The influence of landscape, climate and history on spatial ... · 1999). Marion Island is a...

Molecular Ecology. 2019;28:3291–3305. wileyonlinelibrary.com/journal/mec  | 3291© 2019 John Wiley & Sons Ltd

Received:2April2019  |  Revised:13May2019  |  Accepted:14May2019DOI:10.1111/mec.15147

O R I G I N A L A R T I C L E

The influence of landscape, climate and history on spatial genetic patterns in keystone plants (Azorella) on sub‐Antarctic islands

John H. Chau1  | Céline Born2 | Melodie A. McGeoch3 | Dana Bergstrom4,5 | Justine Shaw6 | Aleks Terauds4 | Mario Mairal2 | Johannes J. Le Roux7 | Bettine Jansen van Vuuren1

1DepartmentofZoology,CentreforEcologicalGenomicsandWildlifeConservation,UniversityofJohannesburg,AucklandPark,SouthAfrica2DepartmentofBotanyandZoology,StellenboschUniversity,Stellenbosch,SouthAfrica3EcologyResearchGroup,SchoolofBiologicalSciences,MonashUniversity,Melbourne,Vic.,Australia4AustralianAntarcticDivision,Kingston,Tas.,Australia5GlobalChallengesProgram,UniversityofWollongong,Wollongong,NSW,Australia6EnvironmentalDecisionGroup,SchoolofBiologicalSciences,UniversityofQueensland,Brisbane,Qld.,Australia7DepartmentofBiologicalSciences,MacquarieUniversity,Sydney,NSW,Australia

CorrespondenceBettineJansenvanVuuren,DepartmentofZoology,CentreforEcologicalGenomicsandWildlifeConservation,UniversityofJohannesburg,AucklandPark,SouthAfrica.Email:bettinevv@uj.ac.za

Funding informationNationalResearchFoundation,Grant/AwardNumber:GUN110728;SouthAfricanNationalResearchFoundation;UniversityofJohannesburg;StellenboschUniversity;AXAResearchFund;SouthAfricanDepartmentofEnvironmentalAffairsandTourism:AntarcticaandIslandsthroughtheSouthAfricanNationalAntarcticProgram;AustralianAntarcticDivision;DepartmentofSustainability,Environment,Water,PeopleandCommunity,Australia;TasmanianParksandWildlifeService

AbstractThedistributionofgeneticvariation inspecies isgovernedbyfactorsthatactdif-ferentlyacrossspatialscales.Toteaseapartthecontributionofdifferentprocesses,especiallyatintermediatespatialscales,itisusefultostudysimpleecosystemssuchasthoseonsub‐Antarcticoceanicislands.Inthisstudy,wecharacterizespatialge-neticpatternsoftwokeystoneplantspecies,Azorella selagoonsub‐AntarcticMarionIslandandAzorella macquariensisonsub‐AntarcticMacquarieIsland.Althoughbothislandsexperienceasimilarclimateandhaveasimilarvegetationstructure,theydif-fersignificantlyintopographyandgeologicalhistory.Wegenotypedsixmicrosatel-lites for1,149 individuals from123sitesacrossMarion Islandand372 individualsfrom42sitesacrossMacquarieIsland.Wetestedforspatialpatternsingeneticdi-versity, includingcorrelationwithelevationandvegetation type, andclines indif-ferentdirectionalbearings.Wealsoexaminedgeneticdifferentiationwithinislands,isolation‐by‐distancewithandwithoutaccountingfordirection,andsignalsofde-mographicchange.MarionIslandwasfoundtohaveadistinctnorthwest–southeastdivide,withlowergeneticdiversityandmoresiteswithasignalofpopulationexpan-sioninthenorthwest.Weattributethistoasymmetricseeddispersalbythedomi-nantnorthwesterlywinds,andtopopulationpersistenceinasouthwesternrefugiumduringtheLastGlacialMaximum.Noapparentspatialpattern,butgreatergeneticdiversityanddifferentiationbetweensites,wasfoundonMacquarie Island,whichmaybeduetothenarrowlengthoftheislandinthedirectionofthedominantwinds

3292  |     CHAU et Al.

1  | INTRODUC TION

Distributionsofspeciesareshapedbyvariousfactorsincludingen-vironmental properties, biological interactions and species traits,whichmayactdifferentlyacrossspatialscales.Atlargespatialscales,long‐distance dispersal, geological processes and physical barriersshape the biogeography of species (González‐Wevar et al., 2018;Lee et al., 2017; Postaire, Gélin, Bruggemann, & Magalon, 2017;Sanmartín&Ronquist,2004;Whittaker,Triantis,&Ladle,2008).Atfinerspatialscales,localdispersalbecomesincreasinglyimportant,drivenprimarilybythebiologyofthespeciesandthelocalenviron-mentasexperiencedbyindividuals(LaRue,Holland,&Emery,2018;Manel,Schwartz,Luikart,&Taberlet,2003).Incontrast,therelativecontributionsofprocessesshapingdistributionsatintermediatespa-tialscales(typicallytenstohundredsofkilometres)aremoredifficulttoteaseapart,asdispersalislessfrequentandlandscapesbecomespatiallyheterogeneous(Lenoiretal.,2012;Linetal.,2013;Mertes& Jetz, 2018). To this end, scale‐dependent approaches are oftenneededtofullyappreciatetheimportanceofdispersalongeneticdi-versitypatterns.Whiledispersalinanimalspeciesdependsprimarilyonthemovementandbehaviourofindividuals,forplantsitcanbemorecomplex; forexample,patternsofgeneflowat intermediatespatial scalesmayreflectdifferences inpollenversusseeddisper-sal dynamics (Dick, Hardy, Jones, & Petit, 2008; Hamilton, 1999;Loveless&Hamrick,1984).

Giventhecomplexityoftheprocessesthatunderliespeciesdistri-butionsatintermediatespatialscales,itisofteneasiertodisentanglethecontributionsoffactorsinfluencingthespatialdynamicsofspeciesinsimplerecosystems.Theecosystemsofsub‐Antarcticislandspro-videgoodmodelsforinvestigatingevolutionaryprocessesatinterme-diatescalesastheytypicallycompriseasmallnumberofspecieswitha simple trophic structure (Chown&Convey,2007), andyet islandlandscapes are sufficiently complex tohavebothbiotic and abioticheterogeneity (Bergstrom&Chown, 1999). Formost sub‐Antarcticislands,naturalcolonizationresultsfromlow‐frequencylong‐distancedispersalevents(Hardouinetal.,2010;JansenvanVuuren&Chown,2007;Kalwij,Medan,Kellermann,Greve,&Chown,2019;Mortimer,JansenvanVuuren,Meiklejohn,&Chown,2012;Stevens,Greenslade,Hogg,&Sunnucks,2006),whichmayactasastronghomogenizingforce on intra‐island genetic diversity. Long‐distance dispersal alsoplaysanimportantroleinshapingthebiogeographicalpatternsofthe

region'sindigenousflora,withmoresimilarflorasoccurringonislandsthataregeographicallycloser (Greve,Gremmen,Gaston,&Chown,2005;Shaw,Spear,Greve,&Chown,2010;Wace,1960).Atsmallspa-tialscales(<3km),anisotropic(i.e.,directionallydependent)short‐dis-tancedispersalhasbeenfoundtoplayapivotal role inshapingthegeneticdiversityofthecushionplantAzorella selagoonMarionIsland(Born,LeRoux,Spohr,McGeoch,&JansenvanVuuren,2012),withthedispersaldynamics largelydependentonthedirectionandstrengthof localprevailingwinds.Atthe intermediate,or island,scale,thesepatternsaremoredifficulttointerpret.Forexample,studiesoninver-tebratespeciesonMarionIsland(Grobler,Bastos,Treasure,&Chown,2011;McGaughran,Convey, Stevens,&Chown,2010;Mortimer&JansenvanVuuren,2007;Mortimeretal.,2012;Myburgh,Chown,Daniels,&JansenvanVuuren,2007)haverevealedcomplexgeneticpatterns,whichhintat the interplayofvariousprocesses, includingpastglaciations,volcanicactivity,persistenceinrefugiaandbarrierstodispersalbygeologicalstructures.

Inthisstudy,wefocusontwocongenerickeystoneplantspeciesinthesub‐Antarctic,A. selagoHook.f.(Apiaceae)onMarionIslandandAzorella macquariensisOrchardonMacquarieIsland.Bothspeciesarelow‐growing, compactplantswithacushiongrowth form.Theyplayanimportantroleaspioneerspecies,colonizingunstablescoriaceous,gravelandpeatslopes,recentlavaflows,andtheforelandsofretreat-ingglaciers.Theyareconsideredkeystonespeciesbecausetheyhostdiverseepiphyteand invertebratecommunitiesand facilitate theoc-currenceofmanyofthesespeciesathigherelevations(Bergstrometal.,2015;Hugo,McGeoch,Marshall,&Chown,2004;McGeoch,LeRoux,Hugo,&Nyakatya,2008;LeRoux&McGeoch,2010).Thestructureofthesmall,flat,lightweightfruitssuggeststhattheyarewind‐dispersed(Haussmann,McGeoch,&Boelhouwers,2010;Orchard,1989).Theirpollination biology is unknown, but the depauperate insect commu-nityontheseislandssuggeststhatwind‐pollinationorselfingislikely(Chown&Marshall,2008;Convey,2007;Lord,2015).Azorella selago iswidely distributedon sub‐Antarctic islands and in southern SouthAmerica.OnMarion Island, it is found inmostvegetation typesandiscommonfromsealevelto~650ma.s.l.(Phiri,McGeoch,&Chown,2009).PalynologicalstudiesconfirmthepresenceofA. selagoatvar-iouslocalitiesacrossMarionIslandverysoonafterthelastglacialre-treat(Schalke&VanZinderenBakker,1971;Scott,1985;Scott&Hall,1983).Whether the species survived in one ormore refugia duringtheLastGlacialMaximum(LGM)orwhetheritrecolonizedtheisland

and longerpopulationpersistencepermittedbythe lackofextensiveglaciationontheisland.Together,ourresultsclearlyillustratetheimplicationsofislandshapeandgeography,andtheimportanceofdirection‐dependentdrivers,inshapingspatialge-neticstructure.

K E Y W O R D S

direction‐dependentdispersal,geneticdiversity,MacquarieIsland,MarionIsland,microsatellites,spatialgeneticstructure

     |  3293CHAU et Al.

fromasourcepopulationonanothersub‐Antarcticisland,orevenfromSouthAmerica,remainsunknown.ApreviousstudybasedonalimitednumberofsamplescollectedmostlyfromcoastalsitesonMarionIslandrevealednogenetic variation for theplastid trnH–psbA region,whilesignificantstructurewasdetectedwithamplifiedfragmentlengthpoly-morphisms(Mortimer,McGeoch,Daniels,&JansenvanVuuren,2008).Giventhesecontrastingresultsandthelimitedsamplesizesincludedintheiranalysis,theseauthorsstressedtheneedforincreasedsamplingeffortbeforedevelopingconclusionsabouttheprocessesthatstruc-turegeneticdiversityinthesepopulations.A. macquariensisisendemictoMacquarieIsland,whereitdominatesfellfieldvegetationontheup-landplateaubutoccasionallyoccursatlowerelevations(Bergstrom&Selkirk,1999;Copson,1984;Selkirk,Seppelt,&Selkirk,1990;Taylor,1955).ItwasrecognizedasadistinctspeciesfromA. selagoduetodif-ferencesinleafandfloralmorphology(Orchard,1989).However,thisdistinctiveness has been questioned (Martinez, 1993), and a recentmolecularphylogenyinfersaclosesisterrelationshipbetweenA. selago and A. macquariensis(Plunkett&Nicolas,2017).

Herewequantifyandcomparethespatialgeneticstructureofthese two keystone plants across Marion andMacquarie islands.Both islands are strongly influencedby theAntarcticCircumpolarCurrentanditsassociatedWest‐WindDrift,buthaveverydifferentgeologicalhistoriesandphysicaldimensions (Bergstrom&Chown,1999).MarionIslandisavolcanicislandlocatedintheIndianOceansector of theSouthernOcean and ismore than1,500 kmdistantfrom the nearest continental landmass (Figure 1a,c). MacquarieIslandislocatedmidwaybetweenAustraliaandAntarcticaandisan

emergedoceanic crust complex (Figure1b,c).Therearemajordif-ferences in the landscapesof the islands:Marion Island is roughlycircular inshape,withastrongelevationalgradientfromitscoast-line to a central plateau with its highest peak reaching 1,230 m;Macquarie Island is a low‐lying ribbon of landwith itsmajor axisrunning roughly north–south and with steep coastal slopes ris-ingtoacentralundulatingplateauwithamaximumheightofonly433m(Figure1).Althoughbothislandsarethoughttobelessthan500,000yearsold(Adamson,Selkirk,Price,Ward,&Selkirk,1996;McDougall,Verwoerd,&Chevallier,2001;Quilty,2007), therearenotabledifferences inglacialhistoriesthatmaytranslate intocon-trasting biogeographical histories for species found on these is-lands.MacquarieIslandwasneverglaciatedtoanysignificantextent(Adamson,Selkirk,&Colhoun,1988;Ledingham,&Pedersen,1984).MarionIsland,ontheotherhand,experiencedintenseglacialperi-odsandwas largelycoveredby iceduringtheLGM(Boelhouwers,Meiklejohn,Holness,&Hedding,2008;Hall,2004;Hall,Meiklejohn,&Bumby,2011).However, ice‐freeareasdidexistduringtheLGM(Halletal.,2011)andmayhaveactedasrefugiaforspecies.

OuraimhereistodocumentspatialgeneticpatternsinAzorella speciesacrosstwosub‐Antarcticoceanic islands,andtoplaceourresultswithin the abiotic setting of these islands.We extensivelysampledAzorellaacrossMarionandMacquarieIslands,andpresentthefirstmolecularecologyresults foranyorganismonMacquarieIsland. Using genotypic data from highly informative microsatel-litemarkers,ouraimswereto:(a)assessspatialpatternsingeneticdiversity, genetic differentiation, spatial genetic structure and

F I G U R E 1  GeographicalorientationofMarionandMacquarieislands.Topographyof(a)MarionIslandand(b)MacquarieIsland,and(c)theirlocationsintheSouthernOcean.ThepositionoftheAntarcticPolarFrontalZoneisindicatedbythedashedline.ElevationdataprovidedbyDavidHedding(UniversityofSouthAfrica)forMarionIslandandtheAustralianAntarcticDivisionforMacquarieIsland,andworldcoastlinedataprovidedbyNaturalEarth

(a) (b)

(c)

3294  |     CHAU et Al.

demographic change inAzorella on the two islands; and (b) evalu-ateinteractionsofthesegeneticpatternswithcurrentandhistoricalfactors,includingwindpatterns,topography,glaciationhistory,ele-vationandvegetationtype.

2  | MATERIAL S AND METHODS

2.1 | Study system

Marion Island (46.9°S, 37.8°E) is part of thePrinceEdward IslandarchipelagoalongwiththesmallerPrinceEdwardIsland,andissit-uatedwithin the Polar Frontal Zone of theAntarctic Circumpolar

Current (Figure1).Roughlyelliptical inshape,withthe longeraxisorientedeast–west,Marion Island is290km2 inarea,withcoastalplains, ridges and valleys surrounding central highlands (Chown,Gremmen,&Gaston,1998;Hänel&Chown,1998).TheclimateonMarionIslandishyperoceanic,characterizedbylowbutstabletem-peratures (meantemperature~6°C,meandailytemperaturerange<3°C,mean seasonal temperature range ~4°C), high precipitation(2,000–3,000mmannually)andhumidity(~80%),andnearcompletecloudcoveronmostdays (Rouault,Mélice,Reason,&Lutjeharms,2005; Le Roux, 2008). The dominant wind direction is from thewest,asexpectedfromtheisland'slocationinabeltofstronglarge‐scalewesterlyatmosphericcirculation.Northwesterlywindsare,on

F I G U R E 2  SamplingsitesandgeneticdiversityofAzorella selagoonMarionIslandandAzorella macquariensisonMacquarieIsland.Locationandexpectedheterozygosity(HE)ofsamplingsiteson(a)MarionIslandand(b)MacquarieIsland.(c)ForMarionIsland,valuesofR2fromlinearregressionsbetweenHE or AR(numberofallelesadjustedbyrarefactionto10samplespersite)andmeasuresofsitepositiontransformedbydirectionalbearings.Filledsymbolsindicatebearingswithasignificantcorrelation(p<.05).Sitepositionsweretransformedusingtheequation:L = x(cosθ)–y(sinθ),wherex and ycorrespondtothelongitudeandlatitudeofeachsite,respectively,andθcorrespondsto(d)18fixedbearingsfrom0°to170°at10°intervals,with0°correspondingtotheeast–westdirection.The120°bearingishighlighted.(e)ForMarionIsland,forthe120°bearing,whichhadthehighestR2value,scatterplotshowingthecorrelationbetweenHEandtransformedmeasuresofsiteposition,andregressionline[Colourfigurecanbeviewedatwileyonlinelibrary.com]

(a) (b)

(d)

(c) (e)

     |  3295CHAU et Al.

average,thestrongest(>10m/s)followedbysouth‐westerlywinds(7m/s)(Rouaultetal.,2005;LeRoux,2008).Theislandfrequentlyexperiences gale forcewinds (>15m/s).Marion Island's terrestrialvegetationhasbeenclassified intovarious types includingcoastalvegetation,bioticherbfield,mire,drainage lineandfernbrakeveg-etationinlowlands,andfellfieldandpolardesertathigherelevations(Gremmen&Smith,2008;Huntley,1971;Smith&Mucina,2006).Azorella selagooccursinmostvegetationtypesandiscommonfromsealevelto~670ma.s.l.(Phirietal.,2009).

MacquarieIsland(54.6°S,158.9°E)isabout1,000kmsouthwestofNewZealandandissituatedtothenorthofthePolarFrontalZone(Figure1).Itformsanarrowrectangle,+34kmlongand5kmwide,withthelongaxisorientednearlynorth–south.Theislandis128km2 in area and consistsmostly of a plateauwhich rises from 150 to433ma.s.l. (Chownet al., 1998;Selkirket al., 1990). LikeMarionIsland,theclimateishyperoceanic,characterizedbylowbutstabletemperatures(mean~6°C,meandailyrange~3.5°C,meanseasonalrange~3.2°C),moderateprecipitation (958mmannually),highhu-midity (~80%)andnearcompletecloudcoveronmostdays.Overtwo‐thirdsofwindsarefromadirectionbetweenwest(~10m/s)andnorthwest(~9m/s),andgaleforcewindsoccuroften(Adams,2009;McGlone,2002).Thevegetationhasbeenclassifiedintograssland,herbfield,mire,fernbrakeandfeldmark(fellfield),withgrasslandandfeldmarkdominatingtheplateauuplands(Selkirketal.,1990;Taylor,1955).Azorella macquariensisismostlyfoundontheplateauat200–400ma.s.l.altitude(Bricher,Lucieer,Shaw,Terauds,&Bergstrom,2013;Selkirketal.,1990),butoccasionallyoccursat lowereleva-tions,usuallyasisolatedcushions.Thespeciesisthedominantvas-cularspeciesinfeldmarkhabitatsonMacquarieIsland(Bergstrom&Selkirk,1999;Copson,1984;Taylor,1955).

2.2 | Sampling

LeafsamplesofA. selagoweretakenfrommultipleindividualsatsitesacrossMarion Island (1,149 individuals from123sites) (Figure2a),and leaf samplesofA. macquariensiswere taken frommultiple in-dividualsatsitesacrossMacquarie Island (372 individuals from42sites)(Figure2b).Sitesweresampledopportunisticallytocoverthespecies’distributionsacrosseach islandandspan theirelevationalranges.GeographicalpositionandelevationofeachcollectionsitewererecordedwithaGarmineTrexVistaorTrimbleDifferentialGPSunit.Leafsamplesweredriedandconservedonsilicagel.

2.3 | DNA extraction, genotyping and data quality control

TotalgenomicDNAwasextractedusingaNucleoSpinPlantIIDNAextractionkit(Macherey‐Nagel).Samplesweregenotypedusingsixspecies‐specific microsatellite markers: azo5, azo6, azo11, azo13,azo17andazo23(MolecularEcologyResourcesPrimerDevelopment,2010).Of theeightpolymorphicmarkersoriginallydescribed, two(azo14andazo21)wereexcludedfromanalysesafterapreliminary

studyshowedtheymaypresentnullalleles.MarkerswereamplifiedfollowingtheprotocolofCerfonteyn,LeRoux,JansenvanVuuren,andBorn (2011).Genotypingwas performedon anABI 3730 au-tomated sequencer (AppliedBiosystems) using theGS500LIZ sizestandard(AppliedBiosystems).

Scoringwas doneusinggenemapper 3.7 (AppliedBiosystems),andthefulldatasetwasdouble‐scoredbytwopeople.Genotypingerrors, including null alleles, stuttering and large allele dropout,wereestimatedwithmicro‐checkerversion2.2.3(VanOosterhout,Hutchinson,Wills,&Shipley,2004).WetestedfordeparturefromHardy–Weinberg equilibrium and for linkage disequilibrium be-tween all pairs of loci using exact tests in genepop 4.7 (Rousset,2008).Inaddition,weassessedtheinformativenessofourmicro-satellite data set by calculating probabilities of identity, a mea-sureofamarker'sabilitytodistinguishindividuals,ingenalex 6.5 (Peakall&Smouse,2012).

2.4 | Genetic diversity

Wecalculated thenumberof allelesper locus (A), observedhete-rozygosity(HO),expectedheterozygosity(HE)andinbreedingindex(FIS) for each site usinggenalex 6.5 (Peakall& Smouse, 2012).Wealsocalculatedthenumberofallelesperlocusaccountingforsam-ple size variation among sites using rarefaction (AR) inhp‐rare 1.1 (Kalinowski, 2005). To determine if these measures were signifi-cantlydifferentbetweenislands,Welcht‐testswereperformedinr version3.5.1(RCoreTeam,2018).

Weassessedwhethergeneticdiversity(HE)wascorrelatedwithelevationand/orvegetationtypeasthesehavebeenshowntoaf-fectthepopulationdensityofAzorellaonsub‐Antarcticislands(Phirietal.,2009).WesuperimposedoursampledsitesonthevegetationmapofSmithandMucina(2006)todeterminethevegetationtype(coastal,mire‐slope,fellfieldorpolardesert)ofeachsiteonMarionIsland.WedidnottestforeffectsofvegetationtypeonMacquarieIslandasAzorellaistypicallyassociatedonlywithfellfieldvegetationthere.Weusedtorocor1.0(Hardy,2009)todetermineifvariablesdisplayed spatial autocorrelation and quantified it usingMoran's I statisticforquantitativevariables(elevationandHE)oritsequivalentforcategoricalvariables(vegetationtype)(Hardy,2009).Eachvari-ablewastestedusingcompleterandomizations(9,999permutationsamong samples). The association between elevation and HE wasquantifiedbyPearson'scorrelationcoefficient,andtheassociationbetweenvegetationtypeandHEbyanintraclasscorrelationcoeffi-cient.Toaccountforspatialautocorrelation,associationsbetweenHEandenvironmentalvariablesweretestedusingtorus‐translationrandomizations (9,999 permutations) on Marion Island, and asso-ciations betweenHE and elevation using complete randomization(9,999permutations)onMacquarieIsland(Harms,Condit,Hubbell,&Foster,2001).Toperformtorus‐translationrandomizations,siteswerepositionedontoeightequalspatialgrids,andwithineachgridsiteswere locatedwithinone tonine transectsdependingon thenumberofsitespergrid.

3296  |     CHAU et Al.

We also assessed whether genetic diversity showed spatialclineswithdirectionalbearingsoneach island.Sitepositionsweretransformedusingtheequation:L = x(cosθ)–y(sinθ),wherex and y correspondto the longitudeand latitudeof thesite inUTM, re-spectively,andθcorrespondsto18fixedbearingsfrom0°to170°at10°intervals,with0°correspondingtotheeast–westorientation(Figure2d).Correlationsbetweentransformedmeasuresofsitepo-sitionforeachbearingandHE or ARwerecalculatedusinglinearre-gressioninmicrosoft excel 2013.Todeterminethedirectionofthecline,weassessedwhichbearingswereassociatedwithasignificantcorrelation(p<.05)andwhichhadthehighestdeterminationcoef-ficient(R2).

2.5 | Genetic differentiation

Weexaminedthepartitioningofgeneticvariationamongandwithinsitesandindividualsforeachislandseparatelybyperforminganaly-sesofmolecularvariance(AMOVAs)with1,000permutationsinar‐lequinversion3.5.2.2(Excoffier&Lischer,2010).Foreachisland,wealsocalculatedpairwiseFSTbetweensites,with999permutationstoassesssignificance,usinggenalex6.5(Peakall&Smouse,2012).

2.6 | Spatial genetic structure

Weassessedspatialgeneticstructure(SGS)oneachislandfollowingtheprocedureofVekemansandHardy(2004)asimplementedinthesoftwarespagedi1.3(Hardy&Vekemans,2002).Givenlimiteddis-persal,geneticdifferentiationbetweensitesisexpectedtoincreasewiththespatialdistancebetweenthem(Hardy&Vekemans,1999;Rousset,1997;Vekemans&Hardy,2004).Toassessthespatialge-neticstructure,valuesofFST/(1−FST)betweensites,ameasureofgenetic differentiation, were regressed on the natural logarithmof the spatial distancebetween sites (ln(dij)) toget the regressionslope(bLd).Totestthesignificanceoftheobservedspatialgeneticstructure values, spatial positions of individuals were permuted9,999 times toobtain the frequencydistributionofbLd under thenullhypothesisthatpairwiseFST/(1−FST)andln(d)areuncorrelated.Tovisualizethespatialstructure,pairwiseFST/(1−FST)valueswereaveraged over a set of distance intervals (d: 0–1, 1–2, 2–3, 3–4,4–5,5–6,6–7,7–8,8–9,9–10,10–12.5,12.5–15,15–17.5,17.5–20,20–25,25–35km)andplottedagainstmeanspatialdistanceineachdistanceinterval.

Bearing analyses were also performed to test for the pres-ence of directional patterns in the spatial genetic structure rela-tionship(Falsetti&Sokal,1993).Apositivecorrelationcoefficient(r)betweenFST/(1−FST)andspatialdistancewasshownforthosebearings(θ)wheregeneflowinthedirectionofthebearingisweak(Bornetal.,2012).Forthisanalysis,thematrix(D)ofthenaturallog-arithmsofspatialdistancesbetweeneachpairofsites(ln(dij))wastransformed into18newmatrices (D0 toD170)byweightingeachdistancebythesquaredcosineoftheangleαij(thearcbetweenthevectorconnectingsitesi and jandareferencevector[θ=0°to170°,

at10°intervalsrotatedanticlockwise,with0°indicatingtheeast–westdirection]).Thistransformationweightseachspatialdistancebyitsalignmentwithatestdirection.RegressionsbetweenadatamatrixofvaluesofFST/(1−FST)betweensitesandD0toD170 ma-triceswereevaluatedviaManteltestswithsignificancedeterminedbypermutationtestsusingpassage2(Rosenberg,2000;Rosenberg&Anderson,2011).

2.7 | Demographic change

Wetestedforevidenceofrecentpopulationexpansionandreduc-tionateachsiteusingtheWilcoxonsignedranktestforheterozy-gositydeficiencyandheterozygosityexcess,respectively (Cornuet& Luikart, 1996;Girod, Vitalis, Leblois, & Fréville, 2011).We per-formed tests in the program bottleneck 1.2.02 (Piry, Luikart, &Cornuet,1999)undertheone‐stepstepwisemutationmodel(SMM)andinfiniteallelemodel(IAM)andwith1,000replicates.

3  | RESULTS

3.1 | Data

For Marion Island, the final data set comprised 1,149 individualsfrom 123 sites (7–10 individuals per site; mean: 9.34) genotypedfor six microsatellite (simple sequence repeat [SSR]) markers. ForMacquarieIsland,thefinaldatasetcomprised372individualsfrom42sites (6–15 individualsper site;mean:8.86)genotyped for fivemicrosatellite(SSR)markers(Figure2a,b;TablesS1andS2).Markerazo13wasexcludedfromthedatasetforMacquarieIslandbecauseitdisplayedahighproportionofnullalleles.Thegenomesofmem-bersofthegenusAzorellaarehighlyconserved,andrelativelyfewpolymorphic markers were found for Azorella (Molecular EcologyResourcesPrimerDevelopment,2010).However,themicrosatellitesusedherewereallhighlyinformativeandreliablefordistinguishingindividualsasestimatedbyprobabilitiesofidentity(p<.05).AllelefrequenciesatmostsitesonMarionandMacquarieislandsconformto those expected under Hardy–Weinberg equilibrium, with onlysixsitesonMarionandninesitesonMacquarieshowingsignificantdepartures fromHardy–Weinberg equilibrium across loci. No sig-nificantlinkagedisequilibriumbetweenlociacrosspopulationswasdetectedforeitherspecies.

3.2 | Genetic diversity and its spatial attributes

Thenumberofallelesperlocusrangedfromtwoto13for Azorella se‐lagoonMarionIsland,andfrom5to11forAzorella macquariensis on MacquarieIsland(Table1).Meanvaluesofgeneticdiversitymeas-ures(A,AR,HO and HE)foreachislandarepresentedinTable2andforeachsiteinTableS1.Expectedheterozygosityrangedfrom.102to.514onMarionIsland(mean±SD=.315±.089)andfrom.167to.511onMacquarieIsland(mean±SD=.377±.073).MeanvaluesofA,AR,HO and HEwereslightlybutsignificantlyhigher(p<.05)forsites

     |  3297CHAU et Al.

onMacquarieIslandthanMarionIsland.Veryfewsitesoneitheroftheseislandsdisplayedinbreeding,withtheinbreedingindexrangingfrom−.556to.362onMarionIsland(mean±SD=−.071±.155)andfrom−.358to.522onMacquarieIsland(mean±SD=−.003±.214)(Table2;TableS1).

Toassessthelevelofcorrelationbetweengeneticdiversityandtheenvironmentalvariables,elevationandvegetationtype,weper-formedcorrelationtestsaftercorrectingforspatialautocorrelationwithinvariables.OnMarionIsland,HE,elevationandvegetationtypeweresignificantlyspatiallyautocorrelated,withtheregressionslopesforHE=−.024(p< .01),elevation=−.010(p< .01)andvegetationtype=−.007 (p < .01).No significant associationswere foundbe-tweenHEandelevationorvegetationtype.Pearson'scorrelationco-efficientbetweenHEandelevationwas.122(p >.05),andintraclasscorrelation coefficient betweenHE and vegetation type was .047(p >.05).OnMacquarieIsland,onlyelevationwasfoundtobesig-nificantlyspatiallyautocorrelated,withtheregressionslope=−.009(p<.01).NosignificantcorrelationwasfoundbetweenHE and eleva-tion(r=.099,p >.05).

Weassessedwhethervariation ingeneticdiversityhasadi-rectional orientation. ForA. selago onMarion Island,HE valuescorrelated significantly with transformed site position valuesforbearings0°–20°and80°–170° (p< .05), forwhichR2valuesranged from .05 to .27. AR values correlated significantly withtransformedsitepositionvaluesforbearings70°–170°(p<.05),forwhichR2valuesrangedfrom.04to.16.ForbothHE and AR,the strength of the regressions followed a periodic functionagainstcompassdirection,withthehighestR2valueassociatedwiththebearing120°forHEand110°forAR(Figure2a,c;TableS3). These bearings indicate a cline in genetic diversity alongthe northwest–southeast direction, with HE and AR increas-ing towards the southeast (Figure 2e). ForA. macquariensis on MacquarieIsland,nobearingpresentedasignificantcorrelationbetweenHE or ARandtransformedsitepositionvalues(p >.05),indicating the absence of a cline along any directional bearing(Figure2b;TableS3).

3.3 | Genetic differentiation

Analyses of molecular variance indicated that the vast major-ity of genetic variation was within individuals (Marion: 93.5%,Macquarie: 86.0%), with a much smaller yet significant portionfound between sites (Marion: 7.6%, Macquarie: 10.6%; p = .001)(Table 3).OnMarion Island, pairwiseFST values ranged from 0 to.399(mean±SD=.073±.067).Forindividualsites,theproportionofpairwiseFSTvaluesthatweresignificant(p<.05)variedfrom.139to.984(mean±SD= .493±.202)(TableS1),buttherewasnoap-parentspatialpattern inthe locationofsiteswithparticularlyhighor low proportions of significant pairwise FST values (Figure S1a).On Macquarie Island, pairwise FST values ranged from 0 to .471(mean ± SD = .104 ± .085). For an individual site, the proportionof pairwise FST values thatwere significant varied from .317 to 1(mean±SD=.653±.189)(TableS1),butagainwedidnotdetectanyapparentspatialpatterninthevariation(FigureS1b).

3.4 | Spatial genetic structure

For A. selago on Marion Island, a significant positive relation-shipwas foundbetweenFST/(1 −FST) and the distance betweensites (i.e.. isolation‐by‐distance [IBD]), with the regression slopebLd=.021(p<.001).Incontrast,forA. macquariensisonMacquarieIsland, no significant IBDwas found, with bLd = −.004 (p = .81)(Figure3).

Todeterminewhetherthespatialgeneticstructurehasadirec-tionalcomponent,weperformedSGSanalysesafter transformingdistance matrices for different bearings. On Marion Island, SGSanalyseswithbearingtransformationsrevealedpositiveandsignifi-cantrelationshipsbetweenthematrixofFST/(1−FST)valuesandD0,D10 and D100–D170spatialdistancematrices(p<.01)(Figure4a).OnMacquarieIsland,bearinganalysesdetectedpositiveandsignificantrelationships between the genetic distancematrix andD100–D120 matrices(p<.05)(Figure4b).Thisindicatesthatthereisapositiverelationship between genetic differentiation between sites and

TA B L E 1  GenotypiccharacteristicsofmicrosatellitelociforAzorella selagoonMarionIslandandAzorella macquariensisonMacquarieIsland

Locus

Marion Macquarie

Total AMean A per site (range) Mean HO Mean HE Total A

Mean A per site (range) Mean HO Mean HE

azo5 13 2.98(1–6) .38 .37 6 3.05(1–5) .45 .44

azo6 10 3.31(1–7) .39 .40 11 4.57(2–8) .70 .64

azo11 10 2.54(1–5) .58 .46 5 1.55(1–3) .05 .10

azo17 5 1.79(1–3) .15 .15 5 2.24(1–3) .32 .31

azo23 6 1.72(1–3) .14 .13 8 3.00(1–5) .39 .40

azo13 2 1.97(1–2) .39 .38 — — — —

Abbreviations:A,numberofalleles;HO,observedheterozygosity;HE,expectedheterozygosity.

3298  |     CHAU et Al.

distancebetweensitesgenerallyinthenorthwest–southeastdirec-tiononbothislands.

3.5 | Demographic change

MultiplesitesonbothMarionandMacquarieislandsshowedasignalofpopulationexpansion,ascharacterizedbyasignificantdeficiencyinheterozygosity(p<.05).OnMarionIsland,thesesiteswerecon-centratedinthewestandnorthoftheisland,with24sitesshowingsignificantheterozygositydeficiencyundertheSMMmodelandsizsitesunder the IAMmodel (Figure5a;FigureS2a).OnMacquarieIsland,eightsitesundertheSMMmodelandsixsitesundertheIAMmodelshowedsignificantheterozygositydeficiency,mostlyintwogroupsinthenorthandcentralpartoftheisland(Figure5b;FigureS2b).AfewsitesonMarionIsland,mostlyinthesoutheast,showedasignalofpopulationbottleneck,ascharacterizedbyasignificantexcessinheterozygosity(p<.05)(Figure5a;FigureS2a).NositesonMacquarieIslanddisplayedasignalofpopulationbottleneckundertheSMMmodel,buttherewerefoursites inthenorthernhalfoftheislandundertheIAMmodel(FigureS2b).

4  | DISCUSSION

ThedifferentspatialgeneticpatternsrecoveredforAzorella selago on MarionIsland(distinctnorthwest–southeastgradientinbothgeneticdiversity,geneticstructureandsignaturesofdemographicchange)andAzorella macquariensisonMacquarieIsland(nospatialgradientinge-neticdiversityordemographicchangebutspatialgeneticstructureinanarrownorthwest–southeastdirection)maybeattributedtoseveralmechanisms:asymmetricdispersalcausedbydominantnorthwesterlywindsacrossthespecificlandscapesoftheislands,historyofglaciationorlackthereof,andenvironmentaldifferencesbetweenthewesternandeasternsidesoftheislands(perhapsmorepronouncedforMarioncomparedtoMacquarie).Also,forMarionIslandspecifically,thearrivaloflong‐distancemigrantsinthesoutheastleewardsideoftheislandmayhavecontributedtothespatialpatterns.Thesefactorsarenotmu-tuallyexclusive,andarediscussedinmoredetailbelow.

4.1 | Spatial genetic patterns

Astrongspatialsignalwasdetectedinthepatternsofgeneticvari-ation ofAzorella onMarion Island, with the southeast portion oftheislandbeinggenerallydistinctfromtherestoftheisland.Thesesoutheasternsiteshadhighergeneticdiversityandweremorelikelyto showevidenceofpopulationbottlenecks,whereas sites acrosstherestoftheislandhadlowergeneticdiversityandweremorelikelytoshowsignalsofpopulationexpansion.SGSanalysesdetectedsig-nificantIBDonMarionIsland,suggestingthatdispersalacrosstheislandmaybelimitedandthatthehighestgeneflowpredominantlyoccursbetweenneighbouringsites.

Todate,island‐widegeneticstudiesonMarionIslandhavemostlyfocused on terrestrial microarthropods, with complex patterns andTA

BLE

2 SamplingandgenotypiccharacteristicsforA

zore

lla se

lagoonMarionIslandusingsixmicrosatellitelociandA

zore

lla m

acqu

arie

nsisonMacquarieIslandusingfivemicrosatelliteloci

Isla

ndS

Tota

l NN

AA R

HO

HE

F ISG

loba

l FST

Pairw

ise

F ST

Marion

123

1,149

9.341(7–10)

2.382(1.33–3.33)

2.078(1.27–2.72)

.338(.08–.58)

.315(.10–.51)

–.071(–.56–.36)

.076

.073(0–.399)

Macquarie

42372

8.857(6–15)

2.881(1.60–3.40)

2.502(1.53–3.11)

.384(.18–.63)

.377(.17–.51)

–.003(–.36–.52)

.106

.104(0–.471)

Not

es: MeanvaluesandrangesacrosslocipersiteareshownforN,A,A

R,H

O,H

E,F IS,andpairwise

F ST.

Abbreviations:A,numberofalleles;A

R,numberofalleles,adjustedbyrarefactionto10samplespersite;FIS,fixationindex/inbreedingcoefficient;H

E,expectedheterozygosity;H

O,observedheterozy

-gosity;N,numberofindividualssampledpersite;S,numberofsites;totalN,totalnumberofindividualssampled.

     |  3299CHAU et Al.

significant spatial structure linked to high genetic diversity beingreported for weevils, springtails and mites (see Grobler, Janse vanRensburg, Bastos, Chimimba, & Chown, 2006; McGaughran et al.,

2010;Mortimeretal.,2012;Myburghetal.,2007).AlthoughwedonotdetectsuchcomplexpatternsforA. selago,onestrikingsimilaritywithpreviousworkisthenoticeabledifferencebetweenthewesternandeasternpartsofMarion Island, inpartdrivenbyclimaticdiffer-ences.However, the exactmechanisms behind the contrasting lev-elsofspatialcomplexityandgeneticdiversitybetweenAzorella and arthropodsremainunclearandspeculative.Possiblecontributorsin-cludefasterratesofmolecularevolutioninarthropodscomparedtoAzorella,differencesinmodeofdispersalincludingactivemovementinarthropodscomparedtopassivemovementinAzorella,differencesinrangesizeandhabitatspecificity,andpossibleoccurrenceofself‐pol-linationinAzorellaversusthedominanceofoutcrossinginarthropods.

In the first molecular ecology study of any organism onMacquarieIsland,wefoundnostrongspatialgeneticpatternforA. macquariensis, with an absence of clinal patterns in geneticdiversity.Whenwe includedspatialorientation inourSGSanal-yses, only a small number of bearings in a northwest–southeastdirection showed signal of IBD, indicating the influence of thestrongwesterlywinds. Although some sites displayed signals ofpopulationexpansion,thesesiteswerescatteredinseveralgroupsaroundthe island,andprobablyreflect localcatastrophiceventsand organismal responses rather than general trends across theisland.

Marion Macquarie

Source of variation df vc (%) Source of variation df vc (%)

Betweensites 122 7.59 Betweensites 41 10.56

Betweenindividualswithinsites

1,026 −1.04 Betweenindividualswithinsites

330 3.34

Withinindividuals 1,149 93.45 Withinindividuals 372 86.09

Total 2,297 100 Total 743 100

Abbreviations:df,degreesoffreedom;vc,variancecomponent.

TA B L E 3  Partitionofgeneticvariationbyanalysisofmolecularvariance(AMOVA)forAzorella selagoonMarionIslandandAzorella macquariensis on MacquarieIsland(p=.001)

F I G U R E 3  Spatialgeneticstructure,orisolation‐by‐distance,in Azorella selagoonMarionIslandandAzorella macquariensis on MacquarieIsland.MeanvaluesofFST/(1−FST)indifferentdistanceclassesbetweensites(meandistancesplotted)forMarionIslandandMacquarieIsland,andregressionlines.Onlydistanceclasseswithgreaterthan50%ofsitesrepresentedareshown(allexcept20–25and25–35kmforMarionIslandand0–1and25–35kmforMacquarieIsland)

F I G U R E 4  Direction‐dependentspatialgeneticstructureinAzorella selagoonMarionIslandandAzorella macquariensisonMacquarieIsland.ValuesofthecorrelationcoefficientrfromlinearregressionsbetweenFST/(1−FST)andtransformeddistancematricesthataccountfordirectionalbearingsfor(a)MarionIslandand(b)MacquarieIsland.Distancematricesweretransformedforbearingsfrom0°to170°at10°intervals,with0°correspondingtotheeast–westdirection.Filledsymbolsindicatebearingswithasignificantcorrelation(p<.05)

(a) (b)

3300  |     CHAU et Al.

4.2 | Contemporary influences on spatial genetic patterns: Shape of the island and wind

WhileMarion andMacquarie islands experience similar climates, in-cluding strongwinds experienced for themajority of the time fromnorthwesterly(Adams,2009;LeRoux,2008;Schulze,1971;Selkirketal.,1990),theislandsdiffermarkedlyinshapeandtopography.Windandwinddirection, inconjunctionwithislandshape,probablyplayamajorroleindrivingspatialgeneticpatternsinAzorella.ForA. selago on thecircular‐shapedMarionIsland,thereisastrongclinalcomponenttothedistributionofgeneticdiversity,withincreasingheterozygosityandallelic richness followinganorthwest–southeastgradient,whichcor-responds to theprevailingwinddirection.Dispersal isclearlyoneofthemostimportantfactorsdrivingspatialstructureinthisspecies,withIBDalsodetectedinanorthwest–southeastdirection.Dominantwindinonedirectioncancauseasymmetricgeneflowsothatsiteslocateddownwindaccumulateallelesdispersedfromelsewhere,andthushar-bour higher levels of genetic diversity comparedwith sites upwind.Consistentwiththisexpectation,A. selagopopulationsinthesoutheastofMarion Islandhave thehighest levels of genetic diversity. Similarpatternsofgenetic accumulationhavebeen reported forplants andinvertebrates thatareunidirectionallydispersedviawaterways,withdownstreampopulationstypicallycontaininghighergeneticdiversitythantheirupstreamcounterparts (Alp,Keller,Westram,&Robinson,2012;Kikuchi,Suzuki,&Sashimura,2011).Tothebestofourknowl-edge,thishasnotyetbeendemonstratedforwind‐dispersedspecies.

OurfindingattheislandscalecloselycorroboratethosereportedbyBornetal.(2012)forAzorellaatsmallspatialscales(tenstohun-dredsofmetres),whofoundthatdirectionaldispersalassociatedwithwindpatternsisaprincipalfactorindeterminingfine‐scalepopulationgeneticstructure.TheyfoundIBDtobeweakerbetweensitesexperi-encingstrongerwinds,oftenweakestinthedirectionoftheprevailingwinds,presumablybecauseofgreaterwinddispersal;andthebearingwiththeweakestsignalofIBDvariedbetweensites,possiblyduetodifferencesinthedominantwinddirectioncausedbylocalturbulenceandtopography.Theimpactofsuchlocal,fine‐scaleprocessesonis-land‐scalepatternsremainsunclearanddeservesfurtherstudy.

IncontrasttoMarionIsland,wedidnotfindastrongdirectionalclineingeneticdiversityforA. macquariensisonMacquarieIsland.Thisfindingmayalsobetheconsequenceoftheprevailingwindinteract-ingwiththeelongateshapeoftheisland.MacquarieIslandisanarrowrectangle(~34×5km),withtheshortaxisorientedWNW–ESE,inlinewith thedirectionof theprevailingwinds (Selkirketal.,1990).Theshortdistancebetweenthewesternandeasternedgesoftheislandmaynotbesufficientforthedevelopmentofanobviouscline.

4.3 | Historical influences on spatial genetic patterns: Glaciation history and refugia

AlthoughtheshapeoftheislandslinkedtoprevailingwindsprobablyplaysamajorroleinshapingspatialgeneticdiversityofAzorellaspe-cies onMarion andMacquarie islands,we cannot rule out possible

F I G U R E 5  DemographicchangeinsitesofAzorella selagoonMarionIslandandAzorella macquariensisonMacquarieIsland.Siteswithsignificantheterozygositydeficiencyorheterozygosityexcess,evidenceforrecentpopulationexpansionorbottleneck,respectively,inferredunderthestepwisemutationmodel(SMM)for(a)MarionIslandand(b)MacquarieIsland[Colourfigurecanbeviewedatwileyonlinelibrary.com]

(a) (b)

     |  3301CHAU et Al.

contributionsfromthegeologicalhistoryoftheislands.MarionIslandhas experienced at least five episodes of glaciation during the LateQuaternary (McDougall et al., 2001). The extent of glaciation dur-ing the LGM (~35–11,000years before present) is uncertain, but itis believed to have beenwidespread and covered almost the entireisland(Boelhouwersetal.,2008;Hall,2004;Halletal.,2011).Severalprominent areas on the island, however, lack evidence of glaciationandmayhaveservedasice‐freerefugia;thesesitesarelargelycoastal(Boelhouwersetal.,2008).PalynologicalstudiesindicatethatthesameplantsfoundonMarionIslandtoday,includingAzorella,werepresentduring interglacial periods (Schalke & Van Zinderen Bakker, 1971;Scott,1985;Scott&Hall,1983).Itisthereforeconceivablethatrefugiaplayedanimportantroleinthesurvivalandpersistenceoftheisland'sfloraduringglaciations,althoughrepeatedrecolonizationvialong‐dis-tancedispersalfromothersourceareasafterdeglaciationisalsopos-sible(VanderPutten,Verbruggen,Ochyra,Verleyen,&Frenot,2010).

SpatialgeneticpatternsinA. selagoonMarionIslandareconsis-tentwithahistoryofsurvivalinasoutheastrefugiumduringglacialperiods (seeMortimer et al., 2012 for similar suggestions for themiteHalozetes fulvus).Geneticdiversity,measuredasexpectedhet-erozygosityandallelic richness, ishighest in thesoutheast,wherethe Feldmark Plateau, a refugial area during the LGM, is located(Boelhouwersetal.,2008).Fromthesoutheast,colonizationoftherestoftheislandcouldhaveoccurredafterglacialretreatviacon-secutivefounderevents,whichwouldresultinlowergeneticdiver-sityalongthedirectionofpopulationexpansion.SimilarpatternsofhighergeneticdiversityinglacialrefugiaarefrequentlyobservedontheEurasianandNorthAmericancontinents(Conroy&Cook,2000;Widmer&Lexer,2001).Wealsofoundevidenceforpopulationex-pansioninsitesoutsidethesoutheast,whichisconsistentwithre-centcolonizationofthoseareas,althoughtheexacttimingofthesedemographicchangesisuncertain.

On Macquarie Island, recent interpretation of geomorpholog-ical evidence suggests that glaciation was very limited (Hodgsonetal.,2014;McGlone,2002;Selkirketal.,1990).Detailedpalyno-logicalstudies,whichincluderecordsofAzorella,extendbackonly7,000–9,000years (Bergstrom,Stewart, Selkirk,&Schmidt,2002;McGlone,2002;Selkirk,Selkirk,Bergstrom,&Adamson,1988),sothecompositionofthefloraduringPleistoceneglacialperiodsisun-known.ThelackofspatialgeneticstructureinA. macquariensis on MacquarieIslandisconsistentwithitspersistenceacrosstheislandthroughouttheLateQuaternary.ThehighergeneticdiversityinA. macquariensisonMacquarieIslandcomparedtoA. selagoonMarionIslandmayalsobeduetothepersistenceofalargerpopulationonMacquarieIslandthroughoutthisperiod.

4.4 | Other contemporary influences on spatial genetic patterns: Long‐distance dispersal and ecological drivers

Anotherimportantconsiderationisthattheareaofarrivalandini-tialcolonizationoflong‐distancemigrantscarriedbywindtosub‐Antarcticislandsmaynotberandom.ChownandAvenant(1992)

theorized that for smallorganismsand inareaswithhighwinds,settlingoutoftheaircolumnismostlikelytooccurontheleewardsideoftheislandduetoturbulencecreatedbyhighertopographyin the island'scentre.Onsub‐Antarctic islands, the leewardsideis typically in the southeast. In support of this hypothesis, sev-eralrecentnaturalintroductionstosub‐Antarcticislandswereini-tiallydiscoveredintheleewardsideofislands,includingthemothPlutella xylostellaonMarion Islandandseveral insectson ÎleauxCochons in theCrozet archipelago (Chown&Avenant,1992). InseveralarthropodspeciesonMarionIsland,sitesinthesoutheastweregeneticallysignificantlydifferent fromothers,whichmightbeduetothearrivalofwind‐borneindividualsfromoutsidetheis-landtothatarea(Myburghetal.,2007).OnMacquarieIsland,sev-eral long‐distance insectmigrantsweredetectedon theeasternsideof the island,althoughthispatterncouldalsohaveresultedfromnonrandomsamplingbyresearchers(Greenslade,Farrow,&Smith, 1999). The spatial genetic patternofA. selago onMarionIsland,withhighergeneticdiversityinthesoutheast,isconsistentwith the arrival of long‐distance immigrants in the southeast oftheisland.

Environmental heterogeneity greatly influences dispersalpatterns, as dispersal alone is not a reliable indicator of poten-tial range expansion or occupancy given that propagules needto encounter favourable habitats to settle in. Both Marion andMacquarieislandshavelocalclimatevariability.ForMarionIsland,climaticdifferencesarefoundalonganeast–westgradientaswellas an altitudinal gradient (Le Roux, 2008), while for MacquarieIsland,thesearepresentbutperhapsnotaspronounced(Davies&Melbourne,1999;Selkirketal.,1990).OnMarionIsland,ithasbeen reported that many ecological traits, including plant size,leafsize,trichomedensityandstomataldensity,aremorestronglyassociatedwith island side (i.e., leewardvs.windward) thanele-vation(McGeochetal.,2008;Nyakatya,2006).Differencesincli-matewithislandside,forexampleinprecipitation,solarradiationandwind intensity (Nyakatya&McGeoch, 2008; Rouault et al.,2005;LeRoux,2008;Schulze,1971),ratherthanwithelevation,mayresult instrongereffectsonthesespecies’traits (Nyakatya,2006).Inlinewiththis,wedidnotfindanysignificantcorrelationsbetweengeneticdiversityandelevation,orwithvegetationtype.Wedid not expect to detect any signatures of adaptation giventheneutralmarkersthatweusedhere.However, it is imperativethatfuturestudiesshouldfocusonidentifyingareasofecologicalimportance. This is especially important in the face of a rapidlychangingclimate(Rouaultetal.,2005)andthepossibilityofspe-ciesonlysurvivingchangeinlocalmicrorefugiaintothefuture.

5  | CONCLUSIONS

Species ranges are dynamic, altered by environmental and evolu-tionarychangeanddispersaldynamics.Onislands,rangesarecon-strained by the scale of the island and the availability of suitablehabitat, and the spatial distribution of genetic variation becomes

3302  |     CHAU et Al.

especially important for evolutionary and ecological processes. InourstudyofkeystoneAzorellaspeciesonsub‐Antarcticislands,wefound that local climate, specifically wind patterns, in interactionwithislandshape,aswellashistoricalpatternsofpopulationpersis-tence,probablycontroldispersalandthusspatialpatternsingeneticdiversityandstructure.Thesefindingscanhelpinformthemanage-mentandconservationofbiologicaldiversityontheseuniquepolarecosystems,especiallyinthefaceofrapidlychangingenvironments.

ACKNOWLEDG EMENTS

This researchwas fundedby theSouthAfricanNationalResearchFoundation (through a South African National Antarctic Programresearch grant to B.J.V.V. and M.A.M.). J.C. and M.M. are sup-portedthroughgrant‐holderpostdoctoralbursariestoB.v.V.(SouthAfrican National Antarctic Program), with supplementary fundingfrom theUniversityof Johannesburg andStellenboschUniversity.C.B.wasthebeneficiaryofapost‐doctoralgrantfromStellenboschUniversity, the NRF and the AXA Research Fund. Sampling forMarionIslandwasconductedduringthe2006,2007and2009re-lief voyages,whichwere logistically supportedand fundedby theSouth AfricanDepartment of Environmental Affairs and Tourism:AntarcticaandIslandsthroughtheSouthAfricanNationalAntarcticProgram.SamplingonMacquarieIslandwasundertakenduringthe2010/11 summer field season logistically supported and fundedbytheAustralianAntarcticDivision,DepartmentofSustainability,Environment, Water, People and Community, Australia, and per-mission to conduct researchwasgrantedbyTasmanianParksandWildlife Service.We thank Kate Kiefer, Jess Bramley‐Aves, PeterleRoux,JesseKalwij,StevenChown,EthelPhiri,TessRautenbach,BruceDyer,GregoryMcClelland,JamesWilshire,MariusRossouwandMashuduMashau for assistancewith sample collections.Wealso thank Cécile Berthouly‐Salazar for interesting discussions onthemanuscript,HattieChauforassistancewith figuresandDavidHedding(UniversityofSouthAfrica)forhelpwithgeneratingmaps.Input from the anonymous reviewers further strengthened themanuscript.

AUTHOR CONTRIBUTIONS

C.B.,M.A.M.andB.J.V.V.designedthestudy.C.B.,M.A.M.,D.B.,J.S.,A.T.andB.J.V.V.collectedsamples.C.B., J.H.C.andB.J.V.V.gener-atedandanalysedthedata.Allauthorscontributedtothewriting.

DATA ACCE SSIBILIT Y

GPScoordinates,elevationandvegetationtypeofeachsiteandmi-crosatellitegenotypesofeachsampleareavailableinTablesS1andS2onDryad:https://doi.org/10.5061/dryad.cr12t51.

ORCID

John H. Chau https://orcid.org/0000‐0002‐8913‐6451

Bettine Jansen van Vuuren https://orcid.org/0000‐0002‐5334‐5358

R E FE R E N C E S

Adams,N.(2009).ClimatetrendsatMacquarieIslandandexpectationsoffutureclimatechangeinthesub‐Antarctic.Papers and Proceedings of the Royal Society of Tasmania,143(1),1–8.https://doi.org/10.26749/rstpp.143.1.1

Adamson,D.A.,Selkirk,P.M.,&Colhoun,E.A.(1988).Landformsofae-olian,tectonicandmarineoriginintheBauerBay‐SandyBayregionofsubantarcticMacquarieIsland.Papers and Proceedings of the Royal Society of Tasmania,122,65–82.

Adamson,D.A., Selkirk,P.M.,Price,D.M.,Ward,N.,&Selkirk, J.M.(1996). Pleistocene uplift and palaeoenvironments of MacquarieIsland: evidence from palaeobeaches and sedimentary deposits.Papers and Proceedings of the Royal Society of Tasmania,130(2),25–32.https://doi.org/10.26749/rstpp.130.2.25

Alp, M., Keller, I., Westram, A. M., & Robinson, C. T. (2012). Howriver structure and biological traits influence gene flow: A pop-ulation genetic study of two stream invertebrates with differingdispersal abilities. Freshwater Biology, 57, 969–981. https://doi.org/10.1111/j.1365‐2427.2012.02758.x

Bergstrom, D.M., Bricher, P. K., Raymond, B., Terauds, A., Doley, D.,McGeoch,M.A.,…Ball,M.C.(2015).Rapidcollapseofasub‐Antarcticalpineecosystem:Theroleofclimateandpathogens.Journal of Applied Ecology,52,774–783.https://doi.org/10.1111/1365‐2664.12436

Bergstrom,D.M.,&Chown,S.L.(1999).Lifeatthefront:History,ecologyandchangeonsouthernoceanislands.Trends in Ecology and Evolution,14,472–477.https://doi.org/10.1016/S0169‐5347(99)01688‐2

Bergstrom,D.M.,& Selkirk, P.M. (1999). Bryophyte propagule banksin a feldmark on subantarctic Macquarie Island. Arctic, Antarctic and Alpine Research, 31, 202–208. https://doi.org/10.1080/15230430.1999.12003299

Bergstrom,D.M.,Stewart,G.R.,Selkirk,P.M.,&Schmidt,S.(2002).15Nnaturalabundanceoffossilpeatreflectstheinfluenceofanimal‐de-rivednitrogenonvegetation.Oecologia,130,309–314.https://doi.org/10.1007/s004420100807

Boelhouwers, J. C.,Meiklejohn, K. I., Holness, S. C., &Hedding, D.W.(2008).Geology,geomorphologyandclimatechange.InS.L.Chown,&P.W.Froneman (Eds.),The Prince Edward Islands. Land‐sea interac‐tions in a changing ecosystem (pp.65–96).Stellenbosch,SouthAfrica:SunPress.

Born,C.,LeRoux,P.C.,Spohr,C.,McGeoch,M.A.,&JansenvanVuuren,B. (2012).Plantdispersal in thesub‐Antarctic inferred fromaniso-tropicgeneticstructure.Molecular Ecology,21,184–194.https://doi.org/10.1111/j.1365‐294X.2011.05372.x

Bricher,P.K.,Lucieer,A.,Shaw,J.,Terauds,A.,&Bergstrom,D.M.(2013).Mappingsub‐Antarcticcushionplantsusingrandomforeststocom-bineveryhighresolutionsatelliteimageryandterrainmodelling.PLoS ONE,8,e72093.https://doi.org/10.1371/journal.pone.0072093

Cerfonteyn,M. E., Le Roux, P. C., Jansen van Vuuren, B., & Born, C.(2011).CrypticspatialaggregationofthecushionplantAzorella sel‐ago(Apiaceae)revealedbyamultilocusmolecularapproachsuggestsfrequent intraspecific facilitation under sub‐Antarctic conditions.American Journal of Botany, 98, 909–914. https://doi.org/10.3732/ajb.1000460

Chown,S.L.,&Avenant,N.(1992).StatusofPlutella xylostellaatMarionIslandsixyearsafteritscolonisation.South African Journal of Antarctic Research,22,37–40.

Chown,S.L.,&Convey,P.(2007).Spatialandtemporalvariabilityacrosslife'shierarchiesintheterrestrialAntarctic.Philosophical Transactions

     |  3303CHAU et Al.

of the Royal Society B, 362, 2307–2331. https://doi.org/10.1098/rstb.2006.1949

Chown,S.L.,Gremmen,N.J.M.,&Gaston,S.L.(1998).Ecologicalbio-geography of Southern Ocean islands: Species‐area relationships,humanimpacts,andconservation.American Naturalist,152,562–575.

Chown, S. L.,&Marshall,D. J. (2008). Terrestrial invertebrates of thePrinceEdwardIslands.InS.L.Chown,&P.W.Froneman(Eds.),The Prince Edward Islands. Land‐sea interactions in a changing ecosystem (pp.400–409).Stellenbosch,SouthAfrica:SunPress.

Conroy,C.J.,&Cook,J.A.(2000).Phylogeographyofapost‐glacialcolo-nizer:Microtus longicaudus(Rodentia:Muridae).Molecular Ecology,9,165–175.https://doi.org/10.1046/j.1365‐294x.2000.00846.x

Convey,P.(2007).Influencesonandoriginsofterrestrialbiodiversityofthesub‐Antarcticislands.Papers and Proceedings of the Royal Society of Tasmania,141,83–93.https://doi.org/10.26749/rstpp.141.1.83

Copson, G. R. (1984). An annotated atlas of the vascular flora ofMacquarieIsland.ANARE Research Notes,18,1–70.

Cornuet,J.M.,&Luikart,G. (1996).Descriptionandpoweranalysisoftwotestsfordetectingrecentpopulationbottlenecksfromallelefre-quencydata.Genetics,144,2001–2014.

Davies,K.F.,&Melbourne,B.A. (1999).Statisticalmodelsof inverte-brate distribution on Macquarie Island: A tool to assess climatechangeandlocalhumanimpacts.Polar Biology,21,240–250.https://doi.org/10.1007/s003000050359

Dick, C. W., Hardy, O. J., Jones, F. A., & Petit, R. J. (2008). Spatialscales of pollen and seed‐mediated gene flow in tropical rain for-est trees.Tropical Plant Biology,1, 20–33. https://doi.org/10.1007/s12042‐007‐9006‐6

Excoffier,L.,&Lischer,H.E.L.(2010).ARLEQUINsuitever3.5:Anewseries of programs to performpopulation genetics analyses underLinuxandWindows.Molecular Ecology Resources,10,564–567.https://doi.org/10.1111/j.1755‐0998.2010.02847.x

Falsetti,A.B.,&Sokal,R.R.(1993).Geneticstructureofhumanpopula-tionsintheBritishIsles.Annals of Human Biology,20,215–229.https://doi.org/10.1080/03014469300002652

Girod,C.,Vitalis,R.,Leblois,R.,&Fréville,H.(2011).Inferringpopulationdeclineandexpansionfrommicrosatellitedata:Asimulation‐basedevaluationoftheMsvarmethod.Genetics,188,165–179.https://doi.org/10.1534/genetics.110.121764

González‐Wevar, C. A., Segovia, N. I., Rosenfeld, S., Ojeda, J., Hüne,M.,Naretto, J.,… Poulin, E. (2018).Unexpected absence of islandendemics: Long‐distance dispersal in higher latitude sub‐AntarcticSiphonaria(Gastropoda:Euthyneura)species.Journal of Biogeography,45,874–884.

Greenslade,P.,Farrow,R.A.,&Smith,J.M.B.(1999).Longdistancemi-grationof insects to a subantarctic island. Journal of Biogeography,26,1161–1167.https://doi.org/10.1046/j.1365‐2699.1999.00356.x

Gremmen,N.J.M.,&Smith,V.R.(2008).Terrestrialvegetationanddy-namics.InS.L.Chown,&P.W.Froneman(Eds.),The Prince Edward Islands. Land‐sea interactions in a changing ecosystem (pp.215–244).Stellenbosch,SouthAfrica:SunPress.

Greve, M., Gremmen, N. J. M., Gaston, K. J., & Chown, S. L. (2005).Nestedness of Southern Ocean island biotas: Ecological perspec-tivesonabiogeographical conundrum.Journal of Biogeography,32,155–168.https://doi.org/10.1111/j.1365‐2699.2004.01169.x

Grobler,G.C.,Bastos,A.D.S.,Treasure,A.M.,&Chown,S.L.(2011).Crypticspecies,biogeographiccomplexityandtheevolutionaryhistoryoftheEctemnorhinus group in the sub‐Antarctic, including adescriptionofBothrometopus huntleyi,n.sp.Antarctic Science,23,211–224.

Grobler,G.C.,JansevanRensburg,L.,Bastos,A.D.S.,Chimimba,C.T.,&Chown,S.L.(2006).Molecularandmorphometricassessmentofthe taxonomic statusofEctemnorhinusweevil species (Coleoptera:Curculionidae, Entiminae) from the sub‐Antarctic Prince EdwardIslands. Journal of Zoological Systematics and Evolutionary Research,44,200–211.https://doi.org/10.1111/j.1439‐0469.2006.00358.x

Hall, K. (2004). Quaternary glaciation of the sub‐Antarctic islands. InJ. Ehlers,&P. I.Gibbard (Eds.),Quaternary glaciations – extent and chronology, part III (pp. 339–345). Amsterdam, the Netherlands:Elsevier.

Hall,K.,Meiklejohn,I.,&Bumby,A.(2011).MarionIslandvolcanismandglaciation.Antarctic Science,23, 155–163. https://doi.org/10.1017/S0954102010000878

Hamilton, M. B. (1999). Tropical tree gene flow and seed dispersal.Nature,401,129–130.https://doi.org/10.1038/43597

Hänel,C.,&Chown,S.(1998).An introductory guide to the Marion and Prince Edward Island Special Nature Reserves 50 years after annexation.Pretoria,SouthAfrica:DepartmentofEnvironmentalAffairsandTourism.

Hardouin, E. A., Chapuis, J. L., Stevens, M. I., Jansen van Vuuren,B., Quillfeldt, P., Scavetta, R. J., … Tautz, D. (2010). Housemouse colonization patterns on the sub‐Antarctic KerguelenArchipelago suggest singular primary invasions and resilienceagainst re‐invasion.BMC Evolutionary Biology, 10, 325. https://doi.org/10.1186/1471‐2148‐10‐325

Hardy,O.J.(2009).TOROCOR: A program to assess the association between spatially autocorrelated variables using a torus‐translation test on mul‐tiple grids.Retrievedfromhttp://ebe.ulb.ac.be/ebe/Software.html

Hardy,O.J.,&Vekemans,X.(1999).Isolationbydistanceinacontinuouspopulation:Reconciliationbetweenspatialautocorrelationanalysisandpopulationgeneticsmodels.Heredity,83,145–154.https://doi.org/10.1046/j.1365‐2540.1999.00558.x

Hardy, O. J., & Vekemans, X. (2002). SPAGEDI: A versatile computerprogram to analyse spatial genetic structure at the individual orpopulation levels.Molecular Ecology Notes,2, 618–620.https://doi.org/10.1046/j.1471‐8286.2002.00305.x

Harms, K. E., Condit, R., Hubbell, S. P., & Foster, R. B. (2001).Habitat associations of trees and shrubs in a 50‐ha neotrop-ical forest plot. Journal of Ecology, 89, 947–959. https://doi.org/10.1111/j.1365‐2745.2001.00615.x

Haussmann, N. S., McGeoch, M. A., & Boelhouwers, J. C. (2010).Contrasting nurse plants and nurse rocks: The spatial distributionofseedlingsoftwosub‐Antarcticspecies.Acta Oecologica,36,299–305.https://doi.org/10.1016/j.actao.2010.02.001

Hodgson,D.A.,Graham,A.G.C.,Roberts,S.J.,Bentley,M.J.,Cofaigh,C.Ó.,Verleyen,E.,…Smith,J.A. (2014).Terrestrialandsubmarineevidence for the extent and timing of the Last Glacial Maximumand the onset of deglaciation on themaritime‐Antarctic and sub‐Antarcticislands.Quaternary Science Reviews,100,137–158.https://doi.org/10.1016/j.quascirev.2013.12.001

Hugo,E.A.,McGeoch,M.A.,Marshall,D.J.,&Chown,S.L.(2004).Finescale variation inmicroarthropod communities inhabiting the key-stone species Azorella selago on Marion Island. Polar Biology, 27,466–473.https://doi.org/10.1007/s00300‐004‐0614‐4

Huntley, B. J. (1971). Vegetation. In E.M. Van Zinderen Bakker, J.M.Winterbottom,&R.A.Dyer(Eds.),Marion and Prince Edward Islands (pp.98–160).CapeTown,SouthAfrica:Balkema.

JansenVanVuuren,B.,&Chown,S.L.(2007).Geneticevidenceconfirmstheoriginofthehousemouseonsub‐AntarcticMarionIsland.Polar Biology,30,327–332.https://doi.org/10.1007/s00300‐006‐0188‐4

Kalinowski,S.T.(2005).HP‐RARE:Acomputerprogramforperformingrarefactiononmeasuresofallelicdiversity.Molecular Ecology Notes,5,187–189.

Kalwij,J.M.,Medan,D.,Kellermann,J.,Greve,M.,&Chown,S.L.(2019).Vagrantbirdsasadispersalvectorintransoceanicrangeexpansionofvascularplants.Scientific Reports,9,4655.https://doi.org/10.1038/s41598‐019‐41081‐9

Kikuchi,S.,Suzuki,W.,&Sashimura,N.(2011).Geneflowinanendan-geredwillowSalix hukaoana (Salicaceae) innaturalandfragmentedriparian landscapes. Conservation Genetics, 12, 79–89. https://doi.org/10.1007/s10592‐009‐9992‐z

3304  |     CHAU et Al.

LaRue,E.A.,Holland,J.D.,&Emery,N.C.(2018).Environmentalpredic-torsofdispersaltraitsacrossaspecies’geographicrange.Ecology,99,1857–1865.https://doi.org/10.1002/ecy.2402

LeRoux,P.C.(2008).Climateandclimatechange.InS.L.Chown,&P.W.Froneman(Eds.),The Prince Edward Islands. Land‐sea interactions in a changing ecosystem(pp.39–64).Stellenbosch,SouthAfrica:SunPress.

LeRoux,P.C.,&McGeoch,M.A. (2010). Interaction intensityand im-portancealongtwostressgradients:Addingshapetothestress‐gra-dienthypothesis.Oecologia,162,733–745.https://doi.org/10.1007/s00442‐009‐1484‐9

Ledingham,R.,&Petersen,A.(1984).Raisedbeachdepositsandthedis-tributionof structural lineaments onMacquarie Island.Papers and Proceedings of the Royal Society of Tasmania,118,223–235.

Lee,J.R.,Raymond,B.,Bracegirdle,T.J.,Chadès,I.,Fuller,R.A.,Shaw,J.D.,&Terauds,A.(2017).ClimatechangedrivesexpansionofAntarcticice‐free habitat.Nature,547, 49–54. https://doi.org/10.1038/nature22996

Lenoir, J., Virtanen, R., Oksanen, J., Oksanen, L., Luoto, M., Grytnes,J. A., & Svenning, J. C. (2012). Dispersal ability links to cross‐scale species diversity patterns across the Eurasian Arctic tun-dra. Global Ecology and Biogeography, 21, 851–860. https://doi.org/10.1111/j.1466‐8238.2011.00733.x

Lin,G. J.,Stralberg,D.,Gong,G.Q.,Huang,Z.L.,Ye,W.H.,&Wu,L.F. (2013).Separatingtheeffectsofenvironmentandspaceontreespecies distribution: From population to community.PLoS ONE,8,e56171.https://doi.org/10.1371/journal.pone.0056171

Lord, J.M. (2015).Patterns in floral traits andplantbreeding systemson Southern Ocean Islands. AoB Plants, 7, plv095. https://doi.org/10.1093/aobpla/plv095

Loveless,M.D.,&Hamrick,J.L.(1984).Ecologicaldeterminantsofgeneticstructureinplantpopulations.Annual Review of Ecology and Systematics,15,65–95.https://doi.org/10.1146/annurev.es.15.110184.000433

Manel,S.,Schwartz,M.K.,Luikart,G.,&Taberlet,P.(2003).Landscapegenetics: Combining landscape ecology and population genetics.Trends in Ecology & Evolution,18,189–197.https://doi.org/10.1016/S0169‐5347(03)00008‐9

Martinez, S. (1993). Sinopsis del genero Azorella (Apiaceae,Hydrocotyloideae).Darwiniana,32,171–184.

McDougall,I.,Verwoerd,W.,&Chevallier,L.(2001).K‐ArgeochronologyofMarion Island, SouthernOcean.Geological Magazine,138, 1–17.https://doi.org/10.1017/S0016756801005039

McGaughran, A., Convey, P., Stevens, M. I., & Chown, S. L. (2010).Metabolic rate, genetic and microclimate variation among spring-tailpopulationsfromsub‐AntarcticMarionIsland.Polar Biology,33,909–918.https://doi.org/10.1007/s00300‐010‐0767‐2

McGeoch,M.A.,LeRoux,P.C.,Hugo,E.A.,&Nyakatya,M.J. (2008).Spatialvariationintheterrestrialbioticsystem.InS.L.Chown,&P.W.Froneman(Eds.),The Prince Edward Islands. Land‐sea interactions in a changing ecosystem (pp.245–276).Stellenbosch,SouthAfrica:SunPress.

McGlone,M.S. (2002).TheLateQuaternarypeat, vegetationand cli-mate history of the Southern Oceanic Islands of New Zealand.Quaternary Science Reviews, 21, 683–707. https://doi.org/10.1016/S0277‐3791(01)00044‐0

Mertes,K.,&Jetz,W.(2018).Disentanglingscaledependenciesinspe-cies environmental niches and distributions. Ecography, 41, 1604–1615.https://doi.org/10.1111/ecog.02871

Molecular Ecology Resources PrimerDevelopment (2010). PermanentgeneticresourcesaddedtoMolecularEcologyResourcesDatabase1August2009–30September2009.Molecular Ecology Resources,10,232–236.

Mortimer,E.,&JansenvanVuuren,B.(2007).PhylogeographyofEupodes minutus(Acari:Prostigmata)onsub‐AntarcticMarionIslandreflects

the impactofhistoricalevents.Polar Biology,30,471–476.https://doi.org/10.1007/s00300‐006‐0205‐7

Mortimer,E., JansenvanVuuren,B.,Meiklejohn,K. I.,&Chown,S. L.(2012). Phylogeography of a mite, Halozetes fulvus, reflects thelandscape history of a young volcanic island in the sub‐Antarctic.Biological Journal of the Linnean Society, 105, 131–145. https://doi.org/10.1111/j.1095‐8312.2011.01770.x

Mortimer,E.,McGeoch,M.A.,Daniels,S.R.,&JansenvanVuuren,B.(2008).GrowthformandpopulationgeneticstructureofAzorella se‐lagoonsub‐AntarcticMarionIsland.Antarctic Science,20,381–390.

Myburgh,M.,Chown,S.L.,Daniels,S.R.,&JansenvanVuuren,B.(2007).Population structure, propagulepressure, and conservationbioge-ography in the sub‐Antarctic: Lessons from indigenous and inva-sivespringtails.Diversity and Distributions,13,143–154.https://doi.org/10.1111/j.1472‐4642.2007.00319.x

Nyakatya, M. J. (2006). Patterns of variability in Azorella selago Hook. (Apiaceae) on sub‐Antarctic Marion Island: Climate change implications (Unpublished M.Sc. thesis).UniversityofStellenbosch.

Nyakatya,M.J.,&McGeoch,M.A.(2008).TemperaturevariationacrossMarion Island associated with a keystone plant species (Azorella selago Hook. (Apiaceae)). Polar Biology, 31, 139–151. https://doi.org/10.1007/s00300‐007‐0341‐8

Orchard,A.E.(1989).AzorellaLamarck(Apiaceae)onHeardandMacquarieIslands,withdescriptionofanewspecies,A. Macquariensis. Muelleria,7,15–20.

Peakall, R.,& Smouse, P. E. (2012).GENALEX6.5:Genetic analysis inExcel.Populationgeneticsoftwareforteachingandresearch‐anup-date. Bioinformatics, 28, 2537–2539. https://doi.org/10.1093/bioinformatics/bts460

Phiri,E.E.,McGeoch,M.A.,&Chown,S.L.(2009).Spatialvariationinstructuraldamagetoakeystoneplantspeciesinthesub‐Antarctic:Interactions between Azorella selago and invasive house mice.Antarctic Science,21,189–196.

Piry,S.,Luikart,G.,&Cornuet,J.M.(1999).BOTTLENECK:Acomputerprogramfordetectingrecentreductionsintheeffectivepopulationsizeusingallelefrequencydata.Journal of Heredity,90,502–503.

Plunkett,G.M.,&Nicolas,A.N.(2017).AssessingAzorella(Apiaceae)anditsallies:Phylogeneticsandanewclassification.Brittonia,69,31–61.https://doi.org/10.1007/s12228‐016‐9446‐0

Postaire,B.,Gélin,P.,Bruggemann,J.H.,&Magalon,H.(2017).Onespe-ciesforoneisland?Unexpecteddiversityandweakconnectivityinawidelydistributedtropicalhydrozoan.Heredity,118,385–394.https://doi.org/10.1038/hdy.2016.126

Quilty,P.G.(2007).Originandevolutionofthesub‐Antarcticislands:Thefoundation.Papers and Proceedings of the Royal Society of Tasmania,141,35–58.https://doi.org/10.26749/rstpp.141.1.35

RCoreTeam(2018).R: A language and environment for statistical comput‐ing.Vienna,Austria:RFoundationforStatisticalComputing.

Rosenberg,M.S.(2000).Thebearingcorrelogram:Anewmethodofan-alyzingdirectionalspatialautocorrelation.Geographical Analysis,32,267–278.https://doi.org/10.1111/j.1538‐4632.2000.tb00428.x

Rosenberg, M. S., & Anderson, C. D. (2011). PASSAGE: Patternanalysis, spatial statistics and geographic exegesis. Version2. Methods in Ecology and Evolution, 2, 229–232. https://doi.org/10.1111/j.2041‐210X.2010.00081.x

Rouault,M.,Mélice,J.L.,Reason,C.J.C.,&Lutijeharms,J.R.E.(2005).Climate variability at Marion Island, Southern Ocean, since 1960.Journal of Geophysical Research‐Oceans, 110, C05007. https://doi.org/10.1029/2004JC002492

Rousset, F. (1997). Genetic differentiation and estimation of geneflow from F‐statistics under isolation by distance. Genetics, 145,1219–1228.

Rousset,F.(2008).Genepop’007:Acompletere‐implementationoftheGenepopsoftwareforWindowsandLinux.Molecular Ecology Resouces,8,103–106.https://doi.org/10.1111/j.1471‐8286.2007.01931.x

     |  3305CHAU et Al.

Sanmartín, I., & Ronquist, F. (2004). SouthernHemisphere biogeogra-phy inferredby event‐basedmodels: Plant versus animal patterns.Systematic Biology, 53, 216–243. https://doi.org/10.1080/10635150490423430

Schalke,H.J.W.G.,&VanZinderenBakker,E.M.(1971).Historyofthevegetation.InE.M.VanZinderenBakker,J.M.Winterbottom,&R.A.Dyer (Eds.),Marion and Prince Edward Islands (pp. 89–97). CapeTown,SouthAfrica:Balkema.

Schulze,B.R.(1971).TheclimateofMarionIsland.InE.M.VanZinderenBakker, J.M.Winterbottom,&R.A.Dyer (Eds.),Marion and Prince Edward Islands(pp.16–31).CapeTown,SouthAfrica:Balkema.

Scott,L.(1985).PalynologicalindicationsoftheQuaternaryvegetationhistoryofMarionIsland(Sub‐Antarctic).Journal of Biogeography,12,413–431.https://doi.org/10.2307/2844951

Scott, L., & Hall, K. J. (1983). Palynological evidence for interglacialvegetationcoveronMarionIsland,sub‐Antarctic.Palaeogeography Palaeoclimatology Palaeoecology,41,35–43.https://doi.org/10.1016/ 0031‐0182(83)90074‐3

Selkirk,D.R.,Selkirk,P.M.,Bergstrom,D.M.,&Adamson,D.A.(1988).RidgetoppeatsandpalaeolakedepositsonMacquarieIsland.Papers and Proceedings of the Royal Society of Tasmania,122,83–90.

Selkirk, P. M., Seppelt, R. D., & Selkirk, D. R. (1990). Subantarctic Macquarie Island. Environment and biology.Cambridge,UK:CambridgeUniversityPress.

Shaw,J.D.,Spear,D.,Greve,M.,&Chown,S.L.(2010).Taxonomicho-mogenizationanddifferentiationacrossSouthernOceanIslandsdif-fer among insects and vascular plants. Journal of Biogeography,37,217–228.https://doi.org/10.1111/j.1365‐2699.2009.02204.x

Smith, V. R., & Mucina, L. (2006). Vegetation of subantarctic MarionandPrinceEdwardIslands.InL.Mucina,&M.C.Rutherford(Eds.),The vegetation of South Africa, Lesotho and Swaziland (pp.698–723).Pretoria,SouthAfrica:SANBI.

Stevens, M. I., Greenslade, P., Hogg, I. D., & Sunnucks, P. (2006).SouthernHemispherespringtails:CouldanyhavesurvivedglaciationofAntarctica?Molecular Biology and Evolution,23,874–882.https://doi.org/10.1093/molbev/msj073

Taylor,B.W. (1955).A.N.A.R.E. Reports, Series B, Volume II, Botany: The flora, vegetation and soils of Macquarie Island.Melbourne,Australia:AntarcticDivision,DepartmentofExternalAffairs.

VanderPutten,N.,Verbruggen,C.,Ochyra,R.,Verleyen,E.,&Frenot,Y.(2010).Subantarcticfloweringplants:Pre‐glacialsurvivorsorpost‐glacialimmigrants?Journal of Biogeography,37,582–592.https://doi.org/10.1111/j.1365‐2699.2009.02217.x

Van Oosterhout, C., Hutchinson, W. F., Wills, D. P. M., & Shipley, P.(2004).MICRO‐CHECKER:Software for identifyingandcorrectinggenotypingerrorsinmicrosatellitedata.Molecular Ecology Notes,4,535–538.https://doi.org/10.1111/j.1471‐8286.2004.00684.x

Vekemans,X.,&Hardy,O.J.(2004).Newinsightsfromfine‐scalespatialgeneticstructureanalysesinplantpopulations.Molecular Ecology,13,921–935.https://doi.org/10.1046/j.1365‐294X.2004.02076.x

Wace,N.M.(1960).BotanyoftheSouthernOceanicIslands.Proceedings of the Royal Society of London Series B, Biological Sciences,152,475–490.

Whittaker,R.J.,Triantis,K.A.,&Ladle,R.J.(2008).Ageneraldynamictheoryofoceanic islandbiogeography. Journal of Biogeography,35,977–994.

Widmer,A.,&Lexer,C.(2001).Glacialrefugia:Sanctuariesforallelicrich-ness,butnotforgenediversity.Trends in Ecology and Evolution,16,267–269.https://doi.org/10.1016/S0169‐5347(01)02163‐2

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.   

How to cite this article:ChauJH,BornC,McGeochMA,etal.Theinfluenceoflandscape,climateandhistoryonspatialgeneticpatternsinkeystoneplants(Azorella)onsub‐Antarcticislands.Mol Ecol. 2019;28:3291–3305. https://doi.org/10.1111/mec.15147