UBC Faculty Research and Publications

Use of Ecopath with Ecosim to Evaluate Strategies for Sustainable Exploitation of Multi-Species Resources Pauly, D. (Daniel); Weingartner, Gunna 1998

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_Use_ecopath_ecosim.pdf [ 476.3kB ]
JSON: 52383-1.0348095.json
JSON-LD: 52383-1.0348095-ld.json
RDF/XML (Pretty): 52383-1.0348095-rdf.xml
RDF/JSON: 52383-1.0348095-rdf.json
Turtle: 52383-1.0348095-turtle.txt
N-Triples: 52383-1.0348095-rdf-ntriples.txt
Original Record: 52383-1.0348095-source.json
Full Text

Full Text

 Fisheries Centre Research Reports 1998   Volume 6   Number 2     Use of Ecopath with Ecosim to Evaluate Strategies for Sustainable Exploitation of Multi-Species Resources    ISSN 1198-6727  Fisheries Centre, University of British Columbia, Canada                            ISSN 1198-6727  Use of Ecopath with Ecosim  to Evaluate Strategies for  Sustainable Exploitation of  Multi-Species Resources  Fisheries  Centre  Research  Reports 1998   Volume  6   Number  2  Use of Ecopath with Ecosim to Evaluate Strategies for Sustainable Exploitation of Multi-Species Resources:  Proceedings of a Workshop held at the  Fisheries Centre University of British Columbia,  Vancouver, B.C., Canada, March 25-27, 1998        edited by  Daniel Pauly   with the assistance of Gunna Weingartner     Published by   The Fisheries Centre,  University of British Columbia 2204 Main Mall, Vancouver, B.C. Canada, V6T 1Z4    ISSN 1198-6727  1 ABSTRACT   This report presents background material to and the main conclusions of a workshop, held on March 25-27, 1998 at the Fisheries Centre, UBC, and devoted to the use of the Ecopath with Ecosim software as a tool for evaluating different strategies for fisheries resource man-agement in a multispecies, i.e., ecosystem con-text. Summaries of lectures describing the latest version of Ecopath with Ecosim (V. Christen-sen), the background, capabilities and short-comings of Ecosim (C. Walters), including the use of Ecosim in an economic context (R. Su-maila) to assess the effectivity of marine pro-tected areas (R. Watson) and for rebuilding ecosystem (T. Pitcher) are presented.  The main features of Ecospace, a spatial version of Eco-path recently developed by C. Walters, are briefly outlined. The requirements of FAO - the main sponsor of the workshop - for field use of the package are presented (K. Cochrane), as are the results of tests and simulations by the workshop partici-pants. The report concludes with a general discussion of the type of prediction (safe, tentative, guesses) that can be expected to result from Ecosim/Ecospace applications.    2Director’s Foreword   Fisheries science, based since the Second World War almost exclusively on single species popu-lation dynamics, has been conspicuously unable to answer, and often has failed even to pose, questions about the impacts of fishing on ma-rine ecosystems. In fact, changes to ecosystems after the collapse of stock have generally caught fisheries scientists by surprise. Contrary to the previous view that fishing has hardly any effect on either the structure or composition of ma-rine ecosystems, it is gradually being realized that the historical impacts of fishing have been large, dramatic and difficult to reverse. Fishing has seriously depleted biodiversity within and among species, reduced trophic linkages, caused local extinctions and compromised the economic value of marine resources. Fisheries scientists are only just beginning to recognize that these questions are the most important of our day, since without quantitative evaluations of ecosystem changes under alternative fishing policies, we will be powerless to reverse trends that, in the face of modern fishing gear technol-ogy, will likely result, within a generation, in the devastation of our oceans.  More than twenty researchers from interna-tional organizations, fisheries research institu-tions and academia, and graduate students, gathered for a workshop sponsored by FAO at the UBC Fisheries Centre from March 25-27, 1998. The aim was a preliminary exploration of the potential of some new analytical tools based on ecosystem models for comparing policy ob-jectives in multispecies fisheries. FAO intends to follow this up with a second workshop, at another location, in about one year.  This was the second Fisheries Centre workshop based on the Ecopath modeling system. The first, in November 1995, led directly to the de-velopment of Ecosim by Carl Walters (pub-lished in 1997) in which the set of simultaneous linear equations estimated by Ecopath is used to parameterize the differential equations which, when integrated, allow dynamic re-sponses to changes in mortality due to fishing to be modeled. The use of Ecopath, itself in an improved version, and integrated with Ecosim, was the primary focus of the meeting.  The present workshop also saw the launching of Ecospace, the first spatial modeling tool based on whole ecosystems. Ecospace will likely revo-lutionize the planning and design of marine reserves.  The ‘Use of Ecopath with Ecosim to Evaluate Strategies for the Sustainable Exploitation of Multispecies Resources’ is the tenth in a series of workshops sponsored by the UBC Fisheries Centre. The workshop series aims to focus on broad multidisciplinary problems in fisheries management, to provide a synoptic overview of the foundations and themes of current research, and identifies profitable ways forward. Edited reports of the workshops are published as Fish-eries Centre Research Reports and distributed to all workshop participants. Further copies are available on request for a modest cost-recovery charge.    Tony J. Pitcher  Professor of Fisheries Director, UBC Fisheries Centre     3 Preface and Acknowledg-ments   The purpose of the workshop documented in this report was to investigate the use of a re-cently developed ecosystem tool, Ecosim, to study the biological and economic impact of different harvesting regimes, based on files representing trophic models of a range of aquatic system types, previously constructed using the Ecopath approach and software.  This report contains (1) brief descriptions of Ecopath (vers. 4.0) and its Ecosim routine, with particular emphasis on the features important in simulating multi-species exploitation and its impacts; (2) descriptions of simulated fishing regimes and their impacts; (3) descriptions of problems encountered during the simulations and of the means these problems were or could in principle be overcome; and (4) discussions of the strength and weaknesses of Ecosim as a tool for simulating fisheries impacts on ecosystems, and of possible ways to improve the software and its underlying theory.  The participants of this workshop were largely drawn from the Fisheries Centre, UBC, where Ecopath is widely used, and where Ecosim was developed, but also included invited partici-pants from further afield, most familiar with Ecosim, or at least with Ecopath.  The valuable result obtained during this work-shop is a clear understanding of the potential usefulness of Ecopath, Ecosim and Ecospace, tempered by a realistic understanding of their limitations. Just right for a three-day event!  I wish to thank the participants for their enthu-siasm, and particularly Kevern Cochrane, of FAO, for the clear goals he provided, Villy Christensen and Felimon ‘Nonong’ Gayanilo, for the timely completion of an alpha version of Ecopath 4.0, including Ecosim and Ecospace, Rashid Sumaila and Reg Watson for their lec-tures, and Carl Walters for his outstanding presentations of the background of Ecosim and Ecospace, and for leading the workshop’s con-cluding discussion. Also, I wish to thank Ms. Gunna Weingartner for her preparation of and organizational support during the workshop. Funding for this event and the attendance of several participants from abroad was provided by the Food and Agriculture Organization of the United Nations (FAO), though the Trust Fund Project GCP/INT/ 643/JPN sponsored by the Government of Japan as part of its follow-up to the Kyoto Conference on the Sustainable Con-tribution of Fisheries to Food Security.  The David and Lucille Packard Foundation provided additional funding for some partici-pants from the US Northwest, while ICLARM provided funding for the participants from the Philippines, and the European Commission for a European participant. I thank all organiza-tions for their support, and hope that this re-port makes palpable some of the excitement generated during the workshop, which their generosity allowed us to organize.  The Editor Vancouver, May 1998  4Table of Contents  ABSTRACT………………………………………………………………………………………   1 DIRECTOR’S FOREWORD …………………………………………………..……….   2 PREFACE AND ACKNOWLEDGMENTS……………………………………….   3 TABLE OF CONTENTS…………………………………………………………………..   4 LIST OF EXHIBITS………………………………………………………………………..   5 INTRODUCTION FAO’S INTEREST IN AND EXPECTATIONS OF THIS WORKSHOP (K.COCHRANE)………   6 ECOPATH/ECOSIM APPLICATIONS IN THE EASTERN BERING SEA (A. TRITES)…………  8 LECTURES THE NEW ECOPATH WITH ECOSIM (V.CHRISTENSEN)…………………………………………………..  9 ECOSIM AND ECOSPACE: BASIC CONSIDERATIONS (C. WALTERS)……………………………………  11 ECOSIM AND MPAS: A QUASI-SPATIAL USE OF ECOSIM (R. WATSON AND C. WALTERS)………… 15 BIOECONOMICS AND THE ECOPATH/ECOSIM FRAMEWORK (R. SUMAILA)…………………………… 19 ECOSYSTEM SIMULATION MODELS AND THE NEW ‘BACK TO THE FUTURE’ APPROACH TO FISHERIES MANAGE-MENT (T. PITCHER) …………………………………………………………………………..  21 COMMENTS STRATEGIES FOR SUSTAINABLE USE OF MULTI-SPECIES RESOURCES (A. BUNDY)……………………… 24 FIRSTS THOUGHTS ON USES, STRENGTH, AND WEAKNESSES OF ECOPATH (K.COCHRANE)…………….. 24 BACK TO THE FUTURE WITH ECOPATH AND ECOSIM (N. HAGGAN)……………………….………….. 26 ECOPATH AND ECOSIM APPLICATIONS TO MARINE MAMMALS AND BIRDS (K. HEISE)………………… 26 HABITAT CONSIDERATION FOR USING ECOPATH/ECOSIM (A. JARRE-TEICHMANN)………………….. 27 ECOPATH, ECOSIM, MPAS, AND PELAGIC SYSTEMS (J. KITCHELL)…………………………………….. 28 EVALUATING STRATEGIES OF SUSTAINABLE EXPLOITATION (P. LIVINGSTON)………………………… 29 THE NEED FOR ALERT USERS (J.J. MAGUIRE)……………………………………………………………  30 ECOSIM APPLICATION TO LAKE VICTORIA (J.MOREAU)………………………………………………..  30 THE PRINCE WILLIAM SOUND MODEL (T. OKEY)………………………………………………………  32 MODELING THE EASTERN CENTRAL PACIFIC OCEAN (R. OLSON)…………………………………….. 34 ECOPATH, ECOSIM AND EVALUATING POLICY IN AN ECOLOGICAL CONTEXT (A. PARMA)……………. 34 IMPROVING FOOD WEB DESCRIPTIONS  FOR USE IN DYNAMIC SIMULATIONS (D. PAULY )………...…. 36 DISCUSSION POLICY USE AND LIMITATIONS OF THE ECOPATH APPROACH (R. WATSON AND D. PAULY)……….…. 38 REFERENCES…………………………………………………………………………………….  44 APPENDIX 1 Workshop Schedule…………………………………………………………………………………   47 APPENDIX 2  LIST OF PARTICIPANTS…………………………………………………………………………………  48  5 List of Exhibits   Box 1 Basic equations, assumptions and parameters of the Ecopath approach……………. 10 Box 2 Inputs required to turn Nile perch into a split pool………………………………………….. 31  Fig. 1 Simplified representation of trophic interactions in the Central South China Sea…. 11 Fig. 2 A feeding triangle in the South China Sea ecosystem. ………………………………….. 12 Fig. 3 How, in Ecosim, prey groups are split into available, and unavailable biomass. ….. 12 Fig. 4 Representing the linkages of a grid cell in Ecospace…………………………………….. 13 Fig. 5 How Ecopath groups may be partitioned into MPA and non-MPA portions………… 15 Fig. 6 Exchange of biomass between MPA and non-MPA biomass portions………………… 16 Fig. 7 Response of biomass after 10 years in Lingayen Gulf, Philippines…………………… 17 Fig. 8 Catches with MPAs of varying ‘sizes’ after 10 years, Thai 10 model…………………. 17 Fig. 9 Biomass response to a slow migration rate, Thai 10 model……………………………. 18 Fig. 10 Biomass response after 10 years of fast migration rate, Thai 10 model…………… 18 Fig. 11 Schematic representation of present and past ecosystems……………………………. 23 Fig. 12 Coarse-grid map of Prince William Sound, Alaska……………………………………….. 33 Fig. 13 Food web incorporating subsystems at lower trophic levels………………………….. 37   6INTRODUCTION   FAO Fisheries Department Inter-est in and Expectations of this  Workshop  Kevern L. Cochrane Fishery Resources Division, FAO   Amongst many other important principles, the FAO Code of Conduct for Responsible Fisheries (FAO, 1995) highlights the importance of multi-species approaches to fisheries management. For example, Paragraph 12.5 recommends that “States should be able to monitor and assess the state of stocks under their jurisdiction, includ-ing the impacts of ecosystem changes resulting from fishing pressure, pollution or habitat al-teration. They should also establish the research capacity necessary to assess the effects of cli-mate or environmental change on fish stocks and aquatic ecosystems.” This issue was also identified as being of par-ticular importance at the Kyoto Conference held in Japan in 1995. The Kyoto Declaration (Anon, 1995) arising from this meeting called upon signatories, among other things:  • To conduct integrated assessments of fish-eries in order to evaluate opportunities and strengthen the scientific basis for multi-species and ecosystem management • To promote allocation of human and finan-cial resources for an international pro-gramme to investigate the effectiveness of multi-species management of commercial fishery resources.  After the Kyoto Conference, the Japanese Gov-ernment established a trust fund to be adminis-tered by the Fisheries Department of FAO, to follow up on a number of recommendations contained within the declaration, including that above, to promote a strengthened scientific basis for multi-species and ecosystem manage-ment. Progress in understanding of multi-species dynamics in fisheries has been slow and many processes and principles of ecosystem function-ing are still very poorly understood. Neverthe-less, some progress has been made. In fisheries, there has been some progress in developing methods to increase our ability to assess fisher-ies as multi-species systems and hence also manage them as such (Walters et al. 1997). Probably the most comprehensive of these ap-proaches is that of Multi-Species Virtual Popu-lation Analysis (MSVPA). The major drawback of MSVPA is that it requires a large amount of data and information for application. The Eco-path approach, which relies on a ‘snap-shot’ of biomass pools and flows between them, as well as exports and imports, has been developed to require much less data and hence to be applica-ble in a much wider range of fisheries systems. In recent years, substantial progress has been made with the Ecopath approach, both in terms of the number of systems to which it has been applied and in the types of applications. Within the latter category, Ecoranger (allowing incor-poration of uncertainty), Ecosim (allowing simulation of ecosystem variables over time) and Ecospace (adding a spatial dimension to Ecopath) have been particularly important de-velopments (see Christensen, this vol., and Wal-ters, this vol. for brief descriptions of these tools). With reference to Ecosim, Walters et al. (1997; 1998) have suggested that is has the following potential uses:  • testing hypotheses about ecosystem func-tions; • policy screening for proposed ecosystem management strategies; • consistency checking for hypotheses about impact of long-term regime shifts; • evaluation of possible trophic causes for non-stationarity in single-species recruit-ment relationships.  The second of these potential uses, policy screening, is particularly relevant to the multi-species requirements of the Code of Conduct and to the Kyoto Declaration. FAO therefore approached the Fishery Centre, University of British Columbia, where several of the key sci-entists working on Ecopath are based, to host a workshop to evaluate the current status and capabilities of the Ecopath suite of assessment tools for potential application in multi-species assessment and management. The objectives of the workshop were defined as:  • to investigate the use of the Ecopath suite of software as a tool to study the impacts of different harvesting approaches on simu-lated ecosystems, with a view to application of suitable approaches in actual multi-species fisheries; and   7 • to identify and document the strengths and weaknesses of the Ecopath approach in this role.  If the Ecopath models are found, in some or all cases, to provide acceptably realistic represen-tations of real ecosystems, then it would open up the possibility of using Ecopath or Ecopath-type models to explore different management strategies and to identify the most appropriate strategy for implementation. This concept could lead to the use of such an ecosystem model as an operating model in a multi-species manage-ment procedure. A management procedure has been defined as a set of rules which specifies exactly how a man-agement recommendation (e.g., TAC, length of closed season, size of closed area etc.) is set and what data are used for this purpose (Butter-worth et al. 1997). These rules are selected based on their anticipated performance in the medium term (e.g. 10 - 20 years), as estimated by simulation on an ‘operating model’ of the resource and fishery or fisheries. Performance is defined in terms of selected indicators related to the resource and the desired benefits (typi-cally including indices of risk of undesirable impacts on the resources, benefits to the users and inter-annual variability in these benefits). In developing a management procedure, an integral and essential part of the process con-sists of ensuring that the selected set of rules is robust to likely uncertainties in the forecasts and data or observations. Based on this more formal approach to using an ecosystem model to guide management strate-gies, the broad objectives of the workshop can be broken down into more specific questions to be addressed at this workshop:  • Does the Ecopath with Ecosim approach encompass sufficient understanding of the dynamics of a multi-species ecosystem for a user to have a reasonable expectation that the real system will respond to a manage-ment strategy in the same way as estimated by Ecopath with Ecosim? • As a part of the previous question, can the Ecopath structure simulate adequately the fishery (or other use) as a component of the system, including sufficient information on e.g. age and species selectivities of gear? • Can the major sources of uncertainty be included and considered in an Ecopath analysis? • Once all reasonable uncertainty has been considered, is there any ‘signal’ remaining which will enable robust forecasts of eco-system response to a management strategy? • Can the common indicators for perform-ance criteria be included in Ecopath with Ecosim, and generated as an output from the system?  Clearly there are no absolute answers to these questions and the answers will vary amongst ecosystems and depend on the existing knowl-edge of the system and on the management strategies that are being considered. Neverthe-less, it is hoped that this workshop will provide adequate answers to these questions to aid peo-ple considering using the Ecopath suite, or an equivalent approach, to decide whether this will assist them in their attempts to understand and manage multi-species fisheries.    8Ecopath/Ecosim applications in the Eastern Bering Sea  Andrew Trites Marine Mammal Research Unit, Fisher-ies Centre   For the past eight months, a team of research-ers from the Fisheries Centre has been working with collaborators from the University of Alaska and the US National Marine Fisheries Service (NMFS) to construct ecosystem models of the eastern Bering Sea (1950s; and late 1980s - early 1990s) using Ecopath. The work, involving Pat Livingston (NMFS; see Livingston, this vol.) and the author, has been supported by the David and Lucille Packard Foundation and is in the final stages of write-up. The more recent model representing the Eastern Bering Sea is one of the most detailed ecosystem models con-structed to date and should become a useful tool for fisheries managers charged with apply-ing ecosystem concepts to the Bering Sea fisher-ies. Ecosystem modeling is still in its infancy, but stands to become a central tool in fisheries management. It is therefore important that ecosystem models, such as ours, convey insights and uncertainties to managers and fishers if they are to be used to enhance the conservation of marine life. There is probably no better way to ensure this than to draw on the collective experiences and insights gained by others using Ecopath and Ecosim. We were particularly pleased, therefore, that funds from the Packard Foundation became available to support the travel and participation of three researchers constructing ecosystem models of the Bering Sea, the Gulf of Alaska and the Pribilof Islands. The lessons learned by comparing our approaches and findings to those of researchers working on other ecosys-tems will be invaluable in ensuring that our results end up on the management table and become a useful tool for fisheries management in the North Pacific.  9 LECTURES   The new Ecopath with Ecosim,  version 4.0 Alpha.  Villy Christensen ICLARM   Over the last two years, work has been in pro-gress to develop a new version of Ecopath which integrates the Ecosim module for dynamic simulation modeling based on mass-balanced Ecopath models (Box 1).  This development has involved Carl Walters and Daniel Pauly at the Fisheries Centre, and Villy Christensen at ICLARM. At this Ecosim workshop, an incomplete Alpha version was used. The version incorporates Ecopath with most of its modules, plus Ecosim and the newly developed Ecospace module for spatial model-ing (see Walters, this vol.).  In my opening lecture at the first workshop session, when I gave an overview of the new version of Ecopath, the following features were highlighted:  • The new version is programmed for 32-bit Window system, and cannot be used with Window 3.1. Previously constructed models are now saved in a MBD-format database allowing for straightforward communica-tion with other databases, notably FishBase (Froese and Pauly 1998). For each model it is possible to save scenarios run with Ecoranger, Ecosim and Ecospace. The new database format is downward compatible with the previous ‘EII’ file format;  • The new version allows for entry of more detailed description of species included in the ecosystem groupings;  • It is possible to enter up to 10 gears or fleets in each model. For each gear, landings, dis-cards, market prices, fixed and variable costs can be entered. In addition a non-market value can be given for each ecosys-tem group. The breakdown in gears and in-clusion of simple bio-economic parameters is of relevance especially for fishery policy analyses using Ecosim (see Sumaila, this vol.);  • The Ecoranger module for parameterization of models using distributions or ranges for all basic input parameters and for address-ing uncertainty in a Bayesian context, has been improved. Also, it is now possible, when Ecoranger cannot find any balanced model, to save the best unbalanced model (BUM);  • The ‘Ecowrite’ system for adding and stor-ing remarks and references and document-ing inputs has been considerably expanded and now includes a system for documenting results as well. Also, the module for extract-ing and editing remarks and references has been improved.  The version used at the workshop was a test version, and a number of bugs were found dur-ing the workshop. This did not have any major significance for the course of the workshop and most participants were able to explore the soft-ware, its characteristics and abilities. Several of the participants had a good knowledge of the software prior to the workshop, which enabled them to work at an advanced level.  The bugs that were identified will be fixed be-fore the Beta version is sent for testing. This will be done as soon as the development and docu-mentation process has been completed. Mean-while the Alpha version can be downloaded from www.ecopath.org, or is available through Villy Christensen  at v.christensen@cgnet.com.  10  Box 1. Basic equations, assumptions and parameters of the Ecopath approach  The mass-balance modeling approach used in this workshop combines an approach by Polovina and Ow (1983) and Polovina (1984, 1985) for estimation of biomass and food consumption of the various elements (species or groups of species) of an aquatic ecosystem (the original ‘ECOPATH’) with an approach proposed by Ulanowicz (1986) for analysis of flows between the elements of ecosystems The result of this synthesis was initially imple-mented as a DOS software called ‘ECOPATH II’, documented in Christensen and Pauly (1992a, 1992b), and more recently in form of a Windows software, Ecopath 3.+ (Christensen and Pauly 1995, 1996). The ecosystem is mod-eled using a set of simultaneous linear equations (one for each group i in the system), i.e. Production by (i) - all predation on (i) - nonpredation losses of (i) – biomass accumulation of (i)  - export of (i) = 0, for all (i). This can also be put as Pi-M2i - Pi (1-EEi) BaccI - EXi = 0                                                     …1) where Pi is the production of (i), M2i is the total predation mortality of (i), EEi is the ecotrophic efficiency of (i) or the proportion of the production that is either exported or predated upon, (1-EEi) is the ‘other mortality’, Bacci is the biomass accumulation of (i), and EXi is the export of (i). Equation (1) can be re-expressed as  Bi*P/Bi - ΣjBj*Q/Bj*DCij-P/Bi*Bi(1-EEi) -Bacci -EXi =0              ...1) or Bi*P/Bi*EEi - ΣjBj*Q/Bj*DCij - Bacci -EXi = 0                                     ...2) where Bi is the biomass of (i), P/Bi is the production/biomass ratio, Q/Bi is the consumption/biomass ratio and DCij is the fraction of prey (i) in the average diet of predator (j). Based on (2), for a system with n groups, n linear equations can be given in explicit terms: B1P/B1EE1 - B1Q/B1DC11-B2Q/B2DC21 - ...-BnQ/BnDCn1 - Bacci -EX1 = 0  B2P/B2EE2 - B1Q/B1DC12 - B2Q/B2DC22 - ...-BnQ/BnDCn2 - Bacci -EX2 = 0  BnP/BnEEn - B1Q/B1DC1n - B2Q/B2DC2n - ...-BnQ/BnDCnn - Bacci -EXn = 0 This system of simultaneous linear equations can be solved through matrix inversion. In Ecopath, this is done using the generalized inverse method described by MacKay (1981), which has features making it generally more versatile than standard inverse methods. Thus, if the set of equations is overdetermined (more equations than unknowns) and the equations are not consis-tent with each other, the generalized inverse method provides least squares estimates which minimize the discrep-ancies. If, on the other hand, the system is undetermined (more unknowns than equations), an answer that is con-sistent with the data (although not unique) will still be output. Generally only one of the parameters Bi, P/Bi, Q/Bi, or EEi may be unknown for any group i. In special cases, however, Q/Bi may be unknown in addition to one of the other parameters (Christensen and Pauly 1992b). Exports (e.g., fisheries catches) and diet compositions are always required for all groups. A box (or “state variable”) in an Ecopath model may be a group of (ecologically) related species, i.e., a functional group, a single species, or a single size/age group of a given species.  11 Ecosim and Ecospace: basic  considerationsb   Carl Walters Fisheries Centre, UBC   This brief contribution, adapted from the mate-rial presented at two lectures, includes only pointers to the more detailed descriptions of Ecosim (Walters et al. 1997; Walters et al. 1998), and Ecospace (Walters et al.; see below for Abstract.), which should be consulted for further details.  Fig. 1 Simplified representation of tro-phic interactions in the Central South China Sea, indicating the biomass of some groups  (t• km2) and the fluxes between them  (t• km2• year-1).                                                            b Editorial note: this contribution is based on the PowerPoint presentation used by Carl Walters for the two lectures he gave at the Workshop, with text added by the editor to smooth the transition between ideas and/or graphs previously in separate exhibits.  The main elements of Ecosim are:  • Ecopath is used for estimation of parame-ters, based on the assumption of mass-balance; • Biomass and size structure dynamics: are represented by a mix of differential and dif-ference equations; • Variable speed splitting is used to model the dynamics of both ‘fast’ (e.g., plankton) and ‘slow’ (e.g., top predators) groups; • Micro-scale behavior is represented by al-lowing differentiation between top-down, intermediate and bottom-up control of pre-dation. Apex predators0.05Mesopelagics2.6BathypelagicsMicrozooplanktonBenthic fishBenthosDetritusPhytoplanktonLargezooplanktonEpipelagicnekton 0.50.5620.1120.4 12Interaction parameters between, e.g. the Apex predators, the Epipelagic nekton and the Mesopelagics (Fig.2) can be computed from the data in Fig. 1, viz.   Q(epipelagics to Apex) = 0.562  a(epipelagics, Apex) = 0.562 / (0.5*0.05) for Lotka-Volterra model, i.e., top-down control, and  Q = aViBj = avijBiBj / (2vij+aBj) for prey vulner-ability limitation.   Fig. 2 A feeding triangle in the South China Sea ecosystem, with data required to compute interaction parameters (see Fig.1).  Representing limited prey vulnerability is achieved, in Ecosim, by splitting each group’s biomass into an available and an unavailable component (Fig. 3).     Fig. 3 Illustrating how, in Ecosim, the biomass of a prey group is split into an available and an unavailable component. The symbols stand for: B = Total prey biomass; V = Vulnerable prey biomass; v = Behavioral exchange rate; P = Total predator biomass; a = Predator rate of search. Note that fast equilibration be-tween B-V and V implies V = vB / (2v+aP).  The dynamics of biomass is, in Ecosim, repre-sented by differential equations of the form   dB/dt = (Consumption)  - (Predation) +(Immigration)  - (Emigration)  - (Fisheries catches).  Their terms are defined by :  (Consumption) = Σ (micro-scale rates);  (Predation) = Σ (micro-scale rates); and  (Micro-scale rate)= aBpredvVprey    = aBpredBpreyv/(v’+aBpred).  Size-structured dynamics are considered only in ‘split pools’, which include the juveniles and adults of the same pool. For these, we have:  • Juvenile size/age structure by monthly cohorts, density- and risk- dependant growth; • Adult numbers, biomass, mean size ac-counting via delay-difference equations; and  • Recruitment relationship as an emergent property of competition/predation interac-tion of juveniles.  The remaining critical gaps and weaknesses of Ecosim are:  a) the parameter estimation does not account for highly seasonal environments; b) meso-scale spatial relationships, i.e., migra-tion are not accounted for; c) effects of change in habitat quality on trophic relationships can be represented only crudely; d) the articulation of policy option leaves much to be desired; and  e) emergent novelty cannot be considered; the model predicts opportunities for the growth of populations already included in the model, while, at least in highly disturbed sys-tems, increased vulnerability to invasion should be predicted.  PredatorPAvailable preyVUnavailable preyB-VaVPvVv(B-V)Apex predators(0.05)Mesopelagics2.6Epipelagicnekton 0.50.5620.1120.4 13 We should be able to address several of these deficiencies in the near future, notably (a) – (d). Item (e) , on the other hand, will continue to plague us, as models such as discussed here have difficulties dealing with novelty. The next routine to present is Ecospace, re-cently developed to provide a spatial dimension to the Ecopath approach. As it presently stands, Ecosim has the following features:  • Replicates Ecosim dynamics over a coarse grid of ‘homogenous’ cells; • Spatial cells are linked through dispersal, and the allocation and movement of fishing effort; • Spatial differences in primary productivity are represented; and  • Habitat ‘preferences’ are represented by differential dispersal, feeding, and preda-tion rates.  The Ecospace dispersal linkage may be repre-sented as in Fig. 4, where the m’s are assumed equal (symmetrical mixing), except at shores, and toward preferred habitat (a smoothing procedure generates the gradient used to ex-tend the ‘reach’ of preferred habitat).         Fig. 4 Representing the linkages of a grid cell in Ecospace.   The computational method involved in Eco-space involves a huge system of equations (20x20 grid for 10 pool model results in 4000 differential equations). Solving such a system of equations requires either much patience or a numerical approximation scheme. Lacking the former, I have included in Ecospace a numeri-cal approximation by linearization involving a matrix exponential solution method. This pro-duces rapidly converging, successive approxi-mations of spatial equilibrium. The method is efficient, but it is a good thing to always test for step size effects. Given the present unavailability of documenta-tion for Ecospace, I include below the abstract of a paper titled: “Ecospace: a software tool for predicting mesoscale spatial patterns in trophic relationships of exploited ecosystems, with special reference to impacts of marine protected areas”, by Carl Walters, D. Pauly and V. Chris-tensen, which will be presented at Theme Ses-sion (S) on ‘Visualization of Spatial (including Survey) Data’, of the ICES Annual Science Con-ference, Cascais, Portugal, September 1998. Here we go:  The growing disillusion with the predictive capability of single species assessment meth-ods, and the realization that the management approaches they imply will always fail to protect bycatch species, has led to growing interest in the potential of marine protected areas (MPAs) as a tool for protecting such species, and allow-ing for rebuilding populations of target species and damaged habitat.  Evaluating MPAs’ abilities to meet these re-quirements will demand both field experiments and simulations. However the tools required for the latter need not be as detailed as is often thought, and particularly, need not include links between resource species and physical processes.  Ecospace is a spatially explicit model for policy evaluation which allows considering the impact of MPAs in an ecosystem (i.e., trophic) context, and which relies on the Ecopath mass-balance approach for most of its parametrization.  Bijmi,j+1Bij 14Additional inputs are movement rates, used to compute exchanges between grid cells, the set-tings (top-down vs. bottom up control) also required for Ecosim, the dynamic simulation routine derived from the system of linear equa-tions in Ecopath and habitat preferences for each of the functional groups included in the model.  Convergence from the homogenous distribution assumed in the Ecopath base model to highly patterned distributions, simultaneously ac-counting for the habitat preference and food requirements of predators and preys, the distri-bution of fishing effort (driven by local abun-dances and fishing costs) and the existence of MPAs is extremely rapid, due to an integration scheme with different step sizes for the ‘fast’ and ‘slow’ groups, allowing the former to track the population changes of the latter.  An application example for coastal waters off Brunei, (Southeast Asia) is presented, docu-menting the ability of Ecospace to generate realistic spatial distributions of  functional groups, under constraints of habitat preference, distribution of fishing effort, etc.    ‘Cascade’ effects, wherein prey organisms are low where predators are abundant, e.g. in areas onto which high fishing costs have been mapped, or in MPAs are discussed; it is then shown that the potential benefits of local effort reductions can be easily negated by high movement rates, and especially by the concen-tration of fishing effort at the edge of the MPAs, where cascade effects generate prey gradients which attract predators out of the protected areas.  Despite various limitations (e.g., no explicit consideration of seasonal changes or directed migration), the outward simplicity of Ecospace, and the information-rich graphs it generates, coupled with the increasingly global availability of the required Ecopath files, should ensure a wide use for this approach, both for generating hypotheses about ecosystem function and evaluating policy choices.   15 Ecosim and MPAs: a quasi-spatial use of Ecosim  Reg Watson Fisheries Western Australia, Perth and Carl Walters Fisheries Centre, UBC   While the first version of Ecosim (Walters et al. 1997) offered many facilities to mana-gers, it did not provide a means of describing the spa-tial relations of biomass and fishing mortalities which are required to examine the potential impacts of marine protected areas (MPAs). To overcome this, we devised a simple modifica-tion to Ecosim which allows the biomass of Ecopath groups to be partitioned into two por-tions (Figure 5) with exchange processes oper-ating between them (Figure 6).       Fig. 5 Schematic representation of how the biomass of Ecopath groups may be partioned into MPA and non-MPA por-tions. One biomass portion would be assumed to be within an MPA and subject to different levels of fishing mortality (at least for some groups) than the other portion. If the biomass of Ecopath groups is assumed to be uniformly distributed in space, then the proportion of the biomass assumed to be in the MPA is also the proportion of the area of the marine system (described by the Ecopath) that is included in the MPA. This assumption allowed us to observe the impact of MPA ‘size’ and biomass exchange rates on the calculated biomasses and catches of Ecopath groups.    The rate of biomass movement of each Ecopath group out of the MPA was defined as   Rout = X / P  where X is a user supplied value and P is the proportion of biomass or the portion of the fishery described by the Ecopath model in-cluded in the MPA.   The balanced movement of biomass into the MPA is defined as   Rin = X / (P (1 - B))   The response of biomass and catch for a range of published Ecopath models were examined for MPA proportions ranging up to 80% of the total fishing ground. For each, a 10-year period was simulated and the harvest from the fishery was maintained by allowing the fishing mortality acting on the non-MPA biomass pools to in-crease as the MPA portion increased (up to a maximum of three times the original Ecopath model value). Large fishMedium-sizefishBenthosNon-MPA MPALarge fishMedium-sizefishBenthos 16Non-MPA MPABiomass BBiomass 1 - BX /   PX  P/(1-B)√   Fig.6 Exchange of biomass between MPA and non-MPA biomass portions. Parameter X is a user-supplied migration value and P is the proportion of the Ecopath group’s biomass within the MPA.      As expected, Ecopath models with low fishing mortalities did not exhibit large changes in biomass or catch from Ecopath levels even when a large proportion of the biomass was protected from fishing mortality within an MPA. In others, a range of responses was ob-served ranging from a simple linear increase or decrease (Figure 7), to that of a dome-shaped curve with an MPA ‘size’ corresponding to a maximum biomass and catch (Figure 8). Most impacted were heavily-fished top predators and their prey.  The biomass and catch responses of Ecopath models to MPA size was sensitive to assump-tions about X, the user-supplied migration rate, but this depended on the model or group under consideration. For a given MPA size, low migra-tion rates allowed greater biomass increases (Figure 9), while higher migration rates ex-posed biomass to fishing mortality and reduced the impact of the MPA (Figure 10). Higher mi-gration rates required larger MPAs, and the scale of the response was determined by the level of fishing mortality assumed in the Eco-path model. Early indications from work with Ecospace, a true spatial model (see Walters, this vol.), suggest that our findings are overly opti-mistic because the spatial aggregation of fishing effort at the borders of an MPA will reduce the average biomass response within the MPA.  17                           Fig.7  Response after 10 years of biomasses in Lingayen Gulf Philippines, as a function of the fraction of MPA to total ecosystem area [file: LINGAYEN] 0204060801001201400204060801001201401601800.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8Fraction of MPAPhytoplanktonMolluscs/JellyCephalopodsZooplanktonCrustaceansBenthosL. zoob.feed.Sm.dem.fish.Interm.pred.Large pred.Sm. pelagicsMedium pel.Large pelagicsSmall crustac.DetritusProfitabilityRelative biomass Fig. 8 Catch response with MPAs of varying ‘sizes’ after 10 years, inshore waters of the Gulf of Thailand [file Thai10] 050 1001502002500.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8Fraction of MPA  Scallops Misc.demersals LeiognathidsSergestids Small pel.Formio niger SauridaScombridsBarracuda Carangids LoligoMugil  sp.Phytoplankt.ZooplanktonZoobenthosJuvenile fishDetritus Profitability     Relative biomass  18051015202530354045.00 .10 .20 .30 .40 .50 .60 .70 .80Fraction of MPAPhytoplanktonMolluscs/JellyCephalopodsZooplanktonCrustaceansBenthosL. zoob.feed.Sm.dem.fish.Interm.pred.Large pred.Sm. pel.Medium pel.Large pel.Small crustac.DetritusProfitabilityRelative biomass Fig. 9 Biomass response to a slow migration rate (X = 0.5) of Thai10 Ecopath model to MPAs of varying ‘sizes’ after 10 years  0102030405060.00 .10 .20 .30 .40 .50 .60 .70 .80Fraction of MPAPhytoplanktonMolluscs/JellyCephalopodsZooplanktonCrustaceansBenthosL. zoob.feed.Sm.dem.fish.Interm.pred.Large pred.Sm. pelagicsMedium pel.Large pelagicsSmall crustac.DetritusProfitabilityRelative biomass Fig. 10 Biomass response of Thai 10 Ecopath model to MPAs of varying ‘sizes’ after 10 years of a fast migration rate (X = 4). Note that calculations can break down when both MPA fraction and migration rate are large.  19 Bioeconomics in the Eco-path/Ecosim Framework  Ussif Rashid Sumaila Chr. Michelsen Institute, Bergen and Fisheries Centre, UBC   This contribution does three things. First, it addresses the question of how to extend the Ecopath/Ecosim framework to allow the per-formance of bioeconomic analysis at the level of the ecosystem. The paper identifies two ways to do this: either by undertaking a basic or an advanced analysis. Second, the presentation covers what has been done so far to introduce bioeconomics into the framework by way of an example. Finally, a proposal is put across on how to proceed with the agenda of extending the Ecopath/Ecosim framework to allow the bioeconomic evaluation of different scenarios of exploiting fishery resources in an ecosystem. To pursue the objective of the presentation, the talk is divided into the following four sub-topics, (i) basic bioeconomic analysis, (ii) ad-vanced bioeconomic analysis, (iii) an example, and (iv) next steps: a proposal.  Basic bioeconomic analysis  The simplest way to introduce bioeconomics into the Ecopath/Ecosim framework is to take the biological results (catches and fishing effort) generated by Ecopath/ Ecosim under different scenarios, and apply appropriately determined unit prices for the fish landed, the cost of ex-ploiting the fish, and the discount rate. In this way, we are able to compute the net discounted economic rent that can be achieved under the different scenarios, which in turn allows us to determine the scenario that produces the best ecologically sustainable economic outcome. The point should be stressed here that this approach is only a basic and simple bioeconomic analysis, mainly because economic motivations do not enter into the decisions made regarding how much of what species to harvest, and when the harvest should be taken. All we do, which is very useful thing to do, is to evaluate alternative biological outcomes using economic parameters (that is, prices, costs and discount rates).  Advanced bioeconomic analysis  Advanced analysis deals with the limitations of the basic analysis by incorporating into the analysis the regulatory body’s and/or fishers’ behavior and motivations for fishing. This method does so by identifying who the stake-holders are, and incorporating what they care about. This is done by defining and incorporat-ing the objective functions and decision vari-ables of the participants into the analysis. It is then possible, through some optimization pro-cedures, to determine the harvest mix that gives the highest economic benefits without disturb-ing the ecological balance and impairing the sustainability of the ecosystem. Fundamental questions that need to be an-swered in order to design the analysis here in-clude (i) who controls the fishery? Is it a single, powerful, benevolent controller who decides how much of what species to harvest? (ii) Or do the fishers with the right to exploit the ecosys-tem go about the exploitation on their own? (iii) If the answer to question (ii) is yes, do the fish-ers work together in a cooperative manner or in a noncooperative one? Answers to these questions will set the stage for designing a truly bioeconomic model of the ecosystem based on Ecopath/ Ecosim. What is more, the answers will allow us to analyze the distribution of the economic benefits to be de-rived under different management strategies or policies (see Sumaila, 1997).  An example using Hong Kong Fisheries  An Ecopath/Ecosim model for Hong Kong fish-eries, with a network of MPAs and artificial reefs (ARs) is developed in this example (see Munro and Sumaila, in press; Sumaila, in press). Different scenarios were created and the model run to produce various biological results in the form of biomass, catches and fishing effort levels, for different species of fish in Hong Kong waters. A basic bioeconomic analysis of this fishery was carried out (i) by determining appropriate unit prices for each group of fish in the fishing habi-tats of Hong Kong using available data; (ii) working cost of landing a unit of a particular group of fish using a given vessel type; and (iii) determining an appropriate discount rate for Hong Kong. Combining (i)-(iii) with the catch and effort levels obtained from Eco-path/Ecosim, a basic bioeconomic analysis was done, generating the net economic gains that can be achieved by incorporating different sizes of MPAs and ARs. Thereby, making it possible to determine the MPA and AR combinations that give the best economic outcomes (Munro and Sumaila, in press).   20Next steps: A proposal In the short term, I think we should concentrate on developing and improving the basic bio-economic analysis. This improvement can be brought about by developing good databases on prices and costs. However, the long term goal should be to incorporate the behavior and mo-tivations of fishers and fisheries managers into the bioeconomic analysis. In fact, work on the latter can form a good basis for a Ph.D. thesis, and interested candidates may contact the au-thor, or either of Drs. Pauly and Pitcher to ex-plore this further.   21 Ecosystem Simulation Mod-els and the new Back to the  Future Approach to Fisher-ies Management  Tony J. Pitcher  The rebuilding of resources, rather than sus-tainability, represents a new policy goal for fisheries management (Pitcher and Pauly 1998). Such a policy likely represents the only hope for the future for fisheries targeting wild living resources, which have been progressively and seriously depleted (e.g. Pauly et al. 1998). This approach attempts to reverse the ratchet-like- ecological processes caused by human fishing, which have been largely ignored by a fisheries science primarily concerned with single species population dynamics (Pauly and Pitcher 1998; Pitcher in press a).   In the 'Back to the Future' (BTF) approach, scientific tools are used to construct and evalu-ate present and past ecosystems. The policy objective for management becomes the rebuild-ing of the past system that would, if restored, maximise economic benefit to society. The ap-proach is fundamentally different from a policy goal of sustainability, which leads only to sustaining our present misery.  In summary, the BTF agenda for fisheries man-agement comprises six elements: 1. model construction of present and past aquatic ecosystems; 2. evaluation of economic and social benefits for each system; 3. choice of system that maximises benefits to society; 4. design  of instruments to achieve this policy goal; 5. evaluation of costs of these management measures; 6. adaptive implementation and monitoring of management measures.  A marine ecosystem model of the present day is used as the starting point for a reconstruction of the system as it might have been prior to the start of modern industrial fishing. Models of several past systems are constructed using data from archives, government and universities including fisheries management databases, the traditional environmental knowledge (TEK) of Indigenous Peoples, the local environmental knowledge of commercial and sport fishers (LEK), and the archaeological record (Fig. 11). Economic evaluations may take into account present values in global seafood markets, reve-nue foregone because of overfishing and stock collapses (see Sumaila, this vol.), and social, cultural, amenity and conservation values. The social benefits include reduced inter-sectoral conflict. Evaluation also includes the costs of implementing and monitoring restoration. Practical restoration techniques to achieve the new policy goal call upon Marine Protected Areas in addition to conventional fishery man-agement (Watson and Walters, this vol.; Pitcher in press a)   Workshops to help build models of past systems can act as a neutral forum where opposing sec-tors meet and share knowledge in the interest of long-term conservation.  Comparing species levels predicted by the model with TEK and other perspectives on past abundance provides both a talking point and a means of cross-validation. Focussing on past abundance high-lights what could be achieved, as opposed to fighting over present scarcity (see also Haggan, this vol.). Moreover, when such policy goals are identified, an ecosystem-based agenda means that, during rebuilding, the public can act as sentinels of progress, and many diverse groups, including industry, aboriginal peoples, schools and colleges can have roles in providing data (Pitcher in press b). A sense of ownership of the process and goals fosters cooperation and re-duces conflict. Restoration accords with the natural resource philosophy of many aboriginal peoples. Additionally, the BTF agenda provides an economic rational for restoration that can benefit all sectors.  Ecosystem modeling, such as that covered in this workshop, has to be an integral part of the methodology required for the BTF approach. Mass-balance Ecopath and Ecosim modeling has advantages in making clear the impacts of harvest, comparing the effects of different gear types, and in being able to provide estimates of unknown biomasses. It can therefore validate anecdotal information on presence or absence, and relative abundance of fish species (Haggan, this vol.). A disadvantage is that in its present state of development, ecosystem modeling is not itself able to provide single species quotas because, generally, many species have to be combined into one 'box'. Conventional stock assessment methods will continue to be needed, but biomass values for single species will have to be constrained by the results of the ecosys-tem model. Ways of merging current sophisti-cated single species stock assessment methods  22with ecosystem modeling of the impacts of har-vest need to become an active research area. Ecospace modeling can play an important part in designing reserves and management tactics needed to achieve BTF policy goals. The Back to the Future approach to fisheries management is in its infancy, and many details remain to be worked out, but a pilot study has been carried out in the Strait of Georgia, B.C. Canada, and work is in progress in the nearby Hecate Strait, and in Hong Kong. Also, proposals for BTF work are being developed for several areas in Southeastern Indonesia. Watch this space!                                                                           Next page:  Fig. 11 Schematic representations of present (right) and past (left) ecosystems, reconstructed using the Ecopath software and different types of source (written documents, oral history, archeo-logical evidence). Dotted lines indicate range of reliable extrapolations using Ecosim, limited by structural changes such as caused by extinctions and invasions.  23    500 YEARS BP 100 YEARS BP PRESENT DAY ECOPATH MODEL ECOSIM LIMITSTRADITIONAL  KNOWLEDGE EXTINCTION ABUNDANCE HISTORICALDOCUMENTS ARCHAE-BBAACCKK  TTOO  TTHHEE  FFUUTTUURREE      ➢➢      SS TTRRAAII GGHHTT  OOFF  GGEEOORRGGII AA   24COMMENTS   Strategies for Sustainable Exploitation of Multi-species Resources  Alida Bundy DFO St. Johns, Newfoundland  The following questions were asked by K. Coch-rane (this vol.): Does Ecosim encompass ade-quate understanding of the system? Can it ade-quately simulate the fisheries? Can major un-certainties be combined? Is there sufficient certainty to give adequate robustness? Does Ecosim estimate indicators for performance?  I present a few comments regarding these ques-tions. However, since the input parameters for Ecosim are taken from mass-balanced Ecopath models, I have focused on the use of Ecopath and the nature of inputs to this model.   In an applied fisheries situation where data is to be collected to construct an Ecopath model (as opposed to using models that are already con-structed), there are considerable data require-ments. Although, compared to models such as MSVPA, the data requirements are relatively small, the demands of this part of the Eco-path/Ecosim approach should not be underes-timated. Carl Walters (unpublished data) has shown that, in particular, reliable diet composi-tion data make the dynamic simulations more robust (see also Pauly, this vol.). At this stage, investing time into model parameterization is worthwhile. This serves at least two purposes. First of all it produces a better model and sec-ondly it gives the model greater legitimacy.  This latter point is relevant where construction of the model involves accessing outside exper-tise and ‘selling’ the model to managers, fishers and the public.  I also make this point because if Eco-path/Ecosim is to be used as a policy tool by FAO, then we are no longer dealing with ideal-ized systems built from guesses based on ex-perience with more or less comparable systems. Since the evaluation of policy strategies will be site-specific, a reasonable attempt should be made to obtain a good parameterization of the Ecopath model. In addition, if parameters are continually transferred from one to the other model, then the same few systems will be repli-cated and our arguments will become circular.  I think that the Ecopath/Ecosim approach is a very exciting development, allowing for wide exploration of many multispecies and ecosys-tem issues in fisheries. It opens a world simply not accessible before. The impact of fishing by different types of fishing gear on a multispecies resource can be explicitly examined under dif-ferent hypotheses of flow regime (‘bottom up’, ‘top down’…). With the addition of Ecospace, it has become possible to examine these interac-tions at a spatial level and to investigate the effect of MPAs.  At a practical level I found the software rela-tively straightforward to use, although like Heise (this vol.), I did have difficulty in input-ting the parameters for the split pools, which had worked in an earlier version of Ecosim. It would have been good to have time to explore the new software in more detail.     First thoughts on uses, strengths and weaknesses of the Ecopath  suite   Kevern Cochrane Fisheries Department, FAO  These comments are based largely on listening to the presentations and discussion during the Workshop, and some exploratory examination of the different components.  I did not have the opportunity to undertake detailed simulations and evaluate the results for an ecosystem with which I am familiar.  1. Limitations  In common with any modeling approach, the Ecopath suite has its limitations and it is impor-tant that these are carefully considered when Ecopath, Ecosim and Ecospace are used to as-sist in evaluation of management strategies.    25 Some of these are: 1.1 The models are based on food webs, and do not incorporate other features of ecosystems which may also be impor-tant.  Examples of these include the role of physical factors in driving ecological processes (e.g. long-term trends in tem-perature), ecological interactions involving resources other than food, e.g. competition for space, etc; 1.2 The level of aggregation commonly encountered in food web data sets, par-ticularly at the lower trophic levels, may mask resource partitioning (see Pauly, this vol.), resulting in incorrect simula-tions of interactions (e.g. food web analysis may indicate anchovy and sar-dine as important competitors. How-ever, experience has indicated that these two species favour different eco-logical conditions even though the dif-ferences have not been clearly deline-ated; see also Jarre-Teichmann, this vol.); 1.3 Inevitably, there are high levels of un-certainty in the food web structure, arising from estimates of all the inputs.  It is vital that the impact of these uncer-tainties is considered explicitly when using the models to guide management decisions. Ecoranger provides a means of quantifying some of these uncertain-ties and needs to be used in this role.  The effect of assumptions on the results should also be checked; 1.4 At present, individual stocks can only be disaggregated into two age/size groups.  This means that age effects, e.g. of fishing, can only be considered in an approximate manner, in contrast to many single-species approaches which allow explicit consideration of age/size structure.  2. Potential Uses  If the limitations of the Ecopath approach, such as those presented above, are considered in interpreting model output, there can be no doubt that the suite of models represents a ma-jor step forward in enabling routine considera-tion of management issues and plans in an eco-system context.  Depending on the system un-der consideration and the information avail-able, the following applications can be envis-aged in relation to fisheries management:  2.1  Ecosim provides a very useful approach for evaluating the multi-species impacts of fisheries.  Simulations under a vari-ety of conditions, critically interpreted, would enable evaluation of ecosystem ramifications, providing input to plan-ning of multi-species management ap-proaches.  It should be used with out-put from Ecoranger, or other informa-tion on uncertainties, to enable some robustness testing of management ap-proaches;  2.2 Ecospace is a potentially widely appli-cable tool and will enable rapid screen-ing of the spatial dynamics of, e.g. closed areas, heterogeneous distribu-tion of fishing effort, impacts of changes in oceanographic features which influence primary production, etc.  This is crucial, as the importance of considering the spatial characteris-tics of ecosystems, fisheries and stocks is increasingly being recognized. For many countries and stocks, the infor-mation and time may not be available for constructing site-specific spatial models.  In these cases, Ecospace pro-vides a user-friendly and easy to use tool for at least preliminary examina-tions of many of these problems.  Overall, I believe the package is extremely use-ful and, critically applied, can provide very use-ful information complementary to existing as-sessment/simulation methods. Its great value lies in the fact that it will enable at least some consideration to be given, within a structured framework, to ecosystem and spatial effects, in a wide range of cases where, until now, such features have been essentially overlooked.   26Back to the future with Ecopath and Ecosim   Nigel Haggan Fisheries Centre, UBC   Back to the Future (BTF) is a new approach which proposes rebuilding rather than sustain-ability as the proper goal of fisheries manage-ment (Pitcher and Pauly, 1998). The rationale is found both in recent work documenting the decline in trophic level brought about by indus-trial fishing (Pauly et al. 1998), and the long-standing concerns of indigenous and artisanal fishers about the effect of industrial harvest (Haggan, 1998). The Ecopath mass-balance approach to aquatic ecosystem modeling has parallels with the way longstanding fishing communities view the environment. Both are more concerned with relationships, interactions and connections within an ecosystem than with achieving a deep understanding of isolated elements (Haggan, 1996). BTF uses Ecopath to re-construct the species composition, relative abundance and productive capacity of marine ecosystems at some past level, say before the onset of modern industrial fishing. For example, a recent BTF project of the UBC Fisheries Centre and the UBC First Nations House of Learning developed models of the Strait of Georgia ecosystem as it might have been 100 years and 500 years ago. The first step is to create an Ecopath model of the present system. This can either be done as a student project, or as a major workshop bring-ing together experts in the various ecosystem components. Either way, the model focuses discussion and input from government science, university science, the traditional environ-mental knowledge (TEK) of indigenous com-munities, the knowledge of commercial and sport fishers, archival sources and the archaeo-logical record. For almost the first time, the BTF methodology provides the TEK of aboriginal peoples and maritime communities with a valu-able, direct function in resource management.  Perhaps even more importantly, three elements combine to promote cooperation between a diverse group of stakeholders. First, a univer-sity-based unit, such as the UBC Fisheries Cen-tre, can act as a neutral forum where frequently opposing sectors can meet and share knowledge in the interest of long-term conservation. Sec-ond, comparing the abundance of species (or functional groups) in an Ecopath mass-balance model with TEK and other views provides a starting point for discussion, and an element of cross-validation. Third, the abundance in the ‘good old days’ may provide a useful contrast to the present, often inequitable access to the re-sources. Economic evaluations of past ecosystems (see Sumaila, this vol.) can then be compared with the present. Restoration goals, which can be simulated using Ecosim, can be based on the economic, social, and cultural values attainable by rebuilding. It requires no great stretch of the imagination to see the same interests agreeing on rebuilding goals and working together on ways to get there.  Current BTF projects initiated through the Fisheries Centre, UBC include a reconstruction of the Hecate Strait ecosystem of northern Brit-ish Columbia in cooperation with the Tsimshian and Haida Nations. Also, a re-construction of the Hong Kong fishery as it might have been 50 years ago, prepared by T. Pitcher and R. Watson will form the basis for a major workshop in Hong Kong (see also Pitcher, this vol.).    Ecopath and Ecosim applications to marine mammals and birds  Kathy Heise Marine Mammal Research Unit, Fisher-ies Centre  It was really helpful to see how Ecopath has evolved since the 1995 Workshop (see Pauly and Christensen, 1996) and I have a better sense of the potential that Ecopath and Ecosim have to provide insight into how changes in biomass at one trophic level can be transferred through the food web. I felt in 1995, and con-tinue to feel that going through the experience of developing a mass-balance model is a tre-mendously valuable heuristic process. In terms of evaluating the software as a man-agement tool, there were obvious limitations that were directly related to bugs in the present version of the software that make it difficult to provide intelligent comments. However, I can highlight a few areas that were of interest to me and which I would like to study further, once these bugs are removed. First, I would like to explore linking juvenile and adult age classes and test under which conditions the additional level of data that this separation implies is worth collecting. The juvenile age class of many fish species are important prey to marine mammals and seabirds and using Ecopath to see what the effects are of altering juvenile bio-masses in higher trophic level predators would be very interesting. Because I couldn’t link the  27 juvenile and adult classes during the workshop, I couldn’t examine the effects of MPAs on to recruitment; I think this should be worth pursu-ing further. My own interest would then be to experiment with the placement of seabird colo-nies within the MPA and to find out at what distance the birds would receive benefits, if any. I also appreciated the option that Ecospace has to alter movement rates, and thought that this could be useful for experimenting with the ef-fects of diffuse vs. tightly schooled fish and their vulnerability to predation by marine mammal and birds.    Habitat Consideration for using Ecopath/Ecosim  Astrid Jarre-Teichmann Danish Institute for Fisheries Research / Chair, ICES Habitat Committee  The models which serve as background to this brief contribution were presented in Jarre-Teichmann (1995), and Jarre-Teichmann et al (in press) and refer to the Baltic Sea, and the Southern Benguela upwelling system, respec-tively. Both models represent comparatively simple ecosystems with regard to species composition. They also share the feature that their compo-nents are largely determined by oceanographic settings. Further, the database for building both of these Ecopath models was relatively good. The split-pool groups in the Southern Benguela model were the two hake species with a ‘small’ (0-2 years) and ‘large’ (age 3+) group each. For the Baltic Sea, detailed age structures for the commercial fish species (sprat, herring and cod) were available from MSVPA assessment, itself based on long-term stomach time series of cod, the top predator. Therefore, the Ecopath model included 4 age groups for cod (0,1,2,3+) and 3 each for sprat and herring (0,1,2+), where the plus group represents the adult stock. Accord-ingly, the juvenile groups were aggregated for the Ecosim analyses.  Congratulations to Villy, Carl and Daniel for making available a tool which, already in its alpha version, runs better than the released versions of Ecopath 3, and for enabling ecosys-tem modeling to make a tremendous leap for-ward. I particularly value the potential of Eco-path/Ecosim to open a discussion between ‘hard-core’ fisheries (stock assessment) scien-tists and marine ecologists-oceanographers -chemists-ecotoxicologists. This discussion will be crucial to any development towards ecosys-tem management.  As mentioned repeatedly during the workshop discussions, Ecosim runs do not show realistic behaviour for groups, whose habitat (or niche) is not primarily defined by trophic interactions. Sardine and anchovy, for example, compete for food in most upwelling systems. Hence, in Eco-sim, one species can be favored strongly ex-ploiting the other. However, when anchovy- or sardine-dominated regimes are observed in the real world, it is usually found to be due to wind-induced structuring of their reproductive habi-tat (Bakun, 1996). Similarly, recruitment suc-cess of cod is critically dependent on the oxygen conditions in the deep basins of the Baltic Sea (Jarre-Teichmann et al., submitted), which in turn depend on saltwater inflow into the Baltic from the North Sea.  To increase the ability of Ecosim to credibly respond to management questions, it would be helpful to allow it to explicitly address habitat-related issues, such as pollution; eutrophica-tion, oxygen depletion, etc., and their effect on the size of suitable habitat.   28Approaches for dealing with these issues may include: 1) developing the approach taken by J. Dalsgaard (Fisheries Centre, UBC, MSc. thesis, in prep.) for tracing radioactivity through the ecosystem into a generic, Ecopath-based routine for tracing the movements of pollutants through a food web; 2) allowing for inclusion of ‘rules’ in the simulation such as resulting from (a) threshold values for pollutants above which productivity would decrease and/or mortality would increase, or from (b) critical biomass levels of some com-ponents in the system, whose presence may be beneficial to others, but in a non-trophic way, e.g. by providing shelter; 3) recommending to users that environ-mental aspects of fishing, e.g. benthic habitat impacts through destruction of corals, sponges, etc., should be explicitly included in the simulations, along with the ‘catches’ and discarded by-catch thus generated. This would allow the related fluxes to be explicitly included in food webs, and thus provide a basis for ad-dressing the issue of ‘shelters’, i.e. refuge from predation.    Ecopath, Ecosim, MPAs, and pe-lagic systems  James F. Kitchell University of Wisconsin  This brief report offers an evaluation of my experience during the Ecopath/ Eco-sim/Ecospace Workshop held at the UBC Fish-eries Centre during 25-27 March 1998. First, I am strongly supportive of this effort and its continued development. I believe that the com-bination of models is a very powerful approach: Ecopath offers the benefit of a solid foundation created by delineating the components and their quantitative interactions; Ecosim allows the expansion of that condition through simula-tion of the response to future conditions; while Ecospace allows even greater capacity in view-ing the spatial context of dynamic interactions and the consequences of management actions imposed on landscape-scale units of habitat. As a learning exercise, I sought to use an Eco-path model we have constructed for the Central Pacific, then apply that to a general problem pertinent to questions about Marine Protected Areas (MPAs) in a pelagic environment. I set up a series of simulations designed to evaluate the interactions between refuge size, mobility of the apex predators and differences in basic produc-tivity of refuge habitat. Similar proportions of the habitat were set aside in each case, but set out as many individual units grading to a single, large unit. Each refuge was defined as an area where fishing was not pursued and, in the sec-ond series, where primary production was greater than in the surrounding habitat. The main conclusions from that exercise are that:  • Apex predators with low swimming veloci-ties effectively link a series of small and proximate MPAs. Populations of those predators in the general region of a refuge are enhanced above that of the effect within the refuge areas alone. Immobile apex predators increase within each of the dis-crete refuges. Highly mobile apex predators disperse the local effects and the effect of the refuges gradually disappears as preda-tor mobility is increased. Similar, but recip-rocal, effects occur among key prey re-sources;  • As refuge size was increased, positive ef-fects on apex predators increased and nega-tive effects on their prey became more ap-parent. In other words, large refuges am-plify the effects of reduced exploitation on food web interactions;  • If the refuge habitat is characterized by higher productivity, trophic cascade effects appear at intermediate to large refuge sizes; these is also evidence of a long-term cyclical oscillation in all components of the food web. The period of that oscillation roughly corresponds to the life history characteris-tics (i.e., generation time) of the apex predators.   29 In summary, the workshop environment is im-portant on two counts. First, the presentations by the developers of this software actually ex-plained how it works, the conceptual framework embedded in its equations, and, equally impor-tant, how it won’t work. In my experience, that understanding cannot be fully derived from readings, or from simply downloading and us-ing the software. Second, the workshop pro-vided opportunity for exchange of ideas and anecdotes that expanded the perspective of participants. My only concern about this workshop focuses on the present state of the software. The user must have confidence that output of the model-ing effort actually derives from the scenarios developed by the user and is not biased by un-known programming problems.  This condition must be met in future workshops and, we hope, will soon be resolved through access to the completed beta version.     Evaluating strategies for sustain-able exploitation  Pat Livingston Alaska Fisheries Science Center, Seattle  The models I have used to evaluate Ecopath and its extensions are recently constructed models of the eastern Bering Sea shelf in the 1950s and 1980s (see Trites, this vol.). Previous exercises with these models indicated an inability to pro-ject from the 1950s state using Ecosim and to achieve the 1980s state, in both a quantitative and qualitative sense. The reasons for this are many and include the usual list of suspects: incomplete knowledge of the 1950s state and a model that does not contain any details on the early life history of gadoids implicated in the changes from a cold to a warm regime. Another cause might have been some unaccounted-for spatial dynamics. I am presently examining different levels of pollock in the 1950s that might come close to obtaining a pollock-dominated ecosystem in the 1980s. Other scenarios I want to examine further are: (1) projecting the 1980s model forward to the present to see how closely it matches the pre-sent ecosystem state; and (2) using Ecospace to examine some hypotheses about spatial over-laps between certain key predators and prey. If we do not spend sufficient time examining model configurations and our ability to accu-rately predict multispecies changes either in a qualitative or quantitative sense under known conditions, then I do not see how we could hope to convince management that advice obtained from this modeling framework is useful.  Of course, I find these models extremely interest-ing and useful from a scientific point of view in developing and testing hypotheses about eco-system structure and function. But making the leap from providing advice on future research efforts to providing advice to guide manage-ment actions will require meeting a certain burden of proof.  I would like to see some model facilities added to the package that could possibly aid in this endeavor. An iterative procedure that would try different combinations of parameter changes to go from a given historical Ecopath state to at-tempt to match a future observed state would be useful. Traditional stock assessment scien-tists often show how model estimates compare with observations and may also provide a way to incorporate time series of observations about the state of the population into the assessment procedure itself. Perhaps there is a way to in-corporate time series of observations about biomass levels of certain ecosystem components into Ecosim’s projections from the past, i.e., from a previously observed state to the present state.  I am still trying to sort out what management questions and what time scales can be ad-dressed with Ecosim. In the Bering Sea, we have relatively conservative fishing regimes com-pared to other systems and this system defi-nitely has seen cyclic fluctuations of important fish species in the last 25 years or so. Tony Pitcher's characterization of sustainability “sus-taining our present misery” (see Pitcher, this vol.) is difficult to apply to this system because we are still trying to understand how the rela-tively light fishing effort can have induced the observed population fluctuations. Also, there is no clear ratcheting down of species. Thus, sepa-rating human effects from climate change ef-fects is a very high priority for the Bering Sea, especially with regard to questions about what would happen to pollock production if we moved back into a cold regime. Other questions that fishery managers and stock assessment scientists are asking include: what are the ef-fects of unequal harvesting rates on individual components of the groundfish complex on the resulting community composition? How does pulse fishing impact upper trophic level preda-tors? I need to do some further exploration of the behavior of the Ecopath/ Ecosim/ Ecospace package, and of its fit to observations, before I can judge whether it could be used to answer  30these questions. I see the possibility that many interesting scientific and management ques-tions might be addressed with Ecospace and look forward to the model exploration and vali-dation that will be required for further moves in this direction.    The need for alert users  Jean-Jacques Maguire Halieutikos / Chair ICES  ACFM  Ecopath modeling is a useful way of summariz-ing and verifying information on ecosystem structure and functioning. The Ecopath/Ecosim combination makes it possible to investigate potential changes to the ecosystem as a result of fisheries management measures. Creative in-terpretation of the parameters can significantly expand the number and type of questions that can be investigated with Ecopath/Ecosim.  The Ecosim/Ecopath/Ecospace suite of soft-ware is a powerful analytical tool because it can produce results even if only sparse data are available. When used by skilled and knowledge-able operators, this characteristic will represent an impressive advantage. However, it may be-come a liability when novice and/or unskilled users utilize the approach with inappropriate data or with improper assumptions. It would therefore seem reasonable to encourage train-ing in the use of the approach. Moreover, it would be prudent to associate the principal developers in any advisory process making use of the methodology. It would take only a few misuses of the approach to permanently dis-credit the Ecosim/Ecopath/Ecospace suite with fishery management authorities.    Ecosim application to Lake Victo-ria  Jacques Moreau Ecole Nationale Supérieure Agronomique de Toulouse  The introduction of Nile perch into Lake Victoria had an immense impact on the structure of that ecosystem. A previous contribution presented quantitative box models for two different periods in the history of Lake Victoria (Moreau et al. 1993). The first model, describing the late 1960s and early 1970s, emphasized the role of haplochromine species in the Lake Victoria food web. The second model, describing the mid to late 1980s, showed the ecological importance of Nile perch as it became the dominant predator [Note that these describe only the shallow and intensively exploited Kenyan sector of the lake and thus do not apply to the lake as a whole]. This note describe some of the steps that were taken to adapt these previous models for use with Ecosim, and to test whether Ecosim can simulate the transition from the late 1960s situation to that prevailing in the late 1980s, thus complementing the account on Lake Victoria Nile perch in Walters et al. (1997). Box 2 provides details on how the VICTOR85 file originally created by Moreau et al. (1993) was modified such that Nile perch could be treated as a split pool. The results of the various runs then performed were as follows:  Equilibrium runs  Except for a small labeling error immediately fixed by V. Christensen, this routine worked flawlessly. Its results confirmed the observation of Walters et al (1997) that the relationship of catch vs. fishing mortality is flat-topped for a wide ranges of fishing mortalities, thus suggesting a Beverton-and-Holt recruitment curve. Walters et al. (1997) suggest this to be due to the effect of adult Nile perch not only consuming juvenile Nile perch, but also the latter’s competitors. By this, they increase the food available to the juveniles, thus increasing their growth rate and decreasing the time they spent in a vulnerable stage.  Simulation runs  The runs were performed both with effort aggregated, and with total effort split into different gear, i.e., gill nets, mainly yielding large Nile perch and Nile tilapia (Oreochromis niloticus); longlines (for Nile perch and bottom fish of minor importance); beach seines (for juvenile Nile perch, tilapiine and haplochromines); and ‘mosquito’ nets (for Rastrineobola, small Nile perch and miscellaneous other species). The runs with (aggregated) effort increasing over time generated unsurprising patterns of rapid decline in most groups, except for the tilapiines, which declined less than expected.  The responses of the various runs using effort disagreggated by gear types were too varied to be summarized here. Suffice to say that they were largely realistic, thus justifying the decision to make available, in Ecopath 4.0, a routine allowing such disagreggation. This is particularly true as  31 Ecospace now allows for spatial disagreggration of effort as well. Indeed, this complements neatly the split-pool option of Ecosim, which has allowed overcoming the cannibalism almost invariably generated within groups that have piscivorous adults. This leads to the notion that functional groups in the Ecopath suite should be defined by: (1) one food type; (2) one habitat type; (3) one gear type. This would perhaps resolve the issue of dynamic stability raised by Pauly (this vol.), and discussed in Watson and Pauly (this vol.).  Further, one way to reduce the resulting proliferation of functional groups would be to combine the prey groups contributing less than 1% to the diet of major predators, especially if these groups also contribute little to the fisheries catches. The other suggestions I have concerning consideration of habitat changes are similar to those of Astrid, and hence I refer to her contribution (Jarre-Teichmann, this vol.).      Box 2  Inputs required to turn Nile perch into a split pool. The maximum size of the juveniles was assumed to be 40 cm, the length at which they tend to escape predation by adults (see Ogari and Dadzie, 1988); also their feeding habits at that size begin to resemble those of the adults. Note that up to this size, young Nile perch are observed mostly in the littoral areas of the lake (Hughes, 1986), and that 40 cm TL is the minimum size fish are caught by the gill nets fleet (at least in the mid 1980s). The P/B ratio of the juveniles thus defined was estimated using the Compleat ELEFAN Software (Gayanilo et al. 1989), based on length frequency distributions for the relevant range of sizes from Asila and Ogari (1987), and  C. Rabuor (pers. comm). The estimate is 3.5 year-1. Q/B was estimated separately for the juveniles and adults using the MAXIMS software of Jarre et al. (1990), and parameters from Moreau et al. (1993) for the 85-86 period, viz. W∞ = 76 000 g; K = 0.36 year-1; Z = 0.85 year-1 ; Wr= 2000 gr; Wmax = 72 000 g.  This led to Q/B = 5.03 year-1; gross efficiency (GE) = 0.16, and ‘beta’ = 0.10 for the adults. The same inputs were used for the juveniles, except for Z = P/B = 3.5 year-1; Wr = 25 g; and Wmax = 2000 g, the last two values representing the lower and upper limits of the integration performed by MAXIMS. This led to Q/B = 11.8 year-1 and GE = 0.29 for the juveniles [Note that GE is higher for the juveniles than for the adults, in agreement with theoretical and empirical considerations (Pauly, 1986; Pauly and Palomares, 1987)]. The value for juveniles EE was set at 0.98, given the strong predation pressure on this group, while the diet composition for the juveniles and adults was adapted for data in Ogari and Dadzie (1988), and Hughes (1986), and considering the distinct spatial distribution of the two stages. Ecopath 4.0, when balanced with Nile perch as a split group, generated biomass estimates nearly the same as presented in Moreau et al. (1993), where Nile perch was a single group.  32The Prince William Sound Model  Tom Okey Fisheries Centre, UBC  An Ecopath model of Prince William Sound (PWS), Alaska is being developed by Fishery Centre scientists through a collaboration of researchers with expertise in various compo-nents of the Prince William Sound ecosystem. The development of this model is funded through the Exxon Valdez Oil Spill (EVOS) Res-toration Council, and thus will be somewhat focused on questions relating to the 1989 EVOS. A variety of researchers and research groups have been gathering information about the PWS ecosystem during the years since the oil spill, and some groups have developed models of certain portions of the ecosystem (Dalsgaard and Pauly, 1997). A great deal of information has been collected, but the overall EVOS re-search program has been recently criticized for producing a small amount of useful information relative to the amount of money spent to date (Paine et al, 1996). As a result, program peer reviewers are calling for initiatives that will synthesize the information collected thus far. The Fisheries Centre, in collaboration with the University of Tennessee (specifically Dr. Stuart Pimm), is now leading the ecosystem synthesis efforts based on the suggestion that the Ecopath approach is well suited for accomplishing the sort of synthesis desired by EVOS program architects. However, a variety of other benefits and applications may be derived from the Eco-path approach such as answering questions related to fisheries exploitation in the region.  Ecopath and Ecospace  I am particularly impressed and excited about the new Ecospace component of the Ecopath package because of the additional realism it introduces to the work with Ecopath.  Organisms do not interact homogeneously in space or time. Furthermore, spatial refugia enable increased prey presence or persistence in the real world, while concentrating interac-tions in space. Top-down and bottom-up effects are probably both important in ecosystems, but refugia can be thought of as a ‘third effect’ that plays a crucial role in structuring communities, shaping the interactions within these communi-ties, and enabling stability, whether these refu-gia are biotic or abiotic. With this in mind, it seems silly to expect realistic predictions or simulations from models that are not spatially explicit. The new Ecospace enables food-web simulations within a spatial and habitat con-text. Although it is still unclear to me just how accurately current Ecospace models can repre-sent real world interactions, given current con-straints on desktop computing power, it is clear to me that a framework for such realistic model-ing exists within Ecopath with Ecosim and Eco-space, and that this is a breakthrough. Figure 12 illustrates a coarse-gridded spatial representation of the Prince William Sound, as required by Ecopath. This representation con-sists of a 31 x 36 square grid in which PWS land and water have been defined. This diagram was created by overlaying position-referenced grid-lines over a map of Prince William Sound, and it has been transferred into the Ecospace grid system for analysis of PWS data. At the PWS Ecopath Workshop held on March 2-5, 1997, in the EVOS Restoration Council office in Anchor-age, this map led to a fierce debate about the spatial resolution of Ecospace, due to the sound’s extreme geographic heterogeneity rela-tive to the diagram. However, since the grid is used for distributing trophic interactions during model runs rather than for estimating input parameters, the method does nothing but heighten realism relative to alternative spatially homogeneous models. Perhaps as important as the realism advances referred to above is the accessibility that Eco-path with Ecosim provides. Although it is help-ful for front end users (those compiling site-specific models) to understand the underlying algorithms and processes, the concepts them-selves are accessible to general users as a learn-ing tool. Indeed, Ecopath models can be used as an educational tool for school children of vari-ous ages, and even by fishery decision makers. In any given management setting, however, a primary question is how to use this modeling approach to address existing dilemmas or ques-tions. Before models are constructed, larger questions about goals must be asked. For ex-ample, is restoration to a previous desired state the main goal? Or is prediction of stock size trajectory in response to particular exploitation regimes a more important goal? Alternatively, is the goal to evaluate particular management or     33 conservation measures e.g. marine protected areas? These questions should influ- ence aspects of research design and also deter-mine whether historical models should be de-veloped to compare to current models.    61.2'   148 W   147 W   146 W61 N60.4'60.2'60 N    Fig. 12 Coarse-grid map of Prince William Sound, Alaska, as created to apply Ecospace to that system. Note straightening of arms and coastlines, to allow unimpeded migration along uninter-rupted series of grid squares (see text).    34Modeling the Eastern Central Pa-cific Ocean  Robert J. Olson Inter-American Tropical Tuna Commis-sion  As a person with only modest familiarity with Ecosim, my participation in the workshop was primarily a learning experience. I am preparing to lead an effort at IATTC to assemble an Eco-path model for the eastern tropical Pacific Ocean (ETP) to explore the ecological implica-tions of three different fishing strategies cur-rently employed by the tuna purse-seine fish-ery. In the ETP, the top levels of the food webs consist of large tunas, dolphins, sharks, bill-fishes, and others. The purse-seine fishery tar-gets yellowfin tuna, but substantial catches of other apex predators also occur. The species composition and magnitude of the associated by-catch, and the size-age distribution of the yellowfin catch, are distinctly different for the three aggregation types and fishing strategies. We intend to incorporate historical and recent data from the ETP into Ecospace. We will work closely with Chris Boggs, NMFS Honolulu, and Jim Kitchell, University of Wisconsin, who are developing an Ecosim model for the Central North Pacific (CNP; see Kitchell, this vol.). During the workshop I examined two models, the latest version of the CNP model, and the Central South China Sea model, distributed as a test file (OCEANSCS) with Ecopath. I tried, with mixed success, to split the apex-predator pool and to add sharks as a functional group. That exercise provided me a quick lesson on the component requirements for mass-balance. The CNP model provided me a pertinent system to manipulate because it has similar components to the pelagic ETP. Jim Kitchell and I made simulation runs in Ecosim and Ecospace that provided interesting results. J. Kitchell tried a hindcasting of sorts by ratcheting down fishing effort to correspond to the onset of the Asian longline fisheries after the Second World War. That exercise showed that yellowfin tuna bio-mass prior to fishing was 3-fold the current biomass, which was the same equilibrium for yellowfin biomass predicted at zero fishing mortality using the equilibrium routine in Eco-sim. This was an encouraging result. Then, us-ing the appropriate routine of Ecospace, I ‘sketched in’ the Central and South American coastline (although I did not re-scale the grid properly) and added the high-productivity zone associated with the upwelling caused by the eastern boundary currents. I also associated fishing with that zone. The predators aggre-gated fast to the zone and had drastic effects on the ecosystem. The exercise convinced me that Ecospace represents a useful tool to simulate the food web impacts resulting from the differ-ent fishing strategies in specific regions within the eastern Pacific. I am not yet convinced that Ecosim provides sufficient certainty about the system, at least with some system types. The biomass and life history characteristics of many components of open-ocean systems are virtually unknown. However, my skepticism has decreased, and as my limited experience with the model grows, I believe my confidence in the model will solidify.    Ecopath, Ecosim and evaluating policy in an ecological context  Ana Parma International Pacific Halibut Commis-sion  My impressions about the potential of Ecopath-Ecosim/space as a tool for developing harvest-ing strategies are largely based on what I learned during the workshop. I did not have any prior experience dealing with Ecopath, and I have little experience in ecosystem modeling. My background is on assessment and manage-ment of single species, and the focus of my re-search has been on developing assessment and harvesting strategies that are robust to the ma-jor uncertainties we face in fisheries manage-ment. To address the basic question posed at the start of the workshop, there are two components of Ecosim that I would like to discuss separately: the component dealing with the ecology of the system; and the component dealing with the policies that can be evaluated.   35 The policy component:  The policies that Ecosim is readily set up to evaluate are harvesting policies, that is, policies that regulate the gear and amount of harvest of the different harvestable stocks in the system. The development of Ecospace expands the range of choices considerably, as spatially ex-plicit policies can now be assessed without hav-ing to resort to pseudo-spatial models. This still excludes some very important approaches for management of multi-species resources, i.e.,  policies that involve manipulating the habitat, and policies that attempt to modify the behavior of the fishers through individual incentives and/or penalties. The latter may be the only way to manage industrial fisheries on multi-species assemblages. Getting to exploit the most productive species without overfishing the less productive ones may involve a selective alloca-tion of effort in space at a spatial resolution much smaller than we can hope to achieve by quota-by-area management. I am, however, not too concerned about the policy aspect. First, the limitations are obvious and so there is little danger of them being over-looked. Second, considering the speed at which the approach has evolved over the last few years, I would expect the policy component to be rapidly expanded, especially considering the range of possibilities opened by Ecospace.  The ecology component:  In contrast to the situation with policy evalua-tion, the limitations of the ecological basis of the model may not be so apparent to the users. The risks of the users ‘believing’ model predic-tions is real no matter how loud and clear the developers of the model may state that the tool is not intended to produce detailed ecological predictions. This is unfortunate, because Eco-sim/Ecospace may be a great tool to have if one is clear about its limitations.  as a tool for helping in the design of adaptive policies and monitoring schemes, for construct-ing scenarios for policy evaluation, for generat-ing qualitative predictions as a way to validate some of the underlying ecological hypotheses which have strong management implications, and for challenging I am very impressed with Ecosim/Ecospace ideas and expectations. One of the things I heard repeatedly over the work-shop was that people were often surprised by the results, which is good. The problem of believing model projections is certainly not exclusive to Ecosim. However, I tend to think that it may be worse here than in other models in which uncertainty is more ap-parent. While Ecopath, through the Ecoranger routine, has a way to translate input uncertainty into output uncertainty, much of that is lost in the transit to Ecosim. This, of course, need not be the case. Users could generate many differ-ent projections from the Ecoranger posterior distribution, but I suspect the tendency will be to base all analyses on the ‘best’ Ecopath model. The power of Ecopath/Ecosim is that you can build models with very limited data, thanks to the mass-balance assumption. But to under-stand better the strengths and limitations of the approach, we will need the collective experience from those that use Ecopath in data-rich situa-tions, where there is enough information to contradict model predictions and to drive model development. For example, one of the limitations of the approach is that the equilib-rium assumption may lead to parameter esti-mates that are wrong and misleading.  This could be explored by varying the biomass ac-cumulation of the Ecopath master equation using available information on population trends in data-rich situations. My impression is that as we are just starting to see Ecosim applications, there is still much to learn about model development from data-rich situations. The only way to take advantage of this powerful tool is to understand its weak-nesses and explore them. Thus, endorsement of Ecosim as a tool for policy design should be very explicit about the limitations of the ap-proach.     36Improving food web descriptions of use in dynamic simulations  Daniel Pauly Fisheries Centre, UBC  Use of food webs as a key input for time- and space-structured simulations imposes more constraints on their quality than their use in the context of Ecopath proper, which does not con-sider their stability in a dynamic context, and only requires them to be mass-balanced. Following discussions with various colleagues, notably Stuart Pimm, of the University of Ten-nessee, I would like to suggest that food webs prepared for use in Ecosim and Ecospace should have the following properties:  1. at least 20 ‘boxes’ (i.e., ‘pools’, or state vari-ables) representing all major groups and trophic levels; 2. cannibalism should be avoided, or at least not contribute more than 1-2 % of a group’s diet. This can be achieved by disaggregating groups into split pools (i.e., juveniles and predators, see Walters, this vol.), or into separate functional groups, of which one is the predator, the other the prey; 3. cycles must be avoided wherein group i feeds  mainly on group j, and group j mainly on group i (use the ‘Cycles’ routine of  Eco-path to identify such groups); 4. the base of the food web (primary produc-ers, e.g. phytoplankton or seagrass, etc., and herbivores, e.g., zooplankton) should be separated by habitat within the ecosys-tem that is being modeled, e.g., inshore vs. offshore, or rocky bottom vs. mud bottom (Fig. 13).  Consideration of these four points will not re-solve all problems that may be generated by a questionable food web; however, some patho-logical behaviors will be avoided, and the mod-els, when run with Ecosim and Ecospace, will not self-simplify as readily as when these sug-gestions are not implemented.  37      Fig. 13 Example of a food web incorporating subsystems at lower trophic levels, i.e., inshore and offshore subsystems. Such separation leads to more realistic predictions by Ecosim, and facilitates the assignments of different groups to habitats, as required by Ecospace   Large predatorsLarger fishSmall pelagicsOmnivorouszooplanktonHerbivorouszooplankton IIPhytoplankton IIHerbivorouszooplankton IPhytoplankton Iparrot fishsea grassMerging of inshoreand offshore web athigh trophic levelsP r o c e s sI N S H O R E  H A B I T A TO F F S H O R E  H A B I T A T 38DISCUSSION   Policy Uses and Limitations of the Ecopath Suite and Approacha  Reg Watson Fisheries Dept., Western Australia and Daniel Pauly Fisheries Centre, UBC  This report documents a free-ranging discus-sion on issues raised during the workshop, structured by Carl Walters, who acted as facili-tator. Specifically, this involved grouping the predictions of Ecosim/ Ecospace into three classes: 1) predictions which can trusted; 2) predictions, that require local test-ing/verification; and 3) predictions which, for various reasons, cannot be trusted.  Discussants are identified by their initials, and their interventions were regrouped and edited, thus making our discussion look more formal than it was.  1. Predictions that can be trusted.  CW: concerning predictions, there are two that are fairly robust: - primary production limits possible total production, through it will not determine whether a system is stable or not. Here, Eco-path and its dynamic routines, Ecosim and Eco-space can help; - equilibrium community composi-tion responds to fishing: as you alter fishing mortality, a number of changes in abundance are predicted. Notably, the system will tend to become top heavy when fishing is very low. Conversely, at very high fishing mortalities, long-lived species will be lost, while the changes at intermediate level of fishing mortality are largely indeterminate, i.e., we cannot assess their reliability. In fact some of the results for intermediate levels of fishing are goofy, with several stable states of which we don’t know if they are realistic or not. But the extreme predic-                                                          a Present during the discussion were E. Buchary, A. Bundy, V.Christensen, K. Cochrane, F. Gayanilo, A. Jarre-Teichmann, J. Kitchell, J. Moreau, T. Okey, R. Olson, D. Pauly (Editor), T. Pitcher, A. Pongase, R. Sumaila, C. Walters (Facilitator), and R. Watson (Rapporteur).  tions are reliable, including e.g., prediction of cascades. DP: If so, one should be able to say that first-order predictions are largely reliable: groups that are exploited decline, and their major prey increase. OK? CW: Yes, furthermore, we found, in some cases, that the fishing mortality which maxi-mizes yield is much smaller than M, in some others it is larger. Gulland’s rule, that Fmax = M is #@%!!.  TP: How do you explain these differences? Perhaps we should stick to the most conserva-tive relationship between Fmax and M. KC: We certainly need policies that are robust to uncertainty of this sort. DP: Not only do we need to be cautious about the value of Fmax, but also about the absolute value of the predicted yield. The point here is that in the ‘development decades’ of the 1960s and 1970s, what was done in several countries, notably in Southeast Asia, was to divide the catch of a typical trawler into the yield potential predicted by Gulland’s equation, then deploy the number of boats that came out of the divi-sion. TP: Does Carl’s point about goofy results being obtained at intermediate levels of F imply that that our ecosystem work is worthless, or no better than single species assessments? KC: We should do both type of work, as there is enough overlap for both to benefit from the other’s results. We should also benefit from attempts to explain the causes of observed dif-ferences. TP: If there are differences, we cannot say both sets of results are valid. Moreover, industry will also go against the more conservative results which tend to come out when you consider feeding interactions. DP: In fact, you get a reduction of predicted yields in multispecies situations even when you don’t consider feeding interactions; all you need is fixed ratios of F between the different species, as often occurs with trawlers. John Pope showed that very nicely in a study he did of the Gulf of Thailand trawl fishery. Incidentally, this is the very reason why we lose large, long-lived bycatch species. CW: Trophic models cannot really provide practical answers to multispecies management problems. TP: Then how about Multispecies VPA, which is used in Europe? DP: They have made real progress with that in the North Sea, even if they don’t use it directly to determine TAC for the various species. What they do is use MSVPA to refine the inputs to the single species assessments, e.g. the M values  39 used for young fishes, which are now set much higher than before. KC: But there are still uncertainties; for exam-ple, the relationship between sardine and an-chovy in upwelling system is competition, but this cannot be captured by trophic interactions. This creates huge uncertainties. TP: Yes, but adaptive management, including ‘pushing the system’ is hard to sell as a way to find out more. Multispecies/ ecosystem models should be inherently more predictive than sin-gle-species methods. In addition to predictions of the type provided by single-species models, they make predictions which single-species models simply cannot make. We have to get people used to these kinds of predictions. CW: I agree that we can make broad predic-tions. We can predict that prey fishes will be affected if we remove their prey. This is not what some people want, though. What they want are detailed predictions of precisely what the biomass of each group will be next year. And this we cannot do. Let’s move on to an-other category of predictions, concerning the efficacy and time requirements for transition policies. This involves, among other things, estimating what it takes for experimental pro-grams, etc. to work. Some policies will never show effects, e. g. the setting up of very small MPAs.d DP: In Alaska, the Exxon Valdez Oil Spill Res-toration Council must certify whether the vari-ous elements of the Prince William Sound eco-system have bounced back or not. We are now working on that using a consensus Ecosystem model as our starting point (see Okey, this vol.). In the oiled area, the slow groups (marine mammals, birds) have not returned to pre-spill levels, and I expect we will be able to find this to be the aftermath of a shock to the food web, i.e., there might be no need to assume continued effects of oil residues. KC: I have a related question: what do you monitor in multispecies management, and how is the monitoring done? CW: We should monitor F directly, by tagging everything. KC: All species? CW: All species that need direct monitoring. Estimating F is important because multispecies problems are due to the effects of a combination                                                           d Editors note: I can’t resist pointing out here that some people believe (indeed: have shown) that very small MPAs can increase the biomass of relatively large, potentially very mobile  fish, such as parrot-fishes and snappers. See C.M. Roberts and J.P. Hawkins. 1997. How small can a marine reserve be and still be effective? Coral Reefs 16:150. of gear, and attempts at effort reduction must not be limited to one gear. Multispecies prob-lem occurs when one gear is cut back and other gears move in, like the sport fishery, and cause new problems. JK: We should be testing for the effects of key-stone species, and for effects caused by feeding triangles such as the Norway pout-euphausiids-copepod case that Villy Christensen presented (in Pauly et al. 1998). Alida Bundy’s analysis of San Miguel Bay dynamics, where species inter-actions appear to be very strong is another case (Bundy, 1997). CW: We will just have to accept that some type of changes is just not predictable. Also, we must be aware that Ecoranger can deal only with some types of uncertainties; it cannot deal with the uncertainty associated with the equilibrium predictions of Ecosim, which are due to the structure built into the systems’ description. Anyway, let’s now turn to our next topic.  2. Predictions that need local testing.  CW: As mentioned before, there are some complex webs for which the outcomes of Eco-sim are indeterminate, i.e., the directions of response to gear changes can go either way. A case in point is Alida’s model of the heavily exploited San Miguel Bay, where clear predic-tions could not be achieved for most groups, given policy changes. This is particularly pro-nounced for the intermediate trophic levels, while the top and lower levels behave as one would expect. DP: This might be due to overaggregated pools at intermediate levels, with diet compositions that are too broad. CW: It is true that, with overgeneralized pis-civores, problems will occur, but Alida’s mid-trophic level pools were not that generalized. DP: Our partner in the Prince William Sound project, Stewart Pimm, says that overaggre-gated lower trophic levels are the main cause of instability and self-simplification in food webs, and that this is the first thing one should look for when problems occur; also feeding cycles between pools at the same trophic level should be avoided. Stuart said he is working on a diag-nostic system, later to be incorporated into Ecopath, which will identify these problems (see Pauly, this vol.). CW: This has been known a while; but this will not resolve the problem of indeterminacy in the intermediate trophic levels. KC: In upwelling systems, notably in the Ben-guela system, sardine vs. anchovy competition is not well understood and is not predictable.  40Note that this is a general problem of lack of knowledge, not a problem with Ecopath. CW: People working on coral reefs think parti-tioning is the key problem to study, because it is what maintains reef diversity. Space, not food is the problem they focus on. DP: I don’t agree that it is their consensus. In fact, many of them, especially in Australia, be-lieve recruitment limitation and variability to be the major structuring elements for coral reefs. In any case, don’t we agree that space being important, we should construct food webs with subwebs referring to different parts of one’s system, and especially separate the phyto- and zooplankton groups? CW: This is unnecessary, and probably wrong; there are better ways to achieve stability. TO: I believe it is wise to follow Pimm’s advice; we are including over 40 groups in the Prince William Sound model we are constructing to analyze the impact of the Exxon Valdez Oil Spill (Okey, this vol.), and use these to define well-separated inshore and offshore webs, merging only at the top. VC: Another way to generate stability is by using low movement rates in Ecospace; this leads to clear spatial separation. CW: Space-structured models can allow co-existence of about anything at some scales, and prevent self-simplification of food webs. DP: I still think Pimm’s suggestions are useful, especially since they generate food webs whose subwebs correspond to the very spatial struc-ture that you say is required. CW: Even if you follow Pimm’s suggestions, some webs will still self-simplify; there are many cases where we don’t know what keeps the system stable. Clearly, spatial separation plays a role at some scales, but this is not the whole story. There are key processes we still do not understand. Even Villy’s detailed Ecopath model of the North Sea is not stable when run on Ecosim. KC: We have to avoid the reductionist trap, and not get into an endless process of digging deeper and deeper and never getting to the key process; clearly each model we use must fit specific circumstances. DP: If we match our spatial separation with our food web, we should be OK. The point is to avoid the bias we have as fisheries people, to lump zooplankton and phytoplankton into great big boxes, because we don’t know about them – though the planktologists do. CW: Sub-models may be the answer, with link-ages at the top, e.g. through the marine mam-mals feeding in the different sub-models. DP: This is what has been done, in effect, in some of the coral reef models that have been published so far. There, subsystems were de-fined (sea grass, lagoon, crest, slope) and linked by groups that feed in two or more of the sub-systems. In fact, I believe we should return to the Ecopath models that do not include this type of structures, and fix them before we use them for spatial modeling. VC: Let’s be cautious before we make such changes; first we must check that they really do what Pimm says they do. CW: I checked the number of pools in Ecopath models vs. the maximum value of the vulner-ability multipliers that could be accommodated without self-simplification of the food webs, which more of less corresponds to S. Pimm’s test of stability. There was a general trend to-ward loss of mid-trophic level pools, except in models with very detailed diets. Medium-sized models were unstable. DP: Pimm’s routine will do more or less the same thing, for any model we want to analyze, as a part of the Ecopath diagnostic system. CW: The way to go about this problem of sta-bility is not necessarily via better ecology, but through a better look at the policy questions and the inherent credibility of answers. KC: Some will want to assure themselves that the ecology included in their model is as good as possible. VC: Certainly users should check for cycles in their models. This can be done easily using  a routine of Ecopath. KC: But we don’t want a cookbook either, or people mindlessly generating numbers. Better perhaps to have a number of guidelines, such as Carl’s check of the effect of changing the vul-nerability schedules. DP: But some of the guidelines we teach to Ecopath users do have cookbook character, e.g., ‘avoid cannibalism’, ‘do not include less than 12-15 boxes spread over the whole food web’, etc. CW: Let’s now move to the third group of pre-dictions.  3. Predictions that cannot be trusted.  CW: The biggest question we have is that relat-ing to animals with trophic changes during their life history. We have tried to resolve this through split pools, but generally, these split pools, in Ecosim, tend to predict too much compensatory change. This is disappointing. It could be due to a failure to describe the factors affecting survival or changes in life history, etc. JK: Behaviour is context dependent, and rules should not be inflexible. One useful rule though is that the P/B value tells us how fast unex- 41 pected things can happen; groups with low P/B cannot change as fast as others. KC: I can imagine, that there might be policy situations in which it would help to distinguish more than the two size/age groups presently allowed as ‘split pools’ in Ecosim. CW: I don’t believe having more size/age groups would make much of a difference. Be-sides, the data and computational requirements would be so enormous, we would lose all pre-sent advantages of the package. In any case, it would be very difficult to have a highly detailed size/age distribution consistent with the rest of Ecopath/Ecosim/Ecospace. In fact even our present, relatively simple representation leads, through various amplifications, to highly com-plex behavior, e.g., in the case of Jim Kitchell’s model of the Central Pacific. Another source of problem is when we use the time-shapes in Ecosim to represent changes in productivity. DP: If we build models without physical forc-ing, we can’t well introduce such forcing though the back door, as it were, and expect it to work well. VC: Yet we do that with the economic compo-nent, and it appears to be of some use. KC: Is it really true, Carl, that these predictions are completely “hopeless”? CW: They are ‘just so stories’ and cannot be validated. Just like the various empirical mod-els that link recruitment and some environ-mental parameter, which all break down the year after they are published. DP: Seems to me that the Cury-Roy (Cury and Roy, 1989) hypothesis of dome-shaped recruit-ment windows did not break down. In fact, it has so far survived every test to which it was put. TP: Could not the ‘time shaping’ in Ecosim be replaced by a proper, if somehow generic oceanographic model to simulate production at the lower trophic levels? DP: The folks working on Prince William Sound have a good physical model, which nicely predicts phytoplankton blooms, the growth of the zooplankton that feeds on it, and the effect on the juveniles of some fishes. The problem is that they cannot put the system higher up in the same modeling framework. Thus, they cannot deal, e.g., with killer whales. TP: Something like that would make lots of sense in Peru and other upwelling ecosystems. DP: Such model would make sense there, as not much of interest to the fishery happens at the highest trophic levels, now that the birds, etc. are gone. VC: Clearly, a model must respond to a specific need for predictions. CW: Exactly; in our Grand Canyon work, we must model things on an hourly basis, because this is the scale at which interventions (water releases) happen, and most of the ecological impact flows from there. KC: Can we not use Ecoranger to deal with some of the uncertainties here? JK: I don’t think so; the uncertainty is the structure of the system itself, so the true uncer-tainty will no be captured by a sensitivity or Monte-Carlo analysis of an existing model. DP: So we delude ourselves when we use Ecoranger to define prior distribution, generate  ‘random’ models, look at the posterior distribu-tions, etc? JK: It’s like painting with a broad brush. It gives a broad picture; whether this is ‘real’ or not is another question. Perhaps it is ‘halfway’ correct. CW: This makes me think of these various es-timates of speed of light: all had confidence intervals about them. Yet the next estimate was invariably outside of the interval. This is similar to ecosystems, where qualitatively new behav-iors emerge which are outside of the range of prediction of previous models. DP: In Ecoranger, we can resample not only parameters such as biomass, P/B ratios, etc, but also the diet compositions. This means that by randomly generating new linkages between groups (where such linkages are possible), we can, in principle generate new system behav-iors. However, nobody has used this routine to that end yet. KC: So we all agree that model uncertainty cannot be overcome, but that Ecoranger is use-ful, in that it provides a measure of the mini-mum amount of uncertainty one has to accept in a given model. VC: I certainly agree with that. Indeed, Ecoranger has now been accepted by lots of colleagues who previously were critics of the Ecopath approach. DP: This should conclude our debate of Carl’s three types of predictions. Let’s now briefly talk about what comes next. For one, Ecospace will be presented at the next annual ICES Science Conference, in Portugal (see abstract in Wal-ters, this vol.). Also, Ecosim and Ecospace will now be incorporated as elements of the Ecopath training courses to be given in the context of a large international project funded by the Euro-pean Commission. Perhaps I should ask Villy to briefly describe this. VC: The project Daniel just mentioned is called “Placing fisheries in their ecosystem context”. It involves 31 institutions in Europe, West and South Africa, the Caribbean and Latin America as partners. It is the intention over the next 4  42years to arrange a number of training work-shops and conferences aimed initially at devel-oping Ecopath models, and next on comparing published Ecopath models across latitude, de-gree of exploitation, and so on. Colleagues in-terested in the activity are very welcome to con-tact me for details. DP: Then there is the conference, sponsored by IOC/SCOR’s Working Group 105, and ICES, on the Ecosystem Impacts of Fisheries, to be held next year in Montpellier. I should chair its ses-sion on ‘Trophic Impacts’. My own contribution to that will probably be to review the trophic level concept, so important in Ecopath. VC: There is also the FAO-sponsored workshop to which the present workshop was the prepara-tion. I think this should be organized like an ICES working group, and involve very knowl-edgeable colleagues, willing and capable to per-form a systematic examination of the features of key predictions. TP: This will require a more formal structure than we had during this workshop, with steps clearly outlined beforehand. KC: I agree. Indeed, there should be dummy policies for people to test, resembling those which are usually evaluated. AP: Recently, the U.S. National Research Council released a report on methods for fish stock assessment in which the methods were all applied to a dummy data set. VC: The participants, if from the ICES area, will need data sets which enable comparisons be-tween the outputs of Multispecies VPA and those of the Ecopath suite. They might other-wise not be interested. KC: But real Ecopath users will also be needed. TP: Experts having experience with other methods are important, lest the workshop might be preaching to the converted. The NRC approach that Daniel mentioned might be best, as it introduces a degree of objectivity in the evaluations.  DP: So what do we conclude for this workshop? RS: If we can trust the type of predictions Carl identified as trustworthy, then we should be in good shape. AP: But we need more than Carl’s opinion. CW: Here it comes, nevertheless: one thing we will have to watch is the possibility of bugs as explanation for some of the strange patterns we got with split pools. But in any case, we should not give too much attention to time transient patterns. Generally, I am not surprised by our results, which are mostly O.K. JK: When split pools gave us problems, we assigned the two stages to separate ‘species’. This helped. Another observation I have is that the MPA scale determined by Ecospace seems to work, and can be used to screen policies and eliminate bad thinking about economic trade-offs. KC: My comment about the workshop as a whole is that I’m happy with the way it went, and with its results. JK: I can only recommend that the software now be made ready for beta testing by as many users as possible, so its remaining bugs can be ironed out. Then, let’s organize blind round robins for further definition of its capabilities, as suggested before. The software will then grow and reach its potential.  VC: We will do that. We would be thankful if people sent us detailed bug reports, which we will fix in the version that can be downloaded from the Web. The address from which to download Ecopath is www.ecopath.org. or you can contact me at v.christensen@cgnet.com. AB: Will there be a separate user guide, as for the DOS version, or an online guide, as for ver-sion 3.0 and 3.1? VC: The beta test version, to be distributed some months from now, will have an online guide, updated from that in version 3.1. This workshop and some others we ran provided ideas as to what to add to the available material. We will have to reverse-engineer a few parts of the text, based on the routines we now have, and their documentation in the primary litera-ture. KC: This has been a very successful workshop and I thank Carl and Villy for their contribu-tions and Daniel and Gunna for the workshop organization. Far more has been achieved than I anticipated, and I can see that we now have a useful tool at hand, which will make it to possi-ble for critical ‘what-if’ questions to be asked in a multispecies spatial context, without the big guns who developed Multispecies VPA having to be consulted. Making this power widely available will be very valuable. I’m very happy and learnt a lot and am looking forward to the final report soon. DP: We thank you and FAO for having made this possible. We are quite proud that such an important player as FAO expressed interest in the work done at the Fisheries Centre; I would like to thank Gunna for her help with the work-shop preparations and Carl Walters, for the fascinating lectures, and for having turned the Ecopath suite into the dynamic tool it now is. Finally, I would like to thank those colleagues who came on their own, such as Jim Kitchell, to share their ideas with us.  43  Referencesc   Anon, 1995. The Kyoto Declaration and Plan of Action. International conference on the sustainable contribution of fisheries to food security. Kyoto, Japan, 4-9 Sep-tember 1995. Fisheries Agency, Japan. 21 p. [6]  Asila, A. and J. Ogari.. 1987. Growth Parameters and Mortality rates of Nile perch (Lates niloticus), estimated from length-frequency data in the Nyanza Gulf (Lake Victoria). FAO Fish. Rep. 389:1272-287. [31]  Bakun, A. 1996. Patterns in the ocean: ocean processes and marine population dy-namics. Centro de Investigaciones Biologica del Nordeste, La Paz, Mexico/ University of California Sea Grant, San Diego, USA. [27]  Bundy, A. 1997. Assessment and Management of Multispecies, Multigear Fisheries: A case study from San Miguel Bay, the Philippines. Ph.D. Thesis. University of British Columbia, Vancouver, B.C. Can-ada. 395 p. [39]  Butterworth, D.S., Cochrane, K.L. and J.A.A. de Oliviera. 1997. Management proce-dures: a better way to manage fisher-ies? The South African Experience. In: E.K. Pikitch. D.D. Huppert and M.P. Sissenwine (eds.). Global Trends: Fish-eries Management. American Fisheries Society, Symposium 20, Bethesda. [7]  Christensen, V. and D. Pauly 1992a. Ecopath II - A system for balancing steady-state ecosystem models and calculating net-work characteristics. Ecol. Modeling 61: 169-185. [10]  Christensen, V. and D. Pauly 1992b. A guide to the Ecopath II software system (version 2.1). ICLARM Software 6, 72 p. [10]  Christensen, V. and D. Pauly 1995. Fish produc-tion, catches and the carrying capacity                                                           c The page(s) where each reference is (are) given in square brackets, and thus these references may be used as an authors’ index. of the world ocean. Naga, the ICLARM Quarterly. 18(3): 34-40. [10]  Christensen, V. and D. Pauly 1996. Ecological modeling for all. Naga, the ICLARM Quarterly. 19(2):25-26. [10]  Cury, P. and C. Roy. 1989. Optimal environ-mental window and pelagic fish re-cruitment success in upwelling areas. Can. J. Fish. Aquat. Sci. 46: 670-680. [41]  Dalsgaard, J. and D. Pauly 1997. Preliminary Mass-Balance Model of Prince William Sound, Alaska for the Pre-Spill Period 1980-1989. Fisheries Centre Research Report 1997, 5(2), 34 p. [32]  FAO 1995. Code of Conduct for Responsible Fisheries. FAO, Rome. 41p. [6]  Froese, R. and D. Pauly (Editors). 1998. FishBase 98, Concepts, Design and Data Sources. ICLARM. [9]  Gayanilo, F.C., M. Soriano and D. Pauly. 1989. A draft guide to the Compleat ELEFAN. ICLARM Software 2, 65 p. [31]  Haggan, N. 1996. Integration of Traditional Environmental Knowledge p. 88-89 In: D. Pauly and V. Christensen (eds.) Mass-Balance Models of North-eastern Pacific Ecosystems. Fisheries Centre Research Reports 4(1). [26]  Haggan, N. 1998. Reinventing the Tree. p. 19-39. In: T. Pitcher, D. Pauly & P. Hart (Eds.) Reinventing Fisheries Manage-ment. Chapman & Hall. London. [26]  44Hughes, N.F. 1986. Changes in the feeding habits of Nile perch Lates niloticus (L.) in Lake Victoria, East Africa since its introduction in 1960 and its impact on the native fish community of the Nyanza Gulf. J. Fish Biol. 29: 541-548.[31]  Jarre, A., M.L. Palomares, M.L. Soriano, V.C. Sambilay, Jr. and D. Pauly. 1990. MAX-IMS: A computer program for estimat-ing the food consumption of fishes from diel stomach contents data and popula-tion parameters. ICLARM Software 4 27 p.[31]  Jarre-Teichmann, A. 1995. Seasonal mass-balance models of carbon flow in the central Baltic Sea, with emphasis on the upper trophic levels. ICES C.M. 1995 T:6.[27]  Jarre-Teichmann, A., K. Wieland, B.R. MacKenzie, H.H. Hinrichsen, M. Plikshs and E. Aro. 1997. Stock re-cruitment relationship for Central Bal-tic cod (Gadus morhua) incorporating environmental variability. ICES Sym-posium on the Recruitment Dynamics of Exploited Marine Populations: Physical-Biological Interactions, 22-24 September 1997, Baltimore, USA. [submitted to ICES Mar. Sci. Symp.] [27]  Jarre-Teichmann, A., L.J. Shannon, C.L. Molo-ney and P. Wickens. Comparing trophic flows in the Southern Benguela to other upwelling ecosystem. South African Journal of Marine Science (in press) [27]  MacKay, A. 1981. The generalized inverse. Prac-tical computing. September 1981: 108-110. [10]  Moreau, J., W. Ligtvoet and M.L.D. Palomares. 1993. Trophic relationship in the fish community of Lake Victoria, Kenya, with emphasis on the impact of Nile perch (Lates niloticus), p. 144-152. In V. Christensen and D. Pauly (eds.) Tro-phic models of aquatic ecosystems. ICLARM Conf. Proc. 26. [30,31]  Munro, G. and U.R. Sumaila. 1998. Assessing the economic benefits of artificial reefs and marine protected areas in Hong Kong waters. In T. Pitcher et al. (eds) Fisheries Centre Research Reports (in press). [19, 20]   Ogari, J. and S. Dadzie 1988. The food of Nile perch Lates niloticus, after the disappearance of the haplochromine cichlids in the Nyanza Gulf of Lake Victoria Kenya. J. Fish. Biol. 32, 571-577. [31]  Paine, R.T., J.L. Ruesink, A. Sun, E.L. Sou-lanille, M.J. Wonham, C.D.G. Harley, D.R. Brumbaugh and D.L. Secord. 1996. Trouble on Oiled Waters: lessons from the Exxon Valdez oil spill. Annu. Rev. Ecol. Sys. 27:197-235 [32]  Pauly, D. 1986. A simple method for estimating the food consumption of fish population from growth data and food conversion experiments. U.S. Fish Bull. 84: 827-840. [31]  Pauly, D. and M.L. Palomares. 1987. Shrimp consumption by fish in Kuwait water: a methodology, preliminary results and their implication for management and research. Kuwait Bull. Marine Sciences 9:101-125. [31]  Pauly, D. and V. Christensen. 1996. Mass-Balance Models of North-eastern Pa-cific Ecosystems. Fisheries Centre Re-search Reports 4(1) 131 p. [26]  Pauly, D., V. Christensen, J. Dalsgaard, R. Froese and F. Torres Jr., 1998. Fishing down marine food webs. Science 279: 860-863. [21, 26, 39]  Pitcher, T. and D. Pauly, 1998. Rebuilding eco-systems, not sustainability, is the proper goal of fisheries management. p. 311-329. In: T. Pitcher, D. Pauly & P. Hart (eds.) Reinventing Fisheries Man-agement. Chapman & Hall. London. [21, 26]   45 Pitcher, T.J. Fisheries management that aims to rebuild resources can help resolve dis-putes, reinvigorate fisheries science and encourage public support. In: Beckett  (ed.) (in press a). [21]  Pitcher, T.J. Rebuilding as a New Goal for Fish-eries Management: Reconstructing the Past to Salvage the Future. In: Eck-schmitt, K. (ed.) Managing Ecological Systems: Challenges, Successes and Failures. John Wiley, NY. (in press b). [21]  Polovina, J.J. 1984. Models of a coral reef eco-system I: the Ecopath model and its application to French Frigate Shoals. Coral Reefs 3(1): 1-11. [10]  Polovina, J.J. 1985. An approach to estimating an ecosystem box model. U.S. Fish. Bull. 83(3):457-560. [10]  Polovina, J.J. and M.D. Ow 1983. Ecopath: a user’s manual and program listings. NMFS/NOAA, Honolulu Admin. Rep. H-83-23, 46 p. [10]  Roberts, C.M. and J.P. Hawkins. 1997. How small can a marine reserve be and still be effective. Coral Reefs 16:150. [39]  Sumaila, U.R. 1997. Cooperative and non-cooperative exploitation of the Arcto-Norwegian cod stock in the Barents Sea. Environm. Res. Econ. 10:147-165. [19]  Sumaila, U.R. Protected marine reserves as fisheries management tools: A bio-economic analysis. Fish. Res. (in press). [19]  Ulanowicz, R.E. 1986. Growth and develop-ment: ecosystems phenomenology. Springer Verlag, New York, 203 p. [10]  Walters, C., V. Christensen and D. Pauly. 1997. Structuring dynamic models of ex-ploited ecosystems from trophic mass-balance assessments. Rev. Fish. Biol. Fish. 7, 139-172. 6, 11, 15, 30, 31]  Walters, C., D. Pauly, V. Christensen and J. Kitchell. 1998. Representing density dependent consequences of life history strategies in an ecosystem model. Eco-sim II. (under review by CJAFS). [11]  Walters, C., D. Pauly and V. Christensen. 1998. Ecospace: a software tool for predicting mesoscale spatial patterns in trophic relationships of exploited ecosystems, with special reference to impacts of ma-rine protected areas. To be presented at Theme Session (S) on ‘Visualization of Spatial (including Survey) Data’, of the ICES Annual Science Conference, Cas-cais, Portugal, September 1998. [6, 11, 13]   46Appendix 1   Workshop Schedule  March 25-27, 1997 Fisheries Centre, University of British Columbia __________________________________________________________________  Wednesday 25  09h30   Introduction by convener (Daniel Pauly) 09h45   FAO's expectations for the workshop (Kevern Cochrane) 10h00   COFFEE BREAK  10h30 Lecture on “Mass balance modelling and Ecopath, with emphasis on new fea-tures of Vers. 4.0”. (Villy Christensen) 11h30   On screen Ecopath demonstrations  12h00   LUNCH 14h00 Participants finalize their Ecopath 4.0 files, with particular attention to split pools (help provided by Villy Christensen, Astrid Jarre-Teichmann, Daniel Pauly and Fisheries Centre’s students). 15h30   COFFEE BREAK   Tuesday 26  09h00   Lecture on "Ecosim: opportunities and limits" (Carl Walters) 10h00   COFFEE BREAK 10h30 Lecture on "Introducing bioeconomics into the Ecopath/Ecosim framework" (Rashid Sumaila) 11h15   Basic analysis of Ecosim runs 12h00   LUNCH 14h00 Lecture on "Population embedding: scaling the strength of trophic interactions" (Carl Walters) 14h45   Ecosim exercises with different levels of predator control 15h30   COFFEE BREAK  15h45   Lecture on “Ecosim and Marine Protected Areas” (Reg Watson) 16h30   Lecture on “Spatial modeling using Ecospace” (Carl Walters)  17h15 Ecosim (and MPA) policy exploration and start of writing up participants' ob-servations.   Friday 27  09h15 Lecture on “Back to the Future: an agenda for reconstruction of past ecosys-tems” (Tony Pitcher) 10h00   Exercises with Ecosim and Ecospace 11h00   COFFE BREAK 11h30   Further exercises and report writing 12h30   LUNCH 14h00   General Discussion: Ecosim/Ecospace predictions    Rapporteur: Reg Watson 15h30   COFFEE BREAK 16h00   Workshop closure (Kevern Cochrane and Daniel Pauly)  47 Appendix 2  List of Participants   Aydin Kerim Fisheries Research Institute, University of Washington Box 357980 Seattle, WA 98195, USA E-mail: kerim@fish.washington.edu  Eny Buchary Fisheries Centre University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: eny@fisheries.com  Alida Bundy Groundfish Division Dept. of Fisheries and Oceans North West Atlantic Fisheries Centre PO Box 5667 St. John’s, NFLD, Canada E-mail: bundy@athena.nwafc.nf.ca  Villy Christensen ICLARM MCPO Box 2631 0718 Makati City, Philippines E-mail: v.christensen@cgnet.com  Lorenzo Ciannelli School of Fisheries University of Washington Box 357980 Seattle, WA 98195, USA E-mail: lorenzo@fish.washington.edu  Kevern Cochrane FAO Via delle Terme di Caracalla 100 Rome, Italy E-mail: kevern.cochrane@fao.org  Johanne Dalsgaard Fisheries Centre University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: johanne@fisheries.com     Felimon Gayanilo ICLARM  MCPO Box 2631 0718 Makati City, Philippines E-mail: f.gayanilo@cgnet.com  Nigel Haggan Fisheries Centre University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: nhaggan@fisheries.com  Kathy Heise Dept. of Zoology University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: heise@zoology.ubc.ca  Leonardo Huato Fisheries Centre University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: huato@fisheries.com  Astrid Jarre-Teichmann Danish Institute for Fisheries Research North Sea Centre PO Box 101 DK-9850 Hirtshals, Denmark E-mail: ajt@dfu.min.dk  Jim Kitchell Centre for Limnology University of Wisconsin 680 N. Park Street Madison, WI 53706, USA E-mail: kitchell@macc.wisc.edu  Pat Livingston Alaska Fisheries Science Centre 2600 Sand Point Way N.E. Seattle, WA 98115, USA E-mail: pat.livingston@noaa.gov   48 Jean-Jaques Maguire Halieutikos Inc. 1450 Godefroy Sillery Quebec G1T 2E4, Canada E-mail: jj_maguire@compuserve.com  Jaques Moreau INP / ENSAT BP 354 F-31006 Toulouse, France E-mail: moreau@ensat.fr  Tom Okey Fisheries Centre University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: tokey@fisheries.com  Bob Olson Inter-American Tropical Tuna Comission Scripps Institution of Oceanography 8604 La Jolla Shores Drive La Jolla, CA 92037-1508, USA E-mail: rolson@iattc.ucsd.edu  Ana Parma International Pacific Halibut Commission PO Box 95009 Seattle, WA 98145-2009, USA E-mail: ana@iris.iphc.washington.edu  Daniel Pauly Fisheries Centre  University of British Columbia Vancouver, B.C. V6T1Z4, Canada E-mail: pauly@fisheries.com  Tony Pitcher Fisheries Centre University of British Columbia Vancouver, B.C. V6T 1Z4. Canada E-mail: tpitcher@fisheries.com  Rashid Sumaila Fisheries Centre, University of British Columbia Vancouver, B.C. V6T 1Z4, Canada & Chr. Michelsen Institute, Bergen E-mail: sumaila@fisheries.com  Andrew Trites Fisheries Centre Marine Mammal Research Unit University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: trites@zoology.ubc.ca  Marcelo Vasconcellos Fisheries Centre University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: marcelo@fisheries.com  Scott Wallace Resource Management and Environmental Studies #436E, 2206 East Mall University of British Columbia Vancouver, B.C. V6T 1Z3 E-mail: sscott@unixg.ubc.ca  Carl Walters Fisheries Centre University of British Columbia Vancouver, B.C. V6T 1Z4, Canada E-mail: walters@fisheries.com  Reg Watson Fisheries Western Australia PO Box 20 North Beach, Australia 6020 E-mail: rwatson@omen.com.au 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items