Bioresource Technology 243 (2017) 922–931

Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Effect of reflux ratio on nitrogen removal in a novel upflow microaerobicsludge reactor treating piggery wastewater with high ammonium andlow COD/TN ratio: Efficiency and quantitative molecular mechanism

http://dx.doi.org/10.1016/j.biortech.2017.07.0520960-8524/� 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: ljz6677@163.com (J. Li).

Jia Meng a, Jiuling Li b, Jianzheng Li a,⇑, Kaiwen Deng a, Jun Nan a, Pianpian Xu a

a School of Municipal and Environmental Engineering, State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology, 73 Huanghe Road,Harbin 150090, PR ChinabAdvanced Water Management Centre, The University of Queensland, St Lucia, Brisbane, QLD 4072, Australia

h i g h l i g h t s

� To treat piggery wastewater with high NH4+ density and low C/N ratio.

� An innovative upflow microaerobic sludge reactor was constructed.� Effect of reflux ratio (RR) and mechanism of nitrogen removal were investigated.� RR 34 was favorable for NH4

+ and TN removal with anammox as the dominant mechanism.� Decrease in RR stimulated ammonia-oxidization but inhibited anammox.

a r t i c l e i n f o

Article history:Received 25 May 2017Received in revised form 6 July 2017Accepted 9 July 2017Available online 11 July 2017

Keywords:Piggery wastewaterMicroaerobic processNitrogen removalReflux ratioMolecular mechanism

a b s t r a c t

A novel upflow microaerobic sludge reactor (UMSR) was constructed to treat manure-free piggerywastewater with high NH4

+-N and low COD/TN ratio. In the light of the potential effect of effluent refluxratio (RR) on nitrogen removal, performance of the UMSR was evaluated at 35 �C and hydraulic retentiontime 8 h with RR decreased from 45 to 25 by stages. A COD, NH4

+-N and TN removal of above 77.1%, 80.0%and 86.6%, respectively, was kept with a RR over 35. To get an effluent of TN not more than 80 mg/L with aTN load removal above 0.88 kg/(m3�d), the RR should be at least 34. Real-time quantitative polymerasechain reaction of functional bacteria revealed that the RR of less than 34 stimulated ammonium oxidationbut badly inhibited anammox, the dominant nitrogen removal pathway, resulting in the remarkabledecrease of nitrogen removal in the reactor.

� 2017 Elsevier Ltd. All rights reserved.

1. Introduction

The increasing demand for pork has stimulated a continualexpansion of pig-breeding, resulting in increasing discharge of pig-gery wastewater from large-scale pig farms, especially in develop-ing countries such as China. If not disposed properly, it wouldresult in significant health and environmental consequences(Zhao et al., 2014). The strength of piggery wastewater variesgreatly due to the manure collection methods (Wang and Guo,2009). As a traditional method, urine-free manure collection iswidely used and the flushed wastewater of piggery is defined asmanure-free piggery wastewater (MFPW) (Zhao et al., 2014). ThisMFPW is characterized by high ammonium (NH4

+-N) and low ratio

of chemical oxygen demand (COD) to total nitrogen (TN), posing achallenge to traditional nitrification-denitrification processes fornitrogen removal (Zhang et al., 2016). Anaerobic-aerobic combinedprocess has been developed for nitrogen removal from sewage andproved to be robust (Bernet et al., 2000; Canals et al., 2013). Thecombined process could also be used for piggery wastewater treat-ment, despite of the costly requisite for physical chemistry pre-treatment to get a feasible COD/TN ratio for biological nitrogenremoval (Bernet et al., 2000). It is essential to develop a novel treat-ment processes that is more efficient and economical for treatingMFPW.

Microaerobic condition is regarded as a transitional statebetween aerobic and anaerobic with a dissolved oxygen (DO) rang-ing from 0.3 to 1.0 mg/L (Zheng and Cui, 2012; Zitomer, 1998). Pre-viously, a novel upflow microaerobic sludge reactor (UMSR) wasdeveloped to treat MFPW (Meng et al., 2015). Ammonia-

J. Meng et al. / Bioresource Technology 243 (2017) 922–931 923

oxidizing bacteria (AOB), nitrite oxidizing bacteria (NOB), hetero-trophic denitrifiers, autotrophic denitrifiers and phosphate accu-mulating organisms were found to coexist in the UMSR, and theirsynergistic action improved microaerobic process in the syn-chronous removal of COD, NH4

+-N, TN and total phosphorus (TP)(Meng et al., 2016). The microaerobic condition in the UMSR wasachieved by recirculating part of the effluent that was aerated toa DO of about 3 mg/L (Meng et al., 2015). Obviously, reflux ratio(RR) plays a crucial role in controlling the microaerobic conditionin the reactor. In fact, the microbial community structure can beremarkably affected by the change of RR, which would result in avariation in pollutant removal efficiency (Jin et al., 2012). An exces-sive RR would not only restrict anaerobic denitrifiers due to theincreased DO, but also increase the treatment cost (Yuan et al.,2002). On the other hand, a RR too low would lead to an insuffi-cient DO for NH4

+-N oxidation, also resulting in a bad TN removalin biological nitrogen removal processes (He et al., 2006). There-fore, a feasible RR is essential for the microaerobic treatment sys-tem to enhance TN removal and cut down the treatment cost.However, few researches can be found to investigate the effect ofRR on microaerobic processes.

In the present research, the UMSR was continually performedfor another 176 days, by stages, to evaluate the effect of RR onthe synchronous removal of COD, NH4

+-N, TN and TP. To understandthe mechanism for biological nitrogen removal in the reactor, func-tional microbial populations under different RR were investigatedusing Illumina Miseq platform, while the nitrogen removal path-ways were also analyzed based on the absolute abundances offunctional bacteria identified by real-time quantitative polymerasechain reaction (qPCR).

2. Materials and methods

2.1. Microaerobic treatment system

The lab-scale UMSR process, a microaerobic treatment system,was illustrated in Fig. 1. The UMSR was a 0.5-meter-high plexiglasscolumn with a working volume of 4.9 L, while with a 0.5 L circularcone attached to the bottom. A 3 L solid-liquid-gas separator wasdesigned on the top of the column with the off-gas discharge aftera water lock. Four sampling ports at 10 cm interval from each otherwere allocated along the vertical height of the column. The temper-

Fig. 1. Schematic representation of the lab-scale UMSR process.

ature in the reactor was kept at 35 ± 1 �C by a temperature con-troller (Jetten et al., 2001). The effluent was collected by a 10 Ltank, in which part of the effluent was aerated to a DO of about3 mg/L and then recirculated into the reactor from the bottom.

2.2. Wastewater and inoculum

The raw MFPW was collected from a local pig breeding farm inHarbin, China. The wastewater quality fluctuated following thechanges in breeding seasonality with a COD, NH4

+-N, nitrite (NO2�-

N), nitrate (NO3�-N), TN, TP and pH ranging from 241 to 459,

237.4 to 343.3, 0.0 to 2.2, 0.0 to 5.8, 285.4 to 458.2, 6.4 to 21.4mg/L and 7.0 to 8.4, respectively. Obviously, the high concentrationof NH4

+-N contributed mainly to the TN in the wastewater.The seed sludge inoculated in the UMSR was anaerobic sludge

collected from an upflow anaerobic sludge blanket (UASB) treatingswine wastewater in the State Key Laboratory of Urban WaterResource and Environment, Harbin Institute of Technology, China.The initial biomass in the UMSR was 5.68 g/L in terms of mixedliquor suspended solids (MLSS), i.e. 1.95 g/L in terms of mixedliquor volatile suspended solids (MLVSS). The UMSR previouslystarted up and performed for 200 days (Meng et al., 2015). At thestart of the present research, MLVSS in the UMSR was 3.5 g/L witha MLVSS/MLSS ratio of 0.66. The sludge was dominated by flocs,among which some small-sized granular sludge (<2 mm) couldalso be observed.

2.3. Operation of the UMSR

In the present research, the UMSR was continued to be operatedfor another 176 days in a 4-stage procedure with a consistenthydraulic retention time (HRT) of 8 h. Quality of the raw MFPWand the other controlling parameters in the 4 stages are illustratedin Table 1. The influent and effluent were sampled daily for analy-sis. Activated sludge in the reactor was sampled at the end of eachstage for measurement of biomass (MLVSS and MLSS) and analysisof high-throughput pyrosequencing and qPCR.

2.4. Analytical methods

COD, NH4+-N, NO2

�-N, NO3�-N, TP and biomass (MLSS and MLVSS)

were detected by closed reflux titrimetric method, Nessler’sreagent colorimetric method, colorimetric method, ultravioletspectrophotometric screening method, vanadomolybdophosphoricacid colorimetric method and gravimetric method, respectively,according to the Standard Methods (APHA, 2005). DO and pH weremeasured with a dissolved oxygen meter (Taiwan Hengxin, AZ8403) and a pH meter (Switzerland Mettler Toledo, DELTA320),respectively. TN was measured on a total nitrogen analyzer (Ger-many Analytikjena Multi, N/C 2100S).

2.5. High-throughput pyrosequencing

Sludge sampled from the UMSR at each end of the 4 stages(Table 1) was named S1, S2, S3, and S4 in sequence, and the totalDNA of the samples were extracted, respectively, using the BacteriaDNA Isolation Kit Components (MOBIO) according to the manufac-turer’s instruction. Bacterial 16S rRNA genes for the V3-V4 regionswere amplified with primer pairs 341f (50-CCTACGGGAGGCAGCAG-30) and 805r (50-GACTACHVGGGTATC TAATCC-30)(Herlemann et al., 2011; Hugerth et al., 2014). Composition ofthe PCR products of V4 region were determined by pyrosequencingon the Illumina Miseq sequencer platform (Shanghai Sangon,China).

Table 1Stages and operating parameters of the UMSR.

Stage 1 Stage 2 Stage 3 Stage 4

Reflux ratio 45 35 30 25Days 1–67 68–109 110–141 142–176NH4

+-N (mg/L) 299.7 ± 16.4 304.1 ± 19.9 298.7 ± 20.4 299.2 ± 17.5NO2

�-N (mg/L) 0.2 ± 0.3 0.2 ± 0.2 0.2 ± 0.1 0.0 ± 0.0NO3

�-N (mg/L) 0.9 ± 0.4 2.6 ± 1.3 0.5 ± 0.3 1.5 ± 0.9TN (mg/L) 366.9 ± 19.9 370.8 ± 33.6 352.2 ± 29.8 365.0 ± 23.4COD (mg/L) 307 ± 35 350 ± 46 285 ± 29 342 ± 40COD/TN 0.84 ± 0.10 0.95 ± 0.14 0.82 ± 0.10 0.94 ± 0.11NLRa (kg/(m3�d)) 1.10 ± 0.06 1.11 ± 0.10 1.06 ± 0.09 1.09 ± 0.07OLRb (kg/(m3�d)) 0.92 ± 0.11 1.05 ± 0.14 0.86 ± 0.09 1.02 ± 0.12pH 8.0 ± 0.1 7.7 ± 0.2 8.0 ± 0.2 7.8 ± 0.2

a Nitrogen loading rate in terms of TN.b Organic loading rate in terms of COD.

924 J. Meng et al. / Bioresource Technology 243 (2017) 922–931

2.6. qPCR analysis

To determine the amount of nitrifiers and denitrifiers in thesludge samples, seven independent qPCR assays were conductedin a real-time PCR system (USA Foster, ABI 7500) by quantifyingthe 16S rRNA of total bacteria, anammox bacteria, Nitrobacterspp., Nitrospira spp., and gene of amoA, nirS and narG, respectively.The primers used are illustrated in Table 2. Each of the PCR reac-tions was performed in a 20 lL system loaded with 2 lL templateDNA (at 22–88 ng/lL), 0.4 lL forward and 0.4 lL reverse primers(with the same 10 mM), 10 lL SYBR Premix (Applied Biosystems),and 7.2 lL PCR-grade sterile water. The PCR program was as fol-lows: 3 min at 95 �C followed by 40 cycles each including 15 s at95 �C, 20 s at 55 �C and 30 s at 72 �C in sequence. The PCR was con-trolled by a reaction without template and each of the qPCR wasperformed in triplicate.

All of the plasmid standards of the 16S rRNAs and the functionalgenes were constructed by Sangon Biotechnology IncorporatedCompany (Shanghai, China). The standard curves were preparedby ten-fold serial dilutions of the plasmid DNA with known copynumbers, and the correlation coefficients were all above 0.997.The PCR amplification efficiency for the genes was between 90%and 110%.

2.7. Data analysis

The absolute abundance of the seven functional genes (i.e. totalbacterial, anammox, Nitrobacter spp., Nitrospira spp., amoA, nirSand narG) obtained from each of the triplicate tests were averaged,respectively, and then used as the basic candidate variables in non-linear regression analyses (SPSS 22, USA) for the correlations withnitrogen transformation rates. Standard deviations were calculated

Table 2Primers in the qPCR.

Target Primer Seq

Bacterial 16S rRNA 338F ACT518R ATT

Anammox 16S rRNA AMX809F GCAMX1066R AA

amoA gene amoA-1F GGamoA-2R CCC

Nitrobacter spp. 16S rRNA Nitro-1198F ACCNitro-1423R CTT

Nitrospira spp. 16S rRNA NSR 1113F CCTNSR 1264R GTT

nirS gene nirS-1F TACnirS-3R GC

narG gene narG 1960m2F TAYnarG 2050m2R CG

using the three replicate measurements and plotted as error barsfor analyzing data variations and measurement errors.

3. Results and discussion

3.1. Pollutant removal in the UMSR

After the previous 200-day performance (Meng et al., 2015), theUMSR was continued to be operated for another 176 days follow-ing a progressive decrease in RR (Table 1) with a consistent HRTof 8 h at 35 ± 1 �C. The performance in pollutant removal withinthe 176 days was illustrated in Fig. 2. The results showed that, instage 1 with the RR 45, NH4

+-N removal in the reactor exhibited arapid decrease phase within the first 6 days, and then increasedgradually in fluctuation thereafter and reached a steady phasewithin the last 13 days (55th–67th day) (Fig. 2A). The averageNH4

+-N removal in the steady phase reached 86.2% with a residualof about 40.2 mg/L in effluent. The NH4

+-N removal was remarkablyaffected and decreased sharply to 37.7% when the RR decreased to35 in the first day of stage 2, but rebounded and reached a newsteady state since the 96th day with the removal and effluent con-centration averaging 77.1% and 67.4 mg/L, respectively. When RRwas further decreased to 30 in stage 3 and 25 in stage 4, the samevariation pattern of NH4

+-N removal with time were also observed.The NH4

+-N removal in the steady phase of stage 3 (130th–141thdays) and stage 4 (164th–176th days) was about 73.1% and53.3%, with a NH4

+-N of about 81.3 and 137.8 mg/L in effluent,respectively. Obviously, the decrease of RR had badly affected theammonium oxidation in the reactor.

Variation of TN removal (Fig. 2B) in the UMSR found the sametrend as that of NH4

+-N removal, with the steady states obtainedsynchronously. In the steady phase of stage 1, 2, 3 and 4, the TNremoval averaged 87.2%, 80.0%, 75.4% and 60.9%, with an effluent

uence (50-30) References

CCTACGGGAGGCAGCAG Muyzer et al. (1993)ACCGCGGCTGCTGGCGTAAACGATGGGCACT Tsushima et al. (2007)CGTCTCACGACACGAGCTGGGTTTCTACTGGTGGT Rotthauwe et al. (1997)CTCKGSAAAGCCTTCTTCCCTAGCAAATCTCAAAAAACCG Graham et al. (2007)CACCCCAGTCGCTGACCGCTTTCAGTTGCTACCG Dionisi et al. (2002)TGCAGCGCTTTGTACCGCACCCSGARCCGCGCGT Braker et al. (1998)

CGCCGTCRTGVAGGAAGTSGGGCAGGARAAACTG López-Gutiérrez et al. (2004)

TAGAAGAAGCTGGTGCTGTT

0

10

20

30

40

50

60

70

80

90

100

0

5

10

15

20

25

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180

Rem

oval

(%)

TP (m

g/L)

Time (d)

InfluentEffluentRemoval

stage 1 stage 2 stage 3Reflux ratio 45 Reflux ratio 35 Reflux ratio 30

stage 4Reflux ratio 25

B

C

A

D

0

10

20

30

40

50

60

70

80

90

100

0

50

100

150

200

250

300

350

400

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180

Rem

oval

(%)

NH

4+ -N

(mg/

L)

Time (d)

Influent Effluent Removal

0

10

20

30

40

50

60

70

80

90

100

0

50

100

150

200

250

300

350

400

450

500

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180

Rem

oval

(%)

CO

D (m

g/L)

Time (d)

Influent

Effluent

Removal

0

10

20

30

40

50

60

70

80

90

100

0

50

100

150

200

250

300

350

400

450

500

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180

TN re

mov

al (%

)N

Ox-

N(m

g/L)

TN (m

g/L)

Time (d)

Inf. TN Eff. TN TN removal Eff. nitrite Eff. nitrate

Fig. 2. Removal of NH4+-N (A), TN (B), COD (C) and TP (D) in the UMSR with the variation of reflux ratio.

J. Meng et al. / Bioresource Technology 243 (2017) 922–931 925

concentration of about 45.8, 73.2, 89.2 and 140.6 mg/L, respec-tively. The average TN load removal (TNLR) in the four steadyphases was calculated to 0.94, 0.88, 0.82 and 0.66 kg/(m3�d),

respectively. The inhibited NH4+-N oxidization by the decreased

RR (Fig. 2A) was considered as the main reason for the decreaseof TN removal because no obvious NOx

�-N (including NO2�-N and

926 J. Meng et al. / Bioresource Technology 243 (2017) 922–931

NO3�-N) acclimation was found in the four steady phases. However,

NO2�-N accumulation could be found in the first days all of the

stages. In stage 1, for example, NO2�-N accumulated to 50.4 mg/L

on the 31th day, but decreased rapidly with time and no observ-able accumulation was detected in the steady phase. The samephenomenon was also observed in the next 3 stages, though theaccumulation of NO2

�-N decreased slightly with the stages. Thisperformance accorded with the cultural characteristics of anam-mox bacteria (van Dongen et al., 2001), indicating anammox wasan important approach to nitrogen removal in the UMSR.

All through the 176-day operation, COD removal in the UMSRwas above 65% with an effluent COD less than 125 mg/L (Fig. 2C).It was found that COD removal in the steady phases of stage 1and stage 3 were almost the same (77.9% and 74.5%, respectively),though the RR (45 and 30, respectively) was the remarkable differ-ence. The COD removal in the steady phases of stage 2 (RR 35) andstage 4 (RR 25) were also similar (86.6% and 83.8%, respectively). Itwas noticed that COD/TN ratio in stage 1 (0.84) and 3 (0.82) werealmost the same, as well as those in stage 2 (0.95) and 4 (0.94). Theresults suggested that COD/TN ratio had a stronger effect than RRon the COD removal in the UMSR. Within the 176-day operationof the UMSR, no excess sludge was discharged and the biomassin terms of MLVSS at each end of the four stages was 5.20, 7.15,4.11, and 6.26 g/L, respectively. Evidently, the higher COD/TN ratioin stage 2 and stage 4 had stimulated the growth of chemo-heterotrophic bacteria, resulting in the better COD removal in thereactor (Li et al., 2016).

The stimulated growth of biomass was favorable for TP removalin the UMSR. As shown in Fig. 2D, TP removal in the second steadyphase reached 83.5% with the highest biomass (7.15 g/L). With alower biomass in stage 1 (5.20 g/L), stage 3 (4.11 g/L) and stage 4(6.26 g/L), the TP removal decreased to 46.8%, 52.5% and 66.6% intheir steady phases, respectively. The effluent TP in the four steadyphases averaged 5.1, 2.1, 4.4 and 5.0 mg/L, respectively. Observ-ably, TP removal in the UMSR was positively correlated with thebiomass, indicating sludge growth was the most importantapproach for TP removal in the microaerobic treatment system.

Discharge standards for various wastewater have been promul-gated in many countries. According to the Discharge Standard ofPollutants for Livestock and Poultry Breeding issued by the Min-istry Environmental Protection of China (MEPPRC, 2001), the dis-charge standard for NH4

+-N should not be more than 80 mg/L. Allthrough the 176-day operation of the UMSR, the effluent CODand TP could meet the standard required very well, while the efflu-ent NH4

+-N and TN could meet the standard only with a RR not less

y = -0.0724x2 + 6.334x - 51R² = 0.981

y

0

10

20

30

40

50

60

70

80

90

100

20 25 30 35

Nitr

ogen

rem

oval

(%)

reflux ra

Ammonium removal

TN removal

Total nitrogen load removal

Fig. 3. Effect of reflux ratio on nit

than 35 (Fig. 2). To get an economical RR for treating manure-freepiggery wastewater, the influence of RR on nitrogen removal (NH4

+-N and TN) and TNLR in the four steady phases were analyzed asillustrated in Fig. 3, in which an inhibitory threshold for nitrogenremoval was defined when TN removal dropped to around 78.0%,with an effluent TN more than 80 mg/L. Based on the quadraticpolynomial regression equations (Fig. 3), the RR threshold shouldbe 34, at which a TNLR of 0.88 kg/(m3�d) could be obtained in theUMSR.

3.2. Diversity of microbial community and synchronous pollutantremoval

To understand the biological mechanism of synchronousremoval of COD, NH4

+-N, TN and TP in the UMSR, DNA extractionand PCR amplification of the sludge sampled at the end of eachstage (Table 1) were carried out. As shown in Table 3, 41,342,36,383, 37,906 and 19,515 reads were obtained from the sludgesampled from stage 1 (S1), stage 2 (S2), stage 3 (S3) and stage 4(S4), respectively. The identified operational taxonomic units(OTUs) in the four sludge samples reached 5632, 5211, 6542 and1459, respectively. The results indicated that better species diver-sity and richness had been kept in the UMSR, though the Chao1and ACE index, as well as the Shannon index, varied with thedecrease of RR.

The microbial community structure and composition under dif-ferent RR were characterized by phylogenetic classification. Asshown in Fig. 4A, Proteobacteria was the most predominant phy-lum, followed by Firmicutes and Bacteroidetes, all the time in theUMSR. Fig. 4B indicated that the main subgroups of phyla Pro-teobacteria in S1, S2 and S4 were the same b-Proteobacteria, whilec-Proteobacteria dominated the S3. It is reported that b- and c-Proteobacteria play a critical part in the biological removal of nitro-gen and phosphorus (Yang et al., 2014).

Phylogenetic classification of the four sludge samples at genuslevel (Fig. 5) showed that the microbial community structure inthe UMSR was distinguished by the difference of RR. Even so, het-erotrophic bacteria (including aerobic and anaerobic heterotrophicbacteria), nitrifiers (including AOB and NOB), and denitrifiers(including heterotrophic and autotrophic denitrifiers) were foundto coexist and flourish in the microaerobic system, resulting inthe synchronous removal of COD, NH4

+-N and TN. However, no gen-era related to anammox had been identified from the activatedsludge samples by high-throughput pyrosequencing (Fig. 5). Thus,the nitrogen removal mechanism had to be further investigated.

y = -0.096x2 + 8.282x - 92.449R² = 0.9628

.429

= -0.0009x2 + 0.0772x - 0.6953R² = 0.988

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

40 45 50

Tota

l nitr

ogen

load

rem

oval

(kg/

(m3 ·

d))

tio

rogen removal in the UMSR.

Table 3Diversity of microbial community in the activated sludge sampled from the UMSR.

Stage (reflux ratio) Steady phase (days) Sludge sample* Reads 3% distance

Raw reads Clean reads OTUs Chao1 ACE Shannon Good’s coverage (%)

1 (45) 55th–67th S1 41342 41328 5632 12481 18929.92 6.83 90.922 (35) 96th–109th S2 36383 36373 5211 12728 19749.52 6.63 90.623 (30) 130th–141th S3 37906 37882 6542 15204 23992.14 6.92 88.484 (25) 164th–176th S4 19515 19512 1459 2619 3112.58 5.68 96.43

* S1, S2, S3 and S4 was the sludge sample taken from the UMSR at the end of stage 1, stage 2, stage 3 and stage 4, respectively.

Fig. 4. Bacterial community structure of the activated sludge at phylum (A) and the classes in Proteobacteria (B). S1, S2, S3 and S4 was the sludge sample taken from the UMSRat the end of stage 1, stage 2, stage 3 and stage 4, respectively.

J. Meng et al. / Bioresource Technology 243 (2017) 922–931 927

3.3. Nitrogen removal pathways

Nitrification-denitrification, shortcut nitrification-denitrificationand anammox are well known as the approaches to the TN removal

in biological denitrification processes (Xie et al., 2016). The reduc-tion of nitrate to nitrogen gas (N2 or N2O) needs more carbonsource to supply electron than that of nitrite reduction, and nitrousoxide (N2O) production would need less carbon source than the

Fig. 5. Bacterial community structure at genus level of the activated sludge sampled from the UMSR at the end of the first stage (A), the second stage (B), the third stage (C)and the fourth stage (D).

928 J. Meng et al. / Bioresource Technology 243 (2017) 922–931

1.0E+00

1.0E+01

1.0E+02

1.0E+03

1.0E+04

1.0E+05

1.0E+06

1.0E+07

45 35 30 25

Cop

ies/

ngD

NA

Reflux ratio

bacteria anammox amoA NitrobacterNitrospira nirS narG

Fig. 6. Copies per ngDNA of nitrogen functional genes with the reflux ratios in theUMSR.

J. Meng et al. / Bioresource Technology 243 (2017) 922–931 929

production of N2. Theoretically, without assimilation, 2.86 g or1.71 g COD is needed to reduce 1 g NO3

�-N or 1 g NO2�-N to N2,

respectively (Li et al., 2012; Saggar et al., 2013). For the reductionof 1 g NO3

�-N or 1 g NO2�-N to N2O, 2.29 g or 1.14 g COD would be

consumed respectively. Based on the cooperation of the theoreticalCOD/TN ratio with the measured CODremoved/TNremoved ratio in theUMSR, the theoretical TN removal efficiencies via various path-ways in the microaerobic system were calculated and illustratedin Table 4. The results showed that the TN removal in the UMSRwas remarkably lower than that of the actual efficiency even allof the removed COD were used for the denitrification via nitritereduction or nitrate reduction to N2. Even for nitrite reduction toN2O, the least carbon-source-consumption denitrification, the the-oretical TN removal was also observably lower (stage 1, stage 2 andstage 3) or similar (stage 4) to that of the actual efficiency in theUMSR. In the steady phase of stage 1, for example, the theoreticalTN removal efficiency via N2O production by nitrite reduction, N2

production by nitrite and nitrate reduction were 59.0%, 39.1%and 23.5%, respectively, considerably lower than the actual 87.2%.Hence, the autotrophic anammox must be one of the mostimportant pathways to nitrogen removal in the UMSR.

To confirm the existence of anammox and understand the nitro-gen removal mechanism in depth, the absolute abundances ofamoA, nirS, narG and Nitrobacter 16S rRNA, Nitrospira 16S rRNA,anammox 16S rRNA and total bacteria 16S rRNA in the sampledsludge (S1, S2, S3 and S4) were quantified using qPCR assays andthe results were illustrated in Fig. 6. Among the seven functionalgenes, amoA was responsible for NH4

+-N conversion and NO2�-N

production, while nirS for NO2�-N reduction, narG for NO3

�-N reduc-tion, both Nitrobacter 16S rRNA and Nitrospira 16S rRNA for NO2

�-Noxidation, and anammox 16S rRNA for NH4

+-N and NO2�-N con-

sumption. Absolute abundances of the seven functional genes wereused as basic candidate variables for nonlinear regression analyses(SPSS 22, USA) to get the correlation with nitrogen conversionrates. Then the following equations are obtained.

NHþ4 � N ¼ �4:706ðamoA=ðanammoxþ Nitrobacter

þ NitrospiraÞÞ þ 0:780 ð1Þ

TN ¼ �2:881ðamoA=ðanammoxþ Nitrobacter þ NitrospiraÞÞþ 23:625ðnirSþ narGÞ=Bacterialþ 0:831 ð2Þ

The R2 in Eqs. (1) and (2) was 0.977 and 0.989, respectively. Itwas found that the NH4

+-N removal had a negative correlation withamoA/(anammox + Nitrobacter + Nitrospira). The NH4

+-N regressionmodel suggested that NH4

+-N removal in the UMSR was not onlyinfluenced by amoA gene but also by anammox, Nitrobacter andNitrospira 16S rRNA, and the increased amoA would reduce theNH4

+-N removal in the UMSR. TN removal in the UMSR was relatedto denitrification, too (Li et al., 2016). Both nirS/(total bacteria) andnarG/(total bacteria), involved in NO2

�-N and NO3�-N reduction

respectively, were positively correlated with TN removal in the

Table 4Actual TN removal efficiency in the UMSR comparing with the theoretical values.

COD removala

(kg/(m3�d))TN removeda

(kg/(m3�d))CODrem./TNrem.ratioa Actual TN

removala (%)

Stage 1 0.72 0.94 0.76 87.2Stage 2 0.89 0.88 1.01 80.0Stage 3 0.63 0.82 0.77 75.4Stage 4 0.86 0.66 1.30 60.9

a The average during steady phase of the stage.b Denitrification via nitrous oxide according to the theoretical 1.14 g COD/g N withouc Denitrification via nitrite according to the theoretical 1.72 g COD/g N without assimd Denitrification via nitrate according to the theoretical 2.86 g COD/g N without assim

UMSR (Eq. (2)). Above all, the qPCR assays indicated that shortcutnitrification, anammox, and denitrification coexisted in the UMSR,and their combined action resulted in the better TN removal with-out supply of extra carbon source.

3.4. Quantitative molecular mechanism of pollutant removal

As illustrated in Fig. 5, though the RR decreased from 45 to 25by stages, the heterotrophic bacteria were the foremost popula-tions all along the 176-day performance of the UMSR. The flourish-ing heterotrophic population resulted in the better COD removal inthe UMSR all through the four operation stages (Fig. 2C). Amongthe heterotrophic bacteria, Variovorax (Prasad and Suresh, 2012),Novosphingobium (Kertesz and Kawasaki, 2010), Stenotrophomonas(Binks et al., 1995) and Altererythrobacter (Park et al., 2011) arestrict aerobic bacteria, and the total proportion decreased from4.66% in S1 to 2.27% in S4 with the decreased RR from 45 to 25 stageby stage. Obviously, it was the decreased RR that had reduced theDO concentration and resulted in the decline of aerobic hetero-trophic bacteria in the UMSR.

It was interesting to find that the decrease of RR enriched AOBand NOB in the reactor. With the RR 45, genera related to AOB in S1were Sphingomonas (Zheng et al., 2005), unclassified_Nitrosomon-adaceae (Prosser et al., 2014), Nitrosococcus (Stein et al., 2013),Nitrosomonas (Bock et al., 1995) and Nitrosospira (Hollibaughet al., 2002), with a total proportion of 1.07% (Fig. 5A). When theRR was decreased to 35 and then to 30, the total proportion ofAOB in S2 (Fig. 5B) and S3 (Fig. 5C) was enhanced to 1.53% and1.43%, respectively. Meanwhile, the total proportion of NOB wasalso improved from 0.17% in S1 to 0.25% in S2 and then to 0.43%in S3. The results showed that the microaerobic condition in theUMSR was feasible for the growth of aerobic autotrophic bacteria.However, the total proportion of NOB was remarkably decreased to0.08% with the reduction of RR to 25 in stage 4, though the

TN removal via N2Oproduction from NO2

�-Nb (%)TN removal via nitritereduction to N2

c (%)TN removal via nitratereduction to N2

d (%)

59.0 39.1 23.570.9 47.0 28.350.9 33.8 20.369.4 46.0 27.7

t assimilation (Saggar et al., 2013).ilation (Li et al., 2012).ilation (Li et al., 2012).

930 J. Meng et al. / Bioresource Technology 243 (2017) 922–931

proportion of AOB was further increased to 2.91% at the same time(Fig. 5D).

As aerobic polyphosphate accumulating bacteria (Barker andDold, 1996), Acinetobacterwas also found in the UMSR. The propor-tion in total bacteria was remarkably decreased from 0.84% to0.05% by stages, indicating Acinetobacter was more sensitive toDO. However, as shown as Fig. 2D, TP removal in the four steadyphases of the UMSR was 46.8%, 83.5%, 52.5% and 66.6%, respec-tively. Obviously, the TP removal in the reactor was not positivelycorrelated with the abundance of Acinetobacter, but with the bio-mass as expressed in Section 3.1.

In fact, aerobic bacteria in the UMSR had played an importantrole in consuming the limited DO in forming an anoxic environ-ment that allowed anaerobic bacteria located deep in the activatedsludge flocs (Meng et al., 2016). As illustrated in Fig. 5, hetero-trophic denitrification was one of the most important approachesto nitrogen removal in the UMSR. In stage 1 with the RR 45, theproportion of heterotrophic denitrifiers in S1 was 10.44%(Fig. 5A). With the RR decreased to 35, 30 and 25 by stages(Table 1), the total oxygen supplement to the microaerobic systemdecreased and resulted in an enhancement of heterotrophic deni-trifiers to 16.01%, 10.76% and 15.54%, respectively(Fig. 5B, 5 C and 5D). In addition to the heterotrophic denitrifiers,Thiobacillus as autotrophic denitrifiers (Claus and Kutzner, 1985)was also observed in the microaerobic system. Its proportiondecreased from 0.83% to 0.68% stage by stage, positively correlatedwith the decrease of TN removal (Fig. 2B).

qPCR analysis of the four sampled activated sludge (Fig. 6)showed that: 1) the total bacteria in the UMSR decreased graduallyfrom 7.87 � 105 copies/ngDNA in S1 to 2.35 � 105 copies/ngDNA inS4 with the decreased RR from 45 to 25 by stages, 2) the amoA genein S4 was 5.6 times as that of 5.28 � 101 copies/ngDNA in S1, 3)Nitrospira 16S rRNA decreased obviously from 2.10 � 102 copies/ngDNA in S1 to only 1.10 copies/ngDNA in S4, 4) Nitrobacter 16SrRNA also showed a decreased trend with 10 copies/ngDNA in S4.With the decrease of RR in the UMSR, both absolute abundancevariation of amoA and the total 16S rRNA of Nitrobacter and Nitro-spira were consistent with the change of AOB and NOB revealed bythe high throughput (Fig. 5). It is known that AOB has a better oxy-gen affinity than NOB (Blackburne et al., 2008). Thus, it could beunderstood that the ratio of amoA/(Nitrobacter + Nitrospira)increased with the decrease of RR. It was found that anammox16S rRNA was the most abundant nitrogen functional gene(Fig. 6) with the copies of 6.00 � 103, 4.19 � 103, 5.81 � 103 and4.44 � 103 in S1, S2, S3, S4, respectively. This result indicated thatanammox was the most important approach for NH4

+-N and TNremoval in the UMSR.

Since nirS gene was more abundant than counterpart nirK genein wastewater treatment system (Kim et al., 2011; Zhi et al., 2015),nirS along with narG were selected as the denitrifying genes in thepresent study. Because the dominant anammox would consumethe most NH4

+-N in the MFPW, only a small amount of NOx�-N

was produced (Fig.2B). Without enough NOx�-N as substrate for

denitrifiers, both narG and nirS genes exhibited a downward trendin the UMSR with the decrease of RR by stages (Fig. 6). Therefore,anammox was identified as the main mechanism for nitrogenremoval in the UMSR, and the decrease of RR had a significant neg-ative effect on anammox bacteria, resulting in a less removal forboth NH4

+-N and TN.

4. Conclusion

RR had a remarkable effect on the nitrogen removal in theUMSR. The TN removal presented a decreasing trend with thedecrease of RR from 45 to 25 by stages. To get an effluent TN of

not more than 80 mg/L with a TN load removal above 0.88 kg/(m3�d), the RR should be at least 34. The RR of less than 34 badlyinhibited the growth of anammox bacteria. Though the lower RRenriched AOB, the nitrogen removal remarkably decreased in thereactor due to the inhibited anammox that was the dominantnitrogen removal pathway.

Acknowledgements

The authors gratefully acknowledge the China PostdoctoralScience Foundation funded project (Grant No. 2017M611376),the National Natural Science Foundation of China (Grant No.51478141) and the State Key Laboratory of Urban Water Resourceand Environment, Harbin Institute of Technology (Grant No.2016DX06) for valuable financial support.

References

APHA (American Public Health Association), 2005. Standard Methods for theExamination of Water and Wastewater (19th edition). Washington DC, USA.

Barker, P.S., Dold, P.L., 1996. Denitrification behaviour in biological excessphosphorus removal activated sludge systems. Water Res. 30, 769–780.

Bernet, N., Delgenes, N., Akunna, J.C., Delgenes, J., Moletta, R., 2000. Combinedanaerobic-aerobic SBR for the treatment of piggery wastewater. Water Res. 34,611–619.

Binks, P.R., Nicklin, S., Bruce, N.C., 1995. Degradation of hexahydro-1, 3, 5-trinitro-1,3, 5-triazine (RDX) by Stenotrophomonas maltophilia PB1. Appl. Environ.Microbiol. 61, 1318–1322.

Blackburne, R., Yuan, Z., Keller, J., 2008. Partial nitrification to nitrite using lowdissolved oxygen concentration as the main selection factor. Biodegradation 19,303–312.

Bock, E., Schmidt, I., Stüven, R., Zart, D., 1995. Nitrogen loss caused by denitrifyingNitrosomonas cells using ammonium or hydrogen as electron donors and nitriteas electron acceptor. Arch. Microbiol. 163, 16–20.

Braker, G., Fesefeldt, A., Witzel, K.P., 1998. Development of PCR primer systems foramplification of nitrite reductase genes (nirK and nirS) to detect denitrifyingbacteria in environmental samples. Appl. Environ. Microbiol. 64, 3769–3775.

Canals, O., Salvadó, H., Auset, M., Hernández, C., Malfeito, J., 2013. Microfaunacommunities as performance indicators for an A/O shortcut biological nitrogenremoval moving-bed biofilm reactor. Water Res. 47, 3141–3150.

Claus, G., Kutzner, H.J., 1985. Physiology and kinetics of autotrophic denitrificationby Thiobacillus denitrificans. Appl. Microbiol. Biotechnol. 22, 283–288.

Dionisi, H.M., Layton, A.C., Harms, G., Gregory, I.R., Robinson, K.G., Sayler, G.S., 2002.Quantification of Nitrosomonas oligotropha-like ammonia-oxidizing bacteriaand Nitrospira spp. from full-scale wastewater treatment plants by competitivePCR. Appl. Environ. Microbiol. 68, 245–253.

Graham, D.W., Knapp, C.W., Van Vleck, E.S., Bloor, K., Lane, T.B., Graham, C.E., 2007.Experimental demonstration of chaotic instability in biological nitrification.ISME J. 1, 385–393.

He, L.S., Liu, H.L., Xi, B.D., Zhu, Y.B., 2006. Effects of effluent recirculation in vertical-flow constructed wetland on treatment efficiency of livestock wastewater.Water Sci. Technol. 54, 137–146.

Herlemann, D.P., Labrenz, M., Jürgens, K., Bertilsson, S., Waniek, J.J., Andersson, A.F.,2011. Transitions in bacterial communities along the 2000 km salinity gradientof the Baltic Sea. ISME J. 5, 1571–1579.

Hollibaugh, J.T., Bano, N., Ducklow, H.W., 2002. Widespread distribution in polaroceans of a 16S rRNA gene sequence with affinity to Nitrosospira-like ammonia-oxidizing bacteria. Appl. Environ. Microbiol. 68, 1478–1484.

Hugerth, L.W., Wefer, H.A., Lundin, S., Jakobsson, H.E., Lindberg, M., Rodin, S.,Engstrand, L., Andersson, A.F., 2014. DegePrime, a program for degenerateprimer design for broad-taxonomic-range PCR in microbial ecology studies.Appl. Environ. Microbiol. 80, 5116–5123.

Jetten, M.S., Wagner, M., Fuerst, J., van Loosdrecht, M., Kuenen, G., Strous, M., 2001.Microbiology and application of the anaerobic ammonium oxidation(anammox) process. Curr. Opin. Biotechnol. 12, 283–288.

Jin, R.C., Yang, G.F., Ma, C., Yu, J.J., Zhang, Q.Q., Xing, B.S., 2012. Influence of effluentrecirculation on the performance of Anammox process. Chem. Eng. J. 200, 176–185.

Kertesz, M.A., Kawasaki, A., 2010. Hydrocarbon-Degrading Sphingomonads:Sphingomonas, Sphingobium, Novosphingobium, and Sphingopyxis. Springer,Berlin Heidelberg, pp. 1693–1705.

Kim, Y.M., Cho, H.U., Lee, D.S., Park, D., Park, J.M., 2011. Influence of operationalparameters on nitrogen removal efficiency and microbial communities in a full-scale activated sludge process. Water Res. 45, 5785–5795.

Li, H., Zhou, S., Ma, W., Huang, G., Xu, B., 2012. Fast start-up of ANAMMOX reactor:operational strategy and some characteristics as indicators of reactorperformance. Desalination 286, 436–441.

Li, J., Meng, J., Li, J., Wang, C., Deng, K., Sun, K., Buelna, G., 2016. The effect andbiological mechanism of COD/TN ratio on nitrogen removal in a novel upflowmicroaerobic sludge reactor treating manure-free piggery wastewater.Bioresour. Technol. 209, 360–368.

J. Meng et al. / Bioresource Technology 243 (2017) 922–931 931

López-Gutiérrez, J.C., Henry, S., Hallet, S., Martin-Laurent, F., Catroux, G., Philippot,L., 2004. Quantification of a novel group of nitrate-reducing bacteria in theenvironment by real-time PCR. J. Microbiol. Meth. 57, 399–407.

Meng, J., Li, J., Li, J., Antwi, P., Deng, K., Wang, C., Buelna, G., 2015. Nitrogen removalfrom low COD/TN ratio manure-free piggery wastewater within an upflowmicroaerobic sludge reactor. Bioresour. Technol. 198, 884–890.

Meng, J., Li, J., Li, J., Sun, K., Antwi, P., Deng, K., Wang, C., Buelna, G., 2016. Efficiencyand bacterial populations related to pollutant removal in an upflowmicroaerobic sludge reactor treating manure-free piggery wastewater withlow COD/TN ratio. Bioresour. Technol. 201, 166–173.

MEPPRC (Ministry Environmental Protection of People’s Republic of China), 2001.Discharge Standard of Pollutants for Livestock and Poultry Breeding. Beijing,China.

Muyzer, G., De Waal, E.C., Uitterlinden, A.G., 1993. Profiling of complex microbialpopulations by denaturing gradient gel electrophoresis analysis of polymerasechain reaction-amplified genes coding for 16S rRNA. Appl. Environ. Microbiol.59, 695–700.

Park, S.C., Baik, K.S., Choe, H.N., Lim, C.H., Kim, H.J., Ka, J.O., Seong, C.N., 2011.Altererythrobacter namhicola sp. nov. and Altererythrobacter aestuarii sp. nov.,isolated from seawater. Int. J. Syst. Evol. Micr. 61, 709–715.

Prasad, B., Suresh, S., 2012. Biodegradation of phthalate esters by Variovorax sp.Apcbee Procedia 1, 16–21.

Prosser, J.I., Head, I.M., Stein, L.Y., 2014. The Family Nitrosomonadaceae. In: TheProkaryotes. Springer, pp. 901–918.

Rotthauwe, J.H., Witzel, K.P., Liesack, W., 1997. The ammonia monooxygenasestructural gene amoA as a functional marker: molecular fine-scale analysis ofnatural ammonia-oxidizing populations. Appl. Environ. Microb. 63, 4704–4712.

Saggar, S., Jha, N., Deslippe, J., Bolan, N., Luo, J., Giltrap, D., Kim, D.G., Zaman, M.,Tillman, R., 2013. Denitrification and N2O:N2 production in temperategrasslands: processes, measurements, modelling and mitigating negativeimpacts. Sci. Total Environ. 465, 173–195.

Stein, L., Campbell, M., Klotz, M., 2013. Energy-mediated vs. ammonium-regulatedgene expression in the obligate ammonia-oxidizing bacterium, Nitrosococcusoceani. Front. Microbiol. 4, 277.

Tsushima, I., Kindaichi, T., Okabe, S., 2007. Quantification of anaerobic ammonium-oxidizing bacteria in enrichment cultures by real-time PCR. Water Res. 41, 785–794.

van Dongen, U., Jetten, M.S., Van Loosdrecht, M., 2001. The SHARON-ANAMMOXprocess for treatment of ammonium rich wastewater. Water Sci. Technol. 44,153–160.

Wang, L., Guo, Z.Y., 2009. The biological fermentation bed raises pigs technical theapplication present situation. Livestock and Poultry Industry 3, 006.

Xie, G.J., Cai, C., Hu, S., Yuan, Z., 2016. Complete nitrogen removal from syntheticanaerobic sludge digestion liquor through integrating anammox anddenitrifying anaerobic methane oxidation in a membrane biofilm reactor.Environ. Sci. Technol. 51, 819–827.

Yang, S., Guo, W., Chen, Y., Zhou, X., Zheng, H., Feng, X., Yin, R., Ren, N., 2014.Simultaneous nutrient removal and reduction in sludge from sewage wasteusing an alternating anaerobic-anoxic-microaerobic-aerobic system combiningozone/ultrasound technology. RSC Adv. 4, 52892–52897.

Yuan, Z., Oehmen, A., Ingildsen, P., 2002. Control of nitrate recirculation flow inpredenitrification systems. Water Sci. Technol. 45, 29–36.

Zhang, Y., Wang, X.C., Cheng, Z., Li, Y., Tang, J., 2016. Effect of fermentation liquidfrom food waste as a carbon source for enhancing denitrification in wastewatertreatment. Chemosphere 144, 689–696.

Zhao, B., Li, J., Leu, S.Y., 2014. An innovative wood-chip-framework soil infiltrator fortreating anaerobic digested swine wastewater and analysis of the microbialcommunity. Bioresour. Technol. 173, 384–391.

Zheng, S., Cui, C., 2012. Efficient COD removal and nitrification in an upflowmicroaerobic sludge blanket reactor for domestic wastewater. Biotechnol. Lett.34, 471–474.

Zheng, X.S., Yang, H., Li, D.T., 2005. Change of microbial populations in a suspended-sludge reactor performing completely autotrophic N-removal. World J. Microb.Biot. 21, 843–850.

Zhi, W., Yuan, L., Ji, G., He, C., 2015. Enhanced long-term nitrogen removal and itsquantitative molecular mechanism in tidal flow constructed wetlands. Environ.Sci. Technol. 49, 4575–4583.

Zitomer, D.H., 1998. Stoichiometry of combined aerobic and methanogenic CODtransformation. Water Res. 32, 669–676.