Host-derived population genomics data provides …Host-derived population genomics data provides...

49
1 Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper 1 *, Jaelle C. Brealey 1 *, Tom van der Valk 1 , Antton Alberdi 2 , John W. Durban 3 , Holly Fearnbach 4 , Kelly M. Robertson 3 , Robin W. Baird 5 , M. Bradley Hanson 6 , Paul Wade 7 , M. Thomas P. Gilbert 2,8 , Phillip A. Morin 3 , Jochen B.W. Wolf 9,10 , Andrew D. Foote 11 **, Katerina Guschanski 1 ** *These authors contributed equally to this work. **These authors contributed equally to this work and are co-senior authors. 1 Animal Ecology, Department of Ecology and Genetics, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18D, 752 36, Uppsala, Sweden 2 Centre for GeoGenetics, Natural History Museum of Denmark, University of Copenhagen, Øster Volgade 5-7, Copenhagen K 1350, Denmark 3 Marine Mammal and Turtle Division, Southwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric Administration, 8901 La Jolla Shores Dr., La Jolla, CA 92037, U.S.A. 4 SR3, SeaLife Response, Rehabilitation, and Research P.O. Box 1404 Mukilteo, WA 98275, USA 5 Cascadia Research, 4th Avenue, Olympia, Washington 98501, USA 6 Northwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric Administration, 2725 Montlake Boulevard East, Seattle, Washington 98112, USA 7 National Marine Mammal Laboratory, Alaska Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric Administration, 7600 Sand Point Way NE, Seattle, Washington 98115, USA 8 NTNU University Museum, N-7491 Trondheim, Norway 9 Science of Life Laboratories and Department of Evolutionary Biology, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18D, Uppsala SE-752 36, Sweden 10 Section of Evolutionary Biology, Faculty of Biology, LMU Munich, Großhaderner Strasse 2, Planegg-Martinsried 82152, Germany 11 Molecular Ecology and Fisheries Genetics Laboratory, School of Biological Sciences, Bangor University, Bangor, Gwynedd, LL57 2UW, UK Correspondence Rebecca Hooper, Tremough House, University of Exeter Cornwall Campus, Penryn, Cornwall, TR10 9FE, UK. Email: [email protected] Jaelle Brealey, Animal Ecology, Department of Ecology and Genetics, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18D, 752 36, Uppsala, Sweden. Email: [email protected] . CC-BY-NC-ND 4.0 International license not certified by peer review) is the author/funder. It is made available under a The copyright holder for this preprint (which was this version posted July 26, 2018. . https://doi.org/10.1101/282038 doi: bioRxiv preprint

Transcript of Host-derived population genomics data provides …Host-derived population genomics data provides...

Page 1: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

1

Host-derived population genomics data provides insights into bacterial and diatom

compositionofthekillerwhaleskin

RebeccaHooper1*,JaelleC.Brealey1*,TomvanderValk1,AnttonAlberdi2,JohnW.Durban3,

HollyFearnbach4,KellyM.Robertson3,RobinW.Baird5,M.BradleyHanson6,PaulWade7,M.

ThomasP.Gilbert2,8,PhillipA.Morin3,JochenB.W.Wolf9,10,AndrewD.Foote11**,Katerina

Guschanski1**

*Theseauthorscontributedequallytothiswork.

**Theseauthorscontributedequallytothisworkandareco-seniorauthors.

1AnimalEcology,DepartmentofEcologyandGenetics,EvolutionaryBiologyCentre,UppsalaUniversity,Norbyvägen18D,75236,Uppsala,Sweden2Centre forGeoGenetics,NaturalHistoryMuseumofDenmark,UniversityofCopenhagen,ØsterVolgade5-7,CopenhagenK1350,Denmark3MarineMammalandTurtleDivision,SouthwestFisheriesScienceCenter,NationalMarineFisheriesService,NationalOceanicandAtmosphericAdministration,8901LaJollaShoresDr.,LaJolla,CA92037,U.S.A.4SR3,SeaLifeResponse,Rehabilitation,andResearchP.O.Box1404Mukilteo,WA98275,USA5CascadiaResearch,4thAvenue,Olympia,Washington98501,USA6NorthwestFisheriesScienceCenter,NationalMarineFisheriesService,NationalOceanicandAtmosphericAdministration,2725MontlakeBoulevardEast,Seattle,Washington98112,USA7National Marine Mammal Laboratory, Alaska Fisheries Science Center, National MarineFisheriesService,NationalOceanicandAtmosphericAdministration,7600SandPointWayNE,Seattle,Washington98115,USA8NTNUUniversityMuseum,N-7491Trondheim,Norway9ScienceofLifeLaboratoriesandDepartmentofEvolutionaryBiology,EvolutionaryBiologyCentre,UppsalaUniversity,Norbyvägen18D,UppsalaSE-75236,Sweden10SectionofEvolutionaryBiology,FacultyofBiology,LMUMunich,GroßhadernerStrasse2,Planegg-Martinsried82152,Germany11MolecularEcologyandFisheriesGeneticsLaboratory,SchoolofBiologicalSciences,BangorUniversity,Bangor,Gwynedd,LL572UW,UKCorrespondenceRebeccaHooper,TremoughHouse,UniversityofExeterCornwallCampus,Penryn,Cornwall,TR109FE,UK.Email:[email protected],AnimalEcology,DepartmentofEcologyandGenetics,EvolutionaryBiologyCentre,UppsalaUniversity,Norbyvägen18D,75236,Uppsala,Sweden.Email:[email protected]

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 2: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

2

AndrewFoote,MolecularEcologyandFisheriesGeneticsLaboratory,SchoolofBiologicalSciences,BangorUniversity,Bangor,Gwynedd,LL572UW,UK.Email:[email protected],AnimalEcology,DepartmentofEcologyandGenetics,EvolutionaryBiologyCentre,UppsalaUniversity,Norbyvägen18D,75236,Uppsala,Sweden.Email:[email protected]

Abstract

Recentexplorationintotheinteractionsandrelationshipbetweenhostsandtheirmicrobiota

hasrevealedaconnectionbetweenmanyaspectsofthehost’sbiology,healthandassociated

microorganisms.Whereasampliconsequencinghastraditionallybeenusedtocharacterise

themicrobiome,theincreasingnumberofpublishedpopulationgenomicsdatasetsofferan

underexploitedopportunity to studymicrobial profiles from thehost shotgun sequencing

data.Here,weusesequencedataoriginallygeneratedfromkillerwhaleOrcinusorca skin

biopsiesforpopulationgenomics,tocharacterisetheskinmicrobiomeandinvestigatehow

hostsocialandgeographicfactors influencethemicrobialcommunitycomposition.Having

identified 845microbial taxa from 2.4million reads that did notmap to the killerwhale

referencegenome,wefoundthatbothecotypicandgeographicfactorsinfluencecommunity

compositionofkillerwhaleskinmicrobiomes.Furthermore,weuncoveredkeytaxathatdrive

themicrobiome community composition and showed that they are embedded in unique

networks, one ofwhich is tentatively linked to diatompresence andpoor skin condition.

CommunitycompositiondifferedbetweenAntarctickillerwhaleswithandwithoutdiatom

coverage, suggesting that the previously reported episodic migrations of Antarctic killer

whalestowarmerwatersassociatedwithskinturnovermaycontroltheeffectsofpotentially

pathogenic bacteria such as Tenacibaculum dicentrarchi. Our work demonstrates the

feasibility of microbiome studies from host shotgun sequencing data and highlights the

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 3: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

3

importanceofmetagenomicsinunderstandingtherelationshipbetweenhostandmicrobial

ecology.

KEYWORDS

Cetacea,Contamination,Metagenomics,Microbiota,Orcinusorca

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 4: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

4

1|INTRODUCTION

Theskinmicrobiomeisanecosystemcomprisedoftrillionsofmicrobessculptedbyecological

andevolutionary forces actingonboth themicrobes and their host (McFall-Ngai,&Eliot,

2005; Byrd et al., 2018). Recent explorations have revealed a tight connection between

almosteveryaspectofthehost’sbiologyandtheassociatedmicrobialcommunity(Reviewed

byMcFall-Ngai,&Eliot,2005;Bordenstein,&Theis,2015;Alberdietal.,2016;Koskellaetal.,

2017;Byrdetal.,2018).Althoughnumerousintrinsicandextrinsicfactorsthatinfluencethe

skinmicrobiomecompositionhavebeenidentified,therelativeimportanceofthesefactors

often appear to differ even between closely related host taxa (Kueneman et al., 2014;

McKenzie,Bowers,Fierer,Knight,&Lauber,2012;Wolzetal.,2017).Intrinsically,thehost’s

evolutionary history, age, sex, and health appear significant (Leyden, McGiley, Mills, &

Kligman,1975;Cho&Blaser,2012;McKenzieetal.,2012;Phillipsetal.,2012;Apprilletal.,

2014;Yingetal.,2015;Chngetal.,2016).Extrinsically,bothenvironmentalfactors,wherea

sub-selectionofenvironmentalmicrobescolonisehostskin(Apprilletal.,2014;Wolzetal.,

2017;Walkeetal.,2014;Yingetal.,2015),andsocioecologicalfactors,suchasahost’ssocial

groupandthelevelofinteractionwithconspecifics(Jain,2011;Laxetal.,2014;Songetal.,

2013;Kolodnyetal.2017),canplayimportantroles.

Mostmicrobiomestudiestodatearebasedon16SribosomalRNAgenesequences,ahighly

conservedregionofthebacterialandarchaealgenome(Hamady&Knight,2009).However,

inaddition topotentialbiases inPCRamplification, inwhich lowreliabilityofquantitative

estimations ariseduemismatches inprimerbinding sites, PCR stochasticity, anddifferent

numbersof16Sgenecopiesineachbacterialspecies(Alberdietal.,2018),analysisofthe16S

regioncanlimitfunctionalandtaxonomicclassification(Quince,Walker,Simpson,Loman,&

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 5: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

5

Segata, 2017). In contrast, shotgun metagenomics can facilitate both high-resolution

taxonomicandfunctionalanalyses(Ranjan,Rani,Metwally,McGee,&Perkins,2016;Quince,

Walker,Simpson,Loman,&Segata,2017;Koskellaetal.,2017).Theadventofaffordablehigh-

throughputsequencinghasseenanever-increasingnumberofpopulationgenomicsstudies

inawiderangeofstudysystems(e.g.Jonesetal.,2012;Poelstraetal.,2014;DerSarkissian

et al., 2015; Nater et al., 2017). This affords an unprecedented opportunity to exploit

sequencingdata to secondarily investigate themicrobial communitiesassociatedwith the

sampled tissue of their host (Salzberg et al., 2005; Ames et al., 2015; Zhang et al., 2015;

Manguletal.,2016;Lassalleetal.,2018).

Here, we explore the relative importance of extrinsic factors on the epidermal skin

microbiome of free-ranging killer whales (Orcinus orca) using shotgun sequencing data

derived from skin biopsy samples of five ecologically specialisedpopulations, or ecotypes

(Foote et al., 2016). Given thewidespread geographic range (Forney&Wade, 2007) and

variationinecologicalspecialisationofkillerwhales,eveninsympatry(Durbanetal.,2017;

Fordetal.,1998),thisspeciesprovidesagoodstudysystemforexploringtheeffectsofboth

geographiclocationandecotype(aproxyforbothsocialityandphylogenetichistory)onthe

skinmicrobiome.However,theopportunisticuseofsuchdataisalsofraughtwithpotential

pitfalls.Wethereforedescribeindetailmeasurestakentodisentanglepotentialsourcesof

contaminationfromthetrueskinmicrobiome,thusprovidingauseful roadmapfor future

host-microbiomestudiesthatexploithost-derivedshotgunsequencingdata.

2|MATERIALSANDMETHODS

2.1|Studysystem

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 6: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

6

ThroughoutmostofthecoastalwatersoftheNorthPacifictwoecotypesofkillerwhalesare

foundinsympatry:themammal-eating‘transient’andfish-eating‘resident’ecotype(Fordet

al.,1998;Saulitisetal.,2000;Matkinetal.,2007;Filatovaetal.,2015).Fourdecadesoffield

studieshave found that theyare socially andgenetically isolated (Hoelzel&Dover,1991;

Barrett-Lennard,2000;Hoelzeletal.,2007;Ford,2009;Parsonsetal.,2013;Filatovaetal.,

2015;Morinetal.,2015;Foote&Morin,2016).Killerwhaleshavealsodiversifiedintoseveral

ecotypesinthewatersaroundtheAntarcticcontinent,includingaformcommonlyobserved

hunting seals in the pack-ice of the Antarctic peninsula (type B1), a form that feeds on

penguinsinthecoastalwatersoftheAntarcticpeninsula(typeB2),andadwarfformthought

toprimarilyfeedonfishinthedensepack-iceoftheRossSea(typeC)(Pitman&Ensor,2003;

Pitman&Durban,2010,2012;Durbanetal.,2017;Pitmanetal.2018).

2.2|Samplecollectionanddatageneration

Weusedtheunmappedreadsfromapublishedpopulationgenomicsstudyofkillerwhale

ecotypes(EuropeanNucleotideArchive,www.ebi.ac.uk/ena,accessionnumbers:ERS554424-

ERS554471;Footeetal.2016),whichproducedlowcoveragegenomesfromatotalof49wild

killerwhales,correspondingtofiveecotypes;10sampleseachoftheNorthPacificfish-eating

resident and sympatric mammal-eating transient ecotypes, and 8, 11 and 10 samples,

respectivelyfromAntarctictypesB1,B2andC(seeFigure1forthesamplinglocations).DNA

wasextractedfromepidermalbiopsiescollectedbyfiringalightweightdartwithasterilised

stainless-steelcuttingtipfromasterilisedprojector(e.g.Palsbølletal.,1991;Barrett-Lennard

etal.,1996)at theflankof thekillerwhale.Asastudyoncaptivekillerwhales found low

variability in the taxonomiccompositionof the skinmicrobiome fromdifferentbodysites

(Chiarelloetal.2017),smallvariationintheexactlocationontheflankfromwhichthebiopsy

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 7: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

7

wastakenshouldnotbiasourresults.Biopsieswerestoredinsteriletubesat-20°C.Atno

pointwerebiopsysamplesindirectcontactwithhumanskin.

DNAextraction,librarybuilding,andsequencinghavebeenpreviouslydescribed(Footeetal.,

2016). All lab work was conducted in a sterile flow hood to prevent contamination.

SequencingwasdoneattheDanishNationalHigh-ThroughputDNASequencingCentrewithin

theUniversityofCopenhagen.ThefacilityisspecificallygearedforlowDNAquantitylibrary

sequencingfromancientandenvironmentalDNA.Samplesofthesameecotypewerepooled

andsequencedacrossmultiplesequencinglanes.Samplesofdifferentecotypeswerealways

runondifferent sequencing lanes,with theexceptionof several typeB1 andB2 samples,

whichwere initially grouped as ‘type B’ (Pitman& Ensor, 2003) and some sampleswere

thereforesequencedonsharedlanes.

2.3|Sequencingreadpre-processing

Asameanstoenrichthedatasetforbacterialsequences,wefirstusedSAMtoolsv1.5(Liet

al.,2009)toremoveallsequencingreadsthatmappedtothekillerwhalereferencenuclear

genome (Oorca1.1, GenBank: ANOL00000000.2; Foote et al., 2015) and mitochondrial

genome (GU187176.1) with BWA-mem (Li & Durbin, 2009). The remaining reads were

adapter-trimmedusingAdapterRemovalV2.1.7(Schubert,Lindgreen,&Orlando,2016).We

then removed duplicates generated during library indexing PCR by merging reads with

identical sequences using in-house python scripts (Dryad doi:10.5061/dryad.c8v3rv6). All

readswithanaveragequalityscore<30werefilteredoutusingprinseqV0.20.4(Schmieder

&Edwards,2011),andallreadsof<35bpwereremovedusingAdapterRemoval.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 8: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

8

2.4|Investigatingcontamination

Despitetheprecautionsoutlinedabove,contaminationcanbeintroducedatseveralstages

ofthesequencedatageneration,andsubsequentlymistakenforthegenuinehost-associated

microbiomesignal.ContaminatingDNAcanbepresentinlaboratoryreagentsandextraction

kits(Lusketal.,2014;Salteretal.,2014).Forexample,silicainsomecommercialDNAspin

columns is derived from diatom cells and therefore can be a potential source of

contaminationwithdiatomDNA(Naccacheetal.,2013).However,theQiagenQIAquickspin

columnsused in this studydonotcontainsilica frombiologicalmaterial,according to the

manufacturer.Crosscontaminationcanalsooccurbetweensamplesprocessedinthesame

sequencing centre (Ballenghien et al., 2017). The impact of contamination increases in

sampleswithsmallamountsoftrueexogenousDNAandcanswampthesignalfromthehost’s

microbiome (Lusk et al., 2014; Salter et al., 2014). Contamination can be assessed using

negativecontrols(e.g.Davisetal.,2017).However,thedatausedinthisstudywasinitially

produced with the sole focus on the host organism. Including extractions and library

preparationblanksisnotaroutineprocedureinpopulationgenomicsstudiesbasedonhigh-

qualityhosttissuesamplesand,assuch,blankswerenotincludedinthelaboratoryworkflow

and hence not sequenced. Therefore, we instead implement an ad hoc workflow that

attemptstodifferentiatebetweencontaminantandrealexogenousDNAfromhostspecies

shotgunsequencingdata.

2.4.1|PhiXcontamination

The contaminationofmicrobial referencegenomesbyPhiX,which isusedas a control in

Illumina sequencing, is a known potential source of error inmetagenomics studies using

shotgunsequencingdata(Mukherjeeetal.,2015).Therefore,toavoiderroneousmappingof

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 9: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

9

PhiX-derived reads to contaminatedgenomes,we removedall readsmapping to thePhiX

genome used by Illumina (NC_001422) with BWA-mem 0.7.15 (Li, 2013) with default

parameters.

2.4.2|Environmentalandlaboratorycontamination

If the amount of contamination (derived from laboratory reagents or environment) is

relatively equal among samples, we expect the relative proportion of contaminant

sequencingreadstobeinverselycorrelatedtothequantityofsample-derivedDNA:i.e.low

DNAquantity sampleswillbedisproportionatelyaffectedbycontaminantDNAsequences

comparedwithhigh-quantitysamples (Lusketal.,2014;Salteretal.,2014).Wetherefore

estimatedthecorrelationbetweentheproportionofthetotalsequencingreadsassignedto

eachmicrobial taxon(seebelowforhowtaxonomicassignmentwasconducted)andtotal

DNAreadcountpersample(priortoremovalofhostDNAandbeforePCRduplicateremoval).

Microbialtaxaforwhichthereadcountwassignificantlynegativelycorrelatedwiththetotal

numberofreadspersample(includinghostDNA), i.e. thosethatconsistently increased in

abundanceinlow-quantityDNAsamples,wereflaggedaspotentialcontaminants.

2.4.3|Humancontamination

Toaccountforthepossibilityofcontaminationwithhuman-associatedmicroorganisms,we

nextquantifiedtheamountsofhumanDNAinoursamplesandusedthisasaproxyofhuman-

derived microbial contamination (see Supporting Information for the details of read

processing).Onlyreadsuniquelymappingtoasingleregionofthegenomewithhighquality

(samtools -q30 -F4 -F256)wereretained,andweremovedallduplicatesusingsamtools

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 10: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

10

rmdupinsingle-endmode.Humancontaminationlevelswereestimatedbycalculatingthe

percentageoffilteredreadsmappedtothehumangenome(TableS1).Weincludedthese

values as a covariate in statistical models as a way to, at least partially, control for

contaminationwithhuman-associatedmicroorganisms.

2.4.4|Knownbacterialcontaminants

Next,weinvestigatedwhetherspecificbacterialtaxathathavepreviouslybeenreportedto

belikelycontaminantsarepresentinourdataset.Followingread-basedanalyses,wefound

that our samples were dominated by Cutibacterium (Propionibacterium) acnes, which is

abundantonhumanskin(Byrdetal.,2018)andaknowncontaminantofhigh-throughput

sequencing data (Lusk et al., 2014;Mollerup et al., 2016).We therefore investigated the

distribution of sequence identity between ourC. acnes reads and theC. acnes reference

genomes,with the expectation that human or laboratory contaminantswould showhigh

(close to 100) percent identity, whereas killer whale-derived C. acnes would be more

divergent.

Additionally,weanalyseddatafromaNorthPacifickillerwhalesequencedat~20xcoverage

inapublishedstudy,inwhichsamplecollection,DNAextraction,andsequencingwasentirely

independentofourdataproduction(accessionnumber:SRP035610;Mouraetal.,2014).IfC.

acneswaspresentinthesedata,itwouldsuggestthateitheritwasarealcomponentofthe

killerwhaleskinmicrobiome,oritwasindependentlyintroducedascontaminationinboth

studies.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 11: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

11

Contaminanttaxaareunlikelytobeintroducedinisolation.C.acneswasconfirmedtobea

likelycontaminant(seebelow)andwethereforeremovedalltaxawithwhichitsignificantly

co-occurred.Usingnetassoc0.6.3(Morueta-Holmeetal.,2016),wecalculatedco-occurrence

scoresbetweenalltaxonpairsintherawtaxadataset.Wesetthenumberofnullreplicates

to999,andcorrectedp-values formultiplecomparisonsusing theFDRmethod.Fromthe

resultingmatrix,weselectedtaxawiththetop10%absolutesignificantco-occurrencescore

withcandidatecontaminanttaxa,andremovedthesetaxafromdownstreamanalyses,along

withC.acnes.

2.4.5|Investigatingsourcesofcontamination

Finally,toascertaintheauthenticityofourdataandtoestimatethelevelandpossiblesource

ofcontamination,weusedSourceTrackerV2.0.1(Knightsetal.,2011),atoolthatimplements

a Bayesian classification model to predict the proportion of taxa derived from different

potentialsourceenvironments.Thisapproachallowedustocomparethecompositionofthe

free-rangingkillerwhaleskinmicrobiometoothermarinemammalskinmicrobiotaandtoa

number of potential contaminating and environmental sources. We obtained data from

publicrepositoriesand includedmicrobialcommunitiesreflectingthemarineenvironment

(oceanwater from Southern Ocean and the North Pacific, Sunagawa et al., 2015), other

marinemammalskin(captivebottlenosedolphinsTursiopstruncatusandkillerwhalesalong

withtherespectivepoolwatersamples,andfree-ranginghumpbackwhales,Chiarelloetal.,

2017,Bierlichetal.,2018),likelycontaminantssuchashumanskinandgut(Ohetal.,2014,

Lloyd-Priceetal.,2017,Meiseletal.,2016),andlaboratorycontaminationfromcommonly

usedreagents(sterilewater,Salteretal.,2014)(TableS2).Weattemptedtospecificallyselect

sourcesthatwereobtainedwiththeshotgunsequencingapproachtoavoidpotentiallocus-

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 12: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

12

specific effects that can produce distinct microbiome profiles in amplicon-based studies.

However,only16SrRNAamplicondatawasavailableforthemarinemammalskinandthe

laboratorycontaminants,eachstudytargetingadifferentregionwithinthislocus(TableS2).

Therefore,tocontrolforlocus-specificeffects,wealsoincludedsamplesfromahumanskin

16Sampliconstudy(Meiseletal.,2016)andlimitedourdatatoreadsmappingtothe16S

rRNA gene for those comparisons (see Supporting Information for more detailed

methodologyofreadprocessing).

WeusedtheRpackageVeganV2.4.6(Oksanenetal.,2017)tocalculatedistancesbetween

microbiome profiles derived from these different datasets. After Total Sum Scaling (TSS)

normalization, abundance-based Bray-Curtis and presence/absence-based binary Jaccard

distanceswerecalculatedandvisualisedusingprincipalcoordinatesanalysis.Subsequently,

a subsetof sourceswasused in SourceTracker andweusedour killerwhaledata as sink

without applying rarefaction to either sink or source samples. We also repeated the

SourceTrackeranalysisusingfree-ranginghumpbackwhalesasthesinksamples.

2.5|Taxonomicassignment

We used MALT (MEGAN Alignment Tool) version 0.3.8 (Herbig et al., 2016) to create a

reference database of bacterial genomes downloaded from the NCBI FTP server

(ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA,accessed26thJanuary2017).Weperformeda

semi-global nucleotide-nucleotide alignment against the reference database. Semi-global

alignmentsaremoresuitableforassessingqualityandauthenticitycriteriacommontoshort-

readdata,andarealsousefulwhenaligning16SrRNAdataagainstareferencedatabasesuch

asSILVA(Herbigetal.,2016).Sequenceidentitythresholdwassetto95%asperVågeneet

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 13: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

13

al.,(2018),butwithamoreconservativethresholdofincludingonlytaxawithfiveormore

alignedreadsinsubsequentanalysis,

ThenucleotidealignmentsproducedinMALTwerefurtheranalysedinMEGANversion6.7.6

(Husonetal.,2016).Genomeswiththepresenceofstackedreadsinsomegenomicregions

and/or large gapswithout anymapped reads,were flaggedusing a custompython script

(Dryad doi:10.5061/dryad.c8v3rv6), and manually assessed in MEGAN. This step was

necessarytoidentifyspuriousandincorrectlysupportedbacterialtaxa,whichwereremoved

from furtheranalysis if they representedhighlyabundant species (Warinneret al., 2017).

TaxonomiccompositionofthesampleswasalsointeractivelyexploredinMEGAN,andthe

numberofreadsassignedtoeachtaxonwasexportedforsubsequentanalysis.

Taxonomic assignment was also carried out using an assembly-based approach. Filtered

metagenomicsequencesofallsamplesweremergedtoperformaco-assemblyusingMegahit

1.1.1(Lietal.,2015)withdefaultsettingsandk-list:21,29,39,59,79.Assemblyqualitywas

assessed using Quast 4.5 (Gurevich et al., 2013). Contigs were subsequently mapped to

reference bacterial genomeswithMGMapper (Petersen et al., 2017) using bestmode to

assigntaxonomy.TheassemblyfilewasindexedusingBWA-indexandSamtools-faidx.BWA-

memwassubsequentlyusedtomapthereadsofeachsamplebacktotheassemblycontigs

tofinallyretrievethemappedreadsusingSAMtools-view.Individualcoveragevalueswere

calculatedwithBedtools2.26.0(Quinlanetal.,2010)andcontigcoveragetablenormalised

usingCumulativeSumScale(CSS)asimplementedinMetagenomeSeq(Paulsonetal.,2013).

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 14: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

14

Thesequencingdatausedisthisstudyisrathershallowintermsofcoverageofmicrobialtaxa,

correspondingtolowcoveragekillerwhalegenomes(meanof2x).Therefore,weexplored

howlowsequencingdepthmayaffecttheinferredbacterialprofiles.Tothisend,weusedan

independentlysequenced20xcoverageresidentkillerwhalegenome(Mouraetal.,2014).By

drawing a random subset of reads from this genome using SAMtools, we compared the

taxonomiccompositionofthemicrobiomeofthesameindividualat20x,10x,5xand2xmean

sequencecoveragedepth.

2.6|Diversityanalyses

WecalculatedalldiversitymeasuresinVegan(Oksanenetal.,2017),usingreadsthatwere

assignedtothespecieslevelinMEGAN.Byfocussingontaxaatthespecies-level,wewere

able to explore the skinmicrobiome at a high resolution, an advantage of shotgun over

amplicon-basedanalyses.However,resultsofthisanalysisshouldbeinterpretedinlightofa

species-levelfocus,whereweareexploringasmallyetwell-resolvedrepresentationofthe

microbiome,whichmaypotentiallybeenrichedwithpathogensandcommonenvironmental

bacteria,ratherthanaholisticrepresentationoftheentireecosystem.

Tocontrol forbias introducedbyvaryinggenomesize (specieswith largergenomesshow

higherreadcounts,whicharetranslatedintohigherabundancescores;Warinneretal.,2017),

wedividedallreadcountsbythesizeoftherespectivefullbacterialgenome.Ifthetaxonwas

mappedtothelevelofthestrain,wedividedthereadcountbythepublishedgenomesizeof

thatstrain;ifidentifiedtothespecieslevel,wedividedthereadcountbytheaveragegenome

sizeacrossallpublishedstrainsofthatspecies.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 15: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

15

Beta diversity was explored using two dissimilarity matrices in the R-package Vegan:

abundance-based Bray-Curtis and presence/absence-based binary Jaccard distances. To

assess the strength and significance of ecotype and geographic location (longitude and

latitude) in describing variation in community composition, we conducted permutational

multivariateanalysisusingthefunctionAdonisinVegan.Wecontrolledfordifferingdepths

ofcoveragebetweensamplesusingtwotechniques.First,weusedgenomesize-controlled

data (see above) and included the number of reads mapping to the species level as a

covariate. Second, TSS normalisation of the genome-size controlled data was conducted,

followedbyconversiontotheBray-Curtisdistancematrix.TSSnormalisationisirrelevantfor

presenceabsencedata,asonlyspeciespresence,ratherthanspeciesabundance,isretained

inthebinarypresenceabsencematrix.Asaresult,threemodelswereexplored:twoBray-

Curtismodelswith differing depth-control techniques, and one Jaccardmodel using read

counts as covariate. Eachmodel consistedof the following covariates: latitude (numeric),

longitude(numeric),ecotype(categorical),andpercentagehumancontamination(numeric),

with library size included only when TSS normalisation was not used. For each model,

residualswerepermuted9999times.WeusedthefunctionBetadisper(Vegan)followedby

ANalysisOfVAriance(ANOVA)totestforhomogeneityofgroupdispersions.Betadispercan

beusedtoensure that i) theAdonismodel resultsarenotconfoundedbyheterogeneous

variances (Anderson, 2001) and ii) to make biological inferences about between-group

varianceincommunitycomposition.

Weused the function capscale from theVeganpackage to performPrincipal Co-ordinate

Analysis(PCoA).ThefourbacterialtaxathatdescribedthemostvariationonPCoA1,andthe

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 16: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

16

four that described the most variation on PCoA2, were designated as ‘driving taxa’. We

thereforeclassifiedatotalofeightuniquedrivingtaxathatdescribeindividualdifferencesin

microbiomecomposition(TableS4).

2.7|Networkanalysis

To venture beyond single microbial taxa and explore microbial interactions that include

interspecific dynamics, we expanded our analyses to networks of bacterial communities

associatedwiththedrivingtaxaidentifiedthroughthePCoA.Usingnetassoc(Morueta-Holme

etal.,2016),wecomparedtheobservedpartialcorrelationcoefficientsbetweentaxawitha

nulldistributionestimatedfromidenticalspeciesrichnessandabundancesastheobserved

data.Again, taxaco-occurrencescoreswerecalculatedbetweenall taxonpairs in theraw

dataset,withnull replicatesset to999.TheFDRmethodwasusedtocorrectp-values for

multiple comparisons. From the resulting matrix of significant co-occurrence scores, we

selectedthe20taxawiththehighestabsoluteco-occurrencescoreforeachofthe8unique

drivingtaxa.Wecreatedanewmatrixincludingonlythesetaxaandvisualisedco-occurrence

networks.

2.8|Functionalprofiling

Communitycompositioncanbeapoorpredictorofthefunctionaltraitsofthemicrobiome,

duetoprocessessuchashorizontalgenetransfer(HGT)betweenbacterialtaxa,whichcan

decouple species compositionand function (Koskellaetal.,2017). Shifting focus fromthe

taxonomiccompositiontothegeniccompositionofthemicrobiomereducestheimpactof

HGTonfunctionalcharacterisation(Koskellaetal.,2017).

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 17: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

17

To explore functional profiles of the samples, we used DIAMOND v0.9.10 with default

parameters(Buchfink,Xie&Huson,2015)tocreateareferencedatabaseofnon-redundant

proteinsequencesfromfullysequencedbacterialgenomesdownloadedfromtheNBCIFTP

server (https://ftp.ncbi.nlm.nih.gov/genomes/genbank/ accessed 9th March 2017).

Nucleotide-to-amino-acid alignments of the sample reads to the reference database was

performedinDIAMONDandthetop10%ofalignmentsperqueryreported.TheMEGANtool

daa-meganizerwasthenusedtoassignreadstoproteinsbasedontheDIAMONDalignments

andtoassignfunctionalrolestotheseproteinsusingtheSEED(Overbeeketal.,2005)and

eggNOG(Huerta-Cepasetal.,2017)databases.Sinceoneproteincanhavemorethanone

function,itispossibleforonereadtobeassignedtomultiplefunctionalsubsystems.Theraw

countdata(numberofreadsassignedtofunctionalsubsystem)wereexportedfromMEGAN

and further processed in R. To control for differences in library depth, read counts per

functionalgroupwerenormalisedbytotalreadnumbersmappingtoSEEDoreggNOGterms.

Weusedprincipalcomponentsanalysis(PCA)performedintheRpackageprcomptovisualise

differencesinfunctionalgroupsbetweenindividuals.

Weadditionallyperformedanassembly-basedfunctionalprofilingtoovercometheindividual

weaknessesofbothassemblyandread-basedmethodologies(Quinceetal.,2017).Abinitio

genepredictionwasperformedoverthemetagenomicassemblyusingProdigal2.6.3(Hyatt

etal.,2010).ThelistofpredictedgenesequenceswasindexedusingBWAandSAMtoolswas

usedtomapthereadsofeachsamplebacktothegenesequences.WeusedBedtools2.26.0

(Quinlan et al., 2010) to calculate individual coverages. Gene coverage table was

subsequentlyCSSnormalisedusingMetagenomeSeq(Paulsonetal.,2013).

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 18: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

18

2.9|Diatom-associationanalyses

Antarctickillerwhalesareoftenobservedtohaveayellowhue,whichhasbeenattributedto

diatom coverage (Berzin & Vladimirov, 1983; Pitman & Ensor, 2003) and identifiable

individualshavebeenobservedtotransitfromthisyellowskincolourationtoa‘clean’skin

condition (Durban & Pitman, 2012). This change is hypothesised to occur during brief

migrationstosubtropicallatitudes,whereturnoveroftheouterskinlayertakesplacewitha

reduced thermal cost (Durban & Pitman, 2012). If this hypothesis is correct, diatom

abundanceshouldbecorrelatedtoskinageandcolouration(Hart,1935;Konishietal.2008;

Durban&Pitman,2012).Inter-individualvariationinmicrobiomeprofileswithintheAntarctic

ecotypescouldthereforereflectvariationintheageoftheouterskinlayer.Duringnetwork

analysis,weidentifiedapossibleassociationbetweenakeybacterialtaxadrivingbetween-

sample differences in community composition (Tenacibaculum dicentrarchi) and bacterial

taxa associated with diatoms. Following from our observations that three samples from

AntarcticecotypeshadhighabundancesofT.dicentrarchiandthatinthePCoAthesesamples

weredifferentiated frommostother samples,we investigated the linkbetweenobserved

diatom coverage, abundance of T. dicentrarchi and abundance of other algae-associated

bacterialtaxa.WeconductedqualitativecolourgradingoftypeB1andtypeB2 individuals

usingphotographs takenat the timeofbiopsy collection, ranging from ‘clean’ through to

‘prominent’yellowcolouration.

Onlyalimitednumberofdiatomreferencegenomesareavailableandthereforeweusedtwo

methodologiestoquantifythelevelofdiatomDNAinoursamples.First,weusedMALTand

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 19: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

19

MEGAN in the same taxonomic pipeline as previously described, but with a reference

database comprised of NCBI RefSeq nucleotide sequences from the diatom phylum

Bacillariophyta (downloaded 30th October 2017). To date, only seven diatom reference

genomesareavailable,thusidentificationatthespecieslevelwasnotattempted.Instead,

numbersofreadsmappingtoBacillariophytawereexportedandfurtherprocessedinR.Raw

diatomreadcountswereconvertedtoproportionoftotalnumberofsequencingreadsper

sample.Second,wealignedallreadsagainsttheSILVArRNAdatabase(release128,Quastet

al.,2013)usingBWA-mem0.7.17,retainedreadsmappingwith>10mappingqualitywith

SAMtools and used uclust (Edgar, 2010) inQIIME 1.9.1 (Caporaso et al., 2010) to assign

taxonomybasedontheSILVA18Sdatabaseat97%similarity.FromtheresultingOTUtable

weretainedreadsthatmatchedtoknowndiatomtaxonomy.

We explored the correlation between latitude (by grouping North Pacific and Antarctic

ecotypes)andproportionofreadspersamplemappingtodiatomsusingageneralisedlinear

modelwithaquasipoissonerrorstructureandloglink.Ascovariates,weincludedlongitude,

numberofreadsmappingtothebacterialspeciesleveltocontrolforlibrarysize,andnumber

ofhumanreadstocontrol forhuman-associatedmicrobialcontamination.Usingthesame

model structure,we then tested the correlationbetweenproportionof readsper sample

mapping to diatomsand the presence/abundance ofT. dicentrarchi reads, aswell as the

correlationwithpresenceofknownalgae-associatedbacterialtaxa(includingT.dicentrarchi,

Cellulophagabaltica,Formosa sp.Hel1_33_131,Winogradskyella sp.,Marinovumalgicola,

Agarivorans gilvus, Pseudoalteromonas atlantica and Shewanella baltica: Bowman 2000;

Aminetal.,2012;Goeckeetal.,2013a,b,incorporatedasabinaryvariable).

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 20: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

20

3|RESULTS

Metagenomicprofilesfromtheskinmicrobiomeof49killerwhalesfromfiveecotypes(Figure

1)weresuccessfullyreconstructedusingshotgunsequencingdatafromDNAextractedfrom

skinbiopsies.Ofthereadsretainedfollowingourstringentfilteringprocedure,butbeforeour

investigationsintoC.acnesasapossiblecontaminant,8.20%(n=7,984,195)wereassigned

tomicrobialtaxausingtheread-basedapproach,with2.45%(n=2,384,587)assignedatthe

specieslevel(seeDryadrepositorydoi:10.5061/dryad.c8v3rv6).Overall,845taxaofmicrobes

were identified. The co-assembly yielded a 33.01 Mbp-long metagenome comprised of

45,934contigs(N50=970bp,average=730bp,max=48,182bp).Taxonomywasassigned

to41.73%ofthecontigs.Resultsfromtheassembly-basedapproachwereconcordantwith

theread-basedresults,wethereforereportonlythelatter.

3.1|Investigatingcontamination

Onaverage0.16%ofreads(range0.01-5.43%)readsmappedtothehumangenome(Table

S2),suggestingpresenceofhumancontaminationandmakingitpossiblethathuman-derived

bacteria were present in our dataset. After correcting for multiple testing, we found no

significantnegativecorrelationbetweentheproportionofreadsassignedtoeachbacterial

taxonandthetotalnumberofsequencedreads(SupportingInformation,Fig.S1).Negative

trends (although not significant) between some bacterial taxa and the total number of

sequenced reads were largely driven by one outlier sample with the lowest coverage

(B1_124047).Followingthededuplicationstepofourprocessingpipeline,thesetaxawereno

longerpresentinthedataset,astheyfellbelowourdefinedthresholdoffivealignedreadsin

MALT(Fig.S2).

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 21: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

21

C.acneswasidentifiedasthemostabundantbacterialtaxon,withanaverageabundanceof

39.57%(standarddeviation=24.65;Fig.S3),butitmayhavebeenintroducedviahumanor

laboratorycontamination(Lusketal.,2014).Percentidentitytothehuman-derivedC.acnes

genomewas100%for245andover97%for505ofthe527contigsidentifiedasC.acnesby

MGMapper(Fig.S4),supportingtheideaofalikelyexogenoussourceofC.acnes.Killerwhale

samplespooledbyecotypeweresequencedacrossmultiplesequencinglanes,allowingusto

investigateifcontaminationwithC.acneswasintroducedatthesequencingstep.RelativeC.

acnes abundanceper samplewashighly similar between sequencing lanes (Coefficient of

Variation=0.076;Fig.S5),suggestingthatthecontaminationoccurredpriortosequencing.

However,C.acneswasalsopresenttoahighabundance(18.06%ofreadsaligningatspecies

level)intheindependentlysequencedresidentkillerwhale(Mouraetal.,2014),suggesting

thatcontaminationwithC.acneswasnotspecifictoourworkflow.Weconcludedthatthere

wasahighprobabilitythatC.acneswasalabcontaminantandthereforeremovedallC.acnes

reads/contigsfromourdatasetbeforecontinuingwithanalysis.

3.1.1|NetworkanalysisresultsforC.acnesassociatedtaxa.

Followingitsidentificationasalikelycontaminant,weusednetworkanalysistoidentifyand

remove the top 10% of species which significantly co-occurred with C. acnes, which

correspondedtoco-occurrencescoresabovetheabsolutevalueof1000(Fig.S6).Overall,82

specieswereremoved(Dryaddoi:10.5061/dryad.c8v3rv6),manyofwhichareknownhuman-

associatedbacterialtaxa.Followingthisfilteringstep,onetypeCsamplehadnoremaining

taxa.Wethereforeexcludedthissamplefromfurtheranalyses.

3.1.2|Metagenomicaffinitiesofwildkillerwhaleskinmicrobiome

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 22: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

22

Only10killerwhalesampleshad50ormore16SreadswithassignedSILVAtaxonomy(eight

killerwhalesamplesremainedafterfilteringforC.acnesassociatedtaxa,Figure2).Overall,

priortoC.acnesfiltering,thekillerwhaledatasethad273taxaincommonwiththedataset

of2,279bacterialtaxaderivedfromsources(e.g.human,marinemammalandenvironmental

samples,seesection2.4.5).AfterfilteringforC.acnesandassociatedtaxa,236ofthe273

killerwhaleassociatedtaxaremained.Free-rangingkillerwhaleandhumpbackwhaleskin

microbiomesoverlappedontheprincipalcoordinates,independentoftheapplieddistance

measureandpresenceofC.acnesassociatedbacteria(Figure2a,Fig.S7a,b).Incontrast,data

fromthecaptivestudy,includingkillerwhaleandcaptivedolphinskinsamplesandtheirpool

water, clustered separately from all other studies. General separation by sequencing

approach (i.e. shotgun versus amplicon) was not observed: For instance, amplicon- and

shotgun-sequencedhumansamplesgroupedtogether(Figure2a,b).Itisthereforepossible

that theseparationof thecaptivestudysamples isduetoeithertheuseofaspecific16S

targetlocusorotherfactorsassociatedwithcaptiveversuswildenvironments(notethatthe

poolwasfilledwithseawaterfromtheMediterraneanSea;Chiarelloetal.,2017).

The threemarinemammal species formed one cluster irrespective of study on the third

dimensionintheabundance-basedBray-Curtisdistanceanalysis(Fig.S7c,d),suggestingthat

thereisacommonfactortothemarinemammalskinmicrobiomecomposition.Importantly,

thefree-rangingkillerwhalemicrobiomeprofilesgenerallygroupedawayfromthehuman

skinsamples,gutsamplesandlaboratorycontaminants.Theywerealsoseparatedfromthe

oceanwatersamples,suggestingthatthekillerwhaleskinmicrobiomescharacterisedinour

studyrepresentamicrobialcommunitythatisclearlydistinctfromsurroundingoceanwater.

Here,itisnoteworthythatfilteringofourdataforC.acnesassociatedtaxaatthegenuslevel

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 23: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

23

ishighlyconservativeandalsoremovesanumberofmicrobialtaxathatareabundantinthe

marineenvironment,astheybelongtothesamegeneraassomeC.acnesassociatedspecies.

Samplesrepresentinglaboratorycontaminationconsistentlyclusteredwiththehumanskin

samples(Figure2a,b,Fig.S7),suggestingthatonesourceofcontaminantsinlaboratorywork

are human-associated skinmicrobes. All results presented above were confirmed with a

largerdatasetthatincluded16killerwhalesampleswithatleast20bacterial16Sreadswith

SILVAtaxonomyassignment(Fig.S8).

Based on the principal coordinates analysis and for greater clarity of presentation, we

restrictedtheselectionofsamplesthatwereusedassourcesintheSourceTrackeranalysisto

captivedolphinskin(n=4),captivekillerwhaleskin(n=4),waterfromthecaptivekillerwhale

pool(n=4),wildhumpbackwhaleskin(n=4),SouthernOceanwater(n=4),humangut(n=4),

shotgun-derivedhumanskindatafromasebaceoussite(n=4),andlaboratorycontamination

(n=3;thefourthsamplehad<2016Sreadsandwasexcludedfromanalysis)(TableS2).The

SourceTrackerresultssupportedthoseoftheprincipalcoordinatesanalysis(Figure2c,d),with

humanskintaxacontributingonaverageonly3.4%tothewildkillerwhaleskinmicrobiome

(range0.0-18.4%).Thispercentagedecreasedto2.2%(range0.0-9.6%)afterfilteringoutC.

acnes associated taxa. The contribution of lab contaminantswas also low (average 4.2%,

range0.0-28.6) inallbutoneresidentkillerwhale individual (31868),whichwasremoved

afterC.acnesfilteringduetolow(<50)readnumbers(average1.7%,range0.0-7.1%after

removalofC.acnesassociatedtaxa).Thesourcescontributingthemosttothefree-ranging

killerwhale skinmicrobiomes after removingC. acnes associated taxa included Southern

Ocean (mean 32.3%, range 4.5-69.4%), humpback whale skin (11.9%, range 0-36.7% in),

captivekillerwhaleskinandcaptivedolphinskin(mean13.2%,range2.1-64.8%andmean

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 24: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

24

12.5%, range 0.2-40.8%, respectively). A high proportion of taxa observed in free-ranging

killerwhalescouldnotbeassignedtoanyofthesourcesincludedintheanalysis(‘Unknown’,

mean>25%).Thesetaxamayrepresentuncharacteriseddiversityspecifictothewildkiller

whale skinmicrobiome, a source thatwas not included in our analysis, e.g. oceanwater

collectedatthesametimeasthekillerwhaleskinbiopsies,ormarinemammalskintaxathat

are poorly characterised by the 16S locus targeted in othermarinemammalmicrobiome

studies.

ToverifytheSourceTrackerresultsforfree-rangingkillerwhalesamplesstudiedhere,wealso

ranSourceTrackerusingthefourwildhumpbackwhalesasthesinksampleswhileassigning

free-rangingkillerwhalesasasource(Fig.S9).Twohumpbackwhalessampledearlyinthe

foraging season around the Antarctic Peninsula closely resembled the wild killer whale

profiles, containing amixture of taxa attributed to the wild killer whale skin (41.7% and

65.3%),thecaptivedolphinskin(31.1%and2.7%)andunknownsources(21.3%and24.5%).

In contrast, the microbiome of the two humpback whales sampled late in the Antarctic

foragingseasonweredominatedbySouthernOceantaxa(both>95%).Thisisconsistentwith

thetemporalvariationinthecompletehumpbackwhaledatasetreportedbyBierlichetal.,

(2018). Overall, the detailed analyses of contributing sources of the killer whale skin

microbiome revealed a large proportion of taxa that are also found on the skin of other

marinemammalsandanimportantcontributionofenvironmentaloceanwatertaxa.Thisis

inlinewithpreviousreportsthatfoundasignificantcontributionofseawaterto,yetdistinct

composition of, marinemammalmicrobiomes (Bik et al., 2016). Expected contaminating

sources, such as human skin and laboratory contaminants, contributed only a small

proportiontoourkillerwhaleskinmicrobiomedataobtainedfromhostshotgunsequencing.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 25: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

25

3.2|Taxonomicexploration

Read-basedandassembly-basedapproachesproducedconcordanttaxonomicprofiles.The

mostabundantconstituentsof thekillerwhaleskinmicrobiomeat thephylum levelwere

Proteobacteria, Actinobacteria, Bacteroidetes and Firmicutes (Fig. S3a), which have been

identifiedinpreviousstudiesofbaleenwhaleskinmicrobiota(Appriletal.,2014;Shottset

al.,1990),includingthrough16Samplificationofskinswabsfromcaptivekillerwhalesunder

controlledconditions(Chiarelloetal.,2017).Atthespecies level,wefoundahighlevelof

inter-individual variation (Figure 3a, Fig. S3b), as previously found for four captive killer

whaleshousedinthesamefacility(Chiarelloetal.,2017).

Subsettingan independentlysequencedresidentkillerwhalegenometo lowersequencing

depth,weinferredthatwhilefivemostcommontaxawerefoundinsimilarproportionsin

highandlowcoveragedata,theidentificationofrarertaxabecamemorestochasticatlower

sequencingdepths(TableS3).Ourresultsmaythereforesufferfromthisbiasassociatedwith

lowcoveragedata,whichwouldbemostprominentinpresence/absence-basedanalyses.As

ameanstocontrolforthisbias,weincludelibrarysizeasacovariateinmodelsinvestigating

betadiversity.

3.3|Diversityanalyses

Human contaminationwas not a significant driver in themodels exploring beta diversity

(Table1),explainingatmost2%ofthevariation intaxonomiccomposition ineachmodel.

Ecotypewasasignificantvariable inallmodels,explaining10-11%ofvariation inthedata

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 26: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

26

(Table1).LatitudewassignificantinbothBrayCurtismodelsbutnotintheJaccardpresence-

absencemodel.Where significant, it explained 4 – 5% of variation in the data (Table 1).

Longitudewasnotsignificantinanyofthemodels.Betadisperanalysisrevealednosignificant

heterogeneity in the variation of community composition between ecotypes (non-TSS

normalisedBray-Curtis:df=4,F=0.52,p=0.72;TSSnormalisedBray-Curtis:df=4,F=1.74,

p=0.16;binary Jaccard:df=4,F=0.63,p=0.64).This suggests thatbetween-individual

variationinmicrobialcompositionissimilaramongecotypes.

TheBray-CurtisPCoAexplainedmorevariationthanJaccard(24.13%versus16.06%onthe

firsttwoaxes),wethereforefocusontheBray-Curtisresults.Anetworkbasedonsignificant

co-occurrencesbetweeneightbacterialtaxadrivingvariationattheindividuallevel(TableS4)

andthetop20co-occurringtaxaforeachofthedrivingtaxa,showedclearlydifferentiated

and distinct community groups (Figure 3). Further investigation into the top 20 taxa that

significantly co-occurred with the driving taxa found that three of the taxa showing the

highest co-occurrence scores with the driving taxon T. dicentrarchi (Formosa sp.

Hel1_33_131,Cellulophagaalgicola andAlgibacteralginolytica)areassociatedwithalgae

(Bowman,2000;Beckeretal.,2017;Sunetal.,2016).

3.4|T.dicentrarchianddiatoms

Bothapproachestodiatomidentificationproducedconcordantresults(Fig.S10,TableS5).

AntarcticecotypeshadasignificantlyhigherabundanceofdiatomDNAthanNorthPacific

ecotypes (β = 0.65, SE = 0.29, p = 0.03; Figure 4a). Individuals with “prominent” yellow

colourationshowedhigherdiatomabundance(Figure4b),supportingthelinkbetweenskin

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 27: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

27

colouranddiatompresenceinAntarctickillerwhales.Furthermore,theabundanceofdiatom

DNApersamplewassignificantlypositivelycorrelatedwiththeabundanceandpresenceof

T. dicentrarchi reads (number ofT. dicentrarchi reads: β = 0.014, SE = 0.003, p = <0.001;

presenceofT.dicentrarchi:β=0.915,SE=0.207,p<0.001;Figure4c)andthepresenceofat

leastonealgae-associatedbacterialtaxa(β=0.98,SE=0.17,p=<0.001;Figure4d).

3.5|Functionalanalysis

Intheread-basedfunctionalanalysisatotalof3,611,441readsmappedtoeggNOGfunctions

and1,440,371readsmappedtoSEEDfunctions.Inthecontig-basedfunctionalanalysiswe

identified 56,042 potential genes in ourmetagenome, out ofwhich EggNog functionwas

assignedto35,182.Bothapproachesidentifiedenergyproductionandconversion(classC)

andaminoacidmetabolismandtransport(classE)asthemostabundanteggNOGfunctions

inourdataset.TheeggNOGPCArevealedhighvariabilitybetween individuals (Figure4e),

however,aclusterofAntarcticwhaleswasobservedinprincipalcomponent2.Thesesamples

had high abundances of T. dicentrarchi (Figure 4e) and were associated with functions

corresponding to the COG functional categories J (translation, ribosomal structure and

biogenesis,β=0.007,SE=0.002,p=<0.001),F(nucleotidetransportandmetabolism,β=

0.005,SE=0.002,p=0.004)andI(lipidtransportandmetabolism,β=0.004,SE=0.002,p=

0.03). The same cluster of high abundance T. dicentrarchi Antarctic samples was also

identifiedintheSEEDPCA(Fig.S11).Thesesampleshadincreasednumbersofreadsmapping

toDNAmetabolism,aminoacidsandderivativesandcofactors/vitamins,althoughnoneof

thesefunctionsweresignificantlycorrelatedwithT.dicentrarchiabundance.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 28: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

28

4|DISCUSSION

Ourstudyhighlights thatcommunitiesofexogenousorhost-associatedmicrobiotacanbe

geneticallycharacterisedfromshotgunsequencingofDNAextractedfromthehosttissue.

However, dedicated analysis and treatment of contamination is necessary and requires

careful consideration in studies such as this, whereby samples were not collected nor

sequencedwiththeintentionofgeneticallyidentifyingmicrobiota.Insuchcases,thenormal

stringentcontrolmeasureswhichareroutineinmicrobialstudies,suchasthesequencingof

blanks, may not be possible. We have therefore presented an array of approaches for

estimating the proportion and sources of contamination and accounting for it in shotgun

studies.Overall,ouranalysessuggestthatwithcarefulconsideration,theminingofmicrobial

DNAfromhostshotgunsequencingdatacanprovideusefulbiological insightsthat inform

future targeted investigations intomicrobiome composition and function under stringent

laboratoryconditions.

After carefully filtering our data,wewere able to identify species interactions, ecological

networksandcommunityassemblyofthemicrobesanddiatomsthatcolonisekillerwhale

skinbyutilisingunmappedreadsfromshotgunsequencingdatageneratedfromskinbiopsies.

Akeyadvantageof this approachoveramplicon-based sequencing is theability toassess

functionalvariationbasedongenecontent,andtoidentifytaxatospecieslevel(Koskellaet

al., 2017; Quince, Walker, Simpson, Loman, & Segata, 2017). However, despite ongoing

efforts to describe bacterial species diversity, the breadth of the reference database is a

limitingfactorintheunbiasedcharacterisationofbacterialcomposition.Thus,taxaidentified

inouranalysesarenecessarilylimitedtospecieswithavailablegenomicinformation,andin

somecasesare likely to represent their closephylogenetic relatives (Tessleret al., 2017).

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 29: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

29

Hence,wereferto‘taxa’ratherthan‘species’whereappropriate.Wealsodemonstratethe

impactofcontaminationonthelownumbersofreadsfromtruehost-associatedmicrobes,

whichcandilutethesignalofbiologicallymeaningfulvariationamongsamples.

Socialandgeographicfactorshavebeenfoundtoinfluencemicrobialdiversityinterrestrial

andsemi-terrestrialanimals(Koskellaetal.,2017).However,thereislessunderstandingof

howthesefactorsinterplayinawide-rangingsocialmarinemammaliansystem(Nelsonetal.,

2015).We found that beta diversity of the killerwhale skinmicrobiomewas significantly

influencedbyecotypeandlatitude.Temperaturehasbeenshowntobeakeydeterminantof

marinemicrobialcommunitystructureataglobalscale(Salazar&Sunagawa,2017;Sunagawa

etal.,2015).However,theeffectofecotypeasthemostimportanttestedvariablehighlights

the significance of social and phylogenetic factors in shaping microbiome richness and

composition.Inaddition,itunderscoresthatalthoughkillerwhaleskinisinfluencedbythe

localenvironment(Romano-Bertrandetal.,2015),itrepresentsauniqueecosystemthatis

separatefromthatofthesurroundinghabitat.Concordantwithourresults,astudyofthe

microbiomeoffourcaptivekillerwhalesandtheseawaterfromtheirpoolfoundthatthe

skin microbiota were more diversity and phylogenetically distinct from the sea water

microbialcommunity(Chiarelloetal.,2017).Killerwhalesarehighlysocialmammals(Baird,

2000; Ford, 2009), and thus are likely to have a high potential for horizontal transfer of

microbes between individuals during contact (Nelson et al., 2015). Ecotype-specific social

behaviour, organisation and population structure, as well as other variables related to

ecotypeecology,suchasrangesize,anddiet(duetotransmissionofbacteriafromdifferent

preyspecies;Menkeetal.,2017),arealllikelytoaffectthediversityofmicrobialspeciesthat

individuals areexposed to, andalso influence the levelofhorizontal transferofmicrobes

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 30: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

30

betweenwhales.Thestrongsocialphilopatryinkillerwhales(Baird,2000;Ford,2009)and

thephylogeneticandphylogeographichistoryofecotypesisalsolikelytoplayarole,whereby

dueto limitedsocial transmissionbetweenecotypes, thephylogenyofbacterialspecies is

likelytoreflectthatofthehost(Leyetal.,2008,butseeRothschildetal.,2018).Itisalsolikely

to be influenced by the host’s evolutionary history, including secondary contact between

ecotypes(Foote&Morin,2016),wherebothverticalandhorizontaltransmissionofmicrobes

betweenecotypesispossible.

Despitethesignificanceof‘ecotype’asadriverofskinmicrobiomediversityinkillerwhales,

atleast79%ofthevariationinthemicrobiomeisunexplainedbythedriversconsideredin

ourmodels (Table 1). There is a strong overlap among ecotypes in the PCoA (Figure 3b),

suggesting a shared coremicrobiomewhichmay be partially sharedwith other cetacean

species(Figure2).Additionally,thePCoAshowssubstantialvariationwithinecotypes(Figure

3b), furtherhighlighting the roleof someotherdriver(s) ofmicrobiomevariation.Among

Antarcticecotypes,individualvariationwasassociatedwithdiatompresenceandadiscrete

subnetworkofmicrobialtaxa.Theoccurrenceofa‘yellowslime’attributedtodiatomsonthe

skinofwhales,includingkillerwhales,wasrecordedasearlyasacenturyago(Bennett1920;

Pitman et al., 2018). The extent of diatom adhesion on Antarctic whales is thought to

correlatewithlatitudeandthetimethewhalehasspentincoldwaters(Hart,1935;Konishi

etal.2008).Theskinmicrobiomeofhumpbackwhaleshasbeenreportedtochangethrough

theAntarcticforagingseason(Bierlichetal.,2018),andourSourceTrackeranalysisfoundthat

humpback whales sampled during the late foraging season (i.e. individuals who had

presumablyspentlongerintheSouthernOceanwatersatthetimeofsampling)hadmore

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 31: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

31

similarity toSouthernOceanmicrobial communities than those collectedduring theearly

foragingseason.Thisraisestheintriguingquestionastowhetherthetimespentinthefrigid

Antarcticwaterscouldbeadriverofvariation in theskinmicrobiomeanddiatom loadof

Antarctickillerwhales.

Satellite tracking of type B1 and B2 killer whale movements, Durban & Pitman (2012)

documentedrapidreturnmigrationstosubtropical latitudes, inwhichindividualstravelled

upto9,400kmin42days.Basedonthestrongdirectionalityandvelocityoftravelduring

thesemigrations,Durban&Pitman(2012)hypothesisedthattheywerenotassociatedwith

breedingorfeedingbehaviour.Instead,theyarguedthatthesemigrationscouldbedrivenby

theneedtoleavethefrigidAntarcticwatersandtemporarilymovetowarmerwaters,toallow

forphysiologicalmaintenanceincludingtheregenerationoftheouterskinlayer(Durban&

Pitman,2012).TheidentificationofthesameindividualsinAntarcticwaters,sometimeswith

a thick accumulation of diatoms, and at other times appearing ‘clean’, supports the

hypothesisthatskinregenerationisanintermittentratherthancontinuousprocess(Durban

&Pitman,2012).Furthermore,bitemarksfromcookiecuttersharksIsistiusspp.,whichare

typicallyfoundinwarmandtemperatewaters,areregularlyobservedontypeCkillerwhales

(Pitman et al., 2018),which have also been directly tracked towarm sub-tropicalwaters

(Durbanetal.,2013).

WepresentgeneticsupportforthehypothesisofDurban&Pitman(2012),that‘clean’and

yellow-tintedtypeB1andB2killerwhalesrepresentdifferencesindiatomload.Inaddition,

we provide the first evidence that the extent of diatom coverage is also associatedwith

significantvariationintheskinmicrobiomecommunity.WefoundthatAntarctickillerwhales

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 32: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

32

with the highest diatom abundance also had skinmicrobiomesmost similar to Southern

Oceanmicrobialcommunities,suggestingthatatthetimeofsamplingtheseindividualshad

spent longer in theAntarcticwaters,consistentwiththehypothesis thatdiatomcoverage

accumulateswithtimespentinthecoldSouthernOceanwaters.Perhapsmostsignificantly,

diatomabundancewaspositivelycorrelatedwiththeabundanceofT.dicentrarchi,aknown

pathogeninseveralfishspecies,whichisassociatedwithskinlesionsandseveretailandfin

rot(Piñeiro-Vidaletal.,2012;Habibetal.,2014;Avendaño-Herreraetal.,2016).

Ouranalyses revealed that sampleswithhighabundancesofT.dicentrarchi showdistinct

functionalprofiles.Functionalanalysesremainexploratoryatthisstage,constrainedbythe

difficultyoftranslatingbroadfunctionalcategories intobiologicalmeaning.However,with

more data that links individual health status and microbiome composition, functional

analyses may provide a tool for identifying individuals at risk. Therefore, whether T.

dicentrarchi represents a pathogen to killerwhalehosts remainsunknown. Type B1 killer

whales in apparently poor health andwith heavy diatom loads have been observedwith

severeskinconditions(skinpeelingandlesions;Figure4f),however,Tenacibaculumsp.have

beenreportedinupto95%ofhumpbackwhalessampledinrecentstudies,whichincluded

apparently healthy individuals (Appril et al., 2011, 2014; Bierlich et al., 2018). Skin

maintenancemaythusrepresentabalancingactforAntarctickillerwhalesofmanagingthe

costsofpathogenload,thermalregulation,reducedforagingtimeandlong-rangemovement.

Researchintotheskinmicrobiomeshouldthereforecontinuetoformacomponentofthe

ongoing holistic and multidisciplinary research programme to investigate the health of

Antarctic killer whale populations and more broadly in studies on the health of marine

mammals(e.g.Appriletal.,2014;Ravertyetal.,2017).

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 33: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

33

Ongoing field efforts provide the opportunity to further explore the relationships and

interactionsbetweenkillerwhalehosts,theirskinmicrobiome,otherexogenoussymbionts

suchasdiatoms,andtheenvironment.Ourcommunity-basedanalysessuggestthepresence

ofadistinctenvironmentaltaxanetworkcentredonP.haloplanktisasadrivingtaxon(Fig.

3c). Collection andmetagenomic characterization of environmental samples, such as sea

water, alongside host biological samples would allow further explorations into the

contributionoflocalecologicalfactorstothehostmicrobiome.Asameansofreducingthe

impact of contamination with DNA from laboratory environment, microbiome

characterisation can be conducted by means of RNA sequencing. This has an additional

advantage of generating metatranscriptomic data, which, in combination with the

metagenomic data, can facilitate the comparison/contrast between community function

(usingRNAtranscript)andcommunitytaxonomiccomposition(usingDNAsequence;Koskella

et al., 2017). Thismay further reduce thepotential impactof common lab contaminants,

allowing the exploration of the bacterial functional repertoire that is in use in a given

ecologicalcontext, includingreconstructionofmetabolicpathways(Bashiardes,Zilberman-

Schapira,Elinav2016).Contaminationinthelaboratorycouldbefurthercontrolledforand

characterisedthroughinclusionofextraction,librarypreparationandPCRblanksasnegative

controls(Lusketal.,2014;Salteretal.,2014)andmeasuressuchasdouble-indexing(Rohland

&Reich,2012;vanderValketal.,2018),whichcantheninformtheemergingdownstream

filteringmethods for separating truemicrobiomes from contamination (Delmont & Eren,

2016; Davis et al., 2018). Lastly, the advances in long-read sequencing using portable

nanopore-based platforms make it possible to generate data suitable for reconstructing

completebacterialgenomeswhileinthefield(Parkeretal.,2017),includingintheAntarctic

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 34: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

34

(Johnsonetal.,2017).Thisisapromisingdevelopmentwithrespecttoimprovingthebreadth

ofhost taxa fromwhichbacterial taxaarederivedandshould improve futuremappingof

metagenomicsdataandtaxonomicassignment.

AcknowledgementsThesuggestionofharvestingtheskinmicrobiomefromhostshotgundatawasfirstmootedbyGeraldPaooftheSalkInstituteduringameetingbackin2012whenwefirstembarkingontheshotgunsequencingproject,andwearegratefultoGeraldforsowingthisseed.WewouldliketothankBobPitmanwhowasinvolvedinthecollectionofmanyofthesamplesusedinthisstudyandgreatlycontributedthroughmanydiscussionsonthevariationamongkillerwhaletypes.We’dfurtherliketothankDavidStudholmeforpointingouttherichliteraturesurroundingdiatommicrobiomes,aswellaspotentialdiatom-relatedcontamination.JamesFellows Yates, Linda Rhodes, Morten Limborg and the microbiome journal club of theEvogenomicssectionattheCentreforGeoGeneticsprovidedvaluablefeedbackonthisworkand an earlier draft of this manuscript. We acknowledge The Danish National High-ThroughputDNASequencingCentreforsequencingthesamplesand,particularly,AndaineSeguin-Orlando, Lillian Petersen, Cecilie Demring Mortensen, Kim Magnussen and IanLissimore for technical support. Sample collection from killer whales in Antarctica wassupportedby the LindbladExpeditions-NationalGeographicConservation Fund. Thisworkwas funded by European Research Council grant ERCStG-336536 to J.B.W.W.; a DanishNationalResearchFoundationgrantDNRF94toM.T.P.G,theWelshGovernmentandHigherEducationFundingCouncilforWalesthroughtheSêrCymruNationalResearchNetworkforLowCarbon,EnergyandEnvironment,andfromtheEuropeanUnion’sHorizon2020researchandinnovationprogrammeundertheMarieSkłodowska-CuriegrantagreementNo.663830andaBritishEcologicalSocietygrant(SR17\1227)toAF,theFORMASgrant(project2016-00835) to KG.A.A.was supportedby theDanishCouncil for IndependentResearch -DFF(grant5051-00033)andLundbeckfonden(grantR250-2017-1351).Finally,wewouldliketothank the Linnaeus Scholarship Foundation at the University of Uppsala for providing ascholarshiptoRH,theErasmusMundusMasterProgrammeinEvolutionaryBiology(MEME)for facilitating the collaborative supervision of RH, and to Vanessa and Peter Hooper forprovidingfinancialsupporttoRHthroughoutthisproject.ReferencesAlberdi,A.,Aizpurua,O.,Bohmann,K.,Zepeda-Mendoza,M.L.,&Gilbert,M.T.P.(2016).Dovertebrategutmetagenomesconferrapidecologicaladaptation?.Trendsinecology&evolution,31,689-699.Alberdi,A.,Aizpurua,O.,Gilbert,M.T.P.,&Bohmann,K.(2017).Scrutinizingkeystepsforreliablemetabarcodingofenvironmentalsamples.MethodsinEcologyandEvolution, 9,134–147.AmesS.K.,GardnerS.N.,MartiJ.M.,SlezakT.R.,GokhaleM.B.,AllenJ.E.(2015)Usingpopulationsofhumanandmicrobialgenomesfororganismdetectioninmetagenomes.GenomeResearch,25,1056–1067.Amin,S.A.,Parker,M.S.,&Armbrust,E.V.(2012).Interactionsbetweendiatomsandbacteria.MicrobiologyandMolecularBiologyReviews,76(3),667-684.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 35: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

35

Anderson,M.J.(2001).Anewmethodfornon-parametricmultivariateanalysisofvariance.Australecology,26(1),32-46.Apprill,A.,Robbins,J.,Eren,A.M.,Pack,A.A.,Reveillaud,J.,Mattila,D.,Moore,M.,Niemeyer,M.,Moore,K.M.&Mincer,T.J.(2014).Humpbackwhalepopulationsshareacoreskinbacterialcommunity:towardsahealthindexformarinemammals?.PLoSOne,9(3),e90785.Avendaño-Herrera,R.,Irgang,R.,Sandoval,C.,Moreno-Lira,P.,Houel,A.,Duchaud,E.,...&Ilardi,P.(2016).Isolation,characterizationandvirulencepotentialofTenacibaculumdicentrarchiinsalmonidculturesinChile.Transboundaryandemergingdiseases,63(2),121-126.Baird,R.W.(2000).Thekillerwhale.Cetaceansocieties:fieldstudiesofdolphinsandwhales,127-153.Ballenghien,M.,Faivre,N.,&Galtier,N.(2017).Patternsofcross-contaminationinamultispeciespopulationgenomicproject:detection,quantification,impact,andsolutions.BMCbiology,15(1),25.Barrett-Lennard,L.G.(2000).Populationstructureandmatingpatternsofkillerwhales(Orcinusorca)asrevealedbyDNAanalysis(Doctoraldissertation,UniversityofBritishColumbia).Barrett-Lennard,L.,Smith,T.G.,&Ellis,G.M.(1996).Acetaceanbiopsysystemusinglightweightpneumaticdarts,anditseffectonthebehaviorofkillerwhales.MarineMammalScience,12(1),14-27.BashiardesS,Zilberman-SchapiraG,ElinavE.(2016).UseofMetatranscriptomicsinMicrobiomeResearch.BioinformaticsandBiologyInsights,10,19–25.Becker,S.,Scheffel,A.,Polz,M.F.,&Hehemann,J.H.(2017).Accuratequantificationoflaminarininmarineorganicmatterwithenzymesfrommarinemicrobes.Appliedandenvironmentalmicrobiology,83(9),e03389-16.Belden,L.K.,Hughey,M.C.,Rebollar,E.A.,Umile,T.P.,Loftus,S.C.,Burzynski,E.A.,Minbiole,K.P.,House,L.L.,Jensen,R.V.,Becker,M.H.&Walke,J.B.(2015).Panamanianfrogspecieshostuniqueskinbacterialcommunities.Frontiersinmicrobiology,6,1171.Bennett,A.G.(1920).Ontheoccurrenceofdiatomsontheskinofwhales.InProc.R.Soc.Lond.B(Vol.91,No.641,pp.352-357).Berzin,A.A.,&Vladimirov,V.L.(1983).Anewspeciesofkillerwhale(Cetacea,Delphinidae)fromtheAntarcticwaters.ZoologicheskyZhurnal,62(2),287-295.Bik,E.M.,Costello,E.K.,Switzer,A.D.,Callahan,B.J.,Holmes,S.P.,Wells,R.S.,…Relman,D.A.(2016).Marinemammalsharboruniquemicrobiotasshapedbyandyetdistinctfromthesea.NatureCommunications,7,10516.doi:10.1038/ncomms10516Bista,I.,Carvalho,G.R.,Tang,M.,Walsh,K.,Zhou,X.,Hajibabaei,M.,...&Christmas,M.(2018).Performanceofampliconandshotgunsequencingforaccuratebiomassestimationininvertebratecommunitysamples.Molecularecologyresources.BolgerAM,LohseM,UsadelB(2014)Trimmomatic:aflexibletrimmerforIlluminasequencedata.Bioinformatics,30,2114–20.Bordenstein,S.R.,&Theis,K.R.(2015).Hostbiologyinlightofthemicrobiome:tenprinciplesofholobiontsandhologenomes.PLoSbiology,13(8),e1002226.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 36: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

36

Bowman,J.P.(2000)DescriptionofCellulophagaalgicolasp.nov.,isolatedfromthesurfaceofAntarcticalgae,andreclassificationofCytophagauliginosa(ZoBellandUpham1944)Reichenbach1989asCellulophagauliginosacomb.nov.InternationalJournalofSystematicandEvolutionaryMicrobiology,50,1861–1868.Buchfink,B.,Xie,C.,&Huson,D.H.(2015).FastandsensitiveproteinalignmentusingDIAMOND.Naturemethods,12(1),59.Buller,N.B.(2014).Bacteriaandfungifromfishandotheraquaticanimals:apracticalidentificationmanual.Cabi.Byrd,A.L.,Belkaid,Y.,&Segre,J.A.(2018).Thehumanskinmicrobiome.NatureReviewsMicrobiology.Caporaso,J.G.,Kuczynski,J.,Stombaugh,J.,Bittinger,K.,Bushman,F.D.,Costello,E.K.,...&Huttley,G.A.(2010).QIIMEallowsanalysisofhigh-throughputcommunitysequencingdata.Naturemethods,7(5),335.Chng,K.R.,Tay,A.S.L.,Li,C.,Ng,A.H.Q.,Wang,J.,Suri,B.K.,Matta,S.A.,McGovern,N.,Janela,B.,Wong,X.F.C.C.&Sio,Y.Y.(2016).Wholemetagenomeprofilingrevealsskinmicrobiome-dependentsusceptibilitytoatopicdermatitisflare.Naturemicrobiology,1(9),16106.Csardi,G.,&Nepusz,T.(2006).Theigraphsoftwarepackageforcomplexnetworkresearch.InterJournal,ComplexSystems,1695(5),1-9.Davis,N.M.,Proctor,D.,Holmes,S.P.,Relman,D.A.,&Callahan,B.J.(2017).Simplestatisticalidentificationandremovalofcontaminantsequencesinmarker-geneandmetagenomicsdata.bioRxiv,221499.Delmont,T.O.,&Eren,A.M.(2016).Identifyingcontaminationwithadvancedvisualizationandanalysispractices:metagenomicapproachesforeukaryoticgenomeassemblies.PeerJ,4,e1839.Durban,J.W.,Fearnbach,H.,Burrows,D.G.,Ylitalo,G.M.,&Pitman,R.L.(2017).MorphologicalandecologicalevidencefortwosympatricformsofTypeBkillerwhalearoundtheAntarcticPeninsula.PolarBiology,40(1),231-236.Durban,J.W.,&Pitman,R.L.(2012).Antarctickillerwhalesmakerapid,round-tripmovementstosubtropicalwaters:evidenceforphysiologicalmaintenancemigrations?.BiologyLetters,8(2),274-277.DurbanJW,PitmanRL(2013)OutofAntarctica:divedatasupport‘physiologicalmaintenancemigration’inAntarctickillerwhales.20thBiennialConferenceontheBiologyofMarineMammals.SocietyforMarineMammalogy(9–13December2013,Dunedin,NewZealand).Edgar,R.C.(2010).SearchandclusteringordersofmagnitudefasterthanBLAST.Bioinformatics,26(19),2460-2461.Filatova,O.A.,Borisova,E.A.,Shpak,O.V.,Meshchersky,I.G.,Tiunov,A.V.,Goncharov,A.A.,Fedutin,I.D.&Burdin,A.M.(2015).ReproductivelyisolatedecotypesofkillerwhalesOrcinusorcainseasoftheRussianfareast.BiologyBulletinoftheRussianAcademyofSciences,42,674-681.Foote,A.D.,&Morin,P.A.(2016).Genome-wideSNPdatasuggestcomplexancestryofsympatricNorthPacifickillerwhaleecotypes.Heredity,117(5),316.Foote,A.D.,Vijay,N.,Ávila-Arcos,M.C.,Baird,R.W.,Durban,J.W.,Fumagalli,M.,...&Robertson,K.M.(2016).Genome-culturecoevolutionpromotesrapiddivergenceofkillerwhaleecotypes.Naturecommunications,7,11693.Ford,J.K.,Ellis,G.M.,Barrett-Lennard,L.G.,Morton,A.B.,Palm,R.S.,&BalcombIII,K.C.(1998).Dietaryspecializationintwosympatricpopulationsofkillerwhales(Orcinusorca)incoastalBritishColumbiaandadjacentwaters.CanadianJournalofZoology,76(8),1456-1471.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 37: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

37

Ford,J.K.(2009).Killerwhale:Orcinusorca.InEncyclopediaofMarineMammals(SecondEdition)(pp.650-657).Forney,K.A.,&Wade,P.R.(2006).Worldwidedistributionandabundanceofkillerwhales.Whales,whalingandoceanecosystems,145-162. Goecke,F.,Labes,A.,Wiese,J.,&Imhoff,J.F.(2013a).Phylogeneticanalysisandantibioticactivityofbacteriaisolatedfromthesurfaceoftwoco-occurringmacroalgaefromtheBalticSea.Europeanjournalofphycology,48(1),47-60.Goecke,F.,Thiel,V.,Wiese,J.,Labes,A.,&Imhoff,J.F.(2013b).Algaeasanimportantenvironmentforbacteria–phylogeneticrelationshipsamongnewbacterialspeciesisolatedfromalgae.Phycologia,52(1),14-24.Gurevich,A.,Saveliev,V.,Vyahhi,N.,&Tesler,G.(2013).QUAST:qualityassessmenttoolforgenomeassemblies.Bioinformatics,29(8),1072-1075.Habib,C.,Houel,A.,Lunazzi,A.,Bernardet,J.F.,Olsen,A.B.,Nilsen,H.,...&Duchaud,E.(2014).MultilocussequenceanalysisofthemarinebacterialgenusTenacibaculumsuggestsparallelevolutionoffishpathogenicityandendemiccolonizationofaquaculturesystems.Appliedandenvironmentalmicrobiology,80(17),5503-5514.Hamady,M.,&Knight,R.(2009).Microbialcommunityprofilingforhumanmicrobiomeprojects:Tools,techniques,andchallenges.Genomeresearch,19(7),1141-1152.Hart,T.J.(1935).Onthediatomsoftheskinfilmofwhales,andtheirpossiblebearingonproblemsofwhalemovements.UniversityPress.Herbig,A.,Maixner,F.,Bos,K.I.,Zink,A.,Krause,J.,&Huson,D.H.(2016).MALT:FastalignmentandanalysisofmetagenomicDNAsequencedataappliedtotheTyroleanIceman.bioRxiv,050559.Hoelzel,A.R.,&Dover,G.A.(1991).Geneticdifferentiationbetweensympatrickillerwhalepopulations.Heredity,66(2),191.Hoelzel,A.R.,Hey,J.,Dahlheim,M.E.,Nicholson,C.,Burkanov,V.,&Black,N.(2007).Evolutionofpopulationstructureinahighlysocialtoppredator,thekillerwhale.MolecularBiologyandEvolution,24(6),1407-1415.HolmesI,DavisRaboskyAR.(2018)Naturalhistorybycatch:apipelineforidentifyingmetagenomicsequencesinRADseqdata.PeerJ6:e4662https://doi.org/10.7717/peerj.4662Huerta-Cepas,J.,Forslund,K.,Coelho,L.P.,Szklarczyk,D.,Jensen,L.J.,vonMering,C.,&Bork,P.(2017).Fastgenome-widefunctionalannotationthroughorthologyassignmentbyeggNOG-mapper.Molecularbiologyandevolution,34(8),2115-2122.Huson,D.H.,Auch,A.F.,Qi,J.,&Schuster,S.C.(2007).MEGANanalysisofmetagenomicdata.Genomeresearch,17(3),377-386.Hyatt,D.,Chen,G.L.,LoCascio,P.F.,Land,M.L.,Larimer,F.W.,&Hauser,L.J.(2010).Prodigal:prokaryoticgenerecognitionandtranslationinitiationsiteidentification.BMCbioinformatics,11(1),119.Johnson,S.S.,Zaikova,E.,Goerlitz,D.S.,Bai,Y.,&Tighe,S.W.(2017).Real-timeDNAsequencingintheAntarcticdryvalleysusingtheOxfordNanoporesequencer.JournalofBiomolecularTechniques:JBT,28(1),2.Jones,F.C.,Grabherr,M.G.,Chan,Y.F.,Russell,P.,Mauceli,E.,Johnson,J.,Swofford,R.,Pirun,M.,Zody,M.C.,White,S.etal.(2012).Thegenomicbasisofadaptiveevolutioninthreespinesticklebacks.Nature,484(7392),55.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 38: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

38

Kircher,M.,Sawyer,S.,&Meyer,M.(2011).DoubleindexingovercomesinaccuraciesinmultiplexsequencingontheIlluminaplatform.Nucleicacidsresearch,40(1),e3-e3.Kolodny,O.,Weinberg,M.,Reshef,L.,Harten,L.,Hefetz,A.,Gophna,U.,Feldman,M.W.,&Yovel,Y.(2017).Whoisthehostofthehost-associatedmicrobiome?Colony-leveldynamicsovershadowindividual-levelcharacteristicsinthefurmicrobiomeofasocialmammal,theEgyptianfruit-bat.bioRxiv,232934.Kong,H.H.,Oh,J.,Deming,C.,Conlan,S.,Grice,E.A.,Beatson,M.A.,Nomicos,E.,Polley,E.C.,Komarow,H.D.,Murray,P.R.&Turner,M.L.(2012).Temporalshiftsintheskinmicrobiomeassociatedwithdiseaseflaresandtreatmentinchildrenwithatopicdermatitis.Genomeresearch,22(5),850-859.Kong,H.H.,&Segre,J.A.(2012).Skinmicrobiome:lookingbacktomoveforward.JournalofInvestigativeDermatology,132(3),933-939.Konishi,K.,Tamura,T.,Zenitani,R.,Bando,T.,Kato,H.,&Walløe,L.(2008).DeclineinenergystorageintheAntarcticminkewhale(Balaenopterabonaerensis)intheSouthernOcean.PolarBiology,31(12),1509-1520.Koskella,B.,Hall,L.J.,&Metcalf,C.J.E.(2017).Themicrobiomebeyondthehorizonofecologicalandevolutionarytheory.Natureecology&evolution,1(11),1606.Kueneman,J.G.,Parfrey,L.W.,Woodhams,D.C.,Archer,H.M.,Knight,R.,&McKenzie,V.J.(2014).Theamphibianskin-associatedmicrobiomeacrossspecies,spaceandlifehistorystages.MolecularEcology,23(6),1238-1250.Lax,S.,Smith,D.P.,Hampton-Marcell,J.,Owens,S.M.,Handley,K.M.,Scott,N.M.,Gibbons,S.M.,Larsen,P.,Shogan,B.D.,Weiss,S.,&Metcalf,J.L.(2014).Longitudinalanalysisofmicrobialinteractionbetweenhumansandtheindoorenvironment.Science,345(6200),1048-1052.Lemieux-Labonté,V.,Tromas,N.,Shapiro,B.J.,&Lapointe,F.J.(2016).Environmentandhostspeciesshapetheskinmicrobiomeofcaptiveneotropicalbats.PeerJ,4,e2430.Ley,R.E.,Lozupone,C.A.,Hamady,M.,Knight,R.,&Gordon,J.I.(2008).Worldswithinworlds:evolutionofthevertebrategutmicrobiota.NatureReviewsMicrobiology,6(10),776.Leyden,J.J.,McGiley,K.J.,Mills,O.H.,&Kligman,A.M.(1975).Age-relatedchangesintheresidentbacterialfloraofthehumanface.JournalofInvestigativeDermatology,65(4),379-381.Li,H.(2013).Aligningsequencereads,clonesequencesandassemblycontigswithBWA-MEM.arXivpreprintarXiv:1303.3997.Li,H.,&Durbin,R.(2009).FastandaccurateshortreadalignmentwithBurrows–Wheelertransform.Bioinformatics,25(14),1754-1760.Li,H.,Handsaker,B.,Wysoker,A.,Fennell,T.,Ruan,J.,Homer,N.,...&Durbin,R.(2009).Thesequencealignment/mapformatandSAMtools.Bioinformatics,25(16),2078-2079.Li,H.,Qu,J.,Li,T.,Li,J.,Lin,Q.,&Li,X.(2016).Pikapopulationdensityisassociatedwiththecompositionanddiversityofgutmicrobiota.Frontiersinmicrobiology,7,758.Lusk,R.W.(2014).Diverseandwidespreadcontaminationevidentintheunmappeddepthsofhighthroughputsequencingdata.PloSone,9(10),e110808.MangulS.,MOldeLoohuisL.,OriA.,JospinG.,KoslickiD.,YangH.T.,WuT.,BoksM.P.,Lomen-HoerthC.,Wiedau-PazosM,…&Ophoff,R.A.(2016)TotalRNASequencingrevealsmicrobialcommunitiesinhumanbloodanddiseasespecificeffects.bioRxiv.057570

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 39: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

39

Matkin,C.O.,Barrett-Lennard,L.G.,Yurk,H.,Ellifrit,D.,&Trites,A.W.(2007).Ecotypicvariationandpredatorybehavioramongkillerwhales(Orcinusorca)offtheeasternAleutianIslands,Alaska.FisheryBulletin,105(1),74-88.McFall-Ngai,M.J.,Henderson,B.,&Ruby,E.G.(Eds.).(2005).Theinfluenceofcooperativebacteriaonanimalhostbiology(Vol.9).CambridgeUniversityPress.McKenzie,V.J.,Bowers,R.M.,Fierer,N.,Knight,R.,&Lauber,C.L.(2012).Co-habitingamphibianspeciesharboruniqueskinbacterialcommunitiesinwildpopulations.TheISMEjournal,6(3),588.Li,D.,Liu,C.M.,Luo,R.,Sadakane,K.,&Lam,T.W.(2015).MEGAHIT:anultra-fastsingle-nodesolutionforlargeandcomplexmetagenomicsassemblyviasuccinctdeBruijngraph.Bioinformatics,31(10),1674-1676.MartinezArbizu,P.(2017).pairwiseAdonis:Pairwisemultilevelcomparisonusingadonis.Rpackageversion0.0.1.Menke,S.,Melzheimer,J.,Thalwitzer,S.,Heinrich,S.,Wachter,B.,&Sommer,S.(2017).Gutmicrobiomesoffree-rangingandcaptiveNamibiancheetahs:diversity,putativefunctions,andoccurrenceofpotentialpathogens.Molecularecology.Mollerup,S.,Friis-Nielsen,J.,Vinner,L.,Hansen,T.A.,Richter,S.R.,Fridholm,H.,...&Mourier,T.(2016).Propionibacteriumacnes:disease-causingagentorcommoncontaminant?Detectionindiversepatientsamplesbynext-generationsequencing.Journalofclinicalmicrobiology,54(4),980-987.Morin,P.A.,Archer,F.I.,Foote,A.D.,Vilstrup,J.,Allen,E.E.,Wade,P.,...&Bouffard,P.(2010).Completemitochondrialgenomephylogeographicanalysisofkillerwhales(Orcinusorca)indicatesmultiplespecies.Genomeresearch,20(7),908-916.Morueta-Holme,N.,Blonder,B.,Sandel,B.,McGill,B.J.,Peet,R.K.,Ott,J.E.,...&Svenning,J.C.(2016).Anetworkapproachforinferringspeciesassociationsfromco-occurrencedata.Ecography,39(12),1139-1150.Moura,A.E.,JansevanRensburg,C.,Pilot,M.,Tehrani,A.,Best,P.B.,Thornton,M.,...&Dahlheim,M.E.(2014).KillerwhalenucleargenomeandmtDNArevealwidespreadpopulationbottleneckduringthelastglacialmaximum.Molecularbiologyandevolution,31(5),1121-1131.Mukherjee,S.,Huntemann,M.,Ivanova,N.,Kyrpides,N.C.,&Pati,A.(2015).Large-scalecontaminationofmicrobialisolategenomesbyIlluminaPhiXcontrol.Standardsingenomicsciences,10(1),18.Naccache,S.N.,Greninger,A.L.,Lee,D.,Coffey,L.L.,Phan,T.,Rein-Weston,A.,...&Chiu,C.Y.(2013).Theperilsofpathogendiscovery:originofanovelparvovirus-likehybridgenometracedtonucleicacidextractionspincolumns.Journalofvirology,87(22),11966-11977.Nater,A.,Mattle-Greminger,M.P.,Nurcahyo,A.,Nowak,M.G.,deManuel,M.,Desai,T.,...&Lameira,A.R.(2017).Morphometric,behavioral,andgenomicevidenceforanewOrangutanspecies.CurrentBiology,27(22),3487-3498.Nelson,T.M.,Apprill,A.,Mann,J.,Rogers,T.L.,&Brown,M.V.(2015).Themarinemammalmicrobiome:currentknowledgeandfuturedirections.MicrobiologyAustralia,36(1),8-13.Oh, J., Byrd, A. L., Deming, C., Conlan, S., Barnabas, B., Blakesley, R., … Segre, J. A. (2014). Biogeography and individuality shape function in the human skin metagenome. Nature, 514(7520), 59–64. https://doi.org/10.1038/nature13786 Oksanen, J., Guillaume Blanchet, F., Kindt, R., & Legendre, P. (2017). others. 2016. vegan: Community ecology package. R package version 2.3–5.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 40: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

40

Overbeek, R., Begley, T., Butler, R. M., Choudhuri, J. V., Chuang, H. Y., Cohoon, M., ... & Fonstein, M. (2005). The subsystems approach to genome annotation and its use in the project to annotate 1000 genomes. Nucleic acids research, 33(17), 5691-5702.Palsbøll,P.J.,F.Larsen,AndE.S.Hansen.1991.Samplingofskinbiopsiesfromfree-ranginglargecetaceansinWestGreenland;Developmentofnewbiopsytipsandboltdesigns.Rep.Int.WhalingComm.,SpecialIssue,13:71–79.Parker,J.,Helmstetter,A.J.,Devey,D.,Wilkinson,T.,&Papadopulos,A.S.(2017).Field-basedspeciesidentificationofclosely-relatedplantsusingreal-timenanoporesequencing.Scientificreports,7(1),8345.Parsons,K.M.,Durban,J.W.,Burdin,A.M.,Burkanov,V.N.,Pitman,R.L.,Barlow,J.,...&Wade,P.R.(2013).GeographicpatternsofgeneticdifferentiationamongkillerwhalesinthenorthernNorthPacific.JournalofHeredity,104(6),737-754.Paulson,J.N.,Stine,O.C.,Bravo,H.C.,&Pop,M.(2013).Differentialabundanceanalysisformicrobialmarker-genesurveys.Naturemethods,10(12),1200.Petersen,T.N.,Lukjancenko,O.,Thomsen,M.C.F.,Sperotto,M.M.,Lund,O.,Aarestrup,F.M.,&Sicheritz-Pontén,T.(2017).MGmapper:Referencebasedmappingandtaxonomyannotationofmetagenomicssequencereads.PloSone,12(5),e0176469.Phillips,C.D.,Phelan,G.,Dowd,S.E.,McDonough,M.M.,Ferguson,A.W.,DeltonHanson,J.,Siles,L.,Ordóñez-Garza,N.I.C.T.É.,SanFrancisco,M.&Baker,R.J.(2012).Microbiomeanalysisamongbatsdescribesinfluencesofhostphylogeny,lifehistory,physiologyandgeography.Molecularecology,21(11),2617-2627.Piñeiro-Vidal,M.,Gijón,D.,Zarza,C.,&Santos,Y.(2012).Tenacibaculumdicentrarchisp.nov.,amarinebacteriumofthefamilyFlavobacteriaceaeisolatedfromEuropeanseabass.Internationaljournalofsystematicandevolutionarymicrobiology,62(2),425-429.Pitman,R.L.,&Durban,J.W.(2010).KillerwhalepredationonpenguinsinAntarctica.PolarBiology,33(11),1589-1594.Pitman,R.L.,&Durban,J.W.(2012).Cooperativehuntingbehavior,preyselectivityandpreyhandlingbypackicekillerwhales(Orcinusorca),typeB,inAntarcticPeninsulawaters.MarineMammalScience,28(1),16-36.Pitman,R.L.,&Ensor,P.(2003).Threeformsofkillerwhales(Orcinusorca)inAntarcticwaters.JournalofCetaceanResearchandManagement,5(2),131-140.Pitman,R.L.,Fearnbach,H.&Durban,J.W.(2018).AbundanceandpopulationstatusofRossSeakillerwhales(Orcinusorca,typeC)inMcMurdoSound,Antarctica:EvidenceforimpactbycommercialfishingintheRossSea?PolarBiologyDOI:/10.1007/s00300-017-2239-4Poelstra,J.W.,Vijay,N.,Bossu,C.M.,Lantz,H.,Ryll,B.,Müller,I.,...&Wolf,J.B.(2014).Thegenomiclandscapeunderlyingphenotypicintegrityinthefaceofgeneflowincrows.Science,344(6190),1410-1414.QuastC,PruesseE,YilmazP,GerkenJ,SchweerT,YarzaP,PepliesJ,GlöcknerFO(2013)TheSILVAribosomalRNAgenedatabaseproject:improveddataprocessingandweb-basedtools.OpensexternallinkinnewwindowNucl.AcidsRes.41(D1):D590-D596.Quince,C.,Walker,A.W.,Simpson,J.T.,Loman,N.J.,&Segata,N.(2017).Shotgunmetagenomics,fromsamplingtoanalysis.Naturebiotechnology,35(9),833.Quinlan,A.R.,&Hall,I.M.(2010).BEDTools:aflexiblesuiteofutilitiesforcomparinggenomicfeatures.Bioinformatics,26(6),841-842.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 41: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

41

Ranjan,R.,Rani,A.,Metwally,A.,McGee,H.S.,&Perkins,D.L.(2016).Analysisofthemicrobiome:advantagesofwholegenomeshotgunversus16Sampliconsequencing.Biochemicalandbiophysicalresearchcommunications,469(4),967-977.Rohland,N.,&Reich,D.(2012).Cost-effective,high-throughputDNAsequencinglibrariesformultiplexedtargetcapture.Genomeresearch,22(5),939-946.Romano-Bertrand,S.,Licznar-Fajardo,P.,Parer,S.,&Jumas-Bilak,E.(2015).Impactdel’environnementsurlesmicrobiotes:focussurl’hospitalisationetlesmicrobiotescutanésetchirurgicaux.RevueFrancophonedesLaboratoires,2015(469),75-82.Rothschild,D.,Weissbrod,O.,Barkan,E.,Korem,T.,Zeevi,D.,Costea,P.I.,...&Pevsner-Fischer,M.(2018).Environmentalfactorsdominateoverhostgeneticsinshapinghumangutmicrobiotacomposition.Naturedoi:10.1038/nature25973Salazar,G.,&Sunagawa,S.(2017).Marinemicrobialdiversity.MolecularEcology,27,R431-R510.Salter,S.J.,Cox,M.J.,Turek,E.M.,Calus,S.T.,Cookson,W.O.,Moffatt,M.F.,...&Walker,A.W.(2014).Reagentandlaboratorycontaminationcancriticallyimpactsequence-basedmicrobiomeanalyses.BMCbiology,12(1),87.DerSarkissian,C.,Ermini,L.,Schubert,M.,Yang,M.A.,Librado,P.,Fumagalli,M.,...&Petersen,B.(2015).EvolutionarygenomicsandconservationoftheendangeredPrzewalski’shorse.CurrentBiology,25(19),2577-2583.Saulitis,E.,Matkin,C.,Barrett-Lennard,L.,Heise,K.,&Ellis,G.(2000).Foragingstrategiesofsympatrickillerwhale(Orcinusorca)populationsinPrinceWilliamSound,Alaska.Marinemammalscience,16(1),94-109.Scharschmidt,T.C.,&Fischbach,M.A.(2013).Whatlivesonourskin:ecology,genomicsandtherapeuticopportunitiesoftheskinmicrobiome.DrugDiscoveryToday:DiseaseMechanisms,10(3-4),e83-e89.Schmieder,R.,&Edwards,R.(2011).Qualitycontrolandpreprocessingofmetagenomicdatasets.Bioinformatics,27(6),863-864.Schubert,M.,Lindgreen,S.,&Orlando,L.(2016).AdapterRemovalv2:rapidadaptertrimming,identification,andreadmerging.BMCresearchnotes,9(1),88.Segata,N.,Izard,J.,Waldron,L.,Gevers,D.,Miropolsky,L.,Garrett,W.S.,&Huttenhower,C.(2011).Metagenomicbiomarkerdiscoveryandexplanation.Genomebiology,12(6),R60.ShottsJr,E.B.,Albert,T.F.,Wooley,R.E.,&Brown,J.(1990).Microfloraassociatedwiththeskinofthebowheadwhale(Balaenamysticetus).JournalofWildlifeDisease,26,351-359.Song,S.J.,Lauber,C.,Costello,E.K.,Lozupone,C.A.,Humphrey,G.,Berg-Lyons,D.,Caporaso,J.G.,Knights,D.,Clemente,J.C.,Nakielny,S.&Gordon,J.I.(2013).Cohabitingfamilymemberssharemicrobiotawithoneanotherandwiththeirdogs.elife,2.Sunagawa,S.,Coelho,L.P.,Chaffron,S.,Kultima,J.R.,Labadie,K.,Salazar,G.,Djahanschiri,B.,Zeller,G.,Mende,D.R.,Alberti,A.,etal.(2015).Structureandfunctionoftheglobaloceanmicrobiome.Science348,1261359.Sun,C.,Fu,G.Y.,Zhang,C.Y.,Hu,J.,Xu,L.,Wang,R.J.,...&Zhang,X.Q.(2016).IsolationandcompletegenomesequenceofAlgibacteralginolyticasp.nov.,anovelseaweed-degradingBacteroidetesbacteriumwithdiverseputativepolysaccharideutilizationloci.Appliedandenvironmentalmicrobiology,82(10),2975-2987.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 42: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

42

Tessler,M.,Neumann,J.S.,Afshinnekoo,E.,Pineda,M.,Hersch,R.,Velho,L.F.M.,...&Mason,C.E.(2017).Large-scaledifferencesinmicrobialbiodiversitydiscoverybetween16Sampliconandshotgunsequencing.Scientificreports,7(1),6589.Tung,J.,Barreiro,L.B.,Burns,M.B.,Grenier,J.C.,Lynch,J.,Grieneisen,L.E.,...&Archie,E.A.(2015).Socialnetworkspredictgutmicrobiomecompositioninwildbaboons.Elife,4.Vågene,Å.J.,Herbig,A.,Campana,M.G.,García,N.M.R.,Warinner,C.,Sabin,S.,...&Bos,K.I.(2018).Salmonellaentericagenomesfromvictimsofamajorsixteenth-centuryepidemicinMexico.NatureEcology&Evolution,2,520.vanderValkT,VezziF,OrmestadM,DalénL,GuschanskiK.LowrateofindexhoppingontheIlluminaHiSeqXplatform.Biotechniques.BioRxiv,doi:10.1101/179028).Walke,J.B.,Becker,M.H.,Loftus,S.C.,House,L.L.,Cormier,G.,Jensen,R.V.,&Belden,L.K.(2014).Amphibianskinmayselectforrareenvironmentalmicrobes.TheISMEjournal,8(11),2207.Warinner,C.,Herbig,A.,Mann,A.,FellowsYates,J.A.,Weiß,C.L.,Burbano,H.A.,...&Krause,J.(2017).Arobustframeworkformicrobialarchaeology.Annualreviewofgenomicsandhumangenetics,18,321-356.Wright,E.S.,&Vetsigian,K.H.(2016).Inhibitoryinteractionspromotefrequentbistabilityamongcompetingbacteria.Naturecommunications,7,11274.Wright,E.S.,&Vetsigian,K.H.(2016).QualityfilteringofIlluminaindexreadsmitigatessamplecross-talk.BMCgenomics,17(1),876.Wolz,M.,Carly,R.,Yarwood,S.A.,Grant,E.H.C.,Fleischer,R.C.,&Lips,K.R.(2017).EffectsofhostspeciesandenvironmentontheskinmicrobiomeofPlethodontidsalamanders.JournalofAnimalEcology.Ying,S.,Zeng,D.N.,Chi,L.,Tan,Y.,Galzote,C.,Cardona,C.,Lax,S.,Gilbert,J.&Quan,Z.X.(2015).Theinfluenceofageandgenderonskin-associatedmicrobialcommunitiesinurbanandruralhumanpopulations.PloSone,10(10),e0141842.ZhangC.,ClevelandK.,Schnoll-SussmanF.,McClureB.,BiggM.,ThakkarP.,SchultzN.,ShahM.A.,BetelD(2015)Identificationoflowabundancemicrobiomeinclinicalsamplesusingwholegenomesequencing.GenomeBiology,16(1).DataAccessibilityAllsequencingdataarearchivedattheEuropeanNucleotideArchive,www.ebi.ac.uk/ena,accessionnumbers:ERS554424-ERS554471.ScriptsandmetadataareavailableinDryadrepositorydoi:10.5061/dryad.c8v3rv6.AuthorContributionsR.H., J.B., T.V. and A.A. analysed the data; J.W.D. and H.F. conducted the photographicgrading.A.F.andK.G. conceivedandcoordinated the study,whichwasdeveloped fromasuggestionbyGeraldPaoof theSalk Institute. J.W.D.,H.F.,R.W.B.,M.B.H.andP.W.wereinvolvedinsamplecollectionandDNAwasextractedbyK.M.R.R.H.,J.B.,A.F.,K.G.wrotethemanuscriptwithinputfromT.V.,A.A.,J.W.D.,H.F.,R.W.B.,M.T.P.G.,P.A.M.andJ.B.W.W.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 43: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

43

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 44: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

44

FIGURES

FIGURE1.Summaryofgeographicsampleoriginsandmicrobialdiversitymeasures.Mapofsampling locations of the 49 samples of five killer whale ecotypes, from which skinmicrobiomeswereincludedinthisstudy.TheAntarcticecotypesprimarilyinhabitwaters8–16°CcolderthantheNorthPacificecotypes.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 45: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

45

Figure 2. Composition of the wild killer whale skin microbiomes and other publishedmicrobiomes, for sampleswith≥50 taxonomyassigned16S reads. Principal coordinatesanalysisofJaccardbinarypresence/absencedistancesbefore(a)andafter(b)filteringofC.acnesassociatedtaxafromthewildkillerwhaledata.Proportionsofsourcescontributingtoeachkillerwhalesample,representedbycolumns,fromSourceTrackeranalysisbefore(c)andafter (d) filteringofC.acnesassociatedtaxa.* in (c)denotessamplesthatwereexcludedafterC.acnesfilteringduetolowreadnumbers.

* *

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 46: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

46

FIGURE3. (a)Proportionofdrivingbacteriaper individual,afterdata filtering. Individuals,representedbycolumns,aregroupedbyecotype,andtherelativeproportionsofbacterialtaxa are indicatedby column shading (1,Tenacibaculumdicentrarchi; 2,Paraburkholderiafungorum; 3, Pseudoalteromonas haloplanktis; 4, Pseudoalteromonas translucida; 5,Acinetobacter johnsonii; 6, Pseudomonas stutzeri; 7, Stenotrophomonas maltophilia; 8,Kocuriapalustris;9,Other).(b)BetadiversitybetweenecotypesillustratedasaBray-CurtisPCoA estimated from read counts. (c) Positive co-occurrence network built from a co-occurrencematrixofallspecies,subsettedtothe8drivingtaxa(blacknodesnumberedasabove) and their top 20 positive and significant co-occurring species.Only specieswith asignificantco-occurrencescoreof>800areshown.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 47: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

47

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 48: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

48

FIGURE4Theinfluenceofdiatomabundanceonskinmicrobiomecommunitycompositionandmicrobial functional profiles. (a) Relative diatomabundance is significantly higher inAntarctic killerwhales thanNorth Pacific, but this is largely driven by a subset of outlierAntarcticindividuals.(b)WithinAntarctictypeB1andtypeB2specimens,therelativediatomabundanceissignificantlyassociatedwithskincolourationofthehostkillerwhale,withtheyellowishhuebeingareliableindicatorofdiatomload.InsetimagesareofthesametypeB2killerwhaleindividualdisplayingextremevariabilityindiatomcoverage,bothphotographsbyJohnDurban.Relativediatomabundanceissignificantlyassociatedwith(c)thepresenceofTenacibaculum dicentrarchi and (d) several algae-associated bacteria, including T.dicentrarchi. (e) PCA of variation in functional COGs between individuals, coloured by T.dicentrarchiabundance.IndividualswithhighrelativeabundancesofT.dicentrarchigenerallycluster with high values in principal component 2. The top 10 COGs contributing to PCAvariationareshowningreyarrows(J:translation,ribosomalstructureandbiogenesis,I:lipidtransportandmetabolism,F:nucleotidetransportandmetabolism,H:coenzymetransportandmetabolism,U:intracellulartrafficking,secretionandvesiculartransport,N:cellmotility,P: inorganiciontransportandmetabolism,T:signaltransductionmechanisms,M:cellwallmembraneenvelopebiogenesis,G:carbohydratetransportandmetabolism).(f)PhotographofatypeB1killerwhaleintheGerlacheStraitoftheAntarcticPeninsulaonthe4thDecember2015withhighdiatomcoverageandpoorskincondition.PhotographbyConorRyan.

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint

Page 49: Host-derived population genomics data provides …Host-derived population genomics data provides insights into bacterial and diatom composition of the killer whale skin Rebecca Hooper1*,

49

TABLESTable1.Factorsinfluencingthekillerwhaleskinmicrobiome.ResultsofAdonismodelsusinggenome-size controlled species data. (a) Bray Curtismodelwith library size included as acovariate; (b)TSSnormalisedBrayCurtismodel; (c)Binary Jaccardmodelwith librarysizeincludedasacovariate.Significantfactorsarehighlightedinbold.

(a) BrayCurtis(b)BrayCurtis

(TSSnormalised) (c)BinaryJaccard

F r2 p F r2 p F r2 p

Latitude 1.8 0.04 0.01 2.35 0.05 <0.01 1.4 0.03 0.05Longitude 0.89 0.02 0.62 0.89 0.02 0.61 0.99 0.02 0.45Ecotype 1.36 0.11 <0.01 1.35 0.11 0.02 1.23 0.10 0.03

Librarysize 1.33 0.03 <0.01 - - - 1.20 0.02 0.23Humancontamination 1.13 0.02 0.35 0.59 0.01 0.89 1.04 0.02 0.42

Residuals 0.79 0.81 0.80

Total 1 1 1

.CC-BY-NC-ND 4.0 International licensenot certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which wasthis version posted July 26, 2018. . https://doi.org/10.1101/282038doi: bioRxiv preprint