Nectaryeastsoftwosouthernspanishplants:The Rolesof Immigrationand

NectaryeastsoftwosouthernSpanishplants:the rolesof immigrationand physiologicaltraitsincommunityassembly

Marı´aI.Pozo1,2,Marc-Andre´ Lachance2CarlosM.Herrera1

1Estacio´nBiolo´gicadeDon˜ana,ConsejoSuperiordeInvestigacionesCientı´ficas(CSIC),Sevilla,Spain;and2DepartmentofBiology,Universityof

WesternOntario,London,ON,Canada

Correspondence:Marı´aI.Pozo,Estacio´n Biolo´gicadeDon˜ana,ConsejoSuperiorde InvestigacionesCientı´ficas(CSIC),Avenida Ame´ricoVespucios/n,E-41092Sevilla,Spain. Tel.:(0034)954466700; fax:(0034)

954621125;

Keywords

nectar;communityassembly;Metschnikowia;

yeast.

Abstract

Recentstudieshaveshownthatdenseyeastpopulationsoftenoccurringinflo- ralnectararenumericallydominated byafewspeciesfromtheflower–insect interfacespecializedgenusMetschnikowia, whilegeneralistyeastspeciescom- monlyoccurringonleafsurfaces,soil,freshwater,andairwererarelyisolated fromnectarsamples.Thisstudywasdesignedtounderstand themainfactors responsiblefortheassemblyofnectaryeastcommunities,bycombiningfield experimentswithlaboratorytestscharacterizingthephysiologicalabilitiesofall yeastspeciesformingthepoolofpotentialcolonizersfortwoSpanishflowering plants(DigitalisobscuraandAtropabaetica).Yeastfrequencyandspecies rich- nesswereassessedinexternalsources(beeglossae,air,plant phylloplane)as wellasin pollinator rewards(pollen, nectar). Yeastsweremost frequent in externalsources(air, flower-visitinginsects),lesssoin the proximate floral environment(phylloplane),andleastinpollenandnectar.Nectarcommunities appeared to beconsiderablyimpoverishedversionsofthoseininsectglossae andphylloplane.Nectar,pollen,andinsectyeastassemblagesdifferedinphysi- ologicalcharacteristicsfromthoseinothersubstrates.NectarivorousMetschnik- owiawere not more resistant than other yeastspeciesto plant secondary compoundsandhighsugarconcentrationstypical ofnectar,buttheirhigher growthratesmaybedecisivefortheirdominanceinephemeralnectarcommu- nities.

Introduction

Microbial communitydynamics inplantsandonplant surfaceshavebeenstudiedforoveracentury(e.g.Rui- nen,1963; LastPrice,1969; PhaffStarmer,1987; Thompson etal.,1993;LindowBrandl,2003;Sampaio etal., 2007),butmanyecologicalprocessesthataffectthe microbialdiversitystillremain to beelucidated(Kinkel,

1997).Individualplantsprovideamultitudeofmicrohab- itatswithdifferenttopographicalfeatures,nutrients,water availability,andarangeofmicroclimaticconditions that select fordiversemicrobialcommunities(Andrews& Harris, 2000). Some yeast microhabitats within plants havebeendescribedtobespecific.Theseincludedecaying tissues,flowers,nectar,orfruits(PhaffStarmer,1987; Lachanceetal.,1989;Starmer etal.,1990;Rosaetal.,

1992,1994;Spenceretal.,1992).

Insomestudiescarriedoutrecently insouth-eastern Spain,yeastcommunitiesofnectarwereshowntopossess highpopulationdensitiesandaclear predominanceof members of the genus Metschnikowia (Herrera etal.,

2009,2010;Pozoetal.,2011).Thefewadditional yeasts foundinnectarconsistedofgeneralistspecies thatare knowntooccurinothermicroenvironmentssuchasleaf surfaces,soil,freshwater,and air (Herzbergetal.,2002; Brysch-Herzberg,2004).

Microbialpopulation dynamicsin floralnectar,asin anyotherhabitat,maybeconsideredacomplexfunction offour processes:immigration, emigration, growth,and death(FonsecaIna´cio,2006).Itisthereforereasonable toaskwhetherfloralnectarisdepletedatthesamerate astheoverall incomingyeastcommunityorwhetherit representsaset of identical individualmicrohabitatsthat constraincolonizationtoasmallgroupofhighly specialized

yeast species. In comparison with other plant parts, spring or summer flowershave a very short lifespan. Insidethe flowers,nectar isahighlyfluctuatinghabitat that experiencesevaporation at hightemperatures, dilu- tionbyrain,anddepletionfrompollinatorvisits.Nectar colonizationbyyeastsisconstrainedbytheircloseassoci- ation withthe flower-visitinginsects,whichboth intro- ducethe yeaststhrough their mouthparts and consume themalongwiththenectar(Gilbert,1980; Sandhu& Waraich, 1985; Brysch-Herzberg, 2004; Canto etal.,

2008).Thecomplexinteraction associatedwithcoloniza- tion history iscompounded further with yeastsurvival and growthprocesses.On theonehand, thehighsugar content common to floralnectars (Nicolson Thorn- burg,2007)mayserveasasourceofnutrients that can support microbial development and at the same time mayrenderthehabitatselectiveinfavourofhighlyosmo- tolerant yeasts.Furthermore, nectarivorousdietsarealso characterizedbyalownitrogencontent(Nicolson,2007). Ontheotherhand,nectarsofmanyplantspeciescontain plantsecondarycompounds (Adler,2000;Mansonetal.,

2007;Nicolson,2007;Gonza´lez-TeuberHeil,2009)that

arethought toactasadeterrentfornectarrobbers(e.g. inefficientpollinator insects,nectar-consuming microbial communities), thusfavouring‘legitimate’visitors.Inthis context,yeastsdepletenectarsugars,devaluatethefloral reward(Herrera etal.,2008),andconsequentlymightbe affected bythepresenceofthesesubstances.Inplant speciesthatproducedefensivecompounds, asisthecase withthoseusedinthisstudy,thesecompoundsmaylimit thenumber ofyeastspeciesthat finallyconstitute nectar communities. The failureof microorganisms to survive infloralnectarmaythenbeduetonectarphysicalcondi- tionsthemselves, asithasbeennoted above,orto competitionprocesses.Giventhatfloralnectarisbothan overpopulated (Herrera etal., 2009) and a resource- limitedhabitat,thecompetitiveadvantage ofMetschnikowia speciesmayalsobetheresultofasuperiorrateofrepro- duction andsurvival,whichinturn leadstoareduction incompetitors.

Themainobjectiveofthisstudywastobetterunder- stand the origin and composition of the yeastspecies poolpotentiallyarrivinginto floralnectarintwodiffer- ent plant speciesat the plant population scale.To this end, several microhabitats are evaluated as potential sources of yeastto floralnectar, and yeastfrequencies aremeasured at differentlevelsfollowinga hierarchical sampling scheme in each individual plant. The levels range from external sourcessuch asflower-visitingbee glossae,air,bracts,andcorollasurfacestofloralrewards suchasnectarandpollen.Measuresofyeast species transferenceto floralnectar areindicativeofthe degree ofmicrosite specialization,whichraisesthe question of

themechanismsunderlyingnectaryeast community assemblage.Accordingly, thisstudyexaminedwhether interspecificdifferencesin carbon and nitrogen sources utilizationpatterns,resistance toinhibitorsandplant defensive compounds, osmotolerance,andgrowthrate contribute alsotoexplainingfeaturesofthenectaryeast community.

Materialsand methods

Fieldsampling

Atotalof234yeastisolateswereobtainedfromthetwo study species,willow-leavedfoxglove(Digitalisobscura, Plantaginaceae;Digitalishereafter) and Atropa baetica (Solanaceae;Atropahereafter),in2009,inthe Sierrade Cazorla region, a well-preservednatural area in Jae´n Province,south-easternSpain.Eachplantspeciescontrib- utedonepopulation, andthetwoweresampledsequen- tiallyaccordingto their blooming period, whichranges fromearlyJunetolateJuly.Sampleswerecollectedfrom eachlocalitywithin4consecutivedays.TheDigitalissam- plingsitecontributed 162yeastisolates,andtheother72 wererecoveredfrom Atropa. The two localitieswere located7.9kmapart.

Yeast isolates were obtained in a spatially nested scheme thatrangedfromexternalsources suchasbees’ glossaeandairsamplestoplant-relatedsamplesobtained from20randomlyselectedplantindividualsineachsam- plingsite.Twoflowersweresampledfromeachindivid- ual,fromeachofwhichonesubsamplewas systematically collected frombracts,corollaouterandinnersurfaces, pollen,andnectarsamples.Toavoidartefactscausedby different yeast colonization times, mature flowersthat were 3days old were systematicallycollected. Digitalis plants produce protandrous flowersthat lastfor up to

5days.Incontrast,flowersofAtropaaremarkedlypro- togynousandlastforupto3–4days.Nectarproduction at both speciestend to be higher at the intermediate phaseoffloral development,whichcoincidedwiththe femalephasein Digitalisand the malephasein Atropa flowers(M.I.Pozo,pers.obs.).

Three different treatments were performed in this study.Thefirstconsistedof10individualflowersnatu- rallyexposedto insectvisitation in eachsamplingsite, and the secondconsistedofflowersof10other plants excluded from pollinator visitation by bagging at the bud stage(N=20flowersin eachtreatment; ‘exposed’ and ‘insect-visitors-excluded’,hereafter).Aftercompleting Digitalisexperiments, it wasdetected that nectar from insect-visitors-excludedplantsstillharbouredyeast ina significantfraction,and sotwoadditional buds in each of the ten previouslybaggedplants werecoveredwith

cellophane envelopes (‘airflow-prevented’ samples,

N=20flowers).

Eighteenand15beeswerehand-netted whiletheyfor- agedonDigitalisandAtropasamplingsites,respectively. ThreeBombusspecieswerecollectedfromDigitalisplants (Bombuspascuorum,Bombus pratorum, andBombuster- restris), comprisingboth workersand males;B.terrestris (males), B.pascuorum (workers), and the solitary bees AnthidiumflorentinumandAnthophora(Amegilla)quadri- fasciatawererecordedinAtropaflowers.Basedonpolli- nationcensuses carriedoutsimultaneouslyforboth samplingsites(seeHerreraetal.,2001forcensusesmeth- odology),bothbeepoolswerearepresentativesampleof thewholesetofvisitors(Supporting Information, Table S1).Immediatelyuponcapture,beeswereplacedindivid- uallyin sterilecontainers and anaesthetized by placing themfor2mininsideafreezerat—20°C.Theglossaof eachindividualbeewascarefullyextended,usingsterile forceps,beyondthetipofthemaxillarygaleaeandcare- fullyrubbedagainstthesurfaceofanagarplate.Airborne yeastsampleswereobtainedbyplacinganopenagarplate undereachselectedplantfor5min.

Entireflowersandtheirmostproximatebractwerecol- lectedinthefieldandkeptrefrigeratedinasterilecon- tainer until they were brought to the laboratory. The bractsand corollasweregentlyswabbedforinoculation onagarmedium.Innerandoutersurfacesofthecorollas werealsosampledbyimprinting them onto agarmed- ium.Pollenwassampledbyimmersingamature anther in1mLofsterilewaterfor48hand streak-inoculating

1lLofthisinoculumontoanagarplate.Nectarsamples weretakenwith1-lLcalibratedmicrocapillaries and transferredontoagarplatesforstreakinoculation.

Yeast isolation,identification,and physiologicalcharacterization

Allsampleswere cultured onto yeastmalt agar plates (1.0% glucose,0.5%peptone, 0.3%malt extract, 0.3% yeastextract,and2.0%agar)with0.01%chloramphenicol andincubatedat25°C.

Isolateswerepurified and characterizedfollowingthe

standard methods of Yarrow(1998). Growth responses wereevaluatedbyreplicaplatingonto115testmedia devel- opedbyLachance(1987)orspecificallyconceivedforthis study.Thisincludedgrowthon44carbonsources, 20 inhibitors, fivenitrogen sources,fourfermentation tests, three testsofhydrolysis,three halotolerance tests,three testsofmorphologicalcharacteristics, twocarbonand nitrogensourcestests,twogrowthfactors,andindividual testsforosmotolerance,acidproduction, colourreaction, andstarchproduction, and25growthtemperatures.The D1/D2domainsofthelargesubunitribosomalRNAgene

weretwo-way-sequencedforalltheisolatesas describedin thestudyofLachanceetal.(1999),usingtheprimerpair NL1/NL4.Gblocks(Castresana,2000)wasusedtotrimthe resultingalignment,and theDNAsequenceswerecom- paredwiththoseintheGenBankdatabase(lastaccessed9

June2010)byquerying withtheBLASTtool.Usingthesame procedure asPozo etal.(2011), operational taxonomic units(OTUs)werealsoevaluated,definedonthebasisof DNAsequencesimilaritywiththeprogramDOTURusingthe

3% similaritythreshold(distance-basedOTUandrichness, SchlossHandelsman,2005).

Plantsecondarycompoundsassays

Agardiffusionwasusedtoevaluatethepossibleeffectof plantdefensive compounds presentinnectaronyeast growth.Threeofthemajorcompoundsineachplantwere tested,namelythetropanealkaloidsatropine,tropine,and scopolaminefromAtropa(Za´rateetal.,1997;Adler,2000), andthedigitoxose-type cardenolidesdigitoxin,digitoxi- genin,and gitoxigeninfrom Digitalis(GavidiaPe´rez- Bermu´dez,1997;C.M.Herrera, unpublished data). The potentialeffectofthesecompoundswasassessedon12yeast strains,eachfromadifferentspecies:Candida bombi,Can- didafloricola, Cryptococcuslaurentii,Cryptococcusvictoriae, Debaryomyces maramus, Lachanceathermotolerans,Metsch- nikowiareukaufii,Metschnikowia gruessii,Metschnikowia kunwiensis,Rhodotorulacolostri,Starmerellabombicola, andSporobolomyces roseus. PlatesofYeastNitrogenBase (Difco)plus1%ofglucoseand 1.5%agarwereinocu- latedwith dilute yeastsuspensionsofeachyeaststrain, and 5lL ofdimethylsulfoxide(DMSO)solutions con- taining1mgofeachcompound wasaddedasindividual drops. LackoftoxicityofDMSOwascorroborated for yeaststrainsbyexposingyeaststo 5lL ofthis solvent. Theplateswereincubatedat24°Candexaminedperiod- icallyfortheevidenceofinhibitionzonesinducedbysec- ondarycompounds.

Osmotoleranceand optimalgrowth temperatureassays

Thegrowthresponsestoincreasingproportionsofglucose inagarmediawastested for144strainsbelongingtothe29 yeastspeciesrecoveredfrom the differentmicrohabitats studied.Themediacontained1%yeastextract,1.5% agar, and40,45,50,or55%(w/w)glucose,andtheplateswere incubatedat24°C.Growthofthesamestrainswasevalu- atedonYeastMalt(YM)agaratfour,six,eight,10,and

12°Candfrom26to42°Cwithincrementsofonedegree.

Yeastgrowthwasrecordedonanordinalscaleafterthree, seven,and18days.Fortaxarepresentedbymorethanone specimen,themostcommonresponseisreported.

Dataanalyses

Sample-basedoccurrencedata wereusedto obtain observedspeciesrichness(MaoTaufunction,Maoetal.,

2005)andtocalculatespeciesrichnessestimators(ICEand Chao2) withESTIMATESversion8.2(Colwell,2005).This programwas usedalsotoconductanalysesofsharedspe- ciesbetweenmicrohabitatsusingsample-basedabundance data.TheChao-sharedestimatorwasusedtocomparespe- ciesco-occurrencebetweenmicrohabitats.Fortheseanaly- ses,beeglossaeandairsampleswereconsideredtogetheras

‘externalsources’samples,andbractsandthetwocorolla surfacesas‘phylloplane’samples.Pollenandnectarisolates werecombined as‘pollinator rewards’samples.Forthe purposeofanalysingthephysiologicalfeatures oftheyeast isolates,an 82-testsubsetfrom the replica-platingseries wasselected bychoosingthosetestsforwhichmorethan

5%ofthestrainsdiffered ingrowthresponse.Thissubset comprised39carbonsources,16temperatures,12inhibi- tors,five nitrogensources,threehalotoleranceandone osmotolerancetests,threetestsoflyticactivity,onetestof carbonandnitrogensource,andindividualtestsforvita- minindependenceandglucosefermentation.Thegrowth

responsesforthesetestsfor50yeaststrainswereanalysed

(above70%),lesssoinsamplesbelongingtotheproxi- mateenvironmentoftheflower,andthelowestinnectar andpollensamples, eventhoughthetwoplantspecies studied differed in yeast frequency at each microsite (Fig.1).

In insect-visitors–excludedplants,phylloplanesamples continuedcarryingroughlythesameproportion ofyeasts, butpollensamplesharboured yeastsata50%lowerfre- quency(Fig.2).Thetwoplant speciesshoweddifferent patterns with respect to nectar yeast frequency in excludedandexposedtreatments.ExposedDigitalisnectar containedyeastsin60%ofsamples,thisfigurefallingto

30%wheninsectvisitswereprevented. Nevertheless,in Atropanectar,yeastswereonlyrecoveredinbaggedflow- ers. Whenbothairbornemicrobialcontaminationand pollinator visits were prevented at the same time, no yeastswererecoveredfrom pollen and nectar samples, andcorollasurfacescarriedyeastin15%ofsamples.

usingprincipalcomponent analysis (PCA)withcross- validationon thecovariancematrix(STATISTICA7.0;Stat- Soft). Interpretationofthedatawasmadebyinspectionof thescoresandloadingplotsforthefirsttwocomponents, whichaccountedfor41%ofthetotalvariance.

Twenty-nineyeastspecieswereclusteredaccordingto growthresponsestoincreasingsugarconcentrationsbycal- culatingtheEuclideandistancematrixasimplementedin PCORDversion4.0ClusteringAnalysis.AnUPGMAtreewas computedusingPHYLIPNeighbouronthisdistancematrix.

Inviewofthetemperaturerangetowhichyeastswere

naturally exposed (C.M. Herrera, unpublished data) at bothsamplingsites duringthesamplingperiod,yeast growth slopes were represented from eight (minimum field temperature during experiments) to 30° (maxi- mum).Althoughyeastgrowthwasfollowedover18days, consideringtheephemeralnatureofflowersasyeasthabi- tats and, more specifically,that flowersofDigitalisand Atropalastup 4–5days(M.I. Pozo,unpublished data), onlydataongrowthratewithin3daysofincubation at eachtemperaturewere includedintheanalyses.

Results

100

Digitalisobscura

Atropabaetica

Occurrenceofyeastspeciesindifferent microhabitattypes

Thefrequencyofsamplescontainingyeastswasthehigh- est in external sources such as air samples and bees

Microsite

Fig.1.Frequencyofsamplescontainingyeastsineachmicrohabitat forthetwosamplingsitesstudied.

100

Digitalisobscura

Open

Pollinatorsexcluded

Speciesrichnessdidnotfollowthesamepatternasdid the yeastfrequenciesin samples.Beeglossaeharboured eightand fiveyeastspeciesin Digitalis and Atropa, respectively,and fiveyeastspecieswererecoveredfrom airsamplesatboth sites.Thehighestyeastspeciesrich- nesswasfoundincorollasamplesandthelowestwasin pollenand nectar,withtwospeciesineachsubstrateat bothsites.Rarefactionanalysesshowedthatthecoverage ofyeastspeciescouldbegenerallyimprovedbyadditional sampling,exceptfor Atropalocalphylloplaneand Digi- talisnectar,whosecurvesappeartohavereachedapla- teauwiththeexistingsamplingeffort(Fig.3).

100

AtropabaeticaOpen

Pollinatorsexcluded

Airflowprevented

PhylloplanePollenNectar

Microsite

Comparativecompositionofyeast

assemblagesinplants,insects,and airsamples

Yeastcommunities isolatedfrom corollasurfacesat the two localitieswere remarkably similar, judging by the totalnumberoftheChao-sharedestimatedspecies(7).In addition, the yeastsofbractsand beeglossaelocalsets wouldshareuptotwoyeastspecieswhencomparingthe twosamplingsites.

Whenconsideringthetwosamplingsitesseparately,the presenceofsharedspecieswas alsousedtostudypatterns atyeast species’degreeofinterchangebetweenmicrosites.

Fig.2.Frequencyof samplescontainingyeastsin the phylloplane (both corollasurfacesand bract mean±SD),pollen, and nectar samplesfrom the two plant species in sequentialbagging experiments:exclusionofpollinator-vectored yeasts(upperpanel)and airbornepluspollinator-vectoredyeasts(bottompanel).

Spatialdistributionofyeastspecieswithin plants

The234yeastisolates obtainedatthetwostudysites belonged to 17 genera and 36 species(Table1). Both plant specieswerecharacterizedby similar amounts of yeastspeciesrichness,with22(Digitalis)and23(Atropa) species,respectively,intotal.TheDOTUR-basedanalysis ofDNAsequencesyieldedfairlysimilarresultsinAtropa samplingsite,withatotalof26OTUs.InDigitalis,the method recognizedup to 38distinct OTUs,whichsug- geststhatsomeundescribedspecies mayoccurinthe samples.

The yeastassemblagesassociatedwith the two plants weresimilarinshowingamarkednumericaldominance ofafewspecies,namelytheubiquitous phylloplanespe- cies Aureobasidium pullulansplus M.gruessii that was recoveredfromallmicrohabitatsexcepttheair(Table1). Metschnikowia reukaufiiwas frequently recovered from bees,andCryptococcusspeciesdominated theairandthe phylloplane(Fig.3).

Thetwoplantspecieswerecharacterizedbydifferentpat-

ternsofspeciessimilaritybetweenmicrosites.InDigitalis, beeglossaeand phylloplanesamplessharedup to eight yeastspecies.Somepollenyeastspecieswouldbeshared (uptotwo)withphylloplane,beeglossae,andnectarsam- ples.InAtropa,phylloplaneandairsampleswerehighly similarinspeciescomposition,with31sharedspecies.In thissamplingsite,beevisitorswouldexchangealowpro- portionofspecieswithphylloplane(1–2).

Micrositesand physiologicalprofiles

Therelationshipbetweenphysiologicalprofileandmicro- siteoforiginofyeastisolateswasexploredbyPCA.Thefirst twoaxesaccountedfor41%ofthetotalvariation.Themain factorsunderlyingthefirstcomponent (24%ofvariance, PC1,Fig.4)were growthonL-arabinose,2-ketogluconate, inositol, raffinose,and gluconate (r0.7). The second component (PC2,17%oftotal variance)wascorrelated withnitrate,nitrite,andphenylalanine(r0.4)asnitrogen sourcesandwasmoderatelycorrelated with lipid hydrolysis andgrowthonD-glucuronate(seeTableS2inSupporting Informationforfurtherinformation).Yeastspeciesdidnot showawell-definedclusteringinatwo-dimensionalspace definedbythefirsttwoaxesofthePCA,but thereisa microhabitataggregationtrend,inwhichthosestrainsfrom airandphylloplane tendtodisaggregatefromthosestrains recoveredfromnectar,pollen,andbees’ glossae.

Table1. Speciesoffilamentousfungiandyeastsrecovered fromair,beeglossae,phylloplane,nectar,andpollensamples, asafunctionofthe plantspeciessurveyed

DigitalisobscuraAtropabaetica

Fungalspp.

Bees Air Bract

Outer corolla

Inner

corollaPollen Nectar Bees Air Bract

Outer corolla

Inner

corollaPollen Nectar

Aureobasidiumpullulans13 1421112715511

Candidabombi31

Candidafloricola5

Candidafriedrichii 11

Coniochaetaleucoplaca1

Cryptococcusadeliensis1

Cryptococcusaerius511

Cryptococcusfestucosus1 1 1 1

Cryptococcuslaurentii1

Cryptococcusmagnus11

Cryptococcusoeirensis1121

Cryptococcussaitoi 1 1

Cryptococcusstepposus22

Cryptococcusterreus 1

Cryptococcusvictoriae1322

Debaryomyceshansenii1

Debaryomycesmaramus1111

Debaryomycesnepalensis 2

Dothichizapythiophila1

Dothioraelliptica1

Lachanceathermotolerans 1

Metschnikowiagruessii21224619

Metschnikowiakunwiensis124

Metschnikowiareukaufii111211

Ogataeazsoltii 1

Rhizosphaerapini1

Rhodotorulamucilaginosa1

Rhodotorulasp.1

Starmerellabombicola 1 2

Sydowiapolispora / 1 / 1 / 1
Trichosporonmontevidensis / 1
Trichosporonmoliniforme / 1
Ustilagomaydis / 1 / 1
Ustilagosp. / 2 / 1
Zygosaccharomycesmellis / 1
Zygosaccharomycesrouxi / 1
Numberofdistinctspecies / 8 / 5 / 4 / 10 / 9 / 3 / 2 / 5 / 5 / 5 / 11 / 10 / 2 / 2
Numberofyeastisolates / 38 / 11 / 18 / 34 / 26 / 9 / 26 / 21 / 12 / 10 / 14 / 11 / 2 / 2

Shadedboxesindicatethepresenceandthenumberofisolatesofaspeciesinagivensampletype.

Osmotoleranceand responsetosecondary compounds

No noticeable heterogeneity among yeast species was foundintheirresistancetosix secondary compounds assayedinthesubsetexamined.CardenolidesfromDigi- talisdid not induce anyinhibitory responseirrespective of dose or strain tested. Asto alkaloids,all the yeast speciestestedwereequallysusceptibleto concentrations ofatropine and tropine above150μgg—1.Contrary to ourinitialhypothesis,Metschnikowiaspeciesdidnotexhi-

bitagreaterresistanceto thetoxiccompounds assayed. In fact,only Metschnikowia kunwensisand M.reukaufii strainsweresensitivetoScopolamine150μgg—1.

Osmotolerancetestsrevealed that16ofthe29yeast speciesexaminedfailedtogrowat40%glucose.Agreat gradationinyeastgrowthresponseunder increasingglu- cose concentrationsmaybediscernedforthesubsetof osmotolerant yeastspecies(Table2).Forexample,Zygo- saccharomycesandDebaryomyceswerereadilycapable to growat55%.Otheryeast species,suchasStarmerellaand C.bombireachedthesameupperlimit,buttheyshowed

Genus

AureobasidiumCandidaMetschnikowia

Other

CryptococcusDebaryomyces

16Digitalisobscura

Rhodotorula

16Atropabaetica

051015202530

Samples,n

Fig.3.Species accumulationcurves(rarefactionanalysis)separately bymicrohabitatsinDigitalis andAtropalocalities:external sources (solid lines),phylloplane(dottedlines),andpollenandnectarcommunities(longdashlines).Piechartsdisplaythetaxonomicdistributionofisolates

amonggeneraateachmicrohabitatcompartment(externalsources,a;phylloplane,b;pollenandnectar,c).

adelayedgrowthresponseabove45–50%glucose.Mets- chnikowiastrainsmostlyfailed togrowatglucoseconcen- trations above50%.Aclassificationofspeciesbasedon theirsimilaritiesintoleranceto highglucoseconcentra- tions did not revealanydistinct segregationofisolates fromnectarandbeeglossae (relatedtohighglucose microenvironments) withrespecttoisolates fromthe phylloplaneandtheair(Table2).However,highlyosmo- tolerantyeastspecies suchasZygosaccharomycesrouxii, Debaryomyceshansenii, Candida friedrichii,and C.bombi werefound onlyamongphylloplaneisolates,andtwoof thethreenectaryeastspecies didnotgrowin40%of glucosemedium(Fig.S1).

Growthratesand temperatureresponses

Asmany as 38% of the speciestested failedto grow during thefirst3daysafterplating,irrespectiveoftem- perature. Species with delayed growth responses comprised six Cryptococcus species (C. festucosus, C.laurentii,C.magnus, C.oeirensis, C.stepposus, and C.terreus), alongwiththeyeast-likefungusConiochaeta leucoplaca and the plant pathogens Dothioraelliptica

and Dothichizapythiophila. Only17%of theyeastspecies grewattemperaturesbelow12°Cwithin3days(Fig.5). Specifically, D.hansenii and C.friedrichiihad minimum growthtemperaturesof8°C,andthethreeMetschnikow- iaspeciesweretheonlyisolatesabletogrowbelowthis temperature. Astotheupper limitoftemperature toler- ance,threespeciesfailedtogrowatorabove30°C(C. friedrichii,C.victoriae,andD.nepalensis). Onlyfourspe- ciesgrowovertheentirerangeobservedinthefield(8–30 ° C). Those were D.hansenii, M.gruessii, M.kunwiensis, and M.reukaufii.In terms of celldensity, fivespecies reached the highest value of growth in this 3-day trial, namely C.bombi,D.hansenii,M.reukaufii,and Ogataeazsoltii.

Discussion

Despite a substantial body of literature on air- and insect-bornepropagulesandtheirsignificancetodispersal ofepiphyticmicroorganisms(e.g.Andrews,1991;Kinkel,

1997;Lachanceetal.,2001;Starmer etal.,2003),there havebeenfewattempts to correlatequantitativelyshifts in air and insectyeastinocula with specificchangesin

CrA

DoE

CrS

phylloplaneandfloralyeastcommunities(e.g.forairand phylloplane interchange, Andrews etal., 1987). It was found that external sources, namely atmosphere and flower-visitinginsects,werequantitativelyimportant yeast

1ZM

CoL

RhM

CrAe

CrT

CrF

DoE

CrM CrV

donors to thephylloplaneand flowersofthe twoplant speciesstudied.Ourresultsshowedthatnectarandpollen

yeastswere mainlyvectoredbybees’mouthparts andonly

CFl

–1

–2

MRMK

DH CFr

DM DN

Air

Beeglossae Nectar Phyllosphere Pollen

secondarilyfromtheatmosphere.

Theyeastassemblagesrecordedinthisstudycouldbe dividedintothreemaingroupsaccordingtotheirmicro- habitats.Communities ofmostlyascomycetous,osmotol- erantspecieswereassociatedwithinsectmouthparts (i.e. Debaryomyces,Metschnikowia, Starmerella, orZygosacchar- omyces spp.), basidiomycetous(Cryptococcus spp.) yeasts

–2–10123

PC1

Fig.4.Plotrepresentingthe principalcomponentsanalysisof the physiologicalcharacteristicsof 50 yeaststrainsand their natural micrositeoforigin.Thefirsttwoprincipalcomponents (PC1andPC2) accountedfor41% thetotalvariationobserved.The fungalspecies representedhereareCandidabombi (CB),Candidafloricola(CFl), Candidafriedrichii(CFr),Coniochaetaleucoplaca(CL),Cryptococcus adeliensis(CrA),Cryptococcusaerius(CrAe),Cryptococcusfestucosus (CrF),Cryptococcus magnus(CrM),Cryptococcus terreus(CrT), Cryptococcusvictoriae (CrV), Debaryomyces hansenii (DH), Debaryomyces maramus (DM), Debaryomycesnepalensis(DN), Dothichizapythiophila(DoP), Dothioraelliptica(DoE), Lachancea thermotolerans(LT),Metschnikowiagruessii(MG), Metschnikowia kunwiensis (MK),Metschnikowiareukaufii(MR),Rhodotorula mucilaginosa(RhM),Ogataeazsoltii(OZ),Starmerella bombicola(SB), Zygosaccharomyces mellis(ZM),andZygosaccharomyces rouxii(ZR). Curvedlinesaredrawn on chartto separateyeastspecieswhen

needed.

wereisolatedprimarilyfromphylloplaneandairsamples, and floralnectar harboured highlyrestricted communi- ties,represented byM.gruessii in Digitalisand bytwo fungalspeciesinAtropa,composedbythebasidiomyce- tousRhodotorula mucilaginosaandtheascomyceteyeast- likefungusC.leucoplaca.

Yeastsfrom natural environments may be classified according to their growth characteristics (Davenport,

1976).In thisstudy,yeastspeciesassociatedwithinsect visitorsandtheirfloralrewardscouldbediscernedfrom generalisticspeciesfrom the phylloplaneand the airon thebasisoftheircarbon and nitrogen sourceutilization patterns.Thisdivisionwaslargely attributable tothe abundanceofbasidiomycetousyeastspeciesonplantsur- facesand the predominance ofascomycetousspeciesin beesandnectar.Consistentwiththisdistinction,theyeast communities of phylloplane and bee mouthparts were

Table2. Comparativeevaluationofosmotoleranceof13yeastspecies*basedontheirgrowthresponse† onmediacontaining40–55%glucose.

Growthindexwithincreasingglucoseconcentration

(%w/w)

Associationwithhigh-sugar

Fungalspp. / environment (beeand/ornectar) / 40% / 45% / 50% / 55%
Debaryomycesnepalensis / B / 5 / 5 / 5 / 5
Zygosaccharomycesmellis / B / 5 / 5 / 5 / 4
Zygosaccharomycesrouxii / 5 / 5 / 5 / 4
Candidabombi / 5 / 5 / 3 / 3
Debaryomycesmaramus / B / 5 / 5 / 4 / 2
Metschnikowiakunwiensis / B / 6 / 5 / 3 / 2
Candidafloricola / B / 5 / 5 / 5 / 0
Metschnikowiareukaufii / B / 5 / 5 / 4 / 1
Starmerellabombicola / B / 5 / 4 / 3 / 3
Lachanceathermotolerans / B / 5 / 5 / 3 / 1
Metschnikowiagruessii / B+N / 5 / 5 / 3 / 1
Candidafriedrichii / 5 / 4 / 3 / 1
Debaryomyceshansenii / 5 / 4 / 3 / 1

*Thepresentlistisasubsetofthe29yeastspeciesusedinosmotolerancetest.Therestones,whicharethosenotcapabletogrowinmedia containing40%glucose,arelistedinFig.S1.

†Growthindex:missingdata,0;absenceofgrowth,1;weak,2–4;apparentgrowth,5–6.

Candidabombi

4Candidafloricola

Candidafriedrichiii

(a)(b)

Cryptococcusadeliensis

4Cryptococcusaerius

Cryptococcusvictoriae

1015202530

Debaryomyceshansenii

Debaryomycesmaramus

Debaryomycesnepalensis

(c)(d)

Metschnikowiagruessii

4Metschnikowiakunwiensis

Metschnikowiareukaufii

1015202530

Lachanceathermotolerans

Ogataeazsoltii

Rhodotorulamucilaginosa

Starmerellabombicola

(e)(f)

Zygosaccharomycesmellis

4Zygosaccharomycesrouxii

10 / 1520 / 25 / 30 / 10 / 15 / 20 / 25 / 30
Temperature(˚C) / Temperature(˚C)

Fig.5.Growth curvesof 29 yeastspecieseachbelongingto the following genera:Candida(a),Cryptococcus(b),Debaryomyces (c),

Metschnikowia(d),others(e),andZygosaccharomyces(f)onYMmediumatchambertemperaturesfrom8to30°C.Growthindex:absenceof growth,0;weak,1;apparentgrowth,2–4.

dissimilar.Typically, insect-vectoredspecies,suchas Metschnikowiaspp., Zygosaccharomyces spp., and S.bombicola (RosaLachance,1998;Rosaetal.,2003), havebeenalsoisolatedatalesserfrequencyfromcorolla surfaces,whichmight beinterpreted asasignofniche

specialization, but this also indicates that phylloplane strainscoveraverybroadrangeofphysiologicalabilities. Thephylloplaneisgenerally consideredoligotrophic (AndrewsHarris,2000),anddependingonthesystem studied, carbon compounds alone or both carbon and

nitrogen compounds wereshown to belimiting factors for bacterialand yeastpopulations on leaves(Bashi& Fokkema,1977; MercierLindow,2000).Froma microbe’sperspective,the phylloplaneisacontinuously fluctuatingphysical environment,bothspatially andtem- porally(HiranoUpper,2000;LindowBrandl,2003), and soahighphysiologicalplasticityatthecommunity level maybeconsideredanindispensablestepforits colonization.

Typicalphylloplanespecieswereisolatedinthenectar ofAtropa.Inthisplantspecies,nectarsamplesharboured yeastsonlywheninsectvisitswereexcluded,andsoitis not surprisingthat nectarandbeeglossaedidnot share anyspeciesinthiscase.Thefungalspeciesappearingin nectar(R.mucilaginosaandC.leucoplaca)havebeenpre- viouslyisolatedasendophytes(Unterseher Schnittler,

2010),suggestingthatadditionalworkshouldbedoneto assessthe possibilitythat nectar yeastsin Atropanectar mayalsooriginatefromtheplanttissuesthemselves.Itis worthnoting,however,thatR.mucilaginosa isthoughtto bethemostubiquitousyeast known(Sampaio,2010). Moreover, the two pigmented speciesfound in Atropa nectarwere isolatedduringsummermonths,whenpre- vailingtemperatures, moisture levels, daylength,and intensityofsunexposuremaybefavouringthoseyeasts that presumably are better adapted: C.leucoplaca is a saprobicyeast-likefungusthatisabletogrowindrycon- ditionsandwhosemaximumgrowthtemperatureis34°C (Cannon Kirk, 2007, pp. 88), and R.mucilaginosa growsattemperaturesupto36°C(datanotshown).

The presentstudysupportstheconclusionthat,in comparisonwithflower-visiting beesandphylloplane, nectarhasanotablyloweryeastfrequencyandharbours averylimitedsetofyeastspecies.Thelowspeciesrich- nessofnectaryeastcommunitiesprovidessupportforthe

‘nectar-filteringhypothesis’,asproposedinHerreraetal. (2010).Accordingtothishypothesis,thehighsugarcon- tentandthepresenceofplantsecondarycompoundsthat usuallycharacterizenectar samples,aswasnoted inthe introduction, mayhaveastrongimpact on the survival ofincomingyeasts.ThemainnectarivorousMetschnikow- iaspecies,asdescribedbythisstudyandpreviouswork conductedinthesameregion(Herreraetal.,2009,2010; HerreraPozo,2010;Pozoetal.,2011),maythenpos- sesssomesuiteofspecificphysiologicaltraitsthat allow themtosuccessfullyovercomenectarfiltering.

Secondarycompoundsand sugars

ThespecificresistanceofmembersoftheMetschnikowia- ceaecladetoplant secondarycompounds and highglu- cose concentrations will be discussed below, by comparing nectar-inhabiting specieswith the potential

poolofspeciesarrivingtofloralnectarinthetwoplant populationssurveyed. Inourattempttodetectspecific resistanceofnectaryeaststotoxiccompounds produced bytheplantspecies examinedinthisstudy,Digistalis cardenolidesdidnot showanyinhibitoryeffectonyeast growth, and Atropaalkaloidswere equallytoxic to all yeastcolonizers.Giventhatthepresentassay tested concentrations ofthesecompounds that gowellbeyond the proposed threshold for plant parts of Atropasp. (Za´rateetal.,1997),theinhibitoryeffectofalkaloidsmay havebeen overestimated.It can be certainlyconcluded thatnectarivorousyeastspecies didnotshowanyadvan- tagewithrespecttoresistancetosecondarycompounds, asit waspointed out byManson etal.(2007).Despite thelackofinhibitoryeffectofglycosidesonyeastgrowth inourassays,itcannotberuledoutthatacomplexmix- ture ofsecondarycompounds and primary metabolites, orevenamixtureofthesesubstancesandinorganicions, could beeffectiveagainstyeasts.It isalsopossiblethat the commercial substances used in those experiments wereslightlydifferentfromthoseoccurringinnature.All the same, both plant speciesharboured yeastsin their nectarsamples,suchthatitcannotbeconcludedthatthe plantsecondary compoundsusedinourassaywereinhib- itorsinnatural conditions fornectar-depletingyeasts,as ithasbeenproposedbythe‘antimicrobial hypothesis’ (Adler,2000;Golonka,2002,ascited inAntonovics, 2005; Irwinetal.,2004;Mansonetal.,2007).

Theprincipal factorthought to limit nectar commu- nitycomposition istheinabilityofahighproportion of colonizerstosurvivethehighsugarconditionsofnectar. Atropanectarischaracterizedbyaveryhighsugarcon- centration (per cent of total sugars, w/w, mean±SD;

46.5±22.1),andDigitalisnectarpossessesmoremoder-

ate sugar concentration (16.2±2.9). In spite of this, mostyeastspecies foundtobeosmophilicwerenot recoveredfrom nectar samplesfrom either of the two plant species surveyed. The exception was M.gruessii, whichisabletogrowinthepresenceofupto50%glu- coseand wasfound innectarsamplesexaminedinour study.

High sugar content and plant secondarycompounds alonethusdonotexplaintheprevalenceofMetschnikowia speciesinfloralnectar,whichraisesthequestionofwhat additionalfactorsshouldbeconsideredinthe‘nectar-fil- tering’hypothesis.Thepossibility cannotberuledout, however,thatsomeofournegativeresultswerethecon- sequenceofinsufficient statisticalpowerowingtorela- tivelysmallsamplesizes(i.e.committingtypeIIerroror failingtorejectafalsenullhypothesis).Theinformation availableonly allowsto speculate that if such effects doactuallyexist, theyprobablyareneitherstrongnor pervasive.

Recentstudieshavesuggestedthat interspecificdiffer- encesinthermaltolerancelimitsorgrowthresponsescan playacentralroleinyeast communityorganizationby influencingthespecies compositionofcommunities (Lachanceetal.,2003;Sweeneyetal.,2004; Goddard,

2008;SampaioGonc¸alves,2008).Differentialthermal tolerancemaydeterminewhatyeastspecies survivein naturalpopulationsforagivenseason,butthisdoesnot explainwhythelowestspeciesrichnesswasinfactfound in nectar samples.Moreover,draining and direct radia- tionmaystrengthenthiseffectonthephylloplane.

Yeast speciesgrowthrate

Rapidgrowthshouldbeasignificantadvantagefornec- tar colonizers, given the ephemeral nature of this resource. Nectar secretion by individual flowersat the twospring–summerbloomingplant speciesexaminedin this study lastsfor up to 3days.Yeastspecieswith a delayed growth response, in such a short time span, should be‘swept’ bycompetition from the faster-grow- ingMetschnikowiaspecies. Evenatlowtemperatures, Metschnikowia speciesreach a high growth rate, higher than that of other species that is capable to grow between8and31°Cwithin3days,suchasD.hansenii. Thetemporal constraintsimposedbytheshortduration ofthefloral microhabitatcouldalso explainthelower frequencyofyeastsinnectarsamplescomparedtoexter- nalsourcesandthephylloplane.Assuming thatthe ephemeralnature offlowersand/or the nectar secretion phasemaybeastronglimitingfactorfornectar-coloniz- ingyeasts, thisduration-restricted effect maybe strengthened in plant specieswith high visitation rates. In Digitalisplants, four speciesofsocialbeeswerethe mainvisitorsduringourexperiment.IntheAtropapop- ulation,moreinsectvisits werecounted,andvisitors covered abroaderspectrumthatincludedsolitarybees, socialbees,and Lepidoptera.When pollinator composi- tion,nectaryeastfrequency,andspecies richnesswere exploredin abroader rangeofplant speciesand study sites in the same region (one population each of 13 plant species,M.I.Pozoand C.M.Herrera,unpublished results), the frequency of total flower visits by social beeswascorrelated positivelywith yeastfrequenciesin thenectarsamples(rs=0.55,P=0.05).Incontrast,the frequencyofvisits bysolitarybeescorrelatedpositively with nectar yeast species richness (13 populations, rs=0.65;P=0.01).