P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 - PDF document

P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 Annu.Rev.Ecol.Syst.2000.31:425Ð39Copyright2000byAnnualReviews.AllrightsreservedCOLOGICALESILIENCEHEORYPPLICATION LanceH.GundersonDept.ofEnvironmentalStudies,EmoryUniversity,Atlanta,Georgia30322;e-mail:lgunder@emory.eduKeyWordsresilience,stability,stablestates,biodiversity,adaptivemanagementIn1973,C.S.Hollingintroducedthewordresilienceintotheecolog-icalliteratureasawayofhelpingtounderstandthenon-lineardynamicsobservedinecosystems.EcologicalresiliencewasdeÞnedastheamountofdisturbancethatanecosystemcouldwithstandwithoutchangingself-organizedprocessesandstructures(deÞnedasalternativestablestates).Otherauthorsconsiderresilienceasareturntimetoastablestatefollowingaperturbation.Anewterm,adaptivecapacity,isintroducedtodescribetheprocessesthatmodifyecologicalresilience.TwodeÞnitionsrecognizethepresenceofmultiplestablestates(orstabilitydomains),andhenceresilienceisthepropertythatmediatestransitionamongthesestates.Transitionsamongstablestateshavebeendescribedformanyecosystems,includingsemi-aridrangelands,lakes,coralreefs,andforests.Inthesesystems,ecologicalresilienceismaintainedbykeystonestructuringprocessesacrossanumberofscales,sourcesofrenewalandreformation,andfunctionalbiodiversity.Inpractice,maintainingacapacityforrenewalinadyna-micenvironmentprovidesanecologicalbufferthatprotectsthesystemfromthefail-ureofmanagementactionsthataretakenbaseduponincompleteunderstanding,anditallowsmanagerstoaffordablylearnandchange.INTRODUCTIONIthasbeenalmostthreedecadessincethetermresiliencewasintroducedtotheliteraturebythetheoreticalecologistC.S.Holling(22).Sincethattime,multiplemeaningsoftheconcepthaveappeared(15,40).Sincemostmanagementactionsarebaseduponsometypeoftheory,thesemultiplemeaningsofresiliencecanleadtoverydifferentsetsofpoliciesandactions.Thisreviewisdividedintothreeparts.TheÞrstsectionreviewsconceptsandmultiplemeaningsofresilienceastheyhaveappearedintheliterature.ThatsectionreviewsexamplesofmodelingandÞeldexperimentsthatenrichourunderstandingofecologicalchange.Thesecondsectionincludesanassess-mentofhowresilienceisrelatedtootherkeyecosystemproperties.Thereview P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 426GUNDERSON concludeswithasectiononhowecologicalresilienceiskeytomanagementofcomplexsystemsofpeopleandnature.Resilience,StabilityandAdaptiveCapacityResilienceofasystemhasbeendeÞnedintheecologicalliteratureintwodifferentways,eachreßectingdifferentaspectsofstability.Holling(22)ÞrstemphasizedthesedifferentaspectsofstabilitytodrawattentiontothedistinctionsbetweenefÞ-ciencyandpersistence,betweenconstancyandchange,andbetweenpredictabilityandunpredictability.Holling(22)characterizedstabilityaspersistenceofasys-temnearorclosetoanequilibriumstate.Bycontrast,resiliencewasintroducedtoindicatebehaviorofdynamicsystemsfarfromequilibrium,bydeÞningresilienceastheamountofdisturbancethatasystemcanabsorbwithoutchangingstate.Themultiplemeaningsofresiliencearerelatedtoassumptionsaboutthepresenceofeithersingleormultipleequilibriainasystem(26),asdescribedinthefollowingResilienceandGlobalEquilibriumManyauthorsdeÞnethetermresilienceasthetimerequiredforasystemtoreturntoanequilibriumorsteady-statefollowingaperturbation(28,39,40,44,51).ImplicitinthisdeÞnitionisthatthesystemexistsnearasingleorglobalequilibriumcondition.Hencethemeasureofresilienceishowfarthesystemhasmovedfromthatequilibrium(intime)andhowquicklyitreturns(35).Otherauthors(22,26,35)considerreturntimesasameasureofstability.Holling(26)describedthereturntimedeÞnitionofresilienceasÔengineeringresilience.ÕThereturntimedeÞnitionarisesfromtraditionsofengineering,wherethemotiveistodesignsystemswithasingleoperatingobjective(9,41,54).Ononehand,thatmakesthemathematicstractable,andontheother,itaccommodatesanengineerÕsgoaltodevelopoptimaldesigns.ThereisanimplicitassumptionofglobalstabilityÑi.e.thereisonlyoneequilibriumorsteadystateor,ifotheroperatingstatesexist,theyshouldbeavoidedbyapplyingsafeguards.OtherÞeldsthatusethetermresilience,suchasphysics,controlsystemdesign,ormaterialengineering,allusethisdeÞnition.ResilienceandMultipleEquilibriumThesecondtypeofresilienceemphasizesconditionsfarfromanysteadystatecondition,whereinstabilitiescanßipasystemintoanotherregimeofbehaviorÑi.e.toanotherstabilitydomain(22).Inthiscase,resilienceismeasuredbythemagnitudeofdisturbancethatcanbeabsorbedbeforethesystemredeÞnesitsstructurebychangingthevariablesandprocessesthatcontrolbehavior.Thishasbeendubbedecologicalresilienceincontrasttoengineeringresilience(22,57).Onekeydistinctionbetweenthesetwotypesofresilienceliesinassumptionsregardingtheexistenceofmultiplestablestates.Ifitisassumedthatonlyonestablestateexistsorcanbedesignedtosoexist,thentheonlypossibledeÞnitionand P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 RESILIENCE-THEORYANDPRACTICE measuresforresiliencearenearequilibriumonesÑsuchascharacteristicreturntimeasdeÞnedabove.Theconceptofecologicalresiliencepresumestheexistenceofmultiplestabilitydomainsandthetoleranceofthesystemtoperturbationsthatfacilitatetransitionsamongstablestates.Hence,ecologicalresiliencereferstothewidthorlimitofastabilitydomainandisdeÞnedbythemagnitudeofdisturbancethatasystemcanabsorbbeforeitchangesstablestates(22,35).Thepresenceofmultiplestablestatesandtransitionsamongthemhavebeendescribedinarangeofecologicalsystems.Theseincludetransitionsfromgrass-dominatedtowoody-dominatedsemi-aridrangelandsinZimbabwe(60)andAustralia(36,59).Inthesecasesthealternativestatesaredescribedbydomi-nantplantforms,andthedisturbanceisgrazingpressure(59).Otherexamplesincludetransitionsfromclearlakestoturbidones(3,46);alternativestatesareindicatedbydominantassemblagesofprimaryproducersinthewaterorrootedmacrophytesanddisturbancesincludephysicalvariablessuchaslightandtem-perature.Alternativestatesarealsodescribedinpopulationslevelscreatedbyinteractionsamongpopulations(10,11,48,62).Carpenteretal.(3,5)andScheffer(46)haveusedtheheuristicofaballandacuptohighlightdifferencesbetweenthesetypesofresilience.Theballrepresentsthesystemstateandthecuprepresentsthestabilitydomain(Figure1).Anequilibriumexistswhentheballsitsatthebottomofthecupanddisturbancesshakethemarbletoatransientpositionwithinthecup.EngineeringresiliencereferstocharacteristicsoftheshapeofthecupÑtheslopeofthesidesdictatethereturntimeoftheballtothebottom.Ecologicalresiliencesuggeststhatmorethanonecupexists,andresilienceisdeÞnedasthewidthatthetopofthecup.ImplicitinbothofthesedeÞnitionsistheassumptionthatresilienceisastaticpropertyof Figure1Ballandcupheuristicofsystemstability.Valleysrepresentstabilitydomains,ballsrepresentthesystem,andarrowsrepresentdisturbances.Engineeringresilienceisde-terminedbytheslopesinthestabilitylandscapes,whereasecologicalresilienceisdescribedasthewidth.Adaptivecapacityreferstotheabilityofthesystemtoremaininastabilitydomain,astheshapeofthedomainchanges(asshownbythethreeslicesorlandscapes). P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 428GUNDERSON systems.Thatis,oncedeÞned,theshapeofthecupremainsÞxedovertime.Butrecentworkindicatesthatstabilitydomainsaredynamicandvariable.AdaptiveCapacityManyofthemanifestationsofhuman-inducedstatechangesinecosystemsresultfromalterationofthekeyvariablesthatinßuencetheunder-lyingstabilitydomains.ThekeyvariablesthatconÞgurethesestabilitydomainschangeatrelativelyslowrates(withouthumanintervention).Examplesincludenutrientsinwetlandsandlakes(5,46),speciescompositionsinrangelands(56,59)ortrophicrelationships(4).Usingtheballincupheuristic,theshapeofthecupissubjecttochange,alteringbothstability(returntime)andresilience(widthofstabilitydomain).Schefferetal.(46)depictthisasmultiplestabilitylandscapes(threeslicesinFigure1).Thepropertyofanecosystemthatdescribesthischangeinstabilitylandscapesandresilienceisreferredtoasadaptivecapacity(19).EcosystemDynamicsandMultipleStableStatesTheprevioussectionoutlinedacontrastamongthreeviewsofresilience.Alldescribeaspectsofchangeinecosystemsandthedegreeofthatchange.Butmuchoftheliteratureoverthelast30yearshasaddressedwhethermultiplestablestatesexistinecosystems,andifsowhatmediatestransitionamongthem.Thereisagrowingbodyofliteraturethatdocumentstransitionsamongstabilitydomainsinavarietyofecosystems(4,21,35,38,59,60).Manyofthosesystemsareinßuencedbyhumanactivities,whichhasledtoaconfoundingproblemaroundecologicalresilience.Someauthors(49)suggestthatalternativestablestatesdonotexistinsystemsuntouchedbyhumans,whileothers(10,47)indicatethattheseareandhavebeenpartofthedynamicsofsystemswithorwithouthumans.Withouttreadingonthequestionofwhetherpeopleareorarenotnaturalpartsofecosystems,threeexamplesarepresentedsuggestingthatpeopledochangetheresilienceofsystem.Oneinvolveslakesystems,anotherwetlands,andtheothersemi-aridrangeland.Ineachexample,thealternativestatesarediscussed,asarethemechanismsthatresultinthetransitionsandthoseprocessesthatcontributeordetractfromecologicalresilience.ShallowLakesLimnologistshavelongrecognizedtheexistenceofqualitativedifferencesinthestateoflakes.Inshallowlakes,twoalternativestatescanbecharacterizedas(a)clearwaterandrootedmacrophytesor(b)turbidwaterwithplanktonicalgae(45,46).Eachofthesestatesisrelativelystableduetointeractionsamongnutrients,thetypesofvegetation,andlightpenetration(Figure1).Intheclearwaterstate,sediments,andnutrientcyclingarestabilizedbyrootedvegetation(45,46).Theturbidstatepersistswhenwind-drivenmixingresuspendssediments.Thesedimentsandphytoplanktoninthewatercolumndecreaselightpenetration,therebycurtailingestablishmentofbenthicvegetation(45,46).Transitionsbetweenthesetwostatescanbemediatedbytrophicrelationships.DecreasingstocksofplanktivorousÞshcancreateashiftfromaturbidtoaclear P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 RESILIENCE-THEORYANDPRACTICE lake.Aspredationonherbivorouszooplanktondecreases,theirpopulationsin-crease,leadingtoanincreaseinherbivoryandareductioninphytoplanktonbiomass.Increasedlightpenetrationandavailablenutrientsthenleadtoestab-lishmentofrootedvegetation(45).Intheotherdirection,shiftsfromthecleartoturbidstatecanresultfromovergrazingofbenthicvegetationbyÞshorwaterfowl(45).TheshiftbetweenstablestatesishystereticÑthedisturbancesthatinßuencechangeinonedirectiondonothavesimilarimpactsintheoppositedirectionWetlandsNutrientenrichmentinthefreshwatermarshesoftheEvergladescausedthelossofresilience.TheEvergladesisanoligotrophicwetland,limitedprimar-ilybyphosphorus(50).Forthepast5000yearsorso,theecosystemeffectivelyself-organizedaroundthislownutrientstatus,pulsedbyannualwet/drycyclesandbydecadalrecyclingassociatedwithÞres(17).Theresultinglandscapemosaicwascomprisedofsawgrassmarshesandwetprairiesinterspersedwithsmalltreeislands(7,33,50).Inthelate1970sandearly1980s,large-scalevegetationchangeswerenoticedintheregionsdownstreamfromtheEvergladesagriculturalarea.SawgrassmarshesandwetprairieshadbecomedominatedbyasinglespeciesÑcattail(8).Thecon-versionwasattributedtoaslowincreaseintheconcentrationofsoilphosphorus,andadisturbance,suchasÞre,droughtorfreeze.Keyecosystemprocessesandstructuresoccuratvariousspatialandtemporalscales.Thevegetationstructuresrepresentthemostrapidlychangingvariables,withplantturnovertimesontheorderof5to10years(8).Firesoperateonre-turnfrequenciesof10to20y(20,53).Otherdisturbancessuchasfreezesanddroughtsoccuronmultipledecadereturntimes(20,53).Thesoilphosphorusconcentrationsaretheslowestofthevariables,withturnovertimesontheorderofcenturies(8).Theresilienceofthefreshwatermarshesisrelatedtothesoilnutrientcontent.Thealternativestabilitydomainsarecharacterizedbythedominantplantspecies;sawgrassorwetprairiecommunitiesdominateonsiteswithlownutrients,andcattaildominatesonsiteswithhighersoilphosphorusconcentrations.Followingadisturbance,itisthesoilphosphoruslevelthatdetermineswhichofthesespeciesSemi-aridRangelandsSavannarangelandsarefoundinclimaticregimesofhot,rainysummersandmild,drywinters.Thesesystemshavehighproductivityandsupportadiverseassemblageofperennialandannualgrassesandfewwoodyplants.Keybiophysicalprocessesinthesesystemsincludevariationinrainfall,Þres,andgrazing.Walker(59)andLudwigetal.(36)identifyalternativestablestatesaseitherwoody/grasscoverage,orwoodythicket.Thetransitionbetweenthesestatesismediatedbygrazingpressuresthatremoveeitherdrought-tolerantorperennialgrasses(36,59).Ifgrazingpressuresarehigh,theperennialgrassabundanceisdecreased,leadingtoanincreasedabundanceofwoodyplants.Once P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 430GUNDERSON thewoodycommunityisestablished,Þresburnlessfrequently(ifthethicketsburnatall)andthewoodycommunitypersistsfordecades.WalkerandLudwigetal(36,59)suggestthatthewoody/grassassemblagecanbereestablished,eitherthroughcollapseofmaturedwoodyplantsorthroughmanipulationofÞreanddrainageprocesses(36).Theroleoffunctionalbiodiversityincontributingtotheecologicalresilienceisdiscussedinthenextsection.ResilienceinEcosystemsÑPatternsandProcessResilienceisanemergentpropertyofecosystemsandisrelatedtoself-organizedbehaviorofthoseecosystemsovertime.Inthissense,self-organizationistheinteractionbetweenstructureandprocessthatleadstosystemdevelopment,re-gardlessofinitialconditions.Self-organizationalsoimpliesthatforcertainscaleranges,structureandprocessarenoteasilyseparableandinteractinanorganicwaytogenerateemergentpatterns.Theadaptivecycleorfour-phasemodelofHolling(24,25)describeshowpatternsandprocesseschangeovertimeinmanyecosystems,especiallythosewithdisturbanceregimes.Holling(24,25)describesecosystemsuccessionintheÞrsttwophasesofthecycle.Theexploitativephaseischaracterizedbyrapidcolonizationofrecentlydisturbedareas.Thisphasegiveswaytoaconservationphase,asmaterialandenergyareaccumulatedandstored.Theexploitativephaseischaracterizedbyrapidgrowthinanarenaofscramblecompetition,whilethespeciesthatdomi-nateinthelaterphasetendtohaveslowergrowthratesandsurviveinanarenaofexclusivecompetition.Thematureorconservativephaseisfollowedbyaphasewhenadisturbanceinßuencesthestructurethathasaccumulatedinprevi-ousphases.Thisphaseiscalledcreativedestruction.DisturbanceagentssuchasforestÞres,insectpests,orintensepulsesofgrazingsuddenlyreleaseaccumulatedecologicalcapital.Thesystementersthefourthphase,orreorganization.Thesys-tempassesthroughthereorganizationphaseandthenentersanotherexploitativeResilienceisrelatedtothephaseofHollingÕsadaptivecycle.Duringtheex-ploitationphase,ecologicalresilienceishighÑthesystemcanabsorbawiderangeofdisturbances.Whenthesystemisreachingthelimitstoitsconservativegrowth,itbecomesincreasinglybrittleanditsaccumulatedcapitalisreadytofuelrapidstructuralchanges.Thesystemisverystable,butthatstabilityislocalandnarrow.Asmalldisturbancecanpushitoutofthatstabledomainintocatastrophe.Thenatureandtimingofthecollapse-initiatingdisturbancedetermines,withinsomebounds,thefuturetrajectoryofthesystem.Thereforethisbrittlestatepresentstheopportunityforachangeatasmallscaletocascaderapidlythroughtheover-connectedsystem,bringingaboutitsrapidtransformation.Eitherinternalconditionsorexternaleventscaninitiatecollapse,buttypicallyitisinternallyinducedbrittleness(becauseofhighconnectivityandaccumulatedcapital)thatsetstheconditionsforcollapse.ThesystembecomesÒanaccidentwaitingtohappenÓ(24) P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 RESILIENCE-THEORYANDPRACTICE Duringthereorganizationphase,asystembecomesmostvulnerabletochangingstabilitydomains.Thereislittlelocalregulationandstability,sothatthesystemcaneasilybemovedfromonestatetoanother.Resourcesforgrowtharepresent,buttheyaredisconnectedfromtheprocessesthatfacilitateandcontrolgrowth.Insuchaweaklyconnectedstate,randomseedingscangeneratemultipletrajectories,whichthenestablishtheexploitativepathalongwhichthesystemdevelops.Theprevioussystempatternmayreassertitself,orthesystemmayreorganizeitselfintoanovelstructure.ResilienceandBiodiversityTherelationshipbetweenbiologicaldiversityandecologicalstabilityhasbeenanongoingdiscussioninecology(37,51,52).Tilman(51,52)hasdemonstratedthat,overecologicallybriefperiods,anincreaseinspeciesnumberincreasestheefÞciencyandstabilityofsomeecosystemfunctions,butdecreasesthestabilityofthepopulationsofthespecies.Whilethisworkisimportantandinteresting,itfocusesuponhowanecosystembehavesnearsomesteadystate.Theroleofecologicaldiversityoveramuchbroaderrangeofvariationsandespeciallytherelationshipbetweendiversityandresiliencehaveonlybeenrecentlyaddressed(31,43,55,56).WhengrapplingwiththisbroaderrelationshipbetweendiversityandresilienceWalker(58)developedadriverandpassengersanalogy.Walkerproposedthatfunc-tionalgroupsofspeciescanbedividedintoÔdriversÕandÔpassengers.ÕDriversarekeystonespeciesthatcontrolthefutureofanecosystem,whilethepassen-gersliveinbutdonotaltersigniÞcantlythisecosystem.However,asconditionschange,endogenouslyorexogenously,speciesshiftroles.Inthismodel,removingpassengershaslittleeffect,butremovingdriverscanhavealargeimpact.Eco-logicalresilienceresidesbothinthediversityofthedrivers,andinthenumberofpassengerswhoarepotentialdrivers.Walkerhasmorerecentlyshownhowthediversityoffunctionalgroupsalsomaintainstheresilienceofecosystemstructureandfunction(56).Suchdiversityprovidesrobustnesstoecosystemfunctionsandresiliencetothesystembehavior.Moreover,thisseemsthewaymanybiologicalprocessesareregulatedÑoverlappinginßuencesbymultipleprocesses,eachoneofwhichisinefÞcientinitsindividualeffectbuttogetheroperatinginarobustmanner.Recentmodelsindicatethatbiodiversityprovidesacross-scaleresilience(43).SpeciescombinetoformanoverlappingsetofreinforcinginßuencesthathelpspreadrisksandbeneÞtswidelytoretainoverallconsistencyinperformanceinde-pendentofwideßuctuationsintheindividualspecies.Becauseoftherobustnessofthisredundancywithinfunctionalgroups,andthenon-linearwaybehaviorsuddenlyßipsfromonepatterntoanotherandonesetofcontrolstoanother,graduallossofspeciesinvolvedincontrollingstructureinitiallywouldhavelittleperceivedeffectoverawiderangeoflossofspecies.Aslossofthosespeciescon-tinued,differentbehaviorwouldemergemoreandmorefrequentlyinmoreand P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 432GUNDERSON moreplaces.Totheobserver,itwouldappearasifonlythefewremainingspecieswerecriticalwheninfactalladdtotheresilience.Althoughbehaviorwouldchangesuddenly,resiliencemeasuredasthesizeofstabilitydomains(22)wouldgraduallycontract.Thesystem,ingraduallylosingresilience,wouldbecomeincreasinglyvulnerabletoperturbationsthatearliercouldbeabsorbedwithoutchangeinfunction,pattern,andcontrols.Ashumansstruggletomanagethesekeyecosystemattributesoffunctionandpattern,thereisagrowingrecognitionoftheimportanceofecologicalresilience.Ecologicalresilienceisrecognizedasthepropertythatallowsformanagersandotheractorstolearnfromandadapttotheunpredictabilityinherentintheseecosys-tems.ThisideaisdevelopedintheÞnalsectioninwhichtheinteractionbetweenecosystemmanagementandresilienceisdiscussed.ManagingforResilienceinPolicyandPracticeAgrowingnumberofcasehistoriesoflarge-scale,bureaucraticresourcesystems(16,29)andtraditionalmanagementsystems(2)demonstratepatternsofsurprise,crises,andreformation.Shiftsbetweenalternativeconditionsareusuallysignaledasaresourcecrisis.Thatis,acrisisoccurswhenanecosystembehavesinasurprisingmannerorwhenobservationsofasystemarequalitativelydifferentfromexpectationsofthatsystem.Suchsurprisesoccurwhenvariationinbroadscaleprocesses(suchasahurricaneorextremedrought),intersectswithinternalchangesinanecosystemduetohumanalteration.Examplesofhumaninducedshiftsincludewoodyinvasionofsemi-aridrangelands(59,60)oralgaebloomsinfreshwaterlakes(15)asdescribedearlier.Theseshiftsinstabilitydomainsarechronicledasresourcecrisis.Understandinghowandwhypeoplechosetoreactiskeytomanagingforresilience.Whenfacedwithshiftingstabilitydomainsandresultingcrises,managementoptionsfallintooneofthreegeneralclassesofresponse.TheÞrstistodonothingandwaittoseeifthesystemwillreturntosomeacceptablestate.OneconsequenceofthisoptionisthatthesocialbeneÞtsofthedesiredstateareforegonewhilewaitingtoseeifthesystemwillreturntothedesiredstate.Thesecondoptionistoactivelymanagethesystemandtrytoreturnthesystemtoadesirablestabilitydomain.Thethirdoptionistoadmitthatthesystemisirreversiblychanged,andhencetheonlystrategyistoadapttothenew,alteredsystem.Theecologicalresilienceofthesystemprovidessomemeasureofeaseoftran-sitionamongstatesandisakeyconsiderationregardinghowmanagementactionsshouldorcanbestructured.Thisthemeisdevelopedinthefollowingtwosections,oneonhowbuildingunderstandingprovidesresilienceandthesecondonhowtomaintainorrestoreresilienceinmanagedsystems.Uncertainty,Understanding,andResilienceDuringmostofthe20century,thegoaloftechnologically-basedresourceman-agementhasbeentocontroltheexternalsourcesofvariabilityinordertoseek P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 RESILIENCE-THEORYANDPRACTICE asinglegoal,suchasmaximizationofyield(trees,Þsh)orcontrollinglevelsofpollution.ThisÔcommandandcontrolÕapproachfocusesoncontrollingatargetvariable,butthenslowlychangesotherpartsofthesystem(27).Thatis,isolatingandcontrollingthevariablesofinterest(i.e.assumingthattheuncertaintyofnaturecanbereplacedwiththecertaintyofcontrol)hasresultedinerosionofresilience.Themanifestationofthaterosionisthepatternofpolicycrisisandreformationasmentionedaboveandelsewhere(16).Muchofsubsidizedagriculture,wherein-centivesaresetuptodealwithchangesinmarketsandcosts,aswellasvariabilityfromnature,fallsintothiscategory(6).MuchoftheÔcommandandcontrolÕresourcemanagementthatleadstolossofecologicalresilienceisbaseduponthepresumedpredictabilityofcomplexeco-logicalsystemsanddrivenbythemyththatdisciplinarysciencewillresolvemostuncertaintiesofmanagement.ButtherehasbeenagrowingsensethattraditionalscientiÞcapproachesarenotworking,and,indeed,maketheproblemworse(34).OnereasonwhyrigidscientiÞcandtechnologicalapproachesfailisbecausetheypresumeasystemnearequilibriumandaconstancyofrelationships.Inthiscase,uncertaintiesarisenotfromerrorsintoolsormodelsbutfromlackofappropriateinformationforthemodels.Anotherreasonforfailureisthatfewapproachesaccountforinherentcomplexrelationshipsamongvariablesthatleadtoinherentunpredictabilitiesinecologicalsystems.ScientiÞcdisciplinestendtobreakthemanagementissueintopartsforanalysis,andhavehistoricallygeneratedpiece-mealsetsofpoliciesassolutions(16,61).Yet,recentmodelsbyCarpenterandothers(5)thatintegrateecologic,economic,andsocialdynamicsaroundaßippinglakesystemindicatethatecosystemresiliencemustbecontinuallyprobed.Differentviewsofsciencecancontributetoalossofecologicalresilience.Onemodeofsciencefocusesonpartsofthesystemanddealswithexperimentsthatnarrowuncertaintytothepointofacceptancebypeers;itisconservativeandun-ambiguousbybeingincompleteandfragmentary.Theotherviewisintegrativeandholistic,searchingforsimplestructuresandrelationshipsthatexplainmuchofnatureÕscomplexity.Thisviewprovidestheunderpinningsforanapproachtodealingwithresourceissuescalledadaptivemanagement,whichassumessur-prisesareinevitable,thatknowledgewillalwaysbeincomplete,andthathumaninteractionwithecosystemswillalwaysbeevolving(16,23,30,61).Adaptivemanagementisanintegrated,multidisciplinarymethodfornaturalresourcesmanagement(23,61).Itisadaptivebecauseitacknowledgesthatthenaturalresourcesbeingmanagedwillalwayschange,sohumansmustrespondbyadjustingandconformingassituationschange.Thereisandalwayswillbeuncertaintyandunpredictabilityinmanagedecosystems,bothashumansex-periencenewsituationsandasthesesystemschangebecauseofmanagement.Surprisesareinevitable.Activelearningisthewayinwhichthisuncertaintyiswinnowed.AdaptivemanagementacknowledgesthatpoliciesmustsatisfysocialobjectivesbutalsomustbecontinuallymodiÞedandbeßexibleforadaptationtothesesurprises(16,23,30,61).AdaptivemanagementthereforeviewspoliciesashypothesesÑthatis;mostpoliciesarereallyquestionsmasqueradingasanswers. P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 434GUNDERSON Sincepoliciesarequestions,thenmanagementactionsbecometreatmentsintheexperimentalsense.Theprocessofadaptivemanagementincludeshighlightinguncertainties,developingandevaluatinghypothesesaroundasetofdesiredsystemoutcomes,andstructuringactionstoevaluateorÔtestÕtheseideas(16,23,30,61).Althoughlearningoccursregardlessofthemanagementapproach,adaptiveman-agementisstructuredtomakethatlearningmoreefÞcient.Trialanderrorisadefaultmodelforlearningwhilemanaging;peoplearegoingtolearnandadaptbythesimpleprocessofexperience.JustasthescientiÞcmethodpromotesefÞcientlearningthrougharticulatinghypothesesandtestingthosehypotheses,adaptivemanagementproposesasimilarapproachtoresolvinguncertaintiesofresourceAuniquepropertyofhumansystemsinresponsetouncertaintyisthegener-ationofnovelty.Noveltyiskeytodealingwithsurprisesorcrises.Humansareuniqueinthattheycreatenoveltythattransformsthefutureovermultipledecadestocenturies.Naturalevolutionaryprocessescausethesamemagnitudeoftransfor-mationovertimespansofmillennia.ExamplesarethecreationofnewtypesandarrangementsofmanagementinstitutionsafterresourcecrisesintheEvergladesorColumbiaRiverBasin(30,32).RestorationandMaintenanceofResilienceAtleasttwoaspectsofmanagingforresiliencecanbeidentiÞed;strategiesthatpeopleemployinordertorestoreormaintainecologicalresilienceandpropertiesthatcontributetoresilienceinhumanorganizations.Inordertoaddresiliencetomanagedsystems,atleastthreestrategiesareemployed:increasingthebufferingcapacityofthesystem,managingforprocessesatmultiplescales,andnurtur-ingsourcesofrenewal(18).Mostactivitiesforbufferingtendtoaddresstheengineeringtypeofresilience,thatis,mitigatingtheeffectsofunwantedvaria-tioninthesysteminordertoshortenthereturntimetoadesiredequilibrium.Inmanyagriculturalsystems,resistancetochangeisdealtwithbyacombina-tionofbarrierstooutsideforces(tariffs,fences,etc.)andinternaladjustmentssuchaswaterorcostcontrolmechanisms(6).Waterresourcesystemscanbedesignedforresiliencebyincreasingthebufferingcapacityorrobustnessthroughredundancyofstructures(andßexibilityofoperations)ratherthanfewer,largerstructuresandrigidoperationalschemes(12).Folke,Berkesandcollaborators(1,2,13,14)suggestthattraditionalapproaches(theydeÞneastraditionalecolog-icalknowledge)buffersystemsbyallowingsmallerscaleperturbationstoenterthesystem,therebylesseningtheimpactofunpredictableorlargeperturbations.OnesuchexampleisthattheCreeÞshermanuseamixed-sizemeshnettoharvestmultipleagesclasses,therebypreservinganageclassstructurethatmimicsanat-uralpopulation(1).ThisagestructurehelpsbufferwidelyvaryingreproductiveResourcesystemsthathavebeensustainedoverlongtimeperiodsincreasere-siliencebymanagingprocessesatmultiplescales.ReturningtotheexampleoftheCreeinnorthernCanada,Berkes(1)arguesthatmultiplespatialdomainsarepart P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 RESILIENCE-THEORYANDPRACTICE oftheirÞshingpracticesandmultipletemporaldomainsintheirhuntingpractices.WhileÞshingwithinaseason,theCreemonitorcatchperuniteffort.Whentheynoticetheratedropping,theyimmediatelymovetodifferentareasanddonotÞshthoseareasforanumberofyears.Overlongertimeframes,theyrotateÞshingefforttomoreremotesites(1).Similarly,theyretaininformationthroughbeliefsystems,e.g.thatcaribouwillreturnforhuntingatannualanddecadalcyclesorperiods.Similarly,theEvergladeswatermanagementsystemhaschangedtoman-ageacrossmultipletimeframes.Inthemid1970swaterdeliveriestoEvergladesParkwerebaseduponaseasonallyvariable,butannuallyconstantvolumeofwa-ter.Thissystemwaschangedinthemid1980stoastatisticalformulationthatincorporatedinterannualvariationintothevolumetriccalculation(32).FolkeandBerkes(2)arguethatlocalcommunitiesandinstitutionsco-evolvebytrialanderrorattimescalesintunetothekeysetsofprocessesthatstructureecosystemswithinwhichthegroupsareembedded.ManyofthecriseschronicledbyGundersonetal.(16)werecreatedbyaninherentfocusononescaleforman-agement,andreformationsoflearningrecognizedthemultiplescalesbywhichtheecosystemwasfunctioning.Institutions(deÞnedbroadlyasthesetofrulesandstructuresthatallowpeopletoorganizeforcollectiveaction)canaddresiliencetoasystem.YoungandMcCay(63)arguethataddingßexibilityandrenewablestructuretopropertyrightsregimeswillincreaseecosystemresilience.Theyindi-catethatmarket-basedpropertyrightschemes(licenses,leases,quotasorpermits)shouldincludeterminationschemes,withstablearrangements(entitlements,obli-gations)intheinterimyears.TheseprinciplescomplementOstromÕs(42)Þndingsthatsuccessfulinstitutionsallowstakeholderstoparticipateinchangingrulesthataffectthem.Afewkeyinstitutionalingredientsappearnecessarytofacilitatethemovementofsystemsoutofcrisisthroughareformation.Inthereviewofmanagementhisto-riesinwesternsystems(16)theseincludedfunctionsoflearning,engagementandtrust.KaiLee(30)callsthisÒsociallearningÓ,bycombiningadaptivemanage-mentframeworkswithinaframeworkofcollectivechoice.Otherauthors(2,14)describethisassocialcapital;comprisedoftheinstitutions,traditionalknowledge,andcommonpropertysystemsthatarethemechanismsbywhichpeoplelinktotheirenvironment.SUMMARYANDCONCLUSIONSResilienceinengineeringsystemsisdeÞnedasareturntimetoasingle,globalequilibrium.Resilienceinecologicalsystemsistheamountofdisturbancethatasystemcanabsorbwithoutchangingstabilitydomains.Adaptivecapacityisdescribedassystemrobustnesstochangesinresilience.Inecologicalsystems,resilienceliesintherequisitevarietyoffunctionalgroupsandtheaccumulatedcapitalthatprovidesourcesforrecovery.Resiliencewithinasystemisgeneratedbydestroyingandrenewingsystemsatsmaller,fasterscales.EcologicalresilienceisreestablishedbytheprocessesthatcontributetosystemÔmemoryÕofthoseinvolved P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 436GUNDERSON inregenerationandrenewalthatconnectthatsystemÕspresenttoitspastandittoitsneighbors.Manyhumanactivitiesshrinkecologicalresiliencebyattemptingtocontrolvariabilityinkeyecosystemprocesses.Thislossofresilienceisoftenaccompa-niedbyachangeinsystemstate,signaledasaresourcecrisis.Whenasystemhasshiftedintoanundesirablestabilitydomain,themanagementalternativesaretorestorethesystemtoadesirabledomain,allowthesystemtoreturntoade-sirabledomainbyitself,oradapttothechangedsystembecausechangesareirreversible.Resilienceismaintainedbyfocusingonkeystonestructuringprocessesthatcrossscales,onsourcesofrenewalandreformation,andonmultiplesourcesofcapitalandskills.Nosinglemechanismcanguaranteemaintenanceofresilience.Strategiesthataddressrequisitevarietyofpurposesandconcentrateonrenewalcontributetoresilience.Institutionsshouldfocusonlearning,andunderstandingofkeycross-scaleinteractions.Learning,trustandengagementarekeycomponentsofsocialresilience.Sociallearningisfacilitatedbyrecognitionofuncertainties,monitoringandevaluationbystakeholders.ThemostdifÞcultissuestodealwitharethosewhoseconsequenceswillberealized10to50yearsinthefutureoverbroadscales.ACKNOWLEDGMENTSThisworkwassupportedbyagrantfromtheJohnD.andCatherineT.MacArthurFoundation,andisacontributionoftheResilienceNetwork,aJointProgramoftheUniversityofFloridaandBeijerInternationalInstituteforEcologicalEconomics.Asalways,C.S.Hollingcontinuestoprovideinspirationandnoveltyinclearlyar-ticulatinghisimaginativetheories.BrianWalker,SteveCarpenter,MartinSchefferhavebeenmosthelpfulineducatingmeastothecomplexdynamicsofecosys-tems.ThisworkhasbeneÞtedfromdiscussionswithGarryPetersonandRustyP.Pritchard,Jr.VisittheAnnualReviewshomepageatwww.AnnualReviews.orgLITERATURECITED1.BerkesF.1995.Indigenousknowledgeandresourcemanagementsystems:anativeCanadiancasestudyfromJamesBay.InPropertyRightsinaSocialandEcologi-calContext,ed.SHanna,MMunasinghe,pp.35Ð49.Washington,DC:BeijerInt.Inst.&WorldBank2.BerkesF,FolkeC.eds.1998.LinkingSo-cialandEcologicalSystemsManagementPracticesandSocialMechanismsforBuild-ingResilience.NewYork:CambridgeUniv.Press.459pp.3.CarpenterSR,CottinghamKL.1997.Resi-lienceandrestorationoflakes.Conser-vationEcology.Vol.1.(url:http://www.consecol.org)4.CarpenterSR,KitchellJF,eds.1993.TrophicCascadeinLakes.Cambridge:CambridgeUniv.Press.385pp.5.CarpenterSR,LudwigD,BrockWA.1999. P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 RESILIENCE-THEORYANDPRACTICE Managementofeutrophicationforlakessubjecttopotentiallyirreversiblechange.Ecol.Appl.6.ConwayG.1993.Sustainableagriculture:thetrade-offswithproductivity,stabilityandequitability.InEconomicsandEcol-ogy,NewFrontiersandSustainableDe-,ed.EBBarbier,pp.57Ð68.London:Chapman&Hall7.CraigheadFC,Sr.1971.TheTreesofSouth.Vol.I.TheNaturalEnvironmentsandTheirSuccession.CoralGables,FL:Univ.MiamiPress.212pp.8.DavisSM.1989.Sawgrassandcattailpro-ductioninrelationtonutrientsupplyintheEverglades.InFreshWaterWetlands&Wildlife,9thAnnualSymposium,SavannahRiverEcologyLaboratory,24Ð27March,.Charleston,SC:USDept.Energy9.DeAngelisDL.1980.Energyßow,nutrientcyclingandecosystemresilience.Ecology10.DublinHT,SinclairARE,McGladeJ.1990.ElephantsandÞreascausesofmul-tiplestablestatesintheSerengeti-marawoodlands.J.Anthropol.Ecol.11.EstesJA,DugginsDO.1995.SeaottersandkelpforestsinAlaska:Generalityandvari-ationinacommunityecologicalparadigm.Ecol.Monogr.12.FieringMB.1982.AlternativeindicesofWaterResourcesRes.13.FolkeC,BerkesF.1995.Mechanismsthatlinkpropertyrightstoecologicalsystems.PropertyRightsandtheEnvironmented.SHanna,MMunasinghe,pp.50Ð62.Washington,DC:BeijerInt.Inst.&World14.FolkeC,HollingCS,PerringsC.1996.Biologicaldiversity,ecosystems,andthehumanscale.Ecol.Appl.15.GrimmV,WisselC.1997.Babel,ortheecologicalstabilitydiscussionsÐaninven-toryandanalysisofterminologyandaguideforavoidingconfusion.Oecologia16.GundersonLH,HollingC,LightS,eds.Barriers&BridgesfortheRenewalofEcosystemsandInstitutions.NewYork:ColumbiaUniv.Press.593pp.17.GundersonLH.1994.Vegetation:deter-minantsofcomposition.InEverglades:TheEcosystemandItsRestoration,ed.SMDavis,JOgden,pp.323Ð340.DelrayBeach,FL:St.Lucie18.GundersonLH.1999.SteppingBack:As-sessingforUnderstandingincomplexregionalsystems.InBioregionalAssess-ments:ScienceattheCrossroadsofMan-agementandPolicy,ed.NKJohnson,FSwanson,MHerring,SGreene,pp.127Ð40.Washington,DC:Island19.GundersonLH,HollingCS,PritchardL,PetersonG.1996.ResilienceinEcosys-tems,InstitutionsandSocieties,Proposi-tionsforaResearchAgenda.Stockholm:BeijerInst.Discuss.Pap.20.GundersonLH,SnyderJR.1994.Firepat-ternsinthesouthernEverglades.InEver-glades:TheEcosystemandItsRestorationed.SMDavis,JOgden,pp.291Ð306.Del-rayBeach,FL:St.Lucie21.HanskiI.1995.Multipleequilibriainmeta-populationdynamics.Nature22.HollingCS.1973.Resilienceandstabil-ityofecologicalsystems.Annu.Rev.Ecol.23.HollingCS,ed.1978.AdaptiveEnvi-ronmentalAssessmentandManagementLondon:Wiley&Sons.377pp.24.HollingCS.1986.Theresilienceofter-restrialecosystems:localsurpriseandglobalchangeInSustainableDevelop-mentoftheBiosphere,ed.WCClark,REMunn,pp.292Ð317.Cambridge:CambridgeUniv.Press.25.HollingCS.1992.Cross-scalemorphol-ogy,geometryanddynamicsofecosys-Ecol.Mongr.26.HollingCS.1996.Engineeringresiliencevs.ecologicalresilience.InWithinEcologicalConstraints,ed.PC P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 438GUNDERSON Schulze,pp.31Ð43.Washington,DC:Natl.27.HollingCS,MeffeGK.1996.Commandandcontrolandthepathologyofnatu-ralresourcemanagement.Conserv.Biol.28.IvesAR.1995.MeasuringresilienceinEcol.Monogr.29.JohnsonKN,SwansonF,HerringM,GreeneS.1999.BioregionalAssessments.Washington,DC:Island.398pp.30.LeeKN.1993.CompassandGyroscopeWashington,DC:Island.243pp.31.LevinS.1995.Biodiversity:interfacingpopulationsandecosystems.InEcologicalPerspectiveofBiodiversity,ed.MHigashi,TAbe,SKevin,Kyoto:KyotoUniv.Press.295pp.32.LightSS,GundersonLH,HollingCS.1995.TheEverglades:evolutionofman-agementinaturbulentecosystem.InBarriersandBridgestotheRenewalofEcosystemsandInstitutions,ed.LHGun-derson,CSHolling,SSLight,pp.103Ð68.NewYork:ColumbiaUniv.Press33.LovelessCM.1959.Astudyofthevege-tationoftheFloridaEverglades.Ecology34.LudwigD,HilbornR,WaltersC.1993.Un-certainty,resourceexploitation,andcon-servation:lessonsfromhistory.35.LudwigD,WalkerB,HollingCS.1996.Sustainability,stabilityandresilience.Conserv.Ecol.36.LudwigJ,TongwayD,FreudenbergerD,NobleJ.1997.LandscapeEcologyFunc-tionandManagementPrinciplesfromAus-traliaÕsRangelands.Collingwood,Aus-tralia:CSIRO.158pp.37.MayRM.1973.StabilityandComplexityinModelEcosystems.Princeton,NJ:Prince-tonUniv.Press.235pp.38.McClanahanTR.1995.Acoralreefecosystem-Þsheriesmodel:impactsofÞshingintensityandcatchselectiononreefstructureandprocesses.Ecol.Model.39.MittelbachG,TurnerA,HallD,RettigJ.1995.PerturbationandresilienceÑalong-term,whole-lakestudyofpredatorextinc-tionandreintroduction.Ecology40.NeubertMG,CaswellH.1997.Alterna-tivestoresilienceformeasuringthere-sponsesofecological-systemstoperturba-Ecology41.OÕNeillRV,DeAngelisDL,WaideJB,AllenTFH.1986.AHierarchicalConceptofEcosystemsMonogr.Pop.Biol.Vol.23,Princeton,NJ:PrincetonUniv.Press.25442.OstromE.1995.Designingcomplexitytogoverncomplexity.InPropertyRightsandtheEnvironment,ed.SHanna,MMunas-inghe,pp.245Ð75.Washington,DC:BeijerIntl.Inst.&WorldBank43.PetersonG,AllenC,HollingCS.1998.Ecologicalresilience,biodiversity,and44.PimmSL.1991.TheBalanceofNature?Chicago:Univ.ChicagoPress.434pp.45.SchefferM.1998.EcologyofShallowLakes.London:Chapman&Hall.357pp.46.SchefferM,HosperSH,MeijerML,MossB.1993.Alternativeequilibriainshallowlakes.TrendsEvol.Ecol.47.SinclairARE,ed.1995.SerengetiII:Dy-namics,Management,andConservationofanEcosystem.Chicago:Univ.ChicagoPress.665pp.48.SinclairARE,OlsenPD,RedheadTD.1990.Canpredatorsregulatesmallmam-malpopulations?EvidencefromhousemouseoutbreaksinAustralia.Oikos49.SousaWP,ConnellJH.1983.Ontheev-idenceneededtojudgeecologicalsta-bilityorpersistence.Am.Nat.50.StewardKK,OrnesWH.1975.Theaute-cologyofsawgrassintheFloridaEver-Ecology P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 RESILIENCE-THEORYANDPRACTICE 51.TilmanD,DowningJA.1994.Biodiver-sityandstabilityingrasslands.Nature52.TilmanD,WedinD,KnopsJ.1996.Pro-ductivityandsustainabilityinßuencedbybiodiversityingrasslandecosystems.ture53.WadeDD,EwelJJ,HofstetterR.1980.FireinSouthFloridaEcosystems.Asheville,NC:SEFor.Exp.Stat.,USDep.Agri.,For.Serv.54.WaideJB,WebsterJR.1976.Engineeringsystemsanalysis:applicabilitytoecosys-tems.InSystemsAnalysisandSimulationinEcology,ed.BCPatton,pp.329Ð71.NewYork:Academic55.WalkerB.1995.Conservingbiologicaldi-versitythroughecosystemresilience.serv.Biol.56.WalkerB,KinzigA,LangridgeJ.1999.Plantattributediversity,resilienceandecosystemfunction:thenatureandsig-niÞcanceofdominantandminorspecies.57.WalkerBH.1981.IssuccessionaviableconceptinAfricansavannaecosystems?InForestSuccession:ConceptsandApplica-,ed.DCWest,HHShugart,DBBotkin,pp.431Ð47.NewYork:Springer-Verlag58.WalkerBH.1992.Biologicaldiversityandecologicalredundancy.Conserv.Biol.59.WalkerBH,LangridgeJL,McfarlaneF.1997.ResilienceofanAustraliansavannagrasslandtoselectiveandnonselectiveper-Aust.J.Ecol.60.WalkerBH,LudwigD,HollingCS,Pe-termanRM.1981.Stabilityofsemi-aridsavannagrazingsystems.J.Ecol.61.WaltersCJ.1986.AdaptiveManagementofRenewableResources.NewYork:Mc-GrawHill.374pp.62.WeaverJ,PaquetPC,RuggieroL.1996.Resilienceandconservationoflargecar-nivoresintheRockymountains.Conserv.63.YoungM,McCayBJ.1995.Buildingequity,stewardshipandresilienceintomarket-basedpropertyrightssystems.InPropertyRightsandtheEnvironment,ed.SHanna,MMunasinghe,pp.210Ð33.Washington,DC:BeijerInt.Inst.&World P1:FXYSeptember15,200020:11AnnualReviewsAR113.18 RESILIENCE-THEORYANDPRACTICE measuresforresiliencearenearequilibriumonesÑsuchascharacteristicreturntimeasdeÞnedabove.Theconceptofecologicalresiliencepresumestheexistenceofmultiplestabilitydomainsandthetoleranceofthesystemtoperturbationsthatfacilitatetransitionsamongstablestates.Hence,ecologicalresiliencereferstothewidthorlimitofastabilitydomainandisdeÞnedbythemagnitudeofdisturbancethatasystemcanabsorbbeforeitchangesstablestates(22,35).Thepresenceofmultiplestablestatesandtransitionsamongthemhavebeendescribedinarangeofecologicalsystems.Theseincludetransitionsfromgrass-dominatedtowoody-dominatedsemi-aridrangelandsinZimbabwe(60)andAustralia(36,59).Inthesecasesthealternativestatesaredescribedbydomi-nantplantforms,andthedisturbanceisgrazingpressure(59).Otherexamplesincludetransitionsfromclearlakestoturbidones(3,46);alternativestatesareindicatedbydominantassemblagesofprimaryproducersinthewaterorrootedmacrophytesanddisturbancesincludephysicalvariablessuchaslightandtem-perature.Alternativestatesarealsodescribedinpopulationslevelscreatedbyinteractionsamongpopulations(10,11,48,62).Carpenteretal.(3,5)andScheffer(46)haveusedtheheuristicofaballandacuptohighlightdifferencesbetweenthesetypesofresilience.Theballrepresentsthesystemstateandthecuprepresentsthestabilitydomain(Figure1).Anequilibriumexistswhentheballsitsatthebottomofthecupanddisturbancesshakethemarbletoatransientpositionwithinthecup.EngineeringresiliencereferstocharacteristicsoftheshapeofthecupÑtheslopeofthesidesdictatethereturntimeoftheballtothebottom.Ecologicalresiliencesuggeststhatmorethanonecupexists,andresilienceisdeÞnedasthewidthatthetopofthecup.ImplicitinbothofthesedeÞnitionsistheassumptionthatresilienceisastaticpropertyof Figure1Ballandcupheuristicofsystemstability.Valleysrepresentstabilitydomains,ballsrepresentthesystem,andarrowsrepresentdisturbances.Engineeringresilienceisde-terminedbytheslopesinthestabilitylandscapes,whereasecologicalresilienceisdescribedasthewidth.Adaptivecapacityreferstotheabilityofthesystemtoremaininastabilitydomain,astheshapeofthedomainchanges(asshownbythethreeslicesorlandscapes).