Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Available online at www.sciencedirect.com

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

Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure Thomas V. Bugge a, Poul Larsen a, Aaron. M. Saunders a, Caroline Kragelund a,b, Lisbeth Wybrandt a, Kristian Keiding a,c, Morten L. Christensen a, Per H. Nielsen a,* a

Department of Biotechnology, Chemistry and Environmental Engineering, Aalborg University, Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmark b Danish Technological Institute, Chemistry and Biotechnology, Kongsvang Alle´ 29, DK-8000 Aarhus, Denmark c Grundfos Holding, Poul Due Jensens vej 7, DK-8850 Bjerringbro, Denmark

article info

abstract

Article history:

In the conventional activated sludge process, a number of important parameters deter-

Received 6 April 2013

mining the efficiency of settling and dewatering are often linked to specific groups of

Received in revised form

bacteria in the sludge e namely floc size, residual turbidity, shear sensitivity and compo-

29 August 2013

sition of extracellular polymeric substances (EPS). In membrane bioreactors (MBRs) the

Accepted 2 September 2013

nature of solids separation at the membrane has much in common with sludge dewater-

Available online xxx

ability but less is known about the effect of specific microbial groups on the sludge characteristics that affect this process.

Keywords:

In this study, six full-scale MBR plants were investigated to identify correlations be-

Membrane bioreactors

tween sludge filterability, sludge characteristics, and microbial community structure. The

Filterability

microbial community structure was described by quantitative fluorescence in situ hy-

Extracellular polymeric substances

bridization and sludge filterability by a low-pressure filtration method. A strong correlation

FISH

between the degree of flocculation (ratio between floc size and residual turbidity) and

Microbial community structure

sludge filterability at low pressure was found. A good balance between EPS and cations in the sludge correlated with good flocculation, relatively large sludge flocs, and low amounts of small particles and single cells in the bulk phase (measured as residual turbidity), all leading to a good filterability. Floc properties could also be linked to the microbial community structure. Bacterial species forming strong microcolonies such as Nitrospira and Accumulibacter were present in plants with good flocculation and filtration properties, while few strong microcolonies and many filamentous bacteria in the plants correlated with poor flocculation and filtration problems. In conclusion this study extends the hitherto accepted perception that plant operation affects floc properties which affects fouling. Additionally, plant operation also affects species composition, which affects floc properties and in the end fouling propensity. ª 2013 Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: þ45 9940 8503; fax: þ45 9814 1808. E-mail address: [email protected] (P.H. Nielsen). 0043-1354/$ e see front matter ª 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2013.09.009

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

2

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

List of symbols k P Pa Vd Vs

1.

e1

Rate constant describing shear sensitivity (min ) Pressure difference across the filter cake (Pa) Characteristic pressure (Pa) Drainage velocity (m/s) Sedimentation velocity (m/s)

Introduction

Membrane bioreactor (MBR) technology has improved significantly during the last decade with respect to energy efficiency and is becoming a competitive alternative to the conventional activated sludge (CAS) process for municipal wastewater treatment (Drews, 2010; Judd, 2008; Meng et al., 2009). However, membrane fouling is still the main operational issue, which needs to be addressed, and much research is targeted towards the most important factors for improved fouling control. So far, there is a progressive consensus that extracellular polymeric substances (EPS), and most significantly, soluble microbial products (SMP) have significant impact on the fouling propensity and should be limited to avoid severe fouling of the membranes (Arabi and Nakhla, 2010; Drews, 2010; Liang et al., 2007; Meng et al., 2009; Trussell et al., 2006; Ahmed et al., 2007). Also bulk sludge properties such as the mixed liquid suspended solids (MLSS) levels (Lousada-Ferreira et al., 2010), floc size (Wisniewski and Grasmick, 1998), number of filamentous bacteria (Meng et al., 2006) and concentration of cations (Arabi and Nakhla, 2009; van den Broeck et al., 2011) are important for the performance of MBRs. This is very similar to the important parameters in the solideliquid separation process of the CAS process (Bruus et al., 1992; Jin et al., 2003). In CAS systems, it has been demonstrated that some physico-chemical floc properties are determined by microbiological activity (e.g. Wile´n et al., 2000) and species composition (e.g. Klausen et al., 2004; Larsen et al., 2006, 2008a). This is, at least in part, due to the production of different EPS components by the different species (Dominiak et al., 2011b; Larsen et al., 2008b). A few similar links between physicochemical floc properties and microbial community composition have been identified in MBR systems, where especially the presence of filamentous bacteria have been associated with poor floc properties and increased fouling propensity (Gil et al., 2011; Kim and Jang, 2006; Meng et al., 2006, 2007; Su et al., 2011; Tian et al., 2011). However, other studies have found negligible or even positive effect of filamentous bacteria on membrane filtration (Al-Halbouni et al., 2008; Li et al., 2008; Parada-Albarracı´n et al., 2012). Improved membrane filtration were caused by certain species of filamentous bacteria that degrade dissolved EPS, which otherwise would increase fouling propensity (Miura et al., 2007a, b; Wang et al., 2012). Other bacterial groups have also been associated with membrane fouling, e.g. nitrifiers (Drews et al., 2007). Such studies indicate that there is a link between the presence of specific microorganisms, the filtration properties and fouling propensity. A prerequisite for investigating this in

Greek symbols Average specific resistance to filtration (m/kg) aav Average specific resistance of a theoretically nona0 compressible cake (m/kg) Infinite turbidity () sinf Initial turbidity () s0

greater detail is that there is manageable number of species to investigate and fortunately, recent research has shown this is the case. In 25 Danish full-scale CAS plants with biological nitrogen and phosphorus removal, 60e90% of the total bacterial biovolume is comprised by a common core of only 30e40 species/genera (Nielsen et al., 2010, 2012), and the microbial composition in each plant is very stable over time (Mielczarek et al., 2012, 2013). Furthermore, information on several abundant bacterial groups and their floc-properties exists (Dominiak et al., 2011b; Larsen et al., 2008b) and as many of the same microorganisms are present in many full-scale MBR systems with similar operation (Saunders et al., 2013; Silva et al., 2012), the correlation between the presence of specific microorganisms and filtration properties relevant to MBRs can be investigated. The aim of this study was to investigate activated sludge at six full-scale municipal WWTP MBRs to identify correlations between sludge filterability at low pressure, sludge characteristics and microbial community structure. Sites with varying operational designs and wastewater characteristics were chosen as a first attempt to identify potential correlations for further in-depth studies.

2.

Materials and methods

2.1.

Wastewater treatment plants

Six municipal wastewater treatment plants operated with submerged membrane bioreactors were included in the study. Activated sludges from two sites in UK (Westbury and Swanage), two in France (Deauville and Avranches) and two in Germany (Nordkanal and Hu¨nxe) were analyzed (Table 1). All plants have N-removal and the two French sites also enhanced biological phosphorus removal (EBPR). In Westbury, the treated wastewater was partly industrial whereas the other sites treated only municipal wastewater. The sites varied in size with Hu¨nxe treating 8800 person equivalents (PE) to Deauville treating 115,000 PE, and also the operational design and type of membrane modules varied from site to site.

2.2.

On-site analysis and sampling

Conductivity, pH, total suspended solids (TSS), diluted sludge volume index (DSVI), shear sensitivity and sludge filterability were all analyzed on-site within two days at each site. In addition, the following samples were taken for further analysis. A 500 mL sludge sample in a 1 L bottle was used for analysis of particle size distribution, EPS extraction and FISH.

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

3

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Table 1 e Overview of membrane configuration, size and loads of the six municipal MBR WWTPs. The suspended solids are measured values of the sludge samples used in the analyses whereas the load values are based on average data for the given plant. Wastewater treatment plant

Westbury

Nordkanal

Hu¨nxe

Swanage

Avranches

Deauville

Membrane configuration () Plant size (P.E.) SS (kg/m3) BOD load (kg/d) N load (kg/d) P load (kg/d)

FS

HF 80,000 10.3 5250 597 123

FS 8800 10.5 e e e

FS 28,000 13 1524 200 e

HF 40,000 8.3 1560 390 104

HF 115,000 9.5 8350 1480 300

e 15.1 3436 238 53

FS: flat sheet, HF: hollow fiber.

For analysis of ion contents, a sludge sample of 4 mL was added to 16 mL of 0.5 M HCl. For total protein analysis, 25 mL sludge was mixed with 1 mL 200 mM NaOH, and for total carbohydrate analysis, 25 mL was mixed with 250 mL 4 M H2SO4. All samples were sent to the laboratory in Aalborg, Denmark in a cool box (<4  C) and handled within 3e5 days after sampling. Samples for FISH were taken from the 500 mL sludge sample and fixed according to Nielsen et al. (2009) upon arrival at the laboratory. The specific analysis methods are described below.

2.3.

Physico-chemical sludge characteristics

TSS was measured in accordance with Standard Methods (APHA et al., 2005). pH was measured with a pH meter (Radiometer Analytical, PHN 200) and conductivity with a conductivity meter (Radiometer Analytical, CDM 210). The size distribution of the sludge flocs and particles were analyzed using a Microtrac II (Model 7997-10, Leeds & Northrup, UK) measuring particles in the interval from 0.9 to 700 mm. The contents of Ca and Fe were determined by atomic absorption spectrometry. For analysis of EPS composition, EPS was extracted from the sludge flocs using a cation exchange resin (Dowex, Marathon C) as described by Frølund et al. (1996). The supernatant from the first centrifugation, before the extraction, was used for determination of dissolved EPS concentrations. The content of proteins, humics, and carbohydrates was determined in homogenized sludge samples (total sludge) and in two EPS fractions (dissolved and extracted) by the modified Lowry method (Frølund et al., 1996; Lowry, 1951) and a modified anthrone method (Gaudy, 1962; Raunkjær et al., 1994). Bovine serum albumin, humic acid and glucose were used as standards.

2.4.

Shear sensitivity analysis

Shear sensitivity of the sludge was analyzed according to the procedure developed by Mikkelsen et al. (1999). A 700 mL sludge sample was stirred at a constant shear intensity (G ¼ 800 s1) in a baffled reactor for 2 h. As the sludge was stirred, 5 mL samples were taken out at given time intervals and centrifuged for 2 min at 3000 rpm (Sigma Laboratory Centrifuges, Model 3e15). The turbidity of the supernatant was measured as the absorbance at 650 nm (Thermo Spectronic, Helios Epsilon). The shear sensitivity was determined as the rate constant k which is found by fitting the function

given in Eq. (1) to the measured turbidities from the experiment. s ¼ s0 þ sinf , 1  ekt




(1)

s0 is the initial turbidity of the sample and sinf is the turbidity at the end of the shearing experiment.

2.5.

Sludge filterability

Sludge filterability was determined by a low pressure deadend filtration setup. The setup and method were developed by Dominiak et al. (2011a) and Sveegaard et al. (2012). A 600 mL sludge sample was diluted with permeate to a concentration of 5 g/L and filled into a transparent plastic container (8  8 cm) placed on top of a filter paper (no. 41, cut-off 20e22 mm, Whatman, Maidstone, UK) and a permeate void, which was connected to a vacuum pump enabling varying suction pressures from 0 to 400 mbar. A small camera followed the sludge blanket, sludge settling level, and the water level in the cylinder during filtration. Filtrations were carried out with 0, 100, 200, 300 and 400 mbar. From these filtrations the settling velocity (Vs) and the drainage velocity (Vd) at the different pressures were obtained and used to calculate the average specific resistance aav as described by Dominiak et al. (2011a). The empirical relation for the specific resistance and filtration pressure given in Eq. (2) has been shown to be applicable for the low range of filtration pressures applied in this method (Bugge et al., 2012). Therefore, it has been fitted to the data sets of specific resistance and pressures measured at each site to be able to compare pressure dependence of the filtration for the analyzed sludge samples. aav ¼ a0 þ

a0 P Pa

(2)

a0 can be interpreted as the specific cake resistance of a theoretically non-compressible filter cake and will thus be affected by various factors, e.g. the packing density of the cake. Pa defines the pressure dependence of the specific cake resistance and can be interpreted as the pressure where the specific resistance reaches 2  a0, i. e. low Pa values means that the specific resistance is highly pressure dependent. The diluted sludge volume index (DSVI), which describes the settling properties of the sludge, was determined by

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

4

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Table 2 e Filtration properties, floc characteristics and composition of extracted/dissolved EPS, total sludge and divalent cations of samples from the six European full-scale MBR wastewater treatment plants. The plants are ordered according to decreasing drainage velocity (Vd). Wastewater treatment plant VD  05 (m/s) VS  105 (m/s) a0 (m/kg) Pa (Pa) Floc Mean floc size (mm) characteristics Filament index () DSVI (mL/g SS) Initial turbidity () Shear sensitivity k (mine1) Extracted and Protein extracted dissolved EPS, (mg/gSS) total sludge Humics extracted and cations (mg/gSS) Carboh. extracted (mg/gSS) Protein dissolved (mg/gSS) Humics dissolved (mg/gSS) Carboh. dissolved (mg/gSS) Protein total sludge (mg/gSS) Humics total sludge (mg/gSS) Carboh. total sludge (mg/gSS) Ca2þ (mg/gSS) Fe2þ/3þ (mg/gSS) Filtration properties

Westbury Nordkanal

Hu¨nxe

Swanage Avranches Deauville CAS av. CAS range

17.8 15.2 0.14  1011 1100 80 2 52 0.050 0.177 71.6

6.5 5.5 0.68  1011 2920 101 3.5 82 0.177 0.171 61.3

2.1 3.1 1.8  1011 860 65 2.5 91 0.171 0.020 79.8

1.5 1.5 1.5  1011 2200 59 3 159 0.153 0.100 80.5

0.9 0.4 2.4  1011 1110 35 3 259 0.131 0.107 72.4

0.3 1.1 5.6  1011 820 48 3.5 134 0.211 0.035 70.9

e 14.0 e e b 139 a 2 a 119 e c 0.062 b 42.9

e 0.54e90 e e b 55e311 a 1e5 a 31e211 e b

28e56

7.8

6.7

8.2

10.4

6.4

7.3

b

34.7

b

17e51

2.5

6.8

10.4

13.4

4.9

13.7

b

12.9

b

5.7e40

0.4

1.4

1.1

0.7

0.4

0.5

e

e

0.2

0.2

0.3

0.3

0.2

0.3

e

e

<1

<1

<1

<1

<1

<1

e

e

147

145

178

146

255

238

b

268

b

191e353

11.1

25.4

31.7

15.3

26.8

20.6

b

144

b

73e195

119

110

131

113

98.5

95.2

b

72.7

b

55e93

8.6 110

24.0 76.6

29.1 31.1

41.7 4.0

27.8 39.6

43.3 61.3

b

11.4 17.5

b

6.2e14 3.0e71

a

b

a

-

b

a Average data and range of filtration properties, filament index and SVI from seven Danish CAS wastewater treatment plants calculated from Dominiak et al. (2011b). b Average data and range of mean floc size, EPS components and cation concentrations from seven Australian CAS wastewater treatment plants calculated from Jin et al. (2003). c Average of activated sludge samples from 10 Danish CAS plants (Mikkelsen and Keiding, 2002).

settling a 1 L sludge sample in a cylinder (diameter 6 cm) for 30 min. The settled sludge volume was found for sludge diluted with permeate in different ratios (1:1, 1:2, 1:4, 1:7) depending on the MLSS at the given site and used to calculate DSVI.

2.6.

Phase contrast microscopy

The activated sludge flocs were characterized by macroscopic examination according to descriptions by Eikelboom (2002) using phase contrast microscopy. The size (small, middle, large) and shape (round, compact, irregular) of sludge flocs were recorded and the impact of filamentous bacteria on the sludge floc structure was evaluated (no impact, open structure, bridging between flocs). The abundance and identity of the filamentous bacteria were determined according to Eikelboom (2002). The filament abundance was estimated as filament index (FI) ranging from 0 (no filaments) to 5 (very many). The identity of the dominating filaments present was also recorded based on the morphological characteristics

including length, cell shape, storage products and staining properties.

2.7.

Quantitative fluorescence in situ hybridization

FISH was performed according to Nielsen et al. (2009) with the probe list shown in Table S1 in supplementary material. A CLSM (LSM 510 META; Carl Zeiss) equipped with an Ar ion laser (488 nm) and a HeNe laser (543 nm) was used to record digital images for quantification of the FISH positive fraction of the activated sludge samples. The digital image analysis was performed with custom-made macros for ImageJ (available at http://rsb.info.nih.gov/ij; developed by W. Rasband, National Institutes of Health, Bethesda, MD). Samples with high background signal were corrected by quantification of the signal from the nonEUB probe, which was subtracted from the probe with increased background. As some members of Chloroflexi are negative with the EUBmix probe, a factor accounting for this was calculated for each sample, and EUBmix corrected calculating the fraction of the remaining groups. This was also done for Microthrix parvicella as it was

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

underestimated due to poor FISH signal. The factor was calculated by quantifying three fractions in each sample: the overlay between the specific and the broad probe, the broad probe alone, and the specific probe alone. The true EUBmix area was then calculated as “area of the specific probe” plus “area of the broad probe” minus “area of the overlay”. The factor was calculated by dividing the true EUBmix with the quantified EUBmix area from image analysis. This factor was multiplied to the EUBmix area for the remaining probes.

3.

Results

3.1. Physico-chemical sludge characteristics and filterability of MBR sludge The filtration properties at low pressure, floc characteristics, and concentrations of EPS and cations for all six MBR plants are presented in Table 2 along with data from typical CAS plants for comparison. The plants are ordered with decreasing drainage velocity from left to right. A high drainage velocity implies a low specific resistance of the filter cake formed during the settling phase of the experiments and hence good filterability. Thus, the best filterability in the applied filtration setup was measured for the sludge from Westbury and the poorest filterability at the two French plants. Settling and drainage velocities (Vs and Vd) were strongly correlated with DSVI (Fig. 1a). Poor settling sludge with a DSVI above approx. 100 mL/g had very low values of Vs and Vd. Interestingly, a linear correlation was found between settling velocity and the drainage velocity. For all plants, it was found that the average specific resistance (aav) of the filter cake increased linearly as function of filtration pressure. Data from Hu¨nxe is shown as an example (Fig. 1b), where aav is plotted as function of applied vacuum in the filterability tests. With this linear pressure dependence, Eq. (2) applies and the specific resistance at zero pressure (a0) and the pressure dependence of the filter cake resistance (Pa) can be identified. The a0 values showed the same trend as the drainage velocities, as the lowest specific resistances were obtained at Westbury and Nordkanal and the highest at Avranches and Deauville. The Pa values were all low with values ranging from 820 to 2920 Pa implying that the sludge was extremely compressible at all plants. The variations of sludge filterability from plant to plant are supported by the sludge floc characteristics (Table 2). The plants with poor filterability had small mean floc size (w40 mm), high initial turbidity and high DSVI indicating a poor flocculation of the sludge, whereas the plants with best filterability had comparable large sludge flocs (w90 mm) and lower DSVI, comparable to the levels seen in typical CAS plants. The amounts of extractable EPS were comparable to that in CAS plants, but with more proteins compared to humic substances. The total amount of protein and carbohydrate in the sludge was similar to CAS plants, whereas the content of humic substances was 6e7 times lower. The cation concentrations, and most notably the iron concentrations, differed significantly from plant to plant, which was probably a result

5

of different local water hardness and varying dosage of flocculation chemicals. Avranches is placed in a region with low water hardness and the Ca concentration at this site was also in the lower range. The other sites are placed in regions with high water hardness. The sludge from Westbury had a significantly lower concentration of Ca compared to the other plants but the incoming wastewater at this site was partly industrial, which might explain the low concentration even though the water hardness in this region should be high. The addition of flocculation chemicals (e.g. FeCl3) depends on the individual operators at the different sites and data for this addition was not available, so it was not possible to link the differences in Fe concentrations directly to addition of flocculation chemicals.

3.2.

Floc morphology and main components

The morphology of sludge flocs was highly comparable to CAS sludge with the bacteria residing in the floc as single cells, microcolonies, and filamentous bacteria with or without epiphytes (Fig. 2aef). The floc structure in the six plants was highly variable ranging from tight, dense flocculated sludge in Westbury with relatively low levels of filamentous bacteria

Fig. 1 e a: Settling (6) and drainage (C) velocity as function of DSVI. b: Specific resistance to filtration (aav) as function of the applied vacuum pressure (0e40000 Pa) for Hu ¨ nxe sludge (r2 [ 0.97).

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

6

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Fig. 2 e aef: Phase contrast microcopy images of the sludge samples. The images are arranged by the drainage velocity with decreasing values from a to f. Scale bar represents 100 mm.

(Filament index, FI, of 2), to poorly flocculated sludge flocs in Deauville and Avranches, exhibiting an open structure with high levels of filamentous bacteria (FI 3e3.5)(Fig. 2e). The impact of the residing filamentous bacteria on sludge floc structure was low in Westbury and Swanage and high in the remaining plants.

3.3.

Microbial composition and growth pattern

The species composition in the full-scale MBR plants was investigated using morphological characterization and quantitative FISH (Table 3). In Westbury and Swanage the dominating filamentous morphotypes belonged to Mycolata, and in Nordkanal, M. parvicella accompanied Mycolata as the predominating filament. These two groups caused the foaming problems in these plants. In the remaining plants, a mix of Mycolata, M. parvicella and Type 0041 (both with and without attached epiphytic growth) were identified as dominating the filamentous morphotype.

When the FISH analysis was carried out, the applied set of gene probes targeted more than 60% of the bacterial biomass in the different samples, thus covering most of the abundant species. The dominant bacterial functional groups e nitrifiers, denitrifiers, and polyphosphate accumulating organisms (PAOs) e were present in similar abundance to that typically measured in Danish CAS EBPR plants. The growth pattern of the probe-defined bacterial populations within the flocs was categorized as a) single cells/weak porous cell aggregate, b) strong microcolonies with tightly packed cells typically organized in spherical structures, or c) filaments. The growth pattern was highly species-specific and consistent and this influenced the floc properties in different ways. Among the nitrifiers, the broad probe Nso190 targeting ammonium oxidizing bacteria (AOB) among Betaproteobacteria showed that primarily Nitrosomonas were most abundant with presence of small Nitrosospira populations (results not shown). Nitrite oxidizing bacteria (NOB) belonged to the genus Nitrospira. The PAOs were Accumulibacter (primarily clade II) and

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

7

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Table 3 e Microbial composition in MBR and CAS with EBPR (partly from Saunders et al., 2013). It was not possible to quantify the sample from Westbury due to high background autofluorescence. The different bacterial groups are listed according to their general observed tendency to reside in the flocs as strong microcolonies, filamentous bacteria or single cells. The plants are ordered according to increased drainage velocity. Wastewater treatment plant Forming strong microcolonies

Weak colonies or single cells

Filaments

Ammonium oxidizing bacteria (most) Nitrospira Halomonas Curvibacter Rhodobacter Azoarcus Zoogloea Thauera Accumulibacter Total % micro colony forming Competibacter Tetraphaera Saprospiraceae Streptococcus Alphaproteobacteria Total % weak colonies or single cells TM7 H. hydrossis Chloroflexi Microthrix Total % filamentous bacteria Total %

Function Nordkanal (%)

Hu¨nxe (%)

Swanage Avranches Deauville CAS EBPR CAS EBPR (%) (%) (%) (%) range (%)

AOB

0.1 (ND)

11.0 (3.3)

4.3 (0.4)

4.8 (1.6)

9.4 (1.7)

3.8

1.2e8.2

NOB PAO Denit Denit Denit Denit Denit PAO

1.2 (0.2) 29.5 (2.6) 6.2 (0.3) 0.1 (ND) 8.6 (0.6) 2.4 (0.8) 5.4 (1.2) 13.6 (2.3) 67.1

5.8 (0.2) 0.1 (ND) 12.2 (1.5) 2.0 (0.5) 0.1 (ND) 2.8 (0.7) 3.1 (0.5) 9.5 (0.2) 46.5

7.9 (0.8) 3.8 (0.8) 10.4 (1.5) 3.4 (0.9) 5.5 (0.6) 0.1 (ND) 1.2 (0.4) 2.6 (0.2) 39.2

5.0 (0.9) 0.1 (ND) 7.6 (0.3) 2.7 (0.8) 0.1 (ND) 0.1 (ND) 1.1 (0.015) 5.1 (0.7) 26.7

3.2 (0.6) 8.5 (3.7) 9.3 (0.8) 3.8 (1.0) 1.0 (0.6) 0.1 (ND) 4.6 (1.1) 4.4 (0.13) 44.3

3.1 2.1 7.1 1.2 3.1 2.1 3.1 3.6

0.2e5.7 0.0e5.7 1.5e14.1 0.2e3.4 0.8e9.3 0.2e5.2 0.2e6.2 1.0e10.8 ND

GAO PAO Hyd Ferment NA

0.1 (ND) 0.1 (ND) 8.4 (1.2) 0.1 (ND) 9.2 (1.1) 9.5

6.8 (0.5) 8.5 (0.5) 0.1 (ND) 0.1 (ND) 2.5 (0.5) 11.1

0.1 (ND) 0.1 (ND) 0.7 (0.9) 0.1 (ND) 2.8 (0.3) 3.0

2.0 (0.3) 12.9 (1.5) 0.1 (ND) 0.1 (ND) 2.0 (0.4) 15.0

1.7 (0.43) 5.5 (1.2) 0.1 (ND) 3.6 (0.4) 2.3 (0.1) 11.4

1.0 8.4 6.1 2.3 ND ND

0.0e8.0 1.8e20.7 1.2e12.2 0.2e6.7 ND ND

Hyd NA Hyd Hyd

0.1 (ND) 0.1 (ND) 8.6 (0.9) 10.9 (0.3) 19.8

2.8 (3.1) 2.0 (0.3) 11.0 (0.5) 0.1 (ND) 15.9

0.1 (ND) 0.1 (ND) 16.9 (1.8) 1.7 (0.5) 18.7

0.1 (ND) 1.1 (0.5) 33.1 (6.5) 2.8 (0.5) 37.1

0.1 (ND) 1.1 (0.07) 21.1 (0.92) 10.2 (1.1) 32.5

5.0 ND 10.1 6.0

0.3e14.0 4.0e6.0 0.3e24.0 0.3e16.0 ND

104.8

80.4

61.8

80.9

90.1

ND

ND

AOB: Ammonium oxidizing bacteria; NOB: Nitrite oxidizing bacteria; Denit: Denitrifier; PAO: Polyphosphate accumulating bacteria; GAO: Glycogen accumulating bacteria; Hyd: Hydrolysing; Ferment: Fermenting; ND: Not determined. Numbers in brackets are standard error of replicates.

Tetrasphaera. The number of filamentous bacteria was high in most plants constituting 20e35% of the biomass. They were primarily Microthrix, species belonging to phylum Chloroflexi, some low numbers of H. hydrossis, and species belonging to candidate phylum TM7. Using specific probes, the Chloroflexi filaments were dominated by Type 0092 that was hidden inside flocs in most plants. In addition, Mycolata morphotypes were identified using phase contrast microscopy, and they were the dominating filament in Nordkanal, Westbury and Swanage. It was not possible to further identify or quantify Mycolata due to poor FISH signal, which is usually due to their thick sheath that stops probe penetration.

3.4. Correlations between species composition, floc properties, and filtration propensity It was investigated whether specific bacterial species could be correlated to floc and filtration properties (Figs. 3 and 4). There was a tendency that increased abundance of strong microcolony-forming bacteria in the different plants coincided with decreased a0 (Fig. 3a) whereas there was less clear correlation between the occurrence of floc formers and Pa

(Fig. 3b). The effect of strong microcolonies on settling and drainage velocity was relatively pronounced with a positive correlation in both cases (Fig. 3c). Focusing on individual species, the presence of Accumulibacter was positively correlated to Pa (Fig. 3d) and AOB were negatively correlated to shear sensitivity (Fig. 3e). To test for the effect of single cells or weakly bound bacterial clusters, the sum of these species was plotted against the filtration data and floc properties (Table 2), but the variation in abundances was too small to identify clear trends (between 9.5 and 15% in 4 out of 5 plants). The total abundance of filamentous bacteria did not correlate well to the filtration resistance (a0) or Pa (results not shown). However, an increasing abundance of filamentous bacteria (both inside and outside the flocs) clearly impaired the settling velocity (Fig. 4a). The drainage velocity was also affected by the abundance of filaments, but to a lesser degree (Fig. 4a). The observed effect on settling and drainage velocity was particularly pronounced for bacteria belonging to Chloroflexi (Fig. 4b) where an increased abundance coincided with a strongly reduced settling and drainage velocity.

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

8

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Fig. 3 e a: Correlation between average specific resistance of a theoretically non-compressible cake (a0) and the sum of strong microcolonies (r2 [ 0.85). Strong microcolonies are defined in Table 3. Deauville (B) is considered as an outlier and not included in the linear regression. b: Correlation between characteristic pressure (Pa) plotted and the sum of strong microcolonies defined in Table 3 (r2 [ 0.64). c: Correlation between settling velocity (6, r2 [ 0.89) or drainage velocity (C, r2 [ 0.75) and the sum of strong microcolonies defined in Table 3 d: Correlation between Pa and the abundance of Accumulibacter (r2 [ 0.97). e: Correlation between shear sensitivity (k) and abundance of ammonium oxidizing bacteria (AOB) (r2 [ 0.99).

4.

Discussion

Only few studies have shown that specific bacterial species correlate with the performances of MBR systems and these studies relate primarily to the abundance of filamentous bacteria that can impair the filtration properties (Gil et al., 2011; Kim and Jang, 2006; Meng et al., 2006, 2007; Pan et al., 2010). This effect is supported in this study by the effect of filaments on DSVI and drainage velocity (which is a measure

reflecting filterability), highlighting the negative effect of bulking sludge in both CAS and MBR systems. Interestingly, this study also indicates that species composition in MBR plants determines the formation of strong or weak flocs, which in turn affects the physicochemical characteristics and thus the filtration properties of the sludge. There was a clear trend that poor filtration properties were linked to presence of small flocs, high residual turbidity and high DSVI, whereas good filtration properties were found with large flocs, low residual turbidity and low

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Fig. 4 e a: Correlation between settling velocities (6) and drainage velocities (C) and the sum of filamentous bacteria defined in Table 3 b: Correlation between settling velocities (6) or drainage velocities (C) and abundance of Chloroflexi.

DSVI. It is well-known from the CAS process that poor flocculation, i.e. weak flocs and many single cells, hamper drainage (Dominiak et al., 2011b) and high-pressure filterability (dewaterability) (Sørensen et al., 1995) and similar results have been found in pilot MBR studies (Tian and Su, 2012). An increased number of single cells and colloidal particles presumably increases the fouling resistance both due to blocking of the membrane pores and due to blinding, i.e. attachment of these particles within the fouling layer and blocking of the pores within the cake layer structure resulting in a high a0. It is also well described that the concentrations of EPS and presence of divalent cations are critical parameters for the degree of flocculation of the sludge (Bruus et al., 1992; Jin et al., 2003) since cations are needed to stabilize the weak bindings between the EPS in the sludge flocs. Additionally, the amount of EPS in the bulk phase is a critical parameter for the filterability of the sludge (Meng et al., 2009; Wang et al., 2008). The total amount of EPS has been positively correlated with the DSVI and the filament index (Meng et al., 2006, 2007; Sun et al.,

9

2011), but a similar relationship was not found in our study. This could be due to the lack of “extreme” samples with either no or very abundant fractions of filaments as the interval of the filament index of the plants in this study was limited to 2e3.5. It was not possible to find strong correlations between the individual physico-chemical parameters (floc size, initial turbidity, extractable EPS, concentration of cations) and the individual filtration property parameters (Vd, Vs, a0, Pa). This is probably due to the complex interrelations between the measured parameters and the varying conditions from site to site, e.g. shear levels, organic loading rates, wastewater composition. Instead, the measured physico-chemical parameters were combined in ratios related to the degree of flocculation and the conditions for flocculation. The ratio between initial turbidity and mean floc size can be considered an indicator for the degree of flocculation (Bruus et al., 1992; Jin et al., 2003), where a high value indicates poor flocculation; i.e. high turbidity (single cells, colloidal particles in bulk etc.) and/or small sludge flocs. The ratio between total extracted EPS amount and total calcium plus iron was considered as an indicator for the conditions for flocculation with a low value indicating good conditions for flocculation; i.e. sufficient concentration of cations for stabilization of the weak EPS bindings within the flocs. These two ratios are plotted in Fig. 5a, and a strong correlation was found between the degree of flocculation and conditions for flocculation. Further, the degree of flocculation showed a strong correlation with the measured filterability in terms of the specific resistance at zero pressure (a0) as shown in Fig. 5b. A high ratio between initial turbidity and mean floc size resulted in a high specific resistance to filtration and poor filterability. Hence, by using these ratios it was found that the degree of flocculation correlated with the filtration properties, and that the degree of flocculation correlated with the balance between amounts of extractable EPS and cations present in the sludge. Another important parameter affecting the filtration properties is the compressibility of the sludge cake on the membrane surface (Bugge et al., 2012). As given in Eq. (2), the pressure dependence of the specific resistance to filtration is given by the ratio between a0 and Pa. The values of Pa were very low at all the plants indicating that the sludge is highly compressible. The fact that sludge is highly compressible is well known for CAS sludge (Sørensen and Sørensen, 1997), and it has recently also been shown in laboratory-scale studies that compressibility should be considered when interpreting filtration data from MBRs (Bugge et al., 2012) even though the filtration pressures are very low. The importance of compressibility for the constant flux operation of MBRs, where the filtration pressure is increased gradually to maintain the flux as the fouling resistance increases, was recently demonstrated by Christensen and Keiding (2012). With highly compressible sludge floc particles in the fouling layer, the increased operational filtration pressure, which will be well above the values of Pa, implies an increase in the fouling layer resistance. Since the pressure dependence is also dependent on a0, an increased number of colloidal particles and single cells will also have an impact on the pressure dependence as these affect a0 as described above.

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

10

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Fig. 5 e a: Plot of the factor “Initial turbidity/Mean floc size” as function of the factor “Sum extracted EPS/Sum ([Fe]D [Ca])” (r2 [ 0.98). Deauville (B) is considered as an outlier. b: Plot of a0 as function of the factor “Initial turbidity/Mean floc size” (r2 [ 0.95). Deauville (B) is considered as an outlier.

Overall, the morphological characterization of the sludge fitted well with the determined filtration properties. Tight dense flocs in Nordkanal and Westbury had high drainage velocity and low resistance to filtration and the total abundance of strong microcolony-forming bacteria inside the flocs correlated positively to drainage and settling velocity. In CAS sludge it is known that a high abundance of strong microcolony-forming bacteria indicates floc properties associated with good settling and dewatering properties. Some studies have demonstrated that the abundance of filamentous bacteria negatively affects settling or filtration (Gil et al., 2011; Meng et al., 2006, 2007). However, this is the first time a positive correlation has been demonstrated between the abundance of strong microcolony-forming bacteria and good settling and filtration properties. Only limited information is available as to which bacteria are strong and weak floc formers and all of these studies were of CAS flocs. The high microcolony strength of AOB demonstrated in other studies (Larsen et al., 2008a) seemed to cause strong flocs as seen by the negative correlation between AOB and the shear sensitivity, k. Also in full-scale MBR systems, the presence of nitrifiers has been suggested to improve the

filtration performance (Trussell et al., 2009). Accumulibacter is also known to form strong microcolonies (Larsen et al., 2006) and the positive correlation between its abundance and Pa should be investigated further. It was surprising that the influence of species growing as single cells or weak aggregates did not correlate with any of the filtration parameters. It was expected that this group of bacteria would cause blinding of the filter cake and hence reduce filtration properties. However, the definition of this group shown in Table 3 is based on knowledge from CAS systems and especially the group of loosely attached bacteria may be different in MBR systems due to a different selection pressure. Hence, there might exist other loosely attached bacterial species in MBR, which are not targeted by the applied FISH probes, and more detailed analysis of weakly bound bacteria and filtration properties is needed. The filamentous species influenced the floc properties in different ways. M. parvicella and Caldilinea-species in Chloroflexi mostly resided in the bulk liquid making bridges between the flocs causing negative effects on settling properties because these loose aggregates of flocs aided by bulk filaments have a relatively low density. Type 0092 was primarily found inside the flocs forming a backbone for strong flocs. However, severe growth of Type 0092 may cause an open/porous structure of the floc (Jenkins et al., 2004), and eventually cause poor settling flocs. It was not possible to find any clear relationship between filamentous bacteria and the filtration properties in terms of a0 and Pa in this study. As stated in the introduction, other studies have found both positive and negative effects of filamentous bacteria and since the different species affect the floc properties in different ways they must still have an impact on the filterability. Thus, further studies are needed to investigate the effect of the common filamentous species on the filtration properties under conditions comparable to the cross-flow filtration mode of MBRs. The study shows that there is a link between the presence of specific microorganisms, the filtration properties and fouling propensity. This is primarily due to the phenotypical differences in the bacterial species growing as strong microcolonies, weakly bound cells, and filamentous bacteria. When such links are further established, it should also be possible to “manage” the microbial populations in MBR systems in order to optimize filtration properties by selecting for microorganisms promoting this. The proportion of these microbial fractions is influenced by operational parameters such as wastewater type, F/M ratio, and oxygenset-point. Such indented management is well known in CAS systems for bulking and foaming control (Nielsen et al., 2009, 2010) and can likely be carried out also in MBR systems.

5.

Conclusion

Activated sludge from six European full-scale MBR municipal WWTPs was investigated with respect to sludge filterability, physico chemical characteristics and microbial community structures.

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

 There was a strong positive relationship between sludge filterability and degree of flocculation.  The degree of flocculation correlated well with the balance between EPS and divalent cations.  The MBR sludges were highly compressible which may impact the filtration properties  The microbial community was largely similar to comparable CAS plants  The relative abundance of bacteria forming strong microcolonies correlated positively with floc properties and thus also with filtration properties.  Presence of filamentous bacteria including M. parvicella and Chloroflexi affected the sludge floc properties significantly and decreased the drainage properties.  In conclusion, the operational conditions of MBRs do not only affect the physico-chemical properties directly but will also determine the microbial composition, which also affect the physico-chemical properties and filterability of the sludge. Hence, there were correlations between the microbial community structures, physical chemical characteristics and sludge filterability that should be investigated in greater detail including seasonal variations of these parameters and relations to the actual filtration properties of the membranes installed in the WWTP.

Acknowledgments This study was a part of the EcoDesign MBR Centre founded by the Danish Council for Strategic Research. The authors would like to thank the staff at the six MBR plants for being open for our visit and helpful during the stays.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2013.09.009.

references

Ahmed, Z., Cho, J., Lim, B.-R., Song, K.-G., Ahn, K.-H., 2007. Effects of sludge retention time on membrane fouling and microbial community structure in a membrane bioreactor. J. Memb. Sci. 287, 211e218. Al-Halbouni, D., Traber, J., Lyko, S., Wintgens, T., Melin, T., Tacke, D., Janot, A., Dott, W., Hollender, J., 2008. Correlation of EPS content in activated sludge at different sludge retention times with membrane fouling phenomena. Water Res. 42, 1475e1488. APHA, AWWA, WEF, 2005. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, American Water Works Association, Water Environment Federation, Washington D.C. Arabi, S., Nakhla, G., 2009. Impact of cation concentrations on fouling in membrane bioreactors. J. Memb. Sci. 343, 110e118.

11

Arabi, S., Nakhla, G., 2010. Impact of molecular weight distribution of soluble microbial products on fouling in membrane bioreactors. Separ. Purif. Technol. 73, 391e396. Bruus, J.H., Nielsen, P.H., Keiding, K., 1992. On the stability of activated sludge flocs with implications to dewatering. Water Res. 26, 1597e1604. Bugge, T.V., Jørgensen, M.K., Christensen, M.L., Keiding, K., 2012. Modeling cake build-up under TMP-step filtration in a membrane bioreactor: cake compressibility is significant. Water Res. 46, 4330e4338. Christensen, M.L., Keiding, K., 2012. Effects of compressibility on MBR sludge filtration. In: Proceedings of the 11th World Filtration Congress, Graz, Austria, April 2012. Dominiak, D., Christensen, M.L., Keiding, K., Nielsen, P.H., 2011a. Gravity drainage of activated sludge: new experimental method and considerations of settling velocity, specific cake resistance and compressibility. Water Res. 45, 1941e1950. Dominiak, D., Christensen, M.L., Keiding, K., Nielsen, P.H., 2011b. Sludge quality aspects of full-scale reed bed drainage. Water Res. 45, 6453e6460. Drews, A., 2010. Membrane fouling in membrane bioreactors e characterisation, contradictions, cause and cures (review). J. Memb. Sci. 363, 1e28. Drews, A., Mante, J., Iversen, V., Vocks, M., Lesjean, B., Kraume, M., 2007. Impact of ambient conditions on SMP elimination and rejection in MBRs. Water Res. 41, 3850e3858. Eikelboom, D.H., 2002. Process Control of Activated Sludge Plants by Microscopic Investigation. IWA Publishing, UK. Frølund, B., Palmgren, R., Keiding, K., Nielsen, P.H., 1996. Extraction of extracellular polymers from activated sludge using a cation exchange resin. Water Res. 30, 1749e1758. Gaudy, A.F., 1962. Colorimetric determination of protein and carbohydrate. Indust. Water Wastes 7, 17e22. Gil, J.A., Krzeminski, P., van Lier, J.B., van der Graaf, J.H.J.M., Wijffels, T., van den Broeck, R., Smets, I.Y., van Impe, J.F.M., Prats, D., 2011. MBR performance: operational problems in industry. Filtr. Separ. 48, 38e41. Jenkins, D., Richard, M.G., Daigger, G.T., 2004. Manual on the causes and control of activated sludge bulking, foaming, and other solids separation problems, Lewis Publishers, London. Jin, B., Wilen, B.-M., Lant, P., 2003. A comprehensive insight into floc characteristics and their impact on compressibility and settleability of activated sludge. Chem. Eng. J. 95, 221e234. Judd, S., 2008. The status of membrane bioreactor technology (review). Trends Biotechnol. 26, 109e116. Kim, I.S., Jang, N., 2006. The effect of calcium on the membrane biofouling in the membrane bioreactor (MBR). Water Res. 40, 2756e2764. Klausen, M.M., Thomsen, T.R., Nielsen, J.L., Mikkelsen, L.H., Nielsen, P.H., 2004. Variations in microcolony strength of probe-defined bacteria in activated sludge flocs. FEMS Microbiol. Ecol. 50, 123e132. Larsen, P., Eriksen, P.S., Lou, M.A., Thomsen, T.R., Kong, Y.H., Nielsen, J.L., Nielsen, P.H., 2006. Floc-forming properties of polyphosphate accumulating organisms in activated sludge. Water Sci. Technol. 54, 257e265. Larsen, P., Nielsen, J.L., Otzen, D., Nielsen, P.H., 2008a. Amyloidlike adhesins produced by floc-forming and filamentous bacteria in activated sludge. Appl. Environ. Microbiol. 74, 1517e1526. Larsen, P., Nielsen, J.L., Svendsen, T.C., Nielsen, P.H., 2008b. Adhesion characteristics of nitrifying bacteria in activated sludge. Water Res. 42, 2814e2826. Li, J., Li, Y., Ohandja, D.-G., Yang, F., Wong, F.-S., Chua, H.-C., 2008. Impact of filamentous bacteria on properties of activated sludge and membrane-fouling rate in a submerged MBR. Separ. Purif. Technol. 59, 238e243.

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009

12

w a t e r r e s e a r c h x x x ( 2 0 1 3 ) 1 e1 2

Liang, S., Liu, C., Song, L., 2007. Soluble microbial products in membrane bioreactor operation: behaviours, characteristics, and fouling potential. Water Res. 41, 95e101. Lousada-Ferreira, M., Geilvoet, S., Moreau, A., Atasoy, E., Krzeminski, P., van Nieuwenhuijzen, A., van der Graaf, J., 2010. MLSS concentration: still a poorly understood parameter in MBR filterability. Desalination 250, 618e622. Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R.J., 1951. Protein measurement with the folin phenol reagent. J. Biol. Chem. 193, 265e275. Meng, F., Shi, B., Yang, F., Zhang, H., 2007. Effect of hydraulic retention time on membrane fouling and biomass characteristics in submered membrane bioreactors. Bioproc. Biosyst. Eng. 30, 359e367. Meng, F., Zhang, H., Yang, F., Li, Y., Xiao, J., Zhang, X., 2006. Effect of filamentous bacteria on membrane fouling in submerged membrane bioreactor. J. Memb. Sci. 272, 161e168. Meng, F., Chae, S.R., Drews, A., Kraume, M., Shin, H.S., Yang, F., 2009. Recent advances in membrane bioreactors (MBRs): membrane fouling and membrane material (review). Water Res. 43, 1489e1512. Mikkelsen, L.H., Keiding, K., 1999. Equilibrium aspects of the effect of shear and solids content on aggregate deflocculation. Adv. Colloid. Interface. Sci. 80, 151e182. Mikkelsen, L.H., Keiding, K., 2002. Physico-chemical characteristics of full scale sewage sludges with implications to dewatering. Water Res. 36, 2451e2462. Mielczarek, A.T., Kragelund, C., Eriksen, P.S., Nielsen, P.H., 2012. Population dynamics of filamentous bacteria in Danish wastewater treatment plants with nutrient removal. Water Res. 46, 3781e3795. Mielczarek, A.T., Nguyen, H.T.T., Nielsen, J.L., Nielsen, P.H., 2013. Population dynamics of bacteria involved in enhanced biological phosphorus removal in Danish wastewater treatment plants. Water Res. 47, 1529e1544. Miura, Y., Watanabe, Y., Okabe, S., 2007a. Significance of Chloroflexi in performance of submerged membrane bioreactors (MBR) treating municipal wastewater. Environ. Sci. Technol. 41, 7787e7794. Miura, Y., Hiraiwa, M.N., Ito, T., Watanabe, Y., Okabe, S., 2007b. Bacterial community structures in MBRs treating municipal wastewater: relationship between community stability and reactor performance. Water Res. 41, 627e637. Nielsen, P.H., Kragelund, C., Seviour, R.J., Nielsen, J.L., 2009. Identity and ecophysiology of filamentous bacteria in activated sludge. FEMS Microbiol. Rev. 33, 969e998. Nielsen, P.H., Mielczarek, A.T., Kragelund, C., Nielsen, J.L., Saunders, A.M., Kong, Y., Hansen, A.A., Vollertsen, J., 2010. A conceptual ecosystem model of microbial communities in enhanced biological phosphorus removal plants. Water Res. 44, 5070e5088. Nielsen, P.H., Saunders, A.M., Hansen, A.A., Larsen, P., Nielsen, J.L., 2012. Microbial communities involved in enhanced biological phosphorus removal from wastewater e a model system in environmental biotechnology. Curr. Opin. Biotechnol. 23, 452e459. Pan, J.R., Su, Y.C., Huang, C., Lee, H.C., 2010. Effect of sludge characteristics on membrane fouling in membrane bioreactors. J. Memb. Sci. 349, 287e294. Parada-Albarracı´n, J.A., Marin, E., Pe´rez, J.I., Moreno, B., Go´mez, M.A., 2012. Evolution of filamentous bacteria during urban wastewater treatment by MBR. J. Environ. Sci. Health. Part A, Toxic/hazard. Subst. Environ. Eng. 47, 863e872.

Raunkjær, K., Hvitved-Jacobsen, T., Nielsen, P.H., 1994. Measurement of pools of protein, carbohydrate and lipid in domestic wastewater. Water Res. 28 (2), 251e262. Saunders, A.M., Larsen, P., Nielsen, P.H., 2013. Comparison of nutrient-removing microbial communities in activated sludge from full-scale MBRs and conventional plants. Water Sci. Technol. 68, 366e371. Silva, A.F., Carvalho, G., Oehmen, A., Lousada-Ferreira, M., van Nieuwenhuijzen, A., Reis, M.M., Crespo, M.T.B., 2012. Microbial population analysis of nutrient removal-related organisms in membrane bioreactors. Appl. Microbiol. Biotechnol. 93, 2171e2180. Su, Y., Pan, J.R., Huang, C., Chang, C., 2011. Impact of sludge retention time on sludge characteristics and microbial community in MBR. Water Sci. Technol. 63, 2250e2254. Sun, C., Fiksdal, L., Hanssen-Bauer, A., Rye, M.B., Leiknes, T., 2011. Characterization of membrane biofouling at different operating conditions (flux) in drinking water treatment using confocal laser scanning microscopy (CLSM) and image analysis. J. Memb. Sci. 382, 194e201. Sveegaard, S.G., Keiding, K., Christensen, M.L., 2012. Compression and swelling of activated sludge cakes during dewatering. J. Memb. Sci. 46, 4999e5008. Sørensen, P.B., Christensen, J.R., Bruus, J.H., 1995. Effect of small scale solids migration in filter cake filtration of wastewater solids suspension. Water Environ. Res. 67, 25.32. Sørensen, B.L., Sørensen, P.B., 1997. Structure compression in cake filtration. J. Environ. Eng. 123, 345e353. Trussell, R.S., Merlo, R.P., Hermanowicz, S.W., Jenkins, D., 2006. The effect of organic loading on process performance and membrane fouling in a submerged membrane bioreactor treating municipal wastewater. Water Res. 40, 2675e2683. Trussell, R.S., Jang, N., Merlo, R.P., Kim, I.S., Hermanowicz, S.W., Jenkins, D., 2009. Changes in mixed liquor and organic foulant properties affect membrane fouling for non-nitrifying and nitrifying conditions. Water Environ. Res. 81, 255e264. Tian, Y., Chen, L., Zhang, S., Cao, C., Zhang, S., 2011. Correlating membrane fouling with sludge characteristics in membrane bioreactors: an especial interest in EPS and sludge morphology analysis. Bioresour. Technol. 102, 8820e8827. Tian, Y., Su, X., 2012. Relation between the stability of activated sludge flocs and membrane fouling in MBR: under different SRTs. Bioresour. Technol. 118, 477e482. van den Broeck, R., Krzeminski, P., van Dierdonck, J., Gins, G., Lousada-Ferreira, M., van Impe, J.F.M., van der Graaf, J.H.J.M., Smets, I.Y., van Lier, J.B., 2011. Activated sludge characteristics affecting sludge filterability in municipal and industrial MBRs: unraveling correlations using multi-component regression analysis. J. Memb. Sci. 378, 330e338. Wang, Q., Wang, Z., Wu, Z., Ma, J., Jiang, Z., 2012. Insights into membrane fouling of submerged membrane bioreactors by characterizing different fouling layers formed on membrane surfaces. Chem. Eng. J. 179, 169e177. Wang, Z., Wu, Z., Yin, X., Tian, L., 2008. Membrane fouling in a submerged membrane bioreactor (MBR) under sub-critical flux operation: membrane foulant and gel layer characterization. J. Memb. Sci. 325, 238e244. Wisniewski, C., Grasmick, A., 1998. Floc size distribution in a membrane bioreactor and consequences for membrane fouling. Colloid. Surf. A: Physicochem. Eng. Aspect. 138, 403e411. Wilen, B., Nielsen, J.L., Keiding, K., Nielsen, P.H., 2000. Influence of microbial activity on the stability of activated sludge flocs. Colloid. Surf. B: Biointerfaces. 18, 145e156.

Please cite this article in press as: Bugge, T.V., et al., Filtration properties of activated sludge in municipal MBR wastewater treatment plants are related to microbial community structure, Water Research (2013), http://dx.doi.org/10.1016/ j.watres.2013.09.009