MPA effectiveness
Evaluating Marine Protected Area (MPA) effectiveness: What are the potential ecosystem-wide effects of implementing different MPA designs (size, location, restrictions) on biomass, biodiversity, and fisheries?
Learning Objectives
By the end of the session participants will be able to:
- Build and run a spatial-temporal Ecospace scenario and extract region-specific biomass and catch indicators.
- Analyze how alternative MPA configurations (size, number, placement, access rules) interact with species dispersal to shape inside-vs-outside outcomes (“spill-over”).
- Diagnose model behaviour by linking parameter choices (especially dispersal, habitat preference and effort redistribution) to ecological mechanisms and observed results.
- Compare scenarios quantitatively and articulate trade-offs among conservation, fisheries yield and bycatch reduction in a form useful to managers.
- Critically assess model uncertainty and identify additional data or analyses (e.g., field tagging, sensitivity tests, spatial optimization) needed before applying results to real-world policy.
This overarching question can lead to an in-depth exploration of various aspects of MPA design and effectiveness using EwE. It addresses current challenges in MPA implementation and management, and the results may provide insights for policymakers and marine resource managers.
The core question for evaluating efficiency of MPAs is: What’s the spill-over effect? That is, does the production in the protected area increase and does it spill-over to outside? In Ecospace, the key parameter that decides the spill-over rate is the dispersal rate (km year-1) – and dispersal rates are difficult to quantify. For spatially-explicit studies it is therefore pertinent to evaluate what impact uncertainty about dispersal rates have on the outcome.
Model choice
We can, once again, turn to our favourite place, Anchovy Bay. For MPA evaluations, we will need a spatially-explicit version of the model. Given that below there are policy questions related to bycatch, we can use a modified version of the model with two bycatch species added. The added species are sneaker shark (Carcharadon endangerous) and goldfish (Solea euro) both caught as bycatch in the trawl fishery. Both are overexploited in the model, have different distributions within Anchovy Bay, and the goldfish have fisheries interest as they are a valuable bycatch. Part of the objective for the MPA closures is to evaluate alternative ways of limiting the bycatch of these two species.
In preparation for the questions below, download the modified version of the Anchovy Bay model from this link. Unzip, and open your version of EwE. Load the file (File > Open model, or click the blue No-fish-is-an-island in the icon row below the top menu), go Ecospace > Load scenario at the top menu (or click the green globe icon) and load the BayOfAnchovies Ecospace scenario. You can now go straight in and run Ecospace, (Ecospace > Output > Run Ecospace, on the left-side Navigator, click Run). The Ecospace scenario is quite coarse (20 x 20 spatial cells), but will be fine for demonstrating how to work with Ecospace to address MPA questions.
Adding more spatial resolution (more cells) will make for prettier maps, slow down runs, and not necessarily produce different results. Best advice is to use coarse resolution for the model development, and only change to finer-scale maps for the final production runs.
But before we move ahead with the policy/research questions below, a bit of background. A crucial question for evaluating MPA efficiency is how much spillover there is? Spillover is indeed one of the best arguments for MPAs, in principle, they let biomass build up and this may lead to higher reproduction, which in turn will lead to spillover to neighbouring areas. The big question is how much? When we model this in Ecospace, the crucial parameter that sets how much organisms moves around randomly (that’s what leads to spillover, at least for non-territorial species) is the dispersal rate (location: Ecospace > Input > Dispersal, unit: km/year). We need estimates for the dispersal rates, if they are set too high, MPAs will have no effect, and if too low, biomass will build up in MPAs till they reach their local carrying capacity. Neither is likely to be common features in ecosystems.
The default value for dispersal rates in Ecospace is 300 km/year. The cell size in the Anchovy Bay model is 5.65 by 5.65 km, so the default implies roughly that on the average, an organism will leave a cell every week. Keep that in mind when working on the questions below.
Potential policy questions
- Spillover effects: How do different MPA sizes and designs influence the spillover of biomass to adjacent areas, and what is the optimal MPA configuration to maximize both conservation and fishery benefits?
- No-take vs. multi-use MPAs: What are the comparative ecosystem-wide effects of strictly no-take MPAs versus multi-use MPAs that allow some regulated fishing activities?
- MPA network design: How does the spatial arrangement of multiple smaller MPAs compare to fewer larger MPAs in terms of biodiversity protection, ecosystem resilience, and fisheries sustainability?
- Habitat representation: How does including a diversity of habitat types within MPA boundaries affect overall ecosystem function and biodiversity compared to more homogeneous protected areas?
- Climate change resilience: How can MPAs be designed to enhance ecosystem resilience to climate change impacts, such as species range shifts and changing productivity patterns?
- Trophic cascades in MPAs: How do different MPA designs influence the recovery of apex predators and the subsequent trophic cascade effects on lower trophic levels?
- MPA size thresholds: Is there a minimum size threshold for MPAs to effectively protect ecosystem processes and maintain viable populations of key species?
- Temporal dynamics of MPA effects: How do the ecosystem-wide effects of MPAs change over time, and what are the short-term versus long-term trade-offs in terms of conservation and fisheries outcomes?
- Enforcement and compliance: How do varying levels of MPA enforcement and stakeholder compliance affect the ecological outcomes and overall effectiveness of protected areas?
- Connectivity and larval dispersal: How does the placement of MPAs affect larval connectivity between protected and unprotected areas, and what are the long-term implications for population sustainability and genetic diversity?
- Socio-economic impacts: How do different MPA designs affect local fishing communities in terms of catch displacement, fishing effort redistribution, and overall economic outcomes?
- Mobile MPAs: How effective are dynamic mosaic MPAs that shift location based on seasonal patterns or changing environmental conditions compared to static MPAs?
- Bycatch limitation: Can dynamic mosaic spatial restrictions or closures be used effectively to limit bycatch?
Spillover effects: How do different MPA sizes and designs influence the spillover of biomass to adjacent areas, and what is the optimal MPA configuration to maximize both conservation and fishery benefits?
Open the Anchovy Bay Spatial Mosaic database (see above), and the BayOfAnchovies Ecospace scenario. Load the anchovybay time series (Ecosim > Input > Time series > Load), which is stored in the model database.
By loading the Ecosim time series, we set the model up to overexploit the species that are caught in the trawl fishery. By next introducing MPAs, we run a retrospective analysis, in essence asking: what if they had introduced MPAs in Anchovy Bay early ony? Would the time trajectory be different? Are there MPAs designs that would have helped? (and hence, may help in the future).
We often run this kind of retrospective analysis. The advantage is that we know a lot about the ecosystem history – compared to the ecosystem future – and we can use the retrospective to test management procedures.
Before running the model, let’s consider how to get hold of results. We’d like to know what happens within versus outside the MPA. We can use Ecospace’s Region facility for that. On the Ecospace input map page, click Regions on the right-hand side. You should now see a map of the bay with MPA 1 indicated. We need to define regions next, click the pen to the right of Regions, and on the pop-up screen, change the Number of regions to 1, click the radio button From MPAs, and OK. You should now have a region corresponding to your MPA. When you later extract results, you will have the option of extracting biomasses and catches within the defined region as well as for the Undefined area, i.e. the area not in any region.
We’re ready for runs now. First a baseline without any MPAs. Run the Ecospace model (Ecospace > Output > Run Ecospace > Run). The plot shows how biomasses change over time. You’ll see deviations from unity (i.e. the biomasses change from the Ecopath baseline), notably showing how the increase in trawler effort crashes the species caught by the gear, but also how seals increase considerably as they rebound after the stop to culling.
Note that even if an Ecosim run generates a flat-line (when there are no drivers changing the conditions), Ecospace will never to that when there are defined preferences for habitats or environmental conditions defined for a model. This is because of species distributions across the maps. If a predator has insufficient overlap with its prey, the predator will decline and the prey increase (and vice versa). We can ignore that for now, but in a model use for management, we would improve on the distributions before accepting a model as OK – i.e., good enough to be used.
If you want to see biomass distributions during a run, click on the Map above the time plot. You’ll see nice colourful maps, rather fascinating, actually. But be aware that when you display the maps during runs, it increases the runtime quite considerably.
After the run, go to Ecospace > Output > Ecospace results > Group caught by where you can see biomass, catch and value for the first and last year of the simulation (actually for any year, just change the years on the form). Copy the table to a spreadsheet.
Next, create an MPA. For this, go to Ecospace > Input > Maps where there’s a MPAs layer accessible on the right-hand side. Know that you can have any number of MPAs (add more by clicking the pen to the right of MPAs) , they can be open or closed for fishing in all or part of the year (Ecospace > Input > Ecospace fishery > Marine Protected Areas), and MPAs can be enforced or not for individual fleets (Ecospace > Input > Ecospace fishery > MPA enforcement). But back to Ecospace > Input > Maps > MPAs, where with your mouse you can sketch a closed area of perhaps 10 cells by 10 cells for MPA 1 somewhere in the centre of the bay.
Now run the model again. As before, go to Ecospace > Output > Ecospace results > Group caught by and extract results as before. Further, click Region, where you can extract the similar information broken down for the Undefined area and for Region 1, i.e. the MPA1 area. Compare the result without and with MPA. Any differences?
Remember the all-important displacement rates. So, next, let’s go to Ecospace > Input > Dispersal, click Base dispersal rate in the top row, and enter 30 in the box to the left of Apply in the top-right corner of the form. Then click Apply. Run the model again with the lower dispersal rate (30 km/year). Extract the results. Are they different?
Try once more, this time setting the dispersal rate to 3 km/year for all groups. Different?
What do you conclude from this exercise?
How do we get estimates of dispersal rates? We can guess, or there may be estimates from field studies – asking for dispersal rates may actually lead to field studies. There’s also a possibility to use an individual based model (IBM) to estimate it from swimming behaviour, see the model description in the EwE User Guide, and here’s a link to an online app implementing the dispersal rate model.
No-take vs. multi-use MPAs: What are the comparative ecosystem-wide effects of strictly no-take MPAs versus multi-use MPAs that allow some regulated fishing activities?
Our MPA1 is by default a no-take MPA implemented for all species, though the bigger problem in Anchovy Bay really is the trawl fishery. What if the MPA1 was only for that fishery? Go to Ecospace > Input > Ecospace fishery > MPA enforcement, and un-check the All MPAs enforced for all groups but the trawlers. Run the model again and see if the results are different now from what you go above.
MPA network design: How does the spatial arrangement of multiple smaller MPAs compare to fewer larger MPAs in terms of biodiversity protection, ecosystem resilience, and fisheries sustainability?
This question relates directly to the classic Single Large Or Several Small? (SLOSS) debate.[1] What’s best, a large MPA or many small? Again, dispersal rates are an important factor to consider, but reflect on whether the MPA sites are to protected specific habitat types with local occurrence, or if the MPAs are to rebuild populations of more widely distributed critters. Then go for modelling.
We can use the Anchovy Bay model again. For the first question above, we made a big MPA with 10×10 cells. Try to distribute the same number of cells across the maps into smaller MPAs. Maybe check the species distributions you can see during (and after) model runs to consider where it would be good to have MPAs. Run and compare results to what you got from the first question.
Habitat representation: How does including a diversity of habitat types within MPA boundaries affect overall ecosystem function and biodiversity compared to more homogeneous protected areas?
There’s a version of the Anchovy Bay model that can be used to explore that! No big surprise. The tutorial on spatial optimization has a version with a number of species and habitats that are of conservation concern, including, but which are not included (and indeed don’t need to be) in the Anchovy Bay model, so how do we go about modelling their protection with MPAs? The spatial optimization module of EwE is designed with that in mind. The first task is to obtain distribution maps for the groups of concern, and read those into Ecospace. Subsequently, we can define an objective function based on economic, social and ecological factors, and search for a protection scheme that will optimize conservation concern at the least possible cost.
We suggest you work your way through the tutorial and read the descriptive chapter as a step toward making actual evaluations of the present question.
Climate change resilience: How can MPAs be designed to enhance ecosystem resilience to climate change impacts, such as species range shifts and changing productivity patterns?
Trophic cascades in MPAs: How do different MPA designs influence the recovery of apex predators and the subsequent trophic cascade effects on lower trophic levels?
MPA size thresholds: Is there a minimum size threshold for MPAs to effectively protect ecosystem processes and maintain viable populations of key species?
Temporal dynamics of MPA effects: How do the ecosystem-wide effects of MPAs change over time, and what are the short-term versus long-term trade-offs in terms of conservation and fisheries outcomes?
Enforcement and compliance: How do varying levels of MPA enforcement and stakeholder compliance affect the ecological outcomes and overall effectiveness of protected areas?
Connectivity and larval dispersal: How does the placement of MPAs affect larval connectivity between protected and unprotected areas, and what are the long-term implications for population sustainability and genetic diversity?
Socio-economic impacts: How do different MPA designs affect local fishing communities in terms of catch displacement, fishing effort redistribution, and overall economic outcomes?
Mobile MPAs: How effective are dynamic mosaic MPAs that shift location based on seasonal patterns or changing environmental conditions compared to static MPAs?
Bycatch limitations: Can dynamic mosaic spatial restrictions or closures be used effectively to limit bycatch?
- Salomon, A.K., Waller, N.P., McIlhagga, C. Yung RL, Walters C. Modeling the trophic effects of marine protected area zoning policies: A case study. Aquatic Ecology 36, 85–95 (2002). https://doi.org/10.1023/A:1013346622536. ↵
