Unix implementation

Fortran, R and MatLab versions

Platform-independent version coming

MSY

EwE modelling framework

What can EwE be used for?

Terrestrial systems?

Tutorial: Build a conceptual model

Tutorial: Fleet dynamics in Anchovy Bay

MPA Challenge

Tutorial: Wind farm

Tutorial: Price elasticity

Tutorial: Lionfish in Anchovy Bay

See folder “invasives” has model from lionfish paper. Needs to be implemented in Anchovy Bay model.

Multiple stable states & saddle points

See spreadsheet: F may comparison time series bias (sort of)

The MSP Challenge simulation platform has been developed to help decision makers, stakeholders, and students understand and manage the maritime (blue) economy and marine environment. It uses advanced game technology and aspects of gameplay to engage and facilitate planners and stakeholders in their dialogue and support their learning. Multiple participants in the MSP Challenge explore the marine spatial planning process by designing spatial plans to obtain specific objectives whilst negotiating and resolving conflicts. In a virtual environment, players experiment with, and learn about, the interplay between conflicting planning activities, society, and ecology (Figure 1). The games capture existing activities and ecology in an area of interest such as the North Sea and the Baltic Sea, for which games have already been developed .

To enhance the ability to represent ecological effects of spatial plans, the computational engine of the EwE software has been integrated into the MSP simulation platform[footnote]Steenbeek, J., Romagnoni, G., Bentley, J., Heymans, J., Serpetti, N., Gonçalves, M., Santos, C., Warmelink, H., Mayer, I., Keijser, X., Fairgrieve, R., Abspoel, L., 2020. Combining ecosystem modeling with serious gaming in support of transboundary maritime spatial planning. Ecology and Society 25. https://doi.org/10.5751/ES-11580-250221[/footnote]. Part of the development of an MSP Challenge game for a particular area involves the creation of a corresponding spatiotemporal Ecospace model that represents the food web with focus on local species of interest. The MSP Challenge and Ecospace are connected by the impacts that planning activities have on the ecosystem. These impacts take the form of environmental pressures (noise, surface disturbance, and bottom disturbance), changes in availability of artificial substrate, and changes in de-facto fisheries restrictions. At every simulation time step, the distribution and intensity of MSP environmental pressures, artificial substrate, and fisheries restrictions are converted to Ecospace maps of environmental drivers, habitat layers and MPAs, respectively. Via the habitat foraging capacity model and the fishing effort gravity model, the food web responds to MSP Challenge activities and then sends aggregated model output back to the MSP Challenge for consideration by players (Figure 2)[footnote]Steenbeek et al., 2020, op. cit.[/footnote]

Figure 2. An example of how shipping in the MSP Challenge platform generates disturbances that displace cetaceans

The integration of a complex, mechanistic and scientifically vetted ecosystem model into an advanced interactive planning support system is a novel and necessary breakthrough to enable the implicit use of scientific models by audiences that need the insights but lack the scientific skills to operate the models. We see this integration as a blueprint for making ecosystem sciences more accessible to stakeholders[footnote]Heymans, J.J., Bundy, A., Christensen, V., Coll, M., de Mutsert, K., Fulton, E.A., Piroddi, C., Shin, Y.-J., Steenbeek, J., Travers-Trolet, M., 2020. The Ocean Decade: A True Ecosystem Modeling Challenge. Front. Mar. Sci. 7, 554573. https://doi.org/10.3389/fmars.2020.554573[/footnote], and as a necessary step towards galvanizing ocean sciences for the future[footnote]Ryabinin, V., Barbière, J., Haugan, P., Kullenberg, G., Smith, N., McLean, C., Troisi, A., Fischer, A., Aricò, S., Aarup, T., Pissierssens, P., Visbeck, M., Enevoldsen, H.O., Rigaud, J., 2019. The UN Decade of Ocean Science for Sustainable Development. Frontiers in Marine Science 6. https://doi.org/10.3389/fmars.2019.00470[/footnote], and hope that case studies such as these will inspire the scientific community to make their science more widely accessible[footnote]Depellegrin, D., Hansen, H.S., Schrøder, L., Bergström, L., Romagnoni, G., Steenbeek, J., Gonçalves, M., Carneiro, G., Hammar, L., Pålsson, J., Crona, J.S., Hume, D., Kotta, J., Fetissov, M., Miloš, A., Kaitaranta, J., Menegon, S., 2021. Current status, advancements and development needs of geospatial decision support tools for marine spatial planning in European seas. Ocean & Coastal Management 209, 105644. https://doi.org/10.1016/j.ocecoaman.2021.105644[/footnote].

It has been traditional in fisheries science that modellers should design, program, and implement their own programs. This is, as a rule, a good practice and the model construction process is indeed both valuable and informative. There are, however, cases, notably related to data access, reporting and infrastructure overheads that make using an existing approach and software package a wise choice. The increasing flexibility of a number of the existing models means that they have become a framework for model creation and use rather than a monolithic model. This means that the user can benefit from the package’s overhead handling while not being simultaneously locked into rigid assumptions. Careful design and application, raises the stakes for and capabilities of the modelling programs, while making ecosystem modelling more accessible for a wider range of scientists. In the current era where EAF is being introduced and developed, this is indeed a facilitating factor.

That being said, care must be taken when using pre-existing packages. It is necessary with these to carefully examine assumptions and requirements, and to investigate how different parameterizations and implementations impact model findings. There is never one model formulation that is 'correct'; alternatives must be examined. Importantly, models should not be used as simple black-box formulations. Ecosystem models are tools, and as such are valuable only if used with thought.

• Used goodness of fit measures to compare spatial distributions of model outputs with field observations. 
• Model fit evaluations were only for habitat-forming groups
  • Dynamic small area, 54.6 km², ⅓ is intertidal, 4.5 m tides, major river estuary
• First EwE model where goodness-of-fit measures has been applied? 

• Derives a test statistic based on the proportion of locations at which predicted occurrences match observed occurrences, and similarly with predicted and observed absences. 
• PCC is estimated from # cells where (1) a functional group is present in field data and in model (PC for present correct), and (2) a functional group is absent in the field data and in model (AC for absent correct). 
  • PCC = (PC + AC) / n 
• Problem: “if the test data consist of 95% absences and 5% presences, and this reflects the true prevalence of the species on the landscape a 'null' model that predicted absence in all cases would be correct 95% of the time”

• Similar to PCC but accounts for agreement occurring by chance. It measures the difference between the model's PCC and the expected PCC from a random model.
• Kappa = (PCC - Expected PCC) / (1 - Expected PCC)

• Percent Correct Presence (Sensitivity) 
  • Predicts a species to occur in an area where it is known to occur
  • Sensitivity = TP / (TP + FN)
• Percent Correct Absence (Specificity)  
  • Predicts a species does not occur in an area where it indeed isn’t found
  • Specificity = TN / (TN + FP)