Evaluating the potential efficacy of invasive lionfish (Pterois volitans) removals

PAGE 1

EvaluatingthePotentialEfficacyofInvasiveLionfish ( Pteroisvolitans )Removals AndrewB.Barbour 1 ,MichaelS.Allen 1 ,ThomasK.Frazer 1 ,KristaD.Sherman 2 1 FisheriesandAquaticSciencesProgram,SchoolofForestResourcesandConservation,TheUniversityofFlorida,Gainesville,Florida,UnitedStat esofAmerica, 2 Parks andScience,TheBahamasNationalTrust,Nassau,TheBahamas Abstract Thelionfish, Pteroisvolitans (Linnaeus)and Pteroismiles (Bennett),invasionoftheWesternAtlanticOcean,CaribbeanSea andGulfofMexicohasthepotentialtoalteraquaticcommunitiesandrepresentsalegitimateecologicalconcern.Several localremovalprogramshavebeeninitiatedtocontrolthisinvasion,butitisnotknownwhetherremovaleffortscan substantiallyreducelionfishnumberstoamelioratetheseconcerns.Weusedanage-structuredpopulationmodelto evaluatethepotentialefficacyoflionfishremovalprogramsandidentifiedcriticaldatagapsforfuturestudies.Weusedhigh andlowestimatesforuncertainparametersincluding:lengthat50%vulnerabilitytoharvest( L vul ),instantaneousnatural mortality( M ),andtheGoodyearcompensationratio( CR ).Themodelpredictedanannualexploitationratebetween35and 65%wouldberequiredtocauserecruitmentoverfishingonlionfishpopulationsforourbaselineparameterestimatesfor M and CR (0.5and15).Lionfishquicklyrecoveredfromhighremovalrates,reaching90%ofunfishedbiomasssixyearsaftera 50-yearsimulatedremovalprogram.Quantifyinglionfishnaturalmortalityandthesize-selectivevulnerabilitytoharvestare themostimportantknowledgegapsforfutureresearch.Wesuggestcompleteeradicationoflionfishthroughfishingis unlikely,andsubstantialreductionofadultabundancewillrequirealong-termcommitmentandmaybefeasibleonlyin small,localizedareaswhereannualexploitationcanbeintenseovermultipleconsecutiveyears. Citation: BarbourAB,AllenMS,FrazerTK,ShermanKD(2011)EvaluatingthePotentialEfficacyofInvasiveLionfish( Pteroisvolitans )Removals.PLoSONE6(5): e19666.doi:10.1371/journal.pone.0019666 Editor: HowardBrowman,InstituteofMarineResearch,Norway Received February25,2011; Accepted April2,2011; Published May10,2011 Copyright: 2011Barbouretal.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermits unrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited. Funding: ThispublicationwassupportedbyagrantfromtheNOAANationalSeaGrantCollegeProgram,andtheNationalScienceFoundationGraduate ResearchFellowshipundergrantNo.DGE-0802270toABB,theNOAAAquaticInvasiveSpeciesProgram,theNOAANationalCentersforCoastalOceanScien ce, andtheNationalMarineFisheriesService'sSoutheastFisheriesScienceCenter.Publicationofthisarticlewasfunded,inpart,bytheUniversityo fFloridaOpenAccessPublishingFund.Thefundershadnoroleinstudydesign,datacollectionandanalysis,decisiontopublish,orpreparationofthemanuscript. CompetingInterests: Theauthorshavedeclaredthatnocompetinginterestsexist. *E-mail:snook@ufl.edu Introduction InvasiveIndo-Pacificlionfishes, Pteroisvolitans (Linnaeus)and Pteroismiles (Bennett),areestablishedintheoffshorewatersofthe SoutheastUnitedStates,Caribbean,andarepresentlyinvading theGulfofMexicoandSouthAmerica[1,2,3,4].Thelionfish invasionisconcerningduetothedangerofhumanhealthrisksby venomouslionfishspinesandbecauseofnumerouspotential ecologicaleffectsonnativehard-bottom,mangrove,seagrass,and coralreefcommunities.Forexample,lionfishhavebeenshownto reducenativefishrecruitmentonexperimentalpatchreefsinThe Bahamas[5]andreductionsinreeffishrecruitmentmaybe exacerbatedbylionfishpredationuponjuvenilenativefishin importantnurseryhabitatssuchasmangrovesandseagrassbeds [6]possiblylimitingthesupplyofeconomicallyimportantreeffish recruits[7]. Throughthesemechanismslionfishmaybecontributingto widespreadregime-shiftsonCaribbeancoralreefsbyconsuming herbivoresresponsibleforcontrollingmacroalgalproduction [5,6,8].Theeffectsofthelionfishinvasionwilllikelycontinueto spread,aslionfishhaveextensivedispersalcapabilities[7,9]and arethoughtonlytobelimitedinrangebytemperaturesbelow 10.0 u C[10].Furthermore,knowninstancesofpredationupon lionfishintheWesternAtlanticarerareandlimitedtoincidental naturaloccurrencesofpredationbysuchspeciesasgroupers (Serranidae)[11]andgreenmorayeels, Gymnothoraxfunebris (Ranzani)(KDSherman,pers.obs.).Moreover,predationon juvenilelionfishbycommonreefpredatoryspeciesinlaboratory trialssuggestslowpredationvulnerability[12].Thissuggests lionfishpopulationslackatop-downcontrolmechanismto regulatetheirpopulationsizeintheirinvasiverange. Asaresultofthisestablishedanddestructiveinvasion,many countrieshaveinstitutedlionfishremovalprograms.These programsincludeinitiativessuchascreatingaspeciallicenseto allowthespearingoflionfishonnearshorereefsandlionfishkill ordersintendedtoinvolvethegeneralpublicinremovalefforts [13].Thelargestinitiativesinvolveusingrecreationaldiversto removelionfishduringderbyevents,andfocusingcommercial diversandfishersonharvestinglionfishasafoodfish[7]. Developinglionfishasacommercialorrecreationalfisheryhas beenproposedasapotentiallong-termsolution[7],butitisnot yetfullyunderstoodwhatlevelofexploitationmightberequiredto controllionfishpopulations. Todate,onlyoneevaluationhasexploredthelevelofharvest requiredtosubstantiallyreducelionfishpopulationsize.Thisstudy [14]utilizedastage-basedmatrixlionfishpopulationmodeland indicatedthatdecreasinglionfishabundancewouldrequire monthlyremovalsof27%oftheadultlionfishpopulation.The studyalsoreportedthatthisrequiredadultexploitationratecould besignificantlyreducedifjuvenileswereremovedfromthe PLoSONE|www.plosone.org1May2011|Volume6|Issue5|e19666

PAGE 2

population[14].Theirmodelwasdensity-independent,whichis appropriateforarecentlyintroducedlionfishpopulation.Thus, themodelassumednocompensationinrecruitmentafterfishing oraftertheparticularpopulationreachedalevelwheredensitydependencewouldoccur,aswouldbeexpectedforamaturing andwell-establishedinvasivespeciespopulation.Lionfishhave nowbeenintheAtlanticbasinforover10yearsandhavereached highdensities( 450fishperhectare)insomelocations[7,15], however,populationassessmentsofabundancearegenerally lacking.Neverthelesslionfishrecruitmentperadultwouldbe expectedtoincreaseasadultabundanceisreducedbyremoval effortsviarecruitmentcompensationasistypicalofestablishedfish populations[e.g.16].Thus,itisessentialthatremovalpractices andpoliciesbeevaluatedforscenarioswhererecruitment compensationoccurs. Inthisstudyweusedanage-structuredpopulationmodelto evaluatethepotentialforremovalprogramstooverfishlionfish populations,whileidentifyingkeydatagapstoguidefuture research.Existinglionfishlifehistoryinformationwascompiledto developthemodel,andvariousharvestrateswereappliedto evaluatetheefficacyofharvestasatop-downcontrolmechanism. Harvestrateswereevaluatedforupperandlowerestimatesof uncertainandunknownparameters.Theresultsofthemodelcan beusedtoinformthebestpossiblemanagementstrategiesunder currentknowledgewhileguidingfutureworktoreducekey uncertainties. Methods ModelDescription Onlysmallnumbersof Pteroismiles havebeendocumentedalong theSoutheastUnitedStates[17]withnocapturestodateinThe Bahamas[18].Lionfisharehereafterreferredtoasinclusiveof both P.miles and P.volitans .Forthepurposeofthismodeling exercise,lifehistoryparameterswerederivedprimarilyfrom P.volitans .Weassumedthatgiventhetaxonomicsimilarity between P.volitans and P.miles (twocloselyrelatedsympatric species),therewouldbenosubstantiallifehistorydifferences[see 19]betweenthetwospeciesthatmightaffecttheoveralloutcome ofthisstudy. Thepopulationmodelstructurewasidenticaltothatpublished previously[20,21],anditpredictedequilibriumrecruitmentand age-specificabundanceunderavarietyofharvestrates.Survival schedulesincorporatednaturalandharvestmortalities.Harvest wasdrivenbyastatedexploitationrateandlength-based vulnerabilitytoremovalefforts.Fecunditywasexpressedasa functionoffishweightandthecollectivefecundityforagivenyear wasreducedbyallmortalitysources.Themodelincludedages1 20andwasconstructedinExcel H EquilibriumrecruitmentwascalculatedusingaBotsford modificationofaBeverton-Holtstockrecruitmentfunction [22,23,24]asdescribedelsewhere[20].Thissimpleformulation predictsequilibriumrecruitmentasafunctionofthefishing mortalityrate.Themodelpredictedtheequilibriumage-1recruits (R eq )ofanexploitedpopulationandissummarizedas[20] R eq ~ R 0 CR { ( W 0 = W f ) CR { 1 1 where R 0 isthenumberofage-1recruitsoftheunfished populationatequilibrium,and CR istheGoodyearcompensation ratio[25].Itisunknownifthecurrentpopulationisnearthe asymptoticunfishedabundance,however,becauselionfishpopulationshavebeenestablishedforovertenyearsintheAtlantic coastalwaters,weinitiatedthesimulationatunfishedequilibrium. Thisissupportedbysimulationrunsinitializedatverylow populationsize,whichreachedequilibriumrecruitmentinfourto sixyearsdependinguponthevaluefor CR .The CR isdefinedas theratiooftherecruitsperspawneratverylowpopulationabundancerelativetotherecruitsperspawnerintheunfishedequilibriumcondition[25].Theparameter R 0 istheunfishedage-1 recruitmentatequilibriumandissimplyascalingparameterthat doesnotinfluencemodelpredictions. Themodelusedsurvivorshipcurvestocalculatethesurvivors perrecruittoeachage.Survivorshiptoage a intheabsenceof fishingwasfoundas l a ~ S a l a { 1 2 where S a istheage-specificfiniteannualnaturalsurvival(i.e., e 2 M ). Oursurvivorshipschedulesinthefishedconditionincorporated naturalmortalityandharvestas l fa ~ l fa { 1 S a (1 { UV a { 1 ), 3 where l fa isthesurvivorshipinthefishedcondition, U isthefinite annualexploitationrate,and V a istheage-specificvulnerabilityto harvest.Wespecifiedtheproportionoffishvulnerabletoharvest as V a ~ 1 1 z e { ( TL { L vul ) SD vul 4 where TL isthemeantotallengthatage a ascalculatedfromthe vonBertalanffygrowthmodel, L vul isthetotallengthat50% vulnerabilitytocapture,and SD vul isthestandarddeviationofthe logisticdistributionfor L vul .Theterm V a modelsincreasing vulnerabilitywithlength,and SD vul specifiesthesteepnessofthe curve.Age-specificabundance( N a )wasestimatedastheproductof thenumberofage-1recruits( R eq )andtheage-specificsurvivorship schedule. Meanfishweight-at-agewasusedasanindexoffecundity(egg production)asfecundityisdirectlyproportionaltoweight-at-age. Theage-specificfecundity( f a )wassettozeroifweight-at-agewas lessthanweight-at-maturity.Toaccountforthecumulativeeffects offishingonthereproductivecapacityofthepopulation,weused theincidencefunctionfortheunfished( W 0 )andfished( W f )egg productionperrecruit[20].Theseincidencefunctionswere calculatedas W o ~ X a f a l a and 5 W f ~ X a f a l fa 6 where f a representsage-specificfecundity,and l a and l fa arethe survivorshipschedulesoftheunfishedandfishedstates.Weused theweightedspawningpotentialratio(SPR)toevaluatetheextent towhichfishingmortalitycanreducereproductiveoutputof lionfish SPR ~ W f R W o R o 7 OverfishingLionfish PLoSONE|www.plosone.org2May2011|Volume6|Issue5|e19666

PAGE 3

where W 0 and W f aredefinedinEqs.(5)and(6),and R isthe recruitmentatequilibriuminthefishedcondition.Theweighted SPRmeasuresthepopulationforagivenleveloffishingmortality relativetotheunfishedcondition,whichisacommonlyused referencetoassessfisheriessustainability[26].Recruitment overfishingisgenerallytermedtooccurwhenSPRisbelowabout 0.4[27].Forthisstudy,wedefinerecruitmentoverfishingas occurringwhenSPR # 0.35. ParameterEstimates Parameterestimatesusedinthemodelsimulationareshownin Table1.Duetoalackofempiricaldataon CR forlionfish,we selectedtwovaluesfor CR ,5and15.Thesevaluesrepresented biologicallyreasonablehighandlowestimatesforspecieswith similarlifehistorytraits(e.g.,relativelyshortlivedpredators), baseduponpastmeta-analyses[16,28]. Ageandgrowthdatatodeterminevaluesfor L and K (Table1) wereobtainedfromlionfishcollectedinoffshorewatersofNorth Carolinabyspearfishing,handnets,hookandline,andtrapping during20042009(dataprovidedbyJ.Potts,NMFS[29]). Collectionsitesrangedbetween2745mdepthandwere characterizedashard-bottomhabitatcomprisedofrockyoutcroppings.Sagittalotolithswereremoved,dried,andembeddedin epoxy.Otolithswereseriallysectionedonalow-speedsaw.The resultingsectionswereadheredtomicroscopeslidesandcovered withaliquidcover-slip. Opaquezoneswereenumeratedforeachfishandwidthofthe marginwasnoted.Opaquezonecountswereconvertedto calendaragesbasedontimingofopaquezonecompletingand dateofcapture.Atotalof774fishwereagedbyasingleperson. Qualityassurancewasassuredusingasecondblind''readerto maintain 95%agreement.AvonBertalanffygrowthcurvewas fittedusingmaximumlikelihoodestimationofthenormal distribution(Figure1),butwefixed t o at 2 0.5becauseoflikely overestimationofmeanlengthatage-0duetogearbias.Thefitted valueof L =425mmisconsistentwithdatareportedpreviously; forexample,apaststudy[30]collected 1,000lionfishwiththe largestfishmeasuring424mmTL;andReefEnvironmental EducationFoundation's(REEF)lionfishderbies,whichhave resultedinthecollectionofn=2,349lionfish,indicatethatthe largestfishmeasured434mmTL. Topredictfishweightfromlength, a and b parameterswere estimatedfromthedatasetof774lionfishcollectedfromthe offshorewatersofNorthCarolina.Usingmaximumlikelihoodof thenormaldistribution,valuesof2.89 6 10 2 5 and2.89were estimatedforthe a and b growthparameters(Figure2).Lionfish sizeat50%maturityhasbeenestimatedat100mmTLformales (n=927)and175mmTLforfemales(n=718)through examinationofgonadaltissue[12].Ageat50%maturitywas specifiedasage-1,whichcorrespondedtoamodel-predictedtotal lengthof159mmandaweightat50%maturity( W mat )of0.07kg (Table1). Instantaneousnaturalmortality( M )isunknownforlionfish.We usedvaluesof0.2and0.5for M (Table1).Avalueof M =0.5is typicalofshort-livedfishandwassimilartothevalueusedinapast study[14].Thevalueof M =0.2istypicaloflonger-livedspecies andwouldbeindicativeofafishwitha1520yearlifespan:in captivity,themaximumlifespanoflionfishhasbeenrecordedas thirtyyears[29].Weincludedthisrangein M duetouncertainty inlionfishmaximumage.Littledataexistforlionfishintheir nativerange,andtheyhavenotbeenpresentintheinvadedrange forenoughtimetoallowestimationofmaximumage,andthus naturalmortality. Thetotallengthat50%vulnerabilitytoremovalefforts( L vul )has notbeenstudied.Weassumedthatsmallfishwouldbelessvisible andlesslikelytobespottedduringremovaleffortsthanlarger animals;alargerfishbiasistypicalofnearlyallfishinggears. Lionfishareeasilyspeared,butdifficulttocatchonhook-and-line. Itisalsodifficulttospearsmalllionfish,meaningmostcaptureof juvenileswillrequirehandnettingorothermethods.Therefore, weevaluatedharvestscenariosundertwopossiblelengthsat50% vulnerability:159mmTL(age-1)and259mmTL(age-2, Table1. Parametersusedinthesimulationmodel. Parameter ValueDataSource NaturalMortality M instantaneousadultnaturalmortality(yr 2 1 )0.2and0.5Inferred FishingMortality U annualharvestexploitationrate0.00to1.0 Vulnerability L vul lengthat50%capturevulnerability(mm)159(age-1)and259(age-2)Inferred SD low standarddeviationof50%capturevulnerability10%of L vul Inferred Growth L asymptoticlength(mm)425Thisstudy K metaboliccoefficient(yr 2 1 )0.47Thisstudy Length-Weight a length-weightcoefficient2.89 6 10 2 5 Thisstudy b length-weightexponent2.89Thisstudy Recruitment R o averageannualunfishedrecruitment100Scalingparameter CR Goodyearcompensationratio5and15Inferred W mat weightatmaturity(kg)0.07EmpiricalData doi:10.1371/journal.pone.0019666.t001 OverfishingLionfish PLoSONE|www.plosone.org3May2011|Volume6|Issue5|e19666

PAGE 4

Table1).Fishrecruitedtoremovaleffortsaccordingtothelogistic functioninthemodel(Eq.4),andfullyvulnerablefishwere harvestedattherateofannualexploitation( U ). SimulationProtocol Toevaluatetheefficacyoflionfishharvestasaremovaltool,we simulatedarangeofexploitationrates( U =0.051.0in0.05step increments).Weappliedtheseexploitationratestoavarietyof scenariosinordertomodelthefullrangeofuncertaintiesinmodel parameters.First, L vul wassetateither159mmTL(age-1)or 259mmTL(age-2).Then,valuesfor M (0.2or0.5)and CR (5or 15)wereselected.Allpossiblecombinationswereevaluated,and theequilibriumoutputmetricwastheannualfiniteexploitation rate( U )requiredtorecruitmentoverfishthestock( U SPR 0.35 ).For eachscenariowecalculatedthenumberofyearstolionfish recoveryfollowingtheremovalofharvest,definedasareturnto 90%oftheunfishedbiomass.Calculatingrecoverytimeis importantasmanyinvasiveremovalprogramsarefundedonly forshortperiodsoftime.Additionally,todeterminethesensitivity ofSPRtouncertaintyinmodelparameters,wemodeledtheeffect ofa10%increaseinindividualparameterestimatesonSPR. Results Themodelindicatedahighdegreeofvariabilityintheannual finiteexploitationrate( U )requiredtocauseoverfishinginlionfish Figure1.LengthatageforlionfishcollectedfromNorthCarolina. ThevonBertalanffygrowthcurveisshownascalculatedbytheequation: TL a ~ 425 : 2(1 { e { 0 : 47( a z 0 : 5) ) doi:10.1371/journal.pone.0019666.g001 Figure2.Lionfishlength-weightrelationshipforlionfishcollectedfromNorthCarolina. Lionfishtotallength(mm,xaxis)andtotalweight (g,yaxis)relationshipandestimatesof a (2.89 6 10 2 5 )and b (2.89)growthparameters.Model predictedvaluescalculatedas: TW ~ (2 : 89 | 10 { 5 ) TL 2 : 89 doi:10.1371/journal.pone.0019666.g002 OverfishingLionfish PLoSONE|www.plosone.org4May2011|Volume6|Issue5|e19666

PAGE 5

populations.Thisvariabilitydependedontheuncertainparameter valuesof L vul M ,and CR (Table2).Oftheunknownparameters, L vul hadthelargestimpactonexploitationratesthatachieved overfishing(Table3).Lower L vul valuesresultedinasubstantially lower U beingrequiredforoverfishingthanhighervaluesof L vul (Table2).Thus,fishinggearsthatselectsmalllionfishwouldbe moreeffectiveatcausingoverfishingthanthoseremovingonly largefish.Instantaneousnaturalmortality( M )hadthesecond largestimpact(Table3),withahigh M of0.5requiring considerablyhigher U than M =0.2.Thisoccursbecausewhen naturalmortalityislow,exploitationstronglychangesabundance. Whennaturalmortalityishigh,fishnaturallydieatahighrate,so exploitationhasadampenedeffectontotalabundance[20]. Therewasconsiderablevariabilityinyearstorecoveryafterthe cessationofexploitation.Lionfishpopulationsrecoveredfairly quickly(67years)when M andrecruitmentcompensation( CR ) werehigh(0.5and15).Recoverytook1525yearswhenlower parameterestimates( M =0.2and CR =5)wereused.Ifthehigher parameterestimatesrepresentrealisticconditionsforlionfishinthe invadedrange,highlevelsofsustainedfishingmortalitywillbe requiredtocauseoverfishing.Giventhattheoldestlionfishagedin thisstudywasage-8,itislikelylionfishareshort-livedpredators andthatourestimatesof M =0.5and CR =15areappropriate. Withtheseparametervaluesandan L vul of159mmSL,anannual exploitationofatleast35%wouldberequiredtocause recruitmentoverfishing.Atan L vul of259mmSL,a65%annual exploitationwouldberequiredtocauserecruitmentoverfishing. Resultsofthesensitivityanalysisshowedthatuncertaintyin L L vul M ,and K hadthegreatesteffectonSPR(Table3).Thisstudy hadestimatesof L and K fromage-growthinformation,butno dataexiston L vul or M .Thevalueof L vul islargelydependenton thetypeofremovaleffortsused.Speargunsareselectiveforlarger lionfish(S.Green,pers.comm.),andatruefisherywouldbebiased towardslargerlionfishduetohighermeatcontent.Thesesize biasesmightbemitigatedwithtargetedhandnettingofsmall lionfish,ascatchesoflionfishwithacombinationofspearingand handnettinghasbeenfoundtoconcurwithvisualestimatesof lionfishsizedistribution(S.Green,pers.comm.).Anothersource ofuncertaintyistheGoodyearcompensationratio( CR ),but sensitivityanalysissuggestsonlyaminorchangeinSPRwitha 10%increasein CR (Table3). Discussion Modelresultssuggestedthatahighlevelofsustainedremoval wouldberequiredtoreducelionfishpopulationsizesbelowthe SPRthresholdofrecruitmentoverfishing.Scalingtheannual exploitationratetoalionfishperhectareremovalfigurebased uponpublisheddataonlionfishdensity[7,15],suggestsayearly removalof157293lionfishperhectarewouldberequiredto causerecruitmentoverfishingforapopulationbasedon M and CR valuesof0.5and15.Thus,thecontroloflionfishpopulations throughtargetedremovaleffortswillbecostly,anderadication throughremovaleffortsishighlyunlikely.Intensiveremovalsare probablyonlyfeasibleatrelativelysmallspatialscaleswherevery highexploitationrates(i.e., 50%)canbeobtained[14]. Resourcemanagersmaybeabletocontroltheinvasionina waythatlimitstheimpactoflionfishonpreyspeciesandprotects ecosystemfunctionality,therebyprotectingbiodiversityand fisheriesatlocalscales.However,beforeanyremovalprogramis implemented,measurablegoalsandtargetexploitationrates shouldbeclearlydefined,andpilotstudiesshouldbeconducted todetermineifthedesiredresultsareattainable. Localandlarge-scalemethodsofexploitinglionfishcurrently exist,butneedfurtherdevelopment.Onalocalscale,lionfish removaleventsintheUnitedStatesandvariouscountriesofthe Caribbeanhavebeenhighlysuccessfulatinvolvingthepublicand generatingawareness,butestimatingtheexploitationratefrom theseeventsisneededtomeasureefficacy.Onalarge-scale,the creationofafisherywithahighexploitationratemayproduce sustainableandmeasurableresults,buttheinfrastructureand demandforsuchafisherydoesnotcurrentlyexist.However,the efficacyoffisheryremovalswouldbedependentonthesizeat vulnerability.Furthermore,suchalionfishfisherywouldbelimited toshallowwater( 30m)spearfishingandhandnettingaslionfish havealowvulnerabilitytocapturebyhookandline[7].Thisgear anddepthlimitationprovidespotentialrefugiafromfishing, potentiallymakingremovaleffortslesseffective.Lionfisharebeing capturedregularlyasbycatchinreeffishtrapfisheries[7],but feasibilityofalionfishspecifictrapcapableofremovinghigh densitiesoflionfishwithouthighbycatchofnativespeciesis questionable. Thisstudyrevealedkeyknowledgegapsthatshouldguidefuture datacollection.Changesinasymptoticlength( L )andlengthat 50%vulnerabilitytoharvest( L vul )causedthegreatestchangein SPR.Weobtaineddatafor K and L fromNorthCarolina,but Table2. Modelresultsforallcombinationsofpossible L vul M and CR parametervalues. L vul (mm)MCR U SPR 0.35 Recovery(yrs)after U SPR Fishing 1590.5150.356 1590.550.310 1590.2150.2012 1590.250.1516 2590.5150.656 2590.550.509 2590.2150.2511 2590.250.216 Modeloutputsinclude:(1) U SPR ,definedasthefiniteannualexploitationrate( U ) requiredtoreduceSPRtoorbelow0.35;and(2)recovery(inyears)after U SPR doi:10.1371/journal.pone.0019666.t002 Table3. Modelsensitivitytoincreasinggivenparametersby 10%. ParameterSPR%Change M 8% L vul 11% SD vul 0.0% CR 1% K 2 6% a 0% b 2 3% L 2 12% W mat 0.0% Sensitivityanalysisperformedwithstartingvaluesof L vul =159; M =0.5; CR =15; and U =0.35(Table2). doi:10.1371/journal.pone.0019666.t003 OverfishingLionfish PLoSONE|www.plosone.org5May2011|Volume6|Issue5|e19666

PAGE 6

growthparametersforlionfishcoulddifferatmoresouthern latitudes,andthus,moreageandgrowthinformationisneeded throughouttheSouthAtlanticandbroaderCaribbeanregion.No dataexistfor L vul ,andourvalueswerebasedonalogical frameworkforfishesofthissize,aswellaspersonalexperiencesin capturinglionfish.Taggingstudiesshouldbeconductedto evaluatethevulnerabilityoflionfishtovariousfishingand collectionefforts,anditislikelythatvulnerabilitywilldifferby region(e.g.NorthCarolinaversusTheBahamas)aswellashabitat type(e.g.mangrovesversusreefversusartificialstructure).Our modelclearlyshowedthatremovaleffortsshouldfocuson methodstocollectsmalllionfish,whichisinagreementwith othermodels[14]. Ourmodelresultswerealsosensitivetochangesinnatural mortality( M )andtheGoodyearcompensationratio.Nodataexist onlionfishnaturalmortality;therefore,uncertaintysurrounding the M parameterestimateishigh.Thisstudyfollowedpast examples[14]inusingthegeneralliteratureon M tochoosea valueforthemodel.Gatheringdataon M isaclearneedand researchpriority.Estimatesof M couldbeobtainedwithtagging studiesorfromagecompositiondata(i.e.catchcurves)inareas wherelionfisharefullyestablishedbutremovaleffortshavenot occurred(i.e.totalmortality=naturalmortality).Additionally, thereisaneedfordataontheGoodyearcompensationratio( CR ). Althoughthisparameterdidnotaffectthemodelpredictionsas stronglyastheparametersM,L ,and L vul ,nolionfish CR data exist,introducingfurtheruncertaintyinthemodelresults. Themodelwasbasedonasimplifiedviewoflionfishlifehistory, whichincreasedsimulationuncertainty.Lionfishareabletospawn almostcontinuously[7,12,19],andexploitedpopulationsmay receiverecruitsfromdistantsourcepopulationsduetolonglarval duration[7,9,31].Oursimulationdidnotincludecontributionsof larvalsupplyfromareasoutsideofthelocalpopulationtargetedby removalefforts;therefore,populationrecoverycouldoccurmore rapidly.Thesource-sinkdynamicsofthelionfishinvasionneedto bebetterunderstood,asrecruitmentoverfishingwillnotbe possibleifrecruitscomefromsourcepopulationsthatarenot fished.Inthecurrentmodel,thisoccurswhensmalllionfishwith lowvulnerabilitytoharvestareabletospawnbeforecapture. Additionally,thiscouldoccurwithalackofremovalsfromlarge spatialareas,orifsourcepopulationsexistinareasthatare inaccessibletoremovaltechniques(e.g.lionfishlivingatdepths 100m:M.Lesser,pers.comm.). Colonizationfromdistantsourceshasbeendemonstratedbythe successofthelionfishinvasiontodate[2].Recolonizationby removedinvasivespeciesistypicallyrapidandlikelylinkedtotheir reproductivesuccess[32],resultingincostlylong-termcontrol programs,suchaswiththemelaleucatree, Melaleucaquinquenervia (Cavanilles)[33],andzebramussel, Dreissenapolymorpha (Pallas) [34].Conductingconcurrentremovalprogramsinallinvaded areaswouldmitigatethiseffect,butwouldrequirelarge investmentsandmaybeinfeasibleduetotheexpansiveand highlyconnectedinvadedrange. Becauseofthedifficultyofsubstantiallyreducinglong-term lionfishabundancethroughremovals,theeffectsandgoalsof removalprogramsshouldbedeterminedbeforeproceeding.It cannotbeassumedthatanyleveloflionfishremovalwillbe beneficialtonativeaquaticcommunities.Forexample,nobenefits tonativefisheswerefoundafteradecadeofnorthernpike Esox lucius andchannelcatfish Ictaluruspunctatus removalsinthe ColoradoRiverandaninvestmentofseveralmilliondollars [32].Additionally,adultremovalsmaycauseashifttosmaller, morenumerousinvasivepredatorswiththeabilitytoconsume nativefishatearlierlifestages[32,35].Ifthisshiftweretooccurin lionfishoncoralreefs,post-settlementmortalityofnativefish wouldlikelyincrease,potentiallydrivingfutureabundancesdown duetothepopulationstructuringeffectofpost-settlement mortality[36,37,38].Studiesonthebiologyandecologyof lionfish,interactionsintheinvadedcommunity,andtheefficacyof removaleffortsmustbeconductedbeforecommittingresourcesto potentiallyineffectiveremovalprograms. Areasonableandmeasurablegoalforlionfishremovaleffortsis toincreasegrowthand/orabundanceofnativepopulationsthat havebeennegativelyimpactedbytheinvasion.Lionfishare knowntoconsumeandcompeteforfood[30]andpossiblyshelter [6]withnativefishes.Althoughnoresearchontheeffectoflionfish onnativefishgrowthrateshasbeenpublished,itislikelythatthe presenceoflionfishreducespopulationand/orindividualgrowth fornativefishes.Onestudydocumentedincreasedpopulation growthinendangeredCaliforniaclapperrail Ralluslongirostris obsoletus followinginvasiveredfox Vulpesvulpes removal,suggesting thatremovaleffortscanbeeffectiveatreversinganegative populationgrowthratetrajectory[39].Lionfishremovalefforts couldpotentiallyimproveabundanceofnativefaunabyreduced predationandcompetition. Thecomplexityoftheecosystemsinvadedbylionfishmustalso beconsideredbeforeenactingremovalprograms.Ontogenetic habitatshiftsbynativereeffishleadtotheuseofmultipleessential habitattypes[40,41,42]andmakeprotectionofecosystem functionalityanimportantgoal.Manyreeffishspeciesuse seagrassandmangroveasjuvenilehabitat[43,44,45].Lionfishin ajuvenilenurserymayreducetherecruitmentpoolavailableto colonizereefsthroughpredationorcompetition[6]actingin concertwithlionfishpredationoncoralreefs[5,30]tofurther stressreeffishpopulations.Additionally,lionfishmaydifferentially usehabitatsthroughouttheirontogeny.Lionfishinmangrove habitat,forexample,maybesmallerthaninreefhabitat[6], suggestingmangrovesmayfunctionaslionfishnurseries.Iftrue, targetinglionfishinmangrovehabitatwouldfocusremovalsonthe importantjuvenilestage,whilealsoreducingpredationstresson nativesusingthehabitatasanursery.Thisstudycouldbeusedas aguidetodeveloptargetsforsuchcontroleffortsbeforeagencies investinremovalprograms. Acknowledgements WethankK.Shertzer,E.Williams,andP.Marraroandtwoanonymous reviewersformanuscriptreviewandtheCapeEleutheraInstituteand DiscoveryDivingforuseofboatsandequipment,andP.Whitfieldandthe manyNOAAdiversthatassistedwithlionfishcollections.Theviews expressedarethoseoftheauthorsanddonotnecessarilyreflecttheviewof theseindividualsororganizations. AuthorContributions Conceivedanddesignedtheexperiments:ABBMSL.Performedthe experiments:ABBMSL.Analyzedthedata:ABBMSLTKFKDS. Contributedreagents/materials/analysistools:ABBMSL.Wrotethe paper:ABBMSLTKFKDS. References 1.WhitfieldPE,GardnerT,VivesSP,GilliganMR,CourtenayWR,Jr.,etal. (2002)BiologicalinvasionoftheIndo-Pacificlionfish Pteroisvolitans alongthe AtlanticCoastofNorthAmerica.MarEcolProgSer235:289297. 2.SchofieldPJ(2009)Geographicextentandchronologyoftheinvasionofnonnativelionfish( Pteroisvolitans [Linnaeus1758]and P.miles [Bennett1828])inthe WesternNorthAtlanticandCaribbeanSea.AqInv4:473479. OverfishingLionfish PLoSONE|www.plosone.org6May2011|Volume6|Issue5|e19666

PAGE 7

3.Aquilar-PereraA,Tuz-SulubA(2010)Non-native,invasiveRedlionfish( Pterois volitans [Linnaeus,1758]:Scorpaenidae]),isfirstrecordedinthesouthernGulfof Mexico,offthenorthernYucatanPeninsula,Mexico.AqInv5(Suppl.1):912. 4.SchofieldPJ,MorrisJA,Jr.,LangstonJN,FullerPL(2010) Pteroisvolitans/miles U.S.GeologicalSurveyNonindigenousAquaticSpeciesDataBase,Gainesville, FL.http://nas.er.usgs.gov/queries/FactSheet.asp?speciesID=963.Accessed24 October2011. 5.AlbinsMA,HixonMA(2008)InvasiveIndo-Pacificlionfish( Pteroisvolitans ) reducerecruitmentofAtlanticcoral-reeffishes.MarEcolProgSer367: 233238. 6.BarbourAB,MontgomeryML,AdamsonAL,D az-FergusonE,SillimanBR (2010)Mangroveusebytheinvasivelionfish Pteroisvolitans .MarEcolProgSer 401:291294. 7.MorrisJA,Jr.,WhitfieldPE(2009) Biology,ecology,controlandmanagementofthe invasiveIndo-Pacificlionfish:anupdatedintegratedassessment .NOAATechnical MemorandumNOSNCCOS99.57p. 8.HughesTP(1994)Catastrophes,phaseshifts,andlarge-scaledegradationofa Caribbeancoralreef.Science265:15471551. 9.AhrenholzD,MorrisJA,Jr.(2010)Larvaldurationofthelionfish, Pteroisvolitans alongtheBahamianArchipelago.EnvironBioFish88:305309. 10.KimballME,MillerJM,WhitfieldPE,HareJA(2004)Thermaltoleranceand potentialdistributionofinvasivelionfish( Pteroisvolitans / miles complex)ontheeast coastoftheUnitedStates.MarEcolProgSer283:269278. 11.Maljkovic A,VanLeeuwenTE,CoveSN(2008)Predationontheinvasivered lionfish, Pteroisvolitans (Pisces:Scorpanidae),bynativegroupersintheBahamas. CoralReefs27:501. 12.MorrisJA,Jr.(2009) ThebiologyandecologyofinvasiveIndo-Pacificlionfish .PhD Thesis,NorthCarolinaStateUniversity,Raleigh,NC.168p. 13.MorrisJA,Jr.,AkinsJL,BarseA,CerinoD,FreshwaterDW,etal.(2009) Biologyandecologyoftheinvasivelionfishes, Pteroismiles and Pteroisvolitans Proceedingsofthe61 st GulfCaribbeanFisheriesInstitute29:409414. 14.MorrisJA,Jr.,ShertzerKW,RiceJA(2011)Astage-basedmatrixpopulation modelofinvasivelionfishwithimplicationsforcontrol.BiologicalInvasions13: 712. 15.GreenSJ,Co te IM(2009)RecorddensitiesofIndo-PacificlionfishonBahamian coralReefs.CoralReefs28:107. 16.MyersRA,BowenKG,BarrowmanNJ(1999)Maximumreproductiverateof fishatlowpopulationsize.CanJFishAqSci56:24042419. 17.HamnerRM,FreshwaterDW,WhitfieldPE(2007)Mitochondrialcytochrome banalysisrevealstwoinvasivelionfishspecieswithstrongfoundereffectsinthe westernAtlantic.JFishBio71:214222. 18.FreshwaterDW,HinesA,ParhamS,WilburA,SabaounM,etal.(2009) MitochondrialcontrolregionsequenceanalysesindicatesdispersalfromtheU.S. EastCoastasthesourceoftheinvasiveIndo-Pacificlionfish Pteroisvolitans inthe Bahamas.MarBio156:12131221. 19.MorrisJA,Jr.,SullivanCV,GovoniJJ(2011)Oogenesisandspawnformationin theinvasivelionfishes, Pteroismiles and Pteroisvolitans .ScientiaMarina75(1): 147154. 20.WaltersCJ,MartellSJD(2004)FisheriesEcologyandManagement.Princeton: PrincetonUniversityPress.399p. 21.AllenMS,BrownP,DouglasJ,FultonW,CatalanoM(2009)Anassessmentof recreationalfisheryharvestpoliciesforMurraycodinsoutheastAustralia.Fish Res95:260267. 22.BotsfordLW,WickhamDE(1979)Populationcyclescausedbyinter-age, density-dependentmortalityinyoungfishandcrustaceans.Proceedingsofthe 13 th EuropeanMarineBiologySymposium13:7382. 23.BotsfordLW(1981)Theeffectsofincreasedindividualgrowthrateson depressedpopulationsize.AmNat117:3863. 24.BotsfordLW(1981)Optimalfisherypolicyforsize-specific,density-dependent populationmodels.JMathBio12:265293. 25.GoodyearCP(1980)Compensationinfishpopulation.In:CHHocutt,JR StaufferJr.,eds.Biologicalmonitoringoffish.Lexington:LexingtonBooks,DC HeathandCompany.pp253280. 26.GoodyearCP(1993)Spawningstockbiomassperrecruitinfisheries management:foundationandcurrentuse.Canadianspecialpublicationsin fisheriesandaquaticscience120:6781. 27.MacePM(1994)Relationshipsbetweencommonbiologicalreferencepoints usedasthresholdsandtargetsoffisheriesmanagementstrategies.CanJFishAq Sci51:110122. 28.GoodwinNB,GrantA,PerryAL,DulvyNK,ReynoldsJD(2006)Lifehistory correlatesofdensity-dependentrecruitmentinmarinefishes.CanJFishAqSci 63:494509. 29.PottsJC,BerraneD,MorrisJA,Jr.(inpress)Ageandgrowthoflionfishfromthe WesternNorthAtlantic.Proceedingsofthe63 rd GulfCaribbeanFisheries Institute;Inpress. 30.MorrisJA,Jr.,AkinsJL(2009)Feedingecologyofinvasivelionfish( Pteroisvolitans ) intheBahamianarchipelago.EnvBioFish86:389398. 31.HareJA,WhitfieldPE(2003)Anintegratedassessmentoftheintroductionof lionfish(Pteroisvolitans/milescomplex)tothewesternAtlanticOcean.NOAA TechnicalMemorandumNOSNCCOS2.21p. 32.MuellerGA(2005)Predatoryfishremovalandnativefishrecoveryinthe ColoradoRivermainstem:whathavewelearned?Fisheries30:1019. 33.Serbesoff-KingK(2003)MelaleucainFlorida:aliteraturereviewonthe taxonomy,distribution,biology,ecology,economicimportanceandcontrol measures.JAqPlantManagement41:98112. 34.JohnsonLE,CarltonJT(1996)Post-establishmentspreadinlarge-scale invasions:dispersalmechanismsofthezebramussel Dreissenapolymorpha .Ecology 77:16861690. 35.DavisJE(2003)Non-nativespeciesmonitoringandcontrol.SanJuanRiver 19992001.ProgressreportfromtheSanJuanRiverrecoveryimplementation program.USFishandWildlifeService,Albuquerque,NewMexico. 36.ShulmanMJ,OgdenJC(1987)Whatcontrolstropicalreeffishpopulations: recruitmentorbenthicmortality?AnexampleintheCaribbeanreeffish Haemulonflavolineatum .MarEcolProgSer39:233242. 37.ConnellSD,JonesGP(1991)Theinfluenceofhabitatcomplexityon postrecruitmentprocessesinatemperatereeffishpopulation.JExpMarBiol Ecol151:271294. 38.AdamsAJ,EbersoleJP(2009)Mechanismsaffectingrecruitmentpatternsoffish anddecapods.In:LNagelkerken,ed.Ecologicalconnectivityamongtropical coastalecosystems.,Netherlands:Springer.pp85228. 39.HardingEK,DoakDF,AlbertsonJD(2001)Evaluatingtheeffectivenessof predatorcontrol:thenon-nativeredfoxasacasestudy.ConBio15:11141122. 40.BeckMW,HeckKL,AbleKW,ChildersDL,EgglestonDB,etal.(2001)The identification,conservation,andmanagementofestuarineandmarinenurseries forfishandinvertebrates.BioScience51:633641. 41.AdamsAJ,DahlgrenCP,KellisonGT,KendallMS,LaymanCA,etal.(2006) Nurseryfunctionoftropicalback-reefsystems.MarEcolProgSer318:287301. 42.DahlgrenCP,KellisonGT,AdamsAJ,GillandersBM,KendallMS,etal. (2006)Marinenurseriesandeffecti vejuvenilehabitats:conceptsand applications.MarEcolProgSer312:291295. 43.NagelkerkenI,RobertsCM,vanderVeldeG,DorenboschM,canRielMC, etal.(2002)Howimportantaremangrovesandseagrassbedsforcoral-reeffish? Thenurseryhypothesistestedonanislandscale.MarEcolProgSer244: 299305. 44.BeckMW,HeckKL,Jr.,AbleDL,ChildersDL,EgglestonDB,etal.(2003)The roleofnearshoreecosystemsasfishandshellfishnurseries.IssuesinEcology11: 112. 45.MumbyPJ,EdwardsAJ,Arias-GonzalesJE,LindemanKC,BlackwellPG,etal. (2004)Mangrovesenhancethebiomassofcoralreeffishcommunitiesinthe Caribbean.Nature427:533536. OverfishingLionfish PLoSONE|www.plosone.org7May2011|Volume6|Issue5|e19666