Using clay to control harmful algal blooms: deposition and resuspension of clay/algal flocs

16 Pages·2004·0.26 MB·English

HarmfulAlgae4(2005)123–138
Using clay to control harmful algal blooms: deposition and
resuspension of clay/algal flocs
Stace E. Beaulieua,∗, Mario R. Sengcob, Donald M. Andersonb
aAppliedOceanPhysicsandEngineeringDepartment,MailStop#7,WoodsHoleOceanographicInstitution,
WoodsHole,MA02543,USA
bBiologyDepartment,MailStop#32,WoodsHoleOceanographicInstitution,WoodsHole,MA02543,USA
Received28June2003;receivedinrevisedform6October2003;accepted21December2003
Abstract
Harmfulalgalblooms(HABs)maybelegitimatetargetsfordirectcontrolormitigation,duetotheirimpactsoncommercial
fisheriesandpublichealth.OnepromisingcontrolstrategyistherapidsedimentationofHABsthroughflocculationwithclay.
Theobjectiveofthisstudywastoevaluateflowenvironmentsinwhichsuchacontrolstrategymightbeeffectiveinremoving
harmfulalgaefromthewatercolumnanddepositingalayerofclay/algalflocsontheseafloor.Wesimulatedthenatural
environmentintwolaboratoryflumes:astraight-channel“17mflume”inwhichflocssettledinastill-watercolumnanda
“racetrackflume”inwhichflocssettledinflow.The17mflumeexperimentsweredesignedtoestimatethecriticalbedshear
stressforresuspensionofflocsthathadsettledfordifferenttimeperiods.Theracetrackflumeexperimentsweredesigned
toexaminethedepositionandrepeatedresuspensionofflocsinasystemwithtidalincreasesinflowspeed.Allflumeruns
were conducted with the non-toxic dinoflagellate Heterocapsa triquetra and phosphatic clay (IMC-P4). We repeated the
experimentswithacoagulant,polyaluminumhydroxychloride(PAC),expectedtoenhancetheremovalefficiency(RE)ofthe
clay.Ourexperimentsindicatedthatatlowflowspeeds(≤10cms−1),phosphaticclaywaseffectiveatremovingalgalcells
fromthewatercolumn,evenafterrepeatedresuspension.Oncealayerofflocsaccumulatedonthebed,theconsolidation,or
dewatering,ofthelayerovertimeincreasedthecriticalshearstressforresuspension(i.e.decreasederodibility).Resuspension
of a 2mm thick layer that settled for 3h in relatively low flow speeds (≤3cms−1) would be expected at bed shear stress
of∼0.06–0.07Pa,ascomparedtoupto0.09Paforalayerthatwasundisturbedfor9or24h.Forthesameexperimental
conditions,theadditionofPACdecreasedtheremovalefficiencyofalgalcellsinflowandincreasedtheerodibilityofflocs
from the bottom. By increasing the likelihood that flocs remain in suspension, the addition of PAC in field trials of clay
dispersal might have greater impact on sensitive, filter-feeding organisms. Overall, our experiments suggest that the flow
environmentshouldbeconsideredbeforeusingclayasacontrolstrategyforHABsincoastalwaters.
©2004ElsevierB.V.Allrightsreserved.
Keywords:Flume;Harmfulalgalbloom;Heterocapsatriquetra;Phosphaticclay;Polyaluminumhydroxychloride;Resuspension;
Sedimentation
1. Introduction
Harmfulalgalblooms(HABs)poseaseriousthreat
∗ Correspondingauthor.Tel.:[phone]; to commercial fisheries and aquaculture, human
fax:[phone]. health, and coastal aesthetics. These phenomena are
E-mailaddress:[email protected](S.E.Beaulieu). caused by rapid population growth of microscopic
1568-9883/$–seefrontmatter©2004ElsevierB.V.Allrightsreserved.
doi:10.1016/j.hal.2003.12.008
124 S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138
algae, some of which produce toxins. Because of the algal and clay particle concentration (Avnimelech
potentially significant impacts of HABs, these phe- etal.,1982;SoballeandThrelkeld,1988).Laboratory
nomena may be legitimate targets for direct control studies of harmful algal species again showed that
or mitigation. Perhaps the most promising control some clay minerals were more effective at removing
strategy, with respect to maximizing effectiveness cells from the water column (e.g. Yu et al., 1995;
and minimizing costs and environmental impacts, is Sengco et al., 2001), with species-specific variability
flocculation and sedimentation of HABs with clay inremovalefficiency(RE)(Sengco,2001).
(Anderson,1997;Boeschetal.,1997;Andersonetal., Although the above studies indicated that clays
2001; Sengco, 2001). Over the past 25 years, clays can be effective in removing algal cells from a static
havebeeninvestigatedinseveralcountriesasameans watercolumn,controlledexperimentssimulatingnat-
of removing harmful algae from the water column ural flow conditions must be performed to facilitate
(e.g.Shirota,1989;Yuetal.,1994;Baeetal.,1998). estimates of the effectiveness of clay application in
Japan (Shirota, 1989), South Korea (Choi et al., the field. A recent series of laboratory flume experi-
1998),andAustralia(Atkinsetal.,2001)alreadyhave mentswithphosphaticclayindicatedthattheremoval
attemptedclaydispersaltocontrolHABsinthefield. efficiency of the dinoflagellate Heterocapsa triquetra
Rapid sedimentation of algal cells may be caused was enhanced in low flow (∼2cms−1) and greatly
by the flocculation (i.e. coagulation; Jackson and reduced in higher flow (∼13cms−1) conditions
Lochmann, 1993) of clay particles with algal cells, (Archambault et al., 2003). Flow conditions would
the entrainment of algal cells into settling clay influence four major processes that must be consid-
flocs, and/or the loss of algal cell motility due to ered in evaluating whether clay can mitigate HABs:
physico-chemical interactions with the clay particles. flocculation, advection, deposition, and resuspension
Uponaccumulationinafloclayerontheseafloor,al- (Fig. 1). Flocculation, or the formation of clay/algal
galcellswouldbetrappedamongclayparticles,likely aggregates (flocs), would result from particle contact
resulting in cell mortality (Sengco et al., 2001; A. (collision) followed by adhesion. The rate of particle
Li et al., unpublished data; but see Burkholder, 1992 contact, in the case of clay/algal flocculation, would
for formation of temporary cysts). A wide variety of beinfluencedbyturbulenceintensity,differentialset-
laboratory studies in static (i.e. still water) systems tling of particles and flocs, and motility of the algal
have confirmed that clays can effectively remove cells (Jackson and Lochmann, 1993; Sengco, 2001).
algal cells from the water column. Early test-tube High rates of shear might also limit the maximum
studies demonstrated that clay/algal flocculation was size of flocs, thus limiting settling velocity (e.g. Hill,
influenced by mineral type, algal species, and both 1998).Flocswithlowersettlingvelocitieswouldhave
1 Clay
Algae
Water column
2
FLOW
4
3
Sea floor
Fig. 1. A scenario for clay/algal flocculation, deposition, and resuspension. The dashed square indicates a study volume in which
environmental impacts are monitored. (1) Clay is sprayed on the water surface. (2) Clay and single-celled algae flocculate in the water
column;someflocsareadvectedlaterallyoutofthestudyvolume.(3)Clay/algalflocssettleandaccumulateontheseafloor.(4)Increased
flowleadstoresuspensionandlateraladvectionofflocs.
S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138 125
a higher probability of remaining suspended in the velocity (u∗crit) can be defined at the threshold(s) of
water column and would be subject to advection, or particle transport (e.g. Self et al., 1989; Sanford and
lateral transport, by currents. These processes, floc- Halka,1993).Oncedepositedontheseafloor,entrain-
culation and advection, were not examined directly mentintothewatercolumnwouldbeafunctionofthe
in this study. Here, we examined the deposition of flocs(density,size,shape,packing,sorting,andstick-
clay/algal flocs in flowing water and the potential for iness), larger-scale bed roughness, the fluid (density
flocstoberesuspendedafterdeposition(events3and and viscosity), and the flow conditions (mean flow,
4inFig.1). oscillatoryflow,turbulence;listmodifiedfromMiller
By inhibiting deposition and/or promoting resus- et al., 1977). Because of the cohesive nature and het-
pension, physical forcing from currents and waves erogeneous composition of clay/algal flocs, critical
has the potential to prolong the residence time of shearstressesmustbedeterminedempirically.Typical
clay/algal flocs in the water column. Initial flume ex- thresholdcurves(e.g.Shieldsdiagram)forinitialmo-
periments conducted with juvenile bivalves indicated tionofsedimentassumenon-cohesivesedimentgrains
that suspended clay/algal flocs negatively impacted ofasinglesizeandhomogeneouscomposition(Miller
growth of these sensitive filter-feeding organisms etal.,1977;MiddletonandSouthard,1984).Theflume
(Archambault et al., 2002). Resuspension also po- experiments in this study were designed to estimate
tentially would re-seed the water column with viable criticalshearstressesfordepositionandresuspension
cells. In test-tube experiments with Karenia brevis of clay/algal flocs, as well as to further study RE in
and phosphatic clay, cell mortality was not observed flow.Ourmainobjectivewastopredictflowenviron-
after 2.5h in clay loadings up to 0.5gl−1 (thickness mentsinwhichalayerofclay/algalflocswouldaccu-
ofpelletnotnoted;Sengcoetal.,2001).Ontheother mulate on the sea floor. An additional objective was
hand, in low flow conditions, a mass deposition of to determine the effect of PAC on the deposition and
flocs potentially would create a layer of flocs on the resuspensionofclay/algalflocs.Weemployedtwoap-
seafloor.Organicmaterialinthelayerwoulddecom- proaches:(1)settlinginastill-watercolumn(asmight
pose, potentially leading to anoxia and deleterious occurduringslacktideorinaprotectedembayment);
effectsonbenthicfauna.Useofclayinthefieldwould and (2) settling in flow. These flume studies are an
involve a balance between removing harmful algae importantintermediatestepbetweenexperimentswith
from the water column and minimizing environmen- test tubes and settling columns in the laboratory and
talimpactsonbenthicfauna.Somelaboratorystudies theactualdispersalofclayinthefield.
in static systems have indicated that cell RE can be
enhanced and clay loadings decreased by dispersing
coagulants such as polyaluminum hydroxychloride 2. Methods
(PAC) in very low concentrations prior to the appli-
cation of clay (Yu et al., 1995; Sengco et al., 2001). At the Coastal Research Laboratory of Woods
For similar RE of K. brevis, pretreatment with 5ppm Hole Oceanographic Institution, two different flumes
PAC reduced clay loadings by an order of magnitude wereusedinthetwoseparatesetsofexperimentsde-
instillwater(Sengcoetal.,2001). scribed below. Fries and Trowbridge (2003) illustrate
Inthenaturalenvironment,theconcentrationofsus- and describe the mechanics of both of the flumes
pended particles in the benthic boundary layer would (for photographs of the flumes, see http://www.whoi.
beinfluencedbydeposition,bedconsolidation(orde- edu/institutes/coi/facilities/coastal lab.htm). Smooth-
watering), and erosion (Mehta, 1989). In accordance turbulent benthic boundary layers developed in both
with sediment transport theory, the processes of de- flumes (fully-developed at the test sections). Due
positionandresuspension(erosion)ofclay/algalflocs to concerns about toxins in the flumes, experiments
wouldbecontrolledbythebottom,orbed,shearstress were conducted with the non-toxic dinoflagellate
(τ ) induced by the boundary-layer flow. Bed shear Heterocapsa triquetra. This species is easy to cul-
0
stress (τ ) often is described in terms of shear ve- ture in the laboratory, has a similar size (14.7(cid:1)m
0
locity (u∗), given by τ0 = ρu2∗ (in which ρ is fluid equivalent spherical diameter; Archambault et al.,
density).Criticalshearstress(τ )andcriticalshear 2003)andshapetoseveralwell-studiedharmfulalgal
0crit
126 S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138
Table1
Summaryofflumeexperiments
Experiment Experiment Settling Flowspeed T(◦C) Salinity Heterocapsa PAC Arealloading Concentration
description runno. time(h) (cms−1) (ppt) cells ((cid:1)ll−1) ofclay ofclay(gl−1)
duringsettling (ml−1 ×103) (gm−2)
17mflume
Clay 1 3 0 21.5 33 2.00 n/a 33.13 0.41
Clay 2 3 0 21.5 33 2.00 n/a 33.13 0.41
Clay 3 3 0 21.5 33 2.00 n/a 33.13 0.41
Clay 1 9 0 22 33 1.89 n/a 44.96 0.56
Clay 2 9 0 22 33 2.53 n/a 44.96 0.56
Clay 3 9 0 22 33 2.43 n/a 38.22 0.48
Clay 1 24 0 23 33 2.35 n/a 56.80 0.71
Clay 2 24 0 23 33 3.28 n/a 47.33 0.59
Clay 3 24 0 23 33 3.23 n/a 47.33 0.59
Clay+PAC 1 3 0 23.5 33 2.68 5 33.13 0.28
Clay+PAC 2 3 0 23.5 33 1.64 5 33.13 0.28
Clay+PAC 3 3 0 23.5 33 1.64 5 33.13 0.28
Clay+PAC 1 9 0 23.5 33 1.80 5 37.86 0.32
Clay+PAC 2 9 0 23.5 33.5 2.46 5 37.86 0.32
Clay+PAC 3 9 0 23.5 33.5 2.43 5 37.86 0.32
Clay+PAC 1 24 0 24 33.5 2.39 5 47.33 0.39
Clay+PAC 2 24 0 24 33.5 2.77 5 47.33 0.39
Clay+PAC 3 24 0 24 33.5 3.36 5 47.33 0.39
Racetrackflume
Clay 1 3,3,9 3 23 30 0.75 n/a 33.21 0.28
Clay+PAC 1 3,3,9 3 23 30 0.33 5 33.21 0.28
Clay 1 3,3,9 10 21 30 0.51 n/a 33.21 0.28
Clay+PAC 1 3,3,9 10 22 30 0.70 5 33.21 0.28
Clay 2a 3 20b 20.5 30 0.15 n/a 33.21 0.28
All experiments were conducted with IMC-P4 clay and the dinoflagellate Heterocapsa triquetra. Clay loadings are reported as dry mass;
n/ameansnotapplicable.
aRun1lostduetopoweroutage.
bInitiallyat20cms−1 andthendecreasedby2cms−1 every3h.
species(e.g.Kareniabrevis;Sengco,2001),andother range of algal taxa in test-tube experiments (Sengco,
species within this genus are known to cause HABs 2001). Flume experiments were repeated with the
(Matsuyama et al., 2001). Cultures were maintained addition of polyaluminum hydroxychloride, or PAC
in15lcarboysinf/2mediumwithfilteredseawaterat (Superfloc9001,CytecIndustriesInc.,WestPaterson,
◦
constant, high light intensity at 20 C. All flume runs NJ, USA), a coagulant used in treatment of drinking
were conducted at ∼20◦C so as not to shock the al- water. Neither the clay nor PAC had acute or chronic
gae (Table 1). Phosphatic clay was used in the flume toxic effects on common estuarine invertebrates and
experiments (IMC-P4; IMC Phosphates Company, fish(Lewisetal.,2003).
LakeForest,IL,USA).Phosphaticclay,aby-product
ofphosphateminingoperations,isamixtureofclays 2.1. Flocssettlinginstillwater
(esp.montmorillonite)andotherminerals(Bromwell,
1982),andinpractice,mayberelativelyinexpensive, 2.1.1. 17mflumeexperimentaldesign
easytoworkwith,andlocallyavailableinthecaseof Thefirstsetofexperimentswasconductedina17m
HABs of the Florida coast. Phosphatic clay (<2(cid:1)m long,60cmwide,straight,open-channelflume(17m
equivalent spherical diameter; Archambault et al., flume;ButmanandChapman,1989).The17mflume
2003) showed high removal efficiency for a broad experimentsweredesignedtocomparethecriticalbed
S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138 127
shearstress(τ )forresuspensionofclay/algalflocs imental design was a 3 × 2 factorial design with
0crit
that settled for different time periods. A removable three replicate flume runs for each of the six treat-
bottom panel with a recessed 20cm×20cm×2cm ments (Table 1). The analysis of critical shear stress
deep box was placed in the flume, centered at 12.9m (τ ) began when the fence was removed, expos-
0crit
downstream. The box was filled with sieved sand ing the floc layer to the overlying flow. The test
(250–500(cid:1)m), leveled ∼2mm below flush to create bed surface was illuminated from above with a light
the test bed. At the beginning of each experiment, sheet (∼2cm wide at the bed), and a video cam-
theflumewasfilledto12cmwaterdepthwith10(cid:1)m era recorded a section of the flow directly above
filteredseawater,movingattheslowestpossibleflow the test bed. As the flow speed was increased in
speed (∼3cms−1). An acrylic fence, covered with the flume, the following visual criteria were used
plastic wrap (or extending above the water surface in for determining three stages of particle transport:
theexperimentswithPAC),wasplacedaroundthetest initial motion—when particles (i.e. clay/algal flocs)
bed. A wide-mouthed pipette was used to introduce started rolling on the test bed; bedload—when most
algae (then PAC) then clay to the enclosed volume particles were rolling and some were saltating; and
of water over the test bed, and flocculation occurred resuspension—when particles were continuously lift-
in a relatively still-water column. Before introducing ing from and not returning to the bed (for video
100ml of algal culture into the enclosure, 1ml was clips of particle transport, see http://www.whoi.edu/
removedandpreservedforcellcountsat200×under science/B/people/sbeaulieu/FIPR/).Foreachmodeof
a compound microscope. Resulting cell concentra- particle transport, a laser doppler velocimeter (LDV)
tions, determined from triplicate counts of 0.1ml was used to record a 4min record of along-channel
aliquots,were∼2×103ml−1 intheenclosedvolume velocityat8cmabovebottom(a.b.).
(Table1).Forafinalconcentrationof5ppm(mgl−1)
PAC in the enclosure, stock solution was first diluted 2.1.2. Dataanalysesfor17mflumeexperiments
with 100ml of deionized water to prevent precipita- Bed shear stress (τ0) in the 17m flume derived
tion which occurs in high concentrations in seawater. mainly from the mean flow in combination with a
The clay was prepared in 200ml of filtered seawater standingwavecreatedbyflowreboundingagainstthe
byblendingandstrainingthrougha63(cid:1)msieve.The weir at the downstream end. Detailed methods for
clayconcentrationsusedintheflumerunsnecessarily estimating mean shear stress (τ¯0) and time series of
differed among treatments (0.28–0.71gl−1; Table 1), wave-forced shear stress) τ˜0(t) in the 17m flume are
for two reasons. First, in static conditions the thick- describedbyBeaulieu(2003).Basedontheanalysisof
ness of a floc layer decreased with time as the layer flow profiles previously measured in the 17m flume,
dewatered.Exploratoryflumerunswereconductedto the mean flow at 8cm a.b. (u¯8) was related to shear
determinethemassofclaynecessaryforaflushlayer velocity(u∗)viaadraglawforboththemeancurrent
on the test bed (Anderson et al., 2003). Second, an (u¯∗ =0.0419u¯8+0.0574;n=85profiles,r2 =0.99)
objective of these experiments with PAC was to sim- and wave boundary layers (u˜∗ = 0.005u¯8 +0.204;
ulate lower clay concentrations in the water column, n = 39 records at 1.2cm a.b., r2 = 0.78). Critical
with the same total mass depositing on the test bed shear velocity at each stage of particle transport then
(to control for the thickness of the layer on the bed). wasestimatedasu∗crit =((u¯∗)2+(u˜∗)2)1/2,withcrit-
Inallexperimentsthearealloadingsofclaywereless ical bed shear stress estimated as τ0crit =ρu2∗crit. The
than the 200gm−2 suggested by Shirota (1989) but techniqueforestimatingu∗critwascheckedbyrunning
more appropriate for such a shallow water column additional experiments with the test bed filled with
(33–57gm−2; Table 1). For all of the algal and clay sandofthefollowinggrainsizes:125–250,250–500,
concentrations tested, ∼80% RE was expected based and 500–1000(cid:1)m. Estimates of u∗crit for initial mo-
on previous laboratory tests in still water (Sengco, tion of sand compared well to the expected thresh-
2001). oldvalues(Milleretal.,1977).Thecriticalbedshear
During each experiment a floc layer accumu- stress (τ0crit) for resuspension is best described as a
lated on the test bed and consolidated over the “resuspension stress” corresponding to a high rate of
treatment period (3, 9, or 24h). Overall, the exper- erosion,ratherthanathresholdofmotion.
128 S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138
Two-way analysis of variance (ANOVA) was used the flow at 3.4 and 8cm a.b. to collect water samples
to compare τ among the six treatments. We used (10–50ml,dependingonclayconcentration)tocheck
0crit
the least significant difference (LSD) procedure for the OBS data with actual measurements of dry mass.
planned multiple comparisons of the mean τ val- ThesyringesamplerconsistedofaPVCsupportwith
0crit
ues.AssumptionsofANOVAwerecheckedwithnor- draw cord that held a 60ml syringe horizontal above
malprobabilityplotsandtheHartleyF -testforho- theflumebed(2mmdiametersampleopening).
max
mogeneityofvariances.Statisticswereconductedwith To begin each experiment, several carboys of algal
Statistica 5.1 software, tested for significance at α = culture (50l) were added to the flume, with paddle
0.05,andaredescribedbySokalandRohlf(1981). speed at 10cms−1. After ∼30min, syringe samples
(50ml) were collected at 6cm a.b. and preserved to
2.2. Flocssettlinginflow determine the initial cell concentration via triplicate
countsof1mlaliquotsunderacompoundmicroscope.
2.2.1. Racetrackflumeexperimentaldesign Initialalgalconcentrationsintheracetrackflumewere
In the second set of experiments, the natural envi- similarbutslightlylessthaninthe17mflumeexper-
ronmentwassimulatedbydispersingclayonthesur- iments (Table 1). For the flume runs with PAC, stock
faceofflowingwaterseededwithdinoflagellatesand PAC solution was diluted in 18l of deionized water
then increasing/decreasing flow speed. These experi- andsprayedat5lmin−1onthewatersurfaceforafinal
mentsweredesignedtoestimatethecriticalbedshear concentration of 5ppm in the flume. The PAC solu-
stressesfordepositionanderosionofflocs,aswellas tionwasdispersedinasweepingmotionover∼1/3of
to study removal efficiency in flow. The experiments theplanareaofthewatersurfaceusingasumppump,
were conducted in an oval, open-channel flume with hose, and “flat” spray nozzle. The paddle speed then
channelwidth76cmandplanarea17.1m2(“racetrack was set to 3 or 20cms−1 (or remained at 10cms−1)
flume”). Flow in the racetrack flume is controlled by for 30min prior to adding clay to the flume. For a
paddles which enter the water vertically. The total of final concentration of 0.28gl−1 in the flume, 1.2kg
six experiments explored three different initial flow (wet;47.33%drymass)ofIMC-P4claywasblended,
speeds with and without PAC (Table 1). Replication strainedthrougha63(cid:1)msieve,suspendedinatotalof
wasnotpossibleduetotheneedforlargeculturevol- 54lof10(cid:1)mfilteredseawater,andsprayed(asabove)
umestoseedthe2052lflumevolume.Toreducevari- at5lmin−1 onthewatersurface.
abilityinourmeasurements(suchasalgalcelllossor For the experiments that started at 3 or 10cms−1,
changesinbottomroughness),thetestbedwassolely flocs settled for 3h prior to the first resuspension
theplasticbottompanels. “event” in which paddle speed was increased every
Prior to each experiment, the flume was filled to 5min in 1cms−1 increments to 25cms−1. The sec-
12cm water depth with 10(cid:1)m filtered seawater. We ond and third resuspension “events” occurred after 3
placed optical backscatter sensors (OBS-3; D&A In- and 9h settling periods, respectively. For the exper-
strumentCompany,PortTownsend,WA,USA)at3.4 iments initially at 20cms−1, the paddle speed was
and 8cm above bottom to measure suspended mass stepped down by 2cms−1 every 3h. Because turbid-
inthewatercolumnevery10sduringtheentireflume ity in the flume during these experiments precluded
run.Duetoturbidityintheflume,visualcriteriacould the use of an LDV to measure flow speeds, flow
not be used to determine critical shear stresses. Prior profiles at paddle speeds 3–25cms−1 were recorded
toplacementintheflume,theOBSsensorswerecal- after the flume was cleaned and filled to 12cm depth
◦
ibrated to suspended mass by mixing clay in five in- withseawaterat20 C.Alsoduetoturbidity,wewere
crementsof0.07gl−1inaseparatetankfilledwithfil- unable to determine the size spectra of flocs during
teredseawater(seeBuntetal.,1999fordiscussionof the experiments with a Laser In Situ Scattering and
OBScalibration).Watersamples(10–50ml)fromthe Transmissometry instrument (LISST-100; Sequoia
tankwerefilteredoverpre-weighedGFFfilters,rinsed Scientific Inc., Bellevue, WA, USA). Water samples
◦
with Milli-Q water, and dried overnight at 60 C be- for cell counts were withdrawn just prior to the addi-
forere-weighingfordrymass.Also,duringtheflume tionofclayandjustpriortoeachresuspensionevent.
runs a “syringe sampler” was directed upstream into Aliquots (10–25ml) of samples with low cell counts
S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138 129
weresettledinUtermöhlchambersandcountedonan 2001). Although shear stress is not equal across the
invertedmicroscope.Wedidnothaveenoughculture entire area of the flume bottom (e.g. near the walls),
to run a control experiment for a 24h period without at 3cms−1 a floc layer accumulated everywhere ex-
theadditionofPACorclay. cept for a ∼15cm section immediately underneath
thepaddlethatextendedclosesttothebottom.Tocal-
2.2.2. Dataanalysesforracetrackflumeexperiments culateremovalefficiencyofalgalcellsfromthewater
As shown for other cohesive sediments at concen- column, we used the following equation: %RE =
trations <0.5gl−1 (Maa et al., 1992), OBS output 100×(C0−Ct)/C0,inwhichC0isthecellconcentra-
voltageswerelinearlyproportionaltosuspendedmass tion just prior to clay addition and Ct is cell concen-
at the concentrations used in the racetrack flume ex- trationatalatertime(e.g.aftera3hsettlingperiod).
periments(r2 =0.99fortankcalibrations).TheOBS
valuesalsowerewell-correlatedwithsuspendedmass
measuredfromsyringesamplesduringtheflumeruns 3. Results
(n=24,r =0.95,P<0.01).Therefore,weusedthe
raw data from the OBS time series to determine the 3.1. Flocssettlinginstillwater
timeduringeachresuspensioneventatwhich:(1)sus-
pended sediment concentration first increased above Duringthe17mflumeexperiments,thebulkofthe
background level; and (2) maximum input of sedi- flocs in the enclosed volume settled to the test bed
menttothewatercolumnoccurred,priortolimitation in ∼15min. Once exposed to flow, initial motion oc-
ofsedimentonthebed(indicatedbymaximumslope curred at about 25% the resuspension stress (half the
in the OBS record). We calculated the slope of the flowspeed)andbedloadatabout40%theresuspension
OBS data via linear fits to all 2min segments of the stress (Table 2). Overall, resuspension stress ranged
time-seriesrecords.Weassumedthatshearstressdur- by a factor of 1.5, from 0.059Pa (u∗ = 0.76cms−1;
ing (1) above was the critical shear stress (τ0crit) for 3hsettlingwithPAC)to0.088Pa(u∗ =0.93cms−1;
erosionoftheflocs,andshearstressduring(2)above 24h settling without PAC). Critical shear stresses
was the stress (τb) at which the maximum erosion (τ0crit) for bedload and resuspension of clay/algal
rate occurs before supply limitation. Similarly, τ flocs differed among the six experimental treatments
0crit
for deposition, or τ , of flocs was estimated when (Table 2). Both factors (settling time and PAC) were
d
suspended sediment concentration first decreased significant in the two-way ANOVA for resuspension
during the flume runs in which flow was stepped stress (with no interaction effect; Table 3). Multiple
downfrom20cms−1,withtheminimumslopeinthe comparisons of the mean values for resuspension
OBS record corresponding to the maximum rate of stress (plotted in Fig. 2) revealed the following, with
deposition. treatments listed in order of τ0crit increasing to the
Tocalculatebedshearstressduringtheaboveobser- rightandnonsignificantdifferencesconnectedwithan
vations,thepaddlespeed(u )wasassociatedtobotha underline: 3PAC < 9PAC324PAC < 924. These re-
p
mean(u¯∗)andwave-forced(u˜∗)shearvelocityviathe sultssuggestthatthegreatestincreaseinconsolidation
following drag laws: u¯∗ = 0.04up+0.0726 (n = 10 for the 2mm thick layer occurred between 3 and 9h.
profiles,r2 =0.99)andu˜∗ =0.023up+0.1081(n= For a given settling time, τ0crit was lower in the runs
11 records at 1.2cm a.b., r2 = 0.97). Time series of with PAC, although this factor was not significant in
wave-forcedshearstressweredeterminedatfrequen- atwo-wayANOVAofτ0crit forbedload(P =0.08).
cies based on the paddle speed divided by spacing
between the paddles. Total shear velocity (u∗) and 3.2. Flocssettlinginflow
bed shear stress (τ ) then were determined using the
0
equations given previously. For each of the resuspen- Afloclayeraccumulatedontheracetrackflumebed
sionevents,wefitthedatatoastandardlinearerosion during experiments with the paddle speed at 3 and
model, given by E = M(τb/τ0crit −1), in which E 10cms−1.Fortheseexperimentswithsubsequentin-
is erosion rate (kgm−2s−1) and M is an empirical creases in flow speed to simulate tides, OBS records
constant(SanfordandHalka,1993;SanfordandMaa, clearly show three periods of settling followed by
130 S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138
Table2
Criticalbedshearstress(τ0crit)forthreestagesoftransportofclay/algalflocsinthe17mflume
Experimentdescription Settlingtime(h) τ0crit (Pa)
Initialmotion Bedload Resuspension
Clay 3 0.017±0.001 0.028±0.003 0.069±0.006a
Clay 9 0.019±0.001 0.029±0.000 0.081±0.002
Clay 24 0.019±0.001 0.032±0.002 0.088±0.005
Clay+PAC 3 0.018±0.000 0.026±0.001 0.059±0.002
Clay+PAC 9 0.017±0.000 0.027±0.002 0.068±0.004
Clay+PAC 24 0.019±0.001 0.031±0.001 0.071±0.002
Values are mean±S.D. for three replicate flume runs. Results for resuspension stress are plotted in Fig. 2. For critical shear velocity,
u∗crit=(τ0crit/ρ)1/2,substituteρ=1.0225gcm−3.
aContainsanoutlierbasedonpost-experimentobservationsofvideo.
Table3 resuspension “events” (Fig. 3A–D; only one of the
Two-way ANOVA for the 17m flume results for resuspension OBS sensors yielded accurate data without drift). At
stressforclay/algalflocs(plottedinFig.2) 3cms−1 the bulk of the flocs settled in <1h. At
Source Degreesof SS MS F P-value 10cms−1 manyoftheflocshadnotsettledevenafter
freedom 9h(especiallyevidentintherunwithPAC).Asanex-
Settling 2 0.000776 0.000388 23.99 0.000064 ampleofOBS“slopeanalysis,”Fig.4showshowτ
0crit
time andτ atthemaximumerosionrateweredetermined
b
PAC 1 0.000751 0.000751 46.42 0.000019
during the first resuspension event for the flume run
Interaction 2 0.000036 0.000018 1.12 0.36 thatstartedat3cms−1withoutPAC.Thefirstincrease
Error 12 0.000194 0.000016
in slope in Fig. 4 occurred when τ = 0.040Pa
0crit
Total 17 0.0018 (u∗crit = 0.63cms−1; paddle speed 11cms−1). The
The two sources of variation in the 3×2 factorial design were maximumslopeinFig.4,indicatingmaximumerosion
settlingtime(3,9,or24h)andPAC(presenceorabsence).
rate prior to sediment limitation on the bed, occurred
at τb = 0.060Pa (u∗ = 0.76cms−1; paddle speed
14cms−1).Table4liststheseshearstressesforallthe
0.10 resuspensioneventsduringtheracetrackflumeexperi-
No PAC
ments,aswellastheslope,M,calculatedforthelinear
With PAC
) 0.09 erosion rate model. For each experimental treatment,
Pa therewasverylittletonodifferenceinτ0critamongthe
ress ( 0.08 tdhurreiengretshuespreunnssiothnaetvsetnartste.dInatge3ncemrals,−t1hewτe0rceritslvigalhutelys
st
n greater than τ for bedload in the 17m flume ex-
o 0crit
si 0.07 periments, and the τ values during the runs that
n 0crit
pe started at 10cms−1 were slightly less than resuspen-
s
u
es 0.06 sionstressinthe17mflumeexperiments(compareto
R Table2).Fortherunsthatstartedat3cms−1,although
therewasnodifferenceinτ withpresence/absence
0crit
0.05
of PAC, the addition of PAC yielded maximum ero-
3 9 24
sion rates at lower shear stresses (higher values for
Settling time (hr)
M; Table 4). In contrast, for the runs that started at
10cms−1,theadditionofPACloweredτ andalso
Fig.2.Resuspensionstressforclay/algalflocsinthe17mflume. 0crit
Eachsymbolindicatesmean±S.D.forthreereplicateflumeruns lowered the erosion rate constant, M (however, note
(valueslistedinTable2). the much lower increase in suspended sediment con-
S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138 131
Fig. 3. Optical backscatter (OBS) plots for racetrack flume runs. Linear fits to 2min subsets of data are superimposed on plots A, B, D,
andE.(AandB)Paddlespeed3cms−1.OBSat3.4cmabovebottom.(A)NoPAC.ThehighlightedboxindicatesdataplottedinFig.4.
(B) With PAC. (C and D) Paddle speed 10cms−1. (C) No PAC. OBS at 6cm a.b. Gaps indicate manual displacement of the OBS to
3.4and8cma.b.duringresuspensionevents.(D)WithPAC.OBSat8cma.b.(E)Paddlespeed20cms−1.NoPAC.OBSat3.4cma.b.
Paddlespeedwasdecreasedby2cms−1 every3h.Arrowsindicatedecreasesto12,10,and8cms−1.
centration due to reduced deposition during the run dle speed at 10cms−1, the minimum slope of the
withPAC;Table4). OBS record indicated the maximum deposition rate
For the experiments that started at 20cms−1, atτb =0.034Pa(u∗ =0.58cms−1;51%massdepo-
significant deposition did not occur until the pad- sition). After the 3h settling period at 8cms−1, 84%
dle speed was stepped down to 12cms−1 (arrow in ofthemassdeposited.
Fig. 3E; one of the experiments ended prematurely IntheabsenceofPACatboth3and10cms−1,algal
with a power outage). At 12cms−1, a layer began cell counts were greatly reduced during all settling
to accumulate on the bed, indicating τ for de- periods, with removal efficiencies exceeding ∼80%
0crit
position, or τd = 0.046Pa (u∗crit = 0.67cms−1; (Fig. 5 and Table 5). However, removal efficiencies
14–18% mass deposition). Very little (2–8%) mass were considerably lower when the experiments were
deposited while the paddle speed was at 14cms−1 repeatedwithPAC(15–78%;Table5).Intheabsence
(τ0 = 0.060Pa; u∗ = 0.76cms−1). With the pad- of PAC, algal cell RE was slightly greater than the
132 S.E.Beaulieuetal./HarmfulAlgae4(2005)123–138
100 1000
e
op 50
sl
S
B 0
O
) 100
-50 -1L
m
OBS s
2000 Stress 0.16 ell
c
(
10
a
aw data (mV) 11050000 00..0182 ear stress (Pa) Heterocaps 1 Flow spe3ed (cm s-1)
r h
BS d s 3 PAC
O 500 0.04 Be 10
10 PAC
20
0 0 0.1
0 20 40 60 80 100 Initial Prior 1st 3hr 2nd 3hr 9hr
to clay
Time elapsed (min)
Fig. 5. Cell counts for racetrack flume runs. Symbols indicate
Fig. 4. OBS “slope analysis” for the first resuspension event in
mean±range of two or three counts of each 50ml sample of
Fig.3A.Upperpanelshowstheslopeoflinearfitsto2minsubsets
flumewater.Notelogscale.Meanvalueswereusedtodetermine
of OBS data. Dashed lines indicate first increase in OBS slope
removalefficiencyreportedinTable5.
andmaximumslope(upperpanel)andintersectionwithbedshear
stresscalibratedtopaddlespeedintheflume(lowerpanel).Each
stepintheplotforbedshearstressrepresentsa1cms−1 increase percentage of clay mass that deposited to the flume
inpaddlespeed(from3to25cms−1). bed(Table5).Incontrast,inthepresenceofPAC,algal
cellREtendedtobelessthanthepercentageofmass
deposited(Table5).Bothexperimentsat3cms−1had
Table4
UsingracetrackflumeresultstomodelerosionrateasE=M(τb/τ0crit−1)
Experiment Flowspeed Resuspension Criticalstress, Stress(τb)at Increaseinsediment Changein M
description (cms−1) event τ0crit (Pa) maxsediment concentration(gl−1) time(h) (mgcm−2h−1)
duringsettling input(Pa)
Clay 3 First3h 0.040 0.060 0.068 0.27 6.1
Second3h 0.034 0.060 0.108 0.37 4.6
9h 0.034 0.060 0.115 0.42 4.3
Clay+PAC 3 First3h 0.034 0.053 0.081 0.36 4.8
Second3h 0.034 0.053 0.107 0.33 6.9
9h 0.034 0.053 0.101 0.35 6.2
Clay 10 First3h 0.060 0.083 0.097 0.32 9.4
Second3h 0.053 0.083 0.102 0.43 5.0
9h 0.060 0.092 0.080 0.36 4.9
Clay+PAC 10 First3h 0.053 0.083 0.058 0.33 3.7
Second3h 0.046 0.083 0.042 0.45 1.4
9h 0.046 0.092 0.038 0.52 0.9
Values are single estimates derived from OBS “slope analysis” (depicted in Fig. 4). Total volume 2052l and area 17.1m2 in racetrack
flume.MispresentedinunitscomparabletoSanfordandHalka(1993).