Characterisation of Arctic Bacterial Communities in the Air above Svalbard

22 Pages·2017·4.03 MB·English

biology
Article
Characterisation of Arctic Bacterial Communities in
the Air above Svalbard
LewisCuthbertson1,HerminiaAmores-Arrocha1,LucieA.Malard1,NoraEls2,BirgitSattler2
andDavidA.Pearce1,*
1 DepartmentofAppliedSciences,FacultyofHealthandLifeSciences,
UniversityofNorthumbriaatNewcastle,EllisonBuilding,Newcastle-upon-TyneNE18ST,UK;
[email protected](L.C.);[email protected](H.A.-A.);
[email protected](L.A.M.)
2 InstituteofEcology,AustrianPolarResearchInstitute,UniversityofInnsbruck,Technikerstrasse25,
6020Innsbruck,Austria;[email protected](N.E.);[email protected](B.S.)
* Correspondence:[email protected];Tel.:[phone]
AcademicEditor:ChrisO’Callaghan
Received:27December2016;Accepted:21April2017;Published:6May2017
Abstract: Atmosphericdispersalofbacteriaisincreasinglyacknowledgedasanimportantfactor
influencing bacterial community biodiversity, biogeography and bacteria–human interactions,
includingthoselinkedtohumanhealth.However,knowledgeaboutpatternsinmicrobialaerobiology
is still relatively scarce, and this can be attributed, in part, to a lack of consensus on appropriate
samplingandanalyticalmethodology. Inthisstudy,threedifferentmethodswereusedtoinvestigate
aerial biodiversity over Svalbard: impaction, membrane filtration and drop plates. Sites around
Svalbardwereselectedduetotheirrelativelyremotelocation,lowhumanpopulation,geographical
locationwithrespecttoairmovementandthetraditionandhistoryofscientificinvestigationonthe
archipelago,ensuringthepresenceofexistingresearchinfrastructure.Theaerialbacterialbiodiversity
foundwassimilartothatdescribedinotheraerobiologicalstudiesfrombothpolarandnon-polar
environments,withProteobacteria,Actinobacteria,andFirmicutesbeingthepredominantgroups.
TwelvedifferentphylaweredetectedintheaircollectedaboveSvalbard,althoughthediversitywas
considerablylowerthaninurbanenvironmentselsewhere. However,only58of196bacterialgenera
detected were consistently present, suggesting potentially higher levels of heterogeneity. Viable
bacteriawerepresentatallsamplinglocations,showingthatlivingbacteriaareubiquitousinthe
airaroundSvalbard. Samplinglocationinfluencedtheresultsobtained,asdidsamplingmethod.
Specifically,impactionwithaSartoriusMD8producedasignificantlyhighernumberofviablecolony
formingunits(CFUs)thandropplatesalone.
Keywords: aerobiology; bioaerosol; Arctic; polar; ecology; bacteria; marine; terrestrial; culture
dependent;cultureindependent
1. Introduction
Microbialdispersalintheatmosphererepresentsakeybiologicalinput,directlyinfluencingthe
genepool[1]. Thedispersalrateofbacteriaintheatmospherehasbeenshowntobedirectlylinkedto
weatherevents,suchasduststorms,thatliftlargeamountsofmicrobialmatterintotheatmosphere[2].
Therearetwomechanismsbywhichbacteriaaretransportedthroughtheatmosphere: freefloating
andattachedtolargerairborneobjects. Freefloatingbacteriaintheatmosphereareunlikelytocome
intocontactwithothermicroorganismsfrequently;however,bacteriaassociatedwithlargerairborne
particlescouldbesubjecttoincreasedhorizontalgeneflow[3]. Infact,itisthishorizontalgeneflow
Biology2017,6,29;doi:10.3390/biology6020029 www.mdpi.com/journal/biology
Biology2017,6,29 2of22
andtheabundanceofbacteriawithintheatmospherewhichhasdrawnattentiontotheenvironment
asapotentialsourcefornewantibiotics[4].
Whilstseveralstudieshavefocusedonthemovementofbacteriathroughtheatmosphere,the
majorityofthesestudieshavefailedtoconsidertheviabilityofthesecolonistsuponarrivalintheir
new environments [5]. Microbial matter can be transported through the atmosphere potentially
over global scale, allowing long distance colonization. A large number of bacteria also remain
viable in the atmosphere for extended periods of time, even under intense selection pressure [6].
These viable microorganisms carry out multiple functions whilst suspended in the atmosphere;
these include cloud formation by ice nucleation [7,8], nitrogen processing [9], the degradation of
organiccarbon-basedcompounds[10]andphotosynthesis[11]. Viablecolonistshavethepotentialto
interactwithmicrobiomesatthesiteofdepositioninanantagonisticorsynergisticway. Forexample,
suspended nitrifying bacteria that are deposited in nutrient poor locations could provide a novel
sourceofnutrientsbenefittingtheecosystem;conversely,thesamemechanismcanprovedisruptivein
othercircumstances,causingtoxicalgalblooms,whichcanbedevastating[12]. Migratingbacteriaalso
poseapotentialpathogenicthreattohumanhealth,globalecosystemstability[13–15]andagriculture
duetothehomogeneityofmoderndaycrops[16].
Atmosphericbacterialabundancegenerallyrangesfrom104to106cellsperm3[17],but,thisvaries
throughouttheyear[18],andcanbeaffectedbyweather(winddirection,windspeed,temperature,
etc.) [19]. Bacterial abundance can decrease by as much as half with increasing altitude, although
viable bacteria have been found in the stratosphere at altitudes as high as 7.7 km [2,20]. Bacteria
foundintheatmospherearediverse. Airbornebacterialassemblagesinbothterrestrialandmarine
environmentscontainmorethan150generaofbacteria[21–23],alevelofdiversitycomparabletoother
nutrientpoorenvironmentssuchasAntarcticsnow,whichhasbeenshowntocontainintheregionof
250generaofbacteria[24]. Barberánetal.[25]collatedover1000samplingeffortsandfoundmore
than110,000differentspeciesofairbornebacteriaintheUSAalone. Mostbacterialcommunitiesinthe
atmospherecomprisefourmainphyla: Actinobacteria,Bacteroidetes,Firmicutes,andProteobacteria,
afactthatremainsconsistentintheatmospheresurroundingbothmarineandterrestrialhabitats[23,26].
However,aerialmicrobialdiversityatgenuslevelismorevariableanddependsonenvironmental
conditions,suchasproximitytoagriculturalsites,meteorologicalconditionsandseason[18,22].
Patterns of diversity in airborne bacterial communities are central to the emerging field of
atmosphericbiogeography. Indeed,untilrelativelyrecentlywhethermicrobialbiogeographyexistedin
theatmosphereatallwascontentious[27]. However,anincreasingnumberofstudieshaveshownthe
inter-continentaldispersalofbacteriaacrosscontinentsseparatedbybothpolitical(EuropeandAsia)
and geographical (North America and Asia) borders [25,28]. Furthermore, distinct geographical
featuresgiverisetodistinctairbornemicrobialcommunities,forexamplemarinecoastalcommunities
aredifferenttocontinentalterrestrialones[25,29]. Despitethesefindings,atmosphericbiogeography
has received little attention as the atmosphere is considered a transport route rather than a stable
habitat [30]. The development of aerobiology as a field and improved techniques should help
understand whether at the ecological level, microbes interact and evolve within the atmosphere,
astheydoinotherhabitats.
TheArcticcanbedefinedastheareaabovetheArcticCircle. TheNorwegianArcticarchipelagoof
Svalbardisoneofthenorthernmostinhabitedlocationsintheworldat79◦N.Svalbardischaracterised
byitsremarkablylowhumanpopulationwithonly2185registeredSvalbardinhabitantsin2015[31].
Thislowpopulationdensitytranslatesintoreducedanthropogenicenvironmentalalterationssuchas
thoselinkedtoagriculture. Thus,theArcticrepresentsanoptimallocationtostudynaturalpatternsof
airbornedispersalanditsinfluenceshapingnaturalcommunities. AerobiologicalstudiesintheArctic
datebackasfarasthelate1940s[32]. Studiesofthisnaturearesparsebetweentheseearlyeffortsand
thepresent,withveryfewstudiestakingadvantageofnovelmoleculartechniques. Tothebestofour
knowledge,theonlyrecentterrestrialstudyofbioaerosols(airborneparticlesofbiologicalorigin)inthe
ArcticwascarriedoutbyHardingetal[33],onWardHuntIslandlocatedintheCanadianhighArctic.
Biology2017,6,29 3of22
Hardingetal. foundsimilaritiesbetweenairandsnowcommunitiesandthosebacterialcommunities
found in the surrounding Arctic Ocean, drawing the conclusion that local sources are the largest
contributorswhichinfluencebacterialcommunityassemblages. Theirstudyalsofoundorganisms
notnormallyassociatedwiththehighCanadianArctic,microbesfromotherArcticlocations,aswell
as some Antarctic microorganisms, supporting the theory of long distance atmospheric dispersal.
Thesefindingsareconsistentwiththoseofpreviousstudiesthathavestatedthedominantgroupsof
bacteriaincoldecosystemstobeProteobacteria(alpha,betaandgamma),Firmicutes,Bacteroidetes,
andActinobacteria[34,35]. However,whileaerobiologicalstudiesintheArcticarescarce,thenumber
ofstudiesintheAntarctichasincreased[1,36]. Tothisend,acomparativeanalysisofaerobiological
dataovertheArcticandtheAntarcticwillallowthestudyofbipolardiversityandpotentially,the
globalatmosphericdistributionofmicrobes.
OrganismsintheArcticatmosphereareexposedtoextremelylowtemperaturesandhurricane
strengthwinds, seasonalfreeze–thawcycles, extremeexposuretoUVandextremelylowlevelsof
nutrients. Thus,organismsinhabitingthisregionarereferredtoasextremophilesandtendtoexploit
featuressuchastheabilitytoformspores,whichallowthemtosurvivetheharshconditions. Similar
tothosemicrobesinhabitingtheArctic,organismssurvivingintheatmospherealsoendureextreme
temperatures,UVexposureandpoornutrientlevels.
Sampling techniques for terrestrial and aquatic microbial ecology studies are highly variable
but based on common principles, established and used consistently. In contrast, a wide range of
techniquesareavailabletoaerobiology, despitethelownumberofstudiesinthefield. Ingeneral,
samplingmethodsinvolveimpaction,impingement,membranefiltrationorthedropplatemechanism,
theresultsofwhicharenotdirectlycomparableduetostrongmethodologicalbiases. Furthermore,
thestrengthofthebiasisstillunknown,duetothelackofstudiescomparingdifferentmethodologies,
althoughrecenteffortshavebeenmadetowardsestablishingastandardmethodology[37].
Analyticaltechniquescanalsovaryconsiderablyamongstudies,compromisingcomparability
even further. To date, most aerobiological studies use colony-forming units (CFU) count per unit
volumeofairsampledtomeasurethedensityofcultivablemicroorganismsintheatmosphere. These
studies report density changes over space, time and varying environmental conditions; however,
culturebasedstudiesonlyprovideapartialpictureoftheoverallmicrobialdiversity[30]. Culture
dependentstudiesarealsobiasedtowardsGram-positivebacteria, whilemolecularbasedstudies
show the opposite trend, with a large proportion of Gram-negative bacteria populating the aerial
environment[38]. Forthisreason,fluorescencemicroscopyisincreasinglyusedforcellcountsand
taxonomicidentification,combinedwithmoleculartechniquessuchashighthroughputsequencing.
Temporal, spatial and meteorological variations also lead to differences in the aerial communities
identified[39,40],reducingfurthertheabilitytodescribebiogeographicalpatterns.
Setagainstthisbackground,inthisstudy,theinfluenceofdifferentsamplingtechniques,sampling
locationandtotalsamplevolumeontheidentificationofaerialbacterialcommunitiesintheArctic
wasexplored,basedonculturedependentandindependentanalyticalmethods,thuspresentinga
preliminarypictureofthemicrobialcommunityintheairoverSvalbard.
2. MaterialsandMethods
2.1. SiteDescription
AirbornemicrobialsampleswerecollectedinJuly2015aboveSvalbard(Figure1). Svalbardis
hometoarelativelysmallhumanpopulationandplayshosttoveryfewmammals. Themajorityofthe
humanpopulationofSvalbardresidesinLongyearbyen;implyingthat,weresamplessubjecttohuman
influence,itwouldmostlikelyoccurhere. ThewestcoastofSvalbardisinfluencedbytheAtlantic
OceanandisaffectedbywarmercurrentsthantheEastCoast,orientedtowardstheBarentsSea.
Biology2017,6,29 4of22
Biology 2017, 6, 29 4 of 21

Figure 1. Svalbard location and sampling sites (map adapted with courtesy of the © Norwegian Polar
Figure1.Svalbardlocationandsamplingsites(mapadaptedwithcourtesyofthe©NorwegianPolar
Institute (http://www.npolar.no/no/).
Institute(http://www.npolar.no/no/).
Samples were collected between 6 and 23 July 2015 above both marine and terrestrial locations
Sampleswerecollectedbetween6and23July2015abovebothmarineandterrestriallocations
using a range of techniques (Table 1). Marine samples were collected aboard the research ship (Viking
usingarangeoftechniques(Table1). Marinesampleswerecollectedaboardtheresearchship(Viking
Explorer) and aboard a zodiac. The terrestrial sites were on the roof of The University Center in
Explorer) and aboard a zodiac. The terrestrial sites were on the roof of The University Center in
Svalbard (UNIS (78°13′ N, 15°39′ E)) located in central Longyearbyen, Mine (Gruve) 7, Deltaneset,
Svalbard(UNIS(78◦13(cid:48) N,15◦39(cid:48) E))locatedincentralLongyearbyen,Mine(Gruve)7,Deltaneset,
Gipsdalen and Bjørndalen; these locations were chosen to represent a large terrestrial geographic
GipsdalenandBjørndalen;theselocationswerechosentorepresentalargeterrestrialgeographicrange.
range. The marine sites were located in the surrounding fjords at Billefjorden, Isfjorden,
ThemarinesiteswerelocatedinthesurroundingfjordsatBillefjorden,Isfjorden,Sassenfjordenand
Sassenfjorden and Adventfjorden bay (Figure 1).
Adventfjordenbay(Figure1).
Table 1. Summary of sample locations and regimes.
Table1.Summaryofsamplelocationsandregimes.
Sample Flow Rate
Environment Sampling Mechanism Date Duration (min)
Location (L m−1)
Sample FlowRate Duration
LocBajtøironndalen EnvTireorrnemstreianlt SampDlrionpg pMlateesc hanism 13 July 20D15a te - (Lm−1) 15 (min)
Deltaneset Terrestrial Impaction onto media 13 July 2015 50 20
Bjørndale n Terrest rial DDrorpo pplpatleast es 13 Jul1y3 2J0u1l5y 2015 - - 15 15
DeltaGnipessdeatlen TeTrreersretrsitarilal IImmppaactcitoino nonoton tmoemdiead ia 13 Jul1y3 2J0u1l5y 2015 50 50 20 20
DDrorpo pplpatleast es 13 Jul1y3 2J0u1l5y 2015 - - 15 15
GipLsodngayleenarbyen TeTrreersretrsitarilal IImmppaactcitoino nonoton tmoemdiead ia 16 Jul1y3 2J0u1l5y 2015 30, 50 50 20, 40, 60, 8200
MembDrraonpe fpilltraatteiosn 06, 19, 2113–2J3u Jluyly2 015 ~20 - 30, 60, 120, 300, 31 5days
Longyearbyen Terrestrial Impactionontomedia 201165 July2015 30,50 20,40,60,80
Mine (Gruve) 7 Terrestrial Impaction onto media 130 J6u,ly1 920,1251 –23July 50 3020, 60,120,300,
MemDrborpa pnleatfieslt ration 13 July 2015 - ~20 15
2015 3days
Adventfjorden Marine Impaction onto media 17 July 2015 50 20
Mine(Gruve)7 Terrestrial Impactionontomedia 13July2015 50 20
Billefjorden Marine Impaction onto media 17 July 2015 50 20
Dropplates 13July2015 - 15
Isfjorden Marine Membrane filtration 11 July 2015 ~20 480
Adventfjorden Marine Impactionontomedia 17July2015 50 20
Sassenfjorden Marine Impaction onto media 17 July 2015 50 20
Billefjorden Marine Impactionontomedia 17July2015 50 20
Isfjorden Marine Membranefiltration 11July2015 ~20 480
2.2S. aMsseetnefojorrodleongical DaMtaa rine Impactionontomedia 17July2015 50 20
Seven-day back trajectory models were calculated for sampling days where sequencing was
2.2. MeteorologicalData
carried out at air mass arrival heights of 10 m, 500 m and 1500 m (Figure 2) using National Oceanic
and SAetvmeno-sdpahyerbica cAkdtmraijencistotrraytimono d(NelOsAwAer)e Hcaylscpulliat tMedodfoerl s[a4m1] palnindg tdhaey Gslwobhaelr eDsaetqa uAenssciimngilawtiaosn
cSayrrsiteedmo (uGtDatAaSir1)m aarscshaivrreidv adlahteai gfihlet.s Ionf g1e0nmer,a5l0, 0pmockanetds 1o5f 0a0irm at( Failgl ualrteit2u)duessi nagrrNivaetdio fnraolmO ac enaonritchaenrdly
A(Atmrcotsicp OhecreiacnA) ddmireinctiisotrna,t hioonw(eNvOerA hAig)hH ayltsiptulidteM aoird pelo[c4k1e]tsa nadt 1t5h0e0G mlo wbaelrDe amtaorAe sesaimsteilralyti oinnflSuyesntecmed
(tGhDanA Sth1e) alrocwhievre daldtiatutadfiesle. .OInn g6e nJeurlayl,, pthoec kleotwso falatiirtuadtea llaairl timtuadsessesa r(r1iv0 emd,f ro50m0 amn)o rwtheerrel ye(aAstrecrtliyc.
OTceemapne)rdaitruercetsio anv,ehroagweedv 8e r°Chi gachroasltsi taulld seaamirpplioncgk deatsysa tw1i5th00 omnlyw oenree pmreocriepeitaasttieornly evineflnut teontcaelldintgh a0n.1
tmhemlo owcceurrarlitnitgu dones 1.7O Jnul6yJ. uWlyi,ntdh esploeweda vltaitruiedde baeirtwmeaesns e1s0( a1n0dm 2,25 0k0mmh-)1 wanedre heuamstiedrliyty. Taevmerpaegreadtu 6r7e%s
a(vTearbalgee 2d).8 ◦Cacrossallsamplingdayswithonlyoneprecipitationeventtotalling0.1mmoccurring
on17July. Windspeedvariedbetween10and22kmh−1andhumidityaveraged67%(Table2).
Biology2017,6,29 5of22
Biology 2017, 6, 29 5 of 21

FFiigguurree 22.. BBaacckk ttrraajjeeccttoorryy mmooddeellss wweerree ccaallccuullaatteedd uussiinngg tthhee NNOOAAAA HHyysspplliitt MMooddeell [[4411]].. TThhrreeee aarrrriivvaall
hheeiigghhttss wweerree uusseedd 1100 mm ((ttrraannsseecctt mmaarrkkeedd bbyy ttrriiaanngglleess)),, 550000 mm ((ttrraannsseecctt mmaarrkkeedd wwiitthh ssqquuaarreess)) aanndd
11550000 mm ((ttrraannsseeccttss mmaarrkkeedd wwiitthh cciirrcclleess)).. SSaammpplliinngg llooccaattiioonn iiss mmaarrkkeedd bbyy aa bbllaacckk ssttaarr..
Table 2. Meteorological conditions on sampling days at Svalbard airport (The Weather Company
(Atlanta, GA, USA)).
21–23 July
06 July 11 July 13 July 16 July 17 July 19 July 21 July 22 July 23 July Average across All
2015
2015 2015 2015 2015 2015 2015 2015 2015 2015 Sampling Days
(Average)
Average
temperature 8 10 8 6 8 8 10 8 6 8 8
(°C)
Total
precipitation 0 0 0 0 0.1 0 0 0 0 0 0
(mm)
Average wind
13 20 10 20 14 22 18 12 12 14 16
speed (kmh-1)
Biology2017,6,29 6of22
Table2.MeteorologicalconditionsonsamplingdaysatSvalbardairport(TheWeatherCompany(Atlanta,GA,USA)).
21–23July2015 AverageacrossAll
06July2015 11July2015 13July2015 16July2015 17July2015 19July2015 21July2015 22July2015 23July2015
(Average) SamplingDays
Averagetemperature(◦C) 8 10 8 6 8 8 10 8 6 8 8
Totalprecipitation(mm) 0 0 0 0 0.1 0 0 0 0 0 0
Averagewindspeed(kmh−1) 13 20 10 20 14 22 18 12 12 14 16
Averagehumidity(%) 63 75 90 57 68 59 65 68 61 65 67
Pressure(hPa) 1025 1022 1019 1009 1013 1015 1015 1013 1006 1012 1016
Biology2017,6,29 7of22
2.3. CultureDependent
DropplatescontainingR2Amedia(Sigma-Aldrich,St. Louis,MO,USA)wereplacedopenat
Gipsdalen,Mine(Gruve)7,DeltanesetandBjørndalenfor15min;plateswereincubatedfor10daysat
roomtemperature;followingincubationtheplateshadcolonycountsanddistinctcolonycountstaken.
Additionally,aportableAirPortMD8(Sartorius,Göttingen,Germany),comprisingadisposable
gelatine filter membrane, was used to compare sampling efficiency and cultivability at two flow
ratesanddifferentsamplingvolumes. Samplingsiteswerechosentocomparewithterrestrialplate
dropsitesbutalsotoassessforthedifferencesatmarinesites. Terrestrialsampleswerecollectedat
Mine(Gruve)7,Deltaneset,GipsdalenandcentralLongyearbyen(UNISroof)andmarinesamplesat
Billefjorden,SassenfjordenandAdventfjorden,respectively. Thesamplerwasusedatrespectiveflow
ratesanddurationsranging30–50Lm−1and20–80Lm−1on13,15,16and17July2015. Thegelatine
filterscollectedatallsiteswereplaceddirectlyontothesurfaceofR2Aagarplates(Sigma-Aldrich,
St.Louis,MO,USA).Theseplateswerethenincubatedatroomtemperaturefor10days. TotalCFU
anddistinctcolonynumberswerecounted.
2.4. CultureIndependent
As gelatine filters are not amenable to culture independent techniques (due to the presence
of gelatine), airborne bacteria from both terrestrial and marine sites were collected via membrane
filtration. AWelchWOB-Lvacuumpump(Welch,Mt. Prospect,IL,USA)wassetupataflowrateof
~20Lm−1connectedtoSartoriusfiltrationunit(Göttingen,Germany)containinga47mm×0.2µm
poresizecellulosenitratemembranefilter(GEHealthcareLifeSciences,Chicago,IL,USA).
AmarinesamplewascollectedatIsfjordenonthe11July2015witharespectivesampleduration
andvolumeof8hand~9600LandtheterrestrialsamplewastakenincentralLongeyearbyen(UNIS
roof)atthefollowingdates,durationsandvolumes,respectively: 6July2015for30(600L),60(1200L),
120(2400L)and300(6000L)min;19July2015for30(600L),60(1200L),120(2400L)and300(6000L)
min;and21–24July2015forthreedays(~86,000L)continuously(Table1).
ThecellulosenitratemembranefiltersweresenttoMrDNA(MrDRNA,Shallowater,TX,USA)for
extractionandsequencing. DNAwasextractedfromsamplesusingtheMoBioPowerSoilkit(MoBio,
Vancouver, BC, Canada) following the manufacturer’s protocol with an additional 1 min of bead
beatingtoaccountforthefilterpaper.Extractedsampleswerethenamplifiedusing16SrRNAuniversal
primers27Fmod(AGRGTTTGATCMTGGCTCAG)and519Rmodbio(GWATTACCGCGGCKGCTG)
andbarcodeswereattachedatthe5(cid:48) end. A28-cyclePCRusingtheHotStarTaqPlusMasterMixKit
(Qiagen,Germantown,MD,USA)wascarriedoutunderthefollowingconditions: 94◦Cfor3min,
followedby28cyclesof94◦Cfor30s,53◦Cfor40sand72◦Cfor1min,afterwhichafinalelongation
stepat72◦Cfor5minwasperformed. Afteramplification,PCRproductswerecheckedin2%agarose
geltodetermineamplificationsuccess. Sampleswerethenpooledbasedontheirmolecularweightand
DNAconcentrations,purifiedandilluminaDNAlibrarieswereprepared. Pairedendsequencingof
theV4regionwasthenperformedonaMiSeqfollowingthemanufacturer’sguidelines. Theresultant
datawereanalysedusingQIIMEv1.9.1[42]. The776,315rawsequencereadswerequalitytrimmed
andcheckedforchimerasusingUSEARCH6.1[43], clusteredatanidentitythresholdof97%and
assignedtoOperationalTaxonomicUnits(OTUs)usingUCLUST[43]andtheGreengenesreference
database[44]wasusedtoassigntaxonomy. SequenceswerethenalignedusingPyNAST[45]anda
phylogenetictreewasbuiltusingFastTree[46].
2.5. StatisticalAnalysis
StatisticalanalyseswereperformedusingPAST[47]totestfordifferencesinmeans,medians,
variancesanddistributionsandMSExcel(2013)tocalculatecorrelationcoefficients,thecoefficientof
varianceandproducegraphsoftheanalyses;statisticaltestswerecarriedoutatanassumedsignificance
of alpha: 0.05. When calculating diversity indices, to avoid statistical bias due to differences in
Biology 2017, 6, 29 7 of 21
2.5. Statistical Analysis
Statistical analyses were performed using PAST [47] to test for differences in means, medians,
variances and distributions and MS Excel (2013) to calculate correlation coefficients, the coefficient of
Biology2017,6,29 8of22
variance and produce graphs of the analyses; statistical tests were carried out at an assumed
significance of alpha: 0.05. When calculating diversity indices, to avoid statistical bias due to
dseiqffueerenncicnegs idne psethquaellnscaimngp dleespwthe raelln soarmmpalleisse wdetorea ndoerpmthaloisfe2d6 ,t1o9 0a rdeeapdtsh. Roaf r2e6f,a1c9t0io rneacudrsv. eRsa,rdeifvaecrtsioitny
cinudrviceess, (dSihvaenrnsiotyn ainnddicSeims p(Sshoannsnroecni parnodca Sl)i,mBprasyo-nCsu rreticsipOrTocUala),n dBruanyw–Ceiugrhtitse dOUTnUi Farnacd puhnywloegiegnhetetidc
UdinstiFanracce pmheytlroicgse,naentdic PdCisotAanscwe mereetprircosd, uancedd PuCsoinAgs QwIeIMre Epr[4o2d]u.ced using QIIME [42].
33.. RReessuullttss
33..11.. CCuullttuurree DDeeppeennddeenntt
VViiaabbllee bbaacctteerriiaaw weereref ofuonudndin inal laollf othf ethsaem sapmlesp.leCsl.e Carledairff edriefnfecreesnwceesr ewaeprpea arepnptairnentht einm tehaen mCFeUans
CfrFomUst hfreotmw othceu lttwuroe cduelptuenred ednetpmenedtheondt smuestehdo(dFsi guusreed3 )(.FAiguKrreu s3k).a lA–W Karlulissktaels–tWfoarleliqsu taelsmt feodri aenqsuoafl
mCFeUdisaannsd omf oCrFpUhosl oagnidca lmlyodrpishtionlcotgCicFaUllsy wdaisstuinncdt erCtaFkUesn wtoaass suenssdethrteadkreonp tpol aatessreespsl itchaete sd,rtohpe rpelsautlet
rdeipdlnicoatteshs,o twhes irgensiufilcta dnitdd nifofetr eshnocews (sKigrunisfkicaal–nWt dalilfifseprelantceefsa l(lKCruFUsk:apl–=W0.a0l9li5s, ppllaattee ffaallllm CoFrUph: op lo= g0i.c0a9ll5y,
pdliasttien fcatlCl mFUorsp:hpo=lo0g.1ic2a3l)l.y distinct CFUs: p = 0.123).

Figure 3. Mean colony-forming units (CFU) and morphologically distinct CFU counts for drop plate
Figure3.Meancolony-formingunits(CFU)andmorphologicallydistinctCFUcountsfordropplate
and Sartorius MD8 data.
andSartoriusMD8data.
Comparing the differences in drop plate and MD8 results for the locations where data were
Comparing the differences in drop plate and MD8 results for the locations where data were
available for both methods, the MD8 showed much higher CFU yields, although this difference is not
availableforbothmethods, theMD8showedmuchhigherCFUyields, althoughthisdifferenceis
obvious when looking at the number of morphologically distinct CFUs i.e. CFUs of different
not obvious when looking at the number of morphologically distinct CFUs i.e., CFUs of different
appearance (Figure 3). Statistical analyses show a significant difference of mean CFUs sampled at the
appearance(Figure3). StatisticalanalysesshowasignificantdifferenceofmeanCFUssampledatthe
same location using different methods (p < 0.05); no differences in variances, medians or coefficient
samelocationusingdifferentmethods(p<0.05);nodifferencesinvariances,mediansorcoefficient
of variations, but a significant difference in equality of distributions (Kolmogov–Smirnov: p < 0.05).
ofvariations,butasignificantdifferenceinequalityofdistributions(Kolmogov–Smirnov: p<0.05).
For the morphologically distinct CFUs, however, there were no significant differences for any of the
ForthemorphologicallydistinctCFUs,however,therewerenosignificantdifferencesforanyofthe
mentioned parameters. When looking at the overall variance and efficiency of both culture dependent
mentionedparameters. Whenlookingattheoverallvarianceandefficiencyofbothculturedependent
methods,onlyconsideringtheMD8samplescollectedat50Lm−1 for20mini.e.,1000Lsampling
volume (Figure 4), there were obvious differences in the mean CFUs, but not for morphologically
distinctCFUs. Anindependentt-testcomparingthetwomethodsshowedasignificantdifferencein
Biology 2017, 6, 29 8 of 21
methods, only considering the MD8 samples collected at 50 L m−1 for 20 min i.e., 1000 L sampling
Biology2017,6,29 9of22
volume (Figure 4), there were obvious differences in the mean CFUs, but not for morphologically
distinct CFUs. An independent t-test comparing the two methods showed a significant difference in
tthhee mmeeaann CCFFUUss ffrroomm ddrroopp ppllaattee aanndd MMDD88 ssaammpplleess ((pp << 00..000011)),, nnoo ssiiggnniiffiiccaanntt ddiiffffeerreenncceess iinn vvaarriiaanncceess,,
bbuutt wwiitthh ssiiggnniiffiiccaanntt ddiiffffeerreenncceess iinn ccooeeffffiicciieennttss ooff vvaarriiaattiioonn ((pp << 00.0.00055),) ,mmedediainans s(M(Manann–nW–Whihtnitenye yUU p
p≤≤ 0.00.0010)1 )anandd ddisitsrtirbibuutitoionns s(p(p << 00.0.00011).) .LLooookkiningg aat ttthhee ssttaattiissttiiccaall aannaallyyssiiss ooff tthhee mmoorrpphhoollooggiiccaallllyy
ddiissttiinncctt CCFFUUss,,t htheererew wasasn onosi gsinginfiicfaicnatndti fdfeifrfeenrceencine tihne tmhee amnesafnrosm frodmro pdprolapt epslaanteds tahnedM tDhe8 M(pD>80 .(0p5 )>,
w0.i0t5h),n woistihg nniofi sciagnntidfiicfafenrte dnicfefesriennvceasr iiann vceasr,iamnecdesia, nmse,ddiiastnrsib, duitsiotrnibs,uotirocnose, fofirc iceonetfsfiocfievnatrsi aotfi ovna.riTahtieosne.
rTehsuesltes rsehsouwltst hshaotwth etrheaits thaesrieg nisifi ac saingtndifiifcfearnetn dcieffbeertewnceee nbethtweetweno tmhee tthwood sm,tehtheoMdDs, 8thyei eMldDin8g yaielladrginegr
nau lamrgbeerr nofuCmFbUers .of CFUs.
35
30
25
s
U
F
C 20
f
o
r
e
b 15
m
u
N
10
5
0
Drop Plate (15 min exposure time) MD8 (50 L m-1, 20 min (i.e. 1000 L))
CFUs (mean) Morphologically distinct CFUs (mean)

Figure 4. Mean CFUs and mean morphologically distinct CFUs for drop plates and 1000 L MD8
Figure 4. Mean CFUs and mean morphologically distinct CFUs for drop plates and 1000 L
samples.
MD8samples.
Comparing MD8 results for terrestrial and marine samples, the mean CFUs were lower at marine
ComparingMD8resultsforterrestrialandmarinesamples,themeanCFUswereloweratmarine
sites than terrestrial sites (Figure 5). Statistical analysis showed no significant difference in CFUs for
sitesthanterrestrialsites(Figure5). StatisticalanalysisshowednosignificantdifferenceinCFUsfor
marine and terrestrial sites (p = 0.070). This also held for the morphologically distinct CFUs, there was
marineandterrestrialsites(p=0.070). ThisalsoheldforthemorphologicallydistinctCFUs,therewas
no significant difference for either of the mentioned parameters between terrestrial and marine sites.
nosignificantdifferenceforeitherofthementionedparametersbetweenterrestrialandmarinesites.
To account for the variable scales across the samples, the normalised coefficient of variation was
Toaccountforthevariablescalesacrossthesamples,thenormalisedcoefficientofvariationwas
used, and this showed the highest variation in the CFUs in the drop plate and MD8 samples collected
used,andthisshowedthehighestvariationintheCFUsinthedropplateandMD8samplescollected
at UNIS. In the MD8 marine sample the variability of morphologically distinct CFUs was the highest.
atUNIS.IntheMD8marinesamplethevariabilityofmorphologicallydistinctCFUswasthehighest.
The CFUs at the terrestrial sites varied the least (Figure 6).
TheCFUsattheterrestrialsitesvariedtheleast(Figure6).
At UNIS, where volume and flow rate were varied, a clear trend of increasing CFUs with
AtUNIS,wherevolumeandflowratewerevaried,acleartrendofincreasingCFUswithincreasing
increasing volume was evident (Figure 7). The 30 L m−1 flow rate sample, however, had slightly
volumewasevident(Figure7). The30Lm−1flowratesample,however,hadslightlyhigherCFUs,
higher CFUs, despite lower volume. There was a clear correlation between CFUs and the sampled
despitelowervolume. TherewasaclearcorrelationbetweenCFUsandthesampledvolumeofair
volume of air (R² = 0.933, Figure 8A), not including the 30 L m−1 sample, and still a very high positive
(R2=0.933,Figure8A),notincludingthe30Lm−1sample,andstillaveryhighpositivecorrelationof
correlation of (R² = 0.906) when this sample was included. For the morphologically distinct CFUs,
(R2=0.906)whenthissamplewasincluded. ForthemorphologicallydistinctCFUs,therewasaslight
there was a slight negative correlation (R² = −0.256) in number of CFUs with increasing volume
negativecorrelation(R2=−0.256)innumberofCFUswithincreasingvolume(Figure8B),leavingout
(Figure 8B), leaving out the exceptional value of 30 L m−1 showed a considerable negative correlation
theexceptionalvalueof30Lm−1showedaconsiderablenegativecorrelation(R2=−0.640).
(R² = −0.640).
Biology2017,6,29 10of22
Biology 2017, 6, 29 9 of 21
Biology 2017, 6, 29 9 of 21
35
3350
sU3205
F
C
sU f2250
FCo r
feb2105
om
rebuN1150
m
u
N 105
50
0 tesenatleDtesenatleD neladspiGneladspiG 7 )evurG( en7 )evurG( eniM SINUSINU nedrojftnevdnedrojftnevdA nedrojfelliBnedrojfelliB nedrojfnessanedrojfnessaS
TerrestrialiM A Marine S
CFTUersrestriaMl orphologically distinct CFUs Marine

CFUs Morphologically distinct CFUs
FFiigguurree 55.. CCoouunnttss oofft tootatall( b(blalacckk))a annddm moroprhpohloolgoigciaclalyllyd idstiisnticntc(tg r(geyre)yC)F CUFsUins iena cehacsham sapmlepsleep saerpaaterda tbeyd
by environment (terrestrial and marine).
environment(terrestrialandmarine).
Figure 5. Counts of total (black) and morphologically distinct (grey) CFUs in each sample separated
by environment (terrestrial and marine).
0.35
n
o
it0.03.53
a
noira
itV 0.3
a 0.25
ir fo
aV tn0.25
fe 0.2
o tnicif
eicfeo 00.1.25
iffeoC de0.01.51
Cz
dila
em 0.1
z 0.05
ilamroN
0.05
r 0
o
N
Drop Plate MD8 MD8 MD8 Marine MD8 UNIS
0
Terrestrial
Drop Plate MD8 MD8 MD8 Marine MD8 UNIS
Method
Terrestrial
Method
CFUs Morphologically distinct CFUs

CFUs Morphologically distinct CFUs
Figure 6. Normalised coefficient of variation (a ratio of mean and standard deviation without unit, to
compare different scales, here normalised to account for small sample size).
FFiigguurree 66.. NNoorrmmaalliisseedd ccooeeffffiicciieenntt ooff vvaarriaiatitoionn (a(a rartaitoi oofo mfmeaena nanadn dstastnadnadradr ddedveiavtiiaotnio wniwthiothuot uutnuitn, tito,
tcoomcopmaprea rdeifdfeifrfeenret nstcaslceasl,e hs,ehree rneonromramliasleidse tdo taocacoccuonutn ftorfo srmsmalla sllasmamplpe lseizseiz).e ).