ww.sciencedirect.com

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 2

Available online at w

journal homepage: www.elsevier .com/locate/watres

Molecular indicators ofNitrobacter spp. population and growthactivity during an induced inhibition event in a bench scalenitrification reactor

Shawn Hawkins a,*, Kevin Robinson b, Alice Layton c, Gary Sayler c

aThe University of Tennessee, Department of Biosystems Engineering and Soil Science, 310 Biosystems Engineering and Environmental

Sciences Office, 2506 E.J. Chapman Drive, Knoxville, TN 37996-4531, USAbThe University of Tennessee, Department of Civil and Environmental Engineering, 219A Perkins Hall, 1506 Middle Drive, Knoxville,

TN 37996-2010, USAcThe University of Tennessee, Center for Environmental Biotechnology, 676 Dabney-Buehler Hall, 1416 Circle Drive, Knoxville,

TN 37996-1605, USA

a r t i c l e i n f o

Article history:

Received 7 September 2011

Received in revised form

16 December 2011

Accepted 28 December 2011

Available online 11 January 2012

Keywords:

Nitrification

Nitrite oxidation

Nitrobacter

rRNA transcript

rRNAt/rDNA ratio

qPCR

* Corresponding author. Tel.: þ1 865 974 772E-mail address: shawkins@utk.edu (S. Ha

0043-1354/$ e see front matter ª 2012 Elsevdoi:10.1016/j.watres.2011.12.053

a b s t r a c t

The Nitrobacter spp. ribosomal RNA gene (rDNA) and transcript (rRNAt) abundance were

quantified in a bench scale nitrification reactor during baseline periods of high nitrification

efficiency and an intervening staged inhibition event. The transcript to gene ratio (rRNAt/

rDNA) was highly sensitive to changes in the reactor nitrite oxidation rate. During high

nitrification efficiency, the rRNAt/rDNA metric displayed a range from 0.68 to 2.01 with

one-sided (a¼ 0.10) lower and upper prediction intervals of 0.70 and 1.78, respectively.

When nitrification was inhibited by disabling the reactor pH control system, this

activity metric declined an order of magnitude to w0.05, well below the lower prediction

interval reflecting high nitrification efficiency. The decline was rapid (2 h) and preceded

a significant drop in reactor nitrification performance, which occurred as ammonia accu-

mulated. The rRNAt/rDNA ratio remained low (w0.05) for several days after the pH control

system was re-enabled at a setpoint of 8.0, which otherwise induced rapid oxidation of

accumulated ammonia and produced high free ammonia concentrations. The timing of

a subsequent increase in the rRNAt/rDNA ratio, which transiently exceeded the upper

prediction interval established during the baseline period of high nitrification efficiency,

was not coincidental with resumption of pH control at 7.2 that lowered free ammonia

concentrations to non-inhibitory levels. Rather, nitrite oxidation resumed and the rRNAt/

rDNA ratio increased only after oxidation of accumulated ammonia was complete, which

was coincidental with reduced reactor oxygen demand. In summary, the Nitrobacter rRNAt/

rDNA activity metric reflected timely and easily recognizable changes in nitrite oxidation

activity, illustrating that molecular data can be used to diagnose poor biological waste-

water treatment performance.

ª 2012 Elsevier Ltd. All rights reserved.

2; fax: þ1 865 974 4514.wkins).ier Ltd. All rights reserved.

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 21794

1. Introduction 2. Materials and methods

Ammonia is conventionally removed from low-strength

industrial and municipal wastewater using nitrification in

biological reactors, thereby preventing undue oxygen demand

and toxicity effects on receiving streams. Nitrification is

a two-step aerobic treatment process in which ammonia is

first oxidized to nitrite by ammonia oxidizing bacteria and/or

archaea (AO). This rate-limiting step is followed by the rapid

oxidation of nitrite to nitrate by nitrite oxidizing bacteria

(NOB) (Huang et al., 2010). Even though nitrification

is commonly employed in wastewater treatment plants,

performance is prone to perhaps mutualistic declines in

nitrite and ammonia oxidizing activity (Anthonisen et al.,

1976) that are frequent and not well understood (Rittmann

and McCarty, 2003).

Less conventional biological treatment techniques are

increasingly being used to remove ammonia and nitrogen

from high strength wastewaters such as digester effluent

(Hellinga et al., 1998) and landfill leachate (Ganigue et al.,

2009). Ammonia but not nitrite oxidation is promoted in

these processes and coupled with denitritification (the use

of nitrite as a terminal electron acceptor) and/or anaerobic

ammonia oxidation (annamox) (Van Hulle et al., 2010).

While these biological reactions more efficiently remove

ammonia and nitrogen, it is historically been difficult to

maintain low NOB activity (Fux et al., 2004; Turk and

Mavinic, 1989).

Engineers increasingly have at their disposal molecular

research tools to monitor bacterial sub-population and in-

situ activity levels to help diagnose lagging biological

wastewater treatment performance. The ultimate goal is to

more fully understand and improve the reliability of

conventional treatment processes such as nitrification and

to ease the startup and long term implementation of

advanced technologies such as anammox (Van Hulle et al.,

2005). This effort demands that focus be given to the

development of techniques that make molecular data more

productive for reactor design and startup and for diagnosing

lagging performance. A prime target to advance these

causes in relation to nitrogen treatment is the NOB

because nitrite oxidation activity determines the

efficiency of both conventional and advanced treatment

techniques.

The objective of the research herein was to demonstrate

the usefulness of the real time quantitative polymerase chain

reaction (qPCR) to improve understanding of lagging

nitrification treatment performance. This was accomplished

by measuring the dominant NOB (Nitrobacter spp.) population

and growth activity levels, measured as the ribosomal gene

(rDNA) and ribosomal gene transcript (rRNAt) abundance,

respectively. The study establishes a “normal” range of the

Nitrobacter spp. rRNAt/rDNA activity metric during two inde-

pendent baseline periods of high nitrification efficiency. The

usefulness of Nitrobacter spp. rRNAt/rDNA activity metric data

to signal lagging performance in a timely manner was inves-

tigated during an intervening staged inhibition event, as was

the ability to provide improved understanding for the under-

lying inhibition causes.

2.1. Bench scale nitrification reactor

A 10 L fill and draw, complete mix, bench scale nitrification

reactor (BSNR) was the subject of this study. The BSNR was

maintained at 30� 1 �C in a temperature control room (Model

NSCP2S; Norlake Scientific, Hudson, WI) and continuously

received 1 L/day of poorly buffered influent: 54 mM (NH4)2SO4

(1500 mg-N/L), 1.5 mM K2HPO4, 1.5 mM KH2PO4, 0.75 mM

MgSO4, 0.20 mM CaCl2, 16.6 mM EDTA, 9.9 mM FeSO4, and

0.5 mMCuSO4. This reactor was operated over one year prior to

conducting the experiments herein and typically exhibited

very high nitrification efficiency (97� 3%). The dominant NOB

in this reactor were previously identified as Nitrobacter spp.

(Hawkins et al., 2006).

Humidified air (z800 cm3/min) was directed through

a 0.2 mm, 37 mm inline bacterial air vent (Uniflo-50T, Gelman

Sciences; Ann Arbor, MI) and metered using airflow rate

control tubes (Model 082-02G, Aalborg Instruments; Orange-

burg, NY) before being sparged into the reactor through

stainless steel fine bubble diffusers. A constant airflow rate

was achieved using an inline backpressure regulator (Model

100HR, Controlair; Amherst, NH). The BSNR dissolved oxygen

(DO) concentrationwas typicallymaintained at or above 3 mg/

L during normal operating conditions (high nitrification effi-

ciency). This concentration exceeds the oxygen affinity

constants for nitrifying bacteria (Guisasola et al., 2005).

Nitrification consumes alkalinity at a rate of 7.1 mg CaCO3

permgNH4þ-N oxidized (Grady et al., 1999). By design the BSNR

received a poorly buffered influent and was outfitted with

a control system to maintain the pH (normal setpoint 7.2) that

consisted of an analyzer/controller and fouling resistant elec-

trode (Model WDP300and WEL-PHF-NN, respectively, Wal-

chem Corporation; Holliston, MA). During normal operation,

when the pH fell below 7.2 a relay in the controller energized

a pump (P/N 10570105B2A126, Barnant Company, Barrington,

IL) that injected a 5% Na2CO3 solution into the reactor. This

solution quickly increased the pH slightly above the setpoint,

at which time the analyzer shut power off to the pump.

2.2. Physical and chemical analyses

Soluble nitrogen was quantified in filtered reactor mixed

liquor samples using ion chromatography as previously

described (Hawkins et al., 2010) and used to determine the

nitrification efficiency

Nitrificationefficiencyð%Þ¼

NO�3

NHþ4 þNH3þNO�

2 þNO�3

!�100%

(1)

Concentrations of free ammonia in the BSNR were calcu-

latedusingmeasuredvaluesof the reactorpHand temperature

and the total ammonia concentration (Anthonisen et al., 1976)

NH3ðmg �N=LÞ ¼ 1714

� Total ammoniaðmg �N=LÞ � 10pH

Kb=Kw þ 10pH; (2)

where Kb/Kw ¼ e(6344/273 þ �C)

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 2 1795

The reactor pH, DO, and temperature were continuously

recorded using a model 556 sonde (YSI, Inc.; Yellow Springs,

OH) during an induced inhibition experiment described in

Section 2.5 below. This data was used to corroborate

and refine trends in ammonia and nitrite oxidation

activity inferred with soluble nitrogen concentration data.

Reactor mixed liquor samples were analyzed for volatile sus-

pended solids (MLVSS) using Standard Method 2540E (APHA,

1998) as a grossmeasure of the reactor biomass concentration.

2.3. Nitrobacter spp. ribosomal gene and ribosomal genetranscript quantification

AqPCRassaywasusedto targetaconservedregionbetweenthe

16Sand23S ribosomalRNAgenes in theNitrobacter spp.genome

(Hawkins et al., 2006). In reactormixed liquor DNA extracts, the

ribosomal RNA gene (rDNA) was quantified and served as

a direct measure of the Nitrobacter spp. population level

(Navarro et al., 1992). In reactor mixed liquor RNA extracts, the

ribosomal RNA transcript (rRNAt) abundance was quantified

followinga reverse transcription step and servedasanabsolute

measure of Nitrobacter spp. growth activity (Lu et al., 2009)

Triplicate biomass samples were collected from the BSNR

three times per week during two baseline periods of high

nitrification efficiency and more frequently during an inter-

vening induced inhibition event. These samples were

preserved, nucleic acids were extracted, and qPCR was per-

formed to measure the Nitrobacter spp. ribosomal gene (rDNA)

and transcript (rRNAt) abundance as previously

described (Hawkins et al., 2006). The detection system tar-

geted a genus specific intergenic spacer region between the

16S and 23S ribosomal RNA genes and consisted of the

primers NITISRf (50-CCATTCACTATCTCCAGGTC-30) and

NITISRr (50-TGATTAGAAAGACCAGCTTGC-30) and a fluores-

cent probe NITISRp [50-TCGAACCGATAGCGAGGCGG-30]. Nitrobacter spp. rDNA abundance was quantified in 1:5

dilutions of the sample DNA extracts using a real time PCR

reaction containing: 12.5 ml of QuantiTect� Probe PCR master

mix (Qiagen; Valencia, CA), 5.125 ml of nuclease free water,

400 nMof each primer (1 ml), 150 nMof probe (0.375 ml), and 5 ml

of template. The rDNA assay temperature protocol was as

follows: 50 �C for 2 min, Taq activation at 95 �C for 15 min, and

40 cycles with melting at 94 �C for 15 s and annealing/exten-

sion at 61 �C for 1 min. Sample amplification efficiency was

typically 91% with a detection limit of 50 copies/reaction.

Nitrobacter spp. rRNAt abundance was quantified in 1:5 dilu-

tions of the sample RNA extracts by real time RT-PCR using

the following reaction mix prepared on ice: 12.5 ml of

QuantiTect� Probe RT-PCR master mix (Qiagen; Valencia, CA),

4.875 ml of RNase free water, 400 nM of each primer (1 ml),

150 nM of probe (0.375 ml), 0.25 ml of RT enzymemix, and 5 ml of

template. The rRNAt assay temperature protocol was as

follows: RT reaction at 50 �C for 30 min, Taq activation at 95 �Cfor 15 min, and 40 cycles with melting at 94 �C for 15 s and

annealing/extension at 61 �C for 1 min. Sample amplification

efficiency was typically 95% with a detection limit of 500

copies/reaction. Both assays were performed on a MJ DNA

Engine Opticon thermocycler in triplicate using the average

fluorescence for cycles 3e7 for baseline subtraction, a fluo-

rescence threshold of 0.005, and external standard curves.

2.4. BSNR baseline activity assessment

The BSNR was monitored for 65 days prior to an induced

inhibition experiment during which time the nitrification

efficiency and MLVSS were measured three times per week

and the Nitrobacter spp. rDNA and rRNAt concentrations were

measured once per week. These same parameters were

measured for 19 additional days following full recovery from

the induced inhibition event. The objective for monitoring

these baseline periodswas to establish expected values for the

Nitrobacter spp. rDNA/rRNAt ratio using non-parametric lower

and upper prediction intervals during documented periods of

high nitrification efficiency.

2.5. BSNR induced inhibition experiment

Nitrification inhibition was induced in the BSNR following 65

days of continuous high nitrification efficiency. The inhibition

event lasted twoweeks and consisted of three phases initiated

by a single manipulation of the reactor pH control system,

after which certain responses in the AO and NOB activity

levels were expected (Table 1). Hypotheses were made for the

rDNA/rRNAt activity metric for each stage of the inhibition

experiment (Table 1). The experimentwas analogous to a drop

in influent alkalinity in a reactor treating highly nitrogenous

wastewater (Grady et al., 1999).

2.6. Statistical analyses

An analysis of variance (ANOVA) of the log transformed

Nitrobacter spp. rDNA and rRNAt abundance measurements

(a¼ 0.05) was used to assess whether the Nitrobacter pop-

ulation and activity varied significantly across periods of high

and lownitrification efficiency. Groups of statistically different

data were established using a post hoc Tukey Honestly Signifi-

cant Difference (HSD) analysis (a¼ 0.05). In this study, data

were assigned to Tukey HSD groups using letter designations;

data points not sharing a common letter were significantly

different. Trends in the reactormolecular and soluble nitrogen

data over time were established non-parametrically using the

slope of a Kendall-Thiel robust line (a¼ 0.05). Correlations

between the reactor MLVSS, rDNA, and rRNAt data were

established non-parametrically using Kendall’s s (a¼ 0.05).

The maximum and minimum expected rRNAt/rDNA ratio

values under steady state, high nitrification efficiency were

established with upper and lower non-parametric prediction

intervals (a¼ 0.1), respectively. Statistical analyses were per-

formed as previously described (Helsel andHirsch, 2002) using

the JMP software package (SAS Institute, Inc.; Cary, NC) or an

Excel spreadsheet (Microsoft, Inc.; Redmond, CA).

3. Results and discussion

3.1. Baseline periods of high nitrification efficiency

3.1.1. Soluble nitrogen concentrationsReactornitrogenconcentrationsweremeasuredduringa65day

timeperiodprior to, anda19day timeperiodafter recover from,

a staged inhibition event (Fig. 1-A). The total ammonia, free

Table 1 e Hypotheses for a three phase inhibition experiment in the bench scale nitrification reactor.

Phase Perturbation Expected response NOB activity interpretation Nitrobacter rRNAt/rDNAhypotheses

1 pH control

disabled

Ammonia but not nitrite

accumulates, nitrate declines

AO inhibited; NOB

substrate starved

Declines below the lower

prediction interval established

during baseline periods of

high nitrification efficiency

2 pH control enabled

at setpoint 8.0

Free ammonia immediately

spikes, nitrate and total

ammonia decline,

nitrite accumulates

AO become active; NOB

inhibited by high AO

activity and/or free ammonia

Remains below the lower

prediction interval established

during baseline periods of high

nitrification efficiency

3 pH control setpoint

lowered to 7.2

Free ammonia lowered,

ammonia oxidized,

nitrite oxidation

eventually resumes,

nitrate eventually increased

NOB become active, the

timing of which indicates

the importance of free

ammonia inhibition

during phase 2

Increases above the upper

prediction interval established

during baseline periods of high

nitrification efficiency

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 21796

ammonia, and nitrite concentrations were low and the nitrate

concentrations were high. These data confirm that prolonged

periods of high, steady state nitrification efficiency (�97%)

occurred both before and after the staged inhibition event.

A

B

Fig. 1 e Bench scale nitrification reactor (A) soluble nitrogen an

suspended solids data collected during a 65 day time baseline

recovery from, a staged inhibition event.

3.1.2. Nitrobacter spp. rDNA and rRNAt abundance and therRNAt/rDNA ratioBaseline Nitrobacter spp. rDNA, rRNAt, and the rRNAt/rDNA

ratio data are provided in Fig. 1-B. ANOVA rejected the null

d (B) Nitrobacter spp. molecular and mixed liquor volatile

period prior to, and a 19 day baseline time period after

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 2 1797

hypothesis that the Nitrobacter spp. rDNA abundance

measured at various times during this study were similar

(F¼ 20; p-value< 0.001). However, a subsequent Tukey HSD

analysis revealed that a clear majority of the Nitrobacter rDNA

concentrations during the baseline periods of high nitrifica-

tion efficiency was similar, including all the measurements in

the five weeks prior to and the three weeks following recovery

from the staged inhibition event (Table 2 e Tukey HSD group

E). Interestingly, the mean rDNA abundance and MLVSS data

were not significantly correlated (Kendall’s s¼�0.091;

p-value¼ 0.697).

Overall, the Nitrobacter rDNA levels were similar to those

found for Nitrospira rDNA in full scale municipal (12 samples

collected monthly: 3� 2� 1010 copies/L) and industrial

(3 samples: 3� 2� 1010 copies/L) wastewater treatment plants

(WWTPs), even though the BSNR influent ammonia concen-

tration (1500 mg-N/L) wasmuch higher than for themunicipal

(z20 mg-N/L) and industrial (45 mg-N/L) reactors (Dionisi

et al., 2002). Nitrospira rDNA abundance in four bench scale

reactors with widely different volatile suspended solids

concentrations (350e2100 mg/L: 1.0� 109e3.6� 1010 copies/L)

were also similar to the Nitrobacter spp. concentrations

Table 2 e Statistical analysis of Nitrobacter spp. rDNA and rRNAevent in the bench scale nitrification reactor.

Operationalday

Inhibition experiment(h since perturbations)

Nit

Phase 1 Phase 2 Phase 3 Mean�

2 2.28� 0.2

9 1.81� 0.0

16 1.84� 0.0

23 1.70� 0.4

30 1.64� 0.2

37 1.39� 0.1

44 1.19� 0.0

51 9.11� 0.7

58 8.84� 1.0

65 �0.1 9.60� 2.3

2.1 9.44� 2.3

3.1 1.25� 0.1

4.0 9.81� 1.9

5.5 9.67� 1.1

8.1 7.80� 3.4

24.7 8.41� 1.6

72.1 �0.1 4.69� 0.8

7.9 5.47� 0.3

20.0 3.37� 0.2

31.7 4.84� 0.6

44.1 �0.1 3.17� 0.3

11.7 6.24� 0.8

22.0 3.80� 0.7

33.9 4.77� 1.0

45.8 4.81� 1.0

69.9 6.32� 1.3

120.0 1.26� 0.1

168.0 1.84� 0.0

87 216.0 1.18� 0.0

96 9.12� 0.3

103 9.53� 0.8

observed in the BSNR during high nitrification

efficiency (Dionisi et al., 2003).

ANOVA rejected the null hypothesis that the Nitrobacter

spp. rRNAt abundance measured at various times during this

study were similar (F¼ 204; p-value< 0.001). However,

a subsequent Tukey HSD analysis revealed that the vast

majority of the Nitrobacter rRNAt concentrations during both

baseline periods of high nitrification efficiency was not

statistically different (Table 2). The relative range of transcript

abundance during steady state, high nitrification efficiency

(2.8-fold) was similar to the relative range observed for the

gene abundance (2.6-fold). The BSNR Nitrobacter rRNAt range

of values at steady state N-treatment performance was lower

than that observed for Pseudomonas abietaniphila rRNA (4-fold

range) in a complete mix bench scale reactor at steady state

(Muttray and Mohn, 1998). Other studies have generally re-

ported that the ribosomal transcript abundance was

decidedly more variable than rDNA and/or cell abundance at

steady state conditions (Cangelosi and Brabant, 1997; Licht

et al., 1999; Oerther et al., 2000). The mean rRNAt abundance

was not significantly correlated with MLVSS or the mean

rDNA data.

t data collected before, during, and after a stated inhibition

Statistical summary

robacter spp. rDNA Nitrobacter spp. rRNAt

SD TukeyHSD groups

Mean� SD TukeyHSD groups

0� 1010 A 3.48� 0.20� 1010 AB

9� 1010 ABC 2.16� 0.34� 1010 BCDE

3� 1010 AB 1.26� 0.27� 1010 E

2� 1010 ABCD 1.31� 0.06� 1010 DE

8� 1010 ABCDE 2.17� 0.30� 1010 BCDE

4� 1010 ABCDEF 2.46� 0.05� 1010 BCD

8� 1010 BCDEFG 1.39� 0.24� 1010 DE

5� 109 DEFGHI 1.62� 0.03� 1010 DE

7� 109 EFGHIJK 1.35� 0.08� 1010 DE

0� 109 DEFGH 1.50� 0.10� 1010 DE

3� 109 DEFGH 5.85� 0.30� 109 F

3� 1010 ABCDEFG 5.76� 0.13� 109 F

2� 109 CDEFGH 3.28� 0.72� 109 FG

9� 109 CDEFGH 1.74� 0.37� 109 GH

1� 109 GHIJKL 7.98� 3.16� 108 I

7� 109 FGHIJKL 4.63� 0.54� 108 IJK

1� 109 LMN 3.78� 0.48� 108 JK

8� 109 HIJKLMN 7.02� 0.62� 108 IJ

7� 109 MN 2.78� 0.27� 108 K

5� 109 IJKLMN 4.93� 1.02� 108 IJK

9� 109 N 4.72� 0.55� 108 IJK

4� 109 HIJKLM 8.94� 1.37� 108 HI

5� 109 MN 3.64� 1.05� 108 K

4� 109 KLMN 3.41� 0.53� 108 K

5� 109 JKLMN 5.79� 0.99� 109 F

9� 109 HIJKLM 1.86� 0.40� 1010 BCDE

7� 1010 ABCDEFGH 3.40� 0.23� 1010 ABC

9� 1010 AB 4.88� 0.26� 1010 A

2� 1010 BCDEFG 1.79� 0.52� 1010 CDE

9� 109 DEFGHIJ 1.83� 0.31� 1010 BCDE

4� 109 DEFGHIJ 1.47� 0.06� 1010 DE

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 21798

A primary goal in conducting this study was to demon-

strate whether molecular data could be used to quickly and

reliability predict reactor treatment performance. Our

approach was to calculate prediction intervals for Nitrobacter

spp. rRNAt/rDNA activity metric, a normalized measure of

growth activity, during baseline periods of documented high

nitrification efficiency. Metric values during the subsequent

staged inhibition event were compared with the prediction

intervals to unambiguously and rigorously establish unusu-

ally high or low growth activity.

The Nitrobacter spp. rRNAt/rDNA ratio ranged from 0.68 to

2.01 during the baseline periods of high nitrification

efficiency before and after the staged nitrification inhibition

event (Fig. 1-B). These thirteen data points were combined to

establish non-parametric lower (0.70) and upper (1.78)

prediction intervals. It was hypothesized that during nitrifi-

cation inhibition the ratio would fall below the lower predic-

tion interval, and during recovery from inhibition the metric

would exceed the upper prediction interval (Table 1). The

lower prediction interval for the Nitrobacter spp. rRNAt/rDNA

in this study was well above the values for this metric during

starvation and exposure to diverse inhibitors in previous

batch experiments (Hawkins et al., 2008, 2010).

3.2. Staged inhibition experiment

3.2.1. Reactor pH and DOPhase 1 of the inhibition experiment was initiated by disabling

the reactor pH control system on operating day 65 (Fig. 1).

Immediately thereafter the reactor pH dropped rapidly (�0.17

pH units/h) for 3.5 h (Fig. 2-A). The steep decrease in pH was

accompanied by a gradual increase in the reactor DO

concentration from 3 to 3.5 mg/L. After 3.5 h, the pH tempo-

rarily stabilized at 6.7 for 30 min, during which time the DO

increased sharply from 3.5 to 6.2 mg/L. Thereafter, the pH

continued to decrease at a much slower rate (�0.04 pH units/

h) and eventually stabilized at the influent pH value of 5.6. The

main biochemical reaction in the BSNR was nitrification

which produces acidity and consumes dissolved oxygen.

Thus, the pH and DO data confirm that nitrification activity

gradually decreased and then abruptly ended 3.5 h after the

pH control system was disabled.

Phase 2 of the inhibition experiment was initiated by re-

enabling the reactor pH control system at a setpoint of 8.0.

This setpoint was attained within 15 min and maintained in

a narrow range thereafter (Fig. 2-A). The reactor DO concen-

tration fell sharply after the pH control system was enabled

(from 6.3 to 4.8 mg/L in 1.1 h). Thereafter, the DO declined to

z2.25 mg/L after 44 h at pH 8.0, with the concentration

dropping more quickly as phase 2 of the experiment pro-

ceeded (Fig. 2-A). This indicated aerobic biological activity

began concomitantly with the pH change and increased

gradually over time.

Phase 3 of the inhibition experiment was initiated by

changing the reactor pH setpoint from 8.0 to 7.2. The pH

dropped quickly at a rate of 0.19 units/h until pH 7.5 was

attained after 2.7 h (Fig. 2-A). This rate of decline was similar

to that observed after pH control was disabled to initiate phase

1 and was likely controlled by the acidity produced during

ammonia oxidation. Between 2.7 and 5.2 h after the setpoint

change, the rate of pH decline slowed to 0.10 pH units/h. After

5.2 h the pH control system became active and thereafter

maintained the pH between 7.14 and 7.39.

The DO concentration decline that began in phase 2

continued into phase 3 (Fig. 2-A). During phase 3, the DO

further decreased from 2.3 to 0.9 mg/L after 22.7 h had

elapsed, and remained below 2.0 mg/L until approximately

37.5 h had elapsed. Between 37.5 and 44 h after phase 3 began,

the DO concentration increased sharply to approximately

3.8 mg/L. This suggested that a significant source of aerobic

activity ended during this time frame. Thereafter, the DO

concentration was maintained between approximately 3 and

4 mg/L.

3.2.2. Soluble nitrogen concentrationsThe total ammonia concentration increased monotonically

(Kendall’s s¼ 1; p-value¼ 0.0014) from less than 1 mg-N/L to

28.7 mg-N/L within 8.1 h after phase 1 of the inhibition

experiment began (Fig. 1-B). However, it would have been

difficult to justify a prediction of deteriorating treatment

performance using only the total ammonia trend and

concentration at that time. After 72 h without pH control, the

total ammonia concentration was very high (403 mg-N/L; 41%

of soluble nitrogen) and indicated nitrification was severely

inhibited. Because the pH was low, the free ammonia

concentration remained insignificant and would not inhibit

ammonia or nitrite oxidation (Anthonisen et al., 1976), there-

fore the low pH was clearly causative (Hawkins et al., 2008).

Nitrite concentrations remained below 0.1 mg-N/L throughout

phase 1, indicating that ammonia rather nitrite oxidation was

inhibited (Ruiz et al., 2003). The reactor nitrate concentration

dropped from 1298 to 995 mg-N/L during phase 1 of the inhi-

bition experiment.

The rate of nitrate decline, as well as the rate of ammonia

accumulation, suggested that these concentrations were

controlled hydraulically in the absence of nitrification during

phase 1 of the inhibition experiment. A nitrogenmass balance

was formulated under the assumption that ammonia

oxidizing activity ended 4 h after the pH control system was

disabled. Using the total ammonia and nitrate concentrations

measured at that time, along with the influent flow rate and

the sample and wastage volumes, the nitrate decline and

ammonia accumulation were projected forward. The results

(982 mgNO3�-N/L; 377 mgNH4

þ-N/L) agreed well with the final

phase 1 measurements (995 mgNO3�-N/L; 402 mgNH4

þ-N/L),

confirming that the ammonia and nitrate concentrationswere

controlled hydraulically in the absence of nitrification during

most of phase 1 of the inhibition experiment.

While the total ammonia concentration increased rapidly

during phase 1 of the experiment due to hydraulic washout,

the trend for increasing ammonia concentrations ended

abruptly when pH control was re-enabled (Fig. 2-B). During the

first 8 h of phase 2, the total ammonia concentration remained

approximately constant (396� 17 mg-N/L). Thereafter, the

concentration decreased to 301 mg-N/L after 44 h. This indi-

cated that during the first 8 h after enabling the pH control

system, the ammonia oxidization rate approximately

equaled the influent ammonia flow rate, but toward the end of

phase 2, ammonia oxidizing activity exceeded the influent

loading rate.

Fig. 2 e Bench scale nitrification reactor (A) pH and DO, (B) soluble nitrogen, and (C) Nitrobacter spp. molecular and mixed

liquor volatile suspended solids data during a three phase staged inhibition event.

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 2 1799

As the pH increased from 5.6 to 8.0, the free ammonia

concentration increased from 0.1 to 30 mg-N/L due to a rapid

acid/base equilibrium reaction (Equation (2)). Thereafter, the

concentration decreased gradually to 23.4 mg-N/L after 44 h

due to a drop in the total ammonia concentration. This free

ammonia concentration range is well above many previously

reported threshold inhibition levels for un-acclimated NOB

inhibition (Abeling and Seyfried, 1992; Anthonisen et al., 1976;

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 21800

Suthersan and Ganczarczyk, 1986; Yang et al., 2004). Because

significant volatile losses of ammonia gas due to air stripping

do not occur at <pH 8.5 (Ford et al., 1980), the decrease in the

ammonia concentration can be attributed to AO activity.

The nitrite concentration increased from below detection

limits to 212 mg-N/L with an increasing rate of accumulation

as phase 2 proceeded (Fig. 2-B). Thus, the nitrite data confirm

that ammonia was oxidized during phase 2. The nitrate

concentration decreased steadily when pH control was absent

in the reactor, a trend that was disrupted for the first 8 h of

phase 2, during which time the concentration remained

relatively constant (971� 4.0 mg-N/L) (Fig. 2-B). However,

thereafter the nitrate concentration resumed a decline down

to 847 mg-N/L at approximately the same rate noted during

phase 1 of the inhibition experiment. Because the nitrate

concentration declined during most of phase 2, NOB activity

clearly remained low.

The free ammonia concentration decreased rapidly almost

an order ofmagnitude in the first 6 h after the pH setpoint was

adjusted to 7.2 to start phase 3 of the inhibition experiment

(Fig. 2-B). If the high free ammonia concentration inhibited

nitrite oxidation during phase 2 of the experiment, the lower

concentration established in the first 6 h of phase 3 would

have been expected to stimulate NOB activity early in phase 3.

These changes certainly had little effect on ammonia oxida-

tion because the decline in the reactor total ammonia

concentration that began in phase 2 continued into phase 3

(Fig. 2-B). In fact, within the first 46 hours of the pH setpoint

change to 7.2, the total ammonia concentration decreased

from 301 to 5 mg-N/L, confirming high ammonia oxidizing

activity was present. Interestingly, this high level of ammonia

oxidizing activity occurred during a continued DO concen-

tration decline that began in phase 2 and continued through

phase 3 (Fig. 2-A). Also, a sharp rise in DO concentration that

occurred between 37.5 and 44 h coincided with the disap-

pearance of excess ammonia, confirming that the loss in

oxygen demandwas due to declining AOB activity after excess

ammonia was removed from the reactor.

Nitrite levels continued to increase with the declining

ammonia concentration during the first 58 h of phase 3, even

though the free ammonia concentration was below inhibitory

levels (Fig. 2-B). The subsequent recovery of NOB activity was

coincidental with the removal of excess ammonia and the rise

in DO concentration in the reactor, rather than with the

significant free ammonia reduction that occurred 40 h earlier.

This indicated free ammonia was not the principle inhibiting

factor for nitrite oxidation in phase 2 (Hawkins et al., 2010;

Simm et al., 2006). Rather, the data indicate that NOB inhibi-

tion was probably due to poor competition for oxygen with

highly active AOB, despite the relatively high bulk DO

concentration. Once NOB activity resumed, nitrite was quickly

removed from the reactor. Between 58 and 192 h after the pH

setpoint was changed to 7.2, the nitrite concentration

decreased from 586 to 1.3 mg-N/L.

Nitrate concentrations corroborate the assessment of AOB

and NOB activity formulated based on the ammonia, and

nitrite data. The washout of nitrate that began in phase 1

continued through phase 2 and 46 h into phase 3 (Fig. 2-B). At

this point the lowest nitrate concentration of the experiment

occurred. Thereafter, nitrate increased for the first time since

the inhibition experiment was initiated, indicating that NOB

activity began between 46 and 58 h into phase 3. As excess

nitrite was removed over the next 134 h, the nitrate concen-

tration increased concomitantly to 1198 mg-N/L, a concentra-

tion similar to that measured at the beginning of phase 1.

3.2.3. Nitrobacter spp. rDNA and rRNAt abundance and therRNAt/rDNA ratioTheNitrobacter rDNA concentrationsmeasured during the first

24 h of phase 1were not significantly different from the values

measured in the reactor over the three week time period prior

to initiating the inhibition experiment (Fig. 2-C; Table 2, Tukey

HSD group G). However, the concentration measured 72 h

after eliminating pH control was significantly lower than at

the beginning of the experiment. The relatively modest total

decrease in rDNA abundance (2.7-fold) was similar to the

range of rDNA concentrations measured during the baseline

time periods (Fig. 1-B) and to the sensitivity range for qPCR

(Dionisi et al., 2003; Hawkins et al., 2006). This clearly indicates

that bacterial population levels would serve as a poor diag-

nostic tool for failing biological reactor treatment

performance.

The Nitrobacter rRNAt concentration displayed a negative

monotonic correlation with time during the first 8 h after

phase 1 began (Kendall’s s¼�1; p-value¼ 0.0014)

and declined at a very high rate (�1.43� 109 copies/L/h)

(Fig. 2-C). Further, the absolute decline in transcript abun-

dance during this short time frame was very large (z20-fold).

The Tukey HSD analysis of the log transformed assay mean

data revealed that a mere 2 h after the pH control system was

disabled, the transcript concentration had significantly

declined (Table 2), even though the DO and pH data indicated

residual nitrification activity was likely present. Thus, the

Nitrobacter rRNAt concentration delineated a potential inhib-

itory condition (loss of pH control) much sooner than the

nitrogen data and even before the pH and DO data. Over the

entire time course of phase 1, the transcript concentration

declined monotonically and was lowest in the last sample

taken (Table 2). The rapid transcript abundance decline in this

study was similar to that observed for P. abietaniphila rRNA in

a bench scale reactor following a pH change from 6.5 to

12.5 (Muttray et al., 2001).

As hypothesized (Table 1), the rRNAt/rDNA ratio declined

below the lower BSNR prediction interval within 2 h of elimi-

nating pH control. Metric values at 8, 24.7, and 72 h after the

pH control system was disabled were all one order of magni-

tude below the lower prediction interval (Table 2). The results

clearly demonstrate this metric could be used to quickly and

unambiguously identify lapses in treatment performance.

The relative change in the Nitrobacter rRNAt/rDNA ratio

during phase 1 (31-fold) was larger than reported for a rRNA/

rDNA metric (5-fold) used to infer activity of a P. abietaniphila

strain in a bench scale reactor following a pH change from 6.5

to 12.5 (Muttray et al., 2001). Further, the changes herein were

larger than for the Acinetobacter spp. rRNAt/rRNA ratio in

bench scale sequencing batch reactors treating filtered

wastewater and during activated sludge transition from the

aeration basin to the return sludge basin at a full scale WWTP

(Oerther et al., 2000). Although this study focuses on Nitro-

bacter spp., the relative rRNAt abundance in biological

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 2 1801

wastewater treatment reactors appears to be a broadly

sensitive indicator of bacterial activity (Lu et al., 2009) that

could be used to diagnose lapses in treatment performance.

The Nitrobacter rDNA concentrations did not vary signifi-

cantly after the pH control system was re-enabled to begin

phase 2 (Table 2; Tukey HSD Group N). However, the abun-

dance at the end of phase 2 was the lowest observed in the

entire experiment. Further, when the phase 1 and phase 2

data were combined, a significant monotonic decline was

present (Kendall’s s¼�0.697; p-value¼ 0.001). The Nitrobacter

rRNAt concentrations remained low after the pH control

system was enabled (Fig. 2-C), reflecting the evident lack of

NOB activity in the reactor. As hypothesized (Table 1), the

rRNAt/rDNA ratio remained well below the lower BSNR

prediction interval throughout phase 2 of the experiment

(Fig. 2-C).

The Nitrobacter rDNA did not increase during the first 46 h

after the pH setpoint was changed from 8.0 to 7.2, beginning

phase 3 of the inhibition experiment, even though the MLVSS

increased steadily during this time frame (Fig. 2-C). A Tukey

HSD analysis of the log transformed rDNA assay mean data

confirmed that during this time the Nitrobacter rDNA concen-

trations were not significantly different than during the first

two phases of the inhibition experiment (Table 2; Tukey HSD

Group M). However, the Tukey HSD analysis did indicate that

the Nitrobacter rDNA concentration increased substantially

between 46 and 120 h after phase 3 was initiated (Table 2;

Tukey HSD group A). It was during this time frame that nitrite

oxidation resumed and excess nitrite was removed from the

reactor (Fig. 2-B).

The Nitrobacter rRNAt concentration was low during the

first 33 h of phase 3, but increased over an order of magnitude

between 33 and 46 h into phase 3 (Fig. 2-C). This sudden

increase in absolute Nitrobacter spp. growth activity was

concomitant with excess ammonia removal and a significant

increase in the DO concentration, and occurred immediately

prior to a renewal in nitrite oxidation activity between 46 and

58 h into phase 3 (Fig. 2-B). Thereafter, the rRNAt concentra-

tion increased steadily to 4.88� 0.03� 1010 copies/L, during

which time the accumulated nitrite was removed. In fact,

168 h after phase 3was initiated the transcript abundancewas

significantly higher than at the beginning of phase 1 (Table 2).

Thus, the Nitrobacter rRNAt abundance was low at the begin-

ning of phase 3 when nitrite oxidation was inhibited,

increased monotonically as excess nitrite was removed, and

declined slightly after the excess nitrite was exhausted,

accurately reflecting nitrite oxidation activity.

The Nitrobacter rRNAt/rDNA ratio was well below the

baseline lower prediction interval during the first 33 h of

phase 3, confirming low NOB activity (Fig. 2-C; Table 2). Only

12 h later, the ratio increased above the lower baseline

prediction interval to 1.0, confirming that nitrite oxidation and

Nitrobacter spp. activity had significantly increased. As

hypothesized (Table 1), the ratio then exceeded the upper

prediction interval as accumulated nitrite was oxidized,

indicating that the NOB activity level was higher than would

be expected in the BSNR when steady state, high nitrification

efficiency was maintained. After the excess nitrite was

removed, the ratio returned to between the upper and lower

prediction intervals.

4. Synthesis of results

The abundance of rRNAt has been linked to growth activity in

both fast and slow growing bacteria (Cangelosi and Brabant,

1997; Cangelosi et al., 1996; Stroot and Oerther, 2003) and the

production rate of rRNAt is generally correlated with bacterial

growth rates (Cutter and Stroot, 2008). Our work with Nitro-

bacter spp. has shown that while the ribosomal rDNA gene

abundance is a poor indicator of short term growth or nitrite

oxidation activity, it is useful as a population indicator to

normalize rRNAt abundance to assess growth and nitrite

oxidation activity as long term changes in reactor population

levels occur (Hawkins et al., 2008). The normalized ratio of the

Nitrobacter spp. transcript to gene abundance (rRNAt/rDNA)

was sensitive to a variety of inhibitors with different mecha-

nisms in batch assays (Hawkins et al., 2008, 2010), but has not

been studied in a continuous flow reactor.

In this work, a typical nitrification inhibition event

(ammonia buildup, followed by ammonia oxidation and

nitrite buildup, then full recovery) was staged in a nitrifier

enriched bench scale reactor (Van Hulle et al., 2010) to provide

an interpretive context for the usefulness of the rRNAt/rDNA

activity metric. NOB activity was well documented using

a combination of pH, DO, and nitrogen data. However, the

reactor design (high influent ammonia concentration,

absence of reduced carbon, and aerobic environment), and the

fact that nitrification is relatively simple (NH3/NO2�/NO3

�),enhanced the effectiveness of these parameters in assessing

activity. If the influent had contained organic substrates,

variable influent ammonia concentrations, or if more

complicated N-conversion pathways were promoted, the

effectiveness of these traditional activity metrics would have

been greatly dimensioned and molecular tools would have

been vital to correctly interpret activity trends.

As hypothesized, the Nitrobacter rRNAt/rDNA ratio re-

flected NOB physiological activity throughout the course of

the staged inhibition experiment. After pH control was

disabled, the metric signaled inhibition within 2 h (well before

excessive ammonia accumulated) through comparison with

a lower prediction interval established during independent

baseline periods of high nitrification performance both before

and after the staged inhibition experiment. This

demonstrated a technique that could be used by engineers to

proactively prevent and ameliorate the effect of nitrification

inhibition using molecular activity data, though sample pro-

cessing times with qPCR currently limit this ability. As ex-

pected, the rRNAt/rDNA ratio remained below the lower BSNR

prediction interval while nitrite oxidation

remained inhibited throughout phase 2. After the pH setpoint

was changed to 7.2, the rRNAt/rDNA ratio remained low even

though the free ammonia concentration, a potential NOB

inhibitor, was drastically reduced. A subsequent substantial

increase in the metric value, above the upper prediction

interval established during high nitrification performance,

occurred concomitantly with exhaustion of excess reactor

ammonia and an obvious decrease in oxygen demand within

the reactor. Thus, the metric data clearly supported

a hypothesis that poor competition for oxygen with highly

active AOB was responsible for NOB inhibition during the

wat e r r e s e a r c h 4 6 ( 2 0 1 2 ) 1 7 9 3e1 8 0 21802

experiment, and dismisses, alongwith other studies (Hawkins

et al., 2010; Simm et al., 2006), the long held notion that free

ammonia significantly inhibits nitrite oxidation (Anthonisen

et al., 1976).

r e f e r e n c e s

Abeling, U., Seyfried, C.F., 1992. Anaerobic-aerobic treatment ofhigh-strength ammonium wastewater e nitrogen removal vianitrite. Water Science and Technology 26 (5e6), 1007e1015.

Anthonisen, A.C., Loehr, R.C., Prakasam, T.B.S., Srinath, E.G.,1976. Inhibition of nitrification by ammonia and nitrous acid.Journal Water Pollution Control Federation 48 (5), 835e852.

APHA, 1998. Standard Methods for the Examination of Water andWastewater, 20th ed. American Public Health Association,New York, NY.

Cangelosi, G.A., Brabant, W.H., 1997. Depletion of pre-16S rRNA instarved Escherichia coli cells. Journal of Bacteriology 179 (14),4457e4463.

Cangelosi, G.A., Brabant, W.H., Britschgi, T.B., Wallis, C.K., 1996.Detection of rifampin- and ciprofloxacin-resistantMycobacterium tuberculosis by using species-specific assays forprecursor rRNA. Antimicrobial Agents and Chemotherapy 40(8), 1790e1795.

Cutter, M.R., Stroot, P.G., 2008. Determination of specific growthrate by measurement of specific rate of ribosome synthesis ingrowing and nongrowing cultures of Acinetobacter calcoaceticus.Applied and Environmental Microbiology 74 (3), 901e903.

Dionisi, H.M., Harms, G., Layton, A.C., Gregory, I.R., Parker, J.,Hawkins, S.A., Robinson, K.G., Sayler, G.S., 2003. Power analysisfor real-time PCR quantification of genes in activated sludgeand analysis of the variability introduced by DNA extraction.Applied and Environmental Microbiology 69 (11), 6597e6604.

Dionisi, H.M., Layton, A.C., Harms, G., Gregory, I.R., Robinson, K.G.,Sayler, G.S., 2002.QuantificationofNitrosomonas oligotropha-likeammonia-oxidizing bacteria and Nitrospira spp. from full-scalewastewater treatment plants by competitive PCR. Applied andEnvironmental Microbiology 68 (1), 245e253.

Ford, D.L., Churchwell, R.L., Kachtick, J.W., 1980. Comprehensiveanalysis of nitrification of chemical processing wastewaters.Journal Water Pollution Control Federation 52 (11), 2726e2746.

Fux, C., Huang, D., Monti, A., Siegrist, H., 2004. Difficulties inmaintaining long-term partial nitritation of ammonium-richsludge digester liquids in amoving-bed biofilm reactor (MBBR).Water Science and Technology 49 (11e12), 53e60.

Ganigue, R., Gabarro, J., Sanchez-Melsio, A., Ruscalleda, M.,Lopez, H., Vila, X., Colprim, J., Balaguer, M.D., 2009. Long-termoperation of a partial nitritation pilot plant treating leachatewith extremely high ammonium concentration prior to ananammoxprocess. BioresourceTechnology100 (23), 5624e5632.

Grady, C.P.L., Daigger, G.T., Lim, H.C., 1999. Biological WastewaterTreatment, second ed. Marcel Deffer, Inc., New York.

Guisasola, A., Jubany, I., Baeza, J.A., Carrera, J., Lafuente, J., 2005.Respirometric estimation of the oxygen affinity constants forbiological ammonium and nitrite oxidation. Journal ofChemical Technology and Biotechnology 80 (4), 388e396.

Hawkins, S.A., Robinson, K.G., Layton, A.C., Sayler, G.S., 2006.A comparison of ribosomal gene and transcript abundanceduring high and low nitrite oxidizing activity using a newlydesigned real-time PCR detection system targeting theNitrobacter spp. 16Se23S intergenic spacer region.Environmental Engineering Science 23 (3), 521e532.

Hawkins, S.A., Robinson, K.G., Layton, A.C., Sayler, G.S., 2010.Limited impact of free ammonia on Nitrobacter spp. inhibitionassessed by chemical and molecular techniques. BioresourceTechnology 101 (12), 4522e4528.

Hawkins, S.A., Robinson, K.G., Layton, A.C., Sayler, G.S., 2008.Response of Nitrobacter spp. ribosomal gene and transcriptabundance following nitrite starvation and exposure tomechanistically distinct inhibitors. Environmental Scienceand Technology 42 (3), 901e907.

Hellinga, C., Schellen, A., Mulder, J.W., van Loosdrecht, M.C.M.,Heijnen, J.J., 1998. The SHARON process: an innovativemethod for nitrogen removal from ammonium-rich wastewater. Water Science and Technology 37 (9), 135e142.

Helsel, D.R., Hirsch, R.M., 2002. Statistical Methods in WaterResources.U.S.GeologicSurvey,U.S.Departmentof the Interior.

Huang, Z., Gedalanga, P.B., Asvapathanagul, P., Olson, B.H., 2010.Influence of physicochemical and operational parameters onNitrobacter and Nitrospira communities in an aerobic activatedsludge bioreactor. Water Research 44 (15), 4351e4358.

Licht, T.R., Tolker-Nielsen, T., Holmstrom, K., Krogfelt, K.A.,Molin, S., 1999. Inhibition of Escherichia coli precursor-16S rRNAprocessing by mouse intestinal contents. EnvironmentalMicrobiology 1 (1), 23e32.

Lu, T., Stroot, P.G., Oerther, D.B., 2009. Reverse transcription of16S rRNA to monitor ribosome-synthesizing bacterialpopulations in the environment. Applied and EnvironmentalMicrobiology 75 (13), 4589e4598.

Muttray, A.F., Mohn, W.W., 1998. RNA/DNA ratio as an indicatorof metabolic activity in resin acid-degrading bacteria. WaterScience and Technology 37 (4e5), 89e93.

Muttray, A.F., Yu, Z.T., Mohn, W.W., 2001. Population dynamicsand metabolic activity of Pseudomonas abietaniphila BKME-9within pulp mill wastewater microbial communities assayedby competitive PCR and RT-PCR. FEMS Microbiology Ecology 38(1), 21e31.

Navarro, E., Simonet, P., Normand, P., Bardin, R., 1992.Characterization of natural-populations of Nitrobacter spp.using PCR/RFLP analysis of the ribosomal intergenic spacer.Archives of Microbiology 157 (2), 107e115.

Oerther, D.B., Pernthaler, J., Schramm, A., Amann, R., Raskin, L.,2000. Monitoring precursor 16S rRNAs of Acinetobacter spp. inactivated sludge wastewater treatment systems. Applied andEnvironmental Microbiology 66 (5), 2154e2165.

Rittmann, B.E., McCarty, P.L., 2003. Environmental Biotechnology:Principles and Applications. McGraw-Hill, New York, NY.

Ruiz, G., Jeison, D., Chamy, R., 2003. Nitrification with high nitriteaccumulation for the treatment of wastewater with highammonia concentration. Water Research 37 (6), 1371e1377.

Simm, R.A., Mavinic, D.S., Ramey, W.D., 2006. A targeted study onpossible free ammonia inhibition of Nitrospira. Journal ofEnvironmental Engineering and Science 5 (5), 365e376.

Stroot, P.G., Oerther, D.B., 2003. Elevated precursor 16S rRNAlevels suggest the presence of growth inhibitors inwastewater. Water Science and Technology 47 (11), 241e250.

Suthersan, S., Ganczarczyk, J.J., 1986. Inhibition of nitriteoxidation during nitrification: some observations. WaterPollution Research Journal of Canada 21, 257e266.

Turk,O.,Mavinic,D.S., 1989.Maintainingnitrite buildup ina systemacclimated to free ammonia.WaterResearch23 (11), 1383e1388.

Van Hulle, S.W.H., Van den Broeck, S., Maertens, J., Villez, K.,Donckels, B.M.R., Schelstraete, G., Volcke, E.I.P.,Vanrolleghem,P.A., 2005.Construction,start-upandoperationofa continuously aerated laboratory-scale SHARON reactor in viewof coupling with an anammox reactor. Water SA 31 (3), 327e334.

Van Hulle, S.W.H., Vandeweyer, H.J.P., Meesschaert, B.D.,Vanrolleghem, P.A., Dejans, P., Dumoulin, A., 2010.Engineering aspects and practical application of autotrophicnitrogen removal from nitrogen rich streams. ChemicalEngineering Journal 162 (1), 1e20.

Yang, S.F., Tay, J.H., Liu, Y., 2004. Respirometric activities ofheterotrophic and nitrifying populations in aerobic granulesdeveloped at different substrate N/COD ratios. CurrentMicrobiology 49 (1), 42e46.