UBC Faculty Research and Publications

[Methods for Evaluating the Impacts of Fisheries on North Atlantic Ecosystems] Pauly, D. (Daniel); Pitcher, Tony J. 2000

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-Pauly_D_et_al_Methods_evaluation_ecosystems.pdf [ 3.84MB ]
JSON: 52383-1.0348147.json
JSON-LD: 52383-1.0348147-ld.json
RDF/XML (Pretty): 52383-1.0348147-rdf.xml
RDF/JSON: 52383-1.0348147-rdf.json
Turtle: 52383-1.0348147-turtle.txt
N-Triples: 52383-1.0348147-rdf-ntriples.txt
Original Record: 52383-1.0348147-source.json
Full Text

Full Text

iiAbstractThe contributions in this report stem from aworkshop held in April 2000 to review themethodology deployed by the research team ofthe Sea Around Us Project.  This project, fundedby The Pew Charitable Trusts, Philadelphia,USA, is designed to provide an integratedanalysis of the impacts of fisheries on marineecosystems, and to devise policies that canmitigate and reverse harmful trends whilstensuring the social and economic benefits ofsustainable fisheries. The data–rich NorthAtlantic was selected as the target area for casestudies to be conducted in the first two years ofthe project, with other areas to follow insubsequent years. The methodology deployed bythe project includes: (1) the development of aspatially explicit catch and effort informationsystem that allows in-depth analysis of fisheriescatches for various large marine ecosystems, i.e.,reported landings, nominal catches, unreportedcatches, misreported catches, and discarded by-catch, sorted by species and sector; (2) thequantification of the biological and economicimpacts of the present fishing trends or a changethereof on the ecosystems, with reference to pastecosystems reconstructed from time series ofscientific data and the Ecopath with Ecosimsoftware; (3) the quantitative evaluation of thestatus of fisheries by sector, gear type andlocation using a robust and simple system ofrapid appraisal (Rapfish) that may be applied topast, present and alternative future fisheries; (4)approaches for scaling all results to a basin-widescale; and (5) quantification of the economic andother benefits to be gained from re-establishinghealthy ecosystems, relative to the lossesexpected from a continuation of the status quo.An important feature of the methodologyassembled to meet these requirements is that itdoes not compete with the elaborate single-species methodology conventionally applied tothe management of fisheries, and whichgenerally pertain to geographic and temporalscales much smaller than the basin-wide scaleconsidered by the Sea Around Us Project.Daniel PaulyPrincipal InvestigatorSea Around Us ProjectProfessor of FisheriesUBC Fisheries CentreiiiDirector’s ForewordThe Fisheries Centre at the University of BritishColumbia supports research that first clarifies,and then finds ways to mitigate, the impacts offisheries on aquatic ecosystems.  Only with suchinsight of how whole aquatic ecosystemsfunction can management policies aim toreconcile the extraction of living resources forfood with the conservation of biodiversity, withthe maintenance of ecosystem services, withamenity and with other multiple uses of aquaticecosystems.  Indeed, the present dire state ofmarine ecosystems and their fisheries around theglobe signals a pressing need for what may betermed the “ecosystem imperative.”Although ecosystem agendas of this kind haverecently become embodied in the legislativegoals of many nations, and are an integral part ofthe FAO Code of Conduct for ResponsibleFisheries, in practice there have been fewattempts to work out how it might actually bedone.  In sponsoring the Sea Around Us Project,the Pew Charitable Trusts of Philadelphia, USA,have devoted a significant amount of funding toan ambitious pilot project that focusing on theNorth Atlantic that aims to address thisquestion.  A research team of senior scientists,postdoctoral research assistants, graduatestudents, consultants and support staffcommenced work in late 1999.Members of this team are excited and challengedby the unprecedented scope of the researchwork.  Moreover, most of the methods used totackle the problem magnitude are new.  It seemsthat in concentrating on the perfection ofquantitative methods that set catch quotas forlarge heavily-industrialized fisheries, traditionalfisheries science has avoided trying to addressecosystem-based questions since the days of thepioneers in the early 20th century.This report presents the edited output of aworkshop held in May 2000 that examined themethodological bases of the research for the SeaAround Us Project.  Each of the papers has beensubjected to peer review by at least two referees,and has been scrutinized before and during theworkshop by a visiting team of experts from FAOand major fisheries management agencies inCanada, the USA and Europe (See Appendix 2).Summary comments about the project and itsmethods from these visiting experts are reportedin Appendix 3.  The Sea Around Us researchteam are especially pleased with the overallsupport that the project is receiving from FAO,DFO, ICES and others.The report is the latest in a series of FisheriesCentre Research Reports published by the UBCFisheries Centre.  A full list is shown on our website at www.fisheries.ubc.ca, and the series isfully abstracted in the Aquatic Sciences andFisheries Abstracts.  The research report aims tofocus on broad multidisciplinary problems infisheries management, to provide a synopticoverview of the foundations and themes ofcurrent research, to report on research work-in-progress, and to identify the next steps and waysthat research may be improved.Edited reports of workshops reported inFisheries Centre Research Reports aredistributed to all project or workshopparticipants.  Further copies are available onrequest for a modest cost-recovery charge.Please contact the Fisheries Centre by mail, faxor email to office@fisheries.ubc.ca.Tony J. PitcherProfessor of FisheriesDirector, UBC Fisheries CentreAssessment and Mitigation of Fisheries ImpactsivPreface and AcknowledgementThe contributions included in this reportoriginate from a workshop held from April 1st to5th, 2000 at Dunsmuir Lodge, Sydney, VancouverIsland, B.C., and devoted to reviewing themethodology to be deployed by the research teamof the Sea Around Us Project.This project, fully funded by the Pew CharitableTrusts, Philadelphia, USA, is designed to providean integrated analysis of the impacts of fisherieson marine ecosystems, and to device policies thatcan mitigate and reverse harmful trends whilesensuring the social and economic benefits ofsustainable fisheries.  The data-rich NorthAtlantic was selected as the target area for caststudies to be conducted in the first two years ofthe project, with other areas to follow insubsequent years.The Sea Around Us Project aims to collate andanalyze catch and ecosystem information usinganalytical tools being developed at the FisheriesCentre, in partnership with a global network ofscientists providing data, evaluating and peerreview.  These elements are required indeveloping strategies and action plans to managemarine ecosystems.Thus, the methodology deployed by the projectincludes:1. The development of a catch and effortinformation system that allows in-depthanalysis of fisheries catches for eachecosystem, i.e., reported landings, nominalcatches, unreported catches, misreportedcatches, discarded by-catch, kill by ghost-fishing, sorted by species and sector;2. The quantification of the biological andeconomic impacts of the present fishingtrends or a change thereof on theecosystems, with reference to pastecosystems reconstructed from time series ofscientific data;3. The quantitative evaluation of the status offisheries by sector, gear type and locationusing a robust and simple system of rapidappraisal that may be applied to past,present and alternative future fisheries;4. Approaches for scaling all results to a basin-wide scale;5. Quantification of the benefits to be gainedfrom re-establishing healthy ecosystems,relative to the losses expected from acontinuation of the status quo.An important feature of the methodologyassembled to meet these requirements is that itdoes not compete with the elaborate single-species methodology conventionally applied tothe management of fisheries, and which generallypertain to geographic and temporal scales muchsmaller than those considered by the Sea AroundUs Project.  Thus, we were able to build on theresults of traditional approaches in fisheriessciences to derive our methodology, which wehope will be seen as complementary to traditionalapproaches.In fact, the Sea Around Us Project has muchprogressed since the workshop documented herewas held, and already, some of the methods inthis report have been modified after they wereapplied to a wide range of concrete situations.Interested readers are advised therefore toconsult the project web page (atwww.fisheries.ubc.ca/projects/SAUP) for currentversions, and sample results.We conclude by thanking the Pew CharitableTrusts for their support of the Sea Around UsProject.  Thanks are also due to the dedicated staffof the Sea Around Us Project, and to our panel ofinvited experts:  Lee Alverson, Kevern Cochrane,Poul Degnbol, Paul Fanning, Richard Graingerand Jay Maclean.We are most grateful to the following externalreferees for providing their comments in a timelyand insightful manner:  Ragnar Arnason, TrondBjorndal, John Blaxter, Cutler Cleveland, MichaelFogarty, Kenneth Frank, Quentin Grafton,Normal Hall, Rognvaldur Hannesson,Paul Hart, Simon Levin, Pamela Mace, PaulMedley, Leif Nottestad, David Pimentel, DavidRamm, Saul Saila and Michael Sinclair.Daniel Pauly and Tony PitcherSeptember 2000.Assessment and Mitigation of Fisheries Impacts1ASSESSMENT AND MITIGATION OFFISHERIES IMPACTS ON MARINEECOSYSTEMS:A MULTIDISCIPLINARY APPROACH FORBASIN-SCALE INFERENCES,APPLIED TO THE NORTH ATLANTIC1D. Pauly and T.J. PitcherFisheries Centre, University of British Columbia,Vancouver, Canada, V6T 1Z4ABSTRACTThe aim of the Sea Around Us Project is to quantify,in ecological and economic terms, the impact offisheries on the marine ecosystems of the NorthAtlantic, and to evaluate the costs and benefits ofvarious scenarios of mitigation, such as status quo,rebuilding of depleted resources andimplementation of closed areas. Dealing with theseissues requires a methodological package related to,but different from, that typically used in fisheriesmanagement, notably because of its ecosystem focusand the much larger temporal and spatial scales,relative to standard fisheries assessments. Thispaper summarizes the methodology deployed by theproject by introducing a suite of papers in which itsrationale and operational details are provided.First, we review the relationships between scale andmethodology choices in marine science.  Then, theprinciple modules of the Sea Around Us Projectmethodology are described as follows:1) The North Atlantic as study area, where wereport a new ecosystem classification schemethat is compatible hierarchically with previouswork and with all statistical divisions;2) North Atlantic fisheries catches in time andspace, where we present the project’s catch andeffort database, discuss the problems inestimating total extractions, and outlinemethods used to overcome them;3) Fish distribution transects, where the biologyand migrations of key commercial NorthAtlantic species are used to link catches byshallow-water and offshore fisheries;4) Bio-economic analyses of fisheries sectors,where the effect of competition between smalland large –scale fisheries is quantified using amulti-species, multi-gear yield per recruit                                                       model and the combination of effort producinga Nash equilibrium is identified;5) Ecosystem modeling, discussing the use ofECOPATH with ECOSIM and ECOSPACE torepresent present and past North Atlanticecosystems with their embedded fisheries, toevaluate ecosystem status, and to simulate likelyresponse to change;6) Evaluating alternative ecosystem-basedmanagement regimes to quantify the benefits ofdifferent ecosystem-based managementscenarios;7) Energy consumption and the ecologicalfootprint of North Atlantic fisheries, to contrastthe energy incorporated in landed fishes to thatrequired to catch them;8) Rapid interdisciplinary appraisal of fisheriesstatus and compliance analyses using RAPFISH,to compare and characterize North Atlanticfisheries in terms of their sustainability (inecological economic technological and socialfields), analysis of their ethical status, and toscore their compliance with the FAO Code ofConduct for Responsible Fisheries, togetherwith the compliance of North Atlantic countriesvis-à-vis their internationally agreedcommitments.9) Mapping the fate of fisheries landings from theNorth Atlantic, to identify possible pressurepoints for intervention by fish productconsumers;We present a diagram expressing the articulation ofthe various methodological components listedabove. The synthesis to emerge from integrating theresults of these modules may contain manysurprises, both in terms of the ecological damageand economic waste presently generated by theNorth Atlantic fisheries, and in clarifying theforegone benefits that could be regained, were theseeconomic and ecological issues to be addressed.Sea Around Us Project Methodology Review2INTRODUCTIONThe task of the Sea Around Us Project, funded bythe Pew Charitable Trusts, Philadelphia, andexecuted at the Fisheries Centre, University ofBritish Columbia, Vancouver, is to provide asynthesis of the impacts of fisheries on marineecosystems of the North Atlantic. More precisely,the questions to be answered are:1. What are the total fishery catches from theecosystems? Total fishery catches includesboth reported and unreported landings anddiscards at sea.2. What are the biological impacts of thesewithdrawals of biomass for the remainingliving components of the ecosystem?3.  What would be the likely biological andeconomic impacts of a continuation ofcurrent  fishing trends (i.e., a maintenanceof the status quo)?4.  What were former states of this ecosystemlike before the expansion of large-scalecommercial fisheries?5.  How does the present-day ecosystemevaluate on a scale from ‘healthy’ to‘unhealthy’?6.  What specific policy changes andmanagement measures should beimplemented:(a) to avoid continued worsening of thepresent situation?(b) to improve ecosystem ‘health’, asdefined in (5)?Each of these questions, though straightforward-looking at first, leads to further questions, manyseemingly without answers. Nevertheless, theproject staff has developed a ‘methodology package’for providing the best possible answers to thesequestions. This package differs from that normallyused to assess local fish populations and localfisheries in that our methods are scalable to theentire North Atlantic basin, and indeed, eventually,to the world ocean. This package therefore,emphasizes aspects of fisheries and other marinescience that are usually given short thrift in localstudies. Conversely, we do not attempt to assess theexploitations status of exploited single-species fishpopulations. As we shall attempt to demonstrate,methods concerned with local or single-speciesstudies and those in our methodology packagesupport and complement each other.Before we present the various elements of thismethodology package, we shall briefly contrast twoviews of (marine) sciences, and provide reason why,given the present, much depleted state of NorthAtlantic fish populations, and the ruinous state ofthe fisheries depending thereon, we have chosen toidentify with one of these views.Two views of (marine) sciencesOur reading of the history of science in general, andmarine science in particular suggests two basic waythat advances are made:1. Through what, for lack of a better term, weshall call Smart New Tricks (SNT), or2. Through assimilation of large sets of pre-existing data, and, based thereon, throughthe creation of New Mental Maps (NMM).Examples of SNT in fisheries were the invention ofVirtual Population Analysis (usually attributed toGulland, 1965), or of Bayesian risk analysis(reviewed by Punt and Hilborn 1997). SNT usuallyresolve one problem (often one that was not evenperceived as such), and do this in a new way that isoften regarded as ‘neat’ or ‘elegant’. On the negativeside, we should add that SNT can also be seen as‘techno-fixes’, resolving the technological aspect of aproblem but usually leaving wide open theunderlying process that generated the problem. Inthe case of the two examples above, the problemswere how to estimate fishing mortality, and how topresent management options to politicians,respectively. Their downside as techno-fixes wasthat the former quickly bred a misplaced confidencein its outputs (see Walters and Maguire 1996,Pitcher and Hart 1982), whilst the latter, eventhough labelling them as such, provided ultra-riskyoptions to industry and politicians (Mace 2000),decision-takers who, by the nature of theirprofessions, tend to prefer risky options to saferones.The alternative to the SNT, the NMM cansometimes build on one or several small SNT. Theimportant feature of the NMM, however, is that itinvolves the assimilation (or meta-analysis) of large(sometimes enormous) data sets. Our best exampleis the realization by U.S. Navy Commander MathewF. Maury, in the mid-1800s, that marinerscollectively held in their head enough informationon currents and winds (e.g., in the North Atlantic),to generate maps which would improve navigation,i.e., shorten the route between Europe and theAmericas (Maury 1963).Maury thus promised cooperating mariners copiesof his planned maps, should they agree tocontribute their individual knowledge on mostAssessment and Mitigation of Fisheries Impacts3favorable routes.  These data (and depth soundingshe also gathered) enabled him not only to produce,after lots of painstaking work, the best navigationmaps then in existence (‘applied’ science), but alsoto be the first to perceive the existence of mid-Atlantic ridge (‘basic’ science). Moreover, single-handedly created the mode of interactions betweenmariners and naval offices that still prevails, andwhich has enabled the emergence of modernphysical oceanography as a discipline wherein dataare shared. Hence the existence and collaboration,even during the coldest years of the Cold War, ofData Center A (in Washington, D.C.) and B (inMoscow). This, incidentally, is also the reason whyoceanographic data can be used to verify theoccurrence of global changes: the data are availablesince the late 19th Century.Which brings us to marine biology and fisheries.Here, like Maury since the end of the 19th Century,we inherit a mountain of data on the variousorganisms, from phyto-plankton (net samples, C14measurements, satellite oceanography) andzooplankton (Hensen nets samples, Hardy samplerstime series), trawl and benthic surveys, catch timeseries, landing and price data, etc. – an enormous,ever-growing data set. Yet we are very often told byfisheries scientists and others that there are “nodata” upon which to make inferences about the stateof North Atlantic ecosystems, and on remedialactions regarding their depletion, and on the futureof the commercial species therein. We are told thatwhat we need is ‘new, better data’, or indeed that weshould hope for a SNT to somehow resolve theproblem(s) that led to the mountain of data beingaccumulated in the first place.Yet major NMM are based on assimilation ofexisting data, even in areas with which all arefamiliar. Thus, for example, it is relatively wellknown that the report which convinced the USauthorities, and the US public, and later others inother parts of the world, that cigarettes are bad forsmokers did not present a SNT. Rather, it was meta-analysis of a large number of small studies, eachperhaps not very convincing by itself, but jointlyproviding incontrovertible evidence. Further, morerecent meta-analyses added the effects of second-hand smoke, now leading to widespread restraintson smoking in enclosed spaces, both public andprivate. What changed here is the position ofcigarettes in peoples’ mental maps.In a similar way, Rachel Carson’s Silent Springassimilated into a coherent whole a large number ofpreviously unconnected observations, and thiscreated a NMM wherein the location of DDT andother pesticides was radically different from itprevious position (Lear 1997).The Convention of Biological Diversity (CBD)requires that all the countries of the world makeinventories of their biodiversity, and take measuresto protect it. Where does small country X get aglobal reference list of the plants and animals thathave so far been described (and of which the speciesin country X must be a subset)? Such a list still doesnot exist, despite the straightforward nature of thescience that would be required (just as for Maury’smaps).In the late 1980s, work on a large database intendedto provide a rigorous nomenclature andclassification for all the fishes in the world, and keyfacts for each of these 25,000 species. Ten yearslater, the job is largely done (see www.fishbase.org):the countries of the developing world now thus havea tool that enables them to get started on meetingtheir obligations vis-à-vis the CBD, at leastconcerning the fishes (presently, the Internetversion of FishBase gets over half a million visits permonth, several orders of magnitude more than forany comparable product). Moreover, the databasethus created, in a collaborative mode resemblingMaury’s, has many elements serving as model forSpecies 2000, which aims at producing a list of allorganisms so far described (seewww.species2000.org).An excellent example of a meta-analysis is the seriesof contributions by R.A. Myers and collaborators onthe stock-recruitment relationships of fishes, basedon their vast compilation of time series of publishedstock and recruitment time series. This workrecently culminated in Myers et al. (1999) and hasthe potential to produce massive changes in themental maps of fisheries biologists. Myers’ studyshows conclusively that the common feature ofstock-recruitment relationships across species (anarrow range of slopes near the origin, indicative ofa narrow range of reproductive potentials ofindividual female fish) was not seen previouslybecause nobody bothered to standardize, over alarge number of cases, the scales of plots ofrecruitment versus parent stocks. Rather, earlierauthors emphasized the ‘uncertain’, even ‘chaotic’nature of stock-recruitment relationships, entirelymissing what turns out to be highly predictablerelationships. It is as if Maury had complainedabout the ‘complex’ nature of mariners’ knowledge,rather than assemble his maps.These items are examples of NMM, major pieces ofwork that make available to practitioners tools thatassimilate much of the work previously done in agiven area. In each case data was already availablein principle, but was not assimilated within arigorous framework. So we can ask: why are thereSea Around Us Project Methodology Review4not more of these collaborative exercises in marinebiology and fisheries, given their potential impact?One reason might be that, in the context ofgovernment-funded research, such work can bedone only after a consensus has emerged about theresearch to be conducted, the idea being that suchNMM should emerge from the bottom up.  Theproblem here is the tendency for collective andcommittee-led research to reduce new sets of ideas,‘visions‘ as it were, to a least common denominator:voluminously documented research proposalsfavoring safe science over risky new approaches.The methodology package we have assembled toanswer the questions above, related to the impactsof fisheries on the ecosystems of the North Atlantic,thus reflect our vision, not yet widely shared, thatsuch questions can be tackled at basin-wide scales.The methodology is devoted to assimilating, inrigorous, quantitative terms, a large amount ofprevious work and to involving multiplecollaborative arrangements. However, we shallmaintain standards such that coherent productsemerge.Such approach, from the top down is, we believe,the only way products can emerge which are usefulat scales above that at which marine and fisheriesbiologists typically operate, usually that defined bythe boat of a university research station, or by thecommercial vessels used in a fishery under study.The North Atlantic as Study AreaAs defined by the Sea Around Us Project, the NorthAtlantic includes all marine waters North of Miami,Florida in the West and North of Cape Bojador,Morocco in the East. This area is identified in Fig. 1,which also identifies the Biogeochemical Provinces(BGCP), which are compatible with the LargeMarine Ecosystems (Sherman and Duda 1999) ofthe North Atlantic (see below). These articulate, atdifferent levels, the ecosystem classification adoptedby the project (see Pauly et al. 2000). Note that thisdefinition excludes the Mediterranean from thescope of the project. Moreover, for variouspragmatic reasons, we also exclude the Balticproper, though not its connections with the NorthSea, the Kattegat and Skagerrak. Except for aSouthern border a bit further south, and theomission of the Baltic, our definition of the NorthAtlantic thus overlaps with the area jointly coveredby FAO areas 21 (Eastern North Atlantic) and 27(Western North Atlantic), themselves largelyoverlapping with the area for which ICES, andNAFO, respectively, are responsible.The questions posed of the Sea Around Us Project,referring to the ecosystem impact of fisheries,require that we identify the ecosystems of the NorthAtlantic. In the spirit of the foregoing, whichemphasizes the need to assimilate large amount ofpre-existing data, we have adopted, for the SeaAround Us Project, the large Marine Ecosystem(LME) concept and definitions developed in the last15 years by K. Sherman and co-workers, andrecently summarized in Sherman and Duda (1999).This decision was facilitated by the discovery thatthe LMEs so far defined can be easily mapped onto,and re-expressed as components of coastalBiogeochemical Provinces (BCGP), the largerFigure 1. (A) ( top) Map of North Atlantic showing thatSea Around Us Project area (southern boundary is thickhorizontal line) overlaps four major FAO statisticalareas. (B)  (bottom) The nine major biogeochemicalprovinces in the Sea Around Us Project area.ARCT=Atlantic Arctic Province (in two regions); NECS= Northeast Atlantic Shelves Province; SARC = AtlanticSubarctic Province; NADR = North Atlantic DriftProvince; NASE = North Atlantic Subtropical GyralProvince (East); NASW = North Atlantic SubtropicalGyral Province (West); GFST = Gulf Stream Province;CHSB = Chesapeake Bay Province; MEDI =Mediterranean. For further details of how these zonesare conflated with Sherman’s Large Marine Ecosystems,ICES and NAFO management areas, and USA andCanadian statistical zones, using a half-degree squareSea Around Us database, see Pauly et al. (2000).ABAssessment and Mitigation of Fisheries Impacts5ecosystem units proposed by Longhurst (1995,1998) to provide a stratification of the world ocean.Indeed, this redefinition of LME provides the lowerrungs of a hierarchy ranging from ‘biomes’, i.e.,large, circum-terrestrial entities with similar climate(Polar; Westerlies; Trades; and Coastal Boundary)to 56 BGCP and about 80 LME (see Pauly et al.2000). Moreover, the LME themselves can befurther subdivided, especially for modelingpurposes (see Pauly et al. 2000 and Christensen andWalters 2000).This structure for ecosystem classification,proposed as a consensus of several research groupsworking on this type of issue (Pauly et al. 2000),appears well suited for the stratification required forbasin-level estimates of various states and rates andto address the issue of variability of scalesemphasized by Levin (1990). Moreover, using thisscheme, fisheries may be mapped onto theecological entities, the ecosystems, that generate thefish caught, and not the artificial boundaries ofcountries, EEZs, and jurisdictions, our next topic.NORTH ATLANTIC FISHERIES CATCHES IN TIME ANDSPACEAccurate time series of fisheries catches, hereunderstood as all animals killed by fishing gears,and not only those that are landed, are at the heartof the Sea Around Us Project. However, contrary towhat may be believed, assembling such time seriesfor the North Atlantic is not a matter of setting up anew program for sampling primary data in thecountries bordering the North Atlantic. Rather, it islargely a matter of identifying, for each of thesecountries, those elements (if any) that prevent theirofficial catch statistics from reflecting the trueeffects of fishing gears.In many cases, even landings are incompletebecause the data collecting entity is not mandated tocollect data from certain types of gear (often small-scale gear, or sport fishers), notwithstanding thepotential impacts of a large number of such gear.In other cases, obvious sources of biases, notablymassive discarding of by-catch are not considered incompiling catch statistics. This also applies to illegalcatches, even when, as occur in some fisheries, allthose involved – including government scientists -know of their existence, and even their magnitude.Watson et al. (2000) review these and relatedissues, and thereby present the database structureand methodology we shall use to obtain, for theNorth Atlantic, figures that will better reflect truecatches (i.e., all withdrawals) than those presentlyavailable, illustrated by an example of cooperationwith a government agency. Moreover, Pitcher andWatson (2000) explore this issue further, byestimating percentage in each category ofunreported catches, following in time the changes inlegal instruments, including the Law of the Sea, thatprovide disincentives to accurate reporting. Theanalysis is presented such that it can be easilyrefined by further work.However, even the first round of estimates resultingfrom these considerations should contribute tomaking our catch figures more realistic. Thiscontrasts with the assumption of zero in thosecategories, the common default position of publicagencies, and one that is neither useful noracceptable to the public itself.We are well aware that the data set thus assembledwill remain fragmentary and incomplete, and thatfar better data sets will exist on local scales. At theLME and basin-wide scale, however, we expect thatour data set will be the most accurate, in that allsources of fishing mortality will be accounted for.Pauly et al. (2000) present the method by which theglobal FAO fisheries catch data set will be re-expressed on a global LME map. The keycomponent of the method proposed therein is that itwill proceed ‘by subtraction’, i.e., by first assigningfishes with clear affinities to depth ranges, habitattypes and/or certain LME, e.g. the anchovetaEngraulis ringens to the inshore part of theHumboldt Current LME, or the neritic fishesreported for Bangladesh to the shelf component ofthe Bay of Bengal LME, etc., each time subtractingthe assigned fish groups from the database. Severalrounds of subtraction will lead to small amounts ofunallocated landings, pertaining mainly to fishlanded in countries with distant water fleets (orproviding flags of convenience to such fleets).Assigning the residual landings to the LME wherethese fleets are known to occur (see Bonfil et al.1999 and references therein), in proportion to thecatches per half-degree square previously allocated,will be sufficient for a first-pass allocation,especially since misallocations should generatevisible patterns in the maps thus generated.For the North Atlantic, this crude approach can bereplaced by one in which the catch reported byspecies, from distinct ICES or NAFO sub areas isassigned to the half-degree squares in each area as afunction of the mean depth of each square, and theobserved depth distributions in the species inquestion, as plotted on the ‘depth transects’presented below (see also Zeller and Pauly 2000).Here again, misallocations should generate visiblepatterns in the maps thus generated, and thus leadto improvements of the allocation rules.Sea Around Us Project Methodology Review6FISH DISTRIBUTION TRANSECTSAs mentioned above, the Sea Around Us Project willnot attempt to perform assessments of single-species fisheries, and not generally question suchassessments as performed by various colleagues.However, we do require connecting our work withkey aspect of the distribution of major commercialspecies, for two reasons:1) These distributions can help assign catches toareas (see above); and2) The depth and distance from the coast of majorpopulation components determines theirrelative vulnerability to coastal (often small-scale) and offshore (often large-scale) gear andhence the existence and intensity of interactionsand (potential) conflicts between these differentfisheries.The format we have developed for these transectfulfils these requirements by integrating the keyinformation on the distribution and migration offish in a single graph (see Zeller and Pauly 2000).Using such a graph, catches of both small- andlarge- scale fisheries, both inshore and offshore, canbe partitioned and their impacts evaluated.Bio-economic analyses of fisheries: small vs. largeFew, if any studies have quantified the economicrent lost from competition between the large- andsmall-scale sectors of a fishery.  Here we havechosen an approach with three important keyfeatures:1) Easily scalable from local fisheries to theentire North Atlantic;2) Provides management alternatives byemphasizing, where possible, thesubstitutability of large by small scalefisheries (and vice versa);3) Should lead to a reliable estimate ofeconomic losses (waste) due to excesscapacity and non-cooperative behaviorbetween different elements of the fisheriessector (see Nash 1951, 1953).This approach uses a multispecies, multifleet yield-per-recruit analysis to estimate, based on thepresent, calculated recruitment ( = influx of youngfishes and invertebrates to the fishing grounds), thefeatures of a ‘small scale’ and a ‘large-scale’ fleetwhich maximize the gross value of the catches ofboth fleets.  These features are the level of effortrelative to present, and the selection curves of eachgear relative to each species. Then, under theassumptions that the present fisheries are at or neartheir bioeconomic equilibrium point (where totalcosts equal gross total returns; Gordon 1954), andthat fishing mortality scales linearly to fishing cost,we identify the equilibrium point at whichmaximum net returns can be obtained if the fleetsadjusted their fishing mortality such that their jointnet benefit is maximized (Munro 1979; Sumaila1997).The difference from between these optional returnsand the Nash Frontier to the present position of thefleet allows the loss (=economic waste) due to non-cooperation and mismanagement. Finally, wepartition benefits by sector and identify the Nashbargaining solution (Nash (1953) associated withthe equilibrium point (Binmore 1982).  Thisprocedure can be applied successively to a largesample of representative North Atlantic fisheries,thus yielding, by addition, an overall estimate ofeconomic losses, and, more importantly of theeconomic gains that would result from improvedmanagement (Ruttan et al. 2000). We expect thesenumbers to be very large, especially when scaled upto our reference area, through the ratio of the sumof all catches in the sample fisheries to the totalNorth Atlantic catches.The Achilles’ heel of this approach is, of course, theassumption that for each fishery, relativerecruitment, as obtained by dividing yield perrecruit into average catches, will remain constantwhile fishing mortality varies. We note, however,that the approach we propose will tend to associatethe Nash equilibrium with levels of fishing mortalitylower than those commonly presently occurring inreal fisheries (which tend to suffer from growthoverfishing). This implies that recruitment would beassumed to remain constant over a small range of F-values only.Moreover, the proposed method will treat eachfishery independently from the others, usingdistinct mixes of species, each with their own sets ofrelative recruitment, and growth and selectionparameters. Thus, given the Central Limit Theorem,our global estimate of economic loss will tend to beaccurate, even if the estimates for certain fisheriesare not.Another aspect of our comparative bioeconomicstudies of small-scale vs. large-scale fisheries is thatthey should provide a framework for evaluatinggovernment policies which purport to benefitemployment, or other social goods: small-scale andlarge-scale fisheries often sharply differ in theAssessment and Mitigation of Fisheries Impacts7employment opportunities or other social benefitsthey provide (see section below on RAPFISH; andAlder et al. 2000).ECOSYSTEM MODELLINGEmbedding the fisheries that generate the catchesand economic returns discussed above intoecosystems will be achieved by constructing at leastone ECOPATH model for each of the LME in theNorth Atlantic. The rationale for ECOPATH asmodeling tool is that it is the only approach so fardemonstrated to be widely applicable for modelingmarine ecosystems, notwithstanding a commonmisunderstanding as to the ready availability ofalternative approaches. Christensen and Walters(2000) review ECOPATH as used in the context of theSea Around Us Project, with emphasis on this andother misunderstandings regarding the capabilitiesand limits of the approach it embodies.Presently, ECOPATH models exist for numerous partsof the world (see Pauly et al., 2000).  However, only20 of these represent ecosystems of the NorthAtlantic basin, hence precluding simple raising ofbiomass flows from ecosystem to basin scales. Thus,a stratification scheme is required, based on thegeographic structure outlined above, which can beused to scale models from the sampling area of thefield data used to parameterize the models to thewider area that is assumed represented by thesesame models.LMEs are seen here as providing the key level forecosystem model construction. For each LME, anECOPATH model will be constructed to describe theecosystem resources and their utilization, and toensure that the total fisheries catch of each LME isused as output constraint (just as their primaryproduction will be used as input constraint). Inaddition, the stratification scheme used must besuch that it can straightforwardly accommodate anynumber of additional ECOPATH models for eachLME. This can be done so as to simultaneouslyaddress the issue of parameter uncertainty, asdescribed in Pauly et al. (2000).The LME ECOPATH models require information onabundance, production and consumption rates anddiets for all ecosystem groupings. Such informationcan be obtained from the following sources:• Abundance, production and consumptionrates, and diets of marine mammals areavailable from the Sea Around Us Projectdatabase for all (117) species of marinemammals and on a seasonal basis;• Fishery catches: available from the spatiallystructured catch database generated asdescribed above (see also Watson et al.2000), and covering all species groups;• Occurrence, biology and ecology of marinefishes: available from FishBase(www.fishbase.org) at LME level for theNorth Atlantic, as a result of cooperationbetween the Sea Around Us Project andFishBase projects.• For marine invertebrates: only limitedinformation (beyond the catches in the FAOdatabase) is available from electronicdatabases, but a variety of publicationsprovide extensive information. Productionrates can be estimated from the well-founded empirical relationships of Brey(1999), now included in ECOPATH;• Primary production estimates:establishment of a global database aimed atsupplying fine grid level satellite basedestimates of primary production ispresently underway through a cooperationbetween the Space Applications Institute,EC Joint Research Centre, Ispra, Italy, andseveral members of the Sea Around UsProject.The LME-level ECOPATH models will serve as thebackbone for addressing issues related to fisheriesimpacts, to derive indices related to ecosystemhealth (Rapport et al. 1998a; 1998b; Costanza andMageau. 1999), to evaluate, using ECOSIM andECOSPACE (Walters et al. 1997, 1999; Walters andChristensen 2000), the likely effects of changes infishing patterns, including setting up of marineprotected areas, and to estimate the expectedeconomic benefits of such interventions.Moreover, these LME-level ECOPATH models,representing the present states of the systems inquestion, will also serve as templates for models ofselected areas (notably the Gulf of Maine,Newfoundland and the North Sea) forreconstructions representing these systems prior tothe onset of large scale mechanized fisheries, andthe ensuing resource depletion. Thus, LME-levelECOPATH models of past ecosystems will provide thebasis for estimating the benefits that would obtainfrom rebuilding strategies, as required to addressQuestion 6 in the Introduction (see also Figure 5).The next section provides more details on this issue.Sea Around Us Project Methodology Review8Evaluating alternative ecosystem-basedmanagement regimesTo complement the analysis of small- and large-scale fisheries as outlined above, leading to anestimate of potential economic gains from improvedmanagement, we will simulate the results of variousmanagement regimes, and evaluate their results inthe framework of fisheries economics, extended tomake it applicable to ecosystem analysis.The extended theory is then applied to explore anumber of questions including (i) to what extent isit worth society’s while to restore currentecosystems to their past states? (ii) What is theoptimal approach path to the past ecosystem? Is itoptimal to invest (disinvest) rapidly in restoring theecosystem, or should investment (disinvestment)proceed more slowly?ECOPATH and ECOSIM models will form theecological basis for our analysis, while ecologicaleconomics valuation techniques will help determinethe economically feasible restoration plans andpaths (see Munro and Sumaila 2000).Mapping the fate of fisheries landings from theNorth AtlanticThe validity of the analyses described abovedepends on the markets presently existing for fishproducts, and their likely evolution. We proposetherefore, that a spreadsheet-based framework canhelp track the flow of fish landings within the NorthAtlantic region (details in Sumaila et al. 2000).Starting with the total fish landings from the watersof each major fishing nation within the NorthAtlantic region, a map can be developed showinghow these landings flow into the major productforms under which they are marketed, i.e., fresh,frozen, salted and smoked. In addition, the portionof the product forms are consumed in the domesticversus the export market can be determined.Finally, the results derived can be used to identifythe sectors or product forms which capture most ofthe economic benefits from the fishes of the NorthAtlantic.Energy consumption and ecological footprint of theNorth Atlantic fisheriesOne way to express the overcapitalization of NorthAtlantic fisheries (i.e., the excess of catchingcapacity) is to relate the energy dissipated ingenerating present landings to the energy containedin the landings.This appears more straightforward than estimatingfleet ‘capacity’, which is not only hard to measure,but even hard to define.  Energy expenditures, onthe other hand are easily defined, and can beestimated reasonably well from the size of thevessels, which relates strongly to that of theirengines, and hence to their fuel consumption.Hence our choice of Horsepower∙days as measure ofeffort, a choice having the further advantage ofallowing comparisons between otherwise widelydifferent boat/fishing gear combination (seeWatson et al. 2000).The estimation of energy consumption by thefishing fleets of the North Atlantic, and the relatedestimation of their ecological footprint(Wackernagel and Rees 1996), are presented byTydmer (2000), who provides details, as well, onthe required distinction between variable energycosts (associated with running vessels) and fixedcosts, associated with the construction and eventualretirements of the vessels comprising a fleet.We anticipate that the aggregate energy costs offishing, in the North Atlantic, will be very high,relative to the energy (and commercial value) of thelandings, the difference being met by varioussubsides.RAPFISH and compliance analysesEvaluations in the Sea Around Us Project employ anew multi-disciplinary, rapid appraisal technique,called RAPFISH, that focuses on the comparativesustainability of fisheries (Pitcher and Preikshot1998; Pitcher et al. 1998a; Pitcher et al. 1998b;Preikshot and Pauly 1998; Preikshot et al. 1998;Pitcher and Preikshot, in press). RAPFISH can beperformed even when the rigorous survey data thatenables conventional stock assessment are notavailable, as is the case for many North Atlanticfisheries.As such, RAPFISH is a typical SNT, a smart new trickas defined above. It is however, suitable for the SeaAround Us Project because it allows us to quantifyaspects of fisheries thought before to beunquantifiable, and thus allows for comparisons.Moreover, the method can be applied at all scalesrelevant to the Sea Around Us Project, from thefisheries of a small bay of gulf, to those of countries,or of the entire North Atlantic. As well, RAPFISH canbe used to compare gears, and thus to contribute itsunique perspective to the comparisons betweensmall- and large-scale fisheries mentioned above.Assessment and Mitigation of Fisheries Impacts9In RAPFISH analyses, sets of attributes, chosen toreflect sustainability within each discipline, arescored on a ranked or binary scale. Where data aresparse or uncertain, scores may be refined whenbetter information becomes available. Ordinationsof sets of attributes are performed using multi-dimensional scaling followed by scaling androtation. The leverage of each attribute on theresults can be estimated with a step-wise procedure.The ordinations are anchored by fixed referencepoints that simulate the best (= ‘good’) and worst(‘bad’) possible fisheries using extremes of theattribute scores, while other anchors secure theordination in a second axis normal to the first.Significant differences are defined by Monte Carlosimulation of errors attached to the original scores.Raw plots of the results show fisheries status inrelation to ‘bad’ and ‘good’.Separate RAPFISH ordinations are performed inevaluation fields (disciplines) that express status interms of ecological, economic, social, technologicaland ethical (Pitcher and Power 2000) sustainability:a further field evaluates compliance with the FAOCode of Conduct for Responsible Fisheries (Pitcher1999).  Status results may be combined in ahierarchical way in ‘kite diagrams’ (see Figure 2) tofacilitate comparison of fisheries by gear type,country, ecosystem or size category, and data maybe constructed to represent the outcomes ofalternative policies (Alder et al. 2000).At this stage in the SAU project, we present a paperreporting preliminary RAPFISH analyses of fisheriesin two major North Atlantic areas, the Gulf of Maineand the North Sea (Alder et al. 2000). By the end ofthe SAU project all major fisheries will be coveredby RAPFISH evaluations. This will allowexamination, for each country, of fisheriescompliance with the FAO Code of Conduct forResponsible Fisheries. Compliance scored in thisway will be also be evaluated using a matrixexpressing international fisheries conventions towhich each country in the North Atlantic issignatory.Figure 2. Diagram illustrating how RAPFISHevaluation fields for different modalities ofsustainability can be considered together as scoreson the axes of a kite diagram. Boxes represent theattributes used to ordinate fisheries within eachevaluation field. Connections, arrows and kite apicesrepresent a score between 0% and 10% from eachfield. The outer rim of the kite is equivalent to 100%scores ( = ‘good’) in each field, while the centre of thekite represents scores of 0% ( = ‘bad’). Six evaluationfields are illustrated here, one of which, for the Codeof Conduct, is comprised hierarchically of a five-fieldRAPFISH.Sea Around Us Project Methodology Review10CONCLUSIONSThe relationships among the various elements of theSea Around Us Project are summarized in Figure 5.We anticipate that the synthesis to emerge fromintegrating the results of these modules will containmany surprises, both in terms of the ecologicaldamage and economic waste presently generated bythe North Atlantic fisheries, and the benefits thatcould be gained, were these economic and ecologicalissues addressed.CATCH/EFFORT DATABASE Comprehensive time seriesaccounting for small-scale catches, discards, unreported catch etc.,by areaECOPATH MODELS  One model per area,incorporating current  data & understanding of each ecosystemRAPFISH ANALYSIS Status of fisheries interms of ecological, economic, technological, social and ethical sustainability MAPS Ecosystem definitions and propertiesFISHBASE  Spp. by  ecosystems )(subcontractTRANSECTS  Depth distributionof key spp.TOTALEXTRACTIONS  By species and ecosystemECOSYSTEMHEALTH  FISHERIESINTERACTIONS  losses from competition between small- and large-scale fisheriesFISHINGEFFORT  Energy consumptionand ecological footprint of fleets PAST ECOYSTEMS  Models of past, reconstructed ecosystemsMARKETS  Market pathways, profits  andsubsidies  COMPLIANCE  with treaties and the FAO Code of Conduct for Responsible Fisheries  EVALUATION  Benefits from ecosystem-based managementFigure 3.  Conceptual diagram illustrating the relationships of the various methodological elements of the SeaAround Us Project.Assessment and Mitigation of Fisheries Impacts11ACKNOWLEDGMENTSThe Sea Around Us Project is supported by theEnvironment Program of The Pew CharitableTrusts. We would particularly like to thank Dr.Joshua Reichert, for his support of the ideas whichled to the development of this project. Also, wethank Drs Reg Watson, Rashid Sumaila, and DirkZeller for detailed comments on the draft of thiscontribution.REFERENCESAlder, J. Pitcher, T.J., Preikshot, D., Kaschner, K. andFerriss, B. (2000) Rapfish estimates - how goodis good? In: Pauly, D. and Pitcher T.J. (eds)Methods for evaluating the impacts of fisherieson North Atlantic ecosystems. Fisheries CentreResearch Reports 8(2), 136-182.Binmore, K. (1982)  Fun and games: a text on gametheory. Chancellor Press, London.Brey, T. (1999) A collection of empirical relations for usein ecological modelling. Naga, the ICLARMQuarterly, 23(3), 24-28.Christensen, V. and Walters, C. (2000) Ecopath withEcosim: Methods, capabilities and limitations.In: Pauly, D. and Pitcher T.J. (eds)   Methods forevaluating the impacts of fisheries on NorthAtlantic ecosystems.  Fisheries Centre ResearchReports 8(2), 79-105.Costanza, R. and M. Mageau. (1999) What is ecosystemhealth? Aquatic Ecology 33, 105-115.Gordon, H.S. (1954) Economic theory of a commonproperty resource: the fishery. Journal ofPolitical Economy 62, 124-142.Gulland, J.A. (1965) Estimation of mortality rates. Annexto the Report of the Arctic Fisheries WorkingGroup, ICES C.M. 1965/3, 9 pp.Lear, Linda. (1997) Rachel Carson: witness for nature.Henry Holt and Company, New York. 634 pp.Levin, S.A. (1990) Physical and biological scales and themodelling of predator-prey interactions in LargeMarine Ecosystems. In: K. Sherman, L.M.Alexander, and B.D. Gold (eds.). Large MarineEcosystems: patterns, processes and yields.Amer. Ass. Adv. Sci. Washington, D.C., 179-187.Longhurst, A.R. (1998) Ecological Geography of the Sea.Academic Press, San Diego  398 pp.Longhurst, A.R. 1995. Seasonal cycles of pelagicproduction and consumption. Progress inOceanography 36, 77-167.Mace, P.M. (2000) MSY reborn, but with a new identity:Fish and Fisheries. (in press.)Maury, M.F. (1963) The physical geography of the sea andits meteorology. The Belknap Press of HarvardUniversity Press, Cambridge, 432 pp.Munro, G.R. (1979) The optimal management oftransboundary renewable resources. CanadianJournal of Economics, 12(8), 355-376.Munro, G. and Sumaila, U.R. (2000) Investment incapture Fishery resources: An ecosystemapproach.  In: Pauly, D. and Pitcher, T.J. (eds)Methods for evaluating the impacts of fisherieson North Atlantic ecosystems. Fisheries CentreResearch Reports 8(2), 106-113.Myers, R.A., K. G. Bowen, and N.J. Barrowman. (1999)Maximum reproductive rate of fish at lowpopulation sizes. Can. J. Fish. Aquat. Sci. 56,2404-2419.Nash, J.F. (1951): Non-cooperative games. Annals ofMathematics 54, 286-295Nash, J.F. (1953): Two person cooperative games.Econometrica 21, 128-295.Pauly, D., Watson, R., Longhurst, A., Sherman, K. andPlatt, T. (2000) Mapping Fisheries onto MarineEcosystems.  In: Pauly, D. and Pitcher T.J. (eds)Methods for evaluating the impacts of fisherieson North Atlantic ecosystems. Fisheries CentreResearch Reports 8(2), 13-22.Pitcher, T.J. (1999) Rapfish, A Rapid Appraisal TechniqueFor Fisheries, And Its Application To The CodeOf Conduct For Responsible Fisheries. FAOFisheries Circular 947, 47 pp.Pitcher, T.J. and Preikshot, D. (1998) Rapid Appraisal ofthe Status of West African Distant Water FleetFisheries Relative to Home Fleets using theRapfish Technique. In: Bonfil, R., Munro, G.,Sumaila, U.R., Valtysson, H., Wright, M.,Pitcher, T., Preikshot, D., Haggan, N. and Pauly,D. (eds.) Distant water fleets: an ecological,economic and social assessment. FisheriesCentre Research Reports 6 (6), 90 – 93.Pitcher, T.J. and Preikshot, D.B. (2000) Rapfish: A RapidAppraisal Technique to Evaluate theSustainability Status of Fisheries. FisheriesCentre Research Reports. (in press).Pitcher, T.J. and Power, M.P. (2000) Fish Figures:Quantifying the Ethical Status of CanadianFisheries, East and West. In: Coward, H.,Ommer, R. and Pitcher, T.J. (eds). Just Fish: theethics of Canadian fisheries. Institute of Socialand Economic Research Press, St John's,Newfoundland, 225-253Pitcher, T.J. and Hart, P.J.B. (1982) Fisheries Ecology.Chapman and Hall, London. 414 pp.Pitcher, T. and Watson, R. (2000) The basis for changePart 2: Estimating the total extractions frommarine ecosystems. In: Pauly, D. and Pitcher,T.J. (eds) Methods for evaluating the impacts offisheries on North Atlantic ecosystems. FisheriesCentre Research Reports 8(2), 40-53.Pitcher, T.J., Bundy, A., Preikshot, D., Hutton, T. andPauly, D.  (1998a) Measuring the unmeasurable:a multivariate interdisciplinary method for rapidappraisal of health of fisheries. In: Pitcher, T.J.,Hart, P.J.B. and Pauly, D. (eds.) ReinventingFisheries Management. Kluwer, Dordrecht,Holland, 31-54.Sea Around Us Project Methodology Review12Pitcher, T.J., Mackinson, S., Vasconcellos, M.,  Nøttestad,L. and Preikshot, D.  (1998b) Rapid appraisal ofthe status of fisheries for small pelagics usingmultivariate, multidisciplinary ordination. In:Quinn, T.J. II, Funk, F., Heifetz, J.,  Ianelli, J.N.,Powers, J.E., Schweigert, J.F., Sullivan, P.J. andZhang, C.I. (eds) Fishery Stock AssessmentModels. Alaska Sea Grant, Fairbanks, USA, 759-782Preikshot, D.B. and Pauly, D. (1998) A multivariateinterdisciplinary assessment of small-scaletropical fisheries. In: Quinn, T.J. II, Funk, F.,Heifetz, J., Ianelli, J.N.,  Powers, J.E.,Schweigert, J.F., Sullivan, P.J. and Zhang, C.I.(eds) Fishery Stock Assessment Models. AlaskaSea Grant, Fairbanks, USA, 803-814Preikshot, D.B., Nsiku. E., Pitcher, T.J. and Pauly, D.(1998) An interdisciplinary evaluation of thestatus and health of African lake fisheries using arapid appraisal technique. J. Fish Biol. 53 (SupplA), 382-393.Punt, A.E. and Hilborn, R. (1997) Fisheries stockassessment and decision analysis: the Bayesianapproach. Rev. Fish. Biol. Fish. 7, 35-63.Rapport, D. J., Costanza, R. and McMichael, A.J. (1998)Assessing ecosystem health: challenges at theinterface of social, natural and health sciences.Trends in Ecology and Evolution 13,397-402.Rapport, D., Costanza, R., Epstein, P., Gaudet, C. andLevins, R. (1998) Ecosystem Health.  BlackwellScientific, New York, 372 pp.Ruttan, L.M., Gayanilo, F.C. Jr. and Pauly, D. (2000)Small versus large scale fisheries: a multispecies,multifleet model for evaluating their potential toconserve fishes and fisheries. In: Pauly, D. andPitcher, T.J. (eds) Methods for evaluating theimpacts of fisheries on North Atlanticecosystems. Fisheries Centre Research Reports8(2), 64-78.Sherman, K. and Duda, A.M. (1999) An ecosystemapproach to global assessment and managementof coastal waters.  Mar. Ecol. Progr. Ser. 190,271-287.Sumaila, U.R. (1997) Cooperative and non-cooperativeexploitation of the Arcto-Norwegian cod stock inthe Barents Sea. Environmental and ResourceEconomics 10, 147-165.Sumaila, U.R., Chuenpagdee, R. and Munro, G. (2000)Tracking fisheries landings in the North Atlantic.In: Pauly, D. and Pitcher, T.J. (eds) Methods forassessing the impact of fisheries on marineecosystems of the North Atlantic. FisheriesCentre Research Reports 8(2), 114-122.Tyedmers, P. (2000) Quantifying the Energy ConsumedBy North Atlantic Fisheries.  In: Pauly, D. andPitcher, T.J. (eds) Methods for assessing theimpact of fisheries on marine ecosystems of theNorth Atlantic. Fisheries Centre ResearchReports 8(2), 123-135.Wackernagel, M. and Rees, W.E. (1996) Our EcologicalFootprint: Reducing Human Impact on theEarth. New Society Publishers, Gabriola Island,British Columbia. 160 pp.Walters, C. Pauly, D. and Christensen, V. (1999)Ecospace: prediction of mesoscale spatialpatterns in trophic relationships of exploitedecosystems, with emphasis on the impacts ofmarine protected areas. Ecosystems 2, 539-554.Walters, C. Christensen, V. and Pauly, D. (1997)Structuring dynamic models of exploitedecosystems from trophic mass-balanceassessments. Rev. Fish Biol. Fish.  7(2), 139-172.Walters, C.J. and Maguire, J. J. (1996) Lessons for stockassessment from the northern cod collapse. Rev.Fish Biol. Fish. 6, 125-137.Watson, R., Guenette, S., Fanning, P. and Pitcher,T.J.(2000) The basis for change: Part 1 -Reconstructing fisheries catch and effort data.In: Pauly, D. and Pitcher, T.J. (eds) Methods forevaluating the impacts of fisheries on NorthAtlantic ecosystems. Fisheries Centre ResearchReports 8(2), 23-39.Zeller, D. and Pauly, D. (2000) Life history patterns anddepth zone analysis.  In: Pauly, D. and Pitcher,T.J. (eds) Methods for evaluating the impacts offisheries on North Atlantic ecosystems. FisheriesCentre Research Reports 8(2), 54-63.Sea Around Us Project Methodology ReviewMapping the Globe’s Fisheries13MAPPING FISHERIES ONTO MARINEECOSYSTEMS:A PROPOSAL FOR A CONSENSUSAPPROACH FORREGIONAL, OCEANIC AND GLOBALINTEGRATIONSa)Daniel Pauly, Villy Christensen, Rainer Froese,Alan Longhurst, Trevor Platt, ShubhaSathyendranath, Kenneth Sherman and RegWatsonABSTRACTResearch on ecosystem-based fisheriesmanagement, marine biodiversity conservation,and other fields requires appropriate maps of themajor natural regions of the oceans, and theirecosystems.It is proposed here that a classification systemproposed by T. Platt and S. Sathyendranath andimplemented by A.R. Longhurst, defined largelyby physical parameters, and which subdivides theoceans into four ‘biomes’ and 57 ‘biogeochemicalprovinces’ (BGCPs), could be merged with thesystem of 50 Large Marine Ecosystems (LMEs)identified by K. Sherman and colleagues, whichwould represent subunits of the provinces. Thisarrangement enhances each of the systems, andrenders them mutually compatible. For theLMEs, subprovinces are pragmatically defined toserve as a framework for the management ofcoastal fisheries, and other purposes, while theBGCPs have rigorous physical definitions,including borders defined by natural features.Moreover, incorporating the 50 defined LMEsinto the framework of BGCPs will allowstraightforward scaling-up of LME-specific flowestimates (including fisheries catches) up to basinand ocean scales. The combined mapping willallow the computation of GIS-derived propertiessuch as temperature, primary production, etc.,and their analysis in relation to fishery catch datafor any study area.A further useful aspect of the proposed scheme isthat it will enable us to quantify the EEZ ofvarious countries in terms of the distribution ofmarine features (e.g., primary production, coralreef areas) which has yet to be straightforwardlyaassociated with coastal states.                                                a presented as C.M. 2000/T:14 at the Annual Science Conference ofthe International Council for the Exploration of the Sea.Applications to shelf, coral reef and oceanicfisheries, and to the mapping of marinebiodiversity are briefly discussed.INTRODUCTIONThere is a broad consensus in the scientificcommunity that fisheries management should beecosystem-based, but very little agreement as towhat this means (NRC 1999). Also, there is a needto analyze biodiversity data at larger scales thangenerally done so far, as demonstrated by, e.g.,Sala et al. (2000) for terrestrial and freshwaterbiomes.Clearly, when dealing with such complex issues,the first task, as in all science-based approachesto a problem, is to define the object(s) of concern,and to develop a consistent method to show howthese objects are interrelated. Here, the objectsare the marine ecosystems within which fisheriesand biodiversity are to be analyzed, and marinelife in general, is embedded.Fortunately, establishing a consensus on theclassification of marine ecosystems may berelatively easy, given the compatibility, so farnever elaborated upon, of two classificationschemes proposed in recent years. Both of theseintegrate enormous amount of empirical data,and are sensitive to previous analyses of marineecology. These two schemes are (1) the globalsystem of 57 ‘biogeochemical provinces’ (BGCPs)developed by Platt and Sathyendranath (1988,1993), Platt et al. (1991, 1992), Sathyendranath etal. (1989), Sathyendranath and Platt (1993),implemented by Longhurst (1995, 1998), anddefined at scales appropriate for understandingphysical forcing of ocean primary production andrelated processes; and (2) the 50 coastal LargeMarine Ecosystems (LMEs) gradually defined bySherman and co-workers (see e.g. Sherman et al.1990, 1993), whose size and on-shelf locationmakes them particularly suitable for addressingmanagement issues, notably those pertaining tofisheries on continental shelves, and coastal areamanagement (Sherman and Duda 1999).After reviewing selected features of these twoschemes, we suggest how the partition of oceanregions that they imply can be made mutuallycompatible. The joint classification which thenemerges is presented in form of a spatialhierarchy, and as maps, each emphasizing a keyfeature of the classification. Overall, theintegrated scheme we propose allows explicitconsideration of different scales, as discussed e.g.by Levin (1990).Sea Around Us Project Methodology Review14BIOGEOCHEMICAL PROVINCESThis partition of the ocean is derived from asuggestion of Platt and Sathyendranath (1988) forthe recognition of natural regions of the ocean,having characteristic physical forcing to whichthere is a characteristic response of the pelagicecosystem.  These regions were to be dynamicbiogeochemical provinces (‘dynamic’ becausetheir boundaries would respond to annual andseasonal changes in physical forcing, and‘biogeochemical’ because within each the biotawould respond to those characteristicgeochemical processes which determine nutrientdelivery to the euphotic zone).  This concept hasbeen used to partition both global and basin-scaleanalyses of primary productivity, though the‘dynamic’ boundary aspect of the system remainsto be exploited:  so far, most applications of thepartition have assumed that boundaries betweenprovinces were fixed at locations representingaverage conditions, though dynamic boundarieshave been used for analysis of Arabian Seaproductivity.The central principle in locating boundariesbetween provinces is that of the critical depthmodel of Sverdrup (1953), which remains themost useful formulation relating phytoplanktongrowth to surface illumination, and to the verticaldensity structure of the water column. Itsuccessfully predicts, for example, the timing ofthe North Atlantic spring bloom. A proposedpartition of the North Atlantic into 18 BGCPs(Platt, et al. 1995) was followed by a partition ofall oceans and adjacent seas into 57 provinces(Longhurst et al. 1995 and Longhurst, 1998).The global partition was arrived at byexamination of 26,000 archived chlorophyllprofiles to determine Gaussian parametersdescribing the regional/seasonal characteristicprofiles, surface chlorophyll from 43,000 grid-points from monthly Coastal Zone ColourScanner images, and about 23,000 monthly meanmixed layer depths, together with otheroceanographic variables. This analysis suggestedthat a two-level partition would be requiredadequately to represent regional differences inthe expression of the Sverdrup model. The firstpartition is into a small number of biomes,following the usage of this term by terrestrialecologists to mean a region of relatively uniformdominant vegetation type, with its associatedflora and fauna: grassland, tundra, steppe, humidforest and so on (Golley 1993).  Secondly, thesebiomes are each be partitioned into a number ofregional entities, the biogeochemical provinces.The four biomes (Figure 1) are defined by thedominant oceanographic process that determinethe vertical density structure of the water column,which itself is what principally constrains thevertical flux of nutrients from the interior of theocean.In the Polar biome, vertical density structure isvery largely determined by the flux of fresh orlow-salinity water derived from ice-melt eachspring and which forms a prominent halocline inpolar and sub-polar oceans.  In oceanographicterms, this occurs in each hemisphere polewardsof the Oceanic Polar Front, whose location in eachocean is determined by the characteristiccirculation of each.  Though looming large onMercator maps, the Polar biome occupies onlyabout 6% of the ocean’s surface.Between the polar Fronts and the subtropicalconvergence in each ocean lies the Westerliesbiome.  Here, large seasonal differences in mixed-layer depth are forced by seasonality in surfaceirradiance and wind stress.  Biological processesconsequently may have sufficiently strongseasonality so that a spring bloom characterizesthe plankton calendar.Across the equatorial regions, between the borealand austral subtropical convergences lies theTrade-wind biome.  Here, the conjunctionbetween low values for the Coriolis parameter, astrong density gradient across the permanentpycnocline and weak seasonality in both windstress and surface irradiance result in relativelyuniform levels of primary production throughoutthe year.Upper continental slopes, continental shelves andmarginal seas comprise the Coastal Boundarybiome. This is constrained between the coastlineitself and (usually) the oceanographic frontcharacteristically found at the shelf-edge. Thesingle generalization that characterizes this biomeis that nutrient flux in the water column is forcedby a great variety of processes: coastal upwelling,tidal friction, fresh-water outflow from rivermouths, etc. In the partitions discussed above,subdivision of this biome into provinces was notcarried as far as might be useful for somepurposes. One of the objectives of the presentstudy is to do just that, through the introductionof subprovinces and their identifications withLMEs.The boundaries between the biomes thus definedcertainly vary seasonally and between years, ascan readily be inferred from satellite images, anddynamic boundaries that respond to thisMapping the Globe’s Fisheries15variability are discussed for primary productionand related studies by Platt and Sathyendranath,(1999). However, such dynamic schemes areneither practical nor necessarily useful forbiodiversity and fisheries studies. For example,one of the tasks facing biodiversity studies are thecreation of global maps documenting thedistribution of hundred of thousands of marinespecies. Requiring that these distributions areassigned to habitats with variable boundarieswould make even simple, first-order assignmentsof species extremely difficult and postpone thedelivery of products whose need is already keenlyfelt by students of biodiversity.Thus, in the case of fishes, of which about 15,000species are marine, the assignment withinFishBase (see www.fishbase.org) of species toclimate type (as defined in the insert of Figure 1),required us to distinguish tropical from non-tropical species (see Pauly 1998), and this taskalone required several person-months worth ofwork to complete.Moreover, there are numerous types of floral orfaunal assemblages whose location does not vary,though their habitat is part of, or affected by asurrounding or overlying pelagic ecosystem.Thus, the reef fishes of the Galapagos do notchange their location when an El Niño eventstrikes the archipelago. Rather, it is theirabundance which is affected (Grove 1985, Groveand Lavenberg 1997). A similar argument appliesto benthic communities, whose boundaries willtend to reflect the long term average location ofthe boundaries of the overlying pelagic systems,rather than tracking their changing location(Ekman 1967).The ecosystem classification scheme proposedhere is thus deliberately fixed in space. On theother hand, we anticipate that its use by variousauthors will quickly lead to the identification andquantification of changes in species compositions,thus reintroducing the dynamic element requiredat various spatial and temporal scales (Levin1990).Oceanographic conditions within the four biomesare obviously not uniform, and each can besubdivided further using the same set ofprinciples as determined the biomes themselves.For example, in both the westerlies and tradesbiomes there are definable ocean regions whereheavy tropical rainfall or excessive continentalfresh water runoff lead to the existence of a quasi-permanent low salinity ‘barrier-layer’ occupyingthe upper portion of the thermally-stratifiedsurface layer. This has important biologicalconsequences and suggests that these regionsshould be recognized as individual partitions.Using such methods, based on close examinationof regional physical oceanography, the fourprimary biomes can be further partitioned into 57provinces, the BGCPs discussed above. Figure 2Figure 1. Map of the world ocean’s biomes, the highest category in the proposed classification of the world oceans.Note its overall similarity to a conventional map of the atmospheric climate (inset, adapted from Anon. 1991).(Polar is lightest, Coastal is next darkest, followed by Westerlies and Trades)Sea Around Us Project Methodology Review16illustrates these provinces as defined byLonghurst et al. (1995). This schema has beenused to stratify the world ocean in two studies,pertaining to the global distribution of primaryproduction (Longhurst et al. 1995) and tunacatches (Fonteneau 1998), with more forthcoming(Platt and Sathyendranath 1999, Pauly 1999).Also, as part of the collaboration between the SeaAround Us project (details atwww.fisheries.ubc.ca) and the FishBase project(Froese and Pauly, 2000), the world’s marinefishes are presently being assigned to BGCPs, ifsomewhat tentatively in a few cases. We note thatthis work, which relies on a large number of localichthyo-fauna lists, will require about 12 person-months to complete. However, it would requiremuch longer were it  necessary to compile first aglobal list of fish species, and to assign themdirectly to the BGCP, without prior assignment toFAO areas, countries, and oceanic islands, as isprovided by FishBase.This point is important with regards toinvertebrate groups, whose global distributionwill have to be mapped, in the long term, in amanner compatible to that used for fishes. Thisshould, for example, be an important componentof an Ocean Biogeographic Information Systemcurrently under consideration (Grassle andStocks, 1999).LARGE MARINE ECOSYSTEMSIn recent years, the formerly generic term ‘LargeMarine Ecosystem’ (LME) has become specific,and is now mainly used for regions of ocean spaceencompassing coastal areas out to the seawardboundary of continental shelves and the outermargins of coastal current systems. As such,LMEs are regions of the order of 200,000 km2 orgreater, characterized by distinct bathymetry,hydrography and productivity patterns (Sherman1994; Sherman and Duda 1999).The 50 LMEs identified by Sherman and Duda(1999) are the source of about 95 % of the world’sannual marine fisheries yields. Also, most of theglobal ocean pollution, overexploitation, andcoastal habitat alteration occur within these 50LMEs. They provide, therefore, a convenientframework for addressing issues of naturalresources management. Moreover, given thatmost of them border developing countries, LMEsalso provide a framework for addressing issuesrelated to issues of economic development.Various development agencies, notably the GlobalEnvironment Facility (GEF), the United NationsDevelopment Programme, the UN EnvironmentProgramme, and the World Bank have endorsedthe LME concept as framework for several of theirFigure. 2. Map of the world ocean’s 57 biogeochemical provinces, the second level in our proposed classification ofthe world oceans. (The borders of a few disjunct provinces, notably ARCH, will be simplified; detailed file availablefrom www.fisheries.ubc.caMapping the Globe’s Fisheries17international development projects, for examplein the Gulf of Guinea, with more such projectsforthcoming (Sherman and Duda 1999). Giventhis considerable amount of interest, it isfortunate that a number of BGCP, i.e., those inthe coastal domain, can easily be divided into‘sub-provinces’ congruent with the 50 LMEs inthe list of Sherman and Duda (1999). Thus,Figure 3 illustrates, for the North Atlantic, howthe 15 LMEs occurring therein (including theBaltic Sea) can be mapped onto BGCPs of Figure2, with some LMEs identified by two components(e.g. ‘Southern’ and ‘Northern’) when theystraddle two provinces, and new subprovincesnamed where appropriate, i.e., for the parts ofprovinces not included in a defined LME. Asimilar map for the entire ocean, including all 50LMEs in Sherman and Duda (1999) is currently inpreparation (details on www.fisheries.ubc.ca).This mapping provides, we believe, the elementsthat had been lacking within each of the systemsthus rendered compatible. For BGCPs, we identifysub-provinces that are pragmatically defined toserve as framework for fisheries, coastal area andother applied research. As for the LMEs, theyobtain, via their incorporation into the scheme ofbiomes and BCGPs discussed above, explicitphysical definitions, including borders (hereimplemented in steps of half-degree squares),that allow GIS-based computation of systemproperties, such as mean depth, temperature,primary production, etc.Another consideration is that our scheme forembedding LME and other subprovinces intoBCGPs can be used as an ecological complementto the coarse stratification scheme used by theFood and Agriculture of the United Nations(FAO) to present global marine fisheries data,and which relies on 18 FAO statistical areas (7 forthe Atlantic ocean, 3 for the Indian Ocean and 8for the Pacific Ocean).To facilitate comparisons between catch datastratified by these two schemes, we split the fivecircumpolar BCGP into ocean-specificsubprovinces. This procedure enables ‘closure’ ofthe Atlantic, Indian and Pacific oceans and thusallows direct comparisons, at least at ocean-levelscale, between catch data stratified within thescheme proposed here, and that used by FAO forits global catch database.  Note that our next task,in this context, is to assign the catches in theFigure 3. Map of the North Atlantic, illustrating how the LMEs identified for this basin (in Sherman and Duda 1999)can be identified with parts of biogeochemical provinces (Figure 2). Note that some LMEs may be subdivided, andtheir subcomponents assigned to different provinces. Also note definition of various sub-provinces for areas notcurrently covered by LMEs).Sea Around Us Project Methodology Review18global FAO data set to BGCP and sub-provinces(and/or LMEs), pending its gradual replacement,starting with the North Atlantic, by locally-derived data sets. Among other things, this willallow for rapidly arraying fisheries catches andrelated data for comparative analyses, i.e., datanow usually assembled on an ad hoc basis (seee.g. Caddy et al. 1998, or Pauly et al. 1998a), atscales that are often inappropriate for theintended results.EXCLUSIVE ECONOMIC ZONESAllocating freshwater species and their catches tocountries is straightforward, as the internationalborders of countries are usually well defined. Thisis more difficult in the marine realm, where thefishes and invertebrates caught off the coast of agiven country may be caught outside its territorialwaters. The International Law of the Seaprovides, at least in principle, a solution to this, inform of Exclusive Economic Zones, usuallyreaching 200 miles into the open ocean, andlinking countries with much of the productiveareas, i.e., the shelves adjacent to their coasts.However, not all countries have EEZ accepted bytheir neighbors, and in certain areas, such as theSouth China Sea, the same rocky outcrops areclaimed by up to half a dozen countries(McManus 1992).  It cannot be expected that thisand similar situations in other parts of the worldwill be resolved soon, nor peacefully for thatmatter, and we cannot expect therefore, thatofficial maps of the EEZ will appear that could beused for assigning fisheries catches to thecountries of the world.Nevertheless, various scholars, and institutionshave published EEZ maps of various parts of theworld (see e.g. Mahon 1987, for the Caribbean),based on the rules for definition of EEZestablished by the Law of the Sea Convention(Charney and Alexander 1993). We propose thatsuch maps can be used to derive a coherent singlemap for the EEZ of the world, especially if care istaken to incorporate into such map thedelimitations so far agreed though bilateral ormultilateral treaties (as compiled, e.g., in Charneyand Alexander 1993).The advantage of such map is that, unlike like themap of LME and provinces mentioned above, itwill enable the assignment of fish and otherspecies, and of fisheries catch statistics tocountries. This will enable comparisons of variousfeatures of the use and productivity of variouscountries’ EEZ, with enough degrees of freedomfor multivariate analyses, as now routinelyperformed for the land-based resources of variouscountries. It is clear, of course, that such adesignation is unofficial and for scientificpurposes only, and that it has no bearing, implicitor explicit, on the status of any disputes betweensovereign nations about EEZ.GLOBAL DISTRIBUTION OFCORAL REEF SYSTEMSCoral reefs, though presently under threatthroughout much of their range (Buddemeier andSmith 1999), support important fisherieswherever they occur (Munro 1996). However,quantifying these catches in reliable fashion hasproven particularly difficult. One reason is thatmost countries with coral reefs are developing,with administrative infrastructures that precludedetailed monitoring of their fisheries.As suggested by the pioneering work of Smith(1978), who performed the first analysis of thistype, global assessment of present and potentialfisheries yields from coral reefs would be muchimproved by comparative studies wherein thecoral reef fish and invertebrate catches fromvarious countries EEZ would be matched againstthe surface area of coral reefs within these sameEEZ.However, while it is possible to assign to coralreefs, at least roughly, a fraction of the catches ofeach country with reefs in the global FAOfisheries catch database, a matching set of coralreef area per country is not available, despitevarious global reviews of coral reefs (see e.g.Wells 1988; Polunin and Roberts 1996).The model of Kleypas et al. (1999) can be used,however, to estimate expected coral reef area forany part of the world ocean with a well defineddepth, temperature and light regime, and thuscan be used to predict coral reef areas within eachof the EEZ defined above. We anticipate, oncethis model becomes widely available, that plots ofcoral reef fish and/or invertebrate catches vs. reefarea will allow us to identify countries withproblematic catch data, and/or estimated reefareas, and thus to gradually improve theunderlying databases and models.SPATIAL EXPRESSION OFFISHERIES CATCH DATAFisheries catches are usually not reported on per-area basis (e.g. as t∙km-2∙year-1), though the areasfrom which they are derived are often specified.Maps of catch per area are rare, and indeed existMapping the Globe’s Fisheries19only for local studies, often pertaining to single-species fisheries.Thus, one additional reason for the hierarchicalsystem proposed above is that would allow, andmake worthwhile, consistent, basin-scale andocean-wide mapping of catches onto theecosystems from which they originate.We anticipate the emergence of such maps, at theglobal level, from two successive steps:1) Mapping the global FAO statistics onto their(presumed) ecosystem of origin, for each ofthe 18 FAO statistical areas, by half-degreesquare;2) Improving the map in (1) through successivereplacement, by LME, of the FAO data bylocal data sets.As (1) is only to provide a ‘default’ map, i.e., thebackground for locally-enriched, presumablymore accurate data sets, there seem no need toallocate massive resources to this step.Our proposed approach therefore, is to proceedby successive ‘subtractions’, i.e., by first assigningfishes with clear affinities to certain LME, e.g. theanchoveta Engraulis ringens to the HumboldtCurrent LME, or the neritic fishes reported forBangladesh to the Bay of Bengal LME, etc., eachtime subtracting the assigned fish groups fromthe database.Several rounds of subtraction should quickly leadto small amounts of unallocated landings,pertaining mainly to fish landed in countries withdistant water fleets (or providing flags ofconvenience to such fleets).  Here, we assume thatassigning the residual landings to the LME wherethese fleets are known to occur (see Bonfil et al.1999 and references therein), in proportion to thecatches per half-degree square previouslyallocated, would be sufficient for a first-passallocation, especially since misallocations shouldgenerate visible patterns in the maps thusgenerated.Note that this procedure, whose application totunas would be very problematic, does not in factneed to be applied to this group, as Fontenau(1998), based on detailed catch data from ICCAT,IATTC, and IPTP, has already allocated globaltuna catches to their BGCP of origin. Similarly,the fraction of fishes in the FAO databasepreviously assigned to the coral reefs of differentcountries (see above) would not require thisprocedure, as they would have been previouslysubtracted, along with Fonteneau’s tunas.Once (1) is completed, i.e., it will bestraightforward to implement (2), i.e., to improvethe maps for certain areas with better coveragethan provided by the FAO catch statistics, e.g. theNorth Atlantic, where international data sets,from ICES and NAFO, and national data sets,from institutions such as DFO in Canada, NMFSin the USA, or IFREMER in France, are available.ECOSYSTEM DESCRIPTION USING ECOPATHThe ECOPATH with ECOSIM (& ECOSPACE)modeling approach has been recently reviewed inseveral contributions (Christensen and Pauly1992, Walters et al. 1997, 1999, Pauly et al. 2000,Christensen and Walters, 2000), and there is noneed here to present its working or outputs.ECOPATH models exist for numerous parts of theworld (details in www.ecopath.org), including theNorth Atlantic. Currently, well over 100 modelshave been published, and more than 1800colleagues in nearly 100 countries have registeredas users of the ECOPATH software system.However, the ecosystem model coverage ofvarious ocean basins is still spotty at best, henceprecluding simple raising of flows and rates fromecosystem to basin scales. Thus, a stratificationscheme is required, based on the geographicstructure outlined above, and which can be usedto scale models from the sampling area of thefield data used to parameterize the models to thewider area that is assumed represented by thesesame models. The strata for the North Atlanticare presented in Figure 4.LMEs (and other subprovinces) are seen here asproviding the key level for ecosystem modelconstruction. For each LME, an Ecopath modelmust be constructed to describe the ecosystemresources and their utilization, and to ensure thatthe total fisheries catch of each LME is used asoutput constraint (just as their primaryproduction will be used as input constraint). Inaddition, our stratification scheme must be suchthat it can straightforwardly accommodate anynumber of additional ECOPATH models for eachLME. This can be done so as to simultaneouslyaddress the issue of parameter uncertainty, asbriefly described below.The LME ECOPATH models require information onabundance, production and consumption ratesand diets for all ecosystem groupings. Suchinformation can be obtained from the followingsources:Sea Around Us Project Methodology Review20NE U.S.Continental ShelfSE U.S.Continental ShelfScotianShelfNewfoundlandShelfNW AtlanticShelvesSouthernNorth SeaCeltic-Biscay ShelfNorwegianShelfSkagerak-KattegatBaltic SeaNE AtlanticShelvesChesapeakeBayChesapeakeBayCoastalLabradorShelfEastGreenland ShelfWesternGreenland ShelfBaffinBayHudsonBayBorealPolar CentralBorealPolarNorthIceland ShelfArcticEastArcticWestAtlanticArcticAtlanticArcticBarentsSeaSouthIceland ShelfFaroePlateauNorthernNorth SeaPolarAtlanticAtlanticSubarcticPolarCantabrianShelfN AtlanticDriftN AtlanticDriftGulfStreamGulfStreamPortugueseCoastNorthCanary CurrentN AtlanticSubtropical GyreN AtlanticSubtropical GyreWesterliesFigure 4. Proposed hierarchy of biomes, biogeochemical provinces and LME/subprovinces in the North Atlantic. Noteimplicit stratification, for use when, e.g., scaling up, from part of an LME, to basin or ocean-wide estimates; see alsoFigure 3).• Abundance, production and consumptionrates, and diets of marine mammals areavailable from the Sea Around Us databasefor all (117) species of marine mammals (seealso Pauly et al 1998b, Trites and Pauly1998);• Fishery catches: available from the spatiallystructured catch database generated asdescribed above, and covering all speciesgroups;• Occurrence, biology and ecology of marinefishes: soon to be available from FishBase(www.fishbase.org), presently available bothat the BGCP level, and the LME/subprovincelevel as well. The relevant FishBase searchroutine option in question was designed foroptimizing extraction of ECOPATH-relevantinformation, and is a result of the ongoingcooperation between FishBase and SeaAround Us projects;• For marine invertebrates: only limitedinformation (beyond the catches in the FAOdatabase) is available from electronicdatabases, but a variety of publicationsprovide extensive information. Productionrates can be estimated from the well-foundedempirical relationships of Brey (1999), nowincluded in ECOPATH;• Primary production estimates: establishmentof a global database aimed at supplying finegrid level satellite based estimates of primaryproduction is presently underway through acooperation between the Space ApplicationsInstitute, EC Joint Research Centre, Ispra,Italy, and several authors of the presentcontribution.The origin of each set of data (5 rate or statevariable for each of the often 20-40 functionalgroup in a model, plus a diet matrix) can bedescribed and a related confidence intervalassigned to each of the input parameters.Confidence intervals can also be estimated, as‘posterior distributions’ for the output parametersof models. In addition a module of ECOPATH isdesigned to describe the ‘pedigree’ of ECOPATHmodels, i.e., the degree to which the models arerooted in locally sample and reliable data,(described in more details by Christensen andWalters, 2000). This module estimates, based onthe pedigree of its input data, an overall qualityindex for each model, which in turn can serve asweighting factor, as required when dealing withdiscrepancies (e.g. between local vs. LME-widecatches), i.e., when raising one or severalmodel(s) to the LME/subprovince level.The LME/subprovince-level ECOPATH models willthus make up the backbone of our approach foraddressing province, basin and global issuesrelated to abundance, productivity, interactionand impact for ecosystem resources e.g., bytrophic levels. Being based on the best availableestimates of productivity and utilization of theupper trophic levels, and on productivity for theprimary producers, the models are constrainedfrom the top as well as from below.Where possible the LME-level models will besupplemented with additional models. Theprocedure for this is:• New models are assigned to strata, based onthe proportion of area covered that fallswithin each of the depth strata < 10 m, 10-50m, 50-200 m, 200-1000 m, and > 1000 m;• For each new model, the confidence intervalsof input and output parameters are estimatedalong with the pedigree index of the model;• The LME/subprovince-level model isassigned to depth strata based using weightsbased on the relative primary productivity ineach of the depth strata;Mapping the Globe’s Fisheries21• Within each of the depth strata productivity,abundance, etc., are raised to theLME/subprovince level using the qualityindex of the models as weighting factors forthe associated confidence intervals.With this structure in place, it will be easy to addnew models as they become available, and it isfeasible to assign confidence intervals to allestimates derived from the analysis.CONCLUSIONSThe ecosystem classification proposed here is notmeant as a panacea that will solve all ourbiogeographical problems, or all spatial problemsof fisheries. It should not be necessary to stressthis; however, it is likely that some readers willthink we believe it. We don’t. However, we knowthat no telephone registry would ever emerge, ifregular debates were held as to the optimal way toarrange the letters in the alphabet.The ecosystem classification proposed here willsoon be implemented globally by FishBase, whichwill thus assign all marine fish species so fardescribed to their LME(s) or subprovince(s). Itwill also be used to give a geographic structure toan unofficial, spatialized, version of the FAOdatabase of global fisheries catches (see above),thus complementing the atlas of tuna catchescompiled by Fonteneau (1998), and allowing bothto be related to estimates of primary productionfor example, mapped in similar fashion byLonghurst et al. (1995).Moreover, this classification is fully compatiblewith the LME approach of Sherman and co-workers, which has led to an extensivedocumentation of management issues at LMEscale (see references in Sherman and Duda 1999),and a number of field projects designed toaddress these issues, funded by variousinternational granting agencies.Thus, we invite colleagues to join us in expressingtheir results using the classification anddefinitions proposed here. To support thiscollaboration, we will supply, via the Internet,tables presenting the details of the classificationby half-degree squares.ACKNOWLEDGMENTSThe members of the Sea Around Us Project thankthe Pew Charitable Trusts for their support of thework leading to this contribution. Also, the firstauthor acknowledges support of a grant from theCanadian National Science and EngineeringResearch Council.REFERENCESAnon. (1991) Bartolomew Illustrated World Atlas. HarperCollin, Edinburgh.Bonfil, R., Munro, G., Sumaila, U.R., Valtysson, H., Wright,M., Pitcher, T., Preikshot, D., Haggan, N. and Pauly,D. (1998) Impacts of distant water fleets: anecological, economic and social assessment. pp. 11-111 In: The footprint of distant water fleet on worldfisheries. Endangered Seas Campaign, WWFInternational, Godalming, Surrey, 111 pp.Brey, T. (1999) A collection of empirical relations for use inecological modelling. Naga, the ICLARM Quarterly,23(3), 24-28.Buddemeier, R.W. and Smith, S.V. (1999) Coral adaptationand acclimatization: a most ingenious paradox.Amer. Zool. 39:1-9.Caddy, J.F., Carocci, F. and Coppola, S. (1998) Have peakfishery production levels been passed in continentalshelf areas? Some perspectives arising for historicaltrends in production per shelf area. J. North Atl.Fish. Sci. 23, 191-219.Charney, J.I. and Alexander, L.M.A. (eds) (1993) InternationalMaritime Boundaries. The American Society ofInternational Law. Kluwer, Dordrecht.Christensen, V. and Pauly, D. (1992) The ECOPATH II - asoftware for balancing steady-state ecosystemmodels and calculating network characteristics.Ecological Modelling 61, 169-185.Christensen, V. and Walters, C.J. (2000) Ecopath withEcosim: Methods, capabilities and limitations In:Pauly, D. and Pitcher, T.J. (eds) Methods forevaluating the impacts of fisheries on North Atlanticecosystems.  Fisheries Centre Research Reports8(2), 79-105.Ekman, S. (1967)  Zoogeography of the Sea. Sidgwick andJackson, London, 417 pp.Fonteneau, A. (1998) Atlas of tropical tuna fisheries. EditionORSTOM, Paris.Froese, R. and Pauly, D. (eds) FishBase 2000: Concepts,Design and Data Sources. ICLARM. [seewww.fishbase.org]Golley. F.B. (1993) A history of the ecosystem concept inecology: more than the sum of its parts. YaleUniversity Press, New Haven and London, 254 pp.Grassle, J.F. and Stocks, K.I. (1999) A global oceanbiogeographic information system (OSIS) for theCensus of marine life.  Oceanography 12(3),  12-16.Grove, J.S. (1985) Influence of the 1982-1983 El Niño eventupon the ichthyofauna of the Galápagosarchipelago. In: G. Robinson and del Pino, E.M.(eds.) El Niño in the Galápagos Islands: the 1982-1983 event. Publication of the Charles DarwinFoundation for the Galápagos Islands, Quito, 191-198.Grove, J.S. and Lavenberg, R.J. (1997) The Fishes of theGalápagos Islands. Stanford University Press,Stanford, 863 pp.Kleypas, J.A., McManus, J.W. and Meñez, L.A.B. (1999)Environmental limits to coral reef development:where do we draw the line? Amer. Zool. 39, 146-159.Levin, S.A. (1990) Physical and biological scales and themodelling of predator-prey interactions in LargeMarine Ecosystems. In: K. Sherman, K., Alexander,L.M. and Gold, B.D.  (eds) Large MarineEcosystems: patterns, processes and yields. Amer.Ass. Adv. Sci. Washington, D.C. , 179-187. Longhurst, A.R. (1995) Seasonal cycles of pelagic productionand consumption. Progress in Oceanography 36, 77-167.Sea Around Us Project Methodology Review22Longhurst, A.R. (1998) Ecological Geography of the Sea.Academic Press, San Diego. 398 pp.Longhurst, AR., Sathyendranath, S., Platt, T. and Caverhill,C.M. (1995) An estimate of global primaryproduction in the ocean from satellite radiometerdata. Journal of Plankton Research 17,  1245-1271.Mahon, R. (ed). (1987) Report and Proceedings of the ExpertConsultation on Shared Fishery Resources in theLesser Antilles. FAO Fish. Tech. Pap. 383, 278 pp.McManus, J. (1992) The Spratley Islands: a marine parkalternative. NAGA, the ICLARM Quarterly 15(3), 4-8.Munro, J.L. (1996) The scope of coral reef fisheries and theirmanagement. In: Polunin, N.V.C. and Roberts, C.M.(eds) Reef Fisheries. Chapman & Hall, London, 1-14.NRC (1999) Sustaining Marine Fisheries. National ResearchCouncil. National Academy Press, Washington, D.C.164 pp.Pauly, D. (1998) Tropical fishes: patterns and propensities. In:Langford, T.E., Langford, J. and Thorpe, J.E. (eds)Tropical Fish Biology. J. Fish Biol. 53 (Suppl. A), 1-17.Pauly, D. (1999) Review of A. Longhurst’s “EcologicalGeography of the Sea.” Trends in Ecology andEvolution 14(3), 118.Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. andTorres, F.C. Jr. (1998a) Fishing down marine foodwebs. Science 279, 860-863.Pauly, D., Trites, A., Capuli, E. and Christensen, V. (1998b)Diet composition and trophic levels of marinemammals. ICES Journal of Marine Science. 55, 467-481.Pauly, D., Christensen, V. and Walters, C.J. (2000) Ecopath,Ecosim and Ecospace as tools for evaluating theecosystem impact of fisheries. ICES J. Mar. Sci.  57,697-706.Polunin, N.V.C. and Roberts, C.M. (eds). Reef Fisheries.Chapman & Hall, London, 477 pp.Platt, T., Caverhill, C.M. and Sathyendranath, S. (1991) Basin-scale estimates of oceanic primary production byremote sensing: the North Atlantic. Journal ofGeophysical Research 96, 15147-15159.Platt, T. and Sathyendranath, S. (1988) Oceanic primaryproduction: estimation by remote sensing at localand regional scales. Science 241, 1613-1620.Platt, T. and Sathyendranath, S. (1999) Spatial structure ofpelagic ecosystem processes in the global ocean.Ecosystems 2,  384-394.Platt, T., Sathyendranath, S., Ulloa, O., Harrison, W.G.,Hoepffner, N. and Goes, J. (1992) Nutrient controlof phytoplankton photosynthesis in the WesternNorth Atlantic. Nature 356, 229-231.Sala, O.E., Chapin, F.S., Armesto, J.J., Berlow, E., Bloomfield,J., Dirzo, R., Huber-Sanwald, E., Huenneke, L.F.,Jackson, R.B., Kinzig, A., Leemans, R., Lodge, D.M.,Mooney, H.A., Oesterheld, M., Poff, N.L., Sykes,M.T., Walker, B.H., Walker, M. and Wall, D.H.(2000) Global Biodiversity Scenarios for the Year2100. Science 287, 1770-1774.Sathyendranath, S. and Platt, T. (1993) Remote sensing ofwater-column primary production. ICES Mar. Sci.Symp. 197, 236-243.Sathyendranath, S., Platt, T., Caverhill, C.M., Warnock, R.E.and Lewis, M.R. (1989) Remote sensing of primaryproduction: computations using a spectral model.Deep-Sea Research 36, 431-453.Sherman, K., Alexander, L.M. and Gold, B.D. (eds). (1990)Large marine ecosystems: patterns, processes andyields. Amer. Assoc. Adv. Sci. Washington D.C. 242pp.Sherman, K., Alexander, L.M. and Gold, B.D. (eds). (1993).Stress, mitigation and sustainability of large marineecosystems. Amer. Assoc. Advan. Sci., WashingtonD.C., 376 pp.Sherman, K. and Duda, A.M. (1999) An ecosystem approach toglobal assessment and management of coastalwaters. Marine Ecology Progress Series. 190, 271-287.Smith, S.V. (1978). Coral reef area and the contribution ofcoral reefs to processes and resources of the world’socean. Nature 273,  225-226.Sverdrup, H.U. (1953) On the conditions for vernal bloomingof the phytoplankton. J. Cons. Perm. Int. Explor.Mer. 18, 287-295.Trites, A. and Pauly, D. (1998) Estimates of mean body weightof marine mammals from measurements ofmaximum body length. Can. J. Zool.  76, 886-896.Walters, C., Christensen, V. and Pauly, D. (1997) Structuringdynamic models of exploited ecosystems fromtrophic mass-balance assessments. Rev. Fish Biol.Fish.  7(2), 139-172.Walters, C., Pauly, D. and Christensen, V. (1999) Ecospace:prediction of mesoscale spatial patterns in trophicrelationships of exploited ecosystems, withemphasis on the impacts of marine protected areas.Ecosystems 2,  539-554.Wells, S.M. (ed) (1988) Coral Reefs of the World.International Union for the Conservation of Nature,Gland, and United Nations EnvironmentProgramme, Nairobi. 3 Volumes.Reconstructing Fisheries Catch and Effort23THE BASIS FOR CHANGE: PART 1RECONSTRUCTING FISHERIES CATCHAND EFFORT DATAR. Watson, S. Guénette, P. Fanningaand T.J. PitcherFisheries Centre, UBC  &  aDFO, HalifaxABSTRACTRational examination of marine policy requiresan analysis of changes in the abundance ofspecies and marine community structure withrespect to past policy decisions. Abundanceestimates themselves rely heavily on catch andeffort statistics. There are official statistics of fishlandings for many fisheries of the world. Fishingeffort data is generally less available.Unfortunately, for a variety of reasons, landingsdata do not always reflect actual catches well. Forexample, discarded catches are left out of officialstatistics, which developed primarily todemonstrate the value of commercial landings.Illegal or unmandated (not subject to regulatedreporting requirements) catches are seldomdocumented except in candid stock assessmentdiscussions of major species. Through anexhaustive compilation of existing data sourcesand with the assistance of expert local consultantsand/or partnerships, we can develop databasesthat present a more complete and accuratepicture of the catches of marine species, includingthose of limited commercial significance. Theimportance of this process is demonstrated by ourexample from the Canadian North Atlanticfisheries. In this case a partnership arrangementhas allowed the inclusion of the discards of fishes,crustaceans and marine mammals based onobserver data. An outline of the database requiredto include and document ‘adjustments’ to officialstatistics is presented. This work will be extendedto the entire North Atlantic region and beyond.INTRODUCTIONThis paper describes the methods employed inthe collection, organisation and adjustment offisheries catch and effort data used in the ‘SeaAround Us Project’. We will elaborate on ourgeneral approach using examples from Canada’sDepartment of Fisheries and Oceans (DFO) andthe Northwest Atlantic Fisheries Organisation(NAFO), and the International Commission forthe Exploration of the Sea (ICES).Many countries, particularly those fishing in theNorth Atlantic, have an excellent record ofcollecting and reporting fisheries statistics. Therehave, of course, been inevitable shifts in theformat and content of these reports over the longtime that seafaring nations have been reporting.These changes have caused numerous problemsfor the interpretation and analysis of this valuableinformation. Impacts have affected: speciesaggregation (typically with commercial speciesgroups now being separated into species), speciesidentification, the units of measure, definition ofstatistical areas (or their replacement with newsystems), degree of coverage of the data collectionsystem and other important measures.Commonly fishing effort data suffer more, asstatistical systems evolve with developments invessels, gear, and fishing practices. Fishing tacticsand techniques change over time with targets andfishing areas, which means that ‘days at sea’ mayhave different interpretations. Some importantmeasures are difficult to obtain.Above all, the purpose for which the statisticswere collected has changed in many cases.Initially to show the value and development offisheries, these statistics are now used to manageand maintain stocks. This shift in objectives hascaused many distortions, but fundamentally thechange is, that before we wanted to know whatwas landed, now we want to know what waskilled. That is, we require the inclusion of allsources of fishing mortality needed to assess theresource, especially if this is to be done in someecosystem-based way1. Patterns of discarding(often altered by quotas or market factors), aswell as unreported, misreported and unmandatedcatches are now important. The spatialdistributions of fish stocks and fishing havebecome important as our understanding of thefishery and the biological processes hasdeveloped. These distributions have becomevaluable as we seek to manage individual fishstocks independently. Our ability to discernspatial patterns is limited by the spatial resolutionof catch data and this has varied over time.Quotas, other management measures andinternational arrangements can all influence thespatial distribution of fishing, regardless of theunderlying distribution of the biological stock.The historical statistics from most countries havefocused on the major commercial fisheries, andhave largely ignored small scale, artisanal and                                                       1 Especially if the impacts of fisheries on marine ecosystemsare to be evaluated. For example, total fish extractions arerequired as input to an ECOPATH model.Sea Around Us Project Methodology Review24recreational fisheries. These fisheries aresometimes under jurisdictions other than thetraditional national/international reportingbodies. Local knowledge is valuable in obtainingand interpreting statistics on these fisheriessectors and will be vital in reconstructing anhistorical record for them. The statistics have alsofrequently ignored or combined the informationon catches of less valuable or less abundantspecies. Again, local knowledge may contribute toenhancing our knowledge of catches from thesesecondary speciesThe challenge is to apply informed procedures toimprove and unify these historical statistics.Substituting a good guess as a default value for anelement of the catch is likely more accurate thanassuming by default that it was zero.Documentation of procedures for dataadjustment, the basis for the estimates and theauthority of the advice used in making theadjustments is essential. The process must betransparent, allowing identification of thereported data and the adjustments made to arriveat the final figures. Only with a transparent andfully documented procedure will it be possible forthe agencies that are the primary owners of thedata to assess and provide input on themagnitude and quality of the adjusted figures.Many times these agencies have considerableknowledge about discarding, misreporting andother sources of differences between the totalcatch and the landed catch.The current uses of fisheries catch and effort dataare many and varied. We are attempting to placethe data from the North Atlantic in a system thatis extendible to the fisheries of the world. As such,our choice of coding and structures has reflected adesire to incorporate fisheries from the extremelysmall scale to the largest of the factory ships. Weare seeking to facilitate analysis of the energyconsumption versus production in the fisheries(see Tyedmers, 2000) and this has demanded anapproach to reporting fishing effort that is widelyapplicable (horsepower-days) and based on datathat are widely available. Users of catch data willwant to reconstruct historical and spatial patternsof exploitation, and in the case of modelling,fishing mortalities. For these purposes, we wantto know all sources of fishing mortality (includingthose not reported in official landings), in thelocation in which they occurred. The spatial scaleof data required for these models varies, but wewill strive to produce statistics at the smallestpractical scale to allow them to be integrated withlarger scale ecological/oceanographic processes(see Pauly et al., 2000).Our terminology for the various data resources wewill discuss is as follows. A dataset refers to thedata holdings of a given agency and may include anumber of databases. For example, the DFOdataset includes a catch/effort database, anobserver database, survey database and manymore. A database is a single, coherent collectionof data records that will usually be stored inseveral tables with relational links. Within adatabase all records will share common codingschemes, units and other standards.Starting with the global dataset from the Foodand Agriculture Organisation (FAO) for worldfisheries landings, we have merged in other majordatasets that have a geographically narrowerscope and finer resolution such as those fromDFO, NAFO and ICES. In addition to the greaterdetail and resolution, these often provideinformation on fishing effort. These datasets mayalso indicate different values for the landingsdata.To this composite of databases we must make‘informed’ adjustments and additions. Theseconsist of additional data such as estimates ofdiscards and other unreported catches.Justifications for these adjustments to the‘official’ statistics come in a variety of forms. Inthe best cases, these adjustments will come fromreliable and documented sources such as observerprograms, but which are not included in officiallandings. In other cases, they may arise from ageneral discarding rate estimated for a specificfishery, or from estimates of illegal catches fromindustry or government sources. In some cases,such as ICES stock assessments, these additionalsources of fish mortality have been compiled andare used in the stock assessments but are notavailable in official statistics. In many of thesearrangements, the statistics supplied by memberstates cannot be officially altered even when theystretch the bounds of credibility. In all these casesit is our intention to make the appropriateadjustment, and to credit the source anddocument the methods used.Data typesFisheries data sources contain information ofdifferent types, including estimates of landings,measures of effort and a variety of classifiersdescribing the effort, such as the gear used, thearea fished and others. In addition, some datasources attach estimates of economic value orprice to the estimates of catch. Integration of datafrom the various sources, and subsequentadjustments, depends on the standardisation ofReconstructing Fisheries Catch and Effort25measures and definitions for all the data typesand sources.Catch and LandingsThe most important and fundamentalinformation about fisheries for managementpurposes is the total catch (Gulland, 1983; Pauly,1998). Catch is usually classified by species, area,fishing gear used, and other factors. What isofficially reported as ‘catch’ should be nominalcatches (the live weight equivalent of thelandings). Data on the weight or numbers ofanimals that were taken but later discarded, evenif collected, are not included as these statisticsthat were designed to describe the contribution offisheries to the food supply and nationaleconomies. The reported catches may be theresult of a census of fishing vessel landings,survey sampling, reporting by fishers, orestimated by proxies such as fishing effort. Forthis paper the reported landings, are nominalcatches, and are treated in metric tonnes (t) oflive weight (mass) equivalents.Catch statistics are important for three reasons(1) the gathering of statistics increases knowledgeof the fishery (tracking of vessels engaged infishing, dockside sampling of these same vessels,etc.), (2) total catches determine the scale of thefisheries, both within and between sectors, interms of their production and value; and (3)examining time series of catches allows for first-order assessment of fisheries, and of the status ofthe species and populations (stocks) upon whichthe fisheries depend (see Grainger and Garcia,1996). Finally, assessments of fisheries and theirimpacts on fish stocks and the environment haveevolved to include other sources of information.However, basic catch statistics are still essentialto the process (see Alder et al., 2000).Fisheries catches may be separated into threecomponents: (1) nominal catches, reported to(and by) a monitoring agency (e.g. by membercountries to FAO), (2) discarded bycatch, thenon-targeted part of a catch, often consisting ofthe juveniles of targeted or other species, caughtdue to the unselective nature of the gear used,and usually thrown overboard rather than landed;and (3) an unreported component, consisting ofcategories not covered by the reporting system inquestion (examples may include sport fisheries,artisanal fisheries, or illegal catches). Thus, thislast group may be composed of catches that agiven agency is not mandated to gather andreport (‘unmandated catches’), and of catchesthat are misreported by fishers or others. A majortask of our current work is to estimateunmandated and misreported catches, with bothrequiring the development of new protocols (seePitcher and Watson, 2000).Each fishery statistical system we deal with hasevolved a set of procedures and conversionfactors for reconstituting the original weight offish landed in a wide range of product forms. Theconversion factors (e.g. COFREPECHE, 1996)that are used in each agency’s statisticalprocessing are not explicitly considered in ouradjustments. It is obvious, however, that ifinappropriate conversion factors were used by theagencies providing the catch data, this would leadto significant errors in the live weight equivalents(e.g. converting lobster tails to whole bodyweights). Note that under quota managementsystems there may be a tendency for industry toseek adjustment of conversion factors(downward) in circumstances where live weight isbeing over-estimated. There is, however, noincentive for them to seek any adjustment in thecase where live weight equivalents are beingunder-estimated.Value and PriceMuch of the original incentive for governments tosystematically monitor fisheries was to determinethe value and economic development of thefishing industry. In some national systems (e.g.Canada), the estimated value of catches (orequivalently, average price) is recorded with thecatch data. In other systems (e.g. FAO), economicstatistics are generated and reportedindependently of the catch data.EffortAs with economic information, fishing effort maybe measured and classified, by area, gear, etc, inthe same process that records the catch data, orestimated independently. In either case, the effortmust be matched to the corresponding catcheswithin the basic statistical system.The definition of fishing effort, unlike catch, isdependent on the nature of the fishing unit (e.g.boat, trap) and the amount of resources expendedby that unit. The specific effort resourcesexpended are routinely measured in units of timeand/or amount of gear used but alternativedefinitions abound. Our work will use three unitsof effort. The conventional units ‘days fished’ and‘days at sea’ will be compiled directly from thestatistical sources. An alternative unit of‘horsepower-days’ will be the product of thenumbers of days at sea times the averageSea Around Us Project Methodology Review26horsepower of the vessels in the given block ofeffort.Gear specific effort units, such as hours trawledor thousands of hook-hours seem to offer anapparent finer resolution of effort but they are notused here. Although such detail is available formany of the North Atlantic fisheries, this is notthe case for many, or possibly most, fisheries inother parts of the world. Where it is available, theaccuracy of gear-specific effort measures has beenchallenged for many reasons and often by thefishers themselves. Fishers have claimed they hadfalsified the original logbook data to appear tocomply with management restrictions. Regardlessof such a concern, the numbers of days fished is astatistic relatively easy to obtain and difficult tofalsify. Finally, gear-specific effort units are alsodifficult or impossible to aggregate across geartypes, e.g. relating total number of trawl-hours tohook-hours. On the other hand, horsepower-daysoffers a comparatively robust measure that can becompared across most fisheries of the world, eventhose where no vessels are involved.PRIMARY DATA SOURCESThis methodology review paper deals with datafrom four primary sources: FAO, DFO, NAFO,and ICES. Each has its own strengths andweaknesses. In addition, information from otherdatasets, such as those from the U.S. government,the tuna commissions, etc. will be included in ourproject database. These data will be augmentedby smaller, tightly focused datasets, prepared byconsultants for a range of European inshore,small-scale and recreational fisheries.FAOOnly one global database of fish catches presentlyexists from which inferences can be made thatpertain to entire ocean regions: the databaseassembled by the FAO from reports supplied bymember countries (FAO, 1980). This databaseconsists mainly of annually updated catch timeseries, by countries and regions, for the year 1950to the present. The quality of the data therein ishighly variable, and ranges from accurate data ona single species basis for some countries to crudeand over-aggregated estimates for others.Moreover, the catches are not assigned to theExclusive Economic Zones (EEZ) of the countriesfor which they originated, but to the large FAOstatistical areas for which it is reported.Few scientists outside the FAO have made use ofthese statistics to draw inferences on fish stockstatus over large areas of the world’s oceans (butsee Alverson and Dunlop, 1998 for exceptions),but content themselves with citing assessmentsmade by FAO staff. There has been littleindependent validation of this database againstoriginal or other data sources. Perhaps there islittle criticism and crosschecking because somany countries and institutions contribute to thisdataset that has engendered a strong sense ofownership. Nevertheless, its weaknesses in theface of current needs are understood. The FAOAdvisory Council on Fisheries Research hasadmitted; “the current statistics collection systemis limited primarily to landings and commoditystatistics, whereas there is a critical need for datarelevant to fleet capacity, participation infisheries, economic performance anddistribution” (Anon 1997). There have been callsfor the FAO reporting areas and speciesgroupings to be changed to reflect currentfisheries practises, which would facilitate analysisof the economic efficiencies of fisheries(Pontecorvo, 1988). Such changes would probablyfacilitate improved biological analysis as well.Canada Department of Fisheries and Oceans(DFO)This dataset includes records of Canadian (that isCanadian vessels only) commercial catch andeffort per species (marine finfish, invertebratesand plants) for Eastern Canada for years 1986-1998 broken down by spatial statistical regionscalled ‘unit areas’. This dataset is obtained in theZonal Interchange File Format (ZIFF) and hasbeen compiled within the Atlantic Canadianfisheries regions to ensure consistency of codingand units from the four different statistical officesthat operate in the zone. It includes date, targetspecies, unit area, tonnage for each specieslanded, vessel characteristics (tonnage, tonnageclass, length, horsepower) and gear. Records maynot include complete vessel characteristics, andtherefore, horsepower or tonnage may bemissing. The Sea Around Us Project willaggregate the DFO data to the level of effort bymonth, unit area, tonnage class, fishery type andgear type. The catch is further classified byspecies. The DFO catch data includes all fisherycatches with the exception of recreational catches(generally considered small with respect to thecommercial fisheries). There are several small-scale fisheries that either have not collected effortdata or have only begun to do so recently i.e., inthe 1990’s.Reconstructing Fisheries Catch and Effort27Northwest Atlantic FisheriesOrganisation (NAFO) dataThe NAFO dataset includes monthly catch(marine finfish and invertebrates) and effort bydivisions only (which may comprise several unitareas in the DFO system) for Canadian andforeign vessels. The data is structured by fishingcountry, vessel and gear types, and speciestargeted. The information gathered by NAFO is acompilation of the catch and effort as declared byeach member country. NAFO and its predecessorthe International Commission for NorthwestAtlantic Fisheries (ICNAF) provide a consistentstatistical data series since 1960.  In order toprevent duplication of records with the DFOdataset we have removed records of catch/effortby Canadian vessels after 1985. This dataset doesnot include any information for vesselhorsepower. A significant number of foreignvessels have been recorded at one time or anotherin the DFO records of vessel characteristics,including horsepower, and these records will beused to estimate horsepower from the tonnageand other characteristics available in the NAFOvessel records.International Commission for the Exploration ofthe Sea (ICES)North Sea data:ICES data for the North Sea comes from twosources. Electronic data sets exist for all ICESareas (including the North Sea) landings back to1973. These data are broken down by statisticalarea and reporting country for the major species.Data provided to us did not include fishing effortor vessel descriptions. There is no officialelectronic dataset of landings prior to 1973; wetherefore used the records provided in ICES’Bulletin Statistique des Pêches Maritimes (despays du nord de l’Europe) to enter landings forthe North Sea from 1903 until 1974. From thiswritten record we also extracted what exists offishing effort records including breakdowns bytonnage class, and more rarely by vesselhorsepower.Consistency of Data SourcesConsistency comparisons between NAFO andDFO datasets will be made, as will DFO, NAFOand ICES with FAO. This will help to determine ifthe national and international reporting systemsare treating data consistently and completely.Comparisons will be limited to the large-scaleaggregates used in the FAO dataset. Howeverdiscrepancies can be investigated with the moredetailed data available in the other sources.ConsultationsOfficial catch and effort statistics are available formost areas of the world. An aggregated set of thisdata is usually provided to the FAO for inclusionin their global dataset. In order to providecomplete details of fish effort and fishing fleetcomposition, it is usually necessary to accessnational databases directly. In the case ofEuropean Union countries, these data is compiledacross member states and are available on theinternet.Obtaining records of small-scale (typicallyinshore), artisanal, and recreational fishery catchand effort statistics is more problematic. Thisusually requires either a co-operative orconsultative arrangement with someagency/individuals within the country inquestion.  Our project has engaged consultants toreport on the inshore, small-scale andrecreational catch for the majority of maritimenations in the North Atlantic region. At presentwe have consultants working on fisheries inIceland, the U.K., the Irish Sea, Denmark,Norway, France, the Netherlands, Germany,Spain, Portugal and Morocco. Plans exist toextend these efforts soon to Belgium, Russia, andthe Azores and Faeroe Islands.Our co-operative arrangement with the DFO hasallowed estimates of discarding to be made basedon their observer program. These valuablecollaborative arrangements, however, are rare.Alverson et al. (1994) provides a range ofdiscarding for major fisheries that can be appliedwhere appropriate. Such extrapolations fromsimilar fisheries (with respect to gear and targetspecies) must be carefully applied. However,these estimates, untested as they are in mostcases, most likely yield a fairer interpretation oftotal mortalities than ignoring discards wherethey are known to occur.Illegal catches are probably the most difficultinformation to obtain, as they are seldomdiscussed in official statistics (Creed 1996). Someattempts to make allowances for ‘misreported’catches through modelling look promising(Patterson, 1998). Usually these catches can beinferred from other fisheries.  Typically, however,interviews must be conducted with informedsources within the fishery or monitoring agency.Personal networks are invaluable for this. Ageneralized approach based on historical changesin fisheries management or other factors affectingSea Around Us Project Methodology Reviewincentivesto cheat can be informative here (Pitcher andWatson, 2000). Though many official statisticswith the scope of our coverage. Indeed some ofour imperatives required a smaller, less-commercial approach (such as the choice of MSFigure 1. Overall data acquisition, data processing and data management supporting the Sea Around Us Project’sreconstruction of actual fish catches from the North Atlantic.ICES StatisticsCatch, EffortValue…ClassifiersClassified and summarised to SAUP categories for the ICES areaCodestics – excl. Canada 86-98Classified and summarised to SAUP categories for  all NAFO except Canada 86-98CodesCatch, EffortValue…ClassifiersCanadian Statistics – ZIFF dataClassifiedand summarised to SAUP categories for Canada 1986-98CodesCatch, EffortValue…ClassifiersSAUP Master Statistics DatabaseClassified and summarised to SAUP categories with standard coding and unitsCodesCatch, EffortValue…ClassifiersFishBase/FAO Statistics DatabaseClassified and summarised to FAO caISSCAAP standard coding and unitsCodesCatch, EffortValue…ClassifiersSAUP Revised Statistics DatabaseClassified and summarised to FAO categories with ISSCAAP standard coding and unitsCodesCatch, EffortValue…ClassifiersSAUP adjustments can be made using the FishBase analysis routines.Fisheries Observer DatabasesClassified and cards  and other statistical  CodeCatch, EffortDiscards … Satellite databases constructed from data obtained from the relevant statistical officesAdjustmentMisreportedUnmandate dUnreportedaNAFO Statistics – excl. Canada 1 6-98to SAUP categories for  all NAFO except Canada 1986-98– ZI F data - 1998Catch, Value…EffortClassifiersIndividual processes to convert satellite database formats to SAUP Master database formats.to FAO categories with Classified and summarised ISSCAAP standard coding and unitsCompar isons to the FAO statistics and the revised statistics based on estimates of discards  and other statistical  anomaliesdjust entsdOther Data Sources Catch, Value…EffortClassifiersCatch, Value…EffortClassifiersCatch, Value…EffortClassifiersCatch, Value…EffortClassifiersCatch, Value…EffortCl ssifiersCatch, Value…EffortCla sifiersto FAO categories with Classified and summarised ISSCAAP standard coding and unitsClas ified and summarised  and summarised Classified and summarised may be difficult to access the trend is for this tochange. ‘Freedom of information’ acts haveremoved legal impediments, and improvedAccess® as our database).The first imperative was the strong desire to28information technology has contributed towidespread and simplified access in severalcountries including Canada and the U.S.Nevertheless the work of key individuals in eachcountry being reviewed is invaluable. It providesa means of contacting local artisanal andrecreational fisher groups who can be very co-operative if properly approached. It provides ameans of accessing available port records, someof which have impressive historical spans. Mostimportantly, having a person within the countryallows fishers and government officials to beinterviewed ‘off the record’ so that estimates ofillegal fishing, discarding and other vital‘unofficial’ statistics can be elucidated.DATABASE REQUIREMENTSDesign of the project’s principal, i.e. ‘Master’,database was constrained by several imperatives.Unlike a conventional database developed by agovernment department or a business we couldnot scale our resources (money, personal etc.)provide an output of summarised catch datacompatible with the 15,000 species of marinefishes included in FishBase (Froese and Pauly2000; www.fishbase.org). This would allow thewealth of descriptive data (taxonomic, life history,occurrence etc) and the significant investment inanalytical procedures (trophic level comparisonsetc.) to be utilised. Likewise we also wished tomaintain, as far as possible, compatibility withthe FAO global dataset including its ISSCAAPspecies codes. Updates from FAO will be valuableto our future work, as no other agency has amandate or the resources to produce a trulyglobal dataset.The second imperative was for the database toallow allocation of catch and effort to spatialstrata representing functional ecological entitiessuch as large marine ecosystems (see Pauly et al.2000). Meta-analysis of spatial data wouldcertainly require the use of geographicalinformation systems. Our database must facilitatethe use of the data by experts developing modelsReconstructing Fisheries Catch and Effort29of marine ecosystems.Thirdly, we wanted to maintain a system of‘satellite’ databases (Figure 1) recording the bestestimates of catch and effort as supplied by thesource agencies. Each satellite database willretain the codes, units and standards of its sourceagency, but the records will be processed to theSea Around Us Project codes, units and standardsinto our master database. Thus each originalsatellite database record will be associated  with aMaster database record where all the subsequentadjustments and additions can be made in arigorously documented manner. In this way an‘audit trail’ will exist to link incoming data to ourfinal estimates. It should be noted that theresolution of the data in the satellite databases(spatial, temporal, effort, gear, etc.) may vary,however, all will be processed to the standardresolution of the project master database as theyare loaded.The fourth and final imperative was that asrapidly as possible the information would beavailable to all, preferably on the Internet andwith a map-based interface. In this way it wouldbe used/improved by experts, and contribute todebates on the state of marine systems andmarine policy in general.DATABASE STRUCTURECatchSpecies CodesThe database utilises the ISSCAAP codes used byFAO, but will allow synonymy with other codingsystems. The codes, broadly compatible with thefish classification in FishBase (itself based on thatof the California Academy of Science), will allowidentification at a variety of taxonomic levels andallow processed products to be differentiatedfrom whole products.Catch ValueCatch values, based on average prices notcorrected for inflation, for three broad periods(1950s, 1970s and 1990s), as well as their majormarkets (Sumaila et al. 2000), will be includedfor each taxonomic group to allow estimation ofcatch values.Fishing EffortThe method of describing fisheries and fisherieseffort used in this project was designed to beextendible to other fisheries around the world,some of which are very different from those dealtwith in the North Atlantic. The ‘taxonomic’approach allows any fishery to be characterizedby its basic gear type, its location, the tonnageclass of vessels used (if any), and the major targetspecies. This system draws upon the descriptionsof world fishing gear by Brandt (1984).   Thosefisheries that can be confused by two of thesedescriptors can be separated by the third. Weplan to further characterize these fishing effortgroupings by the average ‘catching power’, that isthe amount of the target species typically landedfor each fishing day or day at sea (whenabundance is high) with the usual number andconfiguration of gear units (hooks, nets,whatever) employed. This will facilitatecomparison of small and large-scale fisheries (seeRuttan et al., 2000).Time PeriodsThough some data sets provided to us, such asthat from DFO Canada, contain detailed fishingeffort aggregated to month, we have furtheraggregated these records to annual records as weare primarily interested in examining changesover longer periods. The original monthlyinformation will assist in studies of the seasonalaspects of these fisheries, and allow us toformulate a more precise spatial allocation ofcatches and fishing effort. Some of the Canadiandata is available by fishing gear set by the date ofthe fishing activity rather than the date oflanding.Fishing AreasAs with catch data, it is important to be able toaggregate fishing effort into spatial definitionssuch as large marine ecosystems (see Pauly et al.,2000) that we believe to be the correct scale toexamine the impact of management changes.Data from DFO Canada was provided by ‘unitarea’, these are smaller areas that nest withinNAFO statistical areas. ICES data were brokendown into ICES statistical areas. Where possibleand appropriate, expert consultation will be usedto determine appropriate rules to allocate catchand effort to the smaller units which will facilitatetheir re-aggregation into units of ecological ormanagement significance (for example LargeMarine Ecosystems).Gear typeAlthough all the statistical systems record a widevariety of fishing gears, we have grouped theminto a much smaller number (see list below)which ignores the details of gear construction butis based on the primary mode of fish capture orgear operation. For example, hand line andlongline are in the same category because theirefficiencies do not depend upon a particular boatSea Around Us Project Methodology Review30size. Harpoon and spear are quite rare in this dataset and include sealing and swordfish. The dredgegroup includes both hand-held and mechanicaldevices because the hand-held one is rare, andeach of them will be used with distinctive vesselsizes.Gear types are:• bottom trawls,• midwater trawls,• mobile seines,• surrounding nets,• gillnets and entangling nets,• hooks and lines, trap and lift nets,• dredges,• grappling/wounding, harpoons andspears, and• other gear.Fishery TypesAlthough approximately 50 species account for95% of the nominal catch reported to FAO fromthe North Atlantic since 1950, there are still manymore species that are caught but not landed orreported. In most catch statistics, the targetspecies is in fact the more abundant species on atrip by trip basis.  This is sometimes called the‘main species caught’. There are exceptions,however, such as tropical shrimp fisheries, whichoften take many times the quantity of smalldemersal fin fishes than shrimp target species.The number of target species for the fisheries ofthe world is potentially a very long list. Thus, it ismore useful to group the target species intobroader fishery types that reflect the choices thatfishers are really making: fishing for groundfish,small pelagics, squids, etc. (Table 1).With these categories, the assignment of effort tofishery types is less subject to interpretation thanthe assignment to species sought or even ‘mainspecies caught’ groupings. The fishery typesdefined here serve as links to observer data thatwill provide a minimum estimate of discardsproduced by each fishery.Vessel SizeThe most widely available descriptor of vessel sizeis its overall length. Unfortunately, trends invessel design, at least for the North Atlantic, haveresulted in large increases in the displacement ofvessels within regulated length groups. As aresult, the relative fishing power over time is bestdescribed by tonnage. There are long standingtonnage classes (Table 2) in use on both sides ofthe Atlantic and they are used here. Wherenecessary, we will convert from vessel length tovessel tonnage.AdjustmentsAdjustments made to catch and effort data (asimported from ‘satellite’ databases) will bedocumented on a species-time-area-gear basis sothat changes in values can be review and updated.CASE STUDY:CANADIAN NORTH ATLANTIC(DFO AND NAFO DATA)Our approach can be illustrated by the process ofreconstructing the catch and effort for the NorthAtlantic region under the jurisdiction of Canada’sDFO and NAFO. This case demonstrates what ispossible under co-operative arrangements in‘data-rich’ fisheries.  In other circumstanceswhere the unaggregated data are not available,approaches based on more generalconsiderations, e.g. average rates of discarding foraggregated areas or times (Alverson et al., 1994),are necessary. Even in circumstances whereconventional datasets are complete and wellmaintained, the difficult task of estimating totalsfor discards, misreported, and unreported catchesmay call for a different approach (Pitcher andWatson, 2000).Table 1. Fishery types for the North Atlantic.Groundfish Demersals e.g. cod,flounders, redfishSmall Pelagics herring, mackerelLarge Pelagics tuna, swordfishSharks and Skates Porbeagle, dogfishFreshwater or Diadromous alewife, smelt, eelsBivalves clams, quahaugsScallopsSquidLobsterShrimpCrabMiscellaneous Seaweeds, lumpfishTable 2. North Atlantic vessel tonnage classes.Not known (for Canada most are 0-24.9)0-49.90-24.925-49.9This split used in Canadaonly50-149.9150-499.9500-999.91000-1999.92000 orgreaterReconstructing Fisheries Catch and Effort31Species IdentificationSpecies codes and names were rendered uniformacross three data sets: the DFO research,Canadian commercial catch, and NAFO. Codinginconsistencies were traced and corrected whenpossible. Because they are the direct links withthe observer data, the DFO research species codeswere kept. Nine categories were added to theresearch species list: marine plants, sub-productsof already accounted for catches (e.g. seal and codliver), and unconvertible products (Table 3). Theunconvertible products category refers toproducts for which the yield of processedproducts varies significantly across area, fishers,and time, so as to make difficult the estimation ofan accurate live weight. Marine mammals catcheswere not part of the fishery data and aredescribed in a specific section (see Appendix 1).Catch reconstructionFor the years 1986-1998, the Canadian catch wasobtained from the ‘zonal interchange files’ (ZIFfiles) while the foreign catch came from NAFOdata. The foreign catch is defined as the catchreported by vessels registered in other countries.For the years 1960-1985, all catches wereobtained from the NAFO data set.Catch was compiled by year, month, country,NAFO division, unit area, vessel size, gear type,fishery type, and species.Not all aspects of marine harvest are coveredequally by the DFO database. One componentthat is missing is the take of marine mammals,especially seals. A reconstruction of seal harvestsis described in Appendix 1.Effort ReconstructionCanadian data 1986-1998Effort is defined as the number of horsepower-days, that is, the number of days spent fishing(includes searching and fishing) or days at sea(includes fishing days plus travel time),structured by gear, year and month. The directapproach would normally be to match vesselcharacteristics to each fishing trip. However,because of frequent missing vessel characteristics,and because small Newfoundland vessels werenot individually linked with their catches, severalintermediate steps were necessary to generateestimates of horsepower.Where missing, the vessels horsepower wasreplaced by an estimated value based on a linearregression using vessel length and tonnage.Vessels present in the database (actually fishingor not) were used if complete information fortheir length, tonnage and horsepower wasavailable. A preliminary exploration of the datashowed a skewed distribution for horsepower,warranting a fourth root transformation tostabilize the variance. The resulting linearpredictor of transformed horsepower wasHPl = 1.844+ 0.0379 * length - 0.0017  *tonnage… 1)Retransformation, accounting for theretransformation bias givesHP = HPl 4 + 6σ 2HPl 2  + 3σ 4 …2)An alternative estimation of the horsepower usinga generalised linear model and appropriate linkmay provide for improved precision without theproblems of retransformation bias. A comparisonof these two estimation approaches will be madefor these data.For each trip, horsepower was obtained by usingthe horsepower attributed to the vessel thatreported that catch. Missing horsepower werereplaced by the average value computed for eachstratum (combination of year, month, tonnageclass, gear type, and fishery type). The remainingmissing values were replaced by averaginghorsepower over progressively largercombinations of categories (blocks of effort) untilall missing values were estimated (Table 4.)Effort was then computed as the amount ofhorsepower multiplied by the number of fishingdays spent. Because effort was often missing, adistinction is made between catch associated withand without effort so that total effort could bescaled from reported effort.Effort adjustment for catch without effortTotal effort will be computed for each effort cellas the total catch for all species divided by theTable 3. Categories of products unconvertible intolive weight biomass.Description ExamplesSub-products of catchesalready accounted forelsewhereseal and cod liver,seal oilUnconvertible products sea urchins roeMarine plants kelp, Irish moss,rockweedSea Around Us Project Methodology Review32catch rate for all species. Each cell will have aunique factor and will be referenced in thedatabase to our methods. Application will dependon the data source as some may have alreadyapplied effort prorating.DiscardingObserver ProgramsThe use of at-sea observers is a widespreadpractice in large-scale fisheries. In the NorthwestAtlantic there are observer programs operated byCanada, based in Nova Scotia, Newfoundland andthe Gulf of St. Lawrence, and by NAFO. At-seaobservers supplement the much more limitedamount of surveillance conducted by the fisheriesenforcement agencies. Observer data contains avoluminous amount of information but caution isrequired in the analysis. Observers are notdeployed in a random manner, nor in fact is thereusually a sampling design intended to minimisevariance or control bias. Observers are oftendeployed in a 'tactical' manner, meaning theenforcement agency is concerned about aparticular area or the fishing practices of aparticular vessel, and send an observer inresponse. Observer data has been challenged overthe years with accusations of corruption.However, very few of these have ever beensubstantiated. The greatest challenge in analysisof observer data is the effect that the presencealone of an observer has, or may have, on thefishing practices of a vessel, i.e., an observereffect. One expected observer effect would be forcaptains to not commit infractions of theregulations while carrying the observer.Fisheries observers routinely spend a full trip onboard vessels. They record the positions fished,the effort used and the composition and fate ofthe catch taken. The data is far more detailedthan is possible to collect with a logbook and isindependent of either willful or negligentinaccuracies on the part of the ship's Captain. Thecharacteristics of the vessel and gear are recordedat the beginning of the trip and any gearmodifications are recorded as the trip progresses.The effort is recorded by date, time and position(latitude/longitude), the amount of gear andduration fished, conditions of weather and seaduring fishing and any gear damage or otherevents arising during fishing. The catch isobserved and total catch for each species isestimated, including amounts kept and discarded.As many vessels operate 24-hours a day, theobservers update their records from the logbookwhenever they were off-duty.Many species that never appear in the reportedcatch statistics are recorded by the observers, forexample, on the eastern Scotian Shelf (NovaScotia, Canada) in 1986 there were 125 speciesobserved in the catch, while there are only 42reported in the corresponding NAFO database.Some of these 42 include groups of species suchas sharks that are routinely separated byobservers but others, such as skates, arecompletely unaccounted for.A description and application of the approach tocatch adjustment using observer data is workedout for a particular block of data in the followingsections. The data is from 1986 and covers thegroundfish fishery on the eastern part of theScotian Shelf in NAFO divisions 4V and 4W(Figure 2). The catch statistics are obtained fromthe NAFO databases maintained by the CanadianDepartment of Fisheries and Oceans (DFO).These data have been included in The Sea AroundUs database as described above. The observerdata comes from the DFO observer databasemaintained at the Bedford Institute ofOceanography by Marine Fish Division,Maritimes Region of DFO. Most of the analysiswas completed using database queries inORACLE although the same results could beobtained by various other means. The block ofdata was selected for this example because it isdata-rich and allows a good demonstration of themethods. It is acknowledged that many otherfisheries, areas and times are not as well covered,and adjustments to such catches will be based onless data and broader application of the meancatches from other places and times.Table 4. Procedure used to estimate the averagehorsepower in each block of effort.Descriptors used Remainingmissing valuesYear, month, NAFO divisions,tonnage class, gear type, fishery47,127 (41%)Year, NAFO divisions, tonnageclass, gear type, fishery9,736 (9%)NAFO divisions, tonnage class,gear type, fishery5,061 (4%)NAFO divisions, tonnage class,gear type674    (<1%)Tonnage class, gear type 0Reconstructing Fisheries Catch and Effort33A further point regarding the example block ofdata is that during 1986 it was not illegal todiscard fish, in any quantity and of any species.For this reason, there was less likely to be anobserver effect limiting this behaviour. It hasbeen reported that fishing captains were stillreluctant to discard excessively when observerswere present, even in the absence of any specificregulation against it. Consequently, an observereffect, inhibiting discarding practices, cannot beruled out and so the estimated discards, even forthis period of time, must be consideredminimum estimates, especially for target, high-valued species.Estimation of Observer CoverageProportionsAll results from at-sea observers must beinterpreted carefully in light of variable andoften low coverage levels for certain fisheries.We have defined coverage rate to be theproportion of the reported landings for a givenunit of data, i.e., country, area, monthcombination, which was observed as retainedcatch by the observers. The proportions reflectthe total amount of retained catch of all speciesby weight, observed at sea with respect to thetotal amount of landings reported for thecorresponding country, area and month.Proportions greater than 1.0 reflect observedcatches in excess of the total reported landings.Thus, observer coverage proportion, Oc,a,m, on acatch basis is:Oc,a,m=  Σs keptc,a,f,y,m,s / Σs Cc,a,m,s                                       …3)where keptc,a,f,y,m,s is the total observed landingsand Cc,a,m,s is the nominal catch (landings) in theNAFO data.Table 5. Coverage proportions by Canadian fisheries at-sea observers during 1986, for the groundfish trawler fisheryin NAFO 4VW in 1986. Empty cells have no observer coverage, ‘no catch’ indicate that fishing was observedbut no catch was reported to NAFO, grayed cells mean that there was neither reported nor observed catch.Figures represent the ratio of a nation’s observed catch to that nominal catch reported by that nation for thesame month and statistical. Values exceeding 1.0 indicate that the observed catch exceeded the catchsubsequently reported to NAFO.Canada Cuba France Japan USSRMonth 4V 4W 4V 4W 4V 4V 4W 4V 4W1 0.07 1.342 0.09 0.28 0.103 0.14 0.00 0.00 1.40 no catch4 0.02 0.01 0.47 0.775 0.18 0.06 0.58 1.13 0.516 no catch 0.06 1.58 no catch 0.427 0.02 0.02 no catch 1.00 1.94 no catch 0.238 0.09 0.00 0.53 1.68 no catch9 0.14 0.0110 0.08 0.06 1.6511 0.06 0.0012 0.00Figure 2. Area of discard estimation (NAFO’s Divisions 4Vand 4W) as outlined on the eastern portion of the ScotianShelf off Nova Scotia, Canada. 64° 62° 60° 58° 56° 44° 46° 48° 4W 4V Sea Around Us Project Methodology Review34Table 5 highlights several of the problemsinherent in this approach. Cells highlighted ingrey contained neither reported catch norobserved catch, so do not represent a problem.Empty white cells correspond to catches with noobserver coverage at all. Cells indicated as ‘nocatch’ correspond to instances of observersreporting fishing, with retained catches, fromtimes and areas but for which the country inquestion did not report any catch at all. In the sixcases below, it is likely that the vessels were inpart fishing in adjoining areas, i.e., 4V or 4W, andthat their catches were reported as such. Theother problem revealed is the occurrence of casesin which the observed catch exceeds the reportedcatch.Estimation of discard catch ratesIn this example, the observer data (Table 6) andthe corresponding reported catches for thegroundfish trawler fishery in NAFO 4VW in 1986was the compiled weight of both kept anddiscarded catch by month (m), country (c), NAFOarea (a), and species caught (s). Estimates ofdiscard rates, dc,a,m,s, were obtained by linking theobserver program data and the reported catch foreach block of effort,dc,a,m,s = discc,a,m,s / Σs keptc,a,m,s                                           …4)where discc,a,m,f,y,s is the total observed discardsfor species s and keptc,a,m,s is the observedlandings. The estimated discards, Dc,a,m,s, arecomputed as:D c,a,m,s =  dc,a,m,s * Σs Cc,a,m,s                                    …5)Table 6. Estimates of discards (tonnes) by month and species in the groundfish trawler fishery in NAFO area 4VW in1986.  (* refer to unspecified species).CommonName Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecAnnualTotalAlewife 0.05 0.12 0.00 0.00 0.17Americanplaice45.92 9.63 103.54 0.10 14.61 1.80 3.60 26.01 12.17 10.22 0.00 227.59Argentine 0.01 0.06 0.14 1.83 0.02 7.57 0.00 0.07 9.70Bigeye tuna 0.00 0.00 0.00Bluefin tuna 1.28 0.00 0.00 0.00 1.28Cetaceans 3.69 3.69Cod 68.70 234.49 281.96 0.01 55.72 1.20 170.69 292.89 39.22 50.66 0.17 0.00 1195.72Crustaceans 1.80 0.41 0.47 0.44 6.25 15.94 5.79 2.49 0.32 0.19 34.10Cusk 0.00 2.08 0.00 0.15 0.11 0.51 0.00 2.60 0.00 0.18 0.00 0.00 5.63Dogfish* 27.16 164.92 1608.20 462.03 476.99 33.03 2.56 23.36 0.40 0.67 2799.33Flounder* 0.00 0.29 0.97 0.00 0.37 0.00 0.00 0.00 0.00 0.00 1.63Grenadier 0.00 0.00Groundfish 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00Haddock 48.10 109.63 93.55 0.00 29.06 2.89 50.66 162.14 30.96 89.54 38.00 0.00 654.53Hake* 0.01 0.00 0.30 0.00 0.00 0.00 2.22 0.06 2.60Halibut 0.00 0.25 0.02 0.00 0.00 0.16 0.00 12.60 0.00 0.00 0.00 0.00 13.03Herring 0.00 0.17 0.00 0.67 0.48 0.00 2.53 0.01 11.69 8.00 23.54Mackerel 0.04 0.00 0.09 7.89 9.21 5.36 0.04 0.03 0.00 0.00 22.65Monkfish 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00Other fish sp. 10.84 8.94 1.32 2.64 34.53 11.46 73.77 103.87 13.03 38.45 2.69 301.55Other inverts 0.12 0.01 0.40 0.00 2.91 30.20 0.06 759.98 0.16 4.11 797.94Pollock 11.26 115.62 64.99 0.11 61.17 0.94 60.60 19.64 4.93 6.25 0.00 0.00 345.51Porbeagle 0.89 0.65 1.06 0.12 1.56 4.29Red hake 0.00 75.15 4.58 0.71 1.02 5.54 0.01 3.58 0.02 0.11 90.72Redfish 7.73 3.52 13.16 0.00 6.01 0.40 1.62 85.94 1.59 9.69 0.35 0.00 130.00Seals* 0.11 0.32 4.48 0.79 4.77 0.32 10.78Sharks* 0.04 12.33 1.19 2.98 1.79 0.37 26.66 45.36Silver hake 5.23 17.09 1.52 32.68 47.85 200.94 11.10 11.60 1.90 81.18 0.00 0.00 411.10Skates* 174.03 383.03 501.45 327.25 266.96 6.22 414.42 224.60 244.19 313.38 86.29 2941.81Swordfish 0.00 0.48 0.00 0.18 0.00 0.00 0.00 0.66Turbot 0.00 0.00 0.00 0.00 0.00 0.04 0.00 0.32 0.36White hake 0.01 0.93 0.34 8.69 0.89 5.99 1.52 45.73 2.33 10.45 0.00 0.00 76.88Winterflounder0.00 0.00 0.00Witch flounder 14.96 1.03 54.18 0.02 0.41 0.03 0.04 0.62 0.39 0.03 0.00 0.00 71.72Wolffish* 0.98 1.46 0.54 0.01 1.01 0.10 0.05 27.72 1.47 3.64 0.00 0.00 36.98Yellowtail fl. 2.11 0.02 0.14 0.00 10.46 1.19 90.21 77.68 55.62 71.23 0.00 0.00 308.67 Totals 419.08 1129.97 2735.83 835.72 1031.69 338.99 900.50 1900.51 411.35 730.38 135.50 0.00 10569.30* unspecifiedReconstructing Fisheries Catch and Effort35where Cc,a,m,s is the reported catch in the NAFOdata. The individual estimates of discarded catchare summarised by species and month in Table 6,by month and country in Table 7 and totalled byspecies (Table 8) below. The difference betweenthe total in Table 7 (10592) the other two (10569)occurred because a small amount of catch (23 t)had no species identity assigned to it. This is oftenaccounted for in a category called NEI (notelsewhere included).The analysis presented here will be extended tobetter estimate discards from cells withoutobserver estimates through application ofgeneralized linear modelling. One specificapproach may be a logistic model for the rates.However, other alternatives will also beinvestigated. This approach opens the way tousing the EM algorithm for filling in missing cells.Interviews of fishers participating in variousfisheries provide a semi-quantitative means ofestimating discard. However, these are usuallyspecific to particular times and areas, and greatcare must be exercised when applying them tolarge aggregates of data. For this reason, datafrom this source should be given greater weight asa means of setting ‘anchor points’ in the morequalitative discard estimation of Pitcher andWatson (2000).Illegal CatchesEnforcement and surveillance programEstimates of illegal catches taken by both foreignand domestic vessels could potentially beestimated from fishing vessel surveillance data(DFO Conservation and Protection Branches inthe Atlantic regions). Their data is confidentialand considered sensitive but if kept anonymous,it may be possible to use and analyse their data toTable 7. Summary of example results of observer-based estimates of total discards (tonnes) for all species combined, forthe groundfish trawler fishery in NAFO 4VW during 1986.Canada Cuba France Japan USSRMonth 4V 4W 4V 4W 4V 4V 4W 4V 4WMonthlyTotals1 400 0 19 4192 852 212 67 11303 1090 1632 0 14 0 27364 330 0 505 0 0 0 8365 411 31 311 1 277 10326 0 0 0 243 5 0 92 3407 868 42 0 0 4 0 8 9218 713 1149 18 21 0 19019 387 24 0 0 0 41110 598 128 5 73111 69 66 0 0 13512 0 0 0 0 Totals 5717 3284 0 1078 87 49 0 0 377 10592Table 8. Comparison of estimated discards andreported landings for the groundfish trawlerfishery in NAFO area 4VW for 1986, allcountries combined. The table is ordered indescending order of the proportion (percent)of discards in the total catch (tonnes).Common Name Catch Discard PercentGrenadier 1 0.0 0.0Bigeye tuna 10 0.0 0.0Groundfish (unspec) 76 0.0 0.0Monkfish 2081 0.0 0.0Swordfish 231 0.7 0.3Silver hake 82466 411.1 0.5Mackerel 2202 22.7 1.0Halibut 1132 13.0 1.1Cod 79084 1195.7 1.5Redfish 7621 130.0 1.7Cusk 326 5.6 1.7Pollock 19296 345.5 1.8Flounder (unspec) 68 1.6 2.3Witch flounder 2382 71.7 2.9Turbot 12 0.4 2.9White hake (Squirrel) 2341 76.9 3.2Haddock 16384 654.5 3.8Wolffish (unspec) 309 37.0 10.7Bluefin tuna 9 1.3 12.4Red hake (Squirrel) 257 90.7 26.1Yellowtail flounder 692 308.7 30.8Alewife 0 0.2 100.0Hake (unspe 0 2.6 100.0Cetaceans (unspec) 0 3.7 100.0Porbeagle 0 4.3 100.0Argentine 0 9.7 100.0Seals (unspec) 0 10.8 100.0Herring 0 23.5 100.0Crustaceans (unspec) 0 34.1 100.0Sharks (unspec) 0 45.4 100.0American plaice 0 227.6 100.0Other fish species 0 301.6 100.0Other invertebrates 0 797.9 100.0Dogfish (unspec) 0 2799.3 100.0Skates (unspec) 0 2941.8 100.0Total/Average 216980 10569.5 4.6Sea Around Us Project Methodology Review36obtain an estimate of illegal catches includingareas outside the 200 miles limit (but see Pitcherand Watson, 2000). Such estimates wouldprovide the basis for adjustments to the catchesreported to NAFO.Consultants are currently engaged by the Projectto obtain catch records from the home ports ofPortuguese and Spanish fleets which have fishedin the Northwest Atlantic. These will be matchedto records available from NAFO. Processes likethis will be used to obtain better estimates ofunreported and illegal fishing activities.DiscussionOur approach is ambitious and relies uponconsiderable skilled collaboration. However, theneed for the best, most complete records of totalcatch (=mortality estimates) and fishing effort iscritical for the rational re-examination of fisheriespolicies in the light of historical stock collapsesand current concerns. The collection of basiccatch and fishing effort statistics is expensive andrequires local knowledge to overcome errors incoding and interpretation. The role of consultantsfamiliar with the fisheries in question isimportant to addressing any shortcomings in theofficial datasets and, particularly for inshore,artisanal, or recreational fisheries. Their data willbe matched to records available from NAFO andother agencies.Doubtless, mistakes were made when the officialdatasets were compiled but these are almostinevitable, and pale in comparison to deliberateomissions caused by functional or jurisdictionallimitations. By functional limitations we meanthat the data (at least in a usable form) do notexist, so it cannot be reported (because of thelimited resources available). An example of this isillegal fishing; since estimates of such catches arenot often made, they cannot be reported. Incontrast, where data  available, such as estimatesof discarding, it may not be within the mandate ofthe reporting agency to include them in officialcatches. This is understandable as the need toreport the value of landed catch underlay thegenesis of most the world’s fisheries statisticalsystems and this purpose still dominates allothers. Jurisdictional boundaries, for examplebetween tiers of government, may make theproduction of a comprehensive database, one thataccounts for all sources of fishing mortality at alllife-history stages, very difficult. Our Canadiancase study from the Northwest Atlanticdemonstrates that it is possible to make estimatesof discarding for even non-commercial species ifobserver programs are in place. This would bevery difficult based on the scaling of reportedcommercial landings alone. The estimates wehave reported here can be improved through theuse of general linear models or similarapproaches that would allow estimates of blocksof time and space where no observer data exists.Using these methods it may also be possible tomake estimates of discards for years when theobserver program did not operate.Our estimates confirm that discarding was not aminor phenomenon for vessels operating onCanada’s Scotian Shelf (Table 8). Mortalityestimates for many species would be significantlyincreased if discarding were included. Even soour estimate is acknowledged to be a minimum,especially for target species where we may havesignificant ‘observer’ effects. Our total of 10,569tof discards includes fishes, crustaceans, andmarine mammals. Overall, however, the overalldiscard rate was only 4.9% (total discards/totallandings). There was discarding of major targetspecies such as cod and silver hake. We believethat the discarding estimates for non-target(generally non-quota) species are a minimalestimate. However, it is very likely that observerpresence has caused estimates of the discardingrate of target species to be greatlyunderestimated. These results are, however, onlyfor a small portion of the total North Atlanticfishing grounds and for only one year. Moreover,only the groundfish trawl fishery was examinedhere. Nevertheless estimating discarding rates oftarget and non-target species, even in light ofthese problems, will be required before total catchestimates can be attributed to ecosystems andnations’ EEZs (see Pitcher and Watson, 2000).Alverson et al. (1994) estimated that there werenearly 686,000t per year discarded in theNorthwest Atlantic alone, but unlike our estimate,these did not include marine mammals. Based onour minimal estimates, discards in this region,based on the groundfish fishery alone, would haveexceeded 120,000t for 1997. To reach the total ofdiscards estimated by Alverson et al. (1994) forthis region, our overall discarding rate wouldhave to be nearer 30% than the 4.9% wecalculated for the groundfish fisheries instatistical area 4VW in 1986. Likewise, estimatesof discarding for species never landed would bevery difficult to include without an observerprogramme. Nevertheless, individual estimates ofdiscarding rates for the Northwest Atlanticgroundfish trawl fisheries ranged from one in thetop twenty of those fisheries with the highestReconstructing Fisheries Catch and Effort37“recorded” discard ratios, to four estimates in thelowest ten overall. The highest estimate was 5.28to 1 (i.e. more than  5 kg of discards for each kg oftarget species landed). Of the several NorthwestAtlantic fisheries listed, including the HakeTrawl, Cod Trawl, Redfish Trawl, and PlaiceTrawl, all but the last had ratios below 10%(Alverson et at. 1994). The values reported inAlverson (1994) do not represent overalldiscarding rates but are simply the availableindividual estimates (pers. comm. D. L.Alverson). With 5.8% discarding by weight, theNorthwest Atlantic Cod Trawl fishery haddiscarding rates comparable to those wecalculated from observer data.In addition to changes in the abundance and sizestructure of these species, changes to fishingpolicy, gear configuration, and fishing practisecan greatly alter the numbers discarded. Rationaldiscussion of fisheries policy, and its impacts,requires a reliable time series of discardingestimates. We have shown one example of howestimates of discards can be made with existingdata sources. In ‘data-poor’ fisheries, we mustrely on estimates from similar fisheries or ourknowledge of the marine community that is likelyto be impacted. In many cases the marinecommunity will already have been highly altered– fortunately these changes also can beanticipated.Changes to government policy in many countriesof the world, whether mandated by internationalagreements or otherwise, have meant thatestimates of bycatch and discarding are moreavailable and are now discussed openly whenstock management (for the major species) isconsidered. Illegal catches, on the other hand, arestill taken as an admission of mismanagement,enforcement failure, or industry malfeasance, andare usually not included in official figures. In thecases where these estimates must be made andincluded in discussions, such as in internationalassessments of major commercial species, care istaken not be too specific as to which country’svessels have taken the catch and where it wastaken. Unfortunately not all species currentlyhave even this level of candid analysis. We are inthe process of obtaining estimates of illegal andunmandated catch for the major North Atlanticfisheries but results are slow in coming asnetworks of trust are established. The need toinclude estimates of these catches is not yetuniversally accepted.It is a well-accepted axiom, that first we must findout what is happening before we can plan to doanything about it. So it is also with marine policy.Transparent improvements to catch and effortdata series will improve the quality of the debateand contribute to the sustained development thatthe majority of countries have already adopted asnational policyACKNOWLEDGEMENTSThe authors wish to thank the support of the PewCharitable Trusts to the Sea Around Us Project.For assistance with the Canadian data used in ourcase study, we thank Becky Sjare, Garry Stenson,and Joe Firth (DFO St. John's, Nfld). We aregrateful for the assistance of D. Pauly and thesuggestions of reviewers.REFERENCESAlder, J., Pitcher, T., Preikshot, D., Kaschner, K. and Ferriss,B. (2000) Rapfish estimates - how good is good? In:Pauly, D. and Pitcher, T.J. (eds) Methods for evaluatingthe impacts of fisheries on North Atlantic ecosystems.Fisheries Centre Research Reports 8(2), 136-182.Alverson, D.L. and Dunlop, K. (1998) Status of world marinestocks. Fisheries Research Institute, University ofWashington, FRI-UW-9814, 29 pp.Alverson, D.L., Freeberg, M.H., Murawski, S.A. and Pope, J.A.(1994) A global assessment of fisheries bycatch anddiscards. FAO Fisheries Technical Paper No. 339, 233 pp.Anon (1997)  A report on the First Session of the AdvisoryCommittee on Fishery Research, Rome, Italy, 25-26November 1997. FAO Fisheries Report (571), Rome, 34pp.Anon (1998) Report of the joint ICES/NAFO working groupon harp and hooded seals. ICES, ICES Headquarters,Copenhagen, Denmark, ICES CM 1998/Assess:3, 37 pp.Anon (1999) Report of the joint ICES/NAFO working groupon harp and hooded seals. ICES, ICES Headquarters,Copenhagen, Denmark,, ICES CM 1999/ACFM:7, 33 pp.Brandt, A. von (1984) Fish catching methods of the world.Farnham: Fishing News 3rd Ed. 418 pp.COFREPECHE (1996) Comparative study of conversioncoefficients used for estimating the live weight of fishcaught by Community fishing vessels. Final Report ofProject 95/02, Centre Ifremer de Brest, BP. 70 F-29280PLOUZANE, Commission of the European Communities,Directorate General for Fisheries. 127 pp.Creed, C. (1996) Social responses to ITQs: cheating andstewardship in the Canadian Scotia-Fundy inshoremobile gear sector. In: Report of the second workshop onScotia-Fundy groundfish management. Burke, D.L.,O’Boyle, R.N. and Sinclair, M. (eds) Canadian TechnicalReport of Fisheries and Aquatic Sciences 2100,  72-82Froese, R. and Pauly, D. (eds) (1998) FishBase 2000:Concepts, design and data sources. ICLARM, Manila.FAO (1980) Yearbook of fishery statistics: catches andlandings. Vol. 50. Food and Agriculture Organization ofthe United Nations, Rome. 386 pp.Grainger, R.J.R. and Garcia, S. (1996) Chronicles of marinefisheries landings (1950-1994): trend analysis andfisheries potential. FAO Fisheries Technical Paper No.359, 5 pp.Sea Around Us Project Methodology Review38Gulland, J.A. (1983) Fish stock assessment: a manual of basicmethods. Vol. 1. FAO/Wiley Series on Food andAgriculture. 223 pp.Hammil, M. O. and Stenson, G. B. (in press) Estimated preyconsumption by Harp seals (Phoca groenlandica), Greyseals (Halichoerus grypus), Harbour seals (Phocavitulina) and Hooded Seals (Cystophora cristata) in theAtlantic Canada. Journal of  Northwest Atlantic FisheriesScience.ICNAF (1970) Sealing statistics for 1937-1968. StatisticalBulletin 18, 124-140.Patterson, K.R. (1998) Assessing fish stocks when catches aremisreported: model simulation tests, and application tocod, haddock, and whiting in the ICES area. ICESJournal of Marine Sciences  55, 878-891.Pauly, D. (1998) Rationale for reconstructing catch timeseries. European Commission Fisheries CooperationBulletin 11(2),  4-7.Pauly, D., Christensen, V., Longhurst, A., Sherman, K. andWatson, R. (2000) Mapping fisheries onto marineecosystems:  proposal for a consensus architectureallowing inferences at local, basin and global scales.  In:Pauly, D. and Pitcher, T.J. (eds) Methods for evaluatingthe impacts of fisheries on North Atlantic ecosystems.Fisheries Centre Research Reports 8(2),  13-22.Pitcher, T.J. and Watson, R. (2000) The basis for change 2:estimating the total extractions from marine ecosystems.In: Pauly, D. and Pitcher, T.J. (eds) Methods forevaluating the impacts of fisheries on North Atlanticecosystems. Fisheries Centre Research Reports 8(2), 40-53.Pontecorvo, G. (1988) The state of worldwide fisheriesstatistics: a modest proposal. Marine ResourceEconomics 5,  79-81.Ruttan, L.M., Gayanilo, F. C. Jr., Sumaila R. U. and Pauly, D.(2000) Small versus large scale fisheries in the Gulf ofMaine and George's Bank (U.S.A.): a multispecies,multifleet model for evaluating their potential toconserve fishes and fisheries. In: Pauly, D. and Pitcher,T.J. (eds) Methods for evaluating the impacts of fisherieson North Atlantic ecosystems. Fisheries Centre ResearchReports 8(2), 64-78.Sergeant, D. E., (1991) Harp seals, man and ice. Can Spec PublFisheries and Aquatic Sciences 114,  153p.Stenson, G. B., Sjare, B. and Wakeham, D. (1999) Catch-at-ageof northwest Atlantic harp seals. DFO Atlantic FisheriesResearch Documents, 99/105,  21p.Sumaila, R., Chuenpagdee, R., Munro, G. (2000) Trackingfisheries landings in the North Atlantic. In: Pauly, D. andPitcher, T.J. (eds) Methods for evaluating the impacts offisheries on North Atlantic ecosystems. Fisheries CentreResearch Reports 8(2), 114-122.Tyedmers, P. (2000) Estimating the industrial energy inputsand related ecological costs associated with thecommercial fishing fleets of the North Atlantic.  In:Pauly, D. and Pitcher, T.J. (eds) Methods for evaluatingthe impacts of fisheries on North Atlantic ecosystems .Fisheries Centre Research Reports 8(2), 123-135.Zeller, D. and Pauly, D. (2000) Life history patterns and depthzone analysis.  In: Pauly, D. and Pitcher, T.J. (eds)Methods for evaluating the impacts of fisheries on NorthAtlantic ecosystems.  Fisheries Centre Research Reports8(2), 54-63.APPENDIX 1:  MARINE MAMMALSMarine mammal catches are not recorded in acentral database at DFO as happens for fish andinvertebrates. The majority of seal huntingactivity is aimed at harp and hood seals. The greyseal was the object of a bounty hunt from 1967 to1984. The catches of other species of seals wereaggregated as “other seal”.  The principal task wasto obtain the data from different sources bycontacting researchers working on one or severalspecies. Second, the catch information that wasavailable was the number of animals in the catchby age groups. The grouping differs depending onthe hunting grounds and the species. In order toconvert the catch in number to yield, the meanbody weight for each broad size category wasderived from growth curves in the best cases, andfrom more general data on adult sizes in othercases. Additional sources of removals, largelyunaccounted for, are the animals wounded orkilled but never recovered (“struck and lost” inseal hunt terminology), and are included in thecatch statistics. Research has been undertaken toestimate the number of harp seals it represents(G. Stenson, pers. comm.).Reconstructing Harp Seal CatchData sourcesHarp seal are the most abundant catch of marinemammals in western Atlantic Canada. Since harpand hooded seals are migrating from the Gulf ofSt. Lawrence and Newfoundland to Baffin Bay,southeastern Greenland and Hudson Strait, thecatch of both Canadian and Greenland watersshould be considered as sources of mortality onthe same population. The Joint ICES/NAFOWorking Group on harp and hood seals considerscatches from West Greenland and half ofSoutheast Greenland derived from the NorthwestAtlantic harp seal stock (Anon., 1998). Effort datafor Norwegian and Russian hunting directed onWest Greenland Ice were obtained from AppendixIV of the ICES document  (Anon., 1999). Catch atage for the years 1952-1998 for each region, Gulf,eastern Arctic, and Greenland, were obtainedfrom (Stenson et al., 1999). Catches for years1950-1951 were obtained from ICNAF (1970).Table 9. Weight for each commercial category for Harpseal.Stage Age Weight t(kg) Rationalepups 15 –30days30 Stewart and Lavigne1980; Sergeant, 1991p.27adult ages 5-10mostlikelyassumesex ratio50%57-84 Based on catch at agedata and Gompertzgrowth equation(April weight)Reconstructing Fisheries Catch and Effort39Catch biomassThe harp seal catches are now recorded in twosize categories, pups and 1+. Pup weight wasobtained from Sergeant, 1991 (his Table 8). Meanweight of the catch was obtained by using thecatch-at-age data for years 1952 to 1998 (Stensonet al. 1999) and the weight at age Wa computedfrom a Gompertz curve (Hammil and Stenson, inpress).Biomassy = Catchpup,y*Wpup + Σ ropa,y*Wa*Catchywhere propa,y is the estimated age composition ofthe yearly catch (Catchy), expressed inpercentages.  The age composition for years 1950-1951 was assumed to similar to that of the year1952. The Gompertz curve (Figure 3) wascomputed from specimens examined in Aprilwhen seals are leaner than in the winter(Sergeant, 1991). The resulting mean weight ofadult seals varied from 57 to 84 kg over the years(Table 8). Pup weight was estimated at 30 kgwhere seals are 15 to 30 days old. The huntingperiod varies within Canadian regions but theerror in taking the April weight is probablysmaller than error in the estimation of the catchat age (G. Stenson, DFO, St. John’sNewfoundland, pers. comm.).0 5 10 15 20 25 30 35Age (years)050100150200250300Weight (kg)Grey MGrey FHooded MHooded FHarbour MHarbour FHarp MHarp FFigure 3. Weight at age relationship for Canadian seal                      species.Sea Around Us Project Methodology Review40THE BASIS FOR CHANGE 2:ESTIMATING TOTAL FISHERYEXTRACTIONS FROM MARINEECOSYSTEMS OF THE NORTHATLANTICTony J. Pitcher  and  Reg WatsonFisheries Centre, University of British ColumbiaAbstractThe reason for estimating total extractions of fishis to able to account for their impacts on marineecosystems. Such an evaluation has not beenattempted before, since ecosystem modellingtechniques suitable for this purpose have onlyrecently become available. Putting a figure ontotal extractions entails the difficult task ofestimating, in addition to reported landings,discards, illegal, and unmandated catches,including disreported catches. These unreportedextractions cast various types of shadows, manyof which may be tracked and estimatedquantitatively. Official figures often have animplicit assumption that such categories are zero,an unacceptable option for an ecosystem-basedproject. Some examples of adjustments forunrecorded catches are reported. We describe aninnovative, well-funded NGO that tracks andpublicizes illegal catch in the Southern Ocean andwhich may provide a model for other areas of theworld such as the North Atlantic. We present anadjustment procedure based on a simplespreadsheet, divided into categories ofunreported annual catch. Adjustment factors arebased on reports from observers, confidentialcorrespondents and on information published ina variety of sources. Over time the adjustmentfactors respond to changes in regulatory regimeand hence the incentives and disincentives tomis-report. Once in place, this method providespreliminary estimates that may be refinedwithout disruption. Preliminary estimates, set upas a ‘straw man’ for Atlantic Canada, suggestaverage figures since 1960 of around 30% forunreported extractions of cod and over 100% forherring. Although at first sight an adjustmentprocedure for illegal catch may appearcontroversial, we argue that such transparency isnot only an essential part of a new fisheriesregime that mimimizes deleterious impacts tomarine ecosystems, but is also in conformity withthe treatment of other kinds of fraud incontemporary society.“Shame to him that speaks not forth: for neverwas the time so good as now”Robert le Coq, Bishop of Laon, 1356, denouncingthe anarchy that  prevailed under misrule by theDauphin of France.INTRODUCTIONIn order to evaluate the impacts of fisheries onNorth Atlantic ecosystems, the total annualamount of fish killed, from all species and by eachfishery, has to be estimated. Obtaining thesefigures is not a trivial exercise because someitems are not recorded, for a variety of reasons, inpublished catch statistics. In this paper we aim topresent a methodology for making such estimatesof unreported catch, following on from thedatabase methodology presented in The Basis forChange 1 (Watson et al. 2000). Our analyses willtouch on controversial topics, and can beexpected, in some cases, to be at variance withconventional assessments or official positions.CATEGORIES OF UNREPORTED CATCHFisheries catches may be separated into threecomponents:1) nominal catch, that reported to a monitoringagency, generally to national body that itselfreports to the FAO (Food and AgricultureOrganization of the United Nations);2) discarded by-catch, the non-targeted part of acatch, often consisting of the juveniles oftargeted or other species, caught due to theunselective nature of the gear used, andusually thrown overboard rather than landedand generally. At least in recent years, inNorth Atlantic fisheries, this is estimated bysome sort of observer program;3) unreported catch, consisting of categories notcovered by the reporting system in questionCategory 3, unreported catch, as illustrated inFigure 1, may be composed of:1) unreported discards: fish of species or sizesnot wanted by the fishing vessel. Discardsmay be in excess of quota, high grading, andmay or may not be illegal, but are amountsnot reported by observers.2) unmandated catches: catches that a givenagency is not mandated to report, either onaccount of the small size of the vessel (catch isnot recorded from small inshore vessels inthe UK), or the nature of the species (a by-The Basis for Change41catch of dogfish sharks often goesunreported). It may include discards ofspecies not considered important enough torecord, such as pelagic species like herring insome groundfish fisheries. A further exampleis catch from sport fisheries, which is oftenunmandated (it is not included in the FAOdatabase) but can have significant impacts(see Walters 1995).3) illegal catch: catches that contravene aregulation from the regulatory body. May beunreported by being landed away from thehome port, or trans-shipped to foreignflagged vessels at sea. Includes disreportedcatches: catches whose identity (by species orsize) may be deliberately misreported andconcealed. Disreporting usually concealsquota violations, such as haddock reported ascod, or salmon concealed under surface layersof hake.In the developed countries of the North Atlantic,the catch of fish of each commercially importantspecies is routinely estimated by sampling at theports of landing (Shepherd 1988), but this can bea difficult task, especially with scattered small-scale artisanal fisheries. Most of the abovecategories are missed by official fisheries catchstatistics gathered in many countries, whosestatistical systems were generally set up to tracklandings for economic purposes rather than theamount of fish killed by fisheries. Log books andsales figures kept by fishing captains or ownersprovide an alternativesystem, which has theadvantage of also givingdata on fish discarded atsea before landing, and onfishing effort. Interviewswith fishers may providehistorical information(Pauly 1998). But even themost plausibly diligentfishers can make mistakesunder difficult conditions,and data from poorly-paidofficials or observersemployed to recordlandings can be less thanaccurate.An assumption of zero isunacceptableWhere landing or catchdata does not provideamounts of discards, orestimate unreportedcatches such as llegal and unreported catch,transshipments, or unmandated catches, it isimportant to realize that an implicit assumptionhas been made that such categories are zero.  It isnot our purpose to comment on the effect thatsuch assumptions may have on conventionalstock assessments, and in fact estimates of somecatches, sometimes called ‘unassigned’, are oftenmade and used in both the ICES and NAFOarenas at closed stock assessment workshops.Presumably for fear of embarrassing stategovernments, these figures generally remainconfidential, or lie concealed in semi-privatestock assessment working papers. In any event,they are not attributed to nations or locations butonly to the fish stocks under examination. Butleaving these figures at zero, as databases in thepublic domain tend to do, is unacceptable whentrying to examine the impact of fisheries onmarine ecosystems where total extractions mustbe estimated. Political pressures maybe such thateven FAO’s own, well-founded  study of discards(Alverson et al. 1994) are omitted from thepublished FAO catch database.Hence, the assumption of a zero adjustment toreported landings should not be used (Pauly1998). Any percentage estimate of unreportedcatch by category, based on validatedinformation, will be closer to the truth, and soshould be used as a default in estimating the totalcatch figure for North Atlantic ecosystemsmodelled in Sea Around Us project. It is hopedCommerical Landings             AdjustmentAssign to Ecosystems and EEZRecreational ArtisanalSmall-Scale   DiscardsUnmandatedUnreportedIlegalFigure 1. Illustrating how various categories of unreported catches maybe used toadjust reported landings and discards to estimate total extractions from a marineecosystem.Sea Around Us Project Methodology Review42that improvements to our default figures may wellbe stimulated by its publication.As well as unreported and illegal catches, the totalmortality experienced by a stock also includesghost (‘cryptic’) fishing mortality and otherunaccounted sources of mortality. This topic iscomprehensively reviewed by Alverson et al.(2000), building on the work of ICES (1995), andis not considered in detail here.EXAMPLES OF HOW UNREPORTEDCATCH HAS BEEN DEALT WITHLake Malawi• In Lake Malawi, usipa, a small, streamlined,silvery pelagic zooplanktivore belonging tothe carp family, is the subject of aconsiderable artisanal seine net fishery. Thefish are caught at dusk and through the nightwith the aid of lights. There are small localmarkets for the fresh fish, but the bulk of thecatch is sun-dried and exported from the lakeshore, the local variety of a traditional andimportant food commodity known in centralAfrica as 'kapenta'. Official FAO statisticsrecord a total catch of 3,000 to 5,000 tonnesof usipa per year, but this figure seemed lowaccording to the suspicions of experiencedfishery biologists.For eight months in 1985/6, Lewis andTweddle (1990) stationed observers on theonly two roads leading out of the Nankumbapeninsula, situated in the heart of the usipafishery, who censused all trucks and theirsacks of dried usipa. Local consumption andusipa exported by lake steamer was alsoestimated. The catch from the peninsula,which represents only 5% of the lakeshoreline, was calculated as five times greaterthan the official catch for the whole lake.Scaling up the Nankumba catch to anestimate for the whole lake involved anumber of assumptions, but the total catch in1985/6 was probably between 50,000 and100,000 tonnes, contrasting with the officialfigures of 5,573 tonnes from beach recorders.Ecuador• In the late 1980s the tropical chub mackerelfishery in Ecuador landed over 500,000tonnes per year, caught by a fleet of smallvessels of 20 to 350 tonnes, most of which selltheir fish directly to fishmeal factories atthree ports along the coast. Official landingfigures were suspect and a log book systemhad proved unreliable. Since catches andcatch-per-unit-effort for this economicallyimportant fishery have been decliningmarkedly, an accurate assessment of thefishery using reliable catch data was urgent(Patterson 1990, Pitcher and Stokes 1990)and indeed the stock collapsed soonafterwards (Patterson et al. 1993). The catchwas cleverly estimated from the numbers ofsacks of fishmeal output from the fishmealfactories (Patterson et al. 1990). The weightof fish input to the fishmeal process wasback-calculated from the conversion ratios ateach stage of the industrial process. Thenumber of fishing vessels in each month wasestimated from official permits issued eachday ('zarpes'). Knowledge of the fleetstructure allowed an estimate of the catchwhich did not go through this route(approximately 15%). Not only were the finalcatch estimates about double the officialcatch statistics, but disconcertingly there waspoor correlation between the two sets offigures.Peru• During the heyday of the Peruvian anchovyfishery, in the late 1960s and early 1970s, itwas realized that official statistics massivelyunderestimated true catches, and thatfishmeal plants were operating at much lessthan their mandated conversion efficiency.While the official figures were never revised(and are still cited, D. Pauly, pers. comm.),structured interviews of 40 formerparticipants in the industry by one of theformer participants pointed out the need torevise the official catch figure from 12 milliontonnes in 1970 to 16 million tonnes, the actualvalue. Indeed, only the corrected catches arecompatible with the true conversionefficiency of the reduction plants, and withfishmeal exports (Castillo and Mendo 1987).North Atlantic• In 1997 it is estimated that more than 75 % ofthe reported Spanish catch of 37,000 tonnesof swordfish was illegal. ICCAT’s own recordsshow that Spain exceeded its catch limit inboth the North and South Atlantic in everyyear from 1996 when the ICCAT quotas wereintroduced.  For Bluefin tuna, Spain exceededthe catch limits of about 8000 tonnes by 19%in 1995, 58% in 1996 and 51% in 1997.Moreover, France, Italy, Japan and MoroccoThe Basis for Change43are reported as having illegal catches forBluefin tuna and swordfish as large as thoseof Spain (Raymakers and Lynham 1998).• Patterson (1998) used an  “adapt”  type ofnonlinear-least-squares tuned VPA model incomparison with standard ICES VPA in orderto estimate unreported catch. The Pattersonmodel is able to provide good estimates ofstock size and therefore catch, even whencatches are under-reported. The method wasused with three gadoid fisheries, North Seacod and west Scotland cod and whiting.Patterson concluded that the West Scotlandstocks, but not those in the North Sea, hadbeen substantially under-reported since 1991by a factor of 30-60%.• In Scotland and France, large quantities of25-30 cm cod are illegally landed as “bluegreens”, and under a different name, inFrance [2 correspondents].• In western Ireland, the catch of largemidwater trawlers targeting herring andmackerel is estimated to be at least 100% ofthe reported catch, with the consequence thatthe true catch was likely double the quota of50,000 tonnes [1 correspondent].• At least 50% of the catch of Scottish purseseiners is said to be illegal [1 correspondent].• Unreported catch is said to equal reportedcatch for Humberside fisheries, and higherfigures applied to historical periods of distantwater fleets before the EEZs. [1correspondent].• In Denmark, cod landings are oftendisreported as dogfish shark. [1correspondent].• In Canada, the arrest of a Spanish trawler(the Estai) in 1995, revealed a secretspecially-constructed hold that concealedunreported, illegal and undersized catch.There were two sets of log books, eachreporting different catch figures. From theskipper’s secret logbook, total catch wasfound to be 100% underreported [Harris1998]. Moreover, 98% of the catch wasundersized (and hence illegal).• A significant amount of catch from the Estaiwas recorded in the logbook of anotherSpanish vessel, the Patricia Nores [Harris1998]• 45% of all Spanish catches of flounder aresaid to be discarded at sea and not reported[Harris 1998].• In the late 1980s, every haul of the trawl byRussian vessels was estimated to be under-reported by at least 10 tonnes [Internal DFOdocument, quoted by Harris 1998].Harris (1998), who appears to have had access toa considerable amount of privileged information,reports many instances of discards anddisreported catch. His book can therefore be usedto provide preliminary figures for Canadianwaters. We are preparing a corrigenda from hisbook that may be used to tune estimates ofdiscards and illegal catch for his region. Werealize that it is easy to journalists’ reports, butwe would hope for better figures from those whohave better knowledge.An NGO tracking illegal fish catchThe 1996/7 annual quota for Patagonian toothfish(Dissostichus eleginoides), served as ‘Chilean SeaBass’ in expensive seafood restaurants world-wide, was set at 17,000 tonnes by CCAMLR(Commission for the Conservation of AntarcticLiving Marine Resources), illegal catches takenaround Heard and McDonald Island (Australia),Kerguelen Island (France) and Prince Edwardsand Marion Island (South Africa), appear to haveexceeded the legal quota by a factor of 500%.These illegal catches and sales of toothfish havebeen traced by an NGO, ISOFISH (InternationalSouthern Oceans Longline Fisheries InformationClearing House).Based in Hobart, Tasmania, and associated withCCAMLR, ISOFISH is funded by the Australianfishing industry. ISOFISH aims to track andreport the unlicensed fishing activities oftoothfish longliners and monitor the trade inillegally caught fish in cooperation with nationalauthorities and the international regulatory body,CCAMLR.The ISOFISH web site lists over 90 namedindividual boats and their owners, many withdetailed records of their illegal activities. Anewsletter dated March 1999 examines theChilean fishing industry and names the ‘pirateking’ of the industry, (Roberto Verdugo, formerUnder-Secretary of State for Fisheries in aChilean government) worth US$100 million inexports (80% to Japan) from Chile in 1997. Alongwith seven other Chilean companies, over 50fishing vessels sell illegal toothfish catches. ASea Around Us Project Methodology Review441999 report states, “ISOFISH has enoughevidence to publicly identify these companies asknowingly and persistently involved in andbenefiting from toothfish poaching.” By 1998, toits credit, government counter-measures in Chilewere aimed at exposing the trade. However, aconsequence was the re-flagging of many of thesevessels in Belize, Panama, and Honduras.Moreover, port and trade authorities in Uruguay,Mauritius, Mozambique, Namibia and the Frenchisland of Réunion are identified as “providingunquestioning support” to the poachers, andbeing involved in trans-shipments of illegallycaught fish.ISOFISH is a good model of what may achieved,with adequate funding, in identifying specificillegal fishing and tracking the trade in illegally-caught fish that drives such activities.Proposed Method for the SAU ProjectBasis of the adjustment methodWe present an adjustment procedure based on asimple spreadsheet, divided into categories ofunreported annual catch (Figure 2). Adjustmentfactors are based on reports from observers,confidential correspondents and on informationpublished in a variety ofsources. Over time, theadjustment factors respond tochanges in regulatory regimeand hence the incentives anddisincentives to mis-report.Once in place, this methodprovides preliminary estimatesthat may be refined withoutdisruption, and offers a basis forcollaboration and discussion.Figure 2 illustrates the generalprinciples of the procedure. InTable 1 (a to f) we show ahypothetical example of theadjustment process. In eachcase we show five sections ofcatch adjustment: discards thatare reported by observers (or insome other fashion); discardsthat are unreported (forexample in the absence ofobservers); unmandated catches(defined as above); disreportedcatches and illegal landings (fishultimately landed and soldsomewhere in the world).For each species these categoriesare shown for domestic and forforeign fleets. Table 1a lists a set of influences onmisreporting, mapping the ‘incentive climate’ asit were, tabulated in 5-year periods. Table 1bcontains some estimates used as anchor pointsthat have some reasonable validation,  obtainedfrom surveillance, informants or other sources.Each anchor point is documented as to its source(as far as is possible). Table 1c shows adjustmentfactors interpolated between the point estimatesof 1b using influences from 1a. Interpolationshere are simply performed linearly between thepoints with information – obviously moresophisticated statistical methods could be used.Total officially reported landings are listed inTable 1d: this data is extracted from officialdatabases. Missing catches in Table 1e areestimated by multiplying the factors from1c bythe landings in 1d. Hence Table 1f providesestimates of total extractions.The most difficult part of the work is developingTables 1a and 1b. It is important to emphasizethat all anchor points at stage b are explicitlyfootnoted, even if exact sources cannot berevealed in some cases. Beyond this point themethod flows fairly automatically and in such away that most criticism is forced by the scheme toDecadeType of OmissionInfluencesDecadeType of OmissionEstimatesDecadeType of OmissionSeriesAdjustmentsxx xxxxxx+++ +++++++_++_ManagementHistory etc.+++Reports etcOfficial StatisticsBest Estimate ofComplete CatchFigure 2. Illustrating adjustments to landings data to construct total fisheryextractions from a marine ecosystem.  A shifting climate of influences and pointestimates at top lead to adjustment factor matrix at bottom of diagram.The Basis for Change45Table 1. Illustrating the catch adjustment process. (a) climate of factors influencing misreporting; (b) documented pointestimates (anchor points) of misreporting from informants or others; (c) interpolated adjustment factors; (d)landings (and recorded discards) data; (e) missing catch data; (f) estimated total fishery extractions from ecosystem.Species Jurisdiction Type Period(a) INFLUENCE FACTORS 1960s 1970-74 1975-79 1980-84 1985-89 1990-94 1995-99Species A Domestic Obs discards None Some Lots Heaps ? Lots SomeObs effect discards None Some Lots Heaps ? Lots SomeUnmandated ? Some Lots Heaps ? Lots SomeDisreported None Some Lots Heaps Lots Lots SomeIllegal None Some Lots Heaps Lots Lots SomeForeign Obs discards None ? Lots Heaps Lots ? ?Obs effect discards None Some Lots Heaps Lots Lots ?Unmandated None ? Lots Heaps Lots ? ?Disreported None Some Lots Heaps Lots Lots ?Illegal None Some Lots Heaps Lots Lots ?Species B Domestic Obs discards None Some Lots Heaps ? Lots SomeObs effect discards None Some Lots Heaps ? Lots SomeUnmandated ? Some Lots Heaps ? Lots SomeDisreported None Some Lots Heaps Lots Lots SomeIllegal None Some Lots Heaps Lots Lots SomeForeign Obs discards None ? Lots Heaps Lots ? ?Obs effect discards None Some Lots Heaps Lots Lots ?Unmandated None ? Lots Heaps Lots ? ?Disreported None Some Lots Heaps Lots Lots ?Illegal None Some Lots Heaps Lots Lots Some(b) ANCHOR POINTS (%)Species A Domestic Obs discardsL 7AObs effect discards 10BUnmandated 10CDisreported 0D 100E 25FIllegal 30GForeign Obs discardsLObs effect discardsUnmandatedDisreportedIllegal 40HSpecies B Domestic Obs discardsLObs effect discardsUnmandated 25IDisreportedIllegalForeign Obs discardsLObs effect discardsUnmandated 0JDisreported 80KIllegalNotes on sources for Anchor Points  (examples)(A) Informant  A. (B) DFO surveillance reports. (C) Harris (1998).  (D) Harris (1998).  (E) Informant B.  (F) Informant A.  (G)DFO estimate, Anon.  (H) Harris (1998).  ( I) Word Bank Study 1990.  (J)  Informant A.  (K) Informant A. (L) This is theportion of the observer discards that are discarded when no observer is present.Sea Around Us Project Methodology Review46be constructive by way of improving theinterpolations. Moreover, the revised totalextractions are not so controversial because theyare no longer identified by country of origin,rather, they are articulated upon the ecosystem inquestionA preliminary example: Atlantic CanadaA preliminary influence table for fishery catchesin Atlantic Canada is shown in Table 2. This tableis an example: a more complete table has to beassembled with more information for a widerrange of species. Similar tables will be drawn upfor each major area of marine ecosystems in theNorth Atlantic.c)  INTERPOLATIONSSpecies A Domestic Obs discards 0.07 0.07 0.07 0.07 0.07 0.07 0.07Obs effect discards 0.10 0.10 0.10 0.10 0.10 0.10 0.10Unmandated 0.00 0.10 0.30 1.00 0.30 0.40 0.25Disreported 0.00 0.10 0.30 1.00 0.30 0.40 0.25Illegal 0.00 0.10 0.30 1.00 0.30 0.40 0.25Foreign Obs discards 0.07 0.07 0.07 0.07 0.07 0.07 0.07Obs effect discards 0.10 0.10 0.10 0.10 0.10 0.10 0.10Unmandated 0.00 0.10 0.30 1.00 0.30 0.40 0.25Disreported 0.00 0.10 0.30 1.00 0.30 0.40 0.25Illegal 0.00 0.10 0.30 1.00 0.30 0.40 0.25Species B Domestic Obs discards 0.07 0.07 0.07 0.07 0.07 0.07 0.07Obs effect discards 0.10 0.10 0.10 0.10 0.10 0.10 0.10Unmandated 0.00 0.10 0.25 1.00 0.30 0.40 0.25Disreported 0.00 0.10 0.25 1.00 0.30 0.40 0.25Illegal 0.00 0.10 0.25 1.00 0.30 0.40 0.25Foreign Obs discards 0.07 0.07 0.07 0.07 0.07 0.07 0.07Obs effect discards 0.10 0.10 0.10 0.10 0.10 0.10 0.10Unmandated 0.00 0.10 0.25 1.00 0.30 0.40 0.25Disreported 0.00 0.10 0.25 0.80 0.30 0.40 0.25Illegal 0.00 0.10 0.25 0.80 0.30 0.40 0.25(d) LANDINGSSpecies A Domestic Landings 12000 12000 12000 12000 12000 12000 12000Foreign Landings 8000 8000 8000 8000 8000 8000 8000Species B Domestic Landings 11500 11500 11500 11500 11500 11500 11500non-CDN Landings 400 400 400 400 400 400 400(e) MISSING CATCHSpecies A Domestic Obs discards 840 840 840 840 840 840 840Obs effect discards 84 84 84 84 84 84 84Unmandated 0 1200 3600 12000 3600 4800 3000Disreported 0 1200 3600 12000 3600 4800 3000Illegal 0 1200 3600 12000 3600 4800 3000Foreign Obs discards 560 560 560 560 560 560 560Obs effect discards 56 56 56 56 56 56 56Unmandated 0 800 2400 8000 2400 3200 2000Disreported 0 800 2400 8000 2400 3200 2000Illegal 0 800 2400 8000 2400 3200 2000Species B Domestic Obs discards 805 805 805 805 805 805 805Obs effect discards 80 80 80 80 80 80 80Unmandated 0 1150 2875 11500 3450 4600 2875Disreported 0 1150 2875 11500 3450 4600 2875Illegal 0 1150 2875 11500 3450 4600 2875Foreign Obs discards 28 28 28 28 28 28 28Obs effect discards 3 3 3 3 3 3 3Unmandated 0 40 100 400 120 160 100Disreported 0 40 100 320 120 160 100Illegal 0 40 100 320 120 160 100(f) ESTIMATED TOTAL EXTRACTIONSSpecies A Total 21540 27540 39540 81540 39540 45540 36540      Percentage Unreported 7.70 37.70 97.70 307.70 97.70 127.70 82.70Species B Total 12816 16386 21741 48356 23526 27096 21741      Percentage Unreported 7.70 37.70 82.70 306.35 97.70 127.70 82.70The Basis for Change47Table 3 presents our first attempt to quantify theeffects of the factors presented in general terms inTable 2 for two species caught in the Scotian Shelffishery, cod and herring.  In Table 3b it isimportant to try to have at least one anchor pointin each row of the table. In this example,unmandated cod and herring catches do not exist,so all the values, and the anchor point, are zero.Note that herring are targeted by the pelagicpurse seine fishery but are also caught as largelyunreported bycatch in the demersal trawl fishery.Our percentage figure refers here to the targetherring fishery, not the trawl fishery in whichherring are a bycatch. This is different toTable 2.  Summary of influences on the incentives to misreport fishery catches from Atlantic Canada from 1960 to presentday (with thanks to Sylvie Guénette).1960s 1970-74 1975-79 1980-84 1985-89 1990-94 1995-99RegulatoryregimesICNAF quotasoverestimatedEEZs NAFO quotas 1992 codmoratoriumcod fisherystill closedNon-CanadiancatchNo incentive tomisreport.Slightdiscarding ofjuveniles.Discardinghigh for someunusedspecies1Highermisreporting100%unreportedturbot catchoutside EEZ(from arrestedSpanish vesselEstai)Canadianunreportedcatch2Moderatediscarding byinshore fisherywhen plantcapacityexceeded.Discardingmay be high forsome unusedspecies1Offshore vessels:strong incentiveto discard afterenterpriseallocations put inplace.Inshore:moderate  coddiscards atwharf3Offshore:high incentiveto discard;Inshore:Gill nets inwater too long,increased soaktimesdecreasedproportion ofmarketablefish. Largediscards atwharf- butdecrease inminimum fishsize acceptedby buyers3.Illegal catch ofcod duringmoratoriumLow discardsIllegal catchof codduringmoratorium(lower after‘sentinel’and foodfisheryopened)unmandatedcatchlanternfishandmonkfish inscallopfisherydisreportedcatchDisreportedcatch of codHigh forCanadianand non-Canadian(outside theEEZ) forgroundfishsectorNotes:1. Skates for example, on Georges Bank in 1951, the average capture rate for Barndoor skates was as high as 21 per tonne of cod trawled(Bigelow, H.B., Schroeder, W.C., in Casey and Myers, 1998 (this has decreased now as their abundance has decreased)2. Unreported catch defined as: fish in bad condition, for gill nets the catch is retained for household use, for traps, the fish are too small orare dumped when the processing plant’s capacity is exceeded.3. From Hutchings and Ferguson (ms submitted).Sea Around Us Project Methodology Review48 (b) ANCHOR POINTS (%)Cod Domestic Obs discards 2AObs effect discards 1000BBUnmandated 0CDisreported 0.5DIllegal 0.5E 1E 1.5EForeign Obs discards 2EObs effect discards 1000FUnmandated 0CDisreported 0.5EIllegal 5.0EHerring Domestic Obs discards 7GObs effect discards 10HUnmandated 0CDisreported 0CIllegal 1EForeign Obs discards 5EObs effect discards 50EUnmandated 0CDisreported 0IIllegal 5ENotes on sources for anchor points  (examples)(A) Informant  A.  (B) DFO surveillance reports. (C) Unmandated category not applicable to cod in this region.  (D) Informant B. (E) Harris (1998).(F) Informant A.  (G) DFO estimate, Anon.  (H) estimate based on similar fisheries reported elsewhere.  (I) Disreporting for herring fromInformant C.Table 3. Estimations of total extractions of cod (Gadus morhua) and herring (Clupea harengus) from the 4VW region of Atlantic Canada from 1960 topresent day.Species Jursidiction Type Period(a) INFLUENCE FACTORS 1960s 1970-74 1975-79 1980-84 1985-89 1990-94 1995-99Cod Domestic Obs discards Low Low Medium High Medium Low LowObs effect discards High High High High High High HighUnmandated None None None None None None NoneDisreported None None None None None Low LowIllegal Lots Lots Some Low Low More MoreForeign Obs discards Medium Medium Medium High Medium Low LowObs effect discards High High High High High High HighUnmandated None None None None None None NoneDisreported None None Low Low Low Medium MediumIllegal Some Lots Huge huge Lots Some SomeHerring Domestic Obs discards Lots Lots Lots Lots Lots Lots LotsObs effect discards High High High High High High HighUnmandated None None None None None None NoneDisreported None None None None None None NoneIllegal Low Low Low Low Low Low LowForeign Obs discards Lots Lots Lots Lots Lots Lots LotsObs effect discards High High High High High High HighUnmandated None None None None None None NoneDisreported None None None None None None NoneIllegal Low Low Low Low Low Low LowThe Basis for Change49c)  INTERPOLATIONSCod Domestic Obs discards 0.005 0.005 0.02 0.05 0.02 0.005 0.005Obs effect discards 10.0 10.0 10.0 10.0 10.0 10.0 10.0Unmandated 0.0 0 0 0 0 0 0Disreported 0 0 0 0 0 0.005 0.005Illegal 0.04 0.04 0.01 0.005 0.005 0.001 0.015Foreign Obs discards 0.02 0.02 0.02 0.05 0.02 0.01 0.01Obs effect discards 10 10 10 10 10.0 10 10Unmandated 0.0 0 0 0 0 0 0Disreported 0 0 0.01 0.01 0.01 0.05 0.05Illegal 0.05 0.1 0.3 0.3 0.1 0.05 0.05Herring Domestic Obs discards 0.07 0.07 0.07 0.07 0.07 0.07 0.07Obs effect discards 10.0 10.0 10.0 10.0 10.0 10.0 10.0Unmandated 0.0 0 0 0 0 0 0Disreported 0 0 0 0 0 0 0Illegal 0.001 0.001 0.001 0.001 0.001 0.001 0.001Foreign Obs discards 0.05 0.05 0.05 0.05 0.05 0.05 0.05Obs effect discards 50.0 50.0 50.0 50.0 50.0 50.0 50.0Unmandated 0.0 0 0 0 0 0 0Disreported 0 0 0 0 0 0 0Illegal 0.05 0.05 0.05 0.05 0.05 0.05 0.05(d) LANDINGS (1000 tonnes per annum)Cod Domestic 225 144 164 269 261 110 8Foreign 51 126 76 19 13 18 23Herring Domestic 1144 759 320 384 217 65 2Foreign 214 281 121 180 72 28 0(e) MISSING CATCH (1000 tonnes per annum)Cod Domestic Obs discards 1 1 3 13 5 1 0Obs effectdiscards11 7 33 134 52 5 0Unmandated 0 0 0 0 0 0 0Disreported 0 0 0 0 0 1 0Illegal 9 6 2 1 1 0 0Foreign Obs discards 1 3 2 1 0 0 0Obs effectdiscards10 25 15 9 3 2 2Unmandated 0 0 0 0 0 0 0Disreported 0 0 1 0 0 1 1Illegal 3 13 23 6 1 1 1Herring Domestic Obs discards 80 53 22 27 15 5 0Obs effectdiscards801 531 224 269 152 45 2Unmandated 0 0 0 0 0 0 0Disreported 0 0 0 0 0 0 0Illegal 1 1 0 0 0 0 0Foreign Obs discards 11 14 6 9 4 1 0Obs effectdiscards536 702 302 451 180 69 0Unmandated 0 0 0 0 0 0 0Disreported 0 0 0 0 0 0 0Illegal 11 14 6 9 4 1 0Sea Around Us Project Methodology Review50conventional fishery work where there could beno percentage discard estimate as there is nocatch reported by that particular fishery. Thepercentage figure here refers to the percentage ofunreported by-catch of herring extracted from theecosystem, by whatever gear may catch it.The final results show an average of 30% and forcod and 157% for herring over the whole timeperiod, although in the most recent half-decadewith data reports these figures are 17% and 77%respectively. We emphasize again that thesevalues are intended here only as ‘straw men’ to berefined and improved by those moreknowledgeable about these fisheries than us.ConclusionsUnreported extractions cast various kinds ofshadows on fisheries and their associatedactivities. These shadows can help us track them.Patterson (1998) tracked the numerical shadowsof illegal catch using a VPA technique. Illegalcatch generates profits that may be revealed withsuitable financial scrutiny. Transshipments maybe observed directly by aerial surveillance or maycreate unexpected landings at ports complicit insuch dealings (like the deep-sea Antarctictoothfish landings in tropical Mauritius). WithoutVMS or human observer schemes, the shadow ofdiscards at sea may be more difficult to track, asoften the only direct observers are seabirds andmarine mammals. But even here, over time,mass-balance ecosystem models may revealshadows of extractions that need to be explained.As set out in this paper, our method of attemptingto quantify unreported catches has someadvantages. When setting the anchor points, forexample, informants may be asked to rank theseverity of unreported catches. In fact humans arequite good at ranking things presented in pairs,asking the question “which is the better andwhich is the worse?” A series of paired questionsmight be developed for a more formal protocolshere.The method has its difficulties, for example, inthat we use a percentage of the reported catch.How do we deal with the problem where no catchis reported, yet discards and illegal catch areknown to occur? Patterson (1998) considers iteasier to estimate catch ‘reporting efficiency’ (i.e.,accuracy) than to make absolute estimates ofunreported catch. But the key here is that we areinterested in an annual value for wholeecosystems. And this in itself makes some of theissues raised by identifying the sources of anchorpoint estimates less controversial. Thereforefigures in tonnes can be raised to annual valuesand compared with the annual catch of thespecies over the whole system.Publicizing or covering up illegalcatches in the North Atlantic?Creating an organization similar to ISOFISH inthe North Atlantic would be of great value.Keeping illegal catch under wraps is whatgovernments tend to want to do for fear, it seems,of causing political embarrassment to allies.  EvenCanada, famous for the 1995 arrest, instigated bythe fisheries minister Brian Tobin, of a Spanishtrawler, whose secret, specially constructed holdconcealed 100% unreported, illegal andundersized catch, is coy about revealing illegalfishing activities. When asked, Australia rapidlyprovided lists of other vessels arrested for illegalfishing such information, but this information isdifficult to obtain. One study on illegal catch inScotland (data summaries reported inBeddington et al. 1997) is a confidentialdocument, and not obtainable by the public orother scientists.Murawski (1996) has looked at factors influencingdiscards in data from the US and Canada. Generallinear models were fitted to discard rates, totalcatch, species richness, species diversityevenness, together with operational variablesassociated with the fishing process (codend mesh,vessel size, tow duration, total catch, targetspecies, year, month, depth and statistical area).Variances were high, but fisheries managed bymesh and fish size generally had higher discardrates. Year classes with high abundanceinfluenced discard rates disproportionately.Murawski worked with observer estimates ofdiscards, whereas the focus of this paper is tosuggest a method to use when such data is notavailable.In the ICES area, estimates of illegal fishing areroutinely made by the stock assessment working (f) ESTIMATED TOTAL EXTRACTIONS (1000 tonnes per annum)Cod Total 312 324 318 453 338 139 37Percentage unreported 12.8 20.0 32.5 57.5 23.0 8.2 17.4Herring Total 2797 2356 1001 1330 643 214 4Percentage unreported 106.0 126.5 127.2 135.6 122.7 132.1 77.1The Basis for Change51parties that regularly perform single-specie stockassessment. Yet, it is an unwritten but strictlyimposed tradition that the basis of suchadjustments are not made public, even whenofficials have direct knowledge of specific events.Such a policy of secrecy would likely be news forthe public of the countries involved. Covering upfor illegal fishing would be unthinkable if thiswere illegal drug running in North Atlanticcountries. Bank staff who defraud the public ofmillions of dollars are not protected by a shield ofanonymity – so why should this protection beafforded to illegal fishers?Evaluation by FAO of IUU fishingWhile our work was in progress, and following aseries of discussions in international fora such asthe International Maritime Organization (IMO),FAO convened a working group with mandate toevaluate, ‘illegal, unreported and unregulated’catch (IUU: Bray 2000). Leading this initiative,Bray reviews IUU experience world wide, andpoints the finger at flag states for not providingadequate human and financial resources to tacklethe problem.Unfortunately, the three FAO categories do notmap easily into the operational categories we usein our algorithm. Illegal catch includes both areported element (disreported), an estimatedelement (e.g. observer and other estimates ofdiscards) and an unreported component.Moreover the unregulated catch category seemsill-defined, and overlaps with our unmandatedcategory. The term ‘unauthorized‘ fishing is alsoused, but also does not easily link to ourcategories, except as an overarching term for allunreported and misreported catches.In this work, however, FAO has published a verystrong message concerning the criticalimportance of IUU fishing to the sustainability ofbenefits from capture fisheries. For example,Evans (2000) considers that IUU fishing distortsand devalues information from compliantfisheries, lowers allowable catches set using theprecautionary approach, and increasesuncertainty and the risk of overexploitation.Evans considers that, at national scales, there isoften complacency about the intractability of theproblem, echoing our concerns expressed above.Evans considers some fisheries, where newtechnology has recently made deepwater ormarginal stocks vulnerable, to be underreportedby as much as 75% , and in the case of stocks onthe high seas, over 100%. Evans sees compliancewith FAO Code of Conduct for ResponsibleFisheries (see Doulman 1998; Edeson 1996) as anessential first step in improving the situation.Doulman (2000) also considers IUU to be majorflaw in present fisheries management, leading toa loss of economic and social benefits, and, inextreme cases, to the collapse of stocks.  Doulmancalls for a protocol that can operate regionally,sub-regionally and nationally, and be applicableto different types of fisheries and stockdistributions. Hence we offer the method set outin draft here as candidate.Finally, Edeson (2000) reviews the legal remediesavailable to combat IUU fishing. In particular, thepossible role of the FAO Code of Conduct as aninstrument of international law and a part of anInternational Plan of Action. Within the EEZs ofnations, although some national laws might beimproved, the problem is more a lack ofimplementation of existing regulations. Edesonconsiders this situation might be improved byexplicit adoption of the FAO Code of Conduct.The possibility of enforcement by the flag state ofthe vessel is also under discussionBenefits from a transparent new methodObtaining estimates of the total extractions froman ecosystem as essential for a rational evaluationof the impact of fisheries When total extractionsfrom an ecosystem are estimated, ECOPATH andECOSIM modelling can reveal anomalies whenmodels fail to balance, or simulated hindcasts donot fit biomass survey data. These methods cansuggest alternative values for stock biomass. Insome cases existing catch and biomass figuresmay be mutually incompatible where trophicwebs cannot support them. We anticipate anumber of anomalies of this kind arising from ourtotal catch estimations.Transparency is the only way that the manydifficulties this new method will face can bereduced to a minimum. The database for SAU,together with its assumptions and modifiers usedto infer total catches will be available on theWorld Wide Web, in order to allow the retracingof each step involved in arriving at certainconclusions. In so doing, the SAU team makes itsconclusions not only reproducible in principle, asscientists always should, but also in practice. Theonly exception to this would be to protect theanonymity of certain informants, e.g., concerningillegal catches.Cheating is widespread in fisheries, and thepenalties are low, and the risk of detection isSea Around Us Project Methodology Review52often low as the participants are well aware.Unfortunately, political disincentives lead manyconcerned with fisheries to downplay theirknowledge of this cheating. Where governmentand official sources have strong links, and evenfunding, from industry, we may expect thesedisincentives to be stronger. Fraud on this scalehas not only contributed to the depletion of NorthAtlantic ecosystems and contributed to disastrousstock collapses, but has foreclosed options for thefuture generation of wealth and sustainablebenefits from marine resources. Like any othercriminal act, we need to estimate its truemagnitude and encourage its disclosure.CONCLUSIONOur method stands or falls by the explicitness andquality of the anchor points. These need to bedefendable scientifically and to withstandscrutiny by scientists, fisheries, regional andgovernment agencies, managers and informants.Ideally, in the public interest, an analysis wouldobtain the support of all of these constituentsACKNOWLEDGEMENTSWe are most grateful for discussion of the basis ofthis paper with Jaqueline Alder, Sylvie Guenette,Paul Fanning, Nigel Haggan and Daniel Paulyand to two referees for their helpful comments.REFERENCESAlverson, D.L., Freeberg, M.H., Murawski, S.A. andPope, J.G. (1994) A global assessment of fisheriesby-catch and discards. FAO Fisheries TechnicalPaper 339, 233 pp.Alverson, D.L. et al. (2000) Report of the FishingTechnology and Fish Behaviour Working Group(FTFB) on unaccounted mortality in fisheries.ICES, FTFB, Haarlem, the Netherlands, April2000. 76 pp.Beddington, J., Lorenzen, K. and Payne, I. (1997)Limits to exploitation of capture fisheries. In:Hancock, D.A., Smith, D.C., Grant, A. and  Beumer,J.P. (eds) Developing and Sustaining WorldFisheries Resources: the State of Science andManagement. CSIRO, Collingwood, Australia, 529-536.Bray, K. (2000) A global review of illegal, unreportedand unregulated (IUU) fishing. Expert consultationon illegal, unreported and unregulated fishing,FAO: IUU/2000/6,  53 pp.Casey, J.M. and Myers, R.A. (1998) Near extinction of alarge, widely distributed fish. Science 281, 690-692.Castillo, S. and Mendo, J. (1987) Estimation ofunregistered Peruvian anchoveta (Engraulisringens)  In: official catch statistics, 1951 to 1982.In: Pauly, D. and Tsukayama, I. (eds) The Peruviananchoveta and its upwelling ecosystem: threedecades of change. ICLARM Studies and Reviews15, 109-116.Doulman, D. (1998) The Code of Conduct forResponsible fisheries: the requirement forstructural change and adjustments in the fisheriessector. [www.fao.org].Doulman, D. (2000) Illegal, unreported andunregulated fishing: mandate for an internationalplan of action. Expert consultation on illegal,unreported and unregulated fishing, FAO:IUU/2000/4, 16 pp.Edeson, W.R. (1996) Current legal development: TheCode of Conduct for Responsible Fisheries: anintroduction. Intl J. Mar. Coastal Law 11: 233-38.Edeson, W. (2000) Tools to address IUU fishing: thecurrent legal situation. Expert consultation onillegal, unreported and unregulated fishing, FAO:IUU/2000/8, 13 pp.Evans, D. (2000) The consequences of illegal,unreported and unregulated fishing for fishery dataand management. Expert consultation on illegal,unreported and unregulated fishing, FAO:IUU/2000/12,  9 pp.Harris, M. (1998) Lament for an Ocean: the collapse ofthe Atlantic cod fishery. McLelland and Stuart,Toronto, Canada. 342 pp.ICES (1995) Report of the study group on unaccountedmortality in fisheries. ICES CM 1995/B:1, 27 pp.ISOFISH (1999) The Chilean Fishing Industry: itsinvolvement in and connections to the illegal,unreported and unregulated exploitation ofPatagonian toothfish in the Southern Ocean.ISOFISH. Hobart, Tasmania, ISOFISH OccasionalReport No. 2, 98 pp.Lewis, D.S.C. and Tweddle, D. (1990) The yield of usipa(Engraulicypris sardella) from the NankumbaPeninsula, Lake Malawi (1985-1986). Coll. Rep.Fish. Res. Malawi, Occ. Pap. 1, 57-66.Murawski, S. (1996) Factors influencing by-catch anddiscard rates: analyses from multispecies/multifishery sea sampling J. Northw. Atl. Fish. Sci.19, 31-39.Patterson, K.R. (1998) Assessing fish stocks whencatches are misreported: model simulation test,and application to cod, haddock and whiting in theICES area. ICES Journal of Marine Sciences 55,878-891.Patterson, K.R., Castello, M., Santos, M. and Barcia, M(1990) [Validation and recomputation of catch andeffort data in the nearshore fleet of Ecuador for theperiod 1981 to 1989]. Report to INP, Guayaquil,Ecuador, 47 pp. [in Spanish]Patterson, K.R., Pitcher, T.J. and Stokes, T.K.  (1993)  Astock collapse in a fluctuating environment: thechub mackerel Scomber japonicus (Houttuyn) inthe eastern central Pacific. Fisheries Research 18,199-218.Pauly, D. (1998) Rationale for constructing catch timeseries. European Community FisheriesCooperation Bulletin 11 (2), 4-5.Pitcher, T.J. and Stokes, T.K. (1990) Assessment ofEcuadorean Commercial Pelagic Fisheries.Contract Report to the Overseas DevelopmentThe Basis for Change53Administration, United Kingdom. MarineResources Assessment Group, Imperial College,London, 60 pp.Raymakers, C. and Lynham, J. (1999) Slipping the net:Spain’s compliance with ICCAT recommendationsfor Swordfish and Bluefin Tuna. World WildlifeFund Endangered Seas Campaign,  58 pp.Shepherd, J.G. (1988) Fish stock assessments and theirdata requirements. In: Gulland, J.A. (ed) FishPopulation Dynamics, Wiley, London, 35-62.Walters, C.J. (1995) Fish on the Line: the future ofPacific fisheries. Suzuki Foundation Report,Vancouver, 87 pp.Watson, R., Guénette, S., Fanning, P. and Pitcher, T. J.(2000) The Basis for Change I: reconstructingfisheries catch and effort data. In: Pauly, D. andPitcher, T.J. (eds) Methods for assessing theimpact of fisheries on North Atlantic ecosystems.Fisheries Centre Research Reports 8(2), 23-39.Sea Around Us Methodology Review54HOW LIFE HISTORY PATTERNS ANDDEPTH ZONE ANALYSIS CAN HELPFISHERIES POLICYDirk Zeller and Daniel PaulyFisheries Centre, University of British Columbia,2204 Main Mall, Vancouver, B.C., Canada, V6T1Z4ABSTRACTThe life-history patterns of fish species arecomplex. But much of this complexity can becaptured in simple diagrams of coastal transects,where juveniles usually occur in larger numbersin shallow waters, while adults generally inhabitdeeper, offshore waters. Such coastal transectscan be used to show how different fisheriessectors (e.g. small versus large scale) may exploitdifferent parts of the life history of the samespecies or stock. Thus, a species may ‘connect’small with large scale fishery sectors throughtheir life history patterns. We show how this canbe visualized through iconographicrepresentations of generalized life historypatterns and depth profiles, with specific key life-history parameters.  Relevant patterns includespawning areas, nursery/juvenile distributions,adult distributions and spawning migrations.Four preliminary case studies presented hereillustrate some general patterns with regard towater depth and distance from shore. Thediagrams allow us to incorporate intomanagement the concept of life historyinterconnectivity between different fisherysectors. This contributes to sustainableecosystem-based approaches by informing policyoptions when faced with decisions to rationalizeovercapitalized fisheries.INTRODUCTIONThe stock of an exploited species may be utilizedby more than one fisheries sector (such asinshore, small-scale fisheries and offshore, large-scale fisheries) during different stages in thespecies life history (see Ruttan et al. 2000). Lifehistory patterns are generally viewed as multi-dimensional in scale, with complex interactionsbetween components defined by ecology,oceanography, time and geography.  Often thiscomplexity has made it difficult to assimilatepotential effects of multiple fishery sectors on aspecies and the industry it supports.  This may beeither due to the perception of multi-dimensionalcomplexity thought to be intractable, or becauseof an oversight of basic patterns.Here, we argue that this multi-dimensionalcomplexity can be reduced to a simpler,generalized two-dimensional life history pattern,while still capturing the essential information.Both Charles Darwin and Alexander vonHumboldt used the method of reduceddimensionality to focus one’s attention to the keyissues while capturing most of the significantinformation concerning the topic at hand.  Forexample, after reviewing much literature, Darwinconcluded that “latitude is a more importantelement than longitude” for explaining thedistribution of organisms (Barrett et al. 1987).This concept has recently been revisited in alatitudina