This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

Desalination 286 (2012) 120–130

Contents lists available at SciVerse ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

Rapid sand filtration pretreatment for SWRO: Microbial maturation dynamics and
filtration efficiency of organic matter
Edo Bar-Zeev a,⁎, 1, Natalia Belkin a, 1, Boris Liberman b, Tom Berman c, Ilana Berman-Frank a
a
Bar Ilan University, Mina & Everard Goodman Faculty of Life Sciences, Ramat Gan, Israel
b
I.D.E. Kadima, Israel
c
Kinneret Limnological Laboratory, Israel Oceanographic and Limnological Research, P.O.B. 447, Migdal 14950, Israel

a r t i c l e i n f o a b s t r a c t

Article history: Rapid sand filtration (RSF) is used today as an effective pretreatment procedure to enhance water quality
Received 4 July 2011 prior to reverse osmosis (RO) membranes in desalination plants. RSF in newly operated desalination facilities
Received in revised form 3 November 2011 requires a maturation period of about three months before the feedwater may be filtered efficiently. To date
Accepted 4 November 2011
the desalination industry RSF has regarded RSF mainly as a physical barrier effectively retaining particles
Available online 7 December 2011
larger than 0.35 mm.
Keywords:
In this study we assessed the potential of the RSF as a biological filter by following the dynamics of bacterial
Rapid sand filtration colonization and metabolic activity within the filter bed and determining filtration efficiency in respect to
SWRO pretreatment particulate and dissolved organic carbon, chlorophyll a and transparent exopolymeric particles (TEP). Bacte-
Biofilm rial abundance and diversity were shown by DGGE and SEM analysis to increase gradually over a three month
Biodegradation sampling period. Bacterial metabolic activity within the filter bed interstitial water increased erratically with
Maturation period time. With the outgrowth of a microbial population and biofilm development on the filter bed medium, sig-
Filtration efficiency nificant removal of organic carbon from the source water was always observed. Our results indicate that cur-
rent RSFs function as biofilters with moderate efficiency. Innovative design is needed to maximize the
biofiltration potential of these filters.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction
Limitations of global freshwater supplies have stimulated the
application of desalination technology with desalinized water coming
online worldwide at a rate of 40 to 50 million m 3 d − 1 [1]. Currently,
about 50% of global desalination is based on filtration through reverse
osmosis (RO) membranes requiring effective pretreatment proce-
dures upstream to reduce fouling, maintain performance and extend
membrane lifetime and to ensure the manufacturers requirements
for membrane recovery yield [1,2].
Due to its relative simplicity, low energy consumption, and rela-
tively low operational costs, rapid sand filtration (RSF), based on
granular gravity filtration, is the most common pretreatment present-
ly used in large-scale SWRO facilities. RSF with flow rates ranging
between 5 and 30 m 3 h − 1 is a robust technology to physically remove
suspended solids larger than 0.35 mm from water by size exclusion
and adsorption. Flocculation and adsorption prior to the RSF are gen-
erally used to enhance the removal of particulate aggregates formed
by coagulation procedures, while backwashing is the common proce-
dure to clean and reset the RSF performance.
⁎ Corresponding author. Tel.: + 972 544856110; fax: + 972 3 5318214.
E-mail address: edobarzeev@gmail.com (E. Bar-Zeev).
1
Contributed equally to this work.
Slow sand filtration (SSF), with flow rates ranging between 0.1
and 0.2 m 3 h − 1, has been a standard biofiltration treatment for
decades in the wastewater industry. In these systems, biofiltration is
accomplished by a diverse microbial community that colonizes and
forms biofilm on the particle surfaces within the filter bed medium
and acts to decompose various types of organic material [3–5]. The
initial degradation steps are accomplished by extracellular enzymatic
hydrolysis of macromolecules to smaller substrates, which can then
be transported into the biofilm [6]. Further degradation takes place
by a wide range of specific enzymes proliferated by a diverse micro-
bial biofilm community and is completed by oxidation of organic
carbon to CO2 with the concurrent utilization of electron acceptors
such as oxygen or sulfate [7].
Presently, little is known about the biological functioning of RSF
systems. Due to the rapid flow of water through these filters (up to
~300 fold in comparison with SSF) it has been tacitly assumed that
the biological impacts on water passing through RSF are minimal.
The water quality at various pretreatment stages is usually followed
by measuring parameters such as silt density index (SDI), available
organic carbon (AOC), biodegradable dissolved organic carbon
(BDOC) and turbidity. However, recent work [8,9] has pointed out
the need for monitoring other water quality characteristics such as
Chlorophyll a (Chl a), transparent exopolymer particles (TEP), par-
ticulate and dissolved organic carbon (POC and DOC respectively)

0011-9164/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.desal.2011.11.010
 Author's personal copy
E. Bar-Zeev et al. / Desalination 286 (2012) 120–130
that can affect the RO biofouling [10,11]. Chlorophyll a is a standard
method to evaluate algal concentrations while TEP are microscopic
(0.4–300 μm), organic gelatinous particles consisting mainly of
acid polysaccharides, detected by staining with Alcian Blue [12],
and are important components of the natural organic matter
(NOM) pool. Current findings within the water treatment and desa-
lination industries indicate a significant role for TEP in fouling RO
membranes [9–11,13].
In this study we evaluated the potential of a RSF as a biofilter at
the newly operational Hadera SWRO facility, Israel. We followed
the spatial and temporal dynamics of bacterial colonization using
DGGE and SEM visualization and heterotrophic bacterial activity
within the filter bed media during the first three months of operation
to assess the development of the microbial community during matu-
ration of the RSF. The filtration efficiency of the RSF with respect to
Chl a, TEP and dissolved and particulate organic carbon (DOC and
POC, respectively) was measured at several times during the matura-
tion period to determine the extent to which the RSF was functioning
as a biofilter.
2. Materials and methods
2.1. Sampling strategies
Water and filter bed media samples were collected from a newly
constructed, dual media RSF during the first three months of opera-
tion (11 sampling dates) from October 2009 to January 2010 com-
prising the “maturation period”, at the Hadera SWRO desalination
facility (Israel). The sampling was divided into two stages, 1) the
first 4 weeks (0 to 27 days) when a single, 100 cm deep, fine layer
of filter bed medium (grain size 0.5–1 mm ø) was in place, and
2) from 42 to 85 days, when two substrate layers were present; a
crude substrate (grain size 1–3 mm ø) 100 cm thick overlying a
fine layer, (100 cm). (Note: this two stage process is a standard industry
procedure). Throughout the entire maturation period the RSF was sub-
jected to frequent backwashing; for consistency, sampling was always
performed ~15 h after backwashing.
The sampling locations were: 1. Plant intake (feedwater refer-
ence), 2. Pre-RSF. Water after treatment with coagulant and immedi-
ately overlying the RSF, 3–6. Four depths within the single-medium
fine substrate filter bed (A, B, C, D) during the first stage, and four
depths within the dual media filter bed (A*,B*,C*–D*) during the sec-
ond stage, 7. RSF-effluent (filtrate). Sampling dates and locations are
given in Table 1 and Fig. 1.
2.2. Water and particulate sampling within the filter bed
We used a specially constructed sampler/corer [14] to obtain sam-
ples of interstitial water and particulate media at different depths
within the filter bed Particulate samples were stored in 15 ml poly-
carbonate tubes (Falcon) at 4 °C for denaturing gradient gel electro-
phoresis (DGGE) and scanning electron microscopy (SEM) analyses.
Interstitial water (3 L) was pumped directly from each sampling
depth into clean flasks and analyzed within several hours at the
laboratory.
2.3. Dissolved Oxygen (DO) measurements within the filter bed
To measure dissolved oxygen (DO) concentrations at various
depths in the filter media, we used a custom designed corer that con-
tained a pre-calibrated combined DO optode and temperature sensor
connected to a Microx TX3 oxygen meter (PreSense Inc. OXY1) and
portable PC [14].
121
Table 1
Detailed description of the sampling depths within the RSF during the two operational
stages, 1. single (fine grained) medium layer, and 2. two layer of media composite
(crude and fine-grained).
Stage Media depth Media Key Description
(cm) type
First ~ 100 Fine grains A Media near-surface
(single-medium) B Upper-mid filter bed
C Lower-mid filter bed
D Bottom of the filter bed
Second ~ 200 Crude grains A* Upper media near-surface
(dual media) B* Bottom of upper-media
layer
Fine grains C* Lower media near-surface
D* Bottom of the filter bed
2.4. Nucleic-acid extraction and PCR amplification
Filter bed media grain samples were rinsed with sterile sea
water (10 ml) to remove planktonic bacteria. The grains were re-
suspended with 1 ml lysis buffer (40 mM EDTA, 50 mM Tris pH = 8.3
and 0.75 M Sucrose) in 2 ml cryotubes and stored at −80 °C until ex-
traction. Nucleic acids were extracted from the media using a phenol–
chloroform extraction method modified according to Massana et al.
[15] and Brinkof et al. [16]. The extracted DNA was amplified by poly-
merase chain reaction (PCR) using GM5F-GC-Clamp and 907R [17].
Amplification conditions for the PCR included an initial denaturation
step of 95 °C for 4 min, followed by 34 cycles of 94 °C for 30 s, 60 °C
for 30 s, 72 °C for 30 s and a final extension step of 72 °C for 10 min.
2.5. Denaturing gradient gel electrophoresis (DGGE) and analysis
DGGE was performed with a 20 slot Hoefer TM SE 600 Standard
dual cooled gel electrophoresis unit (Hoefer Scientific Instruments,
San Francisco, CA, USA) and gel gradient (5 ml min − 1) using a peri-
staltic pump (Cole-Parmer Instrument Co.). Gels (18 × 16 cm), con-
taining 6% (w/v) polyacrylamide (37.5:1 acrylamide/bis-acrylamide)
and 50X TAE buffer (Tris/acetic acid/EDTA, pH 8.0), were 0.75 mm
thick. A linear gradient from 20% to 60% denaturant was used for all
analyses [18]. (100% denaturant contained 7 M urea and 40% form-
amide). Forty microliters of PCR product was loaded into the gel
which was run at 65 °C at 100 V for 16 h. After electrophoresis the
gels were stained for 50 min with a mix of 1× TAE buffer and Gel
Star (Invitrogen Carlsbad, USA) and photographed using a Kodak
KDS digital camera (Kodak Co., New Haven, CT, USA).
2.6. Scanning electron microscopy (SEM) and image processing
Filter-bed media samples (~ 0.5 ml of grains in an Eppendorf
tube) were immediately fixed using 0.7 ml of Karnovsky solution
and kept at 4 °C until SEM preparation. Samples were prepared for
SEM imaging according to Gamliel [23]. When dry, samples were
carbon coated for 60 s using a Denton Desk Vacuum IV, (LLC Inc.).
From each sample, 3 to 5 grains were examined with a JEOL 840
scanning electron microscope at 4000 × magnification. Counts of
bacterial colonies were made at 5 random sites on single grains and
averaged for 1 mm 2 of grain surface. When the abundance of colo-
nies became too dense to enable discrete counting, the surface area
(mm 2) covered by random colonies was evaluated by measuring
the diameter (μm) of each colony.
2.7. Bacterial production (BP) as bacterial activity (BA)
BP was determined from water samples using the 3H-leucine in-
corporation method [19,20] as modified by Smith and Azam [21].
All samples were run in triplicate with zero time controls. Leucine
 Author's personal copy
122 E. Bar-Zeev et al. / Desalination 286 (2012) 120–130
Fig. 1. Schematic overview of the different sampling locations. First stage; four sampling depths within the fine (100 cm thick) layer: A — surface, B — top mid, C — mid and
D bottom. Second stage; four sampling depths within the crude substrate (100 cm thick,) overlying a fine layer (200 cm) layers: A* — surface, B* — bottom of the crude layer,
C* — top fine layer, D* — bottom of the fine layer.
incorporation was converted to BP (expressed as mg C m − 3 d − 1)
using a factor of 3.1 kg C mol − 1 with an isotope dilution factor of
2.0 [20].
2.8. Chlorophyll a (Chl a) determination
Chl a concentrations are a conveniently measured proxy for algal
biomass, a frequent cause of fouling. Samples (100 ml) were filtered
on Whatman GF/F fiber filters and Chl a was extracted in acetone
(90%) overnight in the dark. Chl a was measured fluorometrically in
duplicate samples according to Holm-Hansen et al. [22].
2.9. Particulate organic carbon (POC)
Samples (500 ml) were filtered onto pre-combusted (4 h, 450 °C)
GF/F filters which were then dried overnight at 60 °C and stored in a
desiccator until further analysis. POC was determined with a CHN
analyzer (Perkin Elmer CHNS/O elemental analyzer, PE 2400) after
carbonate removal from the filters by HCl fumes overnight.
2.10. Dissolved organic carbon (DOC)
Samples were filtered through combusted GF/F filters to remove
particulates and collected in acid-washed, 40 ml glass ampules. HCl
(37%) was added (40 μl) to each ampule which was then stored at
4 °C until analysis by TOC analyzer (Shimadzu ASI-L Autosampler).
Total organic carbon (TOC) was determined as POC + DOC.
2.11. Silt density index (SDI) measurements
SDI of the RSF effluent (filtrate) was measured using a commercial
Multi-Channel On-Line Quality Water Analyzer (MABAT Chemical
Systems Ltd SDI 2200); all the tests were performed according to
the manufacturers' specifications.
2.12. TEP determination
Water samples (100 ml) were filtered gently (~100–150 mbar)
onto 0.4 μm polycarbonate filters and TEP concentrations expressed
as μg Gum Xanthan (GX) equivalents l − 1, were measured according
to Passow and Alldredge [24].
2.13. Estimation of RSF filtration efficiency
The overall efficiency of the RSF in relation to the various water
quality parameters was determined as the difference (ΔT) of POC,
DOC, TOC, Chl a, and TEP, measured at Pre-RSF and at RSF-effluent
sampling locations.
The efficiency of organic carbon filtration within the filter bed was
estimated as the difference between the concentrations of DOC and
POC between intermediate horizontal layers (ΔInt). Each sampling
depth was taken to represent the end-point of a depth layer (During
the first stage: A to D; during the second stage: A* to D*).
3. Results
3.1. Feedwater characteristics
The fluctuations in concentrations of Chl a, POC, DOC and TEP in
the intake seawater over this period are described in detail in
Table 2. In Fig. 2 we also show the variation in temperature, Chl a
and TEP over an annual cycle to indicate the range of fluctuation
that might be expected during routine operation of this desalination
facility.
3.2. Temporal and spatial biofilm dynamics within the RSF
3.2.1. First stage, fine medium only (0 to 42 days)
Close to the onset of the RSF operation (Day 7) no bacterial colo-
nies were present on the fine medium grains (Figs. 3 and 4), similar
to the control (pristine medium grains). The first distinct develop-
ment of coccoid bacterial colonies forming an initial biofilm appeared
after 7 days and increased gradually up to Day 21 (from 0.05 to
1.4 colonies mm − 2). The patches of biofilm that formed on the sur-
faces and in cracks and crevices of the grains were present at all levels
within the filter bed but peaked at depths B and C (mid filter bed).
Bacterial colony abundances ranged from 0.9 to 1.3 colonies mm − 2
(Fig. 4). At this time, DGGE analysis showed only 1 or 2 dominant
bands, suggesting a very limited diversity of the microbial population
composing these biofilm patches (Fig. 5). By Day 27, the numbers of
bacteria and biofilm patches within the grain interstices had in-
creased sharply and were too dense to enable counting of discrete
bacterial colonies. Subsequently, the biofilm coverage of medium
grains was quantified (Fig. 4) as biofilm patch diameter (μm). Peaks
of biofilm coverage were observed at layers B and D with biofilm cov-
erage of 91 and 67 μm, respectively. No changes in the bacterial
 Author's personal copy

E. Bar-Zeev et al. / Desalination 286 (2012) 120–130 123

Table 2
Changes with time in the measured physical and biological characteristics of feedwater (SW Intake) during the RSF maturation period. The biological parameters include
Chlorophyll a (Chl a, proxy for algal biomass), Transparent Exopolymer Particles (TEP), particulate (POC) and dissolved (DOC) organic carbon, and bacterial activity (BA). The
time scale represents the days from the onset of the RSF operation.

Day 7 15 21 27 42 48 56 65 69 79 85

Stage Stage one Stage two

Temp °C 23.9 22.4 21.1 20.6 19 18.6 nd 18.5 18.5 17.7 16.6
Chl a (μg l− 1) 0.8 ± 0.01 2 ± 0.6 2.5 ± 0.7 0.5 ± 0.05 0.3 ± 0.01 0.2 ± 0.03 0.1 ± 0.01 0.4 ± 0.01 0.1 ± 0.01 0.6 ± 0.02 0.6 ± 0.1
TEP (μg GX l− 1) 194 ± 7 465 ± 4 215 ± 7 299 ± 5 258 ± 24 349 ± 24 169 ± 18 402 ± 23 358 ± 36 195 ± 44 334 ± 50
POC (mg l− 1) 0.2 ± 0.01 nd nd 1 0.2 ± 0.02 nd 0.2 ± 0.01 0.2 ± 0.03 nd nd nd
DOC (mg l− 1) 1.6 ± 0.03 nd nd 1.1 ±0.2 0.9 ± 0.04 nd 1.1 ± 0.03 1.3 ± 0.1 nd nd nd
BA (μg C l− 1 d− 1) nd 4.8 ± 0.1 3.9 ± 0.5 5.1 ± 0.2 2.2 ± 1.2 2.3 ± 1 2.5 ± 0.2 7.1 ± 0.2 1.9 ± 0.1 3.7 ± 0.6 7.2 ± 1.1

No available data (nd).

species composing the biofilm were apparent from the DGGE
analyses at this sampling date (Fig. 5).
3.2.2. Second stage: dual media with crude and fine layers (Day 42 to 85)
A 100 cm layer of crude medium was added to the filter bed on
Day 35. In samples taken on Day 42 we observed that the new
crude substrate was still biofilm-free while biofilm coverage over
the fine grain substrate had sharply declined (Figs. 3 and 4). Never-
theless, microbial diversity at the surface layer, A*, as indicated by
DGGE, increased from 1–2 to 4–9 bands with a maximum (9 bands)
(Fig. 5). By Day 48, the biofilm coverage by different types of coccoid
bacteria was well established (71–73 μm), most prominently in the
fine layer (C*, D*). DGGE analysis indicated that although these
coccoid bacteria appeared to be similar morphologically, they were
genetically diverse (Fig. 5). By Day 69 biofilm patches throughout
the entire filter-bed covered most or all of the media grain surfaces.
Coccoid forms appeared mainly in the upper fine layer (A*). For the
first time, numerous filamentous and cone shaped bacteria were
also observed, predominating the deeper (B* to D*) medium layers
(Fig. 3). Overall, as compared to the first stage, bacterial diversity in
the biofilm increased significantly (7–9 bands) throughout the entire
filter bed depth (Fig. 5). Media samples taken on Day 85 (Figs. 3 and
4) were also heavily covered with biofilm in which both filamentous
and coccoid bacteria were prominent. Maximum biofilm coverage
was observed at the deeper media layers (B* to D*) although DGGE
analysis indicated that the most diverse microbial community
(12–16 bands) was situated within the shallow layer (A*).
3.3. RSF filtration efficiency
3.3.1. Overall RSF efficiency
In Table 3 we summarize the concentrations of Chl a, TEP, TOC and
DO in the Pre-RSF water and the overall filtration efficiency (ΔT) of
the RSF in respect to these parameters over the course of this study.
Also shown in Table 3 are the values of SDI and BA measured in the
RSF-effluent on different sampling dates. As noted above, during the
entire sampling period the RSF was subjected to frequent backwash
flushes that undoubtedly affected processes occurring within the
filter bed. Presently we have no data to assess the impact of back-
washes on the overall maturation process. We were careful to sample
routinely each time around 15 h after each backwash cycle.
Over the entire trial period, DO levels were almost unchanged by
passage through the RSF (ΔT values from − 2 to 15). The overall filtra-
tion efficiency for Chl a was expressed in a continuous steady removal
(increasing ΔT) with time resulting in 91% removal efficiency by
Day 85 (Table 3). Initially (Days 7, 15), TEP appeared to be released
(−ΔT) from the filter bed medium. Thereafter ΔT for TEP was positive
but fluctuated considerably reaching a maximum of 91% removal on
Day 56 (Table 3).
A slight reduction (17%) in TOC was observed by Day 7. By the
end of the first stage (Day 27), TOC removal had increased to 37%.
Following the addition of the crude layer, (Day 42) there was a fur-
ther improvement in TOC filtration efficiency (55%). However, by
the final sampling date (Day 65) only a 21% reduction in TOC was
measured (Table 3).
SDI, a key water quality indicator in the desalination industry, was
initially high in the RSF-effluent (5.2 to 6.3) but dropped to levels
generally regarded as acceptable (~3.5) by Day 21 and subsequently
remained relatively stable (Table 3). In contrast, BA in these samples
ranged from tenfold in activity.
3.3.2. Spatial variability of filtration efficiency within the filter bed
To compare between different sampling dates with varying feed-
water concentrations, the concentrations of Chl a, TEP, BP, and POC
and DOC at each sampling point were normalized to those in the
water overlying the RSF (Pre-RSF), taken as 100% (Fig. 6). This figure
shows the spatial variability of Chl a, TEP, and BA within the filter bed
as percentage values relative to the overlying Pre-RSF ambient con-
centrations (see Table 3) taken as 100%. During the first stage, Chl a
was partly removed (average 48%) throughout the entire filter-bed;
subsequently, in the second stage Chl a was effectively filtered (aver-
age 81%) mostly within in the upper crude layer (Fig. 6a). Initially fil-
tration of TEP was ineffective but from Day 48 onwards (except Day
79) increasing amounts of TEP were removed within the surface
and the upper fine layers, A*–C* (Fig. 6b). Bacterial activity (BA) in
the RSF interstitial water showed a sharp increase towards Day 27
in the mid fine layer, C. Following the addition of the crude layer, an-
other peak of BA was recorded towards Day 69 at the bottom of the
crude and fine layers, B* and D* (Fig. 6c).

3.4. Filtration efficiencies of TOC, POC, and DOC, within horizontal layers
of the filter bed

A schematic illustration of the different processes controlling dis-

Fig. 2. Coastal feedwater: Annual changes (2009–2010) of temperature, Chlorophyll a, solved and particulate organic carbon transformations within the
and TEP concentrations. The gray background represents the RSF sampling period. filter-bed of biofiltration systems is shown in Fig. 7. In order to follow
 Author's personal copy
124 E. Bar-Zeev et al. / Desalination 286 (2012) 120–130
Fig. 3. Representative scanning electron microscope images showing the spatial (crude and fine media) and temporal (filter onset, end of the first stage and second stage) fluctuations
of the microbial coverage on the filtration media.
the dynamics of RSF maturation as a biofilter in more detail, we
determined transformations between the different organic carbon
fractions and filtration efficiencies within horizontal layers of the
filtration bed (ΔInt). Removal of organic carbon (increased efficiency)
was classified as positive ΔInt (i.e. concentrations in the upper layer
are higher than in the lower layer) while organic carbon addition (re-
duced efficiency) was registered as negative ΔInt (concentrations in
the upper layer are less than in the lower layer).
The data for organic carbon filtration were limited to 5 sampling
dates only, unfortunately no data were available after Day 65
(Table 3). The changes in POC, DOC and TOC concentrations measured
between intermediate layers of the filter bed on 4 sampling dates are
shown in Fig. 8. Although TOC removal (positive ΔT) was measured
over the entire sampling period, there were considerable fluctuations
of ΔInt POC and ΔInt DOC indicating both removal and addition of
organic carbon within different filter bed layers.
At first (Day 7), both ΔInt POC and ΔInt DOC were low and positive
for all depths (Fig. 8a). Towards the end of Stage 1 (Day 27) both pos-
itive and negative ΔInt values increased (ΔInt values) in different
layers (Fig. 8b). The shallowest layer (A.) was responsible for most
of POC filtration (ΔInt POC = 104 μg C m − 2). By contrast, DOC concen-
trations increased in the intermediate layer (Layer B; ΔInt DOC:
−390 μg C m − 2), but then dropped sharply with depth (Lower mid
layer C; ΔInt DOC = 378 μg C m − 2). Near the beginning of the second
stage, (Day 42), effective reduction of both POC and DOC occurred
in the lower crude layer (B*; ΔInt POC = 111.7 μg C m − 2, ΔInt DOC =
124 μg C m − 2) while in the deeper fine grain layers (C*), ΔInt values
were low indicating diminished organic carbon removal (Fig. 8c).
By Day 65, although there was high DOC removal in the upper fine
layer (C*; ΔInt DOC = 228.6 μg C m − 2), we measured net DOC release
(D*; ΔInt DOC = − 104.3 μg C m − 2) at the bottom of the crude layer
(D*) (Fig. 8d).
 Author's personal copy

E. Bar-Zeev et al. / Desalination 286 (2012) 120–130 125

Fig. 4. Extent of microbial biofilm on media grains. (Quantification of the scanning electron microscope images, see Methods). During the first stage, the striped bars represent the
number of distinct colonies mm− 2, while during the second stage the gray bars represent the mean diameter of the bacterial biofilm coverage. The dashed vertical lines delineate
separate sampling dates as indicated in the x-axis legend. The single black bar emphasizes that at this depth the media was totally covered by microbial biofilm.

4. Discussion
4.1. Feedwater fluctuations
The quality of feedwater flowing into SWRO plants is far from
constant with time (Fig. 2). Israeli coastal waters are characterized
by large seasonal physical and chemical fluctuations coupled to bio-
logical changes [25,26]. Even during the limited time of our study,
which spanned 85 days through autumn to winter, we measured
considerable variability of the inflowing seawater (Table 2), which
undoubtedly affected the RSF maturation process. For example, the
steadily decreasing sea-water temperatures that characterized this
period probably lowered microbial activity and extended the time
for bacterial colonization and biofilm development. The seasonal
changes in algal concentrations (Chl a), TEP and organic carbon in
the intake water may also have impacted RSF filtration efficiency al-
though this remains to be shown.
4.2. Temporal and spatial dynamics of microbial communities within the RSF
In wastewater treatment slow sand filters (SSF) the presence of
well-developed biofilms and associated microorganisms is required
for effective nutrient cycling and biodegradation of organic compounds,
under both in aerobic and anaerobic conditions [27]. Under the condi-
tions of this study, initial bacterial colonies and biofilm structures
were observed within the RSF filter bed medium after about 3 weeks.
However, about three months were required to establish full biofilm
coverage and high bacterial diversity.
The initial stages of “microbial” maturation were characterized by
growth of coccoid bacteria mainly within sheltered interstices of the
filter bed medium grains. At this time meager and sporadic bacterial
colonization of the exposed grain surfaces may have been due to the
high flow velocities and strong shear forces acting on pristine media
grains (Figs. 3 and 4). Generally, pioneer marine biofilm forming col-
onizers were from the Roseobacter or Alteromonasgroups [28–31]
These bacteria have exceptionally high metabolic activity, growth
rates and levels of EPS secretion that enhance surface adhesion and
biofilm formation. Eventually these pioneer biofilm species alter
the surface characteristics of the medium, making it rougher, more
adhesive, and richer in organic material thus stimulating coloniza-
tion by other bacterial species [28,30–32].
In the present study, this process was reflected in the DGGE
results. These results indicated that while only a very limited num-
ber of bacterial species participated (most probably pioneer spe-
cies) in the early colonization of the RSF medium, these early
colonizers may have generated new micro-niches with different

Fig. 5. Spatial and temporal changes of bacterial populations on the filtration media grains as observed by DGGE analysis. Each band represents a distinct bacterial population. White
bars: bands from media grains sampled between the surface to the mid top layer (fine and crude depending on the maturation stage, A, B and A*), Gray bars: bands from grains
between mid and bottom of the top layer (C, D and B*), Dark Gray bars: bands from grains between top to mid crude layer (C*), Black bars: bands from grains between mid to bot-
tom of the crude layer (D*).
 Author's personal copy

126 E. Bar-Zeev et al. / Desalination 286 (2012) 120–130

Table 3
Changes in the concentrations of water quality parameters in the RSF inflow (Pre-RSF) and overall filtration efficiencies, ΔT, during the maturation period*. The time scale (Days)
shows days from onset of RSF operation. The concentrations shown in this table are used in Fig. 6, normalized to 100%.

Pre-RSF concentrations Day 7 15 21 27 42 48 56 65 69 79 85

Filtration efficiencya

Stage Stage one Stage two

Chl a (μg l− 1) 1.1 ± 0.05 0.6 ± 0.2 0.3 ± 0.03 0.07 ± 0.04 0.08 ± 0.01 0.11 ± 0.04 0.13 ± 0.01 0.21 ± 0.01 0.15 ± 0.01 0.3 ± 0.02 0.3 ± 0.02
ΔT, (%) 30 43 49 68 77 69 80 80 84 88 91
TEP (μg GX l− 1) 159 ± 51 226 ± 94 240 ± 55 340 ± 6 539 ± 78 503 ± 46 416 ± 117 639 ± 44 534 ± 7 246 ± 69 712 ± 22
ΔT, (%) − 21 − 98 19 78 34 53.3 91 45 72 12 51
TOC (mg l− 1) 2.36 nd nd 1.42 1.38 nd 1.1 1.6 nd nd nd
ΔT, (%) 17 37 54 13 22
DO (mg l− 1) 6.8 7.7 7.8 8.2 8.5 8.6 8.3 7.9 8.3 8.7 nd
ΔT, (%) −2 8 −5 12 9 15 2 6 0 2
BA (μg C l− 1 d− 1) nd 4.2 ± 0.4 3.4 ± 0.3 0.9 0.9 ± 0.1 8.4 ± 0.3 3.2 ± 0.2 2 ± 0.1 0.7 ± 0.1 4.4 ± 0.1 8 ± 0.6
SDI 5.2 6.3 3.2 3.9 nd 3.2 2.8 3 2.8 2.7 3.3

nd = no available data.
a
The efficiency of the filtration (in %) was determined as: 100 − (100 * Outflow concentration/Pre-RSF concentration).

organic and inorganic characteristics. This would facilitate wider Furthermore, the hydrolytic enzymatic activities of the pio-
bacterial diversity and further microbial colonization and succes- neering bacteria may have liberated dissolved organic material
sion [28,30]. (DOM) from the breakdown of particulate organic material

Fig. 6. Spatial (see Fig. 1) and temporal (first and second stages) changes within the filtration media in concentrations of a. Chl a, b.TEP and c. BA, (given as %). All parameters were
normalized to the Pre-RSF concentrations given in Table 3, taken as 100%.
 Author's personal copy

E. Bar-Zeev et al. / Desalination 286 (2012) 120–130 127

Fig. 7. Schematic illustration of different processes controlling dissolved and particulate organic carbon transformations within the filter bed of biofiltration systems as outlined
in various studies (Azam and Malfatti, 2007 [36]; Battin and Sengschmitt, 1999 [37]; Bouwer and Zehnder, 1993 [4]; Eichinger et al. [38], 2009; Larsen and Harremos, 1994 [6];
Sempere et al., 2000 [39]). Note, that the size of arrows and fonts are proportional to their relative importance. The difference between the concentrations of DOC and POC between
intermediate horizontal layers (U, upper and L, lower) layers was represented as Δint. When the load of organic carbon inflow into any layer was reduced relative to the outflow
from that layer, the Δ was positive. If the concentration of the outflow increased then Δint was negative. The difference between the concentrations of organic carbon entering
the RSF (U) and the concentrations in the filter effluent (F) was represented as ΔT.

(POM) that could attract other bacteria with chemosensory behav-
ior, leading to higher colonization rates by newcomers [29,33].
These processes may have been reflected in our observation of
a steep increase in colony size, diversity and morphology of bacte-
ria (coccoid, cone shaped and filamentous bacteria) that occurred
during the second stage after the addition of the crude layer,
(Figs. 3 and 4).
4.3. Filtration efficiency and biodegradation processes
4.3.1. Bacterial metabolic activity (BA)
BA in interstitial waters of the filter bed is determined by various
parameters such as cell abundance, biomass, nutrients, organic load,
oxygen availability, and water temperature [7,34]. Two maxima of
BA were observed; during the first stage within the fine layer and
within the bottom of crude and fine layers in the second stage
(Fig. 6c). Both peaks occurred within about a month after the start
of each stage, suggesting that this was the length of time required
to build up a critical bacterial biomass within the filter-bed. Dissolved
oxygen measurements indicated that the conditions within the filter-
bed were aerobic throughout the entire filter-bed.
4.3.2. Chlorophyll a (Chl a)
Severe damage can be caused to RO membranes by algal fouling,
either directly (by biomass deposits, silica frustules of diatoms) or in-
directly by release of organic foulants such as TEP [11]. During the
summer and fall period micro and pico-phytoplankton (2–20 and
0.2–2 μm respectively) dominate the Israeli coastal algal populations
with occasional blooms of diatoms.
The sharpest drop in Chl a concentrations occurred consistently
between the intake and pre-RSF sampling points after coagulation
treatment. Thus the RSF media was subjected to relatively low algal
loads (0.07 to 1.13 μg Chl a l − 1) which were removed with increasing
efficiency during the maturation period (Table 3). Filtration of Chl a
improved with the addition of the second media layer and probably
also by degradation on biofilm surfaces by increasingly diverse
 Author's personal copy

128 E. Bar-Zeev et al. / Desalination 286 (2012) 120–130

Fig. 8. Spatial and temporal changes in dissolved, particulate, and total organic carbon within intermediate horizontal layers of the RSF (ΔInt), overall efficiency of the RSF ΔT, and
bacterial activity at a. First stage onset (Day 7), b. End of the first stage (Day 27), c. Second stage onset (Day 42) and d. Second stage (Day 65). Note no BA was measured at day 7.

bacterial populations. This might account for the greatest reduction in
Chl a concentrations being recorded in the upper media layers (A, A*)
where BA was the highest (Fig. 6a, c).
4.3.3. Transparent exopolymer particles (TEP)
Increasing evidence indicates that TEP is potentially a significant
contributor to biofouling in the desalination industry [11,12,14,35].
Thus, it is important to determine the efficiency of pretreatment
technologies in lowering TEP concentrations. Our study showed
two distinct trends regarding TEP concentrations in the RSF filtrate
during the maturation period. Initially, from 0 to 15 days, TEP con-
centrations in the RSF effluent increased from pre-RSF levels mainly
at the deeper media depth (C,D) and remained high in the filtrate
(Table 3, Fig. 6b). During this period, it seems likely that very little
of the inflowing TEP was removed by mechanical size exclusion in
the immature RSF. Indeed, the rise in TEP observed in the RSF efflu-
ent (Days 8, 16) may indicate that TEP was being generated from
“colloidal TEP” precursors [10] in the feedwater as it flowed through
the filter bed.
After the microbial biofilm was established towards the end of
the first stage, a significant improvement in TEP filtration efficiency
(60% reduction in RSF-effluent) was observed (Table 3, Fig. 6b). The
increased TEP removal was presumably related to widespread bio-
film coverage on the filter bed grains and increased diversity of bac-
teria with a greater plasticity of metabolic activity (Table 3, Fig. 6b).
Fluctuations in the efficiency of TEP removal during the second stage
may have been caused by the interplay of TEP/EPS being sloughed off
from mature biofilm and, at the same time TEP being degraded by the
active bacterial community.
4.3.4. POC, DOC and TOC cycling in RSF filter bed during the maturation
process
Moderate amounts of TOC, ranging from 13 to 54% of Pre-RSF con-
centrations were removed by the RSF. Net removal of TOC was presum-
ably achieved by a combination of adsorption and mechanical
adsorption and the uptake and respiration of DOC and through
the activity of microbial extracellular enzymes that solubilized
POC to DOC as outlined schematically in Fig. 7 [5,7,36–39]. Fur-
thermore, in contrast to more mature biofilms, the developing bio-
films may have released some labile DOC to be taken up by other
microorganisms, stimulating overall microbial proliferation [7,38].
The complexity of processes acting on POC and DOC occurring
within adjacent depth layers of the filter bed was reflected by consid-
erable fluctuations of ΔInt POC and ΔInt DOC showing both removal
 Author's personal copy
E. Bar-Zeev et al. / Desalination 286 (2012) 120–130
(positive Δ) and release (negative Δ) of particulate and dissolved
carbon within the filter bed layers (Fig. 7). On Day 7, with only lim-
ited bacterial colonization and biofilm development in the filter
bed medium, POC and DOC were removed mainly by mechanical
exclusion and adsorption in the fine layer (Fig. 8a). At this stage,
only small fluctuations in POC and DOC (positive and negative ΔInt)
were measured within the filter bed, resulting in low TOC removal
(17% of the feedwater).
Towards the end of the first stage (Day 27) considerable bacte-
rial growth and biofilm were measured at the mid and bottom
substrate layers (B,D) together with relatively high rates of BA
(Figs. 3, 6c, 8b). POC was removed in the upper fine layer (A)
probably via mechanical exclusion, physical adsorption and by
microbial bio-degradation. Microbial activity in the upper fine
layer (A) may have solubilized POC, resulting in an increase
(− ΔInt) in the DOC pool. In the mid layer (B), where BA rates
were maximal, more POC was removed by biological degradation
and a significant DOC fraction accumulated (Fig. 8b), both from
direct degradation of POC to DOC and from release of DOC by pro-
liferating biofilms [7,38]. Subsequently, in the lower mid layer (C),
active microbial uptake and/or mineralization may have caused an
effective removal of the suspended DOC. Thus, by day 27, the com-
bination of physical and biological processes within the RSF re-
duced TOC in the filtrate by 37% in comparison with overlying
feedwater.
Mechanical exclusion and adsorption of POC and DOC were also
the dominant mechanisms following the addition of the new crude
substrate layer (Day 42). At this time, the low bacterial activity mea-
sured within the sand layers may have been responsible for the
lower removal values of POC and DOC that were measured within
the fine layer (C*,D*) (Figs. 3, 6c, 8c). Despite the apparent low
microbial activities within the filter bed after 42 days, the overall
efficiency of TOC removal increased to 55%.
By late phase of the maturation period (Day 65), POC appeared
to accumulate within the entire RSF filter bed. This may have been
a result of high concentrations of TEP and DOC in the feedwater
overlying the RSF (Table 3, Fig. 8d). Although some POC and DOC
were removed by the upper crude substrate layer (A*), it appears
that the DOC and TEP loading served as an available nutrient source
for bacterial growth and resulted in the rapid release and high con-
centrations of POC (negative ΔInt) measured at the bottom of the
crude substrate layer (B*). By contrast, effective biofiltration by a
dense mature biofilm, and possibly some mechanical exclusion
and adsorption, may have been responsible for the significant re-
moval (positive ΔInt) of TOC that was observed in the upper fine
layer (C*). Within the fine substrate layer (D*), DOC values were
slightly enriched (negative ΔInt), possibly due to POC degradation
or breakdown by cell lysis or solubilization. Although the overall ef-
ficiency of TOC removal by the RSF on Day 65 was markedly lower
than on the Day 42, we note that feedwater loads of TEP (part of
the POC pool) and TOC were significantly higher than previously.
Possibly higher levels of organic loading tended to reduce the biofil-
tration efficiency.
Nevertheless, the finding that net TOC values were consistently
lower in the RSF effluent than in pre-RSF water (positive ΔTTOC) in-
dicates that the RSF was indeed functioning as a biofilter.
5. Conclusions
This study on a newly operational RSF, clearly showed that a func-
tional biofilm was developed by diverse microbial organisms within
the filter bed and functioned to breakdown organic material; thus
confirming the biofiltration potential of theses filters.
Over the 85 day maturation period, we observed increasing bio-
film coverage together with a rise in bacterial species diversity on
the grains of the filter bed media that broadly coincided with
129
increases in the RSF filtration efficiency in respect to Chl a, TEP, and
TOC. However, it appears that the present RSF configuration and per-
haps filter bed materials are far from optimal. We suggest that consid-
erable improvements in the efficiency of organic load removal can be
achieved by enhancing the biofiltration capacity of these filters. An
important challenge lies in designing RSF's to optimize their potential
as effective microbiological filters capable of significantly reducing
both particulate and dissolved organic matter load in the water
reaching RO membranes. This approach could minimize the use of
chemical additives to feedwater and would thus enable an important
improvement in SWRO pretreatment technology.
Acknowledgments
We thank I.D.E. for partial funding and for access to the Hadera
desalination Plant. We also thank Niv Schechter for assistance in sam-
pling and SEM analysis. This research is part of the MSc (Natalia
Belkin) and PhD (Edo Bar-Zeev) requirements at Bar Ilan University.
Edo Bar Zeev is supported by a President's Scholarship from Bar Ilan
University.
References
[1] C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, State of the art of reverse
osmosis desalination, Desalination 216 (2007) 1–76.
[2] S. van Hoof, A. Hashim, A.J. Kordes, The effect of ultrafiltration as pretreatment
to reverse osmosis in wastewater reuse and seawater desalination applications,
Desalination 124 (1999) 231–242.
[3] G. Bablon, W.D. Bellamy, M.M. Bourbigot, F.B. Daniel, M. Dore, Fundamental
Aspects. IN: Ozone in Water Treatment: Application and Engineering, Lewis
Publishers, Chelsea, Michigan, 1991.
[4] E.J. Bouwer, A.J.B. Zehnder, Bioremediation of organic compounds — putting
microbial metabolism to work, Tren. Biotech. 11 (1993) 360–367.
[5] H. Sontheimer, R. Gimbel, Floculation–filtration in drinking-water treatment,
J. Water Wastewater Res. 12 (1979) 147–155.
[6] T.A. Larsen, P. Harremoes, Degradation mechanisms of colloidal organic matter in
biofilm reactors, Water Res. 28 (1994) 1443–1452.
[7] C. Arnosti, B.B. Jorgensen, J. Sagemann, B. Thamdrup, Temperature dependence
of microbial degradation of organic matter in marine sediments: polysaccharide
hydrolysis, oxygen consumption, and sulfate reduction, Mar. Ecol. Prog. Ser. 165
(1998) 59–70.
[8] T. Berman, B. Liberman, Analysis and monitoring: MSC — a biologically oriented
approach, Filtr. Sep. 43 (2006) 39–40.
[9] L.O. Villacorte, R. Schurer, M.D. Kennedy, G.L. Amy, J.C. Schippers, The fate of trans-
parent exopolymer particles (TEP) in seawater UF–RO system: a pilot plant study
in Zeeland, The Netherlands, Desalin. Water Treat. 13 (2010) 109–119.
[10] E. Bar-Zeev, I. Berman-Frank, B. Liberman, E. Rahav, U. Passow, T. Berman, Trans-
parent exopolymer particles: potential agents for organic fouling and biofilm
formation in desalination and water treatment plants, Desalin. Water Treat. 3
(2009) 136–142.
[11] T. Berman, R. Mizrahi, C.G. Dosoretz, Transparent exopolymer particles (TEP): A
critical factor in aquatic biofilm initiation and fouling on filtration membranes,
Desalination 276 (2011) 184–190.
[12] A.L. Alldredge, U. Passow, B.E. Logan, The abundance and significance of a class
of large, transparent organic particles in the ocean, Deep-Sea Res. I 40 (1993)
1131–1140.
[13] T. Berman, M. Holenberg, Don’t fall foul of biofilm through high TEP levels, Filtr.
Sep. 42 (2005) 30–32.
[14] N. Belkin, E. Bar-Zeev*, T. Berman, I. Berman-Frank, Two innovative devices for
depth sampling in granular filtration systems, Desalination 286 (2012) 115–119.
[15] R. Massana, A.E. Murray, C.M. Preston, E.F. DeLong, Vertical distribution and
phylogenetic characterization of marine planktonic Archaea in the Santa Barbara
Channel, Appl. Environ. Microbiol. 63 (1997) 50–56.
[16] T. Brinkhoff, G. Muyzer, Increased species diversity and extended habitat range of
sulfur-oxidizing Thiomicrospira spp, Appl. Environ. Microbiol. 63 (1997) 3789–3796.
[17] G. Muyzer, K. Smalla, Application of denaturing gradient gel electrophoresis
(DGGE) and temperature gradient gel electrophoresis (TGGE) in microbial ecology,
Antonie Leeuwenhoek 73 (1998) 127–141.
[18] G. Muyzer, E.C. Dewaal, A.G. Uitterlinden, Profiling of complex microbial-
populations by denaturing gradient gel-electrophoresis analysis of polymerase
chain reaction-amplified genes-coding for 16S ribosomal-RNA, Appl. Environ.
Microbiol. 59 (1993) 695–700.
[19] D. Kirchman, E. Knees, R. Hodson, Leucine incorporation and its potential as a
measure of protein-synthesis by bacteria in natural aquatic systems, Appl. Environ.
Microbiol. 49 (1985) 599–607.
[20] M. Simon, F. Azam, Protein-content and protein-synthesis rates of planktonic
marine-bacteria, Mar. Ecol. Pro. Ser. 51 (1989) 201–213.
[21] D.C. Smith, F. Azam, A simple, economical method for measuring bacterial protein syn-
thesis rates in seawater using 3H-leucine, Mar. Microb. Food Webs 6 (1992) 107–114.
 Author's personal copy
130 E. Bar-Zeev et al. / Desalination 286 (2012) 120–130
[22] O. Holm-Hansen, C.J. Lorenzen, R.W. Holmes, J.D.H. Strickland, Fluorometric
determination of chlorophyll, ICES J. Mar. Sci. 30 (1965) 3–15.
[23] H. Gamliel, Optimum fixation conditions may allow air drying of soft biological
specimens with minimum cell shrinkage and maximum preservation of surface
features, Scann. Electr. Micro. 4 (1985) 1649–1664.
[24] U. Passow, A.L. Alldredge, A dye-binding assay for the spectrophotometric mea-
surement of transparent exopolymer particles (TEP), Limnol. Oceanogr. 40 (1995)
1326–1335.
[25] O. Hyams-Kaphzan, A. Almogi-Labin, C. Benjamini, B. Herut, Natural oligotrophy vs.
pollution-induced eutrophy on the SE Mediterranean shallow shelf (Israel): environ-
mental parameters and benthic foraminifera, Mar. Poll. Bull. 58 (2009) 1888–1902.
[26] Z. Rosentraub, S. Brenner, Circulation over the southeastern continental shelf
and slope of the Mediterranean Sea: direct current measurements, winds, and
numerical model simulations, J. Geophys. Res. 112 (2007) 1–21.
[27] L. Mendoza-Espinosa, T. Stephenson, A review of biological aerated filters (BAFs)
for wastewater treatment, Environ. Eng. Sci. 16 (1999) 201–216.
[28] C.R. Jackson, P.F. Churchill, E.E. Roden, Successional changes in bacterial
assemblage structure during epilithic biofilm development, Ecology 82 (2001)
555–566.
[29] H. Dang, C.R. Lovell, Bacterial primary colonization and early succession on
surfaces in marine waters as determined by amplified rRNA gene restriction
analysis and sequence analysis of 16S rRNA genes, Appl. Environ. Microbiol. 66
(2000) 467–475.
[30] C.R. Jackson, Changes in community properties during microbial succession,
Oikos 101 (2003) 444–448.
[31] E. Lyautey, C. Jackson, J. Cayrou, J.-L. Rols, F. Garabétian, Bacterial community
succession in natural river biofilm assemblages, Microb. Ecol. 50 (2005) 589–601.
[32] H. Ivnitsky, I. Katz, D. Minz, G. Volvovic, E. Shimoni, E. Kesselman, R. Semiat, C.G.
Dosoretz, Bacterial community composition and structure of biofilms developing
on nanofiltration membranes applied to wastewater treatment, Water Res. 41
(2007) 3924–3935.
[33] H.P. Grossart, T. Kiorboe, K. Tang, H. Ploug, Bacterial colonization of particles:
growth and interactions, Appl. Environ. Microbiol. 69 (2003) 3500–3509.
[34] C. Gallert, J. Winter, Bacterial metabolism in wastewater treatment systems, in:
H.J. Jördening, J. Winter (Eds.), Wiley-VCH Verlag GmbH & Co., KGaA, Weinheim,
2005.
[35] L.O. Villacorte, M.D. Kennedy, G.L. Amy, J.C. Schippers, The fate of transparent
exopolymer particles (TEP) in integrated membrane systems: removal through
pre-treatment processes and deposition on reverse osmosis membranes, Water
Res. 43 (2009) 5039–5052.
[36] F. Azam, F. Malfatti, Microbial structuring of marine ecosystems, Nat. Rev. Microbiol.
5 (2007) 782–791.
[37] T.J. Battin, D. Sengschmitt, Linking sediment biofilms, hydrodynamics, and river
bed clogging: evidence from a large river, Microb. Ecol. 37 (1999) 185–196.
[38] M. Eichinger, S.A.L.M. Kooijman, R. Sempere, D. Lefevre, G. Gregori, B. Charriere,
J.C. Poggiale, Consumption and release of dissolved organic carbon by marine bac-
teria in a pulsed-substrate environment: from experiments to modeling, Aquat.
Microb. Ecol. 56 (2009) 41–54.
[39] R. Sempere, S.C. Yoro, F. van-Wambeke, B. Charriere, Microbial decomposition of
large organic particles in the northwestern Mediterranean Sea: an experimental
approach, Mar. Ecol. Pro. Ser. 198 (2000) 61–72.