Comparison of bacterial communities from lava cave microbial mats to overlying surface soils from

27 Pages·2017·3.57 MB·English

RESEARCHARTICLE
Comparison of bacterial communities from
lava cave microbial mats to overlying surface
soils from Lava Beds National Monument,
USA
KathleenH.Lavoie1☯*,AraS.Winter2☯,KaitlynJ.H.Read2,EvanM.Hughes2,Michael
N.Spilde3,DianaE.Northup2☯
1 Biology,StateUniversityofNewYork,CollegeatPlattsburgh,Plattsburgh,NY,UnitedStatesofAmerica,
2 Biology,UniversityofNewMexico,Albuquerque,NM,UnitedStatesofAmerica,3 InstituteofMeteoritics,
a1111111111
UniversityofNewMexico,Albuquerque,NM,UnitedStatesofAmerica
a1111111111
a1111111111
☯Theseauthorscontributedequallytothiswork.
a1111111111 *[email protected]
a1111111111
Abstract
Subsurfacehabitatsharbornoveldiversitythathasreceivedlittleattentionuntilrecently.
OPENACCESS
Accessiblesubsurfacehabitatsincludelavacavesaroundtheworldthatoftensupport
Citation:LavoieKH,WinterAS,ReadKJH,Hughes
extensivemicrobialmatsonceilingsandwallsinarangeofcolors.Littleisknownaboutlava
EM,SpildeMN,NorthupDE(2017)Comparisonof
cavemicrobialdiversityandhowthesesubsurfacematsdifferfrommicrobialcommunities
bacterialcommunitiesfromlavacavemicrobial
matstooverlyingsurfacesoilsfromLavaBeds inoverlyingsurfacesoils.Toinvestigatethesedifferences,weanalyzedbacterial16SrDNA
NationalMonument,USA.PLoSONE12(2): from454pyrosequencingfromthreecolorsofmicrobialmats(tan,white,andyellow)from
e0169339.doi:10.1371/journal.pone.0169339
sevenlavacavesinLavaBedsNationalMonument,CA,USA,andcomparedthemwithsur-
Editor:LorenzoBrusetti,FreeUniversityofBozen/ facesoiloverlyingeachcave.Thesamephylawererepresentedinbothsurfacesoilsand
Bolzano,ITALY
cavemicrobialmats,buttheoverlapinsharedOTUs(operationaltaxonomicunit)wasonly
Received:July23,2016 11.2%.Numberofentrancespercaveandtemperaturecontributedtoobserveddifferences
Accepted:December15,2016 indiversity.Intermsofspeciesrichness,diversitybymatcolordiffered,butnotsignificantly.
Actinobacteriadominatedinallcavesamples,with39%fromcavesand21%fromsurface
Published:February15,2017
soils.Proteobacteriamadeup30%ofphylafromcavesand36%fromsurfacesoil.Other
Copyright:©2017Lavoieetal.Thisisanopen
majorphylaincaveswereNitrospirae(7%)followedbyminorphyla(7%),comparedto
accessarticledistributedunderthetermsofthe
CreativeCommonsAttributionLicense,which surfacesoilswithBacteroidetes(8%)andminorphyla(8%).Manyofthemostabundant
permitsunrestricteduse,distribution,and sequencescouldnotbeidentifiedtogenus,indicatingahighdegreeofnovelty.Surfacesoil
reproductioninanymedium,providedtheoriginal
sampleshadmoreOTUsandgreaterdiversityindicesthancavesamples.Althoughsurface
authorandsourcearecredited.
soilmicrobesimmigrateintounderlyingcaves,theenvironmentselectsformicrobesableto
DataAvailabilityStatement:Allsequences,
liveinthecavehabitats,resultinginverydifferentcavemicrobialcommunities.Thisstudyis
taxonomicassignments,andRscriptfilesare
thefirstcomprehensivecomparisonofbacterialcommunitiesinlavacaveswiththeoverly-
availablefromthezenododatabaseat:https://
zenodo.org/record/58072. ingsoilcommunity.
Funding:FundingwasprovidedbyLavaBeds
NationalMonument(LABE)throughtheColorado
PlateauCooperativeEcosystemStudiesUnit
(http://nau.edu/CEFNS/Forestry/cpcesu/),Award
numberJ8410100028-UNM-56.Funder’srole:
LABEpersonnelShaneFryerandShawnThomas
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 1/27
Surfacevs.lavacavebacterialmatcommunities
helpedintheselectionofstudysites,supplied Introduction
metadataforallsites,andassistedwithfieldwork.
MostlifeonEarthintheaphoticsubsurfaceismicrobial[1],butthereismuchthatwedonot
CompetingInterests:Theauthorshavedeclared
yetknowaboutsubsurfacelife.Cavescanprovideanaturalwaytoaccesssubsurfaceenviron-
thatnocompetinginterestsexist.
mentsrangingfromverydeeplimestonecaves(KruberaCaveintheWesternCaucasusis
morethan2,190mdeep[2]),toshallowcaves,suchaslavacavesthathaveanoverburdenof
upto10m[3].DiscoveryofextensivelavaflowsandlavacavesonMars[4]supportsthecon-
ceptthatEarth’slavacavesmayserveasamodelforthestudyoflifeonotherplanets(astrobi-
ology)[5,6,7].
Lavacaves,formedduringactivelavaflows,containdiversemicrobialmatsthatrangefrom
extensivematscoveringwallsandceilingstosmall,scatteredcolonies(Fig1andFig2).Matcolors
includewhite,yellow,tan,gold,orange,andpink,withshadesinbetween[7–11].Despitetheir
extensivenature,littleisknownaboutmicrobialmatdiversity(reviewedin[11,12]).Studiesof
microbialdiversity[7–15]inlavacaveslagbehindsuchstudiesinkarstcaves.StonerandHowarth
[16]firstdescribedthematsor“slimes”inHawaiianlavacavesusingculture-dependentmethods
forisolationofchemoheterotrophicmicroorganismsandreportedonthepresenceoffungiand
aerobicbacteria.Theysuggestedthatwhiteandbrownslimesmightbeimportantsitesfornutrient
cyclingincaves,particularlynitrogen.
Lavacavesareextremeenvironments,simplifiedbythelackofphotosynthesisinthedeep
ordarkzoneofthecave,resultinginextremelyoligotrophicconditions.Thesimplifiednature
ofcavesmakesthemamodelnaturallaboratorytostudyfactorscontrollingbiologicaldiversi-
fication[17,18].Theisolationofmostcaveslimitstheabilityoforganismstomigrate,resulting
inhighlevelsofendemismamongtroglobiontsandstygobiontsasthenorm[19].Culveretal.
[20]foundthatabout30%ofcave-adaptedinvertebratespeciesinU.S.cavesarefoundinonly
asinglecave.TheresultsofHathawayetal.[21]showthatthetrendcanbeextendedtobacte-
rialdiversityinAzoreanandHawaiianlavacavemicrobialmats.Theauthorswerefarfrom
samplingtotaldiversity,butlessthan5%oftheOTUs(operationaltaxonomicunit)foundin
lavacavesoccurinothercavesorinothervolcanicenvironments.Sequencesweremorelikely
toberelatedtosamplesfromthesamecaveorthesameislandthanbetweenislands.Ifmicro-
bialdistributionisubiquitous,thentheywouldexpectahigherpercentageofsharedOTUs
betweenthetwoislandarchipelagos.
Biospeleologistsoriginallythoughtthatcavemicrobesweresimplyasubsetofsurfacemicrobes
washedintounderlyingcaves[22,23].Ortizetal.[24,25]recentlypublishedwhatwebelievetobe
thefirstcomparisonsofcavemicrobialdiversitywiththeoverlyingsoilmicrobialcommunity.
Theirstudiesfocusedonbacterialdiversityacrosscarbonatespeleothemsurfacessampledby
swabbingfromakarstcave,KartchnerCaverninArizona.Comparisonofbacterialtaxonomic
profilestosurfacesoilsamplesrevealedmajordifferencesandonlya16%overlapbetweencave
speleothemandsurfacesoilOTUs[24].
Ourstudyisthefirstandmostcomprehensivetocomparelavacavebacterialmatcommu-
nitiestobacterialcommunitiesfromtheoverlyingsurfacesoilofeachcave.Weexamineda
rangeofenvironmental,geographical,andchemicalfactorsthatmaycontributetobacterial
diversityinmicrobialmatsofdifferentcolors(tan,white,andyellow)fromlavacavesinLava
BedsNationalMonument,California,USA.
Materialsandmethods
Fieldstudiesethicsstatement
AllsamplingwasdoneunderPermitLABE-2011-SCI-0007issuedtoNorthupbytheNational
ParkService(LavaBedsNationalMonument,1IndianWellHeadquarters,POBox1240,
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 2/27
Surfacevs.lavacavebacterialmatcommunities
Fig1.Overviewofsomeofthelavacavesamplingsitesandcaves,plusaviewofthesurfaceterrain
atLavaBedsNationalMonument,CA(LABE).(A)EntrancetoValentineCave.(B)Surfacesamplestaken
aboveHopkinsChocolateCave.(C)YellowmicrobialmatsamplingsiteinValentineCave.(D)Extensive
yellowmicrobialmatsonwallsofHopkinsChocolateCave.(E)PassageinValentineCaveshowingless
microbialmatcoverageneartheentrance.(F)TanmicrobialmatsampletakeninL-V460Cave.Photos
copyrightKennethIngham(A,D,E)andDianaNorthup(B,C,F).
doi:10.1371/journal.pone.0169339.g001
Tulelake,CA96134).LavaBedsNationalMonumentisafederally-protectedareaunderthe
NationalParkService,DepartmentoftheInterior.Noprotectedspeciesweresampled.Shawn
Thomas,ShaneFryer,andJokeVansweevelthavegiventheirconsenttohavephotosofthem-
selves,takenduringsamplingtrips,usedinthismanuscript.
Samplingsites
LavaBedsNationalMonument(LABE)islocatedinnorthernCaliforniaclosetotheborders
ofOregonandNevada[26].TheMonumentcovers190km2ontheNEflankoftheMedicine
LakeVolcano.Two-thirdsofthelavacamefromtheMammothandModoccratersoverthe
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 3/27
Surfacevs.lavacavebacterialmatcommunities
Fig2.Microbialmatandcolonymorphology.(A)Overviewofpredominantlyyellowandwhitemicrobial
mats,someseparateandsomeintermixed.(B)Overviewoftanandwhitemicrobialmats.(C)Intermixedtan,
white,andyellowmicrobialmats.Closeupsof(D)tanandwhitecolonymorphology,(E)tancolonies,and(F)
yellowcolonies.
doi:10.1371/journal.pone.0169339.g002
lasttwomillionplusyearsandasrecentlyas1,100yearsago.Flowsarelargelyofbasaltwith
smalleramountsofsilica-richbasalticandesite[27].
LABEhasthelargestnumberoflavacavesinNorthAmerica,with778known[26].
Twenty-fiveofthelavacaveshavesignedentrancesanddevelopedtrailsforeaseofvisitation.
Theareaisahigh-elevation(1219–1737mabovesealevel),semi-ariddesertwithaverage
yearlyprecipitationof375mm.Temperaturerangesfromanaveragelowof-5.4˚CinJanuary
toanaveragehighof22.3˚CinJulyandAugust.Someofthelavacavescontainperennialice
[27].
WeworkedwithLABEpersonneltoselectsevenlavacavestocoverarangeofparameters
(Table1):amountofhumanvisitation,ageofthelavaflow,surfacevegetation,elevation,
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 4/27
Surfacevs.lavacavebacterialmatcommunities
Table1. CharacteristicsofthesevenstudycavesitesinLABE.
Cave SurfaceSoil Visitation LavaAge Surface Elevation #Entr Branches(Nodes) Length
Catacombs C15 High 36,000±16,000 Sagebrush 1,524m 1 32 2571m
GoldenDome GD25 High 36,000±16,000 Sagebrush 1,378m 2 7 679m
GE-L350 GE13 Low 16,000 Sagebrush 1,506m 1 9 439m
HopkinsChocolate(HCC) HC15 High 36,000±16,000 Sagebrush 1,506m 4 4 519m
L-V460 L14 Low 12,260 Juniper 1,360m 2 5 700m
S-L280 S10 Low 36,000±16,000 Sagebrush 1,372m 2 5 671m
Valentine V16 High 12,260 Juniper 1,3768m 1 12 651m
Cavenameorcodeforbackcountrycaves.Visitation=numberofvisitors,groupedintoHigh(opentothepublic)andLow(closedtovisitation).Lava
Age=Ageoflavaflow.Surface=surfacevegetation,whereSagebrush=Bigsagebrush-antelopebitterbrushscrublandandJuniper=Juniper-mountain
mahoganyscrubland.Elevationistheelevationinmattheentrancea.s.l.#Entr=numberofaccessibleorfunctionalentrancesintothecave.Branches
(Nodes)isanindicationofthegeometriccomplexityofthecavebycountingthenumberofpassagebranchpoints.Lengthisthemappedlengthofthecave
inm.
doi:10.1371/journal.pone.0169339.t001
numberofentrances,branchingcomplexity,andmappedlengthofthecaveInadditiontothe
cavesamples,asampleofsoiloverlyingeachcaveentrance(e.g.Fig1B)wascollectedforcom-
parisonandphoto-documentedatthecollectionsite(Fig1F).
Cavesamples(Table2)werecharacterizedbycolorofthemicrobialmats(tan,white,and
yellowe.g.Fig2),distancefromthenearestentrance,temperature,humidity,andRH,andpH
whensuitablewaterpoolsordrippingwaterwereavailablenearmats.Allsampleswerephoto-
documentedatthecollectionsite
Temperature,RH,andpHmeasurements
Temperature(webbulbanddrybulbinordertoobtainanapproximateRH)wastakenin
April,2011andinSeptember,2012withanIMCtemperatureprobe(http://www.
imcinstruments.com/),whichwascalibratedatfrequentintervalsinthecavetoimproveaccu-
racy.WetbulbreadingswereobtainedwiththeIMCprobesheathedwithwickingsoakedin
deionizedwaterbeforeeachreading.ForsomeRHsamplesaportableKestrel3000wind
meter(https://kestrelmeters.com/products/kestrel-3000-wind-meter)wasused,whichwascal-
ibratedatthebeginningofeachcave.AJavascriptprogram(http://home.fuse.net/clymer/
water/wet.html)useddryandwetbulbtemperaturestoapproximaterelativehumidity.Read-
ingsforpHweretakenwithaTwinCardypHmeter(SpectrumTechnologies,Inc.,http://
www.specmeters.com/nutrient-management/ph-and-ec-meters/ph/cardy-twin-ph-meter/),
calibratedwithpH7buffer.
SamplecollectionsforDNAandScanningElectronMicroscopy(SEM)
SamplingtookplaceinApril2011withadditionalsampling(GEM2,HCC1,SC2)inAugust
2012toincreasethenumberofwhitematsamples.Yellow,white,andtanmatsweresampled
fromeachcave.SamplesforDNAextractionwerecollectedasepticallywithaflame-sterilized
coldchiselintoasterile50ccFalcontube.Soilsampleswerecollectedfromaboveeachcave
entrancebyremovinganysurfaceplantdetritusandscoopingthetop2cmofsoilintoasterile
50ccFalcontube.AllsamplesforDNAwerecoveredwithinhourswithsucroselysisbuffer
[28]toreleaseandstabilizetheDNA.AllsampleswerebroughttotheNorthupLabattheUni-
versityofNewMexicoforfurtherprocessingandanalysiswithinsevendays.Wecollectedsur-
facesoilsfromabovesevencaves,andhadsevensamplesofwhiteandtanmatsandnine
samplesfromyellowmats,withatleastoneofeachcolorfromeachcave.
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 5/27
Surfacevs.lavacavebacterialmatcommunities
Table2. Cavesamplesbymatcolorandenvironmentalcharacteristicsinthecave.
Cave Sample Color Distancefromentrance T˚C %RH pH
Catacombs C1 T 51m 13.6 61 nd
Catacombs C2 Y 51m 13.6 61 nd
Catacombs C9 W 76m 15.7 45 nd
GoldenDome GD1 Y 278m 9.3 99 8.13
GoldenDome GD2 T 276.5m 9.2 100 7.96
GoldenDome GD3 W 276.5m 9.2 100 7.96
GoldenDome GD16 Y 278m 9.3 99 8.13
GE-L350 GE1 T 64m 9.9 87 7.45
GE-L350 GE2 Y 64m 9.9 87 7.45
GE-L350 GEM2 W 67m 7.1 86.6 nd
HCC HC1 Y 37m 11.5 89 nd
HCC HC6 T 107m 8.9 100 7.98
HCC HC7 Y 107m 8.9 100 7.98
HCC HCC1 W 105m 8.9 100 7.97
L-V460 L1 Y 200m 9.3 96.3 7.31
L-V460 L3 W 200m 9.3 96.3 7.31
L-V460 L9 T 87m 9.8 92.7 nd
S-L280 S2 Y 255m 16.1 82.1 nd
S-L280 SC2 W 135m 16.1 82.1 nd
S-L280 S4 T 145m 13.1 70.6 nd
Valentine V1 Y 135m 11.2 100 7.45
Valentine V2 T 135m 11.2 100 7.45
Valentine V13 W 170m 11.3 100 7.75
C=Catacombs;GEandGEM=G-L350;GD=GoldenDome;HCandHCC=HopkinsChocolate;L=L-V460;SandSC=S-L280;andV=Valentine.
Color:T=Tan,Y=Yellow,W=White.Distanceisthedistanceinmfromthenearestentrancetothesamplingsite.Temperature,RH,andpHdatawere
collectedinApril2011andAug2012.(nd=notdetermined)
doi:10.1371/journal.pone.0169339.t002
Samplesforscanningelectronmicroscopy(SEM)fromrockchipsweremounteddirectly
ontoSEMstubswithsuperglueandplacedinacarryingcasefortransport.Amicrobialmat
sample(L-V460-110425-6),consistingofchipsofthewallrockwithwhitetopaleyellowcolo-
nies,wastakenapproximately200mintoCaveL-V460atthebottomofapillar.Oneofthe
twoyellowmicrobialmatsamples(HC110423-5)analyzedwithSEMwasarockchipwithyel-
lowcoloniesfromthefloorofHopkinsChocolateCave,approximately36mintothecave.The
secondyellowmatsample(S-L280-110427-3)wastakenapproximately60mintoCaveS-L280
and1mabovethefloorand1mbelowtheceiling.
Waterchemistryanalysis
Samplesforchemicalanalysiswerepreservedwith6Nhydrochloricacidinthefield,as
describedin[29].Amountsofchloride,nitrite,nitrate,phosphate,andsulfatewereanalyzed
usingaDionexIonChromatographDX-100(Dionex,Sunnyvale,CA,USA)asdescribed[29].
Molecularphylogeny
DNAextraction,sequencing,andsequenceanalysis. DNAwasextractedfromtriplicate
samplesofrockchipswithmicrobialmatsfromeachcavebymatcolorandfromthesurface
soilsamplesusingtheMoBioPowerSoilTMDNAextractionkitfollowingmanufacturer’s
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 6/27
Surfacevs.lavacavebacterialmatcommunities
protocolexceptweusedbeadbeatingratherthanvortexing,whichtheNorthupLabfindsto
bemoreeffectiveatreleasingDNAfromGrampositivecells.
PolymeraseChainReaction(PCR). PCRwasperformedtoverifythequalityandquan-
tityoftheDNApriortosequencing.Onehundredtwentyfivetothreehundredngofpurified
DNAwasusedtoamplifythe16SrRNAgenefromenvironmentalDNAbyPCRwithuniver-
salprimers,p46forward(5’-GCYTAAYACATGCAAGTCG-3’)andp1409reverse(5’-GTG
ACGGGRGTGGTGTRCAA-3’;[30]andAmpliTaqLD(AppliedBiosystems)withanMJther-
malcyclerusing:4mindenaturationat94˚Cfollowedby35cyclesof45secannealingat55˚C,
extensionfor2minat72˚C,denaturationfor30secat94˚C,withafinal45sec55˚Canda20
min72˚Cextensionstepaftercyclingwascomplete.
Sequencingandphylogeneticanalysis. Sampleswereanalyzedwithnext-generation
sequencingofthe16SSSUgenebacterialV1-3region(primer27F)usingRocheFLXandTita-
nium454technologyconductedbyMRDNA,Shallowater,TX(http://www.mrdnalab.com/).
All454datawereprocessedinQIIME1.9.1[31].Qualitycontrolandtrimmingofthe454
datasetweredoneusingthesplit_libraries.pycommandwithalowerlength(-l)of100bpand
anupperlength(-L)of500.Aqualityscore(-s)of30waschosen.Removaloferroneous
sequences(denoising)andOTUclusteringweredoneusingpick_de_novo_otus.pypipeline
withthesumaclustoption[32].Thesumaclustalgorithmismainlyusefultodetectthe’errone-
ous’sequencescreatedduringamplificationandsequencingprotocols.OTUswereclustered
atthe97%similaritylevelusingsumaclust.Thepick_de_novocommandalsopickstherepre-
sentativesetandassignstaxonomyusinguclust[33]againstthegreengenes13.8database[34].
Thepipelinealsoalignsandbuildsaphylogenetictreeusingpynast[35]andfasttree[36]from
therepresentativesequenceset.ChimeracheckingwasdoneusingUSEARCHtodetectarti-
factscreatedduringsequencing.Good’scoverageshowedthatweweresuccessfulingetting
nearlyallofthediversityfromoursamples.Valuesrangedfrom99.11%to87.13%withan
averagevalueof94.98%.Librarysizesrangedfrom471to4317.Rarefactioncurves(S1Fig)
weregeneratedusingacustomfunctionwrittenbyBelaHausmann(https://github.com/
joey711/phyloseq/issues/143)andisalsoavailableinourzenodoarchive.Theseplotscompare
estimatedtotalrichnesspersampleatagivendepth.Therarefactioncurvesarecalculatedwith
observed(totalrichness),Shannon,Chao1andtheirassociateduncertainties.
DiversitybyphylawiththeProteobacteriaseparatedoutbyclasswascomparedforallsam-
ples.TheL2phyladatawerereducedtoeighteengroups,includingunassignedphylaanda
groupwecalledminorphyla.Theminorphylaareentriesthathadlessthan1000OTUsacross
allsamples.Ofatotalof140,848OTUs,26,609werefromthesurfacesamples,and38,214
fromtan,36,014fromwhite,and40,011fromyellowmats.TheprocesswasrepeatedattheL6
genuslevelresultinginninegroups,anunassignedgroup,andminorgenerawith471taxa.
SequencessubmittedtotheNCBIGenBankdatabase(www.ncbi.nlm.nih.gov/genbank/)
wereassignedAccessionNumbersJX694094-JX702544,andthethreeadditionalwhitesam-
plesKP705489-KP706447.
Statisticalanalysis
Communitydissimilaritywasvisualizedusingthephyloseqpackage[37]andggplot2[38]inR
[39].AlphadiversitywasanalyzedusingobservedOTUs,ShannonandChao1indicesinthe
phyloseqpackage.Alphadiversitymeasureswerecarriedoutontherawdataasrecommended
[37].Observedrichnesswillscalewithincreasinglibrarysize;howeverbothChao1andShan-
nonarerobustmeasuresofrichnessanddiversity,whichaccountfordifferencesinlibrarysize
(rarefactioncurvesinS1Fig).ObservedOTUsistherawnumberofOTUspresentineach
sampleofqualitycontrolledandclusteredsequencesasdescribedabove.Betadiversitywas
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 7/27
Surfacevs.lavacavebacterialmatcommunities
analyzedusingnon-metricdimensionalscaling(NMDS)withtheBrays-Curtisdistanceusing
theveganpackage[40]inR.TheBrays-Curtisdistancewaspickedbecauseitisinvariantto
changesinunits,unaffectedbyadditionsandremovalofsamples,andNMDSrecognizesdif-
ferencesintotalabundanceswhenrelativeabundancesaresimilar.Theordinationofthesam-
pleswasdoneusingcustomRscriptsbyUmerZeeshanIjazavailableat:http://userweb.eng.
gla.ac.uk/umer.ijaz/bioinformatics/ecological.html.
DifferentialabundanceoftaxawascharacterizedusingtheDESeq2package[41]with
parametersfitType=“local”.Anadjustedp-valuethresholdof0.1wasusedtocalculatelog2
foldchangesbetweensurfacesoilsandcavemicrobialmats.
Phylogenetictreeanalysiswascarriedoutinthephyloseqpackage.Thedataweresubsetby
phylathatweredifferentiallyabundantasdeterminedbyDeSeq2.Treefileswithtipslabel
werewrittenoutusingwrite.tree(phy_tree(phyloseq_obj),file="phylum_name.newick").The
tree,tiplabels,andtraits(caveorsurface)wereloadedintoInteractiveTreeofLifev3(http://
itol.embl.de/)[42]forvisualization.Traitswereassignedtothetreetipsasrelativeabundance
oftheOTUspresentinthecave.
Totalbacterialrichnessforcavesampleswastestedusingahierarchalgeneralizedmixed
effectsmodelinRwiththestanarm[43]package.RichnessdatawerenormalizedinQIIME
beforemodelingusingnormalize_table.py-ilavabeds.biom-aDESeq2-oDESeq2_normalize-
d_otu_table.biom–DESeq_negatives_to_zero.Cave,colorofmat,visitation,andcavegeome-
trywereassignedasrandomeffects(levelsofhierarchy).Thefollowingoptionswereusedwith
stan_glmer:family=gaussian,dapt_delta=0.95,prior=normal(location=0,scale=8),
seed=SEED,iter=50000,cores=6.
Results
Waterchemistry
Chemicalanalysisofwatersamplescollectedfromeachcavewithavailablestandingwateris
showninTable3.Chlorinelevelsaveraged(5.086ppm),slightlyhigherthantheEPAMaxi-
mumContaminantLevelGoalsfordrinkingwaterof4ppm[44].EPAlevelsfornitrate
(1ppm)andnitrate(10ppm)werenotexceededinmostcavewatersamples.Thereareno
phosphatelevelstandardssetbytheEPA,butourcavesampleswerealllow.Sulfatewaterstan-
dardsare250ppm,andourswereallverylow.
Table3. WaterchemistryofcavewatersamplesfromLavaBedsNationalMonument. SeeTable1forthecavesampleabbreviations.Bromide(not
shown)allbelowdetectionlimits.
Cavesample Chloride(Clmg/L) Nitrite(NO mg/L) Nitrate(NO mg/L) Phosphate(PO mg/L) Sulfate(SO mg/L)
2 3 4 4
CAT7 12.571 nd 0.6039 nd 0.6103
CAT13 2.184 0.9866 0.5924 nd 0.3251
GD20 3.567 0.5526 0.7286 nd nd
GE26 9.203 nd 0.8325 nd 0.4704
GE10 5.128 0.3320 0.9140 nd 0.4329
HC3 2.116 nd 0.6606 0.6715 0.5391
HC10 1.821 nd 1.5608 nd 1.1231
L-V460-4 4.493 0.5778 2.7485 0.5371 0.5851
L-V460-12 2.031 nd 0.7116 0.5571 0.5505
S-L280-6 3.702 0.4489 8.0036 0.4697 2.3381
VAL3 9.132 0.3113 0.5483 nd 0.5910
nd=notdetected
doi:10.1371/journal.pone.0169339.t003
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 8/27
Surfacevs.lavacavebacterialmatcommunities
Surfacesoilversuscavealphadiversity
Measuresofalphadiversityamongsurfacesoilsamplesandcavesamplesbycolorofmat(Fig
3)showspeciesrichness(ObservedandChao1[45])andrelativeabundance(Shannonand
Simpson’s).Simpsonislessinfluencedbysingletons(i.e.raretaxa)thanShannon’sIndex,but
Shannon’sinlesssensitivetodifferenceinlibrarysize.Themeansareverydifferent,with
muchgreaterbacterialdiversityinsoilthancave.Soilbacterialdiversityismoreevenlydistrib-
utedamongOTUs.Cavesamplesarelessevenlydistributed,withawidedistributionofSimp-
son’svalueswithmanyoutliersincontrasttosoilsamples.In-cavevariationismuchhigher
thansurfacesoilvariation.Intermsofspeciesrichnessindifferentlycoloredmats,tansamples
havehigherdiversity,followedbyyellowandthenwhite.IntermsofShannon’sandSimpson’s
indices,thethreecolorsofmicrobialmatsarenotthatdifferent.Fig3showsthedifference
betweencavemicrobialmatsandsurfacesoilsinrespecttototalrichnessofeachsample
(observedOTUs)andShannonIndex.InallcasessurfacesoilshavemoreOTUsandhigher
Shannonindicesindicatinghigherdiversity.Rarefactioncurvescompareestimatedtotalrich-
nesspersampleatagivendepthandareincludedinthesupplementalinformation(S1Fig).
Therarefactioncurvesshowalinearrelationshipbetweensamplingdepthandspeciesrich-
ness.TheChao1estimatedrichnessandShannonleveloffafter1000and500,respectively.
Taxonomybyphylumandproteobacteriaclass
Compositionbythetopfivephylaandfourproteobacterialclasses,plusminorphylabasedon
OTUs,arepresentedinFig4.Comparisonsofsurfacesoilsampleswiththeunderlyingcave
samplesshowsomedifferences.MostnotableisthelowerpercentageofActinobacteriainsur-
facesoilsamples(21%)versuscavesamples(39%),thereductionintheNitrospirae insurface
(3%)vs.cavesamples(7%),andtheincreaseinAlphaproteobacteriainthesurfacesoilsamples
(17%)comparedtothecavesamples(10%).Smaller,butsignificantdifferencesthatdecrease
Fig3.Alphadiversityindices(richness,Chao1,andShannon).Boxplotsofsurfacesoilsandcavemicrobialmats
bycolor.Surfacesoilsamplesareinblueandcavesamplesinred.
doi:10.1371/journal.pone.0169339.g003
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 9/27
Surfacevs.lavacavebacterialmatcommunities
Fig4.StaggeredbarchartofrelativeabundanceofOTUs.Majorandminorphylabyallcavesamplesandallsurfacesamples.Cavesampleshavea
greaterrelativeabundanceofActinobacteriaandNitrospirae.
doi:10.1371/journal.pone.0169339.g004
fromsurfacetocaveareseenwiththeBacteroidetes(surface8%andcave2%);Gemmatimona-
detes(surface3%andcave<1%);andPlanctomycetes(surface2%andcave1%).TheGamma-
proteobacteriaincreasebetweensurface(18%)andcave(20%),whileBetaproteobacteria(surface
4%andcave3%)andDeltaproteobacteria(surface4%andcave1%)bothdeclineincavesam-
ples.Noneofthephyladifferedsignificantlybymatcolor.Somedifferenceswereobservedby
matcolorintheActinobacteria(tan:37%,white:39%,yellow:44%),Gammaproteobacteria(tan:
19%,white:24%,yellow:20%),Nitrospirae (tan:11%,white:4%,yellow:7%),andBetaproteo-
bacteria(tan:3%,white:6%,yellow:2%).
Molecularphylogenybyfamilyorgenus
ThepercentageofOTUsbygenusorfamilyispresentedinFig5forsurfacesoilandcavemat
samples.Resultsfromtheanalysisofdiversitybygeneraareinterestinginseveralrespects.
Firstofall,alargepercentageofOTUsarenovel.Oftheninemajorgenera,onlyfourareiden-
tifiedtogenuslevel,fourtofamily,andonetoclass.Thereisalsoagroupcalledunassigned,
whichcouldnotevenbeclassifiedtobacteria.TheseunassignedOTUsmayrepresentarchaeal
DNAthatamplifiedwithourprimersormaybeorganismsthatareparticularlydifficultto
classifythrough454sequencing,likemembersoftheVerrucomicrobia. Bergmannetal.[46]
usedbarcodedpyrosequencingwithsurfaceandsubsurfacesoilstoreduceprimerbias,and
foundthatVerrucomicrobia areubiquitousandwereoftenthedominantphylumintheirsam-
ples.Inourstudy,noindividualVerrucomicrobia genusmetourcriteriaforinclusioninFig5.
PLOSONE|DOI:10.1371/journal.pone.0169339 February15,2017 10/27