ARTICLE IN PRESS

Available at www.sciencedirect.com

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

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

�Corresponding autE-mail address: p

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

Zebra mussel filtration and its potential uses in industrialwater treatment

Paul Elliotta,�, David C. Aldridgea, Geoff D. Moggridgeb

aDepartment of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UKbDepartment of Chemical Engineering, University of Cambridge, Pembroke Street, Cambridge CB2 3RA, UK

a r t i c l e i n f o

Article history:

Received 22 July 2007

Received in revised form

15 October 2007

Accepted 16 October 2007

Available online 18 October 2007

Keywords:

Dreissena polymorpha

Zebra mussel

Water treatment works

Biofilter

Water filtration

Flow-through system

nt matter & 2007 Elsevie.2007.10.020

hor. Tel.: +44 1223336617;e206@cam.ac.uk (P. Ellio

a b s t r a c t

The zebra mussel (Dreissena polymorpha) is a notorious freshwater biofouling pest, and

populations of the species can alter aquatic environments through their substantial

filtration capabilities. Despite the ecological importance of zebra mussel filtration, many

predictions of their large-scale effects on ecosystems rely on extrapolations from filtration

rates obtained in static laboratory experiments, not accounting for natural mussel

densities, boundary layer effects, flow rates or elevated algal concentrations. This study

used large-scale industrial flume trials to investigate the influence of these factors on zebra

mussel filtration and proposes some novel industrial applications of these findings. The

flume trials revealed some of the highest zebra mussel clearance rates found to date, up to

574720 ml h�1 g�1 of wet tissue mass. Under low algal concentrations, chlorophyll a

removal by zebra mussels was not proportional to mussel density, indicating that field rates

of zebra mussel grazing may be much lower than previous studies have predicted.

Increasing ambient velocities up to 100 ml s�1 (�4 cm s�1) led to increased clearance rates

by zebra mussels, possibly due to the replenishment of locally depleted resources, but

higher velocities of 300 ml s�1 (12 cm s�1) did not lead to further significant increases in

clearance rate. When additional algal cultures were dosed into the flumes, chlorophyll a

removal increased approximately logarithmically with zebra mussel density and there

were no differences in the clearance of three different species of alga: Ankyra judayi,

Pandorina morum and Cyclotella meneghinia. Some novel industrial uses of these zebra mussel

filtration studies are proposed, such as: (1) helping to inform models that predict the large-

scale grazing effects of the mussels, (2) allowing estimates of zebra mussel densities in

industrial pipelines, and (3) constructing large-scale biofilters for use in water clarification.

& 2007 Elsevier Ltd. All rights reserved.

1. Introduction

Much attention has been drawn to the industrial biofouling

effects of zebra mussels around the world (Claudi and Mackie,

1994), but zebra mussels also have profound ecological effects

on aquatic environments. Filter feeders, such as the zebra

mussel, can be major consumers of phytoplankton, exerting

significant top-down control on phytoplankton levels (Caraco

r Ltd. All rights reserved.

fax: +44 1223336676.tt).

et al., 1997). Because zebra mussels can reach densities of

over 700,000 m�2 (Pathy, 1994), filter large volumes of water

and retain a wide size range of particles (Sprung and Rose,

1988; Silverman et al., 1996), zebra mussel populations are

capable of removing over 90% of organic matter from

the water (MacIsaac, 1996). Since the zebra mussel invasion

of North America, chlorophyll a concentration (an indicator

of algal population density) has dropped by over 90%

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 4 1665

on the north shore of Lake Erie (Nicholls and Hopkins,

1993).

In addition to decreasing phytoplankton biomass, zebra

mussels can cause shifts in phytoplankton community

composition (Caraco et al., 1997; Smith et al., 1998). Labora-

tory studies and modelling have shown that zebra mussel

filtration may differentially affect phytoplankton taxa (Van-

derploeg et al., 1996; Bastviken et al., 1998). Most worryingly,

cyanobacterial blooms have often occurred soon after zebra

mussel establishment, despite large declines in overall

chlorophyll a concentration (MacIsaac, 1996; Vanderploeg

et al., 1996). This can cause concern for water treatment

companies, as some species of cyanobacteria can produce

hepatotoxins, which may pose a serious health risk for

human populations (Jochimsen et al., 1998).

Given the potentially massive effects of zebra mussel

filtration, a number of studies from both Europe and America

have attempted to quantify the filtration and clearance rates

of zebra mussels. The filtration rate (or pumping rate) of a

mussel is defined as the volume of water that passes through

the mussel’s gills per unit time. In contrast, the clearance rate

of a mussel is the volume of water that is completely cleared

of suspended particles per unit time. For zebra mussels, there

is enormous variability in recorded filtration and clearance

rates, which range from 5 to 400 ml mussel�1 h�1 (Ackerman,

1999; Baldwin et al., 2002). This variability may reflect true

differences in filtration rate due to factors such as tempera-

ture (Vanderploeg et al., 1995; Lei et al., 1996), water velocity

(Ackerman, 1999), particle size (Sprung and Rose, 1988) or

resting periods (Morton, 1971).

Differences in clearance rates may also reflect differences

in the experimental techniques used. The majority of studies

on zebra mussel filtration to date have used the clearance

method in which the decrease in particle concentration in an

enclosed chamber containing zebra mussels is measured (e.g.

Silverman et al., 1995; Lei et al., 1996; Horgan and Mills, 1997).

Using this method, studies have found per capita clearance

rates ranging from approximately 20 ml mussel�1 h�1 (Bald-

win et al., 2002) to approximately 287 ml mussel�1 h�1. There

have also been a small number of studies on zebra mussels

using the flow-through chamber method (e.g. Ackerman, 1999;

Baldwin et al., 2002), where the decrease in particle concen-

tration in a water current flowing past the mussels is

measured. These have found some of the highest clearance

rates for zebra mussels to date, up 420 ml mussel h�1 (Baldwin

et al., 2002). An advantage of this method is that algal

concentration can be maintained during the experiment.

Riisgard (2001) questions the reliability of many of the tech-

niques used to measure mussel filtration, asserting that the

clearance method is one of the most likely techniques to provide

reliable estimates of filtration rates when used under optimal

laboratory conditions. Although this may be true, zebra mussels

do not live under optimal laboratory conditions and there is still a

great need to characterise the filtration abilities of zebra mussels

at natural population densities, flow rates and algal concentra-

tions. This study attempts to measure mussel filtration using a

large, flow-through rig constructed on a water treatment facility

in North London, England. This enabled the filtration of raw,

untreated reservoir water to be measured under conditions of

high zebra mussel density and with variable flow rates.

Such studies of zebra mussel filtration may also be of

substantial applied use. Firstly, investigations of the relation-

ship between zebra mussel density and clearance rates may

help inform predictions of the large-scale grazing impacts of

zebra mussels. Previously, population filtration rates have

been estimated by combining laboratory measurements of

the filtration rates of individual zebra mussels, with their

density in the field (MacIsaac et al., 1992; Bunt et al., 1993;

Fanslow et al., 1995). However, there have been no compre-

hensive investigations to validate the assumption that algal

removal from a waterbody scales linearly with zebra mussel

density. Secondly, studies of zebra mussel filtration may be of

use to industries experiencing zebra mussel biofouling of

their intake pipelines: if a relationship between algal clear-

ance and mussel density does exist, algal concentrations at

the end of a pipeline could be used to predict the level of

infestation. Finally, there is the possibility that zebra mussels

could actually be used as industrial biofilters. In Holland,

zebra mussels have been considered a useful tool in the

restoration of eutrophic lakes by biomanipulation (Reeders

and Bij de Zaate, 1990; Reeders et al., 1993). Many water

companies have problems with algal blooms in their raw

water sources, and this investigation considers whether zebra

mussels could be used to help alleviate this problem.

The goals of the work described in this paper are (1) to

measure the clearance rates of zebra mussels in the semi-

natural, flow-through conditions of a large industrial rig; (2) to

explore the effects of zebra mussel density and water flow

rate on zebra mussel clearance under conditions of normal

and elevated algal concentration; and (3) to examine the

potential of three applied uses of these zebra mussel filtration

studies: for predicting large scale grazing effects of zebra

mussels, for estimating densities of zebra mussel infestation

in a pipeline and for developing an industrial biofilter based

on zebra mussel filtration.

2. Methods

2.1. Experimental apparatus

Experiments were conducted in late September 2004 using a

large flow-through rig at water treatment works in North

London (a schematic of the rig is shown in Fig. 1). Raw,

untreated reservoir water from a supplying reservoir was

pumped at a rate of 12.6m3 h�1 into a 4 m3 steel header tank.

The water then flowed by gravity out of 30 large taps in the sides

of the tank and down 30 horizontal, independent, 4 m long

flumes. Each flume consisted of square household guttering, of

base width 68 mm, which was sealed at both ends by stop-end

guttering units. A weir in the stop-end unit ensured a

unidirectional flow of water in each flume and maintained a

water depth of approximately 7 cm. Water drained into capture

tanks at the ends of the flume, from which water was pumped

into the wastewater system of the facility.

The flow in each flume was regulated by taps. Water was

pumped into the header tank at a rate faster than it could

flow from the taps by gravity, the remainder being taken away

by an overflow pipe. Consequently, the head of water in the

tank remained constant, as did flow rates in the flumes.

ARTICLE IN PRESS

Fig. 1 – Schematic and photograph of the experimental rig used in clearance experiments.

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 41666

Fifteen scratched 5 cm�20 cm stainless-steel plates were

deposited consecutively along the bases of each flume,

providing a suitable settlement medium for zebra mussels.

Mussels were collected from the sides of one of the sites’

course sand filter beds using a paint stripper blade and

bucket. This technique minimised damage to mussels during

collection because byssus threads remained intact. The

mussels were immediately deposited in each flume at even

densities of approximately 100 g plate�1, their size distribu-

tions reflecting the natural size distributions in the filter bed.

The density of mussels in each flume could be varied by

covering different numbers of the steel settlement plates (0, 5,

10 or 15) with zebra mussels.

Before the experiments, the mussels were allowed to

acclimate in the flowing flumes for 3 days, after which time

most had formed byssal attachments to the plates.

2.2. Effects of zebra mussel density on clearance rate

To assess the effects of mussel density on the clearance rate

of algae from the river water, five replicate flumes were each

stocked with 0, 5 or 10 plates of zebra mussels or 10 plates of

dead shells (20 flumes were used in total). Flows were

regulated at 100 ml s�1. During each experiment, chlorophyll

a concentration was measured at the top and bottom of each

flume using an in vivo fluorometer (Aquaflor Handheld

Fluorometer, Turner Designs, Sunnydale, CA). The fluorom-

eter was initially calibrated using a chlorophyll a solution of

known concentration, recalibrated against a Turner Designs

solid secondary standard every 12 readings. All experiments

were conducted at ambient water temperatures of between

15.1 and 15.5 1C.

For each experimental run, triplicate chlorophyll a mea-

surements were taken from near the inlet and outlet of each

flume, and the means of these values were used in statistical

tests. The clearance rates (Cl) of the zebra mussels were

determined using the following equation (Riisgard, 2001):

Cl ¼ Fð1� C2=C1Þ, (1)

where F is the flow rate down the flumes, and C1 and C2 are

the concentrations of chlorophyll a before and after passage

down the flumes.

ARTICLE IN PRESS

Fig. 2 – Size–frequency distribution of zebra mussels used in

clearance experiments.

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 4 1667

After experimentation, three plates were removed and the

number of mussels on each plate was counted. In addition,

the length of each mussel was measured to the nearest

millimetre using Vernier callipers; the size–frequency dis-

tribution of the mussels is shown in Fig. 2. The wet weight of a

subsample of 25 mussels was then measured to the nearest

0.001 g. Using an allometric regression of mussel length

against wet weight (wet weight ¼ length1.58�10�0.71), the

average wet biomass of mussels in each flume could be

calculated.

2.3. Effects of flow rates on algal clearance

To assess the effects of flow rate on algal removal, the taps on

the header tank were manipulated to give flows of 50 ml s�1

(�2 cm s�1), 100 ml s�1 (�4 cm s�1) and 300 ml s�1 (�12 cm s�1)

in experimental flumes containing either 10 plates of living

mussels or 10 plates of dead shells. Three replicate flumes

were used to test each flow and mussel/shell combination.

Triplicate chlorophyll a measurements were taken near the

inlet and outlet of each flume using the methodology

described in Section 2.2.

2.4. Clearance rates with additional algae

Three species of algae were spiked into the flumes. The

species were already abundant in the reservoirs supplying the

works:

1.

Ankyra judayi, a long fusiform unicellular alga up to 100mm

in length;

2.

Pandorina morum, a colonial unicellular green flagellate

with cells from 8 to 20 mm long, and colonies up to 50mm

across;

3.

Cyclotella meneghiniana, a small diatom up to 30mm in

diameter.

Algal samples were purchased from the Culture Collection

of Algae and Protozoa (CCAP, SAMS Research Services Ltd.,

Dunstaffnage Marine Laboratory, Oban), along with appro-

priate culture media (Jarwoski’s medium for A. judayi and

P. morum, Diatom medium for C. meneghiniana). The 5 ml

samples of algae supplied by CCAP were cultured up to 1 l

samples of at least 500mg l�1 chlorophyll a using sterile

transfer techniques and UV lighting (www.ife.ac.uk/ccap).

Twelve different trial combinations were conducted, each

with three replicates. Fifty millilitres of each algal suspension

was rapidly poured into the top of a flume containing 0, 5, 10

or 15 plates of zebra mussels. For 10 min after this rapid dose,

the concentration of chlorophyll a at the end of each flume

was measured on a second-by second basis using a real-time

chlorophyll a meter (Chelsea Minitrakka Mk II In-situ minia-

ture fluorometer). For each run, the mass of dosed chlorophyll

a removed from the water by the mussels could be estimated

using the formula

Mr ¼Xt¼te

t¼0

ðCt � CoÞ�F, (2)

where Mr is the total mass of chlorophyll a removed during

the experiment in mg, Ct is the concentration of chlorophyll a

in the water at the outlet of the mussel-infested flume at time

t, Co is the concentration of chlorophyll a in at the outlet of

the empty flume at time t, te is the time at the end of the

experimental run in seconds and F is the flow rate in the

flumes in l s�1.

3. Results

3.1. Effects of zebra mussel density on clearance rate

The total percentage of chlorophyll a removed between the

start and end of the flumes differed significantly across the

treatments (one-way ANOVA on Arcsine-transformed data,

F(3,15) ¼ 51.26, po0.001; Fig. 3a). Tukey’s multiple comparisons

showed that both flumes containing living mussels removed

more than those containing dead or no mussels (po0.05).

There was a significant difference between the percentage of

chlorophyll a removed in the flumes containing 5 plates of

mussels (570 g) and 10 plates of mussels (1140 g), which

removed 35.473.2% and 46.672.55% of chlorophyll a, respec-

tively (po0.05).

The total mass of chlorophyll a removed also differed

significantly across the flumes (one-way ANOVA, F(3,15) ¼

41.42, po0.001, Fig. 3b). Tukey’s multiple comparisons showed

that the 5- and 10-plate treatments removed a significantly

greater mass of chlorophyll a than flumes containing dead or

no mussels (po0.05). Flumes containing 10 plates removed

more chlorophyll a than those containing 5 plates (po0.05).

Chlorophyll a removal was translated into estimated

clearance rates using Eq. (1). Clearance rates were signifi-

cantly higher in flumes containing 5 plates of mussels (540 g)

than they were in flumes containing 10 plates (1140 g)

(unpaired t-test, t ¼ 2.9, d.f. ¼ 5, p ¼ 0.034, Fig. 3c). For the

ARTICLE IN PRESS

Table 1 – Results of two-way ANOVAs to test for the effectof sampling point and flow rate on chlorophyll aconcentration in flumes containing dead shells andliving zebra mussels

Factor F d.f. p

Dead shells

Sampling point 0.07 1 0.795

Flow 0.22 2 0.8

Position:flow 0.1 2 0.908

Living mussels

Sampling point 216.96 1 o0.001

Flow 9.96 2 o0.001

Fig. 3 – Influence of zebra mussel density in field-flumes on:

(a) percentage removal of chlorophyll a; (b) mass of

chlorophyll a removed; and (c) individual mussel clearance

rates.

Fig. 4 – Influence of flume flow speeds on (a) the reduction in

chlorophyll a concentration in flumes containing living and

dead zebra mussels and (b) the estimated clearance rates of

living zebra mussels.

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 41668

5-plate treatment, a mussel of length 22 mm would clear an

estimated 574720 ml h�1.

Position:flow 14.37 2 o0.001

3.2. Effects of flow speed on clearance rate

Chlorophyll a concentrations did not differ significantly

between the start and end of any flume containing dead

zebra mussel shells, no matter what the flow (Fig. 4a; Table 1).

In contrast, in flumes with living mussels, chlorophyll a

concentrations had significantly decreased by the end of the

flumes, with flow rate also having a significant effect on

chlorophyll a (Fig. 4a; Table 1).

Estimated clearance rates significantly differed across flow

velocities (F(2,30) ¼ 20.26, po0.001; Fig. 4b). Tukey’s pairwise

comparisons showed that clearance rates were lowest at a

flow of 50 ml s�1 (po0.05), while the clearance rates at 100 and

300 ml s�1 did not differ significantly (p40.05).

3.3. Clearance rates with additional algae

After algal dosing, chlorophyll a concentration at the ends of

each flume followed a skewed unimodal pattern across time

(Fig. 5). Rather than compare each curve directly, the total

mass of chlorophyll a removed from each flume was

calculated using Eq. (2). The removal of different algal species

could not be compared directly because starting concentra-

tions of chlorophyll a in the different algal solutions were not

identical. Consequently, the total percentage of the dosed

chlorophyll a removed by the mussels was calculated. The

ARTICLE IN PRESS

Fig. 5 – Changes in chlorophyll a concentration with time in flumes exposed to 12 different combinations of experimental

treatments. Fifty millilitres of three different types of concentrated algal solution were dosed into flumes containing 0, 5, 10

and 15 plates of zebra mussels, flowing at 100 ml s�1 (the lines represent replicate measurements).

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 4 1669

percentage of chlorophyll a removed increased significantly

with the natural logarithm of the number of plates of mussels

in the flume, but was not significantly affected by the algal

species used (Fig. 6; Table 2). There was no significant

interaction between the number of mussels in each flume

and the algal species used.

4. Discussion

4.1. Clearance rates in this study

The highest estimated clearance rate shown in this study

(based on wet mass extrapolations for 5 plates of mussels)

was 574720 ml h�1 g�1. This is a value far in excess of most

previous studies (Table 3, considering that mussels of length

19–20 mm weigh approximately 2 g). This may be because

previous studies using laboratory methods tend to under-

estimate filtration rates under natural conditions (Yu and

Culver, 1999). Many techniques that have been used in other

studies can negatively affect filtration rate, including artificial

diets, unnaturally high particle concentrations, short incu-

bation times, physical disturbance and chemical stress

(Mohlenberg and Riisgard, 1979; Reeders et al., 1989). The

values in this study more closely resemble the high values

for clearance found by Ackerman (1999) and Baldwin et al.

(2002), both of whom conducted experiments in a similar

flow-through apparatus; it appears that flow can increase

the effectiveness of the bivalve grazing. This may be because

flow generates increased mixing of the water, reducing

localised resource depletion. Peterson and Beal (1989) re-

viewed many marine studies and found that food particles

are often locally depleted around dense patches of suspen-

sion feeders.

In this study, the flow of the flumes would produce a

constant replenishment of algae around the mussels. This is

in contrast to many previous studies (commonly using the

chamber clearance method), which have often avoided

continuous mixing of the water to minimise disturbance of

the mussels (e.g. Sprung and Rose, 1988; Reeders et al., 1989).

As a consequence, the mussels in these other laboratory

experiments may undergo more re-filtration, reducing their

estimated clearance rates.

The high clearance rates observed in this study may also

result from the high densities of mussels that were used. It is

often assumed that the clearance rates of a population of

zebra mussels will be equal to the sum of the clearance rates

of constituent individuals (Stanczykowska, 1968; MacIsaac

et al., 1992; Bunt et al., 1993). Yu and Culver (1999) argue that

zebra mussel clearance will reach a maximum value beyond

which no further increase will occur with increased popula-

tion density due to localised resource depletion. The high

values for clearance found in the current investigation and a

similar study by Ackerman (1999) would certainly indicate

ARTICLE IN PRESS

Fig. 6 – Percentages of dosed chlorophyll a removed from

flumes containing different numbers of mussels subjected

to a rapid dose of three types of algae.

Table 2 – Results of an ANCOVA to test for the effects ofthe number of mussel infested plates and type of algaedosed on percentage removal of chlorophyll a in flumes(number of plates was Ln+1 transformed)

Treatment F d.f. p

Ln(no. plates+1) 126.49 1 o0.001

Type of algae 1.76 2 0.192

Ln(no. plates+1): algae 0.61 2 0.548

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 41670

that group living may increase clearance rates in lentic

waters. Such increases could be explained by the interactions

of multiple exhalant jets in a population; if zebra mussels can

influence the mixing of water using exhalant currents,

multiple jets may have a cumulative effect on water mixing.

Complex mixing interactions have certainly been documen-

ted in artificial jet systems (Monismith et al., 1990; Yu et al.,

2003). This would in effect mean that the filtration currents of

one mussel may facilitate the feeding of others, leading to an

enhanced clearance rate per mussel.

4.2. Effects of zebra mussel density on clearance rate

In flumes containing zebra mussels, the percentage of

chlorophyll a removed in each flume increased from

35.473.2% to 46.672.6% between the 5- and 10-plate treat-

ments. The mass of chlorophyll a removed increased from

4.0370.41mg l�1 with the 5-plate treatment to 5.1870.40mg l�1

with the 10-plate treatment. Thus, neither mass of chlor-

ophyll a removed nor percentage of chlorophyll a removed in

the flumes scaled linearly with mussel density. This led to

apparent reductions in clearance rates with larger numbers of

mussels. There are two possible reasons for this effect, both

relating to food competition.

Firstly, it may be that the zebra mussels downstream are

reducing their pumping rate in response to the lower

concentrations of algae that result from upstream mussel

filtration (Davenport and Woolmington, 1982). However, this

explanation seems implausible because it is contrary to the

commonly observed phonomenon that bivalve filtration rate

either stays constant or decreases with increasing suspension

concentration within a normal natural range (Sprung and

Rose, 1988; Reeders and Bij de Zaate, 1990; Fanslow et al.,

1995).

Alternatively, pumping rates may be remaining the same,

but at high densities, the effectiveness of particle removal by

zebra mussels is physically reduced (Yu and Culver, 1999). If

the upstream mussels were removing the majority of the

algae that could potentially be removed, the mussels down-

stream would be re-filtering water that had already been

partially cleared of seston. Thus, even if filtering at full

capacity they could not have a large effect on the overall

clearance rate. Around half of the initial mass of chlorophyll

a in each flume was not removed, indicating that a pro-

portion of the algae was not susceptible to filtration. This

would be the case if a proportion of the algae did not

pass within the mussels zone of maximal filtration. Such an

effect was documented by Yu and Culver (1999), who

produced a clearance model predicting that seston removal

by a D. polymorpha population is limited by particle delivery

from the ambient water, reaching a maximum value beyond

which no further increase will occur with zebra mussel

density.

4.3. Effects of flow velocity on clearance rate

The clearance rates at flows of 50 ml s�1 (approximately

2 cm s�1) were 6076 ml g�1 h�1. For a mussel of 22 cm, this

approximates to a value of 180 ml h�1, which falls at the upper

end of the range of values determined in previous studies of

Dreissena polymorpha (Table 3). Increased velocity in the flumes

had a positive effect on clearance rates, up to 100 ml s�1

(�4 cm s�1) when clearance rates reached 14778 ml g�1 h�1. A

further increase in velocity to 300 ml s�1 (12 cm s�1) did not

lead to further increases in clearance rate. In a number of

other bivalves, similar velocity responses have been observed,

although high velocities have been inhibitory to clearance

rates, resulting in a ramp-like or unimodal distribution

(Ackerman, 1999). The results of our study would indicate

that zebra mussel clearance rates do not drop before

12 cm s�1, although they have reached a plateau by this

velocity.

The mechanisms responsible for the inhibitory portion of

the velocity response are unknown, although three possibi-

lities have been suggested (Ackerman, 1999). Firstly, velocity-

induced dynamic pressure differentials may under- or over-

pressurise the inhalant and exhalant siphons. Secondly, high

flow velocities may lead to lift and drag, which leads to zebra

mussel shell closure and siphon retraction. Thirdly, there may

be a physiologically optimal combination of seston concen-

trations and flow rates for efficient handling and processing

of food.

ARTICLE IN PRESS

Table 3 – Literature review of the clearance rates of Dreissena polymorpha

Reference Testing device Food Size of mussel(mm)

Temperature(1C)

Clearance rate(ml mussel h�1)

Reeders et al. (1989) Stirred vessel Lake water 20–22 13–17.7 30–170

Reeders and Bij de

Vaate (1990)

In-situ enclosure Lake water 22 10–21 40–75

Morton (1971) Beaker Colloid graphite 29 5–30 5–180

Mikheev (1967) In situ Clay suspension 22 15 2–50

Kryger and Riisgard

(1988)

Beaker Chorella 22 20 286.8

Silverman et al. (1995) Test tube Esherichia coli 17–25 ? 137

Lei et al. (1996) Beaker Polymer

microspheres

57 mg DW 15 74

Bastviken et al. (1998) Recirc. flow-through

E7 cm s�1

River water 11.5275.89 16–20 44–63

Ackerman (1999)* Recirc. flow-through

o1 cm s�1

Chlorella

pyrenoidosa

11.470.8 20.670.3 60730

Ackerman (1999)* Recirc. flow-through

9 cm s�1

Chlorella

pyrenoidosa

11.470.8 20.670.3 140720

Ackerman (1999)* Recirc. flow-through

20 cm s�1

Chlorella

pyrenoidosa

11.470.8 20.670.3 3075

Ackerman (1999)* Recirc. flow-through

9 cm s�1

Chlorella

pyrenoidosa

32.5 20.670.3 320740

Baldwin et al. (2002) Beaker Chalmydomonas 20–25 20 20–125

Baldwin et al. (2002) Flow through Nanochloris 19–22 20 18–402

This study Industrial flow-

through

River water 4–27 15.1–15.5 574720 (g�1)

Sources followed by an asterisk pertain to Dreissena bugensis and are given for comparison. DW ¼ dry weight.

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 4 1671

4.4. Effects of algal dosing on clearance rates

In all experiments with an addition of algae, the amount of

chlorophyll a removed increased with the number of mussels

in each flume. However, the mass of chlorophyll a removed

did not increase linearly with zebra mussel density; as more

mussels were added, they had less of an effect on the

population clearance rate. This was probably the result of

food competition between zebra mussels (Section 4.1; Yu and

Culver, 1999). Consequently, although a reduction in chlor-

ophyll a can provide an indication of zebra mussel infesta-

tion, it is difficult to make precise quantitative estimates of

zebra mussel density based upon this value alone.

Zebra mussels appeared to remove a greater percentage of

the dosed chlorophyll a (between 60% and 85% with 10 plates)

than they did with raw water alone (46.672.6% for 10 plates;

Section 4.1). In contrast, many other studies have shown

decreasing clearance rates with increasing suspension con-

centration (e.g. Sprung and Rose, 1988; Reeders and Bij de

Zaate, 1990; Fanslow et al., 1995). The elevated algal removal

in this study may reflect an increased level of sedimentation

with algal dosing; in the algal dosing trials, there was an

appreciable loss of algae in flumes with no mussels due to

sedimentation (between approximately 25% and 40%), an

effect that was not seen in our other trials. This increased

sedimentation may partly explain the discrepancy between

this study and those of other authors.

The chlorophyll a removal with the addition of three, very

differently shaped algal species did not differ significantly.

This may suggest that zebra mussels in reservoirs impact on

many algal species equally. Horgan and Mills (1997) also found

that algal shape per se did not affect clearance rates, although

it must be noted that seston quality, such as the inorganic

particle fraction, can affect clearance (review in Kryger and

Riisgard, 1988). Similarly, Reeders et al. (1989) and Reeders

and Bij de Zaate (1990) noted that zebra mussel filtration rates

were unrelated to seston composition.

4.5. Potential applied uses of zebra mussel filtrationstudies

4.5.1. Predicting large-scale grazing effects of zebra musselsThe results of the flume trials in this study provide important

information that should be considered when modelling the

grazing effects of zebra mussels and other bivalves. Firstly,

the high clearance rates of zebra mussels in natural popula-

tion densities and conditions should be considered. Secondly,

the observation that zebra mussel clearance does not

necessarily scale with population density has serious reper-

cussions for studies that have attempted to scale-up filtration

rates measured on individual mussels in the laboratory (e.g.

Bunt et al., 1993; Reed-Andersen et al., 2000). If zebra mussel

densities are high in a water body and water mixing is weak,

food depletion may occur over the mussel bed and the

mussels will start to re-filter water that has already been

cleared of algae. As a result, the reduction in algae in a water

body may not be proportional to zebra mussel density. The

flume experiments indicate that this effect may even occur in

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 41672

shallow flowing streams. It certainly seems clear that future

experiments used to inform models of zebra mussel filtration

should attempt to investigate the precise effects of flow and

zebra mussel density.

4.5.2. Estimating densities of zebra mussel infestation in apipelineThe use of algal monitoring to estimate zebra mussel

densities in industrial pipelines is of limited use. Although

this study shows that chlorophyll a removal increases with

zebra mussel density in flumes, at high densities, removal

becomes limited by particle delivery from the ambient water,

making determination of zebra mussel density increasingly

difficult. In pipelines, algal delivery to zebra mussels will be

affected by specific fluid dynamic parameters of the pipeline,

including the flow velocity (Section 4.2), pipeline diameter

(which affects the proportional availability of water to the

mussels), turbulence (which will affect the mixing of the

water) and the length of the pipeline (which will also provide

increased opportunity for water mixing). In slow flowing,

narrow and turbulent pipelines, zebra mussel infestations

should be easier to monitor by their algal clearance, while in

wide pipes with laminar flow, it will be more difficult to detect

algal removal. However, if industrial facilities monitor their

pipelines during a period of increasing zebra mussel infesta-

tion, it is possible that they will detect changes in algal

population densities.

The dosing of algae into pipelines may certainly give a

greater degree of resolution of zebra mussel densities than

simple monitoring. This is because the dosing of a known

quantity of algae minimises the influence of background

fluctuations in algal concentration, making direct compar-

isons easier. However, it is also worth noting that the

culturing of algae is not easy or cheap, and so algal dosing

may not be financially viable for pipelines with a large flow.

An alternative possibility may be to dose an oxidant such as

sodium hypochlorite into the pipeline: this will cause the

zebra mussels to close and stop filtering (Claudi and Evans,

1993), leading to an increase in algal concentration. However,

such chlorination could lead to water quality issues, espe-

cially through the formation of trihalomethane compounds.

4.5.3. Developing an industrial biofilter based on zebramussel filtrationZebra mussels have been proposed as useful tools in the

water quality management of lakes (Reeders et al., 1989;

Reeders and Bij de Zaate, 1990). The experiments in this study

once again reveal that zebra mussels can remove huge

quantities of organic matter from the water. In addition to

algae, zebra mussels can also filter bacteria (Silverman et al.,

1995) and particle-bound phosphorus (Stanczykowska, 1984).

The infestation of reservoirs and pipelines by zebra mussels

would certainly be expected to have similar beneficial effects

on water quality, which would be of great advantage to the

water-supply industry if the water is polluted or eutrophic.

Indeed, it could be argued that water-treatment works

infested with zebra mussels already have an initial stage of

zebra mussel biofiltration! However, invasive species such as

the zebra mussel should certainly not be introduced for water

quality benefits.

Although zebra mussels could provide potential benefits to

the water industry, they are often perceived as more of a pest

than an advantageous biofiltering organism. This is because

reductions in algae can lead to knock-on reductions in

zooplankton and planktivorous fish (MacIsaac, 1996). Also,

through their selective filtration of some types of algae, zebra

mussels have been proposed to facilitate cyanobacterial

blooms (Heath et al., 1995; MacIsaac, 1996; Vanderploeg

et al., 1996). Most importantly, zebra mussels cause serious

biofouling effects on many industrial systems (see Claudi and

Mackie, 1994; Elliott et al., 2005).

Despite these problems, there are ways in which zebra

mussels could be used as a biofilter without causing

ecological or biofouling problems. One method would be to

incorporate them into a purpose-built filtration facility, much

like the rig used in this study. In our experiments, zebra

mussels could remove around 50% of the suspended chlor-

ophyll in water flowing past them at a rate of over 10,000 l h�1.

The mussels survived for over 3 weeks in the flumes (the

duration of experiments) without much maintenance; this

would indicate that there is the potential for substantial and

long-lasting algal control by zebra mussels. Large beds of

zebra mussels could certainly be used to provide an initial

stage in water treatment, especially if a rapid and shallow

flow of water is maintained. This could minimise the

difficulties and costs and of on-site water treatment.

If used in an on-site rig at a water treatment facility, zebra

mussels would not pose as much of an industrial problem as

might be expected. Firstly, they could not transfer larvae

upstream to infest earlier stages of the supply network.

Secondly, the ecological effects of this treatment would be no

worse than pre-existing treatment regimes such as filtration

and ozonation. Thirdly, if the biofilter is followed by another

treatment strategy, downstream infestation is likely to be

prevented. In England, treatment with filtration, ozonation or

chlorination removes or kills zebra mussel larvae from the

water treatment process (Elliott et al., 2005). Mussels that

survive these treatments do not penetrate the facility further,

as they have substantially reduced food sources. It must be

noted that site-by-site risk assessments would be critical in

ascertaining the downstream effects of such a filtration

facility.

In summary, zebra mussels probably already provide a level

of biofiltration for infested water treatment facilities, but their

effects could be enhanced with minimal biofouling effects if

they are incorporated into an on-site filtration facility.

However, such benefits must be balanced with the economic

and ecological risks associated with the transport of live zebra

mussels to new sites. For this reason, we would only

recommend the development of such biofilters within sys-

tems that have already been invaded by zebra mussels.

5. Conclusions

The following conclusions can be drawn from the results

presented in this paper:

1.

Large-scale flume studies using natural densities of zebra

mussels, ambient algal concentrations and flow rates of up

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 4 1673

to 12 cm s�1 revealed some of the highest zebra mussel

clearance rates found to date, as high as 574720 ml h�1 g�1

of wet tissue mass.

2.

Under naturally occurring algal concentrations, chloro-

phyll a removal by zebra mussels was not proportional to

mussel density, possibly due to the effects of re-filtration.

This means that rates of zebra mussel grazing may be

much lower than previous studies have predicted based on

extrapolations of individual zebra mussel clearance rates

to whole zebra mussel populations.

3.

Increasing ambient velocities up to 100 ml s�1 (�4 cm s�1)

led to increased clearance rates by zebra mussels, possibly

due to the replenishment of locally depleted resources.

Higher velocities of 300 ml s�1 (12 cm s�1) did not lead to

further significant increases in clearance rate.

4.

When additional algal cultures were dosed into the flumes,

chlorophyll a removal increased approximately logarith-

mically with zebra mussel density. At higher densities of

zebra mussels, particle removal became limited by particle

delivery from the water. There were no differences in the

clearance of three different species of alga: A. judayi,

P. morum and C. meneghiniana.

5.

A number of applications for these zebra mussel filtration

studies have been proposed. Firstly, they can help produce

more accurate models and estimates of the large-scale

grazing impacts of zebra mussels by taking into account

the effects of flow and density. Secondly, the filtration

studies show that reductions in algal population density

could be used, to a limited extent, to estimate the densities

of zebra mussels infesting industrial pipelines and chan-

nels. Thirdly, the successful removal of algae from the

experimental flow-through rig shows that zebra mussels

could be used as an on-site industrial biofilter for water

treatment. However, such a treatment strategy should be

used with care, and only in systems that have already been

invaded by zebra mussels.

Acknowledgements

The authors are indebted to the staff at Thames Water, Veolia

Water, Anglian Water, Severn-Trent Water and Yorkshire

Water who generously provided funding and advice during

this study. They are particularly grateful for the help of

Dr. Michael Chipps, Roy Grubb, Alf Ives, Dr. Helen Pickett,

Barrie Holden, Angela Richardson, Derek Wilson, Jenny

Banks, Dr. Paula Agutter and Lindsay Neal. Work was partially

funded by Thames water and an EPSRC Grant to Dr. David

Aldridge and Dr. Geoff Moggridge.

R E F E R E N C E S

Ackerman, J.D., 1999. Effect of velocity on the filter feeding of

dreissenid mussels (Dreissena polymorpha and Dreissena bugen-

sis): implications for trophic dynamics. Can. J. Fish. Aquat. Sci.

56, 1551–1561.

Baldwin, B.S., Mayer, M.S., Dayton, J., Pau, N., Mendilla, J., Sullivan,

M., Moore, A., Ma, A., Mills, E.L., 2002. Comparative growth and

feeding in zebra and quagga mussels (Dreissena polymorpha

and Dreissena bugensis): implications for North American lakes.Can. J. Fish. Aquat. Sci. 59, 680–694.

Bastviken, D.T.E., Caraco, N.F., Cole, J.J., 1998. Experimentalmeasurements of zebra mussel (Dreissena polymorpha) im-pacts on phytoplankton community composition. FreshwaterBiol. 39, 375–386.

Bunt, C.M., MacIsaac, H.J., Sprules, W.G., 1993. Pumping rates andprojected filtering impacts of juvenile zebra mussels (Dreissenapolymorpha) in Western Lake Erie. Can. J. Fish. Aquat. Sci. 50,1017–1022.

Caraco, N.F., Cole, J.J., Raymond, P.A., Strayer, D.L., Pace, M.L.,Findlay, S.E.G., Fischer, D.T., 1997. Zebra mussel invasion in alarge, turbid river: phytoplankton response to increasedgrazing. Ecology 78, 588–602.

Claudi, R., Evans, D.W., 1993. Chemical addition strategies forzebra mussel control in once-through service water systems.In: Nalepa, T.F., Schloesser, D.W. (Eds.), Zebra Mussels: Biology,Impacts and Control. Lewis, Boca Raton, FL, pp. 563–574.

Claudi, R., Mackie, G.L., 1994. Practical Manual for Zebra MusselMonitoring and Control. Lewis, Boca Raton, FL.

Davenport, J., Woolmington, A.D., 1982. A new method ofmonitoring ventilatory activity in mussels and its use in astudy of the ventilatory patterns of Mytilus edulis-L. J. Exp. Mar.Biol. Ecol. 62, 55–67.

Elliott, P., Aldridge, D.C., Moggridge, G.D., 2005. The increasingeffects of zebra mussels on water installations in England.Water Environ. Manage. Dec, 367–375.

Fanslow, D.L., Nalepa, T.F., Lang, G.A., 1995. Filtration rates of thezebra mussel (Dreissena polymorpha) on natural seston fromSaginaw Bay, Lake Huron. J. Great Lakes Res. 21, 489–500.

Heath, R.T., Fahnenstiel, G.L., Gardner, W.S., Cavaletto, J.F., Hwang,S.J., 1995. Ecosystem-level effects of zebra mussels (Dreissenapolymorpha): An enclosure experiment in Saginaw Bay, LakeHuron. Journal of Great Lakes Research. 21, 501–516.

Horgan, M.J., Mills, E.L., 1997. Clearance rates and filtering activityof zebra mussel (Dreissena polymorpha): implications forfreshwater lakes. Can. J. Fish. Aquat. Sci. 54, 249–255.

Jochimsen, E.M., Carmichael, W.W., An, J.S., Cardo, D.M., Cookson,S.T., Holmes, C.E.M., Antunes, M.B.D., de Melo, D.A., Lyra, T.M.,Barreto, V.S.T., Azevedo, S., Jarvis, W.R., 1998. Liver failure anddeath after exposure to microcystins at a hemodialysis centerin Brazil. N. Engl. J. Med. 338, 873–878.

Kryger, J., Riisgard, H.U., 1988. Filtration rate capacities in 6species of European fresh-water bivalves. Oecologia 77, 34–38.

Lei, J., Payne, B.S., Wang, S.Y., 1996. Filtration dynamics of thezebra mussel, Dreissena polymorpha. Can. J. Fish. Aquat. Sci. 53,29–37.

MacIsaac, H.J., 1996. Potential abiotic and biotic impacts of zebramussels on the inland waters of North America. Am. Zool. 36,287–299.

MacIsaac, H.J., Sprules, W.G., Johannsson, O.E., Leach, J.H., 1992.Filtering impacts of larval and sessile zebra mussels (Dreissenapolymorpha) in western Lake Erie. Oecologia (Berlin) 92, 30–39.

Mikheev, V.P., 1967. Filtration nutrition of the Dreissena. In:Voprosy prudovogo rybovodstva, 15, Pischevaya Promyshien-nost Publisher, Moscow, pp. 117–129 (in Russian with Englishsummary).

Mohlenberg, F., Riisgard, H.U., 1979. Filtration rate, using a newindirect technique. Mar. Biol. 54, 143–148.

Monismith, S.G., Koseff, J.R., Thompson, J.K., Oriordan, C.A., Nepf,H.M., 1990. A study of model bivalve siphonal currents.Limnol. Oceanogr. 35, 680–696.

Morton, B.S., 1971. Studies on the biology of Dreissena polymorphaPall; V. Some aspects of filter feeding and the effects of micro-organisms upon the rate of filtration. Proc. Malacol. Soc.London 39, 289–301.

Nicholls, K.H., Hopkins, G.J., 1993. Recent changes in LakeErie (north shore) phytoplankton: cumulative impacts of

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 6 6 4 – 1 6 7 41674

phosphorous loading reductions and the zebra mussel intro-duction. J. Great Lakes Res. 19, 637–647.

Pathy, D.A., 1994. The life history and demography of the zebramussel, Dreissena polymorpha, populations in Lake St. Clair,Lake Erie, and Lake Ontario. Master abstracts 33/02, 489.

Peterson, C.H., Beal, B.F., 1989. Bivalve growth and higher orderinteractions: importance of density, site and time. Ecology 70,1390–1404.

Reed-Andersen, T., Carpenter, S.R., Padilla, D.K., Lathrop, R.C.,2000. Predicted impact of zebra mussel (Dreissena polymorpha)invasion on water clarity in Lake Mendota. Can. J. Fish. Aquat.Sci. 57, 1617–1626.

Reeders, H.H., Bij de Zaate, A.B., 1990. Zebra mussels (Dreissenapolymorpha)—a new perspective for water-quality manage-ment. Hydrobiologia 200, 437–450.

Reeders, H.H., DeVaate, A.B., Slim, F.J., 1989. The filtration rate ofDreissena polymorpha (Bivalvia) in 3 Dutch lakes with referenceto biological water quality management. Freshwater Biol. 22,133–141.

Reeders, H.H., Bij de Vaate, A., Noordhuis, R., 1993. Potential of thezebra mussel for water quality management. In: Nalepa, T.F.,Schloesser, D.W. (Eds.), Zebra Mussels: Biology, Impacts andControl. Lewis, Boca Raton, FL, pp. 439–454.

Riisgard, H.A., 2001. On measurement of filtration rates inbivalves—the stony road to reliable data: review and inter-pretation. Mar. Ecol. Prog. Ser. 211, 275–291.

Silverman, H., Achberger, E.C., Lynn, J.W., Dietz, T.H., 1995.Filtration and utilization of laboratory-cultured bacteria byDreissena polymorpha, Corbicula fluminea, and Carunculina tex-asensis. Biol. Bull. 189, 308–319.

Silverman, H., Lynn, J.W., Dietz, T.H., 1996. Particle capture by thegills of Dreissena polymorpha: structure and function of latero-frontal cirri. Biol. Bull. 191, 42–54.

Smith, T.E., Stevenson, R.J., Caraco, N.F., Cole, J.J., 1998. Changes inphytoplankton community structure during the zebra mussel(Dreissena polymorpha) invasion of the Hudson River (NewYork). J. Plankton Res. 20, 1567–1579.

Sprung, M., Rose, U., 1988. Influence of food size and foodquantity on the feeding of the mussel Dreissena polymorpha.Oecologia 77, 526–532.

Stanczykowska, A., 1968. The filtration capacity of populationsof Dreissena polymorpha in different lakes as a factor affectingcirculation of matter in the lake. Ekologia Polska B. 14,265–270.

Stanczykowska, A., 1984. Role of bivalves in the phosphorus andnitrogen budget in lakes. Verh. Inter. Ver. Theor. Angew.Limnol. 22, 982.

Vanderploeg, H.A., Strickler, J.R., Liebig, J.R., Nalepa, T.F., Fahnen-stiel, G.L., Gardener, W.S., Cavaletto, J.F., Fanslow, D.L., Johen-gen, T.H., 1995. Do zebra mussels promote blue-green andmetaphyton blooms on Sanginaw Bay, and do these bloomsaffect the mussels? In: Proceedings of the 38th Conferenceof the International Association of Great Lakes Research,p. 113.

Vanderploeg, H.A., Liebig, J.R., Nalepa, T.F., Johengen, T.H.,Strickler, J.R., 1996. Zebra mussels may be promoting Micro-cystis blooms in Sanginaw bay and Lake Erie through selectivefiltering and nutrient excretion. In: The Sixth InternationalZebra Mussel and Other Aquatic Nuisance Species Confer-ence, Dearborn, MI, March 1996.

Yu, D., Ali, M.S., Lee, J.H.W., 2003. Experiments on interaction ofmultiple jests in a crossflow. In: 16th ASCE EngineeringMechanics Course. University of Washington, Seattle.

Yu, N., Culver, D.A., 1999. Estimating the effective clearance rateand refiltration by zebra mussels, Dreissena polymorpha, in astratified reservoir. Freshwater Biol. 41, 481–492.