EnvironmentalBiologyofFishes(2005)72:467–480óSpringer2005
Thecenterofthecenterofmarineshore?shbiodiversity:thePhilippineIslands
KentE.Carpentera&VictorG.SpringerbaDepartmentofBiologicalSciences,OldDominionUniversity,Norfolk,VA23529,U.S.A.(e-mail:kcarpent@odu.edu)bDivisionofFishes,DepartmentofZoologyMRC-159,NationalMuseumofNaturalHistory,SmithsonianInstitution,P.O.Box37012,Washington,DC20013-7012,U.S.A.
Received21April2004
Accepted12August2004
Keywords:Indo-Paci?cbiogeography,vicariance,islandintegration,conservationSynopsis
MultipledatasetsshowglobalmaximaofmarinebiodiversityintheIndo–Malay–Philippinesarchipelago(IMPA).Analysisofdistributiondatafor2983speciesrevealsapatternofrichnessona?nerscaleandidenti?esapeakofmarinebiodiversityinthecentralPhilippineIslandsandasecondarypeakbetweenpeninsularMalaysiaandSumatra.Thispatternisrepeatedindiversehabitatandhighertaxaclasses,mostrigorouslyformarineshore?shes,supportinggeohistoricalhypothesesasthemostgeneralunifyingexplanations.Speci?cpredictionsbasedonareaofoverlap,areaofaccumulation,andareaofrefugehypothesessuggestthatpresentdayeasternIndonesia,orWallacea,shouldbethecenterofmarinebio-diversity.Processessuggestedbythesethreehypothesescontributetothediversityinthisregionandarealsoalikelyexplanationforthesecondarycenterofdiversity.Ourstudyindicates,however,thatthereisahigherconcentrationofspeciesperunitareainthePhilippinesthananywhereinIndonesia,includingWallacea.ThePhilippinecenterofdiversityisconsistentwithhypothesesthatthisareaexperiencednumerousvicariantandislandintegrationeventsandthesehypotheseswarrantfurthertesting.Specialattentiontomarineconservatione?ortsinthePhilippinesisjusti?edbecauseoftheidenti?cationofitasanepicenterofbiodiversityandevolution.Introduction
TheIndo-Malay-PhilippinesArchipelago(IMPA)haslongbeenconsideredtheareaofhighestmarinebiodiversity,withdecreasinglatitudinalandlongitudinalgradientsinspeciesrichnessradiatingfromthiscenter(Bellwood&Wainwright2002,Moraetal.2003).Thereisawidevarietyofhypothesesthathavebeenpro-posedtoexplaintheremarkablediversityfoundintheIMPA(Rosen1988).Thesehypothesescanbeclassi?edintofourmaincategories:(1)areaofoverlap,(2)areaofaccumulation,(3)areaofref-uge,and(4)centeroforigin.Thereislittleagree-mentastowhichofthesehypothesesisthemostimportantinshapingdiversityintheIMPA
(Bellwood&Wainwright2002).Reconcilingtherelativeimportanceoftheseopposinghypotheseshasledsometoproposewhatcouldbeconsidereda?fthmajorhypothesis:thatacombinationofallprocessesevokedinthesehypothesesisresponsiblefortheextremebiodiversityoftheIMPA(Palumbi1996,Wallace1997,Carpenter1998,Randall1998,Wilson&Rosen1998,Allen&Adrim2003).PredictionsaboutconcentrationsofdiversitywithintheIMPAcanbemadebasedonthevari-oushypothesesandthegeographyandgeologyoftheIMPA.OneassumptioninthesepredictionsisthatlargeshelfareasoftheIMPAwereexposedandthereforeexperiencedaseriesoflocalmarineextinctionsduringPleistocenesea-levellows(Springer&Williams1990,Voris2000,Figure1)
468
Figure1.TheIndo-Malay-PhilippineArchipelagowiththeareashallowerthanthe120mdepthcontourdelineatedinlightgray.ThisareaestimatestheseabottomthatwouldhavebeenexposedduringPleistocenesea-levellows.MajorlandmassesincludethePalawan–Busuanga–Mindoroarchipelago(PBM),northernLuzon(NL),Visayas–Mindanao(VM),Halmahera(HA),Moluccas(MU),andSulawesi(S).Reproducedandmodi?edwithpermissionfromVoris(2000),copyrightFieldMuseumofNaturalHistory,Chicago.
andpotentiallylimiteddiversityintheseareas.WiththesePleistoceneexposedareasexcluded,thetwomajormarinehabitatareasremainingintheIMPAarethePhilippineIslandsandtheareare-ferredtoprimarilybyterrestrialbiologistsasWallacea.Wallaceaisatermnotfrequentlyusedbymarinebiogeographers(Wallace1997)andistypicallyde?nedastheareabetweenthefamedWallace’s(1860)lineandLydekker’sline(Simpson1977,Figure1).WeuseithereasaconvenienttermtorefertotheareawhichencompassestheislandgroupsofSulawesi,Moluccas,Halmahera,andtheLesserSundaIslands.Theareaofoverlap,areaofaccumulation,andareaofrefugehypoth-esessuggestWallaceaasanareaoflikelypeakmarinediversity.
AreaofoverlaphypothesescontendtheIMPAisanareawheredi?erentfaunasfromdi?erentoceansorlithosphericplatesaggregate.OverlapbetweenPaci?candIndianOceanfaunas(Woodland1983,Donaldson1986)wouldcon-centratediversityintheWallaceancorridorbe-causeitiscentrallylocatedandhasamplehabitattosupportspecies.Throughoutthelaterhalfofthe
Cenozoic,elementsofWallaceawerepositionedacrossthecentraldispersalroutebetweenthePa-ci?candIndianoceans.Upuntilthemid-Miocene,thesouthequatorialcurrentspannedthePaci?candIndianoceans(Kennettetal.1985,Figure2).DuringtheOligoceneandmuchoftheMiocene,elementsofWallaceahadmostlyformedandwereinthepathoftheIndonesianSeawaythatcon-nectedthePaci?candIndianoceans(Hall1998).TheIndianOceansouthequatorialcurrentwouldalsohavebeenincontactwithelementsofWall-acea(Figure2).
AustralianandEurasiancontinentalshelvescollidedandde?ectedthesouthernequatorialcurrentssomewherebetween16and8mya(Kennettetal.1985)andformedanequatorialbarrierbetweenthePaci?candIndianoceans.Thistransformedthesouthernequatorialcurrentsintoboundarycurrentsandthesurfacecurrentpatternsseentodaybegantotakeshape(Figure3).ThroughoutthelateMioceneandPli-ocenethemajorislandgroupsofWallaceawereintheprimarypathwayofcurrents?owingeastwardfromtheIndianOceanandwestwardfromthePaci?cOcean(Hall1998).DuringtheHoloceneandtimesofPleistocenesea-levelhighs,IndianandPaci?cOceanexchangeoccurredprimarily
Figure2.Positionofland(diagonallines),shallowseas(gray),anddeepocean(nocolor)inSoutheastAsia30millionyearsago.Relativepositionofpresent-daylandmassesaredrawnwithrespecttotheseancientfeatures,includingthePalawan–Busuanga–Mindoroarchipelago(PBM),northernLuzon(NL),Visayas–Mindanao(VM),Halmahera(HA),Moluccas(MU),andfourcomponentsofSulawesi(S1–S4).ModifedfromHall(2002).
Figure3.Present-daysurfacecurrentpatternsintheIMPA.(a)Winter,(b)Summer.Lightgrayindicatesareasshallowerthanthe120mdepthcontourthatwouldhavebeenexposedduringextremePleistocenesea-levellows[currentpatternsmodi?edfromMorgan&Valencia(1983),basemapmodi?edfromVoris(2000)].
throughWallacea(Flemminger1986,Fineetal.1994,Figure3).DuringPleistocenesea-levellows,theonlypathwaybetweenequatorialPaci?candIndianoceanswasthroughWallacea(Fleminger1986,Figure3).
TheIMPAisalsoconsideredanareaofover-lapwithdi?erentendemicshorefaunas‘rafting’intotheareathroughthemovementof
469
lithosphericplates(Pandol?1992,Wilson&Ro-sen1998,Santini&Winterbottom2002).Colli-sionzonesbetweenplateswouldconcentratedi?erentfaunas.Aprimaryoverlapareawouldbethetriplejuncturezone,locatedineasternWallacea(Figure4).
Theareaofaccumulationhypothesis(alsocalledthevortexhypothesis)suggestsmostspeci-ationoccurredonremotePaci?cislandsandpre-vailingcurrentsconcentratedspeciesintheIMPA(Ladd1960,Jokiel&Martinelli1992,Connollyetal.2003).ApredictionfromthishypothesisisthatWallaceawouldbeacenterofdiversitywithintheIMPAbasedonprevailingcurrents.ThemiddletolateCenozoicPaci?cequatorialcurrentsprimarily?owtowardselementsofWallacea(Figures2and3,Kennettetal.1985,Hall1998).
RelatedtotheareaofaccumulationhypothesisisthattheIMPAalsoservesasanareaofrefuge(Bellwood&Hughes2001)becauseitencompasses
Figure4.Present-daylocationofmajorlithosphericplatesoftheIMPAasdelineatedbymajorsubductionzones(blacklines).ThedashedlineindicatestheapproximatelocationoftheriftzonethatseparatedtheIndianandAustralianplatesintheearlyCenozoic.
470
themostextensiveanddiversetropicalshallowwatermarinehabitatonearth(Carpenter1998).Thesigni?canceofextensivetropicalrefugiarelatestotheecologicalmaximthatlargerareasgenerallysupportmorespeciesthansmallerareasandthattropicallowlatitudehabitatshavegreateravailableenergytosupportbiomassthancoolerlatitudes(Gaston2000).TheareaofrefugehypothesiswouldpredictthattheIndonesian–MalaysianpartoftheIMPAwillhavethehighestspeciesrichness.Thisareahasthegreatestextentonearthoftropicalshallowwaterhabitat(Carpenter1998),includingthegreatestextentofcoralreefs(Spaldingetal.2001)anddiversityofhabitattypes(Woodland1990,Randall1998).Thecenteroforiginhypothesispredictsthatareasofhighestdiversitywouldbeareasthathostednumerousallopatricspeciationevents(McManus1985,Briggs1999a,Moraetal.2003).Thecenteroforiginhypothesishasalsobeenproposedintermsofnumerouscompetitiveorsympatricspeciationevents(Briggs1999b)butthismechanismhasbeendismissedasanimportantmeansofspeciationintheIMPA(Springer&Williams1990,Santini&Winterbottom2002).BothWallaceaandthePhilippinesarepotentialareasofconcentratedallopatricspeciation.Twomajorvicariantpro-cessesoccurredwithintheIMPAduringtheCenozoicwhenmostoftheshore?shfaunaevolved.Thesewere:(1)potentialisolationofseasduringPleistocenesea-levellows(McManus1985)and(2)geologicaloriginthroughcomplextectonicmovements(Hall2002).AnumberofseaswithinWallaceaandthePhilippineswerepotentiallyisolatedfromoneanothertovaryingdegreesduringPleistocene?uctuationsinseale-vel.TheseincludedtheSouthChina,Sulu,Phil-ippine,Celebes,Molucca,andBandaseas(Figure1).TheoriginsofthePhilippineandWallaceanislandarcsareparticularlycompli-catedbecauseofintegrationofvariedPhilippineplateelements(Hall2002).
ThepurposeofthispaperistoexamineresultsofageographicalinformationsystemanalysisofmarinespeciesfoundinthewesterncentralPaci?cOceanwithrespecttohypothesesproposedtoexplainthebiodiversityfoundintheIMPA.Wealsocommentontheimportanceofthese?ndingswithrespecttoconservation.
Methodsandresults
Thegeographicalinformationsystemdatabaseandanalyses
Weusedageographicalinformationsystem(GIS)overlayof2983generalizeddistributionsofmarinespecies(Carpenter&Niem1998–2001)toexaminethepatternofdiversityintheIMPA.Speciesincludedseaweeds(62),corals(27),bivalves(189),gastropods(249),cephalopods(87),stomatopods(13),shrimps(110),lobsters(47),crabs(73),ho-lothurians(18),sharks(150),batoid?shes(116),chimaeras(6),bony?shes(1775),estuarinecroc-odile(1),seaturtles(6),seasnakes(21),andmarinemammals(33)producedby84specialistsintheirrespectivetaxonomicgroups(Carpenter&Niem1998–2001andauthorscitedtherein).Taxawerefromcontinentalshelfandepipelagicenvi-ronments.Asubsetofthisdatabasecontainsthe2047shore?shspeciesthatincludedElasmo-branchii(sharksandbatoid?shes),Holocephali(chimaeras),Sarcopterygii(Sulawesicoelacanth),andrepresentativesfrom24ordersofTeleostei.Themixofspeciesinthisstudyincludesmostlycommon?sheriesspecieswhosedistributionsarewellknownthroughouttheIMPA.Itlargelyexcludesspeciessuchassmallergobioidorblen-nioid?shesforwhichdistributionsaremainlyderivedfromsamplinginhighlyrestrictedareas.Indonesiahasarichhistoryofresearchon?shesbeginningintheearly19thcentury,includingverydiversecollectionsmadebytheDutchnaturalhistorian,PieterBleeker(Weber&deBeaufort1913).ExtensivecollectionsaroundIndonesiacontinuedthroughthepresentincludingnotablesurveysoftrawled(Gloerfelt-Tarp&Kailola1984)andcoralreef?shes(Allen&Adrim2003).TherewerelimitedstudiesofPhilippine?shesinthe19thcentury(Herre1953)butextensivecollectionsbe-ganintheearly20thcentury(Smith&Williams1999).BecauseoftheextensivecollectionsinbothIndonesiaandthePhilippines,distributionrecordsforcommon?sheriesspecieswouldbethoroughlycoveredinbothcountries.And,sincethedistri-butionrecordsarebasedprimarilyonmuseumcollectionsandnot?sheriesstatistics(taxonomistsconsiderthesemostlyunreliableunlesstiedtoamuseumspecimen),uneven?sheriespressurealsowouldnotin?uencethesedistributions.Wedonot,
therefore,suspectsamplingbiasinourdatasetwithintheIMPA.
Onepotentialsourceofbiasinthecurrentdatasetistaxonomicbiasinthat?shesandverte-bratesarecoveredingreaterproportiontooveralldiversitythaninvertebrates.FarmorespeciesofinvertebratesmakeupthediversityoftheIMPAthanthe?shesbutthesearenotasimportantproportionallyin?sheries.Thisisparticularlythecasewhenitcomestocoralsthatmakeupacon-spicuouspartofthemacrofaunalinvertebratefaunaoftheIMPA.Taxonomicbiashasbeencriticizedinotherbiogeographicstudies(Bairdetal.2002,Robertsetal.2002).Weexamineallthemapsincombinedanalysesandalsoexamineapartitionincludingonlytheshore?shessincethislatergroupwascoveredmorerigorously.
All2983generalizedmapsweredigitizedinPCArcInfoandanalyzedusingArcView(Environ-mentalSystemsResearchInstitute,Inc.)withacellsizesetat10km·10km.Specieswerechosenbasedontheirlikelihoodtoenter?sheriesinthewesterncentralPaci?cFAO?shingarea71andtheSouthPaci?cCommissionArea(Carpenter1998).ThisregioncorrespondstomostofthetropicalandsubtropicalwesternPaci?cfromabout98°Eto122°WlongitudeexcludingtheHawaiianIslandsandJohnstonIsland.
Thedistributionofeachspecieswascategorizedaccordingtoconcordancewithamajorlitho-sphericplate(Figure4).Ifthemajorportionofaspeciesdistributionwasfoundononelithosphericplatebutitextendedmarginallyontoanotherplateitwasconsideredendemiconlytotheplatethatitcoveredmostcompletely.TheseplatesweretheEurasianplate(includingthePhilippineandMoluccanmarginalarcsystems),thePaci?cplate,theIndianplate,andtheAustralianplate.TheIndianandAustralianplatesarecurrentlyconsideredasingleplate.However,theseplateswereseparatedduringtheearlyCenozoic(Hall2002,Figure4)andtheshorefaunasontheseplateswerewidelyseparatedandpresumablyevolvedseparately.Inaddition,theIndianpor-tionoftheIndo-Australianplatewasnotdirectlyencompassedinthisstudy.ThePhilippineSeaplateisalsoamajorlithosphericplatebutismostlyopenseawithfewidenti?ableendemics(Myers1989)whenitsmarginalarcislandsystemsareconsideredpartofEurasianorAustralian
471
shorefaunas.Distributionscoveringmorethanincidentallyonelithosphericplatewerecatego-rizedas:Indian–Eurasian,Eurasian–Australian,WesternPaci?c(distributionsthatencompassedinclusivelytheEurasian,Paci?c,Philippine,andAustralianplates),Indo-westmost(speciesfoundinclusivelyontheIndian,Australian,andEur-asianplatesbutnotextendingontothePaci?cplate),andIndo-WestPaci?c(widespreadonallIndianandPaci?coceanplates).Habitattypewascategorizedastoeitherprimarycoralreef,pri-marysoftbottom,rocky,estuarine,generalizeddemersal,generalizedneritic,andepipelagichab-itattypesifthiswasindicatedinthehabitatsectionofeachspeciesaccount(Carpenter&Niem1998–2001).
ResultsfrompreliminaryGISanalyses
Analysisofthe2983combinedrangesrevealsthecentralPhilippinesastheareaofhighestdiversityandendemism(Figure5a,b).AsecondaryareaofhighdiversityislocatedbetweenthetipofMalaysiaandSumatraandextendsalongnorth-easternSumatraandnorthernJava(Figure5a).Bothdiversitycentersarerepeatedinsubsetsofdatabasedondistribution,habitat,allinverte-bratetaxa,andshore?shes(Figure5c–f,Table1).Familiesandgenerawithpredominantlysmall-bodiedspecies,whichtendtohaveahigherpro-portionofrestricted-rangeendemicsthantaxawithlarge-bodiedspecies,werenotincludedinourstudy.Nevertheless,120restricted-rangeendemicswereincludedbecausetheyenterarea?sheries.Thegreatestconcentrationoftheserestricted-rangeendemicsisinthePhilippines(Figure5b)whichhas38.Indonesia/Malaysiahas19suchendemics,Australia18,NewGuinea/Bismark/Louisade18,CoralSea/NewCaledonia/Vanuatu17,andsevenotherlocalitieshadeitheroneortwoendemics.Analysisofdistributionandhabitattypesintheprimaryandsecondarycenterofdiversity
Toexaminethecomponentsofbiodiversityweanalyzedthefrequenciesofgeneraldistributionandhabitattypesintheprimaryandsecondarycentersofdiversity.Weusedtheassemblageofspeciesinthesingle10km·10kmpixelwiththemostspecies(1736orabout58%ofallspeciesin
Environmental Biology of Fishes (2：鱼类环境生物学(2
