This work is licensed under a

Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International licence

Newcastle University ePrints - eprint.ncl.ac.uk

de Almeida Fernandes L, Pereira AD, Leal CD, Davenport R, Werner D, Mota

Filho CR, Bressani-Ribeiro T, Augusto de Lemos Chernicharo C, Calabria de

Araújo J.

Effect of temperature on microbial diversity and nitrogen removal

performance of an anammox reactor treating anaerobically pretreated

municipal wastewater.

Bioresource Technology 2018, 258, 208-219.

Copyright:

© 2018. This manuscript version is made available under the CC-BY-NC-ND 4.0 license

DOI link to article:

https://doi.org/10.1016/j.biortech.2018.02.083

Date deposited:

03/05/2018

Embargo release date:

23 February 2019

Effect of temperature on microbial diversity and nitrogen removal 1

performance of an anammox reactor treating anaerobically pretreated 2

municipal wastewater 3

4

5

Luyara de Almeida Fernandesa, Alyne Duarte Pereiraa, Cíntia Dutra Leala, Russell 6

Davenportb, David Wernerb, Cesar Rossas Mota Filhoa, Thiago Bressani-Ribeiroa, Carlos 7

Augusto de Lemos Chernicharoa, Juliana Calabria de Araújoa,∗ 8

9

aDepartment of Sanitary and Environmental Engineering, Federal University of Minas Gerais 10 - Antonio Carlos Avenue, 6627, Belo Horizonte, Minas Gerais State, 31270-90, Brazil 11

bSchool of Civil Engineering & Geosciences, Newcastle University, NE1 7RU Newcastle 12 upon Tyne, UK 13

14

15

16

Abstract 17

The effects of temperature reduction (from 35 °C to 20 °C) on nitrogen removal 18

performance and microbial diversity of an anammox sequencing batch reactor were evaluated. 19

The reactor was fed for 148 days with anaerobically pretreated municipal wastewater 20

amended with nitrite. On average, removal efficiencies of ammonium and nitrite were high 21

(96%) during the enrichment period and phases 1 (at 35 °C) and 2 (at 25 °C), and slightly 22

decreased (to 90%) when the reactor was operated at 20 °C. Deep sequencing analysis 23

revealed that microbial community structure changed with temperature decrease. Anammox 24

bacteria (Ca. Brocadia and Ca. Anammoximicrobium) and denitrifiers (Burkholderiales, 25

∗Corresponding author.

Tel.: +55 31 3409 3667; fax: +55 31 3409 1879

E-mail address: juliana@desa.ufmg.br (J.C. Araujo).

Myxococcales, Rhodocyclales, Xanthomonadales, and Pseudomonadales) were favoured 26

when the temperature was lowered from 35 °C to 25 °C, while Anaerolineales and 27

Clostridiales were negatively affected. The results support the feasibility of using the 28

anammox process for mainstream nitrogen removal from anaerobically pretreated municipal 29

wastewater at typical tropical temperatures. 30

31

32

Keywords: Anammox; Anaerobic effluent; Temperature; Ion Torrent sequencing; Nitrogen 33

removal 34

35

36

1. Introduction 37

For decades, aerobic nitrification followed by anoxic denitrification has been used to 38

remove nitrogen from wastewaters. Anaerobic ammonium oxidation (anammox) is a 39

biological process and a very promising alternative for nitrogen removal owing to its 40

sustainable characteristics (low or even no oxygen consumption, no addition of external 41

carbon source, and CO2 consumption), and therefore presents the possibility of wide 42

application (Kartal et al., 2013). Anammox bacteria are chemolithoautotrophic 43

microorganisms that can oxidize ammonium (NH4+) into dinitrogen gas (N2) using nitrite 44

(NO2-) as an electron acceptor under anoxic conditions (Strous et al., 1998). To date, seven 45

genera capable of anammox metabolism have been described in the literature. They belong to 46

the phylum Planctomycetes and orders Brocadiales and Planctomycetales (Pereira et al., 47

2017). 48

The anammox process has been applied mainly to remove nitrogen from ammonium-49

rich wastewaters with low chemical oxygen demand (COD)/N ratios (Ali and Okabe, 2015; 50

Shen et al., 2012; Tang et al., 2010). Nevertheless, some studies have shown that it can be 51

applied to remove nitrogen from anaerobic effluents with high COD/N ratios (Leal et al., 52

2016; Vlaeminck et al., 2012). 53

Temperature is a key factor for microorganism growth and metabolism that directly 54

affects their abundance (Ali and Okabe, 2015; Ma et al., 2016). Higher temperatures (35 °C to 55

40 °C) were usually associated with the maximum anammox biomass activity and cell 56

doubling time; however, extreme conditions (above 45 °C) may irreversibly inhibit cell 57

activity because of cell lysis (Dosta et al., 2008). Gao and Tao (2011) and Strous et al. (1999) 58

reported that the anammox process can occur in the temperature range of 20 °C to 43 °C, with 59

the optimum activity at 40±3 °C. Additionally, Ali and Okabe (2015) reported that 37 °C is 60

the optimum temperature for the anammox process. Yet, Zhu et al. (2008) reported that 26 °C 61

to 28 °C is the optimum temperature range and anammox metabolism decreased considerably 62

at temperatures below 15 °C and above 40 °C. Dosta et al. (2008) also noticed a decrease in 63

metabolism when the temperature was lowered from 20 °C to 15 °C. 64

Currently, the anammox and partial nitritation (PN)/anammox processes have been 65

used to treat warm and concentrated effluents, such as anaerobic sludge digestate, at 66

wastewater treatment plants (Lackner et al., 2014), which is known as sidestream treatment. 67

Nevertheless, these processes have not been applied for mainstream treatment (more diluted 68

effluents), taking into account anaerobically pretreated municipal wastewater. 69

The performance of anammox reactors for nitrogen removal at temperatures under 25 70

°C has been extensively investigated for ammonium-rich wastewaters (with 500 mg N·L-1) 71

(Hendrickx et al., 2012; Vázquez-Padín et al., 2010). However, higher ammonium removal 72

efficiency (90%) was reported by Hu et al. (2013) in a nitration-anammox bioreactor at 12 °C 73

fed with synthetic pretreated municipal wastewater (70 mg·L-1 of ammonium). Additionally, 74

low ammonium concentrations and total nitrogen removal efficiencies around 40% were 75

reported in a moving bed biofilm reactor with PN/anammox operated at 13 °C that was 76

treating diluted anaerobic digestate (at decreasing ammonium concentrations from 496 mg·L-1 77

to 43 mg·L-1). In this case, anammox bacteria outcompeted nitrifying bacteria (Persson et al., 78

2016). Moreover, Gonzalez-Martinez et al. (2015) investigated the performance and bacterial 79

community dynamics of a completely autotrophic nitrogen removal over nitrite (CANON) 80

bioreactor that was treating anaerobic digestate at decreasing temperatures (from 35 °C to 15 81

°C ) and decreasing ammonium concentrations (from 466 mg N·L-1 to 100 mg N·L-1). When 82

the system acclimated from 35 °C to 25 °C, nitrogen removal efficiency showed a moderate 83

decrease, affecting the bacterial community structure by selecting Candidatus Brocadia and 84

Candidatus Anammoxoglobus and increasing the abundance of some genera (Anaerolinea, 85

Acidobacterium, Chloroflexi, Fluviicola, and Prosthecobacter). An additional biomass 86

acclimation step from 25 °C to 15 °C sharply decreased the nitrogen removal efficiency in the 87

CANON bioreactor. 88

Despite previous studies, little is known about the microbial community composition 89

and dynamics in anammox reactors under mainstream conditions, i.e. treating real anaerobic 90

effluent. Furthermore, few studies have dealt with temperature variation impacts on microbial 91

community diversity and process performance. In this sense, Leal et al. (2016) reported that 92

high COD, nitrite, and ammonium removal efficiencies (80%, 90%, and 95%, respectively) 93

were reached with addition of real anaerobically pretreated municipal wastewater 94

(supplemented with nitrite) to a SBR. Moreover, the bacterial community structure changed 95

and DNA sequences related to Ca. Brocadia sinica, Ca. Brocadia caroliniensis and 96

Chloroflexi were identified. Nevertheless, the long-term effects of adding real anaerobic 97

effluent to an SBR were not investigated in this study, and the bacterial community structure 98

was investigated via PCR-denaturing gradient gel electrophoresis (DGGE), which detects 99

dominant members of a bacterial community. 100

Therefore, in the present study, the effect of typical tropical temperature variation (20–101

35 °C) on the microbial community and nitrogen removal efficiency of an anammox SBR fed 102

with anaerobically pretreated municipal wastewater over 148 days was investigated. The 103

microbial community structure and diversity were investigated via PCR-DGGE and high-104

throughput amplicon sequencing (Ion Torrent), which provide more detailed information 105

about microbial communities. Quantitative PCR (qPCR) was also applied to determine the 106

abundance of denitrifiers and anammox bacteria. 107

108

109

2. Methods 110

2.1 Experimental setup and monitoring 111

The inoculum used to enrich anammox bacteria was obtained from excess sludge in an 112

activated sludge plant located in the Brazilian city of Belo Horizonte. According to previous 113

studies, anammox bacteria have been enriched successfully from this sludge (Leal et al., 114

2016). A 2.0-L glass reactor (Benchtop Fermentor & Bioreactor BioFlo/CelliGen 115, New 115

Brunswick Scientific Co., Enfield, CT, USA) was used for anammox bacteria cultivation. 116

This reactor comprised dissolved oxygen and pH probes, as well as acid and base in-flow 117

tubes for pH control. The temperature was maintained at 35 °C (or reduced to 25 °C and 20 118

°C, see Table 1) via a water jacket, and the pH was maintained at 7.5. Anaerobic conditions 119

were assured by bubbling N2 gas (99.99%) through the liquid (in the enrichment period). This 120

gas was also flushed into the feed vessel to maintain anaerobic conditions in the synthetic 121

wastewater. When real anaerobic effluent was used, nitrogen was not flushed in feed vessel or 122

in the reactor. 123

The reactor was monitored for 308 days under different operational conditions 124

regarding the applied temperature (Table 1) and was operated in sequencing batch mode with 125

two cycles, one of 7 h (short cycle) and the other of 17 h (long cycle). Each cycle had four 126

phases: (i) continuous feeding period (40 min for both cycles), (ii) anaerobic reaction period 127

(420 min for the short cycle and 1020 min for the long cycle), (iii) settling period (30 min for 128

both cycles), and (iv) withdrawal period (40 min for both cycles). 129

In the enrichment period, the SBR was fed an autotrophic medium (Dapena-Mora et 130

al., 2004) and operated at 35 °C. Concentrations of ammonium and nitrite in the synthetic 131

medium ranged from 40 mg·L-1 to 80 mg·L-1 and from 30 mg·L-1 to 60 mg·L-1, respectively. 132

The nitrite to ammonium ratio was kept around 1.32. The anammox enrichment period lasted 133

160 days. In the following periods (shown in Table 1), the reactor was fed with anaerobically 134

pretreated municipal wastewater containing on average 109 mg COD·L-1, 38.4 mg N-NH4+·L-135

1 and 170 mgCaCO3·L-1, when operated at 35 °C, 25 °C, and 20 °C (phases 1, 2, and 3, 136

respectively). The characteristics of the anaerobic effluent (from a demo-scale upflow 137

anaerobic sludge blanket (UASB) reactor) were previously described by Leal et al. (2016). In 138

phases 1, 2, and 3, the reactor was fed with real anaerobic effluent amended with nitrite (from 139

60 mg·L-1 to 80 mg·L-1, in order to keep the nitrite to ammonium ratio around 1.4 to 1.8 140

therefore providing sufficient nitrite for anammox and denitrifying bacteria. Considering the 141

stoichiometry of the anammox reaction (reported by Strous et al., 1998), it is possible to 142

estimate an alkalinity consumption of 0.28 mg HCO3- (0.23 mg CaCO3)/mgNH4

+-N oxidized 143

in the anammox reaction. Thus, the alkalinity present in the real anaerobic effluent was high 144

enough to support the growth of anammox bacteria. 145

The temperature range (20–35 °C) was chosen to simulate the typical climate 146

conditions prevailing in tropical regions (e.g. Brazil). In addition, the annual average 147

temperature found in the city of Belo Horizonte, MG, Brazil, where this research was 148

conducted, is around 23 °C. Therefore, the results could further support the implementation of 149

the anammox process at ambient temperatures found in tropical regions worldwide. 150

Influent and effluent samples were collected four times a week to monitor the 151

concentrations of ammonium, nitrite, and COD. Analyses were performed according to 152

Standard Methods for the Examination of Water and Wastewater (APHA, 2012). Biomass 153

samples were taken from the reactor at the end of each operational phase to investigate the 154

microbial diversity inside the reactor. 155

Statistical analyses were performed to assess whether the different temperatures 156

applied to the SBR altered the nitrite, ammonium, and COD removal efficiencies. The data 157

were subjected to the Shapiro–Wilk normality test. Because they fit the normal distribution 158

pattern, they were analysed via ANOVA testing, followed by a multiple comparison test of 159

means (α = 5%), using Statistica 7 software. 160

2.2 Biomass sampling 161

During the experiment, six samples (40 mL) were collected from the anammox SBR: 162

sample I (inoculum), sample II (at the end of anammox enrichment period), sample III (at the 163

end of phase 1 at 35°C), sample IV (at the end of phase 2 at 25°C) and samples V and VI after 164

30 days and 45 days in phase 3 at 20°C. Samples were transferred to 50 mL Falcon tubes, 165

centrifuged (4,000 rpm for 20 min), resuspended in phosphate buffer saline (PBS) solution 166

(1:10 dilution of 70 mM Na2HPO4, 34 mM NaH2PO4, and 1300 mM NaCl, pH 7.2), 167

homogenized, centrifuged again, and stored at -20 °C until use. Total DNA was extracted 168

from the samples with a PowerSoil® DNA Isolation Kit (MOBIO Laboratories, USA) and 169

quantified (Qubit, LifeTechnologies). DNA purity was measured using a spectrophotometer 170

(NanoDrop 1000, Thermo Scientific). 171

172

2.3 Denaturing gradient gel electrophoresis (DGGE) 173

DGGE was used to perform a preliminary analysis of the bacterial community profile 174

and to monitor the operational phases of the SBR, as previously described. To prepare the 175

extracted DNA for DGGE, PCR was performed with universal primers (1055F/1392R-GC) 176

for the V8 region of the 16S rRNA gene, according to the methods of Ferris et al. (1996). The 177

PCR product of each sample was subjected to electrophoresis on 1% agarose. Quantification 178

of the PCR products was performed using the software ImageJ (Thermo Scientific). DNA 179

samples (400 ng) from each pool were used for DGGE (DCode Universal Mutation Detection 180

System, Bio-Rad Laboratories) in an 8% polyacrylamide gel with 45% to 75% denaturing 181

gradient for 16.5 h at 75 volts. 182

The gel was stained with SYBR Gold (LifeTechnologies) and analysed with the 183

BioNumerics 7.1 software (Applied Maths). Band profiles were compared using the Dice 184

similarity coefficient, and a dendrogram was generated using the unweighted pair group 185

method with arithmetic mean (UPGMA), with 1% position tolerance. 186

Bands were excised, eluted in 50 µL of ultrapure water, and incubated at 4 °C 187

overnight. The DNA was then re-amplified with the same primer pair, excluding the GC-188

clamp, as described above. The PCR products were purified (Wizard® SV Gel and PCR 189

Clean-Up System, Promega) and quantified, as described for the first PCR. Sequencing 190

reactions were performed by Macrogen Inc. in a 3730xl sequencer. The obtained sequences 191

were analysed with Geneious 8.04 software (Biomatters Ltd.) and compared to the Ribosomal 192

Database Project (RDP) and National Center for Biotechnology Information (NCBI) 193

databases with RDP Classifier and BLASTn tools. 194

195

2.4 Deep sequencing analysis via Ion Torrent sequencing 196

To investigate the microbial diversity present in the reactor more comprehensively, a 197

deep sequence analysis was carried out. 198

DNA extracted from the inoculum and biomass samples taken from the SBR at the end 199

of each operational phase was used for PCR amplification (with primers 515F and 926R, 200

targeting the V4 and most of the V5 region of the 16S rRNA gene of archaea and bacteria), 201

library construction, and sequencing using the Ion Torrent platform. The sequencing on the 202

Ion Torrent PGM (400 bp) was performed at the School of Engineering & Geosciences of 203

Newcastle University with the 316™ ion chip, following the manufacturer's instructions (Life 204

Technologies, USA). Raw sequences were analysed using the QIIME (v 1.7.0) bioinformatics 205

pipeline. After quality filtering (minimum quality score of 20, perfect match to sequence 206

barcode and primer), the remaining sequences were clustered into operational taxonomic units 207

(OTUs) at a 97% similarity level, and representative sequences were taxonomically assigned 208

using the SILVA database (Quast et al., 2013). The results were given in relative abundance 209

(%). Simpson's reciprocal and Shannon–Weaver diversity indices were calculated using 210

PAST 3.0. Distances between samples were computed based on the Bray–Curtis ecological 211

index. Raw sequences were deposited in the NCBI database (project accession number 212

SUB3374973). 213

2.5 Quantitative PCR (qPCR) 214

The abundance of anammox bacteria and denitrifiers was investigated via real-time 215

quantitative PCR (qPCR) using SYBR Green assays of biomass samples taken from the SBR. 216

qPCR assays were conducted in a real-time PCR thermal cycler (Applied Biosystems 7500 217

instrument) using primers Pla46F and Amx667R for anammox bacteria; and nosZF and 218

nosZ1622R for nitrous oxide reductase gene of denitrifiers. PCR assays were performed as 219

described previously (Leal et al., 2016). 220

221

3. Results and Discussion 222

3.1 Nitrogen removal in the sequencing batch reactor (SBR) during different operational 223

phases 224

The SBR was monitored over a period of 308 days to evaluate the effect of different 225

temperatures and the influence of real anaerobic effluent on the anammox process and 226

ammonium and nitrite removal efficiencies. Four different stages were assessed: the 227

anammox enrichment period (in which the reactor was fed with autotrophic medium at 35 228

°C), and the subsequent phases in which the reactor was fed with real anaerobic effluent 229

supplemented with nitrite and operated at 35 °C (Phase 1), 25 °C (Phase 2), and 20 °C (Phase 230

3). 231

During the anammox enrichment period (160 days), the anammox activity was 232

detected in the reactor after 82 days of cultivation. Additionally, higher average ammonium 233

(96.5%) and nitrite (98.2%) removal efficiencies were achieved (Table 2), indicating that the 234

anammox condition was established. The stoichiometric coefficient (consumption of N-NO2-/ 235

consumption of N-NH4+) determined during the enrichment period was 1.47 (as shown in 236

Table 3), which is close to 1.32 and 1.46 reported previously by Strous et al. (1998) and Quan 237

et al. (2008), respectively. The average value found for the coefficient of N-NO3- 238

production/N-NH4+ consumption was 0.35, close to that (0.26) reported by Strous et al. 239

(1998). 240

241

3.2 Performance of the sequencing batch reactor (SBR) during operational phase 1 242

(temperature of 35 °C) 243

After the anammox enrichment period, the subsequent phases started. The 244

performance of the SBR over the 148 days of phases 1, 2, and 3 is shown in Fig. 1. 245

In phase 1, the reactor was fed with real anaerobic effluent (containing on average 246

42.3 mg·L-1 of ammonium, 130 mg·L-1 of COD, and 0.5 mg·L-1 of nitrate) supplemented with 247

nitrite (78.2 mg·L-1) at a temperature of 35 °C. Over the monitoring period (40 days), the 248

average removal efficiencies for ammonium and nitrite were 97.7% and 93.9%, respectively 249

(Fig. 1), and the average nitrate production value was 14.4 mg·L-1 (Table 2). These results 250

demonstrated that the enriched anammox biomass easily adapted to the transition between the 251

synthetic effluent and the real anaerobic effluent at 35 °C. The stoichiometric coefficient 252

(consumption of N-NO2-/consumption of N-NH4

+) determined during phase 1 was 1.77 (as 253

shown in Table 2), which is higher than that (1.32) reported previously by Strous et al. (1998) 254

for anammox reaction, indicating that more nitrite was being consumed, likely by 255

heterotrophic denitrifiers that were present in the biomass and were using the organic matter 256

(COD) present in the anaerobic effluent. Tang et al. (2010) reported a coefficient of 2.09 257

when investigating the effect of organic matter on nitrogen removal during the anammox 258

process. 259

The COD was monitored during Phases 1, 2, and 3 to determine the organic matter 260

consumption (COD removal). The average COD removal efficiency in phase 1 was 61.8% 261

(Fig. 1 and Table 2), indicating that heterotrophic denitrifiers were consuming COD to reduce 262

nitrite, thus corroborating the higher nitrite consumption observed. Nevertheless, the nitrogen 263

mass balance showed that nitrite consumption via the anammox process was higher (Table 2). 264

265

3.3 Performance of the sequencing batch reactor (SBR) during operational phase 2 266

(temperature of 25 °C) 267

In phase 2, the reactor was operated for 63 days at 25 °C. The temperature of the 268

reactor was decreased from 35 °C (Phase 1) to 25 °C, without any acclimatization step. The 269

aim was to determine if the anammox metabolism would be affected by the temperature decay 270

and if, over the monitoring days, the stability verified previously in the reactor operation 271

would be maintained. The average removal efficiencies for ammonium and nitrite were 100% 272

and 95.7%, respectively (Fig. 1), and the average nitrate production value was 14.2 mg·L-1 273

(Table 2). The stoichiometric coefficient (consumption of N-NO2-/consumption of N-NH4

+) 274

determined during phase 2 was similar (1.70) to that observed in phase 1 (1.77), and the COD 275

removal efficiency was higher (75.7%) than that previously noticed. Therefore, the 276

temperature decay from 35 °C to 25 °C together with the remaining organic carbon in the 277

influent (100 mg COD·L-1) did not cause any adverse effect on the anammox process. 278

Moreover, it favoured nitrite consumption by denitrifiers, as was shown by the nitrogen mass 279

balance calculated for all operational phases (Table 2). 280

The average value found for the coefficient of N-NO3- production/N-NH4

+ 281

consumption was 0.44, which is higher than the stoichiometric values of 0.26 (reported by 282

Strous et al. (1998), and 0.18 reported by Hu et al. (2013), respectively when operating a 283

nitritation-anammox bioreactor at 25 °C. 284

285

3.4 Performance of the sequencing batch reactor (SBR) during operational phase 3 286

(temperature of 20 °C) 287

In phase 3, the reactor was operated for 45 days at 20 °C. The temperature of the 288

reactor was decreased in one step from 25 °C (Phase 2) to 20 °C. The aim was to verify the 289

anammox process resilience at this temperature. The average removal efficiencies for 290

ammonium and nitrite were 98.5% and 89.3%, respectively (Fig. 1). It is important to mention 291

that the influent ammonium concentration was lower (32.3 mg·L-1) compared to that of 292

previous phases (around 40 mg·L-1) and the nitrite concentration added to the reactor was 293

similar to that in phase 2 (around 70 mg·L-1). The stoichiometric coefficient (consumption of 294

N-NO2-/consumption of N-NH4

+) determined during phase 3 was higher (1.96) than that 295

observed in phase 2 (1.70) (Table 2), indicating that more nitrite was being consumed during 296

the denitrification process. Ma et al. (2013) also reported high values for this coefficient when 297

operating an SBR in lower temperatures (16 °C), suggesting that temperature reduction might 298

favour heterotrophic bacteria. Regarding the coefficient of N-NO3- production/N-NH4

+ 299

consumption, the value determined (0.38) was close to the values observed in previous 300

phases. 301

The average COD removal efficiency in phase 3 was 65% (Table 2) with influent 302

COD concentration of 96.7 mg·L-1. Taken together, the results of phase 3 indicated that at 20 303

°C, more nitrite was consumed by the heterotrophic denitrifiers in comparison to that in 304

previous phases, showing that temperature reduction probably enhanced the denitrification 305

process. However, nitrogen removal (ammonium and nitrite) via the anammox process 306

remained high (although some instability in nitrite removal was observed; Fig. 1), suggesting 307

that anammox bacteria were slightly affected by the temperature reduction and the anammox 308

process prevailed in the reactor. 309

When comparing results obtained among experimental phases, it was observed that 310

ammonium removal efficiencies were high for the three phases (average values above 95%). 311

Nevertheless, statistical analysis (ANOVA) detected differences in ammonium removal 312

efficiencies between phase 2 and phase 3 (α = 5%; 0.000031). This result indicated that the 313

temperature decrease (from 35 °C to 25 °C) between phases 1 and 2 did not cause any adverse 314

effect on the anammox process, but the 5 °C decrease (from 25 °C to 20 °C) was able to 315

impact ammonium removal. In this case, more nitrite was consumed in the denitrification 316

process in phase 3. It is important to mention that 20 °C is the minimum temperature reported 317

in the literature for the establishment of a warm anammox process (from 20 °C to 43 °C 318

according to Gao and Tao (2011) and Strous et al. (1999). Tao et al. (2012) compared 319

variations in pH and temperature in a nitritation/anammox reactor and concluded that the 320

effects of temperature variation were higher than those observed for pH variation. 321

Concerning nitrite removal efficiencies, statistical differences between the average 322

values obtained for phase 3 and phases 1 and 2 were detected (α = 5%; p = 0.000001), 323

indicating that the nitrogen removal efficiency observed in phase 3 was different from that of 324

the other phases, which corroborates the mass balance results presented above (Table 2). 325

Therefore, the decrease in temperature from 25 °C to 20 °C affected the nitrite and 326

ammonium removal as well. 327

Regarding COD removal efficiencies, no statistical differences between the phases 328

was detected (α = 5%; p = 0.0566). However, when comparing total nitrogen removal 329

efficiency, statistical differences between the average values obtained from phase 2 and those 330

from phase 1 (α = 5%; p = 0.0025) and phase 3 (α = 5%; p = 0.000001) were detected. Results 331

also showed that phases 1 (T = 35 °C) and 3 (T = 20 °C) did not differ. This information is of 332

great relevance, because most studies have used and applied the anammox process for 333

nitrogen removal at temperatures above 30 °C. This study indicates that the conditions 334

maintained in phase 2 at 25 °C were better than those at 35 °C and 20 °C, showing that the 335

maximum ammonium and nitrite removal efficiencies were achieved at 25 °C. Nevertheless, 336

ammonium and nitrite removal efficiencies were still high at 20 °C (98.5% and 89.3, 337

respectively). 338

In summary, the results indicated the possibility of using the anammox process to 339

remove nitrogen from anaerobically pretreated municipal wastewater at ambient temperatures 340

in tropical regions. Therefore, common seasonal temperature fluctuations would likely not 341

affect the process stability considering future full-scale applications. 342

3.5 Microbial diversity in the sequencing batch reactor (SBR) at each operational phase 343

Sequencing using the Ion Torrent platform generated around 33,422 to 50,000 344

sequences per sample. Table 3 shows the Shannon–Weaver and Simpson’s diversity indices, 345

as well as the number of sequences and OTUs identified for each sample. Once the diversity 346

estimators based on the DGGE band profiles could be greatly influenced by non-ideal band 347

migration, the indices calculated from deep sequencing were considered ideal for diversity 348

analysis. 349

Both indices (Shannon–Weaver and Simpson’s diversity) indicated that the microbial 350

community in the reactor was diverse. Slight variations were noticed among experimental 351

phases, especially between the anammox enrichment period and phases 1, 2, and 3, when the 352

reactor was fed with real anaerobic effluent (Table 3). The number of reads (abundance) and 353

OTUs (richness) varied substantially in the reactor compared to the inoculum, indicating that 354

specific microbial groups involved in nitrogen removal (such as anammox bacteria) were 355

selected during the enrichment period with the synthetic medium. However, in phases 1, 2, 356

and 3 with the addition of real anaerobic effluent (containing ammonium and COD), the 357

richness increased (compared to the enrichment period), indicating that other groups involved 358

in COD and nitrogen removal were favoured. Reactor samples (II and III) showed higher 359

dominance values compared to those of the inoculum, which also showed the dominance of 360

certain microbial communities in the reactor. Nevertheless, the addition of the anaerobic 361

effluent and the temperature decrease (from 35 °C to 20 °C) affected the dominance and 362

increased the diversity, suggesting that other groups were favoured (as will be further 363

discussed below). 364

DGGE sequenced bands and a dendrogram are shown in Fig. 2. Three general clusters 365

were observed, which share 45.5%, 68.3%, and 78.1% similarity. The lower part of the 366

dendrogram consists of sample I (inoculum) and samples II, V and VI. Apparently, these 367

samples were more divergent compared to the samples taken prior to the temperature decrease 368

(II, III, and IV), which showed 78% similarity. In addition, samples from the reactor at 35 °C 369

and 25 °C (samples III and IV) had high similarity with each other (95%). Therefore, the 370

microbial community did not vary widely between phases 1 and 2, although the dendrogram 371

indicated a possible differentiation as a consequence of temperature decay (from 25 °C to 20 372

°C). These results are partially in accordance with the deep sequencing analysis, as discussed 373

below. 374

Distances between samples were calculated based on the Bray–Curtis ecological index 375

(Table 4). The inoculum sample (I) was distant from the reactor samples, which was expected 376

because the enrichment period with synthetic medium selected for specific groups involved in 377

nitrogen removal. The reactor samples from phases 1, 2, and 3 were closer to each other, 378

when compared to that of the enrichment period (Table 4), indicating that the addition of real 379

anaerobic effluent (containing COD) influenced the microbial community dynamics. Some 380

groups of microorganisms were favoured, such as Proteobacteria and Planctomycetes (Ca. 381

Anammoximicrobium), which increased in abundance (Fig. 3). This condition was reflected 382

in the deep sequencing results as a larger distance between sample II and other samples (II, 383

IV, and V). Additionally, the temperature decrease also affected microbial composition, as 384

sample III (at 35 °C) showed around 60% similarity with sample IV (at 25 °C) and sample V 385

(at 20 °C). However, temperature decay from 25 °C to 20 °C seemed not to affect the 386

microbial community, as samples IV (at 25 °C) and V (at 20 °C) were more similar (around 387

80%) to each other than to other samples. 388

389

3.6 Predominant microbial populations detected by denaturing gradient gel 390

electrophoresis (DGGE) 391

Sequences from the DGGE bands (Fig. 2) were compared with the NCBI and RDP 392

databases. The results are summarised in Table 5. There was a significant presence of 393

Proteobacteria (12 of 27 sequences), specifically Betaproteobacteria. Within this class, 394

members of Rhodocyclaceae and Burkholderiales, which include many genera of denitrifying 395

bacteria, dominated. In addition, there were sequences classified as anammox bacteria (Ca. 396

Brocadia, 7 bands) and as Chloroflexi (8 of 27 sequences). The phyla Proteobacteria and 397

Chloroflexi are frequently found in anammox reactors together with Planctomycetes (Leal et 398

al., 2016; Pereira et al., 2017; Persson et al., 2016). Bands 8, 11, 16, 21, and 25 belonging to 399

the genus Denitratisoma (Rhodocyclaceae), represent organisms that can perform 400

heterotrophic denitrification. Denitratisoma oestradiolicum is a denitrifying bacteria that was 401

isolated from a municipal wastewater treatment plant in Germany (Fahrbach et al., 2006). The 402

corresponding bands were detected in all samples (excluding the inoculum), which may 403

indicate the constant presence of these bacteria, even in the anammox enrichment period 404

when the reactor was fed with autotrophic medium. Leal et al. (2016), using an anammox 405

SBR fed with real anaerobic effluent, reported the presence of this genus, suggesting that it 406

might be involved in COD removal. Other sequences related to denitrifying bacteria detected 407

in the present study were Acidovorax sp., Burkholderia sp., and Burkholderiales (Fig. 2 and 408

Table 5). 409

The DGGE results showed that bands 4, 5, 14, 15, 18, 19, and 24, with sequences closely 410

related to the anammox bacteria Ca. Brocadia caroliniensis and Ca. Brocadia sp., were found 411

in every phase of the study (excluding the inoculum), even when the reactor was operated at 412

20 °C (Fig. 2 and Table 5). These species are commonly found in wastewater treatment 413

systems (Hu et al., 2010). Ca. Brocadia caroliniensis has been reported in other anammox 414

reactors fed with anaerobic effluent (Leal et al., 2016) and PN/anammox reactors treating 415

anaerobic sludge digestate (Persson et al., 2016). Thus, it was suggested that Ca. Brocadia has 416

broad ecophysiology and competitiveness, which justify their wide occurrence in different 417

wastewater treatment systems (Oshiki et al., 2011). 418

Bacteria within the phylum Chloroflexi related to Anaerolineacea were observed in the 419

present study in all samples (DGGE bands 3, 6, 10, 12, 13, 23, and 27; Table 4). This phylum 420

includes bacteria with diversified metabolism (Hug et al., 2013) that are frequently found in 421

anammox reactors (Leal et al., 2016; Pereira et al., 2017). They can degrade starch, sugars, 422

and peptides (Hug et al., 2013) and thus might have been involved in the COD removal 423

observed in the present study. Although Anaerolineacea sequences were detected in all 424

samples using the DGGE, the intensity of the DNA bands decreased in samples taken from 425

the reactor at 25 °C and 20 °C, suggesting that temperature decay from 35 °C to 20 °C 426

negatively affected these bacteria. 427

3.7 Microbial composition as revealed by Ion Torrent sequencing 428

Regarding the deep sequencing data, the major phyla of bacteria identified in the SBR 429

samples were Proteobacteria (corresponded to 30.2–47.8% of the total number of sequences), 430

Chloroflexi (16.4% to 36.4%), Chlorobi (7.0% to 14.6%), Planctomycetes (3.5% to 7.2%), 431

and Candidate division WS3 (7.0% to 14.6%), Together, these five phyla accounted for 432

approximately 82.5% to 85.1% of the diversity, and they were present in all the collected 433

samples, as shown in Fig. 3. 434

The Ion Torrent sequencing results confirmed the PCR-DGGE results and showed that 435

temperature decrease (from 35 °C to 20 °C) had a negative effect on the Chloroflexi group, as 436

its relative abundance of sequences decreased from 36% to 16% of the total reads. Members 437

of the phylum Chloroflexi include bacteria with diversified metabolism (Hug et al., 2013). 438

Within this phylum, sequences related to Anaerolineales were detected in high abundance 439

(25% of total reads at the end of phase 1 and 10.6% at the end of phases 2 and 3; see Table 6). 440

This order is composed of anaerobic and heterotrophic bacteria reported in different 441

environments including anammox reactors (Leal et al., 2016; Pereira et al., 2017). Wu et al. 442

(2016) reported that the abundance of Chloroflexi (unknown Anaerolineales) decreased (from 443

32% to 23%) when the temperature of a PN/anammox reactor treating anaerobic sludge 444

digestate was lowered from 35 °C to 13 °C. According to these authors, the majority of 445

bacteria belonging to this phylum is thermophilic and therefore, at low temperatures, the 446

activity and metabolism of these bacteria would be lower and slow. Moreover, in reactors 447

with anammox activity, the main carbon source used by Chloroflexi would come from cell 448

lysis and decay. Nevertheless, cell decay rates can be lower at low temperatures, thus limiting 449

the growth of Chloroflexi (Wu et al., 2016), as was also observed in the present study. 450

Within the phylum Proteobacteria, the most abundant orders were Rhodocyclales, 451

Rhodospirillales, Burkholderiales, Xanthomonadales, Myxococcales, Nitrosomonadales, and 452

Hydrogenophilales (Fig. 4 and Table 6). Members of Burkholderiales, Rhodocyclales, 453

Xanthomonadales, Rhodospirillales, and Pseudomonadales are bacteria capable of 454

performing heterotrophic denitrification (Heylen et al., 2008). In this way, the SBR operation 455

with real anaerobic effluent (containing COD) favoured the growth of denitrifying bacteria, as 456

the relative abundance of the orders Rhodocyclales, Rhodospirillales, Burkholderiales, 457

Pseudomonadales, and Xanthomonadales increased from 13.7% (at the end of the anammox 458

enrichment period) to 30% of the total reads (at the end of phase 3 - temperature of 20 °C), as 459

shown in Table 6. Within Rhodocyclales (Rhodocyclaceae), the dominant genera identified in 460

all the reactor samples were Sulfuritalea (1.7 to 9.8%) and Denitratisoma (1.0 to 5.0%), 461

confirming the PCR-DGGE results. The genus Sulfuritalea was dominant and the reduction of 462

temperature increased the abundance of this group (from 4.7% to 9.8% of the total reads, 463

Table 6). This genus is described as a facultative autotroph capable of oxidizing thiosulfate, 464

elemental sulfur, and hydrogen as sole energy sources for autotrophic growth. These bacteria 465

can also utilize nitrate as an electron acceptor (Kojima and Fukui, 2011). Additionally, in 466

activated sludge, this genus has a broad substrate uptake profile—assimilating pyruvate, 467

acetate, propionate, amino acids, and glucose (Mcllroy et al., 2015). These bacteria are also 468

able to utilize nitrite as an electron acceptor, likely indicating a role in denitrification (McIlroy 469

et al., 2015). Denitratisoma species can perform oxidative metabolism through nitrate and 470

nitrite respiration, indicating their participation as denitrifiers in this reactor. Temperature 471

reduction from 35 °C to 25 °C also favoured this genus (Table 6). 472

The genus Thauera, identified as the functional bacteria for partial denitrification with 473

high nitrite production in previous studies (Cao et al., 2016; Du et al., 2017) was detected in 474

the present study at very low abundance (0.1% in phases 1, 2, and 3), indicating that partial 475

denitrification was likely not occurring in the reactor (and/or it was too low to be noticed). 476

Persson et al. (2016) also reported the presence of Rhodocyclales, Burkholderiales, 477

Rhizobiales, and Xanthomonadales, which have members that can perform denitrification in 478

wastewater treatment systems (Mcllroy et al., 2016). 479

Sequences related to Hydrogenophilales were detected in the SBR when the reactor 480

was fed with anaerobic effluent supplemented with nitrite, but the reduction of temperature 481

decreased the abundance of this group (from 3.1 to 1.2% of the total reads; Table 6). This 482

order contains bacteria, such as Thiobacillus denitrificans, that are capable of performing 483

autotrophic denitrification using hydrogen sulfide as an electron donor. Because the anaerobic 484

effluent used to fed the reactor contained dissolved hydrogen sulfide (from 1.0 to 17.0 mg·L-485

1), this may explain the presence of sulfide-oxidizing bacteria (Sulfuritalea and Thiobacillus) 486

in the SBR. 487

Concerning bacteria involved in autotrophic nitrogen removal, ammonia-oxidizing 488

bacteria (AOB), nitrite-oxidizing bacteria (NOB), and anammox bacteria were detected in all 489

the samples from the reactor (Table 6). All identified AOB sequences were affiliated to 490

Nitrosomonadales (2.0% to 2.5%). The major sequence detected was affiliated to 491

Nitrosomonadaceae (Table 6). Aerobic nitrifiers, such as Nitrosomonas, are facultative 492

anaerobes that can also metabolize ammonium and nitrite anaerobically, but their metabolic 493

rate is at least 30 times lower than that of Ca. B. anammoxidans (Kuenen and Jetten, 2001). 494

All identified NOB were affiliated to Nitrospirales, belonging to the genus Nitrospira (1.2% 495

to 1.5% of total sequences), and were detected in all the samples from the reactor (Table 6). 496

Complete ammonia oxidizers (Comammox) Nitrospira, which oxidize ammonium to nitrate 497

on their own, could also be present among the sequences retrieved from the reactor. However, 498

further analysis should be performed to confirm this by using amoA specific primers, as 499

described by Pjevac et al. (2017). 500

Persson et al. (2016) also detected sequences for the AOB Nitrosomonas europaea/N. 501

eutropha (0.2 to 0.35%) and the NOB Ca. Nitrotoga sp. and Nitrospirales (0.05% to 0.2%), 502

using high-throughput amplicon sequencing of the 16S rRNA gene (V4 region). 503

The anammox bacteria were affiliated with the genus Ca. Brocadia and Ca. 504

Anammoximicrobium (Table 6), with Ca. Anammoximicrobium the dominant anammox 505

population in the reactor after the addition of the real anaerobic effluent. Temperature 506

decrease (from 35 °C to 25 °C) and addition of real anaerobic effluent (containing COD) 507

seemed not to adversely impact the anammox bacteria, on the contrary seemed to select for 508

Ca. Anammoximicrobium. Moreover, the abundance of anammox populations increased from 509

phases 1 (sample III) to 2 (sample IV) (from 0.9% to 3.0%). However, when the temperature 510

was lowered from 25 °C to 20 °C, this group was affected and decreased in abundance (Table 511

6). Khramenkov et al. (2013) reported that the activity of Ca. Anammoximicrobium 512

(determined by ammonium and nitrite consumption rates) was higher at 25 °C when 513

compared to at 35 °C, therefore indicating that temperature can select for different anammox 514

populations and Ca. Anammoximicrobium is favoured at 25 °C, as was observed in the 515

present study. 516

It is important to mention that anammox sequences retrieved by sequencing the DGGE 517

bands (Ca. Brocadia sp. and Ca. Brocadia caroliniensis) were found in low relative frequency 518

(ranging from 0.2% to 0.7% of total reads) through Ion Torrent sequencing. The primer pair 519

used for the DGGE amplified the bacterial V8 hypervariable region, whereas that used for 520

deep sequencing targeted the V4–V5 region of the 16S rRNA gene of both bacteria and 521

archaea. Because different primers can produce varying results and the cell copy number of 522

target genes can influence relative abundances (Albertsen et al., 2015), quantitative real-time 523

PCR (qPCR) analysis was performed. It provided a more realistic determination of anammox 524

bacteria abundance. 525

Concerning possible secondary processes, sequences related to NC 10 phylum bacteria 526

(Ca. Methylomirabilis oxyfera) and related to archaea (Ca. Methanoperedens nitroreducens) 527

known to perform anaerobic methane oxidation coupled to denitrification (DAMO) process 528

were not detected, indicating that this process was not occurring in the reactor despite 529

dissolved methane was present in the anaerobic effluent. In addition, dissimilatory nitrate 530

reduction to ammonium (DNRA), which Ca. Brocadia sapporoensis has the genetic potential 531

to perform (Narita et al., 2017), was likely not occurring in the reactor (and/or it was too low 532

to be noticed) since more nitrate was produced than it was consumed (as shown in Table 2). 533

In general, the deep sequencing results indicated the presence of a metabolically 534

diverse microbial community in the SBR, with sequences related to anammox bacteria, AOB, 535

NOB, autotrophic denitrifiers (such as Thiobacillus) and heterotrophic bacteria (fermenting 536

bacteria such as Firmicutes, Bacteroidetes, Chloroflexi and heterotrophic denitrifiers such as 537

Denitratisoma, among others) (Table 6). Some genera and orders were favoured with the 538

temperature decrease, such as Ca. Brocadia, Ca. Anammoximicrobium, Burkholderiales, 539

Myxococcales, Rhodocyclales (Denitratisoma and Sulfuritalea), Xanthomonadales, 540

Chlorobiales, and Ignavibacteriales, whereas others (Anaerolineales and Clostridiales) were 541

negatively affected (Table 6), indicating that these groups are better adapted to a temperature 542

of 25 °C, rather than 35 °C. Temperature reduction (from 35 °C to 25 °C) seemed not to affect 543

AOB and NOB, since the relative abundance of these groups remained stable (Table 6). 544

Similar results were reported by Gonzalez-Martinez et al. (2015) for a CANON 545

reactor treating diluted anaerobic digestion supernatant. These authors observed that the 546

acclimatization of the biomass from 35 °C to 25 °C selected for Ca. Brocadia and Ca. 547

Anammoxoglobus and increased the abundance of some genera (Anaerolinea, 548

Acidobacterium, Chloroflexi, Fluviicola, and Prosthecobacter). 549

550

3.8 Abundance of anammox and denitrifying bacteria determined by quantitative PCR 551

(qPCR) 552

Anammox and denitrifying bacterial abundances in the SBR subjected to decreasing 553

temperatures were quantified by real-time qPCR performed on the 16S rRNA of anammox 554

and nosZ genes, respectively (as shown in Fig. 5). In the inoculum sample, the concentration 555

of anammox bacteria in relation to that of denitrifiers was three orders of magnitude lower 556

(2.9 x 106 and 2.8 x 109 gene copies per g of sludge, respectively). However, anammox 16S 557

rRNA gene concentration increased after the enrichment period, corresponding to 558

approximately 10% of total bacteria 16SrRNA gene copies, and reached values similar to that 559

for nosZ gene copies (1 x 109gene copies per g of sludge, Fig. 5). Park et al. (2010) operating 560

a 4L CANON reactor fed with raw anaerobic digestate observed, after 400 days of operation, 561

similar anammox bacteria abundance (10% of total bacteria 16SrRNA copies). Cao et al. 562

(2016) reported that Ca. Brocadia was in low proportion (2.37%) in an anammox UASB 563

reactor treating high-strength wastewater, though high efficiency of ammonium to nitrite 564

removal activities was obtained. Du et al. (2017) reported that Planctomycetes was detected 565

with the abundance of 7.39% in a Denitrifying Ammonium Oxidation (DEAMOX) reactor 566

treating domestic wastewater (amended with nitrate). Therefore, the anammox cultivation in 567

the present study was successful (with stable anammox activity, Table 2) and the application 568

of real anaerobic effluent to the reactor favoured the growth of denitrifiers (Table 6). 569

However, anammox bacterial concentrations were still high during phases 1, 2, and 3 (from 570

1.5 x 108 to 1.3 x 109 gene copies per g of sludge), indicating that both groups dominated the 571

bacterial community and coexisted (Fig. 5). A decrease in the number of anammox 16S rRNA 572

gene copies (from 8.6 to 1.5 x 108 gene copies per g of sludge) was observed when the reactor 573

was operated at 20 °C (phase 3), but nitrogen removal efficiency remained high (90%) (Table 574

2). 575

576

3.9 Practical implications of this work and future perspectives 577

Use of the anammox process for mainstream municipal wastewater treatment (i.e. 578

nitrogen removal from anaerobic effluents) seems to be feasible, taking into account the 579

typical tropical temperature variation. It is a remarkable achievement, as anaerobic sewage 580

treatment could be considered a consolidated technology in many warm climate regions 581

(Chernicharo et al., 2015). A recent survey estimated that around 40% of sewage treatment 582

plants (STPs) in operation in the most populated Brazilian region use anaerobic technology 583

(UASB reactors) as the first stage in the treatment process (Chernicharo et al., 2017, in press). 584

This could be considered the biggest anaerobic park of UASB reactors treating sewage around 585

the world. 586

Despite such wide use of UASB reactors in countries such as Brazil, there are some 587

constraints that still need to be released, such as the presence of ammonia in the effluent, in 588

order to improve the performance of these reactors. Therefore, establishing the anammox 589

process in the post-treatment step seems to be an achievable possibility. In this case, for future 590

full-scale STPs, a possible technological flowsheet can be associated with sponge-bed 591

trickling filter (SBTF) post-UASB reactors (Bressani-Ribeiro et al., 2017; MacConnell et al., 592

2015) for nitrogen removal. Recent studies indicated the cultivation of anammox bacteria 593

(Sànchez-Guillèn et al., 2015a) and autotrophic nitrogen removal over nitrite in SBTFs 594

(Chuang et al., 2008; Sànchez-Guillèn et al., 2015b) as proof of concept. Therefore, the 595

current study presents an important complementary contribution to future mainstream 596

anammox applications in warm climate regions. 597

598

4. Conclusions 599

The SBR performance showed that anammox bacteria coexisted with heterotrophic 600

bacteria and despite temperature decrease (35–20 °C), ammonium and nitrite removal 601

efficiencies were high (90%). Temperature decay changed the microbial community structure 602

and diversity. Anammox bacteria (Ca. Brocadia and Ca. Anammoximicrobium) and 603

denitrifiers (Burkholderiales, Myxococcales, Xanthomonadales, Pseudomonadales, 604

Denitratisoma and Sulfuritalea) were favoured when the temperature was lowered from 35 °C 605

to 25 °C; whereas Anaerolineales and Clostridiales were negatively affected. AOB and NOB 606

abundance remained stable. The feasibility of applying the anammox process to mainstream 607

municipal wastewater treatment (nitrogen removal from anaerobic effluents) at typical 608

tropical temperatures was demonstrated. 609

610

Conflict of interest 611

All authors declare that they have no conflict of interest. 612

Acknowledgments 613

We are thankful to the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior 614

(CAPES) [grant number 486-2014]; Fundação de Amparo a Pesquisa do Estado de Minas 615

Gerais (FAPEMIG) [grant number 02669-14]; Conselho Nacional de Desenvolvimento 616

Científico e Tecnológico (CNPq) [grant number 481405-2013-5]; Financiadora de Estudos e 617

Projetos (FINEP); Instituto Nacional de Ciência e Tecnologia em Estações Sustentáveis de 618

Tratamento de Esgoto-INCT ETEs Sustentáveis and Project Global Innovation Partnership to 619

Investigate, Restore and Protect the Urban Water Environment, funded by the British Council. 620

621

References 622

1. Albertsen, M., Karst, S.M., Ziegler, A.S., Kirkegaard, R.H., Nielsen, P.H., 2015. Back to 623 basics–the influence of DNA extraction and primer choice on phylogenetic analysis of 624 activated sludge communities. PLOS ONE 10(7), e0132783. 625

2. Ali, M., Okabe, S., 2015. Anammox-based technologies for nitrogen removal: Advances in 626 process start-up and remaining issues. Chemosphere 141: 144–153. 627 https://doi.org/10.1016/j.chemosphere.2015.06.094 628

3. APHA, 2012. Standard Methods for the Examination of Water and Wastewater, 629 22nd ed. American Public Health Association, Washington, DC, USA. 630

4. Bressani-Ribeiro, T., Brandt, E.M.F., Almeida, P.G.S., Flórez, C.A.D, Chernicharo, 631 C.A.L., 2017. Technological improvements in compact UASB/SBTF systems for 632 decentralized sewage treatment in developing countries. Desalination Water Treat. 91, 633 112–120. 634

5. Cao, S.,Du, R.,Li, B., Ren, N., Peng, Y., 2016. High-throughput profiling of microbial 635 community structures in na ANAMMOX-UASB reactor treating high-strength 636 wastewater. Appl. Microbiol Biotechnol. 100: 6457–6467. 637

6. Chernicharo, C.A.L., Van Lier, J.B., Noyola, A., Bressani-Ribeiro, T., 2015. Anaerobic 638 sewage treatment: state of the art, constraints and challenges. Rev. Environ. Sci. 639 Biotechnol. 14(4), 649–679. 640

7. Chenicharo, A.A.L., Ribeiro, T.B., Garcia, G.B., Lermontov, A., Pereira, C.B., Platzer, 641 C.J., Possetti, G.R.C., Leites, M.A.L., Rosseto, R., 2017. Overview of sewage treatment 642 in the South, Southeast and Midwest regions of Brazil: most used technologies. 643 (Panorama do tratamento de esgoto sanitário nas regiões Sul, Sudeste e Centro-Oeste 644 do Brasil: tecnologias mais empregadas). Revista DAE (in press). (In Portuguese). 645

8. Chuang , H.P., Yamaguchi, T., Harada, H., Ohashi, A., 2008. Anoxic ammonium oxidation 646 by application of a down-flow hanging sponge (DHS) reactor. J. Environ. Eng. Manag. 647 18(6), 409–417. 648

9. Dapena-Mora, A., Van Hulle, S.W.H., Campos, J.L., Méndez, R., Vanrolleghem, P.A., 649 Jetten, M., 2004. Enrichment of Anammox biomass from municipal activated sludge: 650 Experimental and modelling results. J. Chem. Technol. Biotechnol. 79:1421–1428. 651 https://doi.org/10.1002/jctb.1148 652

10. Dosta, J., Fernández, I., Vázquez-Padín, J.R., Mosquera-Corral, A., Campos, J.L., Mata-653 Álvarez, J., Méndez, R., 2008. Short- and long-term effects of temperature on the 654 Anammox process. J. Hazard. Mater. 154:688–693. 655 https://doi.org/10.1016/j.jhazmat.2007.10.082 656

11. Du, R., Cao, S., Li, B., Wang, S., Peng, Y., 2017. Simultaneous domestic wastewater and 657 nitrate sewage treatment by DEnitrifying AMmonium Oxidation (DEAMOX) in 658 sequencing batch reactor. Chemosphere 174: 399–407. 659

12. Enwall, K., Philippot, L., Hallin, S., Salin, S., 2005. Activity and composition of the 660 denitrifying bacterial community respond differently to long-term fertilization. 661 Appl. Environ. Microbiol. 71, 8335–8343. 662

13. Fahrbach, M., Kuever, J., Meinke, R., Kämpfer, P., Hollender, J., 2006. Denitratisoma 663 oestradiolicum gen. nov., sp. nov., a 17 β-oestradiol-degrading, denitrifying 664 betaproteobacterium. Int. J. Syst. Evol. Microbiol. 56, 1547–1552. 665 https://doi.org/10.1099/ijs.0.63672-0 666

14. Ferris, M.J., Muyzer, G., Ward, D.M., 1996. Denaturing gradient gel electrophoresis 667 profiles of 16S rRNA-defined populations inhabiting a hot spring microbial mat 668 community. Appl. Environ. Microbiol. 62(2), 340–6. 669

15. Gao, D.W., Tao, Y., 2011. Versatility and application of anaerobic ammonium-oxidizing 670 bacteria. Appl. Microbiol. Biotechnol. 91(4), 887–894. https://doi.org/10.1007/s00253-671 011-3411-3 672

16. Gonzalez-Martinez, A., Rodriguez-Sanchez, A., Garcia-Ruiz, M.J., Muñoz-Palazon, B., 673 Cortes-Lorenzo, C., Osorio, F., Vahala, R., 2015. Performance and bacterial community 674 dynamics of a CANON bioreactor acclimated from high to low operational temperatures. 675 Chem. Eng. J.287, 557–567. https://doi.org/10.1016/j.cej.2015.11.081 676

17. Hendrickx, T.L.G., Wang, Y., Kampman, C., Zeeman, G., Temmink, H., Buisman, C.J.N., 677 2012. Autotrophic nitrogen removal from low strength waste water at low temperature. 678 Water Res. 46, 2187–2193. https://doi.org/10.1016/j.watres.2012.01.037 679

18. Heylen, K., Lebbe, L., de Vos, P., 2008. Acidovorax caeni sp. nov., a denitrifying species 680 with genetically diverse isolates from activated sludge. Int. J. Syst. Evol. Microbiol. 58, 681 73–77. https://doi.org/10.1099/ijs.0.65387-0 682

19. Hu, Z., Lotti, T., de Kreuk, M., Kleerebezem, R., van Loosdrecht, M., Kruit, J., Jetten, 683 M.S.M., Kartal, B., 2013. Nitrogen removal by a nitritation-anammox bioreactor at low 684 temperature. Appl. Environ. Microbiol. 79(8), 2807–2812. 685 https://doi.org/10.1128/AEM.03987-12 686

20. Hug, L.A., Castelle, C.J., Wrighton, K.C., Thomas, B.C., Sharon, I., Frischkorn, K.R., 687 Wiliams, K.H., Tringe, S.G., Banfield, J.F., 2013. Community genomic analyses 688 constrain the distribution of metabolic traits across the Chloroflexi phylum and indicate 689 roles in sediment carbon cycling. Microbiome 1, 22. https://doi.org/10.1186/2049-2618-690 1-22 691

21. Jenni, S., Vlaeminck, S.E., Morgenroth, E., Udert, K.M., 2014. Successful application of 692 nitritation/anammox to wastewater with elevated organic carbon to ammonia ratios. 693 Water Res. 49, 316–326. https://doi.org/10.1016/j.watres.2013.10.073 694

22. Kartal, B., De Almeida, N.M., Maalcke, W.J., Op den Camp, H.J.M., Jetten, M.S.M., 695 Keltjens, J.T., 2013. How to make a living from anaerobic ammonium oxidation. FEMS 696 Microbiol. Rev. 37(3), 428–461. https://doi.org/10.1111/1574-6976.12014 697

23. Kojima H, Fukui M 2011. Sulfuritalea hydrogenovorans gen. nov., sp. nov., a facultative 698 autotroph isolated from a freshwater lake. Int. J. Syst. Evol. Microbiol. 61(7), 1651–699 1655. doi:10.1099/ijs.0.024968-0. Epub 2010 Aug. 13. 700

24. Khramenkov, S.V., Kozlov, M.N., Kevbrina, M.V., Dorofeev, A.G., Kazakova, E.A., 701 Grachev, V.A., Kuznetsov, B.B., Polyakov, D.Y., Nilolaev, A., 2013. A novel bacterium 702 carrying out anaerobic ammonium oxidation in a reactor for biological treatment of the 703 filtrate of wastewater sludge. Microbiology 82(5), 628–636. 704

25. Kuenen, J.G. and Jetten,M.S.M. 2001. Extraordinary Anaerobic Ammonium-Oxidizing 705 Bacteria. ASM News. 67(9),456–463. 706

26. Lackner, S., Gilbert, E.M., Vlaeminck, S.E., Joss, A., Horn, H., van Loosdrecht, M.C.M., 707 2014. Full-scale partial nitritation/anammox experiences - An application survey. Water 708 Res. 55, 292–303. https://doi.org/10.1016/j.watres.2014.02.032 709

27. Leal, C.D., Pereira, A.D., Nunes, F.T., Ferreira, L.O., Coelho, A.C.C., Bicalho, S.K., 710 Chernicharo, C.A.L., Araújo, J.C., 2016. Anammox for nitrogen removal from 711 anaerobically pre-treated municipal wastewater: Effect of COD/N ratios on process 712 performance and bacterial community structure. Bioresour. Technol. 211, 257–266. 713 https://doi.org/10.1016/j.biortech.2016.03.107 714

28. Ma, B., Peng, Y., Zhang, S., Wang, J., Gan, Y., Chang, J., Wang, S., Zhu, G., 2013. 715 Performance of anammox UASB reactor treating low strength wastewater under 716 moderate and low temperatures. Bioresour. Technol. 129, 606–611. 717 https://doi.org/10.1016/j.biortech.2012.11.025 718

29. Ma, B., Wang, S., Cao, S., Miao, Y., Jia, F., Du, R., Peng, Y., 2016. Biological nitrogen 719 removal from sewage via anammox: Recent advances. Bioresour. Technol. 200, 981–720 990. https://doi.org/10.1016/j.biortech.2015.10.074 721

30. MacConnell, E.F.A.M., Almeida, P.G.S., Martins, K.L., Araujo, J.C., Chernicharo, C.A.L. 722 2015. Bacterial community involved in the nitrogen cycle in a down-flow sponge-based 723 trickling filter treating UASB effluent. Water Sci. Technol. 72, 116–122. 724

31. Mcllroy, S.J., Starnawska, A., Starnawski, P., Saunders, A.M., Nierychlo, M., Nielsen, 725 P.H., Nielsen, J.L., 2016. Identification of active denitrifiers in full-scale nutrient 726 removal wastewater treatment systems. Environ. Microbiol. 18(1), 50–64. 727 https://doi.org/10.1111/1462-2920.12614 728

32. McIlroy, S.J., Awata, T., Nierychlo, M., Albertsen, M., Kindaichi, T., Nielsen, P.H., 2015. 729 Characterization of the In Situ Ecophysiology of Novel Phylotypes in Nutrient Removal 730 Activated Sludge Treatment Plants. PLOS ONE 10(9), e0136424. doi: 731 10.1371/journal.pone.0136424. 732

33. Narita, Y., Zhang, L., Kimura, Z.I., Ali, M., Fujii, T., Okabe, S., 2017. Enrichment and 733 physiological characterization of na anaerobic ammonium-oxidizing bacterium 734 ‘Candidatus Brocadia sapporoensis’. Syst Appl Environ, 40(7), 448–457. 735

34.Oshiki, M., Shimokawa, M., Fujii, N., Satoh, H. Okabe, S., 2011. Physiological 736 characteristics of the anaerobic ammonium-oxidizing bacterium Candidatus Brocadia 737 sinica. Microbiology. 157, 1706 –1713. 738

35. Park, H., Rosenthal, A., Jezek R., Ramalingam K., Fillos J., Chandran K., 2010. Impact of 739 inocula and growth mode on the molecular microbial ecology of anaerobic ammonia 740 oxidation (anammox) bioreactor communities. Water Research 44: 5005–5013. 741

36. Pereira, A.D., Cabezas, A., Etchebehere, C., Chernicharo, C.A. de L., Araújo, J.C., 2017. 742 Microbial communities in anammox reactors: a review. Environ. Technol. Rev. 6(1), 74–743 93. https://doi.org/10.1080/21622515.2017.1304457 744

37. Persson, F., Suarez, C., Hermansson, M., Plaza, E., Sultana, R., Wilén, B.M., 2016. 745 Community structure of partial nitritation-anammox biofilms at decreasing substrate 746 concentrations and low temperature. Microb. Biotechnol. 10(4), 761–772. 747 https://doi.org/10.1111/1751-7915.12435 748

38. Pjevac,P., Schauberger, C., Poghosyan, L., Herbold, C.W., van Kessel, M.A.H.J., 749 Daebeler, A., Steinberger, M., Jetten, M.S.M., Lücker, S., Wagner, M., Daims H., 2017. 750 AmoA-Targeted Polymerase Chain Reaction Primers for the Specific Detection and 751 Quantification of Comammox Nitrospira in the Environment. Frontiers in Microbiology 752 1(8) article 1508. 753

39. Quan, Z.X., Rhee, S.K., Zuo, J.E., Yang, Y., Bae, J.W., Park, J.R., Lee, S.T., Park, Y.H., 754 2008. Diversity of ammonium-oxidizing bacteria in a granular sludge anaerobic 755 ammonium-oxidizing (anammox) reactor. Environ. Microbiol. 10, 3130–3139. 756

https://doi.org/10.1111/j.1462-2920.2008.01642.x 757

40. Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Peplies, J., 758 Glöckner, F.O., 2013 The SILVA ribosomal RNA gene database project: improved data 759 processing and web-based tools. Nucleic Acids Res. 41(D1), 590–596. 760

41. Sànchez-Guillèn, J.A., Cuèllar Guardado, P.R., Lopez Vazquez, C.M., Oliveira Cruz, 761 L.M., Brdjanovic, D., van Lier, J.B., 2015a. Anammox cultivation in a closed sponge-762 bed trickling filter. Bioresour. Technol. 186, 252–260. 763

42. Sànchez-Guillèn, J.A., Jayawardana, L.K.M.C.B., Lopez Vazquez, C.M., Oliveira Cruz, 764 L.M., Brdjanovic, D., van Lier, J.B., 2015b. Autotrophic nitrogen removal over nitrite in 765 a sponge-bed trickling filter. Bioresour. Technol. 187, 314–325. 766

43. Shen, L.D., Hu, A.H., Jin, R.C., Cheng, D.Q., Zheng, P., Xu, X.Y., Hu, B.L., 2012. 767 Enrichment of anammox bacteria from three sludge sources for the startup of 768 monosodium glutamate industrial wastewater treatment system. J. Hazard. Mater. 199–769 200, 193–199. https://doi.org/10.1016/j.jhazmat.2011.10.081 770

44. Strous, M., Heijnen, J.J., Kuenen, J.G., Jetten, M.S.M., 1998. The sequencing batch 771 reactor as a powerful tool for the study of slowly growing anaerobic ammonium-772 oxidizing microorganisms. Appl. Microbiol. Biotechnol. Appl. Environ. Microbiol. 50, 773 589–596. 774

45. Tang, C.J., Zheng, P., Wang, C.H., Mahmood, Q., 2010. Suppression of anaerobic 775 ammonium oxidizers under high organic content in high-rate Anammox UASB reactor. 776 Bioresour. Technol. 101(6), 1762–1768. https://doi.org/10.1016/j.biortech.2009.10.032 777

46. Tao, W., He, Y., Wang, Z., Smith, R., Shayya, W., Pei, Y., 2012. Effects of pH and 778 temperature on coupling nitritation and anammox in biofilters treating dairy wastewater. 779 Ecol. Eng. 47, 76–82. https://doi.org/10.1016/j.ecoleng.2012.06.035 780

47. Vázquez-Padín, J., Mosquera-Corral, A., Campos, J.L., Méndez, R., Revsbech, N.P., 781 2010. Microbial community distribution and activity dynamics of granular biomass in a 782 CANON reactor. Water Res. 44(15), 4359–4370. 783 https://doi.org/10.1016/j.watres.2010.05.041 784

48. Vlaeminck, S.E., De Clippeleir, H., Verstraete, W., 2012. Microbial resource management 785 of one-stage partial nitritation/anammox. Microb. Biotechnol. 786 https://doi.org/10.1111/j.1751-7915.2012.00341.x 787

49. Wu, S., Bhattacharjee, A.S., Weissbrodt, D.G., Morgenroth, E., Goel, R., 2016. Effect of 788 short external perturbations on bacterial ecology and activities in a partial nitritation and 789 anammox reactor. Bioresour. Technol. 219, 527–535. 790

50. Zhu, G., Peng, Y., Li, B., Guo, J., Yang, Q., Wang, S., 2008. Biological Removal of 791 Nitrogen from Wastewater. Rev. Environ. Contam. Toxicol. 192, 159–195. 792

793

794

Figure Captions 795

796

Fig. 1: Performance of the SBR during operation with real anaerobic effluent (supplemented 797

with nitrite) at decreasing temperatures: phase 1 (T = 35 °C), phase 2 (T = 25 °C), and phase 3 798

(T = 20 °C). Ammonium (a), nitrite (b) and COD removal efficiencies of the SBR and 799

influent and effluent concentrations. 800

Fig. 2: DGGE profile of the bacterial community in the SBR (a) and dendrogram based on the 801

DGGE profiles (b). The numbers on the gel picture correspond to band identification at the 802

similarity level. Samples (I) inoculum, (II) end of the enrichment period, (III) end of phase 1 803

(T = 35 °C), (IV) end of phase 2 (T = 25 °C), (V) after 30 days in phase 3 (T = 20 °C), (VI) 804

end of phase 3 (T = 20 °C). 805

Fig. 3 Taxonomic composition of the microbial communities at the phylum level. OTUs with 806

less than 1% abundance (in each sample) were included in the group ‘Others’ to improve data 807

visualization. Biomass was sampled from the reactor at the end of each operational phase (II - 808

enrichment period, III - phase 1- temperature of 35 °C, IV - phase 2 - temperature of 25 °C, 809

and V - phase 3 - temperature of 20 °C). The composition of the inoculum used was also 810

investigated and is shown in this figure (I). 811

Fig. 4 Microbial composition at the order level within the phylum Proteobacteria. Biomass 812

was sampled from the reactor at the end of each operational phase (II - enrichment period, III 813

- phase 1 - temperature of 35 °C, IV - phase 2 - temperature of 25 °C, and V - phase 3 - 814

temperature of 20 °C). The composition of the inoculum used was also investigated and is 815

shown in this figure (I). 816

Fig. 5 Abundance of anammox and denitrifying bacteria (inferred from nosZ gene) in the SBR 817

per qPCR of the samples: inoculum, enrichment period, end of phase 1(T= 35°C), phase 2 818

(T= 25°C) and phase 3 (T= 20° C). 819

820

821

822

Tables

Table 1

Mean values of physico-chemical parameters and experimental phases in the SBR

Experimental phase Duration (days)

Temperature (°C)

NO2 (mg·L-1)

NH4 (mg·L-1)

COD (mg·L-1)

Anammox enrichment perioda

0 to 160 35 47 32 0

Phase 1b 40 35 78 42 130

Phase 2 63 25 72 41 100

Phase 3 45 20 70 32 97

aDuring the enrichment period, the reactor was fed with autotrophic medium. bIn phases 1, 2, and 3, the reactor was fed with anaerobically pretreated municipal wastewater supplemented with nitrite.

Table 2

Summary of the ammonium, nitrite, and COD removal efficiencies obtained in the SBR and nitrogen mass balance calculated for each operational

phase.

Operational phase

NH4+ -Na NO2

--Na NO3--Na CODa Stoichiometry

Influent (mg·L-1)

AMX (mg·L-1)

Removal efficiency

%

Influent (mg·L-1)

AMX (mg·L-1)

DN (mg·L-1)

Removal efficiency

%

AMX production (mg·L-1)

DN (mg·L-1)

Effluent (mg·L-1)

Influent (mg·L-1)

DN (mg·L-1)

Removal efficiency

%

Anammox enrichment 32.4 31.3 96.5 46.7 46.0 - 98.2 11.0 - 11.0 - - - 1:1.47:0.35b

Phase 1 (35 °C)

42.3 41.2 97.7 78.2 60.6 12.5 93.9 14.4 -1.1c 15.5 130.2 86.4 61.8 1:1.77:0.37

Phase 2 (25 °C)

40.6 40.6 100.0 72.2 59.7 9.4 95.7 14.2 -3.7c 17.9 100.1 76.2 75.7 1:1.70:0.44

Phase 3 (20 °C)

32.3 31.8 98.5 70.0 46.7 15.9 89.3 11.1 -1.1c 12.2 96.7 62.8 65.0 1:1.96:0.38

aMean values bThe stoichiometry of removed NH4

+-N:NO2--N:produced NO3

--N obtained during the anammox enrichment phase was used to calculate the nitrite removal via the anammox and denitrification processes during Phases 1, 2, and 3. cNegative value indicates that nitrate mass balance did not close, as more nitrate was determined in the effluent.

AMX: anammox consumption

DN: denitrification consumption

823

Table 3

Diversity and richness indices, number of OTUs, and sequences from Ion Torrent data analysis

Sample Experimental

phase Shannon–Weaver

Simpson (1-D)

Dominance (D)

Number of OTUs

Number of sequences

I Inoculum 4.609 0.977 0.02263 233 33422

II Anammox enrichment

3.779 0.935 0.06499 151 50464

III 1 (T = 35 °C) 3.901 0.934 0.06546 180 43229

IV 2 (T = 25 °C) 4.300 0.972 0.02801 189 45016

V 3 (T = 20 °C) 4.321 0.971 0.02840 187 49973

824

825

826

827

Table 4

Distances between samples calculated based on Bray-Curtis ecological indexa.

Sample I II III IV V

I Inoculum 1 0.19088 0.17354 0.18119 0.19272

II Enrichment 0.19088 1 0.64534 0.51915 0.4948

III Phase I (T=35 °C) 0.17354 0.64534 1 0.66773 0.59842

IV Phase II (T=25 °C) 0.18119 0.51915 0.66773 1 0.81593

V Phase III (T=20 °C) 0.19272 0.4948 0.59842 0.81593 1

a The closer to 1 the more similar the samples

Table 5

Identification of DGGE band sequences

Band RDP Classifier BLASTn Identity (%) a Acc. No 1 Burkholderiales Acidovorax sp. 89 NR_116740.1 2 Saprospirales Haliscomenobacter sp. 91 NR_074420.1 3 Anaerolineaceae Thermomarinilinea sp. 89 NR_132293.1 4 Ca. Brocadiaceae Uncultured bacterium clone Dok57 98 FJ710776.1 5 Ca. Brocadia sp. Uncultured bacterium clone Dok57 91 FJ710776.1 6 Anaerolineaceae Anaerolinea sp. 84 NR_074383.1 7 Rhodocyclales uncultured bacterium 93 KT835629.1 8 Rhodocyclaceae Denitratisoma oestradiolicum 97 KF810114.1 9 Myxococcales Myxococcales 92 FJ551068.1 10 Anaerolineaceae Bellilinea sp. 95 JQ183076.1 11 Rhodocyclaceae Denitratisoma oestradiolicum 97 KF810114.1 12 Anaerolineaceae Anaerolineaceae 94 HE648186.1 13 Anaerolineaceae Bellilinea sp. 93 NR_041354.1 14 Ca. Brocadia sp. Ca. Brocadia caroliniensis 98 KF810110.1 15 Ca. Brocadia sp. Ca. Brocadia sp. 97 AM285341.1 16 Rhodocyclaceae Denitratisoma sp. 97 HM769664.1 17 Anaerolineaceae Bellilinea sp. 91 NR_041354.1 18 Ca. Brocadia sp. Uncultured bacterium clone Dok57 98 FJ710776.1 19 Ca. Brocadia sp. Uncultured bacterium clone Dok57 96 FJ710776.1 20 Rhodocyclaceae Rhodocyclaceae 89 HQ033257.1 21 Rhodocyclaceae Denitratisoma oestradiolicum 98 KF810114.1 22 Burkholderiales Burkholderia sp. 90 NR_118986.1 23 Anaerolineaceae Bellilinea sp. 91 NR_041354.1 24 Ca. Brocadia sp. Ca. Brocadia caroliniensis 98 KF810110.1 25 Rhodocyclaceae Denitratisoma oestradiolicum 98 KF810114.1 26 Burkholderiales Burkholderiales 91 KM083133.1 27 Anaerolineaceae Bellilinea 91 NR_041354.1

aPercentages indicate the identity between the DGGE band sequences and the closest matched sequences in GenBank. Words in bold indicate DNA bands related to anammox bacteria (Ca. Brocadia). The other bands most affiliated to Rhodocyclaceae and Chloroflexi (Anaerolineaceae) are heterotrophic bacteria likely involved in COD removal.

Table 6 Major identified orders or genera (>0.5% relative sequence contribution) of bacteria related to autotrophic nitrogen removal (in bold) and heterotrophic bacteria potentially involved in organic matter removal in the SBR based on Ion Torrent data analysis.

Classification Relative frequency on SBR samples (%)

II a

III (T = 35 °C)

IV (T = 25 °C)

V (T = 20 °C)

Planctomycetes; Ca. Brocadia 0.2 0.2 0.7 0.5

Planctomycetes; Ca. Anammoximicrobium 0 0.7 2.3 0.4 Proteobacteria; Nitrosomonas 0.4 0.3 0.2 0.3

Proteobacteria; Nitrosomonadaceae 2.1 1.6 1.9 2.0

Nitrospirae; Nitrospira 1.3 1.3 1.2 1.5 Proteobacteria; Rhodospirillalesb 7.7 4.6 4.6 4.3 Proteobacteria; Burkholderialesb 0.6 2.3 4.0 4.1 Proteobacteria; Rhodocyclalesb;Denitratisoma 0.7 1.3 4.7 5.0 Proteobacteria; Rhodocyclalesb;Sulfuritalea 1.7 4.7 8.9 9.8 Proteobacteria; Rhodocyclalesb;other 0.1 0.3 0.6 1.7 Proteobacteria; Myxococcales 6.8 2.2 3.7 4.4 Proteobacteria; Pseudomonadalesb 0.4 0.3 0.6 0.3 Proteobacteria; Xanthomonadalesb 3.6 2.9 4.6 4.8 Proteobacteria; Rhizobiales 1.0 1.1 1.0 1.0 Proteobacteria; Hydrogenophilales 0.4 3.1 2.4 1.2 Proteobacteria; Sh765B-TzT-29 2.1 2.9 3.7 2.9 Chloroflexic; Anaerolineales 25.4 24.9 10.9 10.6 Chloroflexi; Caldilineales 1.2 0.8 1.0 1.0

Chloroflexi; TK10; uncultured 0.9 7.7 3.1 2.3

Chloroflexi; Other 3.1 1.8 1.5 0.9

Chlorobi; Chlorobiales 1.2 0.4 1.3 1.4 Chlorobi; Ignavibacteriales 13.4 6.6 7.1 7.9 Bacteroidetesd; Sphingobacteriales 0.7 1.4 1.0 0.9

Firmicutesd; Clostridiales 0.5 1.9 0.9 0.9

Planctomycetes; OM190; uncultured 1.3 0.7 1.2 1.7

Planctomycetes; Physisphaerae; mle1-8 0.1 0.1 0.9 0.5

Planctomycetes; Physisphaerae; Pla1 0.4 0.5 0.7 0.6 Planctomycetes; Planctomycetales 1.0 1.6 3.3 1.3 Actinobacteria; Acidimicrobiales 2.0 1.6 1.3 1.0 Acidobacteria; Subgroup3 0.9 0.9 1.0 1.1

Acidobacteria; Subgroup4 0.7 0.6 0.7 0.9

Acidobacteria; Subgroup6 0.1 0.3 0.5 0.7 Acidobacteria; Subgroup10 0.5 0.3 0.4 0.7 Candidate division WS3; uncultured bacterium mle1-16 1.6 3.4 5.3 5.7

aBiomass sample taken from the reactor operated at 35 °C and fed with autotrophic medium. The other samples were taken 828 when the reactor was fed with real anaerobic effluent amended with nitrite.b Members of Burkholderiales, Rhodocyclales, 829 Xanthomonadales, Rhodospirillales, and Pseudomonadales are bacteria capable of performing heterotrophic denitrification 830 (Heylen et al., 2008).cChloroflexi are anaerobic and heterotrophic bacteria reported in anammox reactors (Pereira et al., 831 2017).dMembers of Bacteroidetes and Firmicutes are capable of fermenting various organic substrates. 832

Fig.1 (a) 833

834 835 Fig 1b. 836

837 838 839 840 841 842 843 844 845

846 Fig 1c. 847

848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876

877 Fig2A 878

879 880 881 882 883 Fig2B 884

885 886 887 888 889 890 891 892 893 894 895 896

Fig 3. 897

898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931

Fig. 4 932

933 934 935 Fig. 5 936

937

1,00E+00

1,00E+01

1,00E+02

1,00E+03

1,00E+04

1,00E+05

1,00E+06

1,00E+07

1,00E+08

1,00E+09

1,00E+10

Inoculum

Nu

mb

er

of

ge

ne

co

pie

s/ g

of

slu

dg

e

16S rRNA gene of Anammox

Enrichment Phase 1 Phase 2

16S rRNA gene of Anammox NosZ gene of Denitrifiers

Phase 3

NosZ gene of Denitrifiers