fh_2007_04.pdf (1.528Mb)

FI SK E N o g H AV ET

No. 4/2007

Environmental information for stock evaluation and management advice purposes

Geir Huse, Bjarte Bogstad, Kjellrun Hiis Hauge, Knut Korsbrekke, Harald Loeng, Kathrine Michalsen, Dankert Skagen, Jan Erik Stiansen, Svein Sundby, Einar Svendsen, Sigurd Tjelmeland and Reidar Toresen

Environmental information for stock evaluation and management advice purposes

0 4 8 12 16

1900 1920 1940 1960 1980 2000

3,5 3,7 3,9 4,1 4,3

By

GeirHuse,BjarteBogstad,KjellrunHiisHauge,KnutKorsbrekke, Harald Loeng, Kathrine Michalsen, Dankert Skagen, Jan Erik Stiansen, SveinSundby,EinarSvendsen,SigurdTjelmelandandReidarToresen

November 2006

2

1. Recommendations

Withaviewtoincreasingtheutilisationofenvironmentalinformationforstock evaluation and management advice purposes, we offer the following recommendations:

1. EstablishecosystemͲdefinedstockͲadviceprojectsstaffedwithawiderange ofexpertiseinordertoincreaseknowledgetransferandtherobustnessoftheadvice offered,andtocapturecommonecosystemprocesses.

2. Improve the flow, operationality and availability of data derived from observationsandmodelsimulations.

3. Identifyandestablishoperationalandquantitativerelationshipsbetween environmentalfactorsandstockvariablessuchasgrowth,recruitmentandnatural mortalityinourmostimportantspecies.

4. Quantifypotentialimprovementsinhistoricalstockevaluationandadvisory situationsbyincludingenvironmentalinformation.

5. Review environmentͲdependent reference points for fish mortality and spawningstockbiomass.

6. Develop andimplement environmental andbehaviourͲbased models for correctingacousticandtrawlsurveys.

7. DrawupanewdataͲgatheringstrategyaimedatmeetingtherequirements ofecosystemͲbasedadviceprovision.

8. Continue to develop and improve the utilisation of environmental informationinexistingmultiͲstockmodels(BIFROST,GADGET,SYSTMOD).

9. Finalisethenumericalmodelforcopepods,improvemonitoringprocesses andinitiateeffortstoestimatezooplanktonstocks.

10. DevelopindividualͲbasedmodelsofmigration,growthandmaturationfor ourmostimportantstocks,particularlywiththeaimofestimatingpredatorͲprey overlaps,andstudymonitoringandmanagementstrategies.

2. Contents

1. Recommendations...2

2. Contents...3

3. Mandate...4

4. Introduction...5

4.1 Background...5

4.2 Ecosystemdynamics...5

4.3 Climaticprognoses...8

5. Examplesoftheapplicationofenvironmentalinformationforstock evaluationandmanagementadvicepurposes...10

5.1 Mackerel...10

5.2 Cod,haddockandcapelin...10

5.3 NorwegianspringͲspawningherring...11

6. ExamplesofknownenvironmentalͲfishrelationshipsthatarenotutilisedfor stockevaluationandmanagementadvicepurposes...11

6.1 Cod,capelinandherring...11

6.2 AnchoviesintheBayofBiscay...12

6.3 Westernhorsemackerel...12

7. Integrationofenvironmentalparametersintocurrentstockevaluationand managementadvicepractices...12

7.1 Stockestimates...13

7.2 ShortͲtermprognosesandtacticaladvice...14

7.3 MediumͲtermprognoses...17

7.4 Designandevaluationofmanagementstrategies...20

8. NewstockͲevaluationandadvisoryconcepts...23

9. OperationalInformation...24

10. Discussion...25

10.1 Quantificationofecosystemdynamics...25

10.2 Organisationofstockadviceprovision...27

10.3 Environmentalinformation...27

10.4 Zooplanktonbiomassmeasurement...28

10.5 RelevantworkinggroupsoftheICES...29

11.References...30

4

3. Mandate

FollowinganinitiativefromHaraldLoeng(sakLg66/2006),theInstituteofMarine Research’smanagementgroup appointedagroup to evaluatetheinclusionof environmentalparametersinstockevaluations.Themandateofthegroupwasto considerthefollowing:

x Currentstatus

ͲWhichclimateparametersarebeingused?

ͲHowaretheseparametersutilised?

ͲWhichclimate/fishrelationshipsdoweunderstandatpresent, either qualitativelyand/orquantitatively?

x Whycanwenotuseeverythingthatweknowforstockevaluationpurposes?

x Howcanweintegrateclimateparametersintostockestimatesandstock prognoses?

x Whatsortofclimateinformationisrelevantforuseinthefuture?

x Howcanweobtainsuchinformation,andwhatareourrequirementsas regardsformatandoperationality?

Thefirstmeetingofthegroupagreedtoslightlymodifytheconceptualframeworkof the mandate. It was decided to adopt a broader perspective and to look at environmentͲfishrelationships(includingstockinteractions)asawhole,ratherthan inthenarrowsenseoftheconcept.Thegroupalsoagreedtoemploytheconceptsof

“stockevaluation”and“provisionofstockadvice”inplaceof“stockestimates”and

“stockprognoses”.Thiswouldgivethestudymoreroomformanoeuvrethanifit hadkepttothemorenarrowlydefinedinitialsetofconcepts.Thestructureofthe reportwasalsofreedsomewhatfromthestructuresetoutinthemandate,while thegroupsoughttocovertheoriginalsetoftopics.Inallotherrespects,thegroup stayedwithinthetermsofitsmandate.

The group comprises a wide range of expertise and includes scientists workingonstockestimation,climaticeffects,oceanographyandbehaviour/ecology:

GeirHuse,chair DankertSkagen SigurdTjelmeland KathrineMichalsen KjellrunHiisHauge EinarSvendsen SveinSundby HaraldLoeng JanErikStiansen ReidarToresen BjarteBogstad KnutKorsbrekke

ThegroupsubmitteditsreporttotheInstituteofMarineResearch’smanagement teamon15November2006.

4. Introduction

4.1 Background

Knowledgeofecosystemdynamicsisessentialifwearetobeabletounderstand, evaluateandpredicthowenvironmentalchangeandchangesinfishingpracticeswill affectthemarineecosystem.Inthisconnection,twofundamentalchallengeswillbe thoseofidentifyinginteractionsanddistinguishingbetweenmajornaturalvariations andhuman impacts. Stockevaluations andprovisionofadvice regardingliving marineresources(fish,crustaceans,marinemammals)arecurrentlybasedalmost entirely oncommercial catchdata and trawlandacoustic data from research surveys. Theseestimates are essentially descriptive, telling us how stocks and harvestingrateshaveevolvedovertime,andhowthefuturedevelopmentofastock willbeaffectedbyharvestingrates,butsaylittleornothingaboutwhyastockhas evolvedasithasdone,apartfromsheddingsomelightontheroleplayedby harvesting.Withoutabroaderunderstandingofthedrivingforcesbehindchangesin stocks,wecanmakeonlyextremelylimitedpredictionsabouthowtheywillevolvein thefuture.

This report concretises the problems involved in incorporating environmental information,usedinthebroadestsenseoftheterm,intostockevaluationand advisoryprocesses.Inthecourseoftime,wehavegeneratedagreatdealof knowledge ofthestructuresandfunctions ofmarine ecosystems,andclimate changemayprovetobejustasimportantasthefisheriesinbringingaboutchanges inthestateofourecosystems,includingthesizeoffishstocks.Thereportcoversto onlyalimitedextenttheconsiderableknowledgethatwealreadypossessaboutour marineecosystems;howeverinwhatfollowsweofferabriefdescriptionofnatural variationsinecosystems,inordertobeabletosaysomethingaboutthepotential benefits of environmental information. This is followed by some examples of environmental information used in the provision of advice, and of known relationshipsthatarepotentiallycapableofbeingemployed,butwhichhavenot beenusedtodate.Wethenpresentanddiscussananalysisinwhichweoutlinehow environmentalinformationcouldbeincorporatedintostockevaluationandadvisory processes. Finally, on the basis of this analysis, we offer some specific recommendations(seeabove)regardingwhattheInstituteofMarineResearch oughttodointermsofincorporatingenvironmentalinformationintoitsstock evaluationandadvisoryprocesses.

4.2 Ecosystemdynamics

Norwegianfisheriesareas,fromtheBarentsSeainthenorthtotheNorwegianSea, thecoastofNorwayandtheNorthSeainthesouth,spandifferenttypesofmarine ecosystems.TheBarentsSeaisanArcticecosystem.Aswemovesouthtowardsthe NorthSea,temperaturesrise,andthespeciescompositionchangesinfavourof moretemperatespecies.Wemovefroman“arctic”ecosysteminthenorthtoan

6

“arcticͲboreal”systemintheNorwegianSeaanda“borealͲtemperate”ecosystemin theNorthSea.Withineachoftheseecosystemswecanidentifyaconsiderable degreeofnaturalvariation.Thisisreflectedtoacertainextentbyfisheriesdataand surveys,butinmanycasesthereisagreatdealtobegainedbylookingatseveral componentsoftheecosystem,ratherthanfocusingpurelyoncurrentstocklevels.

Climaticvariationsareafundamentalsourceofvariationinotherpartsofthe ecosystem,butitcanbeusefultolookatresponsesinmanypartsoftheecosystem, becausethelatterisaffectedbyclimateinsomanyways.Climateaffects,for example,productionandtransportatlowertrophiclevels,thedistributionoffish, mutualinteractionsamongfishstocks,andthestructureoftheecosystemitself.This lastwillbeparticularlyrelevantinthefuturebecauseofthemajorchangesthatare takingplaceintheecosystemsoftheNorthSeaandtheBarentsSea.

Temperatureisanimportantvariableintheoceanclimatebecauseitaffectsevery linkinthefoodchain,fromphytoplanktontofish.Organismsattheselevelsareall poikilotherms, which means that temperature has a direct influence on their metabolism.Anychangeintemperaturewillthusaffectfishbothdirectlyandalso indirectlyviaallorganismsatlowerlevelsofthefoodchain.However,themarine climateisnotsimplyamatteroftemperature;lightlevels,whicharemodifiedby cloudconditions,andturbulence,whichisaffectedbythewinds,areoceanclimate variablesthataffectorganismsatindividuallevel,anditisparticularlythelowerlinks inthefoodchain,i.e.plankton,thatareaffectedbylightandturbulence.Current systems also affect the transport and dispersal of freely drifting plankton at populationlevel.Therearethusalargenumberofpotentialindirecteffectsofocean climate change on, forexample, cod in the marine ecosystem.In correlations betweentemperatureandgrowth ratesof a fishstock,aknowntemperature relationshipmayactuallybeasurrogateforaseriesofotheroceanclimatevariables.

Figure1illustratestheeffectsofvariouslinksinthefoodchainatindividualand populationlevel.ClimatechangealsotakesplaceonarangeofdifferenttimeͲscales, rangingfromannualtodecadalandmultidecadal(Figure2).

AnexampleoftheeffectofclimateonfishstocksisshowninFigure3,where variationsinthesizeoftheNorwegianspringspawning(NSS)herringstockpartially correlatewithlongͲperiodtemperatureoscillationsintheKolaSection.Thecollapse oftheherringstockattheendofthe60swasprimarilytheresultofhighfishing pressure,butthelowbiomassofthestockatthebeginningofthepreviouscentury canscarcelyhavebeenduetotheherringfishery.ThebuildͲupofthestockinthe 30swasprobablylargelyclimateͲdriven;theresultofseveralgoodrecruitmentyears andgoodgrowthconditionsforthespawningstock.

Zooplankton Planktivore fish

Larger fish Marine mammals

Darwinian

Influence on individuals:

SALINITY TEMPERATURE

TURBULENCE LIGHT

Influence on populations:

ADVECTION SPREADING VERTICAL

MIXING

Phytoplankton

Kelvinian

CARBON NUTRIENTS

The microbes

New tonian

Figure1. Effectsofclimateonthemarineecosystem.Theoceanclimatevariablesof salinity,temperature,lightandturbulenceaffectmarineorganismsatindividuallevel.These variablesareincludedinindividualͲbasedmodelsofmarineorganisms.Theoceanclimate processesofadvection,diffusionandverticalmixingaffectmarineorganisms,particularly plankton,atpopulationlevel.Theseprocessesareincludedinnumericalcurrentmodels.

FromSundby(2006a).

Thedynamicsofherringstocks,whichareaffectedbytheclimate,arejustone exampleoftheobviouseffectsoftheenvironmentonfish.Climatealsohasmajor effects on other fish stocks, and it also produces cascade effects; rises in temperature,forexample,increaserecruitmenttoherringstocks,whichinturn meansmoreherringpredationoncapelinfryandthusacollapseincapelinstocks.In manycases,wearealsoabletoquantifyenvironmentaleffectsongrowth,for example,butthisisonlytakenintoaccounttoalimitedextentinstockevaluation andadviceprovision.OneexceptionistherelationshipbetweentheNorthAtlantic Oscillation(NAO)andherringcondition(Holstetal.2004),whichisusedinproviding adviceaboutNSSherring.

Inspiteofthesecleareffects,wemaystillaskourselveswhyweoughttoimplement knowledgeoftheenvironmentintheadvisoryprocess.Putsimply,thisoughttobe doneif itenables ustoimprove theaccuracyand/orrobustness of ourstock estimatesandtheadviceweoffer.Thisisobviouslyanabsoluterequirementfor suchanexpansionofcurrentmethodology.Inanycase,itwillbeimportanttomake majoreffortsinresearchifwewishtoimproveourunderstandingofecosystems, andsucheffortsdonotthereforeneedtobejustifiedintermsoftheirdirect applicabilitytostockevaluationandadviceprovision.Nevertheless,itisaproblem thatresearchresultsofrelevancetostockevaluationandprovisionofadvicearenot

utilisedbecausetheadaptationoftheresultsneededtoturnthemintoaformthat wouldrenderthemdirectlysuitableforthesetasksisnotbeingdone.Responsibility forensuringthatitisdonemustliebothwiththosewhoproducetheresearch resultsandwiththosewhoworkonthestockmodelsthatemploythem.

7HPSHUDWXUH>&@R

Figure2. TemperaturesintheAtlanticsegmentoftheKolaSection;ϬʹϮϬϬm,stations ϯʹϳͿ͘ ϭͿBluecurve:annualmean. ϮͿRedcurve:3Ͳyearrunningannualmean.Thisclearly showsthedecadaloscillations.ϯͿGreencurve:LongͲtermaverageproducedwiththeaidof a30ͲyearlowͲpassfilter.Thisclearlyshowsthemultidecadaloscillations.Originaldatafrom PINRO,Murmansk.

0 4 8 12 16

1900 1920 1940 1960 1980 2000

3,5 3,7 3,9 4,1 4,3

Spawningstockbiomass(mill.tonnes)

Year

Temperature(°C)

Figure3. Biomass of spawning stocks of herring ;ƌĞĚͿ and longͲterm mean temperaturesintheKolaSection;ďůƵĞͿ͘Thetemperaturecurveisinmostrespectssimilarto thegreencurveinFigureϮ͕buttheaveragingmethodissomewhatdifferent.FromToresen

&Østvedt;ϮϬϬϬͿ͘

4.3 Climaticprognoses

TheInstituteofMarineResearchiswelltotheforefrontinthedevelopmentof coupledhydrodynamicoceanclimatemodels,andisamemberofaninternational

Temperature

groupthatisdevelopingtheROMSmodelsystem,aconceptthatisparticularly suitableforlinkingwithmodelsofbiologicalprocessesandindividualͲbasedmodels.

Wehaverecentlycompletedtherunningofa50Ͳyear“hindcast”timeseriesfora globalROMSmodel,withhighresolutionintheNorthAtlantic.Inthefuture,thiswill placeusinabetterpositiontostudytheunderlyingmechanismsofthewellͲ establishedrelationshipsbetweenoceanclimateandfishstocks.However,thereis stillalongwaytogobeforesuchmodelscanbeusedforgeneralclimateprediction purposes.Allcoupledoceanclimatemodels(ocean/ice/atmosphere)areinfluenced bytheatmosphere,whichmeansthatpredictionsoftheoceanclimatecanneverbe betterthanthepotentialpredictivecapacityofouratmosphericmodels.Generally speaking,changesinthefutureclimatecanbedividedintotwocomponents:

1.Naturalclimaticvariations,whichtakeplaceoveracascadeofperiodsranging fromseasonaloscillationstothousandsofyears(e.g.theMilankovitchCycleof 26,000years).

2.Anthropogenic climate change, resulting from the rise in concentrations of greenhousegasesintheatmosphere.

Itisactuallymucheasiertopredictglobalanthropogenicclimatechangesthan naturalchanges.ThisisbecausetheriseinCO₂levelsintheatmosphereproduces quite specific, simple alterations in the global radiation budget, which is the relationshipbetweenincomingshortͲwavelengthsolarradiationandthelongerͲ wavelengthradiationthatleavestheEarthforspace.Atpresent,therearetwentyor soglobalcoupledclimatemodelscapableofsimulatingrisesinatmosphericCO2. Most of them produce fairly similar results: a doubling of atmospheric CO₂ concentrationswillraisethemeanatmospherictemperatureby2–4^oC.TheBergen ClimateModel(BCM),whichisusedbytheBjerknesCentre,isonesuchmodel.For theNorthAtlanticOceanregion,includingtheNorthSea,theNorwegianSeaandthe BarentsSea,itestimatesariseofbetween1and2^oCintheannualmeanseaͲsurface temperatureinwinterby2070,withariseofaround1.5^oCinthecentralBarents Sea(Fureviketal.2002).However,onlythegreenhouseeffectisbeingsimulated here.Naturalvariations,frominterannualtomultidecadaloscillations,cannotbe predictedbythismodel,becausewedonotknowthespecificdrivingforcesthat causetheseoscillationsandthuscannotincorporatethemintothemodel.Onthe basisofexistingtimeseries,however,suchasthosefromtheKolaSectioninthe BarentsSea (Figure 2), we know that both decadal and multidecadal climatic oscillationsareverydistinctinourwaters,andthesehavefurthermorebeenshown tohavecleareffectsonecosystemsandfishstocks.Thereisnoreasontobelieve thatsuchoscillationswillbeabsentfromafuturewarmerclimate.Butonthebasis ofourtimeseries,weknowthattheseoscillationscansuddenlychange,bothin amplitudeandinfrequency.TheNAO,forexample,hasbeenextremelydominant fromthesixtiesuntilthepresentday,asitalsowasatthebeginningofthetwentieth century,butduringotherpartsofthetwentiethcenturyitwasmuchlessmarked.

ThedecadaloscillationsinNorwegianwatersappeartohaveweakenedagainduring thepastfewyears.Inmakingpredictionsofoceanclimate,therefore,weareleftfor thetimebeingwithstatisticalanalysesofclimaticperiods,combinedwithour

10

knowledgeabouthowaclimatesignalinonepartoftheNorthAtlanticspreads throughoutthesystem.

5. Examples of the application of environmental information for stockevaluationandmanagementadvicepurposes

As mentioned above, the application of environmental information for stock evaluationandmanagementadvicepurposesisextremelylimited;howeverafew examplesdoexist.Environmentalinformationisused,forexample,incombination withsurveyestimates(e.g.ofmackerel),inpredationand/orproductionmodelling (e.g.capelinintheBarentsSea,shrimpsoffGreenland),instatisticsͲbasedprocesses andintheformofqualitativeindicators.Thefollowingsectionsoffersomeexamples ofsuchusesofenvironmentalinformationforcertainindividualstocks.

5.1 Mackerel

TheonlyfisheryͲindependentmeasureofstocksizecurrentlyinuseconsistsof biomassestimatesbasedonthetriͲannualeggsurvey,whichcountsthenumberof eggsintheseaattheirearlieststageofdevelopment.Thisnumberisconvertedto givetheproductionofeggsperhenfish.Theprocessofrecalculationmakesuseof thetimethataneggremainsatthefirststage,andisdoneonaroutinebasis togetherwiththeeggmeasurements,butmoreaccurateestimatesoftheambient temperatureinwhichtheeggsgrowwouldimproveproductionestimates.Theuse oftheContinuousUnderwayFishEggSampler(CUFES)inconjunctionwithamodel oftheverticaldistributionofmackereleggs(e.g.Sundby1983)hasthepotentialto raisethelevelofaccuracyofestimatesofeggproductionwhileenablingtheegg surveyitselftobecarriedoutinashortertime.

5.2 Cod,haddockandcapelin

Stockestimatesofcapelin,codandhaddocktakeintoaccountmortalityarisingfrom thepredationofcodonthesespeciesbytakingcodstomachsamples,estimated environmentaltemperatureandexperimentaldataontherateofdigestionincod.

Theexperimentshaveshownthatariseintemperatureof1^oCraisestherateof digestion by 10–15 %. It is therefore very important to use a representative temperatureincalculationsofthissort,andeffortsarebeingmadetoimprovethe methodologyonthispoint.Wearealsoconsideringincorporatingtheconsumption bycodinourstockestimatesofsomeoftheothermostimportantpreyspeciesof cod,e.g.shrimp.

Inthecapelinstockmanagementprocess,expectedpredationbycodisusedto predictthespawningstockbiomass(SSB)ofcapelinfromthecapelinsurveycruisein September–Octoberuntilspawningthefollowingspring,usingamodelbasedoncod

stockestimatesandstomachsamples,andinthiscasetoo,correctenvironmental temperatureestimatesareimportant.

5.3 NorwegianspringͲspawningherring

ThereisacloserelationshipbetweentheobservedNAOindexinwinterandthe biomassofcopepodsinthefollowingsummer.Thereisalsoacloserelationship betweencopepodbiomassintheNorwegianSeaandherringconditionaftera feedingseasonwithlargeamountsofcopepods.Herringgrowthcanthereforebe predictedbyestimatingplanktonbiomassintheyearaftertheobservedwinterNAO index(Holstetal.2004).Inprovidingadviceforherring,inconnectionwiththezones towhichtheybelong,thetemperatureandplanktondistributionobservedduring theMayecosystemcruiseisusedtoprovideadvice(internallywithintheNorwegian delegation) in connection with the negotiations regarding the zones to which NorwegianspringͲspawningherringbelong.

6. ExamplesofknownenvironmentalͲfishrelationshipsthatarenot utilised for stock evaluation and management advice purposes

Therearealsoagoodnumberofknownenvironment–fishrelationshipsthat,for various reasons (see Discussion), are not utilised for stock evaluation and managementadvicepurposes

6.1 Cod,capelinandherring

Inthecourseofthepastfewyearswehaveidentifiedrelationships(statistical models)thatappeartobepromisingformakingpredictionsofcodrecruitmenttwo tothreeyearsahead,basedonobservationsoftemperature,capelinandcod(Huse

&Ottersen2002,Stiansenetal.2002,Stiansenetal.2005)andonmodelsofwater transportandprimaryproductionintheBarentsSea(Svendsenetal.,inpress).We expectthesetobeparticularlyusefulasawayofenablingustoprovideearly warningofrecruitmentfailure.

OneͲyearpredictionsofcapelinrecruitment(oneͲyearͲolds)appeartobepossibleon thebasisofsatellitemeasurementsofseaͲsurfacetemperaturesintheBarentsSea andbiomassestimatesof0Ͳgroupandmaturingcapelin(Stiansenetal.2002,2005).

Onthebasis of satellite measurementsofNorwegianSeasurfacetemperature measurementsandmeasured0Ͳgroupherringindices,herringrecruitment(threeͲ yearͲold)canbepredictedthreeyearsaheadintime(Stiansenetal.2002,2005).

12 6.2 AnchoviesintheBayofBiscay

AnchoviesareashortͲlivedspecies,inwhichrecruitmenthasadecisiveeffectonthe followingyear’sfishery.ThereisawellͲknownrelationshipbetweenrecruitmentand upwelling.Anupwellingindexwasutilisedforseveralyears,butwasabandoned after it resulted in completely unjustified advice to stop the fishery. “InͲyear monitoring”andacousticandeggsurveysarenowutilisedasindicators.Agreatdeal ofworkhasbeenputintounderstandingtherelationshipbetweendistributionand oceancurrent/temperature,bothinordertoimprovetheinterpretationofthe resultsandasabasisforprovidingmanagementadvice.

6.3 Westernhorsemackerel

ThereisagoodcorrelationbetweencatchesofhorsemackerelintheNorthSeain theautumnandinflowsofAtlanticWater.AprognosisfortheNorthSeafisheryis regularlyreportedtotheworkinggroup,butisnotusedindevelopingitsadvice, partlybecausetheNorwegianfisheryintheNorthSeaisunregulated,partlybecause theregulationasawholedoesnotfunctionparticularlywell,partlybecausethe NorthSeafisheryisarelativelysmallpartofthetotalfisheryandfinally,partly becausewelackareliableestimateforthisstock.Tightercontrolsofthisfisheryare beingdeveloped,andhorsemackerelareonthelistofspeciesforwhichNorwayand theEUaretryingtodevelopajointmanagementstrategy.Relationshipsofthissort maycometobeimportantinthefuture.

7. Integration of environmental parameters into current stock evaluationandmanagementadvicepractices

Theabovesurveyisnotexhaustive,butitdoesshowthattherearenotsomany applicationsofenvironmentalinformationinstockassessmentandadviceprovision, in spite of the fact that we know of, and can quantify, a good number of relationships.Thereasonsforthisstate ofaffairs arecomplex,and arepartly historicalandpartlyanaturalconsequenceofthefactthatfisheriesdataandquota adviceshouldbeintegrated.AtthispointwedonotintendtoofferaninͲdepth analysisofwhythingshaveturnedoutastheyhavedone,butwillratherconsider howwecanimprovestockassessmentandadviceprovisionbymakingmoreuseof environmentalinformation.Inordertodiscussthisprobleminmoredetail,itmaybe usefultodividethistaskintofourfacets:

x Stockestimates

x ShortͲtermprognosesandprovisionoftacticaladvice x MediumͲtermprognoses

x Designandevaluationofmanagementstrategies

Thefollowingsectionsdescribetheessenceoftheseconceptsandhowweenvisage thatenvironmentalinformationcouldbeimplementedtoagreaterextent,bothona generalbasisandforspecificstocks.

7.1 Stockestimates

7.1.1 Description

Theseareestimatesofstocksize,usuallyintermsofagestructureandharvesting rates,fromthepresentdayandasfarbackintimeaswehavedata.Eventhoughthe toolsthatweusetomakestockestimatesoftenaredescribedasmodels,theyare reallyonlyanalysesofobserveddata,whichwetrytomakeasindependentas possibleofmodellingassumptions.Wecalculatehowlargethestockmusthavebeen backintimetoallowthereportedcatchestohavebeentaken,alsotakingother causesofmortalityintoaccount.ThemostcentraltoolforthispurposeisVirtual PopulationAnalysis(VPA)anditsequivalents,althoughothermethodsalsoexist.

Cruisedata(e.g.CatchPerUnitEffort:CPUE)isusedtodeterminecurrentstock levelsrelativetopreviouslevels.Thisisquitedifferentfrommodellingpopulation dynamics,whereweanalysemathematicallyhowapopulationwillbehave,given certainassumptionsabouthowproductionisdependentonthestateofthestock.

Productionmodels,whichareutilisedwhenfewdataareavailable,liesomewhere betweenthetwo.Thekeyisanassumptionabouthowproduction(growthand recruitmentminusmortality)isdependentonthesizeofthestock.TheInstituteof MarineResearchhaslargelyavoidedusingmodelsofthistype,asourphilosophyhas beenthatweprefertobaseourestimatesondirectobservationsofstocksand fisheriesratherthanonmodelassumptions,bothbecausedirectobservationsare availabletousandbecausestocksdonotalwaysbehaveaccordingtothetextbooks.

7.1.2 Potentialapplicationsofenvironmentalinformation

Therehavebeenfewattemptstoincorporateenvironmentalinformationinstock the stock estimation process. Nevertheless, certain areas can be identified as potentialareasforadoptingtheuseofenvironmentalinformation;theseincludethe quantificationofinterannualvariationsinnaturalmortality(Recommendation3)and theparameterisationofverticaldistributionandangleoftiltinfish,inorderto correctsurveydata(Recommendation6).

Researchhasshownhowacoustictargetstrength(TS)dataforherringvaryinthe courseofthedayasafunctionofdepthandtiltangle(Huse&Ona1996,Huse&

Korneliussen2000,Vabøetal.2002).Thisresearch,togetherwithastudyofthe depthdistributionofherring(ofdifferentyearclasses)inthehistoricalacoustictime seriescouldprovideabasisforimprovingourobservationmodel,whichinturn wouldprovidemorereliablestockestimates.Similarly,thereexistmajordifferences betweendaytimeandnightͲtimeacousticmeasurementsofbiomassoffishinthe

14

BarentsSea,whereacousticbackscatteringisonlyhalfasstrongatnightasitis duringtheday(Hjellviketal.2004).Thisisprobablyduetochangesinthevertical distributionofthefish,anddifferencesbetweendayandnightTS,butmayalsobea functionofseasonalandenvironmentalfactors.Anenvironment/behaviourͲbased correctionmightbeonewayofimprovingacousticestimatesinsuchcases.

Environmentalinformationisnotcurrentlyusedinestimatingcapelinstocks,which areproducedonthebasisofdirectacousticmeasurements.Astudyofcapelin migrationrelatedtoenvironmentalvariablessuchastemperatureandcurrentcould leadtoimprovementsinsurveydesignandthustomorereliablemeasurementsfor agivensurveyeffort.Systematicvariationsinacousticsignalstrengthcouldalso influenceestimates(Jørgensen&Olsen2002),asdiscussedabove.Thisisparticularly criticaliftherearemajorinterannualvariationsinenvironmentalfactors.

Thelevelofandvariationsinnaturalmortalitycanbeimportant,particularlyfor shortͲlivedspecies,butalsoforyoungerageͲgroupsofotherspecies.Forexample, shrimpsareanimportantitemofthedietofcod,particularlywhencapelinare scarce(Mehl1989,Bogstadetal.2000).Itcanbedifficulttoproducegoodestimates ofnaturalmortality,butbeyondthelarvalstage,predationisthemajorcauseoffish mortality,andthereforethereoughttobesignificantpotentialfordevelopingbetter estimatesofnaturalmortalityifweincreaseoureffortsinthisdirection.

7.2 ShortͲtermprognosesandtacticaladvice

7.2.1 Description

Thesecalculationsarebasedonprojectingstockestimatesforaselectionoffish mortalities,usuallyforoneyearahead,inordertocalculatewhatsuchmortalities are equivalent to in terms of catches.Theseform the basis of annual quota recommendations.Recommendedquotasarethecatchescorrespondingtowhatis regarded as responsible or desirable mortality on the basis of given criteria (precautionary principle and/or management regulations). The number of fish caughtthatisequivalenttothedesirablemortalityisconvertedintotonnesofcatch, andtheremainingstockisexpressedintermsofspawningbiomass,whichisthe measureofstockswhichresourcesmanagementusuallyrefersto.Theaimisusually thatthespawningbiomassshouldbemaintainedatahigherlevelthanthereference pointsBPAandBLIM(seebelow).Thiscalculationcanbemadestochastically(with bootstrapping) or deterministically; the latter method is more usual. The managementauthoritiesneedaspecificfigureforthefollowingyear’squota,asthey have problems in relating to distributions, while for our part, we are rather uncomfortableaboutthefactthatthefigureshavefairlywideconfidenceintervals.

These estimates are based on assumptions about natural mortality, growth, maturationandrecruitment,aswellastheestimatedstockandfishmortalityage profile.Recruitmentplaysanimportantroleifmortalityratesarehigh(shortͲlived

speciesand/orhighfishingpressure),otherwiseitisnotimportant(forsingleͲyear prognoses).

7.2.2 Potentialapplicationsofenvironmentalinformation

Of the factors that are included in shortͲterm prognoses and tactical advice provision,recruitment,naturalmortalityandgrowth/maturationareparticularly important candidates for making estimates on the basis of environmental information(Recommendations3and9),sincealloftheseprocessesarecontrolled byenvironmentalconditions,includingstockinteractions(Recommendation8).The problemhasbeenthatofpredictingtheseonthebasisofknownenvironmental information,andinthisrespectthereispotentialforfurtherdevelopment;itis particularlyimportanttobeabletowarnofchanges.

Sætersdal&Loeng(1987)showedthatcod,haddockandherringrecruitedbetterin warmerthanincolderyears.Thebestyearclassesoccurredwhenevertherewasa changefromacoldtoawarmregime.Ellertsenetal.(1989)showedthathigh temperaturesinspawningandnurserygroundswereanecessarybutnotasufficient conditionforgoodyearclassesofcod.Ithasalsobeenshownthattheenvironment hasagreaterinfluenceonrecruitmentwhencodstocksconsistoffewyearclasses thanofmany.Thisisrelatedtothefactthatwhenmanyyearclassesarespawning, thespawningprocesstakeplaceoveralongerperiodoftimeandinalargerarea, andthereisagreaterprobabilitythatsomeofthelarvaewilloverlapwiththe copepodspawningperiod.Wearethusawareofarangeofdifferentrelationships, andforseveralspeciesthesehavealsobeenquantifiedandpresentedaspredictions forseveralyearsahead(e.g.Huse&Ottersen2002,Stiansenetal.2002,Stiansenet al.2005,Svendsenetal.,inpress).

Individualgrowth(andthusweight/maturation)isdependentontemperatureand theavailabilityoffood.Forsomespecies,weusefoodavailabilitydatatoprovide prognosesofgrowth.Prognosesofthegrowthofherringinthequotayear,for example,canbemadeonthebasisofplanktonbiomassmeasuredinthestock estimationyearandonprognosesofplanktonbiomass(Holstetal.2004).

Naturalmortalitywillbedependentonthenumberofpredators,theamountof otherpreyandtheoverlapbetweenpredatorsandprey.Itisperhapsthelastfactor thatwillbemostdependentontheenvironmentinashortͲtermprognosis.In reality,estimatingpredatorͲpreyoverlap(continuouslyintime)canonlybedoneby usingnumericalmodelsofmigration(withdataassimilation),whichinturnmustbe coupledwithmultispeciesmodelswithinputdatafrombiophysicalmodelsoflower trophiclevels(physicalfactors,plankton,larvae).Inthisconnection,weneedtoput mucheffortintocompletingongoingworkonamodelofCalanusfinmarchicus,and performinglongͲtermsimulations.

Naturalmortalityalsoformspartofanassumedquantityinmakinghistoricalstock estimates,andithasacertainsignificanceforestimatesofcurrentstocklevels,i.e.

16

thepointofdeparturefortheshortͲtermprognosis.Forthefollowingyear’squota recommendations,theassumedlevelofnaturalmortalityplaysalesserrolethanwe mightsuppose.Ifnaturalmortalitychanges,however,theadvicewillbewrong.In thisconnection,therefore,changesinnaturalmortalityaremoreimportantthanthe absolute level, and in one way or another, such changes will be based on environmentalconditions.

TheInternationalCouncilfortheExplorationoftheSea(ICES)isrequiredtoprovide adviceinaccordancewiththeprecautionaryprinciple.ThisconceptisnotaswellͲ definedaswemightwish,andICEShaschosentointerpretitinitsownway.The InstituteofMarineResearchusuallysupportstheICESinterpretation.Thecoreof thisinterpretationisthatspawningstocksshouldnotbecomesolowthatthisin itselfleadstoreducedrecruitment.Thereisalsoanassumptionthatrecruitmentis largelyindependentofbiomass,aslongasthisislargeenough.

ICES has established threshold values for biomass (BLIM) that are assumed to representtheminimumbiomassthatisrequiredtomaintain“normal”recruitment.

ICEShasalsoestimatedamaximumleveloffishmortality(F_LIM)which(atleastin principle)willproducealongͲtermbiomassofB_LIMwithnormalrecruitment.Inorder toallowroomfortheuncertaintyofstockestimatesandprognoses,ICEShaschosen tomakeuseofmoreconservativevalues(B_PAandF_PA).Theideaisthatifweestimate thatthestockisorwillbeclosetoB_PA,itisunlikelythatitwillactuallybeB_LIM.A similarwayofthinkingappliestoFPAandFLIM.Advicewillusuallybebasedonthe ideathatfishmortalityshouldbenohigherthanFPAifthestockafterfishingis expectedtobegreaterthanBPA.Ifitisnot,amortalityratelowenoughtobringitup againtoB_PAinthecourseofoneortwoyearswillberecommended.Forafew stocks,usingthisasabasisforadviceisnotsuitableforvariousreasons,andother criteriahavetobeused.Thecapelinstockisoneexampleofthis(seebelow).

Thisbasisforadviceprovisionhasseveralproblematicaspects.Theseconcernboth howthereferencevaluescanbederived inastatisticallyacceptablewayand whether there exists a wellͲdefined biomass threshold that will guarantee a

“normal”recruitmentlevel,howeverstaticsuchathresholdmightbe.Onecurrent problemiswhetherB_LIMandB_PAshouldbefixedorfunctionsofenvironmental conditions.Itisalsoanopenquestionwhetherfishmortalityshouldbemodifiedas stockproductivitychanges.Thereisagreatdealtobesaidfortheideathatthesame levelofmortalityisoptimalevenifproductionrateschange,butthisquestionisthe subjectofdiscussion.

DevelopmentsarenowmovinginthedirectionoflongͲtermmanagementstrategies (seebelow),amainfeatureofwhichisrationalexploitationoftheproductivityofthe stock.Insuchaconnection,otherindicatorsthancurrentpointsofreferencecould be relevant, and we can envisage that most reference points would become superfluousasthisstrategydevelops.

Thecapelinquotaissetonthebasisofacousticbiomassmeasurementsandon prognosesofcodpredationoncapelin.Thequotaissetonthebasisofprobabilistic

estimatesandBLIM,withouttheuseofBPA.Thismakesiteasiertoincorporate environmental information in a logically consistent way. The most important contributionismadebyincludingaprognosisofyoungherringintheBarentsSeain thequotayear,basedonmeasurementsinthestockestimationyear,andbasing quotarecommendationsonexpectedcapelinrecruitmentinsteadofonminimum spawningbiomass.Expectedpredationbyharpsealscanalsobeincorporated.Here, theoverlapbetweenpredatorandcapelinisacentralelement,andthiscanbeused inprognoseswhenwehavegoodmodelsofmigration.Likecapelin,shrimparealso preyforanumberofpredators,andpredationisadecisivefactorindeterminingthe evolution and population dynamics of stocks. Predation by cod is included in estimatesofshrimpstocksoffGreenland(Hvingel&Kingsley2006),andcould certainlyalsobeimplementedelsewhere.

7.3 MediumͲtermprognoses

7.3.1 Description

Thisiswherestockprojections,typicallyfor2–10yearsahead,aremade.Foreach year,anewyearͲclassmustbeincorporatedandweightsandmaturationestimated.

Projectionsofthissortarealwaysmadestochastically,typicallybybootstrapping, whererecruitmentandgrowthratesandanyotherparametersaredrawnfrom distributions.Thesedistributionsareusuallyderivedfromhistoricalstockestimates, andunlesschangesareincorporatedthedistributionsareassumedtobestationary over time. MediumͲterm prognoses are primarily used in the evaluation of managementregulations, to examine longͲtermyields, risk, and to assess the prospectsofrebuildingstocksinpoorcondition.Howmanyyearsaheadsimulations ofthissortaremeaningfuldependsonhowlongthefishusuallylive,thequalityof the stock estimates, and the degree to which the methods employed take multispeciesinteractionsandenvironmentalfactorsintoaccount.

7.3.2 Potentialapplicationsofenvironmentalinformation

Within this framework there are ample opportunities to utilise environmental information, which needs to be translated into probability distributions for recruitment,growthandnaturalmortality.For2–3Ͳyearprognoses,however,wecan utiliseenvironmentalinformation,suchasrecruitmentpredictionsforcod.Itwillbe particularlyimportanttobeabletoshedlightonlongͲtermchangesinrecruitment, becausestocksandyieldsovertimeareproportionaltoaveragerecruitmentand naturalmortality(Recommendations3,8and9).

Thereexistsanapparatusforprobabilisticprojectionsforfiveandtenyearsahead, butthishasnotbeenutilisedduringthepastfewyears.Thereexistspawningstock–

recruitment relationships based on temperature, and a methodology for

18

incorporatingmodeluncertainty.Forcertainstocks,suchascapelinandshrimp, changesinnaturalmortalitycanbeobtainedbymeansofmultispeciesmodels.

InconnectionwiththechangefrommanagingcapelinaccordingtoB_LIMtomanaging thespeciesaccordingtoexpectedrecruitment,itwouldbenaturalforthestandard producttobeaprobabilisticprojectionforthreetofiveyearsahead.Thereexistsa model(BIFROST)thatdoesthisinaconsistentmultispeciescontext,butharpseal predationhasnotyetbeenimplemented.TheInstituteofMarineResearchshould furtherdevelopitsmethodologyforprojectingallofourmostimportantstocksin consistent multispecies contexts, and the models GADGET (minke whale–cod–

herring–capelin;sametypeofmodelasMULTSPEC, (Bogstadetal.1997))and SYSTMOD(cod–herring–capelin)(Hamre2003)couldbegoodpointsofdeparturefor theBarentsSea(Recommendation8).

Oceanclimateandplanktonproductiondatahavesofaronlybeenutilisedtoa limitedextentinthepredictionoffishstocksinthemediumtolongterm.However, moregoalͲorientedresearchoffersasignificantpotentialfordevelopmentinspiteof thefactthatwearestillunabletopredictthedevelopmentoftheoceanclimate fromoneyeartothenext.Thereasonforthisisthatittakestimefromwhena climaticpulseinfluencesproductionatlowertrophiclevels(planktoniclevels)untilit hasaneffectintermsoffishproductionandrecruitment(seeFigure1).Examplesof thisaretherelationshipbetweentheNAOindexandherringconditiononeanda halfyearslater(Holstetal.2004)andthesignificanceofclimateforthesurvivalof codfry,whichoffersthreeͲyearpredictionsofrecruitmentlevelsofthreeͲyearͲold fish(Stiansenetal.2002,Stiansenetal.2005,Svendsenetal.,inpress),(becausethe yearͲclassstrengthislargelydeterminedatthelarval/frystage).

Ona longer timeͲscale,biologicalmultispecies interactions enable us to make predictions,forexamplewhenastrongyearclassofherringgrazesdownthecapelin stock,inturncreatingproblemsforcod.ItalsonowseemsclearthatmultiͲdecade climatesignalshavealongͲlastingeffectontheproductivityofmarineecosystems (Toresen&Østvedt2000,Drinkwater2006).Thisresultsinlonger(multiͲdecadal) periodsduringwhichanecosystemcantolerategreaterorlowerfishingpressure, dependingonthephaseoftheclimaticperiodinwhichwefindourselves(Sundby 2006b).ThisisillustratedinFigure4,whichshowsthatbluewhitingrecruitmenthas risendramaticallyandhascompletelychangedduringthepastfewyears,inspiteof veryhighouttakesofuptotwomilliontonnes.Similarly,theconditionofNorwegian springͲspawningherringhasvariedratherwidelyinthecourseoftheyears(Figure 5).Someofthesechangesaredependentonstockbiomass,butthereisprobably alsoasignificantenvironmentalcomponentherethatsayssomethingaboutthe productivityoftheecosystematdifferenttimes.Weoughttobeabletousethisto estimategrowthandrecruitmentrates.ThereareindicationsthatthelongͲcyclic productivityofzooplanktonisthekeytothis.Forthisreason,itisimportantthatwe shouldstartmakingstockestimatesofzooplanktonsimilartothosethatwemakeof fishstocks(Recommendation9).

Itwilloftenbedifficulttopredictecosystemchanges,asisthecasewiththerisein bluewhitingrecruitment.However,itisimportantthatweshouldbeabletocapture these changes at an early stage and commit resources to determine causal relationshipsandhowsuchchangesshouldbereflectedintheadvicewegive.The InstituteofMarineResearchfacesanimportantchallengeinthisrespect.The quarterlyseriesofsituationreportsfortheNorthSeathatareissuedbytheNORth SEaPilotProject(NORSEPP),whichareeditedandpromotedbyHeinRuneSkjoldal (Skjoldal2006)isanexampleofanoperationecosystemevaluationthatcouldbe usedtorapidlycapturechangesinecosystems.

Even though environmental information is not directly numerically utilised in managementinstruments,itsmostimportantapplicationmaybetoprovideearly warningofmajorchangesinecosystems.Whensuchchangesareobservedor modelledintheclimate,planktonoratearlystagesinthelifecycle,weneedto possessadequateknowledgeandmethodsthatwillenableustowarnofthelikely effectsonfishstocks,sothatthesecanbetakenintoaccountintheadvicethatwe offer(precautionaryprinciple).

0 10000000 20000000 30000000 40000000 50000000 60000000 70000000

1980 1985 1990 1995 2000 2005

Number of recruits (x10-6)

Year

Figure4. Bluewhitingrecruitment.Blackcurve:annualnumberofoneͲyearͲoldfish.

Redcurve:fiveͲyearrunningaverage.FromICES(2006).

20 0.70

0.75 0.80 0.85 0.90

1950 1960 1970 1980 1990 2000

Condition

Year

Figure5. Condition of Norwegian springͲspawning herring. Black curve: annual condition.Redcurve:fiveͲyearrunningaverage.

7.4 Designandevaluationofmanagementstrategies

7.4.1 Description

Thebasicprerequisiteforanymanagementregimeisthatitshouldnotremove–in thecourseoftime–morefromthestockthanitiscapableofproducing.Insimple terms,productionistheincreaseinbiomassduetothegrowthofindividualfishand the recruitment of new fish, less mortality losses. A complete evaluation of managementstrategiesisacomprehensiveprocessthatmayalsoinvolvesocioͲ economicfactors,butatthispointwewillconsideronlythepurelybiologicalpartsof thisprocess.Inanevaluationofthissortwetypicallyusethesametypeoftoolsas formediumͲtermprognoses,butoftenoveralongerperiodoftime.Withtheaidof numericaltoolsofthistype,wecansimulatearangeofmanagementstrategies(fish mortalities,patternsoffishing,etc).Partoftheevaluationprocessconsistsof checkinghowwellastrategycantoleratedeviationsfromidealconditions,suchas differencesbetweenactualstocksandassumedstockswhendecisionsaretaken, permittedvs.actualouttakes,variationsinproductivityresultingfromchangesin environmentalconditionsorclimaticscenarios.Theresultswillbeevaluatedwith respecttooneormorecriteria,themostusualofwhicharetheprobabilityof exceedingabiologicalreferencepoint(e.g.thatthespawningstockfallsbelowBLIM), averageyield(intonnesoreconomicvalue),andinterannualvariationincatches.

7.4.2 Factorsthatdeterminehowstocksevolve

Theevolutionofastockisdependentbothonmanagementmeasuresandhow theseareactuallyimplemented,andonnaturaldrivingforces,whicharewhatwe usuallyassociatewiththeconceptof“environment”(Figure6).Thereareeffectson recruitment,growthandmortality,andthesecantakeplaceeitherimmediatelyor aftersomedelay.Forexample,alowerlevelofrecruitmentwillresultinareduction intotalbiomassandspawningbiomassafterseveralyears,andseveralmoreyears maypassbeforeeffectivemanagementmeasuresareimplemented.

Managementmeasures Resultingfishery

Impact bynature

Climate

Food availability Other influences

Stock

development:

Recruitment Growth Mortality Immediate/delayed

Stateof the stock

Figure6. Themostimportantfactorsaffectingtheevolutionofastock.

Theinternalrelevanceofthesefactorsvariesfromonestocktoanother.Natural forcesareoccasionallyblamedforafallinastock–sometimesjustifiably–butwe wouldbewrongtobelievethatfisheriesareofnoimportance.Onthecontrary,the point should be to modify fishing pressure according to changes in natural conditions.Ifwecansayanythingabouthownaturalconditionsarelikelytoevolve, wecanmakerecommendationsthattaketheseintoaccount.Ifnot,wemusttryto adopt management strategies thatare capable ofadjusting to changes in the productivityofastockastheseoccur.Todate,thelatterhasbeenthemostusual wayofdesigningmanagementstrategies.Sucharobustadaptivestrategyrequires ustobecapableofrecognisingchangesingoodtime,andexperiencehasshownthat thisismuchmoredifficultifwedonotunderstandtheunderlyingmechanisms involved.Thisisperhapspartofthereasonthattheadviceregardingbluewhiting wasfartoolateintakingintoaccountthefactthatrecruitmenthadchanged dramatically.

7.4.3 Potentialapplicationsofenvironmentalinformation

Environmental information ought to play an important role in the design of managementstrategies.Wecantakeenvironmentalimpactintoaccounteitherin thedesignofmanagementstrategiesand/orbyincludingenvironmentalimpactin theevaluationofamanagementstrategy(Recommendations4,5,8,10).Wecanuse

22

environmentalinformationeithertoexploitastockeithermoreefficientlyormore carefully,accordingtotheprevailingenvironmentalconditions.Foreachstock,we needtoassesswhetherwearecapableofusingenvironmentalinformation(ormore environmentalinformation)insuchawaythatitactuallycontributestoachieving ourmanagementaims(Recommendations3,4,7,10).

Wehaveanumberofalternativesavailablewhenwearedesigningmanagement strategies:

Wecanattempttotakeenvironmentalimpactintoaccountinthedecisionrule,by:

x “Bakingitinto”therule(forexamplethatcapelinshouldbemanagedonthe basisofthefoodrequirementsofcod).Thisisbasedonthegoalofexploiting resourcesinthebestpossiblewayandonacceptingthatthemanagement strategyforonestockmayalsoaffectotherstocks(predators,competitors, mixedfisheries,etc.).

x Environmentalimpactcanbeexpressedintermsofoneormoreindicators, andtheseformpartofthedecisionruleitself.Inthiscase,theremustbe definedthresholdvaluesfortheindicators,andrulesneedtobedrawnupfor theconsequencesthatwillensuewhenthethresholdvaluesareexceeded.In allprobability,thiscouldonlybeusedwhentheaimistotakeextragood careofastock.Thismethodisnotutilisedtoday.

Adhoc:

x In cases where scientists are worried about a stock because of an environmentalfactor.Thiswillnotbeexpressedinthe“quantitative”advice, butthescientistswillrequestthemanagementauthoritiestobeparticularly carefulinsettingaquota.Suchawarningwouldnotbeexpressedintheform ofarule,butratherintheformofastatementthatweoughttodepartfrom theruleandbemoreconservativethanthequantitativeadvicesuggests.The management authorities themselves must work out how the recommendation should be implemented. This is done occasionally in practice.

Thereferencepointsonwhichtheharvestingruleisbasedaredependentonthe environment(thisisnotusedtoday;Recommendation5):

x Definingreferencepointswhoseaimistomaintainagivenspawningstock structure(problem:geneticchangesresultingfromhighfishingpressure).

x The productivity of the spawningstockisdependenton environmental factors(skippedspawning,conditionfactor).

x Movablereferencepointisdependentonenvironmentalfactorsthataffect thefishbeforetheyarerecruitedtothefishablestock.

x Harvestingrulesareoccasionallyevaluatedwhileasingleenvironmental condition is modified. This can be done without the rule itself taking environmentalfactorsintoaccount.

Atthepresenttime,majorchangesaretakingplaceintheecosystem,withfairly largedisplacementsoffishstocksandlongͲtermchangesinrecruitment,forexample

ofbluewhiting(Figure4).Thisneedstobetakenintoaccountintheprovisionof strategicadvice,sinceitaffectsecosystemcarryingcapacity.Arangeofdifferent scenariosintermsofprey,predatorsandclimateoughtthereforetobestudied, ideallyusingmultiͲstockmodelsresolvedinspatialterms.

8. NewstockͲevaluationandadvisoryconcepts

Theaboveanalysisshowshowenvironmentalinformationcanbeincorporatedinto currentstockevaluationsandadviceprovision.Inspiteofthefactthatthecurrent systemhasevolvedinthecourseoftime,thefoundationsofthemethodologiesit employswerelaidmorethan50yearsago.Sincethen,therehavebeendramatic technologicaldevelopmentsthatofferusquitedifferentpossibilitiesthanweusedto have.Atthesame,theextentofmanagementhasdramaticallychanged,andin principleitnowcoversthewholeecosystem.Itmaythereforebeusefultobriefly mentionsomenewanglesofattackthatdifferconceptuallyfromcurrentstock evaluationandadviceprovisionprocesses.

Giskeetal.(2001)consideredtheprospectsofdevelopingspatiallybased,fisheryͲ independent monitoring systems. They evaluated five different concepts and concludedthatsuchsystemscouldbedevelopedusingexistingtechnologyandthat theycouldprovidesignificantinputstostockevaluationandadviceprovision.But thiswouldrequiresignificantnewinvestmentsinobservationplatformsforboth physicalandbiologicalvariables.

InspiteofthefactthattheAMOEBEproject(Svendsenetal.2002),whichwasgiven averygoodinternationalevaluation,wasputintocoldstorage,theeffortinvolvedin describingtheprojectlaidthefoundationsforenvisaginghowafullyoperational spatiallybasedmodellingsystemformarineresearchandfisheriesmanagement couldbedeveloped.TheAMOEBEprojectwaslargelybasedonasystemofmodels basedonobservations,andinwhichsignificantinvestmentswouldbemadeinthe observationinfrastructureintheformoftheoperationalisationofdataflowsfrom vessels,remotemeasurements,andvarioustypesofmooredanddriftingbuoys.

Thesystemwasbasedontherecognitionthatanecologicallybasedapproachto marinemanagementinaccordancewiththeprecautionaryprinciplerequiresaccess tomuchmoreinformationthandoestraditionalsingleͲstockmanagement.Thiscan only be achieved by putting significant efforts into national and international cooperationaimedatintegratingexistingandnewmultidisciplinaryknowledgeand dataviatheextensiveuseofmodels,withtheassimilationofobservations.The understanding,quantificationandpredictionofrecruitment,growth,maturationand natural mortality will require (more or less) threeͲdimensional “continuous”

knowledgeofphysics,plankton,larvaeandfry,migrationsanddistributionsoffish andmarinemammals,overlapsbetweenpredatorsandprey,andwhoeatswhom andhowmuch.However,thiswouldrequirethedevelopmentofanoperational systemofspatiallyresolvedmultispeciesmodelswithdataassimilation(overlapping

24

intimeandspacebetweenpredatorsandprey)andthreeͲdimensionalbiophysical modelsofoceanclimateandlowertrophiclevels(physics,planktonandfishlarvae).

Thespatial3ͲDsystemsdescribedaboveareconceptuallycompletelydifferentfrom today’sadvicesystem.Such3ͲDsystemswouldbecapableofprovidinguswith significantamountsofknowledgeaboutthemarineecosystem,andtheywouldalso havethe potentialtoimproveour advice.However,thedevelopmentof such systemswouldbeextremelyexpensive,anditwasestimatedthattheAMOEBE projectwouldhavecostNOK1billion.Weseenoprospectoffinancingsucha systemwithoutsignificantadditionalfunding,andhavethereforenotevaluated systemsofthistypeinmoredetailhere,sinceourprimarytaskhasbeentoevaluate theincorporationofenvironmentalinformationintotheexistingadvisorysystem.

9. Operational information

Byitsverynature,adviceprovisionisoperational.Thismeansthatweareinvolvedin supplyingproductsand/orservicesatmoreorlessregularintervals,inaformatthat theusercanunderstandandmakeuseof.Iftheyaretobeuseful,thesourcesof environmentalinformationneedtobeextremelyreliableandlongͲterminnature.

Theabilitytoutiliseinformationrapidlywhenitarrivesisachallengethathas scarcelybeenmetbytheexistingsystem.Inthisrespect,theKULTprojectatthe InstituteofMarineResearchisanimportantinitiativeintermsofoptimisingtheflow andavailability ofrelevantdata. Atpresent,togetherwithICES,wearefairly operationalwithrespecttocountingfishandadvisingonindividualstocks,butwe stillhavealongwaytogobeforewecanbecomeoperationalontherelevant environmentalinformationthatisneededforausefuloperationalecosystemͲbased approach.Nevertheless,thereismuchtobegainedfromimprovingtheflowand availabilityofdataattheInstituteofMarineResearch(Recommendation2).One exampleofthisareacousticdata,whicharecurrentlyscarcelyavailableinonline databases,andwhichrequiremajoreffortstoaccess.Onemeansofimprovingthis situationmightbetoincreasethecapacityofNMDtotakeonthissortofwork,so thatalldatafromtheInstituteofMarineResearchwouldactuallybecollectedthere.

Forseveralyears,theEUandESA(EuropeanSpaceAgency)havebeendeveloping GMES(GlobalMonitoringfortheEnvironmentandSecurity),whichwillbethe Europeanframeworkforoperationalenvironmentalinformation.(Onaglobalscale GEOSS(GlobalEarthObservationSystemofSystems)hassincebeensetup(mostly bytheUSA)andGMESmightwellberegardedaspartofGEOSS).GMESwillforma central part of the EU’s 7^th Framework Programme, and the programme is particularlyinterestedinhavingsomeonedeliverarangeofoperationalmarinecore services(MCS)aboutthestateoftheenvironment(past,presentandfuture,on globalandregionalscales)thatwillbeaccessibletoeveryone.TheInstituteof Marine Researchshould playan activerole here inensuring thattheflowof environmental information will be useful in achieving the Institute’s general

objectives.Inthefirstinstance,thecoreserviceswilldealwithclimateandphysics, probablyprimaryproduction,andpossiblythedriftoffishlarvae.

Wealsoneedtodomoretoconsiderwhichessentialprocessesoughttobestudied (inthelaboratoryandinthefield)andmonitored,andtowhatextentthereare nodalpointsintheoceanwherekeyparameterscanbemonitoredwithsufficiently highfrequency.Doweneedtodevelopnewtechnologyorsimplyresourcesthatwill enableustomakebetteruseofexistingmoderntechnology?Anexampleisour ARGOprofilingbuoysintheNorwegianSeawhich,fittedwithsomesimpleacoustic orpossiblyopticalcapabilities,couldbeusedtomonitorwinteringzooplankton stocks.Modelscanbeusedtoamuchgreaterextentthantheyareatpresentto defineoptimisedobservationsystems.

10. Discussion

10.1 Quantificationofecosystemdynamics

AswementionedintheIntroduction,clearrelationshipsexistbetweenclimateand ecosystemproductivity.Fishbiomass,inparticular,iscorrelatedwithlongͲcycle climaticoscillations,sinceclimatehassomanydifferentdirectandindirecteffects ontheecosystem.LongerͲlastingclimatechangeswillthusresultinmoreeasily detectableeffects.Nevertheless,themediumͲtermeffectsareveryimportantfor our understanding of changes in recruitment to fish stocks, and a better understanding of such relationships can be useful in providing better stock prognoses.Climaticvariationsandrelatedecosystemeffectsarethustobefoundon mosttimeͲscales,andtheymayoccurfairlyrapidly.Inthecourseofthepastcouple of years, for example, snake pipefish (Entelurus aequoreus) have established themselvesinlargeareasoftheNorthSea,theNorwegianSeaandtheBarentsSea.

ItisimportantthatmanagementshouldbeoperationalonthistimeͲscale,andthere isajob to be done here in implementingenvironmentalfactorsinour stock evaluationsandadviceprovision,particularlyinprovidingearlywarningsofchanges thataretakingplace.

Inthenameofthe“ecosystemapproach”tomanagement,agooddealofeffortis beingputintoidentifyingindicatorsthatdescribethestateoftheecosystem(in someconnectionorother).Itisquitenaturalforecosystemstochangeinthecourse oftime.Ifwearetomakesignificantadvanceinouruseofsuchindicators,itwillbe importantthatweshouldbeabletoevaluateintegratedeffectsandtodifferentiate toacertainextentbetweenclimateͲdrivenchangesandalterationsthataredueto humanactivity.Thisisfarfrombeingassimpleasitsounds,becausethereisevery reasontobelievethatthereareinteractiveeffectsbetweentheclimateandhuman activity(forcertainperiods,theecosystemcan“tolerate”greaterimpactsthan during others). For this reason, it is important to identify operational and quantitativerelationshipsbetweenenvironmentalfactorsandstockparameterssuch

26

asgrowth,recruitmentandnaturalmortality(Recommendation3).Wemayhope that establishing such relationships will increase the accuracy and reduce the uncertaintyofourannualadvice,andpossiblyallowustoadoptaprecautionary approachifweanticipatereducedgrowthrates.Evenifsuchcorrelationsarenot basedonanunderstandingoftheprocessesinvolved,theycanbeusefultoolsthat willenableustoutiliseinformationabouttheenvironmenttosaysomethingabout thefutureevolutionofstocksthatwillbebetterthanhavinganexpectedaverage value.Wherewealsopossessunderstanding,processͲbasedmodelsareobviously preferable. Multispecies models are partly processͲbased, and the further developmentofsuchmodelsintermsofenvironmentalinformationwillimproveour utilisation of process understanding in stock evaluation and advice provision (Recommendation8).ForshortͲlivedspecieswithhighnaturalmortalityrates,such assandͲeels,wearefacedwithanumberofinterestingchallenges,todealwith which climatic variables may turnout tobeuseful. However, the situation is differentforlongͲlivedspecies,forexamplethecurrentyear’sestimateofnumbers ofthreeͲyearͲoldNorthArcticcod.Thisyearclassisthe2003yearclass,whichhas been(atthetimeofevaluation)observedsixtimes(11timesifwedifferentiate betweentrawlandacousticindices).

Today,thereferencepointsonwhichtheharvestingrulesarebasedareconstant, andindependentofthestateoftheecosystem.IfweseealongͲlastingchangeinthe productivityoftheecosystemorareabletoidentifygoodqualitativerelationships betweenenvironmentalfactorsandgrowthorrecruitment,wemayimaginethatthe pointsofreferencewillbe setasa functionofthestateofthe environment (Recommendation5).Thiswouldallowstockstobeharvestedmoreheavilyinmore productiveperiodsthaninpoorperiods.

VirtualecosystemsarecomputerͲsimulatedecosystemsinwhichweattemptto representthemostimportantelementsofthesystemintermsoftheinteractions, growth,survivalandreproductionofasetofspecies.Suchasystem,ifitemphasised importantfishstocks,wouldbeagoodpointofdepartureformakingthorough analyses of the relationships between stocks, data collection and advice (Recommendation10):Thesystemcouldthenbemanagedinthesamewayasreal stocksandecosystems,withthedifferencethatinthevirtualsystem,wecouldenter allthecomponentsatalltimes.Wecouldthussailcruisesinthevirtualsystemand studytheeffectsofvariousdegreesofcruisecoverageonstockestimatesandon howdifferenttypesofadviceandouttakesoffishwouldaffectthestockinthe courseoftime.Wewouldalsobeabletostudyaccumulatederrorsovertimeand theconsequencesofdifferentspatialpatternsoffishing.Thisapproachisalready wellͲknowninfisheriesmodelling,butonlywitheithernoorhighlysimplifiedspatial resolutionsandlongtimeͲsteps(monthly,quarterly,annual).Developingsucha systemwill require a fairamountofeffort,butwould be quitepossible with currentlyavailabletechnology.Systemsofthissortwereacentralelementofthe AMOEBEprojectapplication(Svendsenetal.2002).

10.2 Organisationofstockadviceprovision

AnimportantorganisationaltrendattheInstituteofMarineResearch,aimedat increasingtheuseofenvironmentalinformation,involvesestablishingecosystemͲ definedstockadviceprojectsbasedonawiderangeofexpertise,staffedbypeople whoworkonstockadviceprovision,statistics,ecology,oceanographyandplankton (Recommendation1).Thiswillalsoincreaseknowledgetransferandtherobustness oftheadviceweprovide,sincemorepeoplewillbedirectlyinvolvedintheprocess.

Thiswayoforganisingouradvicewillalsoenableustobetterabletocapture ecosystemprocessesthathavesharedeffectsongeographicallyoverlappingstocks thanthecurrentsystempermits.Incertaincases,thegeographicalaspectisalready established, but we lack the multiplicity of expertise needed. This type of organisationwillmeanthatmorepeopleareinformedaboutthestateofthestocks andtheecosystem,andwillfeelthattheyarepartoftheadvisoryprocess.

Wemayhopethatthiswillcontributetostrongerconnectionsbetweenresearch projectsattheInstituteofMarineResearchandtheadvisoryprocessthanwehave today,asituationthatwouldbefruitfulforindividualscientists,theadviceprocess andtheInstituteofMarineResearchasaresearchͲbasedadvisoryinstitution.Such anorganisationaltrendwouldalsobeanaturalsteptotakeinconnectionwiththe currentreorganisationinthedirectionofecosystemͲbasedstockmanagement.

10.3 Environmentalinformation

InthecourseofthepastfewyearstheInstituteofMarineResearchhasbeguntosail annualecosystemcruises,inthecourseofwhichwecovertheecosystemsofthe BarentsSeaandtheNorwegianSeasynoptically.Thisisanimportantsteponthe way to ecosystemͲbased management,but it might nevertheless be useful to developastrategyformonitoringecosystemswithaviewtodeterminingwhatsort of environmental information is needed for ecosystemͲbased management (Recommendation7).Inthefieldofsurveymethodology,thereisalsothepossibility of improving stock estimates by incorporating environmental information, for example by correcting for how different environmental conditions produce variationsinfishbehaviour(Recommendation5).

Thephilosophyofbasingstockestimatesoncatchdatamayactasabarrierto makingrapidprogressinanecosystemͲbasedapproachtofishstockmanagement, whilewecanalsoseeagrowthintheuncertaintyofthecatchdataonwhich traditionalVPAanalysesarebased.Theproblemwithsuchastrategyisthatthe methodologyisextremelylimitedintermsoflookingaheadintime,sincewedonot havedataforthefuture.Thismeansthatwewillneedtodevelopandmaintaina verydifferentmethodologyforevaluatingthepastintermsofthefuture.Bytheir verynature,adviceandstrategiesarebasedonbeingtoevaluatethefuture(onthe basisofknowledgeandmodelassumptionsbasedonobservationsfromthepast).

Thevariousmethodsinvolvedarecomplementary,butagreatdealcouldbegained frombeingabletoagreeoncommonmethodsofevaluatingstocksinthepast,the

28

presentandthefuture.Wementionedatthebeginningofthisreportthatthe objectiveofusingenvironmentalinformationmustberelatedtoimprovingthe accuracyorrobustnessofquotaestimates.Forthisreason,itisimportantthatnew methodologyshouldbeintroducedinsuchawaythatweareabletotestthatthis requirementissatisfiedbeforeweadoptit(Recommendation4).

TheInstituteofMarineResearchiscurrentlyrunninganumberofrelevantprojects thatcouldprovidevaluableinputtoefforttoincludeenvironmentalinformationin stockevaluationandadviceprovision,anditwillbeimportanttomakeuseofthis knowledge by actively seeking for information in these projects. The new programmemanagers,particularlythoseinchargeoftheecosystemandclimate–

fishprogrammes,willplaykeyrolesinconnectingupresearchandadvisoryprojects.

TheKULTprojectwillplayacentralroleinimprovingtheflowofdata.Wehavesaid littleabouttheavailabilityofdatainthisreport,sincewehaverathertakenitfor granted that such data as are gathered will rapidly become operational (Recommendation2).Thisisnotnecessarilythecase,andwemusthopethatthe KULTprojectwillleadtoamoresatisfactoryflowandavailabilityofdata.

10.4 Zooplanktonbiomassmeasurement

CopepodsareacentralspeciesintheecosystemsoftheNorwegianSea,theNorth SeaandtheBarentsSea,andithasamajorinfluenceontherecruitmentandgrowth of fish in these ecosystems. Copepod biomass is already usedto predict the conditionofNorwegianspringͲspawningherringayearaheadintime.Becauseof thecentralpositionandinfluenceofcopepodsontherestoftheecosystem,we needtoworkforabetterunderstandingofcopepoddynamics(Recommendation9).

PerformingregularstockestimatesoftheNorwegianSeastockofthisspeciescould beausefultoolforthebetterunderstanding, quantificationandpredictionof growthandrecruitmentofmanyofourmostimportantfishstocks.LongtimeͲseries forzooplankton,inadditiontotheabilitytorelatechangesinzooplanktonbiomass to changes in other parts of the North Atlantic, would make an important contributiontosuchanunderstanding.TheSirAlistairHardyFoundation(SAHFOS) runs a largeͲscale, worldͲwide plankton monitoring scheme using continuous planktonrecorders(CPRs)installedonboardshipsofopportunity,usuallyferriesand liners.TheirlevelofactivityisparticularlyhighintheNorthAtlanticandtheNorth Sea,althoughtheoceanregionthatisourspecificresponsibilityisunfortunately Mareincognituminthisrespect.Theexceptionisa20Ͳyearperiodthatlasteduntil the 70s, when CPRs were operated by the weather ships “Polarfront I” and

“Polarfront II”,ontheroute betweenBergenandStationM. Weregardit as extremelyimportantthatthistimeseriesshouldbetakenupagain,andthatwe shouldalsoinitiatemeasurementsintheBarentsSea,byinstallingCPRseitherona supplyvesselroutetotheoffshoreinstallationsontheTromsøflakoronashipping routebetweenTromsøandLongyearbyen.SAHFOScandothisifNorwaybecomesa memberoftheFoundation.Norwayistheonlymajormarineresearchnationinthe westernworldthatisnotcurrentlyamemberofthisfoundation.