Zooming-inonfloralnectar:afirstexplorationof nectar-associatedbacteriainwildplantcommunities
SergioA´lvarez-Pe´rez,CarlosM.HerreraClaradeVega
Estacio´nBiolo´gicadeDon˜ana,ConsejoSuperiordeInvestigacionesCientı´ficas(CSIC),Sevilla,Spain
Correspondence:SergioA´lvarez-Pe´rez, Estacio´nBiolo´gicadeDon˜ana,Consejo SuperiordeInvestigacionesCientı´ficas(CSIC), Avda.Ame´ricoVespucio,E-41092Sevilla, Spain.Tel.:+34954466700;
fax:+34954621125;
e-mail:
Keywords
bacteria;floralnectar;microbialcommunities;
pollination.
Introduction
Abstract
Floralnectarofsomeanimal-pollinatedplantsusuallyharbourshighlyadapted yeastcommunitieswhichcanprofoundlyalternectarcharacteristicsand,there- fore,potentiallyhavesignificantimpactsonplantreproduction through their effects oninsectforagingbehaviour.Bacteriahavealsobeenoccasionally observedinfloralnectar,buttheirprevalence,phylogeneticdiversityandeco- logical rolewithinplant–pollinator–yeastsystemsremainsunclear.Herewe present the firstreported surveyofbacteriain floralnectar from a natural plantcommunity.Culturablebacteriaoccurringinatotalof71nectarsamples collectedfrom27SouthAfricanplant specieswereisolatedandidentifiedby
16SrRNAgenesequencing.Rarefaction-basedanalyses wereusedtoassess operational taxonomic units (OTUs) richnessat the plant community level usingnectardropsassamplingunits.Ourresultsshowed thatbacteriaare commoninhabitantsoffloralnectarofSouthAfricanplants(53.5%ofsamples yieldedgrowth),andtheircommunitiesarecharacterizedbylowspeciesrich- ness(18OTUsata16SrRNAgenesequencedissimilaritycut-offof3%)and moderate phylogeneticdiversity,withmostisolatesbelongingtotheGamma- proteobacteria. Furthermore, isolatesshowed osmotolerance, catalaseactivity andtheabilitytogrowundermicroaerobiosis,threetraitsthatmighthelpbac- teriato overcomeimportant factorslimitingtheir survivaland/or growthin nectar.
(NicolsonThornburg, 2007).Nevertheless,pollinators actnot onlyaspollenvectorswhilevisitingplants,but
Plantsprovideextraordinarilydiversehabitatsformicro- organisms(Andrews& Harris,2000).Plant-associated habitats,suchasroots,leaves,flowers,fruitsordecaying tissues,differin their localavailabilityofnutrients and physicochemical conditions,thusfilteringtherangeof potentialmicrobialinhabitants(AndrewsHarris,2000; Herreraetal.,2010).Fromamicrobiologicalperspective, only roots havebeen extensivelyinvestigated,especially on topics relatedto rhizosphere microbial communities and mycorrhizalor legume–Rhizobiumsymbioses(Long,
1996; AndrewsHarris,2000; KentTriplett,2002; SmithRead,2008).However,otherplantmicrohabitats remainvirtuallyunexplored.Thislatteristhecaseforflo- ralnectar.
Historically,floralnectarhasbeenregardedmerelyas asweetaqueoussecretion,containing sugarsand amino acids,offeredby floweringplants to attract pollinators
atthesame timetheycanmovemicroorganismsfrom flowertoflower(SandhuWaraich,1985;Brysch-Herz- berg,2004;Herrera et al.,2009).Asarichsourceof nutrients,nectarcouldharboura microbiotathatmay consumesugaractivelyandproduce arangeofmetabo- lites,entailingadecreasein itsattractivenessand ener- geticvaluefrom the viewpoint of pollinators. Thus, it hasbeenpostulatedthatplantsshouldhave evolved mechanismstomaintainnectarfree ofmicroorganisms (Adler,2000;CarterThornburg,2004).Inthisrespect, in recent yearsseveralclassesof antimicrobial proteins and secondary compounds putatively protecting nectar frommicrobialinvasionhavebeenisolated(Carteretal.,
2007;ParkThornburg,2009;Heil,2011).Forexample,
it has been hypothesized that reactive oxygen species, suchashydrogenperoxide(H2O2), keepnectarpalatable for visiting pollinators by limiting microbial growth,
thereby preventingtoxinbuildupandreducingthe breakdownofsugars andothernectarcomponentsby microbialmetabolism(CarterThornburg,2004;Carter et al.,2007).Despitetheseassumptions, severalrecent studieshaverevealed thatfloralnectarofanimal-polli- natedplantsfromdifferentcontinentsandavariety of disparatehabitat typescanharbour highlyadapted yeast communities (Brysch-Herzberg, 2004; Herrera etal.,
2008,2009,2010;deVegaetal.,2009;Pozoetal.,2011; deVega &Herrera,2012).Furthermore, ithasbeen demonstrated that nectarivorous yeastscan profoundly alter both the sugar and amino acid composition and theoverallenergeticcontent ofnectarasaconsequence oftheirmetabolicactivity(Cantoetal.,2007,2008;Her- reraetal.,2008;deVegaetal.,2009;deVegaHerre- ra, 2012;Peay etal., 2012) and warm the flowersof somewinter-bloomingplants(Herrera& Pozo,2010). Therefore, theseeukaryoticmicroorganismscouldhave significanteffects onplantreproduction through their effects on insect foraging behaviour (Herrera etal.,
2010).
Thehandfulofstudiesthat haveaddressedsofarthe topic ofnectar microbiota mostlydealwithyeasts(but seeJunkeretal.,2011;whichstudiesthe microbiota of
identificationoffutureresearch directionsinnectar microbiology.
Materialsand methods
Samplesand microbiologicalanalysis
Seventy-onesamplesoffloralnectarfrom27plantspecies belongingto13familieswereanalysed. Acompletelistof speciessampledand the familyto whichtheybelongis providedinTable1.Floralnectarcollectionwascarried outatseverallocalitiesintheKwaZulu-Natalprovinceof SouthAfricadifferinginecologicalcharacteristicsinclud- ingelevation, soilandtypeofvegetation(forfurther information, seedeVegaetal.,2009).Allplantsbelong- ing to the same species were collected on the same location and, in some cases,plants of different species co-occurred.Sampledindividualswereatleast5mapart.
Table1. Isolationof bacteriafrom nectar samplesof 27 South
Africanplantspecies
Bacteria
floral surfaces). Bacteria have also been occasionally
Plantfamily*SpeciesNo.†
frequency‡
observed in the nectar of some plants (Gilliam etal.,
1983; EhlersOlesen,1997),buttheirprevalenceand phylogeneticdiversityhavenot beenassessed to date. Furthermore, theecologicaland functional roleofthese microorganismsinplant–pollinator–yeastsystemsremains unclear.Bacteriacandegradesugars,transformtheminto compounds whicharedifficulttoassimilate byother microorganisms, produce alcohols, and release a wide arrayofsecondary metaboliteswhicharetoxictoyeasts and/or insects(Latour &Lemanceau,1997;Stro¨metal.,
2002;Barton, 2005;Bode,2009).Hence,their potential impactonplantpollinationcannotberuledout.
Herewepresentthefirstreportedsurveyofbacteriain floralnectarfromanaturalplantcommunity. Our main objectiveswereto assesstheprevalence,speciesrichness andphylogenetic diversityofbacteriainasetofnectar samplesfromwild-growingSouthAfricanplants.Inthis regard, the present report complements our previous workonnectaryeasts associatedwiththesameSouth Africanplantcommunity(deVegaet al.,2009).Addi- tionally,wedeterminedthreephysiologicalcharacteristics ofbacterialisolatesthat mightberelevantfortheirsur- vival infloralnectar,namelyosmotolerance,catalase activity andtheabilitytogrowundermicroaerobiosis. These traitsmighthelpbacteriatoovercome,respectively, highsucroselevels,toxichydrogenperoxideandpossible oxygen limitation(e.g.derivedfrommicrobialmetabo- lism) occurring in nectar. These analysesallowed the
AcanthaceaeAdhatodaandromeda32
Ruelliacordata31
Thunbergianatalensis21
AmaryllidaceaeHaemanthushumilis31
CaryophyllaceaeSilenebellidioides32
EricaceaeEricacerinthoides31
FabaceaeEriosemadistinctum11
IridaceaeDieramaluteo-albidum30
Freesialaxa22
Gladiolusappendiculatus10
Gladioluslongicollis31
Gladiolusparvulus32
Moraeaunibracteata22
Watsonialepida10
Watsoniapillansii30
LamiaceaeAjugaophrydis32
Stachysaethiopica10
OrchidaceaeDisacrassicornis22
OrobanchaceaeCycniumadonense44
Cycniumracemosum42
Graderiascabra11
ProteaceaeProteacaffra44
Protearoupelliae30
RubiaceaeBurchelliabubalina32
ScrophulariaceaeGlumicalyxgoseloides62
XanthorrhoeaceaeKniphofiacaulescens11
Kniphofiasp.32
Total7138(53.5%)
*FamilialclassificationfollowsthatofStevens(2011).
†Numberofnectarsamplesanalysedperplantspecies.
‡Numberofnectarsamplesfromwhichbacteriawereisolated.
Collectedbranches,inflorescences orflowersalready openandexposedtopollinatorswerecarefullyplacedin plastic jars in a portable cooler until taken indoors, and then keptunder refrigeration(4 °C)until further processing.Extractionsofnectarfromindividualflowers, usingsterilecalibratedmicrocapillaries(FisherScientific, Madrid, Spain), wereconducted within 24h afterfield collection.Nectarstandingcropvariedwidelydepending on the plant species, ranging from approximately 1–
3lL perflowerinAdhatodaandromeda,Ajugaophrydis,
Cycnium adonense,Cycnium racemosum, Dieramaluteo- albidum,Disacrassicornis, Eriosema distinctum,Freesia laxa,Gladiolusappendiculatus,Gladiolusparvulus, Glumi- calyxgoseloides, Graderiascabra, Haemanthus humilis, Moraeaunibracteata, Ruelliacordata,Silene bellidioides and Stachys aethiopica; 3–7lL per flowerin Burchellia bubalina,Ericacerinthoides, Kniphofiasp., Thunbergia natalensis, Watsonialepidaand Watsoniapillansii;to morethan 30lL perflowerinGladioluslongicollis, Kni- phofiacaulescens, Proteacaffraand Protearoupelliae. Nectarsugarconcentration wasmeasuredforsomeplant species withalow-volumehandrefractometer(Belling- ham StanleyLtd,Tunbridge Wells,UK),and exhib- ited extensive variation at both the inter- and intraspecificlevel.Forexample,thevaluesofsugarcon- centration (in percentofsucroseequivalents)forsome ofthesespecieswereasfollows:G.longicollis (11–35%, mean 26.5%), K. caulescens (7–16%, mean 12%), P.caffra (2–12%,mean7%),P.roupelliae (2–10%,mean
6%),G.appendiculatus (19–31%,mean 25%),W. lepida
(10–29%,mean 21%) and W.pillansii(8.5–25%,mean
16%).
Nectarsampleswereimmediatelydilutedin500lL of ultrapure deionized water (Purite Select; Purite Ltd, Thame, UK), and stored at 4°C until processed.This procedure has proven to be similar to other methods (e.g.dilutingin0.85–1%NaClsolutions)inmaintaining nectarmicrobiotainoptimalconditions.Twenty-fivemi- crolitresofnectardilutionswas streakedontrypticasesoy agar(TSA;Panreac,CastellardelValle`s,Spain).Cultures were incubated at room temperature (c.25°C) for
7days.Acolonyofeachphenotypicallydistincttypewas picked and separatelysubcultivated on TSAto obtain purecultures.Allisolateswerestoredat—20°CinLuria
–Bertani (LB)broth(Difco,Sparks, MD)containing25%
glycerol(Sigma-Aldrich,Madrid,Spain).
Phenotypiccharacterizationofbacterialisolates
Bacterialisolateswerefurther characterizedbyassessing theirreactiontohydrogenperoxide(catalase activity), sucrosetoleranceandtheability togrowatlowoxygen levels(microaerobiosis),usingthethreefollowingtests:
(1) Fordetectionofcatalaseactivity,abacterialcolony wastakenfromanaxeniccultureonTSAwithamicrobi- ologyloopandwassuspendedin3%hydrogenperoxide (Panreac).Theappearanceofbubbleswasrecordedasa positiveresult(Aslanzadeh,2006).
(2) Sucrosetolerancewasassessedbyculturingisolates atroomtemperatureforupto7days intransparentplas- ticvialscontainingLB broth supplementedwith0% (positive control), 10%, 20% and 30% sucrose (w/v, Sigma-Aldrich).Theappearanceofturbidityinthetubes withrespecttonegativecontrols(i.e.tubescontainingno inoculatedmedia)wasrecordedasapositiveresult.The rangeofsugarconcentrationstestedroughlycorresponds withnaturalvariationobserved infloral nectarsinwild SouthAfricanplants(seeabove).
(3) Growthundermicroaerobiosiswas determinedby culturing isolateson TSAand incubating the plates at room temperature for72hinacandlejar.Theappear- anceofcoloniesontheplateswasrecordedasapositive result.
DNA isolation,PCRamplificationand sequencingofthebacterial16SrRNA gene
GenomicDNAwasisolatedbyboilingbacterialcolonies in 500lL of ultrapure deionized water at 100°C for
20min. Cell debris was removed by centrifuging at
8000gfor2min.
Thebacterial16SrRNAgenewasamplifiedusingthe universalprimer1492R(5′-GGTTACCTTGTTACGACTT-
3′) (Reysenbachetal.,2000)and theeubacterial-specific primer 27F(5′-AGAGTTTGATCMTGGCTCAG-3′, where M=A or C) (Braker etal., 2001). Reaction mixtures contained 5lL of NH4 buffer (109; Bioline,London, UK), 1mM MgCl2, 0.4lM of each primer (Sigma- Aldrich),250lMofeachdNTP(Sigma-Aldrich), 3UBi- otaqDNApolymerase(Bioline)and2–5lLDNAextract. Thefinalvolumewasadjustedto 50lL withultrapure deionizedwater.AmplificationwascarriedoutinaFlex- CyclerPCRthermalcycler(AnalytikJena,Jena, Germany) andconsistedofadenaturation stepof4min at94°C, followedby35cyclesof90sat94°C,90sat50°Cand
2minat72°C,andafinalextensionat72°Cfor2min. PCR products were cleaned up with ExoSAP-IT(USB Corporation, Cleveland, OH),whichdegradesexcess primersandnucleotides.
Sequencing ofampliconswasperformedusingtheABI Prism BigDye Terminator v3.0ReadyReactionCycle Sequencingkit(AppliedBiosystems,Madrid,Spain)and the following six primers (Sigma-Aldrich): 27F, 515F (5′-GTGCCAGCMGCCGCGGTAA-3′,where M=A or C),906F(5′-GAAACTTAAAKGAATTG-3′),519R (5′- GWATTACCGCGGCKGCTG-3′,whereW=AorTand
K=GorT),907R(5′-CCGTCAATTCCTTTRAGTTT-3′, whereR=AorG)and1492R(Reysenbachetal.,2000). Thesequencesweredeterminedonanautomatedsequen- cer(ABIPrism3130xl;AppliedBiosystems),andassem- bledand manuallyedited with the program SEQUENCHER ver. 4.9(GeneCodesCorp.,AnnArbor,MI).TheGen- BankaccessionnumbersoftheDNAsequencesobtained inthisstudyareJN872496–JN872548 (forfurther details seeSupportinginformation,Table.S1).
Dataanalyses
The16SrRNAgenesequencesobtainedfromnectarbac- teriawerecomparedwithreferencesequences fromthe GenBankdatabases,usingBLASTsoftware( nlm.nih.gov/Blast.cgi)and the RibosomalDatabasePro- ject (RDP) website ( Isolates wereassigned togenusorthehighesttaxonomicrank possible,leavingfurther hierarchicaltaxonomyunidenti- fied.
16SrRNAgenesequenceswereincludedinamultiple alignment generated by CLUSTALW (Chenna etal.,2003) andtheresultingalignmentwas trimmedbyGblocks (Castresana,2000)to eliminatepoorlyalignedpositions anddivergentregions.Aphylogenetic treewas con- structedfornectarisolatesandreferencesequencesusing MRBAYESv3.1.2(RonquistHuelsenbeck,2003)asimple- mented on the CIPRESScienceGateway(Miller etal.,
2010).Thesimplestmodelofsequenceevolutionamong thoseavailableinMRBAYESthatbestfitsthesequencedata wasdetermined usingtheAkaikeInformation Criterion. ThistestwasconductedusingtheJMODELTEST0.1.1pack- age(Posada,2008),andresultedinselectionofageneral time-reversiblemodelwithgamma-distributed ratevaria- tion across sites and a proportion of invariable sites (GTR+G+I).Fourchainswererun twice(chaintem- perature=0.2; sample frequency=100) until average standard deviation of split frequencieswasbelow0.01 (c.9.2milliongenerations).A50%majority-ruleconsen- sustreewascalculatedusingthesumtcommandanddis- carding the first 25% of the trees to yield the final Bayesianestimate of phylogeny.The resulting tree was finallydrawnandfurther editedwithTREEGRAPH2(Sto¨ver
& Mu¨ller, 2010).
Determinationof thenumberof distinctoperational taxonomicunits(OTUs)occurringinourset of DNA sequences and assignment of sequences to OTUs was done with the program MOTHUR v.1.17.3(Schlossetal.,
2009).DNAdissimilaritycut-offsof 1% and 3% were usedin theseanalyses.Toassessthe overallrichnessof bacterial OTUs,sample-basedrarefactionmethodswere appliedto presence–absence data.Dueto the limited number ofnectar samplesavailablein this work,OTU
occurrencedatafromallsampleswereanalysedtogether (i.e.irrespectiveoftheplantspeciesand/orfamilyofori- gin).Averagerarefactioncurveswerecomputed withthe ESTIMATESv.8.2.0program (Colwell,2009),using50ran- domizations and samplingwithout replacement. Asour dataarebasedonincidence, theICE andChao2nonpara- metricestimatorsoftheexpectedOTUrichnesswerealso calculated.
Results
Theresultsofthemicrobiologicalanalysis ofnectar samplesare provided in Tables1 and 2.Atotal of53 bacterialisolateswererecoveredfrom 38nectarsamples (53.5%, n=71), and 21 of the plant speciessurveyed (77.8%,n=27).NobacteriawererecoveredfromD.lu- teo-albidum, G.appendiculatus,P.roupelliae, S.aethiopic- a, W. lepida or W.pillansii. All bacterial isolates recovered fromnectarwere abletogrowundermicro- aerobiosisandshowedapositivereactioninthecatalase test (Table3). Most isolates also tolerated 10–30% (w/v) sucrose, the exceptions being some isolates fromthegeneraBurkholderiaandMethylobacterium, and
the familiesSphingomonadaceae and Xanthomonadaceae
(Table3).
IntheMOTHUR analysisoftheDNAsequencedatafor bacterialisolates, 18OTUswereidentifiedatthe3%dis- similaritycut-off(OTUs0.03;Table2).Onlysixadditional OTUswereidentifiedwhenthedissimilaritycut-offwas loweredto1%,thusgivingatotalof24OTUs0.01.Given thesmalldifferenceintotalOTUsobtainedwiththetwo thresholds,onlyOTUs0.03 were retainedforsubsequent analyses,asthisrepresentsthethresholdcommonlyused todistinguishbacterialOTUsatthespecieslevelinenvi- ronmental studies (Lambaisetal.,2006;Teixeiraetal.,
2010).
Whenrarefaction-basedmethodswereappliedtoobtain reliable estimatesoftotalbacterialOTUrichness,the OTU0.03 rarefactioncurvewasclose toreachingaplateau for the number of nectar samples examined (n=71; Fig.1).AlthoughadditionalOTUsareexpectedtoappear byfurtherincreasingthesamplingeffort(and/or lowering theDNAdissimilaritycut-off;seeFig.S1),resultsofthis surveyseemtoprovideareliablebasisforestimatingover- all bacterialOTUrichnessinthefloralnectarofthesetof plants surveyed. OTU0.03 richness estimates were 26.8
Table3. Physiologicalcharacteristicsofnectarbacterialisolates
Taxonomicalaffiliation
Growthunder
Sucrosetolerance
Catalase
ofOTUs0.03* / microaerobiosis / production / 10% / 20% / 30%Actinobacteria
Leifsoniasp.(3) / + / + / + / + / +
Microbacteriaceae(2) / + / + / + / + / +
Micrococcaceae(3) / + / + / + / + / +
Firmicutes
Bacillussp.A(2) / + / + / + / + / +
Bacillussp.B(1) / + / + / + / + / +
Paenibacillussp.(1) / + / + / + / + / +
Proteobacteria
Alphaproteobacteria
Asaiasp.(3) / + / + / + / + / +
Methylobacteriumsp.A(2) / + / + / + / + / V
Methylobacteriumsp.B(1) / + / + / + / – / –
Sphingomonadaceaesp.A(2) / + / + / V / V / –
Sphingomonadaceaesp.B(1) / + / + / + / – / –
Betaproteobacteria
Alcaligenaceae(1) / + / + / + / + / +
Burkholderiasp.(6) / + / + / + / + / V
Gammaproteobacteria
Enterobacteriaceae(2) / + / + / + / + / +
Pantoeasp.(9) / + / + / + / + / +
Pseudomonassp.(12) / + / + / + / + / +
Stenotrophomonassp.(1) / + / + / + / + / +
Xanthomonadaceae(1) / + / + / + / + / –
+,positive;—,negative; V,variable.
*ThenumberofisolatesbelongingtoeachOTU0.03isgiveninparentheses.AsinTable2,toavoidconfusion,differentOTUsbelongingtothe samefamilyorgenuswerenamedassp.Aandsp.B.
Fig.1.Graphicalrepresentationoftherarefactioncurve(solidline) andnonparametricestimatorsofnectarbacteriaOTU0.03richnessfor ourdataset: ICE(longdashes)andChao2(dottedline).
(ICE estimator) and 22.9 species (Chao2 estimator; Fig.1).Therefore,oursamplingrecoveredmorethan67% oftheestimatednumber ofbacterialOTUsoccurringin thenectarofsampledplantspeciesinthestudyarea.
Phylogeneticanalysesshowedadistribution ofisolates amongthreemajorbacterialphyla: Actinobacteria, FirmicutesandProteobacteria(Alphaproteobacteria,Beta- proteobacteria and Gammaproteobacteria) (Table2 and Fig.2),thelast-namedbeingthemostfrequent(77.4%of isolates). Furthermore, Pseudomonas and Pantoea were thetwomostcommon bacterialgenerarecovered,albeit withalowoverallincidence(16.9%and12.7%ofnectar samplesanalysed,respectively).OtherProteobacteriagen- eraidentifiedinphylogeneticanalysiswereAsaia, Burk- holderia, Methylobacterium, Stenotrophomonas,andseveral otherrepresentativesfromthefamiliesAlcaligenaceae,En- terobacteriaceae,Sphingomonadaceaeand Xanthomonada- ceae. Ontheotherhand,15.1%ofbacterialisolates belongedto the phylum Actinobacteria, and wereclassi- fiedwithin the familiesMicrobacteriaceae (including Le- ifsonia sp.) and Micrococcaceae. Finally,membersofthe phylumFirmicutescomprisedonlyanegligiblefractionof isolates(7.5%),and belongedtothegeneraBacillusand Paenibacillus.
Discussion
Wehavepresentedherethefirstanalysis ofbacterial communities associated with floral nectar in a diverse array of wildplants, whichrepresents a necessarystep towardsabetterunderstanding ofmultikingdom interac- tionssurrounding insect-pollinatedflowersinnature.We focuson nectar, the main reward offeredbyplants to theirpollinators,whichisconsideredherethekeyfloral resourceinsupporting abacterialmicrocosm.Thisisin sharp contrast to previous investigations which either focusonthebacterialcommunitiesinhabitingotherfloral
parts (Junker etal.,2011)or do not providedetailson thefloralorgansfromwhichthemicrobiotawassampled (Yamada etal., 2000; Lachance etal., 2003; Yukphan etal.,2004).
Amainfindingfromthepresentstudywasthatbacte- riaarerelativelycommon inhabitants offloralnectarof animal-pollinated SouthAfricanplants,beingpresentin
21plantspeciesandmorethan ahalfofsamplesanaly- sed.ThispictureissimilartothatencounteredbyEhlers
Olesen(1997)inEpipactishelleborineatdifferentloca-
tionsinnorthernEurope.Unfortunately,thereis no additionalinformation ontheprevalence ofbacteriain nectarofwildplants.
Ontheotherhand,thestudiedbacterial microbiota associatedwithfloralnectarwas characterizedbyrela- tivelylowspeciesrichness.EighteenbacterialOTUswere identifiedinMOTHUR-basedanalysesatthe3%dissimilar- itycut-off.Lowering thiscut-offto1%allowedthe identification of six additional OTUs, which did not have a dramatic impact on rarefaction-based estimates oftotal bacterialOTU richness(seesupporting Appen- dix S1 and Fig. S1). Furthermore, although it is expectedthat additional samplingeffortwould increase thenumberofOTUsidentified, rarefactionanalysis revealedthat,forthegroupofplantssampled,werecov- eredahighproportion oftheOTUrichness ofnectar- inhabitingbacteria.Asimilarlowvalueforspeciesrich- ness has been reported for nectar yeast communities (Pozoet al.,2011),butcontrastswiththehighspecies richnessofother plant-associatedenvironments, suchas the rhizosphere (Teixeira etal., 2010; Weinert etal.,
2011)or the phyllosphere(Lambaisetal.,2006).Along thesameline,nectarbacterialcommunitieswerecharac- terizedbyamoderate phylogeneticdiversity,astheiso- latesbelonged toonlythreedifferentbacterialphyla: Proteobacteria, ActinobacteriaandFirmicutes.Withinthe Proteobacteria, nectar bacteria were distributed among theAlphaproteobacteria, Betaproteobacteria andGamma- proteobacteria. Aconsiderableproportion ofisolateswere membersof theGammaproteobacteria,withPseudomonas and Pantoeabeingthepredominant genera.Lowphylo- geneticdiversity isalsoacharacteristicofnectar-associ- atedyeastcommunities (Brysch-Herzberg,2004;Herrera etal., 2010; Pozo etal., 2011; de Vega Herrera,
2012),aconcordancewhichstressesthepotentialroleof nectar as a habitat filter that excludesspeciesthat do not possesshabitat-specificphysiological adaptations. Interestingly, althoughwefoundnodominant bacterial speciesin the setofnectar samplesstudied, one single yeastspecies(Metschnikowia reukaufii)hasbeenrepeat- edly isolatedfromnectaratdifferentlocations(Eisiko- witchetal.,1990;Brysch-Herzberg, 2004;Herrera etal.,
2010;Pozoetal.,2011;deVega &Herrera,2012).
Fig.2.Phylogeneticrelationshipof16SrRNAgenesequencesfromnectarbacteriaretrievedinthisstudyfromSouthAfricanplants(indicatedin boldtypeandwithcollectionreferencenumbers)andreferencesequencesoftypestrainsstoredintheGenBank database, asdeterminedby Bayesianinference.Deinococcus radiodurans servedastheoutgroupforthisanalysis.GenBankaccessionnumbers andfurtherdetailsonnectar isolatesandreferencestrainsareprovidedinTableS1.Numbersabovebranchesshowcladecredibilityvalues(posteriorprobabilities).
Apartfromthepredominance ofGammaproteobacteria inthenectarfromSouthAfricanplantsreported inthis work,membersofthisclasshavebeenalsoidentifiedas thedominant bacterialinhabitants ofother plant-associ- ated environments, such asthe surfaceofleaves(Erco- lani,1991;Thompson et al.,1993;Krimmet al.,2005; Junkeretal.,2011)and petals(Junker etal.,2011),the interior ofpitchersofsomecarnivorousplants(Siragusa et al.,2007;Koopmanet al.,2010),andsugar-richsap exudates(Lagace´ etal.,2004,2006;Filteauetal.,2010). However, while Pseudomonas was the most prevalent genus among nectar isolates (Table2) and has been repeatedlyidentifiedasakey componentofepiphytic bacterialcommunitiesonleaves(Ercolani,1991;Thomp- sonetal.,1993;Krimmetal.,2005;Junkeretal.,2011), the results from a recent investigation show that the surface of petals of some plant species are predomi- nantlycolonizedbymembersofthefamilyEnterobacteri- aceae (Junker etal.,2011).Nevertheless,in the present work we have only identified bacterial isolates up to genuslevel,whichprecludesfurther species-basedcom- parisonsof microbialcommunitycompositionwithpre- vious reports. Furthermore, as nectar-inhabiting and flowerepiphytic microbial communities have not been yetextensivelycharacterized,andintrafloralmicrohabitat heterogeneity remainspoorlyunderstood,adetailed comparative analysis of floral nectar and petals as microbialhabitatscannotbeperformed.
Nearlyallbacteriaisolatedfrom floralnectar in this studyappearto bephysiologicallyableto overcomethe threemainstressorscharacteristicoftheirhabitat,namely highosmoticpressure,lowoxygenlevelsandpresenceof toxichydrogenperoxide.Although thephysiological mechanisms allowingsurvivalofnectar microorganisms inthissugar-richhabitathavenotbeenstudiedindetail to date, they could be similar to those employed for copingwiththeosmoticstressimposedbyhighlevelsof solutesinother environments suchassalterns,hypersa- linelakes,andsaltyorsugaryfoodproducts(reviewedby Beales,2004;Grant,2004).Theabilityofnectarbacteria togrowatlowoxygenlevelsmightberelevantinsitua- tions where oxygendiffusion through nectaries is hindered (e.g.inplantswithrelativelylonghorn-shaped nectaries, whereabiofilmusuallyappearsattheair-nectar interface;C.M.Herrera,pers.obs.)orwhenenvironmen- taloxygenisdepletedbymicrobialmetabolism.Addition- ally,catalaseactivitymight protect nectar bacteriafrom the toxic action of H2O2, as demonstrated for other plant-associatedmicrobes(XuPan,2000).Nevertheless, ithasalsobeennoted that insomebacterialspeciesthe presenceorabsenceofcatalaseisnotcorrelatedwiththe ability of the microorganism to overcome the lethal effectsofH2O2, assusceptibilitytothistoxiccompound
alsodependsonotherfactors(see,forexample,Schwartz etal.,1983;WilsonWeaver,1985).Moreover,Carter et al. (2007) demonstrated that some Proteobacteria, includingstrainsofPseudomonassyringaeandPseudomo- nasfluorescens, weresensitivetotheH2O2 concentrations observedinthefloralnectarofornamentalNicotianaspe- cies.Thus,thehypothesizedrelationshipbetweencatalase activityandsurvivalinnectarclearlydeserves further investigation.
Apartfromthethreestressors mentionedabove, additional factors not considered in this study could limit microbial growth and/or survivalin floralnectar. Forexample,ithasbeenrecentlydemonstrated that the antimicrobial activity of Petuniahybridanectar is not based on H2O2 production but on RNase activities (Hillwigetal.,2010).Antimicrobialproperties havealso beenattributed toaGDSLlipaseofthefloralnectarof Jacarandamimosifolia (Kram etal.,2008).Thisgrowing listofnectarproteinsandsecondary metabolitespoten- tiallyimplicatedinplant antimicrobial defence(seealso Adler,2000)contrasts with the lackofinformation on the physiologicalstrategiesof nectar microbes for adapting to their stressful habitat (but see Herrera etal.,2012,forarecentstudyonnectaryeasts).Inany case,strongselectivepressuresareexpectedtoturn nec- tarintoa potentialmicroorganism-freeenvironment, operating over micro- and macroevolutionary time scales.
Insummary,ourresultshave shownthatbacteriaare common inhabitants of floral nectar of South African plants,andtheircommunitiesarecharacterized bylow species richnessandmoderatephylogeneticdiversity. Moreover,we haveidentifiedosmotolerance,catalase activityandtheabilitytogrowunder microaerobiosisas traitsthatmighthelpbacteriatoovercomeimportant fac- tors limiting their survival and/or growth in nectar. Futureworkshouldclarifytheroleofbacteriawithinthe plant–yeast–pollinatorsystemandmighthelptofill a conspicuousgapinour knowledgeofecologicalinterac- tionsinvolvingmacro-andmicroorganismsattheinter- sectionofseveralkingdoms.
Acknowledgements
WethankProf.S.D.Johnsonforlogisticalsupport and hospitalityandR.G.Albaladejo,H.Burger,V.Ferrero,J. Rodger,P.Rymer,A. ShuttleworthandS.Steenhuisenfor fieldassistance.Thisworkwassupported byfundsfrom theConsejerı´adeInnovacio´n,Cienciay Empresa,Juntade Andalucı´a(ProyectodeExcelenciaP09-RNM-4517toC.M. H.),andMinisteriodeCienciaeInnovacio´n(Juandela CiervaProgrammetoC.dV.). BeatrizGuzma´nisgratefully acknowledgedfor herassistancewith phylogeneticanalyses.
References
AdlerLS(2000)Theecologicalsignificanceoftoxicnectar.
Oikos91:409–420.
AndrewsJHHarrisRF(2000)Theecologyand biogeographyofmicroorganismsonplantsurfaces.Annu RevPhytopathol 38:145–180.
AslanzadehJ(2006)Biochemicalprofile-basedmicrobial identification systems.AdvancedTechniquesinDiagnostic Microbiology(TangY-WStratton CW,eds),pp.84–116. Springer,NewYork.
BartonL(2005)StructuralandFunctionalRelationshipsin
Prokaryotes,pp.529–601.Springer,NewYork.
BealesN(2004)Adaptation ofmicroorganisms tocold temperatures, weakacidpreservatives,lowpH,andosmotic stress:areview.Compr RevFoodSciFoodSaf3:1–20.
BodeHB(2009)Entomopathogenicbacteriaasasourceof secondarymetabolites.CurrOpinChemBiol13:224–230.
BrakerG,Ayala-del-Rı´oHL,DevolAH,FesefeldtATiedje
JM(2001)Community structure ofdenitrifiers,Bacteria,andArchaeaalongredoxgradientsinPacificNorthwestmarinesedimentsbyterminalrestriction fragmentlength polymorphism analysisofamplifiednitritereductase(nirS) and16SrRNAgenes.ApplEnvironMicrobiol67:1893–1901.
Brysch-HerzbergM(2004)Ecologyofyeastsinplant– bumblebeemutualism inCentralEurope.FEMSMicrobiol Ecol50:87–100.
Canto A,Pe´rezR,Medrano M,CastellanosMCHerrera
CM(2007)Intra-plant variation innectarsugar composition intwoAquilegiaspecies(Ranunculaceae): contrasting patterns under fieldandglasshouseconditions. AnnBot99:653–660.
CantoA,HerreraCM,MedranoM,Pe´rezRGarcı´aIM
(2008)Pollinator foragingmodifiesnectarsugar composition inHelleborusfoetidus(Ranunculaceae):an experimentaltest.AmJBot95:315–320.
CarterCThornburg RW(2004)Isthenectarredoxcyclea floraldefenseagainstmicrobialattack?TrendsPlantSci9:
320–324.
CarterC,HealyR,O’ToolNM,NaqviSM,RenG,ParkS, BeattieGA,Horner HTThornburg RW(2007)Tobacco nectariesexpressanovelNADPHoxidaseimplicatedinthe defenseoffloralreproductive tissuesagainst microorganisms.PlantPhysiol143:389–399.
CastresanaJ(2000)Selectionofconservedblocksfrom multiplealignmentsfortheiruseinphylogeneticanalysis. MolBiolEvol17:540–552.
ChennaR,SugawaraH,KoikeT,LopezR,GibsonTJ,Higgins DGThompson JD(2003)Multiplesequencealignment withtheClustalseriesofprograms.Nucleic AcidsRes31:
3497–3500.
Colwell RK(2009)EstimateS:statisticalestimation ofspecies richnessandsharedspeciesfromsamples,Version8.2. User’sGuideandapplication.Availableat: org/estimates.
deVegaCHerreraCM(2012)Relationshipsamongnectar- dwellingyeasts,flowersandants:patterns andincidenceon nectartraits.Oikos,inpress.
deVegaC,HerreraCMJohnsonSD(2009)Yeastsinfloral nectarofsomeSouthAfricanplants:quantification and associationswithpollinatortypeandsugarconcentration. S AfrJBot75:798–806.
EhlersBKOlesenJM(1997)Thefruit-wasproutetotoxic nectarinEpipactisorchids?Flora192:223–229.
EisikowitchD,KevanPGLachanceMA(1990)Thenectar- inhabitingyeastsandtheireffectonpollengerminationin commonmilkweed,AsclepiassyriacaL.IsrJBot39: 217–225.
ErcolaniGL(1991)Distribution ofepiphyticbacteriaonolive leavesandtheinfluenceofleafageandsamplingtime. MicrobEcol 21:35–48.
FilteauM,Lagace´L,LaPointeGRoyD(2010)Seasonaland
regionaldiversityofmaplesapmicrobiota revealedusing community PCRfingerprinting and16S rRNAgeneclone libraries.SystApplMicrobiol33:165–173.
GilliamM,MoffettJOKauffeldNK(1983)Examination of floralnectarofCitrus,cotton, andArizonadesertplantsfor microbes.Apidologie14:299–302.
GrantWD(2004)Lifeatlowwateractivity.ProcRSocLondB
BiolSci359: 1249–1267.
HeilM(2011)Nectar:generation,regulationandecological functions.TrendsPlantSci16:191–200.
HerreraCMPozoMI(2010)Nectaryeastswarmtheflowers ofawinter-blooming plant.ProcRSocLond BBiolSci277:
1827–1834.
HerreraCM,Garcı´aIMPe´rezR(2008)Invisiblefloral larcenies:microbialcommunities degradefloralnectarof bumblebee-pollinatedplants.Ecology89:2369–2376.
HerreraCM,deVegaC,CantoAPozoMI(2009)Yeastsin floralnectar:aquantitative survey.AnnBot103:1415–1423. HerreraCM,CantoA,PozoMIBazagaP(2010)Inhospitable sweetness:nectarfilteringof pollinator-borneinoculaleadsto
impoverished,phylogeneticallyclusteredyeastcommunities.
ProcRSocLond BBiolSci277:747–754.
HerreraCM,PozoMIBazagaP(2012)Jackofallnectars, masterofmost:DNAmethylationandtheepigeneticbasis ofnichewidthinaflower-livingyeast.MolEcol, doi:10.1111/j.1365-294X.2011.05402.x.
HillwigMS,LiuX,LiuG,Thornburg RWMacintoshGC (2010)Petunianectarproteinshaveribonucleaseactivity. JExpBot61:2951–2965.
JunkerRR,LoewelC,GrossR,Do¨tterlS,KellerABlu¨thgen N(2011)Composition ofepiphyticbacterialcommunities differsonpetalsandleaves.PlantBiol13:918–924.
KentADTriplettEW(2002)Microbialcommunities and theirinteractions insoilandrhizosphereecosystems.Annu RevMicrobiol 56:211–236.
Koopman MM,FuselierDM,HirdSCarstensBC(2010) Thecarnivorouspalepitcherplantharbours diverse, distinct,andtime-dependentbacterialcommunities. Appl EnvironMicrobiol76:1851–1860.
KramBW,BainbridgeEA,PereraMACarterC(2008) Identification, cloningandcharacterization ofaGDSLlipase secretedintothenectarofJacarandamimosifolia. PlantMol Biol68:173–183.
KrimmU,Abanda-Nkpwatt D,SchwabWSchreiberL (2005)Epiphyticmicroorganisms onstrawberryplants (Fragariaananassacv.Elsanta):identification ofbacterial isolatesandanalysisoftheirinteraction withleafsurfaces. FEMSMicrobiol Ecol 53:483–492.
LachanceMA,BowlesJMStarmerWT(2003)Metschnikowia
santaceciliae,Candidahawaiiana, andCandidakipukae,three new yeastspeciesassociatedwithinsectsof tropicalmorning glory.FEMS YeastRes3:97–103.
Lagace´ L,Pitre M,JacquesM RoyD (2004)
Identification ofthe bacterial community ofmaple sap by usingamplifiedribosomalDNA(rDNA)restriction analysisand rDNA sequencing. ApplEnvironMicrobiol
70:2052–2060.
Lagace´ L,JacquesM,MafuAARoyD(2006) Compositionsofmaplesapmicroflora andcollection systembiofilmsevaluatedbyscanningelectronmicroscopy anddenaturing gradientgelelectrophoresis.IntJ Food Microbiol109:9–18.
LambaisMR,CrowleyDE,CuryJC,Bu¨llRCRodriguesRR (2006)BacterialdiversityintreecanopiesoftheAtlantic forest.Science312: 1917.
LatourXLemanceauP(1997)Me´tabolismecarbone´ et
e´nerge´tiquedesPseudomonasspp.fluorescentssaprophytesa`
oxidasepositive.Agronomie17:427–443.
LongSR(1996)Rhizobiumsymbiosis:nodfactorsin perspective.PlantCell8:1885–1898.
MillerMA,PfeifferWSchwartzT(2010)CreatingtheCIPRES ScienceGateway forinferenceoflargephylogenetictrees. ProceedingsoftheGatewayComputingEnvironmentsWorkshop (GCE),pp.1–8.NewOrleans,LA.
NicolsonSWThornburg RW(2007)Nectarchemistry. NectariesandNectar(NicolsonSW,NepiMPaciniE, eds),pp.215–264.Springer-Verlag,Dordrecht.
ParkSThornburg RW(2009)Biochemistryofnectar proteins.JPlantBiol52:27–34.
PeayKG,BelisleMFukamiT(2012)Phylogenetic relatednesspredictspriorityeffectsinnectaryeast communities. ProcRSocLond BBiolSci279: 749–758.
PosadaD(2008)jModelTest:phylogeneticmodelaveraging.
MolBiolEvol25:1253–1256.
PozoMI,HerreraCMBazagaP(2011)Speciesrichnessof yeastcommunities infloralnectarofsouthern Spanish plants.MicrobEcol61:82–91.
ReysenbachAL,LongneckerKKirshteinJ(2000)Novel bacterialandarchaeallineages fromaninsitugrowth chamberdeployedataMid-AtlanticRidgehydrothermal vent.ApplEnvironMicrobiol66:3798–3806.
RonquistFHuelsenbeckJP(2003)MrBayes3:Bayesian phylogeneticinferenceunder mixedmodels.Bioinformatics
19:1572–1574.
SandhuDKWaraichMK(1985)Yeastsassociatedwith pollinatingbeesandflowernectar.MicrobEcol 11:51–58.
Schloss PD,WestcottSL,RyabinTetal.(2009)Introducing mothur: open-source, platform-independent,community- supported softwarefordescribingandcomparing microbial communities. ApplEnvironMicrobiol75:7537–7541.
SchwartzCE,KrallJ,Norton L,McKay K,KayDLynchRE (1983)CatalaseandsuperoxidedismutaseinEscherichiacoli. JBiolChem258:6277–6281.
SiragusaAJ,SwensonJECasamattaDA(2007)Culturable bacteriapresentinthefluidofthehooded-pitcherplant Sarraceniaminorbasedon16S rRNAgenesequencedata. MicrobEcol 54:324–331.
SmithSEReadDJ(2008)Mycorrhizal Symbiosis,3rdedn.
AcademicPress,London.
Stevens PF(2011)Angiospermphylogenywebsite,version9. Availableat: APweb/(lastaccessed21October2011).
Sto¨verBCMu¨llerKF(2010)TreeGraph2:combiningand visualizingevidencefromdifferentphylogeneticanalyses. BMCBioinformatics11:7.
Stro¨mK,Sjo¨grenJ,BrobergASchnu¨rerJ(2002) LactobacillusplantarumMiLAB393producestheantifungal cyclicdipeptidescyclo(L-Phe-L-Pro)andcyclo(L-Phe-trans-
4-OH-L-Pro) and3-phenyllacticacid.ApplEnviron
Microbiol68:4322–4327.
TeixeiraLC,PeixotoRS,CuryJC,SulWJ,PellizariVH,Tiedje JRosadoAS(2010)Bacterialdiversityinrhizospheresoil fromAntarcticvascularplantsofAdmiraltyBay,maritime Antarctica.ISMEJ4:989–1001.