Hydrobiologia 395/396: 161–172,1999.D.M. Harper, B. Brierley, A.J.D. Ferguson & G. Phillips (eds), TheEcological Bases for Lake andReservoir Management.© 1999Kluwer AcademicPublishers. Printedin theNetherlands.

161

Biological control of cyanobacteria:principles and possibilities

D.C. Sigee,R. Glenn,M.J.Andrews,E.G.Bellinger, R.D. Butler, H.A.S.Epton& R.D. HendrySchoolof Biological Sciences, University of Manchester, 1.800Stopford Building, Oxford Road,Manchester, U.K.

Key words: actinomycetes, biologicalcontrol,bluegreenalgae,cyanobacteria,protozoa,Streptomyces

Abstract

A rangeof naturally occurring organisms are available for the biological control of cyanobacteria: including vir-uses, bacteria, fungi, actinomycetesandprotozoa.Developmentof these organisms as biological control agentsinvolves isolation from environmental samples, characterisation of anti-cyanobacterial activity, microcosm andlarge-scalefield experimentsandfinal developmentof a biologicalcontrol lake managementstrategy. Two groupsof antagonist areconsideredin detail – actinomycetes(e.g.Streptomycesexfoliatus, modeof action by productionof a lytic agent)andprotozoa(Nuclearia delicatulaandNassula tumida, modeof action by predation).Theeffect-ivenessof biologicalcontrol agentsin thelakeenvironmentdependsonarangeof biologicalandphysico-chemicalfactors.Variousstrategiescanbe implementedto optimisetheir activity.

Intr oduction

Blooms of freshwater cyanobacteria, particularly inthe generaMicrocystis and Anabaena, have causedincreasing problemsin recentyears. These have fre-quently resulted in a deterioration of water quality,with adverse effectson lake ecology, livestock, hu-manwater supply and recreationalamenity. Themostdirectform of control involves theapplication of algi-cides, but this is expensive and potentially damagingto the environment.Even wherechemicaltreatmentshave no immediatedamagingeffectson lakes thereis the risk of accumulationof harmful concentrationsin bottom sediments (Mason, 1996). An alternativeapproachfor the direct elimination of nuisancealgaeinvolvesthe application of biological control agents.In this context, they may be defined as microbial or-ganisms that have the ability to destroy, or limit thegrowth of, target algae.This paperwill consider theinvestigation and potential use of such agents in thecontrol of cyanobacteriain the lakeenvironment.

Rangeand modeof action of potential biocontrolagents

Activity of potential biocontrol agent rangesfromhighlyspecificparasitismto non-specificreleaseof ex-

tracellular lytic products. Potential biocontrol agentsinclude viruses, bacteria, fungi, actinomycetes andprotozoa.

Viral agents

The first isolation and partial purification of a virusof cyanobacteria (cyanophage)was by Safferman&Morris (1963). The cyanophagewas called LPP-1because it lysed speciesof thegeneraLyngbya,Plec-tonema and Phormidium. Subsequentsearchesforcyanophageson aworld-widescalehavemadeit clearthat they are extremely widespreadin both freshwaterandmarineenvironments(Ohki & Fujita, 1996;Padanet al., 1967).Therole of cyanophagesin determiningthecyclic bloomsof cyanobacteria, and their potentialuseasbiocontrol agents, hasbeensuggestedeversincethese viruseswere discovered.

The rapid generation time of cyanophagesmakesthem attractive agents for controlling blooms of cy-anobacteria.In the caseof LPP-DUN1, for example,the generation time is 10 h and the burst size is about100 phageparticlesfor eachinfectedcell makingupthePlectonema boryanumfilament(Daft etal., 1970).

An important consideration in the potential useof cyanophagesas biocontrol agents is the rapid ap-pearanceof resistant host mutants(Padan& Shilo,1973).Thesemayhavechangesin thealgalcell envel-

162

ope,preventing phageadsorption (Padanet al., 1967).Barnetet al. (1981)isolatedtwo typesof Plectonemaresistant to wild-type LPP-DUN1, however resistanthost strainsweresusceptible to attackby mutant cyan-ophagestrainsandit wassuggestedthatmutantphagesmaypresent amethodfor thecontrol of cyanobacteria.Thehigh degreeof host specificity, occurrenceof res-istant host mutantsandeffectof environmental factorsall contribute to the complexity and unpredictabilityof cyanobacterial/phageinteractionsin the field. Dif-ficulties involved with producing large amountsofactive inoculumalso presentproblemsin theeffectiveuse of cyanophagesas biocontrol agents in the lakeenvironment.

Bacterial agents

Bacteriawhichcanlysecyanobacteriahavebeenasso-ciatedwith suddendeclinesin cyanobacterial biomass(Fallon & Brock,1979),althoughnoconclusionswerereachedasto whetherthebacteriaactedastheprimarycausativeagent(pathogens) or whether they were act-ing as saprophytes, decomposing deadalgal materialresulting from other primary processes.Daft et al.(1975)founda directstatistical correlation (5% level)betweenthechlorophyll-a concentration of eightbod-ies of water andthe abundanceof lytic bacteria. Cy-anobacteriolyticbacteriaappearto act in threemajorways: production of extra-cellular products; contactlysisand entrapment lysis.

The culturefiltratefrom a Bacillussp., isolatedbyReim et al. (1974),wasfoundto lyse seven generaofcyanobacteria (including AnabaenaandMicrocystis).The heat stability and small molecular size of thediffusible inhibitory factor suggested that it was anantibiotic substance. In a laterstudy, complex volat-ile products released by Bacillus sp. were shown tobeparticularly inhibitory to filamentouscyanobacteria(Wright & Thompson, 1985). Wright et al. (1991)subsequently implicated isoamyl alcohol (3-methyl-1-butanol)as onecyanobacteriolyticvolatile product.Recently, two speciesof Flexibacter (F. flexilis andF. sancti) were isolated from domestic sewage andfound to lyse the cyanobacterium Oscillatoria willi-amsii (Sallal,1994).Inhibitionof photosyntheticelec-tron transport reactions, glycolate dehydrogenaseandnitrogenase activity were observed. An extracellularmetabolite in the cell filtratecausedgrowth inhibition(Sallal, 1994). Slab gel electrophoresis was used toidentify thecompoundasa lysozyme.

Daft et al. (1985) questionedthe significanceofgenerally released extracellular lytic compoundsinnature due to rapid dilution and subsequentloss ofactivity. Morespecificbacterial/cyanobacterialassoci-ations, wherethe activity of lytic compoundsis morelocalised, may be moreefficient. Two particular typesof association,bacterialattachment(contactlysis) andentrapmentof cyanobacteria (entrapmentlysis), havebeenstudiedin detail.

Daft & Stewart (1971) isolated four strains ofMyxobacteriales (designated CP 1-4) which causedlysis of over 40 strains(from all orders) of cyanobac-teria, causing disruption within 20 min of contact.Although no detectableextracellular productswereproducedby the CP isolates, it was suggested thatenzymeson the bacterial surface might be effect-ive in causing lysis. Lysis of bacterialcell walls hasbeendemonstrated using an enzymeextracted fromMyxobacter (Ensign & Wolfe,1965).

The effectiveness of contact lysis dependson anumberof factors, including population levels of thealgaeandmyxobacteria, andnutrientstatusof thewa-ter medium. Whenintroducedinto laboratory culturesor field samplesof cyanobacteria,at least 106 cellsml−1 myxobacteria were required to cause signific-antalgal cell lysis (Daft & Stewart, 1971;Daft et al.,1975;Shilo, 1970).Fraleigh & Burnham(1988)pos-tulatedthat above 106 cellsml−1predation will occur,with little dependence on inorganic nutrient concen-trations or host density. Below this population levelthe growth of the predator population may be relatedto the inorganic nutrient status of the water. Fallon &Brock(1979)suggestedthatlyticbacterial populationsof <103 cells ml−1were insufficient to cause signi-ficant cell lysis in natural host populations, and thatlow densities of lytic bacteriaencounteredin the en-vironment maybedueto insufficient inorganicfertility(Fraleigh & Burnham,1988).

Burnham et al. (1981) described cyanobacteriallysis by a memberof the Myxococcusgroup(Myxo-coccusxanthus, designated strain PC02), in whichthecyanobacteriumPhormidiumluridumvar. olivaceawas entrappedand encapsulated within colonies ofthe bacterium. The colonial spherules entrappedthecyanobacterial prey and then lysed the cells by thereleaseof a ‘l ysozyme’-likeenzyme.

Fraleigh & Burnham(1988) tested the effective-ness of myxococcalpredationin natural cyanobac-terial populationsover a rangeof natural densitiesandnutrientconcentrationsagainst two sensitive cy-anobacteria(Nostoc muscorum andPhormidium lur-

163

idum). The results suggesteda threshold myxococcaldensity for significant cell lysis by M. fulvus andthat concentrationsof inorganic nutrients in eutrophiclakesmightbeinsufficientto supportgrowth of myxo-coccalpopulationsto this thresholddensity. Burnhametal. (1984)suggestedthatwheninitial predatordens-ities were less than the threshold density, the myxo-cocci may derive their necessary organic nutrientsfrom secretionsproducedby cyanobacterial hosts orby preying on thecyanobacteria withoutsignificantlyreducing thenumberof cells in thehost population.

Daft et al. (1985)consideredmyxococcito bepo-tentially the bestbacterial biologicalcontrol agentsofbloom-forming phytoplankton on the basis of sevenattributes which they consideredwould make an or-ganism agoodpredator. Thesewere:(1) adaptabilityto variationsin physical conditions;(2) ability to searchor trap;(3) capacityand ability to multiply;(4) prey consumption;(5) ability to survive low prey densities;(6) widehost range;(7) ability to respondto changesin thehost.

Fungal pathogensof cyanobacteria

Parasitism of a cyanobacterium(Oscillatoria agardhiivar. isothrix) by the chytridiaceousfungus Rhizo-phidium planktonicum was first demonstrated byCanter & Lund (1951). More recently Daft et al.(1985) consideredthe chytridiaceousfungi to be oflimitedusein thebiologicalcontrol of bloom-formingorganisms because of the apparentobligate natureof theseparasitesand difficulties in their large scaleculture.

Safferman& Morris (1962) tested 142 culturesof non-chytrid fungi and found that 4.2% formedproductswhich demonstratedspecific antagonistic ef-fectson eithergreenalgaeor cyanobacteria.Thestudyof non-chytrid fungiantagonistic to cyanobacteriawascontinuedby Redhead& Wright(1978).Sixty-twooutof 70 pure cultures able to lyse cyanobacteria werefungi representing thegeneraAcremonium, Emericel-lopsis, andVerticilliu m. All isolateslysedAnabaenaflos-aquaeand,in most cases, severalotherfilament-ousandunicellular cyanobacteria.

Lysisof cyanobacteriaby AcremoniumandEmeri-cellopsis sp. was associated with the formation ofdiffusible heat-stable extra-cellular factors. Redhead& Wright (1980)extracted and partially purified the

β-lactamantibiotic cephalosporin C from liquid cul-turesof thefungi Emericellopsis salmosynnemata andAcremoniumkiliense. Theseauthorssuggestedthattheisolatesprobably produced small quantities of anti-biotic, which affect cyanobacterialcells only whenthey arein close proximity to the fungus(antibioticsmay concentrate to antagonistic levels in the mucoussheath surroundingmany cyanobacteria), asoccursintheaggregatesproducedin agitatedliquid culture.

Antagonism of cyanobacteria by actinomycetes

Actinomycete antagonistsof cyanobacteriahave beenisolated from both freshwater and soil environments.Investigations on reservoirs have shown Streptomy-ces to be the most commongenusof actinomycetefound in freshwater sites (Silvey, 1973),occurring inthe bottom sediments associatedwith vegetation andalong theshoreline frequently associatedwith mats ofcyanobacteriaor greenalgae.Growth in the varioustypesof aquatic environmentis also dependenton thenatural biological productivity of the water. Studiesof South-westernUnited States reservoirshaveshownthat the greater the primaryproductivity, as measuredby C14 uptakeandATPmeasurements, themorecopi-ousthe actinomycetedevelopment(Silvey & Roach,1975).Thoughclearly associated with cyanobacteria(andableto use algaeasa nutrientsource)the exactplacethat actinomycetesoccupied in the biodynamiccycle could not be completely evaluated. Daft et al.(1985)also mentionedthe lack of detailed studiesofcyanobacteria-actinomyceteinteractions.

Soil isolatesof actinomycetesmayalso havepotentanti-cyanobacterialproperties. Safferman& Morris(1962)assayed the culture filtrate of soil-isolated ac-tinomycetes for anti-microbial properties and foundthat 213 out of 403 strains tested showed inhibitoryeffects; 90%of these agentsdisplayedspecific effectsagainst thecyanobacteria.Variousactinomyceteisol-atescould effectively eliminate growth of cyanobac-teria (includingspeciesof Anacystis, Fremyella, Lyn-gbya, Nostoc, PhormidiumandPlectonema). Whyteetal. (1985)foundthat Streptomycesachromogeneswasthemainspeciesin soil thatlysedAnabaenacylindricaandTolypothrix tenuis. Al-Tai (1982)describedan ac-tinomycete(AN6) isolatedfrom Iraqi soil which hada very wide host range.Theextracellular products ofAN6 were able to lyse cyanobacteria, fungi, bacteriaand greenalgae.Resistant cyanobacterialcells werenot found.

164

In a generalsurvey, the lytic properties of variousisolatedstrainsof Streptomyceshave beencomparedwith those of other actinomycetesand myxobacteria(Salton,1955).Relatively little appliedwork hasbeencarriedout on the potentialuse of actinomycetesasbiocontrol agents. Work by Martin (1976),in attemptsto develop cheapand environmentally safe algal con-trol procedures, resulted in the isolation of an actin-omycetestrain which was specifically lytic againstcyanobacteriabut not against several bacterial testspecies. It was found that cell-free filtrates from theactinomcyetehadlytic activity against threeAnabaenastrains. A few studies have beencarried out on theeffectivenessof actinomycetesagainst bloom-formingcyanobacteria(Bershovaet al., 1968;Rubenchik et al.,1965).Actinomycetesmayalso be indirectly involvedasbiocontrolagentsin theinhibition of cyanobacteriaby barley straw, wherethey are part of a wide rangeof micro-organisms that produceanti-cyanobacterialsubstances(Newman& Barrett, 1993)

Protozoanpredators of cyanobacteria

Within aquatic ecosystemsprotozoahavean importantrole in the reduction of phytoplanktonpopulationsbygrazing (Canter et al., 1990; Pace& Orcutt, 1981).Cyanobacteria in particular provide a suitable foodsourcefor several generaof protozoa, including theciliate Nassula (Canter et al., 1990), the flagellateOchromonas(Cole & Wynne, 1974) and the amoe-baeAcanthamoeba(Wright et al., 1981), Mayorella(Laybourne-Parryet al., 1987)andNuclearia (Yamo-moto, 1981).

Predation of cyanobacteria by protozoahasbeenobserved in samples taken directly from the naturalenvironment(Canter et al., 1990; Cook et al., 1974;Laybourn-Parry et al., 1987), in laboratory experi-ments (Dryden & Wright, 1987; Yamomoto, 1981)and in biocontrol field experiments(Brabrandet al.,1983).Suchwork hascontributed to our understand-ing of cyanobacteria–protozoaninteractions, provid-ing information on the range of protozoa capableof grazing cyanobacteria (Dryden & Wright, 1987)andtheir potential role in biocontrol (Brabrandet al.,1983). The effectiveness of protozoa as biocontrolagents will dependupon a numberof factors – in-cludingprotozoangrowth andgrazingrates, predationspecificity, cyanobacterialgrowth ratesandpredationrates by higher organisms such as copepodson theprotozoa(Brabrandet al., 1983).

Figure 1. General scheme for the testing and development ofbiocontrolagentsto beused against lake cyanobacteria.Thecentralpart of thescheme (work carriedout in thelaboratory) is illustratedto theright of theflow diagram.

Casestudieson two typesof potential biocontrolagent: actinomycetesand protozoa.

A generalschemefor the testing and developmentofbiocontrol agents to be used against lake cyanobac-teria is outlined in Figure 1. This scheme is broadlysimilar to thedevelopmentof biocontrol agentsagainstothertargetorganisms(e.g.plant pathogenic bacteria;Sigee, 1993), and follows a well-defined sequence.This involves: isolation of naturally occurring ant-agonistic strains from environmental samples (Fig-ure1a,b); subculture, purification and identification ofparticularisolates(Figure1c); characterisationof anti-cyanobacterial activity in the laboratory (Figure1d);small-scale (microcosm) and large-scale field trials(Figure 1e); and finally developmentof a biocon-trol strategy anduse of the biocontrol agents in lakemanagement. The final stage (Figure 1f) would al-most certainly also involve commercial productionof the biocontrol agents, with assessment of cost-effectivenessandbulk production.

The following section considers examples of twoparticular groupsof antagonist, with quite differentmodesof action:actinomycetesandprotozoa.

Actinomycetes

Theability of selectedisolatesof actinomyceteantag-oniststo control populationsof cyanobacteriahasbeenexaminedinboth laboratory (algallawnandliquidcul-ture experiments) andfield (microcosm) experiments.

165

Laboratory studiesprovide a rangeof useful inform-ation on the activity of potential biocontrol agents,including the rangeof algaethat are lysed, changesincyanobacterial andactinomycetepopulationsduringthe interaction sequence,level of antagonist inoculumrequiredfor complete lysisandmodeof action.

Actinomycetes were isolated by applying dilutesamples of soil or freshwater to agar cultures of cy-anobacteria(algallawns), thenstreakingmaterialfromareasshowing lysisonto yeast–dextrosemedium(10gyeast extract; 10 g D-lucose; 1000 ml distilled wa-ter). Actinomycetes were purified by picking singlecolonies and repeatedlysub-culturing. Pure strainswerethenre-tested against agarculturesof the samecyanobacterialspecies. One specific isolate, Strep-tomycesexfoliatus strain RG12, proved particularlyeffective in laboratory experiments and was used asthemajor test organism.

Algallawnsprovidearapidandconvenientmethodto determine the range of algal species that arelysed by particular actinomycete isolates, and havebeenwidely usedby previousworkers(Al-Tai, 1982;Burnhamet al., 1981;1984;Daft et al., 1975;Martin,1976; Stewart & Daft, 1977; Yamamoto & Suzuki,1990).

Thealgal lawn assay canbeusedin variousways.Details of one commonly used approachare showndiagrammatically in Figure 2a,with anexample illus-tratedin Figure2b. Replicatecores(0.5cm diameter)werecut from 3-week-old S. exfoliatus agarculturesandinverted onto lawnsof cyanobacteria, so thatant-agonist and cyanobacterium were in direct contact.Corescut from un-inoculatedplateswereusedascon-trol. Preparationswere incubated at 20◦C, continuous500 lux light andexaminedafter 3 days, any yellow-ing or clearingof the treatmentcomparedto controlbeing recordedaspositive. Thespeedof developmentand diameter of the lysis zone also gives a meas-ure of antagonistic effectiveness underthe prevailingexperimental conditions.

Nine out of 13 cyanobacteria tested were lysedwithin 3 days by S. exfoliatus, including membersofthe major bloom forming generaAnabaena, Micro-cystis and Oscillatoria. Filamentous and unicellularcyanobacteriawereequally susceptible. Coloniesof S.exfoliatus were observed growing over thecyanobac-terial lawn. Clear lysis zoneswere seenbeyond theleading edge of actinomycete growth, indicating anextracellular lytic process(Figure2b).

Antagonist isolateswere tested against a rangeofcyanobacteria in liquid culture, underdifferentphys-

Figure 2. Algal lawn assay for testing actinomycete isolates. (a).Diagramof technique,using antagonist cores. The cyanobacteriallawn comprises auniform layerof algaeoverlying theagarand wasnormally usedwithin 24 h of production.Coresfrom actinomycetecultures wereinvertedontothealgallawn andtheiractivity assessedin terms of lytic activity. (b) Appearanceof lawn assay, 5 daysafter inversion of coresfrom S. exfoliatus culture onto a lawn ofLyngbyasp. Notelysis zonesextendingfrom theedgeof actinomy-cetegrowth. Thecentralcontrolcore(noactinomycete)hasno lysiszone.

ical conditions and algal/antagonist ratios. Resultsfrom a typical liquid culture experimentareshown inFigure3. S. exfoliatus was culturedfor 3 daysin inor-ganic salts–starch medium then5ml (106 CFU ml−1)addedto replicate flasks containing 100-ml culturesof Anabaenacylindrica. Incubation wasat20◦C, con-tinuouslight (1000lux) with continuous(100r.p.m.)stirring.Ascontrols, 5 ml sterileactinomycetegrowthmediumwereaddedto replicateA.cylindrica cultures.

Coloniesof S. exfoliatuswereobserved asdiscretefloccules in thealgal culture, frequently directly asso-ciatedwith algal filaments. Over a period of 7 daysthe presence of actinomycete antagonist sharply re-duced the biomassof A. cylindrica, as measuredby

166

Figure3. Liquid culture assay for testing actinomyceteisolates. (a)Effectof StreptomycesexfoliatusstrainRG12onthegrowth (changein optical density) of Anabaenacylindrica in 100 ml of medium.(b) Appearanceof antagonist treated(left flask) andcontrol (rightflask) 7-day cultures of Anabaenacylindrica described in (a). Lysisof thecyanobacteriumin thepresenceof antagonist results in almostcompleteclearing of thealgalculture.

optical density (Figure3a) andchlorophyll-a concen-tration, leading to almost complete clearing of thecyanobacterial culture (Figure 3b). S. exfoliatus ini-tially resulted in filament fragmentation, followedbycell lysis. Heterocysts were apparently unaffectedbyantagonist treatmentandaccumulatedas single cellsin lysedcultures.In parallel experiments,lysisof Ana-baenacultureswas accomplished by the addition ofas little as0.05 ml S. exfoliatus inoculum (106 CFUml−1) in 100ml of algalculture.

Thepossible use of antagonists for environmentalcontrol of cyanobacteria can be tested progressivelyusing asequenceof experimentalapproaches: laborat-

Figure4. Testing of actinomyceteisolateson environmentallawns.Testing of fouractinomyceteisolateson an algal lawn obtainedfroman environmental (bloom) sample. The lawn (composed largely ofMicrocystis aeruginosa), shows large lysis zonesaround two ant-agonist cores(right/left), but the top and bottom coreshave littleactivity (seealso Figures2a,b).

ory testing of pure cultures of environmental isolates,laboratorytesting of freshly-collectedenvironmental(phytoplankton) samples (Figure 4), small (micro-cosm) and large (mesocosm) enclosed-volume ex-periments in lakes and finally testing in the openwater against natural blooms. In the early stages ofantagonist testing there is probably most value incontrolled, replicatedlaboratory or microcosm exper-iments. Simply adding biological control organismsto openwatersmay createunforseenproblemsduetovariablesnot recreatedin laboratory experiments.

Environmental testing of S. exfoliatus hasso farprogressedvia intermediate laboratory studies to themicrocosm stage. Laboratory trials with freshly isol-ated field samples, using both lawn (Figure 4) andliquid culture assay, have demonstrated clear lysisof lake phytoplankton – including Microcystis andAnabaenabloomsamples.

Microcosmsprovidea useful intermediatestepto-wards larger-scale environmental experiments. Theyinvolve antagonist testing underphysical and biolo-gical conditionsthat approximate to the natural stateand that the effect of these agents can be quantifiedwithin a definedvolumeof lakewater. Thevolumesofmicrocosmsusedto test S. exfoliatusvariedfrom verysmall (350 ml) to 5 l, with all experiments plus con-trols carriedout in triplicate. Addition of antagonistculture to such microcosms containing dominant cy-

167

Figure 5. In situ (microcosm) lake experiment. Streptomycesex-foliatus strain RG12was addedto laboratory-culturedMicrocystisaeruginosa within dialysis tubing(350ml). Algal biomass (determ-ined as chlorophyll-a) showed a 60% decrease in the presenceofantagonist.

anobacteriatypically resultedinasignificantreductionin algalpopulation,normally decreasing to about50%of theinitial level over a 1–2-weekperiod(Figure5).

Protozoa

Studies are currently in progress to assess the po-tential biocontrol activity of culture collection strainsof two species of protozoa: the amoebaNucleariadelicatula (CCAP 1552/1,Figure6a) and the ciliateNassula tumida (CCAP 1650/2, Figure6b), both ofwhich are known to feed on cyanobacteria.So far,the investigation hasmainly involvedlaboratory stud-ies to determinepredation rangeandthedynamicsofprotozoan/algal interactions, thoughsomesmall-scalepreliminaryfieldwork hasbeencarriedout.

The potential use of N. delicatula andN. tumidaas biocontrol agents of differentnuisance algae wasassessedin liquid culture. Cultures of dif ferent algaewere inoculated with protozoaandmaintainedunderstatic conditionsat 20◦C, with a diurnal regimeof 15h light/8 h dark to simulate the natural environment.Changesin the population of algae were monitoredthroughopticaldensity with acolorimeterandthepro-tozoanpopulation determined by direct counts on aSedgewick Rafter countingcell.

The results have shown that N. delicatula grazesa rangeof culturedand environmental isolates of Os-cillatoria sp. and Anabaenaspp. (but not Microcystisaeruginosa), whereasN. tumida hasonly proved act-ive against Oscillatoria spp. The feeding preferencesof the amoeba and the ciliate appear to be filament-ous cyanobacteria, shown in Figure 6a, b, where

Figure 6. Protozoanantagonists of cyanobacteria.(a) Light mi-crographof Nuclearia delicatula grazingon Anabaenaflos-aquae.(×300); (b) Light micrographof Nassula tumidagrazing on Oscil-latoria sp. (×150).

both species of protozoaare attachedto the end ofa cyanobacterial filament which is being ingested.Lack of ingestion of other cyanobacteriamay relateto factorssuchas thenon-filamentoushabit, cell size,insufficient nutrient source and production of algaltoxins.

The grazing efficiencies of the protozoawere in-vestigated in laboratory cultures by varying the ab-solute andrelative populationsof algaeand protozoaunderdif ferentenvironmentalconditions. Results ob-tained from one of theseexperiments, involving N.delicatula and an environmental isolate of Anabaenasp., areshown in Figure7. The experimentwas car-riedoutin agrowthcabinetundersimilar conditionstothosenotedearlier, with threereplicateflasks for eachcombinationof algalandprotozoanpopulations. Dailymeasurements of optical density, chlorophyll ‘a’ andprotozoancounts weremadeover the 10-dayexperi-mentalperiod.Initial populationlevelsof N. delicatula(100, 500 and 1000 cells ml−1) were tested againstthree starter levels of Anabaenasp. (40, 204 and

168

Figure 7. Predator–prey experiments, using laboratory cultures. The ability of different inoculumlevels of Nuclearia delicatula to control thegrowth of Anabaenasp. in laboratorycultureis shown for threestarterlevels of algalpopulation,with respective chlorophyll ‘a’ concentrationsof 40µg ml−1 (a), 204µg ml−1 (b) and 572µg ml−1 (c). In eachcase, protozoawere addedto give overall initial populationlevels of 100,500and1000cellsml−1. Algal populationlevelsare given as themeanchlorophyll-‘a’ concentration (± SD) of threereplicateflasks.

169

572µg chlorophyll ‘a’ l −1, Figure7a,c). Thehighestchlorophyll ‘a’ levelusedwasfar in excessof thatnor-mally occurring in an algal bloom. The results showthat the ability of N. delicatula to reducecyanobac-terial population levels under laboratoryconditionsdependson the initial populationsof bothorganisms.As expected,N. delicatula was typically most effect-ive at high protozoanandlow algal population levels,with rapid reduction of all three algal biomasslevelsat 10000 protozoaml−1. Experiments such as thisgive someideaof the protozoaninoculum requiredtocontrol differentcyanobacterial populationsin a lakeenvironment.

As with the actinomycetestudies,microcosmtest-ing allows differentspecies or strains of antagonistand algae to be brought togetherat defined initialpopulation levels and undercontrolled nutrient con-ditions, but interacting underambient environmentalconditionsof temperatureand light regime.Prelimin-aryexperimentshavebeencarriedatalocalfreshwatersite using submerged dialysis tubes with a capacityof 350 ml. These were suspendedfrom a wire meshfloatingplatform,andwereseparatelyinoculatedwiththecyanobacteriaOscillatoria sp. (chlorophyll ‘a’ 645µg l−1) or Anabaenaflos-aquae(chlorophyll ‘a’ 678µg l−1) pluseithersterile protozoan medium (control)or protozoanculture (treatment). The results showeda clear reduction in the populations of Oscillatoriasp. and Anabaenaflos-aquaefollowing addition of N.delicatula. As with the laboratory-based experiments,the initial algal levels were substantially higher thanthose normally encounteredunderbloom conditions.In eachexperiment,theprotozoanpopulationshowedamarked increase,whichwasmost pronouncedduringthe initial phaseof rapid algal decline.

Roleof biological contr ol in the lake environment

Biological control of cyanobacteria represents a po-tential short-term measure to reducethe population,or preventthebuild-up,of nuisancealgal populations.This is in contrast to long term control strategiessuchasnutrient limitation (bottom-up control) and bioma-nipulation (top-down control), which involve morefundamental alterations to the lake ecosystem. Thebiocontrol agents consideredin this paperhave eitherbeen recently isolated from environmentalsamples(actinomycetes), or were originally derived from thefreshwater environment(protozoa).These native or-ganisms have not undergoneany deliberate genetic

manipulation or gene enhancement. The addition ofsuch agentsto the lake canthusbe viewed asan al-teration to the existing balance of organisms alreadypresent within theecosystem. Thisalteration is invari-ably temporary, with reversion of the ecosystemto itsoriginal statewithin a relatively short period of time.

Biologicalcontrol of cyanobacteria, likebiologicalcontrol of othernuisanceorganisms(plant pathogenicbacteria andfungi, insectpests, weeds) hasa numberof advantagesover the other major type of imme-diate control, chemical control. Biological controlcan be highly specific to the target organism, withno destruction of other organisms and with no directchemicalpollutionthatmight affecthumans. Potentialdisadvantagesinclude limiteddestruction of the targetorganism, limited survival of theagentor its removalby other organisms and problems with large-scaleproduction, storageand application of the biocontrolagent.

Laboratory studiesarean important aspectof thegeneralschemefor developmentof biocontrol agentsoutlined in Figure 1, providing useful information onaspects such as mode of action, effect of physico-chemical factors, progress of control activity andrequired antagonist inoculum levels. In the case ofStreptomycesexfoliatus, for example, laboratory stud-ies showed that activity involved a lytic diffusibleagent, production of which was genetically stable—persisting over repeatedsub-culturing. This is in con-trastto Streptomycesisolatesobtainedby Redhead&Wright (1978)which lost cyanobacteriolytic activityafter four transfers. Althoughformation of the lyticagentin S. exfoliatusoccursindependantly of thepres-ence of cyanobacteria, the ability to destroy theseorganisms is probably an advantage to the antagonistsinceactinomycetegrowth ispromoted.

Althoughlaboratorystudieshaveanimportantpartto play in biocontrol work, results obtainedshould beviewedwith caution if they are to be interpretedin thelake context. Laboratory data cannotsimply beextra-polatedto the freshwater environment. This is shownwith both of the case studiespresentedin this paper,wheretheanti-cyanobacterialeffectivenessof theant-agonists in themicrocosm environmentwasmuchlessthanin laboratory cultures.In thecaseof Streptomycesexfoliatus, addition of 0.05 ml Streptomycesculture(∼106 CFU ml−1) to 100ml of laboratory Anabaenacylindricaculture(50µg l−1chlorophyll-‘a’) achievedcompleteinhibition within 12 days.

Similar (2000-fold) dilutions of antagonist werenot effective in the lake, where inoculum levels of

170

250 ml in 5 l lakewater were required for signific-ant reduction in the cyanobacterial population. Dif-ferencesbetweenthe effectiveness of cyanobacterialantagonists in laboratory culturesand the lake envir-onmentmayrelateto anumberof generalfactors. Theimportant onesare:(1) the effects of the lake environment (physico-

chemical and biotic) on antagonist populations,growth rateandantagonistic activity;

(2) effects of the lake environmenton cyanobacterialresponseto antagonists;

(3) Differences in the distribution of cyanobacteriaandantagonistswithin the lake ecosystem.Separation in thepopulationsof these two groups

of organisms(e.g.lossof antagonistsfrom surfacewa-tersdueto sinking,replacementof surfacealgaefromthe hypolimnion) will limit the biocontrol activity ofantagonists.

The most important test for any potential biocon-trol agentmust beits effectivenessin theenvironment,where the above factors are encountered under theprevailing lake conditions. In spite of these limit-ations, both of the antagonists (actinomycetes andprotozoa)detailed in this paperdid have a significantbiocontrol effect in lake microcosms at the appropri-ate inoculum level. Other workers have also carriedoutdirectenvironmental testing,particularly with viraland bacterial antagonists. Using 5000-gallon tanksJackson & Sladecek(1970) examinedinteractionofcyanophageLPP-1 with Plectonema boryanum. Theyconcludedthat cyanophageshad the potential to pre-vent growth of their hosts undernatural conditions.Daft et al. (1975) tested 15 fresh algal bloomsfromvariouswaters against Myxobacter CP-1, in eachcaseobtaining clearlysis. Two environmental experimentssubsequently confirmedtheeffectivenessof thisantag-onist. In the first, 3 l of Microcystis aeruginosa weretaken from thetop 0.5 m of thereservoir andaddedtoeachof four polythenetrayswhich werefloatedsemi-submergedon thewater. A log phase culture of CP-1wasadded to two of thetraysto giveaninitial concen-tration of 106 ml−1. Within 96h chlorophyll-‘a’ i n thetreatments haddeclinedto negligible amounts. In thesecondexperiment, 200 l of reservoir was portionedoff using polythenesheeting. One area was sprayedwith CP1to 106 ml−1. Within 60h therewasanalmostcomplete lossof chlorophyll from thetreatedarea.

Various field studies have also beencarried outon the use of barley straw to control cyanobacteria,the activity of which may involve microbial action.Althoughthe use of straw has beenproposed as an

inexpensive and environmentallyacceptableform ofcontrol (Ridge & Barrett, 1992; Ridge et al., 1999)practical results have beenerratic, with field studiesproving successful in some cases (Everall & Lees,1996)but notothers(Kelly & Smith, 1996).

Practical applications

In caseswherea particular biocontrol agentis shownto beconsistently effective in environment testing, thenext phase would be full-scale field trials (includingcomplete environmental analysis), commercial costassessmentandevaluation of how thebiocontrol agentcould be used in lake management.In assessing thepractical use of such antagonists for control of cy-anobacteriain thelake,aspectssuchas frequency andtiming of application, mode of application (formu-lation and method of dispersal) and implications forwaterprocessing areall important.

Thevolumeof inoculum requiredper unit areaoflake surfaceis also relevant. In the case of Strepto-mycesexfoliatus, minimum inoculum levels obtainedin laboratory and microcosm experiments (see above)would translateto requiredlevelsof 5×103 and5×105

l of antagonist inoculumperhectare(treatmentof topmetre of water column, total volume approximately10000m3 ha−1). These amounts of biocontrol agentrequireddo not soundencouraging,either in termsofcost or ease of production and application,andwouldsuggest that simple additionof antagonist culture tocontrol establishedgrowthsof cyanobacteriain alargebodyof lakewater is not feasible. Thepotential use ofsuchbiocontrol agents in lakemanagementshould notbedismissed, however, sincesmaller volumesmaybepossiblein somecircumstances. Thefollowingmaybeappropriate:1. Control of developing (pre-bloom)cyanobacterial

populations. Application of such agents may bemost effective during build-up of blooms, ratherthanunderfull bloom conditions. This would re-quire very careful monitoring of the increase ofcyanobacterialpopulationswithin the epilimnion,including the movement of these organisms tosurfacewaters from thehypolimnion.

2. Application to small water bodies, including mu-nicipal ponds and small amenity lakes, wherethe overall amountof inoculum required wouldobviously beless thanlarger lakesand reservoirs.

3. Improved types of biocontrol agent. The effect-iveness of particular biocontrol agents may be

171

improved in various ways, including the useof genetically-engineeredorganisms (e.g. withhigher growth rates, improved synthesis of anti-cyanobacterial compounds).

4. Improved methods of application. One problemnotedin themicrocosmexperimentswith S. exfoli-atuswasthatantagonist flocculestendedto sink inthe water column. Maintenanceof the antagonistin suspension,by attachmentto afloating substratesuch a straw, would ensure that theagentstays inthesurfacealgalzone.

5. Mixture of agents. Combining different typesofbiocontrol agent might offer greaterflexibility inacontinuously changinglakeenvironment. Daft etal. (1985) have pointed out that a such mixture,offering differentmodesof attack, may offer thebest chanceof preventingproblemcyanobacterialbloomsin eutrophic lakes.

6. Use with other techniques. Experiencewith otherbiocontrol situations hasshown that such agentsarebest used inconjunctionwith otherapproaches,andnot in isolation. Theuse of biological controlagents in lake and reservoir managementshouldbe regardedas just oneof a numberof availablecontrol measuresthat form part of an integratedmanagementpolicy.

References

Al-Tai, A. M. S., 1982.Characteristics and microbial interactionsof certain actinomycetesfrom Iraq. Ph.D. Thesis, University ofDundee.

Barnet,Y. M., M. J. Daft, & W. D. Stewart, 1981.Cyanobacteria–cyanophageinteractionsin continuousculture.J. appl. Bact.51:541–552.

Bershova, O. I., Z. P. Kopteva & E. V. Tantsyurenko, 1968. Theinterrelationsbetweentheblue-greenalgae- thecausative agentsof ‘water bloom’ – and bacteria. In A.V Topachevsky (ed.),Tsvetnie Vody: 159–171.

Brabrand,A., B. A. Faafeng,T. Kallquist, & J. P. Nilssen, 1983.Biological control of undesirable cyanobacteriain culturallyeutrophiclakes. Oecologia60:1–5.

Burnham, J. C., S. A. Collart & M. J. Daft, 1984.Myxococcalpred-ation of the cyanobacterium Phormidium luridum in aqueousenvironments. Arch.Microbiol. 137:220–225.

Burnham, J. C., S. A. Collart, & B. A. Highison, 1981. Entrap-ment and lysis of the cyanobacteriumPhormidium luridum byaqueouscolonies of Myxococcusxanthus PCO2. Archives ofMicrobiology 129:285–294.

Canter, H. M. & J. W. Lund, 1951.Fungalparasites of the phyto-plankton.II. (StudiesonBritish ChytridsXII). Ann.Bot. XV 58:129–159.

Canter, H. M., S.I. Heaney & J. W. Lund,1990.Theecologicalsig-nificanceof grazingon planktonicpopulationsof cyanobacteriaby theciliate Nassula. New Phytol.114:247–263.

Cole, G. T. & M. J. Wynne, 1974. Endocytosis of Microcystisaeruginsoa by Ochromonasdanica.Journal of Phycology 10:397–410.

Cook, W. L., D. G. Ahearn,Reinhardt,D. J. & R. J.Reiber, 1974.Blooms of an algophorousamoebaassociatedwith Anabaenaina freshwater lake.Wat. Air Soil Pollut. 3: 71–80.

Daft, M. J., J. Begg, & W. D. Stewart, 1970. A virus of blue-greenalgaefrom fresh-waterhabitatsin Scotland.New Phytol.69: 1029–1038.

Daft, M. J., J. C. Burnham, & Y. Yamamoto, 1985.Algal blooms:consequencesandpotentialcures. J. appl. Bact.175S–186S.

Daft, M. J., S.B. McCord& W. D. Stewart, 1975.Ecologicalstudieson algal lysing bacteriain fresh waters. Freshwat. Biol. 5: 577–596.

Daft, M. J. & W. D. Stewart, 1971.Bacterialpathogensof freshwa-ter blue-greenalgae.New Phytol. 70: 819–829.

Dryden,R. C. & J. C. Wright, 1987.Predationof cyanobacteriabyprotozoa.Can.J. Microbiol. 33:471–482.

Ensign, J. C. & R. S. Wolfe, 1965.Lysis of bacterialcell walls byanenzyme isolatedfrom aMyxobacter. J. Bact.90: 395–402.

Everall,N. C.& D. R. Lees, 1996.Theuseof barley straw to controlgeneraland blue-greenalgal growth in a Derbyshire reservoir.Wat. Res. 30:269–276.

Fallon, R.D. & T. D. Brock,1979.Lytic organismsandphotooxidat-ive effects: Influenceon blue-greenalgae(cyanobacteria) in lakeMendota,Wisconsin. Appl. environ. Microbiol. 38:499–505.

Fraleigh,P. C.& J. C. Burnham, 1988.Myxococcalpredationon cy-anobacterialpopulations: nutrienteffects. Limnol. & Oceanogr.33: 476–483.

Jackson, D. F. & V. Sladeck,1970.Algal viruses– eutrophicationandcontrol potential.Yale Sci. Mag.44:16–22.

Kelly, L. A. & S. Smith, 1996.The nutrientbudgetof a small eu-trophic loch and the effectiveness of straw balesin controllingalgalblooms. Freshwat.Biol. 36: 411–418.

Laybourn-Parry, J., K. Jones, & J. P. Holdich, 1987. GrazingbyMayorella sp. (protozoa: Sarcodina)on cyanobacteria.Funct.Ecol. 1: 99–104.

Martin, E. L., 1976.Biological controlof blue-greenalgae.NationalTechnicalInformationServiceReportPB-288311.Nebraska.

Mason, C. F., 1996. Biology of Freshwater Pollution, 3rd edn.Longman,Essex.

Newman, J. R. & P. R. F., Barrett, 1993. Control of Microcystisaeruginosa by decomposing Barley straw. J. of aquatic PlantManage.31: 203–206

Ohki, K. & Y. Fujita, 1996.Occurrenceof a temperatecyanophagelysogenising the marine cyanophytePhormidium persicum. J.Phycol.32:365–370.

Pace,M. L. & J. D. Orcutt, 1981.The relative importanceof pro-tozoans, rotifers and crustaceansin a freshwater zooplanktoncommunity. Limnol. Oceanogr. 26: 822–830.

Padan,E. & M. Shilo, 1973.Cyanophages-viruses attacking blue-greenalgae.Bacteriol.Rev. 37: 343–370.

Padan,E., M. Shilo, & N. Kislev, 1967.Isolationof ‘Cyanophages’from freshwater pondsand their interaction with Plectonemaboryanum. Virology 32: 234–246.

Redhead,K. & S.J. Wright, 1978.Isolationand propertiesof fungithat lyse blue-greenalgae.Appl. Environ. Microbiol. 35: 962–969.

Redhead,K. & S. J. Wright, 1980. Lysis of the cyanobacteriumAnabaenaflos-aquaeby antibioticproducingfungi.J. Microbiol.119:95–101.

Reim, R. L., M. S. Shane,& R. E. Cannon,1974.The character-isation of a Bacillus capableof blue-greenbactericidal activity.Can.J. Microbiol. 20: 981–986.

172

Ridge, I. & P. R. Barrett, 1992. Algal control with barley straw.Aspectsappl. Biol. 29:457–462.

Ridge, I., J. Walters & M. Street, 1999. Algal growth by ter-restrial leaf litter: a realistic tool? Hydrobiologia395/396(Dev.Hydrobiol. 136):173–180.

Rubenchik,L. I., O. I. Bershova, & Z. P. Knizhnik, 1965.On theinterrelationship of Anabaenawith bacteria andactinomycetes.In: Ecologia i physiologia sinezelenych vodorosleiy. NaukaMoscow: 223–226.

Safferman, R. S.& M. Morris, 1962.Evaluationof naturalproductsfor algicidal properties. Appl. Microbiol. 10:289–292.

Safferman, R. S. & Morris, M. 1963. The antagonistic effectsof actinomyceteson algaefound in waste stabilisation ponds.Bacteriol.Proc.14: A56.

Sallal, A. K., 1994. Lysis of cyanobacteriawith Flexibacter sppisolatedfrom domestic sewage.Microbios77: 57–67.

Salton, M. R. J., 1955. Isolation of Streptomyces spp. capableof decomposing preparationsof cell walls from variousmicro-organismsandacomparison of their lytic activities with thoseofcertainactinomycetesandmyxobacteria.J. Gen.Microbiol. 12:25–30.

Shilo, M., 1970. Lysis of blue-greenalgae by Myxobacter. J.Bacteriol.104:453–461.

Sigee,D. C., 1993.BacterialPlant Pathology: Cell andMolecularAspects. CambridgeUniversity Press, Cambridge.

Silvey, J. K., 1973.Determination of nuisanceorganisms. Proceed-ings AWWA Water Quality TechnologyConference,December1973.

Silvey, J. K. & A. W. Roach,1975.The taste andodourproducingaquaticactinomycetes. CRC Crit. Rev. environ. Control March1975:233–273.

Stewart, W. D. & M. J. Daft, 1977. Microbial pathogensofcyanophyceanblooms. Adv. aquaticMicrobiology 1: 177–219.

Whyte, L. G., A. Maule, & D. R. Cullimore, 1985. Method forisolating cyanobacteria-lysing Streptomycetesfrom soil. J. appl.Bact.58: 195–197.

Wright, S. J. & R J. Thompson,1985.Bacillus volatilesantagonisecyanobacteria.FEMSMicrobiol. Lett.30: 263–267.

Wright, S. J., K. Redhead& H. Maudsley, 1981. Acanthamoebacastellanii, a predatorof cyanobacteria.J. Gen.Microbiol. 125:293–300

Wright, S. J., C. J. Linton, R. A. Edwards, & E. Drury,1991. Isoamyl alcohol (3-methyl-1-butanol), a volatile anti-cyanobacterialand phytotoxic product of some Bacillus spp.Lett. appl.Microbiol. 13:130–132.

Yamamoto, Y., 1981.Observationson the occurrenceof microbialagentswhich cause lysis of blue-greenalgaein Lake Kasumig-aura.Jpn.J. Limnol. 42:20–27.

Yamamoto, Y. & K. Suzuki, 1990. Distribution and algal-lysingactivity of fruiting myxobacteriain lake Suwa. J. Phycol. 26:457–462.