biologically mediated particle removal
mechanism can be rapidly and reversibly
inhibited with azide (Weber-Shirk & Dick,
1997a). The reversible effect of azide led to
the hypothesis that bacterivores are
responsible for removal of a significant
fraction of bacteria, for temporary exposure
to azide would be expected to prevent
feeding by bacterivores but not particle
removal by biofilms.
The potential role of bacterivory in slow
sand filters was reported by Burman and
Lewin (1961); however, the possibility first
was investigated by Lloyd (1973). Two
mechanisms by which predators may assist
in the filtration process have been proposed.
The most commonly proposed mechanism is
that predators graze on bacteria and detritus
attached to sand grains (Huisman & Wood,
1974; Poynter & Slade, 1977). The second
proposed mechanism is that suspension
feeding predators remove suspended
particles as the particles flow through the
filter (Lloyd, 1973). Predators that graze on
attached bacteria potentially free up sites for
future bacteria attachment while suspension
feeding predators directly remove particles
from the mobile phase.
The population of protozoa present in slow
sand filters has been enumerated by Lloyd
and Richards. They suggested a relationship
between bacterial removal and the number
of Vorticella (Lloyd, 1973) or flagellates and
ciliates (Richards, 1974) in the filter. They
did not show, however, that the protozoa
were necessary for, or capable of, significant
bacterial removal by slow sand filters. Here,
we describe methods used to detect, isolate,
and culture a specific bacterivore and
demonstrate the capability of the bacterivore
to produce the observed bacterial removal.
BACTERIVORY BY A
CHRYSOPHYTE IN SLOW SAND
FILTERS
M. L. WEBER-SHIRK AND R. I. DICK
School of Civil and Environmental
Engineering, Cornell University, Ithaca, NY
14853
Abstract—Bacterivory previously was
shown to be responsible for significant
removal of bacteria in slow sand filters. This
research was designed to identify the
responsible bacterivores and to evaluate
their ability to remove a significant fraction
of bacteria. A small (3-µm diameter)
chrysophyte was isolated from slow sand
filter effluent. The ability of a pure culture
of the chrysophyte to rapidly ripen a slow
sand filter was demonstrated.
Key Words—slow sand filtration, bacteria,
protozoa,
chrysophyte,
Vorticella,
heterotrophic, nanoflagellate3
INTRODUCTION
Slow sand filters have been used to treat
public drinking water supplies since
1829(Baker, 1981). They commonly were
used in the cities of Europe and the eastern
United States around the turn of this century
and currently are used in Amsterdam
(Rittmann & Huck, 1989), London (Ellis,
1985), Paris (Bonnet, et al., 1992), and
Zurich (Aeppli, 1990). They have received
renewed interest in the U. S. during the last
ten years due, in part, to increasingly
stringent surface drinking water supply
regulations mandated by the Environmental
Protection Agency.
In two recent papers we have established
that physical-chemical particle removal
mechanisms in slow sand filters are
enhanced by particles retained in the filters
(Weber-Shirk & Dick, 1997b) and that a
MATERIALS AND METHODS
Filtration apparatus
A schematic drawing of the experimental
apparatus is shown in Figure 1. The filter
1
2
M ain feed (Cayuga Lake water) from
batch tank at 785 ml/h for a filtration
approach velocity of 10 cm/h
M anometer/surge tube
M anifold/valve block
Open to atmosphere
Auxiliary feeds
each 0.67% of
main feed
Peristaltic
pumps
Sampling Chamber
To waste
Sampling tube
Lower to collect sample
1 liter
E. coli
feed
1 liter
Ps.
putida
feed
10 cm I.D. filter cell with 18 cm of
0.17 mm diameter glass beads
Figure 1. Schematic
experimental apparatus.
drawing
of
the
bed depth was reduced from the
conventional bed depth of 1 m because
previous research with the filtration
apparatus had shown that only insignificant
bacterial removal occurred below 12 cm
(Weber-Shirk, 1992). The primary feed was
either water from Cayuga Lake (located in
central NY) or a synthetic feed (described
below). Auxiliary feed lines were used to
add bacteria to the main feed. The filter cells
and feed lines were behind a black plastic
curtain to minimize photosynthesis. The
laboratory temperature was 23°C ± 2°C.
Sampling procedure
When filter influent was sampled, flow
through the filter unit stopped, and thus
influent sampling potentially affected filter
performance. This problem was minimized
by always sampling the effluent before
sampling the influent and by only sampling
once per day. Thus, any effects of
temporarily stopping the flow were greatly
diminished before subsequent sampling. E.
coli and particle count sample volumes
generally were 10 mL and the flow rate
through the filters was 13 mL/min; thus the
flow through each filter unit was stopped
only briefly while influent samples were
taken.
Filter performance evaluation
Filter performance was evaluated based on
Escherichia coli and particle removal. E.
coli were chosen as test particles because
they do not multiply significantly under the
low nutrient conditions of slow sand filters
and because their removal is one of the
objectives of water treatment. The average
size of the E. coli as determined with an
electronic
particle
counter
(Coulter
Multisizer II) was approximately 1 µm in
diameter. E. coli were added to all filters at
an influent density of approximately 4/L.
The method of adding a relatively constant
concentration of E. coli to filter influent was
described previously (Weber-Shirk & Dick,
1997b).
The E. coli were enumerated in the filter
influent and effluent using the membrane
filtration technique as described in Standard
Methods for the Examination of Water and
Wastewater (Greenberg, et al., 1992).
Duplicate membrane filter tests were
conducted on each sample. An attempt was
made to place slightly less than 80 E. coli on
each membrane filter by adjusting the
sample volume. Die off did not contribute
significantly to E. coli inactivation or
removal as demonstrated by observing that
the E. coli reduction in viable E. coli
concentration through a filter containing
clean glass beads was less than 20%.
Particle counts
Particles (including bacteria and other
particles) in the 0.75 to 20 µm diameter
range were sized and counted using the
electronic particle counter. Two mL of each
of the samples were transferred into particlefree vials and diluted with Isoton II (Coulter
Diagnostics) to a final volume of 20 mL.
Three 500 µL samples were analyzed from
each vial to verify that results were
consistent.
3
Synthetic feeds
Two different synthetic feeds were used. The
first was developed from Cayuga Lake water
which had been subjected to slow sand
filtration and then filtered through a 0.2 µm
diameter pore size serial nylon membrane
filter cartridge (Cole-Parmer 06479-16). The
second synthetic feed was prepared by
adding trace elements to distilled water. The
trace elements were added according to the
concentrations used in Fraquil, a synthetic
water designed to mimic the trace element
composition of fresh waters (Morel, et al.,
1975). However, vitamins (also used in
Fraquil) were not added. The modified
Fraquil (i.e. not containing vitamins) was
filtered through an activated carbon
cartridge to remove dissolved organic
carbon and through a 0.2 µm diameter pore
size serial nylon membrane filter cartridge to
remove particulate matter.
Pseudomonas putida was added to each of
the synthetic feeds to produce influents with
known bacterial concentrations using
auxiliary feed lines (Figure 1). The target
influent concentration of Ps. putida was
either 40 or 500/µL depending on the
experiment. Pseudomonas putida was
cultured in nutrient broth (Becton Dickinson
Microbiology Systems) at 25°C. A flask
containing 300 mL of nutrient broth was
inoculated from a Ps. putida streak plate and
placed on a shaker at 25°C for
approximately 24 h. The culture was then
spun down with a centrifuge (Beckman J221) at 3,800 g for 10 min and washed twice
with buffered distilled water. The cells were
resuspended in 100 mL of buffered distilled
water and the total number of cells was
determined using the electronic particle
counter. The effective diameter of the
Ps. putida as determined with the electronic
particle counter was approximately 1 µm.
Fluorescent Labeled Bacteria
Fluorescent-labeled bacteria were prepared
from a pure culture of Ps. putida. The
bacteria were washed with buffered distilled
water to remove dissolved organic carbon.
Approximately 3 x 1011 washed cells were
stained in 50 mL of 0.03% acridine orange.
After staining for 10 min, the cells were
centrifuged (3,800 g for 10 min) and washed
twice with buffered distilled water to
remove excess acridine orange.
Chrysophyte Culture
A mixed culture of heterotrophic
nanoflagellates (HNF) was obtained by
culturing the effluent of a slow sand filter
receiving Cayuga Lake water. The mixed
culture was fed 2 x 104 Ps. putida µL-1 d-1.
An unispecific culture of a chrysophyte
(containing Ps. putida and the chrysophyte)
was obtained from the mixed culture by
serial dilution (Cowling, 1991). The
subcultures were prepared in particle free
vials containing 10 mL modified Fraquil and
2 x 108 Ps. putida. The vials were stirred
gently by an orbital shaker and maintained at
23°C. The samples were visually turbid after
addition of the bacteria. Within 2 days, a
decrease in turbidity in cultures containing
protozoans was evident. A culture from the
most dilute inoculum that exhibited reduced
turbidity was selected as the source of the
chrysophyte for further study. Microscopic
examination revealed that the culture
contained only a few bacteria and the
chrysophyte. This unispecific culture was
used to grow a large number of the
chrysophytes for use as a filter inoculum.
One liter of modified Fraquil containing 1 x
1011 Ps. putida was inoculated with 10 mL
of the unispecific culture of the chrysophyte.
The flask containing the culture was stirred
gently with a magnetic stirrer. After 1.7 days
the culture was sampled for examination
with the microscope and enumeration with
the Coulter Multisizer. The chrysophyte
culture had an effective modal particle
diameter of approximately 1.7 µm as
determined using the electronic particle
4
counter (Figure 2) and an average diameter
of 3 µm as measured microscopically. The
modal diameter determined using the
Coulter Multisizer was smaller than that
determined using the microscope due to the
effect of the salt solution on the chrysophyte.
Changes in cell size also have been observed
by researchers using aldehyde fixatives
(Sherr & Sherr, 1991). The peak below 1.3
µm (partially shown in Figure 2) probably
was due to the Ps. putida in the culture.
4000
3500
3000
2500
2000
1500
1000
500
0
1
1.5
2
2.5
3
Particle diameter (µm)
3.5
Figure 2. Particle size distribution of
chrysophyte culture in saline solution as
determined by electronic particle counter.
RESULTS AND DISCUSSION
Results from previous investigations using
sodium azide to inhibit oxidative
phosphorylation (Weber-Shirk & Dick,
1997a) were consistent with the hypothesis
that bacterivory was a significant cause of
bacterial removal in slow sand filters. The
investigations reported here were extensions
of the earlier work designed to detect and
evaluate the potential roles of protozoa in
slow sand filters. Three complementary
techniques were used to detect the protozoa.
The first technique was to selectively stain
bacterivores within the filter bed using
fluorescent labelled bacteria. This technique
led to the detection of protozoans larger than
about 10 µm in diameter. The second
technique was to count particles in filter
effluent, and the third technique was to
microscopically observe samples taken from
filter beds.
Detection of a Vorticella spp.
The first technique for assaying protozoans
was applied to a filter that had been ripened
using the synthetic feed obtained by filtering
Cayuga Lake water. Pseudomonas putida
(40/µL) were added to the filter influent as
food for bacterivores, but bacterial growth in
the feed lines increased the average number
of influent bacteria-sized particles (as
measured with the electronic particle
counter) to 120/µL. After 4 weeks of
operation, a 1-h pulse of fluorescent labeled
bacteria was fed to the filter. The
fluorescent-labeled bacteria were fed to the
filters at an influent concentration of
4,000/µL.
Qualitative information on the types of
bacterivores in the filter bed was established
using a Zeiss Universal Research
microscope. The filter was disassembled and
glass beads from various depths were placed
on microscope slides. When samples were
viewed using epifluorescence (excitation at
450-490 nm and beam splitter at 510 nm) at
a magnification not requiring a cover slip or
oil immersion (less than 320x), the most
frequently observed organisms were
Vorticella spp. However, even at the low
magnification used, the depth of focus was
less than the diameter of the glass beads and
a significant volume of the glass bead
preparation was hidden from view beneath
the glass beads. Because of this, no attempt
was made to determine the number of
Vorticella spp. within the filter.
Detection of a heterotrophic nanoflagellate
E. coli concentrations were observed to
decline more rapidly in effluent from slow
sand filters than in protozoan-free control
samples. This observation suggested
bacterial predators were contained in the
filter effluent. Coulter Multisizer analysis of
5
slow sand filter effluent as ripening
progressed showed emergence of a 1.7 µm
diameter particle (Figure 3) as bacteria-sized
particle concentrations decreased (data not
shown). Given the reduction in cell size due
to the high ionic strength required for
Coulter Counter analysis, it was considered
that this particle might be a bacterial
predator. However, the concentration of the
particle in the filter effluent was too low to
allow direct microscopic verification.
9
8
day 2
7
day 3
6
day 4
5
day 5
4
day 6
3
2
1
0
1.5
1.6
1.7
1.8
Particle diameter (µm)
1.9
2
Figure 3. Particle concentrations in effluent
from a filter fed Cayuga Lake water.
Two methods were used to obtain samples
with higher concentrations of these effluent
particles. The first was to add predator food
source in the form of 2 x 104 Ps. putida µL–
1 –1
d to a 500 mL effluent sample from a
filter receiving unmodified Cayuga Lake
water. Microscopic examination of the
resulting culture after 2 d revealed a
population of protozoa dominated by a
heterotrophic nanoflagellate (a chrysophyte)
approximately 3 µm in diameter. Although
the chrysophyte numerically dominated the
culture, it was recognized that the culture
conditions could have selected for a
different organism than was dominant in the
filter.
The second method was to backwash
resident organisms from a filter that had
been receiving Cayuga Lake water and the
E. coli supplement. The filter cake (a layer
of fine particles that forms at the surface of
the filter bed) was removed prior to
backwashing. Microscopic examination of
the filter cake revealed few organisms
relative to the numbers in the sand
immediately below the filter cake. The filter
was backwashed with distilled water while
being vigorously agitated to dislodge
attached matter from the medium. The first
liter of the backwash water was centrifuged
(3800 g for 10 min) and the concentrated
sample was examined microscopically.
Microscopic examination of the particles in
the backwash water indicated that the
sample also was dominated by a chrysophyte
(Figure 4). Many of the chrysophytes were
attached to debris presumably as they had
been in the filter column. Although it is
generally not possible to identify
nanoflagellates to the species level using
light microscopy, the organisms in the
backwash water appeared to be the same
chrysophyte obtained by culturing filter
effluent.
Figure 4. Photomicrograph of 3 chrysophytes
attached to debris. The sample was obtained by
backwashing a slow sand filter.
The
colorless
chrysophyte
was
approximately spherical in shape, 2.0–4.3
6
µm in diameter, with an average diameter of
about 3.0 µm. The difference between size
as measured by the light microscope (3.0
m) and the electronic particle counter (1.7
m) was due to the high ionic strength of the
solution used for electronic particle
counting. The chrysophyte possessed a long
flagellum oriented anteriorly and a short
second flagellum. These characteristics are
indicative of the genera Spumella and
Paraphysomonas (Patterson, 1992), but the
presence or absence of scales on the cell
surface (diagnostic for Paraphysomonas)
was not determined.
Bacterial removal by a chrysophyte
The ability of the chrysophyte to enhance
bacterial removal in a slow sand filter was
tested by feeding a chrysophyte culture
(isolated from slow sand filter effluent)
x
containing approximately 2
109
organisms (in 4 liters) to a new filter during
the first 5 h of a filter run. The filter
performance was compared with that of a
control filter that did not receive the
chrysophyte culture. Both filters were fed
modified Fraquil amended by the addition of
500 Ps. putida/µL. The addition of the
chrysophyte markedly enhanced removal of
E. coli (Figure 6). The filter that had
received the chrysophyte inoculum achieved
excellent (99.7%) removal of E. coli by the
time the first post-inoculum sample was
taken. The control filter removed 99% of E.
coli approximately 2 days later than the filter
receiving the chrysophyte inoculum. The
ability of the control filter to remove E. coli
in time may have been indicative of the
presence of the chrysophyte in the
experimental apparatus. The apparatus was
not designed to be sterilized and thus small
numbers of the chrysophyte likely were
present in feed lines or in the filter cell and
were able to populate the filter bed when
presented with an adequate supply of
bacteria.
50
45
day 1
40
35
day 2
30
day 3
25
day 4
20
15
10
5
0
1.25 1.5 1.75
2
2.25 2.5 2.75 3 3.25 3.5 3.75
Particle diameter (µm)
4
Figure 5. Effluent particle concentrations from
a filter that received a large inoculum of the
chrysophyte.
1
Control
Chrysophyte
inoculum
0.1
0.01
0.001
0
1
2
Time (days)
3
4
Figure 6. Removal of E. coli in a filter that
received a chrysophyte inoculum compared
with removal in a control filter.
Effluent particle size distributions again
revealed the peak at 1.7 µm characteristic of
the chrysophyte. The filter receiving the
chrysophyte inoculum showed a decrease in
1.7 µm diameter particles from day 1 to day
4 (Figure 5). The peak around 1.7 µm
corresponds to the size of the chrysophyte as
measured by the electronic particle counter
in the original inoculum (Figure 2). The
decrease in peak area from day 1 to 4
7
(Figure 5) may correspond to the decrease in
E. coli removal during the same time (Figure
6). The control filter did not have a
significant peak at 1.7 µm until day 4
(Figure 7) corresponding to the beginning of
effective E. coli removal (Figure 6).
Bacterivory by heterotrophic
nanoflagellates in other environments
In the past 20 years the role of heterotrophic
nanoflagellates (2 – 20 µm in diameter) as
bacterial consumers has been investigated
and verified in many aquatic environments.
Heterotrophic nanoflagellates (HNF) are
omnipresent in aquatic environments
(Fenchel, 1982a) and often have been shown
to be dominant consumers of bacteria. HNF
are the primary consumers of picoplanktonic
(<2 µm) microorganisms in a variety of
aquatic environments (Berninger, et al.,
1991a; Fenchel, 1982c; Holen & Boraas,
1991). The role of HNF as dominant
bacterivores in aquatic food webs is related
to their ability to feed on bacteria-sizedparticles (often <1 µm) more effectively
than most other organisms (Berninger, et al.,
1991a). HNF are the most important
bacterivores in pelagic and benthic
environments (Bak, et al., 1991), and in
freshwater lakes, ponds, bogs, and rivers
(Barcina, et al., 1991; Berninger, et al.,
1991b; Carlough & Meyer, 1991; Finlay, et
al., 1988). The terrestrial environment also
harbors many flagellates (Patterson &
Larsen, 1991).
Enumeration of heterotrophic nanoflagellate
The technical difficulties of enumerating
HNF in porous media have been
documented in detrital and benthic
ecosystems (Patterson & Larsen, 1991).
Quantification of HNF attached to filter
medium is not feasible using standard
microscope techniques especially when the
media diameter is much larger than the
depth of focus. It may be possible to use the
backwash technique described above to
50
45
day 1
40
35
day 2
30
day 3
25
day 4
20
15
10
5
0
1.25 1.5 1.75
2
2.25 2.5 2.75 3 3.25 3.5 3.75
Particle diameter (µm)
4
Figure 7. Effluent particle concentrations from a
control filter.
quantify HNF. However, it will be necessary
to demonstrate that the backwash technique
removes most HNF from the filter medium.
The presence of debris in backwash water
sample precludes use of electronic particle
counters to enumerate HNF and the
propensity of HNF to attach to surfaces
including
debris
will
complicate
microscopic
enumeration.
Fluorescent
staining of backwash samples may facilitate
microscopic enumeration.
The only known count of flagellates in slow
sand filters was obtained by shaking sand
samples with Chalkley's medium and
counting protozoa using the MPN (most
probable number) procedure (Richards,
1974). Flagellates were the most abundant
protozoa counted by Richards, and reached
64,000/cm3 of filter medium after 6 weeks
of filtration. The efficiency of extraction
from the sand and the total number of
flagellates per area of filter bed were not
reported.
Heterotrophic nanoflagellate clearance rate
The clearance rate, the volume of water an
organism can clear of particles per unit time
at low particle concentrations, for the
chrysophyte isolated from slow sand filters
has not yet been determined, but the volume-
8
specific clearance rate (qp = Qp/p where
p is volume of the protozoan and Qp is the
clearance rate) of HNF are 10 to 50 times
higher than those of bacteriovorus ciliates
(Fenchel, 1982b). Fenchel measured the
volume-specific clearance rate for six HNF
species to be 5 x 104 to 106/h. He obtained
a volume-specific clearance rate for
Paraphysomonas vestita of 9.1 x 104/h.
Seale et al. (1990) calculated a volumespecific clearance rate of 7.9 x 105/h for
Spumella. Berninger et al. (1991b) obtained
a volume-specific clearance rate estimate of
8 x 104/h for a diverse community of HNF in
freshwater.
The minimum number of protozoa per bed
area (Np with dimensions [protozoa/L2])
required to filter the water in a slow sand
filter can estimated by dividing the approach
velocity of the water above the filter bed (Va
with dimensions [L/T]) by the individual
protozoa clearance rate (Qp with dimensions
[L3/(protozoa · T)]).
Va
Np = Q
p
(1)
This crude minimum estimate is based on
the unrealistic approximation that all of the
protozoa operate in parallel (that is, all
protozoa
only
process
previously
unprocessed water).
Using the volume-specific clearance rate
measurements for Paraphysomonas vestita
and Spumella (qp = 9.1 x 104/h to 7.9 x
105/h) and substituting into Eq. 1 with Va =
10 cm/h and p = 14 µm3 (Qp = 1.3 x 106
to 1.1 x 107 µm3/h) yields 9.0 x 105 to 7.8
6
2
x 10 HNF/cm required to process water at
the rate applied to a slow sand filter. The
HNF are not all arranged in parallel and thus
the HNF population density required to clear
the water of bacteria in slow sand filters is
6
2
expected to exceed 10 HNF/cm . In this
study 2.5 x 107 HNF/cm2 were used to
successfully inoculate a filter column.
Although some of the applied HNF may
have washed through the filter and although
some increase in HNF may have occurred
prior
to
the
first
post-inoculum
measurements, the ability of 2.5 x 107
HNF/cm2 to clear the water of bacteria is
consistent with the clearance rate
measurements of others.
The same estimation procedure was used to
evaluate the potential role of Vorticella.
Using V a = 15 cm/h as in Lloyd’s (Lloyd,
1973) experiments and Qp = 4.4 x 104 m3/s
as reported by Fenchel (Fenchel, 1986) for
Vorticella elongata gave a minimum
Vorticella density, Np, of 90,000/cm2. This
minimum number is 45 times the Vorticella
density reported by Lloyd (Lloyd, 1973), and
suggests that suspension feeding by
Vorticella is not principally responsible for
bacterial predation.
Volume-specific clearance rates vary
significantly depending on the measurement
technique and the species of HNF. Thus, it
will be necessary to measure volumespecific clearance rates for the species of
HNF found in slow sand filters. Reported
HNF volume-specific clearance rates
obtained thus far have been for HNF in
suspension (Berninger, et al., 1991b;
Bjornsen, et al., 1988; Fenchel, 1982b;
Goldman & Dennett, 1990; Gonzalez, et al.,
1990; Kuuppo-Leinikki, 1990; Monger &
Landry, 1991; Nygaard & Hessen, 1990).
However, HNF preferentially colonize
surfaces (Fenchel, 1991), presumably
because they benefit from doing so
(Patterson & Larsen, 1991). The ability to
capture prey at low population densities is
important and thus we expect the clearance
rate of an attached organism to be greater
than the clearance rate of a free-swimming
organism.
9
Flow relative to attached HNF also may
increase the effective volume-specific
clearance rate by reducing the amount of
flagellum-induced recirculation. In stagnant
water this recirculation may decrease the
effective volume-specific clearance rate due
to reprocessing of fluid previously cleared of
prey. Based on the apparent advantages of
attachment and imposed large scale flow in
slow sand filters, the effective HNF volumespecific clearance rate is expected to be
higher in slow sand filters than in
suspensions of HNF and bacteria.
Implications for slow sand filtration
The small size of the chrysophyte suggests
that it is unable to ingest pathogenic
protozoa such as Cryptosporidium or
Giardia lamblia. This is consistent with
previous results indicating that particles
larger than about 2 µm are not removed by
biological mechanisms in slow sand filters
(Weber-Shirk & Dick, 1997b). Thus,
Cryptosporidium and Giardia lamblia are
unlikely to be removed by predation in slow
sand filters.
The ability of HNF to discriminate between
particles on the basis of qualities other than
size (Fenchel, 1987) leads to the possibility
that slow sand filters may preferentially
remove certain types of bacteria (Simek &
Chrzanoski, 1992) and that the filters may
not remove similar-sized inert particles with
the same efficiency. The appearance of HNF
in filter effluents raises the possibility that
ingested pathogenic bacteria may be
transported through a slow sand filter and
through subsequent disinfection (King, et
al., 1988).
Improved understanding of the role of
predators in slow sand filter performance
offers opportunity for improving filter
design and operation. As demonstrated
herein, an inoculum of the HNF might be
used to rapidly ripen a slow sand filter. Also,
as demonstrated previously (Weber-Shirk &
Dick, 1993), augmentation of water fed to
filters with particulate feed designed to
increase the population density of predators
could enhance bacterial removal.
CONCLUSIONS
(1) A heterotrophic nanoflagellate identified
as a chrysophyte was isolated from the
effluent of a slow sand filter.
(2) The chrysophyte was grown on a pure
culture of Ps. putida.
(3) The chrysophyte was able to remove
more than 99.7% of influent E. coli
within one day of being applied to a new
filter column. In comparison, the control
filter removed less than 10% of the
influent E. coli at the same time.
(4) Bacteria and, potentially, bacteria-sized
particles can be removed by the
chrysophyte in slow sand filters.
(5) The chrysophyte is smaller than
pathogenic protozoa such as Giardia
lamblia and Cryptosporidium oocysts
and thus does not contribute to their
removal.
ACKNOWLEDGMENTS
This research was supported by the United
States Environmental Protection Agency
grant
number
R–816409.
Opinions
expressed are the authors and not necessarily
those of the sponsor. Mention of trade
names or commercial products does not
constitute endorsement or recommendation
for their use.
We thank Carol Rehkugler for isolating and
identifying the E. coli and Sharon Best for
supplying a pure culture of Ps. putida.
REFERENCES
Aeppli, J. (1990). Appearance of Invertebrates in
Slow Sand Filters and Reservoirs of the Zurich
Switzerland Water Supply. Aqua (Oxf), 39, 4855.
Bak, R. P. M., Duyl, F. C. V., Nieuwland, G., & Kop,
A.
J.
(1991).
Benthic
Heterotrophic
Nanoflagellates in North Sea Field-Mesocosm
Bottoms and their Response to Algal
Sedimentation. Ophelia, 33, 187-196.
10
Baker, M. N. (1981). The Quest for Pure Water.
Denver: American Water Works Association.
Fenchel, T. (1986). Protozoan filter feeding. Progress
in Protistology, 1, 65-113.
Barcina, I., Ayo, B., Muela, A., Egea, L., & Iriberri,
J. (1991). Predation Rates of Flagellate and
Ciliated Protozoa on Bacterioplankton in a River.
Fems Microbiol. Ecol., 85, 141-150.
Fenchel, T. (1987). Ecology of Protozoa. Madison,
Wisconsin: Science Technical.
Berninger, U. G., Caron, D. A., Sanders, R. W., &
Finlay, B. J. (1991a). Heterotrophic flagellates of
planktonic communities, their characteristics and
methods of study. In D. J. Patterson & J. Larsen
(Eds.), The biology of free-living heterotrophic
flagellates (pp. 22-38). Oxford: Clarendon Press.
Berninger, U. G., Finlay, B. J., & Kuuppo, L. P.
(1991b). Protozoan control of bacterial
abundances in freshwater. Limnol. Oceanogr., 36,
139-147.
Bjornsen, P. K., Riemann, B., Horsted, S. J., Nielsen,
T. G., & Pock, S. J. (1988). Trophic interactions
between heterotrophic nanoflagellates and
bacterioplankton in manipulated seawater
enclosures. Limnol. Oceanogr., 33, 409-420.
Fenchel, T. (1991). Flagellate design and function. In
D. J. Patterson & J. Larsen (Eds.), The biology of
free-living heterotrophic flagellates (pp. 7-19).
Oxford: Clarendon Press.
Finlay, B. J., Clarke, K. J., Cowling, A. J., Hindle, R.
M., Rogerson, A., & Berninger, U. G. (1988). On
the abundance and distribution of protozoa and
their food in a productive freshwater pond. Eur.
J. Protistol., 23, 205-217.
Goldman, J. C., & Dennett, M. R. (1990). Dynamics
of prey selection by an omnivorous flagellate.
Mar. Ecol. Prog. Ser., 59, 183-194.
Gonzalez, J. M., Sherr, E. B., & Sherr, B. F. (1990).
Size-selective grazing on bacteria by natural
assemblages of estuarine flagellates and ciliates.
Appl. Environ. Microbiol., 56, 583-589.
Bonnet, M. C., Welte, B., & Montiel, A. (1992).
Removal of Biodegradable Dissolved Organic
Carbon in a Water Treatment Plant. Wat. Res.,
26, 1673-1680.
Greenberg, A. E., Clesceri, L. S., & Eaton, A. D.
(Ed.). (1992).
Standard Methods for the
Examination of Water and Wastewater. (18th
Edition ed.). Washington, D.C.:
APHA,
AWWA, WPCF.
Burman, N. P., & Lewin, J. (1961). Microbiological
and Operational Investigation of Relative Effects
of Skimming and in situ Sand Washing on Two
Experimental Slow Sand Filters. J. Inst. Wat.
Engin., 15, 355-367.
Holen, D. A., & Boraas, M. E. (1991). The feeding
behavior of Spumella sp. as a function of particle
size: Implications for bacterial size in pelagic
systems. Hydrobiologia, 220, 73-88.
Carlough, L. A., & Meyer, J. L. (1991). Bacterivory
by sestonic protists in a southwestern blackwater
river. Limnol. Oceanogr., 36, 873-883.
Cowling, A. J. (1991). Free-living heterotrophic
flagellates: methods of isolation and maintenance,
including sources of strains in culture. In D. J.
Patterson & J. Larsen (Eds.), The biology of freeliving heterotrophic flagellates (pp. 477-492).
Oxford: Clarendon Press.
Ellis, K. V. (1985). Slow Sand Filtration. Crit. Rev.
Environ. Cont., 15, 315.
Fenchel, T. (1982a). Ecology of Heterotrophic
Microflagellates. I. Some Important Forms and
Their Functional Morphology. Mar. Ecol. Prog.
Ser., 8, 211-223.
Fenchel, T. (1982b). Ecology of Heterotrophic
Microflagellates. II. Bioenergetics and Growth.
Mar. Ecol. Prog. Ser., 8, 225-231.
Fenchel, T. (1982c). Ecology of Heterotrophic
Microflagellates. IV. Quantitative Occurrence
and Importance as Bacterial Consumers. Mar.
Ecol. Prog. Ser., 9, 35-42.
Huisman, L., & Wood, W. E. (1974). Slow Sand
Filtration. Geneva: World Health Organization.
King, C. H., Shotts, E. B. Jr., Wooley, R. E., &
Porter, K. G. (1988). Survival of coliforms and
bacterial pathogens within protozoa during
chlorination. Appl. Environ. Microbiol., 54,
3023-3033.
Kuuppo-Leinikki, P. (1990). Protozoan grazing on
planktonic bacteria and its impact on bacterial
population. Mar. Ecol. Prog. Ser., 63, 227-238.
Lloyd, B. (1973). The Construction of a Sand Profile
Sampler: Its Use in the Study of the Vorticella
Populations and the General Interstitial
Microfauna of Slow Sand Filters. Wat. Res., 7,
963-973.
Monger, B. C., & Landry, M. R. (1991). Prey-size
dependency of grazing by free-living marine
flagellates. Mar. Ecol. Prog. Ser., 74, 239-248.
Morel, F. M. M., Westall, J. C., Rueter, J. G., &
Chaplick, J. P. (1975). Description of the Algal
Growth Media "Aquil" and "Fraquil" (Technical
Note No. 16). National Science Foundation.
11
Nygaard, K., & Hessen, D. O. (1990). Use of carbon14 protein-labelled bacteria for estimating
clearance rates by heterotrophic and mixotrophic
flagellates. Mar. Ecol. Prog. Ser., 68, 7-14.
Patterson, D. J. (1992). Free-Living Freshwater
Protozoa. Boca Raton, Florida: CRC Press, Inc.
Patterson, D. J., & Larsen, J. (1991). General
introduction. In D. J. Patterson & J. Larsen
(Eds.), The biology of free-living heterotrophic
flagellates (pp. 1-5). Oxford: Clarendon Press.
Poynter, S. F. B., & Slade, J. S. (1977). The Removal
of Viruses by Slow Sand Filtration. Prog. Wat.
Tech., 9, 75-88.
Richards, A. D. (1974). The Distribution and Activity
of Protozoa in Slow Sand Filters. Abstract in
Journal of Protozoology, 21, 451.
Rittmann, B. E., & Huck, P. M. (1989). Biological
Treatment of Public Water Supplies. Crit. Rev.
Environ. Cont., 19, 119-184.
Seale, D. B., Boraas, M. E., Holen, D., & Nealson, K.
H. (1990). Use of bioluminescent bacteria
xenorhabdus-luminescens to measure predation
on bacteria by freshwater microflagellates. Fems
(Fed Eur Microbiol Soc) Microbiol Ecol, 73, 3140.
Sherr, B. F., & Sherr, E. B. (1991). Proportional
distribution of total numbers, biovolume, and
bacterivory among size classes of 2-20 µm
nonpigmented marine flagellates. Mar. Microb.
Food Webs, 5, 227-237.
Simek, K., & Chrzanoski, T. H. (1992). Direct and
Indirect Evidence of Size-Selective Grazing on
Pelagic Bacteria by Freshwater Nanoflagellates.
Appl. Environ. Microbiol., 58, 3715-3720.
Weber-Shirk, M., & Dick, R. I. (1993). Evidence for
Biologically Mediated Bacteria Removal In Slow
Sand Filters. Proceedings American Water Works
Association Annual Conference, (pp. 667-691).
San Antonio, TX: American Water Works
Association, Denver Colorado.
Weber-Shirk, M., & Dick, R. I. (1997a). Biological
Mechanisms in Slow Sand Filters. Jour. AWWA,
89, 72-83.
Weber-Shirk, M., & Dick, R. I. (1997b). PhysicalChemical Mechanisms in Slow Sand Filters. Jour.
AWWA, 89, 87-100.
Weber-Shirk, M. L. (1992) Bacteria Removal
Mechanisms in Slow Sand Filters. Ph.D., Cornell
University.