UBC Community, Partners, and Alumni Publications

Ecological and economic analyses of marine ecosystems in the Bird's Head Seascape, Papua, Indonesia:.. Pitcher, Tony J.; Ainsworth, Cameron H.; Bailey, Megan 2007

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
FCRR_2007_15-5.pdf [ 5.12MB ]
Metadata
JSON: 1.0074756.json
JSON-LD: 1.0074756+ld.json
RDF/XML (Pretty): 1.0074756.xml
RDF/JSON: 1.0074756+rdf.json
Turtle: 1.0074756+rdf-turtle.txt
N-Triples: 1.0074756+rdf-ntriples.txt
Original Record: 1.0074756 +original-record.json
Full Text
1.0074756.txt
Citation
1.0074756.ris

Full Text

                                                                                                                                                ISSN 1198-6727  Ecological And Economic Analyses Of  Marine  Ecosystems In The Bird’s Head  Seascape, Papua, Indonesia: I  Fisheries  Centre  Research  Reports 2007   Volume  15   Number  5    Fisheries Centre Research Reports      2007 Volume  15  Number   5        Ecological And Economic Analyses Of Marine Ecosystems In The Bird’s Head Seascape, Papua, Indonesia: I         Fisheries Centre, University of British Columbia, Canada ISSN 1198-6727   ECOLOGICAL AND ECONOMIC ANALYSES OF  MARINE  ECOSYSTEMS IN THE BIRD’S HEAD SEASCAPE,  PAPUA, INDONESIA:  I     Edited by Tony J. Pitcher, Cameron H. Ainsworth and Megan Bailey                Fisheries Centre Research Reports 15(5)  184  pages © published 2007  by   The Fisheries Centre,  University of British Columbia, 2202 Main  Mall Vancouver, B.C., Canada, V6T 1Z4     ISSN 1198-6727  F I S H E R I E S  C E N T R E  R E S E A R C H  R E P O R T S  V O L U M E  1 5  N U M B E R  5  2 0 0 7   ECOLOGICAL AND ECONOMIC ANALYSES OF MARINE ECOSYSTEMS  IN THE BIRD’S HEAD SEASCAPE, PAPUA, INDONESIA:  I  Edited by  Tony J. Pitcher, Cameron H. Ainsworth and Megan Bailey    CONTENTS   Page Director’s Foreword.................................................................................................................................... 4 Executive Summary ....................................................................................................................................5 Ecosystem Simulation Models For The Bird’s Head Seascape, Papua, Fitted to Field Data  Cameron Ainsworth, Divya Varkey and Tony Pitcher Abstract ...................................................................................................................................................... 6 Introduction................................................................................................................................................ 6  Ecopath with Ecosim............................................................................................................................7  Raja Ampat Islands ............................................................................................................................. 8  Project synthesis.................................................................................................................................. 9  First field trip .....................................................................................................................................10  Ecopath parameterization   Raja Ampat model ...................................................................................................................... 11   Kofiau model............................................................................................................................... 11   Dampier Strait model .................................................................................................................13 Methods   Ecopath (mass balance) .....................................................................................................................13  Ecosim (dynamic simulations) .......................................................................................................... 15  Predator-prey vulnerabilities............................................................................................................. 15  Mediation factors ............................................................................................................................... 15  Ecospace (dynamic spatial simulations)............................................................................................ 17  Ecopath parameterization   Functional group designations ................................................................................................... 17   Fish groups..................................................................................................................................18   Bioeroders ...................................................................................................................................19  Basic parameterization ......................................................................................................................21   Growth parameters .....................................................................................................................21   Estimating consumption rate (Q/B).......................................................................................... 24   Estimating natural mortality (M) for fish ................................................................................. 25   Daily ration ................................................................................................................................ 25   Ingestion rate in deposit feeders ............................................................................................... 25   Estimating P/B of invertebrates ................................................................................................ 26   Group maturity parameters....................................................................................................... 26   Biomass density estimates......................................................................................................... 26   Diet algorithm.............................................................................................................................27   Fisheries ......................................................................................................................................31   Functional group descriptions....................................................................................................37  The 1990 Raja Ampat model   Group biomasses........................................................................................................................ 68   Fisheries ..................................................................................................................................... 69   Fitting to time series .................................................................................................................. 69   Equilibrium analysis.................................................................................................................. 70   Challenges to Ecosim..................................................................................................................72  Ecospace parameterization   Raja Ampat 2006 Ecospace model.............................................................................................76   Kofiau Island model................................................................................................................... 78   Dampier Strait model ................................................................................................................ 80  Fishing policy optimizations............................................................................................................. 80 Bird’s Head Seascape Analyses, Page 3  Results   Time series fitting .............................................................................................................................. 81  Predicted climate anomaly ................................................................................................................82  Equilibrium analysis..........................................................................................................................84  Challenges to Ecosim.........................................................................................................................84  Fishing policy optimizations .............................................................................................................85 Discussion  Fitting the model ...............................................................................................................................89  Fishing policy optimizations .............................................................................................................90  Fisher interview forms.......................................................................................................................90  Stomach content analysis .................................................................................................................. 91 Conclusions ............................................................................................................................................... 91 References .................................................................................................................................................92 Appendix A: EwE parameterization  A1.  Species level data...................................................................................................................... 103  A2.  Fish family data ....................................................................................................................... 129  A3. Ecopath parameters: 2006 RA model ..................................................................................... 133  A4.  Ecopath parameters: 1990 RA model......................................................................................161  A5. Ecosim parameters: 1990-2006 RA model ............................................................................. 163  A6. Time series data........................................................................................................................ 164 Appendix B: EwE results  B1. Ecopath results .......................................................................................................................... 166  B2. Ecosim results ........................................................................................................................... 167  The migrant anchovy fishery in Kabui Bay, Raja Ampat, Indonesia:  catch, profitability and income distribution  Megan Bailey, Christovel Rotinsulu, and U. Rashid Sumaila Abstract ...................................................................................................................................................175 Introduction .............................................................................................................................................175 Area description ...................................................................................................................................... 176 The migrant anchovy fishery ...................................................................................................................177 Interviews................................................................................................................................................ 178 Computations.......................................................................................................................................... 179  Catch ................................................................................................................................................ 179  Revenue............................................................................................................................................ 179  Cost .................................................................................................................................................. 180 Profit and gains from the fishery ............................................................................................................ 180 Results   Catch .................................................................................................................................................181  Revenue.............................................................................................................................................181  Costs................................................................................................................................................. 182  Profit and gains from the fishery..................................................................................................... 182 Conclusion............................................................................................................................................... 183  Suggested Citation: Pitcher, T.J., Ainsworth C.H. and Bailey, M. (eds) (2007) Ecological and Economic Analyses of The Bird’s Head Seascape, Papua, Indonesia: I. Fisheries Centre Research Reports 15(5): 184 pp.  A Research Report from Fisheries Ecosystem Restoration Research & th e Fisheries Economics Research Unit Fisheries Centre Research Reports 15(5) 184 pages © Fisheries Centre, Unive rsity of British Columbia, 2007  FISHERIES CENTRE RESEARCH REPORTS ARE ABSTRACTED IN THE FAO AQUATIC SCIENCES AND FISHERIES ABSTRACTS (ASFA) ISSN 1198-6727 Page 4, Fisheries Centre Research Reports 15(5), 2007 DIRECTORS FOREWORD   This Report presents two contributions of very unequal length on the fisheries of Raja Ampat, in Eastern Indonesia.  The second of these is devoted to a neat account of the economics of an anchovy fishery which developed without being monitored by official statistics, as probably most small-scale fisheries do throughout the world. It is also, apparently, a profitable fishery, and this again, raises questions about the usual neglect of small-scale fisheries. It is, however, the first of these contributions which I want to elaborate on, as it connects very deeply to my personal trajectory.  In 1975 and 1976, I worked in western Indonesia, with a freshly-minted Master of Fisheries, in an ‘aid’ project devoted to the development of trawl fisheries of the Java Sea and adjacent areas.  I did not know then much about fisheries in general, and tropical fisheries in particular, but I realized, upon seeing my very first multi-species trawl haul wiggling on deck that it would be impossible to estimate, using ‘classical methods’ (i.e., those I had been taught), the parameters of growth, natural mortality, etc., required for the (single-species) models that were then in vogue for the management of fisheries.  This realization was the start of my personal research program, devoted to identifying pattern in the growth and mortality parameters across a number of species, which could be used to infer their likely value in the absence of local data, and of methods for their estimation, given a minimum of such data. This program, which coincided with that of many fisheries scientists at the time (including T.J. Pitcher, one of the authors of the contribution commented upon here) was rather successful, as reflected in this very Report. Raja Ampat, in Eastern Indonesia is near the centre of the world’s marine biodiversity, but it is, by any other standard, an extremely peripheral area, notably as science goes. It could be inferred, therefore, that, as the phrase goes, “nothing is known on [whatever] in the area”. But this is not so.  An amazing amount of data is available on virtually all areas of the world, including areas as ‘remote’ as Raja Ampat, even if we go back as far as the 17th century.  The point is to know where to find these seemingly dead data, and to make them alive again. One way to do this is through the compilation and analysis of observations by the naturalists of successive historic expeditions, and the narrative of travellers, as  illustrated in an earlier report on the same area by ‘Deng’ Palomares and ‘Sheila’ Heymans*.  The other approach to overcoming the dictum that “nothing is known…”, documented in this Report, is to combine locally available, but scattered data (which always exits) with general patterns on the distribution, feeding and production of fish and invertebrates, derived from databases such as Fishbase. Reading this document, I have a sense that the work we did the last thirty years actually was useful: we now have the tools to build realistic models, and to propose practical schemes for the management of about any marine ecosystem in the world. Not bad. These tools work only when in good hands, but clearly, this is here the case. I conclude, therefore, by congratulating the authors for a job well done.  Daniel Pauly Director, Fisheries Centre                                                              *  Maria Lourdes D. Palomares and Johanna J. Heymans 2006. Historical Ecology of the Raja Ampat Archipelago, Papua Province, Indonesia. Fisheries Centre Research Reports 14(7): 64 pp.    Bird’s Head Seascape Analyses, Page 5 EXECUTIVE SUMMARY    A growing awareness of the decline in ecosystem health and the depletion of resources world-wide has led researchers to explore the use of ecosystem-based management (EBM), an approach that integrates ecological, social, and economic goals, and explicitly recognizes humans as key components within an ecosystem. EBM is still in its infancy, and a number of research projects have been launched to try to increase understanding, develop EBM tools, and attempt to mitigate or even reverse at least the worst of the present trends.   One such study is within the Coral Triangle, spanning eastern Indonesia, parts of Malaysia, the Philippines and Papua New Guinea, where the highest coral reef biodiversity on earth has been measured. At the heart of the Coral Triangle is the Bird’s Head Seascape, off the west coast of Papua Province, Indonesia.  This is still a relatively remote and pristine area, home to about 75% of the world’s reef-building coral species, and over 1,000 fish species. The Raja Ampat archipelago, where Alfred Wallace made a home in the 1830s, has attracted the interests of conservation groups and scientists, and has been selected as one of the top conservation priorities in the world. The high level of biodiversity has lead to a growing marine tourism sector, and the newly decentralized government is trying to develop the area sustainably for the 31,000 inhabitants. Following a proposal, funding was generously provided by the David and Lucile Packard Foundation to researchers and scientists from Conservation International (CI), The Nature Conservancy (TRC), the World Wildlife Fund (WWF), the State University of Papua (UNIPA), and the Fisheries Centre, University of British Columbia, for a project entitled, “Toward Ecosystem-Based Management in the Bird’s Head Functional Seascape of Papua, Indonesia”. Three teams at the Fisheries Centre are working to provide a synthesis of key ecological, economic and historical components, supporting field teams from UNIPA, TNC, CI and WWF who are sampling and collecting data. This report represents the second of UBC’s contributions to the Bird’s Head Seascape EBM project†.   The first paper in this report, from the Fisheries Ecosystems Restoration Research group, describes the developement of a 98-functional group ecosystem simulation model (Ecopath with Ecosim, EwE) for Raja Ampat, fitted to local time series abundance and CPUE data, and driven by local climate changes. Fine- scaled local models for three areas (Kofiau Island, Misool Island, and the Dampier Strait) are also included. Several novel approaches have been added to this model, including a new algorithm for estimating diets based on fish gape size, body depth and habitat co-occupation. These EwE models will be further refined with data from the field teams on diets and fisheries, and with the results of interviews with fishers on percieved changes in the Raja Ampat ecosystem. The aim is to use the models to develop optimal management scenarios in order to provide EBM advice that can be appraised by stakeholders in the Raja Ampat archipelago.  Illegal, unreported, and unregulated (IUU) fishing is now widely recognized as undermining management goals. The second paper in this report, from the Fisheries Economics Research Unit, uses field observations to estimate the unreported and unregulated catch of anchovies in Kabui Bay, Raja Ampat. The estimates include uncertainty, revenues, costs and the apparent profitability of the fishery. Results suggest that fisheries managers in Raja Ampat could consider capturing some of the fishery rent. The UBC team hopes to provide an IUU estimate for all of Raja Ampat upon the completion of the project, and the anchovy estimate will contribute to this.         Tony J. Pitcher, Cameron H. Ainsworth and Megan Bailey Vancouver, May 2007                                                         † The first contribution was: Palomares, M.L.D. and Heymans, J.J. (2006) Historical Ecology of the Raja Ampat Archipelago, Papua Province, Indonesia. Fisheries Centre Research Reports 14(7): 64 pp.  Page 6, Fisheries Centre Research Reports 15(5), 2007 ECOSYSTEM SIMULATION MODELS FOR THE BIRDS HEAD SEASCAPE, PAPUA, FITTED TO DATA   Cameron Ainsworth, Divya Varkey and Tony Pitcher Fisheries Ecosystems Restoration Research, Fisheries Centre,  University of British Columbia,‡ 2202 Main Mall, Vancouver, BC, Canada, V6T 1Z4  Abstract  Ecopath with Ecosim models are described for the marine ecosystem of the Raja Ampat (RA) archipelago in Papua province, eastern Indonesia.  The models are based on literature and output data emerging from the Birds Head Seascape Ecosystem Based Management (BHS EBM) project, a joint Packard-funded initiative between TNC, CI, WWF and UBC.  A new diet allocation algorithm is developed for use in tropical ecosystems, based on predator gape and prey body size.   The algorithm predicts feeding relationships in order to make better use of FishBase diet information.  Time series of catch, effort and catch-per-unit-effort are developed from governmental fisheries statistics assembled in the field.  A historic model, representing 1990 AD is developed based on this time series information, and a 16-year dynamic Ecosim simulation is fitted to agree with time series.  The model incorporates four mediation functions to capture key non-trophic interactions important in reef ecosystems.  A primary production anomaly is developed that would help explain the difference between observed and predicted biomass dynamics from 1990 to 2006.  The anomaly shows a non- significant negative correlation with sea surface temperature.  An equilibrium analysis and various challenges to Ecosim are used to test the behaviour of the model.  Policy optimizations are conducted to sketch the potential trade-off frontier between economic and ecological harvest benefits available in RA.  Ecospace maps are designed for RA and sub-area models of Kofiau Island and Dampier Strait.  Some comments are made regarding future developments of the UBC spatial modelling component of the BHS EBM project.      INTRODUCTION  This report presents the methodology used to create Ecopath with Ecosim (EwE) and Ecospace ecosystem models of the Raja Ampat Islands in Papua, Indonesia, and provides a preliminary analysis of ecosystem functioning and resource potential of coral reefs.  The models created here are based on scientific data emerging from the research project “Towards Ecosystem-Based Management in the Bird’s Head Functional Seascape of Papua, Indonesia”, being conducted jointly by The Nature Conservancy (TNC), Conservation International (CI), World Wildlife Fund (WWF), and the University of British Columbia (UBC).    Ecosystem models are being developed for the RA region at various spatial scales.  These temporal and spatial dynamic models capture biotic and abiotic interactions in the ecosystem.   By accurately representing ecological processes on coral reefs, they will help us to improve our understanding of reef ecosystem behaviour.  The models can be used to assist ecosystem-based marine policy; with them we can design sustainable fishing strategies that maximize economic benefits while protecting coral reef communities.  Fishing policies developed with these tools can be made robust against various future climate scenarios, and the risk and uncertainty surrounding harvest recommendations can be evaluated and quantified.  Importantly, the spatial models are able to forecast the effects and benefits of spatial management schemes, such as the                                                         Cite as:‡ Ainsworth, C.H., Varkey, D. and Pitcher, T.J.  (2007) Ecosystem simulation models for the Bird’s Head Seascape, Papua, fitted to data. Pages 6–172 in Pitcher, T.J., Ainsworth C.H. and Bailey, M. (eds) Ecological and Economic Analyses of Marine Ecosystems in the Birds Head Seascape, Papua, Indonesia: I.  Fisheries Centre Research Reports 15(5): 184 pp.  Bird’s Head Seascape Analyses, Page 7 application of marine protected areas (MPAs).  The models are built within a flexible framework that can be continually modified and improved as new data becomes available.  The work presented here should provide a starting point for further study of ecosystem-based management (EBM) strategies helpful to the management of the Bird’s Head Functional Seascape (BHS).  Ecopath with Ecosim (EwE)  We have used the family of modelling tools, Ecopath with Ecosim (EwE) and Ecospace to represent the food web of Raja Ampat and simulate trophic interactions of interest to fisheries and conservation.  Invented by Polovina (1984) and advanced by Christensen and Pauly (1992, 1993), Walters et al. (1997, 1998) and Christensen and Walters (2004a) among others, EwE is a mass-balance trophic simulator that acts as a thermodynamic accounting system.  Summarizing all ecosystem components into a small number of functional groups (i.e., species aggregated by trophic similarity), the box model describes the flux of matter and energy in and out of each group, and can represent human influence through fishery removals and other ways.  There are now dozens of published articles that use EwE to describe ecosystems, test hypotheses and demonstrate innovative applications useful for EBM (see review in Christensen and Walters, 2005).  EwE has been used in actual fisheries management, but to a limited extent.  Reviews and criticisms of the EwE approach are provided by Fulton et al. (2003), Christensen and Walters (2004a), and Plagányi and Butterworth (2004).  An EwE model is presented here for the marine ecosystem of Raja Ampat (RA) as it appeared in 2006 AD.  The model utilizes BHS EBM project information and data from literature sources.  New methodologies are developed to make the best use of FishBase data.  For example, a new diet allocation algorithm determines likely prey items based on predator gape-size and processes FishBase diet data to the level of functional groups.    An Ecopath model of RA representing the system in 1990 AD is created based on the 2006 model.  Relative functional group biomass and catch is estimated for these years based on Indonesian governmental statistics, and ecosystem dynamics are tuned to agree with the historic trends from the years 1990-2006.  The data fitting process attempts to capture ecosystem responses to fishing and climate that occurred over the last 16 years.  The dynamic Ecosim model utilizes advanced features such as mediation functions, which capture critical animal behaviours and allow us to represent important non-trophic relationships present in the coral reef environment.  A primary production time series anomaly is determined that may explain the discrepancy between the observed and predicted catch and biomass trends.  The anomaly is compared to various environmental indices.  Insight gained can potentially improve our understanding of regional climate and its affect on marine production.    A comprehensive review of model behaviour is performed using the equilibrium analysis facility in Ecosim.  This routine describes the exploitation status of commercial functional groups, allowing us to judge the accuracy of the baseline model condition against our knowledge of the ecology and fishing history of the region.  The analysis generates catch and biomass curves equivalent to those used by classical fisheries methods.  Diagnostic challenges are presented to the model to test its performance - extreme combinations of fishing practices that reveal the behaviour and stability of the model.  By presenting these early diagnostic outputs in this report, we hope to the draw attention of local experts and enlist their help to establishing realistic model behaviours.  Basic policy optimization is conducted using the tuned Ecosim model, and the socioeconomic/ecological tradeoff frontier is mapped to reveal the sustainable production potential of the ecosystem.  This application should demonstrate the power of policy optimizations in EwE, although the specific values and estimates of resource potential will  Page 8, Fisheries Centre Research Reports 15(5), 2007 continue to change as our understanding of the RA ecosystem improves.  Initial efforts to produce spatially explicit models are described here.  Habitat maps, based on data collected in the BHS EBM project, provide a foundation for the Ecospace models of RA and two smaller-scale models: Kofiau Island and Dampier Strait.  Basic parameterization of the Ecospace models is described.  Finally, we discuss our goals and the current direction of the spatial modelling component for the BHS-EBM project.    Raja Ampat Islands  The RA archipelago extends over 45,000 km2 and consists of approximately 610 islands including the ‘four kings’, Batanta, Misool, Salawati and Waigeo (COREMAP, 2005).  Erdmann and Pet (2002) provide a summary of the major oceanographic features occurring in the Raja Ampat archipelago.  The area encompasses a variety of marine habitats, including some of the most biodiverse coral reef areas on Earth (Donnelly et al., 2003; McKenna et al., 2002a).  It is estimated that RA possesses over 75 percent of the world’s known coral species (Halim and Mous, 2006).    Fisheries by native peoples of Papua have likely persisted for centuries; although there is evidence that long-term ‘chronic’ exploitation of coral reefs had an early impact on reef health in many places throughout the world (Pandolfi et al., 2003).  However, record keeping typically begins long after the major depletion of reef resources occurs (Bellwood et al., 2004).  This is the case in Indonesia and many other countries that manage coral reef resources.  The gradual or early declines may therefore go unnoticed thanks to the shifting-baseline syndrome (Pauly, 1995) in which each generation of scientists and resource users accept a lower standard of abundance as normal. It is therefore difficult to estimate the loss of potential productivity that has occurred, especially since there are few pristine areas remaining with which to form a baseline comparison.  The use of ecosystem models does allow us to predict unexploited biomass levels for critical species, but accurate predictions depend on the quality of the models, and the models can only be vetted against time-series catch and abundance data.  Unfortunately, quantitative data is limited in RA, and much of the knowledge we have about ecosystem changes comes in the form of local ecological knowledge (LEK) from scientists and inhabitants.   Currently, the main marine commodities in the RA archipelago include skipjack tuna (Katsuwonus pelamis), yellowfin tuna (Thunnus albacares) and Spanish mackerel (Scomberomorus commerson), but significant artisanal fisheries also exist for reef-associated fish and invertebrates.  Indonesia is known to have suffered a rapid depletion in recent decades of near-shore fish stocks and coral reef animals, especially sharks, tunas and reef-associated fish (Tomascik et al., 1997).  The pressures on the reef systems in Eastern Indonesia can only be expected to increase as the human population grows.  Overfishing has reduced the average life span of some marine species (Myers and Worm, 2003).  Consequently, marine ecosystems may be increasingly unstable and responsive to environmental fluctuations (Hughes et al., 2005).  The effect is likely to be pronounced in coral reef environments, where large and influential predators and herbivores are targeted (Hughes et al., 2003).  Increased system volatility could potentially be a long-lasting effect if the constant antagonism of fisheries asserts an evolutionary pressure towards early maturation and high turnover rates in exploited species.     Challenges to management of coral reefs now centre on the serious issues of overexploitation (Pandolfi et al., 2003), land-based pollution (McCullock et al., 2003), disease outbreaks (Kaczmarsky et al., 2005) and outbreaks of coralivores such as the crown of thorns starfish (Acanthaster planci) - a source of mass mortality in corals (Chesher, 1969).  Loss of coral cover from these stressors has far reaching impacts throughout the food web, and may result in a long- term loss of fish biodiversity (Wilson et al., 2006).   Bird’s Head Seascape Analyses, Page 9 Project Synthesis  The BHS EBM project contains 17 major scientific components.  The wide diversity of information resulting from these projects can readily be incorporated into various aspects of Ecopath, Ecosim and Ecospace; and novel methodologies are under development that will allow us to use, for the first time in the EwE, the sort of highly resolved biogeographic information emerging from BHS EBM studies.  Generally, data collected for the RA region will help to make the EwE models, presented here in a preliminary and generic form, more relevant to the local context.  The resulting suite of EBM tools should be able to test specific ecological and socioeconomic hypotheses relevant to the management of coral reef ecosystems in RA.  Importantly, the unique opportunity provided by this project, to collate and integrate data resulting from multidisciplinary studies, will strengthen our understanding of coral reef ecology, improve EBM tools, and increase scientific dividends resulting from the BHS EBM project.  TNC reef health monitoring ongoing at Kofiau, Boo and Misool Islands among other sites (see Mous and Muljadi, 2005) is intended to provide coral cover and biomass data for important species of herbivorous fish and large piscivorous fish.  Results from this analysis were not available in time for this report; the final reef monitoring report for Kofiau Island is expected in December 2006 (P. Mous.  TNC-CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia,  pers. comm.).  The biomass estimates we are currently using for herbivorous fish and large piscivorous fish species are based on reef transect data from sites near Weigeo Island (COREMAP, 2005).  Unless the updated biomass information is very different from current estimates, integrating this new data should be straight forward, and should not require extensive reworking of the models.  It will, however, allow us to produce more accurate site-specific versions of the model representing the various field sites.  Depending on the precision and extent of the data, it may also help up in developing scenarios for the novel EwE sub-routine now under development, Ecolocator (contact: C. Ainsworth, UBC Fisheries Centre.  2202 Main Mall, Vancouver BC. Canada.  Email: c.ainsworth@fisheries.ubc.ca).  The CI seascape connectivity analysis may provide us with an independent check of Ecospace dispersal parameters and advection patterns.  In EwE, dispersion represents the tendency of populations to shift or expand their occupied range.  It is not necessarily related to swimming ability or speed of movement, but more closely reflects the fidelity of individuals to their natal habitats, or the ability of planktonic propagules to travel and settle new areas.  Populations that display genetic homogeneity across the study area may therefore be assumed to have higher rates of dispersion, while heterogeneous populations may reveal the action of isolating biogeographic effects.  At the time of this report, connectivity data is forthcoming.  Interviews conducted in the Seascape reproduction study, which identifies and monitors spawning aggregation sites (SPAGS), may provide critical habitat data for Ecospace that will allow us to accurately represent source-sink dynamics of major commercial fish populations.  This information may influence our expectations of the ecological and economic merit of spatial management schemes (Sanchirico et al., 2006).  Adding to this output, oceanographic data resulting from the SPAG vial release program may help us to track advection currents and predict areas of larval settlement.  This may prove to be an important factor, both in determining sustainable exploitation levels, and in the siting of marine protected areas, as mortality during settlement may be a bottleneck for some reef fish species (Doherty et al., 2005; Hughes et al., 2005).  At the time of this report, the SPAGS studies had failed to confirm the existence of any large spawning aggregations sites on Kofiau Island; but further studies are planned for SE Misool.  The WWF Seascape migration and dispersal analysis for turtles is expected to provide habitat and movement information for Ecospace (contact: L. Pet-Soede. Jl. Raya Puputan No. 488, Renon Denpasar, Bali, Indonesia).  The TNC fish stomach content analysis study will provide  Page 10, Fisheries Centre Research Reports 15(5), 2007 valuable diet information directly usable by Ecopath, and should supplement (or render obsolete) the current diet allocation used to parameterize the EwE food web (contact: P. Mous, TNC-CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia).  The CI socioeconomic study will provide cost and price information essential to the socioeconomic optimization facilities in Ecosim (contact: A. Dohar, CI. Jl.Gunung Arfak.45.Sorong, Papua, Indonesia).  The historical ecology study can provide us with more accurate model baselines with which to parameterize fisheries indicators in Ecosim (contact: S. Heymans, UBC Fisheries Centre.  2002 Main Mall, Vancouver, Canada).  The TNC marine resource utilization survey has generated aerial photographs of RA, providing useful habitat data and allowing us to estimate fishing effort distribution; this will be useful for validating Ecospace (contact: P. Mous, TNC-CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia).  Analyses using MARXAN to assess the conservation potential of protected areas will guide the Ecospace research and provide candidate closure scenarios for socioeconomic evaluation (contact: M. Barmawi, TNC-CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia).  Results from the historical ecology study and aerial photography are currently being assessed for integration into the models; results from other project components are forthcoming.  First field trip  The UBC synthesis model Post-Doctoral Fellow, Cameron Ainsworth, traveled to Papua and Bali in Feb. 20 – Apr. 19, 2006.  The purpose of the trip was to meet key personnel involved with the BHS EBM project, collect data from Indonesian repositories, gather preliminary information and literature which had been assembled by project partners, and collect early data resulting from the project.  The first week was spent in Bali liaising with researchers from TNC, CI and WWF.  Meetings were held with senior project staff from the partner organizations in which we planned the strategic direction of the spatial modelling effort and discussed ways to incorporate project information into the models.  We agreed on the general outputs that are expected from the modelling, and determined what outputs might assist the regional management of RA marine fisheries. Contributing to those meetings were Peter Mous (TNC), Lida Pet-Soede (WWF), Jos Pet, Muhammad Barmawi and Abdul Halim (TNC). Cameron Ainsworthwas briefed on the state of major research projects, and collected preliminary GIS data that had been collated, and interview materials from the perception monitoring study.  Traveling to Sorong, Papua provided the opportunity to discuss specific model requirements with experts knowledgeable in the ecology and fisheries of RA.  Functional group structure and fleet design were discussed at length.  By representing the most critical functional elements in the ecosystem, we hoped to provide a suitable basis for the models that was capable of capturing important processes.  The basic structure of the model was designed so that it could provide outputs that would be relevant to the management process and hold resonance with managers, policy makers and the public.    The extensive field experience of TNC and CI scientists, divers and research staff was invaluable to model design.  Particularly, their knowledge of coral reef animals and their habits, biogeographic and oceanographic features laid the foundation for the Ecopath and Ecospace models especially.  Although many researchers contributed to the early design of the models, Peter Mous, Andreas Muljadi and Obed Lense provided particularly valuable assistance in designing the functional group structure and fisheries.  We also acknowledge the contributions of Chris Rotinsulu, Reinhard Poat, Anton Suebu, Adityo Setiawan and other researchers in TNC and CI Sorong offices.  Throughout this report, specific contributions are acknowledged as personal communications.  In Sorong we visited the offices of the Sorong Regency Fisheries Office (Departemen Kelautan dan Perikanan, DKP), the Raja Ampat Regency Fisheries Office, the Trade and Industry Office (Departemen Perinustrian dan Perdagangan) and the Agricultural Quarantine Office (Badan Karantina Pertanian).  Cameron Ainsworthalso had the opportunity to talk with student  Bird’s Head Seascape Analyses, Page 11 researchers from the State University of Papua (UNIPA), who were in the process of collecting information for the socioeconomic evaluation study (CI).  A week spent in Deer Village on Kofiau Island allowed familiarization with the artisanal fishing methods, to witness fishing operations and to interact with residents.  Penny Goodwyn, a student researcher from the University of Canberra provided valuable translation assistance. Opportunity to snorkel and SCUBA dive was provided, and we also released spawning aggregation (SPAG) tracking vials at a suspected grouper aggregation site (Gebe Island) to study local currents and larval settlement patterns.  Some data were collected from the marine use monitoring study.     Returning to Bali, researchers from UBC, TNC, CI, and WWF participated in a modelling coordination workshop, April 10-14 in Sanur.  The model format was presented for review: structure, data sources and preliminary parameters were vetted.  The local knowledge and scientific experience of Mark Erdmann helped set the direction of the UBC modelling study.  Presentations by the UBC modelling study and socioeconomic study, as well as TNC, CI and WWF staff helped to coordinate team members.  Ecopath parameterization   Raja Ampat model  The Raja Ampat model describes the region from 129o 12' E and 0o 12' N to 131o 30' E and 2o 42' S (Fig. 1.1). This large-scale model includes all the waters of Raja Ampat.  The functional groups represent reef-associated fish identified by McKenna et al., (2002b), as well as pelagic and deepwater fish occurring in Eastern Indonesia.  In order to be included in the model, a fish species had to be listed both under the ‘Indonesia’ country code in FishBase (FishBase country code 360) and the ‘Papua New Guinea’ code (FishBase country code 598).  That information is found on the “DemersPelag” (habitat) field of the “Species” table in the FishBase database.    Kofiau model  The Kofiau Island Ecospace model extends from 129o 14' E and 1o 5' S in the north-west corner to 130o 1' E and 1o 20' S in the south east corner (Fig. 1.2).  Kofiau was selected as a study area for a small-scale model based on a number of advantages.  Firstly, it is the most well developed TNC RA field site in the BHS EBM project.  The permanent field office in Deer Village is staffed throughout the year, and there are many marine experts on site and in Sorong that have extensive knowledge of its ecology and biogeography.  Secondly, the process of data gathering is furthest along at this location.  At the time of this report, reef fish abundance counts have been made in transect studies; however, the data is not yet available.  Community interviews for the resource use assessment are underway, the SPAG vial release program has so far only been conducted at Kofiau, and MPA site selection using MARXAN is also furthest advanced for this area (P. Mous, TNC-CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia).  Thirdly, this site provides an excellent example of the valuable reef habitat that is associated with RA, and is responsible for the biodiversity and beauty of the coral triangle.  For example, Wambong Bay on Kofiau Island has the highest number of fish species ever recorded from a single site (208) (Allen, 2000).  The small-scale model representing Kofiau Island is primarily a coral and reef-fish model that has been expanded to include important pelagic elements.  Reef fish species in the Kofiau Island model are based on the 940 species identified by McKenna et al., (2002b) to species level.  The species list for the Kofiau model was expanded to include key pelagic species occurring around Kofiau Island such as tunas (Scombridae), sardines and herrings (Clupeidae), wolf-herring (Chirocentridae), anchovies (Engraulidae), flying fish (Exocoetidae) based on expert communications (Andreas Muljadi, Obed Lense, Reinhart Poat, Adityo Setiawan.  TNC-CTC.  Jl  Page 12, Fisheries Centre Research Reports 15(5), 2007 Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413; Chris Rotinsulu.  CI.  Jl Arfak No. 45.  Sorong, Papua, Indonesia 98413, pers. comm.).  Figure 1.1 - Area represented by Raja Ampat (RA) model.  RA model is delimited at 129o 12' E and 0o 12' N at the northwest corner and 131o 30' E and 2o 42' S at the southeast corner.  From north to south, inset rectangles show areas described by Dampier Strait, Kofiau Island and SE Misool Island models.  Figure 1.2 - Area described by Kofiau Island model.  Kofiau Island model is delimited at 129o 14' E and 1o 5' S at the northwest corner and 130o 1' E and 1o 20' S at the southeast corner.   Bird’s Head Seascape Analyses, Page 13 The pelagic species list also includes species mentioned in Venema (1997).  Individual parameters were set for each of the 940 reef fish species in the model.  Fish families were then divided into functional groups.  The fishing fleet in the model represents near shore and artisanal gear types; the foreign fishing fleet, which operates in RA, is excluded.  Dampier Strait model  The Dampier Strait model extends from 130o 25' 12'' E and 0o 18' S at the northwest corner to 131o 21' 36'' E and 0o 50' S at the southeast corner.  The model includes Waisai Bay in the northwest and incorporates a large extent of the southern coast of Weigeo Island, including Gam Island and Kabui Bay (Fig. 1.3).  Mayalibit Bay, a shallow enclosed body of turbid water occupying a south-central position of Weigeo Island, was excluded from the model as it likely ecologically distinct from the deeper and faster flowing Dampier Strait (Mark Erdmann.  CI.  Jl. Dr. Muwardi. 17 Renon Denpasar, Bali, Indonesia, pers. comm.).  The modelled area is bounded by the convoluted shore of Batanta Island in the south.  Dampier Strait is an important and productive area in Raja Ampat that sustains a major artisanal fishery for anchovy due to a region of strong upwelling.   Figure 1.3 - Area described by Dampier Stra est co it model.  Dampier Strait model is delimited at rner and 131o 21' 36'' E and 0o 50' S at the , 1992) operates under two main assumptions.  he first assumption is that biological production within a functional group equals the su  rtality caused by fisheries and predators, net migration, biomass accumulation and 130o 25' 12'' E and 0o 18' S at the northw southeast corner.   METHODS  Ecopath (Mass Balance)  copath (Polovina, 1984; Christensen and PaulyE T m of other m u o nexplained mortality.  Eq. 1.1 expresses this relationship:  ( ) ( ) ( ) ( ) = −⋅ j ii EEBP 1 1  ( 1.1) ∑ +++⋅⋅+=⋅ n iiiijjjiii BBAEDCBQBYBPB  Page 14, Fisheries Centre Research Reports 15(5), 2007 where Bi and Bj are biomasses of prey (i) and predator (j), respectively; P/Bi is the production/biomass ratio;  Yi is the total fishery catch rate of group (i); Q/Bj is the consumption/biomass ratio; DCij is the fraction of prey (i) in the average diet of predator (j); Ei is the net migration rate (emigration – immigration); and BAi is the biomass accumulation rate for group (i). EEi is the ecotrophic efficiency; the fraction of group mortality explained in the model;  within a group equals the sum of production, espiration and unassimilated food, as in eq. 1.2.  The second assumption is that consumption r  ( ) ( ) ( ) ( ) ( ) GSBQBPTMQGSBPBBQB ⋅+⋅−−⋅−+⋅=⋅ 11/  (1.2)  Where GS is the proportion of food unassimilated; and TM is the trophic mode expressing the egree of heterotrophy; 0 and 1 represent autotrophs and heterotrophs, respectively.  Intermediate values represent facultative consumers. Ecopath uses a set of algorithms (Mackay, 1981) to simultaneously solve n linear equations of th rm in eq. 1.1, where n is the number of functional groups.  Under the assumption of mass- lect their inputs.   uncertain ecosystem ur knowledge of well-understood groups.  It places piecemeal  of data, and it offers g scientists a forum to summarize what is known about the ecosystem cosim (Walters et al., 1997) adds temporal dynamics.  It accounts for the biomass flux between d  e fo balance, Ecopath can estimate missing parameters.  This allows modelers to se Ecopath uses the constraint of mass-balance to infer qualities of components based on o information on a framework that allows us to analyze the compatibility heuristic value by providin and to identify gaps in knowledge.   Ecosim (dynamic simulations)  E groups using coupled differential equations derived from the first Ecopath master equation (eq. 1.1).  The set of differential equations is solved using the Adams-Bashford integration method by default.  Biomass dynamics are described by eq. 1.3.   iiiii n ij n ij i eFMIBBfBBfg dB ++−+−= ∑∑ (),(),( jj B dt ⋅ == ) 11   (1.3) here dB /dt represents biomass growth rate of group (i) during the interval d ; bundance based on the adult pool biomass and on life tage mortality rates, employing a Deriso-Schnute delay difference model. For a complete  w i t gi represents the net growth efficiency (production/consumption ratio); Ii is the immigration rate; Mi and Fi are natural and fishing mortality rates of group (i), respectively; ei is emigration rate; and ƒ(Bj,Bi) is a function used to predict consumption rates of predator (j) on prey (i) according to the assumptions of foraging arena theory (Walters and Juanes 1993; Walters and Korman, 1999; Walters and Martell, 2004). It is modified by the predator-prey vulnerability parameter assigned to the interaction.  Variable speed splitting enables Ecosim to simulate the trophic dynamics of both slow and fast growing groups (e.g., whales/plankton), while multi-stanza pools (Christensen and Walters, 2004a) allow us to represent life histories and model ontogenetic dynamics.  The multi-stanza routine back-calculates juvenile cohort a s  Bird’s Head Seascape Analyses, Page 15 description of the multi-stanza routine see Walters et al. (2000).  Predator-prey vulnerabilities  The principle innovation in Ecosim considers risk-dependant growth by attributing a specific vulnerability term for each predator-prey interaction.  The vulnerability parameter is directly related to the carrying capacity of the system.  Each predator-prey trophic interaction is assigned a vulnerability coefficient, from one to infinity.  The figure is unitless and it describes the aximum increase in predation mortality allowable on that feeding interaction.  By assigning a s.   rol in Ecosim may produce unrealistically smooth changes in prey and redator biomass that fail to propagate through the food web (Christensen et al., 2004), and can  of better information, many modelers assume mid-range vulnerabilities to temper e dynamics (Okey and Wright, 2004).   ediating group.  This can be used to apture important behavioral aspects of populations and more accurately simulate ecosystem predation.  This may increase their vulnerability  other types of predators that attack dense prey schools, such as diving birds.  Dayton (1973) m low value, we imply a donor driven density-dependant interaction.  In foraging arena theory (Walters and Juanes, 1993; Walters and Korman, 1999; Walters and Martell 2004), the prey can remain hidden or otherwise inaccessible during periods of high predator abundance.  Predators are never satiated and handling time or physiological constraints do not limit predation mortality (Essington et al., 2000).  By assigning a high value, we imply a predator driven density-independent interaction, in which predation mortality is proportional to the product of prey and predator abundance (i.e., Lotka-Volterra).  This implies a high flux rate for prey species in and out of vulnerable biomass pool  Strict bottom-up cont p impart an unrealistic degree of resilience to the effects of fishing (Martell et al. 2002).  Strict top- down control may cause rapid oscillations in biomass and unpredictable simulation behaviour (Christensen et al., 2004; Mackinson, 2002) and will often produce a complex response surface that is difficult to work with under policy optimizations (Cheung et al., 2002; Ainsworth, 2006).  In the absence th  The preferable parameterization method is to fit the model’s dynamic behaviour to time series of catch or biomass by altering the vulnerabilities manually, or with the assistance of automated routines in Ecosim (Christensen et al., 2004).  Data fitting is done here using the available time series that we collated from governmental fisheries statistics.  Future revisions to this model will incorporate time series abundance information recently collected in community interviews (contact: C. Rotinsulu.  CI.  Jl Arfak No. 45.  Sorong, Papua, Indonesia 98413), and catch and effort data collected by the CI socioeconomic analysis (contact: A. Dohar, CI. Jl. Gunung Arfak.45.Sorong, Papua, Indonesia).  Mediation factors  Ecosim offers the capability to represent non-trophic effects that have a strong influence on food web dynamics.  Using mediation functions, the vulnerability of a given prey to a given predator can be affected according to the biomass density of a third m c functioning.    The most common types of mediation models applied in Ecosim include facilitation and protection.  An example of facilitation is seen when pelagic piscivores like tuna drive small pelagics to surface waters, increasing their vulnerability to avian predators (e.g., Dill et al., 2003).  This is known as ‘competitor facilitation’ because birds compete with tuna for a common prey type.  Similarly, small pelagics may be corralled into tight aggregations as an anti-predator defense in response to fish or marine mammal to provides a different example of competition facilitation in which urchins, taking insecure footholds to avoid predation by sea stars, are dislodged through wave action and made available to anemone predators.  Strand (1988) provides another example concerning inter-specific  Page 16, Fisheries Centre Research Reports 15(5), 2007 foraging associations; where nuclear-follower behaviour improves hunting success in certain reef species.  Protection effects occur when structure-forming species, such as reef-building corals, provide shelter for reef dwelling fish or invertebrates.  Elimination of the biotic structure by grazing corallivorous fish or crown of thorns starfish for example may regulate the survival of fish and invertebrate species taking refuge within the reefs.    Four mediation functions have been entered into the BHS-EBM models.  The first function describes a major facilitation effect, in which tunas (both ‘skipjack tuna’ and the ‘other tuna’ group) corral small pelagics and anchovy near to the surface, and make them more vulnerable to predation by birds.  The mediation function is entered so that the vulnerability of the prey groups increases in linear proportion to the biomass of tuna, up to a maximum increase of 2X the baseline vulnerability.  The mediating groups, skipjack tuna and other tuna, contribute equally to this effect.  Prey groups subject to this mediation effect are adult and juvenile anchovy, and adult and juvenile small pelagics.  The second mediation function represents a major protection effect, in which hermatypic (reef- uilding) scleractinian corals confer protection against predators to the following groups: small he third mediation function represents a minor protection effect, in which cleaner wrasse uring periods of low cleaner wrasse biomass), and they may decrease to a inimum of 0.5 times the baseline value (during periods of high cleaner wrasse biomass).  By seline value.  The ediating groups, mangroves and sea grasses, do not affect the predator-prey interactions stricted to maximum of 2X the baseline value, egardless of the biomass of the mediating group(s).  This limitation was less of a concern when b and medium reef-associated fish, sub-adult groupers, sub-adult snappers, juvenile and sub-adult Napoleon wrasse, juvenile coral trout and octopus.  The function is modeled so that the vulnerability of the prey species changes in inverse linear proportion to coral biomass.  All the predators of these prey species are affected equally.  The vulnerabilities of these small reef fish are free to increase to a maximum of 2X the baseline value (during periods of low coral biomass) and can decrease to near 1 (during periods of high coral biomass).  T improves the health of large reef-associated fish.  This effect is applied to the adult stanzas for groupers, snappers, large reef-associated fish, coral trout and Napoleon wrasse.  The effect is modelled so that vulnerability of the large reef fish to their predators changes in inverse linear proportion to cleaner wrasse biomass.  All the predators of these large reef fish species are affected equally by this mediation effect.  Vulnerabilities may increase to a maximum of 1.5 times the baseline value (d m using the mediation functions, we are making the assumption that cleaner wrasse improve the health of large reef fish populations, and that this allows them to avoid predation.  The fourth mediation function represents a minor protection effect, in which mangroves and sea grasses provide protection to juvenile groupers and snappers.  The effect is weighted so that the vulnerabilities of these juvenile reef fish species increases to all their predators as the biomass of mangroves and sea grasses goes down; i.e., in inverse linear proportion. The vulnerabilities are allowed to increase to 1.5X the baseline value, or decrease to 0.5X the ba m equally; mangroves have a stronger affect than sea grasses on the order of 3:1.    We have chosen to use simple linear effects for all of the mediation functions pending review by experts.  However, it may be difficult to differentiate the relative effects of behavioural interactions with those of trophic cascades (Carpenter and Kitchell, 1993; Walters et al., 1997).  Even if animal behaviour and ecology is well understood, it may be difficult to prescribe mediation functions based on empirical data, as there are currently significant limitations in the mediation routine.    The increase in vulnerabilities is currently re r the routine was originally integrated into Ecosim, but since the release of Ecopath V5.1 the definition of the vulnerability parameter has changed.  It is now set for each predator-prey  Bird’s Head Seascape Analyses, Page 17 interaction from 1 to infinity (Christensen and Walters, 2004a).  When vulnerabilities are low, as in donor-driven interactions, a relative increase of 2X has a much larger affect than when vulnerabilities are high.  If the same mediation effect is applied to numerous trophic interactions, it may be difficult to forecast the relative impact on each group.   A second limitation in the routine is that each predator-prey interaction can be governed by only one mediation function.  Therefore, we are currently forced to choose only the most influential mediating effect for any given predator-prey interaction.  We cannot, for example, model the protection that coral reefs impart on a reef fish population, while simultaneously representing the advantage conferred on them by cleaner wrasse.  In the present models, cleaner wrasse are assumed to be more important to the adult reef fish stanzas, while reef protection is assumed to e more important to sub-adult or juvenile stanzas. This limitation will be resolved with the n September 2007.   ) and test hypotheses regarding ecological nction and the effect of fisheries.  Previous authors have used Ecospace in this capacity (e.g., inety-eight functional groups are used to represent the marine ecosystem of Raja Ampat.  hese include mammals, birds, reptiles, fish, invertebrates, plants, zooplankton, phytoplankton,  groups such as fishery discards and organic detritus (Table A.3.1).  The models ave been designed to serve at various spatial scales.  Ideally, smaller area models, such as the  groups, these are omplex models, but we believe that this approach is necessary in order to provide sufficient b upcoming release of EwE V6.0 i  Ecospace (dynamic spatial simulations)  Ecospace (Walters et al. 1998) models the feeding interactions of functional groups in a spatially explicit way.  A simple grid represents the study area, and it is divided into a number of habitat types.  Each functional group is allocated to its appropriate habitat(s), where it must find enough food to eat, grow and reproduce - while providing energy to its predators and to fisheries.  Each cell hosts its own Ecosim simulation and cells are linked through symmetrical biomass flux in four directions; the rate of transfer is affected by habitat quality.  Optimal and sub-optimal habitat can be distinguished using various parameters such as the availability of food, vulnerability to predation and immigration/emigration rate.  By delimiting an area as a protected zone, and by defining which gear types are allowed to fish there and when, we can explore the effects of marine protected areas (MPAs fu Walters et al., 1998; Beattie, 2001; Pitcher and Buchary, 2002a/b; Buchary et al., 2002; Pitcher et al., 2001; Salomon et al., 2002; Sayer et al., 2005).  Ecopath parameterization  Functional group designations  N T and non-living h one representing Kofiau Island, would have a group structure especially suited to represent coral reef organisms and their interactions, while the larger area RA model should consider pelagic and deep-water species in more detail.  However, to keep the various models comparable, identical group structures are used.  A compromise solution is therefore used that tends to emphasize reef communities, while providing the basic level of functionality necessary to assist management of pelagic and deep-water resources.  .   High-order food web dynamics are carefully represented in the BHS EBM models in order to provide reliable forecasts concerning the impacts of fisheries on coral reefs.  Important predatory, herbivorous and commercial fish tend to be allotted into highly specialized functional groups, while basal organisms are generally aggregated.  At 98 functional c resolution to capture important processes occurring on coral reefs.       Page 18, Fisheries Centre Research Reports 15(5), 2007 Fish groups  Because of the enormous amount of differentiation in life-history, morphology and feeding itional groups configured to allow the presentation of important commercial, social and ecological interests.  The important tions were determined based on the ecological literature available for coral reef cosystems (e.g. Bellwood et al., 2004) and through expert communication.   nal roles (e.g., grooming by leaner wrasse, algae mediation by herbivorous echinoids), to represent species of commercial l groups were established based on Bellwood et al. (2004) and Ayre and ughes (2004), and modified based on expert opinion (T. Pitcher, UBC Fisheries Centre.  2204 guilds that appears within coral reef fish families, delineating functional groups by fish family or clade is impractical and may be unwise.  Through evolutionary convergence, similar niche specializations can be present in unrelated taxa; or, a single fish family may include multiple functional niches.  The specific group structure in a EwE model is largely subjective and should be tailored to satisfy specific requirements of the investigation.  Therefore, most of the functional groups developed for the preliminary Raja Ampat ecosystem models are based on the functional role that the fishes play in the ecosystem, with add re specializa e  There are 1203 fish species represented in the RA model.  The common and scientific name of each species is presented in Table A.1.1 along with their assigned functional group.  The fish species are apportioned into 57 functional groups; of which 30 represent unique species or species groups.  The remaining functional groups correspond to various juvenile, sub-adult and adult life history stages included in the model to represent ontogenetic feeding, mortality and behaviour.     Fish functional groups may be designed to represent specific functio c interest (e.g., skipjack tuna, groupers) or to cover the wide diversity of fishes in aggregated species groups (e.g., large reef-associated fish).  Fish have been allocated into functional groups based also on body size (e.g., small, medium and large groups), feeding guild (e.g., planktivorous and piscivorous) and habitat (e.g., pelagic, demersal, reef-associated).  The rationale behind functional group designation is provided in Table A.3.1.   Reef fish  Reef fish functiona H Main Mall.  Vancouver, BC; P. Mous, TNC-CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia; A. Muljadi, Reinhart Poat, Obed Lense TNC-CTC. Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413).  The groups were revised again following the TNC, CI, WWF, UBC Modelling coordination workshop (Sanur, Bali, Indonesia April 10-14).  McKenna et al. (2002b) identified 940 species of coral reef fish present in Raja Ampat.  Although all of these species are associated with reefs to some degree, the species were subdivided into pelagic and demersal based on the comment field in the FishBase Ecology table.  Where a single fish species could suitably fit into several aggregate functional groups, it was usually assigned to the most taxonomically specific group.  For example, kawakawa tuna (Euthynnus affinis) are large and piscivorous, and so could fit into the piscivorous ‘large pelagic’ functional group.  Instead, kawakawa is slotted into the more exclusive ‘other tuna’ functional group.  Similarly, the group ‘large planktivorous fish’ includes planktivorous species that are both reef-associated and pelagic, but these were kept apart from those larger aggregate groups to highlight their uncommon feeding mode.   Planktivorous fish  Obligate and facultative planktivorous species are included in the planktivorous functional groups. Where quantitative diet information is unavailable from the FishBase Diet table, assigning fish to planktivorous functional groups may require a judgment call based on  Bird’s Head Seascape Analyses, Page 19 qualitative information as contained in the FishBase Species, Fooditems and Ecology tables.  For a species to be included into a planktivorous functional group a prominent mention of planktivory is required in diet remarks on the Species table.  A comment such as ‘eats mainly zooplankton’ is assumed to indicate planktivory.  The Ecology table provides a simple diet classification in its ‘Mainfood’ and ‘Feeding type’ fields.  Positive indicators for planktivory clude the entry ‘zooplankton’ in the ‘Mainfood’ field, and ‘selective plankton feeding’ or s (FishBase; Allen, 2000).  The size istributions of ‘pelagic fish’, ‘reef-associated/demersal fish’ and ‘planktivorous fish’ are ig. 2.1.  The functional groups ‘pelagic fish’ and ‘reef-associated/demersal fish’ onsist of 133 and 674 species respectively.  The pelagic habitats categorized by FishBase include ciated planktivores) or pelagic fish (in the case of pelagic planktivores).   s (Hutchings, 2002).  Bioeroders can be classified into browers, who crape or rasp the reef substrate feeding on epilithic algae or invertebrates, and grazers who consume much more reef material in search of endolithic prey (Holt, 2003).  Bioeroders can have a positive impact on the reef community by oxygenating the reef substrate and removing dead coral to facilitate settlement and growth of new individuals, but they can also initiate a cascading destruction of the reef if chronic degradation weakens resistance to biological invasion or wave action.  In the BHS EBM models, bioeroding fish are classified into three functional groups based on Bellwood et al. (2004).  Causing the least damage to reefs are the herbivorous ‘macro-algal browers’, selected so based on diet information and qualitative remarks in FishBase.  More damaging are the ‘scraping grazers’, including members of Scaridae (parrotfish), Acanthuridae (surgeonfish), Monacanthidae (filefish), and Tetraodontidae (puffers).  The functional group in ‘filtering plankton’ in the ‘Feeding type’ field.  The Fooditems table lists prey items in order of importance, and a prominent mention of a planktivorous prey item is said to qualify the species for a planktivorous EwE group.  In addition, species may be designated as planktivorous without specific mention of planktivory if their specified prey items are among the more common planktivorous taxa (e.g., copepods, euphausiids, ostracods) and if their diet does not contain a large portion of non-planktonic components.  Subdividing habitat type and feeding guilds by fish size  Fish size was based on maximum length, converted to TL (see Section 2.5.1 - Length-length conversions) since an Lmax could be found for 96.3% of specie d presented in F c ‘pelagic’ and ‘benthopelagic’ zones, while the demersal habitat includes the ‘demersal’ zone.  Fish occurring in the ‘bathypelagic’ or ‘bathydemersal’ zones (i.e., occurring at depths > 200m) are considered to be deep-water species.  These groups, aggregated by habitat type, were further divided into either 2 or 3 size categories (e.g., small, medium and large).  The size category for each species was determined by comparing their length against the length of other species occurring in their habitat.  Fish species that are present in the ‘planktivorous’ functional groups, for example, were divided into small, medium or large size categories based on a comparison against other reef-associated fish (in the case of reef-asso  They were not compared strictly to other planktivores, but to all sympatric species.  This method was preferred in order to maximize the number of species serving as a comparison.  There are eleven sharks in the Raja Ampat model; these are divided based on Lmax into ‘small sharks’ (five species < 200 cm) and ‘large sharks’ (six species > 200 cm).  Bioeroders  Bioerosion is the process whereby certain species of fish, invertebrates, plants, fungi and bacteria cause mechanical and/or chemical erosion of calcareous skeletons of corals and other reef organisms through feeding and burrowing behaviour.  It is known to be a major structuring force in coral reef ecosystem s  Page 20, Fisheries Centre Research Reports 15(5), 2007 ‘eroding grazers’ is reserved for the two most damaging species of parrotfish, which use their specialized beak-like jaws and pharyngeal mill to process coral substrate: doubleheaded parrotfish (Scarus microhinos) and green humphead parrotfish (Bolbometopon muricatum).  These species are thought to have a serious impact on reefs in RA (Adityo Setiawan.  TNC-CTC.  Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413, pers. comm.; Bellwood et al., 2004).  he crown-of-thorns starfish (Acanthaster planci) was given its own functional group to A) Reef associated & demersal (n=688) 0 30 60 90 120 5 35 65 94 124 154 184 214 243 273 Lmax TL (cm) F re qu en cy  B) Pelagic (n=133) 0 5 10 15 20 25 7 57 108 158 208 259 309 360 410 460 Lmax TL (cm) F re qu en cy  C) Deepwater (n=57) 0 2 4 6 8 10 5 25 46 67 88 108 129 150 171 191 Lmax TL (cm) F re qu en cy  D) Planktivorous (n=160) 5 15 20 F y 10 re qu en c 0 5 25 46 67 88 108 129 150 171 191 Lmax TL (cm)  Figure 2.1 - Fish maximum length (Lmax) distribution.  Histograms based on total body length (TL).  A- C) based on habitat type; D) based on feeding guild.  Species are mutually inclusive.   T describe the serious impact that these animals can have on reefs. As a dominant corallivore,  Bird’s Head Seascape Analyses, Page 21 periodic outbreaks of the sea star can have long-lasting impacts on the health of the coral reef community.  Such outbreaks may be a direct or indirect consequence of human activities (Endean, 1969; Randall, 1972) but empirical evidence is scarce (Pratchett, 2005).  Removal of large fish species (e.g., Lethrinidae, Napoleon wrasse Cheilinus undulatus) may also reduce the atural predation mortality on A. planci populations permitting outbreaks, although Sweatman pirically.    in year-1), and ecotrophic efficiency (EE; unitless).  copath also provides an input field representing the ratio of production over consumption n.  Section ‘Functional group arameterization’ addresses each group specifically, reporting where literature values and other  used to set the basic parameters.  Most often, Q/B was set using the mpirical formulae of Pauly (1986); a few species were set using Palomares and Pauly (1998)  length L∞), asymptotic weight (W∞) and the von Bertalanffy growth constant (K) were selected among oC.  An average value was taken for each species for values ithin this temperature range.  When no growth parameters were available from within this he W∞ parameter is utilized by the Q/B regression formula presented below (eq. 2.7), and it is ed by the multi-stanza routine.  W∞ is the asymptotic fish body weight in grams  n (1995) could not confirm it em   The activity of bioeroding species is captured through the diet matrix.  They consume coral groups, including hermatypic scleractinian corals, non-reef building corals and soft corals, as well as calcareous algae.  By used of mediation functions in Ecosim (see Section on Mediation factors) a realistic impact of bioeroders can be modelled, where removal of the substrate impacts the survival of juvenile fish by limiting their refuges and increasing predation mortality  Basic parameterization   The data needs of Ecopath can be summarized as follows.  Four data points are required for each functional group: biomass (in t·km-2), the ratio of production over biomass (P/B; in year-1), the ratio of consumption over biomass (Q/B; E (P/Q; unitless), which users may alternatively use to infer either P/B or Q/B based on the other.  Each functional group requires 3 out of 4 of these input parameters and the remaining parameter is estimated using the mass-balance relationship in eq. 1.1. A biomass accumulation rate may be entered optionally; the default setting assumes a zero-rate instantaneous biomass change. These Ecopath data points are referred to collectively in this report as the basic parameters.  For a more thorough description of Ecopath data needs and parameter definitions please refer to Christensen et al., (2004).    This section ‘Basic parameterization’ describes the general methodology used to assign fish functional groups their basic parameters using FishBase informatio p special data sources were e using tail aspect ratio as modified by Christensen et al. (2004).  P/B was determined based on the sum of the natural mortality rate (M), estimated using the empirical formula of Pauly (1980), and some fishing mortality rate (F), which is an assumed fraction of M.  As a guideline, heavily exploited species were assumed to have an F approximately equal to M, while moderately exploited species were assumed to have an F equal to M/2 or less.     Growth parameters  All growth parameters utilized from the FishBase PopGrowth table, including asymptotic ( values in the temperature range 28 ± 2 w range, an average value of all available parameters, regardless of the temperature, was used for the species.  Some growth data is duplicated in other FishBase tables, for example W∞ occurs in the PopGrowth table and the QB table.  The growth constant K can occur in the FishBase PopGrowth table or the QB table.  In all cases, growth data was taken from the PopGrowth table preferentially, then the ‘QB’ table, then the ‘Species’ table, as illustrated in Figs. 2.2 - 2.4.  Estimating asymptotic weight  (W∞)  T also requir .  Fig.  Page 22, Fisheries Centre Research Reports 15(5), 2007 2.2 illustrates the method used to establish W∞ for fish species.  W∞ is taken directly from  (eq. 2.1), utilizing a and b growth parameters found espectively in the ‘a’ and ‘b’ fields of the FishBase PopGrowth table, and L∞.  L∞ is taken  the ‘aveLinf (TL)’ field of the PopGrowth table.   ld of the FishBase Species table, according to e assumption shown in eq. 2.2§. max = W∞ · 0.95 (2.2) and the L·W relationship q. 2.1) is subsequently used to establish W∞.  A decision flow tree is presented in Appendix A.1 summarizing the data source used to calculate W∞  L  To maximize the number of species contributing data towards parameter values for aggregate nctional groups, average family values for L·W parameters were calculated for some functional  values for at least five species per family. FishBase, if it is available in the ‘aveWinf’ field of the PopGrowth table or the ‘Winf’ field of the QB table.  Where no value is available from FishBase, the parameter is calculated from the length-weight (L/W) relationship r preferentially from  W = a · L b (2.1)  If any of these L/W parameters are unavailable, then W∞ is instead estimated from the maximum weight (Wmax), which occurs in the ‘Max weight’ fie th  W  If Wmax is unavailable, then L∞ is estimated from Lmax, which can be found in the FishBase PopGrowth table, according to eq. 2.3 as found in Pauly et al., (1993), (e . max = L∞ · 0.95 (2.3) fu groups if there were example  Is W    available from FB GROWTH table? ∞ Is W   available from FB Q/B table? ∞ Are a, b and L  av ble?aila W   found∞ Use L•W relationship (convert length to TL) N Y Y N Is Wmax available? Assume Wmax = 0.95•W∞ Assume Lmax = 0.95•L (Pauly et al., 1993) ∞ Cannot estimate W∞ ∞ Are a, b and Lmax available? N N N Y Y Y Start                                                          § If we assume eq. 2.3 is correct, then a more precise estimate is given by Wmax=W∞*0.86  Bird’s Head Seascape Analyses, Page 23 Figure 2.2 - Flow chart showing W∞ parameterization method.  All growth parameters are taken from areas occupying the temperature range 28 ± 2oC; where values were unavailable from within this temperature range, an average value was used for all available parameters regardless of temperature.  Length-length conversions  The empirical formula of Pauly (1980) for estimating M and the formula of Pauly (1986) for estimating Q/B both require L∞ as measured in total length (TL).  Entries for L∞ in FishBase (in both Species and PopGrowth tables) are usually provided in TL.  Where length measurement are given in other formats by the original data sources (e.g. in fork length (FL) or standard length (SL)), FishBase usually provides conversions to TL in the ‘TLinfinity’ field; no conversions are provided for maximum lengths found in the ‘Species’ table.  When required, conversions were performed manually.  To convert FL to TL, the linear empirical relationships of Booth and Isted (1997) were used.  For fish with forked tails, the relationship employed is based on panga (Pterogymnus laniarus), as in eq. 2.4:  FL = 0.901·TL – 0.6848  (2.4)   Q/B found Empirical formula based on aspect ratio (Palomares and Pauly, 1998) Is Q/B available from FB Q/B table? Y Is W available? Cannot estimate Q/B (use family parameter if available) N feeding mode (Pauly, 1990) Empirical formula based Is hd and pf available? aspect ratio available? ∞ Is hd and N YYAre a, b and L∞ Use L•W relationship Y available? to estimate L∞ Y N N N Start  Figure 2.3 - Flow chart showing Q/B parameterization method.  Feeding mode (pf) and diet e lesser gurnard (Chelidonichthys uekerri) as in eq. 2.5:  composition (hd) parameters for empirical formulae were obtained from FishBase Ecology table (Herbivory and FeedingType fields, respectively), or set according to qualitative description of feeding habits in FishBase Species table.   For fish with emarginated tails, the relationship is based on th q  Page 24, Fisheries Centre Research Reports 15(5), 2007       FL = 0.9454·TL + 3.6166  (2.5)  All pelagic, benthopelagic and bathypelagic fish were assumed to have forked tails, while all reef f marginated tails.  Each FishBase pecies is demarked into one of these six habitat classifications according to habitat data stimating consumption rate (Q/B)  /B = 10 6.37 · 0.0313 (1000 / T) · W∞ -0.168 · 1.38 Pf · 1.89 Hd    (2.7) ish, demersal and bathydemersal fish were assumed to have e s indicated in the Habitat field of the FishBase Species table.  Where SL was provided, the conversion factor to TL was applied from Christensen and Pauly (1992) as in eq. 2.6.  TL = 1.1757·SL – 0.1215 (2.6)  E  Q/B was taken preferentially from the literature or as estimated in FishBase.  Estimates of Q/Bs from FishBase sources were accepted if the data is based on a study of similar temperature to Raja Ampat (28oC ± 2oC).  For each fish species, the Q/B value was taken directly from FishBase, if available from the ‘PopQB’ field of the ‘QB’ table.  Otherwise, an empirical relationship was used to estimate Q/B for each species.  The empirical formula of Pauly (1986) based on feeding mode was preferred (eq. 2.7), using W∞ as determined above.    Q  Is M available from FB Q/B table? Is K available from FB GROWTH table? Is K available from FB Q/B table? Cannot estimate M Is L   available?∞ Are a, b and W available? ∞ Use L/W relationship to estimate L M found Empirical formula (Pauly, 1980) N Y Y Y N N N N Y ∞ Y L    may be estimated as 0.95•Lmax∞ Start  owing M parameterization method.   emperature (T) is expressed as 1000 / (ToC + 273.1) where ToC is temperature rs Pf and Hd were set for each species based on qualitative feeding remarks located in rbivory2 and FeedingType fields of the FishBase Ecology table, and in the l  Species table.     - Flow chart shFigure 2.4  he mean annual tT in degrees Celsius (assumed 28oC).  The feeding mode parameter (Pf) is set equal to 1 for predators and zooplankton feeders, and zero for other fish species as per Pauly (1986).  The diet composition parameter (Hd) is set to 1 for herbivores, and 0 for omnivores and carnivores.  arameteP MainFood, He omment field of the  genera c   Bird’s Head Seascape Analyses, Page 25 If W∞ could not be determined, then the empirical formula of Palomares and Pauly (1998) was stead to estimate Q/B based on caudal fin aspect ratio (eq. 2.8).  Here, aspect ratio (A) is efined as (tail height/area)2; it is available from the AspectRatio field of the FishBase 991) and eported by Palomares and Pauly (1998)).  These binary values were set for each species based n the FishBase diet table or on comment fields (e.g., in the  ble). /B = 7.964 · 0.204 log W∞ + 1.965 T + 0.083A + 0.532h + 0.398d (2.8) atural mortality (M) for fish r species that are unexploited; for ated as the sum of M and fishing mortality (F).  Fig. 2.4 eterize M for fish species.  Where available, the M value fy growth constant (K) and the asymptotic er  tab  ∞ Lmax was subst su (2.9) of biomass (Q/B) as in eq. 2.11, (2.11) used in d Swimming table.  Parameters h and the d refer to the types of food consumed (i.e., for herbivores h=1, d=0; for carnivores h=0, d=0; for detritivores d=1, h=0 as defined by Palomares (1 r on diet information provided i Species ta  Q  The decision tree in Fig. 2.3 demonstrates the parameterization method for the consumption rate (Q/B) of fish species.  Q/B was set individually for all species and then averaged to obtain functional group parameters reported in Table A.3.2.  Estimating n  Natural mortality (M) is used to represent the P/B rate fo species with an annual catch, P/B is estim shows the decision tree used to param was taken directly from literature sources or from data tables in FishBase.  Where an estimate could not be found, the regression equation of Pauly (1980) was used to determine M (eq. 2.9), which requires growth information: the von Bertalanf length (L∞).  These values w e obtained for most species from FishBase PopGrowth When L  was unavailable, the maximum specimen length observed le. ituted, as ming that L∞ = 0.95·Lmax.  M = K 0.65 · L∞ -0.279 · T 0.463  Daily ration   Marine mammals  The empirical equation for daily ration of marine mammals, modified from Innes et al. (1987) in Trites and Heise (1996), is used for estimating the consumption per unit of biomass (Q/B) as in eq. 2.10.  R = 0.1 · W 0.8 (2.10)  W is body weight in kg and R is the daily ration in kg·day-1.  Birds  The empirical equation for daily ration for birds given by Nilsson and Nilsson (1976) in Wada (1996), is used for estimating the consumption per unit  log R = -0.293 + 0.85 · log W  W is the body weight in grams and R the ration in grams per day.  Ingestion rate in deposit feeders  An empirical model for the ingestion rate of aquatic deposit feeders and detritivores was used in  Page 26, Fisheries Centre Research Reports 15(5), 2007 the calculation of Q/B for sea cucumbers, as in eq. 2.12.  C = -0.381 · W0.742  (2.12) eters outine in Ecopath (Christensen et al., 2004) requires the following growth and aturity information: the von Bertalanffy growth constant K, recruitment power, relative eq. 2.1.  For all multi-stanza groups, the adult tage is considered to be reproductive, and so the Wmat represents the average body weight at the or species that had multiple data values, the maturation parameters are taken as an average, regardless o ograp th   W e of values is provided f om a single publication, an lu ed.  Maturity data is utilized for 148 reef- a  species and 122 gic/deepwa ecies.    B ity estimates  Biomass density estimates are based directly on COREMA 2005) nsect data for 26 reef- associated fish functional groups out of 48 in the models.  Biomass ates coul e for 17 additional reef-associated fish groups based on the su e a e rank vided by Mc et al. (2002b) (e.g., “common”, “rare”).  Biomass weightin e ed to each ance ranking offered by McKenna et al. (2002b).  The weig were d by comparing McKenna et al.’s (2002b) abundanc ings against biomass densiti ecies estimated from COREMAP (2005).   s xtrapolated bas n the weighting factors.  COREMAP abundance counts are based o reef resource ventor ercept sects ( ices 3 and 6 in REMAP, 2005).  The abund ce data in ss by m h numbers b average species weight. a alcu  the s  an age- ctured mod   The model s a Ri ment ip a lanffy growth function, an  employs species-spe and B  parameters from FishBase symptotic le th (L∞) and owth c t (K).  W me a s hedule for each specie ased on wh r the a oited, l r unexploited.  Species-l l biomass es ates w in rrent functional groups (Table A.3.1), to provide biomass est ates fo . mass d tes determin  from COREMAP (2005) tr cts re bi  reef  Consumption (C) is in mg·day-1 and dry weight (W) is in mg.  Estimating P/B of invertebrates  The P/B ratio for benthic invertebrate functional groups was obtained by an empirical model established by Brey (1995); it is presented in eq. 2.13.  log P/B = 1.672 + 0.993 · log(1/Amax) - 0.0335 · log(Mmax) - 300.447 · - 1/(T+273) (2.13)  Amax is the maximum age in years, Mmax is the maximum individual body mass in grams dry mass (gDM) and T is the bottom water temperature in degrees Celsius.  Group maturity param  The multi-stanza r m biomass accumulation rate, the weight at maturity (Wmat) the asymptotic weight (W∞) and the age at maturity.  W∞ was compiled at the species level for each fish functional group according to the methodology described in the above section.  K was taken as a direct average of FishBase entries in the PopGrowth table.  To calculate Wmat from FishBase length at maturity (Lmat) data, a length-weight relationship was employed as in s transition (i.e., knife edge entry to the reproductive cohort is assumed).    F f ge hic origin of average va e data points. e was accept here a rang r ssociated  pela ter sp iomass dens P ( reef tra estim d be mad bjectiv bundanc ings pro Kenna g factors w hting factors re assign  abund etermined e rank es of pecies are e  known sp The biomasses of unknown ed o n  in y and line int  tran see Append  CO an  is converted to bioma ultiplying fis y an  The aver ge weight is c lated at pecies level using stru el. use cker recruit relationsh nd von Berta d cific A length-weight , a ng  gr onstan e assu vily explimple mortality sc s b ethe groups are he ightly exploited o eve tim ere compiled to the cu im r the groups  The bio ensity estima ed anse presents fish omass on  Bird’s Head Seascape Analyses, Page 27 a re, to calculate an average biomass density e wh lu shore a s, the COREMAP (2005) biomass densit as red  1.75 % iginal r This ratio represents the f area to ma  area r all of I  used b ng et al. (2001).  D  Q t informat  was obtain from the FishBase Diet table for 255 out of 1196 ja Ampat model.  26% of e reef fish a  demer s ilable iet information, while 17% of the pelagic and deep water fish species had data.  Of the 30 fish reas.  Therefo for th ole of RA, inc ding off nd deeper area y w uced to of the or eef area value.  ree rine ratio fo ndonesia y Spaldi iet algorithm uantitative die ion ed species in the Ra  th nd sal fish specie  had ava d groups present in the model, 23 had data on at least one representative species.  The availability of diet information is summarized in Table 2.1.    Table 2.1 - Diet algorithm supporting parameters.  Habitats assigned to EwE fish functional groups for prey item allocation algorithm.  FishBase (FishBase); Raja Ampat (RA).  Fish functional group Number of species in RA model Species with FishBase diet data Speices with diet data (%) Habitat Feeding mode Groupers 46 8 17 reef-associated swallows Snappers 32 16 50 reef-associated swallows Napoleon wrasse 1 0 0 reef-associated swallows Skipjack tuna 1 1 100 pelagic swallows Other tuna 10 9 90 pelagic swallows Mackerel 9 3 33 pelagic swallows Billfish 5 2 40 pelagic swallows Coral trout 6 0 0 reef-associated swallows Large sharks 6 6 100 pelagic bites Small sharks 5 3 60 pelagic bites Whale shark 1 1 100 pelagic swallows Manta ray 1 1 100 reef-associated bites Rays 8 2 25 reef-associated bites Butterflyfish 57 29 51 reef-associated swallows Cleaner wrasse 3 3 100 reef-associated swallows Large pelagic 26 8 31 pelagic swallows Medium pelagic 9 1 11 pelagic swallows Small pelagic 75 0 0 pelagic swallows Large reef associated 212 80 38 reef-associated swallows Medium reef assoc. 175 38 22 reef-associated swallows Small reef associated 206 17 8 reef-associated swallows Large demersal 10 2 20 reef-associated swallows Small demersal 11 0 0 reef-associated swallows Large planktivore 52 15 29 either swallows Small planktivore 62 13 21 either swallows Anchovy 17 1 6 either swallows Deepwater fish 58 4 7 either swallows Macro-algal browsing 3 0 0 reef-associated swallows Eroding grazers 1 0 0 reef-associated swallows Scraping grazers 82 24 29 reef-associated swallows Detritivorous fish 7 0 0 reef-associated swallows   Categories of prey items listed in the FishBase Diet table are imprecise (e.g., ‘bony fish’, ‘benthic invertebrates’) and there are formatting and spelling variations. The FishBase data has therefore been standardized. FishBase prey items are sorted into their corresponding EwE functional  Page 28, Fisheries Centre Research Reports 15(5), 2007 groups, either in equal proportions for non-fish prey items, or in specific proportions for fish prey items calculated using a diet allocation algorithm.  The algorithm determines likely prey species for each predator based on habitat co-occupation and gape size/body depth limitations, determines the fractional contribution of each prey species according to a size-based vulnerability function, and aggregates the values to produce a predator-prey diet matrix at the functional group level suitable for EwE.    FB die t  item f o r  p r e d a t o r Id e n t i fy  p o t e n t i a l p r e y  s p eci es in  p r ey  g r oup fr o m  Raja A m p at sp ecies  list 2 D o e s  p r ed at o r  gr o up swall o w  p r e y  w h o l e ? 3 Does  p r ey  s p eci es co-occupy  hab ita t  wit h  t h e predator grou p ? 3 (p e l ag i c  or d e m e rsal/reef) I s  t h er e a f a mil y - sp eci f i c gap e - l en gth  r e l a ti on shi p  av ail a bl e f o r  t h i s  p r e d a t o r ’ s  f a m i l y ? 4 U s e  f a m i l y  r e l a t i o n s h i p  t o  c a l c u l a t e  m a x i m u m  g a p e  s i z e  bas e d  on  L m ax I s  pr ed ato r  f a m i l y  mai n l y  p i s c i v or ou s ? 5 U s e L a b r id ae gap e - l e n g t h  r e l a t i o n s h i p  t o  cal cu l a t e  gap e  si ze F a m i l y  i s  m a i n l y  p l a n k t i v o r o u s Us e Mu ll id a e ga p e - l e n g t h  r e l a t i o n s h i p  t o  c a l c u l a t e  g a p e  si z e For each FB diet  item Cat e gor i ze ontogeneti c diet  rec o rds  into ‘ j uv e n ile ’ or ‘ a d u l t ’ s t a n z a s  b a s e d  o n  F i s h B a s e  d e s c r i p t i o n Calculating predator gape N Y Y N Y N I s  p r e y  sp eci e s  eel - l i k e  o r  e l o n g a t e d ? 5 I s  pr e y  sp eci es f u s i fo rm or  n o  d a t a ? 5 M a x .  d i m e n s i o n  1 / 8  T L M a x .  di m e ns i o n 1 / 4  T L Pr ey  sp ec ies is d e ep or  fl att e ned .   M a x .  d i m e n s i o n  1/ 2 T L N N Is  pred a t or g a p e  larger th a n  prey  m a x .  d i m e ns ion? I s  p r e y  g r o u p  a  f i s h ? T h e  pe rc e n t c o ntri but i on t h at  each F B  diet it em make s  t o w a r d s a p r e d at o r ’ s  d i e t  i s  di vi de d uniformly ac r o ss  a l l  suit abl e  pr ey  fun c t i onal grou ps 1 For each prey group N Y For each prey species Predat o r  gap e  size M a x .  p r e y  d i m e n s i o n Av erage out  m u l t ipl e  diet rec o r d s  fo r sam e s p eci es / l i fe h i story  st ages (cas e d a ta fr o m  an y  world  area is  used) A l l o cat e a p pro p ri a t e  EwE  functi o n al g r oups  to each FB p r ey  i t em  (e . g . ,  ‘bo n y  fish’) 1 Calculating prey body size Calcul ate prey  body  si ze C alcula te p r e y  b o d y  si ze Sp ec i e s e x c l ude d  fro m p r e d a t o r  d i e t N Determ i ne p r e d ato r f u n ct i ona l  grou p  b a sed on  FishB a se s p ecies  code 2 N Y Y Y Sp ec i e s exc l ude d  fro m p r e d ator d i et P r e y  s p e c i e s  c o n t r i b u t e s  t o  t h e  d i e t  c o m p o s i t i o n  o f  t h e  p r e d a t o r  a c c o r d i n g  t o  a  q u a d r a t i c  r e l a t i o n s h i p  Y   Figure 2.5 - Flow chart showing diet allocation algorithm. 1.) Table 2.2.  2.) Table A.1.1.  3.) Table 2.1.  4.) Karpouzi and Stergiou (2003).  5.) Table A.2.1.   Bird’s Head Seascape Analyses, Page 29 The algorithm used to allocate prey fish species to predators is presented in Fig. 2.5.  For each predator, the algorithm assigns appropriate functional groups to each prey item category as listed in FishBase; the group assignments are presented in Table 2.2.  Ontogenetic FishBase diet ecords for predator fish are characterized into either adult or juvenile entries based on the data f a r r j et composition for certain Ecopath functional groups that are not differentiated into life history stages.  Table 2.2 - Fishbase prey items assigned to EwE functional groups.  Relevant EwE functional groups are assigned to each prey item listed for Raja Ampat fish species in the FishBase Diet table.  Groups 1-98 refer to Ecopath functional groups listed in Table A.3.1; Groups 99, 100 and 101 are diet import, juvenile fish and unidentified items respectively.  Juvenile fish items were distributed evenly across juvenile prey fish groups, unidentified items were omitted from predator diets.  r ield ‘SampleStage’ in the FishBase Diet table.  The FishBase entries ‘larvae’ and ‘recruits/juv.’ re assumed to refer to juvenile fish, while the entries ‘juv/adults’ and ‘adults’ are assumed to efer to adult fish.  These data categories are used to parameterize adult and juvenile stanzas espectively for corresponding EwE predator fish functional groups.  In some cases, adult and uvenile diet records were combined to provide an overall di F B  p r e y  i t e m R e l e v a n t  E w E  g r o u p s b o n y  f i s h 1 0 , 1 1 , 1 2 , 1 3 , 1 4 , 1 5 , 1 6 , 1 7 , 1 8 , 1 9 , 2 0 , 2 1 , 2 2 , 3 3 , 3 4 , 3 5 , 3 6 , 3 7 , 3 8 , 3 9 , 4 0 , 4 1 , 4 2 , 4 3 , 4 4 , 4 5 , 4 6 , 4 7 , 4 8 , 4 9 , 5 0 , 5 1 , 5 2 , 5 3 , 5 4 , 5 5 , 5 6 , 5 7 , 5 8 , 5 9 , 6 0 , 6 1 , 6 2 , 6 3 , 6 4 , 6 5 , 6 6 s q u i d s / c u t t l e f i s h 7 5 n . a . / o t h e r  b e n t h .  c r u s t a c e a n s 7 4 , 7 8 , 7 9 , 8 0 , 8 2 , 8 6 , 8 7 , 8 8 n . a . / o t h e r  m o l l u s k s 7 5 , 7 6 , 8 2 , 8 4 , 8 6 , 8 7 , 8 8 b e n t h i c  a l g a e / w e e d s 9 4 , 9 5 n . a . / o t h e r  p l a n k .  C r u s t a c e a n s 7 4 , 9 0 , 9 1 , 9 2 m y s i d s 9 1 , 9 2 s t o m a t o p o d s 9 0 n . a . / o t h e r  p l a n k .  c r u s t a c e a n s 7 4 , 9 0 , 9 1 , 9 2 b i v a l v e s 8 4 crab s 7 9 , 8 0 i s o p o d s 8 6 , 8 7 , 8 8 s h r i m p s / p r a w n s 7 4 n . a . / o t h e r  p l a n k .  I n v e r t e b r a t e s 7 4 , 7 5 , 8 9 , 9 0 , 9 1 , 9 2 n . a . / o t h e r  p l a n k .  i n v e r t e b r a t e s 7 4 , 7 5 , 8 9 , 9 0 , 9 1 , 9 3 p l a n k .  c o p e p o d s 9 0 , 9 2 n . a . / o t h e r  a n n e l i d s 8 6 , 8 7 , 8 8 ascid i a n s 8 5 e u p h a u s i i d s 9 0 g a s t r o p o d s 8 2 , 8 6 , 8 7 , 8 8 a m p h i p o d s 8 6 , 8 7 , 8 8 , 9 0 , 9 1 , 9 2 b e n t h .  c o p e p o d s 8 7 , 8 8 j e l l y f i s h / h y d r o i d s 8 9 sea b i rd s 5 F B  p r ey  i t e m R el evan t  Ew E g r o u p s i n sect s 9 9 t o a d s / f r o g s 9 9 f i s h  egg s / l a rvae 1 0 0 n . a . / o t h e r s 1 0 1 n . a . / o t h e r  f i n f i s h 1 0 , 1 1 , 1 2 , 1 3 , 1 4 , 1 5 , 1 6 , 1 7 , 1 8 , 1 9 , 2 0 , 2 1 , 2 2 , 2 5 , 2 6 , 2 7 , 2 8 , 3 0 , 3 1 , 3 2 , 3 3 , 3 4 , 3 5 , 3 6 , 3 7 , 3 8 , 3 9 , 4 0 , 4 1 , 4 2 , 4 3 , 4 4 , 4 5 , 4 6 , 4 7 , 4 8 , 4 9 , 5 0 , 5 1 , 5 2 , 5 3 , 5 4 , 5 5 , 5 6 , 5 7 , 5 8 , 5 9 , 6 0 , 6 1 , 6 2 , 6 3 , 6 4 , 6 5 , 6 6 n . a . / o t h e r  b e n t h .  I n v e r t e b r a t e s 7 2 , 7 4 , 7 6 , 7 7 , 7 8 , 7 9 , 8 0 , 8 1 , 8 2 , 8 3 , 8 4 , 8 5 , 8 6 , 8 7 , 8 8 p o l y c h a e t e s 8 7 , 8 8 o s t r a c o d s 8 6 , 8 7 , 9 0 , 9 1 , 9 2 n . a . / o t h e r  r e p t i l e s 6 , 7 , 8 , 9 , 9 9 d i a t o m s 9 3 n . a . / o t h e r  b e n t h .  i n v e r t e b r a t e s 7 2 , 7 4 n . a . / o t h e r  e c h i n o d e r m s 7 4 , 7 6 , 7 7 , 7 8 , 7 9 , 8 0 , 8 1 , 8 2 , 8 3 , 8 4 , 8 5 , 8 6 , 8 7 , 8 8 c h i t o n s 8 7 n o n - a n n e l i d s 8 6 , 8 7 , 8 8 n . a. / o t h er  b e n t h .  C r u s t acean s 7 4 , 7 8 , 7 9 , 8 0 , 8 2 , 8 6 , 8 7 , 8 8 l o b s t e r s 7 8 s ea u r ch i n s 8 3 s e a  s t a r s / b r i t t l e  s t a r s 8 7 d e b r i s 9 7 , 9 8 cl ad o cer an s 9 1 , 9 2 s ea cu cu m b er s 8 3 n . a . / o t h e r  m a m m a l s 1 , 2 , 3 o c t o p i 7 6 sea st a r s/ b r it t e  st ars 8 7 n . a . / o t h e r  c e p h a l o p o d s 7 5 , 7 6 carca ss e s 9 7 , 9 8 n . a . / o t h e r  p h y t o p l a n k t o n 9 3 h a rd  co ral s 6 7 , 6 8 , 6 9 n . a . / o t h e r  p o l y p s 6 7 , 6 8 , 6 9 , 7 0 s p o n g e s 8 5 d i n o f l a g e l l a t e s 9 0 , 9 2 b l u e - g reen  al gae 9 3   The diet allocation algorithm first eliminates potential prey species from the predator’s diet if ey do not occur in the same habitat as the predator.  Predator habitat was determined at the th functional group level as listed in Table 2.1.  Predator habitat classification is divided into two  Page 30, Fisheries Centre Research Reports 15(5), 2007 categories, reef-associated/demersal and pelagic; it is based on the ‘habitat’ field of the FishBase Species table.  Species categorized in FishBase as pelagic, benthopelagic or bathypelagic are assumed to occupy a ‘pelagic’ habitat, while reef-associated and demersal fish are assumed to occupy a ‘reef-associated/demersal’ habitat.  Prey habitat types are similarly simplified from FishBase habitat entries, but at the level of species.  A minimum and maximize prey size is then determined for each predator based on mouth gape size.  These may be important parameters governing population dynamics (Claessen et al., 2002).  In aquatic systems, the lower limit to the consumption relationship may be set by the ncounter rate, the predator’s ability to visually locate prey (Lundvall et al., 1999) or to retain the imple rules were used here to establish the predation window and the consumption rate of fish ach redator family was designated as being mainly piscivorous or planktivorous based on the  consumption by a otential predator, is determined in a separate calculation.  The body morphology is assessed for equal 50% of Lmax.  Fish mily body morphologies used by the algorithm are reported in Table A.2.1.  e assume that the predator-prey consumption rate follows a domed relationship that is minimum and maximum rey sizes available to the predator, and it is highest in the mid-range.  Eq. 2.14 shows the dome e prey after capture (Persson, 1987).  The maximum limit may be determined by mouth gape size (Hoyle and Keast, 1988; Scharf et al., 2000), or by changes in capture and handling efficiency (Christensen, 1996), changes in prey fish behaviour, prey visibility/camouflage (Lundvall et al., 1999), nutritional content, toxicity and other factors.  Both the minimum and maximum prey sizes may also be constrained by the precepts of optimal foraging (Emlen 1966; Schoener 1971).  However, within this ‘predation window’, prey species are vulnerable.    S predators on fish prey.  Predator functional groups that swallow their prey whole are assumed to be constrained through gape size limitations with respect to the size of prey they can consume.  All of the predator fish functional groups are assumed to swallow prey whole except for the functional groups ‘large sharks’, ‘small sharks’, ‘Manta ray’ and ‘rays’ and their corresponding juvenile groups (Table 2.1).  These groups feed by biting or tearing pieces off their prey, and so are assumed able to feed on larger fish than a gape-restricted species of similar size.    To determine the maximum gape size of swallowing predator species, family-specific gape-body length relationships were utilized from Karpouzi and Stergiou (2003) for Synodontidae, Scorpaenidae, Serranidae, Carangidae, Mullidae, Labridae and Scaridae.  Calculating the gape size requires an estimate of body length standardized into TL (see Section 2.5.1 - Length-length conversions).  For other predator families, the maximum gape size was determined by assuming a similar gape-body length ratio as Labridae, in the case of mainly piscivorous predator families, or Mullidae, in the case of mainly planktivorous predator families (see Table A.2.1).  E p predominant feeding mode seen in member species.  The proportion of species within each family exhibiting piscivory is reported in Table A.2.1; this figure applies specifically to species present in RA.  Member species are considered to be planktivorous under the same criteria used for assigning fish into planktivorous EwE functional groups (see Section 2.4.2 - Planktivorous fish).  Briefly, the main food items must be planktonic as reported either quantitatively in the FishBase Diet table, or qualitatively as reported in the Ecology, Fooditems or Species tables.  The smallest body dimension of the prey species, i.e., the dimension limiting p each fish family based on representative members that have morphological information in FishBase.  For ‘eel-like’ or ‘elongated’ fish familes, the smallest body dimension is assumed to be 12.5% of the maximum body length (Lmax in TL, taken at the species level).  For ‘fusiform’ fish or fish with no data, the smallest body dimension is assumed to be equal to 25% of Lmax.  For ‘deep bodied’ or ‘flattened’ fish, the smalled body dimension is assumed to fa W dependant on the relative sizes of the species.  The quadratic model used to predict the consumption is initialized so that the consumption rate is zero at the p shaped quadratic function passing through (0,0) and (0,1).  Bird’s Head Seascape Analyses, Page 31   xxQ ij 44 2 +⋅−=  (2.14)  Q equals the relative consumption of predator (j) on prey (i), and x equals the smallest body imension of prey species (i), divided by the gape size of predd ator (j).   be to use a right-skewed by prey max ave ons are, the more inaccurately will the algorithm predict the y size range of prey consumed, since gape size and prey body length change non-linearly with length.  Additional sampling work could help describe the current population age structure for critical species and address this potential source of error.  The output of the diet allocation algorithm has been modified during the process of balancing and tuning the model to time series data.  The diet matrix used in the 2000 RA model is presented in Table A.3.6.     A dome-shaped vulnerability function may be an appropriate model to describe prey mortality as a function of predator length (Lundvall et al., 1999; Claessen et al., 2002).  Nevertheless, the relationship can be confused by the presence of refugia (Lundvall et al., 1999), which may be an important factor on coral reefs with high substrate complexity.  The lower limit to the predation window may have an especially influential impact on population dynamics through the effects of cannibalism (Claessen et al., 2002; Persson et al., 2000).    n alternative to the quadratic consumption rate equation may A relationship such as a beta distribution, so that a wide range of smaller prey sizes are accessible, but predation mortality falls quickly as prey size approaches the predator gape-size limit.  This may be appropriate if the minimum prey size consumed by the predator does not tend to increase as fast as the maximum gape size, (e.g., Scharf et al., 2000).  Another prospective improvement to the algorithm may be to implement a monotonically increasing consumption rate function for smaller predators, or to employ a dynamic predation mortality function, whose peak shifts right with larger prey sizes (Lundvall et al., 1999).    he diet algorithm in place also assumes that the availability of prey species is affected T abundance.  ‘Abundant’ prey species identified by McKenna et al. (2002b) are assumed to incur 130% of the baseline predation mortality rate; ‘common’ prey incur 120%, ‘moderately common’ prey incur 110% mortality, ‘occasional’ prey suffer 90% mortality and ‘rare’ prey suffer 80% mortality.  All other species are assumed to incur baseline predation rates (100%) and the prey- consumption ratios are normalized for each predator group so that the fractional sum of prey species equals 1.  The consumption rate of a given predator on a given prey is therefore affected by both the relative sizes of the species, and the relative abundance of the prey species.    Another key assumption required by this algorithm is worth discussing.  The assumption made is that, for both predator and prey, Lmax can serve as an adequate proxy for Lave, the average fish length in the population.  However, if predator populations have been reduced significantly from unexploited levels, or if their age structure is shifted towards smaller fish by the influence of fisheries or other factors, then the algorithm will overestimate the range of prey sizes available to the predator.  Conversely, if the prey population is reduced in size or average length, then the ange of prey sizes available to predators will be underestimated.   r  If predator and prey populations are reduced in size or skewed from their unexploited age- structure by a proportionately equal amount, then the L : L  proxy may hold true.  However, e further depressed the populatith likel  Page 32, Fisheries Centre Research Reports 15(5), 2007 Fisheries  Gear types  The preliminary gear types included in the RA model were selected based on discussions with local fisheries experts and on Indonesian fishery records and publications (Departemen Pertanian. Jakarta; Subani and Barus, 1989; Andreas Muljadi, Obed Lense, Reinhart Poat, Arif Pratomo. TNC-CTC.  Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413, pers. comm.).  For the RA model, the gear structure is shown in Table 2.3.    The gear types include spear fishing, reef gleening, shore gillnets, driftnets, permanent and portable traps, spear diving (for fish and invertebrates), diving specifically for live fish, diving with use of cyanide and surface-supplied air, blast fishing using dynamite, trolling, purse seining, pole and line, set lines, lift nets, the foreign fleet and shrimp trawl.  Three diving gear types are used to represent distinct fishing methods, markets, and commodity prices received for product.  Blast fishing using explosives is known to occur throughout the archipelago, although BHS-EMB aerial surveys have not detected any (M. Barmawi; unpublished manuscript. TNC- CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia).  The foreign fleet consists mainly of powered Philippino tuna vessels operating in deeper areas in the north of RA (A. Muljadi.  TNC-CTC.  Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413, pers. comm.). The shrimp trawl fishery is located amost exclusively in the Arafura Sea, and only a small fraction of that area is considered by the RA models.    Catch time series  Fishery statistics were collected from several agency offices in Sorong: the Sorong Regency Fisheries Office (Departemen Kelautan dan Perikanan, DKP), the Raja Ampat Regency Fisheries Office and the Trade and Industry Office (Departemen Perinustrian dan Perdagangan).  The data were collated into catch and effort time series, and converted into standard units for use in Ecosim.  Catch and effort data from 1990 to 1999 are contained in the Sorong Regency Fisheries Office statistics as well as commodity prices; catch and effort data from 2000 to 2004 are contained in the Trade and Industry Office statistics.  The Raja Ampat Regency Fisheries Office had additional fisheries export data for 2005.  Export data was also acquired from the Sorong Quarantine Service for 2002 and 2004-2006.  However, the data were not used because we could not reconcile those export figures with other information from the principal data agencies mentioned above. Information from the Quarantine Service is largely concerned with the activities of specific fishing companies, and so there may be potential for further socioeconomic analysis if the ambiguity can be resolved.    For the purposes of this preliminary report, the data series assembled from DKP and the Trade and Industry Office statistics seem to form a continuous and coherent time series of catch and effort.  Trade and Industry Office data was received in hard copy, as was 2005 data from the RA Regency statistics office (DKP).  Data from the Sorong Regency office (DKP) was received in electronic format as was data from the Quarantine office (Pos Karantina Ikan Sorong).  For some species catch data was taken directly from other literature sources (e.g, Venema et al., 1997).  The collated time series catch data is presented in time series in Figs. A.3.1 and A.3.2.  Interpreting catch statistics  Assumptions must be made in order to translate imprecise and incomplete fishery statistics into useable series for the EwE models.  In some cases, data from Indonesian governmental sources may contain statistical inaccuracies (Dudley and Harris, 1987) due to the complexity of catch reporting in tropical reef-based fisheries, and common resource limitations in the fisheries bureau.  These problems can mean that we have to use some guesswork in some estimates of  Bird’s Head Seascape Analyses, Page 33 catch for Raja Ampat   Table 2.3 - Fishing gear types included in the Raja Ampat model.  Source: (Andrease Muljadi.  TNC- CTC. Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413,  pers. comm.). 1) Reef fish catch includes mainly fusiliers, rabbitfish, parrotfish and jacks; 2) Small pelagic catch includes mainly anchovy and sardine; 3) Hard shell invertebrate catch includes mainly shellfish and snails.       Gear type       Indonesia n name  Ski p jack tu n a  O th er  t u n a  G ro u p er s S n a p p er s N a p o le o n  w ra ss e Reef fi sh   1  M a ck er el  D em er sa ls  S h a rk s S m a ll  p el a gi cs   2  S q u id  C u cu m b er s O ct o p u s H a rd  s h el l in ve rt s.   3  Spear fishing / harpoon Aco / panah                       Reef gleaning  Balobe / Meting                    Shore gillnet   Jaring insang                   Driftnets   Jaring hanyut                  Permanent trap  Sero                      Portable trap  Bubu                   Diving spear and gleen  Molo / Menyelam                     Diving live fish  Molo / Menyelam                 Diving air supply (cyanide) Molo / Menyelam                 Blast fishing   Bom                         Trolling with FAD  Pancing tonda                    Purse seine with FAD  Rumpon                     Pole and line with FAD  Rumpon                    Set lines   Rawai                     Lift net   Bagan Apung                  Illegal foreign fleet  -                     The species names recorded in the catch statistics varied slightly from year to year.  Pelagic fish that were consistently included are anchovy, scad, trevally, sardines, mackerels, Spanish mackerels and tuna.  Anchovy catch was allotted entirely to the adult anchovy EwE group. Based on expert communications, the most important scad in terms of biomass and harvest value in RA  Page 34, Fisheries Centre Research Reports 15(5), 2007 is the oxyeye scad (Selar boops) (Obed Lense, TNC-CTC, Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413), which occurs in the large planktivore group.  All of the scad catch was therefore allotted to this group.   There are nine trevally species in the RA model, and they all occur in the large reef associated group; the adult stanza therefore received 100% of the trevally catch.  Sardine catch was apportioned to the adult small planktivore group.  Mackerels and Spanish mackerel catch was attributed to the adult Mackerel functional group.  Tuna catch was divided between the adult skipjack tuna group (92%) and the adult other tuna group (8%) in the same proportion as landings observed throughout Indonesia (Venema, 1997).  The ‘other’ component was divided evenly among small, medium and large adult pelagic groups.  Demersal species reported in the catch statistics are: Leiognathids, threadfin bream, croakers, hairtails, Polynemus spp., catfishes, Emperor bream, groupers, snappers and others.  Leiognathidae catch was allotted completely to the adult large reef associated functional group.  However, no Leiognathidae (ponyfish) appear in the Raja Ampat species list provided by McKenna et al., (2002b).  According to IFDG (2001), these are a common catch in the Arafura Sea, indicating that the Sorong DKP statistics include landings from the Arafura Sea.  Since the RA model only includes a sliver of the Arafura Sea, the landing density could well be overestimated for our geographic scope.  After having allocated DKP and Trade and Industry Office catch statistics into their most relevant groups, there was a quantity left over representing ‘other’ unidentified species.  This quantity was divided between the functional groups that lacked explicit catch estimates, in the proportion suggested by the total number of species in each group.  We therefore assumed that the catch of each species was equal, and that functional groups possessing many species, such as butterflyfish and the aggregate reef-associated groups, should receive a larger relative fraction of the undetermined catch component.  The catch for ‘other’ reef associated fish was divided between butterflyfish, macro-algal browsing fish, eroding grazers, detritivorous fish and the aggregate groups: large, medium and small reef-associated fish.  There is a large amount of frozen catch recorded in the Trade and Industry Office statistics for the years 2000-2002.  On average, the total frozen quantity is 13% of the total recorded catch.  However, the frozen product is likely bycatch from shrimp trawl fisheries operating in the Arafura Sea (C. Rotinsulu.  CI.  Jl Arfak No. 45.  Sorong, Papua, Indonesia 98413, pers. comm.).  As this is outside of the study area, this amount was not included in the model.  Splitting catch between functional groups  Total catch was first determined for each functional group according to the methodology described above.  Catches were then divided into juvenile, sub-adult and adult stanzas using ratios described in Section 2.5.11 - Functional group descriptions.  Generally, juveniles are assumed to comprise 10% of the total fisheries catch for all reef associated and demersal groups, the remaining 90% is allotted to the adult and subadult stanzas.  For each age class and functional group, the total calculated catch was divided among the 17 gear types in the model according to ratios presented in Table 2.4.  Each functional group was slotted into one of six categories that define the principle gear types used to pursue it.  Catches for each gear type category are divided among EwE fisheries in a unique proportion.  Interactions marked as bycatch in Table 2.4 were assumed to catch half as much as directed landings.  The final EwE landings matrix, including catch and bycatch, is presented in Table A.3.4.  We include bycatch in the catch matrix, rather than the discard matrix, because we assume that it is always sold.  Discards are set as a small fraction of the total estimated catch for each EwE fishery.  Blast fishing is assumed to discard a quantity of hermatypic scleractinian corals equal to 1% of their catch weight; trolling with FAD is assumed to discard a weight of birds equal to 1% of their catch; set lines are assumed to discard 1% of their catch weight in birds, green turtles and  Bird’s Head Seascape Analyses, Page 35 oceanic turtles combined; shrimp trawl is assumed to discard 50% of its catch weight in small demersals, deepwater fish, epifaunal detritivorous and carnivorous invertebrates.    A small discarding of crocodiles was added to account for incidental capture or hunting; this corresponds to about 20 animals per year from RA at 200 kg per animal.  Table 2.4 - Functional group catch distribution by gear type.  Each functional group is assigned into one of six gear type categories (SPEL, DIVING, etc.).  Catch for each group is distributed among 17 EwE fisheries according to unique ratios for each category.  A.) Catch ratios used for each gear type category; tuna catch ratio is based on Indonesian trends (Venema, 1997).  B.) Functional groups pursued by fisheries; D = Directed catch; B = Bycatch.  Bycatch is assumed to catch half as much as directed catch.  SKIP indicates all EwE fisheries catch an equal proportion. A) G e a r  t y p e G e a r  n a m e Sp ear  and  har poon R eef  g le ani ng Sho re  gillne t D rif tne t Pe rm an en t t ra p Por ta ble  tra p Diving  sp ear and gle an Diving  live fis h Diving  a ir su pply cyan id e B la st f is hing Tr ollin g w ith FA D Pu rs e s ei ne wi th  FA D Pole  a nd lin e w ith FA D Se t line Li ft n et Fo re ig n fl ee t Shr imp tr aw l S P E L S m a l l  p e l a g i c  g e a r s 2 0 % 2 0 % 1 5 % 1 5 % 5 % 2 5 % D I V I N G D e m e r s a l  g e a r  &  d i v i n g 5 % 1 0 % 1 0 % 1 0 % 2 5 % 2 5 % 5 % 1 0 % I N V E R T I n v e r t e b r a t e  g e a r s 1 9 % 6 0 % 2 0 % 1 % D E M D e m e r s a l  g e a r s 1 9 % 1 9 % 1 9 % 1 9 % 1 9 % 5 % P E L P e l a g i c  g e a r s 2 5 % 2 0 % 1 5 % 1 5 % 2 5 % T U N A T u n a  g e a r s 2 9 % 8 % 3 8 % 1 2 % 1 3 % E w E  F i s h e r i e s    B) y p e # G r o u p  N a m e D I V I N G 1 0 A d .  g r o u p e r s D D D D D D D D I V I N G 1 1 S u b .  g r o u p e r s D D D D D D D D I V I N G 1 2 J u v .  g r o u p e r s B B B B D I V I N G 1 3 A d .  s n a p p e r s D D D D D D I V I N G 1 4 S u b .  s n a p p e r s D D D D D D I V I N G 1 5 J u v .  s n a p p e r s B B B B B D I V I N G 1 6 A d .  N a p o l e o n  w r a s s e D D D D I V I N G 1 7 S u b .  N a p o l e o n  w r a s s e D D D D I V I N G 1 8 J u v .  N a p o l e o n  w r a s s e B T U N A 1 9 S k i p j a c k  t u n a D D D D D T U N A 2 0 O t h e r  t u n a D D D D D T U N A 2 1 M a c k e r e l D D D D S K I P 2 2 B i l l f i s h D D E M 2 3 A d .  c o r a l  t r o u t D D D D D D D E M 2 4 J u v e n i l e  c o r a l  t r o u t B B B B B B S K I P 2 5 A d .  l a r g e  s h a r k s D S K I P 2 6 J u v .  l a r g e  s h a r k s B S K I P 2 7 A d .  s m a l l  s h a r k s D S K I P 2 8 J u v .  s m a l l  s h a r k s B D E M 3 1 A d .  r a y s D D D D D E M 3 2 J u v .  r a y s D D D D D E M 3 3 A d .  b u t t e r f l y f i s h D D D D D D D E M 3 4 J u v .  b u t t e r f l y f i s h B B B B B B D E M 3 5 C l e a n e r  w r a s s e D D D D D P E L 3 6 A d .  l a r g e  p e l a g i c D D D D D P E L 3 7 J u v .  l a r g e  p e l a g i c B B B B B P E L 3 8 A d .  m e d i u m  p e l a g i c D D D D D P E L 3 9 J u v .  m e d i u m  p e l a g i c B B B B B S P E L 4 0 A d .  s m a l l  p e l a g i c D D D D D D S P E L 4 1 J u v .  s m a l l  p e l a g i c B B B B B B D E M 4 2 A d .  l a r g e  r e e f  a s s o c i a t e d D D D D D D D E M 4 3 J u v .  l a r g e  r e e f  a s s o c i a t e d B B B B B B D E M 4 4 A d .  m e d i u m  r e e f  a s s o c i a t e d D D D D D D D E M 4 5 J u v .  m e d i u m  r e e f  a s s o c i a t e d B B B B B B D E M 4 6 A d .  s m a l l  r e e f  a s s o c i a t e d D D D D D D D E M 4 7 J u v .  s m a l l  r e e f  a s o c i a t e d B B B B B B D E M 4 8 A d .  l a r g e  d e m e r s a l D D D D D E M 4 9 J u v .  l a r g e  d e m e r s a l B B B B D E M 5 0 A d .  s m a l l  d e m e r s a l D D D D D E M 5 1 J u v .  s m a l l  d e m e r s a l B B B B D E M 5 2 A d .  l a r g e  p l a n k t i v o r e D D D D D D D E M 5 3 J u v .  l a r g e  p l a n k t i v o r e B B B B B B D E M 5 4 A d .  s m a l l  p l a n k t i v o r e D D D D D D D E M 5 5 J u v .  s m a l l  p l a n k t i v o r e B B B B B B P E L 5 6 A d .  a n c h o v y D D D D D P E L 5 7 J u v .  a n c h o v y B B B B B D E M 5 8 A d .  d e e p w a t e r  f i s h D D D D D E M 5 9 J u v .  d e e p w a t e r  f i s h B B B B D E M 6 0 A d .  m a c r o  a l g a l  b r o w s i n g D D D D B D E M 6 1 J u v .  m a c r o  a l g a l  b r o w s i n g B B B B B D E M 6 2 A d .  e r o d i n g  g r a z e r s D D D D B D E M 6 3 J u v .  e r o d i n g  g r a z e r s B B B B B D E M 6 4 A d .  s c r a p i n g  g r a z e r s D D D D B D E M 6 5 J u v .  s c r a p i n g  g r a z e r s B B B B B D E M 6 6 D e t r i t i v o r e  f i s h D D D D B S K I P 6 8 H e r m a t y p i c  c o r a l s D S K I P 7 3 P e n a e i d  s h r i m p s D S K I P 7 4 S h r i m p s  a n d  p r a w n s D S K I P 7 5 S q u i d D I N V E R T 7 6 O c t o p u s D D D D I N V E R T 7 7 S e a  c u c u m b e r s D D D D I N V E R T 7 8 L o b s t e r s D D D I N V E R T 7 9 L a r g e  c r a b s D D D I N V E R T 8 0 S m a l l  c r a b s D D D I N V E R T 8 2 G i a n t  t r i t o n D D D I N V E R T 8 3 H e r b i v o r o u s  e c h i n o i d s D D D I N V E R T 8 4 B i v a l v e s D S K I P 8 5 S e s s i l e  f i l t e r  f e e d e r s D I N V E R T 8 6 E p i f a u n a l  d e t .  i n v e r t s . D D D I N V E R T 8 7 E p i f a u n a l  c a r n . i n v e r t s . D D D G e a r  t      Page 36, Fisheries Centre Research Reports 15(5), 2007 Effort time series  DKP statistics included a limited fishing effort series for the years 1994-1999.  In order to produce a continuous effort trend suitable for Ecosim analysis, we have extrapolated fishing effort back to the year 1990 using linear regression.  Similarly, effort was estimated for the years 1999-2006 by assuming a constant annual rate of increase.  The average rate of increase is based on data for all available years; however, we limit the maximum effort increase at 5% per year in the absence of better information.  Gear-effort categories identified in the statistics are as follows: hand line (HL), gill net (GN), lift bag net (LB), lift bag net in raft (LBR), troll (TR), trammel net (TN), pole and line (PL), bottom long line (BL), tidal weir (TW) and fish trap (FT).  Boat-effort categories are: non- motorized (NM), wooden outboard (WO), wooden inboard (WI) and inboard motorboats (IM).  In order to produce a relative effort series for each EwE fishing gear type, we assigned each EwE gear type to one or more appropriate gear-effort categories as listed in the statistics.  The assignments are provided in Table 2.5.  The effort of each EwE gear type is assumed to follow these categories.  Where EwE gear type effort follows more than one DKP effort category the effort series used by EwE represents the average of the relevant DKP categories.  For certain gear types, the effort increase from 1990 to 2006 was assumed to follow the population increase in Papua.  The average annual population increase is recorded as 3.22% per year by Badan Pusat Statistik (BPS) Provinsi Papua for the years 1990-2000 (BPS, 2006).  This assumption was used for the following artisanal fisheries: spear and harpoon, reef gleaning, diving with cyanide and blast fishing. Table 2.5 - Gear effort assignments.  See text for explanation of effort series.  EwE gear type Relevant DKP effort categories Spear and harpoon POP Reef gleaning POP Shore gillnet GN Driftnet GN+TN Permanent trap TW Portable trap FT Diving spear NM+WO Diving live fish WI+WO Diving cyanide POP Blast fishing POP Trolling TR Purse seine WI+WO Pole and line PL Set line BL Lift net LB+LBR Foreign fleet IM Shrimp trawl WI+IM  To produce an effort series at the level of functional groups requires some basic assumptions concerning the relative contribution made by each EwE gear type.  The fishing effort exerted on a particular group is assumed to equal the weighted average of recorded efforts for all gear type that are catching it.  Each gear type contributes to the weighted average in a proportion equal to the relative amount of catch claimed by that gear type.  In order to determine a CPUE trend for functional groups, both for use in parameterizing biomass values of the 1990 model and in fitting temporal dynamics, we simply divide the catch of each functional group estimated from DKP statistics (Section 2.5.10 - Catch time series) by the calculated effort series for each biomass pool.  The resulting trends are provided in Fig. A.6.2.  Prices  For commercial functional groups in the model, an export price is determined from Trade and Industry Office statistics; these represent average ex-vessel prices for the years 2000-2004.  Export prices are determined for groupers, Napoleon wrasse and octopus.  Prices for export product are determined for a further 32 reef-associated functional groups based on generic price listings in the Trade and Industry Office statistics for ‘mixed fish’.  The price of all these groups is assumed to equal by unit weight.  Commodity prices for domestic sale were determined based on 1993-1994 information from the DKP (Sorong Regency Office).  The value of products were  Bird’s Head Seascape Analyses, Page 37 divided into local prices (i.e., vended in Sorong market) and prices received at island markets, which are typically lower.  These were averaged to produce an overall domestic price.  Domestic prices were calculated for groupers, snappers, tuna, shrimp, shark fins, sea cucumbers, mollusks, squid, lobster and crabs.  The prices of aggregate groups (i.e., large, medium and small reef- associated / demersal / planktivorous groups and others) are set based on the generic ‘mixed fish’ price entry.  Prices in Venema (1997) were applied to tunas (export), crabs, jellyfish, seaweed and corals.  Since we had catch estimates for both export and local consumption, prices were set for each commercial group as a realistic weighted average of export and domestic prices.  Where catch data was lacking, the price of groups were assumed to be an average of export and domestic prices.  Juvenile fish always received the local prices, as we assume that they were unsuitable for export.  The price of small pelagics was modelled after anchovy.  Small pelagics are assumed to be sold locally, as no export price was found.  Market prices in the model are presented in (Table A.3.5).    Unreported catch  Most of the catch information available to us originated in Sorong or nearby cities, but fisheries catches occurring in smaller villages, especially those off the mainland, are subject to little or no observation (C. Rotinsulu.  CI.  Jl Arfak No. 45.  Sorong, Papua, Indonesia 98413,  pers. comm.).  Therefore, the catch statistics presented in Fig. A.6.1 probably represent only a fraction of total fisheries catch, considering the disperse and artisanal nature of reef fisheries, and the minimal reporting infrastructure.  Preliminary figures for unreported catch quantities have therefore been entered as placeholders into the model to allow a more accurate representation of energy flow in the system.  As major reef predators are likely harvested in unreported fisheries, the trophic implications of the missing catch could be major.  Artisanal and unreported catch estimates are now being developed by the CI socioeconomic analysis component of the BHS EBM project (contact: A. Dohar, CI. Jl.Gunung Arfak.45.  Sorong, Papua, Indonesia) and the UBC team (e.g., see Bailey et al., this volume for a contribution covering the unreported Waigeo anchovy fishery.)   Functional group descriptions  Mysticetae  The species of the cetacean suborder mysticetae occurring in RA were short-listed based on Kahn (2001) and Kreb and Budiono (2005). The estimated proportions of global abundance for the species found in FAO Area 71 were obtained from Kaschner (2004).  However, the estimates from her model are not meant to be applied to small geographic areas such as RA, and the uncertainties involved are relatively high (K. Kaschner, Forschungs- und Technologiezentrum Westküste, Hafentörn, 25761 Büsum, Germany, pers. comm.).  These were not used for the biomass estimates.  Instead, the EE of the group was fixed at 0.025 and Ecopath was allowed to estimate the biomass as 0.033 t·km-2.  Mysticete P/B is calculated as the average of r/2 (Schmitz and Lavigne, 1984), where r is the intrinsic rate of growth, for Sei whale, Minke whale and Fin whale.  P/B is estimated to be 0.0583 year-1. The r/2 method was also used as a measure for mammal P/B in Guénette (2005).  The average body weight of 6 baleen whale species is taken from Trites and Pauly (1998) (Balanoptera musculus, Balanoptera borealis, Balanoptera edeni, Balanoptera acutorostrata, Balanoptera physalis and Megaptera novaeangliae).  To calculate Q/B, the feeding ration of is determined using the relationship in Innes et al. (1987) as modified by Trites and Heise (1996); Q/B is then averaged among species to obtain a value of 4.850 year-1.   Page 38, Fisheries Centre Research Reports 15(5), 2007 Piscivorous and deep-diving odontocetae  The species of odontocetae visiting the area were short-listed based on Kahn (2001) and Kreb and Budiono (2005). The estimated proportion of the global abundance of the species that can be found in FAO Area 71 were obtained from Kaschner (2004) but the uncertainties involved were relatively high (K. Kaschner, Forschungs- und Technologiezentrum Westküste, Hafentörn, 25761 Büsum, Germany, pers. comm.), these were not used for the biomass estimates. The EE of the group was fixed at 0.0025 and Ecopath was allowed to estimate the biomasses for piscivorous and deep-diving odontocetae as 0.052 and 0.091 t·km-2, respectively.  The P/B for piscivorous odontocetae is calculated as the average of r/2 (Schmitz and Lavigne, 1984), where r is the intrinsic rate of growth, for Stenella longirostris, Tursiops truncates, and Stenella attenuate to be 0.0325 year-1.  The P/B for deep-diving odontocetae is calculated as the average of r/2 for Physeter macrocephalus and Ziphius cavirostris (after Guénette, 2005) to be 0.02 year-1.  The average weight of 13 piscivorous odontocetae species are taken from Trites and Pauly (1998) and Noren and Williams (2000) (long nosed spinner dolphin, Stenella longirostris; bottlenose dolphin, Tursiops truncates; pan-tropical spotted dolphin, Stenella attenuate; Fraser's dolphin, Lagenodelphis hosei; Risso's dolphin, Grampus griseus; common dolphin, Delphinus  spp.; rough toothed dolphin, Steno bredanensis; Indo-Pacific humpbacked dolphin, Sousa chinensis; Irrawady dolphin, Orcella brevirostris; melon headed whale, Peponocephala electra; Pygmy killer whale, Feresa attenuate; dwarf sperm whale, Kogia simus; pygmy/dwarf sperm whale, Kogia  spp.,).  That source also provides the body weights of 5 deep-diving odontocetae species (sperm whale, Physeter macrocephalus; false killer whale, Pseudorca crassidens; Cuvier's beaked whale, Ziphius cavirostris; short finned pilot whale, Globicephala macrorhynus and orca, Orca orca).    For both piscivorous and deep-diving odontocetae, Q/B is based on the feeding ration determined using the relation given by Innes et al. (1987) as modified by Trites and Heise (1996); individual Q/B is averaged among species to obtain a Q/B value of 14.476 year-1 and 8.531 year-1 for piscivorous and deep-diving odontocetae, respectively.  However, these values produce a very low P/Q value ~0.001, and so consumption rates were ultimately reduced for both groups, so that the EE value matched the one employed for Mysticetae.  The resulting Q/B values are 6.1 year-1 and 3.6 year-1, for piscivorous and deep-diving odontocetae, respectively.  Dugongs  The biomass of dugongs was calculated based on a population estimate in Torres Strait by Marsh et al. (1997) (i.e., 24,225 individuals in 30,561 km-2 survey area).  The mean size of an individual is assumed to be 400 kg based on the weight range reported to be between 250 kg and 600 kg by Blanshard (2001).  The biomass is thus estimated to be 0.317 t·km-2. This estimate is scaled to the shelf area in RA, assuming that these animals occur on the shelf, to obtain the final biomass estimate 0.054 t·km-2 used in the model.  The maximum rate of increase in dugong population is approximately 5% per annum (Marsh et al., 1997).  The P/B is calculated to be equal to r/2 = 0.025 year-1, where r is the intrinsic rate of growth.  The average value of 400 kg was also used to calculate the ration based on the empirical relation given by Innes et al. (1987). The Q/B was calculated to be 11.012 year-1. Another estimate by Goto et al. (2004) places consumption of captive dugongs at 14% of their body weight, before maturity, and 7% after maturity.  This leads to Q/B estimates of 51.1 year-1 and 25.6 year-1 for dugong before and after maturity.  The values are considered to be too high in relation to the estimated production rate, and so the lower alternative is used.   Bird’s Head Seascape Analyses, Page 39 Birds  The biomass for the birds in the RA model is estimated to be 0.366 t·km-2. The value based on the biomass of  11 species (black-naped tern, Sterna sumatrana; brown noddy, Anous stolidus; bridled tern, Sterna anaethetus; crested tern, Sterna bergii; brown booby, Sula leucogaster; red-footed booby, Sula sula; great frigatebird, Fregata  minor; white-tailed tropicbird, Phaethon lepturus; red-tailed tropicbird, Phaethon rubricauda; sooty tern, Sterna fuscata; masked booby, Sula dactylatra) from the Banda sea (Karpouzi, 2005). The extent of Banda Sea was obtained from (Britannica, 2006).  The estimated value is high compared to Opitz (1993), who used a biomass density for seabirds of 0.015 t·km-2 for a Caribbean reef.   P/B for Leach’s storm petrel (Oceanodroma leucorhoa) 0.381 year-1 is used as P/B for the group in the model based on Russel (1999).  This value is low compared to the production rate for birds in French Frigate Shoals by Polovina (1984) 5.4 year-1; the same value was used for Caribbean coral reefs by Opitz (1993) and also Vidal and Basurto (2003) for Bahía de la Ascensión.  The Q/B was determined by first calculating the ration using the empirical formula given by Nilsson and Nilsson (1976) in Wada (1996), and then averaging the values for 11 species (i.e., the same species that were used to calculate biomass).  A weighted average was used based on relative biomass of each species to obtain the group Q/B, which is equal to 63.95 year-1.  This high value is comparable to Polovina’s (1984) estimate for Hawaiian reefs of 80 year-1.  Reef-associated, Green and Oceanic turtles  The turtles are grouped into three functional groups based on their habitat and feeding habits: Reef associated (hawksbill turtle, Eretmochelys imbricate; loggerhead turtle, Caretta caretta); green turtle (Chelonia mydas) and oceanic turtles (leatherback turtle, Dermochelys coriacae; olive ridley, Lepidochelys olivacae; flatback turtle, Natator depressus).   The total biomass of turtles is approximately 0.02 t·km-2 (Alias, 2003), this was scaled in a ratio (1:2:2) for reef associated, green turtles and oceanic turtles. Mast and Hutchinson (2005) estimated the leatherback population to be about 650 nesting females in the BHS. Studies of sea turtle nesting site at Jamursba Medi Beach in Raja Ampat estimated 2983 Leatherback nests, 171 green turtle nests, 13 Hawksbill nests and 77 Olive Ridley nests (Putrawidjaja, 1997). These estimates could be used for partitioning the biomass estimate into the three functional groups, however at present, the ratio was maintained at (1:2:2) until better estimates becomes available from additional sites and nesting seasons.  Biomass values are therefore 0.004, 0.008 and 0.008 t·km-2 for reef associated, green and oceanic turtles, respectively.  The latter two groups were overfished in the initial model from the effects of set line discarding, and so a biomass accumulation rate was allowed of -0.02 year-1.  The survival of loggerhead turtle was estimated as 0.8613 year-1 by Chaloupka and Limpus (2002). The P/B is calculated using the relation (M = -ln S) to be 0.1493 year-1.  The survival estimate of green turtle, 0.984 year-1 is obtained from Mortimer et al., (2000) and P/B is calculated, using the same method, as 0.053 year-1.  Opitz (1993) used a higher production rate for marine turtles on Caribbean reefs, 0.2 year-1.  The P/B estimate for green turtles is used for oceanic turtles in the absence of better estimates.  Survivorship estimates were obtained for adult female turtles.  The values were not used in the calculation of P/B, but they are informative about the proportion of hatchlings that reach the adult stage: 0.93 year-1 for flatback (Parameter and Limpus, 1995); 0.61 year-1 for green turtle (Bjorndal, 1980,); 0.43 year-1 for Kemp’s ridley (Marquez et al., 1982b); 0.48 year-1 for olive ridley (Marquez et al., 1982a) and 0.81 year-1 for loggerhead (Frazer, 1983).    A Q/B value of 3.5 year-1 was used for all the turtle groups; the value taken from a trophic model  Page 40, Fisheries Centre Research Reports 15(5), 2007 for the coastal ecosystem of the West Coast of Penisular Malaysia (Alias, 2003).    Crocodiles   The biomass of crocodiles is estimated to be 5.75E-3 t·km-2 based on population estimate of 55 animals (Kushlan, 1980) and individual weight of 230 kg (Pritchard, 1978) in Florida Bay; the area is assumed to be about 2200 km-2 (Healy, 1996).  However, the value is uncertain.  Due to diet matrix conflicts, Ecopath was ultimately allowed to estimate crocodile biomass as 1.33E-3 t·km-2.  The estimate of P/B (0.408 year-1) is based on Davis and Odgen (1994); the estimate of Q/B (6.5 year-1) is based on estimates for American crocodile, Crocodylus acutus, from Day et al. (1990).     In the RA Ecospace model, crocodiles are restricted to shallow water habitat (<10 m).  This habitat type implicitly represents marine and brackish environments.  Crocodiles are limited to these areas using dispersal parameters that are strictly prohibitive to movement.  In the small scale models for Kofiau and SW Misool estuaries are entered as an explicit habitat type; crocodiles are restricted to these regions.  There is no directed catch entered in Ecopath for crocodiles, but there is a small amount of discarding, 0.0001 t·km-2.  This is about 4.5 tonnes for all of RA, or about 20 animals per year at 200 kg per animal.  Although we expect very little crocodile catch from the study area, it is known to occur.  A large male specimen was killed by villagers on Kofiau Island in February of 2006 (C. Ainsworth, pers. observation).  For safety reasons, the villagers attempt to kill every crocodile they encounter, according to their accounts.  We therefore entered this as a discard in Ecopath, so no monetary catch value will be recorded.  Groupers  Groupers are divided into three functional groups representing life history stages: adult, subadult and juveniles.  These groups incorporate information from 46 species and 16 genera of family Serranidae (Table A.1.1).  Grouper biomass is calculated from COREMAP (2005) abundance counts, adjusted for the relative reef area in RA using the reef area to marine area ratio for all of Indonesia (Spalding et al., 2001).  Abundance counts are converted to biomass using an average individual weight obtained from an age-structured model (see Section 2.5.8 - Biomass density estimates).  Biomass density is estimated to be 0.257 t·km-2.  This amount is split by Ecopath among the three life history stages according to the mortality schedule in Table A.3.3.  Ontogenic parameters used by the multi-stanza routine represent species-level averages for RA species determined with FishBase maturity data.  Biomass accumulation rate is set at -2% per year.  The COREMAP (2005) abundance counts suggested a high biomass density in sites near Weigeo Island, 0.256 t·km-2.  This value has been scaled to represent the average biomass density in RA, according to the relative marine area to reef area ratio in Spalding et al. (2001).  The adult, sub- adult and juvenile stanzas receive 72%, 22% and 6% of the biomass respectively by employing the multi-stanza parameter estimates (Table A.3.3).  When similarly scaled for reef area in RA, Wolanski’s (2001) estimate of “Large groupers” biomass is 0.035 t·km-2; Kongchai et al., (2003) estimated only 0.0025 t·km-2 for the Gulf of Thailand.  Allen et al., (2005) provided grouper densities for East Andaman Sea of approximately 0.032 t·km-2 (this value was converted to weight using length-weight parameters for RA serranids).  The P/B of adult groupers was set at 0.225 year-1 after 5 RA species of genus Epinephelus (Grandcourt, 2005).  Subadult and juvenile groupers was set at 0.4 and 1.2 year-1, respectively to provide a realistic age distribution as quantified by Ecopath’s multi-stanza routine.  Opitz (1993) used a production value for large groupers of 0.37 year-1.  The Q/B of adult groupers was  Bird’s Head Seascape Analyses, Page 41 determined to be 9.086 year-1 using the empirical regression of Pauly (1986) based on the average of 41 grouper species out of 46.  Q/B of subadult and juvenile groups is estimated by Ecopath as 13.224 and 26.908 year-1, respectively.  Groupers are pursued by all three diving gear types in the model, as well as blast fishing, spear and harpoon and permanent traps.  Cyanide fishing, which supplies premium live fish to the Hong Kong market, has also resulted in the loss of valuable reef-associated species like Napoleon wrasse (Cheilinus undulatus) and giant grouper (Epinephelus lanceolatus) due to overexploitation (Erdmann and Pet-Soede, 1996; Mous et al., 2000).  Total catch was estimated for this group based on DKP statistics as 0.022 t·km-2, or approximately 990 tonnes annually for all of Raja Ampat.  50% of the total grouper catch was allotted to the adult functional group; 40% was attributed to sub-adults and 10% to juveniles.  Adult groupers were lightly exploited in the initial RA model, and catch was increased to represent the impact of unreported catches occurring in this group.  Assuming that grouper catches reported in Sorong constitute 40% of the entire RA catch provides the following fishery indicators from Ecosim’s equilibrium analysis: Fmsy estimate of 0.21 year-1, F2006 of 0.094 year-1 and MSY of 0.027 t·km-2.  This MSY value is an average for all of RA, but when corrected to represent only the reef area this value equates to roughly 1.35 t·km-2.  This amount compares well with the ‘typical’ grouper MSY estimate offered by Jennings and Polunin (1995) of 1 t·km-2.  We assume that there is no discarding of this valuable species group.    Snappers  Snappers are divided into three functional groups representing life history stages: adult, subadult and juveniles.  These groups incorporate information from 32 species and 9 genera of family Lutjanidae (Table A.1.1).    The biomass for snappers was determined from COREMAP (2005) abundance counts.  Abundance counts are converted to biomass using an average individual weight obtained from an age-structured model (see Section 2.5.8 - Biomass density estimates).  Biomass density is estimated to be 0.152 t·km-2.  This amount is split among the three life history stages by Ecopath according to the mortality schedule in Table A.3.3, with adults, sub-adults and juveniles stanzas receiving 53%, 27% and 20% respectively.  Ontogenetic parameters used by the multi-stanza routine represent species-level averages for RA species determined with FishBase maturity data.  Biomass accumulation rate is set at -10% per year.  The P/B ratio of adult snappers is set at 0.4 year-1.  This represents the average M of 17 species of family Lutjanidae from independent sampling studies, 0.3 year-1 (Marcano, et al, 1996; Burton, 2001; Burton, 2002; Newman et al., 1996; Newman, 2002; Newman et al., 2000; Kamukuru et al., 2005, Wilde and Sawynok 2004), but the value has been increased by one third to account for fishing mortality.  This value is not too different from the one used to represent snappers in EwE models by Vidal and Basurto (2003) and Arreguín-Sánchez et al. (1993); their value is 0.49 year-1.  They did not use age stanzas, and so their value implicitly includes younger age classes and should be higher.  The sub-adult production rate was set higher relative to adults at 1.1 year- 1, while the juvenile production rate was set at 1.47 year-1.  These production rates reflect the M estimate for 18 RA snapper species based on the empirical equation of Pauly (1980); but the values have been increased by 50% and 100% respectively to represent additional predation mortality incurred by the immature stanzas (as well as any fishing mortality).  These rates generate a realistic age-biomass distribution under the species-specific growth and mortality values obtained from FishBase, in which the majority of biomass is concentrated in the adult and sub-adult stanzas.  The consumption rate of adult snappers, 7.105 year-1 is based on the empirical equation of Pauly (1986); this uses species-specific information for 29 species of RA snappers out of 32, and represents an average species value.  It is slightly higher than the consumption rate used to model snappers in the Mexican Caribbean, 5.6 year-1 by Vidal and  Page 42, Fisheries Centre Research Reports 15(5), 2007 Basurto (2003).  Snapper catch is estimated from DKP and Trade and Industry Office statistics as 0.031 t·km-2.  This represents average catches between 2000-2005.  45% of the total snapper catch was allotted to the adult functional group; 45% was attributed to sub-adults and 10% to juveniles.  This catch quantity includes an estimate of unreported artisanal catch equal to 50% of the reported value.  Snappers are represented in the RA model as fully exploited, with an F2006 of 0.15 year-1, which is close to the Fmsy (0.21 year-1).  MSY is predicted to be 8.4 kg·km-2 for RA, or about 0.479 t·km-2 on coral reefs.  Napoleon wrasse  This functional group represents only Napoleon wrasse (Cheilinus undulatus), which is a conspicuously large reef fish species in family Labridae. It is also commonly referred to as humphead wrasse or double-headed Maori wrasse, among other names (Allen, 2000).  The functional group is divided into adult, subadult and juvenile stanzas.    A biomass value for this species could not be calculated based on the reef transects in COREMAP (2005) because Cheilinus is only reported to the genus level (four other Cheilnus are also present in the model in the medium and large reef associated groups).  However, Donaldson and Sadovy (2001) suggested that Napoleon wrasse is uncommon wherever it occurs, and Russell (2004) suggested a typical density of 10 fish per hectare in reef environments and a maximum density of 20 fish per hectare.  Since there is a heavy fishery on Napoleon wrasse in RA, we assume that the standing biomass should fall toward the lower end of that possible range.  Ten fish per hectare equates to 2 t·km-2 on reefs; and when corrected for reef area a possible overall biomass density in RA is determined as 0.035 t·km-2.  This amount was split into adult (33%), subadult (57%) and juvenile groups (10%) using the mortality schedule in Table A.3.3.  The P/B of adult Napoleon wrasse is set at 0.5 year-1.  It is based on the M regression formula of Pauly (1980), but the M value (0.25 year-1) was then doubled to estimate P/B and account for fishing mortality.  A similar P/B value was used for sub-adults, but juveniles were set higher at 1.2 year-1 to represent additional predation mortality suffered by the immature stanzas.  Sampling data for C. undulatus suggests that the natural mortality rate may be lower, 0.11 year-1 (Eckert, 1987).  However, the contribution of fishing mortality to total mortality is in question, and we have therefore made a precautionary assumption that F is at least equal to M.  Q/B rate for adults is set at 8.9 year-1, and the rates for immature stanzas were calculated according to the mortality schedule in place.  A consumption rate could not be found for C. undulatus, and so this value was designed to represent a slightly lower consumption rate than of groupers.  This is appropriate since Napoleon wrasse is among the largest reef-associated fish species, therefore consuming less per unit body mass.  This species is subject to a live reef food fishery supplying high value export product (Mous et al., 2000).  The export of Napoleon wrasse is regulated by CITES Appendix II, of which Indonesia is a signatory.  The fishery in RA is conducted primarily by surface air supplied divers who may use cyanide to stun the fish, and it is also pursued by reef bombing operations (Andreas Muljadi.  TNC-CTC.  Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413, pers. comm.).  Catch of Napoleon wrasse is estimated from DKP and Trade and Industry Office statistics.  It represents an average of the years 2000-2005.  The value was doubled from the official sources to represent unreported artisanal catch.  However, the total catch estimate remains small at only 2.07 kg·km-2.  It is divided between age stanzas: 45% was attributed to the adult functional group, 45% to sub-adults and 10% to juveniles.  With this small amount of catch, Ecosim predicts that F2006 equals 0.085 year-1, which is short of Fmsy (0.23 year-1).  MSY in RA is predicted to be 1.8 kg·km-2 for adults, and approximately 5.3 kg·km-2 for adults and sub-adults together.  This equates to 0.302 t·km-2 on coral reefs.  Bird’s Head Seascape Analyses, Page 43  Skipjack tuna  This group represents only Skipjack tuna (Katsuwonus pelamis).  They were allotted their own functional group because they are heavily exploited in eastern Indonesia and constitute a major commercial resource.  The species also exhibits faster growth and mortality rates than other major tuna stocks in the area, (e.g., yellowfin and bigeye tuna: Thunnus albacares and T. obesus), which are incorporated in the other tuna functional group.    Between the years 1998-2001, biomass of the western and central Pacific Ocean stock was thought to be at the highest levels in 30 years thanks to an upward shift in recruitment rates occurring during the mid-1980s (Langley et al., 2003), and El Ninõ events in the 1990s may have benefited Skipjack tuna recruitment as well (SCTB, 2004).  Skipjack biomass is now thought to lie above the level that produces MSY (BMSY) (SCTB, 2004).  Our RA model predicts that the values are close.  We have allowed Ecopath to estimate Skipjack tuna biomass as 0.699 t·km-2, while BMSY is predicted to be 0.765 t·km-2 by the equilibrium analysis.  There is no biomass accumulation entered.   A range of values are reported in the literature for skipjack tuna mortality rates, a summary is provided by Wild and Hampton (1994).  Those authors cite Bayliff (1977), who suggests an upper limit, 6.48 year-1, while the inter-American Tropical Tuna Commission assumes a lower mortality for management, between 1.39 and 2.30 year-1 (IATTC, 1989).  We assumed an intermediate value for total mortality, 2 year-1, which is applied as the P/B value for skipjack.  This estimate is similar to one derived from Pauly’s (1980) empirical M formula.  When applied, the estimate of M, 0.99 year-1, can be doubled to represent a fully exploited stock, where M=F.  The resulting P/B is 1.99 year-1.    Skipjack tuna received a high Q/B value of 32.57 year-1 from Pauly (1989), and we do expect a high consumption rate for fishes with high-performance physiology like tunas and billfish due to elevated metabolism rates that facilitate their pelagic-hunter niche (Magnuson, 1969; Brill, 1996).  However, Pauly’s (1989) value is very high compared to our aggregate group for large pelagics (5.644 year-1), indicating that skipjack are voracious predators.  The stock evaluated by Pauly’s (1989) in fact represents a Pacific stock at a lower temperature (24 oC) than Raja Ampat (28 oC), and so the consumption rate in RA may be higher still.  However, we have chosen to use a lower value, 6.64 year-1, so that production over consumption (P/Q) ratio approximate equals 0.3. This is a rule-of-thumb applicable to a fast growing pelagic species.  As a highly migratory pelagic species, we have assumed a large amount of diet import in the models (85%) and we have applied a low EE (0.42).  This represents the high rate of mortality caused by fisheries and predation elsewhere in the Pacific, external to the model.  The stock of western and central Pacific Ocean skipjack tuna is thought to be exploited at a modest level relative to its biological potential (Langley et al., 2003).  We have calculated a catch value of 0.347 t·km-2 based on DKP and Trade and Industry Office catch statistics - this represents an average of the years 2000-2005.  The catch record is relatively well documented for skipjack tuna, and so we assume zero unreported catch.  We also assume zero discards for this group.  The equilibrium analysis provides the following fishery indicators: F2006 = 0.548 year-1, Fmsy = 0.479 year-1, MSY = 0.366 t·km-2, predicted MSY for RA is about 16,400 tonnes.  The stock is assumed for management purposes to be contiguous throughout the eastern and central Pacific (Wild and Hampton, 1994).  Therefore, fishery catches elsewhere will affect the abundance of animals occurring in RA.  This limits our ability to predict stock dynamics for this group (see Martell, 2004 for a discussion on modelling migratory species in EwE).  Other tuna   Page 44, Fisheries Centre Research Reports 15(5), 2007 The ‘other tuna’ functional group includes 10 species of scombrids: wahoo (Acanthocybium solandri), bullet tuna (Auxis rochei rochei), frigate tuna (A. thazard thazard), Kawakawa (Euthynnus affnis), dogtooth tuna (Gymnosarda unicolor), albacore tuna (Thunnus alalunga), yellowfin tuna (T. albacares), bigeye tuna (T. obesus), Pacific bluefin tuna (T. orientalis) and longtail tuna (T. tonggol).  The current biomass of bigeye tuna is thought to lie above the MSY level (SCTB, 2004).  The biomass of albacore in the South Pacific may be (as of 2004) at approximately 60% of unexploited biomass B0, while the biomass of yellowfin in the western central Pacific Ocean may be 65-80% of B0 (SCTB, 2004).  Our biomass estimate of 0.604 t·km-2 was calculated by Ecopath by assuming a low EE of 0.4 for this migratory group.  That biomass is approximately 88% of the B0 predicted by the equilibrium analysis**.  A biomass accumulation rate of -5% per year is included.  The production rate P/B (1.408 year-1) is set according to Pauly’s (1980) empirical formula for M, which is calculated at the species level and doubled to represent the contribution of F.  P/B values were averaged for 8 species to provide an estimate for this group.  The value compares well with M estimates for T. albacares and T. obesus obtained from (Hampton, 2000), which average out to 1.2 year-1, once doubled to account for fishing mortality.  The Q/B value for other tuna was estimated using species-specific parameters based on 9 RA species and applying the empirical equation of Pauly (1986).  The original estimate 5.587 year-1 was reduced to 4.693  year-1, so that P/Q equals 0.3.    In RA, the fishery for tuna is primarily conducted by the pole and line fleet.  Catch of other tuna is represented from DKP and Trade and Industry Office statistics.  It was estimated to be very low from government statistics, 0.0263 t·km-2 - this represents an average of the years 2000- 2005.  We increased this amount by 80% to account for unreported catch and represent a fully exploited stock.  We did not include any additional discards.  This results in the following fishery indicators: F2006 = 0.746 year-1, Fmsy = 0.746 year-1, MSY = 0.058 t·km-2.  Predicted MSY for the whole of RA is therefore predicted to be about 2,610 tonnes.    Mackerel  The Mackerel group contains 9 species of scombrids identified in McKenna et al. (2002b) or reported as being present in the area by FishBase records.  The species included are Double- lined mackerel (Grammatorcynus bilineatus), Short mackerel (Rastrelliger brachysoma), Island mackerel (R. faughni), Indian mackerel (R. kanagurta), Blue mackerel (Scomber australasicus), Narrow-barred Spanish mackerel (Scomberomorus commerson), Australian spotted mackerel (S. munroi), Queensland school mackerel (S. queenslandicus) and Broadbarred king mackerel (S. semifasciatus).   Biomass of mackerel, 0.086 t·km-2, is based on an estimate obtained from the relative abundance rankings of McKenna et al., (2002b) for RA.  The species-level abundance rankings were converted to absolute biomass by applying weighting factors.  Weighting factors were calculated based on common species found in both the McKenna et al. (2002b) species list and the COREMAP (2005) biomass transects (see Section 2.5.8 - Biomass density estimates).  No biomass accumulation rate is entered for this group.  The P/B rate for mackerels was set according to the empirical formula for M of Pauly (1980), based on 9 mackerel species and using species-specific growth parameters available from FishBase.  The M morality rate was doubled to represent the contribution of F, so that P/B is set                                                         ** The equilibrium analysis presented in Appendix B shows a lower B0 for ‘other tuna’, 0.472 t·km-2, because it does not consider trophic interactions.  These increase the potential surplus production.  Bird’s Head Seascape Analyses, Page 45 at 2.913 year-1.  Species-level P/B values were averaged to provide an estimate for this group.  Independent mortality estimates from sampling could only be found for one RA mackerel species, Scomberomorus commerson, at 0.59 year-1 (McIlwain, 2005).  This is a low value, even when increased to account for fishing mortality, and it was not used for the group average.  Buchary (1999) used a higher P/B rate for Rastrelliger spp., 4.248 year-1.  Q/B was set at 9.712 year-1 so that the gross efficiency (P/Q) ratio equals 0.3.   The Q/B formula of Pauly (1986) suggested a slightly lower rate, 8.593 year-1, based on 10 mackerel species.  Buchary (1999) maintained a similar P/Q ratio (3.3) as in the present model.  Catch of mackerels was estimated based on DKP and Trade and Industry Office statistics (0.064 t·km-2); this represent average RA catches between the years 2000-2005.  We assume there is zero unreported catch in this group.  Under these assumptions, the equilibrium analysis suggests that the group is now fully exploited: F2006 (0.746 year-1) lies very close to Fmsy (0.746 year-1), while current catches are slightly above MSY (0.058 t·km-2).  Billfish  The billfish functional group includes highly migratory sailfish and billfish species: Indo-Pacific sailfish (Istiophorus platypterus), black marlin (Makaira indica), Indo-Pacific blue marlin (Makaira mazara), shortbill spearfish (Tetrapturus angustirostris), striped marlin (Tetrapturus audax) and swordfish (Xiphias gladius).    The biomass of billfish was estimated by Ecopath as 0.825 t·km-2 based on an assumed EE of 0.2.  This low EE value was used to represent a highly migratory species, where a large fraction of natural mortality (80%) occurs outside the modelled system.  A significant diet import term (approx. 88% of diet) was also included to represent feeding that occurs outside of RA.    The P/B rate of billfish (0.956 year-1) was set at the species level according to the empirical M formula of Pauly (1980), which was doubled to represent the contribution of F.  This value represents the average of 4 billfish species.  Q/B was set so that the P/Q ratio is equal to 0.3.  This assumption results in a Q/B value of 3.187 year-1, which is similar to the estimate derived from the consumption rate formula of Pauly (1986), 3.256 year-1 for 5 species of RA billfish.  There was no data available on billfish landings in the governmental fisheries statistics, and so we assume a small catch for billfish occurring in RA from trolling operations, including recreational fisheries.  A catch of 0.05 t·km-2 in the RA model (~5% of standing biomass) corresponds to an F2006 of 0.06 year-1, or about 40% of Fmsy (0.148 year-1) representing a lightly exploited stock.  MSY is estimated by the equilibrium analysis to be approximately 0.068 t·km-2, equivalent to almost 3,100 tonnes for RA.  Billfish biomass is depleted in the present-day RA model to approximately 75% of the pristine level (B0).  Coral trout  This functional group encompasses six species that are commonly referred to as coral trout: coral hind (Cephalopholis miniata), leopard coralgrouper (Plectropomus leopardus), blacksaddeled coralgrouper (P. laevis), spotted coralgrouper (P. maculates), highfin coralgrouper (P. oligocanthus) and squaretail coralgrouper (P. areolatus).    Coral trout biomass is based on reef transects conducted on Weigeo Island (COREMAP, 2005).  It is calculated to be 0.040 t·km-2, with about 93% of the biomass occurring in the adult group and the remainder in the juvenile group as determined by the multi-stanza routine using mortality parameters in Table A.3.3.  A biomass accumulation rate of -0.07 year-1 was entered to adjust the surplus production potential so that current (2006) fishing mortality lies close to Fmsy, representing a fully exploited stock.    Page 46, Fisheries Centre Research Reports 15(5), 2007  The P/B rate of coral trout is set at 0.35 year-1 for adults and 0.7 year-1 for juveniles.  The adult value is based on P. leopardus (ages 6-8) and P. laevis; it is an average of natural mortalities from sampling (Russ et al., 1998), and it has been increased by 50% to account for fishing mortality.  The juvenile production rate is based on a high value for total mortality (Z) found in the literature for P. maculatus (Ferrira and Russ, 1992), but it has been increased by 25% to account for additional predation mortality incurred by juvenile stanzas.  The M predicted by Pauly’s (1980) formula is 0.5 year-1 for two RA coral trout species, which falls between the values used for our life history stanzas.  Similarly, Gribble (2001) used 0.35 year-1 for coral trout on the Great Barrier Reef, which lies between the adult and juvenile estimates.  The parameters in use generate a realistic age-biomass distribution under assumed maturity parameters.  Coral trout Q/B was estimated from Pauly’s (1986) empirical formula as 6.1 year-1 and is based on 6 species.  This amount was ultimately decreased to 3.3 year-1 for the adult group during balancing in order to more accurately reflect the consumption rates of physiologically comparable groups, such as large reef associated fish.  The consumption rate for the juvenile stanza was estimated by the multi-stanza routine to be 8.393 year-1 based on the adult rate and the given mortality schedule.    Catch of coral trout was estimated directly from DKP and Trade and Industry Office statistics at about 1.8 kg·km-2, which falls just below MSY indicating a fully exploited stock.  If there are significant sources of unreported catch for coral trout, then this functional group may actually be overexploited.  The catch of coral trout is based on the ‘other’ reef fish catch category listed in DKP and Trade and Industry Office statistics. That quantity was divided among reef-associated functional groups whose catch was not quantified explicitly by other catch statistic categories.  The ‘other’ catch was divided between reef associated groups according to their relative number of species, assuming that the more specious groups contribute a greater fraction to the undetermined catch.  In the preliminary models we assumed zero discarding.  Equilibrium statistics are as follows: F2006 = 0.045 year-1, Fmsy = 0.092 year-1, MSY = 1.9 kg·km-2, or about 85.5 tonnes total catch for RA.  This is equivalent to 0.108 t·km-2 on coral reefs.  Large and small sharks  Large sharks include the grey reef shark (Charcharhinus amblyrhynchos), Pondicherry shark (C. hemiodon), blacktip reef shark (C. melanopterus), blue shark (Prionace glauca), whitetip reef shark (Triaenodon obesus) and tawny nurse shark (Nebrius ferrugineus).  Small sharks include the graceful shark (Carcharhinus amblyrhynchoides), Australian sharpnose shark (Rhizoprionodon taylori), smallfin gulper shark (Centrophorus moluccensis), Indonesian speckled carpetshark (Hemiscyllium freycineti) and tasseled wobbegong (Eucrossorhinus dasypogon).  The biomass of large sharks is estimated to be approximately 0.115 t·km-2 in the RA model based on the subjective species-level abundance ratings provided by McKenna et al. (2002b); the biomass of small sharks is estimated to be 0.057 t·km-2.  Biomass weighting factors were assigned to each qualitative abundance rating offered by McKenna et al. (2002b) (e.g., rare, “occasional”, “common”) based on quantitative values provided by COREMAP (2005) for certain species that were common to both lists.  A biomass density is extrapolated for species missing from the COREMAP list according to the subjective biomass rating in McKenna et al. (2002b), and the biomass of the functional groups large and small sharks are calculated as the sum of the biomasses of constituent species.  Group biomass is divided between the adult and juvenile stanzas for large sharks (56% and 44%, respectively) and small sharks (14% and 86%, respectively) according to the mortality schedule in Table A.3.3.    An EE of 0.5 was entered into adult large sharks to represent a migratory group that moves (and dies) outside the model boundaries.  With biomass, production rate, consumption rate and EE entered as input parameters, Ecopath was able to estimate the biomass accumulation rate to be  Bird’s Head Seascape Analyses, Page 47 9% year-1.  Small sharks were assumed to have more restricted ranges, and their EE is estimated by Ecopath to reflect a more sedentary nature (>0.95).  Since the major depletion of large sharks probably occurred before 1980 in RA, the group is not of major trophodynamic consequence in the present-day model. Nevertheless, these top predators have the potential to fulfill a keystone functional role, and so it is important that their dynamics are accurately represented in the model.  Forecasting scenarios will also require an accurate account in order to represent conservation interests.  By design, the biomass of large sharks in the 2006 RA model represents approximately 10% of B0, as determined using the equilibrium routine.  This should provide a realistic scope for growth in restoration studies.  The P/B ratio of adult large sharks was estimated based on Pauly (1980).  His empirical formula predicts an M equal to 0.64 year-1, which we can increase by 50% to represent fishing mortality (0.967 year-1). This value, based on 2 species, was used in the initial model, but it was subsequently increased to 1.1 year-1 during the process of balancing.  Juvenile sharks were set slightly higher, 1.3 year-1.  Polovina (1984) used a lower value for his ‘reef sharks’ group in the French Frigate Shoals, 0.18 year-1 and Opitz (1993) used 0.24 year-1 for her ‘large sharks/rays’ category in the Caribbean.  However, our higher P/B value is appropriate for a heavily exploited stock.  Consumption rate for the adult stanza was set at 3.6 year-1 in order to initialize the gross efficiency P/Q ratio at 0.3.  The juvenile Q/B was estimated based on the mortality schedule in Table A.3.3.  The production and consumption rates of small sharks were set relatively higher than large sharks, at 1.2 year-1 and 4 year-1, respectively for adult stanzas; juvenile small shark consumption rates were estimated by the multi-stanza routine.  Elasmobranchs are an important marine resource for many artisanal fishers in RA.  In fact, Indonesia has among the highest landings of chondrichthyans in the world (Stevens et al., 2000), yet this significant catch goes largely unregulated.  Although the gross catch of sharks is typically small compared to other oceanic resources such as teleosts (FAO 2005), fishing can have a major impact on the species group considering the slow growth rates of large sharks, in particular, and the low fecundity of viviparous species.  Trolling fisheries in RA pursue large sharks for their high value fins, and these animals are also likely to be taken as bycatch in pelagic fishing operations, although no records were found.  Catch of large sharks is estimated from DKP and Trade and Industry Office statistics to be 0.028 t·km-2, of which 90% is assumed to originate from the adult stanza.  78% of the catch is exported, according DKP statistics.  This value represents an average of the years 2000-2005.  There is also a minor bycatch of large sharks entered for the trolling fleet (0.001 t·km-2).  The catch statistics available from the government bureaus referred to ‘shark fins’.  We assumed that this catch was entirely attributable to the large sharks functional group, and we back-calculated the total amount of shark biomass required to provide that quantity of shark fins.  Under the assumed conversion ratio, 1 tonne of fins equates to roughly 24.1 tonnes of sharks.  This rough estimate is based on a remark made for Gulf of Mexico fisheries (P. Ortiz, National Oceanic and Atmospheric Administration, cited in Raloff, 2002).  Despite the positive biomass accumulation rate estimated by Ecopath in 2006, the large shark group stands as overexploited in the model, with F2006 (0.97 year-1) well in excess of Fmsy (0.48 year-1). The MSY of large sharks is estimated to be only 0.01 t·km-2, or 470 tonnes for the whole area of RA.  It is less than half of the current estimated catch.   Only a miniscule catch was estimated for small sharks from governmental fisheries statistics, although this value is highly uncertain.  A revised catch figure was therefore entered for small sharks that would represent a lightly exploited stock, where F2006 is approximately equal to 0.25 Fmsy.  The catch of small sharks is set at 6.24 kg·km-2 in the preliminary RA model.  We assume that small shark species constitute 50% of the domestic catch, while exported catch consists entirely of large sharks.     Page 48, Fisheries Centre Research Reports 15(5), 2007 Whale shark   This group represents the planktivorous whale shark, Rhincodon typus.  Little is known about the abundance of this animal in RA or the health of the stock.  However, sightings recorded by an ecotourism company operating in the Andaman Sea out of Phuket, Thailand suggests that whale shark abundance may have dropped by as much as 96% between 1998 and 2001 (Theberge and Dearden, 2006), although there are a variety of possible explanations.  Biomass was estimated by Ecopath to be 3.2 kg·km-2, providing an estimate of 143 tonnes of whale sharks in RA.  This suggests that there could be very few of these animals in the study area, especially considering that their maximum size may be as large as 36 tonnes per animal (Ritter, 2000), although Wmax is frequently cited as less than 20 tonnes per animal.  The P/B ratio for whale shark was entered in very approximately as 0.068 year-1, based on the empirical relationship for M offered by Pauly (1980), and assuming zero fishing mortality.  Q/B was set at 0.228 year-1 to establish a P/Q ratio of 0.3.  The value used for Q/B is preliminary.  It could be low considering the empirical formula of Pauly (1986) provides a much higher Q/B estimate, 2.022 year-1.  We have applied a very low EE of 0.025, but it is not clearly known the extent to which these animals migrate (Wilson et al., 2005; Colman, 1997). Similarly, an 80% diet import term was applied to represent the potentially wide-ranging habits of these individuals.  There is fishing for these animals throughout the world in countries such as Indonesia, India, Philippines, Pakistan, Iraq and other places (Theberge and Dearden, 2006; Colman, 1997 and references therein).  Catches in Philippines are thought to have declined in recent years, although the cause is unsure (Colman, 1997).  We have entered in a zero catch rate for whale shark and zero discards, pending better information.  Manta ray and Rays  The manta ray group includes the giant manta (Manta birostris).  The ray group, which is divided into adult and juvenile stanzas, represents 7 species of families Dasyatididae, Mobulidae and Myliobatidae: the bluespotted stingray (Dasyatis kuhlii), Bluespotted ribbontail ray (Taeniura lymma), Chilean devil ray (Mobula tarapacana), spotted eagle ray (Aetobatus narinari), painted maskray (Dasyatis leylandi), blackspotted whipray (Himantura toshi) and pygmy devilray (Mobula eregoodootenkee).  The biomass of manta rays is estimated by Ecopath to be low, only 3.166 kg·km-2.  This equates to about 142 tonnes in all of Raja Ampat.  The biomass of rays (0.177 t·km-2) was estimated based on subjective species-level abundance rankings offered by McKenna et al. (2002b).  Weighting factors were determined for each abundance ranking based on the estimated absolute biomass of certain species in common to both the McKenna et al. (2002b) species list and COREMAP (2005).  COREMAP (2005) abundance counts were converted to biomass density using an average species weight obtained from an age-structured model (see Section 2.5.8 - Biomass density estimates).  The P/B for manta rays is set slightly lower than that of rays, at 0.6 year-1.  The P/B for rays is set at 0.96 year-1 in order to establish a realistic age distribution as quantified by the multi-stanza routine.  The multi-stanza routine utilizes species-specific growth and maturity parameters from FishBase.  In the Java Sea, Buchary (1999) estimated a production rate for demersal rays of 1.3 year-1, which compares sufficiently well with our P/B estimate.  The Q/B rate for manta rays is set at 2 year-1 to provide a gross efficiency (P/Q) ratio of 0.3.  The Q/B rate for rays was estimated based on the empirical formula Pauly (1986) to be 3.817 year-1 (based on 5 species), but this was later decreased to 2.416 year-1 to reduce the P/Q ratio.  For comparison, Buchary (1999) used a consumption rate for demersal rays of 8.2 year-1, and Opitz (1993) used a value of  Bird’s Head Seascape Analyses, Page 49 4.9 year-1 for her ‘large sharks/rays’ group.  We assume zero catch of manta rays.  Ray catch is set at 0.021 t·km-2 based on the ‘other’ demersal catch category listed in DKP statistics. That quantity was divided among demersal functional groups whose catch was not quantified in a more precise catch category; those are rays and large demersals.  The ‘other’ catch was divided between rays and large demersal groups according to their relative number of species in each group, assuming that the more specious large demersal group contributed a greater fraction of the undetermined catch.  Butterflyfish  The butterflyfish functional group consists of 57 member species of family Chaetodontidae.  Fourteen genera are represented, but almost half the species in this group belong to genus Chaetodon.  This functional group was designed to capture the unique ability of butterflyfish to predate on sea anemones, although some species may also prey on coral polyps (anthozoids), invertebrates and plant material (Cox, 1994).  Biomass density was calculated for RA to be 0.325 t·km-2 based on 44 species counted in the reef resource inventory of COREMAP (2005).  To convert the coral reef biomass density to an average value for all of RA, we applied a correction factor based on the marine area to reef area ratio of Indonesia presented in Spalding et al. (2001).   Abundance counts are converted to biomass using an average individual weight obtained from an age-structured model (see Section 2.5.8 - Biomass density estimates).  This biomass was divided between adults (79%) and juveniles (21%) by implementing the multi-stanza mortality schedule in Table A.3.3.  The P/B rate for butterflyfish is set at 1.0 year-1 after a natural mortality estimate for Centropyge bicolor (Aldenhoven, 1986).  We determined a higher alternate value, 2.14 year-1, based on Pauly’s (1980) M formula for two RA species.  However, the higher value is not used because it leads to a left-skewed age-biomass distribution under species-specific growth and mortality rates available from FishBase.  The production rate of juveniles was set at 1.6 year-1.  A relatively high rate was required to resolve issues with over-predation of juveniles in the diet matrix.  The Q/B of adult butterflyfish was estimated to be 11.282 year-1 based on FishBase information for 49 RA species of butterflyfish.  However, this rate was ultimately reduced to 6.720 year-1 to reduce the P/Q ratio and allow the Q/B estimate to lie closer to the value for physiologically similar groups, such as medium reef-associated fish.  There is no explicit mention of butterflyfish in governmental fishery statistics, and it is likely that fishery on this group is minor compared to their substantial biomass.  These species are typically solitary, or occur in pairs, and do not tend to form large shoaling aggregations suitable for targeted fisheries.  This group is underexploited in the model.  Total catch for this group is estimated to be 0.017 t·km-2, of which 90% is directed at the adult stanza.  The following fishery indicators are estimated by the equilibrium analysis: F2006 = 0.06 year-1; Fmsy = 0.553 year-1; MSY = 0.079 t·km-2, or about 3550 tonnes for RA.  Cleaner wrasse  The cleaner wrasse functional group includes 3 labrids: the tubelip wrasse (Labrichthys unilineatus), the bicolor cleaner wrasse (Labroides bicolor) and the Bluestreak cleaner wrasse (Labroides dimidiatus).  Cleaner wrasse was given its own functional group to represent the cleaning mutualism effect seen between members of these species and larger reef fish that solicit their services.  The removal of ectoparasites, dead skin and other refuse is assumed to improve the health of reef fish.  This is represented in the BHS EBM models through mediation effects; adult groupers and snappers benefit from high biomass density of cleaner wrasse (see Section 2.2.2 - Mediation factors).  Page 50, Fisheries Centre Research Reports 15(5), 2007  The biomass of cleaner wrasse in the RA model, 0.009 t·km-2, is based on COREMAP (2005) reef resource inventory transects.  The value is adjusted for the relative reef area in RA using the reef area to marine area ratio for all of Indonesia (Spalding et al., 2001).  Abundance counts are converted to biomass using an average individual weight obtained from an age-structured model (see Section 2.5.8 - Biomass density estimates).    Production rate of cleaner wrasse (3.779 year-1) is set after the small reef associated fish group, as there was insufficient data available for cleaner wrasse.  The only independent sampling-based mortality figure located for this group refers to L. dimidiatus (Eckert, 1987).  The adult M is estimated as 0.11 year-1 and the juvenile M is 0.5 year-1.  These values are not used because they are too low compared to the Q/B estimate for this group.  The Q/B calculation used the empirical formula of Pauly (1986).  The figure, 13.097 year-1, is based on L. unilineatus and L. dimidiatus.    A minuscule catch for cleaner wrasse is incorporated, 0.819 kg·km-2·year-1.  The figure is based on the ‘other’ reef fish catch category listed in DKP and Trade and Industry Office statistics. That quantity was divided among reef-associated functional groups whose catch was not quantified explicitly by other catch statistic categories.  The ‘other’ catch was divided between reef associated groups according to their relative number of species in each group, assuming that the more specious groups contribute a greater fraction to the undetermined catch.  This data-poor group would benefit from further investigation.    Large pelagic fish  The large pelagic fish group consists of mainly piscivorous fish.  It is divided into adult and juvenile stanzas.  It is diverse and includes 25 species in the following familes: Belonidae, Bregmacerotidae, Chirocentridae, Coryphaenidae, Elopidae, Exocoetidae, Gonostomatidae, Hemiramphidae, Leiognathidae, Molidae, Myctophidae, Nettastomatidae, Polynemidae, Pristigasteridae, Salmonidae, Scombridae, Sphyraenidae, Stomiidae and Tetragonuridae.    The biomass of large pelagic fish (0.086 t·km-2) was estimated based on the subjective species- level abundance rankings offered by McKenna et al., (2002b).  Weighting factors were determined for each abundance ranking based on the estimated absolute biomass of certain species in common to both the McKenna et al. (2002b) species list and COREMAP (2005).  COREMAP (2005) abundance counts were converted to biomass density using an average species weight obtained from an age-structured model (see Section 2.5.8 - Biomass density estimates).  The biomass of this group was divided between the adult (63%) and juvenile (37%) stanzas according to mortality schedule in Table A.3.3.  The P/B rate of adult large pelagic fish is set at 0.8 year-1, slightly lower than tuna groups or smaller pelagic fish groups.  This rate was modified during balancing from the original value calculated with Pauly’s (1980) empirical formula for M, 1.079 year-1, which was increased by 50% to account for F.  The higher P/B rate, which is based on 9 species out of 26, is instead retained for the juvenile group.  Q/B is set at 2.667 year-1 to maintain a gross efficiency P/Q ratio of 0.3.  The empirical Q/B formula of Pauly (1986) predicted 5.644 year-1.  For comparison, Buchary (1999) used the following values to represent large pelagic predators in the Java Sea: P/B = 1.2 year-1 and Q/B = 8.65 year-1.  Catch of large pelagic fish are calculated based on the miscellaneous or generic catch categories listed in DKP statistics for domestic landings (e.g., other pelagics), and Trade and Industry Office statistics for exported landings (mixed, frozen and smoked fish).  These miscellaneous categories were divided among 8 aggregate groups in the model (all size classes of pelagic, reef-associated and demersal groups).  Total catch for large pelagics is estimated to be about 0.035 t·km-2, or about 186 tonnes for RA.  90% of the catch was assumed to originate from the adult stanza, and  Bird’s Head Seascape Analyses, Page 51 10% from the juvenile stanza.  The equilibrium analysis suggests that this group is fully exploited; current fishing mortality lies very close to Fmsy.  The following fishery indicators are estimated: F2006 = 0.575 year-1; Fmsy = 0.575 year-1; MSY = 0.023 t·km-2, or about 1030 tonnes for RA.  Medium pelagic fish  The medium pelagic fish group contains adult and juvenile stanzas for yellowtail barracuda (Sphyraena flavicauda), herring scad (Alepes vari), leaping bonito (Cybiosarda elegans), Hawaiian lady fish (Elops hawaiensis), slender suckerfish (Phtheirichthys lineatus), long tom (Strongylura krefftii), spottail needlefish (S. strongylura), largescale archerfish (Toxotes chatareus) and silvermouth trevally (Ulua aurochs).  The biomass of medium pelagic fish (0.029 t·km-2) is determined in the same way as large pelagic fish.  It is based on the subjective species-level abundance rankings offered by McKenna et al., (2002b), where an absolute biomass value is calculated based on representative species found in COREMAP (2005) species transects (see Section 2.5.8 - Biomass density estimates).  The biomass of this group was divided between the adult (40%) and juvenile (60%) stanzas according to mortality schedule in Table A.3.3.  The P/B rate of adult medium pelagic fish is set at 1.0 year-1, and the rate for juveniles is set at 1.5 year-1.  These values produce a reasonable age distribution in the multi-stanza routine for an exploited group under species-specific FishBase growth and mortality parameters, and they are in line with respect to physiologically similar groups.  Q/B for the adult group was set at 5.0 year- 1; the estimate was revised downward during balancing from the initial estimate based on Pauly’s (1986) empirical equation, 7.729 year-1.  The P/Q ratio is 2.   Catches for medium pelagic fish were estimated from DKP statistics for domestic landings based on the ‘other’ pelagic fish miscellaneous category, and from Trade and Industry Office statistics for exported landings based on a fraction of the miscellaneous catch categories (frozen, mixed and smoked).  These miscellaneous categories were divided among 8 aggregate groups in the model (all size classes of pelagic, reef-associated and demersal groups).  However, the catch of adult medium pelagic fish was ultimately reduced during balancing to 6.912 kg·km-2, which is 25% of the initial estimate.  Without this amendment, the fishing mortality of the group was predicted to be more than 10 times the predation mortality, which is unrealistic.  The juvenile catch estimate remains unaltered.  Therefore, the overall catch for this group is represented in the RA model as 9.984 kg·km-2 for both stanzas combined, with approximately 2/3 of that amount attributed to the adult group.  The following fishery indicators are estimated: F2006 = 1.383 year-1; Fmsy = 1.276 year-1; MSY = 0.023 t·km-2, or about 1030 tonnes for RA.   Small pelagic fish  The small pelagic fish are divided into adult and juvenile stanzas.  This group contains 75 species and 47 genera in the following families: Atherinidae, Bregmacerotidae, Carangidae, Centrolophidae, Champsodontidae, Clupeidae, Dentatherinidae, Exocoetidae, Gobiidae, Hemiramphidae, Lactariidae, Leiognathidae, Melanotaeniidae, Microstomatidae, Myctophidae, Nomeidae, Pristigasteridae, Pseudomugilidae, Scombridae, Scopelosauridae, Sternoptychidae, Stomiidae and Terapontidae.    The biomass of small pelagic fish (0.178 t·km-2) is determined in the same way as large and medium pelagic fish.  It is based on the subjective species-level abundance rankings offered by McKenna et al., (2002b), where an absolute biomass value is calculated based on representative species found in COREMAP (2005) species transects (see Section 2.5.8 - Biomass density estimates).  The biomass of this group was divided between the adult (40%) and juvenile (60%)  Page 52, Fisheries Centre Research Reports 15(5), 2007 stanzas according to mortality schedule in Table A.3.3.  P/B for small pelagic fish was estimated to be 3.99 year-1 based on an empirical formula for M of Pauly (1980); the value was increased by 50% to account for additional fishing mortality.  This is high compared to previously used values, and so we considered it an upper estimate of total mortality or production and it was applied to the juvenile stanza.  Rates used for similar groups in coral reef models are 1.1 year-1 (small pelagics; Polovina, 1984) and 1.8 year-1 (small schooling fish; Opitz, 1993).  The adult stanza P/B was instead set at 2.0 year-1.  Q/B was estimated using the empirical formula of Pauly (1986) as 18.462 year-1 based on 8 species, but this rate was reduced by about 30% during balancing to 13.267 year-1.    Catches for small pelagic fish were estimated from DKP and Trade and Industry Office statistics and represents an average of the years 2000-2005.  Miscellaneous catch categories (e.g., mixed fish, other pelagics) were divided among 8 aggregate groups in the model (all size classes of pelagic, reef-associated and demersal groups).  Total catch for small pelagics is 0.038 t·km-2, or about 1690 tonnes for RA.  90% of the catch was assumed to originate from the adult stanzas, and 10% from the juvenile stanzas.  The following fishery indicators are estimated: F2006 = 0.824 year-1; Fmsy = 1.154 year-1; MSY = 0.042 t·km-2, or about 1900 tonnes for RA.  Large reef-associated fish  Large reef-associated fish are divided into adult and juvenile stanzas.  This is the most speciose group in the RA model.  It represents 213 species (54 families and 111 genera) not elsewhere included in functional groups.  Since it is a large aggregate group, many life histories and feeding modes are implicitly represented.    Estimates vary greatly in the literature as to the biomass of large reef-associated fish on coral reefs.  Reef transect results from Weigeo Island in COREMAP (2005) lead to an estimate of 11.640 t·km-2; this value has been compiled at the species level and corrected for reef area to represent all of RA.  We divided biomass between the adult (55%) and juvenile stanzas (45%) according to mortalities in Table A.3.3.  Although we have entered this value into the preliminary RA model, it is worth noting that Weigeo may be less exploited than other areas in RA.  This value may therefore represent an upper estimate of large reef fish biomass.  It is high compared to the biomass value of large reef fish 3.5 t·km-2 for Caribbean reefs (Opitz, 1993; corrected for reef area ratio), 3.0 t·km-2 for NW Philippines reefs (based on a compilation of functional group data from Aliño et al., 1993) or 0.5 t·km-2 for the Gulf of Thailand (estimated from Khongchai et al., 2003).  However, the COREMAP value is in line with the biomass density used by Polovina (1984) for a large area surrounding French Frigate Shoals, 23 t·km-2; especially since his value could be reduced somewhat for accurate comparison with respect to species composition and relative reef area coverage.  Project outputs are expected to provide a better estimate of large reef-associated fish biomass.  A negative biomass accumulation rate of -0.15 year-1 was entered to reproduce the observed rate of decline seen in time series abundance estimates.  The P/B ratio of large reef-associated fish was set preliminarily as 0.4 year-1 for adults and 0.6 year-1 for juveniles.  The M formula of Pauly (1980) suggested a high natural mortality rate for RA large reef associated species of 1.29 year-1, to which we can likely add a sizable amount of fishing mortality.  An alternate M estimate for this group, 1.31 year-1, can be based on four member species of families Mullidae, Labridae and Siganidae (calculated from Macpherson et al., 2000; Eckert, 1987; Pajuelo et al., 1997; Ozbilgin et al., 2004; Kaunda-Arara et al., 2003).  However, these high values perturbed the age structure; they lead to a left-skewed distribution and were not used.  Q/B of large reef associated fish (4.0 year-1) was reduced significantly during balancing from an initial estimate based on the formula of Pauly (1986) of over 8.9 year-1.  Further investigation is required to parameterize this influential functional group.   Bird’s Head Seascape Analyses, Page 53 Large reef associated catch estimated from DKP statistics was about 0.069 t·km-2.  This amount was based on a compilation of catch statistic categories listed in governmental records including trevallies, breams, catfish and the ‘other’ unidentified reef fish category.  The latter category was divided between this EwE functional group and other reef-associated groups not explicitly mentioned by catch statistics, in a proportion equal to the relative number of species in each group.  The large reef associated group, having many species, garnered a large fraction (55%) of this undermined catch component.  However, compared with the adult large reef-associated fish biomass estimate from COREMAP (2005) of 6.368 t·km-2, the fishery did not appear to be a major source of mortality.  The calculated catch value is low even compared to Venema (1997), whose catch estimate for ‘coral fish’ in an adjacent area can be converted to 0.509 t·km-2.    However, it is likely that there is a large amount of unreported catch also occurring in this group.  The statistics recorded by the DKP and the Trade and Industry Office refer to fish landed in Sorong.  However, this group is targeted throughout the archipelago by commercial and artisanal fisheries.  Catches may go unreported even when landed in port (M. Bailey, UBC Fisheries Centre, 2202 Main Mall, Vancouver, Canada, pers. comm.).  Until we have a more formal estimate of unreported artisanal catch occurring in this group, we will make the precautionary assumption that 10% of the catch is recorded by government statistics.  Total landings for this group are therefore 0.690 t·km-2, which is applied to the adult stanza (80%) and juvenile stanza (20%).  For the adult large reef-associated group this yields a current fishing mortality approximately equal to 1/3 of Fmsy, representing a lightly exploited stock.  F2006 is determined to be 0.081 year-1, Fmsy is 0.178 year-1 and MSY is 0.343 t·km-2, or about 15,400 tonnes annually from RA.   Medium reef-associated fish  The medium reef-associated fish group includes 176 species, with 26 fish families represented and 79 genera.  The majority of species belong to three families: wrasses (Labridae), damselfish (Pomacentridae) and cardinalfish (Apogonidae).  The biomass of medium reef-associated fish is estimated based on COREMAP (2005) reef transect abundance counts to be about 5.2 t·km-2.  About 55% of biomass is concentrated in the adult stanza and the remainder is in the juvenile stanza according to the multi-stanza mortality schedule (Table A.3.3).  The biomass estimate for the RA model, determined on transect sites near Weigeo Island, has been scaled to represent the relative marine to reef area ratio in Raja Ampat after Spalding et al. (2001).  The P/B ratio of medium reef-associated fish (adults: 0.8 year-1; juveniles: 1.4 year-1) is set arbitrarily to a reasonable value, as there is not enough species-level information to apply an empirical formula.  Independent natural mortality estimates were located for three species in this group, Selaroides leptolepis, Stethojulis strigiventer and Istigobius decoratus (Torres et al., 2004; Eckert, 1987; Kritzer, 2001).  These values average out to 4.64 year-1, but this M value is high compared to the other reef fish groups in the model, and it is probably not representative of medium reef-associated fish.  The Q/B value was set at 5.0 year-1 for adult medium reef associated fish to maintain an intermediate value with respect to the small and large reef associated groups.    The catch of medium reef-associated fish, as estimated from DKP and the Trade and Industry Office statistics, was very low: only 0.027 t·km-2.  The estimate barely constitutes 2% of the standing stock biomass, and the resulting F is only 2% of M.  We have assumed, as with the large reef-associated group, that there is a substantial amount of unreported catch.  Catch of adult medium reef-associated fish was increased in the RA model to 0.3 t·km-2, such that F ≈ 1/4 M.  This represents a lightly exploited stock.  The assumption implies that the annual artisanal and unreported catch (and discards) could be as much as 10 times the reported landings.  However, the methodology used to estimate the catch of this group, based on an assigned fraction of miscellaneous catch categories in governmental statistics, is very approximate.  Therefore, the  Page 54, Fisheries Centre Research Reports 15(5), 2007 ‘reported’ catch figure is also highly uncertain.  Juvenile catches were increased in the same proportion as the adults with respect to the initial fishery estimate.  The following fishery indicators are estimated for this group: F2006 = 0.123 year-1, Fmsy = 0.4 year-1 and MSY = 0.824 t·km-2, or about 37,000 tonnes for RA annually.   Small reef-associated fish  The small reef-associated fish group is divided into adult and juvenile stanzas.  206 fish species are included in 22 families and 92 genera.  About 1/3 of the species in this group are gobies (Gobiidae); the other major familes are cardinalfish (Apogonidae), damselfish (Pomacentridae) and wrasses (Labridae).  The biomass of this group is estimated from COREMAP (2005) abundance transects as 0.394 t·km-2.  We converted the abundance counts into biomass using an average weight for each species, calculated with an age-structured model (see Section 2.5.8 - Biomass density estimates).  This value has been adjusted to represent average biomass density in RA using reef area to marine area ratio for all of Indonesia (Spalding et al., 2001).  The biomass value was split into adult (66%) and juvenile stanzas (34%) according to the multi-stanza mortality schedule (Table A.3.3).  This functional group is energetically less important to the system than the medium or large aggregate reef fish groups due to the size-based criteria used in assigning fish species to groups.  P/B of adult small reef associated fish is based on data for 4 member species (1 cardinalfish and 3 damselfish).  We applied the empirical relationship for M described by Pauly (1980) to estimate a production rate of 3.779 year-1.  We did not explicitly incorporate fishing mortality.  For comparison, Aliño et al. (1993) used 14.02 year-1 for gobies, 3.88 year-1 for cardinal fish and 3.3 year-1 for damselfish in modelling a reef flat in the Philippines.  Independent sampling estimates a production rate of 5.95 - 6.37 year-11 for one species of blenny present in RA (Salarias patzneri, Wilson, 2004); rightfully, this small species should fall higher than the group average.  Q/B for the adult stanza is set at 15.0 year-1.  It was reduced slightly from the estimate based on Pauly’s (1986) formula, 18.3 year-1 (based on 77 species out of 206) in order to keep the P/Q ratio (0.2) similar to other reef-associated groups.  As with large and medium reef associated fish, the catch figure derived from DKP and Trade and Industry Office statistics was low, 0.019 t·km-2.  This number was not used, and a higher catch figure was added to Ecosim (0.165 t·km-2), so that F ≈ 1/4 M.  This represents a lightly exploited stock.  Fishery values are as follows: F2006 = 0.579 year-1, Fmsy = 2.422 year-1 and MSY = 0.371 t·km-2, or about 16,700 tonnes for RA annually.  Large and small demersal fish  The large demersal group includes the following species: Japanese rubyfish (Erythrocles schlegelii), whipfin silverbiddy (Gerres filamentosus), gobies (Amblyeleotris arcupinna, Trimma griffthsi, T. halonevum), freshwater moray (Gymnothorax polyuranodon), barredfin moray (G. zonipectus), spotted armoured gurnard (Satyrichthys rieffeli), Japanese flathead (Inegocia japonica) and Jarbua terapon (Terapon jarbua).  The small demersal group includes: Ocellated waspfish (Apistus carinatus), cardinal fish (Apogon fleurieu, A. ocellicaudus), spotwing flying gurnard (Dactyloptena macracantha), blue speckled prawn goby (Cryptocentrus octofasciatus), wrasse (Choerodon zosterophorus), black-edged sweeper (Pempheris mangula), tuberculated flathead (Sorsogona tuberculata), freshwater demoiselle (Neopomacentrus taeniurus), insular shelf beauty (Symphysanodon typus) and threadfin blenny (Enneapterygius philippinus).     Bird’s Head Seascape Analyses, Page 55 The total biomass for large demersal fish, 0.415 t·km-2, is based on subjective species-level abundance rankings provided by McKenna et al., 2004.  The quantity is divided between adult (48%) and juvenile stanzas (52%) based on the mortality schedule in Table A.3.3.  A weighing factor was assigned to each abundance ranking based on species common to both the McKenna et al. (2004) list and COREMAP (2005).  The biomass for this group represents a sum of member species’ values; it is corrected for area to represent a RA average, using marine to reef area ratios in Spalding et al. (2001).  The biomass compares well to the estimate of Venema (1997), who quoted a biomass density in 1995 equivalent to 0.6 t·km-2, and whose estimate included other specific taxa incorporated here into other functional groups.  Biomass of small demersal fish (0.327 t·km-2) is calculated in the same way as large demersal fish.  It is similarly divided into adult (59%) and juvenile (41%) stanzas.    Pauly’s (1980) equation was used initially to determine natural mortality for large demersals as 1.69 year-1, based on a tigerfish and a silverbiddy.  This value is high compared to the one used by Buchary (1999) for modelling large demersal predators (0.92 year-1).  It also provided an unrealistic age distribution in the multi-stanza routine and so was not used.  A lower value is substituted for adult large demersals (0.6 year-1).  Buchary’s (1999) value was assumed to represent an upper limit for this group, and it was applied to the juvenile stanza.  Similarly, a value of 2 year-1 is set for small demersals, and Buchary’s (1999) value for small demersal predators is applied to the juvenile stanza (2.56 year-1).  Large consumption rates were estimated using the Q/B relationship of Pauly (1986): for large demersals, 8.42 year-1 and for small demersals, 18.5 year-1.  These figures are uncertain and they are only based on 2 and 1 species, respectively.  Buchary used lower values, 6.13 year-1 and 12.84 year-1 for large and small demersals, yet any of these produce unrealistic P/Q ratios.  Lower consumption rates are therefore in place: for large demersals (3.1 year-1) and for small demersals (8.6 year-1), providing P/Q ratios ≈ 0.2.  We estimate that there is a catch of approximately 0.029 t·km-2 for large demersal fish.  This is based on DKP statistics, and it represents an average of the years 2000-2004.  The figure combines recorded catch for croakers, threadfins and miscellaneous groups such as ‘other demersal fish’.  The figure has also been increased by 50% to represent unreported artisanal catch and scaled in proportion to relative reef area (from marine area ratios in Spalding et al., 2001) to provide an average RA estimate.  The fishery catch offered by Venema (1997) for an adjacent area in Eastern Indonesia can be re-stated as 0.297 t·km-2, when corrected for relative reef area in RA; it is an order of magnitude higher than the present estimate, and in fact higher than the standing biomass of the large demersal group.  However, this catch value considers species which have been placed into other functional groups in the RA models.  90% of the large demersal catch is assumed to originate from the adult stanza; the remainder is considered juvenile catch.  Small demersal catch is determined in the same way, as 0.032 t·km-2.  These catch values result in the following fishery indicators: for large demersals F2006 = 0.679 year-1, Fmsy = 0.561 year-1 and MSY = 0.040 t·km-2, or about 670 tonnes for RA annually; for small demersals, F2006 = 0.210 year-1, Fmsy = 2.868 year-1 and MSY = 0.247 t·km-2, or about 11,100 tonnes for RA.  Large demersals are therefore represented as being overexploited, while small demersals are underexploited.  Large and small planktivore fish  The large planktivore fish group contains 52 species (19 families and 31 genera); well- represented are planktivorous species of fusiliers, trevally, jacks, scads and soldierfish.  The small planktivore group contains 62 species (17 families and 39 genera); almost half of the species in this group are in family Pomacentridae (mainly damselfish and demoiselles), with some species of cardinalfish, blennies and gobies as well.  These groups are divided into adult and juveniles stanzas.  The planktivorous functional groups were created to represent an important trophic link on coral reefs, through which energy passes from planktonic secondary  Page 56, Fisheries Centre Research Reports 15(5), 2007 producers to the benthic reef fish community.  Obligate and facultative planktivorous species are included in the planktivorous functional groups. For a species to be included into a planktivorous functional group a prominent mention of planktivory is required in diet remarks on the FishBase Species, Ecology, or FoodItems table (see Section 2.4.2 - Planktivorous fish).    Abundance data available from reef transects at Weigeo Island lead to a very high biomass density for large planktivorous for the RA model, 9.56 t·km-2.  Under the current mortality scheme (Table A.3.3) this would amount to more than 5 t·km-2 of adult fish in RA (about 290 t·km-2 on reefs).  Although this group contains some abundant species, such as the red-bellied fusilier (Caesio cuning) (COREMAP, 2005) and the oxeye scad (Selar boops) (Obed Lense, TNC- CTC.  Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413, pers. Comm..), we considered this value to be too high for an RA average.  The adult biomass was therefore set arbitrarily to 1.0 t·km-2, and the juvenile stanza biomass was calculated by Ecopath as 0.89 t·km-2.  This may be a critical group in the trophic functioning of the RA ecosystem, and we hope that project outputs will allow us to improve this parameter.  The P/B ratio of adult and juvenile large planktivorous fish is determined based on the M relationship described by Pauly (1980).  The value represents an average for 17 RA species.  The calculated value, 2.0 year-1, was applied to the juvenile stanza, while the adult stanza received a lower value, 1.5 year-1.  These figures provided a suitable age-biomass distribution.  The small planktivore group uses a P/B rate of 2.0 year-1 for adult and juvenile stanzas.  Pauly’s (1980) M formula had been used to predict a P/B rate for small planktivorous fish of over 6.0 year-1 but when applied as a mortality rate, this high value produces a left-skewed age-biomass distribution.  The empirical equation of Pauly (1986) predicted a Q/B consumption rate of over 20 year-1 for small planktivorous fish.  This value was reduced substantially, so that P/Q equals 0.33.  Q/B of the adult stanza was similarly set to produce a P/Q ratio of 0.3.  Catch was estimated for large planktivores at a very low quantity from DKP and Trade and Industry Office statistics, less than 0.019 t·km-2.  However, this estimate is based on highly aggregated statistics and may be missing a large amount of unreported catch.  As this amount did not have any noticeable influence on the functional group in preliminary fishery simulations, the figure was discarded in favour of a larger value, 0.33 t·km-2, so that F ≈ 0.4M.  This quantity elicits a more reasonable response from the large planktivore group under a realistic variety of fishing pressures.  Small planktivore catch is set at 0.014 t·km-2.  For both large and small groups, 90% of the catch is assumed to originate from the adult stanza; the remainder from the juvenile stanza.  The following fishery indicators are determined: for large planktivores F2006 = 0.3 year-1, Fmsy = 0.7 year-1 and MSY = 0.478 t·km-2, or about 11,100 tonnes for RA; for small planktivores, F2006 = 0.031 year-1, Fmsy = 0.6 year-1 and MSY = 0.223 t·km-2, or about 10,000 tonnes for RA.  Large and small planktivores are therefore underexploited in the RA model.  Anchovy  The Anchovy group contains 17 Engraulids of genera Stolephorus, Thryssa, Setipinna and Thryssa; Stolephorus is the dominant genus by biomass in shallow habitats surrounding the Raja Ampat islands; especially important is S. indicus (Mark Erdmann.  CI.  Jl. Dr. Muwardi. 17 Renon Denpasar, Bali, Indonesia; Chris Rotinsulu.  CI.  Jl Arfak No. 45.  Sorong, Papua, Indonesia 98413, pers. comm.).  This species supports large artisanal fisheries throughout the RA archipelago.  A major artisanal fishery is located on southern Weigeo Island in Kabui Bay and surrounding areas.  Villagers export the anchovies for bait to northern Weigeo pelagic fisheries, or dry them for local consumption.  The large anchovy population is thought to be supported by a productive upwelling area in central Dampier Strait (M. Erdmann.  CI. Jl. Dr. Muwardi. 17 Renon Denpasar, Bali, Indonesia, pers. comm.).    Wolanski (2001) used an anchovy biomass of 3.122 t·km-2 for an inter-reef / lagoon Great Barrier  Bird’s Head Seascape Analyses, Page 57 Reef model.  As our study area contains a greater proportion of deep areas, the coastal anchovy species will have a lower average biomass density.  We have elected to use a smaller arbitrary value, pending better information.  Adult anchovy biomass is set at 1.5 t·km-2 and juvenile biomass is estimated by the multi-stanza routine, providing a total anchovy biomass of 3.737 t·km-2.  Fishers in Kabui Bay indicated that there has been a recent reduction in the available biomass of anchovies (M. Bailey, UBC Fisheries Centre.  2202 Main Mall, Vancouver, Canada,  pers. comm.).  A negative biomass accumulation rate of -0.2 year-1 was entered to represent this and set baseline surplus stock production so that Fmsy is approximately 2-3 times greater than the current fishing mortality (Fig. B.2.1).  The P/B ratio entered for anchovy is 3.37 year-1, based on M of S. indicus (Torres et al., 2004).  Pauly’s (1980) equation predicts a similar value, M = 3.27 year-1, averaged for eight RA anchovy species present in the model, while an average of 5 world engraulids from independent sampling studies yields M = 5 year-1 (Torres et al., 2004).  These values could be increased somewhat to represent fishing mortality.  However, a greater mortality value than the one used results in an unrealistic left-skewed age-biomass distribution under the estimated maturity parameters.  Moreover, other authors have used even lower production rate values for anchovy, such as Heymans et al., 2004 (1.2 year-1 for Benguela upwelling) and Ainsworth et al., 2001 (1.15 year-1 for Bay of Biscay).  Q/B rate for anchovy (14.625 year-1) is estimated from 9 RA engraulids using the regression relationship of Pauly (1986).  A rough estimation of anchovy catch based on reported catch rates from Waisai fishers by Bailey et al. (this volume) suggests an unreported catch rate of 0.401 t·km-2 for all Weigeo fisheries.  Increasing that value by 20% to consider other area in RA, and adding the official fraction of reported catch (0.028 t·km-2), leads us to an overall catch estimate of 0.509 t·km-2 for RA, which we apply entirely to the adult stanza.  Here we have assumed that Weigeo fisheries constitute the large majority of RA anchovy catch, as is the consensus among field-based researchers.  This provides a similar estimate to an independent catch calculation based on data in Venema (1997).  They reported a catch rate in 1993 of 70 thousand tonnes of small pelagics from Area III.3 (Ceram, Maluku and Tomini), or 1.829 t·km-2.  Assuming that anchovies comprise 30% of this catch we may expect 0.549 t·km-2, which is close to our current estimate.  The following fishery indicators are estimated for anchovy: F2006 = 0.391 year-1, Fmsy = 1.218 year-1 and MSY = 0.887 t·km-2, or almost 40,000 tonnes for RA.    Deep-water fish  Deep-water fish is a large aggregate group representing 58 species that occur at depths greater than 200 m, on the shelf slope or deeper.  Bathypelagic and bathydemersal species were included in this group based on their habitats listed in the FishBase Species table ‘Habitat’ field, and on the depth range reported in the ‘DepthRangeDeep’ field.  This group summarizes data for 26 fish families, but almost half the species in this group belong to families Myctophidae (lanternfishes) and Stomiidae (dragonfishes).  No species of deep-water fish were identified in the available surveys for RA.  However, the rough estimate for adult stanza biomass, 0.6 t·km-2, provides an appropriate EE value (~0.9).  With juvenile biomass estimated by the multi-stanza routine, overall biomass for deep-water fish is set at 1.394 year-1.  The production rate used is 1.13 year-1, and it is based on Myctophum asperum (Palomares and Pauly, 1998).  This value is significantly lower than the natural mortality estimate made using Pauly’s (1980) M formula, 3.94 year-1 based on six RA deepwater species.  The higher value is not used because it results in a left-skewed age-biomass distribution under the assumed maturity parameters.  An alternate estimate for myctophids is 0.91 year-1, which is based on 5 world species (Palomares and Pauly, 1998).  Q/B (3.667 year-1) is set relative to the production rate, so  Page 58, Fisheries Centre Research Reports 15(5), 2007 that the gross efficiency P/Q ratio equals 0.3.  The catch estimate for deep-water fish is based on the entry for hairtails in DKP statistics, as there are 4 species of family Trichiuridae in the models and they all occur in this group.  The figure represents an average of the years 2000-2004.  We assume zero unreported catch and zero discards.  The catch estimate (9.2 kg·km-2) is divided between adult (90%) and juvenile (10%) stanzas.  Ecosim estimates the following fishery indices: F2006 = 0.034 year-1, Fmsy = 0.450 year-1 and MSY = 0.115 t·km-2, or about 5100 tonnes for RA.  Macro-algal browsing  Herbivorous fish in the RA models are divided into three functional groups according to how severely they impact the substrate.  Most damaging are the eroding grazers, followed by scraping grazers and then macro-algal browsing fish.  Macro-algal browsing fish represent an important functional link in coral reef ecosystems.  Together with sea urchins, they regulate the biomass of algae, and may help coral recruits to settle by keeping the substrate exposed (Mous and Muljadi, 2005).  The group represents three herbivorous species: Piaractus brachypomus, Nematalosa erebi and Valamugil buchanani.  Local abundance data is expected in late 2006 for some herbivorous species on reefs from Kofiau Island transects, but it was not available for this report.  No species in this group were identified by the available RA surveys, and so a preliminary biomass value of 0.25 t·km-2 was entered.  It is a relatively low value for this selective group of 3 species.  With juvenile biomass estimated by the multi-stanza routine, total biomass of macro-algal browsing fish is estimated to be 0.75 t·km- 2.    P/B was set at 1.339 year-1 using the M formula of Pauly (1980) based on V. buchanani and P. brachypomus.  Juvenile P/B is set at 1.4 year-1.  Q/B was estimated for all species as 13.76 year-1 using the empirical formula of Pauly (1986).  A small catch was entered in for macro algal browsing fish of about 8 kg·km-2.  This is a small fraction of the large reef associated group catch estimated from the DKP and Trade and Industry Office statistics.  The amount is proportional to the relative number of species occurring in each group.  This group is underexploited in the model.  The fishery indicators for this group are: F2006 = 3E-3 year-1, Fmsy = 0.25 year-1 and MSY = 0.033 t·km-2, or about 1500 tonnes for RA.  Eroding grazers  This group consists of green humphead parrotfish (Bolbometopon muricatum) and the steep head parrotfish (S. microhinos).  The biomass estimate for eroding grazers is derived from COREMAP (2005) reef transects on Weigeo Island, 0.783 t·km-2.  The value has been scaled down based on the relative reef area in RA using ratios in Spalding et al. (2001).  That amount is divided between the adult (67%) and juvenile stanzas (33%) according to the mortality schedule in Table A.3.3.  The abundance data refers to Bolbometropon spp. but we assume the majority of biomass is accounted for by the member species of this group.  The production rate of eroding grazers is based on B. muricatum.  The value, 0.435 year-1 represents M from Pauly’s (1980) equation, but it has been increased by 50% to account for fishing mortality.  Juvenile P/B is set at 1.0 year-1.  We estimated the Q/B of B. muricatum as 4.319 year-1 using the formula of Pauly (1986), but subsequently we accepted a lower value of 1.45 year-1 so that P/Q equals 0.3.    A small amount of catch was entered in the base model representing a fraction of the estimated large reef associated fish catch.  The proportion is set based on the relative number of species in  Bird’s Head Seascape Analyses, Page 59 each group.  The fishery indicators for this group are: F2006 = 3E-3 year-1, Fmsy = 0.25 year-1 and MSY = 0.056 t·km-2, or about 2500 tonnes for RA.  Scraping grazers  Scraping grazers include 82 species of parrotfish (family Scaridae), surgeonfish and unicornfish (Acanthuridae) and filefile (Monacanthidae).  Scraping grazers were given their own functional group in the RA models to represent the important role that these animals play on coral reefs (Bellwood et al., 2004).  The biomass estimate for eroding grazers is derived from COREMAP (2005) reef transects on Weigeo Island, 2.004 t·km-2.  The value has been scaled down based on the relative reef area in RA.  It is divided between adult (17%) and juvenile stanzas (83%) according to the mortality schedule (Table A.3.3).    The P/B ratio is based on Pauly’s (1980) M relationship, and averages 18 fish in the scraping grazers group.  The value, 2.339 year-1 has been increased by 50% to account for additional fishing mortality.  Juvenile P/B is set at 3.0 year-1.  The Q/B value is was estimated from the equation of Pauly (1986) as 12.74 year-1 based on 50 species.  There is a small amount of catch entered for scraping grazers, 0.025 t·km-2.  This represents a fraction of the large reef associated catch proportional to the relative number of species in each group.  Scraping grazers are underexploited in the model.  The fishery indicators for this group are: F2006 = 0.094 year-1, Fmsy = 0.7 year-1 and MSY = 0.092 t·km-2, or about 4100 tonnes for RA.  Detritivore fish  Seven detritivorous fish species were categorized into this group.  To qualify, a prominent mention of detritivory is required for the predator species in the FishBase ‘Diet’ table, or in the ‘Species’ table comment field.  Although many species consume detritus incidentally, or as a minor diet component, only species that rely on detritus as a main food source were included in these functional groups.  Detritivory is based on the FishBase ‘Species’ table ‘comment’ field and ‘Ecology’ table ‘Herbivory2’ field.    The detritivorous fish group contributes relatively little to the overall reef fish biomass in the RA model, as it has been noted that invertebrates, not detritivorous reef fish, are primarily responsible for energy cycling in the ecosystem (A. Muljadi.  TNC-CTC.  Jl Gunung Merapi No. 38, Kampung Baru, Sorong, Papua, Indonesia 98413, pers. comm.).  Reef transects confirm a low biomass of these species, 0.016 t·km-2 (COREMAP, 2005).  This value has been scaled to represent the average of RA.    The P/B rate of detritivorous fish is set at 2.339 year-1; it is set equal to the value used for scraping grazers.  No addition mortality information could be found for these species.  A Q/B value was estimated using Pauly’s (1986) empirical equation, 11.86 year-1, but this was reduced to 8.33 year-1 so that P/Q would lie closer to 0.3.  A small catch is entered for this group in the base model as a fraction of the large reef associated group catch.  The proportion represents the relative number of species in each group.  The group is lightly exploited in the model; F is about 6% of M.  Azooxanthellate corals  Azooxanthellate corals are consumers.  We have allowed Ecopath to estimate their biomass in the system based on the assumption that EE ≈ 0.95.  Biomass is estimated to be 0.6 t·km-2.  A production rate of 1.44 year-1 was used because it is two-thirds of the value used for reef-building corals, and a Q/B rate of 3.6 year-1 provides a P/Q ratio of 0.4.  Page 60, Fisheries Centre Research Reports 15(5), 2007  Hermatypic scleractinian corals  These are the reef-building scleractinian corals.  Hermatypic scleractinian corals are modelled as facultative consumers because they predate on zooplankton, yet also have endosymbiotic zooxanthellae, autotrophic dinoflagellates that provide photosynthetic products to the coral.  There have been a wide range of parameters applied to coral functional groups in previous EwE studies (Table 2.6).  We use a biomass value calculated from Crossland et al. (1991), who suggested a global coral reef biomass density in the range of 10-100 gC·m-2.  Their mean reported value was used, 30 gC·m-2.  This amount was converted to wet weight using a carbon to carbohydrate conversion factor, which we assumed to equal animal dry weight, and a dry to wet weight conversion factor from Atkinson et al. (1984).  This calculation gives a biomass estimate of about 50 t·km-2 on reefs.  Corrected for the relative reef area in RA using the marine to reef area ratio for Indonesia reported by Spalding et al. (2001) gives an overall biomass value for the study area of 0.875 t·km-2.  We hope to replace this approximate value with a better estimate from RA coral cover estimates.  The production rate of reef building corals was calculated from Crossland et al. (1991).  They estimated a daily turnover rate of reef biomass on the order of 0.003 day-1.  This equates to 1.095 year-1.  The Q/B was set at 3.6 year-1, giving a high P/Q ratio of 0.6.  This is appropriate for a facultative consumer.  Table 2.6 - Biomass and production rates used previously in EwE to represent reef-building corals.  Area Biomass (t·km-2) P/B (·year-1) Original group name Source Central Java 17.48 0.1 "Living bottom structure" Nurhakim, 2003 Mexican Caribbean 1-30 0.7-1.8 "Sessile animal feeders" Arias-González, 1998 Java Sea 20 0.1 "Living bottom structure" Buchary, 1999 Caribbean 1000 0.8 "Sessile animals" Opitz, 1993 Bolinao, Philippines 200 0.1 "Sessile invertebrate consumers" Aliño et al., 1993 Hong Kong 0.399 1.09 "Corals" Buchary, 1999 Tiahura, Moorea Island, French Polynesia 19.74 1.92 "Corals" Arias-González, 1997 New Caledonia 1.47 1.47 "Corals/zooxanthellae" Bozec et al., 2004 French Frigate Shoals 289 3 "Heterotrophic benthos" Polovina, 1984 Great Barrier Reef  0.04 ____ Sorokin, 1981 World averages  1.095 ____ Crossland et al., 1991    Bird’s Head Seascape Analyses, Page 61 The loss rate of hard corals from destructive fishing methods and other stressors is not well known, but one estimate from Bolinao, Philippines supposes that 0.4% of the live coral cover is lost each year (McManus et al., 1997); and cyanide fishing is also known to have caused damage to reefs in Raja Ampat.  The damage arises from the toxin’s direct contact on coral polyps, and also from the action of divers breaking coral away to retrieve the stunned fish.  The loss of coral can cause major changes in the reef ecosystem, and it has been associated with a decline in fish biodiversity (Wilson et al. 2006).  A range of possible values for coral loss were identified for Indonesia: a conservative 0.05-0.06% per year estimate to 0.5-0.7% per year (Mous et al., 2000); the authors note that this is a small possible rate of loss compared to coral re-growth rates.  We therefore enter a biomass accumulation rate into this group of -0.5% per year for the 2006 RA model, which has a minimal impact on dynamics.  Non reef-building scleractinian corals and soft corals  Like the hermatypic sceractinian corals, the non-reef building ahermatypic scleractinian corals and the soft corals are both facultative consumers containing symbiotic zooxanthellae.  Until we can find or produce more accurate information for RA, we have allowed Ecopath to estimate the biomass of these groups based on the assumption that EE ≈ 0.95.  For both groups, this produces a biomass value very close to 0.6 year-1.  The production rate of non-reef building scleractinian corals is set arbitrarily at 1.4 year-1, slightly lower than azooxanthellate corals.  Penaeid shrimps and Shrimps and prawns  Biomass of penaeid shrimp and shrimps and prawns is set as 2.0 t·km-2 respectively pending better information.  The P/B value used for penaeid shrimps in the RA model is 3.824 year-1. This value was calculated as the average of the P/B values calculated for 4 species. The P/B for Penaeus duorarum (an Atlantic species), and Metapenaeus monoceros was calculated using Brey’s (1995) equation. The maximum age and maximum weight for Penaeus duorarum is 1.17 year and 38.05 g (Bielsa et al, 1983). Maximum age for Metapenaeus monoceros was estimated by Srivatsa (1953) as 1.59 year. The maximum weight is calculated based on length-based relationship of Abdurahiman et al. (2004).  P/B values of 5.245 year-1 and 3.83 year-1 were used for the average based on Trachypenaeus fulvus and Parapenaeus longipes (Pauly et al., 1984).  For comparison, Buchary (1999) used a P/B ratio for penaeid shrimps of 5 year-1   , which agrees, but Pauly et al. (1993) calculated a higher ratio for shrimps, equivalent to 18.2 year-1.    The Q/B for penaeid shrimps is taken as the average value for Penaeus longistylus and Penaeus esculentus from an Ecopath model of Great Barrier Reef (Gribble, 2003), 37.9 year-1.  Buchary (1999) had used a Q/B ratio of 28.95 year-1 for her adult penaeid shrimp group.  The value for other prawns from Gribble (2003) was used as the Q/B for shrimps and prawns in the RA model, 20 year-1.  The Q/B estimate of Schwamborn and Criales (2000) for juvenile pink shrimp (Farfantepenaeus duorarum) (an Atlantic species), might also be applicable to this group at 48.976 year-1.  The value is likely too high for our use as it refers to juvenile animals.  Pauly et al. (1993) suggested a Q/B ratio for shrimps of 28.94 year-1, which is more in line with our estimate.  Squid and Octopus  The P/B value used for squid in the RA model is 4.348 year-1.  This value was calculated as the average of the P/B value calculated for 7 species using Brey’s (1995) equation.  Maximum age and weights for Photololigo chinensis, Photololigo edulis, Sepioteuthis australis and Sepioteuthis lessoniana is obtained from (BRS, 1999). Maximum ages for Loligo duvauceli and Loligo chinensis is from Jackson (2004) and maximum weights were obtained from Kongprom et al. (2003).  Maximum age for Sepia officinalis is from Zielinski and Portner (2000) and  Page 62, Fisheries Centre Research Reports 15(5), 2007 maximum weight is from FAO (2006).  The production rate used by Buchary (1999) for cephalopods in the Java Sea is slightly lower at 3.1 year-1; the rate used by Optiz (1993) for a Caribbean reef was also 3.1 year-1.  Our value is intermediate though compared to the P/B rate for squid calculated by Pauly et al. (1993) for a Philippines reef, which equates to 10.66 year-1.  The P/B ratio used for octopus in the RA model is 2.327 year-1.  This value was calculated using Brey’s (1995) equation based on the maximum age of Octopus cyanea (Cascorbi, 2004) and maximum weight obtained from FAO (2006).  Pauly et al. (1993) used a higher value for octopus, 4.49 year-1.  The Q/B rate used by Buchary (1999) for cephalopods was 20.318 year-1; the Q/B rate used by Opitz was 11.7 year-1.  The Q/B value used in the RA model is intermediate, at 14.792 year-1.  This value was calculated as a weighted average according to the biomasses of three species: Sepioteuthis lessoniana, Sepia officinalis and Sepiola affinis.  Q/B is based on food intake studies of these species by Wells (1996).  Q/B for octopus was also calculated from the same source to be 13.24 year-1; this value was the average for 5 species: Eledone moschata, E. cirrhosa, Octopus cyanea, O. dofleini, O. maya and O. vulgaris.  An alternative estimate of Q/B is 10.95 year-1 based on Sepioteuthis lessoniana from food intake studies (Rodhouse and Nigmatullin, 1996).  This value was not used in the calculation of the group Q/B estimate.  Pauly et al. (1993) used the following Q/B values for the Bolinao Reef Ecosystem in the Philippines: squids, 16.64 year-1, octopus 7.3 year-1.  Sea cucumbers  Sea cucumber biomass was estimated for the RA model based on reef transect results provided in COREMAP (2005).  The average number of individuals on the Weigeo Island reef top was 2.3 individuals per 40 m2.  We converted this density to weight by assuming an individual animal weight of 965 g, as calculated from Desumont (2003) based on 20 sea cucumber species occurring in Papua New Guinea.  Total biomass density for sea cucumbers in RA then is 0.971 t·km-2, when corrected for reef area using ratios in Spalding et al., (2001).  This amount equates to 55.5 t·km-2 biomass density on coral reefs.  For comparison, the biomass of sea cucumbers was identified by Aliño et al. (1993) as 35.77 t·km-2 on coral reefs.  However, Trobe-Bateman et al., (2004) provided a much higher estimate for sea cucumber biomass in Papua New Guinea.  Using the same assumptions regarding average individual weight, their biomass density on reefs computes to 221.95 t·km-2, which is about four times higher than the RA model estimates currently in place.  The P/B rate used for sea cucumbers, 0.74 year-1, represents an average value for Actinopyga echinites and Holothuria scabra (Shelley, 1985).  Aliño et al. (1993) suggested a higher value for sea cucumbers, 4.45 year-1, and Pauly et al. (1993) suggested 2.66 year-1.  The Q/B value used in the RA model for sea cucumbers is 8.248 year-1. The value is an average of the individual consumption to biomass ratios calculated for 20 species using an empirical model by Cammen (1980). The average weight of sea cucumber species was obtained from species identification cards issued by Papua New Guinea National Fisheries Authority. The average weight was converted to dry weight using a factor 0.11 (Brey, 2006).  Pauly et al. (1993) used a lower Q/B value, 3.58 year-1.  Lobsters  Lobster biomass was estimated for the RA model to be 0.219 t·km-2 based on reef transect results provided in COREMAP (2005). The average number of individuals in reef top transects is 0.5 individuals.  Reef top transect area is 40 m2.  This gives an average density of 0.0125 individuals·m-2.  This density was converted to weight using the average individual weight of 1 kg.  Biomass density on reefs was corrected for reef area based on Spalding et al. (2001).   Bird’s Head Seascape Analyses, Page 63  The P/B value used for squid in the RA model is 0.446 year-1. This value was calculated using Brey’s (1995) equation as the average P/B value for 4 species.  Maximum age and weights for Panulirus ornatus, Thenus spp., Jasus verreauxi and Panulirus cygnus were obtained from (BRS, 1999).  The Q/B value of 15.207 year-1 was calculated from a consumption estimate of 0.1151 gC·m-2·y-1 and a biomass estimate 0.076 gC·m-2 obtained from Florida Bay (Jorgensen et al., 1991).  Large and small crabs  In the Java Sea, Buchary (1999) estimated the biomass of crustaceans to be 0.86 t·km-2.  This value is equally distributed among the three crustacean groups (lobsters, large crabs and small crabs) to obtain the biomass estimate of 0.286 t·km-2 for each.  The P/B for large crabs is calculated to be 1.24 year-1. The value is the average of P/B values for 3 species (Portunus pelagicus, Ranina ranina and Scylla serrata) determined using Brey’s (1995) equation. The maximum weight and age for the species are from BRS (1999).  The P/B for small crabs was calculated as 2.610 year-1. The value is the average of P/B values for 8 species (Uca rapax, U. maracoani, U. cumulanta, U. vocator, Eurytium limosum, Emerita analoga, Pachygrapsus gracilis and Ucides cordatus) from Brey (2006).   The Q/B values of 14.55 year-1 and 20.21 year-1 are calculated for large and small crabs respectively from consumption and biomass estimates in Jorgensen et al. (1991); data is from Florida Bay:  The annual catch made on large and small crabs represents a very rough approximation (2.76 kg·km-2 each).  It was determined by splitting evenly the quantity of ‘other’ catch reported in DKP and Trade and Industry Office between several groups which were not explicitly recorded in other catch categories.  The statistics represent catch in the years 2000-2005.  We are awaiting improved estimates for catch.  Crown of thorns starfish  This functional group contains only the crown-of-thorns starfish (Acanthaster planci), which is a highly influential (keystone) predator species on coral reefs (Pearson, 1981; Moran, 1986).  Biomass for crown of thorns starfish is taken from COREMAP (2005).  They estimated 0.5 individuals on average per 40 m2 of reef top.  Assuming that each animal weighs 1 kg provides a biomass estimate on reefs of 12.5 t·km-2.  Scaling this by the marine area to reef area ratio of Spalding et al., (2001) provides an estimate for RA of 0.218 t·km-2.    Evidence of coral damage from crown-of-thorns starfish in Raja Ampat was weak according to an earlier survey.  McKenna et al. (2002a) observed coral damage in only 6.7% of the sites surveyed in RA.  The P/B value used in the RA model is 0.463 year-1. This value was calculated using Brey’s (1995) equation. The maximum age used was 8 years (Zann, et al., 1990) and maximum weight was calculated using the diameter to weight equation in Birkeland and Lucas (1990). The maximum size was 60 cm from Moran (1990).  The Q/B value is based on consumption by juveniles and adults (Jangoux, 1982); the value is weighted according to the relative percentages of juveniles (23.5%) and adults (76.5%) in the population (Engelhardt et al., 2000) and the average body weight of juveniles and adults (Bass and Miller, 2006).    Page 64, Fisheries Centre Research Reports 15(5), 2007 Giant triton  The giant triton biomass was assumed to be 1% of the bivalve biomass estimated for the RA model and was fixed at 0.05 t·km-2.  The P/B value calculated for epifaunal detrivorous invertebrates was also used for giant triton in the RA model. This value is equal to 1.224 year-1. A P/Q ratio of 0.3 was assumed as this is a slow growing species; the Q/B ratio was thus calculated as 4.08 year-1.  It has been suggested that removal of predators such as the giant triton (Charonia tritonis) may play a role in the periodic crown-of-thorn outbreaks that threaten coral reefs, although the evidence is not conclusive (Sweatman, 1995).  Herbivorous echinoids  Herbivorous urchins probably have a more influential affect on algal cover in the coral reef environment than do herbivorous fishes such as parrot fishes (Scaridae) and surgeon fishes (Acanthuridae) (Levinton, 1982).  The biomass for herbivorous echinoids is based on COREMAP (2005) estimate of 3.3 sea urchins per reef-top transects for sites near Weigeo Island.  Assuming an average weight of 0.5 kg per animal, and correcting for reef area in RA based on the marine area to reef area ratio for all of Indonesia (Spalding et al., 2001), provides a herbivorous echinoids biomass estimate for RA model of 0.722 t·km-2.  The biomass on coral reefs alone is then approximately 41.25 t·km-2, which compares favorably with the value used by Aliño et al., (1993).  They assumed a sea urchin biomass on Philippine reefs of 35.77 t·km-2 based on unpublished data cited therein.  However, for a Caribbean reef, Opitz (1993) assumed a very high biomass value for echinoderms of 600 t·km-2.  The P/B for the RA model is 0.541 year-1. A histogram of the average test diameter of 20 sea urchin species was plotted and 5 cm was found to be the most frequent value for test diameter.  The test diameter was converted to weight using the relationship W= 0.247·D-2.66 (Russo, 1977).  The maximum age estimate is 8 years for the species Brissopsis lyrifera (Hollertz, 2002) is used in Brey’s (1995) equation to obtain the P/B value.  Our production rate is lower than the one used by Pauly et al. (1993) for sea urchins, which is equivalent to 2.34 year-1.  The Q/B value for the model (9.423 year-1) was calculated as the average for 2 species (Tripneustes gratilla and Salmacis sphaeroides) based on feeding ecology of tropical sea urchins (Klumpp et al, 1993).  A consumption rate cited in Pauly et al. (1993) is 3.58 year-1.  A small catch for sea urchins was entered into the RA model, 2.76E-3 t·km-2.  It is based on the ‘other invertebrate’ category identified in DKP fisheries statistics.  That unidentified amount was divided between 5 invertebrate groups that did not have more precise catch information available.  The statistics represent average catch in the years 2000-2005.    Bivalves  Bivalve biomass is estimated for the RA model based on reef transect results for giant clam provided in COREMAP (2005). The average number of individuals in reef top transects is 2.1 individuals.  Reef top transect area is 40 m2.  This gives an average density of 0.053 individuals·m-2.  This density was converted to weight using the average individual weight of 2 kg to obtain a total biomass estimate of 1.8377 t·km-2.  Assuming that giant clam contributed to a fifth of the biomass of bivalves, bivalve biomass was estimated to be 9.189 t·km-2. Biomass density on reefs has been corrected for the relative reef area in RA based on Spalding et al. (2001).   The P/B for bivalves in the model is 2.514 year-1. The value is calculated as the average of 31 warm water species from Brey (2006). The Q/B value of 5.618 year-1 is calculated for bivalves from consumption and biomass estimates from the same source.  Bird’s Head Seascape Analyses, Page 65  A catch is calculated for RA as 5.89E-3 t·km-2.  It is based on entries in DKP fisheries statistics for pearl oyster, unidentified mollusks, clam and abalone.  It is an average of the years 2000- 2004.  Sessile filter feeders  The biomass of the group, 4.58 t·km-2, is based on estimates of sponge biomass from fore-reef, lagoon and back-reef environments on Davies Reef (Wilkinson and Evans, 1989). The P/B value of 1.48 year-1 for sessile filter feeders was borrowed from a Mexican coral reef model (Alverez- Hernandez, 2003). The Q/B value, 5.258 year-1, was calculated from consumption estimates of epireefal sponges from Kötter (2002). A small catch was entered for sessile filter feeders, 0.001 t·km-2.  Epifaunal detritivorous invertebrates  The biomass of the epifaunal detritivorous invertebrate group is based on starfish biomass calculated from COREMAP (2005) reef transect densities.  That document suggests 3.2 individuals per 40 m2 of reef area.  Assuming an average weight of 1 kg per animal, and multiplying the total estimate by five to account for unsampled taxa in this functional group, a biomass estimate is determined for RA as 7.0 t·km-2 (this density has been corrected for reef area in RA based on ratios in Spalding et al., 2001).  This amount was split between epifaunal detritivorous and carnivorous invertebrate groups in the ratio of 1 to 5.  Biomass of epifaunal detritivorous invertebrates is therefore estimated for RA to be 1.4 t·km-2.  The total biomass of epifaunal invertebrates calculated here (7.001 t·km-2) equates 400 t·km-2 on reefs.  This agrees well with the reef biomass density used by Optiz (1993) for miscellaneous mollusks/worms on a Caribbean reef system, 430 t·km-2.  The P/B was calculated as the average of the P/B of grazing, suspension feeding, deposit feeding and scavenger gastropods (13 year-1) and echinoderm species (16 year-1) from (Brey, 2006). Detritivores ingest about 0.01 to 0.4 times their body weight daily (Lopez and Levington, 1987). For the calculation of Q/B, the average was arbitrarily chosen, 0.05, for all the detritivorous invertebrates. This gave a Q/B equal to 18.25 year-1. The consumption rate cited in Alverez- Hernandez (2003) is 15 year-1.   A small catch for epifaunal detritivorous invertebrates (3.08E-3 t·km-2) is entered into the RA model.  It is based on the ‘other invertebrate’ category identified in DKP fisheries statistics.  That unidentified amount was divided between 5 invertebrate groups that did not have more precise catch information available.  The catch used in the RA model also includes figures listed for mancadu, a gastropod.  The statistics represent average catch in the years 2000-2005.     Epifaunal carnivorous invertebrates  As mentioned in the group description above, total epifaunal biomass was estimated to be 7.0 t·km-2 based on starfish abundance counts from COREMAP (2005).  The starfish biomass estimates were inflated by five times to represent other taxa, and this amount was split between epifaunal detritivorous and carnivorous invertebrate groups in the ratio of 1 to 5 (see above entry for more information).  Biomass of epifaunal carnivorous invertebrates is therefore estimated to be 5.833 t·km-2, although that amount was subsequently reduced in balancing the model to 5.6 t·km-2.    The P/B ratio for the predatory echinoderm Asterias forbesi (2.64 year-1) was used (Robertson, 1979). The Q/B value of 10.52 year-1 was calculated for predatory gastropods and echinoderms from consumption and biomass estimates in Jorgensen et al. (1991) from Florida Bay.  Page 66, Fisheries Centre Research Reports 15(5), 2007  A small catch for epifaunal carnivorous invertebrates (3.6E-3 t·km-2) is entered into the RA model.  It is based on the ‘other invertebrate’ category identified in DKP fisheries statistics.  That unidentified amount was divided between 5 invertebrate groups that did not have more precise catch information available.  Catch for this group also includes the figures listed for snails.  The statistics represent average catch in the years 2000-2005.    Infaunal invertebrates  The infaunal biomass was estimated for a coral reef lagoon to be 3.181 gC·m-2. This value was converted to wet weight using the factor 0.116 (Brey, 2006) to obtain the biomass estimate 27.422 t·km-2 for the RA model.  The P/B ratio for large and small macrophagus polychaetes, microphagous polychaetes, crustaceans, bivalves, gastropods, and other infauna (Riddle et al., 1990) was weighted by relative biomass and averaged to obtain a P/B equal to 4.014 year-1. The Q/B was estimated to be 19.267 year-1 from consumption estimates of infauna from shallow and deep zones (Riddle et al., 1990).  Jellyfish and hydroids  The biomass estimate from Buchary (1999) for the Java sea model 0.1 t·km-2 was used for the model. An alternate estimate equal to 0.222 t·km-2 was found for Florida Bay (Uye and Shimauchi, 2005). The P/B ratio 10.230 year-1 (Venier, 1997) was used for the RA model. The Q/B value (25.463 year-1) is based on consumption estimates of Aurelia aurita in the inland sea of Japan (Uye and Shimauchi, 2005).  For comparison, Buchary (1999) used a lower P/B ratio for jellyfish from the Java Sea of 5.011 year-1 but her Q/B value was very similar, 25.050 year-1.  Carnivorous zooplankton  Aliño et al., (1993) used a zooplankton biomass estimate for a Philippines reef system of 2.87 t·km-2, while Buchary (1999) used 0.310 t·km-2.  The value used in this model, 1.0 t·km-2 is intermediate.  The P/B value 63.875 year-1 was based on Borgne (1982) daily P/B estimate in the range 15 to 20% daily for carnivorous zooplankton species.  The mid value of the range was used to calculate the P/B value.  The Q/B value used by Aliño et al., (1993) was 133.33 year-1, although they allowed Ecopath to estimate this figure.  The Q/B value used by Buchary (1999) was similar at 135.05 year-1.  Although the consumption rate used in the RA model for carnivorous zooplankton (196.28 year- 1) was finally estimated by Ecopath assuming an EE of 0.95, estimates of Q/B were located for chaetognaths, mysids, planktonic amphipods and ichtyoplankton. Q/B for chaetognaths is calculated as 53.851 year-1 (Saito and Kiorobe, 2001), mysids consumption is 26.456 year-1 based on a laboratory experiment (Chipps and Bennett, 2002), planktonic amphipod consumption is 23.884 year-1 based on ingestion rate experiments (Ikeda and Shiga, 1999) and ichthyoplankton consumption is 178.487 year-1 based on an empirical relationship (Houde and Schekter, 1980).  Another similar estimate was obtained for Sagitta elegans Q/B equal to 65.675 year-1 was calculated based on consumption estimate by Terazaki (1996); this assumes a mean size of 7.5 to 10 mm and weight of 41 µgC per individual (Saito and Kiorobe, 2001).  Large and Small herbivorous zooplankton  The RA model uses the biomass estimate from Buchary (1999) for the Java sea of 0.56 t·km-2 for large herbivorous zooplankton and 2.43 t·km-2 for small herbivorous zooplankton.  An alternate estimate was made as 6.645 t·km-2 for all the zooplankton groups.  This represents the average of zooplankton biomass for the latitude (0.5N to 2.5N) and longitude (129.5E to 130.5E) (O'Brien, 2005). This estimate agrees with the one used in the RA model when the figure is divided among  Bird’s Head Seascape Analyses, Page 67 the EwE zooplankton groups.    The P/B value for large herbivorous zooplankton 29.2 year-1 is based on Borgne’s (1982) daily P/B estimate in the range 6 to 10% of body weight daily for herbivorous zooplankton species. The mid value of the range was used to calculate the P/B value. The P/B value used for the model is 32.0 year-1; it was increased slightly to lay closer to the carnivorous zooplankton rate, as the carnivorous group would consist of larger species. An alternate estimate was made as 101.284 year-1 as the average P/B for Daphnia galeata and Bosmina longirostris (McCauley et al., 1996), although these high rates are likely inappropriate as a group average.  The value for herbivorous zooplankton used by Christensen (1996) is 27 year-1, close to our estimate.  The P/B value for small herbivorous zooplankton 91.25 year-1 is based on Borgne’s (1982) P/B estimate in the range 22 to 28% daily for small herbivorous zooplankton species. The mid value of the range was used to calculate the P/B value.  Another estimate was calculated based on the average P/B for small herbivorous plankton from 4 sites (Princess Charlotte Bay, CairnseInnisfail, North West Cape shelf and NWC shelf break) in Great Barrier Reef.  That value is 54.743 year-1 (McKinnon et al., 2005). A conversion ratio of 0.29 is used to convert zooplankton weight in mgC to wet weight (Hansen et al., 2004).  The Q/B for large zooplankton is calculated based on ingestion rates estimated by McCauley et al., (1996). The adult weight of Daphnia galeata and Bosmina longirostris is determined using a size-weight table for Daphnia (Gorokhova and Kyle, 2002); it was assumed that Bosmina has the same relation of body size to body weight. The Q/B for small herbivorous zooplankton is calculated based on consumption by copepods estimated by Borgne et al. (1989) to be 265.813 year-1. Another estimate calculated using an empirical relation for ingestion rate for copepods (Huntley, 1988) was too high 2232.537 year-1 and was abandoned.   Phytoplankton  A biomass estimate of 26.1 t·km-2 was calculated from the average of phytoplankton standing biomass during upwelling (3.7 gC·m-2) and downwelling (2.1 gC·m-2) from the Banda Sea (Tomascik et al., 1997).  The P/B value of 109.118 year-1 was calculated by dividing the average PP estimate 2848 g·m-2·year-1 (GoMor SAI, Italy 2006) for the cells used in the RA model by the biomass estimate. Production rates used for phytoplankton in other reef Ecopath models follow: Caribbean: 160 year-1 (Arias-Gonzalez, 1998), 70 year-1 (Opitz, 1993); Java Sea: 135 year-1 (Buchary, 1999).  Ours is therefore an intermediate value.  Macro-algae  The biomass estimate for macro-algae is 39.389 t·km-2 based on algal biomass on Caribbean coral reefs (Odum and Odum, 1995). The biomass estimate was obtained as 2250.4762 t·km-2. This was scaled according to coral reef area (using ratios in Spalding et al., 2001) to obtain the value 39.389 year-1.  The P/B rate for macro-algae was set in the RA model as 10.225 year-1; it is based on benthic algal production in coral reefs reported by Russ and McCook (1999). The same value is cited in Wolanski (2001) for ‘benthic autotroph’ production rates.  By comparison, the P/B rate used for benthic autotrophs in Caribbean reef systems by Opitz (1993) and by Arias-Gonzalez (1998) was 13.25 year-1.  The value used by Buchary (1999) for the Java Sea was 11.885 year-1 for the group ‘benthic producers’; and the value used by Aliño et al., (1993) in the Philippines for ‘seaweeds’ was 15.34 year-1.  These estimates all agree closely.     Page 68, Fisheries Centre Research Reports 15(5), 2007 Sea grass  The biomass was calculated based on the biomass estimates of 9 sea grass species (Enhalus acoroides, Cymodocea rotundata, Cymodocea serrulata, Halophia ovalis, Halodule pinifolia, Halodule uninervis, Syringodium isoetifolium, Thalassia hemprichii and Thalassodendron ciliatum) from Flores Sea (Tomascik et al., 1997). The biomass was calculated to be 3180.952 t·km-2, this value was scaled according to the potential sea grass area in the model to obtain the value 20.157 t·km-2. This amount is similar to the value used by Aliño et al., (1993), who based their value on Fortes (1990) and estimated 702 g·ww·m-2, which converts to about 14 t·km-2 when corrected for the area of coral reefs in RA. Other estimates of sea grass biomass when scaled to seagrass area are 3.680 t·km-2 (DeIongh et al., 1995) and 6.97 t·km-2 (Erftemeijer. 1994).  P/B is calculated to be 13.758 year-1 based on leaf biomass and production of Thallassia hemprichii (Erftemeijer, 1994).  Mangroves  The average dry weight litter production was estimated to be 7.7 t·ha-1·year-1 (Tomascik et al., 1997).  The litter biomass can be calculated by dividing the production estimate by mangrove P/B.  Alternatively, the primary production from mangroves was given by Tomascik et al. (1997) as 25.936 kgC·ha-1·day-1. Biomass was calculated using this production estimate, however the values obtained in both cases were very high (>30,097 t·km-2) so were not used for the model. This amount refers to production in mangrove areas and will be lower when averaged over RA, but we elected to use Ecopath’s estimate. The EE of the mangroves was fixed at 0.02 to account for terrestrial mortality and Ecopath was allowed to calculate the biomass as 19.136 t·km-2.  The P/B was calculated from the leaf litter production over total mangrove biomass (includes the roots, trunk, branches and leaves) for the 2 species Rhizophora mucronata and Ceriops tagal. The average P/B is calculated to be 0.066 year-1 based on biomass and production estimates from Slim et al. (1996).  Alternatively, P/B could be calculated based on net primary production of Rhizophora apiculata as 0.170 year-1 (Christensen, 1978).    Mangroves thrive in close proximity to the shoreline reefs in Mayalibit Passage in Waigeo, there is an excellent mix of mangroves and sheltered reefs in Wayag Islands (Tomascik et al., 1997).  However, Indonesia’s mangrove forests face a variety of threats.  They are harvested for timber and they are also being removed for land reclamation and to make habitat for fish ponds (Priyono and Sumiono, 1997).  Loss of mangroves hurts the shrimp fishery (Martosubroto and Naamin, 1977) and may impact the survival of juvenile fish that congregate to forage and avoid predation (Laegdsgaard and Johnson, 2001).  These behaviours are incorporated in the EwE models through use of mediation functions (Section 2.5.2 - Mediation factors).  Faunce and Serafy (2006) provide a comprehensive review of field studies that consider mangroves as fish habitat.  Fishery discards and detritus  The standing biomass of fishery discards is set at 20 t·km-2.  Detritus biomass is set at 100 t·km-2.    The 1990 Raja Ampat model   Group biomasses  A 1990 RA Ecopath model is designed based on the 2006 RA model.  Biomasses for the 1990 model are shown in Table A.4.1, along with the rationale used to parameterize biomass and catch of each functional group.  Groups for which no relative biomass estimates are available are assumed to have a similar biomass in 1990 as in 2006; this mainly applies to non-commercial  Bird’s Head Seascape Analyses, Page 69 and non-fish groups (i.e., listed as “No change” in Table A.4.1).  The biomasses of some commercial groups are set in the 1990 model according to the relative abundance change suggested by catch per unit effort (CPUE) series (“CPUE” in Table A.4.1).  The CPUE series did not suggest significant biomass declines for groupers or snappers, despite the fact that they have been heavily fished since 1990.  We assume that the decline in these species has been masked by the CPUE series due to their reproductive biology.  Because they congregate in spawning aggregations, and because fishers target those aggregations, the biomass density available to fisheries may remain constant over a wide range of population sizes.  For a discussion on the dangers of using CPUE data as a proxy for abundance see Beverton and Holt (1957), Gulland (1974) or Hilborn and Walters (1992).   Grouper and snapper biomass was therefore assumed to have decreased by 50% since 1990 (“Custom” in Table A.4.1).  In general, the CPUE data suggested a higher abundance of commercial fish predators in 1990 relative to 2006.  This was evident in balancing the 1990 model because many of the invertebrate prey groups appeared overexploited by fish predators.  We therefore allowed Ecopath to estimate the biomass of several basal invertebrate and planktonic groups, by assuming an EE of 0.99 (“Ecopath” in Table A.4.1).  For multi-stanza groups, the total biomass of all age classes combined is assumed to have increased or decreased since 1990 in direct proportion to CPUE, or according to the custom rules used.  Group biomass is divided into age stanzas according to the mortality schedule, which is inherited from the 2006 model for all functional groups except groupers, snappers, coral trout, Napoleon wrasse and large reef associated fish.  Recent fisheries have developed for these groups that might have shifted their population age structures, and so we assume that the 2006 ecosystem contains a greater proportion of immature individuals relative to 1990.  The total mortality of adult stanzas is reduced in the 1990 model by 20% for these groups.  Large sharks are also heavily exploited in RA, but we assume that they have a similar age distribution today as they did in 1990, since it is likely that most of the depletion of this group occurred prior to 1990.    Fisheries  Fishery catches for 1990 were set for commercial groups directly from the trends suggested by the recorded landings in DKP and/or Trade and Industry Office statistics (see Section 2.5.10 - Interpreting catch statistics) (i.e., listed as “Time series” in Table A.4.1).  For some groups, the time series data does not extend as far back as 1990, and so the values used to initialize the 1990 Ecopath model typically record the earliest catch figure available.  Where the time series landings record is uncertain, we have made critical assumptions regarding the annual quantities of group catch over the last 16 years.  A relatively new and major fishery has developed for Napoleon wrasse, for example.  As the time series data was inadequate, we assumed that the catch in 1990 was equivalent to 10% of the current amount (listed as “10%” in Table A.4.1).  Other heavily exploited species also had inadequate time series catch information.  Most often, we made the assumption that the catch in 1990 was equal to 50% of the current landings; this is listed as “50%” in Table A.4.1.  Non-commercial groups are generally assigned “no catch” in Table 1990 model parameters.  Fitting to time series  The 1990 RA model is driven forward 16 years using an independent series of fishing effort.  Refinements were made to the model structure to improve the data fit with respect to the observed catch and CPUE time series.  Coarse corrections were made to basic parameters to correct the model’s dynamic behaviour (see functional group descriptions).  The proper adjustment of P/B ratios for multi-stanza groups is especially critical to Ecosim’s performance.   The diet matrix and the vulnerability matrix were modified, as well as Ecosim’s feeding parameters.  1990 vulnerabilities are presented in Table A.3.6; feeding parameters are in Table A.5.1.    Page 70, Fisheries Centre Research Reports 15(5), 2007  Trophic flow parameters  The vulnerability matrix for the 1990 RA model was parameterized manually and by using the automated vulnerability search routine available in Ecosim (Christensen and Walters, 2004a).  The search routine uses an iterative procedure to first identify predator-prey interactions critical to model functioning.  With a least-squares criterion, it optimizes those key vulnerabilities in order to recreate observed time series of catch, biomass or other input data.  The optimization was performed first on a large number of interactions.  Then, additional searches were used throughout the balancing process on a fewer number of groups that are highly influential in the system.  The most influential predator groups in the 1990 model tend to be mackerel, large and medium reef associated fish, skipjack and other tuna.  Biomass accumulation rates were used widely to manipulate the initial mortality to production ratios of functional groups, and achieve realistic biomass change as suggested by CPUE data.  Generally, commercial groups are made to follow their observed pattern of biomass change through the impacts of fisheries; we coerced other groups to follow time series data by redistributing predation mortality throughout the diet matrix and reshaping the predation mortality trends using vulnerability adjustments.  The assumption that we have made is therefore that the decline seen in commercial groups is attributable primarily to fishing, while changes to non-commercial groups are due to trophodynamics.    In many cases with commercial groups, the catch recorded in governmental statistics is not sufficient to cause the decrease in group biomass suggested by CPUE data.  The time series catch estimates used for fitting were therefore increased over the original DKP and Trade and Industry Office statistics to acknowledge the impact of unreported catches.  For each year between 1990 and 2006, the reported catch was increased by a fixed percentage.  The relative proportion is based on the estimates of unreported catch used in the 2006 and 1990 models (see Section 2.5.11 - Functional group descriptions).  The catch has therefore been scaled so that the time series forms a continuous trend passing through the start and end point model values.  The CPUE trend, which serves as a proxy for biomass, is entered into Ecosim as a relative trend.  The suspected decline in grouper and snapper biomass over the last 16 years is not reflected in the CPUE trends for biological reasons discussed earlier.  These groups were therefore omitted from the vulnerability search criterion; the automated routine did not attempt to fit these groups to data.  Primary production anomaly  We introduce a primary production anomaly trend using Ecosim’s data fitting technique.  Ecosim generated a climate anomaly trend for the years 1990-2006 that would minimize the residuals between observed and predicted catch and CPUE.  The P/B anomaly is applied only to the most variable group, phytoplankton, and it is designed to reduce the sum of squares with regard to all ecosystem components.  Five spline points are introduced to smooth the production trend.    We rescaled and reentered the production modifier so that the predicted annual phytoplankton biomass variability from simulations matched the observed variability, as determined by SeaWifs primary production data (Sea Around Us Project, 2006).  The annual coefficient of variation (CV) is estimated to be 4.7% from the satellite data.  That is an average for all the cells listed in the database for RA, and it represents the average variability of each 5 year period between 1990- 2005.  We used the average 5-year variability so that random environmental fluctuations would be the main cause of interannual biomass change, and not directional biomass reductions caused by fisheries.  The CV is based on data from the years 1998-2002.  The amplitude of the primary production forcing pattern was reduced by 42% to generate the required CV.  Bird’s Head Seascape Analyses, Page 71  By adjusting the primary production anomaly trend in this way, discrepancy between predicted and observed catch and relative biomass for the ecosystem is made slightly worse (sum of squares is increased by 1.7%).  However, Ainsworth (2006) demonstrated that this method can lead to realistic population variability at higher trophic levels.  Accurately representing the variability of production trends throughout the system can greatly improve the output of more advanced analyses in Ecosim.  For example, a Monte Carlo technique can be used in an ecosystem-based population viability analysis to estimate the extinction risk for commercial species associated with various fishing policies (Ainsworth, 2006; Pitcher et al., 2005).  This technique is applied for RA in Section 3.4 - Challenges to Ecosim.  Equilibrium analysis  The equilibrium analysis routine in Ecosim helps us examine functional group dynamics under varying degrees of fishing pressure.  This routine is an invaluable diagnostic tool and it can be used to answer fundamental questions regarding the production potential of stocks and their resilience to fishing.  The routine sketches the surplus yield curve and the biomass equilibrium curve†† (Equilibrium routine: Christensen et al. 2004a).  Increasing fishing mortality stepwise from zero to several times the baseline model value, the automated routine calculates the equilibrium biomass of a subject functional group under a long-term fishing scheme (1000 years).  At the left-most extent, the biomass equilibrium curves tell us what biomass level the group assumes under zero fishing mortality (i.e., pristine biomass or B0).  The catch equilibrium curves are essentially single-species surplus production curves; the maximum height of the curve shows maximum sustainable yield (MSY) of the stock and the fishing mortality at which that occurs, the Fmsy.  It is useful to compare the current fishing mortality, for example in 2006 (F2006), with the Fmsy.  In a properly parameterized model, the baseline fishing mortality of underexploited groups should be less than Fmsy and greater than Fmsy for overexploited stocks.  How a functional group behaves under dynamic simulation will be greatly influenced by the initial relative level of exploitation represented in the basic Ecopath model. The predictions made by this routine can be compared to estimates derived from single-species tools, and presented to fisheries experts in Indonesia for the purposes of validation.  The equilibrium routine offers several settings, including one that holds the biomass of other functional groups in the model at their (static) baseline conditions.  By selecting this option, Ecosim is reduced to a single-species model, and higher order trophic interactions are removed from consideration.  Alternatively, the user can permit the usual predator-prey dynamics to occur that Ecosim is designed to simulate.  In this use, the equilibrium analysis will consider the multi-species context and provide, in principle, a more accurate representation of ecosystem response to fishing that is suitable for EBM.  By excluding these interactions, the analysis serves as a validation tool, by which we can compare Ecosim’s predictions with ‘classical’ single-species analysis models.    Where possible, the equilibrium analysis is performed for subject groups while holding the biomass constant for other functional groups to facilitate comparison with single-species models.  However, where multi-stanza groups were employed, it was sometimes necessary to perform the equilibrium analysis manually in Ecosim to circumvent a current limitation in the equilibrium analysis routine.  The equilibrium analysis routine increments fishing mortality only for the subject functional group being tested (V. Christensen. UBC Fisheries Centre, 2202 Main Mall, Vancouver Canada,                                                         †† Hilborn and Walters (1992) discuss the theory and applications of equilibrium stock assessment models.  Page 72, Fisheries Centre Research Reports 15(5), 2007 pers. comm.).  If the test is run on an adult stanza, then the routine holds fisheries constant for the juvenile and/or sub-adult stanzas.  If there is normally a high fishing rate on the juvenile or sub-adult stanzas, then the equilibrium analysis can be misleading.  It will suggest a high estimate of sustainable adult fishing mortality because the adult pool is being fed by constant recruitment from the immature stanzas.  In reality, an increasing amount of catch on adults will usually be accompanied by increased catch on immature groups, either intentionally or incidentally.  This increase is currently missed by the equilibrium routine.  This has always been a potential source of error, where split-pools were used to describe age stanzas, but the problem was amplified with the addition of the multi-stanza routine.  Now, large amounts of fishing effort may typically be modelled on sub-adult groups, yet the equilibrium analysis routine will continue to assume knife edge entry to the fishery effectively, and increment fishing effort only on the adult stanza.  To provide a more realistic view of surplus stock production for these groups, it is necessary to perform the equilibrium analysis manually, increasing fishing mortality on juvenile and sub-adult groups as well as the adults.  This was done here on key groups, and these runs are indicated by asterisks in Fig. B.2.1. Trophic interactions must necessarily be included in these runs.   Challenges to Ecosim  The 2006 RA Ecosim model is subjected to various challenges to test its behaviour and stability.  By applying extreme fishing scenarios we can see how the model performs when fishery and functional group parameters vary far from their initialization values.  Species interactions, which may appear nominal under baseline conditions, can compound in unexpected ways to cause oscillations or chaotic model behaviour.  If a region of instability exists in the fleet-effort responses of the model, normal fishing forecasts that apply conservative or realistic fishing strategies may or may not be affected.  However, if a combination of fishing efforts exists that can drive the model to instability, the policy search routine will typically become useless as it then only locates unstable solutions that offer impossibly large benefits.  We use three fishing strategies to challenge the model: no fishing, baseline fishing and increasing fishing.  For the ‘no fishing’ scenario, the fishing mortality of each gear type is reduced to zero for all simulation years and target groups (including directed and bycatch mortality).  Under this fishing test, we expect depleted commercial functional groups to rebound, and the prey of these groups should see a corresponding decrease.  The baseline fishing scenario is provided for comparison; it represents the current (2006) fishing mortalities applied into the future.  The ‘increasing fishing’ scenario assumes an annual increase in fishing mortalities of 3.2% across all gear types.  This rate corresponds to the recent increase in the human population in eastern Indonesia (BPS, 2006).  We expect to see the biomass of exploited functional groups decline and a corresponding increase in the biomass of their prey.  All Ecosim simulations are for 16 years, from 2006-2022.  As an additional test of the model’s performance, we have used 50 Monte Carlo simulations to vary Ecopath’s biomass parameters for commercial groups.  For this sensitivity analysis, groups are allowed to vary +/- 20% from their Ecopath biomass values, the Monte Carlo draws from a uniform distribution.  The following commercial groups are varied: adult and sub-adult groupers, adult and sub-adult snappers, adult and sub-adult Napoleon wrasse, skipjack tuna, other tuna, mackerel, billfish, and the adult stanzas of coral trout, large sharks, large, medium and small pelagic, large, medium and small reef-associated, large and small demersals, large and small planktivores, and anchovy.  The Monte Carlo routine allows us to test the sensitivity of initial biomass parameters, establish a range of error for predictions, and determine the depletion risk for functional groups in a population viability analysis.     Bird’s Head Seascape Analyses, Page 73 Ecospace parameterization  An Ecospace model was developed for RA and is presented here.  Ecospace maps are also presented for high resolution models of Kofiau Island and Dampier Strait, as well as habitat classifications (Table 2.7) and fishery locations (Table 2.8).  All of the Ecospace work is preliminary thus far; dynamics have not been analyzed and we present these here only for expert evaluation.  High-resolution models for Kofiau Island, Dampier Strait and SE Misool Island will be integrated into Ecospace in 2007, following the development and refinement of reef-based Ecopath models, where biomass, catch and other parameters are adjusted to represent the smaller study areas using site-specific data.  The RA Ecospace model utilizes GIS information assembled by the BHS EBM project, as well as oceanographic and biological data from the literature to represent the study area in a 2 dimensional grid matrix of spatial habitat cells.  Standard Ecopath and Ecosim parameters are inherited from the 2006 RA model described in this report.     Page 74, Fisheries Centre Research Reports 15(5), 2007  Table 2.7 - Habitats occupied by functional groups in three 2006 Ecospace models.  RA, Kofiau Island and Dampier Strait. Al l 10 m 20 m 200m Deep  (>  2 00 m ) R eef Ar ea o ccu pi ed Al l 20 m 200m Deep  (>  2 00 m ) R eef Encl os ed  lago on Es tuary Mang ro vev Ar ea o ccu pi ed Al l 20 m 200m M es op el ag ic  (2 00- 1000 m ) B at hy pe la gi c ( > 100 0m ) R eef Encl os ed  lago on Mang ro ves Ar ea o ccu pi ed M y s t i c e t a e 0 . 9 0 0 . 9 3 0 . 9 2 P i s c .  o d o n t o c e t a e 0 . 9 6 0 . 9 3 0 . 9 2 D e e p .  o d o n t o c e t a e 0 . 9 0 0 . 8 4 0 . 2 2 Du g o ng s 0 .0 8 0 .1 6 0 .4 4 B i r d s 1 . 0 0 1 . 0 0 1 . 0 0 R e e f  a s s o c .  t u r t l e s 1 . 0 0 1 . 0 0 1 . 0 0 G r e e n  t u r t l e s 1 . 0 0 1 . 0 0 1 . 0 0 O c e a n i c  t u r t l e s 1 . 0 0 1 . 0 0 1 . 0 0 C r o c o d i l e s 0 . 0 2 0 . 0 2 0 . 0 5 A d .  g r o u p e r s 0 . 0 4 0 . 0 6 0 . 0 7 S u b .  g r o u p e r s 0 . 0 4 0 . 0 6 0 . 0 7 J u v .  g r o u p e r s 0 . 0 1 0 . 0 4 0 . 0 2 A d .  s n a p p e r s 0 . 0 4 0 . 0 6 0 . 0 7 S u b .  s n a p p e r s 0 . 0 4 0 . 0 6 0 . 0 7 J u v .  s n a p p e r s 0 . 0 1 0 . 0 4 0 . 0 2 A d .  N a p o l e o n  w r a s s e 0 . 0 4 0 . 0 6 0 . 0 7 S u b .  N a p o l e o n  w r a s s e 0 . 0 4 0 . 0 6 0 . 0 7 J u v .  N a p o l e o n  w r a s s e 0 . 0 1 0 . 0 4 0 . 0 2 S k i p j a c k  t u n a 0 . 9 9 0 . 9 9 0 . 9 9 O t h e r  t u n a 0 . 9 9 0 . 9 9 0 . 9 9 M a c k e r e l 0 . 9 9 0 . 9 9 0 . 9 9 B i l l f i s h 0 . 9 0 0 . 9 3 0 . 9 2 A d .  c o r a l  t r o u t 0 . 0 4 0 . 0 6 0 . 0 7 J u v .  c o r a l  t r o u t 0 . 0 1 0 . 0 4 0 . 0 2 A d .  l a r g e  s h a r k s 0 . 9 0 0 . 9 7 0 . 5 8 J u v .  l a r g e  s h a r k s 0 . 9 0 0 . 9 7 0 . 5 8 A d .  s m a l l  s h a r k s 0 . 9 0 0 . 9 9 0 . 9 9 J u v .  s m a l l  s h a r k s 0 . 9 0 0 . 9 9 0 . 9 9 W h a l e  s h a r k 0 . 9 0 0 . 9 7 0 . 5 8 M a n t a  r a y 1 . 0 0 0 . 9 9 0 . 9 9 A d u l t  r a y s 1 . 0 0 0 . 9 9 0 . 9 9 J u v .  r a y s 1 . 0 0 0 . 9 9 0 . 9 9 A d .  b u t t e r f l y f i s h 0 . 0 4 0 . 0 6 0 . 0 7 J u v .  b u t t e r f l y f i s h 0 . 0 1 0 . 0 7 0 . 0 8 C l e a n e r  w r a s s e 0 . 0 1 0 . 0 6 0 . 0 7 A d .  l a r g e  p e l a g i c 1 . 0 0 0 . 9 5 0 . 9 7 Ju v .  l a rg e  p e la gi c 1 .0 0 0 .9 6 0 .9 8 A d .  m e d i u m  p e l a g i c 1 . 0 0 0 . 9 5 0 . 9 7 Ju v .  m e diu m  pel a g i c 1 .0 0 0 .9 6 0 .9 8 A d .  s m a l l  p e l a g i c 1 . 0 0 0 . 9 5 0 . 9 7 Ju v .  sm a l l  pel a g i c 1 .0 0 0 .9 6 0 .9 8 A d .  l a r g e  r e e f  a s s o c . 0 . 0 4 0 . 9 7 0 . 9 4 J u v .  l a r g e  r e e f  a s s o c . 0 . 0 1 0 . 1 6 0 . 4 4 A d .  m e d i u m  r e e f  a s s o c . 0 . 0 4 0 . 9 9 0 . 7 7 J u v .  m e d i u m  r e e f  a s s o c . 0 . 0 1 0 . 1 6 0 . 4 4 A d .  s m a l l  r e e f  a s s o c . 0 . 0 4 0 . 9 9 0 . 7 7 J u v .  s m a l l  r e e f  a s s o c . 0 . 0 1 0 . 1 6 0 . 4 4 A d .  l a r g e  d e m e r s a l 0 . 3 9 0 . 9 3 0 . 9 2 J u v .  l a r g e  d e m e r s a l 0 . 3 9 0 . 1 3 0 . 4 2 A d .  s m a l l  d e m e r s a l 0 . 3 9 0 . 9 5 0 . 9 7 J u v .  s m a l l  d e m e r s a l 0 . 3 9 0 . 1 3 0 . 4 2 A d .  l a r g e  p l a n k t i v o r e 0 . 0 4 0 . 9 7 0 . 7 2 J u v .  l a r g e  p l a n k t i v o r e 0 . 0 1 0 . 1 6 0 . 4 4 A d .  s m a l l  p l a n k t i v o r e 0 . 0 4 0 . 9 9 0 . 7 7 Ju v .  sm a l l  pla n k tiv o r e 0 .0 1 0 .1 6 0 .4 4 A d .  a n c h ov y 0 .4 1 1 .0 0 0 .5 6 Ju v .  a n cho v y 0 .4 1 1 .0 0 0 .5 6 A d .  d e e p w a t e r  f i s h 0 . 5 9 0 . 8 4 0 . 5 6 J u v .  d e e p w a t e r  f i s h 0 . 9 0 0 . 9 5 0 . 9 7 A d .  m a c r o  a l g a l  b r o w s i n g 0 . 0 4 0 . 9 9 0 . 9 9 J u v .  m a c r o  a l g a l  b r o w s i n g 0 . 0 1 0 . 1 6 0 . 4 4 Raja Ampat K ofia u D ampie r  St.          Bird’s Head Seascape Analyses, Page 75  Table 2.7 - (cont.) Al l 10 m 20 m 20 0m D ee p ( > 20 0m ) upied 20 0m ) go on ev cu pi ed Re ef A rea  o cc Al l 20 m 20 0m D ee p ( > Re ef Enclosed la Estuary Mangro v A rea  o c Al l 20 m 20 0m Mes opela gi c (20 0-1 000 m ) Ba th yp el agic (> 1 00 0m) Re ef Enclosed la go on Mangro ves A rea  o cc upied A d .  e r o d i n g  g r a z e r s 0 . 0 4 0 . 9 9 0 . 7 7 J u v .  e r o d i n g  g r a z e r s 0 . 0 1 0 . 1 6 0 . 4 4 A d .  s c r a p i n g  g r a z e r s 0 . 0 4 0 . 9 9 0 . 7 7 J u v .  s c r a p i n g  g r a z e r s 0 . 0 1 0 . 1 6 0 . 4 4 D e t r i t i v o r e  f i s h 0 . 1 0 0 . 9 9 0 . 7 7 A z o o x a n t h e l l a t e  c o r a l s 0 . 1 0 0 . 1 5 0 . 4 3 H e r m a t y p i c  c o r a l s 0 . 0 1 0 . 0 4 0 . 0 2 N o n  r e e f  b u i l d i n g  c o r a l s 0 . 1 0 0 . 1 5 0 . 4 3 S o f t  c o r a l s 0 . 1 0 0 . 1 5 0 . 4 3 C a l c a r e o u s  a l g a e 0 . 1 0 0 . 1 5 0 . 4 3 A n e m o n i e s 0 . 9 9 0 . 9 9 0 . 9 9 P e n a e i d  s h r i m p s 0 . 9 6 0 . 9 9 0 . 9 9 S h r i m p s  a n d  p r a w n s 0 . 9 6 0 . 9 9 0 . 9 9 S q u i d 0 . 9 0 0 . 9 3 0 . 9 2 O c t o p u s 1 . 0 0 0 . 9 9 0 . 7 7 S e a  c u c u m b e r s 1 . 0 0 1 . 0 0 1 . 0 0 L o b s t e r s 1 . 0 0 1 . 0 0 1 . 0 0 L a r g e  c r a b s 1 . 0 0 1 . 0 0 1 . 0 0 S m a l l  c r a b s 1 . 0 0 1 . 0 0 1 . 0 0 C r o w n  o f  t h o r n s 0 . 0 4 0 . 0 6 0 . 4 3 G i a n t  t r i t o n 0 . 1 0 0 . 0 6 0 . 4 3 H e r b i v o r o u s  e c h i n o i d s 1 . 0 0 1 . 0 0 1 . 0 0 B i v a l v e s 0 . 4 1 0 . 9 9 0 . 9 9 S e s s i l e f i l t er  f e ed er s 0 . 3 9 0 . 9 9 0 . 9 9 E p i f a u n a l  d e t .  i n v e r t s . 1 . 0 0 1 . 0 0 1 . 0 0 p i f a u n a l  c a r n .  i n v e r t s 1 . 0 0 1 . 0 0 1 . 0 0 f a u n a l  i n v e r t s . 0 . 9 9 0 . 9 6 0 . 9 8 l l y f i s h  a n d  h y d r o i d s 1 . 0 0 0 . 9 9 0 . 9 9 C a r n .  z o o p l a n k t o n 1 . 0 0 1 . 0 0 1 . 0 0 L a r g e  h e r b .  z o o p l a n k t o n 1 . 0 0 1 . 0 0 1 . 0 0 S m a l l  h e r b .  z o o p l a n k t o n 1 . 0 0 1 . 0 0 1 . 0 0 P h y t o p l a n k t o n 1 . 0 0 1 . 0 0 1 . 0 0 M a c r o  a l g a e 0 . 4 1 0 . 9 9 0 . 7 7 S e a  g r a s s 0 . 0 8 0 . 0 1 0 . 0 1 M a n g r o v e s 0 . 0 2 0 . 0 1 0 . 0 1 F i s h e r y  d i s c a r d s 1 . 0 0 1 . 0 0 1 . 0 0 D e t r i t u s 1 . 0 0 1 . 0 0 1 . 0 0 R a j a  A m p a t K o f i a u D a m p i e r  S t . E I n J e     Table 2.8 - Designated fishing activity in Ecospace habitat types. All 10 m 20 m 20 0m D ee p ( > 2 00m ) Ree f All 20 m 20 0m D ee p ( > 2 00m ) Ree f En closed lag oo n Estu ary Ma ng rove v All 20 m 20 0m M eso pelagic (20 0- 10 00 m) Bath yp el ag ic (> 10 00 m) Ree f En closed lag oo n Ma ng rove s S p e a r  a n d  h a r p o o n Re ef  g l ea n i n g Sh ore g i l l net Dr i f t n e t P e rm an e n t trap P o rtabl e  trap Di vi n g  s p ea r Di vi n g  l i ve  fi s h D i v i n g  cy anid e Blas t  f i sh in g T r ol lin g P u r s e  s e i n e P o le a n d li ne Set lin e Li ft  n e t F o r e i g n  f l e e t - - - - - - - - - - - - - - - - S h r i m p  t r a w l - - - - - - - - - - - - - - - - Raj a  A m pat K of i a u D ampier St .    Page 76, Fisheries Centre Research Reports 15(5), 2007 Raja Ampat 2006 Ecospace model  The RA Ecospace map is 100 x 120 cells and describes an area about 256 km east to west, and 321 km north to south; each cell represents an area 2.56 x 2.57 km, or approximately 6.57 km2 at mid-latitudes.  The north-westernmost coordinate lies at 129o 12’ E, 0o 12’N, and the south- easternmost coordinate lies at 130o 30’ E, 2o 42’S.  Five aquatic habitat types are used in the RA Ecospace model.  The habitats are based partly on bathymetry, with 10, 20 and 200 m contours represented as light green, orange and dark green cell colours in Fig. 2.6.  A deep-water habitat type describes cells greater than 200 m in depth (blue), and a reef habitat type shows the locations of submerged reefs (red).  Bathymetric information was obtained from Indonesian nautical charts collected by the BHS-EBM project in GIS files (contact: M. Barmawi, TNC-CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia).       Figure 2.6 - Ecospace habitat map of RA.  Land cells are black, and five habitat types summarize major oceanographic zones (light green: 10 m isobath; orange: 20 m isobath; dark green: 200m isobath; blue: deep water (>200 m); red: reef areas).  Map dimensions are 125 x 100 cells.  Cell dimensions are approximately 2.56 x 2.57 km.    Bird’s Head Seascape Analyses, Page 77 Primary production spatial forcing pattern  A primary production spatial forcing pattern is entered for the RA model using data from the Sea Around Us Project ecology database (Sea Around Us Project, 2006); the information is retrieved automatically by an Ecospace sub-routine (contact: V. Christensen, UBC Fisheries Centre.  2202 Main Mall, Vancouver, Canada).  The primary production has been estimated using a model by Platt and Satyendranath (1999) that integrates PP by depth based on chlorophyll pigment concentrations and photosynthetically active radiation (Lai, 2004; Hoepffner et al., 1999).  Ocean colour is provided by the SeaWiFS database at a spatial resolution of approximately 6 minutes, or 11 km (http://oceancolor.gsfc.nasa.gov/SeaWiFS/).    ensing   There is an average primary production of about 315.4 gC·m-2·year-1 in the study area.  The chlorophyll data showed an area of high productivity (>1000 gC·m-2·year-1) in the northern extent of the Arafura Sea, in the southeast region of the study area.  However, there is a large amount of river input to the sea along the mainland coast of Papua, and remote s   Figure 2.7 - Spatial primary production (P/B) for Raja Ampat.  High production (red) corresponds to production values >1000 gC·m-2·year-1; green areas represent medium production ~400-600 gC·m- 2·year-1; blue areas are low production ~200-400 gC·m-2·year-1; white areas are oligotrophic <200 gC·m- 2·year-1.  Resolution: 6 minutes.  Modified from: Sea Around Us Project database (Sea Around Us Project, 2006).  Page 78, Fisheries Centre Research Reports 15(5), 2007 techniques based on ocean colour may become confused as the concentration of optically absorbing particles increases.  Suspended material, n a ension of sedim Similarly, the area of high production indicated to the w turbidity due to suspended sediments from waves, and u  above the reported rate to the east and west of the central pwelling zone.  On average, cells in this zone are in the 90th percentile of production rates ximum production rate in this zone is 723 gC·m-2·year-1, and the region ontributes 1.5% to the total RA primary production.  This modification to the Sea Around Us s approximately 27.8 km by 87.6 km; each cell covers an area 555 m by 80 m, incorporating about 0.32 km2 of sea area per cell.  The north-westernmost coordinate lies dissolved organic matter, and bottom ous nutrient loading will legitimately t the colour signal north of the Arafura ents which affects ocean reflectance.  est of Salawati Island may be biased by pwelling produced in the Sagewin Strait to the NE (M. Erdmann.  CI.  Jl. Dr. Muwardi. 17 Renon Denpasar, Bali, Indonesia, pers. comm.).   We have applied the primary production data directly as obtained from the Sea Around Us Project database, but we offer these caveats.  Further work can test the sensitivity of the Ecospace analyses to our assumptions concerning the distribution of primary production.  There were no elevated levels of primary production reported by Sea Around Us Project (2006) data for the central portion of Dampier Strait, despite the presence of a strong and productive region of upwelling that supports a large anchovy fishery (Mark Erdmann.  CI.  Jl. Dr. Muwardi. 17 Renon Denpasar, Bali, Indonesia, pers. comm.).  To capture this production in the Ecospace model, a region of high productivity was entered in manually.  The most productive region of central Dampier Strait was set arbitrarily at 250% of the reported production rate; the area of high production tapers off to 30% reflectance can influence the data.  Although terrige increase primary production in that area, we caution th Sea may also be biased by the susp u among map cells.  The ma c Project (2006) data is visible in Fig. 2.7 as a yellow-green area in central Dampier Strait.  Note that this modification does not change the overall primary productivity of the Ecopath model.  Ecospace scales the relative cell production rates so that their average equals the phytoplankton P/B defined in the basic parameter set (Table A.3.2).   Kofiau Island model    Figure 2.8 - Ecospace habitat map of Kofiau Island.  Land cells are black, and seven habitat types are used to describe the marine area (orange: 20m isobath; dark green: 200m isobath; blue: deep water (> 200m); red: reef areas; light blue: enclosed lagoon; purple: estuary; teal: mangroves).  Map dimensions are 50 x 150 cells.  Cell dimensions are approximately 0.56 x 0.58 km, or about 0.32 km2.  The Kofiau Island model is represented by 150 cells east to west, and 50 cells north to south (Fig. 2.8).  The modelled area i 5 at 129o 14’ 20’’ E, 1o 5’S, and the south-easternmost coordinate lies at 130o 1’ 20’’E, 1o 20’S.  Seven habitat types are used for the Kofiau Island Ecospace model.  Three are based on bathymetric data from GIS collections: 20 m isobath (orange habitat), 200 m isobath (green habitat) and deep water (>200 m) (dark blue habitat).  Reef areas (red), mangrove areas (teal)  Bird’s Head Seascape Analyses, Page 79 and estuaries (light blue) were identified by the marine use survey; these are all incorporated as dedicated habitat types.    The model relies on GIS data collected by the BHS EBM project (M. Barmawi, TNC-CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia.  Unpublished data), and on habitat exchange with outside areas due to land cell proximity.   By assigning a dedicated habitat type we can further quarantine this area from n surrounding the Kofiau Island group has been entered based on esults from the SPAG vial release program.  286 out of 1000 vials have been recovered.  They Muljadi, TNC-CTC.  l Gunung Merapi No. 38, Kampung information reported in COREMAP (2005).  We include ecologically significant areas identified by expert knowledge and by community interviews in the BHS EBM resource use assessment study.  The habitats consider an enclosed lagoon area (Fig. 2.9) in the Boo Island group west of Kofiau Island.  This area will function as an enclosed lagoon in Ecospace by having little biomass   - An enclosed lagoon on Taudore Island the rest of the system, removing or limiting the trophic influence of oceanic and high sea species, like billfish, large sharks or whales.  A seamount area (Dona Carmalita) (Fig. 2.10) was identified by the marine use survey; it is included in Ecospace as a collection of reef and shallow water habitats.    Figure 2.9 .  Location is west of Kofiau Island, photographed during aerial surveys.  Photo credit: Erdi Lazuardi.  A simple advection patter r were found in disperse places, but all east of the release point.  The majority of vials (255) were recovered south and east of Kofiau, near Wejim Is. (north of Misool). (C. Rotinsulu.  CI.  Jl Arfak No. 45.  Sorong, Papua, Indonesia 98413, pers. comm.).  We therefore applied a southeast advection current.  Ocean currents must be monitored in RA throughout the year, however, so that we can account for seasonal variations.  Vials were also found to the east on Salawati and Batanta Islands, and to the north on Weigeo Island indicating complex current systems.  When the Ecospace models are more advanced, we will try to adjust the dispersal rates of juvenile groups to allow settlement throughout RA.   There are also tidal flows north to south throughout the year passing between the Kofiau Island group and the Boo Island group in a deep central area.  Tidal mixing may sustain a population of small pelagics providing a significant feeding area for sea birds (A.   Figure 2.10 - Dona Carmalita seamount.  Location is south of the Boo Island group (west of Kofiau Island) photographed during aerial surveys.  Photo credit: Andreas Muljadi. J  Page 80, Fisheries Centre Research Reports 15(5), 2007 Baru, Sorong, Papua, Indonesia 98413, pers. comm.).  We will model that interaction when cospace is more fully developed.   m by 80 km.  Each cell is 1.33 km by 1.33 km, r about 1.76 km2.  Seven habitat types were used to represent Dampier Strait.  Two habitat types ssembled GIS data: 20 m isobath (orange habitat), 200 m CTC.  Jl Pengembak 2, Sanur, Bali, Indonesia, narrow channel connects Mayalibit Bay to Dampier Strait E  Dampier Strait model  The Dampier Strait model uses 105 cells east to west, and 60 cells north to south (Fig. 2.11).  The rea covered by the model is approximately 139 ka o are based on bathymetry from a obath (green habitat) (M. Barmawi, TNC-is unpublished data).  Although we did not have accurate depth contour information for areas greater than 200 m, we have assumed that the interior of Dampier Strait consists of a deeper area between 500-1000 m (dark blue).  Mangrove areas (teal) and reef areas (red) are based on recent BHS EBM project outputs (Firman and Azhar, 2006). The enclosed Mayalibit bay (light blue) receives its own habitat type to distinguish the sheltered, shallow bay from the deeper oceanic areas of Dampier Strait.  A (Fig. 2.12).    Figure 2.11 - Ecospace habitat map of Dampier Strait.  Weigeo Island borders on the north, Batana Island is in the south.  Land cells are black, and seven habitat types are used to describe the marine area (orange: 20 m isobath; dark green: 200 m isobath; blue: deep water (>200 m); red: reef areas; light blue: enclosed lagoon; purple: estuary; teal: mangroves).  Map dimensions are 50 x 150 cells.  Cell dimensions are approximately 0.56 x 0.58 km, or about 0.32 km2.   Fishing policy optimizations  The policy search routine in Ecosim (Christensen and Walters, 2004b) iteratively varies the fleet effort and reruns harvest simulations until it finds the optimal combination of fishing mortalities that maximizes harvest benefits. The routine can be used to identify fishing patterns that crease economic, social and ecological benefits from the ecosystem by use of a multi-criterion here to explore the sustainable production poten e RA ecosystem, and quantify the ecological impacts of various optimal fishing policies.  We in objective function.  We use this routine tial of th  Bird’s Head Seascape Analyses, Page 81 use randomly selected F per gear type for initialization, and employ the Fletcher-Powell Fletcher and Powell, 1963).  For ach gear type in the model, a ptimal fishing ortality is calculated and in the imulation to find the best from P).  This is the default ecological objective in the policy m’s (1969) description of mature ecosystems.  Economic  according to their net present value (NPV).  NPV is a term used in cost- enefit analysis to summarize the expected future flow of benefits into a single value, which can atives.  Intergenerational discounting (Sumaila, 2001; sed by default in Ecosim, where the standard discount nerations (δfg) is 10%.  This is a precautionary economic rces will be preserved for future generations (Sumaila, ed that incrementally vary the relative weightings of the weightings are represented by WECON and WECOL in eq.  maximized by the search.  terms evaluate socio-econo nctional group (i), gear type (j) and simu tine will allow us to sketch the trade-off frontier between e the marginal costs and benefits of conservation. ESULTS its to biomass and catch data for functional groups are presented in Figs. B.2.2 and B.2.3, respectively.  Regarding biomass predictions, there is acceptable agreement with data for most functional groups.  Groupers and snappers show a poor fit, but CPUE data is a poor proxy for the biomass of these groups due to their aggregation behaviour.  Because they congregate in spawning aggregations, and because fishers target those aggregations, the biomass density available to fisheries may remain constant over a wide range of population sizes.  The conjugate gradient optimization method ( e single o m applied to each year s equilibrium-level solution in a 16 year forecast, from 2006- 2022.  The objective function used here considers two criteria, the economic value produced the ecosystem, and the ecological health of the system measured using a proxy for functional group longevity (B/ search routine and it is inspired by Odu benefits are assessed   Figure connects Myalibit 2.12 - Myalibit Bay entrance.  A narrow channel  Da pier Strait to the shallow enclosed area of bay dit: Andreas Muljadi. m .  Photo cre b be compared across investment altern 2004; Sumaila and Walters, 2005) is u rate (δ) is 4% and the rate for future ge criterion because it assures that resou 2001).    A range of optimal policies are generat economic and ecological criteria.  The 2.15; OBJ is the objective function to be  OBJ = WECON · Σ NPVij + WECOL · Σ B/Pit   The summed (2.15) mic and ecological benefits of the harvest plan across lation time step (t).   fu  This application of the policy search rou profit and ecosystem health, and calculat   R  Time series fitting  F  Page 82, Fisheries Centre Research Reports 15(5), 2007 populations of groupers and snappers have likely declined throughout RA, and Ecosim predicts this case under the effort assumptions in place (Section 2.5.10 - Effort time series).    One other discrepancy in the time series fitting from 1990 to 2006 is that there is little or no decrease seen  the biomass of large, medium and ite a eported decline in the CPUE during y commercial functional roups, the catch required to cause the biomass decline suggested by CPUE is substantially larger than the catch estimated from government records (Fig. B.2.3); the difference may serve as a first estimate of unreported catch for the study area.  Biodiversity is predicted to have declined from 1990-2006 according to Fig. 3.1.  The biodiversity statistic in use, Q75, is a variant on Kempton’s Q Index (Kempton and Taylor, 1976) that has been designed for use with ecosystem box models‡‡.    The average trophic level of the catch in RA is reported in Fig. 3.1.  The analysis suggests that the average trophic level may have seen a slight downward trend for most of the last 16 years.  However, the pattern has been variable, and the total overall decline from 1990 to 2006 remains small, from 0 to 0.07 TL.  For comparison, Essington et al. (2006) suggested that a mean decline in the trophic level of catch of 0.15 constitutes evidence of ecologically significant ‘fishing down the food web’ (Pauly et al., 1998).  In the model, the trophic level of the catch remains somewhat constant because of an expanding fishery on high trophic level predator fish groups throughout the length of the simulation, and decreased catches of anchovy because of a population decline he optimal primary production anomaly trend determined by Ecosim (broken line; Fig. 3.2)  average.  Applying the P/B forcing pattern to  %.  When we scale the primary production in small reef associated fish, desp r that period.  These groups will be the subject of further tuning after we have processed abundance data from fisher interviews (see discussion), and improved our understanding of the relative biomass change in the RA ecosystem.   For man g (Fig. B.2.3).    Predicted climate anomaly  T suggests a higher than average rate of production for early years in the simulation, 1990-1995, as high as 12% above the mean value.  The trend shows a lower relative rate of production in recent years, 2002-2006, approximately 15% below phytoplankton reduces ecosystem residuals by                                                        5  ‡‡ Kempton’s Q index represents the inter-quartile slop both species evenness and species richness.  Larger valu use of Kempton’s Q Index in ecosystem models, see Ai e e n index in EwE see Christensen et al. (2004).  of the cumulative species log-abundance curve; it evaluates s indicate a more biodiverse system.  For a discussion on the sworth and Pitcher (2006); for implementation of the Q75 6 1 9 9 0 1 9 9 4 1 9 9 8 2 0 0 2 2 0 0 6 Y e a r 3 . 2 0 Figure 3.1 - Raja Ampat ecosystem indicators (1990- 2006).  Biodiversity trajectory predicted for 1990-2006 (shaded area).  Biodiversity is measured using Q75, a 7 8 B io div e 3 . 2 2 3 . 2 4 TL  o variant of Kempton’s Q index.  Trophic level of catch (broken line) may have decreased, indicating ‘fishing down the food web’. 9 1 0 1 1 1 2 rs ity  (Q 75 ) 3 . 2 6 3 . 2 8 3 . 3 0 f c at ch  Bird’s Head Seascape Analyses, Page 83 anomaly trend to reproduce the observed phytoplankton variability, ecosystem residuals are reduced only 3.7% versus the baseline simulation.  Future work may confirm whether the redicted biomass variability of higher trophic level groups has been improved, although data is p limiting.  0 . 1 0 0 . 1 5 - 0 . 2 0 - 0 . 1 5 - 0 . 1 0 - 0 . 0 5 PP  an o 0 . 0 0 1 9 9 0 1 9 9 2 1 9 9 4 1 9 9 6 1 9 9 8 2 0 0 0 2 0 0 2 2 0 0 4 2 0 0 6 Y e a r m  F r ary production a ly.  An is pre to red screpan tween observ ted catch and rela mass; 5  points ed.  Bro e indic timal fo g by Ecosi aded hows  resca  matc ed p p   The es maly shows a weak (non-signifi correlation with the Southern Oscillation Index (SOI) (Spearman  correlation ρ = -0. .05(2),16 = 0.503) (AGBM, 2006) and a s also non-significant neg tive c  surface temperature (SST) (ρ = -0.31) (IRI, 2006).  T w apua has th est primary pro ability in Indonesia (Susanto ).  Theref e outputs of the primary production could potentially  su i in RA.    F . icient of var  for functional group biomass in RA.  The d m 6 are con tive w u oduction ing pattern in place.  Using the optimal forcing patter u  the g e on in residual sus observ causes a high degree o o p a d to tch p op the v lity o is reduced in the m l 0 . 0 5 aly igu e 3.2 - Prim noma omaly dicted uce di cy be ed and predic tive bio  spline  are us ken lin ates op rcin  pattern predicted m; sh area s pattern led to h observ hyto lankton variability. timated ano cant) negative s rank 05; ρ0 tronger but a orrelation with sea he est coast of P e high duction vari et al., 2006 ore, th analysis affect our choice of nable fishing policies sta ig. 3 3 shows the coeff iation yna ics from 1990-200 serva itho t the primary pr  forc n determined by Ecosim res lts in reat st reducti s ver ed data, and also f ec system volatility.  By rescaling the rim ry production tren  ma hyt lankton observations, ariabi f functional groups ode . 0 0.2 0.4 0.6 0.8 1-2 2-3 3-4 4+ Tr op hi c le ve l Coeffic ariation   - Coefficient of variation (CV) of RA nal gro mass ( 006).  orted phic le lack bars indicate imary tion (PP) forcing, grey bars show optimal PP  patter e bars d  that improves phytoplankton variability. ient of v Figure 3.3 functio up bio 1990-2 CV is s by tro vel.  B no pr produc forcing n, whit show rescale  pattern  Page 84, Fisheries Centre Research Reports 15(5), 2007  Equilibrium analysis  Table B.2.1 presents the equilibrium analysis results.  Key fishery indicators for commercial functional groups are summarized in Table 3.1.  Where catch on juveniles is significant, the equilibrium analysis was performed manually.  Table 3.1 - Fishery indicators for major commercial groups in the 2006 RA model.  Values are determined by the equilibrium analysis.  Asterix indicates that the equilibrium values were determined manually and fishing mortality was incremented for  stanzas (see te  MSY Fmsy F2006 F2006/Fmsy 2006 catch # Group (t·km-2) (year-1) (year-1)  (t·km-2) 10 Ad. groupers 0.027 0.207 0.094 0.454 0.017 13 Ad. snappers 0.008 0.210 0.155 0.735 0.014 16 Ad. Napoleon wrasse* 0.002 0.228 0.085 0.372 0.001 17 Sub. Napoleon wrasse* 0.003 0.244 0.048 0.001 19 Skipjack tuna 0.366 0.548 0.348 20 Other tuna 0.058 0.251 0.097 0.385 0.047 21 Mackerel 0.058 0.746 0.746 1.000 0.064 22 Billfish 0.068 0.147 0.061 0.417 0.050 23 Ad. coral trout 0.002 0.092 0.045 0.484 0.002 25 Ad. large sharks 0.010 0.476 0.971 2.041 0.025 33 Ad. butterflyfish 0.079 0.553 0.060 0.109 0.016 36 Ad. large pelagic 0.023 0.575 0.575 1.000 0.031 38 Ad. medium pelagic 0.008 40 Ad. small pelagic 0.042 42 Ad. large reef assoc. 0.343 44 Ad. medium reef assoc.* 0.824 46 Ad. small reef assoc. 0.371 48 Ad. large demersal 0.040 50 Ad. small demersal 0.247 2.868 0.211 0.074 0.028 52 Ad. large planktivore* 0.478 0.700 0.300 0.429 0.339  55 Juv. small planktivore* 0.223 0.600 0.002 0.004 0.001 56 Ad. anchovy 0.887 1.218 0.391 0.321 0.509 0.115 0.450 0.034 0.075 0.022 59 Juv. deepwater fish* 0.158 0.300 0.016 0.055 0.014  all life history xt). 0.197 1.144 0.479 1.276 1.383 1.084 0.014 1.154 0.825 0.714 0.034 0.178 0.081 0.455 0.577 0.400 0.123 0.307 0.438 2.422 0.142 0.059 0.019 0.561 0.679 1.210 0.024 53 Juv. large planktivore* 0.452 0.700 0.034 0.048 0.034 54 Ad. small planktivore* 0.130 0.600 0.031 0.052 0.013 58 Ad. deepwater fish* 60 Ad. macro algal browsing* 0.033 0.250 0.003 0.013 0.001 61 Juv. macro algal browsing* 0.034 0.300 0.000 0.001 0.000 62 Ad. eroding grazers* 0.056 0.250 0.003 0.013 0.000 63 Juv. eroding grazers* 0.036 0.250 0.000 0.001 0.000 64 Ad. scraping grazers* 0.092 0.700 0.094 0.134 0.022 65 Juv. scraping grazers* 0.495 0.800 0.002 0.002 0.002    Challenges to Ecosim  Fig. 3.4 shows synoptic results of the challenges to Ecosim.  It summarizes the biomass change in reef associated fish functional groups (including specific and aggregated reef groups), pelagic fish, predator fish (TL > 3), forage fish (TL 2-3), invertebrates and mammals.  The ‘no fishing’  Bird’s Head Seascape Analyses, Page 85 scenario (0F) completes the simulation with the highest standing biomass for exploited species groups such as reef fish, pelagic fish and high trophic level fish.  The ‘increasing fishing’ scenario Reef-associated fish Fish TL 2-3 (F+) ends with depressed biomasses for ighest in the ‘increased fishing’ ig. B.2.4 shows biomass predictions for three shing scenario challenges.  The error e mean; the mean is indicated by an crease relative to baseline biomass for the ‘no fishin ario.  Only t absolute biomass values of coral trout and large ree appear unsatisfactory; cover under elieved fis ing pressu e.  As the model is improved with field data, cs will be revisited for these groups.  Table 3.2 shows the depletion risk of functional groups associated wit e thre ing scenarios.  The cenario has th ewest n mber of d letions,  the lea evere d pletions.  U  fishing cond tions, whe e current  fish  morta s are c rried on unt nappers drop below 30% of their current biomass value % of trials.  Also th ral trout and large sharks, which ach decline to 40% eir c biomas e conditions in 6% of simulations.  Under increasing fishing mortality, the depletions are more s re.  Snappers are prone to collapse to 15%  current biomass value in 80% of trial while m el, coral trout, sharks and larg  serious depletion risk.  Large reef associated fi erform poorly in most simulatio equires further tuning.  F shing polic s ig. 3.5 shows the relationship between the expected NPV from the optimal harvest policy, and cosystem maturity B/P.  156 optimal policies have been computed based on a random F vector 100% 105% io m as s 100% 105% io m as s BB these exploited groups.  Forage fish, with trophic levels between 2 and 3 are the prey for piscivores, and their biomasses do see an appropriate decrease once predators recover.  Likewise, the invertebrate biomasses are 95% 0F 1F F+  95% 0F 1F F+  Pelagic fish Invertebrates 100.4% 200% h 0% 100%B io m as 0F 1F F+ s  99.2% 99.6%B io m as 100.0% 0F 1F F+ s  Fish TL 3+ Mammals scenario, when predators have been removed from the system, and lowest in the ‘no fishing’ scenario, when predators are allowed to recover.  F functional groups under the 140% fi bars show the variation in biomass trajectories predicted by the Monte Carlo analysis, when Ecopath biomass parameters are allowed to vary for key groups by +/- 20%.  The error bars represent 1 standard deviation around 80% 100% 120% 0F 1F F+ B io m as s  95% 100% B io m as s th open circle.  All groups see an appropriate decline in biomass relative to baseline levels for the ‘increased fishing’ scenario; all groups see an in g’ scen he  f-associated fish  they should re r h r  biomass dynami h th e fish  ‘no fishing’ s e f u ep and st s e nder baseline i r 2006 ing litie a il 2022, adult s  in 6 reatened are co  e  of th urrent s under baselin    eve  of their s, acker 105% 0F 1F F+  Figure 3.4.  Group biomass change following    ss for commercial fish.  Simulations are from 2006-2022; extreme fishing policies (2006-2022).  No fishing (0F), baseline fishing (1F) and increasing fishing (F+).  Mean biomass values are shown that result from a Monte Carlo that varies input bioma biomass is relative to baseline endstate. e pelagics all a sh p ns; this group r i y optimization  F e  Page 86, Fisheries Centre Research Reports 15(5), 2007 starting.  More valuable harvest plan depletion of slow-growing animals su reported in Fig. 3.5 for compariso economic nor ecological benefits ar  Table 3.2 - Group depletion risk (in %) fishing (1F) and increasing fishing (F functional group declined to a given leve depletion is stated relative to baseline m   Group     s tend to result in a lower ecosystem maturity score due to a ch as large predators.  The current ecosystem state is n.  The current RA fisheries appear sub-optimal; neither e being realized to their full potential.    following extreme fishing scenarios.  No fishing (0F), baseline +).  Depletion risk shows the percentage of times that each l of biomass during Monte Carlo simulations (n = 50).  Biomass odel biomass.  Adult (ad.); sub-adult (sub.); juvenile (juv.). End-state biomass (2022) vs. baseline model (2006) 15% 20% 30% 40% 50% 0 F Ad. coral trout     16  Juv. coral trout     12  Juv. large sharks     44 1 F Ad. snappers   6 100 100  Sub. snappers    20 62  Juv. snappers    28 64  Ad. coral trout    6 50  Juv. coral trout     38  Ad. large sharks    6 18 F+ Ad. snappers 80 100 100 100 100  Sub. snappers  8 64 92 98  Juv. snappers  12 72 94 96  Mackerel   2 54 94  Ad. coral trout    10 76  Juv. coral trout     46  Ad. large sharks    8 32  Ad. large pelagic    52 100  Juv. large pelagic    8 56    A convex relationship appears between the two criteria, suggesting that a win-win harvest policy may exist that will generate the greatest economic return while preserving the ecosystem.  However, results could change with model improvements, and no such win-win policy exists with respect to total catch and biodiversity.  Fig. 3.6 re-expresses the benefits of the optimal scenarios in these terms, and indicates a linear relationship.  The correct fishing policy to employ remains a matter of social priority.    Results suggest that under an optimal fishing policy, the RA ecosystem could sustainably deliver more catc 0.0 0.5 1.0 1.5 2.0 2.5 0.99 1 1.01 1.02 1.03 1.04 1.05 B/Popt / B/Pbase N P V op t/N P V ba se  Figure 3.5 - Tradeoff between NPV and B/P.  Points are determined through policy optimizations.  Open circle show current RA fishery (sub-optimal), black points show the trade-off frontier (optimal). h that it currently does.  he high degree of biodiversity estimated for the 2006 RA ecosystem T  Bird