtain a stable denitritation (i.e. by controlling the adequate nitrate to future studies, a proper automatic control strategy of the incoming ammonium ratio) is needed for PD/AMX systems. Further, the PD/AMX ammonium load should be desirable (see [15]). The automatic control of stability at temperatures below 20 ◦ C have not been yet proved, which is the COD addition is a complex problem because requires of online probably linked to the complexity of maintaining a high anammox ac­ measurements of the nitrogen load to the CANDLE reactor, and the tivity at low temperatures due to the low anammox abundance pop­ control strategy and reactor sensors to be used require for further ulations detected within the employed SBR configurations [41,45,46]. research. In comparison to PD/AMX systems, the PN/AMX process coupled to full The use of acetate as an organic source was proposed in this work. heterotrophic denitrification (i.e. the so-called CANDLE configuration Acetate is preferentially used for nitrate reductase rather than nitrite proposed in this study) requires for a 20 to 30 % lower oxygen con­ reductase within the electron transfer chain [38], and thus, beneficial to sumption as partial nitritation instead of nitrification is performed [39]. reduce the produced nitrate by anammox bacteria. Further, acetate has In addition, a higher amount of organic dosage is required for PD/AMX been reported to be mostly converted to PHB as a storage compound than for PN/AMX configurations; for the PD/AMX process, the nitrate rather than being used for biosynthesis [28,38]. This offers an advantage obtained from the nitrification step (50 % of the incoming nitrogen) is for mainstream anammox as it allows to buffer the instability of previous denitrified to nitrite, whereas only the produced nitrate by anammox mainstream stages by reducing the extra incoming nitrate (e.g., PN bacteria needs to be denitrified in PN/AMX systems (i.e. 10 % of the destabilization at low temperatures). However, the use of acetate re­ incoming nitrogen). quires for higher economical costs compared to other organic sources [43]. Future studies should focus on investigating the feasibility of using 4. Conclusions other organic sources such as glycerol or ethanol, among others [41,44]. The direct use of the sewage stream as an organic source is not recom­ • The ad hoc reactor design for Coupling ANammox and Denitrifica­ mended; the incoming nitrogen from the sewage (i.e., ammonia) could tion in a singLE unit (i.e. the CANDLE reactor) allowed the addition not be converted as the reactor configuration of this study was designed of an external organic source for the removal of nitrate produced by to add the C-source beyond the anammox sludge bed, to avoid negative anammox bacteria without compromising the stability of the effects on anammox performance. anammox process. Recently, different studies proposed the application of a partial • The coexistence of anammox and heterotrophic denitrifying bacteria denitritation/anammox (PD/AMX) process to the main water line to in a CANDLE reactor was possible as anammox activity dominated overcome the associated drawbacks of achieving a stable nitrite pro­ within bottom sludge bed sections, while heterotrophic denitrifica­ duction within PN/AMX systems at low ammonium strength and low tion occurred within middle and upper sludge bed sections. temperature conditions. Promising results in terms of attaining high 9 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 • N2O emissions were lower during the addition of an external carbon [13] X. Juan-Díaz, J. Pérez, J. Carrera, Effective dampening of temperature effects in an anammox reactor treating real mainstream wastewater, J. Water Process Eng. source, indicating that heterotrophic denitrifiers acted as an N2O (2021), 101853, https://doi.org/10.1016/j.jwpe.2020.101853. sink. [14] M. Laureni, D.G. Weissbrodt, I. Szivák, O. Robin, J.L. Nielsen, E. Morgenroth, • Microbial diversity results confirmed that both populations (anam­ A. Joss, Activity and growth of anammox biomass on aerobically pre-treated mox and heterotrophic denitrifiers) occupied differentiated niche municipal wastewater, Water Res. 80 (2015) 325–336, https://doi.org/10.1016/j. watres.2015.04.026. compartments in a CANDLE reactor. [15] J. Pérez, E. Isanta, J. Carrera, Would a two-stage N-removal be a suitable • Anammox population was dominated by the genus Candidatus Bro­ technology to implement at full scale the use of anammox for sewage treatment? cadia while the Thauera genus dominated among heterotrophic Water Sci. Technol. 72 (2015) 858–864, https://doi.org/10.2166/wst.2015.281. [16] S. Cao, R. Du, Y. Zhou, Coupling anammox with heterotrophic denitrification for denitrifiers. enhanced nitrogen removal: A review, Crit. Rev. Environ. Sci. Technol. 51 (2021) • The CANDLE reactor has been shown to be a feasible technology to 2260–2293, https://doi.org/10.1080/10643389.2020.1778394. enhance the robustness of the two-stage mainstream partial nitrita­ [17] A. Soler-Jofra, R. Wang, R. Kleerebezem, M.C.M. van Loosdrecht, J. Pérez, Stratification of nitrifier guilds in granular sludge in relation to nitritation, Water tion/anammox process, alleviating the negative effects from the Res. 148 (2019) 479–491, https://doi.org/10.1016/j.watres.2018.10.064. previous treatment stages. [18] C. Zhang, A. Guisasola, J.A. Baeza, Achieving simultaneous biological COD and phosphorus removal in a continuous anaerobic/aerobic A-stage system, Water Res. 190 (2021), 116703, https://doi.org/10.1016/j.watres.2020.116703. [19] H. Duan, L. Ye, Q. Wang, M. Zheng, X. Lu, Z. Wang, Z. Yuan, Nitrite oxidizing Declaration of Competing Interest bacteria (NOB) contained in influent deteriorate mainstream NOB suppression by sidestream inactivation, Water Res. 162 (2019) 331–338, https://doi.org/ The authors declare that they have no known competing financial 10.1016/j.watres.2019.07.002. [20] J. Li, J. Li, Y. Peng, S. Wang, L. Zhang, S. Yang, S. Li, Insight into the impacts of interests or personal relationships that could have appeared to influence organics on anammox and their potential linking to system performance of sewage the work reported in this paper. partial nitrification-anammox (PN/A): A critical review, Bioresour. Technol. 300 (2020), 122655, https://doi.org/10.1016/j.biortech.2019.122655. [21] M. Pijuan, A. Ribera-Guardia, J.L. Balcázar, M.M. Micó, T. de la Torre, Effect of Acknowledgments COD on mainstream anammox: Evaluation of process performance, granule morphology and nitrous oxide production, Sci. Total Environ. 712 (2020), 136372, https://doi.org/10.1016/j.scitotenv.2019.136372. This research was funded by the Spanish Ministerio de Economía, [22] X. Zhang, H. Zhang, C. Ye, M. Wei, J. Du, Effect of COD/N ratio on nitrogen Industria y Competitividad (MINECO), Agencia Estatal de Investigación removal and microbial communities of CANON process in membrane bioreactors, (AEI) and Fondo Europeo de Desarrollo Regional (FEDER, EU), Bioresour. Technol. 189 (2015) 302–308, https://doi.org/10.1016/j. biortech.2015.04.006. CTQ2017-82404-R. [23] G.J.F. Smolders, J. van der Meij, M.C.M. van Loosdrecht, J.J. Heijnen, Model of the anaerobic metabolism of the biological phosphorus removal process: Stoichiometry Appendix A. Supplementary data and pH influence, Biotechnol. Bioeng. (1994), https://doi.org/10.1002/ bit.260430605. [24] E.M. van den Berg, M. Boleij, J.G. Kuenen, R. Kleerebezem, M.C.M. van Loosdrecht, Supplementary data to this article can be found online at https://doi. DNRA and denitrification coexist over a broad range of acetate/N-NO3- ratios, in a org/10.1016/j.cej.2021.134087. chemostat enrichment culture, Front. Microbiol. 7 (2016) 1–12, https://doi.org/ 10.3389/fmicb.2016.01842. [25] APHA, APHA, Standard methods for the examination of water and wastewater, References 21st ed., 2005. [26] M. Majone, M. Beccari, D. Dionisi, C. Levantesi, V. Renzi, Role of storage phenomena on removal of different substrates during pre-denitrification, Water [1] M. Strous, J.J. Heijnen, J.G. Kuenen, M.S.M. Jetten, The sequencing batch reactor Sci. Technol. (2001), https://doi.org/10.2166/wst.2001.0131. as a powerful tool for the study of slowly growing anaerobic ammonium-oxidizing [27] Y. Liu, L. Peng, J. Guo, X. Chen, Z. Yuan, B.J. Ni, Evaluating the Role of Microbial microorganisms, Appl. Microbiol. Biotechnol. 50 (1998) 589–596, https://doi.org/ Internal Storage Turnover on Nitrous Oxide Accumulation during Denitrification, 10.1007/s002530051340. Sci. Rep. 5 (2015) 1–10, https://doi.org/10.1038/srep15138. [2] B. Kartal, J.G. Kuenen, M.C.M. van Loosdrecht, Engineering. Sewage treatment [28] S. Cao, Y. Peng, R. Du, S. Wang, Feasibility of enhancing the DEnitrifying with anammox, Science. 328 (2010) 702–703, https://doi.org/10.1126/ AMmonium OXidation (DEAMOX) process for nitrogen removal by seeding partial science.1185941. denitrification sludge, Chemosphere 148 (2016) 403–407, https://doi.org/ [3] M.S.M. Jetten, S.J. Horn, M.C.M. Van Loosdrecht, Towards a more sustainable 10.1016/j.chemosphere.2015.09.062. municipal wastewater treatment system, Water Sci. Technol. 35 (1997) 171–180, [29] C. Chen, F. Sun, H. Zhang, J. Wang, Y. Shen, X. Liang, Evaluation of COD effect on https://doi.org/10.1016/S0273-1223(97)00195-9. anammox process and microbial communities in the anaerobic baffled reactor [4] A.I. Versprille, B. Zuurveen, T. Stein, The A-B process: a novel two-stage (ABR), Bioresour. Technol. 216 (2016) 571–578, https://doi.org/10.1016/j. wastewater treatment system, Water Sci. Technol. 17 (1984) 235–246. biortech.2016.05.115. [5] B. Böehnke, R. Schulze-Rettmer, S. Zuckut, Cost-effective reduction of high- [30] B. Kartal, M.M.M. Kuypers, G. Lavik, J. Schalk, H.J.M. Op Den Camp, M.S. strength wastewater by adsorption-based activated sludge technology, J. Wat. Eng. M. Jetten, M. Strous, Anammox bacteria disguised as denitrifiers: Nitrate reduction Manag. 145 (1998) 1–34. to dinitrogen gas via nitrite and ammonium, Environ. Microbiol. 9 (2007) [6] B. Böehnke, Das Adsorptions-Belebungsverfahren, Korrespondenz Abwasser. 24 635–642, https://doi.org/10.1111/j.1462-2920.2006.01183.x. (1977) 121–127. [31] M. Jia, C.M. Castro-Barros, M.K.H. Winkler, E.I.P. Volcke, Effect of organic matter [7] H. Siegrist, D. Salzgeber, J. Eugster, A. Joss, Anammox brings WWTP closer to on the performance and N2O emission of a granular sludge anammox reactor, energy autarky due to increased biogas production and reduced aeration energy for Environ. Sci. Water Res. Technol. 4 (2018) 1035–1046, https://doi.org/10.1039/ N-removal, Water Sci. Technol. 57 (2008) 383–388, https://doi.org/10.2166/ c8ew00125a. wst.2008.048. [32] M.J. Kampschreur, N.C.G. Tan, R. Kleerebezem, C. Picioreanu, M.S.M. Jetten, M.C. [8] E.M. Gilbert, S. Agrawal, S.M. Karst, H. Horn, P.H. Nielsen, S. Lackner, Low M. Van Loosdrecht, Effect of dynamic process conditions on nitrogen oxides Temperature Partial Nitritation/Anammox in a Moving Bed Biofilm Reactor emission from a nitrifying culture, Environ. Sci. Technol. 42 (2008) 429–435, Treating Low Strength Wastewater, Environ. Sci. Technol. 48 (2014) 8784–8792, https://doi.org/10.1021/es071667p. https://doi.org/10.1021/es501649m. [33] M. Conthe, P. Lycus, M. Arntzen, A. Ramos da Silva, Å. Frostegård, L.R. Bakken, [9] M. Laureni, P. Falås, O. Robin, A. Wick, D.G. Weissbrodt, J.L. Nielsen, T.A. Ternes, R. Kleerebezem, M.C.M. van Loosdrecht, Denitrification as an N2O sink, Water Res. E. Morgenroth, A. Joss, Mainstream partial nitritation and anammox: Long-term 151 (2019) 381–387, https://doi.org/10.1016/j.watres.2018.11.087. process stability and effluent quality at low temperatures, Water Res. 101 (2016) [34] Metcalf & Eddy, Wastewater Engineering Teatment and Reuse, 5th edition, New 628–639, https://doi.org/10.1016/j.watres.2016.05.005. York, 2003. https://doi.org/10.1016/0309-1708(80)90067-6. [10] E. Isanta, C. Reino, J. Carrera, J. Pérez, Stable partial nitritation for low-strength [35] X.R. Li, B. Du, H.X. Fu, R.F. Wang, J.H. Shi, Y. Wang, M.S.M. Jetten, Z.X. Quan, The wastewater at low temperature in an aerobic granular reactor, Water Res. 80 bacterial diversity in an anaerobic ammonium-oxidizing (anammox) reactor (2015) 149–158, https://doi.org/10.1016/j.watres.2015.04.028. community, Syst. Appl. Microbiol. 32 (2009) 278–289, https://doi.org/10.1016/j. [11] C. Reino, M.E. Suárez-Ojeda, J. Pérez, J. Carrera, Kinetic and microbiological syapm.2009.03.002. characterization of aerobic granules performing partial nitritation of a low-strength [36] D.R. Speth, M.H. In’T Zandt, S. Guerrero-Cruz, B.E. Dutilh, M.S.M. Jetten, Genome- wastewater at 10 ◦ C, Water Res. 101 (2016) 147–156, https://doi.org/10.1016/j. based microbial ecology of anammox granules in a full-scale wastewater treatment watres.2016.05.059. system, Nat. Commun. 7 (2016) 11172. https://doi.org/10.1038/ncomms11172. [12] J. Pérez, M. Laureni, M.C.M. van Loosdrecht, F. Persson, D.J.I. Gustavsson, The role [37] C.E. Lawson, S. Wu, A.S. Bhattacharjee, J.J. Hamilton, K.D. McMahon, R. Goel, D. of the external mass transfer resistance in nitrite oxidizing bacteria repression in R. Noguera, Metabolic network analysis reveals microbial community interactions biofilm-based partial nitritation/anammox reactors, Water Res. 186 (2020), https://doi.org/10.1016/j.watres.2020.116348. 10 X. Juan-Díaz et al. Chemical Engineering Journal 431 (2022) 134087 in anammox granules, Nat. Commun. 8 (2017) 15416, https://doi.org/10.1038/ different organic matter concentrations, Chemosphere 193 (2018) 591–601, ncomms15416. https://doi.org/10.1016/j.chemosphere.2017.11.066. [38] R. Xiao, B.-J. Ni, S. Liu, H. Lu, Impacts of organics on the microbial ecology of [43] Y. Fernández-Nava, E. Marã, J. Soons, L. Castrillón, Denitrification of high nitrate wastewater anammox processes: Recent advances and meta-analysis, Water Res. concentration wastewater using alternative carbon sources, J. Hazard. Mater. 173 191 (2021), 116817, https://doi.org/10.1016/j.watres.2021.116817. (2010) 682–688, https://doi.org/10.1016/j.jhazmat.2009.08.140. [39] B. Ma, W. Qian, C. Yuan, Z. Yuan, Y. Peng, Achieving Mainstream Nitrogen [44] J.A. Torà, J.A. Baeza, J. Carrera, J.A. Oleszkiewicz, Denitritation of a high-strength Removal through Coupling Anammox with Denitratation, Environ. Sci. Technol. 51 nitrite wastewater in a sequencing batch reactor using different organic carbon (2017) 8405–8413, https://doi.org/10.1021/acs.est.7b01866. sources, Chem. Eng. J. 172 (2011) 994–998, https://doi.org/10.1016/j. [40] J. Ji, Y. Peng, X. Li, Q. Zhang, X. Liu, A novel partial nitrification-synchronous cej.2011.07.013. anammox and endogenous partial denitrification (PN-SAEPD) process for advanced [45] R. Du, S. Cao, H. Zhang, X. Li, Y. Peng, Flexible Nitrite Supply Alternative for nitrogen removal from municipal wastewater at ambient temperatures, Water Res. Mainstream Anammox: Advances in Enhancing Process Stability, Environ. Sci. (2020), 115690, https://doi.org/10.1016/j.watres.2020.115690. Technol. 54 (2020) 6353–6364, https://doi.org/10.1021/acs.est.9b06265. [41] R. Du, S. Cao, B. Li, M. Niu, S. Wang, Y. Peng, Performance and microbial [46] Y. Cao, M.C.M. van Loosdrecht, G.T. Daigger, Mainstream partial community analysis of a novel DEAMOX based on partial-denitrification and nitritation–anammox in municipal wastewater treatment: status, bottlenecks, and anammox treating ammonia and nitrate wastewaters, Water Res. (2017), https:// further studies, Appl. Microbiol. Biotechnol. 101 (2017) 1365–1383, https://doi. doi.org/10.1016/j.watres.2016.10.051. org/10.1007/s00253-016-8058-7. [42] M.J. García-Ruiz, P. Maza-Márquez, J. González-López, F. Osorio, Nitrogen removal capacity and bacterial community dynamics of a Canon biofilter system at 11