ContentslistsavailableatScienceDirect
Fisheries
Research
jo u r n al h om ep ag e :w w w . e l s e v i e r . c o m / l o c a t e / f i s h r e s
Harvest
control
rules
for
data
limited
stocks
using
length-based
reference
points
and
survey
biomass
indices
Ernesto
Jardim
a,b,∗,
Manuela
Azevedo
a,1,
Nuno
M.
Brites
a,c,1,2 aIPMA,InstitutoPortuguêsdoMaredaAtmosfera,Av.Brasília,1449-006Lisboa,PortugalbEuropeanCommission,JointResearchCentre,InstitutefortheProtectionandSecurityoftheCitizen,MaritimeAffairsUnit,TP051,21027Ispra,VA,Italy cDepartamentodeMatemática,CentrodeInvestigac¸ãoemMatemáticaeAplicac¸ões,InstitutodeInvestigac¸ãoeFormac¸ãoAvanc¸ada,Universidadede Évora,RuaRomãoRamalho,59,7000-671Évora,Portugal
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received22April2014 Receivedinrevisedform 18November2014 Accepted20November2014 HandlingEditorA.E.Punt Availableonline24December2014 Keywords:
Datalimitedstocks Risk
Surveyindices Lengthfrequencies Harvestcontrolrules LifeHistory
a
b
s
t
r
a
c
t
Therearealargenumberofcommerciallyexploitedstockslackingquantitativeassessmentsandreliable estimatesofstockstatus.ProvidingMSY-basedadviceforthesedata-limitedstocksremainsachallenge forfisheriesscience.Formanydata-limitedstocks,catchlengthcompositionand/orsurveybiomass indicesorcatch-per-uniteffort(cpue)areavailable.Informationonlifehistorytraitsmayalsobeavailable orborrowedfromsimilarspecies/stocks.Inthisworkwepresentthreeharvestcontrolrules(HCRs), drivenbyindicatorsderivedfromkeymonitoringdata.Theseweretestedthroughsimulationusingtwo exploitationscenarios(developmentandover-exploitation)appliedto50stocks(pelagic,demersal,deep seaspeciesandNephrops).WeexaminetheperformanceoftheHCRstodelivercatch-basedadvicethat isriskadverseanddrivesstockstoMSY.TheHCRwithabiomassindex-adjustedstatusquocatch,usedto providecatch-basedadviceforseveralEuropeandata-limitedstocks,showedthepoorestperformance, keepingthebiomassatloworverylowlevels.TheHCRsthatadjustthestatusquocatchbasedonthe variabilityofthebiomassindextimeserieswasabletodrivemostofthestockstoMSY,showinglow tomoderatebiologicalrisk.Therecoveryofbiomassrequiredasymmetricconfidenceintervalsforthe biomassindexandlargerdecreasesinstatusquocatchthanincreases.TheHCRbasedonlengthreference pointsasproxiesfortheFSQ/FMSYratiowasabletoreversethedecreasingtrendinbiomassbutwith
levelsofcatchbelowMSY.ThisHCRdidnotpreventsomeofthestocksdecliningwhensubjectto over-exploitation.Fordata-limitedstocks,theempiricalHCRstestedinthisworkcanprovidethebasisfor catchadvice.Nevertheless,applicationstoreallifecasesrequiresimulationtestingtobecarriedoutto tunetheHCRs.Ourapproachtosimulationtestingcanbeusedforsuchanalysis.
©2014ElsevierB.V.Allrightsreserved.
1. Introduction
The majority of worldwide commercially exploited fish and
shellfishstocksdonothaveaformalstockassessment,dueto
insuf-ficientdatatoestimatestocksizesandfishingmortality(Rosenberg
etal.,2014).IntheNortheastAtlantic,morethanhalfofthe200
stocksforwhichICESprovidesadvicelackaquantitative
assess-ment and estimatesof stock status (ICES,2013a). Thesestocks
∗ Correspondingauthorat:EuropeanCommission,JointResearchCentre,Institute fortheProtectionandSecurityoftheCitizen,MaritimeAffairsUnit,TP051,21027 Ispra,VA,Italy.Tel.:+390332785311.
E-mailaddresses:ernesto.jardim@jrc.ec.europa.eu(E.Jardim),
mazevedo@ipma.pt(M.Azevedo),brites@uevora.pt(N.M.Brites). 1 Tel.:+351213027000.
2 Tel.:+351266745370.
areclassifiedbyICESasdata-limitedstocks(ICES,2012a).Inthe
absenceofestimatesofstockstatus,providingquantitative
MSY-basedadviceremainsachallengeforfisheriesscience,whilefor
managementbodiesitcontinuestobeakeytopicforincreasing
thesustainableharvestingofmarineresources(Anon.,2013,2008,
2007,2006).
Thelastfiveyearshaveseenanincreaseinthedevelopment
andtestingofassessmentmethodstoestimatesustainableyields
andharvestlevelsfordata-limitedstocks(Rosenbergetal.,2014;
MartellandFroese,2013;WetzelandPunt,2011;MacCall,2009).
Thecategoryofdata-limitedstocks,inthesenseofstocks
with-outananalyticalassessment,includesstockswithaconsiderable
amountofinformation,althoughnotalwaysavailableorcompiled.
Forexample,lifehistoryparametersexistformoststocksexploited
commercially(seefishbase.org),lengthfrequenciesoflandingsare
cheaptocollect,and scientific surveyscatchmore speciesthan
thoseforwhichabundanceindicesaretraditionallyderived.Insuch
http://dx.doi.org/10.1016/j.fishres.2014.11.013
casestheproblemtendstoberelatedtothedifficultytoevaluate
thequalityofthedata,and/orthecostsofprocessingthe
informa-tion,and/ortheshorttimeseriesavailable.InEuropeanwaters,all
stockscaughtbyEuropeanfleetsarecoveredbytheEuropeanData
CollectionFramework(CouncilRegulation(EC)No199/2008)and,
hence,suchinformationexists.Moreover,datasetsaregrowing
everyyear,whichwillcontributetomovingstocksupwardsinthe
levelofinformationavailable.
GeromontandButterworth(2014)usedsimulationtestingto
checktherobustnesstouncertainties aboutresourcedynamics,
ofsimpleempiricalTAC-basedharvestcontrolrules(HCRs).These
authorsconsideredtwodata-poorscenariosfordataavailability:
a data-limited scenarioin which the stock had historical catch
data and information onthe mean length of the catch, and a
data-moderatescenariowherea directindexofabundancewas
also available for thestock. The empirical management
proce-duresinvestigatedbyGeromontandButterworth(2014)setfuture
TACsbyadjustingthepreviousyear’sTACupwardsordownwards,
dependingonwhethertherecentmeanlengthisaboveorbelow
atargetmeanlength.Inthecaseofthedata-moderatestocks,the
TACadjustmentwasbasedontherecentslopeoftheabundance
series(cpue)andonthedifferencebetweentherecentcpueanda
targetlevel.
Since 2012, ICES hasapplied a framework to provide catch
advicefortheEuropeandata-limitedstocks(ICES,2013a)within
aprecautionaryapproach.Forstockswherehistoricalcatchdata
andsurveysorotherrelativeabundanceindicesareavailable,catch
adviceisbasedontheapplicationofanHCRusingbiomassor
abun-danceindexadjustedstatusquocatches(ICES,2012b).Withinthis
HCR,itisquitecommontoadoptacomparisonoftheaveragesof
thetwomostrecentindexvalueswiththethreeprecedingones,
althoughthenumberofyearstobeusedtocomputethese
aver-agesmayvaryamongstocks,inordertoaccountforinter-annual
variabilityofsurveysorcpueseries.Simulationtestinghasshown
thatthisHCRdoesnotensureconservativecatchadvicewhena
stockisover-exploited,whileit’sperformancedeteriorateswhen
awell-managedstockbecomesover-exploited(ICES,2013b).
The work presented here has three main objectives: (i) to
developandtestgenericHCRsforstocksforwhichtheamountand
qualityofdataisinsufficienttoperformquantitativeassessments;
(ii)tounderstandthemechanismsthatdrivetheperformanceof
theseHCRs;and(iii)tosuggestasimulationprocedurebasedon
ManagementStrategiesEvaluation(MSE;Butterworthetal.,1997;
Cooke,1999;ButterworthandPunt,1999;Kelletal.,2005;Punt andDonovan,2007)tobeusedfordatalimitedstocks.
ThesimulationprocedureandtheHCRsproposedhereprovide
aframeworkfordesigningandtestingmanagementplansfordata
limitedstocks.However,whendealingwithareal-lifeapplication,
itis importanttotune thecandidate HCRstothespecificstock
andfishery(notattemptedhere)andtestitsimplementationwith
simulations.
2. Methods
Twoalternativestotheharvestcontrolrulesetby(ICES,2012a)
weredeveloped.Oneusessurveyinformationandthestatistical
characteristicsofthebiomassindex,whiletheotherusescatch
lengthcompositionsandlifehistoryparameters.TheseHCRswere
developedtakingintoaccountcommonindicatorsderivedfrom
fisheriesmonitoringprogrammesandinthemselvesdefine
differ-entlevelsof datalimitation.Survey informationisindependent
ofthefisheryandconstitutesa directobservationofthetrends
inbiomass.It’sanexpensivetypeofinformation,butalsoagood
sourceaboutthetrendsinbiomass,ifthesurveyisperformed
fol-lowingthegoodpracticesofdatacollectionandtheindexderived
with sound statistical methods. On the other hand, the length
distributionsofthecatchis fishery-dependentinformation. Itis
cheapertocollectthansurveyinformation,andoncealinkis
estab-lishedbetweenanexploitedstock’sexpectedlengthcomposition
and itsspawningpotentialratio(Hordyket al.,2014)it
consti-tutesanindirectobservationofthestockstatus.Duetotheirdata
requirements,theseHCRscanbeappliedtoawiderangeofstocks
withdistinctlevelsofdatalimitations,standingbetween
catch-onlytypeapproaches(MartellandFroese,2013)andfullanalytical
approaches.
The simulation procedure uses life history parameters and
information about the fishery’s selectivity to create a
simula-tionenvironment fortestingHCRs.Althoughsomeassumptions
arerequired,thesesimulationscanprovideinsight intothe
rel-ative effect of different HCRsand theirrobustness to common
uncertainties, like those about stock recruitment relationships,
implementationerror,etc.
2.1. SimulationsandMSE
Thestocksusedinthecurrentstudyweresimulatedbasedon
lifehistorytraitsthatlooselyrepresentthebiologyofthespecies
(Table1),coveringawiderangeoflifehistorycharacteristics.There
were50stocks:3pelagicstocks(2species:HerringandSandeel),
36demersalstocks(17species:Cod,Haddock,Whiting,Pollack,
Megrim, Anglerfish,Plaice, Sole,Striped redmulet,Lemon sole,
Brill,Dab,Redfish,Goldenredfish,Greenlandhalibut,Turbotand
Witch),6deepseastocks(4species:Ling,Blueling,Greater
sil-versmeltandRoundnosegrenadier)andNephropsintheshellfish
grouping(Table1).ThelifehistoryM/Kratioofthesimulatedstocks
rangedfrom1.09to4.20,withtheexceptionofthedeepseastock
‘rng-comb’thatpresentedahighM/Kratioof8.98(Table1and
Fig.1).
An age-structuredpopulationdynamics model with
recruit-mentgovernedbytheBeverton–Holtstock-recruitfunctionwith
steepness,h,of0.75(tosimulatestockswithmedium
productiv-ity)andvirginbiomass,Bv,of1000twasadopted.Thehvaluewas
adaptedfromMyersetal.(1999),whoindicatedamedianof0.71.
NaturalmortalityforeachstockwascomputedfollowingGislason
etal.(2010)andwasassumedtobetime-invariant.Fishgrowth
wasmodeledwiththevonBertalanffyfunction,withparameter’s
givenbyICES(2012b).Fleetselectivitywasmodeledwithadouble
normaldistribution(Hilbornetal.,2003)withthemodeequalto
thetheageof50%maturity,whilesurveyselectivitywasmodeled
withanasymptoticallyflattopcurvewithmaximumretentionat
age10.Fig.2showsanexampleforastockwithageof50%
matu-rityof5.4years.Bothselectivityfunctionsweresetconstantover
time.
Simulationswereperformedfortwodistincttrendsinfishing
mortalitytotesttheHCR’sperformancein‘development’and
‘over-exploitation’ fishery scenarios. For the‘development’ phase the
stocksweresubject,duringyears−14to0,toalinearincreasing
fishingmortalityfromF=0to2FMSY.The‘over-exploitation’phase
wasobtainedbykeepingfishingmortalityat2FMSY for25years
(years:−24to0),althoughonlythelast15(years:−14to0)were
usedforthesimulationstudy.Inbothcases,theHCRswereapplied
for25years(years:1–25).
Uncertaintywasintroducedin:
• theoperatingmodelconditioning,throughcatch-at-ageandthe
stock-recruitmentrelationship;
• theobservationerrormodel,throughtheobservationoftheindex
ortheratioM/K;
• theimplementationerrormodel,throughtherealizedcatches,
Table1
Lifehistoryparametersusedtosimulatestocks.aandb:length-weightrelationshipparameters;a50:ageof50%retention;a0,K,L∞:vonBertalanffygrowthfunction parameters;FMSY:fishingmortalitythatproducesMSYcatchesandM:naturalmortality.
Stockcode Group a b a50 a0 K L∞ FMSY M
ang-78ab Demersal 0.00001 3.000 5.680 −0.100 0.121 162.000 0.095 0.217 ang-ivvi Demersal 0.00001 3.000 4.320 −0.100 0.160 150.000 0.121 0.261 arg-comb-exVa Deep 0.01000 3.000 3.569 −0.100 0.220 42.400 0.219 0.348 arg-comb-Va Deep 0.01000 3.000 4.586 −0.100 0.170 47.900 0.165 0.324 bli-comb Deep 0.00116 3.273 5.379 −0.100 0.130 140.000 0.089 0.302 bll-2232 Demersal 0.02470 2.930 2.837 −0.100 0.270 50.100 0.263 0.502 cod-coas Demersal 0.01000 3.000 5.035 −0.100 0.140 130.000 0.097 0.351 cod-ewgr Demersal 0.00750 3.040 5.924 −0.100 0.117 150.000 0.068 0.485 cod-farb Demersal 0.00001 3.000 4.594 −0.100 0.156 110.000 0.124 0.280 cod-iceg Demersal 0.01000 3.000 4.692 −0.100 0.150 128.000 0.115 0.303 cod-rock Demersal 0.00001 3.000 3.483 −0.100 0.200 132.000 0.151 0.312 cod-wgr-in Demersal 0.01025 2.979 6.491 −0.100 0.107 150.000 0.065 0.376 dab-2232 Demersal 0.00001 3.000 1.980 −0.100 0.400 32.000 0.497 0.792 dab-nsea Demersal 0.00500 3.140 1.954 −0.100 0.400 36.000 0.455 0.777 GHL-DIS Demersal 0.00333 3.249 9.757 −0.100 0.073 120.000 0.043 0.308 had-iris Demersal 0.00001 3.000 4.697 −0.100 0.160 68.000 0.142 0.330 her-31 Pelagic 0.00360 3.039 2.322 −0.100 0.360 21.000 0.468 0.573 her-nirs Pelagic 0.00360 3.039 2.510 −0.100 0.320 31.000 0.334 0.655 lem-nsea Demersal 0.07560 3.142 2.608 −0.100 0.300 40.000 0.295 0.512
lin-combinSubareasI-II Deep 0.00390 3.074 5.446 −0.100 0.128 150.000 0.097 0.227
lin-combotherareas Deep 0.00390 3.074 5.225 −0.100 0.136 119.000 0.102 0.309
meg-4a6a Demersal 0.00001 3.000 6.455 −0.100 0.120 54.000 0.116 0.299 mut-comb Demersal 0.00440 3.351 4.204 −0.100 0.183 53.340 0.124 0.620 nep-25 Shellfish 0.00001 3.000 4.683 −0.100 0.160 70.000 0.141 0.328 nep-2627 Shellfish 0.00001 3.000 4.931 −0.100 0.150 80.000 0.128 0.301 nep-2829 Shellfish 0.00001 3.000 3.726 −0.100 0.200 70.000 0.176 0.373 nep-30 Shellfish 0.00001 3.000 4.761 −0.100 0.160 60.000 0.145 0.345 nep-31 Shellfish 0.00001 3.000 4.899 −0.100 0.150 85.000 0.126 0.298 ple-2232 Demersal 0.01000 3.000 4.287 −0.100 0.180 52.000 0.169 0.391 ple-7b-c Demersal 0.00001 3.000 6.978 −0.100 0.110 59.400 0.103 0.275 ple-89a Demersal 0.00001 3.000 1.572 −0.100 0.470 54.576 0.605 0.702 ple-celt Demersal 0.00001 3.000 6.983 −0.100 0.110 59.000 0.104 0.280 ple-eche Demersal 0.00001 3.000 3.542 −0.100 0.211 68.500 0.178 0.397 ple-iris Demersal 0.00001 3.000 4.360 −0.100 0.165 100.000 0.134 0.302 pol-89a Demersal 0.00001 3.000 3.844 −0.100 0.190 85.600 0.159 0.342 pol-nsea Demersal 0.00001 3.000 3.928 −0.100 0.186 85.600 0.156 0.342 rng-comb Deep 0.21360 2.987 23.962 −0.100 0.035 28.700 0.015 0.314 san-ns4 Pelagic 0.00001 3.000 1.891 −0.100 0.436 22.000 0.670 0.916 smn-arct Demersal 0.00001 3.000 2.945 −0.100 0.261 49.000 0.248 0.502 smn-con Demersal 0.00001 3.000 2.945 −0.100 0.261 49.000 0.253 0.284 smn-dp Demersal 0.00001 3.000 15.483 −0.100 0.051 49.000 0.051 0.196 smr-5614 Demersal 0.00001 3.000 12.886 −0.100 0.060 59.000 0.062 0.183 sol-7b-c Demersal 0.00001 3.000 6.003 −0.100 0.130 49.800 0.120 0.327 sol-8c9a Demersal 0.00001 3.000 6.978 −0.100 0.110 59.400 0.103 0.275 tur-nsea Demersal 0.00001 3.132 2.576 −0.100 0.290 61.200 0.257 0.385 whg-7e-k Demersal 0.00001 3.000 3.441 −0.100 0.218 65.000 0.188 0.401 whg-89a Demersal 0.00001 3.000 3.928 −0.100 0.190 70.000 0.167 0.363 whg-rock Demersal 0.00001 3.000 1.241 −0.420 0.483 45.000 0.864 0.676 whg-scow Demersal 0.00001 3.000 7.019 −0.100 0.110 56.300 0.105 0.282 wit-nsea Demersal 0.00170 3.390 2.970 −0.100 0.260 47.000 0.220 0.505
Forallcasesindependentlog-normalerrorswithaCVof20%
wereusedand250simulationsranforeachcase(combinationof
stock,scenarioandHCR).
2.2. Harvestcontrolrules
TheHarvestControlRulestestedinthispapertaketheform:
Cy+1=Cy−1·˛
whereCarecatchesinweightandyindexesyears.
Differentformsfor˛,thecatchmultiplier,leadtoalternative
HCRs.Threerulesweretested,onebasedonshorttermtrendsin
surveys(usedfrequentlybyICEStoprovidecatchadvicefor
data-limitedstocks);analternativethatusestheconfidenceintervalof
themeanabundanceofthesurvey,totakeintoaccountlongterm
information;andathirdthatuseslength-basedreferencepoints.
Therulebasedonshorttermchangesinabundance,hereafter
referredtoasHCR1,comparesthetwomostrecentindexvalues
withthethreeprecedingvaluesandtakesthefollowingform:
˛=
y−1 i=y−2Ii/2 y−3 i=y−5Ii/3 ,where I refers tothe stock biomass or abundanceindex and i
indexesyears.
Therulebasedonsurveyconfidenceintervals,namedHCR2,sets
˛as: ˛=
⎧
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎩
˛l ifIy−1<I+zlow√nI I 1 ifI+zlow√I nI ≤Iy−1≤I+zupp I √n I ˛u ifIy−1>I+zupp√nI I,Fig.1.LifehistoryinvariantM/Kforthestocksusedinthisstudy.
Fig.2.Maturity,fleetselectivityandsurveyselectivitymodelsbyageusedinthis study.Fleetselectivitywasmodeledwithadoublenormaldistributionwiththe modeequaltotheageof50%maturity.Surveyselectivitywasmodeledasan asymp-toticallyflatcurvewithmaximumretentionatage10.Bothwereheldconstantover time.Exampleshownforastockwithageof50%maturityof5.4years.
withIthemeanindexabundance,Itheindex’sstandard
devi-ation,nIthelengthoftheindextime-series,andzxthez-statistic
fromthestandardnormaldistributionforwhichP[Z≤zx]=x.zlow
andzuppdefinetheconfidenceintervallimits,whichdon’thaveto
besymmetric,and˛land˛uthecatchmultiplierswhentheindexis
outsidethelowerorupperconfidenceinterval,respectively.Note
thatinthisrulethelengthoftheindexincreaseswithtime,asnew
databecomesavailable,whichintheorymeansthattheprecision
oftheindexmean,I,willincreasewithnewdata.
HCR2wasparametrizedusinganasymmetricconfidence
inter-val,withzlow=z0.33=−0.44andzupp=z0.974=1.96,andasymmetric
changes in TAC, with ˛l=0.75 and ˛u=1.05. For comparison
purposes HCR2was alsoran with symmetric confidence
inter-vals, referred to as ‘HCR2.sym’, where zlow=z0.025=−1.96 and
zupp=z0.975=1.96, and symmetric changes in TAC,˛l=0.75 and
˛u=1.25.
Therulebasedonlength-basedreferencepoints,namedHCR3,
hasthefollowingformfor˛:
˛=LLSQ
F=M
whereFisthefishingmortalityrate,Mthenaturalmortalityrate,
Lreferstoindividuallength,andLSQthestatusquo(current)mean
lengthinthecatch,whichisgivenby:
LSQ=
A a=1Ca,yLa Aa=1Ca,y
withaindexingages.LF=Mreferstothemeanlengthinthecatch
thatsetsfishingmortalityatthesamelevelasnaturalmortality
(BevertonandHolt,1957),givenby:
LF=M=0.75Lc+0.25L∞ (1)
withLc,themeanlengthatfirstcapture,definedby:
Lc=Lac =L∞(1−e−K(ac−a0))
whereL∞,Kanda0arethevonBertalanffygrowthmodel
parame-ters,andactheageoffirstcapture,whichwassetattheageof25%
maturity.
The length-based reference points LSQ and LF=M were used,
respectively, as proxies of current fishing mortality, FSQ,
and the fishing mortality at MSY, FMSY (ICES, 2012c). Thus,
LSQ/LF=M≈FSQ/FMSY,andratiosdifferentfrom1shoulddrivethe
fishery toMSY.Appendix Apresents ageneralized formof this
proxywhereFcanbesettoanyproportionofM.
TheapplicationoftheHCRswithinthemanagementprocedure
wasperformedignoringtheparametersusedinthesimulation,to
maximizetheindependenceofthedecisionmakingprocesswith
respecttothesimulations.Inparticular,thelengthbasedHCRused
thecommonlifehistoryinvariant,K=2M/3,althoughthe
popula-tionsweresimulatedwiththeirownestimatesoftheseparameters.
Thisapproachtriestoreplicatethemostfrequentoptionstakenby
assessmentworkinggroups,whileatthesametimeadding
obser-vationerrortotheindicatorusedforHCR3.
ComputationswerecarriedoutwithR(http://r-project.org)and
FLR(http://flr-project.org;Kelletal.,2007).Thecodeanddatacan
bemadeavailableuponrequesttothefirstauthor.
3. Results
Fig.3showsthemedianSSBtrajectoriesforeachstockover40
yearsbyHCRandexploitationscenarios(developmentand
over-exploitation, named ‘dev’and ‘hi’, respectively). Biologicalrisk,
computedastheratioofSSBovervirginbiomassinyears16-20,
wascategorizedbyiterationas‘low’whenSSB>0.30Bv,‘medium’
Fig.3.Medianspawningstockbiomass(ssb)trajectoriesbystock,scenarioandharvestcontrolrule(HCR).Scenariosarelabeled‘dev’fordevelopment,or‘hi’for over-exploited.
percentageofsimulations(i.eof50stocksby5yearsby250
iter-ations)ineach category reflectsthebiological riskby HCR and
exploitationscenario(Table2).Toevaluatetheperformanceofthe
HCRsinrelationtoyield,thedistributionoftheratiocatch/MSYfor
theyear20,byscenarioandHCRwascomputed(Fig.4).
TheresultsshowthatHCR1isnotabletorecoverthebiomassin
bothscenarios(Fig.3).Theruleresultedinahighbiologicalriskfor
themajorityofthestocks(79%ofthesimulationsinthe‘hi’scenario
and69%inthe‘dev’scenario;Table2).HCRs2and3showedbetter
performanceintermsofSSBrecoveryandbiologicalriskforboth
exploitationscenarios.HCR2showsonly11%ofthesimulationsin
the‘hi’scenarioand5%inthe‘dev’scenariowithhighrisk,while
HCR3shows16%ofthesimulationsinthe‘hi’scenarioand27%in
the‘dev’scenariowithhighrisk.
Fig.4.DistributionofthecatchoverMSYratioinyear20,byscenarioandharvestcontrolrule(HCR).Thedistributionsarecomputedfromthesimulations(250)performed foreachstock(50).Scenariosarelabeled‘dev’fordevelopment,or‘hi’forover-exploited.
Table2
Biologicalrisk,betweenyear16and20,byscenarioandharvestcontrolrule(HCR), computedastheproportionofsimulations(i.e.proportionof50stocksby5years by250iterations)ineachcategoryofbiomasstovirginbiomass(Bv)ratio.The categoriesare:high:<10%Bv,med:10–30%Bv,low:>30%Bv.
Scenario Biologicalrisk HCR1 HCR2 HCR3
dev low 0.01 0.48 0.49 dev med 0.31 0.47 0.24 dev high 0.69 0.05 0.27 hi low 0.00 0.18 0.61 hi med 0.21 0.71 0.23 hi high 0.79 0.11 0.16 Table3
Probabilityofstocksbreakingtheinter-annualcatchconstraint(IACC)of25%, betweenyear16and20,byscenarioandharvestcontrolrule(HCR),computedas theproportionofsimulations(i.e.proportionof50stocksby5yearsby250 itera-tions)above,beloworwithinthe25%catchrangeinrelationtothepreviousyear’s catch. scenario IACC HCR1 HCR2 HCR3 dev below 0.39 0.36 0.38 dev within 0.24 0.24 0.32 dev above 0.37 0.40 0.30 hi below 0.41 0.37 0.36 hi within 0.23 0.23 0.32 hi above 0.36 0.40 0.32
Fig.4showsthatallHCRsdrivemostfisheriestolevelsofcatches
belowMSY,althoughinthecaseofHCR1itisduetolowlevels
ofbiomass,whileforHCR2andHCR3thiseffectisrelatedtolow
levelsoffishingmortality,whichreflectsthemoreprecautionary
approachoftheserules.HCR2seemstobehavebetter,showinga
lowernumberofverysmallcatch/MSYratios.
Table3 presentstheprobability ofviolating aninter-annual
catchconstraint of 25% for years 16-20. AllHCRs have similar
results, although HCR3 shows slightly better results regarding
inter-annualcatchstability.NotethatinthecaseofHCR2these
resultsreflecttheuncertaintyincatchesintroducedthroughthe
implementationerror,giventhefactthatthisrulehasthecatch
limitsembeddedinitsparameters.Additionally,theHCRsoperate
withatwoyearlagononesingleyearcatch(Cy+1=Cy−1·˛);
replac-ingCy−1byanaveragecatchoverarecentperiodmaysmooththe
catchvariability.
To furtherinvestigatethereasons for thepoorperformance
foundinsomecases,weselectedtwostocksofherring(inBothnian
Bay,area31,coded‘her-31’;andintheIrishSea,coded‘her-nirs’)
withsimilarlifehistoryparametersbutwhichperformed
differ-ently.Fig.5showsSSB,catchinweightandthecatchmultipliers
(˛).In thecaseofHCR1themultiplierincreasesassoon asthe
biomassincreases,whichresultsinanincreaseincatcheskeeping
thebiomassatlowlevels.Thisruleclearlystabilizesthebiomass
atlevelssimilartothemostrecentobservedones.HCR2identifies
correctlythatbothstocksareover-exploited,butcatchesarenot
reducedquicklyenoughinitiallytorecoverthebiomassofher-31.
Thisrulehasafixedcatchreductionof 25%.Theopposite
hap-pensforher-nirsandbiomassincreasesaswellasthemultiplier,
yieldinganincreaseincatches.HCR3shows˛above1forher-31,
whichresultsfromwronglyidentifyingthestockasbeing
under-exploited.Thissituationis duetoalowerL∞ (21cmfor her-31
comparedto31cmforher-nirs),andalowerlengthof50%mature,
whichisindirectlydefiningtheageoffirstcapture,ac.Botheffects
resultinalowerLF=Mforher-31which,inthisparticularcase,sets
theratiowiththestatusquolengthabove1.
Fig.6showsacomparisonbetweentheparametrizationusedfor
HCR2inthisstudyandafullysymmetricparametrization(named
‘HCR2.sym’),witha0.05%confidenceintervalandthe˛multipliers
of0.75and1.25,appliedtothestockofplaiceinIberianwaters.SSB,
catchandthe˛multiplierareshown.Themostimportant
differ-encebetweenthetwoisthestabilizationofSSBafterrecoveringin
thecaseofasymmetry,whilethesymmetricHCRdrivesSSBbackto
lowlevels.Thiseffectresultsfromthehighervaluesof˛,thatthis
Fig.5.Trendsinspawningstockbiomass(ssb),catchandcatchmultiplier(˛intheformulations),whichdefinesthemultipliertothecatchinyeary−1thatderivesthe catchadviceforyeary+1,forthestocksofherringintheBothnianBay,area31(‘her-31’),andintheIrishSea(‘her-nirs’).Resultsareforthe‘dev’scenario.
Fig.6.Trendsinspawningstockbiomass(ssb),catchandcatchmultiplier(˛inthe formulations),whichdefinesthemultipliertothecatchinyeary−1thatderives thecatchadviceforyeary+1,forthestockofplaiceinIberianwatersby sce-nario.‘HCR2.sym’reflectsadifferentparametrizationofHCR2,withasymmetric confidenceintervalof5%andmultipliersof0.75and1.25.
parametrizationholds,whichprovideshighercatchesbutalsorisks
forSSB.Theexampleshowsthatthereisawiderangeofoptions
forparameterizingthisHCR,which,inareallifesituation,should
bedoneusingMSEanalysis.
4. Discussion
Inthisstudyweinvestigatedtheperformanceofthreesimple
HCRs,inviewoftheirpotentialapplicationtodata-limitedstocks,
withdifferentdatasituationsregardingtheamountandqualityof
information.HCR1andHCR2canbeusedtoprovidecatchadvice
ifsurveydataexists,whileHCR3canbeusedinmoreseveredata
limitedcaseswhereonlycatchlengthcompositionsareavailable.
Ifbothinformationsourcesexist,combinationsoftheseHCRscan
beexplored.
TheHCRstestedwerenotabletosteersimultaneouslyallstocks
tooptimalexploitation,withSSBsandfishingmortalitiesaround
valuesthatprovideMSY.Suchresultsareexpected,consideringit
isunlikelythatasingleparametrizationcancopewiththe
diver-sityinpopulationlifehistories.Inreal-lifesituations,theseHCRs
shouldbetunedtothespecificitiesofthestockandfisheriessubject
tomanagement.Nevertheless,HCR2withthecurrent
parametriza-tionwasmuchmoreefficientthanHCR1inapproachingMSYlevels,
failinginonecaseonly.HCR3wasabletodriveSSBtosafelevelsin
mostcasesbutleftthefisherieswithlowcatches,mostlikelydue
toapoorapproximationtoFMSYbyLF=M.AsinthecaseofHCR2,
itwasmoreefficientthanHCR1atkeepingthestockswithinsafe
levels.Klaeretal.(2012)exploredtheperformanceofrulessimilar
toHCR3andalsofoundawiderangeofresults,withcaseswhere
theSSBwasdriventoveryhighlevelsinrelationtovirginbiomass
andotherswherethisratiowasverylow.Theauthorsjustifythese
resultswiththedisparitybetweenthenaturalmortalityassumed
bytheassessmentmethodandtherealnaturalmortality.Inour
studywedidn’texplorethiseffect.
NotethatHCR3usesindirectinformationaboutthestatusofthe
stockandlifehistoryparameterstocomputeproxiestoFMSY.Itisa
rulethatcanbeappliedtostocksthathavemoreseveredata
lim-itationsthanHCR1andHCR2,butshowsperformancelimitations,
althoughwiththeadvantageofsteeringthestocktosafebiological
levelsinsteadofstabilizingSSBattheirrecentlevels,likeHCR1.As
inanyotherHCR,add-hocadjustmentscanbeincludedtoimprove
itsperformance,e.g.intheformofadditionaltuningparameters.
Theapproachtosimulationtestingusedcanprovidethe
nec-essaryframeworktotunedatalimited HCRs.Depending onthe
informationavailable,theoperatingmodelwillbeabetterorworse
representation of the real system. Nevertheless, our approach
allows studying the robustness of the HCR to a set of
com-monuncertaintyfactors,suchasstock-recruitmentrelationships,
growth,fleetselectivity,implementationerror,etc.
Analysing theresults obtainedin more detail, one can
con-cludethatHCR1,whichwasdesignedtoprovidestatusquocatch
adviceadjustedbythebiomassindexratio,showedthepoorest
performance,decreasingorkeepingSSBatlowlevelsandcatches
mostlybelowtheMSYtarget.Thisruleadjuststhecatchbasedon
short-termchangesinbiomass,whichprovesefficientat
stabiliz-ingthecatchesandthebiomassesatrecentlevels,butisunable
toimplementthenecessarycatchreductionthatwouldsteerSSB
tosustainablelevels.Improvedperformanceoftherulemaybe
achievedbyreplacingthedenominatorbyabiomassindextarget,
orincludingapenalizingtermaccountingfortheover-exploitation
ofthestock.Thesetweaksrequireassumptionstobemadeabout
therelationshipbetweentheindexandthestateofthestock.Such
assumptionsmaybepossible,ifthesurveytimeseriescoversa
knownperiodofmoderateexploitation,butwillbeoflimited
inter-estifthesurveytimeseriesisshortoronlycoversaperiodwhen
thestockwasalreadyover-exploited.
HCR2,thatadjuststhestatusquocatchbasedonthevariabilityof
thebiomassindextimeseries,showedthebestoverallperformance
inbothscenarios.However,toobtainsuchresults,itwasnecessary
toapplyanasymmetricconfidenceintervalcombinedwithlarger
decreasesinstatusquocatch(0.75)thanincreases(1.05).Theneed
toparametrizethisHCR maybealimitationtoitsapplicability.
Theruleofthumbisthatifthestockislikelytobeover-exploited,
thenanasymmetricalparametrizationwillbemore
precaution-arythanonethatissymmetrical.Thelevelofasymmetrythatwill
increasethechancesofsteeringthefisherytoMSYmustbefound
throughsimulationtesting,althoughtheparametrizationprovided
hereappearedtobeprecautionary.
HCR3uses themeanlengthinthecatch(LSQ)asa proxyfor
currentfishingmortalityandthemeanlengthwhenfishing
mor-talityequalsnaturalmortality(LF=M)asaproxyforFMSY.Therule
isdesignedtoadjustthestatusquocatchdownwardsorupwards
whenthemeanlengthinthecatchisbeloworabovethetarget
(LF=M),respectively.Thesimulationsshowedthatforthemajority
ofthestocks,HCR3resultedinareductionofFtobelowFMSY,which
hadtheeffectofdrivingcatchesofthesestocksbelowMSY(Fig.4).
Inthesecases,LF=MwasapoorapproximationtoFMSY,resulting
in acatchmultiplier(˛)wellbelow one.A commonfeatureof
exploitation(lengthatfirstcapture,Lc)startswellbelowthelength
offirstmaturity(L50%).Infact,forthestocksfailingwithHCR3,the
ratioLc/L50%wasbelow0.25.Moreover,SSBrecoverywasweaker
orstockdepletionmoresevere,forstockswithlatematurity,i.e
whenthelengthatfirstmaturityisclosertomaximumlength.
Fur-thersimulationtestingisrequiredtoinvestigatemodificationsof
HCR3thatalignthetargetmeanlengthinthedenominator,which
dependsonLcandL∞,withtheexploitationandlife-history
char-acteristicsofthestocks,inordertoreducetheadvisedcatchand
promotetherebuildingofthestock.Onepossibilityworth
explor-ingistoaddacorrectionfactortoLF=MlinkedtotheLc,L50%andL∞
forthegivenstocktoimprovetheyield/risktrade-off.Inspiteof
thelowcatches,HCR3wasabletoreverse,relativelyquickly,the
decreasingtrendinSSBinthe‘dev’scenarioandtoincreasethelow
SSBlevelsinthe‘hi’scenario,resultinginlowbiologicalrisks.
Finally,theapproachpresentedheredoesnotreplaceanalytical
assessments.Theinformationprovidedbythesemethodsismore
limitedthantheinformationgivenbyananalyticalmethod.Itis
importanttoavoidthe“data-poortrap”,whereduetothefactthat
catchadviceisbeinggivenforthesestocks,thecollectionofdata
isseenasunnecessary(Bentley,2014).Movingthestocksupon
the“datalevel”isanimportantobjectivethatshouldbekeptin
managementpolicies.
Acknowledgements
Theauthorswouldliketothankthethoroughreviewsoftwo
anonymousreviewers.Part ofthisworkwascarriedoutwithin
GesPeProject(PlanosdeGestaoPesqueira,PROMAR
31-03-01-FEP-0017,EUco-financed).
AppendixA. GeneralizingthelengthbasedHCR
NotethataninterestingdevelopmentaboutHCR3canbemade.
FromBevertonandHolt(1957)themeanlengthfishintheannual
catchisdefinedby ¯Ly=L∞
1− (F+M)(1−e−(F+M+K)) (F+M+K)(1−e−(F+M))e−K(ac−a0) .Considering =a−ac the fishable life-span, then setting
→+∞intheequationaboveresultsin
¯Ly=L∞
1−(F+M)e−K(ac−a0)
F+M+K
. (A.1)
From Ly=L∞(1−e−K(a−a0)) and taking into account that
a=ac⇒Ly=Lc,weobtain
Lc=L∞(1−e−K(ac−a0)),
whichcanbewrittenas
e−K(ac−a0)= L∞−Lc
L∞ . (A.2)
WecannowsubstituteEq.(A.2)intoEq.(A.1)toobtain
¯Ly=KL∞F++FLMc++KMLc.
SettingF=M,with>0andK=Mresultsin
LF=M,K=M=L∞++Lc(+1+1). (A.3)
Whichisageneralizedformthatallowssettingtheproxyfor
FMSY atanylevelofnaturalmortalityandnotonlyF=M.Eq.(1)
holdsifweset=1and=2/3inEq.(A.3).
References
Anon., 2006. Magnuson-Stevens Fishery Conservation and Management Reauthorized Act. Public Law 109-479. NOAA http://www.nero.noaa.gov/ sfd/MSAamended20070112FINAL.pdf
Anon.,2007.CommonwealthFisheries HarvestStrategy:PolicyandGuidelines. Australian Government,Department ofAgriculture, Fisheries and Forestry
http://www.daff.gov.au/ data/assets/pdf file/0004/397264/hsp.pdf
Anon.,2008.HarvestStrategyStandardforNewZealandFisheries.NewZealand MinistryofFisheries.CircularNo.1086http://www.fish.govt.nz/NR/rdonlyres/ 487988D0-F768-4297-ADDE-B5E1DFA53404/0/harveststrategyfinal.pdf
Anon.,2013.Regulation(EC)no1380/2013oftheEuropeanParliamentandof theCouncilof11December2013ontheCommonFisheryPolicy.European Commission.OfficialJournaloftheEuropeanUnionhttp://eur-lex.europa.eu/ LexUriServ/LexUriServ.do?uri=OJ:L:2013:354:0022:0061:EN:PDF
Bentley, N., 2014. Data and time poverty in fisheries estimation: poten-tial approaches and solutions. ICES J. Mar.Sci., http://dx.doi.org/10.1093/ icesjms/fsu023.
Beverton,R.J.H.,Holt,S.J.,1957.OntheDynamicsofExploitedFishPopulations, vol.19.Springer-Science+BusinessMedia,B.V.FishandFisheries.
Butterworth,D.S.,Cochrane,K.L.,DeOliveira,J.A.A.,1997.Managementprocedures: abetterwaytomanagefisheries?TheSouthAfricanexperience.In:Pikitch,E.K., Huppert,D.D.,Sissenwine,M.P.(Eds.),GlobalTrends:FisheriesManagement. AmericanFisheriesSociety,Bethesda,MD,pp.83–90.
Butterworth,D.S.,Punt,A.E.,1999.Experiencesintheevaluationand implementa-tionofmanagementprocedures.ICESJ.Mar.Sci.56,985–998.
Cooke,J.G.,1999.Improvementoffishery-managementadvicethroughsimulation testingofharvestalgorithms.ICESJ.Mar.Sci.56,797–810.
Geromont, H.F., Butterworth,D.S., 2014. Generic management procedures for data-poorfisheries:forecastingwithfewdata.ICESJ.Mar.Sci.,http://dx.doi. org/10.1093/icesjms/fst232.
Gislason,H.,Daan,N.,Rice,J.C.,Pope,J.G.,2010.Size,growth,temperatureandthe naturalmortalityofmarinefish.Fish.Fish.11,149–158.
Hilborn,R.,Maunder,M.,Parma,A.,Ernst,B.,Payne,J.,Starr,P.,2003.COLERAINE:A GeneralizedAge-StructuredStockAssessmentModel.UniversityofWashington, SchoolofAquaticandFisheriesSciencehttp://fish.washington.edu/research/ colerain/pdf/coleraine.pdf
Hordyk,A.,Ono,K.,Sainsbury,K.,Loneragan,N.,Prince,J.,2014.Someexplorations ofthelifehistoryratiostodescribelengthcomposition,spawning-per-recruit, andthespawningpotentialratio.ICESJ.Mar.Sci.,http://dx.doi.org/10.1093/ icesjms/fst235.
ICES,2012a.ICESimplementationofadvicefordata-limitedstocksin2012inits 2012advice(ICESCM2012/ACOM:68),42pp.
ICES,2012b.Reportoftheworkshoponthedevelopmentofassessmentsbasedon life-historytraitsandexploitationcharacteristics(WKLIFE),13–17February, Lisbon,Portugal(ICESCM2012/ACOM:36),140pp.
ICES, 2012c. Report of the workshop to finalize theICES data-limited stock (DLS)methodologies documentation inan operationalform for the2013 adviceseasonandtomakerecommendationsontargetcategoriesfor data-limitedstocks(WKLIFEII),20–22November,Copenhagen,Denmark(ICESCM 2012/ACOM:79),46pp.
ICES,2013a.GeneralContextofICESAdvice.InReportoftheICESAdvisory Commit-tee(Book1).
ICES, 2013b.Report ofthe workinggroup onmethods forfish stock assess-ment (WGMG) , 30 September–4 October, Reykjavik, Iceland (ICES CM 2013/SSGUE:09),130pp.
Kell,L.T.,Mosqueira,I.,Grosjean,P.,Fromentin,J.-M.,Garcia,D.,Hillary,R.,Jardim,E., Mardle,S.,Pastoors,M.A.,Poos,J.J.,Scott,F.,Scott,R.D.,2007.Flr:anopen-source frameworkfortheevaluationanddevelopmentofmanagementstrategies.ICES J.Mar.Sci.64(4),640–646.
Kell,L.T.,Pilling, G.M.,Kirkwood,G.P.,Pastoors,M.,Mesnil,B.,Korsbrekke,K., Abaunza,P.,Aps,R.,Biseau,A.,Kunzlik,P.,Needle,C.,Roel,B.A.,Ulrich-Rescan, C.,2005.Anevaluationoftheimplicitmanagementprocedureusedforsome ICESroundfishstocks.ICESJ.Mar.Sci.62,750–759.
Klaer,N.L.,Wayte,S.E.,Fay,G.,2012.Anevaluationoftheperformanceofa har-veststrategythatusesanaverage-length-basedassessmentmethod.Fish.Res. 134–136,42–51.
MacCall,A.D., 2009. Depletion-corrected averagecatch: a simpleformula for estimatingsustainable yieldsin data-poorsituations. ICES J.Mar. Sci. 66, 2267–2271.
Martell,S.,Froese,R.,2013.AsimplemethodforestimatingMSYfromcatchand resilience.Fish.Fish.14,504–514.
Myers,R.A.,Bowen,K.G.,Barrowman,N.J.,1999.Maximumreproductiverateoffish atlowpopulationsizes.Can.J.Fish.Aquat.Sci.56(12),2404–2419.
Punt,A.E.,Donovan,G.P.,2007.Developingmanagementproceduresthatarerobust touncertainty:lessonsfromtheInternationalWhalingCommission.ICESJ.Mar. Sci.4(64),603–612.
Rosenberg,A.A.,Fogarty,M.J.,Cooper,A.B.,Dickey-Collas,M.,Fulton,E.A., Gutiér-rez,N.L.,Hyde,K.J.W.,Kleisner,K.M.,Kristiansen,T.,Longo,C.,Minte-Vera,C., Minto,C.,Mosqueira,I.,Osio,G.C.,Ovando,D.,Selig,E.R.,Thorson,J.T.,Ye,Y., 2014.Developingnewapproachestoglobalstockstatusandfishery produc-tionpotentialoftheseas,Tech.Rep.CircularNo.1086.FAOhttp://www.fao.org/ docrep/019/i3491e/i3491e.pdf
Wetzel,C.R.,Punt,A.,2011.Modelperformanceforthedeterminationofappropriate harvestlevelsinthecaseofdata-poorstocks.Fish.Res.110,342–355.