EVR 3740 500-word narrative

environmental science writing question and need an explanation and answer to help me learn.

500-word narrative based on 3 articles.7 pages each.
Requirements: 500 words minimum
LETTERSPUBLISHEDONLINE:14JULY2013|DOI:10.1038/NCLIMATE1944Coastalhabitatsshieldpeopleandpropertyfromsea-levelriseandstormsKatieK.Arkema1*,GregGuannel2,GregoryVerutes3,SpencerA.Wood2,AnneGuerry2,MaryRuckelshaus2,PeterKareiva4,MartinLacayo2andJessicaM.Silver2Extremeweather,sea-levelriseanddegradedcoastalecosystemsareplacingpeopleandpropertyatgreaterriskofdamagefromcoastalhazards1–5.Thelikelihoodandmagnitudeoflossesmaybereducedbyintactreefsandcoastalvegetation1,especiallywhenthosehabitatsfringevulnerablecommunitiesandinfrastructure.Usingfivesea-level-risescenarios,wecalculateahazardindexforevery1km2oftheUnitedStatescoastline.Weusethisindextoidentifythemostvulnerablepeopleandpropertyasindicatedbybeingintheupperquartileofhazardforthenation’scoastline.Thenumberofpeople,poorfamilies,elderlyandtotalvalueofresidentialpropertythataremostexposedtohazardscanbereducedbyhalfifexistingcoastalhabitatsremainfullyintact.CoastalhabitatsdefendthegreatestnumberofpeopleandtotalpropertyvalueinFlorida,NewYorkandCalifornia.Ouranalysesdeliverthefirstnationalmapofriskreductionowingtonaturalhabitatsandindicateswhereconservationandrestorationofreefsandvegetationhavethegreatestpotentialtoprotectcoastalcommunities.Globally,coastalfloodingandsealevelareexpectedtoincreasesignificantlybymid-century,withpotentiallysevereconsequencesforcoastalpopulationsaroundtheworld6.IntheUnitedStateswhere23ofthenation’s25mostdenselypopulatedcountiesarecoastalthecombinationofstormsandrisingseasisalreadyputtingvaluablepropertyandlargenumbersofpeopleinharm’sway15.Thetraditionalapproachtoprotectingtownsandcitieshasbeento`harden’shorelines.Althoughengineeredsolutionsarenecessaryanddesirableinsomecontexts,theycanbeexpensivetobuildandmaintain7,8,andconstructionmayimpairrecreation,enhanceerosion,degradewaterqualityandreducetheproductionoffisheries9,10.Overthepastdecade,effortstoprotectpeopleandpropertyhavebroadened11toconsiderconservationandrestorationofmarshes,seagrassbeds,coastalandkelpforests,andoysterandcoralreefsthatbuffercoastlinesfromwavesandstormsurge1214andprovidecollateralbenefitstopeople15.Butapproachesandtoolsforevaluatingthepotentialroleofnaturaldefencemechanismslagbehindthoseforhardeningshorelines15.Prioritizingecosystemsforconservationorrestorationinserviceofnaturalhazardreductionrequiresknowledgeofwherehabitatsaremostlikelytoreduceexposuretoerosionandfloodingfromstormsandfuturesealevels,andprotectvulnerablepeopleandproperty(seeSupplementaryInformationfordefinitions1TheNaturalCapitalProject,StanfordUniversity,890725thAveNESeattle,Washington98115,USA,2TheNaturalCapitalProject,StanfordUniversity,371SerraMall,Stanford,California94305-5020,USA,3TheNaturalCapitalProject,StanfordUniversity,c/oConservationScienceProgramWorldWildlifeFund—US,125024thStreetNW,WashingtonDC20037-1193,USA,4TheNatureConservancy,4722LatonaAveNE,Seattle,Washington98105,USA.*e-mail:karkema@stanford.eduofvulnerabilityandsoon).Previouseffortshavemappedphysicalvulnerabilityofcoastalareasusingdataandforecastsforsea-levelriseandstormsurge16,17andusedsocialmetricsofvulnerability18toidentifywhereconsequencesofphysicalhazardswillbegreatestforpeople2,19.Missing,however,isasynthesisofhazardmodels,climatescenarios,demographicinformationandecologicaldatatoidentifywherehabitatsmaycontributetoprotectionfromcoastalhazards.EventssuchasHurricaneSandy,whichdevastatedthenortheastUnitedStatesinOctober2012,demonstratethedesperateneedforsuchananalysistoinformplanningandyieldcoastalregionsmoreresilienttotheexpectedeffectsofclimatechange20.Toidentifythestretchesofshorelinewherehabitatshavethegreatestpotentialtodefendcoastalcommunitiesagainststormsandsea-levelrise,wecreatedahazardindexthatincorporatestheprotectiveroleofecosystemsfortheUSshorelineata1km2scale(SupplementaryFig.S1).Wecompiledanationwidemapofthemaincoastalhabitats,designedtwohabitatscenarios(withandwithouthabitat)andfivescenariosofcurrentandfuturesealevel,andidentifiedareaswiththehighestexposuretoinundationanderosionusingphysicaldataandmodels16,17,21(seeMethodsandSupplementaryInformation).Next,weconvertedhazardtoimperiledhumanlifeandpropertybymappingexposureofthepeople,poorfamilies,elderlypopulations22andresidentialpropertyvalues23ineach1km2segmentofthecoastline.Todeterminethereductioninriskofdamagesprovidedbyhabitatstocurrentstormintensitiesandthefivescenariosofcurrentandfuturesealevel24,wemodelledthenumberofpeopleandtotalvalueofpropertyhighlyexposedtohazardswithandwithouthabitats.Byquantifyingwhereandtowhatextenthabitatsreducetheexposureofvulnerablepopulationsandproperty,ouranalysesare,tothebestofourknowledge,thefirsttotargetwhereconservationandrestorationofcoastalhabitatsaremostcriticalforprotectinglivesandpropertyonanationalscale.Weassessedcoastalvulnerabilitynowandinthefuturebyestimatingthehazardindexona1km2scalefortheentirecoastlineacrosstenscenariosvaryinginsea-levelriseandpresenceofhabitats(noriseandfourUSNationalClimateAssessmentscenariosofrise24bothwithandwithouthabitat;seeMethods,SupplementaryInformationandSupplementaryFig.S2).Fromthefrequencydistributionof1,007,020(rangingfrom1.05to4.84),weidentifiedtheupperquartile(`highhazard’)asgreaterthan3.36(SupplementaryFig.S3).Today16%oftheUScoastlinecomprisesNATURECLIMATECHANGEjVOL3jOCTOBER2013jwww.nature.com/natureclimatechange913
LETTERSNATURECLIMATECHANGEDOI:10.1038/NCLIMATE1944Figure1jCoastalhabitatsreducebyapproximately50%theproportionofpeopleandpropertyalongtheUScoastlinethataremostexposedtostormsandsea-levelrise.Weestimatepeopleandpropertyexposedtohazardswith(blackbars)andwithout(whitebars)habitatsusingfourmetrics:totalpopulation,elderlypeople,poorfamilies(threeleftyaxes)andresidentialpropertyvalues(rightyaxis).Resultsarerepresentedusingthesamesetofbarsforallmetricsbecauseonthenationalscalethesevariablesarehighlycorrelated.Thecorrelationbreaksdownonmorelocalscales(Figs3,4).Dataareforhighesthazardsegments(index>3:36):`highhazard’areas,harbouring1.3millionpeople,250,000elderly,30,000familiesbelowthepovertylineandUS$300billioninresidentialpropertyvalue(Fig.1).Akeyquestionthatariseswithanindexofmodelledhazardiswhetherobservedandpredictedspatialvariationindamagesarecorrelated.Tocompareourcoastalhazardindexwithfindingsfromempiricalstudies,wesynthesizeddatafromtheSpatialHazardsEventsandLossesDatabasefortheUnitedStates(ref.25).Usingstate-leveldatafrom1995to2010,wefoundahighlysignificantpositiverelationshipbetweenourmodelledestimatesoftotalpopulationexposedtothegreatestcoastalhazard(currentscenarioonly;upperquartile>3:14)andtheobservednumberofcoastalhazard-relatedfatalities(ND21states,R2D0:70,P<0:0001,totalcoastalhazardsD1,270,totalcoastalhazard-relatedfatalitiesD527;seeSupplementaryInformation).Toassessfuturevulnerability,weexaminedresultsfromthehazardindexandestimatedrisktopeopleandpropertyunderfoursea-level-risescenariosfortheyear2100(ref.24).Acrossallfuturescenarios,ourresultssuggestthatmorecoastalsegmentswillbehighlyexposedtohazardsandthattheamountofhighlythreatenedpeopleandpropertywillincreaseby3060%overthecurrentscenario(Fig.1).Givenmodelledsea-levelriseandobservedstormcharacteristics,1.7to2.1millionoftoday'spopulationwillliveinareasexposedtothehighesthazard(Fig.1).Between30,000and40,000familiesbelowthepovertylineandUS$400toUS$500billionofresidentialpropertywillbemostexposedtofuturehazards(Fig.1).Ofcourse,bothpropertyvaluesandpopulationsalongthecoastareexpectedtogrow;thus,ourstudyprobablyunderestimatesthenumberofpeopleandvalueofpropertyexpectedtobeinharm'swayby2100.Becauseouranalysisincludesthevalueofonlyresidentialunits,notcommercialproperties,itunderestimatesthetotalvalueofpropertyexposedtodamagefromcoastalhazards.Todeterminetheextenttowhichhabitatsprovideprotection,wecomparedestimatesofriskforthefivesea-level-risescenarioswithandwithoutthepresenceofninehabitatsthatfringetheUnitedStates:coastalforests(forexample,mangrovesandothercoastaltreesandshrubs),coralreefs,emergentmarsh,oysterreefs,highandlowdunes,seagrassbeds,kelpforestsandadditionalintertidalaquaticvegetation(seeSupplementaryFig.S4).Wemodelledthecompletelossofhabitattoidentifywherehabitatsreducetheexposureofpeopleandpropertytohazards.Atpresent,habitatsprotect67%ofthecoastline,ashazardincreasesintwo-thirdsofallsegmentsinthescenariowithouthabitat.Habitatlosswoulddoubletheextentofcoastlinehighlyexposedtostormsandsea-levelrise(hazardindex>3:36),makinganadditional1.4millionpeoplenowlivingwithin1kmofthecoastvulnerable.Thenumberofpoorfamilies,elderlypeopleandtotalpropertyvaluehighlyexposedtohazardswouldalsodoubleifprotectivehabitatswerelost(Fig.1).VulnerabilitytocoastalhazardsandtheimportanceofnaturalhabitatsvaryacrosstheUnitedStates.Forallclimatescenarios(SupplementaryFig.S5),theeastandgulfcoastsaremorephysicallyvulnerabletosea-levelriseandstormsthanthewestcoast(shownforA2inFig.2).Regionswithgreaterexposuretohazardshaveagreaterpercentageoflow-reliefcoastalareaswithsoftersubstrates(forexample,beaches,deltas),higherratesofsea-levelriseandpotentialforstormsurge(SupplementaryFigsS7andS8).Largeexpansesofcoastalforestsandwetlands,oysterandcoralreefs,dunesandseagrassbeds(SupplementaryFig.S4)arecriticalforprotectingtheeasternseaboardandGulfofMexicofromstormsandsea-levelrise(compareSupplementaryFigsS5andS6).Atthestatelevel,habitatsprotectthegreatestextentofcoastlineinFlorida,NorthCarolinaandAlaska(shownforA2inSupplementaryTableS7).Althoughcoastalecosystemsaremostimportantforreducingexposuretohazardsintheaforementionedstates,theyprovideprotectionforthegreatestnumberofpeople,sociallyvulnerablepopulationsandpropertyinFlorida,NewYorkandCalifornia(seedifferencebetween`withhabitat’and`withouthabitat’inFig.2bandSupplementaryTableS7forothermetrics).Todeterminewherehabitatsarelikelytobecriticalforprotectingthemostvaluablecoastlinenowandunderfuture914NATURECLIMATECHANGEjVOL3jOCTOBER2013jwww.nature.com/natureclimatechange
NATURECLIMATECHANGEDOI:10.1038/NCLIMATE1944LETTERSFigure2jExposureoftheUScoastlineandcoastalpopulationtosea-levelrisein2100(A2scenario)andstorms.Warmercoloursindicateregionswithmoreexposuretocoastalhazards(index>3:36).Thebargraphshowsthepopulationlivinginareasmostexposedtohazards(red1km2coastalsegmentsinthemap)withprotectionprovidedbyhabitats(blackbars)andtheincreaseinpopulationexposedtohazardsifhabitatswerelostowingtoclimatechangeorhumanimpacts(whitebars).LettersonthexaxisrepresentUSstateabbreviations.Datadepictedintheinsetmapsaremagnifiedviewsofthenationwideanalysis.Figure3jNature’sshieldfortotalresidentialpropertyvalue.a,b,Totalpropertyvalueforwhichhabitatsreduceexposuretostormsandsea-levelriseineachcoastalcountyoftheUnitedStatesforthecurrent(a)andfutureA2(b)sea-level-risescenarios.InsetsshowMonroeCountyinFlorida,GeorgetownandHorrycountiesinSouthCarolinaandBrunswickandPendercountiesinNorthCarolina.Reductioninthetotalvalueofpropertyexposedtocoastalhazardsisthedifferenceinthetotalvalueofpropertyexposedtocoastalhazardswithandwithouthabitatsincludedinthehazardindex.Estimatesforeach1km2segmentinthehighesthazardcategory(index>3:36)aresummedbycounty.climatescenarios,weestimatedthedifferenceintotalpropertyvalueexposedtocoastalhazards,withandwithouthabitats,onacountyscale.Variationamongcountiesinthevalueofpropertynowprotectedbycoastalhabitatsissubstantial,rangingfromUS$0(forexample,Jefferson,Florida),tomorethanUS$20billioninSuffolkandKings,NewYork(Fig.3a).TherearealsodifferencesNATURECLIMATECHANGEjVOL3jOCTOBER2013jwww.nature.com/natureclimatechange915
LETTERSNATURECLIMATECHANGEDOI:10.1038/NCLIMATE1944Figure4jNature’sshieldforsociallyvulnerablecounties.a–d,Proportionofpoorfamilies(a,b)andelderlypeople(c,d),relativetothetotalpopulationineachcountrythatareprotectedbyhabitatsfromexposuretocurrent(a,c)andfutureA2(b,d)sea-levelriseandstorms.Cut-offsforhigh(upper25%),medium(centre50%)andlow(lower25%)proportionsarebasedonthequantilesofthetwodistributions(ratioofpoororelderlytototalpopulation)acrossthetwosea-level-risescenarios.inthepotentialimportanceofhabitatsforprotectionassealevelsrise.Forexample,iftheextensivecoral,mangroveandseagrassecosystemsthatlineFloridaatpresentpersistinthefaceofdevelopmentandclimatechange,ouranalysispredictsthesehabitatswillreduceexposureofnearlyUS$4billionworthof2010homepropertyvalueswithin1kmofthecoastlineby2100upfromUS$0.7billionatpresent(Fig.3a,binsets).Inothercountiessea-levelrisewilloverwhelmcoastalhabitats,reducingpropertyprotection(Fig.3insets).Focusingsolelyonpropertyvaluemaycausedecision-makersandplannerstooverlookecosystemsthatprovidedisproportionateprotectiontovulnerablepopulations.Forexample,habitatsprotectmorepoorfamiliesrelativetothetotalpopulationinTexas(Fig.4a,b)andmoreelderlyandtotalpropertyvalueinFlorida(Figs3and4c,d).Thus,onthecountyscale,thegreatesthazardprotectionfromhabitatsforpoorfamiliesalongtheGulfcoastoccurswheretherearedisproportionatelyfewerelderlyandlowertotalpropertyvalue.Thesefindingsreflecttheco-locationofhighpropertyvalueandvulnerablepeopleinsomeregionsandtheirindependenceinotherregions.AroundtheworldandtheUnitedStates,coastaldefenceplanningisbeginningtoincorporateecosystemsalongsidephysicalstructures.IntheaftermathofHurricaneSandy,callsforenhancingtheresilienceofNewYorkCityhaveincludedrestorationofoysterandwetlandhabitats26.Louisiana’s2012masterplantocombinenaturalandengineeredstrategiesforprotection11isexemplaryofsuchefforts.Thesepioneeringinitiativeswillprobablybeemulatedbyotherregions.Ourresultssuggestthattheextenttowhichnaturaldefencemechanismsoperatedependsontherelativelocationofthehazard(forexample,sea-levelrisehotspots)5,habitats,vulnerablepopulationsandproperties.Questionsabouttheadaptation(orlackthereof)ofhabitatstoclimatechange(forexample,wetlandsmigratingwithsea-levelrise)andhownumeroushabitats(forexample,oystersandmarshes)functiontogethertoreduceexposure26deservefurtherattention.Moreworkisneededtoidentifywherecombiningecosystem-basedandengineeredapproacheswillbemosteffectiveforreducingdamages.Owingtodatalimitationsonanationalscale,wecombinedphysicalstructuresandgeomorphologyintoasinglevariable,whichprecludescomparisonsofgreenandgreysolutions(seeSupplementaryInformation).Afullcost-benefitanalysisofalternativeswillbemostusefulonlocalscalesandrequirequantitativeecological,surgeandwavemodelscombinedwithvaluationofasuiteofecosystemservices.TheauthorsareengagedinsuchworkinTexas,USAandBelize.Theindexwedevelopedismostusefulonnationalandregionalscalesforprioritizinghabitatsforcoastaldefence.Ouranalysisilluminatesthatlossofexistingecosystemswillresultingreaterdamagetopeopleandpropertyorwillrequiremassiveinvestmentsinengineereddefences.Identifyingthebestlocationstotargetforecosystem-basedstrategiesdependsonwherehabitatseffectivelyreducehazardsandwherepeoplebenefitthemost,bothnowandunderfutureclimate.916NATURECLIMATECHANGEjVOL3jOCTOBER2013jwww.nature.com/natureclimatechange
NATURECLIMATECHANGEDOI:10.1038/NCLIMATE1944LETTERSMethodsDesignofsea-level-risescenarios.Wedevelopedonecurrentandfourfuturesea-level-risescenariosfor2100forthecoastoftheUSusinglong-termtide-gaugedataandguidancefromthe2013USNationalClimateAssessment:`current’isbasedonobservedratesofsea-levelrise,`trend’representstheprojectionoftheobservedriseto2100,`B1’and`A2’arebasedinpartontheSpecialReportonEmissionsScenarios27and`high’incorporatesglacierandice-sheetcontributions24(seeSupplementaryFig.S1).Tocalculatelocalestimatesofsea-levelriseforeachscenarioweassignedeach1km2segmenttotheclosesttidegauge28.Weestimatedthecurrentsea-level-risescenarioastheincreaseinwaterelevationfrom1992to2006usingthelong-termobservedrateforeachtidegauge28.Predictedoutcomesforthefourfuturescenarioswereglobalrisefor2100predictedbytheNationalClimateAssessment(0.2,0.5,1.2,2m;ref.24),multipliedbyascalingfactor(theratioofthehistoricallocalratetothehistoricalglobalrate(1:8mmyr1);refs24,28).Designofhabitatscenarios.Toevaluatetheroleofcoastalecosystemsinreducingexposuretosea-levelriseandstorms,wedevelopedtwohabitatscenarios.`Withhabitat’includesninehabitatsinthehazardindex(SupplementaryFig.S4).`Withouthabitat’assumesthosehabitatsnolongerprovideprotection.Thehabitatscenarioisassumedtobethecurrentstateofthesystem.The`withouthabitat’scenarioisnotintendedtobeaplausiblereflectionofthefuture.Instead,weusedittoevaluatewhereandtowhatextenthabitatsplayasignificantroleinprotectingpeopleandproperty,andtodeterminewheretheirlosswouldaffectriskfromcoastalhazards.Calculatingcoastalhazard.Toestimatetherelativeexposureofeach1km2segmentoftheUScoastlinein2100andtodaywithandwithouthabitats(foratotalof1,007,020segments),wecalculatedanindexforcoastalerosionandinundationusingthecoastalvulnerabilitymodelinInVEST,anopen-sourcetoolavailableatwww.naturalcapitalproject.org.Thetoolbuildsonpreviousapproaches16,17byspecificallyincludingtheroleofhabitatsinprovidingprotection.Theindexalsoincludestheeffectofstormsonexposurebyincorporatingobserveddataonwind,waves29andsurgepotential,aswellasdataandmodelsforfourotherkeyvariables:habitattype,shorelinetype,reliefandsea-levelrise(SupplementaryInformation).Owingtouncertaintyamongmodelsandstudiesabouttherelationshipbetweenwavesandclimatechange30,wemadethesimplifyingassumptionthatstormintensityandfrequencyin2100willbethesameasthecurrentscenario.WeestimatedcurrentwaveandwindexposurebasedonsixyearsoftheNationalOceanographicandAtmosphericAdministrationWAVEWATCHIIImodelhindcastreanalysisresultsfor20052010(ref.29).WefollowedtheNationalOceanographicandAtmosphericAdministration’senvironmentalsensitivityindexshorelineclassificationschemeandassumedthatseawallshavethesamerankasrockycoastlinesandcliffs(SupplementaryTableS1).Thissimplification,whichineffectcombinesstructuresandgeomorphologyintoshorelinetype,isanartefactofthelimitationsofthenationwidedatasetandanalysis,andshouldbeaddressedinfutureresearch.Usingobservedandmodelleddata,wegeneratedabsolutevaluesforeachvariableforeach1km2segmentofcoastline.Wethenrankedeachvariableforeachsegmentfromlow(rankD1)tohigh(rankD5)exposure(seeSupplementaryTableS1).HazardIndexD(RHabitatsRShorelineTypeRReliefRSLRRWindRWavesRSurgePorential)1=7Weweightedallvariablesequally,afterseveralothercoastalvulnerabilityindices16,17.Theresultsaretherelativeexposuretocoastalhazardofeach1km2segmentcomparedwithallothersegmentsnationwideandacrossthetenhabitat-by-climatescenarios(seeSupplementaryFig.S2).Tomaphazardweclassifiedthedistributionofresultsforallsegmentsandscenarios(rangingfromonetofive)intoquartilesthatdemarcateareasofhighest(>3:36Dupper25%),intermediate(2.363:36Dcentral50%)andlowesthazard(<2:36Dlower25%,seeSupplementaryFig.S2).Quantifyingrisk.Toconverthazardtoimperiledpropertyandhumanlifewecombineditwithmappeddataondemographics22andpropertyvalues23ineach1km2segmentoftheentirecoastline.WeusedZillow'sHomeValueIndex23,whichisthemedianmarketvalueofresidentialpropertiesineachUS2010censusblockgroupandfiveyears(20062010)oftheCensusBureau'sAmericanCommunitySurveydata22.Wedistributeddataforpeopleandpropertiesthroughoutthecensusblockgroupataresolutionof30mwithadasymetricmappingapproach31thatusesland-use,land-coverandlandstewardshipdata(indicatinguninhabitedpubliclands)toidentifywherepeoplearemostlikelytolive.Wethenestimatedthetotalpopulation,numberofpeopleolderthan65years,numberoffamiliesbelowthepovertylineandmedianvalueofpropertiesin1km2segmentsclassifiedashighesthazard.Validationofcurrentcoastalhazardrisk.Toassesstheabilityofthehazardindextocapturerisk,wecomparedourestimatesforpopulationexposedtohighesthazardwiththeobservednumberofcoastalhazard-relatedfatalitiesperstatefromtheSpatialHazardsEventsandLossesDatabasefortheUnitedStates(ref.25).Received21November2012;accepted3June2013;publishedonline14July2013;correctedonline1August2013References1.Day,J.W.etal.Restorationofthemississippidelta:Lessonsfromhurricaneskatrinaandrita.Science315,16791684(2007).2.Shepard,C.etal.Assessingfuturerisk:QuantifyingtheeffectsofsealevelriseonstormsurgeriskforthesouthernshoresofLongIsland,NewYork.Nat.Hazards60,727745(2012).3.CCSPCoastalSensitivitytoSea-LevelRise:AFocusontheMid-AtlanticRegion(USEnvironmentalProtectionAgency,2009).4.Nicholls,R.J.,Hoozemans,F.M.J.&Marchand,M.Increasingfloodriskandwetlandlossesduetoglobalsea-levelrise:Regionalandglobalanalyses.Glob.Environ.Change9(Suppl.1),S69S87(1999).5.Sallenger,A.H.,Doran,K.S.&Howd,P.A.Hotspotofacceleratedsea-levelriseontheAtlanticcoastofNorthAmerica.NatureClim.Change2,884888(2012).6.IPCCClimateChange2007:SynthesisReport(edsPachauri,R.K.&Reisinger,A.)104(IPCC,2007).7.Burgess,K.&Townend,I.TheImpactofClimateChangeuponCoastalDefenseStructures(DepartmentforEnvironment,FoodandRuralAffairs,2004).8.Hillen,M.M.etal.CoastalDefenseCostEstimates:CaseStudyoftheNetherlands,NewOrleansandVietnam(DelftUniversityofTechnology,RoyalHaskoning,UniversitySouthampton,2010).9.Peterson,M.S.&Lowe,M.R.Implicationsofcumulativeimpactstoestuarineandmarinehabitatqualityforfishandinvertebrateresources.Rev.Fish.Sci.17,505523(2009).10.Defeo,O.etal.Threatstosandybeachecosystems:Areview.Estuar.Coast.ShelfS.81,112(2009).11.Louisiana'sComprehensiveMasterPlanforaStainableCoast(CoastalProtectionandRestorationAuthority,2012).12.Mukherjee,N.,Balakrishnan,M.&Shanker,K.inIntegratedCoastalZoneManagement(edsMoksness,E.,Dahl,E.&Støttrup,J.)131144(Wiley-Blackwell,2009);availableathttp://onlinelibrary.wiley.com/doi/10.1002/9781444316285.ch10/summary13.NationalResearchCouncilAdaptingtotheImpactsofClimateChange,America'sClimateChoices(NationalAcademies,2010).14.Barbier,E.B.etal.Coastalecosystem-basedmanagementwithnonlinearecologicalfunctionsandvalues.Science319,321323(2008).15.Jones,H.P.,Hole,D.G.&Zavaleta,E.S.Harnessingnaturetohelppeopleadapttoclimatechange.NatureClim.Change2,504509(2012).16.Hammar-Klose,E.S.&Thieler,E.R.Coastalvulnerabilitytosea-levelrise:ApreliminarydatabasefortheUSAtlantic,PacificandGulfofMexicoCoasts.(USGeologicalSurveyDigitalDataSeries,Vol.68,2001).17.Gornitz,V.Vulnerabilityoftheeastcoast,USAtofuturesealevelrise.J.Coast.Res.201237(1990).18.Cutter,S.L.,Emrich,C.T.,Webb,J.J.&Morath,D.SocialVulnerabilitytoClimateVariabilityHazards:AReviewoftheLiterature(OxfamAmerica,2009).19.Boruff,B.J.,Emrich,C.&Cutter,S.L.ErosionhazardvulnerabilityofUScoastalcounties.J.Coast.Res.215,932942(2005).20.Tollefson,J.Hurricanesweepsintoclimate-adaptationdebate.Nature491,167168(2012).21.Tallis,H.etal.InVEST2.3.0User'sGuide(TheNaturalCapitalProject,2011);availableathttp://ncp-dev.stanford.edu/dataportal/invest-releases/documentation/current_release/22.The20062010AmericanCommunitySurvey5YearSummaryFileTechnicalDocumentation(USCensusBureau,2011);availableathttp://www2.census.gov/acs2010_5yr/summaryfile/ACS_2006-2010_SF_Tech_Doc.pdf23.WhatisaZestimate?(Zillow,2012);availableathttp://www.zillow.com/wikipages/What-is-a-Zestimate/24.Parris,A.etal.GlobalSeaLevelRiseScenariosfortheUSNationalClimateAssessmentTechMemoOARCPO-1(NOAA,2012).25.HazardsandVulnerabilityResearchInstituteTheSpatialHazardEventsandLossesDatabasefortheUnitedStates(UniversityofSouthCarolina,2011);availableathttp://www.sheldus.org26.Feuer,A.ProtectingNewYorkCity,BeforeNextTime.NewYorkTimes(3November2012);availableathttp://www.nytimes.com/2012/11/04/nyregion/protecting-new-york-city-before-next-time.html27.IPCCSpecialReportonEmissionsScenarios(edsNakicenovic,N.&Swart,R.)(CambridgeUniv.Press,2000).28.SeaLevelVariationsoftheUnitedStates(1854to2006)Derivedfrom128NationalWaterLevelObservationNetworkStations(NOAA'sOceanService);availableathttp://www.ncddc.noaa.gov/approved_recs/nos_de/coops/coops/coops/sl_trend.htmlNATURECLIMATECHANGEjVOL3jOCTOBER2013jwww.nature.com/natureclimatechange917 LETTERSNATURECLIMATECHANGEDOI:10.1038/NCLIMATE194429.Tolman,H.L.UserManualandSystemDocumentationofWAVEWATCHIIIversion3.14TechnicalNote(USDepartmentofCommerce,NationalOceanographicandAtmosphericAdministration,NationalWeatherService,NationalCentersforEnvironmentalPredictions,2009).30.Hemer,M.A.,Fan,Y.,Mori,N.,Semedo,A.&Wang,X.L.Projectedchangesinwaveclimatefromamulti-modelensemble.NatureClim.Change3,471476(2013).31.Mennis,J.Generatingsurfacemodelsofpopulationusingdasymetricmapping.Prof.Geogr.55,3142(2003).AcknowledgementsWethanktheGordonandBettyMooreFoundationforfinancialsupportandforhostingtheNationalClimateAssessmentBiodiversity,EcosystemsandEcosystemServicesTechnicalChapterworkinggroup.WethankZillowandthemanyindividualsandinstitutionsthatprovideddata(seeSupplementaryInformationforfulldetails).WealsothankJ.Burke,G.Gelfenbaum,R.Griffin,C.K.Kim,J.Lawler,M.Plummer,P.Ruggiero,J.Samhouri,H.Tallis,J.ToftandG.Zivfordiscussionsduringthisresearch.Linksfordownloadingthecoastalhazardindexanddata,andvisualizingresultsareavailableatwww.naturalcapitalproject.org.AuthorcontributionsP.K.,M.R.,K.K.A.,G.G.,A.G.,S.A.W.andG.V.conceivedtheresearch.G.G.andG.V.developedthecoastalhazardindex.K.A.,G.V.andS.W.carriedoutanalyses.K.K.A.,G.G.,G.V.andS.A.W.collectedthedata.M.L.andJ.M.S.helpedwithdatacollectionandanalyses.K.K.A.wrotethepaperwithcontributionsfromA.G.,G.G,P.K.,M.R.,J.M.S.,G.V.andS.A.W.AdditionalinformationSupplementaryinformationisavailableintheonlineversionofthepaper.Reprintsandpermissionsinformationisavailableonlineatwww.nature.com/reprints.CorrespondenceandrequestsformaterialsshouldbeaddressedtoK.K.A.CompetingfinancialinterestsTheauthorsdeclarenocompetingfinancialinterests.918NATURECLIMATECHANGEjVOL3jOCTOBER2013jwww.nature.com/natureclimatechange Nature Clim. Change http://dx.doi.org/10.1038/nclimate1944 (2013); published online 14 July 2013; corrected online 1 August 2013.In the version of this Letter originally published online, the second sentence of the Acknowledgements section should have read “We thank Zillow and the many individuals and institutions that provided data (see Supplementary Information for full details)”. This error has now been corrected in all versions of the Letter.Coastal habitats shield people and property from sea-level rise and stormsKatie K. Arkema, Greg Guannel, Gregory Verutes, Spencer A. Wood, Anne Guerry, Mary Ruckelshaus, Peter Kareiva, Martin Lacayo and Jessica M. SilverCORRIGENDUM REVIEWdoi:10.1038/nature09678HastheEarth’ssixthmassextinctionalreadyarrived?AnthonyD.Barnosky1,2,3,NicholasMatzke1,SusumuTomiya1,2,3,GuinevereO.U.Wogan1,3,BrianSwartz1,2,TiagoB.Quental1,2{,CharlesMarshall1,2,JennyL.McGuire1,2,3{,EmilyL.Lindsey1,2,KaitlinC.Maguire1,2,BenMersey1,4&ElizabethA.Ferrer1,2PalaeontologistscharacterizemassextinctionsastimeswhentheEarthlosesmorethanthree-quartersofitsspeciesinageologicallyshortinterval,ashashappenedonlyfivetimesinthepast540millionyearsorso.Biologistsnowsuggestthatasixthmassextinctionmaybeunderway,giventheknownspecieslossesoverthepastfewcenturiesandmillennia.Herewereviewhowdifferencesbetweenfossilandmoderndataandtheadditionofrecentlyavailablepalaeontologicalinformationinfluenceourunderstandingofthecurrentextinctioncrisis.Ourresultsconfirmthatcurrentextinctionratesarehigherthanwouldbeexpectedfromthefossilrecord,highlightingtheneedforeffectiveconservationmeasures.OfthefourbillionspeciesestimatedtohaveevolvedontheEarthoverthelast3.5billionyears,some99%aregone1.Thatshowshowverycommonextinctionis,butnormallyitisbalancedbyspeciation.Thebalancewaverssuchthatatseveraltimesinlife’shistoryextinctionratesappearsomewhatelevated,butonlyfivetimesqualifyfor‘massextinction’status:neartheendoftheOrdovician,Devonian,Permian,TriassicandCretaceousPeriods2,3.Thesearethe‘BigFive’massextinctions(twoaretechnically‘massdepletions’)4.Differentcausesarethoughttohaveprecipitatedtheextinctions(Table1),andtheextentofeachextinctionabovethebackgroundlevelvariesdepend-ingonanalyticaltechnique4,5,buttheyallstandoutinhavingextinctionratesspikinghigherthaninanyothergeologicalintervalofthelast,540millionyears3andexhibitingalossofover75%ofestimatedspecies2.Increasingly,scientistsarerecognizingmodernextinctionsofspecies6,7andpopulations8,9.Documentednumbersarelikelytobeseriousunder-estimates,becausemostspecieshavenotyetbeenformallydescribed10,11.Suchobservationssuggestthathumansarenowcausingthesixthmassextinction10,12–17,throughco-optingresources,fragmentinghabitats,introducingnon-nativespecies,spreadingpathogens,killingspeciesdirectly,andchangingglobalclimate10,12–20.Ifso,recoveryofbiodiversitywillnotoccuronanytimeframemeaningfultopeople:evolutionofnewspeciestypicallytakesatleasthundredsofthousandsofyears21,22,andrecoveryfrommassextinctionepisodesprobablyoccursontimescalesencompassingmillionsofyears5,23.Althoughtherearemanydefinitionsofmassextinctionandgrada-tionsofextinctionintensity4,5,herewetakeaconservativeapproachtoassessingtheseriousnessoftheongoingextinctioncrisis,bysettingahighbarforrecognizingmassextinction,thatis,theextremediversitylossthatcharacterizedtheveryunusualBigFive(Table1).WefindthattheEarthcouldreachthatextremewithinjustafewcenturiesifcurrentthreatstomanyspeciesarenotalleviated.DatadisparitiesOnlycertainkindsoftaxa(primarilythosewithfossilizablehardparts)andarestrictedsubsetoftheEarth’sbiomes(generallyintemperatelatitudes)havedatasufficientfordirectfossil-to-moderncomparisons1DepartmentofIntegrativeBiology,UniversityofCalifornia,Berkeley,California94720,USA.2UniversityofCaliforniaMuseumofPaleontology,California,USA.3UniversityofCaliforniaMuseumofVertebrateZoology,California,USA.4HumanEvolutionResearchCenter,California,USA.{Presentaddresses:DepartamentodeEcologia,UniversidadedeSa˜oPaulo(USP),Sa˜oPaulo,Brazil(T.B.Q.);NationalEvolutionarySynthesisCenter,2024W.MainStreet,SuiteA200,Durham,NorthCarolina27705,USA(J.L.M.).Table1|The‘BigFive’massextinctioneventsEventProposedcausesTheOrdovicianevent64–66ended,443Myrago;within3.3to1.9Myr57%ofgenerawerelost,anestimated86%ofspecies.Onsetofalternatingglacialandinterglacialepisodes;repeatedmarinetransgressionsandregressions.UpliftandweatheringoftheAppalachiansaffectingatmosphericandoceanchemistry.SequestrationofCO2.TheDevonianevent4,64,67–70ended,359Myrago;within29to2Myr35%ofgenerawerelost,anestimated75%ofspecies.Globalcooling(followedbyglobalwarming),possiblytiedtothediversificationoflandplants,withassociatedweathering,paedogenesis,andthedrawdownofglobalCO2.Evidenceforwidespreaddeep-wateranoxiaandthespreadofanoxicwatersbytransgressions.Timingandimportanceofbolideimpactsstilldebated.ThePermianevent54,71–73ended,251Myrago;within2.8Myrto160Kyr56%ofgenerawerelost,anestimated96%ofspecies.Siberianvolcanism.Globalwarming.Spreadofdeepmarineanoxicwaters.ElevatedH2SandCO2concentrationsinbothmarineandterrestrialrealms.Oceanacidification.Evidenceforabolideimpactstilldebated.TheTriassicevent74,75ended,200Myrago;within8.3Myrto600Kyr47%ofgenerawerelost,anestimated80%ofspecies.ActivityintheCentralAtlanticMagmaticProvince(CAMP)thoughttohaveelevatedatmosphericCO2levels,whichincreasedglobaltemperaturesandledtoacalcificationcrisisintheworldoceans.TheCretaceousevent58–60,76–79ended,65Myrago;within2.5Myrtolessthanayear40%ofgenerawerelost,anestimated76%ofspecies.AbolideimpactintheYucata´nisthoughttohaveledtoaglobalcataclysmandcausedrapidcooling.Precedingtheimpact,biotamayhavebeendecliningowingtoavarietyofcauses:Deccanvolcanismcontemporaneouswithglobalwarming;tectonicupliftalteringbiogeographyandacceleratingerosion,potentiallycontributingtooceaneutrophicationandanoxicepisodes.CO2spikejustbeforeextinction,dropduringextinction.Myr,millionyears.Kyr,thousandyears.3MARCH2011|VOL471|NATURE|51Macmillan Publishers Limited. All rights reserved©2011 (Box1).Fossilsarewidelyacknowledgedtobeabiasedandincompletesampleofpastspecies,butmoderndataalsohaveimportantbiasesthat,ifnotaccountedfor,caninfluenceglobalextinctionestimates.Onlyatinyfraction(,2.7%)oftheapproximately1.9millionnamed,extantspecieshavebeenformallyevaluatedforextinctionstatusbytheInternationalUnionforConservationofNature(IUCN).TheseIUCNcompilationsarethebestavailable,butevaluatedspeciesrepresentjustafewtwigspluckedfromtheenormousnumberofbranchesthatcomposethetreeoflife.Evenforcladesrecordedas100%evaluated,manyspeciesstillfallintotheDataDeficient(DD)category24.Alsorelevantisthatnotallofthepartiallyevaluatedcladeshavehadtheirspeciessampledinthesameway:somearerandomlysubsampled25,andothersareevaluatedasopportunityarisesorbecausethreatsseemapparent.Despitethelimita-tionsofboththefossilandmodernrecords,byworkingaroundthediversedatabiasesitispossibletoavoiderrorsinextrapolatingfromwhatwedoknowtoinferringglobalpatterns.Ourgoalhereistohigh-lightsomepromisingapproaches(Table2).DefiningmassextinctionsrelativetotheBigFiveExtinctioninvolvesbothrateandmagnitude,whicharedistinctbutintimatelylinkedmetrics26.Rateisessentiallythenumberofextinctionsdividedbythetimeoverwhichtheextinctionsoccurred.Onecanalsoderivefromthisaproportionalrate—thefractionofspeciesthathavegoneextinctperunittime.Magnitudeisthepercentageofspeciesthathavegoneextinct.Massextinctionswereoriginallydiagnosedbyrate:thepaceofextinctionappearedtobecomesignificantlyfasterthanbackgroundextinction3.RecentstudiessuggestthattheDevonianandTriassiceventsresultedmorefromadecreaseinoriginationratesthananincreaseinextinctionrates4,5.Eitherway,thestandingcropoftheEarth’sspeciesfellbyanestimated75%ormore2.Thus,massextinction,intheconservativepalaeontologicalsense,iswhenextinctionratesacceleraterelativetooriginationratessuchthatover75%ofspeciesdisappearwithinageologicallyshortinterval—typicallylessthan2millionyears,insomecasesmuchless(seeTable1).Therefore,todocumentwherethecurrentextinctionepisodeliesonthemassextinctionscaledefinedbytheBigFiverequiresustoknowbothwhethercurrentextinc-tionratesareabovebackgroundrates(andifso,howfarabove)andhowcloselyhistoricandprojectedbiodiversitylossesapproach75%oftheEarth’sspecies.BackgroundratecomparisonsLandmarkstudies12,14–17thathighlightedamodernextinctioncrisisestimatedcurrentratesofextinctiontobeordersofmagnitudehigherthanthebackgroundrate(Table2).AusefulandwidelyappliedmetricBOX1SeveredatacomparisonproblemsGeographyThefossilrecordisverypatchy,sparsestinuplandenvironmentsandtropics,butmodernglobaldistributionsareknownformanyspecies.Apossiblecomparativetechniquecouldbetoexamineregionsorbiomeswherebothfossilandmoderndataexist—suchasthenear-shoremarinerealmincludingcoralreefsandterrestrialdepositionallowlands(rivervalleys,coastlines,andlakebasins).Currentlyavailabledatabases6couldbeusedtoidentifymoderntaxawithgeographicrangesindicatinglowfossilizationpotentialandthenextractthemfromthecurrent-extinctionequation.TaxaavailableforstudyThefossilrecordusuallyincludesonlyspeciesthatpossessidentifiableanatomicalhardpartsthatfossilizewell.Theoreticallyalllivingspeciescouldbestudied,butinpracticeextinctionanalysesoftenrelyonthesmallsubsetofspeciesevaluatedbytheIUCN.EvaluationfollowingIUCNprocedures34placesspeciesinoneofthefollowingcategories:extinct(EX),extinctinthewild(EW),criticallyendangered(CR),endangered(EN),vulnerable(VU),nearthreatened(NT),leastconcern(LC),ordatadeficient(DD,informationinsufficienttoreliablydetermineextinctionrisk).SpeciesintheEXandEWcategoriesaretypicallycountedasfunctionallyextinct.ThoseintheCRplusENplusVUcategoriesarecountedas‘threatened’.AssignmenttoCR,ENorVUisbasedonhowhightheriskofextinctionisdeterminedtobeusingfivecriteria34(roughly,CRprobabilityofextinctionexceeds0.50intenyearsorthreegenerations;ENprobabilityofextinctionexceeds0.20in20yearsorfivegenerations;VUprobabilityofextinctionexceeds0.10overacentury24).Apossiblecomparativetechniquecouldbetousetaxabestknowninbothfossilandmodernrecords:near-shoremarinespecieswithshells,lowlandterrestrialvertebrates(especiallymammals),andsomeplants.Thiswouldrequireimprovedassessmentsofmodernbivalvesandgastropods.Statisticaltechniquescouldbeusedtoclarifyhowasubsampleofwell-assessedtaxaextrapolatestoundersampledand/orpoorlyassessedtaxa25.TaxonomyAnalysesoffossilsareoftendoneatthelevelofgenusratherthanspecies.Whenspeciesareidentifiedtheyareusuallybasedonamorphologicalspeciesconcept.Thiscanresultinlumpingspeciestogetherthataredistinct,or,ifincompletefossilmaterialisused,over-splittingspecies.Formoderntaxa,analysesareusuallydoneatthelevelofspecies,oftenusingaphylogeneticspeciesconcept,whichprobablyincreasesspeciescountsrelativetomorphospecies.Apossiblecomparativetechniquewouldbetoaggregatemodernphylogeneticspeciesintomorphospeciesorgenerabeforecomparingwiththefossilrecord.AssessingextinctionFossilextinctionisrecordedwhenataxonpermanentlydisappearsfromthefossilrecordandunderestimatestheactualnumberofextinctions(andnumberofspecies)becausemosttaxahavenofossilrecord.Theactualtimeofextinctionalmostalwayspostdatesthelastfossiloccurrence.Modernextinctionisrecordedwhennofurtherindividualsofaspeciesaresightedafterappropriateefforts.Inthepastfewdecadesdesignationas‘extinct’usuallyfollowsIUCNcriteria,whichareconservativeandlikelytounderestimatefunctionallyextinctspecies34.Modernextinctionisalsounderestimatedbecausemanyspeciesareunevaluatedorundescribed.Apossiblecomparativetechniquecouldbetostandardizeextinctioncountsbynumberofspeciesknownpertimeintervalofinterest(proportionalextinction).However,fossildatademonstratethatbackgroundratescanvarywidelyfromonetaxontothenext35,86,87,sofossil-to-modernextinctionratecomparisonsaremostreliablydoneonataxon-by-taxonbasis,usingwell-knownextantcladesthatalsohaveagoodfossilrecord.TimeInthefossilrecordsparsesamplesofspeciesarediscontinuouslydistributedthroughvasttimespans,from103to108years.Inmoderntimeswehaverelativelydensesamplesofspeciesoververyshorttimespansofyears,decadesandcenturies.Holocenefossilsarebecomingincreasinglyavailableandvaluableinlinkingthepresentwiththepast48,90.Apossiblecomparativetechniquewouldbetoscaleproportionalextinctionrelativetothetimeintervaloverwhichextinctionismeasured.RESEARCHREVIEW52|NATURE|VOL471|3MARCH2011Macmillan Publishers Limited. All rights reserved©2011 hasbeenE/MSY(extinctionspermillionspecies-years,asdefinedinrefs15and27).Inthisapproach,backgroundratesareestimatedfromfossilextinctionsthattookplaceinmillion-year-or-moretimebins.Forcur-rentrates,theproportionofspeciesextinctinacomparativelyveryshorttime(onetoafewcenturies)isextrapolatedtopredictwhattheratewouldbeoveramillionyears.However,boththeoryandempiricaldataindicatethatextinctionratesvarymarkedlydependingonthelengthoftimeoverwhichtheyaremeasured28,29.Extrapolatingaratecomputedoverashorttime,therefore,willprobablyyieldaratethatiseithermuchfasterormuchslowerthantheaveragemillion-yearrate,socurrentratesthatseemtobeelevatedneedtobeinterpretedinthislight.Onlyrecentlyhasitbecomepossibletodothisbyusingpalaeontologydatabases30,31combinedwithlistsofrecentlyextinctspecies.Themostcompletedatasetofthiskindisformammals,whichverifiestheefficacyofE/MSYbysettingshort-intervalandlong-intervalratesinacomparativecontext(Fig.1).Adatagapremainsbetweenaboutonemillionandabout50thousandyearsbecauseitisnotyetpossibletodateextinctionsinthattimerangewithadequateprecision.Nevertheless,theoverallpatternisasexpected:themaximumE/MSYanditsvarianceincreaseasmeasurementintervalsbecomeshorter.Thehighestratesarerarebutlowratesarecommon;infact,attimeintervalsoflessthanathousandyears,themostcommonE/MSYis0.Threeconclusionsemerge.(1)Themaximumobservedratessinceathousandyearsago(E/MSY<24in1,000-yearbinstoE/MSY<693in1-yearbins)areclearlyfarabovetheaveragefossilrate(aboutE/MSY<1.8),andevenabovethoseofthewidelyrecognizedlate-Pleistocenemegafaunaldiversitycrash32,33(maximumE/MSY<9,reddatapointsinFig.1).(2)Recentaverageratesarealsotoohighcomparedtopre-anthropogenicaverages:E/MSYincreasestoover5(andrisesto23)inless-than-50-yeartimebins.(3)Inthescenariowherecurrently‘threatened’species34wouldultimatelygoextincteveninasmuchasathousandyears,theresultingrateswouldfarexceedanyreasonableestimationoftheupperboundaryforvariationrelatedtointervallength.Thesameappliesiftheextinctionscenarioisrestrictedtoonly‘criticallyendangered’species34.Thisdoesnotimplythatweconsiderallspeciesinthesecategoriestobeinevitablydestinedforextinction—simplythatinaworst-casescenariowherethatoccurred,theextinctionrateformammalswouldfarexceednormalbackgroundrates.Becauseourcomputationalmethodmaximizesthefossilbackgroundratesandminimizesthecurrentrates(seeFig.1caption),ourobservationthatmodernratesareelevatedislikelytobeparticularlyrobust.Moreover,forreasonsarguedbyothers27,themodernrateswecomputedprobablyseriouslyunderestimatecurrentE/MSYvalues.Anotherapproachissimplytoaskwhetheritislikelythatextinctionratescouldhavebeenashighinmanypast500-yearintervalsastheyhavebeeninthemostrecent500years.Whereadequatedataexist,asisthecaseforourmammalexample,theanswerisclearlyno.Themeanper-million-yearfossilrateformammalswedetermined(Fig.1)isabout1.8E/MSY.Tomaintainthatmillion-yearaverage,therecouldbenomorethan6.3%of500-yearbinspermillionyears(126outofapossible2,000)withanextinctionrateashighasthatobservedoverthepast500years(80extinctof5,570specieslivingin500years).Million-yearextinctionratescalculatedbyothers,usingdifferenttechniques,areslower:0.4extinctionsperlineagepermillionyears(alineageinthiscontextisroughlyequivalenttoaspecies)35.Tomaintainthatslowermillion-yearaverage,therecouldbenomorethan1.4%(28intervals)ofthe500-yearintervalspermillionyearshavinganextinctionrateashighasthecurrent500-yearrate.Ratescomputedforshortertimeintervalswouldbeevenlesslikelytofallwithinbackgroundlevels,forreasonsnotedbyref.27.MagnitudeComparisonsofpercentagelossofspeciesinhistoricaltimes6,36tothepercentagelossthatcharacterizedeachoftheBigFive(Fig.2)needtoberefinedbycompensatingformanydifferencesbetweenthemodernandthefossilrecords2,37–39.Seldomtakenintoaccountistheeffectofusingdifferentspeciesconcepts(Box1),whichpotentiallyinflatesthenumbersofmodernspeciesrelativetofossilspecies39,40.Asecond,relatedcaveatisthatmostassessmentsoffossildiversityareatthelevelofgenus,notspecies2,3,37,38,41.Fossilspeciesestimatesarefrequentlyobtainedbycalculat-ingthespecies-to-genusratiodeterminedforwell-knowngroups,thenextrapolatingthatratiotogroupsforwhichonlygenus-levelcountsexist.Theover-75%benchmarkformassextinctionisobtainedinthisway2.Table2|MethodsofcomparingpresentandpastextinctionsGeneralmethodVariationsandrepresentativestudiesReferencesComparecurrentlymeasuredextinctionratestobackgroundratesassessedfromfossilrecordE/MSY*{7,10,15,27,62Comparativespeciesduration(estimatesspeciesdurationstoderiveanestimateofextinctionrate)*{14FuzzyMath*{44,80Interval-ratestandardization(empiricalderivationofrelationshipbetweenrateandintervallengthoverwhichextinctionismeasuredprovidescontextforinterpretingshort-termrates){ThispaperUsevariousmodellingtechniques,includingspecies-arearelationships,toassesslossofspeciesComparerateofexpectednear-termfuturelossestoestimatedbackgroundextinctionrates*{{7,10,14,15Assessmagnitudeofpastspecieslosses{{42,45Predictmagnitudeoffuturelosses.Ref.7exploresseveralmodelsandprovidesarangeofpossibleoutcomesusingdifferentimpactstorylines{{7,14,15,27,36,62,81–84Comparecurrentlymeasuredextinctionratestomass-extinctionratesUsegeologicaldataandhypotheticalscenariostobrackettherangeofratesthatcouldhaveproducedpastmassextinctions,andcomparewithcurrentextinctionrates(assumesBigFivemassextinctionsweresudden,occurringwithin500years,producinga‘worst-casescenario’forhighrates,butwiththepossibleexceptionoftheCretaceousevent,itisunlikelythatanyoftheBigFivewerethisfast){ThispaperAssessextinctionincontextoflong-termcladedynamicsMapprojectedextinctiontrajectoriesontolong-termdiversification/extinctiontrendsinwell-studiedclades{ThispaperAssesspercentagelossofspeciesUseIUCNliststoassessmagnitudeorrateofactualandpotentialspecieslossesinwell-studiedtaxa{Thispaperandrefs6,7,10,14,15,20,36and62UsemolecularphylogeniestoestimateextinctionrateCalculatebackgroundextinctionratesfromtime-correctedmolecularphylogeniesofextantspecies,andcomparetomodernrates85FuzzyMathattemptstoaccountfordifferentbiasesinfossilandmodernsamplesandusesempiricallybasedfossilbackgroundextinctionratesasastandardforcomparison:0.25speciespermillionyearsformarineinvertebrates,determinedfromthe‘kill-curve’method86,and0.21species35to0.46species87permillionyearsforNorthAmericanmammals,determinedfromapplyingmaximum-likelihoodtechniques.Themolecularphylogeniesmethodassumesthatdiversificationratesareconstantthroughtimeandcanbepartitionedintooriginationsandextinctionswithoutevidencefromthefossilrecord.Recentworkhasdemonstratedthatdisentanglementofdiversificationfromextinctionratesbythismethodisdifficult,particularlyintheabsenceofafossilrecord,andthatextinctionratesestimatedfrommolecularphylogeniesofextantorganismsarehighlyunreliablewhendiversificationratesvaryamonglineagesthroughtime46,88.*Comparisonofmodernshort-termrateswithfossillong-termratesindicatehighlyelevatedmodernrates,butdoesnottakeintoaccountinterval-rateeffect.{Assumesthattherelationshipbetweennumberandkindofspecieslostinstudyareacanbescaleduptomakeglobalprojections.{Assumesthatconclusionsfromwell-studiedtaxaillustrategeneralprinciples.REVIEWRESEARCH3MARCH2011|VOL471|NATURE|53Macmillan Publishers Limited. All rights reserved©2011 Potentiallyvaluablecomparisonsofextinctionmagnitudecouldcomefromassessingmoderntaxonomicgroupsthatarealsoknownfromexceptionallygoodfossilrecords.Thebestfossilrecordsarefornear-shoremarineinvertebrateslikegastropods,bivalvesandcorals,andtemperateterrestrialmammals,withgoodinformationalsoavailableforHolocenePacificIslandbirds2,33,35,42–44.However,betterknowledgeofunderstudiedmoderntaxaiscriticallyimportantfordevelopingcommonmetricsformodernandfossilgroups.Forexample,some49%ofbivalveswentextinctduringtheend-Cretaceousevent43,butonly1%oftoday’sspecieshaveevenbeenassessed6,makingmeaningfulcomparisondifficult.Asimilarproblemprevailsforgastropods,exacerbatedbecausemostmodernassessmentsareonterrestrialspecies,andmostfossildatacomefrommarinespecies.Giventhedauntingchallengeofassessingextinctionriskineverylivingspecies,statisticalapproachesaimedatunderstandingwhatwellsampledtaxatellusaboutextinctionrisksinpoorlysampledtaxaarecriticallyimportant25.Foraveryfewgroups,modernassessmentsareclosetoadequate.Scleractiniancorals,amphibians,birdsandmammalshaveallknownspeciesassessed6(Fig.2),althoughspeciescountsremainamovingtarget27.Inthesegroups,eventhoughthepercentageofspeciesextinctinhistorictimeislow(zeroto1%),20–43%oftheirspeciesandmanymoreoftheirpopulationsarethreatened(Fig.2).Thosenumberssuggestthatwehavenotyetseenthesixthmassextinction,butthatwewouldjumpfromone-quartertohalfwaytowardsitif‘threatened’speciesdisappear.Giventhatmanycladesareundersampledorunevenlysampled,magnitudeestimatesthatrelyontheoreticalpredictionsratherthanempiricaldatabecomeimportant.Oftenspecies-arearelationshipsoralliedmodellingtechniquesareusedtorelatespecieslossestohabitat-arealosses(Table2).Thesetechniquessuggestthatfuturespeciesextinctionswillbearound21–52%,similartothemagnitudesexpressedCycadopsidaMammaliaAvesReptiliaAmphibiaActinopterygiiScleractiniaGastropodaBivalviaConiferopsidaChondrichthyesDecapodaBig Five massextinctions0255075100Extinction magnitude(percentage of species)Figure2|ExtinctionmagnitudesofIUCN-assessedtaxa6incomparisontothe75%mass-extinctionbenchmark.Numbersnexttoeachiconindicatepercentageofspecies.Whiteiconsindicatespecies‘extinct’and‘extinctinthewild’overthepast500years.Blackiconsaddcurrently‘threatened’speciestothosealready‘extinct’or‘extinctinthewild’;theamphibianpercentagemaybeashighas43%(ref.19).YellowiconsindicatetheBigFivespecieslosses:Cretaceous1Devonian,Triassic,OrdovicianandPermian(fromlefttoright).Asterisksindicatetaxaforwhichveryfewspecies(lessthan3%forgastropodsandbivalves)havebeenassessed;whitearrowsshowwhereextinctionpercentagesareprobablyinflated(becausespeciesperceivedtobeinperilareoftenassessedfirst).Thenumberofspeciesknownorassessedforeachofthegroupslistedis:Mammalia5,490/5,490;Aves(birds)10,027/10,027;Reptilia8,855/1,677;Amphibia6,285/6,285,Actinopterygii24,000/5,826,Scleractinia(corals)837/837;Gastropoda85,000/2,319;Bivalvia30,000/310,Cycadopsida307/307;Coniferopsida618/618;Chondrichthyes1,044/1,044;andDecapoda1,867/1,867.10,0001,0001001010.10.010107106105104103102101Time-interval length (years)E/MSYCenozoicfossilsCRENVUCRENVUCRENVUPleistoceneExtinctionssince 2010Minus batsand endemicsFigure1|Relationshipbetweenextinctionratesandthetimeintervaloverwhichtherateswerecalculated,formammals.EachsmallgreydatumpointrepresentstheE/MSY(extinctionpermillionspecies-years)calculatedfromtaxondurationsrecordedinthePaleobiologyDatabase30(million-year-or-moretimebins)orfromlistsofextant,recentlyextinct,andPleistocenespeciescompiledfromtheliterature(100,000-year-and-lesstimebins)6,32,33,89–97.Morethan4,600datapointsareplottedandclusterontopofeachother.Yellowshadingencompassesthe‘normal’(non-anthropogenic)rangeofvarianceinextinctionratethatwouldbeexpectedgivendifferentmeasurementintervals;formorethan100,000years,itisthesameasthe95%confidenceinterval,butthefadingtotherightindicatesthattheupperboundaryof‘normal’variancebecomesuncertainatshorttimeintervals.TheshorthorizontallinesindicatetheempiricallydeterminedmeanE/MSYforeachtimebin.Largecoloureddotsrepresentthecalculatedextinctionratessince2010.Red,theend-Pleistoceneextinctionevent.Orange,documentedhistoricalextinctionsaveraged(fromrighttoleft)overthelast1,30,50,70,100,500,1,000and5,000years.Blue,attemptstoenhancecomparabilityofmodernwithfossildatabyadjustingforextinctionsofspecieswithverylowfossilizationpotential(suchasthosewithverysmallgeographicrangesandbats).Forthesecalculations,‘extinct’and‘extinctinthewild’speciesthathadgeographicrangeslessthan500km2asrecordedbytheIUCN6,allspeciesrestrictedtoislandsoflessthan105km2,andbatswereexcludedfromthecounts(under-representationofbatsasfossilsisindicatedbytheircomposingonlyabout2.5%ofthefossilspeciescount,versusaround20%ofthemodernspeciescount30).Browntrianglesrepresenttheprojectionsofratesthatwouldresultif‘threatened’mammalsgoextinctwithin100,500or1,000years.Thelowesttriangle(ofeachverticalset)indicatestherateifonly‘criticallyendangered’speciesweretogoextinct(CR),themiddletriangleindicatestherateif‘criticallyendangered’1‘endangered’speciesweretogoextinct(EN),andthehighesttriangleindicatestherateif‘criticallyendangered’1‘endangered’1‘vulnerable’speciesweretogoextinct(VU).ToproduceFig.1wefirstdeterminedthelast-occurrencerecordsofCenozoicmammalsfromthePaleobiologyDatabase30,andthelastoccurrencesofPleistoceneandHolocenemammalsfromrefs6,32,33and89–97.WethenusedR-scripts(writtenbyN.M.)tocomputetotaldiversity,numberofextinctions,proportionalextinction,andE/MSY(anditsmean)fortime-binsofvaryingduration.Cenozoictimebinsrangedfrom25milliontoamillionyears.Pleistocenetimebinsrangedfrom100,000to5,000years,andHolocenetimebinsfrom5,000yearstoayear.ForCenozoicdata,themeanE/MSYwascomputedusingtheaveragewithin-binstandingdiversity,whichwascalculatedbycountingalltaxathatcrosseach100,000-yearboundarywithinamillion-yearbin,thenaveragingthoseboundary-crossingcountstocomputestandingdiversityfortheentiremillion-year-and-overbin.Formoderndata,themeanwascomputedusingthetotalstandingdiversityineachbin(extinctplussurvivingtaxa).Thismethodmayoverestimatethefossilmeanextinctionrateandunderestimatethemodernmeans,soitisaconservativecomparisonintermsofassessingwhethermodernmeansarehigher.TheCenozoicdataareforNorthAmericaandthePleistoceneandHolocenedataareforglobalextinction;adequateglobalCenozoicdataareunavailable.ThereisnoapparentreasontosuspectthattheNorthAmericanaveragewoulddifferfromtheglobalaverageatthemillion-yeartimescale.RESEARCHREVIEW54|NATURE|VOL471|3MARCH2011Macmillan Publishers Limited. All rights reserved©2011 inFig.2,althoughderivedquitedifferently.Suchmodelsmaybesensi-tivetotheparticulargeographicarea,taxaandspecies-arearelationshipthatisemployed,andhaveusuallyusedonlymoderndata.However,fossil-to-moderncomparisonsusingspecies-areamethodsarenowbecomingpossibleasonlinepalaeontologicaldatabasesgrow30,31,45.Anadditional,newapproachmodelshowmuchextinctioncanbeexpectedundervaryingscenariosofhumanimpact7.Itsuggestsabroaderrangeofpossiblefutureextinctionmagnitudesthanpreviousstudies,althoughallscenariosresultinadditionalbiodiversitydeclineinthetwenty-firstcentury.Combinedrate–magnitudecomparisonsBecauserateandmagnitudearesointimatelylinked,acriticalquestioniswhethercurrentrateswouldproduceBig-Five-magnitudemassextinc-tionsinthesameamountofgeologicaltimethatwethinkmostBigFiveextinctionsspanned(Table1).Theanswerisyes(Fig.3).Currentextinc-tionratesformammals,amphibians,birds,andreptiles(Fig.3,lightyellowdotsontheleft),ifcalculatedoverthelast500years(aconserva-tivelyslowrate27)arefasterthan(birds,mammals,amphibians,whichhave100%ofspeciesassessed)orasfastas(reptiles,uncertainbecauseonly19%ofspeciesareassessed)allratesthatwouldhaveproducedtheBigFiveextinctionsoverhundredsofthousandsormillionsofyears(Fig.3,verticallines).Wouldratescalculatedforhistoricalandnear-timeprehistoricextinctionsresultinBig-Five-magnitudeextinctionintheforeseeablefuture—lessthanafewcenturies?Again,takingthe500-yearrateasausefulbasisofcomparison,twodifferenthypotheticalapproachesarepossible.ThefirstassumesthattheBigFiveextinctionstookplacesuddenlyandaskswhatrateswouldhaveproducedtheirestimatedspecieslosseswithin500years(Fig.3,coloureddotsontheright).(Weemphasizethatthisisahypotheticalscenarioandthatwearenotarguingthatallmassextinctionsweresudden.)Inthatscenario,theratesforcontemporaryextinctions(Fig.3,lightyellowdotsontheleft)areslowerthantheratesthatwouldhaveproducedeachoftheBigFiveextinctionsin500years.However,ratesthatconsider‘threatened’speciesasinevitablyextinct(Fig.3,orangedotsontheleft)arealmostasfastasthe500-yearBigFiverates.Therefore,atleastasjudgedusingthesevertebratetaxa,losingthreatenedspecieswouldsignalamassextinctionnearlyonparwiththeBigFive.AsecondhypotheticalapproachaskshowmanymoreyearsitwouldtakeforcurrentextinctionratestoproducespecieslossesequivalenttoBigFivemagnitudes.Theansweristhatifall‘threatened’speciesbecameextinctwithinacentury,andthatratethencontinuedunabated,terrestrialamphibian,birdandmammalextinctionwouldreachBigFivemagni-tudesin,240to540years(241.7yearsforamphibians,536.6yearsforbirds,334.4yearsformammals).Reptileshavesofewoftheirspeciesassessedthattheyarenotincludedinthiscalculation.Ifextinctionwerelimitedto‘criticallyendangered’speciesoverthenextcenturyandthoseextinctionratescontinued,thetimeuntil75%ofspecieswerelostpergroupwouldbe890yearsforamphibians,2,265yearsforbirdsand1,519yearsformammals.Forscenariosthatprojectextinctionof‘threatened’or‘criticallyendangered’speciesover500yearsinsteadofacentury,massextinctionmagnitudeswouldbereachedinabout1,200to2,690yearsforthe‘threatened’scenario(1,209yearsforamphibians,2,683yearsforbirdsand1,672yearsformammals)or,4,450to11,330yearsforthe‘criticallyendangered’scenario(4,452yearsforamphi-bians,11,326yearsforbirdsand7,593yearsformammals).ThisemphasizesthatcurrentextinctionratesarehigherthanthosethatcausedBigFiveextinctionsingeologicaltime;theycouldbesevereenoughtocarryextinctionmagnitudestotheBigFivebenchmarkinaslittleasthreecenturies.Italsohighlightsareasformuch-neededfutureresearch.Amongmajorunknownsare(1)whether‘criticallyendangered’,‘endangered’and‘vulnerable’specieswillgoextinct,(2)whetherthecurrentratesweusedinourcalculationswillcontinue,increaseordecrease;and(3)howreliablyextinctionratesinwell-studiedtaxacanbeextrapolatedtootherkindsofspeciesinotherplaces7,20,25,34.ThebackdropofdiversitydynamicsLittleexplorediswhethercurrentextinctionrateswithinacladefallout-sideexpectationswhenconsideredinthecontextoflong-termdiversitydynamics.Forexample,analysesofcetacean(whalesanddolphins)extinctionandoriginationratesillustratethatwithin-cladediversityhasbeendecliningforthelast5.3millionyears,andthatthatdeclineisnestedwithinanevenlonger-termdeclinethatbegansome14millionyearsago.Yet,withinthatcontext,evenif‘threatened’generalastedaslongas100,000yearsbeforegoingextinct,thecladewouldstillexperienceanextinctionratethatisanorderofmagnitudehigherthananythingithasexperiencedduringitsevolutionaryhistory46.Thefossilrecordisalsoenablingustointerpretbetterthesignificanceofcurrentlyobservedpopulationdistributionsanddeclines.TheuseofancientDNA,phylochronologyandsimulationsdemonstratethatthepopulationstructureconsidered‘normal’onthecurrentlandscapehasinfactalreadysuffereddiversitydeclinesrelativetoconditionsafewthousandyearsago47,48.Likewise,thefossilrecordshowsthatspeciesrichnessandevennesstakenas‘normal’todayarelowcomparedtopre-anthropogenicconditions10,27,32,33,42,45,49.SelectivityDuringtimesofnormalbackgroundextinction,thetaxathatsufferextinctionmostfrequentlyarecharacterizedbysmallgeographicrangesandlowpopulationabundance38.However,duringtimesofmassextinc-tion,therulesofextinctionselectivitycanchangemarkedly,sothatwidespread,abundanttaxaalsogoextinct37,38.Large-bodiedanimalsandthoseincertainphylogeneticgroupscanbeparticularlyhardhit33,50–52.Inthatcontext,thereductionofformerlywidespreadranges8anddisproportionatecullingofcertainkindsofspecies50–53maybeE/MSY1,000,0001010010.11,00010,000100,000DevonianCretaceousTH, CRTriassicOrdovicianPermian0.01Extinction magnitude (percentage of species)020406080100CriticallyendangeredAlreadyextinctThreatenedFigure3|Extinctionrateversusextinctionmagnitude.Verticallinesontherightillustratetherangeofmassextinctionrates(E/MSY)thatwouldproducetheBigFiveextinctionmagnitudes,asbracketedbythebestavailabledatafromthegeologicalrecord.Thecorrespondinglycoloureddotsindicatewhattheextinctionratewouldhavebeeniftheextinctionshadhappened(hypothetically)overonly500years.Ontheleft,dotsconnectedbylinesindicatetherateascomputedforthepast500yearsforvertebrates:lightyellow,speciesalreadyextinct;darkyellow,hypotheticalextinctionof‘criticallyendangered’species;orange,hypotheticalextinctionofall‘threatened’species.TH:ifall‘threatened’speciesbecameextinctin100years,andthatrateofextinctionremainedconstant,thetimeto75%speciesloss—thatis,thesixthmassextinction—wouldbe,240to540yearsforthosevertebratesshownherethathavebeenfullyassessed(allbutreptiles).CR:similarly,ifall‘criticallyendangered’speciesbecameextinctin100years,thetimeto75%specieslosswouldbe,890to2,270yearsforthesefullyassessedterrestrialvertebrates.REVIEWRESEARCH3MARCH2011|VOL471|NATURE|55Macmillan Publishers Limited. All rights reserved©2011 particularlyinformativeinindicatingthatextinction-selectivityischan-gingintoastatecharacterizingmassextinctions.Perfectstorms?Hypothesestoexplainthegeneralphenomenonofmassextinctionshaveemphasizedsynergiesbetweenunusualevents54–57.Commonfea-turesoftheBigFive(Table1)suggestthatkeysynergiesmayinvolveunusualclimatedynamics,atmosphericcompositionandabnormallyhigh-intensityecologicalstressorsthatnegativelyaffectmanydifferentlineages.ThisdoesnotimplythatrandomaccidentslikeaCretaceousasteroidimpact58,59wouldnotcausedevastatingextinctionontheirown,onlythatextinctionmagnitudewouldbelowerifsynergisticstressorshadnotalready‘primedthepump’ofextinction60.Morerigorouslyformulatingandtestingsynergyhypothesesmaybeespeciallyimportantinassessingsixthmassextinctionpotential,becauseonceagaintheglobalstageissetforunusualinteractions.Existingecosystemsarethelegacyofabioticturnoverinitiatedbytheonsetofglacial–interglacialcyclesthatbegan,2.6millionyearsago,andevolvedprimarilyintheabsenceofHomosapiens.Today,rapidlychangingatmosphericconditionsandwarmingabovetypicalinterglacialtemperaturesasCO2levelscontinuetorise,habitatfragmentation,pol-lution,overfishingandoverhunting,invasivespeciesandpathogens(likechytridfungus),andexpandinghumanbiomass6,7,18,20areallmoreextremeecologicalstressorsthanmostlivingspecieshavepreviouslyexperienced.Withoutconcertedmitigationefforts,suchstressorswillaccelerateinthefutureandthusintensifyextinction7,20,especiallygiventhefeedbacksbetweenindividualstressors56.ViewtothefutureThereisconsiderablymoretobelearnedbyapplyingnewmethodsthatappropriatelyadjustforthedifferentkindsofdataandtimescalesinherentinthefossilrecordsversusmodernrecords.Futureworkneedsto:(1)standardizeratecomparisonstoadjustforratemeasurementsoverwidelydisparatetimescales;(2)standardizemagnitudecomparisonsbyusingthesamespecies(orothertaxonomicrank)conceptsformodernandfossilorganisms;(3)standardizetaxonomicandgeographiccomparisonsbyusingmodernandfossiltaxathathaveequalfossilizationpotential;(4)assesstheextinctionriskofmoderntaxasuchasbivalvesandgastropodsthatareextremelycommoninthefossilrecordbutareatpresentpoorlyassessed;(5)setcurrentextinctionobservationsinthecontextoflong-termclade,species-richness,andpopulationdynamicsusingthefossilrecordandphylogenetictechniques;(6)furtherexploretherelationshipbetweenextinctionselectivityandextinctionintensity;and(7)developandtestmodelsthatpositgeneralconditionsrequiredformassextinction,andhowthosecomparewiththecurrentstateoftheEarth.Ourexaminationofexistingdatainthesecontextsraisestwoimportantpoints.First,therecentlossofspeciesisdramaticandseriousbutdoesnotyetqualifyasamassextinctioninthepalaeontologicalsenseoftheBigFive.Inhistorictimeswehaveactuallylostonlyafewpercentofassessedspecies(thoughwehavenowayofknowinghowmanyspecieswehavelostthathadneverbeendescribed).Itisencouragingthatthereisstillmuchoftheworld’sbiodiversitylefttosave,butdauntingthatdoingsowillrequirethereversalofmanydireandescalatingthreats7,20,61–63.Thesecondpointisparticularlyimportant.Eventakingintoaccountthedifficultiesofcomparingthefossilandmodernrecords,andapplyingconservativecomparativemethodsthatfavourminimizingthediffer-encesbetweenfossilandmodernextinctionmetrics,thereareclearindi-cationsthatlosingspeciesnowinthe‘criticallyendangered’categorywouldpropeltheworldtoastateofmassextinctionthathaspreviouslybeenseenonlyfivetimesinabout540millionyears.Additionallossesofspeciesinthe‘endangered’and‘vulnerable’categoriescouldaccomplishthesixthmassextinctioninjustafewcenturies.Itmaybeofparticularconcernthatthisextinctiontrajectorywouldplayoutunderconditionsthatresemblethe‘perfectstorm’thatcoincidedwithpastmassextinc-tions:multiple,atypicalhigh-intensityecologicalstressors,includingrapid,unusualclimatechangeandhighlyelevatedatmosphericCO2.Thehugedifferencebetweenwherewearenow,andwherewecouldeasilybewithinafewgenerations,revealstheurgencyofrelievingthepressuresthatarepushingtoday’sspeciestowardsextinction.1.Novacek,M.J.(ed.)TheBiodiversityCrisis:LosingWhatCounts(TheNewPress,2001).2.Jablonski,D.Extinctionsinthefossilrecord.Phil.Trans.R.Soc.Lond.B344,11–17(1994).Thispapersummarizes,fromapalaeontologicalperspective,thedifficultiesofcomparingthepastandpresentextinctions.3.Raup,D.M.&Sepkoski,J.J.Massextinctionsinthemarinefossilrecord.Science215,1501–1503(1982).ThisisastatisticalassessmentoftheBigFiveextinctionratesrelativetobackgroundrates.4.Bambach,R.K.Phanerozoicbiodiversitymassextinctions.Annu.Rev.EarthPlanet.Sci.34,127–155(2006).Thispaperdiscussesthedefinitionofmassextinctionsandmassdepletions,andtherelativeroleoforiginationversusextinctionratesincausingthediversityreductionsthatcharacterizetheBigFive.5.Alroy,J.Dynamicsoforiginationandextinctioninthemarinefossilrecord.Proc.NatlAcad.Sci.USA105,11536–11542(2008).6.IUCN.InternationalUnionforConservationofNatureRedListÆhttp://www.iucn.org/about/work/programmes/species/red_list/æ(2010).7.Pereira,H.M.etal.Scenariosforglobalbiodiversityinthe21stcentury.Science330,1496–1501(2010).8.Ceballos,G.&Ehrlich,P.R.Mammalpopulationlossesandtheextinctioncrisis.Science296,904–907(2002).9.Hughes,J.B.,Daily,G.C.&Ehrlich,P.R.Populationdiversity:itsextentandextinction.Science278,689–692(1997).10.Dirzo,R.&Raven,P.H.Globalstateofbiodiversityandloss.Annu.Rev.Environ.Resour.28,137–167(2003).Thispaperisanoverviewofthetaxonomicandspatiotemporalpatternsofbiodiversityandthemagnitudeofthecurrentbiodiversitycrisis.11.Joppa,L.N.,Roberts,D.L.&Pimm,S.L.Howmanyspeciesoffloweringplantsarethere?Proc.R.Soc.Lond.B.doi:10.1098/rspb.2010.1004(2010).12.Leakey,R.&Lewin,R.TheSixthExtinction:PatternsofLifeandtheFutureofHumankind(Doubleday,1992).13.Wake,D.B.&Vredenburg,V.T.Areweinthemidstofthesixthmassextinction?Aviewfromtheworldofamphibians.Proc.NatlAcad.Sci.USA105,11466–11473(2008).14.May,R.M.,Lawton,J.H.&Stork,N.E.inExtinctionRates(edsLawton,J.H.&May,R.M.)Ch.1,1–24(OxfordUniversityPress,1995).Thispapercomparesfossil-backgroundandrecentextinctionratesandexplainsthenumerousassumptionsthatarerequiredforthecomparison.15.Pimm,S.L.,Russell,G.J.,Gittleman,J.L.&Brooks,T.M.Thefutureofbiodiversity.Science269,347–350(1995).ThispaperexplainsandusestheE/MSYmetrictocomparethefossil-background,current,andpredictedfutureextinctionrates.16.Myers,N.Massextinctions:whatcanthepasttellusaboutthepresentandfuture?Palaeogeogr.Palaeoclimatol.Palaeoecol.82,175–185(1990).17.Pimm,S.L.&Brooks,T.M.inNatureandHumanSociety:TheQuestforaSustainableWorld(edsRaven,P.H.&Williams,T.)46–62(NationalAcademyPress,1997).18.Barnosky,A.D.Heatstroke:NatureinanAgeofGlobalWarming1–269(IslandPress,2009).19.Vredenburg,V.T.,Knapp,R.A.,Tunstall,T.S.&Briggs,C.J.Dynamicsofanemergingdiseasedrivelarge-scaleamphibianpopulationextinctions.Proc.NatlAcad.Sci.USA107,9689–9694(2010).20.Hoffmann,M.etal.Theimpactofconservationonthestatusoftheworld’svertebrates.Science330,1503–1509(2010).21.Avise,J.C.,Walker,D.&Johns,G.C.SpeciationdurationsandPleistoceneeffectsonvertebratephylogeography.Proc.R.Soc.Lond.B265,1707–1712(1998).22.Weir,J.T.&Schluter,D.Thelatitudinalgradientinrecentspeciationandextinctionratesofbirdsandmammals.Science315,1574–1576(2007).23.Lu,P.J.,Yogo,M.&Marshall,C.R.Phanerozoicmarinebiodiversitydynamicsinlightoftheincompletenessofthefossilrecord.Proc.NatlAcad.Sci.USA103,2736–2739(2006).24.Vie´,J.-C.,Hilton-Taylor,C.&Stuart,S.N.(eds)WildlifeinaChangingWorld—AnAnalysisofthe2008IUCNRedListofThreatenedSpecies180(IUCN,2009).25.Baillie,J.E.M.etal.Towardmonitoringglobalbiodiversity.Conserv.Lett.1,18–26(2008).26.S¸engo¨r,A.M.C.,Atayman,S.&O¨zeren,S.Ascaleofgreatnessandcausalclassificationofmassextinctions:implicationsformechanisms.Proc.NatlAcad.Sci.USA105,13736–13740(2008).27.Pimm,S.,Raven,P.,Peterson,A.,Sekercioglu,Ç.H.&Ehrlich,P.R.Humanimpactsontheratesofrecent,present,andfuturebirdextinctions.Proc.NatlAcad.Sci.USA103,10941–10946(2006).28.Foote,M.Temporalvariationinextinctionriskandtemporalscalingofextinctionmetrics.Paleobiology20,424–444(1994).Thispaperaddressestheeffectofintervallengthonextinctionmetricsusingsimulations.29.Foote,M.&Raup,D.M.Fossilpreservationandthestratigraphicrangesoftaxa.Paleobiology22,121–140(1996).30.PBDB.ThePaleobiologyDatabaseÆhttp://paleodb.org/cgi-bin/bridge.plæ(2010).31.NEOMAP.TheNeogeneMammalMappingPortalÆhttp://www.ucmp.berkeley.edu/neomap/æ(2010).RESEARCHREVIEW56|NATURE|VOL471|3MARCH2011Macmillan Publishers Limited. All rights reserved©2011 32.Barnosky,A.D.MegafaunabiomasstradeoffasadriverofQuaternaryandfutureextinctions.Proc.NatlAcad.Sci.USA105,11543–11548(2008).33.Koch,P.L.&Barnosky,A.D.LateQuaternaryextinctions:stateofthedebate.Annu.Rev.Ecol.Evol.Syst.37,215–250(2006).34.Mace,G.M.etal.Quantificationofextinctionrisk:IUCN’ssystemforclassifyingthreatenedspecies.Conserv.Biol.22,1424–1442(2008).ThispaperexplainsthemethodologyusedbytheIUCNtoassesstheextinctionrisksofextantspecies.35.Alroy,J.Constantextinction,constraineddiversification,anduncoordinatedstasisinNorthAmericanmammals.Palaeogeogr.Palaeoclimatol.Palaeoecol.127,285–311(1996).36.Stork,N.E.Re-assessingcurrentextinctionrates.Biodivers.Conserv.19,357–371(2010).37.Jablonski,D.Lessonsfromthepast:evolutionaryimpactsofmassextinctions.Proc.NatlAcad.Sci.USA98,5393–5398(2001).38.Jablonski,D.Extinctionandthespatialdynamicsofbiodiversity.Proc.NatlAcad.Sci.USA105,11528–11535(2008).39.Purvis,A.,Jones,K.E.&Mace,G.M.Extinction.Bioessays22,1123–1133(2000).40.Agapow,P.-M.etal.Theimpactofthespeciesconceptonbiodiversitystudies.Q.Rev.Biol.79,161–179(2004).41.Alroy,J.etal.Phanerozoictrendsintheglobaldiversityofmarineinvertebrates.Science321,97–100(2008).42.Steadman,D.W.ExtinctionandBiogeographyofTropicalPacificBirds(UniversityofChicagoPress,2006).43.Raup,D.M.&Jablonski,D.Geographyofend-Cretaceousmarinebivalveextinctions.Science260,971–973(1993).44.Regan,H.M.,Lupia,R.,Drinnan,A.N.&Burgman,M.A.Thecurrencyandtempoofextinction.Am.Nat.157,1–10(2001).45.Carrasco,M.A.,Barnosky,A.D.&Graham,R.W.QuantifyingtheextentofNorthAmericanmammalextinctionrelativetothepre-anthropogenicbaseline.PLoSONE4,e8331(2009).Thispaperusesspecies-arearelationshipsbasedonfossildatatodemonstratethattherecentbiodiversitybaselineformammalsissubstantiallydepressedwithrespecttoitsnormalcondition.46.Quental,T.B.&Marshall,C.R.Diversitydynamics:molecularphylogeniesneedthefossilrecord.TrendsEcol.Evol.25,434–441(2010).47.Anderson,C.N.K.,Ramakrishnan,U.,Chan,Y.L.&Hadly,E.A.SerialSimCoal:apopulationgeneticsmodelfordatafrommultiplepopulationsandpointsintime.Bioinformatics21,1733–1734(2005).48.Ramakrishnan,U.&Hadly,E.A.Usingphylochronologytorevealcrypticpopulationhistories:reviewandsynthesisoffourancientDNAstudies.Mol.Ecol.18,1310–1330(2009).Thispaperusesahypothesis-testingframeworktorevealpopulationhistoriesandcomparepastpopulationstopresentones.49.Blois,J.L.,McGuire,J.L.&Hadly,E.A.Smallmammaldiversitylossinresponsetolate-Pleistoceneclimaticchange.Nature465,771–774(2010).50.Cardillo,M.etal.Multiplecausesofhighextinctionriskinlargemammalspecies.Science309,1239–1241(2005).51.Davies,T.J.etal.Phylogenetictreesandthefutureofmammalianbiodiversity.Proc.NatlAcad.Sci.USA105,11556–11563(2008).52.Isaac,N.J.B.,Turvey,S.T.,Collen,B.,Waterman,C.&Baillie,J.E.M.MammalsontheEDGE:conservationprioritiesbasedonthreatandphylogeny.PLoSONE2,e296(2007).53.Russell,G.J.,Brooks,T.M.,McKinney,M.M.&Anderson,C.G.Presentandfuturetaxonomicselectivityinbirdandmammalextinctions.Conserv.Biol.12,1365–1376(1998).54.Erwin,D.H.ThePermo-Triassicextinction.Nature367,231–236(1994).55.Arens,N.C.&West,I.D.Press-pulse:ageneraltheoryofmassextinction?Paleobiology34,456–471(2008).56.Brook,B.W.,Sodhi,N.S.&Bradshaw,C.J.A.Synergiesamongextinctiondriversunderglobalchange.TrendsEcol.Evol.23,453–460(2008).57.Jablonski,D.inDynamicsofExtinction(ed.Elliott,D.K.)183–229(Wiley,1986).58.Alvarez,L.W.,Alvarez,W.,Asaro,F.&Michel,H.V.ExtraterrestrialcausefortheCretaceous-Tertiaryextinction.Science208,1095–1108(1980).59.Schulte,P.etal.TheChicxulubasteroidimpactandmassextinctionattheCretaceous-Paleogeneboundary.Science327,1214–1218(2010).60.Prauss,M.L.TheK/PgboundaryatBrazos-River,Texas,USA—anapproachbymarinepalynology.Palaeogeogr.Palaeoclimatol.Palaeoecol.283,195–215(2009).61.GBO3.GlobalBiodiversityOutlook394(SecretariatoftheConventiononBiologicalDiversity,2010).62.Mace,G.etal.inMilleniumEcosystemAssessment,EcosystemsandHumanWell-being:BiodiversitySynthesis(edsCeballos,S.L.G.,Orians,G.&Pacala,S.)Ch.4,77–122(WorldResourcesInstitute,2005).63.Cardillo,M.,Mace,G.M.,Gittleman,J.L.&Purvis,A.Latentextinctionriskandthefuturebattlegroundsofmammalconservation.Proc.NatlAcad.Sci.USA103,4157–4161(2006).64.Sepkoski,J.J.inGlobalEventsandEventStratigraphyinthePhanerozoic(ed.Walliser,O.H.)35–51(Springer,1996).65.Sheehan,P.M.TheLateOrdovicianmassextinction.Annu.Rev.EarthPlanet.Sci.29,331–364(2001).66.Sutcliffe,O.E.,Dowdeswell,J.A.,Whittington,R.J.,Theron,J.N.&Craig,J.CalibratingtheLateOrdovicianglaciationandmassextinctionbytheeccentricitycyclesofEarth’sorbit.Geology28,967–970(2000).67.Sandberg,C.A.,Morrow,J.R.&Zlegler,W.inCatastrophicEventsandMassExtinctions:ImpactsandBeyond(edsKoeberl,C.&MacLeod,K.G.)473–387(GeologicalSocietyofAmericaSpecialPaper356,GSA,2002).68.McGhee,G.R.TheLateDevonianMassExtinction1–302(ColumbiaUniversityPress,1996).69.Murphy,A.E.,Sageman,B.B.&Hollander,D.J.EutrophicationbydecouplingofthemarinebiogeochemicalcyclesofC,N,andP:amechanismfortheLateDevonianmassextinction.Geology28,427–430(2000).70.Algeo,T.J.,Scheckler,S.E.&Maynard,J.B.inPlantsInvadetheLand:EvolutionaryandEnvironmentalApproaches(edsGensel,P.G.&Edwards,D.)213–236(ColumbiaUniversityPress,2000).71.Berner,R.A.ExaminationofhypothesesforthePermo-Triassicboundaryextinctionbycarboncyclemodeling.Proc.NatlAcad.Sci.USA99,4172–4177(2002).72.Payne,J.L.etal.Calciumisotopeconstraintsontheend-Permianmassextinction.Proc.NatlAcad.Sci.USA107,8543–8548(2010).73.Knoll,A.H.,Bambach,R.K.,Payne,J.L.,Pruss,S.&Fischer,W.W.Paleophysiologyandend-Permianmassextinction.EarthPlanet.Sci.Lett.256,295–313(2007).74.Hesselbo,S.P.,McRoberts,C.A.&Palfy,J.Triassic-Jurassicboundaryevents:problems,progress,possibilities.Palaeogeogr.Palaeoclimatol.Palaeoecol.244,1–10(2007).75.Ward,P.D.etal.SuddenproductivitycollapseassociatedwiththeTriassic-Jurassicboundarymassextinction.Science292,1148–1151(2001).76.Archibald,J.D.etal.Cretaceousextinctions:multiplecauses.Science328,973(2010).77.Keller,G.Cretaceousclimate,volcanism,impacts,andbioticeffects.Cretac.Res.29,754–771(2008).78.Mukhopadhyay,S.,Farley,K.A.&Montanari,A.AshortdurationoftheCretaceous-Tertiaryboundaryevent:evidencefromextraterrestrialhelium-3.Science291,1952–1955(2010).79.Royer,D.L.CO2-forcedclimatethresholdsduringthePhanerozoic.Geochim.Cosmochim.Acta70,5665–5675(2006).80.McCallum,M.L.Amphibiandeclineorextinction?Currentdeclinesdwarfbackgroundextinctionrate.J.Herpetol.41,483–491(2007).81.Rosenzweig,M.L.Lossofspeciationratewillimpoverishfuturediversity.Proc.NatlAcad.Sci.USA98,5404–5410(2001).82.Thomas,C.D.etal.Extinctionriskfromclimatechange.Nature427,145–148(2004).83.Pimm,S.L.&Raven,P.H.Extinctionbynumbers.Nature403,843–845(2000).84.Sinervo,B.etal.Erosionoflizarddiversitybyclimatechangeandalteredthermalniches.Science328,894–899(2010).85.Roelants,K.etal.Globalpatternsofdiversificationinthehistoryofmodernamphibians.Proc.NatlAcad.Sci.USA104,887–892(2007).86.Raup,D.M.AkillcurveforPhanerozoicmarinespecies.Paleobiology17,37–48(1991).87.Foote,M.Estimatingtaxonomicdurationsandpreservationprobability.Paleobiology23,278–300(1997).88.Rabosky,D.L.Extinctionratesshouldnotbeestimatedfrommolecularphylogenies.Evolution64,1816–1824(2010).89.Barnosky,A.D.&Lindsey,E.L.TimingofQuaternarymegafaunalextinctioninSouthAmericainrelationtohumanarrivalandclimatechange.Quat.Int.217,10–29(2010).90.Turvey,S.T.HoloceneExtinctions(OxfordUniversityPress,2009).91.Faith,J.T.&Surovell,T.A.SynchronousextinctionofNorthAmerica’sPleistocenemammals.Proc.NatlAcad.Sci.USA106,20641–20645(2009).92.Surovell,T.,Waguespack,N.&Brantingham,P.J.Globalarchaeologicalevidenceforproboscideanoverkill.Proc.NatlAcad.Sci.USA102,6231–6236(2005).93.Finlayson,C.etal.LatesurvivalofNeanderthalsatthesouthernmostextremeofEurope.Nature443,850–853(2006).94.Morwood,M.J.etal.ArchaeologyandageofanewhomininfromFloresineasternIndonesia.Nature431,1087–1091(2004).95.Orlova,L.A.,Vasil’ev,S.K.,Kuz’min,Y.V.&Kosintsev,P.A.NewdataonthetimeandplaceofextinctionofthewoollyrhinocerosCoelodontaantiquitatisBlumenbach,1799.Dokl.Akad.Nauk423,133–135(2008).96.Reumer,J.W.F.etal.LatePleistocenesurvivalofthesaber-toothedcatHomotheriuminNorthwesternEurope.J.Vertebr.Paleontol.23,260–262(2003).97.MacPhee,R.D.E.ExtinctionsinNearTime:Causes,Contexts,andConsequences(KluwerAcademic/PlenumPublishers,1999).AcknowledgementsS.Beissinger,P.Ehrlich,E.Hadly,A.Hubbe,D.Jablonski,S.PimmandD.Wakeprovidedconstructivecomments.PaleobiologyDatabasedatawerecontributedbyM.Carrano,J.Alroy,M.Uhen,R.Butler,J.Mueller,L.vandenHoekOstende,J.Head,E.Fara,D.Croft,W.Clyde,K.Behrensmeyer,J.Hunter,R.WhatleyandW.Kiessling.TheworkwasfundedinpartbyNSFgrantsEAR-0720387(toA.D.B.)andDEB-0919451(supportingN.M.).ThisisUniversityofCaliforniaMuseumofPaleontologyContribution2024.AuthorContributionsAllauthorsparticipatedinliteraturereviewandcontributedtodiscussionsthatresultedinthispaper.A.D.B.plannedtheproject,analysedandinterpreteddata,andwrotethepaper.N.M.andS.T.performedkeydataanalysesandinterpretationrelatingtoratecomparisons.G.O.U.W.,B.S.andE.L.L.assembledcriticaldata.T.B.Q.andC.M.provideddata,analysesandideasrelatingtodiversitydynamicsandrate-magnitudecomparisons.J.L.M.helpedproducefiguresandwithN.M.,S.T.,G.O.U.W.,B.S.,T.B.Q.,C.M.,K.C.M.,B.M.andE.A.F.contributedtofinalizingthetext.AuthorInformationReprintsandpermissionsinformationisavailableatwww.nature.com/reprints.Theauthorsdeclarenocompetingfinancialinterests.Readersarewelcometocommentontheonlineversionofthisarticleatwww.nature.com/nature.CorrespondenceshouldbeaddressedtoA.D.B.(barnosky@berkeley.edu).REVIEWRESEARCH3MARCH2011|VOL471|NATURE|57Macmillan Publishers Limited. All rights reserved©2011 insightreview articlesNATURE|VOL 418|8 AUGUST 2002|www.nature.com/nature689Fishing is the catching of aquatic wildlife, theequivalent of hunting bison, deer and rabbits onland. Thus, it is not surprising that industrial-scale fishing should generally not be sustainable:industrial-scale hunting, on land, would not be,either. What is surprising rather, is how entrenched thenotion is that unspecified Ôenvironmental changeÕ caused,and continues to cause, the collapse of exploited fishpopulations. Examining the history of fishing and fisheriesmakes it abundantly clear that humans have had forthousands of years a major impact on target species andtheir supporting ecosystems1. Indeed, the archaeologicalliterature contains many examples of ancient humanfishing associated with gradual shifts, through time, tosmaller sizes and the serial depletion of species that wenow recognize as the symptoms of overfishing1,2.This literature supports the claim that, historically, fish-eries have tended to be non-sustainable, although not unexpectedly there is a debate about the cause for this3, andthe exceptions4. The few uncontested historical examples ofsustainable fisheries seem to occur where a superabundanceof fish supported small human populations in challengingclimates5. Sustainability occurred where fish populationswere naturally protected by having a large part of their distribution outside of the range of fishing operations.Hence, many large old fecund females, which contributeoverwhelmingly to the egg production that renews fish pop-ulations, remained untouched. How important suchfemales can be is illustrated by the example of a single ripefemale red snapper, Lutjanus campechanus, of 61 cm and12.5 kg, which contains the same number of eggs(9,300,000) as 212 females of 42 cm and 1.1 kg each6. Wheresuch natural protection was absent, that is, where the entirepopulation was accessible to fishing gears, depletion ensued,even if the gear used seems inefficient in retrospect7,8. Thiswas usually masked, however, by the availability of otherspecies to target, leading to early instances of depletionsobservable in the changing size and species composition offish remains, for example, in middens9.The fishing process became industrialized in the earlynineteenth century when English fishers started operatingsteam trawlers, soon rendered more effective by powerwinches and, after the First World War, diesel engines10. Theaftermath of the Second World War added another ÔpeacedividendÕ to the industrialization of fishing: freezer trawlers,radar and acoustic fish finders. The fleets of the NorthernHemisphere were ready to take on the world.Fisheries science advanced over this time as well: the twoworld wars had shown that strongly exploited fish popula-tions, such as those of the North Sea, would recover most, ifnot all, of their previous abundance when released from fishing11. This allowed the construction of models of single-species fish populations whose size is affected only by fishingpressure, expressed either as a fishing mortality rate (F, orcatch/biomass ratio), or by a measure of fishing effort (f, forexample, trawling hours per year) related to Fthrough acatchability coefficient12,13 (q): F4qf. Here, qrepresents thefraction of a population caught by one unit of effort, directlyexpressing the effectiveness of a gear. Thus, qshould be monitored as closely as fishing effort itself, if the impact offishing on a given stock, as expressed by F, is to be evaluated.Technology changes tend to increase q, leading to increasesreferred to as Ôtechnology coefficientÕ14, which quickly renders meaningless any attempts to limit fishing mortalityby limiting only fishing effort.The conclusion of these models, still in use even now(although in greatly modified forms; Box 1), is that adjustingfishing effort to some optimum level should generate Ômaximum sustainableÕ yield, a notion that the fishingindustry and the regulatory agencies eagerly adopted Ñ ifonly in theory15. In practice, optimum effort levels were veryrarely implemented (the Pacific halibut fishery is one exception16). Rather the fisheries expanded their reach, bothoffshore, by fishing deeper waters and remote sea mounts17,and by moving onto the then untapped resources of WestAfrica18, southeast Asia19, and other low-latitude and Southern Hemispheric regions20.Fisheries go globalIn 1950, the newly founded Food and Agriculture Organiza-tion (FAO) of the United Nations began collection of globalstatistics. Fisheries in the early 1950s were at the onset of aperiod of extremely rapid growth, both in the NorthernHemisphere and along the coast of the countries of what isnow known as the developing world. Everywhere that indus-trial-scale fishing (mainly trawling, but also purse seiningTowards sustainability in world fisheriesDaniel Pauly, Villy Christensen, Sylvie GuŽnette, Tony J. Pitcher, U. Rashid Sumaila, Carl J. Walters, R. Watson & Dirk ZellerFisheries Centre, University of British Columbia, 2204 Main Mall, Vancouver, British Columbia, Canada V6T 1Z4 (e-mail: d.pauly@fisheries.ubc.ca)Fisheries have rarely been ÔsustainableÕ. Rather, fishing has induced serial depletions, long masked byimproved technology, geographic expansion and exploitation of previously spurned species lower in the foodweb. With global catches declining since the late 1980s, continuation of present trends will lead to supplyshortfall, for which aquaculture cannot be expected to compensate, and may well exacerbate. Reducingfishing capacity to appropriate levels will require strong reductions of subsidies. Zoning the oceans intounfished marine reserves and areas with limited levels of fishing effort would allow sustainable fisheries,based on resources embedded in functional, diverse ecosystems.© 2002 Nature PublishingGroup Throughout the 1950s and 1960s, this huge increase of global fishingeffort led to an increase in catches (Fig. 1) so rapid that their trendexceeded human population growth, encouraging an entire genera-tion of managers and politicians to believe that launching more boatswould automatically lead to higher catches.The first collapse with global repercussions was that of the Peru-vian anchoveta in 1971Ð1972, which is often perceived as having beencaused by an El Ni–o event. However, much of the available evidence,including actual catches (about 18 million tonnes22) exceeding offi-cially reported catches (12 million tonnes), suggest that overfishingwas implicated as well. But attributing the collapse of the Peruviananchoveta to Ôenvironmental effectsÕ allowed business as usual tocontinue and, in the mid-1970s, this led to the beginning of a declinein total catches from the North Atlantic. The declining trend accelerated in the late 1980s and early 1990s when most of the codstocks off New England and eastern Canada collapsed, ending fishingtraditions reaching back for centuries23.Despite these collapses, the global expansion of effort continued14and trade in fish products intensified to the extent that they have nowbecome some of the most globalized commodities, whose priceincreased much faster than the cost of living index24. In 1996, FAOpublished a chronicle of global fisheries showing that a rapidlyincreasing fraction of world catches originate from stocks that aredepleted or collapsed, that is, ÔsenescentÕ in FAOÕs parlance25. Yet,insightreview articles690NATURE|VOL 418|8 AUGUST 2002|www.nature.com/natureand long-lining) was introduced, it competed with small-scale, orartisanal fisheries. This is especially true for tropical shallow waters(10Ð100 m), where artisanal fisheries targeting food fish for localconsumption, and trawlers targeting shrimps for export, and dis-carding the associated by-catch, compete for the same resource21.Box 1Single-species stock assessmentsSingle-species assessments have been performed since the early1950s, when the founders of modern fisheries science12,13attempted to equate the concept of sustainability with the notion ofoptimum fishing mortality, leading to some form of maximumsustainable yield. Most of these models, now much evolved fromtheir original versions (some to baroque complexity, involvinghundreds of free parameters), require catch-at-age data. Hencegovernment laboratories, at least in developed countries, spend alarge part of their budget on the routine acquisition andinterpretation of catch and age-composition data.Yet, single-species assessment models and the related policieshave not served us particularly well, due to at least four broadproblems. First, assessment results, although implying limitation onlevels of fishing mortality which would have helped maintain stocks ifimplemented, have often been ignored, on the excuse that theywere not Ôprecise enoughÕ to use as evidence for economicallypainful restriction of fishing (the Ôburden of proofÕ problem86). Second, the assessment methods have failed badly in a fewimportant cases involving rapid stock declines, and in particularhave led us to grossly underestimate the severity of the decline andthe increasing (ÔdepensatoryÕ) impacts of fishing during thedecline87. Third, there has been insufficient attention in some cases toregulatory tactics: the assessments and models have providedreasonable overall targets for management (estimates of long-termsustainable harvest), but we have failed to implement and evendevelop effective short-term regulatory systems for achieving thosetargets88.Fourth, we have seen apparently severe violation of theassumptions usually made about Ôcompensatory responsesÕ inrecruitment to reduction in spawning population size. We haveusually assumed that decreasing egg production will result inimproving juvenile survival (compensation) so that recruitment(eggs2survival) will not fall off rapidly during a stock decline and willhence tend to stop the decline. Some stocks have shownrecruitment failure after severe decline, possibly associated withchanges in feeding interactions that are becoming known asÔcultivation/depensationÕ effects89. According to this phenomenon,adult predatory fish (such as cod) can control the abundance ofpotential predators and competitors of their juvenile offspring, butthis control lost when these predatory fish become scarce. This maywell lead to alternate stable states of ecosystems, which has severeimplications for fisheries management90.Jointly, these four broad problems imply a need to complementour single-species assessments by elements drawn from ecology,that is, to move towards ecosystem-based management. What thiswill consist of is not clearly established, although it is likely that, whileretaining single-species models at its core, it will have to explicitlyinclude trophic interaction between species91, habitat impacts ofvarious gears50, and a theory for dealing with the optimumplacement and size of marine reserves (see main text). Ecosystem-based management will have to rely on the principles of, andlessons learnt from, single-species stock assessments, especiallyregarding the need to limit fishing mortality. It will certainly not beapplicable in areas where effort or catch limits derived from single-species approaches cannot be implemented in the first place.There are many ways ecosystems can be described, for example interms of the information that is exchanged as their componentsinteract, or in terms of size spectra. But perhaps the moststraightforward way to describe ecosystems is in terms of thefeeding interactions among their component species, which can bedone by studying their stomach contents. A vast historical databaseof such published studies exists27, which has enabled a number ofuseful generalizations to be made for ecosystem-basedmanagement of fisheries. One of these is that marine systems haveherbivores (zooplankton) that are usually much smaller than thefirst-order carnivores (small fishes), which are themselvesconsumed by much larger piscivorous fishes, and so on. This is asignificant difference from terrestrial systems, where, for example,wolves are smaller than the moose they prey on. Anothergeneralization is that the organisms we have so far extracted frommarine food webs have tended to play therein roles very differentfrom those played by the terrestrial animals we consume. This canbe shown in terms of their Ôtrophic levelÕ (TL), defined as 1&themean TL of their prey.Thus, in marine systems we have: algae at the bottom of thefood web (TL41, by definition); herbivorous zooplankton feedingon the algae (TL42); large zooplankton or small fishes, feeding onthe herbivorous zooplankton (TL43); large fishes (for example, cod,tuna and groupers) whose food tends to be a mixture of low- andhigh-TL organisms (TL43.5Ð4.5).The mean TL of fisheries landings can be used as an index ofsustainability in exploited marine ecosystems. Fisheries tend at firstto remove large, slower-growing fishes, and thus reduce the meanTL of the fish remaining in an ecosystem. This eventually leads todeclining trends of mean TL in the catches extracted from thatecosystem, a process now known as Ôfishing down marine foodwebsÕ29.Declining TL is an effect that occurs within species as well asbetween species. Most fishes are hatched as tiny larvae that feedon herbivorous zooplankton. At this stage they have a TL of about3, but this value increases with size, especially in piscivorousspecies. Because fisheries tend to reduce the size of the fish in anexploited stock, they also reduce their TL.Box 2Trophic levels as indicators of fisheries impacts© 2002 Nature PublishingGroup insightreview articlesNATURE|VOL 418|8 AUGUST 2002|www.nature.com/nature691global catches seemed to continue, increasing through the 1990saccording to official catch statistics. This surprising result wasexplained recently when massive over-reporting of marine fisheriescatches by one single country, the PeopleÕs Republic of China, wasuncovered26. Correcting for this showed that reported world fisherieslandings have in fact been declining slowly since the late 1980s, byabout 0.7 million tonnes per year.Fisheries impact on ecosystem and biodiversityThe position within ecosystems of the fishes and invertebrates landedby fisheries can be expressed by their trophic levels, expressing thenumber of steps they are removed from the algae (occupying a troph-ic level of 1) that fuel marine food webs (Box 2). Most food fishes havetrophic levels ranging from 3.0 to 4.5, that is, from sardines feedingon zooplankton to large cod or tuna feeding on miscellaneous fish-es27. Thus, the observed global decline of 0.05Ð0.10 trophic levels perdecade in global fisheries landings (Fig. 2) is extremely worrisome, asit implies the gradual removal of large, long-lived fishes from theecosystems of the world oceans. This is perhaps most clearly illustrat-ed by a recent study in the North Atlantic showing that the biomass ofpredatory fishes (with a trophic level of 3.75 or more) declined bytwo-thirds through the second half to the twentieth century, eventhough this area was already severely depleted before the start of thistime period28.It may be argued that so-called Ôfishing down marine food websÕ isboth a good and an unavoidable thing, given a growing demand forfish29. Indeed, the initial ecosystem reaction to the process may be arelease from predation, where cascading effects may lead to increasedcatches30. Such effects are, however, seldom observed in marineecosystems31,32, mainly because they do not function simply as anumber of unconnected food chains. Rather, predators operate within finely meshed food webs, whose structure (which they helpmaintain) tends to support the production of their prey. Hence theconcept of Ôbeneficial predationÕ, where a predator may have a directnegative impact on its prey, but also an indirect positive effect, byconsuming other predators and competitors of the prey33(and seeBox 1). Thus, removing predators does not necessarily lead to moreof their prey becoming available for humans. Instead, it leads toincreases or outbursts of previously suppressed species, often invertebrates30,34,35, some of which may be exploited (for example,squid or jellyfish, the latter a relatively new resource, exported to eastAsia), and some outright noxious36.The principal, direct impact of fishing is that it reduces the abundance of target species. It has often been assumed that this doesnot impose any direct threat of species extinction as marine fish gen-erally are very fecund and the ocean expanse is wide37. But the pastfew decades have witnessed a growing awareness that fishes can notonly be severely depleted, but also be threatened with extinctionthrough overexploitation38. Among commercially importantspecies, those particularly at risk are species that are highly valued,large and slow to mature, have limited geographical range, and/orhave sporadic recruitment39. There is actually little support, though,for the general assumption that the most highly fecund marine fishspecies are less susceptible to overexploitation; rather it seems thatthis perception is flawed40. Fisheries may also change the evolution-ary characteristics of populations by selectively removing the larger,fast-growing individuals, and one important research question iswhether this induces irreversible changes in the gene pool41. Overall,this has implications for research, monitoring and management, andit points to the need for incorporating ecological consideration infisheries management42,43, as exemplified by the development ofquantitative guidelines to avoid local extinctions44.Another worrisome aspect of fishing down marine food webs isthat it involves a reduction of the number and length of pathwayslinking food fishes to the primary producers, and hence a simplifica-tion of the food webs. Diversified food webs allow predators to switchbetween prey as their abundance fluctuates45, and hence to compen-sate for prey fluctuations induced by environmental fluctuations46.Fisheries-induced food-web simplification, combined with the drastic fisheries-induced reduction in the number of year classes inpredator populations47,48, makes their reduced biomass stronglydependent of annual recruitment. This leads to increasing variability,and to lack of predictability in population sizes, and hence in predicted catches. The net effect is that it will increasingly look likeenvironmental fluctuations impact strongly on fisheries resources,even where they originally did not. This resolves, if in a perverse way,the question of the relative importance of fisheries and environmen-tal variability as the major driver for changes in the abundance offisheries resources49(Fig. 3).It seems unbelievable in retrospect, but there was a time when itwas believed that bottom trawling had little detrimental impact, oreven a beneficial impact, on the sea bottom that it ÔploughedÕ. Recentresearch shows that the ploughing analogy is inappropriate and thatif an analogy is required, it should be that of clear cutting forests in thecourse of hunting deer. Indeed, the productivity of the benthicorganisms at the base of food webs leading to food fishes is seriouslyimpacted by bottom trawling50, as is the survival of their juvenileswhen deprived of the biogenic bottom structure destroyed by thatFigure 1Estimated global fishlandings 1950Ð1999. Figures forinvertebrates, groundfish, pelagicfish and Peruvian anchoveta arefrom FAO catch statistics, withadjustment for over-reporting fromChina26. Fish caught but thendiscarded were not included in theFAO landings; data relate to theearly 1990s83were madeproportional to the FAO landings forother periods. Other illegal,unreported or unregulated (IUU)catches65were estimated byidentifying, for each 5-year block,the dominant jurisdiction and gearuse (and hence incentive for IUU)84;reported catches were then raised by the percentage of IUU in major fisheries for each 5-year block. The resulting estimates of IUU are very tentative (note dotted y-axis), and weconsider that complementing landings statistics with more reliable estimates of discards and IUU is crucial for a transition to ecosystem-based management.150 100 5019901980197019601950YearCatch (million tonnes per year)Pelagic fishGroundfishInvertebratesPeruvian anchovetaDiscardsIUU© 2002 Nature PublishingGroup is retired, but its licence. This means that ÔretiredÕ vessels can still beused to catch species without quota (so-called Ôunder-utilizedresourcesÕ, which are often the prey of species for which there is aquota), or deployed along the coast of some developing country, theaccess to which may also be subsidized18. Clearly, the decommission-ing schemes that will have to be implemented if we are ever to reduceovercapacity will have to address these deficiencies if they are not toend up, as most have so far, in fleet modernization and increased fishing mortality.It is clear that a real, drastic reduction of overcapacity will have tooccur if fisheries are to acquire some semblance of sustainability. Therequired reductions will have to be strong enough to reduce Fby afactor of two or three in some areas, and even more in others. Thismust involve even greater decreases in f, because catches can be main-tained in the face of dwindling biomasses by increasing q(and henceF; see definitions above), even when nominal effort is constant.Indeed, this is the very reason behind the incessant technologicalinnovation in fisheries, which now relies on global positioning systems and detailed maps of the sea bottom to seek out residual fish concentrations previously protected by rough terrain. This tech-nological race, and the resulting increase in q, is also the reason whyfishers often remain unaware of their own impacts on the resourcethey exploit and object so strongly to scientistsÕ claims of reductionsin biomass.If fleet reduction is done properly, it should result in an increase innet benefits (ÔrentÕ) from the resources, as predicted by the basic theory of bioeconomics62. This can be used, via taxation of the rentgained by the remaining fishers, to ease the transition of those whohad to stop fishing. This would contrast with the present situation,where taxes from outside the fisheries sector are used, in form of subsidies, to maintain fishing at levels that are biologically unsus-tainable, and which ultimately lead to the depletion and collapse ofthe underlying resources.Biological constraints to fisheries and aquaculturePerhaps the strongest factor behind the politiciansÕ use of tax moneyto subsidize non-sustainable, even destructive fisheries, and its tacitsupport by the public at large, is the notion that, somehow, theoceans will yield what we need Ñ just because we need it. Indeed,insightreview articles692NATURE|VOL 418|8 AUGUST 2002|www.nature.com/natureform of fishing51. Hence, given the extensive coverage of the worldÕsshelf ecosystems by bottom trawling52, it is not surprising that gener-ally longer-lived, demersal (bottom) fishes have tended to declinefaster than shorter-lived, pelagic (open water) fishes, a trend alsoindicated by changes in the ratio of piscivorous (mainly demersal) tozooplanktivorous (mainly pelagic) fishes53.It is difficult to fully appreciate the extent of the changes to ecosys-tems that fishing has wrought, given shifting baselines as to what isconsidered a pristine ecosystem1,54and continued reliance on single-species models (Box 1). These changes, often involving reductions ofcommercial fish biomasses to a few per cent of their pre-exploitationlevels, prevent us taking much guidance from the concept of sustainability, understood as aiming to maintain what we have3,8.Rather, the challenge is rebuilding the stocks in question.Reducing fishing capacityThere is widespread awareness that increases in fishing-fleet capacityrepresent one of the main threats to the long-term survival of marinecapture-fishery resources, and to the fisheries themselves55,56. Rea-sons advanced for the overcapitalization of the worldÕs fisheriesinclude: the open-access nature of many fisheries57; common-poolfisheries that are managed non-cooperatively58,59; sole-ownershipfisheries with high discount rates and/or high price-to-cost ratios60;the increasing replacement of small-scale fishing vessels with largerones55; and the payment of subsidies by governments to fishers61,which generate ÔprofitsÕ even when resources are overfished.This literature shows that fishing overcapacity is likely to build upnot only under open access62, but also under all forms of propertyregimes. Subsidies, which amount to US$2.5 billion for the NorthAtlantic alone, exacerbate the problems arising from the open accessand/or Ôcommon poolÕ aspects of capture fisheries, including fisheries with full-fledged property rights61,63.Even subsidies used for vessel decommissioning schemes can havenegative effects. In fact, decommissioning schemes can lead to theintended reduction in fleet size only if vessel owners are consistentlycaught by surprise by those offering this form of subsidy. As this is anunlikely proposition, decommissioning schemes often end up providing the collaterals that banks require to underwrite fleet mod-ernizations. Additionally, in most cases, it is not the actual vessel that3.53.73.33.12.92.72.5197019751980198519901995Global freshwater fisheriesGlobal marine fisheriesNorway aquacultureYearMean trophic levelFigure 2Fisheries, both marine and freshwater, are characterized by a decline of themean trophic level in the landings, implying an increased reliance on organisms low infood webs (data from FishBase27, with Peru/Chile excluded owing to the dominance ofPeruvian anchoveta; see also Fig. 1). Freshwater fisheries have lower trophic levelvalues overall, indicating an earlier onset of the Ôfishing downÕ phenomenon29. Thetrend is inverted in non-Asian aquaculture, whose production consists increasingly ofpiscivorous organisms, as illustrated here for Norway (a major producer, yetrepresentative country)85.TimePopulation sizeUnexploited populationExploited population protected by no-take reservesExploited populationwith no reservesPopulation crashEnvironmentalvariationFigure 3Schematic representation of the effects of some environmental variation onan unexploited, exploited but protected, and exploited but unprotected fish population.This illustrates how protection through a marine reserve (and/or stock rebuilding) canmitigate the effects of environmental fluctuations, including Ôregime shiftsÕ49. (Graphfrom J. Jackson, personal communication.) © 2002 Nature PublishingGroup insightreview articlesNATURE|VOL 418|8 AUGUST 2002|www.nature.com/nature693demand projections generated by national and international agen-cies largely reflect present consumption patterns, which by somemeans the oceans ought to help us maintain, even if the global humanpopulation were to double again. Although much of the deep ocean isindeed unexplored and ÔmysteriousÕ, we know enough about oceanprocesses to realize that its productive capacity cannot keep up withan ever-increasing demand for fish.Just as a tropical scientist might look at the impressive expanse ofCanada and assume that this country has boundless potential foragricultural production, unaware that in reality only the thin sliver ofland along its southern border (5%) is arable, we terrestrial alienshave assumed that the expanse and depths of the worldÕs oceans willprovide for us in the ways that its more familiar coastal fringes have.But this assumption is very wrong. Of the 363 million square kilometres of ocean on this planet, less than 7% Ñ the continentalshelves Ñ are shallower than 200 m, and some of this shelf area is cov-ered by ice. Shelves generate the biological production supportingover 90% of global fish catches, the rest consisting of tuna and otheroceanic organisms that gather their food from the vast, desert-likeexpanse of the open oceans.The overwhelming majority of shelves are now ÔshelteredÕ withinthe exclusive economic zones (EEZ) of maritime countries, whichalso include all coral reefs and their fisheries (Box 3). According to the1982 United Nations Convention on the Law of the Sea64, any countrythat cannot fully utilize the fisheries resource of its EEZ must makethis surplus available to the fleet of other countries. This, along witheagerness for foreign exchange, political pressure18and illegal fishing65, has led to all of the worldÕs shelves being trawled for bottomfish, purse-seined for pelagic fishes and illuminated to attract andcatch squid (to the extent that satellites can map the night time location of fishing fleets as well as that of cities). Overall, about 35%of the primary production on the worldÕs shelves is required to sustain the fisheries66, a figure similar to the human appropriation ofterrestrial primary production67.The constraints to fisheries expansion that this implies, combinedwith the declining catches alluded to above, have led to suggestionsthat aquaculture should be able to bridge the gap between supply anddemand. Indeed, the impressive recent growth of reported aquacul-ture is often cited as evidence of the potential of that sector to meet thegrowing demand for fish, or even to Ôfeed the worldÕ.Three lines of argument suggest that this is unlikely. The first isthat the rapidly growing global production figures underlying thisdocumented growth are driven to a large extent by the PeopleÕsRepublic of China, which reported 63% of world aquaculture production in 1998. But it is now known that China not only over-reports its marine fisheries catches, but also the production of manyother sectors of its economy68. Thus, there is no reason to believe thatglobal aquaculture production in the past decades has risen as muchas officially reported.Second, modern aquaculture practices are largely unsustainable:they consume natural resources at a high rate and, because of theirintensity, they are extremely vulnerable to the pollution and diseaseoutbreaks they induce. Thus, shrimp aquaculture ventures are inmany cases operated as slash-and-burn operations, leaving devastat-ed coastal habitats and human communities in their wake69,70.Third, much of what is described as aquaculture, at least inEurope, North America and other parts of the developed world, consists of feedlot operations in which carnivorous fish (mainlysalmon, but also various sea bass and other species) are fattened on adiet rich in fish meal and oil. The idea makes commercial sense, as thefarmed fish fetch a much higher market price than the fish ground upfor fish meal (even though they may consist of species that are consumed by people, such as herring, sardine or mackerels, formingthe bulk of the pelagic fishes in Fig. 1). The point is that operations ofthis type, which are directed to wealthy consumers, use up muchmore fish flesh than they produce, and hence cannot replace capturefisheries, especially in developing countries, where very few canafford imported smoked salmon. Indeed, this form of aquaculturerepresents another source of pressure on wild fish populations71.PerspectivesWe believe the concept of sustainability upon which most quantita-tive fisheries management is based72to be flawed, because there is lit-tle point in sustaining stocks whose biomass is but a small fraction ofits value at the onset of industrial-scale fishing. Rebuilding of marinesystems is needed, and we foresee a practical restoration ecology forthe oceans that can take place alongside the extraction of marineresources for human food. Reconciling these apparently dissonantgoals provides a major challenge for fisheries ecologists, for the public, for management agencies and for the fishing industry73. It isimportant here to realize that there is no reason to expect marineresources to keep pace with the demand that will result from ourgrowing population, and hopefully, growing incomes in now impoverished parts of the world, although we note that fisheriesdesigned to be sustainable in a world of scarcity may be profitable.We argued in the beginning of this review that whatever sem-blance of sustainability fisheries in the past might have had was dueGlobally, 75% of coral reefs occur in developing countries wherehuman populations are still increasing rapidly. Although coral reefsaccount for only 0.1% of the worldÕs ocean, their fisheries resourcesprovide tens of millions of people with food and livelihood92. Yet,their food security, as well as other ecosystem functions theyprovide, is threatened by various human activities, many of which,including forest and land management, are unrelated to fishing93.It has often been assumed that the high levels of primaryproductivity reported for coral reefs imply high fisheries yields94.However, the long-held notion that coral reef fishes are ÔfastturnoverÕ species, capable of high productivity, is being increasinglychallenged95. Yield estimates for coral reefs vary widely, rangingfrom 0.2 to over 40 tonnes kmÐ2yr-1(ref. 96), depending on what isdefined as coral reef area, and as coral reef fishes96,97. Taking yieldsfrom the central part of this range (5Ð15 tonnes kmÐ2yrÐ1) and themost comprehensive reef-area estimate available92, we derive anestimate for total global annual yield of 1.4Ð4.2 million tonnes.Although these estimates represent only 2Ð5% of global fisheriescatches, they provide an important, almost irreplaceable, source ofanimal protein to the populations of many developing countries96.Clearly, maintaining the biodiversity that is a characteristic ofhealthy reefs is the key to maintaining sustainable reef fisheries. Yetcoral reefs throughout the world are being degraded rapidly,especially in developing countries93. Concerns regardingoverexploitation of reef fisheries are widespread1,75,98. The entry ofnew, non-traditional fishers into reef fisheries has led to intensecompetition and the use of destructive fishing implements, such asexplosives and poisons, a process known as ÔmalthusianoverfishingÕ21.Another major problem is the growing international trade for livereef fish99, often associated with mobile fleets using cyanide fishing,and targeting species that often have limited ranges ofmovements100. This leads to serial depletion of large coral reeffishes, notably the humphead wrasse (Cheilinus undulatusLabridae), groupers (Serranidae) and snappers (Lutjanidae), and toreefs devastated by the cyanide applications.These fisheries, which destroy the habitat of the species uponwhich they rely, are inherently unsustainable. It can be expected thatthey will have to cease operating within a few decades, that is,before warm surface waters and sea-level rise overcome what maybe left of the worldÕs coral reefs.Box 3Sustainable coral reef fisheries: an oxymoron?© 2002 Nature PublishingGroup goal82. There is still time to achieve this, and for our fisheries to be put on a path towards sustainability.nndoi:10.1038/nature010171.Jackson, J. B. C. et al.Historical overfishing and the recent collapse of coastal ecosystems. Science293,629Ð638 (2001).2.Orensanz, J. M. L., Armstrong, J., Armstrong, D. & Hilborn, R. Crustacean resources are vulnerable toserial depletionÑthe multifaceted decline of crab and shrimp fisheries in the Greater Gulf of Alaska.Rev. Fish Biol. Fish.8,117Ð176 (1998).3.Ludwig, D., Hilborn, R. & Walters, C. Uncertainty, resource exploitation, and conservation: Lessonsfrom history. Science260,17Ð18 (1993).4.Rosenberg, A. A., Fogarty, M. J., Sissenwine, M. P., Beddington, J. R. & Shepherd, J. G. Achievingsustainable use of renewable resources. Science262,828Ð829 (1993).5.Boyd, R. T. in Handbook of American Indians: Northwest Coast(ed. Suttles, W.) 135Ð148 (SmithsonianInstitute, Washington DC, 1990).6.Bohnsack, J. A. (Subcommittee Chair) NOAA Tech. Memo. NMFS-SEFC-261 (National Oceanic andAtmospheric Agency, Miami, 1990).7.Yellen, J. E., Brooks, A. S., Cornelissen, E., Mehlman, M. J. & Stewart, K. A middle stone-age workedbone industry from Katanda, Upper Semliki Valley, Zaire. Science268,553Ð556 (1995).8.Pitcher, T. J. Fisheries managed to rebuild ecosystems? Reconstructing the past to salvage the future.Ecol. Applic.11,601Ð617 (2001).9.Wing, E. S. The sustainability of resources used by native Americans on four Caribbean islands. Int. J.Osteoarchaeol.11,112Ð126 (2001).10.Cushing, D. H. The Provident Sea(Cambridge Univ. Press, Cambridge, 1987).11.Hardy, A. The Open Sea(Collins, London, 1956).12.Schaefer, M. B. Some aspects of the dynamics of populations important to the management of thecommercial marine fisheries. Bull. Inter-Am. Trop. Tuna Commiss.1,27Ð56 (1954).13.Beverton, R. J. H. & Holt, S. J. On the Dynamics of Exploited Fish Populations(Chapman and Hall,London, 1957; Facsimile reprint 1993).14.Garcia, S. M. & Newton, C. in Global Trends: Fisheries Management(ed. Sissenwine, M. P.) Am. Fish.Soc. Symp. 20, 3Ð27 (American Fisheries Society, Bethesda, MD, 1997).15.Mace, P. M. A new role for MSY in single-species and ecosystem approaches to fisheries stockassessment and management. Fish Fish.2,2Ð32 (2001).16.Clark, W. G., Hare, S. R., Parma, A. M., Sullivan, P. J. & Trumble, R. J. Decadal changes in growth andrecruitment of Pacific halibut (Hippoglossus stenolepis). Can. J. Fish. Aquat. Sci.56,242Ð252 (1999).17.Koslow, J. A. et al.Continental slope and deep-sea fisheries: implications for a fragile ecosystem. ICESJ. Mar. Sci.57,548Ð557 (1999).18.Kaczynski, V. M. & Fluharty, D. L. European policies in West Africa: who benefits from fisheriesagreements. Mar. Policy26,75Ð93 (2002).19.Silvestre, G. & Pauly, D. Status and Management of Tropical Coastal Fisheries in Asia(ICLARM, MakatiCity, Philippines, 1997).20.Thorpe, A. & Bennett, E. Globalisation and the sustainability of world fisheries: a view from LatinAmerica. Mar. Resource Econ.16,143Ð164 (2001).21.Pauly, D. Small-scale Fisheries in the Tropics: Marginality, Marginalization, and Some Implications forFisheries Management(eds Pikitch, E. K., Huppert, D. D. & Sissenwine, M. P.) (American FisheriesSociety, Bethesda, MD,1997).22.Castillo, S. & Mendo, J. in The Peruvian Anchoveta and its Upwelling Ecosystem: Three Decades ofChange(eds Pauly, D. & Tsukayama, I.) ICLARM Stud. Rev. 15, 109Ð116 (ICLARM, Manila,Philippines, 1987).23.Myers, R. A., Hutchings, J. A. & Barrowman, N. J. Why do fish stocks collapse? The example of cod inAtlantic Canada. Ecol. Applic.7,91Ð106 (1997).24.Sumaila, U. R. in Proc. Expo Ô98 Conf. Ocean Food Webs Econ. Product., Lisbon, 1-3 July 1998(edsPauly, D., Christensen, V. & Coelho, L.) ACP-EU Fish. Res. Rep. 5, 87 (ACP-EU, Brussels, 1999).25.Grainger, R. J. R. & Garcia, S. M. Chronicles of Marine Fishery Landings (1950-1994). Trend Analysisand Fisheries PotentialFAO Fish. Tech. Pap. 359 (Food and Agriculture Organization of the UnitedNations, Rome, 1996).26.Watson, R. & Pauly, D. Systematic distortions in world fisheries catch trends. Nature424,534Ð536 (2001).27.Froese, R. & Pauly, D. (eds) FishBase 2000: Concepts, Design and Data Sources(distributed with fourCD-ROMs; updates available at http://www.fishbase.org) (ICLARM, Los Ba–os, Philippines, 2000).28.Christensen, V. et al.in Fisheries Impacts on North Atlantic Ecosystems: Models and Analyses(edsGuŽnette, S., Christensen, V. & Pauly, D.) Fisheries Centre Res. Rep. 9(4), 1Ð25 (also available athttp://www.fisheries.ubc.ca) (Fisheries Centre, Univ. British Columbia, Vancouver, 2002).29.Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. & Torres, F. Jr Fishing down marine food webs.Science279,860Ð863 (1998).30.Daskalov, G. M. Overfishing drives a trophic cascade in the Black Sea. Mar. Ecol. Prog. Ser.225,53Ð63(2002).31.Pace, M. L., Cole, J. J., Carpenter, S. R. & Kitchell, J. F. Trophic cascades revealed in diverse ecosystems.Trends Ecol. Evol.14,483Ð488 (1999).32.Pinnegar, J. K. et al.Trophic cascades in benthic marine ecosystems: lessons for fisheries andprotected-area management. Environ. Conserv.27,179Ð200 (2000).33.Ulanowicz, R. E. & Puccia, C. J. Mixed trophic impacts in ecosystems. Coenoses5,7Ð16. (1990).34.Parsons, T. R. in Food Webs: Integration of Patterns and Dynamics(eds Polis, G. A. & Winemiller, K.D.) 352Ð357 (Chapman and Hall, New York, 1996).35.Mills, C. E. Jellyfish blooms: are populations increasing globally in response to changing oceanconditions? Hydrobiologia451,55Ð68 (2001).36.Van Dolah, F. M., Roelke, D. & Greene, R. M. Health and ecological impacts of harmful algal blooms:risk assessment needs. Hum. Ecol. Risk Assess.7,1329Ð1345 (2001).37.Pitcher, T. J. A cover story: fisheries may drive stocks to extinction. Rev. Fish Biol. Fish.8, 367Ð370 (1998).38.Casey, J. M. & Myers, R. A. Near extinction of a large, widely distributed fish. Science281,690Ð692 (1998).39.Sadovy, Y. The threat of fishing to highly fecund fishes. J. Fish Biol.59,90Ð108 (2001).40.Hutchings, J. A. Collapse and recovery of marine fishes. Nature406,882Ð885 (2000).41.Law, R. Fishing, selection, and phenotypic evolution. ICES J. Mar. Sci.57,659Ð668 (2000).42.Gislason, H., Sinclair, M., Sainsbury, K. & OÕBoyle, R. Symposium overview: incorporating ecosystemobjectives within fisheries management. Ices J. Mar. Sci.57,468Ð475 (2000).insightreview articles694NATURE|VOL 418|8 AUGUST 2002|www.nature.com/natureto their inability to cover the entire range inhabited by the wildlifespecies that were exploited, which thus had natural reserves. We further argued that the models used traditionally to assess fisheries,and to set catch limits, tend to require explicit knowledge on stockstatus and total withdrawal from stocks, that is, knowledge that willinherently remain imprecise and error prone. We also showed thatgenerally overcapitalized fisheries are leading, globally, to the gradual elimination of large, long-lived fishes from marine ecosystems, and their replacement by shorter-lived fishes and invertebrates, operating within food webs that are much simplifiedand lack their former ÔbufferingÕ capacity.If these trends are to be reversed, a huge reduction of fishingeffort involving effective decommissioning of a large fraction of the worldÕs fishing fleet will have to be implemented, along with fisheries regulations incorporating a strong form of the precaution-ary principle. The conceptual elements required for this are in place,for example, in form of the FAO Code of Conduct for ResponsibleFisheries74, but the required political will has been lacking so far, anabsence that is becoming more glaring as increasing numbers offisheries collapse throughout the world, and catches continue todecline.Given the high level of uncertainty facing the management offisheries, which induced several collapses, it has been suggested bynumerous authors that closing a part of the fishing grounds wouldprevent overexploitation by setting an upper limit on fishing mortality. Marine protected areas (MPAs), with no-take reserves attheir core, combined with a strongly limited effort in the remainingfishable areas, have been shown to have positive effects in helping torebuild depleted stocks75Ð77. In most cases, the successful MPAs were used to protect rather sedentary species, rebuild their biomass,and eventually sustain the fishery outside the reserves by exportingjuveniles or adults75. Although migrating species would not benefitfrom the local reduction in fishing mortality caused by an MPA78,79,the MPA would still help some of these species by rebuilding thecomplexity of their habitat destroyed by trawling, and thus decreasemortality of their juveniles80. Enforcement of the no-take zoneswithin MPAs would benefit from the application of high technology(for example, satellite monitoring of fishing vessels), presently usedmainly to increase fishing pressure.There is still much fear among fisheries scientists, especially inextra-tropical areas, that the export of fish from such reserves wouldnot be sufficient to compensate for the loss of fishing ground81.Although we agree that marine reserves are no panacea, the presenttrends in fisheries, combined with the low degree of protectionpresently afforded (only 0.01% of the worldÕs ocean is effectivelyprotected), virtually guarantee that more fish stocks will collapse,and that these collapses will be attributed to environmental fluctuations or climate change (Fig. 3). Moreover, many exploitedfish populations and eventually fish species will become extinct.MPAs that cover a representative set of marine habitats should helpprevent this, just like forest and other natural terrestrial habitatshave enabled the survival of wildlife species which agriculture wouldhave otherwise rendered extinct.Focused studies on the appropriate size and location of marinereserves and their combination into networks, given locale-specificoceanographic conditions, should therefore be supported. This willlead to the identification of reserve designs that would optimizeexport to adjacent fished areas, and which could thus be offered to the affected coastal and fisher communities, whose consent andsupport will be required to establish marine reserves and restructurethe fisheries8. The general public could also be involved, througheco-labelling and other market-driven schemes, and through sup-port for conservation-orientated non-government organizations,which can complement the activities of governmental regulatoryagencies.In conclusion, we think that the restoration of marine ecosystems to some state that existed in the past is a logical policy© 2002 Nature PublishingGroup insightreview articlesNATURE|VOL 418|8 AUGUST 2002|www.nature.com/nature69543.Hutchings, J. A. Numerical assessment in the front seat, ecology and evolution in the back seat: Timeto change drivers in fisheries and aquatic sciences? Mar. Ecol. Prog. Ser.208,299Ð313 (2000).44.Punt, A. E. Extinction of marine renewable resources: a demographic analysis. Popul. Ecol.42,19Ð27 (2000).45.Stephens, D. W. & Krebs, J. R. Foraging Theory(Princeton Univ. Press, Princeton, 1986).46.Neutel, A.-M., Heesterbeek, J. A. P. & de Ruiter, P. C. Stability in real food webs: weak links in longloops. Science296,1120Ð1123 (2002).47.Longhurst, A. Cod: perhaps if we all stood back a bit? Fish. Res.38,101Ð108 (1998).48.Stergiou, K. I. Overfishing, tropicalization of fish stocks, uncertainty and ecosystem management:resharpening OckhamÕs razor. Fish. Res.55,1Ð9 (2002).49.Steele, J. H. Regime shifts in marine ecosystems. Ecol. Applic.8(Suppl. 1), S33ÐS36 (1998).50.Hall, S. J. The Effects of Fisheries on Ecosystems and Communities(Blackwell, Oxford, 1998).51.Turner, S. J., Thrush, S. F., Hewitt, J. E., Cummings, V. J. & Funnell, G. Fishing impacts and thedegradation or loss of habitat structure. Fish. Mgmt Ecol.6,401Ð420 (1999).52.Watling, L. & Norse, E. A. Disturbance of the seabed by mobile fishing gear: a comparison to forestclearcutting. Conserv. Biol.12,1180Ð1197 (1998).53.Caddy, J. F. & Garibaldi, L. Apparent changes in the trophic composition of world marine harvests:the perspective from the FAO capture database. Ocean Coastal Mgmt43,615Ð655 (2000).54.Pauly, D. Anecdotes and the shifting baseline syndrome of fisheries. Trends Ecol. Evol.10,430 (1995).55.Weber, P. Facing limits to oceanic fisheries: Part II: The social consequences. Nat. Res. Forum19,39Ð46 (1995).56.Mace, P. M. in Developing and Sustaining World Fisheries Resources(Proc. 2nd World Fish. Congr.) (edsHancock, D. H., Smith, D. C. & Beumer, J.) 1Ð20 (CSIRO Publishing, Collingwood, VIC, 1997).57.Gordon, H. S. The economic theory of a common property resource: the fishery. J. Polit. Econ.62,124Ð142 (1954).58.Munro, G. The optimal management of transboundary renewable resources. Can. J. Econ.12,355Ð376 (1979).59.Sumaila, U. R. Cooperative and non-cooperative exploitation of the Arcto-Norwegian cod stock inthe Barents Sea. Environ. Resource Econ.10,147Ð165 (1997).60.Sumaila, U. R. & Bawumia, M. in Fish Ethics: Justice in the Canadian Fisheries(eds Coward, H.,Ommer, R. & Pitcher, T. J.) 140Ð153. (Institute of Social and Economic Research, MemorialUniversity, St JohnÕs, Newfoundland, 2000).61.Munro, G. R. & Sumaila, U. R. in Fisheries Centre Research Report 9(5)(eds Pitcher, T. J., Sumaila, U.R. & Pauly, D.) 10Ð27 (also available at http://www.fisheries.ubc.ca) (Fisheries Centre, Univ. BritishColumbia, Vancouver, 2001).62.Clark, C. W. Mathematical Bioeconomics: The Optimal Management of Renewable Resources(Wiley,New York, 1990).63.Milazzo, M. World Bank Tech. Pap. No. 406 (World Bank, Washington DC, 1998).64.United Nations. The United Nations Convention on the Law of the Sea(United Nations, New York, 1982).65.Bray, K. FAO Rep. IUU/2000/6, 53 (Food and Agriculture Organization of the United Nations,Rome, 2000).66.Pauly, D. & Christensen, V. Primary production required to sustain global fisheries. Nature374,255Ð257 (1995).67.Vitousek, P. M., Ehrlich, P. R. & Ehrlich, A. H. Human appropriation of the products ofphotosynthesis. BioScience36,368Ð373 (1986).68.Rawski, T. G. & Xiao, W. Roundtable on Chinese economic statistics: introduction. China Econ. Rev.12,298Ð302 (2001).69.Feigon, L. A harbinger of the problems confronting ChinaÕs economy and environment: the greatChinese shrimp disaster of 1993. J. Contemp. China9,323Ð332 (2000).70.Pullin, R. S. V., Rosenthal, H. & Maclean, J. L. (eds) Environment and Aquaculture in DevelopingCountries(ICLARM, Manila, Philippines, 1993).71.Naylor, R. L. et al.Effect of aquaculture on world fish supplies. Nature405,1017Ð1024 (2000).72.Sainsbury, K. J., Punt, A. E. & Smith, A. D. M. Design of operational management strategies forachieving fishery ecosystem objectives. Ices J. Mar. Sci.57,731Ð741 (2000).73.Cochrane, K. L. Reconciling sustainability, economic efficiency and equity in fisheries: the one thatgot away? Fish Fish.1,3Ð21 (2000).74.Edeson, W. R. Current legal development. The Code of Conduct for Responsible Fisheries: anintroduction. Int. J. Mar. Coast. Law11,233Ð238 (1996).75.Roberts, C. M., Bohnsack, J. A., Gell, F., Hawkins, J. P. & Goodridge, R. Effects of marine reserves onadjacent fisheries. Science294,1920Ð1923 (2001).76.Mosquera, I., C™tŽ, I. M., Jennings, S. & Reynolds, J. D. Conservation benefits of marine reserves forfish populations. Anim. Conserv.3,321Ð332 (2000).77.Murawski, S. A., Brown, R., Lai, H. L., Rago, P. R. & Hendrickson, L. Large-scale closed areas as afishery management tool in temperate marine systems: the Georges Bank experience. Bull. Mar. Sci.66,775Ð798 (2000).78.GuŽnette, S., Pitcher, T. J. & Walters, C. J. The potential of marine reserves for the management ofnorthern cod in Newfoundland. Bull. Mar. Sci.66,831Ð852 (2000).79.Lipcius, R. N., Seitz, R. D., Goldsborough, W. J., Montane, M. M. & Stockhausen, W. T. in SpatialProcesses and Management of Marine Populations(ed. Witherell, D.) 643Ð666 (Alaska Sea GrantCollege Program, Anchorage, 2001).80.Lindholm, J. B. I., Auster, P. J. & Kaufman, L. Habitat-mediated survivorship of 0-year Atlantic cod(Gadus morhua). Mar. Ecol. Prog. Ser.180,247Ð255 (1999).81.Tupper, M. K. et al.Marine reserves and fisheries management. Science295,1233Ð1235 (2002).82.Sumaila, U. R., Pitcher, T. J., Haggan, N. & Jones, R. in Microbehavior and Macroresults: Proc. 10thBiannual Conf. Int. Inst. Fish. Econ. Trade, Corvallis, Oregon, 10-14 July 2000(eds Johnston, R. S. &Shriver, A. L.) (International Institute of Fisheries Economics & Trade, Corvallis, OR, 2001).83.Alverson, D. L., Freeberg, M. H., Murawski, S. A. & Pope, J. G. FAO Fish. Tech. Pap. 339, 233 (Foodand Agriculture Organization of the United Nations, Rome, 1994).84.Forrest, R., Pitcher, T., Watson, R., Valty«sson, H. & GuŽnette, S. in Fisheries Impacts on North AtlanticEcosystems: Evaluations and Policy Exploration(ed. Pauly, D.) Fisheries Centre Rep. 5, 81Ð93 (alsoavailable at http://www.fisheries.ubc.ca) (Fisheries Centre, Univ. British Columbia, Vancouver, 2001).85.Pauly, D., Tyedmers, P., Froese, R. & Liu, L. Y. Fishing down and farming up the food web. Conserv.Biol. Pract.2,25 (2001).86.Dayton, P. K. EcologyÑreversal of the burden of proof in fisheries management. Science279,821Ð822 (1998).87.Walters, C. J. & Maguire, J. J. Lessons for stock assessment from the Northern Cod collapse. Rev. FishBiol. Fish.6,125Ð137 (1996).88.Perry, R. I., Walters, C. J. & Boutillier, J. A. A framework for providing scientific advice for themanagement of new and developing invertebrate fisheries. Rev. Fish Biol. Fish.9,125Ð150 (1999).89.Walters, C. & Kitchell, J. F. Cultivation/depensation effects on juvenile survival and recruitment:implications for the theory of fishing. Can. J. Fish. Aquat. Sci.58,39Ð50 (2001).90.Scheffer, M., Carpenter, S., Foley, J. A., Folke, C. & Walker, B. Catastrophic shifts in ecosystems.Nature413,591Ð596 (2001).91.Walters, C., Pauly, D., Christensen, V. & Kitchell, J. F. Representing density dependent consequencesof life history strategies in aquatic ecosystems: EcoSim II. Ecosystems3,70Ð83 (2000).92.Spalding, M. D., Ravilious, C. & Green, E. P. World Atlas of Coral Reefs(Univ. California Press,Berkeley, 2001).93.Roberts, C. M. et al.Marine biodiversity hotspots and conservation priorities for tropical reefs.Science295,1280Ð1284 (2002).94.Lewis, J. B. Processes of organic production on coral reefs. Biol. Rev. Cambr. Phil. Soc.52,305Ð347 (1977).95.Choat, J. H. & Robertson, D. R. in Coral Reef Fishes: Dynamics and Diversity in a Complex Ecosystem(ed. Sale, P. F.) 57Ð80 (Academic, San Diego, 2002).96.Russ, G. R. in The Ecology of Fishes on Coral Reefs(ed. Sale, P. F.) 601Ð635 (Academic, San Diego, 1991).97.Spalding, M. D. & Grenfell, A. M. New estimates of global and regional coral reef areas. Coral Reefs16, 225Ð230 (1997).98.Russ, G. R. in Coral Reef Fishes: Dynamics and Diversity in a Complex Ecosystem(ed. Sale, P. F.)421Ð443 (Academic, San Diego, 2002).99.Sadovy, Y. J. & Vincent, A. C. J. in Coral Reef Fishes: Dynamics and Diversity in a Complex Ecosystem(ed. Sale, P. F.) 391Ð420 (Academic, San Diego, 2002).100.Zeller, D. C. Home range and activity patterns of the coral trout Plectropomus leopardus(Serranidae). Mar. Ecol. Prog. Ser.154, 65Ð77 (1997).AcknowledgementsThe work forms part of the Sea Around Us project at the Fisheries Centre, University ofBritish Columbia, an activity initiated and funded by the Pew Charitable Trusts. D.P. andC.W. also acknowledge support from the Canadian National Scientific and EngineeringResearch Council. We thank G. Russ and M. L. ÔDengÕ Palomares for various suggestionsthat improved an earlier draft of our paper.© 2002 Nature PublishingGroup

Place this order or similar order and get an amazing discount. USE Discount code “GET20” for 20% discount