ARTICLE IN PRESS

Available at www.sciencedirect.com

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 6

0043-1354/$ - see frodoi:10.1016/j.watres

�Corresponding auE-mail address:

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

Denitrifying phosphorus removal: Linking the processperformance with the microbial community structure

Gilda Carvalhoa,b, Paulo C. Lemosa, Adrian Oehmena, Maria A.M. Reisa,�

aREQUIMTE/CQFB, Chemistry Department, Faculdade de Ciencias e Tecnologia, Universidade Nova de Lisboa, 2829-516 Caparica, PortugalbIBET/ITQB, Apt.12, 2781-901 Oeiras, Portugal

a r t i c l e i n f o

Article history:

Received 24 January 2007

Received in revised form

26 June 2007

Accepted 29 June 2007

Available online 31 July 2007

Keywords:

Denitrifying P removal

Acetate

Propionate

Accumulibacter

Nitrate-DPAO

Nitrite-DPAO

nt matter & 2007 Elsevie.2007.06.065

thor. Tel./fax: +351 212948amr@dq.fct.unl.pt (M.A.M

a b s t r a c t

This study investigated the link between the process performance of two denitrifying

phosphorus (P) removal systems and their microbial community structure. Two sequencing

batch reactors (SBRs) were operated with either acetate or propionate as the sole carbon

source, and were gradually acclimatised from anaerobic–aerobic to anaerobic–anoxic

conditions. It was found that the propionate SBR was able to sustain denitrifying P removal

after acclimatisation, while the enhanced biological phosphorus removal (EBPR) activity in

the acetate reactor collapsed after the aerobic phase was eliminated. The results suggested

that the anoxic glycogen production rate in the acetate SBR was insufficient to support the

anaerobic glycogen demand for acetate uptake. The chemical transformations in each SBR

suggested that different types of polyphosphate-accumulating organisms (PAOs) were

present in each system, possessing different affinities for nitrate. Microbial characterisa-

tion with fluorescence in situ hybridisation (FISH) revealed that Accumulibacter was the

dominant organism in each reactor, although different cell morphotypes were observed.

A coccus morphotype was predominant in the acetate SBR while the propionate SBR was

enriched in a rod morphotype. It is hypothesised that the coccus morphotype corresponds

to an Accumulibacter strain that is unable to use nitrate as electron acceptor but is able to

use oxygen, and possibly nitrite. The rod morphotype is proposed to be a PAO able to use

nitrate, nitrite and oxygen. This hypothesis is in agreement with literature studies focussed

on the identity of denitrifying PAOs (DPAOs), as well as a recent metagenomic study on

Accumulibacter.

& 2007 Elsevier Ltd. All rights reserved.

1. Introduction

Enhanced biological phosphorus removal (EBPR) is a cost-

effective and environmentally friendly technology to remove

phosphorus from wastewater. In this process, the polypho-

sphate-accumulating organisms (PAOs) are able to store

phosphorus through sequential anaerobic–aerobic condi-

tions. Carbon sources, particularly volatile fatty acids (VFA),

are taken up anaerobically and stored as poly-b-hydroxyalk-

anoates (PHA) through the release of phosphorus (P) and

r Ltd. All rights reserved.

385.. Reis).

degradation of glycogen. A higher amount of phosphorus (P)

is then taken up when an electron acceptor is supplied

(normally oxygen, i.e., aerobic conditions) through PHA

oxidation, which is accompanied by biomass growth and

the regeneration of glycogen. Alternatively, nitrate or nitrite

(NOx) can be used as electron acceptors (i.e. anoxic condi-

tions) instead of oxygen, which is advantageous because both

N and P are removed in the same process (Kerrn Jespersen

and Henze, 1993; Kuba et al., 1993). Moreover, when compared

to conventional EBPR, simultaneous denitrification and P

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 64384

removal can save on aeration, minimise sludge disposal and

reduce the demand for the often-limiting carbon sources

(Kuba et al., 1994, 1996).

The activity of denitrifying PAOs (DPAOs) has often been

demonstrated, both in lab-scale and full-scale EBPR systems

(Kuba et al., 1993; Ahn et al., 2002; Zilles et al., 2002; Shoji

et al., 2003; Kong et al., 2004). However, it is still unclear

whether the same organisms are responsible for P removal

under both aerobic and anoxic conditions. Some authors have

shown results supporting the presence of two different types

of PAOs in denitrifying P removal systems. For example, Kerrn

Jespersen and Henze (1993), Meinhold et al. (1999) and Freitas

et al. (2005) performed anaerobic–anoxic–aerobic tests which

demonstrated that P was initially taken up under anoxic

conditions, but anoxic P uptake eventually ceased despite the

continued presence of nitrate. Once oxygen was introduced to

the system, P uptake then re-started. The proposed explana-

tion for this observation was the co-existence of DPAOs,

capable of using nitrate as electron acceptor in the anoxic

phase until their PHA pools were depleted, and non-DPAOs,

only able to use oxygen for phosphorus removal, which would

be responsible for the subsequent aerobic P uptake.

Molecular techniques gave a new, but still inconclusive,

insight into the diversity of PAOs. Ahn et al. (2002) used

denaturing gradient gel electrophoresis to show a substantial

variation in the population depending on the electron

acceptor employed. Falkentoft et al. (2002) and Lee et al.

(2003) verified using group-specific fluorescence in situ

hybridisation (FISH) probes that the microbial populations

were different when using nitrate or oxygen as electron

acceptors. On the other hand, Dabert et al. (2001) found that

the population enriched under anaerobic–aerobic EBPR con-

ditions was very similar to the population obtained after

shifting to anaerobic–anoxic conditions by addition of nitrate,

as shown by PCR single-strand conformation polymorphism.

Furthermore, Zeng et al. (2003a) showed with FISH and

chemical staining that Candidatus Accumulibacter phosphatis

(hereafter referred to as Accumulibacter; Hesselmann et al.,

1999) were the dominant bacteria both in anaerobic–aerobic

and anaerobic–anoxic systems. Zeng et al. (2003a) found that

the populations enriched under anaerobic–aerobic conditions

were capable of using nitrate after only a few hours of

acclimatisation (the time required to induce denitrifying

enzymes), suggesting that PAOs and DPAOs were the same

microorganisms. Kong et al. (2004) demonstrated by MAR

(microautoradiography)-FISH that Accumulibacter were able to

remove P with oxygen, nitrate or nitrite.

Indeed, there seems to be consensus amongst most

literature studies with respect to the importance of Accumu-

libacter in both aerobic and anoxic EBPR systems (Dabert et al.,

2001; Ahn et al., 2002; Lee et al., 2003; Zeng et al., 2003a). This

does not appear to correlate well with the results mentioned

above that demonstrated the existence of two types of PAOs

through analysis of the chemical transformations (Kerrn

Jespersen and Henze, 1993; Meinhold et al., 1999; Freitas

et al., 2005). Clearly, it is necessary to link the information

gained from chemical and microbiological analyses in order

to solve this apparent disagreement found in literature.

Previous research on the identity of DPAOs has mostly been

conducted using acetate as the carbon source. However, it has

been suggested that other carbon sources, namely propio-

nate, offer a competitive advantage to PAOs in relation to their

main competitors, the glycogen-accumulating microorgan-

isms (GAOs) (Pijuan et al., 2004; Oehmen et al., 2005a). While

GAOs have been found to often compete with PAOs for carbon

sources in anaerobic–aerobic systems (Crocetti et al., 2002;

Levantesi et al., 2002; Oehmen et al., 2005a; Meyer et al., 2006),

denitrifying GAOs (DGAOs) enriched with acetate have been

suggested to be less competitive with DPAOs (Zeng et al.,

2003b). To the best of our knowledge, the enrichment of

DPAOs and DGAOs with carbon sources other than acetate

and the impact of carbon source on denitrifying P removal

performance are not yet well established.

This study aims at investigating the putative existence of

PAOs with different affinities for nitrate and/or oxygen

electron acceptors in EBPR systems. Two sequencing batch

reactors (SBR) were operated with acetate or propionate as the

sole carbon source, and were gradually acclimatised from

anaerobic–aerobic to anaerobic–anoxic conditions. Through

combining microbial techniques, such as FISH, with chemical

analysis of the intracellular and extracellular compounds, the

population dynamics and their influence on process perfor-

mance were assessed in each system.

2. Materials and methods

2.1. Reactors operation

Two 0.9L SBRs were operated for P removal with either acetate or

propionate as the sole carbon source. The SBRs were inoculated

with sludge from a parent EBPR reactor, operated with a

sequence of anaerobic–aerobic conditions, which was fed with

equivalent COD fractions of acetate, propionate and butyrate

(Levantesi et al., 2002). The cycle of each SBR consisted of 1h

settling, 0.5h decanting, 0.5h feed and 6h reaction from which

2h were anaerobic and the remainder consisted of anoxic and/or

aerobic periods. Both the acetate and propionate SBRs were

initially operated for 25 days with anaerobic (2h)–aerobic (4h)

conditions and were then gradually acclimatised to denitrifying P

removal conditions. During acclimatisation, the 4h aerobic phase

was converted to anoxic conditions in a stepwise manner as

shown in Table 1. The anoxic conditions were obtained by

pumping 5mL of a solution of NaNO3 of 5.4, 10.9 or 54.6g/L in

steps 1, 2 or 3, respectively, in order to obtain the N concentra-

tions indicated in Table 1 for each acclimatisation step.

Argon was sparged through the reactors at 1.5 L/min during

the anaerobic and anoxic phases to keep the reactors oxygen-

free. Air was sparged at the same flow rate in the aerobic

phase. The hydraulic retention time and sludge retention

time (SRT) were 1 and 10 d, respectively. The reactors were

operated with pH control set at 8.2, with the automatic

addition of 0.2 M HCl when the pH was above the setpoint.

Reactor control and data acquisition (pH and ORP) were done

with the software LabView, National Instruments, USA.

2.2. Culture medium

The acetate and propionate feeds contained the same carbon

content on a C-mM basis as the parent reactor (26.6C-mM: see

ARTICLE IN PRESS

Table 1 – The acclimatisation to denitrifying P removal conditions in the acetate- and propionate-fed SBRs

Acclimatisationstep

SBR anoxic and/oraerobic phase

Nitrate in the reactor at the beginning ofthe anoxic phase

Acetate SBR PropionateSBR

0 4 h AER – 25 d 25 d

1 1 h ANOX+3 h AER 5 mg N/L Days 1–34 (34 d) Days 1–14

(14 d)

2 2 h ANOX+2 h AER 10 mg N/L Days 35–68 (34 d) Days 15–49

(35 d)

3 4 h ANOX 50 mg N/L Days 69–92 d (non-

EBPR after 15 d)

Days 50–87

(38 d)

The anoxic and/or aerobic phase was preceded by a 2 h anaerobic phase in all steps.

WA T E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 6 4385

Levantesi et al., 2002), which was equivalent to 16.1mgCOD/

mgP in the acetate feed and 18.7 mgCOD/mgP in the propionate

feed. The composition of the feeds was therefore (per litre):

1.81 g of CH3COONa � 3H2O or 0.66 g of CH3CH2COOH; 0.16 g of

NH4Cl; 0.6 g of MgSO4 �7H2O; 0.07g of CaCl2 � 2H2O; 0.1 g of EDTA;

0.09 of K2HPO4; 0.19 g of KH2PO4; 0.01g of allylthiourea (ATU) to

inhibit nitrification; and 1 mL of a micro- and macro-nutrient

solution with the following composition (per litre): 1500 mg

FeCl3 � 6H2O, 150mg H3BO3, 150mg CoCl2 � 6H2O, 120 mg

MnCl2 �4H2O, 120 mg ZnSO4 � 7H2O, 60 mg Na2MoO4 � 2H2O,

30 mg CuSO4 � 5H2O and 30 mg of KI. The pH of the feed was

adjusted to 7.2 with sodium hydroxide pellets.

2.3. Batch tests

Bach tests were performed with approximately 0.2 L of waste

sludge taken from either the acetate or propionate SBR, and

diluted to 0.5 L with mineral medium plus feed. After an initial

2 h anaerobic phase, a spike of nitrate (10–50 mg NO3-N/L)

was supplied in order to monitor the sludge capability of

using nitrate as the sole electron acceptor. When P uptake

ceased, the system was aerated to assess the subsequent

aerobic P removal capacity. The anaerobic/anoxic or aerobic

conditions were maintained by sparging argon or air, respec-

tively, as in the SBRs. These batch tests were carried out with

acetate- and propionate-fed sludges at the end of each

acclimatisation step, as described in Table 1. The initial

consumption and production rates of the compounds in-

volved in P removal activity were calculated by linear

regression of their concentrations in these batch tests.

2.4. Chemical analyses

The reactors were monitored through chemical analytical

techniques. Samples were taken regularly from both reactors

during the batch tests and in the parent reactors throughout

the study. Phosphate, nitrate and nitrite were analysed by

segmented flow analysis (Skalar 5100, Skalar Analytical, The

Netherlands). VFA were measured by high-performance liquid

chromatography using a BioRad Aminex HPX-87H precolumn

and column, and a UV detector set at 210 nm. Sulphuric acid

(0.01 M) was used as eluent at a flow rate of 0.6 mL/min and

50 1C operating temperature. PHA was determined by gas

chromatography according to the methodology described by

Lemos et al. (2006). Glycogen was measured as glucose using

an enzymatic kit (UV method for D-glucose, Roche, Germany),

as described by Levantesi et al. (2002). Mixed liquor sus-

pended solids and volatile suspended solids (VSS) were

determined according to standard methods (APHA, AWWA

and WPCF, 1995).

2.5. Microbial characterisation

Sludge samples were periodically fixed in 4% paraformalde-

hyde. FISH was performed according to Amann (1995) to

assess the evolution of the microbial populations in both

systems. FISH samples were observed using an Olympus BX51

epifluorescence microscope. The oligonucleotide probes em-

ployed for FISH are listed in Table 2. PAO462, PAO651 and

PAO846 were applied together (PAOMIX, for Accumulibacter), as

well as EUB338, EUB338-II and EUB338-III (EUBMIX, for all

Bacteria), GAOQ431, GAOQ989 and GB_G2 (GAOMIX, for

Candidatus Competibacter Phosphatis), TFO_DF218 and

TFO_DF618 (TFO_DFMIX, for Cluster 1 of Defluviicoccus-related

GAOs) and DEF988 and DEF1020 (DEFMIX, for Cluster 2 of

Defluviicoccus-related GAOs). FISH quantification of the PAOs

was done by image analysis with the Image J software at the

beginning and end of the acclimatisation using the PAOMIX

and the EUBMIX probes, following the methodology described

in Bouchez et al. (2000) and Crocetti et al. (2002). Forty images

of each sample were taken using a Zeiss LSM 510 Meta

confocal laser scanning microscope (CLSM). The abundance

of PAOs was determined as the mean image area with a

positive signal for both PAOMIX and EUBMIX relative to the

area with a positive signal for EUBMIX. The standard error of

the mean was calculated as the standard deviation divided by

the square root of the number of images.

Methylene blue (Murray et al., 1994) and Nile blue (Ostle and

Holt, 1982) staining were employed to confirm the cycling of

stored polyphosphate and PHA granules, respectively. Gram

staining (Jenkins et al., 1993) was performed to check for the

presence of Gram-positive bacteria.

3. Results

The EBPR performance of the acetate and propionate SBRs

was evaluated throughout the acclimatisation process to

ARTICLE IN PRESS

Table 2 – Oligonucleotide FISH probes employed in this study

Probe Sequence 50–30 Specificity Reference

EUB338 GCTGCCTCCCGTAGGAGT Most Bacteria Amann et al. (1990)

EUB338-II GCAGCCACCCGTAGGTGT Planctomycetales and other Bacteria not detected by EUB338 Daims et al. (1999)

EUB338-III GCTGCCACCCGTAGGTGT Verrucomicrobiales and other Bacteria not detected by EUB338 Daims et al. (1999)

ALF1b CGTTCGYTCTGAGCCAG Alpha-Proteobacteria Manz et al. (1992)

BET42a GCCTTCCCACTTCGTTT Beta-Proteobacteria Manz et al. (1992)

GAM42a GCCTTCCCACATCGTTT Gamma-Proteobacteria Manz et al. (1992)

PAO462 CCGTCATCTACWCAGGGTATTAAC Most Accumulibacter Crocetti et al. (2000)

PAO651 CCCTCTGCCAAACTCCAG Most Accumulibacter Crocetti et al. (2000)

PAO846 GTTAGCTACGGCACTAAAAGG Most Accumulibacter Crocetti et al. (2000)

GAOQ431 TCCCCGCCTAAAGGGCTT Some Competibacter Crocetti et al. (2002)

GAOQ989 TTCCCCGGATGTCAAGGC Some Competibacter Crocetti et al. (2002)

GB_G2 TTCCCCAGATGTCAAGGC Some Competibacter Kong et al. (2002)

TFO_DF218 GAAGCCTTTGCCCCTCAG ‘Defluviicoccus’-related TFO Wong et al. (2004)

TFO_DF618 GCCTCACTTGTCTAACCG ‘Defluviicoccus’-related TFO Wong et al. (2004)

DF988 GATACGACGCCCATGTCAAGGG ‘Defluviicoccus vanus’-cluster D2 related organisms Meyer et al. (2006)

DF1020 CCGGCCGAACCGACTCCC ‘Defluviicoccus vanus’-cluster D2 related organisms Meyer et al. (2006)

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 64386

nitrate. The microbial populations of each reactor were

characterised through microbial and chemical analyses, in

order to establish a link between the different behaviours

observed in the two SBRs and their respective microbial

communities.

3.1. Denitrifying EBPR performance of the acetate andpropionate SBRs

The anaerobic carbon and phosphorus transformations

throughout the acclimatisation study are summarised in

Table 3, and are compared with metabolic models and

literature studies. During SBR operation under anaero-

bic–aerobic conditions (step 0), before the introduction of

the anoxic phase on day 1, the acetate and the propionate

reactors exhibited good EBPR performance. Both reactors had

499% P removal efficiencies at the beginning of the acclima-

tisation study (data not shown). From Table 3, it can be

observed that the anaerobic stoichiometry during step 0 in

both the acetate and propionate SBRs agreed reasonably well

with the expected activity of an enriched PAO reactor.

Nevertheless, in the acetate reactor the glycogen degradation

per VFA uptake was somewhat higher than expected (0.69 vs

0.50, see Table 3), which may have suggested a small amount

of GAO activity inside this reactor. However, after the anoxic

phase was introduced (see Table 1 for operational details), it is

clear from Table 3 that each reactor responded differently

throughout the acclimatisation steps to anaerobic–anoxic

conditions. There was a sharp decrease in the P/VFA ratio

observed upon the introduction of an anoxic period after the

anaerobic period in the acetate reactor, which was not

observed in the propionate reactor. Overall, the anaerobic

stoichiometry in the propionate reactor was more stable

throughout the acclimatisation study in comparison with the

acetate reactor.

In the last step of acclimatisation, with the elimination of

the aerobic phase, a rapid decrease in the EBPR activity was

observed in the acetate SBR. On day 76, 1 week after the

elimination of the aerobic phase, very low phosphorus release

and uptake (o10 mgP/L) were observed in the acetate SBR and

only about 25% of the acetate was consumed in the anaerobic

phase, indicating that the activity of GAOs was not the reason

for the decreased P removal. Two weeks later, no carbon or

phosphorus transformations were observed in acetate-fed

sludge. The reactor was re-inoculated twice more with waste

sludge from acclimatisation step 2, however, a similar

behaviour was observed each time. These results suggest

that the PAOs enriched with acetate could not be sustained in

fully anaerobic–anoxic conditions and started being

washed out after one SRT. This result is similar to that

obtained by Dabert et al. (2001), who reported a breakdown in

P uptake after nitrate replaced oxygen as the electron

acceptor in an acetate-fed EBPR reactor. On the contrary, the

propionate SBR achieved stable denitrifying P removal

performance and maintained it for over three SRT. To the

best of our knowledge, this is the first study showing

sustained EBPR activity with a propionate carbon source

under anaerobic–anoxic conditions.

The evolution of the P uptake capacity in anoxic and

aerobic conditions during the acclimatisation study was

monitored through batch tests, which were performed at

the end of each step of the acclimatisation study (steps 0–3,

Table 1). Typical profiles obtained in these batch tests are

shown in Figs. 1 (step 1) and 2 (step 2). In the batch tests

performed during step 2, several pulses of nitrate were added

in the anoxic phase, to ensure that there was no limitation of

electron acceptor. Initial rates of anoxic biotransformations

from the batch tests are presented in Table 4.

From Table 4, the anoxic P uptake rate in step 0 was very

low in the acetate sludge as compared to the propionate

sludge as well as literature studies carried out with acetate,

suggesting that the dominant PAOs in this reactor were not

readily able to use nitrate as electron acceptor. In the

propionate SBR, the anoxic P uptake rate was initially higher

than in the acetate SBR, but remained relatively steady until

the end of acclimatisation step 2. It should be noted that, in

ARTIC

LEIN

PRES

S

Table 3 – Anaerobic carbon and phosphorus transformations for the acetate and propionate SBRs throughout the acclimatisation study: comparison with metabolicmodels and literature studies

Study Gly/VFA

(C-mol/C-mol)

PHA/VFA

(C-mol/C-mol)

PHB

(%)

PHV

(%)

PH2MV

(%)

Prel/VFAup

(P-mol/C-mol)

Anaerobic

pH

Comments

Acetate SBR

Step 0 (this study) 0.69 1.10 87 13 0 0.52 7.0–8.2

Step 1 (this study) 0.67 1.12 68 32 0 0.16 7.0–8.2

Step 2 (this study) 0.50 1.44 67 33 0 0.16 7.0–8.2

Smolders et al. (1994) 0.50 1.33 100 0 0 0.48–0.71a 7.0–8.2 PAO metabolic model

Zeng et al. (2002) 1.12 1.86 73 25 2 0.00 7.0 GAO metabolic model

Smolders et al. (1994) – 1.22 90 10 – 0.50 7.070.05

Liu et al. (1997) 0.78 1.47 79 18 o3 0.45 7.0–8.0 Likely contained PAOs and GAOs

Filipe et al. (2001) 0.53 1.30 88 12 – 0.57 7.2–7.6 Glycogen and PHA did not

change with increasing pH

Filipe et al. (2001) 0.53 1.30 89 11 – 0.73 8.070.1

Zeng et al. (2003a) 0.64 1.48 91 9 – 0.35 7.070.1 DPAO reactor

Lu et al. (2006) 0.46 1.26 94 6 0 0.62 7.0–8.0 490% Accumulibacter

Propionate SBR

Step 0 (this study) 0.32 0.97 0 41 59 0.40 7.0–8.2

Step 1 (this study) 0.26 0.96 0 37 63 0.37 7.0–8.2

Step 2 (this study) 0.28 0.94 0 39 61 0.29 7.0–8.2

Step 3 (this study) 0.23 0.96 0 33 67 0.32 7.0–8.2

Oehmen et al. (2005c) 0.33 1.22 0 46 55 0.42 7.0 PAO metabolic model

Oehmen et al. (2006) 0.67 1.50 7 39 55 0.00 7.0 GAO metabolic model

Satoh et al. (1992) 0.37 1.20 o3% 50 48 0.33 7.0–8.5

Pijuan et al. (2004) 0.21 0.91 37 63 – 0.42 7.2–7.6

Oehmen et al. (2005b) 0.32 1.23 3 45 53 0.42 7.070.1 Glycogen and PHA decreased

with increasing pH

Oehmen et al. (2005c) 0.25 1.10 5 41 55 0.41 7.570.05

Oehmen et al. (2005c) 0.12 0.95 4 36 61 0.51 8.070.05

Lu et al. (2006) 0.29 1.22 0 46 54 0.44 7.0–8.0 490% Accumulibacter

a P/VFA range calculated from Eq. (8) of Smolders et al. (1994), based on the operational pH range of this study.

WA

TE

RR

ES

EA

RC

H4

1(2

00

7)

43

83

–4

39

64

38

7

ARTICLE IN PRESS

Fig. 1 – Batch kinetic study ((K) acetate or propionate; (’) PHA; ( ) glycogen; (B) phosphorus; (m) nitrate) performed with

(a) acetate and (b) propionate sludge during step 1 of acclimatisation to nitrate (days 31 and 8, respectively).

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 64388

acclimatisation steps 1 and 2, the 5 and 10 mg N/L of nitrate,

respectively, fed to the propionate SBR were rapidly con-

sumed, leaving the sludge in nitrate starvation conditions for

most of the anoxic phase (results not shown), which probably

limited the growth of DPAOs. Between steps 1 and 2, the

anoxic P uptake rate remained relatively constant in the

propionate reactor. This suggests that the increase observed

in the nitrate uptake rate in this period may have been due to

the growth of DGAOs or other heterotrophs able to denitrify

using an internally stored carbon source (Third et al., 2003),

since no external carbon was present at the beginning of the

anoxic phase. In the last step of acclimatisation (nitrate

ARTICLE IN PRESS

Fig. 2 – Batch kinetic study ((K) acetate or propionate; (’) PHA; ( ) glycogen; (B) phosphorus; (m) nitrate; (n) nitrite)

performed with (a) acetate and (b) propionate sludge during step 2 of acclimatisation to nitrate (days 65 and 44, respectively).

WA T E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 6 4389

addition in the SBRs increased to 50 mg N/L), the anoxic P

uptake rate in the propionate SBR increased relative to the

previous period, in parallel with the increase of the nitrate

uptake rate (Table 4). This suggests that, in non-nitrate

limiting conditions, an increase in the DPAO population took

place in the system.

Interestingly, the anoxic glycogen production rate was

approximately 0 in the first batch test with an anoxic phase,

both for the acetate and the propionate sludge. However, this

rate increased more rapidly in the propionate sludge as

compared to the acetate sludge. The slow anoxic glycogen

production in the acetate SBR may have been a reason for its

eventual collapse in EBPR performance. This will be discussed

in more detail later.

3.2. Microbial population analysis

The different denitrifying EBPR performance observed in the

acetate and propionate SBRs could be due to variations in the

metabolic behaviour of the same population when fed with

ARTICLE IN PRESS

Table 4 – Anoxic chemical transformation rates determined from batch tests performed with acetate and propionatesludge throughout the acclimatisation study (R2 and number of points used for regression are given in brackets)

Acclim. step Day N-NO3 uptake

(mmol/gVSS h)

Anoxic P uptake

(mmol/gVSS h)

Anoxic Gly prod.

(C-mmol/gVSS h)

Anoxic PHA deg.

(C-mmol/gVSS h)

Acetate SBR

0 1 0.10 (0.984; 8) 0.02 (0.991; 7) �0 – �0 –

1 31 0.10 (0.978; 7) 0.06 (0.985; 5) �0 – 0.30 (0.926; 6)

2 65 0.45 (0.999; 8) 0.27 (0.975; 5) 0.24 (0.913; 4) 0.90 (0.997; 5)

Propionate SBR

0 1 0.08 (0.993;

10)

0.10 (0.979;

10)

�0 – 0.24 (0.977; 4)

1 8 0.12 (0.993; 5) 0.10 (0.987; 7) 0.29 (0.984; 4) 0.19 (0.932; 5)

2 44 0.24 (0.998; 9) 0.13 (0.987; 6) 0.42 (0.995; 4) 0.60 (0.957; 5)

3 80 0.77 (0.981; 3) 0.63 (0.935; 5) 0.70 (0.970; 4) 0.97 (0.993; 6)

Literature findings (fed with acetate)

Kuba et al. (1993) X1.15 X0.92 – –

Zeng et al. (2003a) 1.08 0.58 0.9 2.0

Saito et al. (2004) 0.59 0.33 – –

Table 5 – Quantification of PAOs and GAOs in the acetate and propionate SBR before and after acclimatisation to nitrate

C source PAOs (Accumulibacter) GAOs

Beforeacclimatisation

End ofacclimatisationa

Before acclimatisation End of acclimatisationa

Acetate 6472%, mostly

coccus

morphology

3772%, mostly

coccus

morphology

D. vanus Cluster 2 most abundant GAOs

(5%). Competibacter also present (1%)

Competibacter and D. vanus Cluster

2 most abundant GAOs (5%)

Propionate 8972%. Cocci and

rods

7672%, mostly of

rod morphology

D. vanus Cluster 2 most abundant GAOs

(5%), Cluster 1 also present (1%)

Only few GAOs detected, mostly

D. vanus Cluster 2 (1%)

a End of step 2 (day 65) for the acetate SBR and end of step 3 (day 80) for the propionate SBR.

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 64390

different carbon sources. Alternatively, it could be explained

by the development of different microbial populations in each

reactor. In order to investigate this possibility, the microbial

communities enriched in each SBR were characterised by

FISH and chemical staining.

FISH analysis showed that Accumulibacter, bound with the

FISH probe PAOMIX, were dominant throughout the whole

acclimatisation process in both SBRs. As shown in Table 5,

this dominance was much higher in the propionate than in

the acetate SBR. After acclimatisation, there was only a slight

decrease in the number of Accumulibacter in the propionate

SBR, while a substantial decrease in these bacteria (of

approximately 25–30%) was observed in the acetate SBR

(Fig. 3). Another group of bacteria belonging to the Actinobac-

teria that have been found in some EBPR systems and are

believed to be PAOs (Lee et al., 2003; Kong et al., 2005), was not

likely to be present in these SBRs, since no Gram-positive cells

were detected. A large fraction of the organisms in the acetate

SBR that did not bind the PAOMIX probe were large cocci

belonging to the g-proteobacteria. The FISH probes presently

available for GAOs did not reveal high numbers of either

Competibacter- or Defluviicoccus-related bacteria in any of the

sludge samples analysed.

Fig. 4A and B shows a more magnified picture (using a 100�

objective instead of a 40� , as used for quantification in Fig. 3)

of the PAOMIX-targeted bacteria in the SBRs. From these

images, it is clear that two different morphotypes were bound

with these FISH probes (Fig. 4). Some Accumulibacter were

cocci or cocci–bacilli, with one or two polyphosphate gran-

ules, and tended to be arranged in spherical clusters (Fig. 4A,

C and E). Other Accumulibacter were rods and often displayed

three–four polyphosphate granules. The rods did not aggre-

gate in geometrically shaped clusters and seemed to be

more loosely bound (Fig. 4B, D and F). To the best of our

knowledge, this is the first report of different morphotypes of

Accumulibacter.

The cells shown in Figs. 3 and 4 were fixed at the end of the

SBR cycle and the dark granules observed inside the cells in

the FISH micrographs (Fig. 4A and B) are likely polyphosphate

granules for both rod and coccus morphologies, as can be

deduced by comparison with images stained with methylene

blue (Fig. 4E and F).

These two Accumulibacter morphologies were not evenly

distributed in the two reactors and varied during the

acclimatisation process (Table 5). At the beginning of the

study, the two morphotypes seemed to be present in both

ARTICLE IN PRESS

Fig. 3 – CLSM micrographs of sludge hybridised with Cy3-labelled PAOMIX probes and Cy5-labelled EUBMIX probes.

Accumulibacter are shown in magenta and all other bacteria are shown in blue. (A and B) Images of the acetate-fed sludge

fixed on day 1 and day 52, respectively. (C and D) Obtained with propionate-fed sludge fixed on days 1 and 80, respectively.

Bar ¼ 10 lm.

WA T E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 6 4391

reactors, but the acetate SBR had a higher content of cocci,

while the propionate SBR was abundant in both types. With

acclimatisation, the propionate reactor became more en-

riched in the rod morphotype, whereas the acetate SBR still

had a higher proportion of coccus morphotype, although the

overall number of Accumulibacter had decreased substantially

in this reactor.

4. Discussion

4.1. Population analysis: linking the microbialcharacterisation with chemical transformations

The populations of the acetate and propionate systems in this

study were characterised through chemical analysis of their

anoxic and aerobic transformations, as well as through

microbial population analysis. These two complementary

types of data were combined in this study to clarify the

putative existence of different PAOs in these systems, and to

investigate their potential to denitrify.

It can be observed from Fig. 1 that, after the anoxic P

removal activity had ceased (with nitrate still available), the

sludge was able to continue taking up P if oxygen was

supplied. These results are in agreement with previous

studies (Kerrn Jespersen and Henze, 1993; Meinhold et al.,

1999; Freitas et al., 2005) which suggested the existence of two

different types of PAOs: those that are able to use either

nitrate or oxygen as electron acceptor (DPAOs) and those that

cannot use nitrate but can use oxygen. Possibly, the DPAOs

that took up phosphorus in the anoxic phase became limited

in internal carbon, since their PHA pools would be depleted

first. However, the other PAOs in the system, which were not

able to remove P with nitrate, would still have stored carbon

available when oxygen was supplied, thereby being respon-

sible for the observed aerobic P uptake. The PHA profile was

consistent with the P uptake profiles, both anoxically and

aerobically, supporting the dual-PAO theory.

In contrast to these results, analysis by FISH suggested that

the primary PAO present in both SBRs throughout the

acclimatisation study was Accumulibacter. However, two dis-

tinct morphologies of Accumulibacter were observed: cocci and

rods. The relative abundance of these two morphotypes

correlated well with the successful denitrifying EBPR perfor-

mance observed in the propionate SBR (mostly rods after

acclimatisation) and the collapse of the acetate SBR (mostly

ARTICLE IN PRESS

Fig. 4 – (A–D) Epifluorescence micrographs of sludge hybridised with Cy3-labelled PAOMIX probes and FITC-labelled BET42a

probe. (A and B) Accumulibacter are shown in yellow and any other b-Proteobacteria are shown in green. (C and D)

Accumulibacter are shown in red. (A and C) acetate-fed sludge, fixed on day 52, displaying the cocci morphology, dominant in

this reactor. (B and D) propionate-fed sludge, fixed on day 80, displaying the rod morphology, enriched in this reactor. (E–F)

Methylene blue staining of acetate-fed sludge (E) and propionate-fed sludge (F) sampled at the end of the SBR cycle, displaying

the poly-P granules. Bar ¼ 10 lm.

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 64392

cocci throughout the study). This suggests that the two

morphotypes could correspond to two different strains that

possess different denitrifying EBPR performances, and per-

haps display different affinities for the two carbon sources.

The existence of different strains of Accumulibacter has been

previously suggested in literature, although the role of each

strain in the EBPR process is still to be explored in greater

depth. A metagenomic study was carried out by Martin et al.

(2006), who sequenced a strain of Accumulibacter that was

dominant in two lab-scale EBPR systems, one fed with acetate

and another with propionate. However, Martin et al. (2006)

also reported the presence of other Accumulibacter strains in

these systems with up to 15% divergent genomes, which

could reflect variations in the phenotype. The distribution of

different Accumulibacter types was studied by He et al. (2006)

in a lab-scale reactor and six full-scale plants displaying

different EBPR performances. This study demonstrated the

existence of various clades within the Accumulibacter group,

based on phylogenetic analysis of the 16S rRNA gene plus

internally transcribed spacer region clone libraries, although

a correlation with their distribution in the WWTP could not be

well established due to the low number of sequences

analysed. Although preliminary, these studies open up the

hypothesis that different types (or strains) of Accumulibacter

possess different metabolic behaviours, which is supported

by the results obtained in this work.

Based on the EBPR activity observed in the batch tests, other

differences could be observed between the acetate and

propionate sludges that suggest the presence of two different

Accumulibacter strains.

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 6 4393

Fig. 2 shows that, in step 2 of the acclimatisation process,

nitrite accumulated in the acetate reactor but not in the

propionate reactor. This temporary nitrite accumulation was

detected in the acetate SBR from day 53 onwards, and it was

also observed by Zeng et al. (2003a) and Saito et al. (2004) in

acetate-fed denitrifying EBPR systems. Nitrite accumulation

demonstrates that the reduction of nitrate to nitrite occurs

faster than the subsequent reduction of nitrite to nitrogen

gas, and it is often observed in denitrifying systems (Drysdale

et al., 2001). This uncoupling of nitrate and nitrite reduction

could be due to differences in the kinetics of each metabolic

step within the same microorganism (Almeida et al., 1995).

This hypothesis is difficult to prove in studies performed with

mixed cultures. A second hypothesis is based on the co-

existence of bacteria that only possess the enzymes to reduce

nitrate to nitrite, while others only perform nitrite reduction

(Drysdale et al., 2001). Differences in the kinetics and/or the

quantity of nitrate and nitrite reducers could result in nitrite

accumulation. Indeed, the near-complete genome of Accumu-

libacter obtained by Martin et al. (2006) revealed that their

dominant Accumulibacter strain lacked the gene nar that

encodes for the nitrate reductase enzyme, although the genes

that encode for the rest of the denitrification pathway were

detected. Based on these findings, it was suggested that this

type of Accumulibacter is not capable of using nitrate as

electron acceptor, but it can use nitrite, which is likely

produced by flanking species of the EBPR community.

In this study, the PAOs present in the acetate SBR may have

belonged to an Accumulibacter strain that is unable to use

nitrate, but able to use nitrite (nitrite-DPAOs). This hypothesis

could explain the nitrite accumulation observed in this

system. Nitrate reduction would be carried out by other

populations present in the community, with the uncoupling

of the nitrate and nitrite reduction rates resulting in nitrite

accumulation. While this strain of Accumulibacter would likely

have been present at the beginning of the acclimatisation

study as well, they would be less able to take up P anoxically

until other nitrate reducers were enriched in the system,

producing the necessary nitrite (which did not accumulate to

detectable levels in the acetate SBR until day 53). In this case,

these bacteria would act like non-DPAOs (able to use only

oxygen) until nitrite was present in the reactor. This could

explain the very low anoxic P uptake rate of the acetate

sludge in batch tests carried out at the beginning of the

acclimatisation study (Table 4) and the subsequent increase

measured in step 2 (day 65), despite the reduction in the

overall PAO population (Table 5). Once the nitrite produced in

the acetate SBR was sufficient to enable nitrite-DPAOs to take

up P anoxically, they would use up their stored PHA in the

anoxic phase, which would explain why there was no further

P uptake in Fig. 2a when oxygen was introduced in the

system, despite the additional phosphorus added in this

batch test at 12.6 h to ensure that P was not limiting. Thus,

one hypothesis is that at least part of the group of organisms

known as non-DPAOs may in fact be capable of using nitrite

as an electron acceptor (i.e. nitrite-DPAOs).

If the same strain of Accumulibacter is assumed to be the

dominant PAO in the propionate SBR, the absence of nitrite

accumulation in this reactor would mean that nitrate was

reduced by flanking populations at the same rate as nitrite

was reduced by nitrite-DPAOs. However, that would imply the

presence of a high number of nitrate reducers at the

beginning of the acclimatisation process in this reactor, since

P was taken up anoxically in the propionate SBR from day 1 at

a rate 5 times higher than the acetate SBR (Table 4). This is

unlikely due to the absence of nitrate before day 1 in both

reactors (ATU was added to prevent nitrification). This

difference in initial anoxic P uptake rate also cannot be

justified by the differences observed in the size of the PAO

population (Table 5), since the number of Accumulibacter in the

propionate SBR were only 1.4 times higher than in the acetate

SBR. Alternatively, it can be hypothesised that a different

strain of Accumulibacter was present in the propionate SBR,

which possessed the metabolic capacity of using nitrate as

the electron acceptor (i.e. nitrate-DPAOs). The enrichment of

these bacteria in this reactor agrees well with the stable

denitrifying EBPR performance observed, and the fact that

nitrite was never detected. This hypothesis is in agreement

with the dual-PAO theory suggested by previous studies in

literature, where non-DPAOs (unable to use nitrate) are here

proposed to be able to use nitrite (nitrite-DPAOs).

The combination of chemical and microbial data suggests

that the PAOs with a rod morphotype, which were more

highly enriched with propionate, correspond to nitrate-

DPAOs. The PAOs with a coccus morphotype, dominant in

the acetate SBR, could correspond to nitrite-DPAOs. Further

investigation is necessary to confirm if these two Accumuli-

bacter morphologies correspond with different Accumulibacter

strains that possess different denitrifying capabilities.

4.2. The impact of carbon source on denitrifying EBPRperformance

The population enriched in the acetate SBR was not able to

maintain good denitrifying EBPR performance, contrary to the

propionate SBR in this study. This difference in performance

may be linked to the glycogen cycling in these systems.

According to the metabolic models developed for PAOs,

acetate requires more glycogen than propionate to produce

sufficient reducing power for anaerobic VFA storage as PHA.

The theoretical stoichiometry of glycogen consumption per

VFA uptake is, on a C-mol basis, 0.50 for acetate (Smolders

et al., 1994) and 0.33 for propionate (Oehmen et al., 2005b).

From the experimental data in Table 3, it was also observed in

this study that the anaerobic glycogen consumption per unit

of VFA in the acetate SBR was almost double that of the

propionate SBR.

Table 4 shows that the acetate SBR had a low anoxic

glycogen production rate when compared with the propionate

SBR throughout the acclimatisation study, or when compared

with the results of Zeng et al. (2003a). According to the

glycogen production rates given in Table 4, the anoxic phase

would have to be longer than 16 h in the acetate SBR in order

to compensate for the glycogen demand in the 2 h anaerobic

phase (4.0 C-mmol/gVSS). In the propionate SBR, however,

enough glycogen would be produced in less than 3 h in step 3,

when 1.5 C-mmol/gVSS of glycogen were consumed anaero-

bically. Figs. 1 and 2 show that the glycogen production rate

was much higher aerobically than anoxically in the acetate

SBR (respectively, 0.74 C-mmol/gVSS h vs approx. 0 for step

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 64394

1—Fig. 1—and 1.26 vs 0.24 C-mmol/gVSS h for step 2—Fig. 2).

This illustrates the necessity of an aerobic phase in this

reactor for maintaining sufficient glycogen storage within

PAO cells. Indeed, in step 3 of the acetate SBR, glycogen was

not observed to be anaerobically consumed or anoxically

produced, and the level was only 12% of the glycogen content

observed during anaerobic–aerobic conditions. This suggests

that the acetate SBR was limited by glycogen in step 3, after

the aerobic period was eliminated.

Another possible reason that the acetate reactor could not

sustain EBPR activity could be due to nitrite inhibition.

Indeed, nitrite has been shown to be potentially inhibitory

to PAOs (Saito et al., 2004). Saito et al. (2004) found that nitrite

levels above 2 mg NO2-N/gVSS inhibited anoxic P uptake.

While the maximum nitrite accumulation observed during

normal SBR operation (excluding batch tests operated with an

extended anoxic phase) was 1.2 mg NO2-N/gVSS, nitrite could

have had a negative effect on the ability of DPAOs to produce

glycogen anoxically.

As shown in Table 3, the anaerobic P/VFA ratio decreased

sharply as soon as an anoxic phase was included in the

acetate SBR cycle. This implies that an alternative energy

source was used to achieve VFA uptake and conversion to

PHA, such as glycogen. However, the glycogen degradation

per unit of VFA did not increase. Notably, there was a

corresponding increase in the fraction of PHV produced in

the acetate reactor, further suggesting a change in the

anaerobic metabolism of the microbial community. It is

highly likely that this change reflected a lower activity of

PAOs, which is supported by the reduction in Accumulibacter

(Table 5), and may also have contributed to the collapse in

EBPR performance. The reason why this change in stoichio-

metry occurred in the acetate SBR is still unclear, but reflects

the instability of the reactor. Indeed, the operation of the

acetate SBR under anaerobic–anoxic conditions was at-

tempted 3 times (the SBR was re-inoculated using waste

sludge from acclimatisation step 2), and each time the reactor

could not sustain denitrifying EBPR activity.

The low abundance of known GAOs in the acetate SBR

suggests that the collapse of this reactor under anaerobic–

anoxic conditions was not due to competition by GAOs for

anaerobic acetate uptake. This is further supported by the

very low anaerobic/anoxic glycogen transformations ob-

served in the SBR. Nevertheless, acetate was completely

removed in the anaerobic phase of the acetate-SBR until

acclimatisation step 3. The aforementioned decrease in the

P/VFA ratio in the acetate SBR during steps 1 and 2 could

possibly have been due to a group of non-GAOs competing

with PAOs for anaerobic acetate uptake.

The high abundance of Accumulibacter in the propionate-fed

reactor throughout the study correlates well with the good

EBPR performance observed in this reactor. The results from

this study suggest that a propionate carbon source led to a

higher level of DPAO activity than acetate, which is consistent

with some previous studies with PAOs operated under

anaerobic–aerobic conditions (Pijuan et al., 2004; Oehmen

et al., 2005a). Nevertheless, the acetate reactor from this study

exhibited inferior performance as compared to previous

studies where DPAO reactors were operated (see Table 4),

which also used acetate as the sole carbon source (Kuba et al.,

1993; Zeng et al., 2003a). Thus, the result that acetate was

inferior to propionate for achieving denitrifying P removal

should not be interpreted as a general rule. The reason for the

low anoxic P removal in the acetate SBR is likely to be linked

with the microbial community present in the reactor, which

differs from the propionate reactor as discussed above, and

possibly from previous DPAO studies as well. The DPAO

reactors operated previously with acetate (Kuba et al., 1993;

Zeng et al., 2003a; Saito et al., 2004) likely possessed

differences in their microbial communities, as reflected by

the varying anoxic kinetics (see Table 4). Indeed, Saito et al.

(2004) observed both PAOs and GAOs inside their denitrifying

EBPR sludge, contrary to this study.

It would be of high interest to develop FISH probes that

could distinguish between PAOs displaying different affinities

for nitrate, thus enabling greater characterisation of the

microbial populations involved in lab- and full-scale EBPR

systems. These probes could also be employed in studies

focussed on determining the operational parameters that

select for an appropriate population to achieve superior

denitrifying P removal.

5. Conclusions

Chemical analysis of anaerobic–anoxic–aerobic batch tests

suggested the existence of different types of PAOs, each

displaying different affinities for nitrate. Through microbial

analysis, a high abundance of Accumulibacter was observed in

both acetate- and propionate-fed reactors, although two

different cell morphotypes were detected: rods and cocci. It

was hypothesised that a rod morphotype is linked with PAOs

that are able to directly use nitrate as electron acceptor, while

a coccus morphotype corresponds with PAOs that might be

able to use nitrite, but not nitrate. While this hypothesis has

yet to be confirmed, it may explain the seemingly contrasting

results that have been observed in previous studies with

respect to the identity of DPAOs, and whether or not they are

different organisms than non-DPAOs. In this study, the SBR

fed with acetate as the sole carbon source did not sustain

denitrifying EBPR activity, unlike the propionate SBR. This is

likely reflective of different microbial populations enriched

with each carbon source, and their corresponding metabolic

activity. It is possible that the Accumulibacter group includes

multiple strains with different phenotypic characteristics,

such as the ability to denitrify. Consequently, the type of

Accumulibacter present in a system, and not only the presence

of Accumulibacter, might be a key factor for achieving good P

and N removal.

Acknowledgements

Gilda Carvalho and Adrian Oehmen acknowledge the

Fundac- ao para a Ciencia e Tecnologia (FCT) for the Post-

Doctoral fellowships SFRH/BPD/30800/2006 and SFRH/BPD/

20862/2004. This work was funded by the FCT Project POCI/

AMB/56075/2004.

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 6 4395

R E F E R E N C E S

Ahn, J., Daidou, T., Tsuneda, S., Hirata, A., 2002. Characterizationof denitrifying phosphate-accumulating organisms cultivatedunder different electron acceptor conditions using polymerasechain reaction-denaturing gradient gel electrophoresis assay.Water Res. 36 (2), 403–412.

Almeida, J.S., Reis, M.A.M., Carrondo, M.J.T., 1995. Competitionbetween nitrate and nitrite reduction in denitrification byPseudomonas fluorescens. Biotechnol. Bioeng. 46 (5), 476–484.

Amann, R.I., 1995. In situ identification of microorganisms bywhole cell hybridization with rRNA-targeted nucleic acidprobes. In: Akkermans, A.D.L., van Elsas, J.D., de Bruijn, F.J.(Eds.), Molecular Microbial Ecology Manual. Kluwer AcademicPublications, Dordrecht, Holland, pp. 1–15.

Amann, R.I., Binder, B.J., Olson, R.J., Chisholm, S.W., Devereux, R.,Stahl, D.A., 1990. Combination of 16S ribosomal-RNA-targetedoligonucleotide probes with flow-cytometry for analyzingmixed microbial-populations. Appl. Environ. Microbiol. 56 (6),1919–1925.

APHA, AWWA, WPCF, 1995. Standard methods for the examina-tion of water and wastewater. In: Eaton, A.D., Clesceri, L.S.,Greenberg, A.E. (Eds.), Standard Methods for the Examinationof Water and Wastewater. Port City Press, Baltimore.

Bouchez, T., Patureau, D., Dabert, P., Juretschko, S., Dore, J.,Delgenes, P., Moletta, R., Wagner, M., 2000. Ecological study ofa bioaugmentation failure. Environ. Microbiol. 2 (2), 179–190.

Crocetti, G.R., Hugenholtz, P., Bond, P.L., Schuler, A., Keller, J.,Jenkins, D., Blackall, L.L., 2000. Identification of polypho-sphate-accumulating organisms and design of 16S rRNA-directed probes for their detection and quantitation. Appl.Environ. Microbiol. 66 (3), 1175–1182.

Crocetti, G.R., Banfield, J.F., Keller, J., Bond, P.L., Blackall, L.L., 2002.Glycogen-accumulating organisms in laboratory-scale andfull-scale wastewater treatment processes. Microbiolo-gy—Sgm 148, 3353–3364.

Dabert, P., Sialve, B., Delgenes, J.P., Moletta, R., Godon, J.J., 2001.Characterisation of the microbial 16S rDNA diversity of anaerobic phosphorus-removal ecosystem and monitoring of itstransition to nitrate respiration. Appl. Microbiol. Biotechnol.55 (4), 500–509.

Daims, H., Bruhl, A., Amann, R., Schleifer, K.H., Wagner, M., 1999.The domain-specific probe EUB338 is insufficient for thedetection of all Bacteria: development and evaluation of a morecomprehensive probe set. Syst. Appl. Microbiol. 22 (3), 434–444.

Drysdale, G.D., Kasan, H.C., Bux, F., 2001. Assessment of deni-trification by the ordinary heterotrophic organisms in anNDBEPR activated sludge system. Water Sci. Technol. 43 (1),147–154.

Falkentoft, C.M., Muller, E., Arnz, P., Harremoes, P., Mosbaek, H.,Wilderer, P.A., Wuertz, S., 2002. Population changes in abiofilm reactor for phosphorus removal as evidenced by theuse of FISH. Water Res. 36 (2), 491–500.

Filipe, C.D.M., Daigger, G.T., Grady, C.P.L., 2001. Stoichiometry andkinetics of acetate uptake under anaerobic conditions by anenriched culture of phosphorus-accumulating organisms atdifferent pHs. Biotechnol. Bioeng. 76 (1), 32–43.

Freitas, F., Temudo, M., Reis, M.A.M., 2005. Microbial populationresponse to changes of the operating conditions in a dynamicnutrient-removal sequencing batch reactor. Bioprocess Bio-systems Eng. 28 (3), 199–209.

He, S., Gu, A.Z., McMahon, K.D., 2006. Fine-scale differencesbetween Accumulibacter-like bacteria in enhanced biologicalphosphorus removal activated sludge. Water Sci. Technol. 54(1), 111–117.

Hesselmann, R.P.X., Werlen, C., Hahn, D., van der Meer, J.R.,Zehnder, A.J.B., 1999. Enrichment, phylogenetic analysis and

detection of a bacterium that performs enhanced biologicalphosphate removal in activated sludge. Syst. Appl. Microbiol.22 (3), 454–465.

Jenkins, D., Richard, M.G., Daigger, G.T., 1993. Manual on theCauses and Control of Activated Sludge Bulking and Foaming.Lewis Publishers, New York.

Kerrn Jespersen, J.P., Henze, M., 1993. Biological phosphorusuptake under anoxic and aerobic conditions. Water Res. 27(4), 617–624.

Kong, Y.H., Ong, S.L., Ng, W.J., Liu, W.T., 2002. Diversity anddistribution of a deeply branched novel proteobacterial groupfound in anaerobic–aerobic activated sludge processes. Envir-on. Microbiol. 4 (11), 753–757.

Kong, Y.H., Nielsen, J.L., Nielsen, P.H., 2004. Microautoradiographicstudy of Rhodocyclus-related polyphosphate-accumulatingbacteria in full-scale enhanced biological phosphorus removalplants. Appl. Environ. Microbiol. 70 (9), 5383–5390.

Kong, Y.H., Nielsen, J.L., Nielsen, P.H., 2005. Identity and ecophy-siology of uncultured actinobacterial polyphosphate-accu-mulating organisms in full-scale enhanced biologicalphosphorus removal plants. Appl. Environ. Microbiol. 71 (7),4076–4085.

Kuba, T., Smolders, G., Vanloosdrecht, M.C.M., Heijnen, J.J., 1993.Biological phosphorus removal from waste-water by anaero-bic–anoxic sequencing batch reactor. Water Sci. Technol. 27(5-6), 241–252.

Kuba, T., Wachtmeister, A., Vanloosdrecht, M.C.M., Heijnen, J.J.,1994. Effect of nitrate on phosphorus release in biologicalphosphorus removal systems. Water Sci. Technol. 30 (6),263–269.

Kuba, T., VanLoosdrecht, M.C.M., Heijnen, J.J., 1996. Phosphorusand nitrogen removal with minimal cod requirement byintegration of denitrifying dephosphatation and nitrificationin a two-sludge system. Water Res. 30 (7), 1702–1710.

Lee, N., Nielsen, P.H., Aspegren, H., Henze, M., Schleifer, K.H.,Jansen, J.L., 2003. Long-term population dynamics and in situphysiology in activated sludge systems with enhanced biolo-gical phosphorus removal operated with and without nitrogenremoval. Syst. Appl. Microbiol. 26 (2), 211–227.

Lemos, P.C., Serafim, L.S., Reis, M.A.M., 2006. Synthesis ofpolyhydroxyalkanoates from different short-chain fatty acidsby mixed cultures submitted to aerobic dynamic feeding.J. Biotechnol. 122 (2), 226–238.

Levantesi, C., Serafim, L.S., Crocetti, G.R., Lemos, P.C., Rossetti, S.,Blackall, L.L., Reis, M.A.M., Tandoi, V., 2002. Analysis of themicrobial community structure and function of a laboratoryscale enhanced biological phosphorus removal reactor. Envir-on. Microbiol. 4 (10), 559–569.

Liu, W.T., Nakamura, K., Matsuo, T., Mino, T., 1997. Internalenergy-based competition between polyphosphate- and gly-cogen-accumulating bacteria in biological phosphorus re-moval reactors—effect of P/C feeding ratio. Water Res. 31 (6),1430–1438.

Lu, H., Oehmen, A., Virdis, B., Keller, J., Yuan, Z.G., 2006. Obtaininghighly enriched cultures of Candidatus Accumulibacter phos-phatis through alternating carbon sources. Water Res. 40 (20),3838–3848.

Manz, W., Amann, R., Ludwig, W., Wagner, M., Schleifer, K.H.,1992. Phylogenetic oligodeoxynucleotide probes for the majorsubclasses of Proteobacteria—problems and solutions. Syst.Appl. Microbiol. 15 (4), 593–600.

Martin, H.G., Ivanova, N., Kunin, V., Warnecke, F., Barry, K.,McHardy, A.C., Yeates, C., He, S., Salamov, A., Szeto, E., et al.,2006. Metagenomic analysis of two enhanced biologicalphosphorus removal (EBPR) sludge communities. Nat. Bio-technol. 24 (10), 1263–1269.

Meinhold, J., Filipe, C.D.M., Daigger, G.T., Isaacs, S., 1999.Characterization of the denitrifying fraction of phosphate

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 4 3 8 3 – 4 3 9 64396

accumulating organisms in biological phosphate removal.Water Sci. Technol. 39 (1), 31–42.

Meyer, R.L., Saunders, A.M., Blackall, L.L., 2006. Putative glycogen-accumulating organisms belonging to Alphaproteobacteriaidentified through rRNA-based stable isotope probing. Micro-biology—Sgm 152, 419–429.

Murray, R.G.E., Doetsch, R.N., Robinow, C.F., 1994. Determinativeand cytological light microscopy. In: Gerhardt, P., Murray, R.G.E.,Wood, W.A., Krieg, N.R. (Eds.), Methods for General andMolecular Bacteriology. American Society for Microbiology,Washington, DC, pp. 21–41.

Oehmen, A., Yuan, Z.G., Blackall, L.L., Keller, J., 2005a. Comparisonof acetate and propionate uptake by polyphosphate accumu-lating organisms and glycogen accumulating organisms.Biotechnol. Bioeng. 91 (2), 162–168.

Oehmen, A., Zeng, R.J., Yuan, Z.G., Keller, J., 2005b. Anaerobicmetabolism of propionate by polyphosphate-accumulatingorganisms in enhanced biological phosphorus removal sys-tems. Biotechnol. Bioeng. 91 (1), 43–53.

Oehmen, A., Vives, M.T., Lu, H., Yuan, Z., Keller, J., 2005c. Theeffect of pH on the competition between polyphosphate-accumulating organisms and glycogen-accumulating organ-isms. Water Res. 39 (15), 3727–3737.

Oehmen, A., Zeng, R.J., Saunders, A.M., Blackall, L.L., Yuan, Z.,Keller, J., 2006. Anaerobic and aerobic metabolism of glycogenaccumulating organisms selected with propionate as the solecarbon source. Microbiology—Sgm 152 (9), 2767–2778.

Ostle, A.G., Holt, J.G., 1982. Nile blue A as a fluorescent stain forpoly-b-hydroxybutyrate. Appl. Environ. Microbiol. 44, 238–241.

Pijuan, M., Saunders, A.M., Guisasola, A., Baeza, J.A., Casas, C.,Blackall, L.L., 2004. Enhanced biological phosphorus removalin a sequencing batch reactor using propionate as the solecarbon source. Biotechnol. Bioeng. 85 (1), 56–67.

Saito, T., Brdjanovic, D., van Loosdrecht, M.C.M., 2004. Effect ofnitrite on phosphate uptake by phosphate accumulatingorganisms. Water Res. 38 (17), 3760–3768.

Satoh, H., Mino, T., Matsuo, T., 1992. Uptake of organicsubstrates and accumulation of polyhydroxyalkanoateslinked with glycolysis of intracellular carbohydratesunder anaerobic conditions in the biological excessphosphate removal processes. Water Sci. Technol. 26 (5-6),933–942.

Shoji, T., Satoh, H., Mino, T., 2003. Quantitative estimation of therole of denitrifying phosphate accumulating organisms innutrient removal. Water Sci. Technol. 47 (11), 23–29.

Smolders, G.J.F., Vandermeij, J., Vanloosdrecht, M.C.M., Heijnen, J.J.,1994. Model of the anaerobic metabolism of the biologicalphosphorus removal process-stoichiometry and pH influence.Biotechnol. Bioeng. 43 (6), 461–470.

Third, K.A., Burnett, N., Cord-Ruwisch, R., 2003. Simultaneousnitrification and denitrification using stored substrate (PHB) asthe electron donor in an SBR. Biotechnol. Bioeng. 83 (6),706–720.

Wong, M.T., Tan, F.M., Ng, W.J., Liu, W.T., 2004. Identification andoccurrence of tetrad-forming Alphaproteobacteria in anaero-bic–aerobic activated sludge processes. Microbiology—Sgm150, 3741–3748.

Zeng, R., Yuan, Z., van Loosdrecht, M.C.M., Keller, J., 2002.Proposed modifications to metabolic model for glycogen-accumulating organisms under anaerobic conditions. Bio-technol. Bioeng. 80 (3), 277–279.

Zeng, R.J., Saunders, A.M., Yuan, Z., Blackall, L.L., Keller, J., 2003a.Identification and comparison of aerobic and denitrifyingpolyphosphate-accumulating organisms. Biotechnol. Bioeng.83 (2), 140–148.

Zeng, R.J., Yuan, Z.G., Keller, J., 2003b. Enrichment of denitrifyingglycogen-accumulating organisms in anaerobic/anoxic acti-vated sludge system. Biotechnol. Bioeng. 81 (4), 397–404.

Zilles, J.L., Peccia, J., Kim, M.W., Hung, C.H., Noguera, D.R., 2002.Involvement of Rhodocyclus-related organisms in phosphorusremoval in full-scale wastewater treatment plants. Appl.Environ. Microbiol. 68 (6), 2763–2769.