w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

Available online at www.sciencedirect.com

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

Characterization of biofouling in a lab-scale forward osmosis membrane bioreactor (FOMBR) Qiaoyun Zhang a, Yap Wei Jie b, Winson Lay Chee Loong b, Jinsong Zhang b, Anthony G. Fane b, Staffan Kjelleberg c,d,e, Scott A. Rice c,d,e, Diane McDougald a,c,e,* a

Advanced Environmental Biotechnology Centre, Nanyang Environment and Water Research Institute, Singapore Singapore Membrane Technology Centre, Nanyang Environment and Water Research Institute, Singapore c Centre for Marine Bio-Innovation, School of Biotechnology and Biomolecular Science, University of New South Wales, Sydney, Australia d Singapore Centre on Environmental Life Sciences Engineering, Nanyang Technological University, Singapore e School of Biological Sciences, Nanyang Technological University, Singapore b

article info

abstract

Article history:

Forward osmosis membrane bioreactors (FOMBR) provide high quality permeate, however

Received 26 November 2013

the propensity for membrane biofouling in FOMBRs is unknown. Here, FOMBRs were

Received in revised form

operated under high and low aeration and the membrane-associated biofilms were char-

13 March 2014

acterized by confocal laser scanning microscopy (CLSM) and rRNA gene-tagged pyrose-

Accepted 18 March 2014

quencing. CLSM images revealed that there was little biofilm formed under high aeration,

Available online 5 April 2014

while thick biofilms were observed on the membranes operated under low aeration. The diversity and richness of bacterial and archaeal communities as assessed by pyrose-

Keywords:

quencing varied under high and low aeration. The composition of the bacterial suspended

Forward osmosis membrane

sludge communities and the sessile biomass on the membrane surface, as assessed by

bioreactor

non-metric multidimensional scaling, was significantly different under high aeration, but

Biofouling

was more similar under low aeration. SIMPER analysis indicated that Pseudomonas, Aero-

Biofilm

monas and Fluviicola preferentially attached to the membrane. The results presented here

Community analysis

provide a comprehensive understanding of membrane biofouling in FOMBRs, which is

Wastewater treatment

essential for the development of effective control strategies.

Membrane fouling

1.

Introduction

Membrane bioreactors (MBRs) are increasingly used worldwide as a wastewater treatment strategy, due to the introduction of more stringent regulations for effluent quality

ª 2014 Elsevier Ltd. All rights reserved.

(Stephenson, 2000). MBRs have a reduced footprint, can provide sufficient biodegradation of organic matters, and crucially, produce high quality product water with complete removal of suspended solids and lower total organics compared to conventional wastewater treatment technologies (Judd, 2006). However, the pore sizes of current

* Corresponding author. Nanyang Technological University, School of Biological Science, 60 Nanyang Drive, Singapore, Singapore. Tel.: þ65 63162818. E-mail address: [email protected] (D. McDougald). http://dx.doi.org/10.1016/j.watres.2014.03.052 0043-1354/ª 2014 Elsevier Ltd. All rights reserved.

142

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

microfiltration and ultrafiltration membranes that are used in MBRs are not sufficient to retain all undegraded organics, which poses threats to downstream reuse of treated water (Ramesh et al., 2006). Therefore, various innovative technologies are under development to overcome the limitations of conventional MBRs. A novel, high retention MBR, the forward osmosis membrane bioreactor (FOMBR), uses osmotic pressure rather than hydraulic pressure as the driving force. FOMBR requires two solutions with different salt concentrations separated by a semi-permeable membrane, which allows the passage of water across the membrane while solute molecules or ions are retained (Cath et al., 2006). It is currently unknown if the FO membrane is prone to fouling as is typical of other types of membrane systems. Membrane fouling has an adverse impact on operational sustainability and the cost of the MBR process. Biofouling has been determined to be the most serious and pervasive problem for the use of MBRs (Mansouri et al., 2010) because it decreases the permeate productivity (Flemming, 2011), causes damage to the membrane module shortening its lifetime and causes increased energy consumption (Schneider et al., 2005) due to the need for increased aeration. In order to develop effective strategies for the control of biofouling, a more thorough understanding of the biofouling process is required. Many studies on biofouling in conventional MBRs have focused on visualization of cake morphology (Meng et al., 2010) and identification of bacterial community structure (Miura et al., 2007a) by conventional molecular techniques such as PCR-DGGE, clone libraries and fluorescence in situ hybridization (FISH). For example, Yun et al. (2006) characterized biofouling in a dye wastewater treatment plant using CLSM and discovered that biofilms formed under anoxic conditions were thinner than biofilms formed under aerobic conditions, but the anoxic biofilm covered the membrane more uniformly which caused higher membrane fouling. In Miura’s study (2007a), the FISH results revealed significant differences between the microbial communities on membrane surfaces and those in the planktonic biomass in the mixed liquor, and both FISH and 16S rRNA gene sequencing showed that the Betaproteobacteria probably played a major role in the development of the mature biofilms. Culture-independent approaches such as DGGE and clone libraries can be used to improve the characterization of microbial communities in various environments, but these techniques can only identify the most abundant community members. Pyrosequencing allows rapid characterization of microbial communities at great depth (Amend et al., 2010) as it can generate 400,000 reads using multiplex barcoding. This technology has been widely applied to a variety of environmental samples, including marine water (Qian et al., 2011), soil (Roesch et al., 2007) and the human body (Costello et al., 2009). Pyrosequencing has been recently used to investigate the microbial diversity and ecosystem function of engineered environments. For example, pyrosequencing revealed clear geographical differences between activated sludge samples from Asia and North America (Zhang et al., 2012). Lim et al. (2012) investigated biofouling in an MBR to determine how the microbial community influences the fouling process. Pyrosequencing indicated that Enterobacter cancerogenus was a

dominant community member of the biocake on the membrane. The authors report that this species had a preference for attachment and growth on the membrane due to its quorum sensing activity, which was closely associated with biofouling. Here, rDNA-tagged pyrosequencing was used to investigate biofouling in an unconventional MBR. The bacterial and archaeal community diversity and composition of the suspended sludge and of the biofilm on the MBR membrane was characterized. In addition, CLSM was used to characterize the structure and morphology of the fouling biomass under different operating conditions. Sample discrimination analysis was performed to determine the organisms responsible for fouling.

2.

Materials and methods

2.1.

The lab-scale FOMBR system

The forward osmosis membrane bioreactor (FOMBR) used in this study was set-up as previously described (Lay et al., 2011). The reactor had a dual-track design that allows modules to be run with similar or different operational parameters depending on experimental demands. Each track contained a 4 L aerobic bioreactor with a membrane area of 372 cm2. A set of monitoring and control systems were connected to the bioreactor and measured parameters such as conductivity, pH and dissolved oxygen (DO) concentration online. The feed and draw compartments were separated by the membrane module, and were recirculated to maintain the stability of the system. Both compartments were filled to prevent hydraulic pressure on the membranes. The feed bioreactor volume was maintained by a water level sensor which regulated a peristaltic pump that controlled the addition of feed solution to the compartment. On the draw side, a constant NaCl concentration was maintained by a conductivity meter connected to the draw reservoir. Flux was calculated based on the overflow from the draw tank to the product tank via an ultrasound level sensor. Final flux values were calculated by overflow normalized to membrane area. A software package (Omron CX-Supervisor, Japan) was installed for measuring flux, adjusting the water level in the bioreactor and the salinity in the draw tank. The composition of the synthetic feed water used in this study was designed to simulate domestic sewage (Lay et al., 2011). There were four FOMBR experiments with different operational parameters as listed in Table 1. An aerobic

Table 1 e Varied operational parameters of the FOMBR. Run Feed Draw e NaCl SRT [d] Length of Average TOCa run [d] DO [mg L1] [mol kg1] [ppm] A B C D a

200 400 400 400

0.5 0.5 1.0 1.0

20 10 10 10

73 35 35 43

8.95 8.53 6.54 4.59

Feed consisted of 200 mg of glucose, 200 mg of sodium acetate, 50 mg of meat extract and 50 mg of peptone per kg tap water.

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

environment was maintained by the use of four aeration channels as shown in Supplementary Fig. 1, so that more than 1 mg L1 of dissolved oxygen (DO) was supplied continuously. The pH and temperature of the mixed liquor were maintained at 6.6e7.9 and 20e22  C, respectively. For a sludge retention time (SRT) of 20 days, approximately 300 mL of mixed liquor were discharged every weekday for physiochemical analysis and replaced with feed solution. For an SRT of 10 days, 600 mL of sludge were discharged every day. During Runs A and B, all four air channels were on maximum aeration, while during Run C, two of the four channels became blocked after 25 days of operation. In order to test the influence of reduced aeration on the performance of the FOMBR, the reactor was run for a further 10 days. During Run D two air channels were manually shut off in the middle of the run, similar to the operation in Run C.

2.2.

Confocal laser scanning microscopy

Membranes were removed from the reactors on the final day of each experiment. Eight pieces with an area of approximately 3.5  3.5 mm, were randomly cut from each membrane. The biofouling layers were stained with LIVE/DEAD BacLight Bacterial Viability Kits (Invitrogen) which utilized propidium iodide (PI) for detection of dead cells, and SYTO9 for live cells. The working concentrations of PI and SYTO9 were 3 and 0.51 mM respectively in 0.85% NaCl buffer. Samples were stained and incubated at room temperature for 20 min. Polysaccharides were analyzed by staining with the fluorescein isothiocyanate conjugated concanavalin A (FITC-ConA) which has an affinity for a-glucose and a-mannose residues (Chu and Li, 2005; Goldstein et al., 1965; Strathmann et al., 2002) This dye was diluted in PBS buffer at room temperature in the dark for 30 min. Examination of stained biofilms was performed by CLSM (Carl Zeiss LSM 710, Germany) using Argon 488 (for dyes SYTO9 and ConA-FITC) and DPSS 561-10 (for PI) lasers for excitation. The CLSM images were generated in multi track mode. Two and three dimensional images of random fields of view were acquired (Zen 2009), and images were analyzed using Imaris (Version 7.2.2, Bitplane) to calculate quantitative parameters such as biovolume and coverage of the substratum.

143

were sequenced by tag-encoded FLX amplicon pyrosequencing (TEFAP) by Research Testing Laboratories (Texas, USA). Sequences were trimmed in Mothur (Schloss, 2009) to remove ambiguous and low-quality sequences, primers and barcodes. A non-redundant database was obtained and robust sequence alignments were created using the SILVA database as previously described (Schloss, 2009). Sequences shorter than 150 bp were removed using Screen and 454 sequencing noise was reduced using Precluster (Huse et al., 2010). Chimeric sequences were detected and removed using ChimeraSlayer (Haas et al., 2011). Both OTU-based and phylotype-based approaches were used to analyze community diversity. Pairwise distances were calculated and all sequences were assigned to OTUs by clustering (cutoff ¼ 0.03). To compare the 8 samples at the same sequencing depth, normalization of the sequence number was performed using Mothur. Rarefaction curves obtained from Mothur were plotted in R 2.15.0 (Copyright (C) 2012 The R Foundation for Statistical Computing). Numbers of OTUs, Chao 1 estimator and Inversimpson estimator, and Shannon Weaver index were used to evaluate the alpha diversity. The relative abundances were transformed (fourth root) when required in order to satisfy the assumption of normality in Primer V6 (PRIMER-E Ltd, UK). Resemblance analysis based on the algorithm of BrayeCurtis similarity was presented as non-metric multi-dimensional scaling (NMDS) plots. CLUSTER analysis was used for hierarchical clustering and sample discrimination (SIMPER) determined the contribution of each species to average dissimilarity between sample groups. Furthermore, Bio-Env tool of Primer correlated the datasets of biological communities with the environmental variables. In addition to the OTU-based analysis, a phylotype-based approach was used to analyze community composition. Effective bacterial sequences were assigned to taxonomic units based on classification against the SILVA database with the consensus confidence threshold set at 51%. A bootstrap of 60% or above was used for assignment of taxonomic units to taxa levels. For archaeal sequences, the Ribosomal Database Project (RDP) was chosen for alignment (Bates et al., 2011) at a confidence threshold of 80%. All the effective sequences were binned into phylotypes using Mothur-formatted RDP archaeal references based on Bayesian statistics.

2.3. Analysis of microbial community diversity and composition Four suspended sludge samples and four biofilms samples were collected for extraction of genomic DNA by a modified cetyltrimethyl ammonium bromide (CTAB) protocol (Lay et al., 2012). The concentration of DNA was determined by use of a spectrophotometer (Thermo Scientific spectrometer, Nanodrop 1000) and adjusted to 50e100 ng mL1 for PCR. RNA was removed by RNase treatment prior to pyrosequencing. For 16S rDNA tagged pyrosequencing, 100 ng mL1 of DNA was used as template for PCR reactions. Two fusion primer pairs were used to amplify the target DNA fragments: Gray 28F (GAGTTTGATCNTGGCTCAG) - Gray 518R (GTNTTACNGCGGC KGCTG) were used for amplification of bacterial 16S rDNA and Arch349F (GYGCASCAGKCGMGAAW) e Arch806R (GGACTACVSGGGTATCTAAT) for archaeal 16S rDNA. PCR products

Fig. 1 e Flux profile across four runs of the FOMBR system e Run A (blue line), Run B (red line), Run C (green line) and Run D (purple line). Flux was calculated by total volume of the overflow normalized to the membrane area. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

144

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

Fig. 2 e CLSM images showing the distribution of biomass (aed) and polysaccharides (eef) on membranes. Samples from Run A (a), Run B (b), Run C (c) and Run D (d) were stained with the LIVE/DEAD BacLight kit. CLSM images of ConA-FITCstained polysaccharides in Run A (e), Run B (f), Run C (g) and Run D ( h).

3.

Results

3.1.

FOMBR reactor performance

In Run A, the flux was 3.2 Lm2 h1 at the start of experiment, and stabilized at approximately 2.7 Lm2 h1 (Fig. 1). For Run B, the flux was between 2.9 and 3.6 Lm2 h1. The initial flux was higher for Runs C and D due to the increased salt concentration of draw solutions. It is interesting to note that the flux in Runs A and B was stable throughout the entire operation, while the

flux in Runs C and D was not stable even before the reduction in aeration. In Run C, the flux ranged from 2 to 5.8 Lm2 h1 for the first 22 d, after which the flux declined from 5 to 2.2 Lm2 h1 within next five days, and stablized afterwards. On days 20 and 21, the flux decreased from 3.7 to 1.2 Lm2 h1 in Run D. This is consistent with previously published studies of an FOMBR, where there was a sudden initial drop in flux, followed by a slow decline in flux (Lay et al., 2010), which could be attributed to the increase in salt on the feed side, resulting in increased osmotic pressure, or due to the blocking of the membrane pores by the initial attachment of foulants.

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

3.2. Characterization of biofouling by confocal laser scanning microscope At the end of each operation, the FO membranes were removed, stained with the BacLight LIVE/DEAD kit and observed under CLSM. During Runs A and B, both live and dead cells were detected in low abundance on membrane surfaces (Fig. 2a and b respectively), while in Runs C and D, thick biofilms formed (Fig. 2c and d respectively). Some single cells were found to be attached to the membranes, but most occurred in clusters. Biofilm formation in Run C was homogeneous as the cells were evenly distributed across the fouling layers. Furthermore, the biofilm in Run C was thicker than those in Runs A and B, but was not as compact as the biofilm formed in Run D. Polysaccharides are one of the major components of the biofilm matrix, thus the distribution and abundance of polysaccharides was determined by staining with ConA-FITC (Fig. 2eeh). Small aggregates of polysaccharides were detected on the membranes from Runs A and B (Fig. 2e and f), while some larger deposits of polysaccharides were detected in Runs C and D (Fig. 2g and h). The distribution of polysaccharides was more uniform than the biomass, which probably resulted in a greater hydraulic barrier than the biomass. The high relative coverage of the membrane by polysaccharides in Runs C and D would have a large effect on membrane permeability. IMARIS (Bitplane Switzerland, V7.3.3.2) was used to analyze the CLSM images to quantify the biovolume and surface coverage of biofilms normalized by area (100  100 mm) (Fig. 3).

Fig. 3 e Quantification of biofilms by IMARIS to determine (a) biomass biovolume and thickness, (b) surface coverage of biomass, (c) polysaccharide biovolume and thickness and (d) percent surface coverage of polysaccharides. Data are presented as the average from more than 8 images.

145

To quantify differences between biofilms formed under high (Runs A and B) and low aeration (Runs C and D), the average values of Runs A and B were compared to the average of Runs C and D. The biomass biovolume and surface coverage under low aeration was 31 and 12 times higher than that under high aeration respectively. The polysaccharide biovolume and surface coverage under low aeration was 10 and 40 times higher than that under high aeration, respectively. Both the biovolume and coverage of polysaccharides were much higher than biomass, suggesting that the matrix was quite extensive.

3.3. Comparison of microbial communities of suspended sludge and biofilms 3.3.1.

Diversity and composition of bacterial communities

After trimming, aligning, screening, filtering, preclustering and chimera slaying, eight bacterial datasets, including suspended sludge samples from day 73 of Run A (Ad73B), day 35 of Run B (Bd35B), day 35 of Run C (Cd35B) day 43 of Run D (Dd43B)

Fig. 4 e Rarefaction curves (a) and NMDS plot of suspended and attached bacterial communities (b) defined by 3% variation among 16S rRNA bacterial sequences. The size of every dataset was normalized to be 1,820 reads for rarefaction curves. Points labeled as ‘*’ represent the suspended sludge communities and ‘A’ represent the biofilm samples.

146

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

and biofilm samples from fouled membranes collected on the same days with suspended sludge at the end of every run: Run A (AMB), Run B (BMB), Run C (CMB), Run D (DMB), 47,807 sequences with an average length of 313 bp were obtained. The sequence reads were clustered into groups of 3% variation within one OTU. All of the sequence datasets were standardized using Mothur, and the size of datasets was set at 1820 sequences as a minimum. The sludge sample from day 35 of Run B (Bd35B) had the richest and most diverse bacterial community with the least sampling coverage (90%), while the bacterial community attached to the membrane in this run (BMB) had the least richness and diversity with the highest sampling coverage (99.2%) (Table 1). Approximately 300 OTUs were detected in each suspended sludge sample under high aeration, with 93 and 26 OTUs in the sessile biomass attached to the membrane in Runs A and B, respectively (Table 1 and Fig. 4a). The Shannon Weaver index indicated that the suspended sludge was 4e10 fold more diverse than the sessile biomass. However, both the number of OTUs and Shannon Weaver index showed similar diversity of suspended and sessile biomass under low aeration. Beta diversity of the communities was determined from an OTU abundance matrix based on 3% variation to evaluate the dissimilarity between communities. The original datasets without normalization were used because the abundance of OTUs was normalized by conversion to percentages. An NMDS plot based on BrayeCurtis similarity demonstrated that the bacterial communities fell into four groups (Fig. 4b). The sludge bacterial communities in the high aeration runs (A and B) clustered together and had a similarity of 41%. The two biofilm communities from Runs A and B were much more distant to each other, but still clustered into one group. Under low aeration conditions, suspended samples had a high similarity with biofilm communities (Run C sludge (Cd35B) and biofilm (CMB), Run D sludge (Dd35B) and biofilm (DMB)). The resemblance analysis showed that dissimilarity between the suspended sludge and the membrane samples in four runs were 79.6, 95.4, 63.1 and 41.64% respectively. Therefore, under high aeration, membrane attached bacterial communities

Fig. 5 e Relative bacterial abundance of 14 major phylogenetic classes retrieved from biofilms and suspended sludge in an FOMBR during four operations (Runs AeD). The abundances are presented as percentages.

were dissimilar to the suspended sludge, while under low aeration the biofilm and sludge communities were similar to each other. The SILVA database was used for taxonomical assignments of the bacterial datasets. Three hundred and twentynine bacterial phylotypes from 8 samples were identified, with 42 being unclassified at the genus level. The distribution of taxonomies at the class level is shown in Fig. 5. With the exception of the sludge in Run C, all of the samples were dominated by Proteobacteria. This agrees with previous studies of bacterial communities from sewage (Zhang et al., 2012; Mclellan et al., 2010), membrane bioreactors (Miura et al., 2007b; Lim et al., 2012) and drinking water distribution systems (Hong et al., 2010). Bacterial populations of biofilms of Runs A and B were composed of Gammaproteobacteria at an abundance of 85 and 96% respectively. Further classification at the genus level revealed that Pseudomonas was the dominant genus (97.4 and 97.3%) and Aeromonas, Dokdonella and Luteimonas constituted the minor proportion of the Gammaproteobacteria. With respect to the other samples, Gammaproteobacteria represented 11.8% of the sequence reads from the sludge bacterial community in Run A, and 75% of the Gammaproteobacteria in Run A sludge belonged to Pseudomonas. However, the abundance of Gammaproteobacteria was low in the Run B sludge bacterial community and in both sludge and biofilm samples of Runs C and D. Alphaproteobacteria dominated the sludge communities in Runs A and B and the biofilm community in Run C at proportions of 34.9, 33.2 and 51.6% respectively. Betaproteobacteria accounted for 54 and 41% of the total sequences from Run D sludge and biofilm respectively, and 41% of the sequences from Run C sludge belonged to Flavobacteria. In addition to the predominant classes in individual samples, there were other classes present at high proportions as well. Run A sludge contained Betaproteobacteria, Flavobacteria, Gammaproteobacteria and Sphingobacteria at frequencies from 10.8 to 14%. The abundance of Betaproteobacteria and Flavobacteria was high in both sludge and biofilm communities of Run B. Run C sludge was comprised of Alpha- and Betaproteobacteria in large percentages. The sludge and biofilm communities from Run D shared Flavobacteria and Actinobacteria as their secondary dominant phyla. There were diverse genera in the dominant phyla mentioned above (Fig. 5). The genus Corynebacterineae, which belonged to the class Actinobacteria, was predominant in the Run D sludge community. Among the nine major genera under the phylum Flavobacteria detected in the FOMBR, Muricauda was the most dominant genus in all of the suspended sludge samples except for Run D. Flavobacterium, Muricauda and Pibocella were detected with abundances of approximately 4% both in suspended sludge and in the biofilm in Run D. In addition, Coenonia accounted for 7.2 and 9.7% in Run C biofilm and sludge communities respectively, while abundance was low in samples from other runs. Of the 329 genera in the FOMBR communities, 101 belonged to the class Alphaproteobacteria. The 11 most dominant genera are listed in Table 2. Within Alphaproteobacteria, Brevundimonas and Paracoccus accounted for 13.6 and 6.4% of sequence reads from Run A sludge respectively. An

147

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

Table 2 e Number of sequences and alpha diversity of sludge and biofilm bacterial communities. Bacterial samples

Raw reads

Trimmed reads

Effective sequences

OTUs

Chao 1

ACE

Invsimp-son

Shannon Weaver

Good’s coverage (%)

AMB Ad73B BMB Bd35B CMB Cd35B DMB Dd43B

7811 7337 9550 6832 4865 4735 15472 3238

7518 7257 9221 6407 4037 3797 15120 3175

6770 5124 8371 5035 3360 2849 13611 2687

93 293 26 300 164 141 129 158

243 723 47 781 503 402 264 233

1.44 26.78 1.13 23.11 9.95 8.27 10.55 14.42

428 1175 113 1218 1307 556 318 299

1.02 4.24 0.36 4.17 3.09 2.80 3.11 3.55

96.54 90.44 99.18 89.89 93.90 95.11 96.59 96.48

unclassified genus of the family Alysiosphaera accounted for 29.4% and the genus Amaricoccus represented 9.4% of biofilm samples in Run B. The diversity of Betaproteobacteria and Gammaproteobacteria in the FOMBR was not as high as for the Alphaproteobacteria, yet some genera in the Betaproteobacteria were predominant in particular samples. Within Betaproteobacteria, Castellaniella represented 47.8% in the Run D biofilm community, and 38.2% in the suspended sludge, Methylobacillus was present at 15.8% in Run B sludge and 20.7% in Run C sludge communities. Although Gammaproteobacteria were only detected in a few samples, the genus Pseudomonas accounted for 83% and 94.4% of sessile biomass of Runs A and B respectively.

3.3.2.

Diversity and composition of archaeal communities

No archaea were found in FOMBR Runs A to C. In Run D, the archaeal datasets of suspended sludge and biofilms contained 134 and 471 reads respectively (Table 4). All the reads clustered to 6 OTUs at the cutoff of 0.03. A phylotype-based approach using the SILVA database revealed that the archaeal reads could be classified at the genus level. Most of the genera belonged to the order Methanomicrobiales. In Run D sludge, 76.9% were assigned to Methanosaeta, while 75.4% of biofilm reads were Methanoculleus.

4.

Discussion

The data presented here shows that under high aeration, the membrane was only slightly fouled, while low aeration resulted in significant biofouling. As reported previously, FO membranes are designed to be smooth and hydrophilic (Tang et al., 2010), and there is no hydraulic pressure exerted on membrane, thus it is relatively resistant to fouling. However, when the membrane surface is covered with foulants, the organic matter and the attached microbes causes the membranes to become more hydrophobic and rough. If the shear force is not strong enough, the membrane can lose fouling resistance, and eventually becomes fouled. The diversity and richness of microbial communities as assessed by 16S gene pyrosequencing varied under high and low aeration. The composition of the bacterial communities in the suspended sludge and the sessile biomass on the membrane surface as assessed by NMDS plots was significantly different under high aeration but was more similar under low aeration. Phylogenetic information gained through pyrosequencing can reveal the contribution of the individual bacteria to membrane fouling and the characteristics of the biofilms formed on the membranes.

3.3.3. Genera that contributed to the dissimilarity between suspended and sessile biomass in an FOMBR

4.1.

Biofilm structure on FOMBR membranes

The SIMPER tool in Primer V6 was used to identify taxa that contributed most to the dissimilarity between the suspended sludge and sessile biofilm communities (Table 3). All the responsible bacterial genera were among the top 50 genera (Table 2) with the exception of Micropruina and Corynebacterium. The discrimination analysis within individual runs revealed that in Runs A, B and C, Pseudomonas was responsible for the dissimilarity. Its abundance in biofilms was significantly higher than in suspended sludge, while in Run D its abundance in the sludge was higher than the biofilms. The genera of Chloroacidobacterium, Terrimonas, Lactococcus, DB1-14 and Chromobacterium were not abundant in any community (>1%), but their occurrence contributed a lot to the dissimilarity. In contrast, Castellaniella had a high abundance, but it contributed little to the sample dissimilarity. The genera of Muricauda in Runs A and B and Methylobacillus and Rhodobacter in Runs B and C were abundant in suspended sludge but rarely found in biofilms. There was only one genus (Pseudomonas) which preferred to attach and grow on the membrane in at least three runs (Table 5).

Similar to biofilms in submerged MBRs, bacterial clusters and polysaccharides contribute to the biofouling in an FOMBR (Lee et al., 2008). Previous studies in a conventional MBR revealed that the biovolume of biomass and polysaccharides were 0.93 and 3.82 mm3/mm2 respectively, on the hollow fiber membranes (Bjorkoy and Fiksdal, 2009). The biofilms in the FOMBR system in this study had a slightly higher biomass biovolume, but slightly lower biovolume of polysaccharides. So there was no significant difference in the amount of biofoulants between the conventional MBR and FOMBR. Furthermore, in an MBR (Chu and Li, 2005) biopolymers not only adsorbed to the membrane surface, but also entered the membrane pores and could not be removed by regular physical washing. However, the fouling layer on the FOMBR membranes was not tightly attached and was removed by immersion in water without any pressure after removal from the bioreactor. Therefore, the structure of the FOMBR biofilms was not significantly different from those in an MBR, but the removal of the fouling layer is much easier in the FOMBR compared to traditional MBRs.

148

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

Table 3 e Distribution of bacterial genera in suspended and sessile sludge samples. Taxonomic levels

Relative abundance (%)

Class

Genera

Acidobacteria Actinobacteria

Bacteroidia Flavobacteria

Sphingobacteria

TM7 ML635J-21(100) Deinococcales Bacilli Clostridia Gemmatimonadetes Alphaproteobacteria

Betaproteobacteria

Gammaproteobacteria

a b c d

AMB

Chloroacidobacterium Corynebacterineaeb Micrococcineae_fa Propionibacterineae_f Paludibacter Fluviicola Coenonia Flagellimonas Flavobacterium Gaetbulibacte Leeuwenhoekiella_f Muricauda Persicivirga Pibocella Terrimonas Chitinophagaceae_f Algoriphagus Aquiflexum Cyclobacteriaceae_f Flexibacter Fulvivirga WCHB1-69 TM7_p ML635J-21 Truepera Lactococcus Proteiniborus Gemmatimonas Brevundimonas DB1-14 Aminobacter Afifella Shinella Amaricoccus Paracoccus Rhodobacter Alysiosphaera MNH4 Sphingopyxis Alcaligenes Castellaniella Janthinobacterium Methylobacillus Chromobacterium Denitromonas Thauera Aeromonas Pseudomonas Dokdonella Luteimonas

d

e 0.09 0.16 0.04 0.15 0.78 0.03 0.01 e e e 0.18 e e e 0.06 0.19 0.04 e 0.01 0.01 e e 3.59 e 0.01 e e 1.52 e e e 0.24 e 0.19 0.18 0.04 e 0.32 e 0.12 3.29 0.15 e 0.07 0.03 1.31 83.0 0.13 0.01

Ad73B

BMB

Bd35B

CMB

Cd35B

DMB

Dd43B

1.19 0.68 4.08 2.50 e 0.02 1.23 0.68 e 0.02 0.18 9.52c 0.02 0.92 e 1.09 1.83 3.08 2.63 2.28 1.48 e 1.56 0.72 0.16 0.49 e 0.86 13.60 1.07 e 0.12 2.95 0.82 6.40 4.41 0.98 0.02 0.72 e e 0.16 5.39 0.60 1.74 1.93 e 8.88 1.13 0.33

e e e e 1.54 e e e 0.01 e e 0.01 e e e e e e e e e e e e 0.01 0.07 e 0.01 0.02 e e e e e 0.01 e e e e e 0.08 1.73 0.01 e e 0.04 1.18 94.4 0.02 e

0.56 0.26 2.38 4.91 0.02 0.26 1.83 1.49 e e 0.18 16.98 e 0.08 0.06 1.19 0.02 3.20 0.52 1.55 0.34 e 0.52 e 0.04 1.03 e 0.79 2.28 0.22 0.10 0.10 7.65 1.09 8.54 5.84 1.55 0.02 1.01 e e e 15.75 0.79 0.54 2.44 e 0.64 1.17 0.50

e 0.08 0.27 0.06 e 1.96 7.21 1.51 0.05 0.87 0.47 21.9 0.46 0.31 1.08 0.06 0.03 e 0.01 0.71 e e e e e e e 0.01 1.75 e 0.01 1.77 0.02 9.41 0.09 e 29.5 5.57 1.12 e 0.02 e 2.20 0.08 6.59 0.12 e 0.17 0.01 0.06

e e 0.88 e e 0.03 9.70 1.78 0.05 0.47 0.53 27.46 0.22 0.14 0.95 e 0.10 0.07 e 5.78 0.09 e e e e 0.51 e 0.04 0.13 e 0.01 0.42 1.30 1.28 0.67 0.19 7.01 0.99 2.03 e 0.01 e 20.67 0.97 12.16 0.58 e e e 0.01

0.03 4.41 5.23 0.81 e e 1.21 0.32 4.87 0.28 0.09 4.04 0.50 3.99 0.16 0.16 e e e 0.80 0.03 1.73 0.48 e 1.87 e 0.99 e 0.50 0.74 2.02 e 0.14 0.08 0.46 e 0.25 0.10 e 1.59 47.8 e 0.06 0.01 3.47 0.19 e e e 1.39

0.11 13.81c 6.18 0.71 e e 1.27 0.15 3.76 0.33 0.19 3.98 0.60 3.98 e 0.15 e e e 1.04 0.07 2.98 0.07 e 1.49 e 6.70 e 0.71 0.41 1.27 e 0.19 0.15 0.82 e 0.04 e e 2.49 38.18 e e e 0.11 0.04 e 0.07 e 0.74

Some reads were only classified to the family (represented by _f), phylum (_o), or class level (_c). Names in bold were taxa abundant in reactor. Values underlined are the proportion of abundant genera. Symbol represents the absence of genera in the samples.

Table 4 e Alpha diversity of archaeal communities of Dd43A and DMA. Samples Dd43A DMA

Effective reads

OTUs

Coverage

Chao 1

Invimpson

ACE

Shannon

471 134

6 5

99.6% 99.3%

6 6

1.63 1.7

13.4 6.5

0.64 0.82

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

Table 5 e Bacteria primarily provided dissimilarity in each run. Sample discrimination

Responsible bacteria

RAM vs RAd73

AMB > Ad73B Pseudomonas Aeromonas

RBM vs RBd35

BMB > Bd35B Pseudomonas

RCM vs RCd35

CMB > Cd35B Fluviicola Alysiosphaera Amaricoccus Pseudomonas DMB > Dd43B Denitromonas Terrimonas MNH4 Methylobacillus TM7 (p)

RDM vs RDd43

AMB < Ad73B Cyclobacteriaceae (f) TM7(p) Muricauda Chloroacidobacterium Pibocella BMB < Bd35B Muricauda Shinella s Methylobacillus Rhodobacter Micropruina Paracoccus CMB < Cd35B Methylobacillus Lactococcus Shinella Rhodobacter DMB < Dd43B Proteiniborus Pseudomonas Corynebacterium

Note: The parentheses following the taxa signified that OTUs could not be classified to any taxonomy at the genus level, so they give the lowest ancient levels: f-family, o-order, p-phylum.

its ability to secrete EPS and form biofilms, is widely used as a model for the study of biofilm formation (Hassett et al., 2002; Yong and Zhong, 2009; Barraud et al., 2006; McDougald et al., 2008). At low aeration in Runs C and D, there was a relatively small dissimilarity between suspended and attached bacterial communities, indicating that the lack of scouring by aeration allowed most species in the reactor to settle on the membrane. However, there was a large difference in the abundance of these organisms in sludge and membrane samples. Methylobacillus is dominant in suspended sludge, but in low numbers on the membrane, while Alysiosphaera dominated on the membrane in Run D while in low abundance in the sludge (29.2% vs 7%). Alysiosphaera are filamentous, non-motile organisms that are very abundant in industrial wastewater treatment plants and are often involved in bulking incidents (Kragelund et al., 2006). The fact that these organisms are filtamentous may explain their partitioning on the membrane. This study reveals that the fouling microbial community in an FOMBR under low and high aeration are different. However, we acknowledge that these organisms may not be important foulers in all systems. Even for the same FOMBR set-up, if the operational parameters were changed, or the source of the sludge is different, the fouling microbial consortium may be significantly different. Therefore, while this study investigates biofouling in novel MBR configurations, more investigations are needed to clarity the roles and functions of the organisms on membranes.

4.3. 4.2. Genera in the suspended sludge and biofilm communities The phylotype-dependent approach indicated that there was a similarity of 20 and 4.6% between the suspended sludge and membrane communities in Runs A and B, respectively. A high similarity between bacterial communities in the suspended sludge and on the membrane was observed in Runs C and D according to OTU-based analysis. Phylotype-based analysis revealed that some taxa were abundant in both sludge and membrane samples, i.e. Muricauda, Coenonia and Denitromonas in Run C, and Castellaniella, Corynebacterineae, Flavobacterium and Muricauda in Run D. These bacteria have a wide metabolic diversity and are adapted to a variety of environments. Some of the dominant bacteria identified here are well adapted to the marine environment (Muricauda, Coenonia, Flavobacterium) (Yoon et al., 2005; Euzeby, 2000) while other species have been isolated from activated sludge or wastewater related environments (Castellaniella, Denitromonas and Corynebacterineae) (Liu et al., 2008). Under high aeration, there was a dissimilarity of 80% between the suspended sludge and the sessile biomass. Thus, under high aeration, only those species which are well adapted to biofilm formation are able to overcome the scouring effects of aeration. Among the bacteria contributing the most to the sample dissimilarities (Table 3), Janthinobacterium and Pseudomonas were abundant on the membrane but were rare in the suspended sludge. Pseudomonas species are widely distrubuted in aquatic environments and due to

149

Biofilm maturation in Runs C and D

In this study, membranes were only collected at the end of every experiment in this study, and thus, the fouling process was not monitored temporally. However, the maturation state of the biofilm can be inferred according to the microbial community composition and reactor performance. In one study (Gao et al. (2011), Alphaproteobacteria were found to be the initial colonizers. This group was replaced by Betaproteobacteria or Deltaproteobacteria in the stable biofouling stage. Miura et al. (2007a) also reported that when the aeration decreased, Betaproteobacteria became the most dominant in biofilms. The authors suggested that the anoxic and anaerobic conditions inside the biocake does not favor most of the Alphaproteobacteria populations, thus their growth and propagation would be negatively affected when aeration decreases, and gradually replaced by other bacterial groups, such as Betaproteobacteria, which are able to multiply in the oxygen-limited environment (Miura et al., 2007a). Similar dominant bacterial populations were detected on the membranes in our reactor. In Run C, 53% of the bacterial sequences from the biofilm were Alphaproteobacteria, while in Run D, 60.6% were Betaproteobacteria (Fig. 5), possibly suggesting that the biofilm was at the colonization stage in Run C, and in the succession or stabilization stage in Run D. Therefore, it is possible that the biofilm in Run C was in early stages of development while the Run D biofilm was mature. In addition to the bacterial community structure, the reactor performance also reflected the fouling situation. The reactor was run for 8 d after a sudden drop of flux in Run C,

150

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

while it was run for 22 d after a dramatic drop from 3.7 to 1.2 Lm2 h1 in Run D (Fig. 1). The sudden decrease in flux may represent the beginning of fouling on the membranes, which resulted in blockage of the membrane pores (Bjorkoy and Fiksdal, 2009). The initial attachment of pioneer microbes and EPS production may provide a favorable environment for the subsequent attachment of other microorganisms on the membranes. Consequently, in the following 22 d, biofilms in Run D accumulated more organic matter and microbes. As a result, the flux further decreased to 0.9 Lm2 h1 by the end of Run D (Fig. 1). In Run C, after 8 d of continuous fouling the operation was ended when flux decreased to around 2 Lm2 h1. In addition, the biofilm in Run C was homogeneous, while in Run D it was more compact (Fig. 2). Therefore, both the flux decline and the CLSM images indicated that biofouling in Run D was more serious than Run C.

4.4.

The influence of aeration on hydraulic resistance

The hydraulic resistance in Runs A and B was negligible, but became significant in Runs C and D. According to the comparison between Runs C and D (Table S1), there was no straightforward correlation between biofilm thickness and hydraulic resistance. Indeed, it has been shown previously that thickness is not proportional to resistance (Dreszer et al. (2013). Here, we show that the thinner biofilm with a higher biovolume in Run D resulted in higher resistance than the thicker biofilm in Run C. A study on nanofiltration and reverse osmosis membrane systems showed that higher aeration led to higher hydraulic resistance (Vrouwenvelder et al., 2011), but the results of our FOMBR were in contrast (Table S1). We find that lower aeration (represented by DO) in Run D induced higher hydraulic resistance. A possible explanation is related to the different driving forces and lower flux. The FOMBR uses osmotic pressure rather than hydraulic pressure as the driving force, which may lead to differences in air transport in the mixed liquor as well as on the membranes surfaces. The flux in the FOMBR was much lower than conventional MBRs, so the convective forces towards the membranes would be lower. In addition, the average DO value does not represent the effective shear forces on the membranes. Therefore, a moderately high aeration is needed as a potential control strategy of fouling in FOMBRs.

5.

Conclusions

Taken together, the results presented here demonstrate that aeration is a crucial factor for prevention of FO membrane fouling. Both the performance of the reactor and the microbial community composition revealed different stages of membrane fouling under different aeration conditions. Under high aeration conditions, the microbial communities of the suspended sludge and the membrane were dissimilar which indicates that only those organisms that are able to withstand the scouring can attach to the membrane. Under conditions of low aeration, most of the organisms present in the sludge were found on the membrane. The data presented here provides a better understanding of biofouling mechanisms and

lays the foundation for the development of biofouling control strategies.

Acknowledgment This research was supported by a research grant (MEWR C651/ 06/177) from the Environment and Water Industry Programme Office of Singapore (gs1). The authors also thank Dr. Torsten Thomas from CMB of UNSW, Dr. Charmaine Ng and Dr. James Guest from AEBC for their help in the optimization of the analysis pipeline for pyrosequencing datasets.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2014.03.052.

references

Amend, A.S., Seifert, K.A., Bruns, T.D., 2010. Quantifying microbial communities with 454 pyrosequencing: does read abundance count? Mol. Ecol. 19, 5555e5565. Barraud, N., Hassett, D.J., Hwang, S.H., Rice, S.A., Kjelleberg, S., Webb, J.S., 2006. Involvement of nitric oxide in biofilm dispersal of Pseudomonas aeruginosa. J. Bacteriol. 188, 7344e7353. Bates, S.T., Berg-Lyons, D., Caporaso, J.G., Walters, W.A., Knight, R., Fierer, N., 2011. Examining the global distribution of dominant archaeal populations in soil. Isme J. 5, 908e917. Bjorkoy, A., Fiksdal, L., 2009. Characterization of biofouling on hollow fiber membranes using confocal laser scanning microcscopy and image analysis. Desalination 245, 474e484. Cath, T.Y., Childress, A.E., Elimelech, M., 2006. Forward osmosis: principles, applications, and recent developments. J. Membr. Sci. 281, 70e87. Chu, H.P., Li, X.Y., 2005. Membrane fouling in a membrane bioreactor (MBR): sludge cake formation and fouling characteristics. Biotechnol. Bioeng. 90, 323e331. Costello, E.K., Lauber, C.L., Hamady, M., Fierer, N., Gordon, J.I., Knight, R., 2009. Bacterial community variation in human body habitats across space and time. Science 326, 1694e1697. Dreszer, C., Vrouwenvelder, J.S., Paulitsch-Fuchs, A.H., Zwijnenburg, A., Kruithof, J.C., Flemming, H.C., 2013. Hydraulic resistance of biofilms. J. Membr. Sci. 429, 436e447. Euzeby, J.P., 2000. The genera Riemerella and Coenonia: a short review. Rev. Med. Vet-Toulouse 151, 63e68. Flemming, H.-C., 2011. Microbial biofouling: unsolved problems, insufficient approaches, and possible solutions. In: Flemming, H.-C., Wingender, J., Szewzyk, U. (Eds.), Biofilm Highlights. Springer Berlin Heidelberg. Gao, D.W., Fu, Y., Tao, Y., Li, X.X., Xing, M., Gao, X.H., Ren, N.Q., 2011. Linking microbial community structure to membrane biofouling associated with varying dissolved oxygen concentrations. Bioresour. Technol. 102, 5626e5633. Goldstein, I.J., Hollerman, C.E., Merrick, J.M., 1965. Proteincarbohydrate interaction. I. The interaction of polysaccharides with concanavalin A. Biochim. Biophys. Acta 97, 68e76. Haas, B.J., Gevers, D., Earl, A.M., Feldgarden, M., Ward, D.V., Giannoukos, G., Ciulla, D., Tabbaa, D., Highlander, S.K., Sodergren, E., et al., 2011. Chimeric 16S rRNA sequence

w a t e r r e s e a r c h 5 8 ( 2 0 1 4 ) 1 4 1 e1 5 1

formation and detection in Sanger and 454-pyrosequenced PCR amplicons. Genome Res. 21, 494e504. Hassett, D.J., Cuppoletti, J., Trapnell, B., Lymar, S.V., Rowe, J.J., Yoon, S.S., Hilliard, G.M., Parvatiyar, K., Kamani, M.C., Wozniak, D.J., et al., 2002. Anaerobic metabolism and quorum sensing by Pseudomonas aeruginosa biofilms in chronically infected cystic fibrosis airways: rethinking antibiotic treatment strategies and drug targets. Adv. Drug. Deliv. Rev. 54, 1425e1443. Hong, P.Y., Hwang, C.C., Ling, F.Q., Andersen, G.L., Lechevallier, M.W., Liu, W.T., 2010. Pyrosequencing analysis of bacterial biofilm communities in water meters of a drinking water distribution system. Appl. Environ. Microbol. 76, 5631e5635. Huse, S.M., Welch, D.M., Morrison, H.G., Sogin, M.L., 2010. Ironing out the wrinkles in the rare biosphere through improved OTU clustering. Environ. Microbiol. 12, 1889e1898. Judd, S., 2006. The MBR Book: Principles and Applications of Membrane Bioreactors in Water and Wastewater Treatment. Elsevier. Kragelund, C., Kong, Y., Van Der Waarde, J., Thelen, K., Eikelboom, D., Tandoi, V., Thomsen, T.R., Nielsen, P.H., 2006. Ecophysiology of different filamentous Alphaproteobacteria in industrial wastewater treatment plants. Microbiol-Sgm. 152, 3003e3012. Lay, W.C., Chong, T.H., Tang, C.Y., Fane, A.G., Zhang, J., Liu, Y., 2010. Fouling propensity of forward osmosis: investigation of the slower flux decline phenomenon. Water Sci. Technol. 61, 927e936. Lay, W.C., Zhang, Q.Y., Zhang, J.S., Mcdougald, D., Tang, C.Y., Wang, R., Liu, Y., Fane, A.G., 2011. Study of integration of forward osmosis and biological process: membrane performance under elevated salt environment. Desalination 283, 123e130. Lay, W.C., Zhang, Q.Y., Zhang, J.S., Mcdougald, D., Tang, C.Y., Wang, R., Liu, Y., Fane, A.G., 2012. Effect of pharmaceuticals on the performance of a novel osmotic membrane bioreactor (OMBR). Sep. Sci. Technol. 47, 543e554. Lee, C.H., Park, P.K., Lee, W.N., Hwang, B.K., Hong, S.H., Yeon, K.M., Oh, H.S., Chang, I.S., 2008. Correlation of biofouling with the bio-cake architecture in an MBR. Desalination 231, 115e123. Lim, S., Kim, S., Yeon, K.M., Sang, B.I., Chun, J., Lee, C.H., 2012. Correlation between microbial community structure and biofouling in a laboratory scale membrane bioreactor with synthetic wastewater. Desalination 287, 209e215. Liu, O.M., Ten, L.N., Im, W.T., Lee, S.T., 2008. Castellaniella caeni sp nov., a denitrifying bacterium isolated from sludge of a leachate treatment plant. Int. J. Syst. Evol. Microbiol. 58, 2141e2146. Mansouri, J., Harrisson, S., Chen, V., 2010. Strategies for controlling biofouling in membrane filtration systems: challenges and opportunities. J. Mater. Chem. 20, 4567e4586. McDougald, D., Klebensberger, J., Tolker-Nielsen, 2008. Pseudomonas aeruginosa: a model for biofilm formulation. In: B, Rehm (Ed.), Pseudomonas. Modal Organism, Pathogen, Workhorse. WILEY-VCH Verlag GmbH and Co, Weinheim. Mclellan, S.L., Huse, S.M., Mueller-Spitz, S.R., Andreishcheva, E.N., Sogin, M.L., 2010. Diversity and population structure of sewage-derived microorganisms in

151

wastewater treatment plant influent. Environ. Microbiol. 12, 378e392. Meng, F.G., Liao, B.Q., Liang, S.A., Yang, F.L., Zhang, H.M., Song, L.F., 2010. Morphological visualization, componential characterization and microbiological identification of membrane fouling in membrane bioreactors (MBRs). J. Membr. Sci. 361, 1e14. Miura, Y., Watanabe, Y., Okabe, S., 2007a. Membrane biofouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater: impact of biofilm formation. Environ. Sci. Technol. 41, 632e638. Miura, Y., Watanbe, Y., Okabe, S., 2007b. Membrane biofouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater: impact of biofilm formation. Environ. Sci. Technol. 41, 632e638. Qian, P.Y., Wang, Y., Lee, O.O., Lau, S.C.K., Yang, J.K., Lafi, F.F., AlSuwailem, A., Wong, T.Y.H., 2011. Vertical stratification of microbial communities in the Red Sea revealed by 16S rDNA pyrosequencing. Isme J. 5, 507e518. Ramesh, A., Lee, D.J., Wang, M.L., Hsu, J.P., Juang, R.S., Hwang, K.J., Liu, J.C., Tseng, S.J., 2006. Biofouling in membrane bioreactor. Sep. Sci. Technol. 41, 1345e1370. Roesch, L.F., Fulthorpe, R.R., Riva, A., Casella, G., Hadwin, A.K.M., Kent, A.D., Daroub, S.H., Camargo, F. a. O., Farmerie, W.G., Triplett, E.W., 2007. Pyrosequencing enumerates and contrasts soil microbial diversity. Isme J. 1, 283e290. Schloss, P.D., 2009. A high-throughput DNA sequence aligner for microbial ecology studies. Plos One 4. Schneider, R.P., Ferreira, L.M., Binder, P., Ramos, J.R., 2005. Analysis of foulant layer in all elements of an RO train. J. Membr. Sci. 261, 152e162. Stephenson, T., 2000. Membrane Bioreactors for Wastewater Treatment. IWA. Strathmann, M., Wingender, J., Flemming, H.C., 2002. Application of fluorescently labelled lectins for the visualization and biochemical characterization of polysaccharides in biofilms of Pseudomonas aeruginosa. J. Micobiol Meth. 50, 237e248. Tang, C.Y.Y., She, Q.H., Lay, W.C.L., Wang, R., Fane, A.G., 2010. Coupled effects of internal concentration polarization and fouling on flux behavior of forward osmosis membranes during humic acid filtration. J. Membr. Sci. 354, 123e133. Vrouwenvelder, J.S., Van Loosdrecht, M.C.M., Kruithof, J.C., 2011. A novel scenario for biofouling control of spiral wound membrane systems. Water Res. 45, 3890e3898. Yong, Y.C., Zhong, J.J., 2009. A genetically engineered whole-cell pigment-based bacterial biosensing system for quantification of N-butyryl homoserine lactone quorum sensing signal. Biosens. Bioelectron. 25, 41e47. Yoon, J.H., Lee, M.H., Oh, T.K., Park, Y.H., 2005. Muricauda flavescens sp. nov. and Muricauda aquimarina sp. nov., isolated from a salt lake near Hwajinpo Beach of the East Sea in Korea, and emended description of the genus Muricauda. Int. J. Syst. Evol. Microbiol. 55, 1015e1019. Yun, M.A., Yeon, K.M., Park, J.S., Lee, C.H., Chun, J., Lim, D.J., 2006. Characterization of biofilm structure and its effect on membrane permeability in MBR for dye wastewater treatment. Water Res. 40, 45e52. Zhang, T., Shao, M.F., Ye, L., 2012. 454 Pyrosequencing reveals bacterial diversity of activated sludge from 14 sewage treatment plants. Isme J. 6, 1137e1147.

Characterization of biofouling in a lab-scale forward osmosis membrane bioreactor (FOMBR).

Forward osmosis membrane bioreactors (FOMBR) provide high quality permeate, however the propensity for membrane biofouling in FOMBRs is unknown. Here,...
1MB Sizes 0 Downloads 3 Views