Phil. Trans. R. Soc. B. article template Phil. Trans. R. Soc. B. doi:10.1098/not yet assigned Modelling MPAs: Insights and Hurdles Elizabeth A. Fulton1,2, Nicholas J. Bax1,3, Rodrigo H. Bustamante4, Jeffrey M. Dambacher5, Catherine Dichmont4, Piers Dunstan1, Keith R. Hayes1, Alistair J. Hobday1,2, Roland Pitcher4, Éva E. Plagányi1, André E. Punt1,6, Marie Savina-Rolland7, Anthony D.M. Smith1,2, David C. Smith1,2 1. CSIRO Oceans & Atmosphere, GPO Box 1538, Hobart, Tasmania, 7001, Australia Centre for Marine Socioecology, University of Tasmania, 20 Castray Esplanade, Battery Point, Tasmania, 7004, Australia 3. Institute for Marine and Antarctic Studies, University of Tasmania, 20 Castray Esplanade, Battery Point, Tasmania, 7004, Australia 4. CSIRO Oceans & Atmosphere, PO Box 2583, Brisbane, Queensland, 4001, Australia 5. CSIRO Digital Productivity, GPO Box 1538, Hobart, Tasmania, 7001, Australia 6. University of Washington, School of Aquatic & Fishery Sciences, Box 355020, Seattlw, WA 98195-5020 7. Laboratoire Ressources Halieutiques, Centre Manche - Mer du Nord, 150, quai Gambetta, BP 699, 62321 Boulogne sur Mer cedex, France 2. Keywords: Spatial management, modelling, MPA Summary Simulation models provide a useful means of gaining insights into conservation and resource management. Their use to date has highlighted conditions where no-take marine protected areas may help to achieve the goals of ecosystem-based management by reducing pressures, and where they might fail to achieve desired goals such as when static reserve designs are applied to mobile species compromised by climate-related ecosystem restructuring and range shifts. Modelling tools allow planners to explore a range of options, such as basing closures on presence of dynamic oceanic features, and to evaluate the potential future impacts of alternative interventions in comparison with “no-action” counterfactuals, under a range of environmental and development scenarios. The control possible in the modelling environment makes it highly useful for exploring how robust indicators and management strategies may be to uncertainties in how the system operates. Moreover, simulation model results can be presented at multiple spatial and temporal scales, as well as relative to ecological, economic and social objectives. In this way, potential ecological ‘surprises’, such as regime shifts and bottlenecks in human responses can be revealed. Using illustrative examples, this paper briefly covers the history of the use of simulation models for evaluating MPA options, and discusses their utility and limitations to inform protected area management in the marine realm. Introduction Marine resources are used for food, energy and recreational purposes, and are valued for existence and cultural reasons. The systems they are part of, also play an important regulatory role in the world’s climate [1-3]. Evidence of impacts from the many pressures on marine ecosystems is both direct [1] and indirect [4], with some impacts mediated through marine food webs and biogeochemical cycles. Fishing, pollution and eutrophication have resulted in clear impacts on coasts, estuaries, and enclosed seas [5]. Open ocean impacts are also being increasingly recognised – in the form of marine debris [6], ocean warming and acidification [7-8]. These pressures are typically being managed through national and international forms of governance using sector- or pressurespecific management measures, such as fisheries management. *Author for correspondence (beth.fulton@csiro.au). †Present address: CSIRO Oceans & Atmosphere, GPO box 1538, Hobart, Tasmania, Australia, 7001 1 R. Soc. open sci. article template 2 Modelling MPAs: Insights and Hurdles Unfortunately, the broad range of pressures inevitably leads to conflicts between sectors, such that individual sectoral objectives may no longer be achievable [9]. Even relatively well-managed marine systems such as Australia’s Great Barrier Reef are recognised as having a poor future due to the cumulative pressures of marine stressors, and impacts from terrestrial land use [10]. In more intensively used regions of the world, the desire to add new industries, including renewable energy or seabed mining, in addition to existing uses is increasing the complexity and quantum of cumulative impact. In response, integrated forms of ocean management are being attempted, with marine spatial planning becoming a more prominent tool of choice [11], particularly in crowded coastal seas [9]. Spatial management is one of several tools available to managers to reduce potential conflict and cumulative impacts, and is one that is relatively straightforward to apply across sectors [12]. Marine protected areas (MPAs), one spatial approach to managing human pressures, have been embraced globally, with goals to include 10% of the world’s seas and oceans in MPAs by 2020 [13]. Modelling can be useful for informing the use of such management tools, helping to develop an understanding of how individual decisions may impact the broader ecosystem. This paper will discuss the strengths and weaknesses of using modelling to investigate the performance of (mostly) no-take MPAs, considered to be the most ‘extreme’ form of spatial management. The clear demarcation of no-take MPAs make their impacts easier to assess, thus they provide an exemplar of the benefits and challenges in modelling spatial management more generally, highlighting the issues relevant to (or even amplified) other spatial management systems. Models of many kinds have been used to design and evaluate MPAs. Perhaps the most common are optimisation tools that support systematic conservation planning, such as Marxan [14] and Zonation [15]. Statistical and geo-statistical approaches have been used to map the distributions of key conservation species [16], identifying bioregions to be captured in representative networks of MPAs [17] and identifying locations that may simultaneously service many conservation objectives [18]. While all of these tools have been used quite widely and effectively to support MPA design there is also a rich literature discussing them [19-20] and we will focus here on dynamic modelling approaches used to predict the impacts of the changed management arrangements. Objectives for modelling Models are a good way of describing and improving our understanding of the how MPAs may perform against objectives, synthesising information and drawing it together in a coherent form so that it can be used to understand particular questions of interest (Table 1). Many forms of model can be used to evaluate MPA performances. Selecting the appropriate model would be a lengthy paper in itself and has been tackled elsewhere in depth [51-54]. In brief, we concur that both the scale of the question and data availability should dictate the form of the model used. In addition, we recognize that models are constrained by the impossibility of simultaneously maximising all three of Levin’s modelling goals: generality, precision and realism [55]. These constraints endure despite the advances in modelling methods and resources – such as data assimilating methods [56] and end-to-end models [43] – which now allow for observations and models to be coupled, and ecosystems to be represented at temporal and spatial scales undreamt of previously. Conceptual models continue to be the fundamental building blocks of all modelling exercises. An end in themselves for synthesising understanding, they are also the means of defining the content of quantitative modelling approaches. The graphical nature of signed diagraphs, increasingly used as a means of codifying conceptual models as qualitative models [57], are a particularly effective way to understand systems that transcend the boundaries between, and backgrounds of, different stakeholder groups. This common understanding of connections and potential feedbacks and trade- Phil. Trans. R. Soc. B. Proc. R. Soc. B article template 3 offs can then act as a useful starting point for planning and discussions of options, which in turn can be further supported by quantitative models. The content and the complexity of quantitative models has typically grown through time, as computational capacity has increased, but is still largely dictated by whether the tool is being used to help design a network of MPAs or to consider aspects of the performance of MPAs. The former may be spatially detailed on quite fine scales, but contain few physical or ecological processes, while the latter can be quite complex once trophic, habitat and human uses and behavioural responses have been added, though data availability might limit their resolution (Figure 1). Ultimately, a diversity of modelling approaches, which together span generality, precision and realism remains the best means of informing decision making around MPAs, with each model addressing specific roles rather than everything simultaneously (Figure 1): using qualitative models to maximise realism and generality conceptually, statistical and minimally realistic models to maximise realism and precision in tactical roles, and large scale mechanistic (e.g. end-to-end) models to maximise precision and reality in strategic roles. Modelling strategies with a broad scope, using model suites that sacrifice a little in all three dimensions, and with acknowledgement of benefits and drawbacks associated with individual models, should be used to explore issues associated with MPAs. Benefits Perhaps the greatest strength of a model-based approach is that the simulation environment can act as common ground for discussions, enabling critical questions to be addressed and acting as an important precursor to evidence-based decision making. Even when the model results are not sufficiently reliable to inform specific decisions, the process of assembling the data will synthesise information and theories, identify missing and contradictory information and highlight beliefs and opinions that are not currently supported by data. This is especially true when retrospective analyses of the effects of real world MPAs is used to inform model dynamics and projections. Models can represent MPAs at scales beyond the capacity of field studies (e.g. at global scales). By moving understanding beyond the sometimes misleading comparison of conditions immediately inside and out of small no-take MPAs, models can help elucidate i) general patterns of performance; ii) the potential for unintended consequences that inadvertently undermine management intentions; and iii) how MPA networks in combination with other management actions can influence system state and service conservation, fisheries and other social, economic and environmental objectives (see Table 2 for examples of model-based MPA findings). These insights can then be used to shape future decision-making and adaptive management. Qualitative modelling [21] is one method for synthesising understanding in the absence of sufficient data for fully quantitative dynamic models. This approach has been usefully applied in support of conservation planning [71], climate change implications [72] and conceptual understanding of the reasons for success and failure of MPAs [46]. Many management relevant modelling results have ensued from model-based counterfactuals, where the evolution of a modelled system is tracked with and without the specific management action of interest, or under differing levels of perturbation. This approach is a means of exploring options in safety, under conditions that are hard to observe or have not yet been experienced. This has made the method particularly appealing as a way to discern the shape of future barriers and the nature of future opportunities, and the trade-offs associated with alternative management options under uncertainty. This leads to an increased willingness to go beyond minor modifications to existing arrangements, as there is an easy freedom to changing management regulations in a model that cannot be matched in reality. The model-based approach also provides a dynamic context for discussing potential outcomes, highlighting how the performance of a management option, such as a Proc. R. Soc. B R. Soc. open sci. article template 4 Modelling MPAs: Insights and Hurdles no-take MPA, may vary through time and depend on context. In addition, the influence on management outcomes of decision uncertainty and ambiguity around system structure and function can be dealt with explicitly [73]. This is typically done by considering the overlap in outcomes across ensembles of models (encapsulating different theories about system processes, including climate). This can indicate whether similar outcomes are repeatedly realised or whether performance is sensitive to poorly understood system details. Understanding built up in this way, from a set of diverse models, can be used to establish a common understanding of a system’s characteristics, promoting more informed discussion over contentious issues, regardless of whether or not one particular model prediction is accepted. End-to-end ecosystem models are increasingly being used to, provide counterfactuals under impending global change. One example is [43]. This study used multiple productivity and food web parameterisations of two Atlantis ecosystem models of southeast Australia to explore nine management strategies, three climate (emission) scenarios (RCP 3, 4.5 and 8.5 [74]) and eight system scenarios – including low, moderate and high industrial development and population growth scenarios, with and without market shifts and catastrophic extreme events. The status quo (business as usual in 2010) was used as reference, to show what would be gained or lost in each simulated future versus “change nothing”. This showed that targeted management options, such as no-take MPAs, can perform well for individual management objectives (e.g. extensive spatial management can lead to improved stock status for large-bodied habitat-associated predatory fish), but they do not successfully meet minimum requirements across multiple objectives (such as the status of prey species, catch composition, equity of access or employment). [43] found that reducing the physical extent of spatial management was a universally poor management action. Although, [43] also found that static spatial zoning currently used as the basis of conservation management in the region was not well suited to the more fluid nature of future marine ecosystems, potentially making some existing reserves less effective and others completely ineffectual. Even when range shifts did not directly degrade the level of protection provided by a no-take MPA, their performance may be degraded by non-compliance – a situation modelling suggests is more likely with an increasing number of uses or the intensity of use (and thus competition for space). Compliance can decline where the community or industry believe that no-take MPAs are unduly constraining their ability to respond to new circumstances and opportunities. Model findings allow stakeholders to anticipate challenges to the effective use of MPAs ahead of time, providing time to prepare or adapt. For instance, many MPAs in Australia are defined to include key ecological features. In no-take MPAs models can help determine whether climateinduced changes in species composition influences the role and importance of the feature – and thus whether it still needs to be protected. For example, there would be pressure to dissolve existing MPAs instituted to help protect particular vulnerable species, and establish them elsewhere if the kind of performance failure seen in [43] was realised in reality. On the other hand, MPAs designed to protect specific vulnerable habitats may retain their value under climate change. There is significant inertia associated with the declaration of MPAs, e.g. in Australia, where their boundaries have to be formally gazetted in parliament and this has led to the scientific proposition that pelagic MPAs may be defined around oceanographic features (or community states) rather than geographic coordinates [75-76]. It is possible for regulatory bodies to translate model findings into practical trials around new management concepts in fast warming areas [77]. Rather than closing off large areas, the Australian Fisheries Management Authority spatially manages Southern Bluefin Tuna by integrating information from habitat models (e.g. based on oceanography), recent industry-species interactions and historical knowledge to dynamically close smaller areas for shorter periods of time. The result is a reduction in unwanted interactions, improved efficiently of operations [76] and a management approach that will be robust to shifting habitats and species under climate change [75, 78]. Models can also assess the costs associated with the use of MPAs and society’s ability and willingness to pay. For example, fairly conventional economics models have traditionally been used to quantify the cost-benefit trade-offs of no-take MPA's, such as benefits from tourism, but also the Phil. Trans. R. Soc. B. Proc. R. Soc. B article template 5 cost to industry of placing no-take MPA's in previously fished areas [69]. Similarly, models can also quantify costs of novel forms of management, which may rely on continuously updating data streams [79] and how those costs may be ameliorated via the use of proxy-based approaches. This is important, as logistics and the cost of informing management are an on-going concern for modern evidence-based decision-making and are difficult to estimate at the system level [80]. Models can further assist by identifying informative monitoring schemes and quantifying associated costs [81]. The Atlantis modelling framework was used to determine appropriate monitoring schemes (in terms of frequency and spatial extent), and indicators for assessing the performance of spatial management that were robust to a wide range of environmental and anthropogenic scenarios [47]. The results of this work indicated that monitoring MPAs under global change may be far from simple. Sampling schemes of low frequency or spatial coverage could detect change inside and outside closures once sufficient time series had accumulated to enable causes of the signal to be evaluated. However, they had little power to detect signals at broader spatial scales and had no power to rapidly detect changes in the system. Moreover, a lack of a temporal dimension in monitoring cannot generally be completely compensated for by periodically applying very intensive surveys across broad spatial scales, as intensive sampling is confounded by natural system variation and shifts through time. The modelling results also showed that ecosystem shifts in response to changing climate drivers, mean that reference points (or indicator-attribute relationships) will need to be adjusted as the system changes (otherwise they run the risk of becoming irrelevant or misguided). Drawbacks Some drawbacks extend beyond the modelling sphere. Chief among them is a lack of clearly defined operational objectives for MPAs [82]. It is almost impossible to demonstrate that objectives as vague as “increasing biodiversity” have been achieved. The formality of modelling clearly identifies such vagueness. Turning to modelling, one of the greatest challenges remains the need to address managementrelevant scales. Despite the difficulty of working at fine scales there is nevertheless a demand for such information and this may require combining a diversity of modelling approaches. Process model based thinking from qualitative and mechanistic (e.g. end-to-end) models is key to developing informative statistical models – i.e. exactly what relationships and parameters should be measured and why [83]. Once developed, statistical models can be useful for setting trigger points and thresholds, identifying trends in indicators and ultimately (in)validating the predictions and correlated responses of the quantitative or qualitative mathematical models [48]. Empirical statistical models have been used to great effect at management relevant scales [24], but they also have their limitations. The quantum of data needed to demonstrate significant effects can be large at the ecosystem scale or for elements with significant variation. Statistical models are also of limited utility by themselves when projecting beyond the bounds of the data used to define them, meaning that models used to inform on future MPA issues are currently often process based (e.g. [43]). However, it is difficult to capture the fine spatial scales typical of many MPAs in process models as there is insufficient data or understanding at such spatial, temporal or taxonomic scales. These technical impediments mean that it is hard to resolve and represent small no-take MPAs (globally the median area of individual MPAs is < 5 km2 [84]) in models, and thus to show that reserves have any effect. It may well be true that such small no-take MPAs have limited effect (especially at broader scales), as suggested by meta-analysis [85]. However, when presented as a model result, the modeller is left wondering whether it is a true reflection of the system or a model artefact. This can lead to confusion in stakeholders and may not provide the kinds of information that MPA managers need. Proc. R. Soc. B R. Soc. open sci. article template 6 Modelling MPAs: Insights and Hurdles The challenge of knowing how much faith to put in model results is a constant concern. This is especially true of models that are not fitted to data (or do not match data well). Unfortunately, the novel nature of potential future ecosystem states means that many models used to consider future states have an unknown veracity, even if well fitted to current data. Models considering strategic and conceptual questions are no more immune to these problems than tactical models. The lack of good empirical data hampers dynamic quantitative modelling of all kinds, as it can be a relatively expensive and data intensive exercise. This is particularly true when trying to inform on future as yet unobserved system states. Even without extreme environmental change, data are sparse or lacking for many species or life history stages, particularly those that are not fishery targets or of particular conservation concern. Fisheries dependent data and remote-sensed data are available and of increasing spatio-temporal density. However, these data are often not representative of variables that are most biologically relevant [86]. Similarly, observer programs mean that, while there is often considerable data for marine mammals and seabirds, entire invertebrate families, orders and even classes can be depauperate of data beyond presence/absence data. This does not mean modelling is impossible, just that it has to be done carefully and with due attention to the handling of uncertainty and clear statements as to its veracity and limitations. Management decisions can be made in the absence of quantitative models but they will also suffer from knowledge gaps, even if this is not explicitly recognised. While easy to perform in data poor conditions, qualitative modelling also has its limitations. For instance, the kind of complex ecological outcomes leading to system-specific relationships between indicator and desirable system attribute uncovered in [47] could not be dealt with a priori with this method. The need to include relevant processes and scales bedevils strategic process based modelling as much as tactical modelling. One way of addressing this problem is to constrain model extent. For instance, models of intermediate complexity [44] are useful tools for rigorously analysing a subset of an ecosystem and focusing on specific questions. However, the relevance of any model result needs to be interpreted in the context of the purpose for which the model was built, as well as the chosen spatial and temporal scales. Thus, intermediate complexity models would not typically be suited to addressing broader biodiversity questions. [41] used an age-structured model to investigate the consequences of no-take MPA’s for the South African deepwater hake Merluccius paradoxus, a relatively mobile species. Area closures were found to have a negligible benefit for the hake fishery in the area. However, a more complex model is required to fully quantify the effects of a closure on aspects of the system and for drawing conclusions regarding potential effects on the broader ecosystem and any concomitant conservation benefits of MPA implementation. The modelling challenge is larger still when contemplating the explicit inclusion of uncertain biological processes (e.g. movement, evolution etc.) and the complexity of human jurisdictional and regulatory arrangements and responses to them. Management of MPAs can be very complicated, involving a large number of ill-defined and potentially contradictory contributions accumulated over time and groups with differing objectives. Fine-scale tactical management detail is difficult to implement in models, but may significantly influence human behaviour and hence the effectiveness of the overall management package. It is possible to model some aspects of human behaviour, thereby reducing some forms of uncertainty [87]. However, other uncertainties remain, as this is still a relatively new discipline; and it shares the common model challenges when projecting forward into novel (previously unobserved) conditions. These impediments do not mean that human responses to the implementation of no-take MPAs should be ignored, as they will be a key determinant of success. Here again, attempting to predict their impact with models informs how MPAs will work in the broader socio-political setting and how compliance could be influenced to improve the benefits of spatial management [31] (Table 2). MPAs, and spatial and ecosystem based management more broadly, is a government process, and science frequently does a poor job at interfacing with that process. Scientists tend to focus on what is Phil. Trans. R. Soc. B. Proc. R. Soc. B article template 7 innovative, how new methods can resolve existing problems and frequently advocate adoption of their new ‘optimal’ approach without considering the broader social-political setting. Conversely, governments embrace established process and look for a variety of scientific options that they can choose between to satisfy a diversity of stakeholders. Models should therefore not only provide scientific information in support of management, but new kinds of models are needed that describe the management process and the key points for the insertion of scientific information. Such models are in their infancy in the conservation and MPA arena [88]. However, extensive experience in fisheries shows such models are possible [89], but they are not simple and still require extensive explanation and communication. When used well, these kinds of models are a very effective means of defining useful questions, identifying measureable operational objectives and supporting evidence-based decision making – with the models playing the role of an honest broker, increasing the number of options available to decision makers so that they can identify the option that also meets their needs [90]. Discussion and Conclusions Modelling may have many constraints and the representation of details of most immediate interest to managers is exceptionally challenging. However, they remain one of the most effective ways science can provide support to the creative freedom necessary to find solutions for the novel situations we face in the future. To address the myriad questions being posed of models will require a diversity of complementary modelling approaches. Coupling simulation and qualitative models, with statistical models has the potential to provide significant insight into the dynamics and state of ecosystems and monitor their responses to the implementation of MPAs. Such combinations are not yet common. Instead there are six rough classes of applications across the literature dealing with models of MPAs. The most common four uses are for MPA design, assessments of potential ecological benefits, bioeconomic assessments (including human responses to the establishment of an MPA) and management evaluations. Dynamic models of no-take MPAs have also been used to design adaptive management experiments [49] or as the basis of discussions of modelling philosophy – around the contextual usefulness of modelling types when aiming to providing insights on MPAs [28]. The evolution of MPA modelling is instructive, as it mirrors both technological capacity and an increasing realisation of how complex the questions of interest are. Models have moved from abstract conceptualisations to issues of MPA design and quantitative evaluation of MPA impact [51]. The use of equilibrium models to explore the potential fisheries and conservation benefits of closures to individual species dominated the model-based consideration of no-take MPAs until the late 1990s [58]; with the consistent finding that reserves benefit overfished stocks, but that these benefits (and costs) were dependent on the rate of fish movement [52] and how fishers reallocated the displaced effort [60]. Improved computing capacity saw more and more species distribution models [91] or connectivity models [92] developed to explore MPA network design. The turn of the 21st century saw a step change in the use of dynamic models for exploring issues associated with MPAs, driven by an increasing focus on spatial management as a conservation tool by non-governmental organizations and international legal requirements [93]. It was at this point that models grew from analytical considerations [94] to include habitats, multiple species or food webs and more sophisticated fleet dynamics [40, 95]. While some conservation questions remain focussed on single species issues, others address more complicated issues associated with one or more aspects of the multitude of marine genotypes, species and ecosystems with their varied distributions, abundances and life histories, i.e. biodiversity. In parallel to the expanding ecological scope, the bioeconomic models have extended beyond straightforward treatments of economic costs of no-take MPAs [33], to consider the role of spatial management as part of integrated management regimes within and outside marine reserves [96]. This approach is taken further still in management strategy evaluations Proc. R. Soc. B R. Soc. open sci. article template 8 Modelling MPAs: Insights and Hurdles that consider MPAs along with other management options [42, 97, 98]; some of these go so far as to consider how MPAs may be designed in the first place [88]. It is a natural next step that these same kinds of models can be used to investigate future management approaches – their potential outcomes, benefits and when they may fail to meet management objectives. The exploration of the utility of MPAs under climate change, and associated regime shifts, has heightened debate around the value of dynamic spatial management, including dynamic MPAs. Given the shifts in marine habitats expected under climate change [75], accounting for future environments has been a recent focus. Climate layers are being added to Marxan to try and find network designs that are robust to future shocks [99], but more attention has been given to adaptive approaches, such as slow-moving MPAs that change along with the environment [75] or the dynamic zoning based on oceanographic features [79]. Modelling is useful in such discussions but is not enough by itself, especially given the messy perception of MPAs and institutional complexity, multiple objectives and uncertainty about future system dynamics. Modelling should be used to start and support discussions around management options for and pressures on a system and ideally model outputs will be an important part of the solution that managers will then adapt to fit their broader reality. Models also allow for comparison of management processes, such as monitoring schemes and quantification of associated costs. Models support and enhance the learning that comes from such joint discussions, especially across diverse stakeholder groups, emphasizing the necessity to set clear objectives around what is desirable, or at least acceptable. What modelling has shown us is that even in a simulated environment (far simpler than the real world), the success of MPAs depends on what you are trying to achieve (i.e. your objectives) and how clearly you express the why and the wherefore. Additional Information Acknowledgments The authors would like to thank their many collaborators on the various modelling projects touched on in this paper, in particular L.R. Little, E. Morello and P. Johnson and to those who have helped shape earlier versions of the manuscript. Funding Statement The Commonwealth Scientific and Industrial Research Organisation (CSIRO) Australia has provided funds in support of all authors. Work by NB, PD, JD, KRH and RP included in this paper has been supported by the Australian Government’s National Environmental Research Program through the Marine Biodiversity Hub. The research by EAF, RHB, CD, AH, ÉEP, AEP, ADMS and DCS discussed in this paper has been supported by funding from the Australian Fisheries Research and Development Corporation. Finally the NSW based work of EAF and MSR was supported by the New South Wales Department of Primary Industries. Competing Interests We have no competing interests. Authors' Contributions All authors participated in the drafting of the outline of this article. EAF drafted the text, which was then revised and approved for publication by all other authors. References 1. 2. 3. Jackson JBC. 2008. Ecological extinction and evolution in the brave new ocean. PNAS 105, 11458–11465 (DOI: 10.1073pnas.0802812105) Reid PC, Fischer AC, Lewis-Brown E, Meredith MP, Sparrow M, Andersson AJ, Antia A, Bates NR, Bathmann U, Beaugrand G, Brix H, Dye S, Edwards M, Furevik T, Gangstø R, Hátún H, Hopcroft RR, Kendall M, Kasten S, Keeling R, Le Quéré C, Mackenzie FT, Malin G, Mauritzen C, Olafsson J, Paull C, Rignot E, Shimada K, Vogt M, Wallace C, Wang Z, Washington R. Chapter 1. Impacts of the oceans on climate change. Adv. Mar. Biol. 56,1-150. (DOI: 10.1016/S0065-2881(09)56001-4). Pelc R, Fujita RM. 2012. Renewable energy from the ocean. Mar. Pol. 26, 471-479. Phil. Trans. R. Soc. B. Proc. R. Soc. B article template 9 4. Waycotta M, Duarte CM, Carruthers TJB, Orth RJ, Dennison WC, Olyarnik S, Calladine A, Fourqurean JW, Heck Jr KL, Hughes AR,. Kendrick GA, Kenworthy WJ, Short FT, Williams SL. 2009. Accelerating loss of seagrasses across the globe threatens coastal ecosystems. PNAS 106, 12377–12381. (DOI: 10.1073pnas.0905620106). 5. Halpern BS, Longo C, Hardy D, McLeod KL, Samhouri JF, Katona SK, Kleisner K, Lester SE, O’Leary J, Ranelletti M, Rosenberg AA, Scarborough C, Selig ER, Best BD, Brumbaugh DR, Chapin FS, Crowder LB, Daly KL, Doney SC, Elfes C, Fogarty MJ, Gaines SD, Jacobsen KI, Karrer LB, Leslie HM, Neeley E, Pauly D, Polasky S, Ris B, St Martin K, Stone GS, Sumaila UR, Zeller D. 2012. An index to assess the health and benefits of the global ocean. Nature 488, 615-620. (DOI: 10.1038/nature11397) 6. Eriksen M, Lebreton LCM, Carson HS, Thiel M, Moore CJ, Borerro JC, Galgani F, Ryan PG, Reisser J. 2014. Plastic pollution in the world’s oceans: more than 5 trillion plastic pieces weighing over 250,000 tons afloat at sea. PLoS ONE 9(12): e111913. (DOI:10.1371/journal.pone.0111913) 7. Schiel DR, Steinbeck JR, Foster MS. 2004. Ten years of induced ocean warming causes comprehensive chanes in marine benthic communicates. Ecol. 85, 1833-1839. 8. Kroeker KJ, Kordas RL , Crim R, Hendriks IE, Ramajo L, Singh GS, Duarte CM, Gattuso JP. 2013. Impacts of ocean acidification on marine organisms: quantifying sensitivities and interaction with warming. GCB 19, 1884-1896. (DOI: 10.1111/gcb.12179). 9. Ban NC, Maxwell SM, Dunn DC, Hobday AJ, Bax N, Ardron J, Gjerde KM, Game ET, Devillers R, Kaplan DM, Dunstan PK, Halpin PN and. Pressey RL. 2014. Better integration of sectoral planning and management approaches for the interlinked ecology of the open oceans. Mar. Pol. 49: 127-136 http://dx.doi.org/10.1016/j.marpol.2013.11.024. 10. Great Barrier Reef Marine Park Authority 2014, Great Barrier Reef Outlook Report 2014, GBRMPA, Townsville. 309 pp. 11. Zaucha J. 2014. Sea basin maritime spatial planning: A case study of the Baltic Sea region and Poland. Mar. Pol. 50, 34-45 (DOI: 10.1016/j.marpol.2014.05.003) 12. Agardy T. 1994. Advances in marine conservation: the role of marine protected areas. TREE 9, 267-270. 13. CBD. 2010. Strategic Plan for Biodiversity 2011-2020. 14. Ball IR, Possingham HP, Watts M. 2009. Marxan and relatives: Software for spatial conservation prioritisation. In Spatial conservation prioritisation: Quantitative methods and computational tools, Moilanen, A., K.A. Wilson, and H.P. Possingham (Eds). Oxford University Press, Oxford, UK. Chapter 14: Pages 185-195. 15. Moilanen A, Kujala H, Leathwick JR 2009. The zonation framework and software for conservation prioritization. In A Moilanen, KA Wilson, HP Possingham (eds), Spatial conservation prioritization. Oxford University Press, Oxford, pp. 196-210. 16. Alessi J, Fiori C. 2014. From science to policy - a geostatistical approach to identifying potential areas for cetacean conservation: a case study of bottlenose dolphins in the Pelagos sanctuary (Mediterranean Sea). J. Coast. Conserv. 18, 449-458. (DOI: 10.1007/s11852-014-0330-3). 17. Last PR, Lyne VD, Williams A, Davies CR, Butler AJ, Yearsley GK. 2010. A hierarchical framework for classifying seabed biodiversity with application to planning and managing Australia’s marine biological resources. Biol. Cons. 143, 1675-1686. (DOI: 10.1016/j.biocon.2010.04.008) 18. Parravicini V, Rovere A, Vassaloo P, Michelo F, Montefalcone M, Morri C, Paoli C, Albertelii G, Fabiano M, Bianchi CN. 2012. Understanding relationships between conflicting human uses and coastal ecosystems status: A geospatial modeling approach. Ecol. Ind. 19, 253-263. 19. Sarkar S, Pressey RL, Faith DP, Margules CR, Fuller T, Stoms DM, Moffett A, Wilson KA, Williams KJ, Williams PH, Andelman S. 2006. Biodiversity conservation planning tools: present status and challenges for the future. Annu. Rev. Environ. Resour. 31,123-59. (DOI: 10.1146/annurev.energy.31.042606.085844) 20. Ban NC, Klein CJ. 2009. Spatial socioeconomic data as a cost in systematic marine conservation planning. Cons. Lett. 2, 206-215. (DOI: 10.1111/j.1755-263X.2009.00071.x) 21. Dambacher JM, Gaughan DJ, Rochet MJ, Rossignol PA, Trenkel VM. 2009. Qualitative modelling and indicators of exploited ecosystems. Fish. Fish. 10, 305-322. 22. Meliadou A, Santoro F, Nader MR, Dagher MA, Indary SA, Salloum BA. 2012. Prioritising coastal zone management issues through fuzzy cognitive mapping approach. J. Environ. Manag. 97, 56 – 68 23. Abecasis D, Afonso P, Erzini K. 2014. Combining multispecies home range and distribution models aids assessment of MPA effectiveness. MEPS 513: 155–169. doi: 10.3354/meps10987 24. Dunstan PK, Foster SD. 2011. RAD biodiversity: prediction of rank abundance distributions from deep water benthic assemblages. Ecography, 34, 798-806. 25. Munroe DM, Klinck JM, Hofmann EE, Powell EN. 2014. A modelling study of the role of marine protected areas in metapopulation genetic connectivity in Delawere Bay oysters. Aquatic Conservation Marine and Freshwater Ecosystems 24(5), 645-666 26. Neubert MG. 2003. Marine reserves and optimal harvesting. Ecol Lett 6, 843–849 27. Tuck GN, Possingham HP. 2000. Marine protected areas for spatially structured exploited stocks. MEPS. 192, 89-101. 28. Codling EA. 2008. Individual-based movement behaviour in a simple marine reserve-fishery system: why predictive models should be handled with care. Hydrobiologia 606, 55-61 (DOI: 10.1007/s10750-008-9345-9) 29. Takashina N, Mougi A, Iwasa Y. 2012. Paradox of marine protected areas: suppression of fishing may cause species loss. Popul. Ecol. 54, 475-485. 30. Guénette S, Meissa B, Gascuel D. 2014. Assessing the Contribution of Marine Protected Areas to the Trophic Functioning of Ecosystems: A Model for the Banc d’Arguin and the Mauritanian Shelf. PLoS ONE 9(4), e94742. doi:10.1371/journal.pone.0094742 31. Walters C. 2000. Impacts of dispersal, ecological interactions, and fishing effort dynamics on efficacy of marine protected areas: how large should protected areas be? Bull. Mar. Sci 66, 745-757. 32. Brochier T, Ecoutin JM, de Morais LT, Kaplan DM, Lae R. 2013. A multi-agent ecosystem model for studying changes in a tropical estuarine fish assemblage within a marine protected area. Aquat. Living Resour. 26, 147–158. DOI: 10.1051/alr/2012028 33. Armstrong CW. 2008. A note on the ecological–economic modelling of marine reserves in fisheries. Ecol. Model. 62, 242-250. 34. Sethi SA, Hilborn R. 2008. Interactions between poaching and management policy affect marine reserves as conservation tools. Biol. Cons. 141, 506 – 516. Proc. R. Soc. B R. Soc. open sci. article template 10 Modelling MPAs: Insights and Hurdles 35. Dowling NA, Wilcox C, Mangel M, Pascoe S. 2012. Assessing opportunity and relocation costs of marine protected areas using a behavioural model of longline fleet dynamics. Fish. Fish. 13, 139–157. DOI: 10.1111/j.14672979.2011.00422.x 36. Sumaila UR. 2002. Marine protected area performance in a model of the fishery. Nat. Resour. Model 15, 439-451. 37. Punt MJ, Weikard HP, Groeneveld RA, Ierland ECV, Stel JH. 2010. Planning marine protected areas: a multiple use game. Nat. Resour. Model. 23, 610-646 38. Barton NL. 2006. Methods for evaluating the potential effects of MPAs on adjacent fisheries. Thesis: Master of Resource Management, Department of Resource and Environmental Management. Project No. 396. Simon Fraser University. pp 97. 39. Stelzenmüller V, Schulze T, Fock HO, Berkenhagen J. 2011. Integrated modelling tools to support risk-based decisionmaking in marine spatial management. MEPS 441,197-212. DOI: 10.3354/meps09354. 40. Little LR, Punt AE, Mapstone BD, Pantus F, Smith ADM, Davies CR, McDonald A.D. 2007. ELFSim—A model for evaluating management options for spatially structured reef fish populations: An illustration of the “larval subsidy” effect. Ecol. Model. 205, 381-396. 41. Edwards CTT, Rademeyer RA, Butterworth DS, Plagányi ÉE. 2009. Investigating the consequences of Marine Protected Areas for the South African deep-water hake (Merluccius paradoxus) resource. ICES J. Mar. Sci. 66, 72-81. 42. Dichmont CM, Ellis N, Bustamante RH, Deng R, Tickell S, Pascual R, Lozano-Montes H, Griffiths S. 2013. Evaluating marine spatial closures with conflicting fisheries and conservation objectives. J. App. Ecol. 50, 1060-1070. (DOI: 10.1111/1365-2664.12110) 43. Fulton EA, Gorton R. 2014. Adaptive Futures for SE Australian Fisheries & Aquaculture: Climate Adaptation Simulations. CSIRO, Australia. 307 pp. 44. Plagányi É, Punt A, Hillary R, Morello E, Thébaud O, Hutton T, Pillans R, Thorson J, Fulton EA, Smith ADM, Smith F, Bayliss P, Haywood M, Lyne V, Rothlisberg, P. 2014. Multi-species fisheries management and conservation: tactical applications using models of intermediate complexity. Fish and Fisheries 15,1-22. DOI: 10.1111/j.14672979.2012.00488.x 45. Pascoe S, Doshi A, Dell Q, Tonks M, Kenyon R. 2014. Economic value of recreational fishing in Moreton Bay and the potential impact of the marine park rezoning. Tourism Management 41, 53-63 (DOI: 10.1016/j.tourman.2013.08.015). 46. Espinoza-Tenorio A, Montaño-Moctezuma G, Espejel I. 2010. Ecosystem-based analysis of a marine protected area where fisheries and protected species coexist. Environ. Manag. 45, 739-750. DOI: 10.1007/s00267-010-9451-0 47. Smith DC, Fulton EA, Johnson P, Jenkins G, Barrett N, Buxton C. 2011. Developing integrated performance measures for spatial management of marine systems. FRDC Project No: 2004/005 Final Report. CSIRO. 48. Pitcher R, Venables B, Browne M, Doherty P, De’ath, G. 2007. Indicators of protection levels for seabed habitats, species and assemblages on the continental shelf of the Great Barrier Reef World Heritage Area. Unpublished report to the Marine and Tropical Sciences Research Facility. Reef and Rainforest Research Centre Limited, Cairns (75pp.). Available from http://www.cmar.csiro.au/e-print/open/2007/pitchercr_x.pdf 49. Campbell RA, Mapstone BD, Smith ADM. 2001. Evaluating large-scale experimental designs for management of coral trout on the Great Barrier Reef. Ecol. App. 11, 1763-1777. 50. McGilliard CR, Punt AE, Hilborn R 2011. Spatial structure induced by marine reserves shapes population responses to catastrophes in mathematical models. Ecol. App., 21, 1399–1409. 51. Pelletier D, Mahévas S. 2005. Spatially explicit fisheries simulation models for policy evaluation. Fish. Fish. 6, 307–349 52. Gerber LR, Botsford LW, Hastings A, Possingham HP, Gaines SD, Palumbi SR, Andelman S. 2003. Population models for marine reserve design: a retrospective and prospective synthesis. Ecol. App. 13 supplement: S47-S64. 53. Plagányi ÉE. 2007. Models for an ecosystem approach to fisheries. FAO Fisheries Technical Paper. No. 477. Rome, FAO. 2007. 108p. 54. Fulton EA, Link JS. 2014. Modeling Approaches for Marine Ecosystem-based Management. In: M.J. Fogarty and J.J. McCarthy (Eds) Marine Ecosystem-Based Management. The Sea: Volume 16. Harvard University Press 55. Levins R. 1966. The strategy of model building in population biology. Am. Sci. 54, 421-431. 56. Dowd M. 2007. Bayesian statistical data assimilation for ecosystem models using Markov Chain Monte Carlo. JMS 68:439–456 57. Dambacher JM, Li HW, Rossignol PA. 2002. Qualitative predictions in model ecosystems. Ecol. Model. 161: 79-93 58. Guénette S, Lauck T, Clark C. 1998. Marine reserves: from Beverton Holt to the present. Rev. Fish. Biol. Fish. 8: 251272. 59. Sanchirico JN, Wilen JE. 2001. A bioeconomic model of marine reserve creation. J. Environ. Econ. Manag. 42, 257– 276. 60. Parrish R. 1999. Marine reserves for fisheries management: why not. California Cooperative Oceanic Fisheries Investigations Reports 40, 77–86. 61. Edwards C, Plagányi ÉE. 2011. Protecting old fish through spatial management: is there a benefit for sustainable exploitation? J. App. Ecol. 48, 853- 863. (DOI: 10.1111/j.1365-2664.2011.01961.x) 62. Pitcher CR, Doherty P, Arnold P, Hooper J, Gribble N, et al . 2007. Seabed Biodiversity on the Continental Shelf of the Great Barrier Reef World Heritage Area. AIMS/CSIRO/QM/QDPI CRC Reef Research Task Final Report. 320 pp 63. Savina M, Condie SA, Fulton EA. 2013. The role of pre-existing disturbances in the effect of marine reserves on coastal ecosystems: a modelling approach. PLoS ONE 8(4), e61207. (DOI:10.1371/journal.pone.0061207) 64. Raemaekers S, Hauck M, Burgener M, Mackenzie A, Maharaj G, Plagányi ÉE, Britz PJ. 2011. Review of the causes of the rise of the illegal South African abalone fishery and consequent closure of the rights-based fishery. Ocean and Coast.Manage. 54, 433-445 (DOI:10.1016/j.ocecoaman.2011.02.001). 65. Plagányi ÉE, Butterworth DS, Burgener M. 2011. Illegal and Unreported Fishing on abalone - quantifying the extent using a fully integrated assessment model. Fish. Res. 107, 221-232 66. Gray R, Fulton EA, Little LR. 2013. Human-ecosystem interaction in large ensemble-model systems. In A Smajgl and O Barreteau (Eds) Empirical Agent-Based Modeling: Challenges and Solutions, Springer:Berlin. 67. Pelletier D, Claudet J, Ferraris J, Benedetti-Cecchi L, Garcìa-Charton JA. 2008. Models and indicators for assessing conservation and fisheries-related effects of marine protected areas. Can J. Fish. Aquat. Sci. 65, 765-779 68. Brown CJ, Mumby PJ. 2014. Trade-offs between fisheries and the conservation of ecosystem function are defined by management strategy. Front. Ecol. Environ. 2014; 12, 324-329. (DOI:10.1890/130296) Phil. Trans. R. Soc. B. Proc. R. Soc. B article template 11 69. Boncoeur J, Alban F, Guyader O, Thébaud, O. 2002. Fish, fishers, seals and tourists: economic consequences of creating a marine reserve in a multispecies, multi-activity context. Nat. Resour. Model 15, 387–411. 70. Sumaila UR. 1998. Protected marine reserves as fisheries management tools: a bioeconomic analysis. Fish. Res. 37, 287–296. 71. Dambacher JM, Hosack GR, Rochester W. 2012. Ecological indicators for the Exclusive Economic Zone of Australia’s east marine region. CSIRO 118 pp. 72. Melbourne-Thomas J, Constable A, Wotherspoon S, Raymond B .2013. Testing paradigms of ecosystem change under climate warming in Antarctica. PLoS ONE 8(2): e55093. (DOI:10.1371/journal.pone.0055093). 73. Finkel AM. 1990. Confronting uncertainty in risk management: A guide for decision makers. Technical report, Center for Risk Management Resources for the Future, Washington, USA. 74. Meinshausen M, Smith SJ, Calvin K, Daniel JS, Kainuma MLT, Lamarque J-F, Matsumoto K, Montzka SA, Raper SCB, Riahi K, Thomson A, Velders GJM and van Vuuren DPP. 2011. The RCP greenhouse gas concentrations and their extensions from 1765 to 2300. Climatic Change 109, 213-241. 75. Hobday AJ. 2011. Sliding baselines and shuffling species: implications of climate change for marine conservation. Mar. Ecol. 32, 392-403. (DOI: 10.1111/j.1439-0485.2011.00459.x) 76. Hobday AJ, Hartog JR, Timmiss T, Fielding J. 2010. Dynamic spatial zoning to manage southern bluefin tuna (Thunnus maccoyii) capture in a multi-species longline fishery. Fish. Oceanogr. 19, 243-253 (DOI: 10.1111/j.13652419.2010.00540.x) 77. Hobday AJ, Pecl GT. 2014. Identification of global marine hotspots: sentinels for change and vanguards for adaptation action. Rev. Fish Biol. Fish. 24, 415-425. DOI 10.1007/s11160-013-9326-6. 78. Lewison, R. L., A. J. Hobday, S. M. Maxwell, E. L. Hazen, J. R. Hartog, D. C. Dunn, D. K. Briscoe, S. Fossette, C. E. O'Keefe, M. Barnes-Mauthe, M. Abecassis, S. J. Bograd, N. D. Bethoney, H. Bailey, D. Wiley, S. Andrews, L. Hazen and L. B. Crowder (in review). Dynamic Ocean Management: 21st century approaches for marine resource management and conservation. Bioscience. 79. Hobday AJ, Maxwell SM, Forgie J, McDonald J, Darby M, Seto K, Bailey H, Bograd SJ, Briscoe DK, Costa DP, Crowder LB, Dunn DC, Fossette S, Halpin PN, Hartog JR, Hazen EL, Lascelles BG, Lewison RL, Poulos G, Powers A. 2014. Dynamic ocean management: Integrating scientific and technological capacity with law, policy and management. Stanford Enviro. Law J. 33, 125-165. 80. Balmford A, Gravestock P, Hockley N, McClean CJ, Roberts CM. 2004. The worldwide costs of marine protected areas. PNAS 101, 9694-9697. 81. Ban NC, Adams V, Pressey RL, Hicks J. 2011. Promise and problems for estimating management costs of marine protected areas. Cons. Lett. 4, 241-252. 82. Kemp J, Jenkins GP, Smith DC, Fulton EA. 2012. Measuring the performance of spatial management in marine protected areas. Oceanog. Mar. Biol. Ann. Rev. 50, 287-314. 83. Warton DI, Foster SD, De’ath G, Stoklosa J, Dunstan PK. In press. Model-based thinking for community ecology. Plant Ecol (DOI: 10.1007/s11258-014-0366-3). 84. IUCN database summarise at http://www.protectplanetocean.org/ accessed December 2014 85. Edgar GJ, Stuart-Smith RD, Willis TJ, Kininmonth S, Baker SC, Banks S, Barrett NS, Becerro MA, Bernard ATF, Berkhout J, Buxton CD, Campbell SJ, Cooper AT, Davey M, Edgar SC, Försterra G, Galván DE, Irigoyen AJ, Kushner DJ, Moura R, Parnell PE, Shears NT, Soler G, Strain EMA, Thomson RJ. 2014. Global conservation outcomes depend on marine protected areas with five key features. Nature 506, 216-220. (DOI:10.1038/nature13022) 86. Hobday AJ, Hartog JR. 2014. Dynamic Ocean Features for use in Ocean Management. Oceanography 27, 134–145. 87. Fulton EA, Smith ADM, Smith DC, van Putten IE. 2011. Human behaviour: the key source of uncertainty in fisheries management. Fish. Fish. 12, 2-17 (DOI: 10.1111/j.1467-2979.2010.00371.x). 88. Rassweiler A, Costello C, Hilborn R, Siegel DA. 2014 Integrating scientific guidance into marine spatial planning. Proc. R. Soc. B 281: 20132252. (DOI: 10.1098/rspb.2013.2252) 89. Smith ADM, Sainsbury KJ, Stevens RA (1999) Implementing effective fisheries management systems - management strategy evaluation and the Australian partnership approach. ICES J. Mar. Sci. 56, 967-979. 90. Pielke RA. 2007. The honest broker: making sense of science in policy and politics. Cambridge: Cambridge University Press. 91. Marshall CE, Glegg GA, Howell KL. 2014. Species distribution modelling to support marine conservation planning: The next steps. Mar. Pol. 45, 330-332. 92. Garavelli L, Kaplan DM, Colas F, Stotz W, Yannicelli B, Lett C. 2014. Identifying appropriate spatial scales for marine conservation and management using a larval dispersal model: The case of Concholepas concholepas (loco) in Chile. Prog. Oceanogr. 124, 42-53. (DOI: 10.1016/j.pocean.2014.03.011) 93. Hilborn R. 2012. The evolution of quantitative marine fisheries management 1985-2010. Nat. Resour. Model. 25, 122144. 94. Guénette S, Pitcher T.J. 1999. An age-structured model showing the benefits of marine reserves in controlling overexploitation. Fish. Res. 39, 295-303. 95. Walters C, Christensen V, Walters W, Rose K. 2010. Representation of multistanza life histories in Ecospace models for spatial organization of ecosystem trophic interaction patterns. Bull. Mar. Sci. 86, 439-459. 96. Yamazaki S, Grafton QR, Kompas T, Jennings S. 2014. Biomass management targets and the conservation and economic benefits of marine reserves. Fish. Fish. 15, 196-208. (DOI: 10.1111/faf.12008) 97. Ainsworth CH, Morzaria-Luna HN, Kaplan IC, Levin PS, Fulton EA. 2012. Full compliance with harvest regulations yields ecological benefits: Northern Gulf of California case study. J. App. Ecol. 49, 63–72. DOI: 10.1111/j.13652664.2011.02064.x 98. Kaplan IC, Horne PJ, Levin PS. 2012. Screening California Current fishery management scenarios using the Atlantis end-to-end ecosystem model. Prog. Oceanog. 10,: 5–18. DOI:10.1016/j.pocean.2012.03.009 99. Levy JS, Ban NC. 2012. A method for incorporating climate change modelling into marine conservation planning: An Indo-west Pacific example. Mar. Pol. 38, 16-24. (DOI: 10.1016/j.marpol.2012.05.015) Figure captions Proc. R. Soc. B R. Soc. open sci. article template 12 Modelling MPAs: Insights and Hurdles Figure 1: Schematic representation of the broad classes of model used to consider MPAs showing the relative location of a range of model types (reference numbers are provided if from an example used in the text or Tables). Black circles represent conceptual models, blue tactical and purple are strategic. Circles of more than one colour have more than one use. Phil. Trans. R. Soc. B. Phil. Trans. R. Soc. B. article template Phil. Trans. R. Soc. B. doi:10.1098/not yet assigned Tables Table 1: Objectives for modelling studies of MPAs and examples of model types used to address the questions. Objective type Conceptual understanding Conceptual MPA design/planning Tactical MPA design/planning Tactical MPA design/planning MPA assessment (ecological); MPA planning; conceptual understanding Strategic Strategic Conceptual MPA assessment (fisheries or bioeconomic) Strategic Conceptual understanding EBM planning & use of MPAs Conceptual Strategic Objective Synthesise understanding (& communication) Determine species vulnerability or relative protection (e.g. overlap of fishing, habitat and species distribution) Determine effective MPA network design Optimal no-take MPA size Assessment of ecological effects of notake MPAs* (often considering the influence of life history parameters or trophic interactions on outcomes) Assessment of fisheries or bioeconomic effects of no-take MPAs* (often aimed at finding optimal harvesting policy in combination with no-take MPAs, or exploring the implications for effort allocation) Assess influence of no-take MPAs on data streams used to inform other industries (e.g. influence on information content of catch statistics) * Evaluation of role of MPAs in fisheries or conservation management, EBM and integrated coastal zone or ocean management 13 Appropriate model types Qualitative model (e.g. signed diagraphs [21] or fuzzy cognitive maps [22]) Species distribution model (e.g. Maxent [23]) Connectivity models (e.g. biophysical larval dispersal model [24, 25]); geostatistical or GIS models (e.g. [16,18]); spatial optimisation (e.g. Marxan [14]) Population (& harvest) model (e.g. [26]) Population models (e.g. [27]), IBM (e.g. [28]) or multispecies models (e.g. predator-prey [29]); implicitly spatially partitioned food web model (e.g. Ecosim [30]); or explicit ecosystem model (e.g. Ecospace [31]; or OSMOSE [32]) Single or multi-species bioeconomic models (e.g. [33]); effort allocation models (e.g. [34, 35] or Ecospace [31]); game theory based behaviour (e.g. fleet cooperation [36]; interactions between countries, industries and objectives [37]) Single or multi-species bioeconomic model or other model containing effort dynamics [38] Empirically based GIS-Bayesian Belief Network models (e.g. [39]). Process based models including: spatial single or multi-species models (e.g. ELFsim [40]); ecosystem models (e.g. models R. Soc. open sci. article template 14 Modelling MPAs: Insights and Hurdles MPA evaluation (overall) Tactical Evaluate performance of MPAs MPA evaluation (economic) Tactical MPA evaluation; conceptual understanding Strategic (for later tactical use) Economic assessment of performance and effects of no-take MPAs Identify performance measures for assessing MPAs Experimental design Evaluate modelling tools Tactical All Adaptive management experiment design Review model types appropriate for modelling MPAs of intermediate complexity [41]); coupled models (e.g. [42]); end-to-end ecosystem models (e.g. Atlantis [43]) Empirical statistical and GIS models (generating system diagnostics & test effects); spatially resolved multispecies or ecosystem models of intermediate complexity (e.g. [44]) Econometric model (e.g. travel cost model [45]) Loop analysis applied to qualitative signed diagraph (e.g. [46]); quantitative population, multi-species or ecosystem model (e.g. Atlantis [47]); spatially finely resolved statistical models (e.g. [48]) Spatial population, multispecies or ecosystem model (e.g. [49]) All (see Figure 1) * These may be hypothetical abstracted representations (e.g. implicitly representing spatial effects or simplifying space to one cell representing an MPA and one cell representing an area open to harvesting, as in [29]) or a highly detailed representation of a real geographic location (e.g. the Great Barrier Reef resolved to individual reefs and shoals [40]); habitat may be represented implicitly via modifications to carrying capacities [33] or explicitly [43]; and they may be run under historical, current or future environmental conditions and external shocks (e.g. [50]). Phil. Trans. R. Soc. B. Phil. Trans. R. Soc. B. article template Phil. Trans. R. Soc. B. doi:10.1098/not yet assigned Table 2: Example of model-based lessons learnt regarding the performance of MPAs. Extensive review of lessons from literature up until 2005-2008 available in [51]. Lesson Example Reference(s) No-take MPAs service the conservation of key habitats (e.g. canyon heads, [29, 52, 58] shelf reefs, seamounts and key substrates), and potentially slow-growing localised or highly aggregated species (or life history stage), but may not guarantee healthy stocks of mobile predator or prey species – particularly under changing large-scale anthropogenic and environmental pressures. Single species spill over from no-take MPAs has the maximum [27, 59] conservation effect when dispersal rates are moderate and source locations are protected Spill over from no take MPAs provides significant contributions to [36, 51] biodiversity conservation and fished stock status if the MPAs are large (hundreds of square kilometres or more than 30-50% of an ecosystem type), well demarcated and well enforced. No-take MPAs may cause the displacement of fishing effort. This effort can [51,60-62] ultimately depress overall productivity, system state or biodiversity if not removed from the area. Removal of fishing pressure due to the introduction of no-take MPAs in [63] highly perturbed systems has clear, positive, and mostly direct, effects on biomass and functional biodiversity. In light to moderately fished systems, the level of disturbance may provide [63] for a higher coexistence of species and the introduction of no-take MPAs can cause both direct positive effects, but also indirect negative effects through trophic cascades, ultimately leading to a drop in overall functional biodiversity. Human behaviour (such as poaching or fishing the edges of MPAs to [51, 64-66] benefit from any spill over) can undermine the performance of MPAs and as such must be accounted for in MPA models and management plans. No-take MPAs may have a dual influence on the assessment of fish stocks [38, 67] or system state, by either (i) providing reference locations that contrast with exploited areas, (ii) degrading information content or terminating data streams (via a ban on collections). The multi-faceted nature of ecosystems and the multitude of potentially [42-43, 46, 68] conflicting objectives held for them means that spatial management is an important part of ecosystem-based management, but that by themselves no-take MPAs cannot deliver across all objectives*. Integrated management across areas inside and outside of reserves is required. No-take MPAs may confer an economic benefit via improved ecosystem [51, 69-70] service status A suite of indicators for monitoring MPAs is required to characterise [47] overall system state, with simple indicators typically performing with more skill than complex or abstracted indicators Regional scale observations and understanding of system dynamics will be [46-47] necessary to define performance measures for MPAs (as indicator-attribute relationships can change on scales of a few hundred kilometres). * These may be combinations of social, economic and ecological objectives from different stakeholder groups, or even simply conservation objectives across interacting species 15 Phil. Trans. R. Soc. B. article template Phil. Trans. R. Soc. B. doi:10.1098/not yet assigned Figures Figure 1 16