SEAwise Tutorial: Fishing impacts in the North Sea using OSMOSE

Authors: Logan Binch, Jan Jaap Poos, Karen van de Wolfshaar

Institute: Wageningen University and Research, Netherlands

Date: 12.09.2025

1 Tutorial Overview

The following document outlines a short tutorial where we first introduce the North Sea OSMOSE model, then explore spatial fishing scenarios, and finally, evaluate biological and ecosystem level responses to changes in fishing effort distribution following the implementation of Marine Protected Areas (MPAs).

This work is predominantly built upon developments made in SEAwise Work Package 5: Spatial management impacts on ecological systems and fisheries, more specifically Deliverable 5.5: SEAwise report on predicting effect of changes in ‘fishable’ areas on fish and fisheries (Basterdie et al., 2023). The model and case described in this tutorial is detailed further in Binch et al. (2025).

1.1 Tutorial Aims

1. Understand the mechanistic processes underpinning the North Sea OSMOSE model and how they represent species interactions and ecosystem dynamics.

2. Investigate fishing effort displacement scenarios resulting from spatial closures and understand their implications for marine ecosystems.

3. Access, download, and run the North Sea OSMOSE model

4. Assess biological and ecosystem-level indicators produced by the model to evaluate spatial fishing impacts.

2 Introduction to OSMOSE

The Object-oriented Simulator of Marine ecOSystEms Exploitation (OSMOSE) model is a spatially explicit, multi-species, individual-based food web model that simulates how fish populations interact through feeding, growth, reproduction, and mortality processes (Figure 1; Table 1) (Shin & Cury, 2001; Shin et al., 2004; Travers et al., 2009; van de Wolfshaar et al., 2021; Binch et al., 2025).

In the model, fish are represented as super-individuals—groups of individuals with the same species, size, weight, and age (Scheffer et al., 1995). Super-individuals grow by feeding on lower trophic levels (zooplankton and benthic invertebrates) or on other fish, depending on spatial overlap and size adequacy. They die due to predation, starvation, background mortality, or fishing. New super-individuals are introduced during spawning events, representing recruitment.

Figure 1. Representation of the mechanistic processes that compose the OSMOSE food web model.

The North Sea OSMOSE model runs on a grid of 1/9 ICES Statistical Rectangles (~20 × 18.5 km), with a bi-monthly (24-timesteps). Movement is simulated as a random walk within each species’ known habitat, allowing super-individuals to shift between neighbouring cells. Lower trophic levels are not explicitly modelled but are provided through spatio-temporal resource maps generated by ERSEM (European Regional Seas Ecosystem Model; Butenschön et al., 2016). These maps define the availability of zooplankton and benthic invertebrates across space and time, ensuring bottom-up food resources for fish in the model.

In this study, the North Sea OSMOSE configuration includes 14 fish species (Table 2), allowing the model to capture the main trophic interactions and responses to fishing and spatial management measures in the North Sea ecosystem.

Table 1. Mechanistic processes represented in the North Sea OSMOSE model and their corresponding mathematical formulations. Superindividuals are denoted by i, prey items by j, and model timesteps by t. All species specific parameter values can be seen in Supplementary Materials (Table S1). For further details see Travers et al. (2009); van de Wolfshaar et al. (2021); and Binch et al., (2025).

2.1 Metier Fishing

Fishing mortality in the model is implemented at the metier level, using effort and landings data from the STECF Fisheries Dependent Information (FDI) database (STECF et al., 2024; Binch et al., 2025). A total of 14 metiers are represented, each defined by gear type, mesh size, and target assemblage, with target species identified when they contribute at least 5% of recorded landings (Table 2).

Catch selectivity is incorporated through species- and mesh-size-specific selectivity ogives, ensuring that fishing mortality reflects the biological characteristics of each super-individual. Fishing effort is applied at a fine spatial resolution (1/9th ICES Statistical Rectangle) and bi-monthly timesteps, with input data disaggregated from ISR- and quarter-level records.

When multiple metiers target the same species in the same area and time, their order of operation is randomised to avoid bias. This framework captures the spatial heterogeneity, gear-specific selectivity, and seasonal dynamics of North Sea fishing activity.

Table 2. Metiers represented in the model with average annual number of fishing days between 2016 – 2020, based on information taken from the FDI STECF database (STECF, 2024). The gear class describes the category of the gear type for a respective metier assigned either Bottom Trawl (BT) or Pelagic (PEL). Effort change applies to the percentage of displaced effort under the three effort scenarios. Target species are selected based on ≥5% of the total species specific landings, in the model area, being represented by the metier. Target species: COD (cod, Gadus morhua); DAB (dab, Limanda limanda); DGS (spurdog, Squalus acanthias); GUG (grey gurnard, Eutrigla gurnardus); HAD (haddock, Melanogrammus aeglefinus); HER (herring, Clupea harengus); NOP (Norway pout, Trisopterus esmarkii); PLE (plaice, Pleuronectes platessa); POK (saithe, Pollachius virens); RJC (thornback ray, Raja clavata); SAN (sandeel, Ammodytes spp.); SOL (sole, Solea solea); SPR (sprat, Sprattus sprattus); WHG (whiting, Merlangius merlangus). Table taken from Binch et al., (2025).

2.2 OSMOSE Parameterisation

The main components of the OSMOSE model are defined in the model_conf.xml file. This is the configuration file that defines things such as the latitudinal and longitudinal range of the model, the number of years per model run, and the number of timesteps in a year. Additionally, the configuration file also contains the link to various parameter files specific to species and metiers (Figure 2). These files contain specific information and parameters that can be adapted. For example in the species files parameters associated with the mechanistic processes above can be changed. In the example below, metier parameters such as temporal effort allocation, spatial effort allocation, target species, and mesh size can all be adjusted to specific use cases.

Figure 2. Representation of the model parameterisation input file for a single metier used in the North Sea OSMOSE model.

3 Scenario Exploration

In this tutorial we will explore the effect of Marine Protected Area (MPA) implementation on the fishing effort distribution of bottom trawling metiers in the North Sea. We will then run the OSMOSE food-web model to explore the effects of changes in the spatial fishing effort allocation on different species groups and the ecosystem as a whole.

Here we use three simple assumptions for what happens to fishing effort from within newly designated MPAs (Figure 3). First we redistribute the fishing effort to the nearest available cell on the ‘boundary’ of MPAs using the following equations:

\[r^* = \min \{ r \mid N_{r,j,k} > 0 \}\]

\[E_{(x,y)}' = E_{(x,y)} + \frac{E_{(j,k)}}{|N_{r^*,j,k}|}, \quad \forall (x,y) \in N_{r^*,j,k}\]

this can be seen as fishing the line. Secondly, a proportional effort redistribution scenario sees the effort from MPAs displaced to all other areas where the metier is active with effort allocation depending on the intensity of current effort levels, represented by:

\[E_{(x,y)}' = E_{(x,y)} + \frac{E_{(x,y)}}{\sum_{(x,y) \in S} E_{(x,y)}} \cdot \sum_{(j,k) \in \text{MPA}} E_{(j,k)}, \quad \forall (x,y) \in S\]

Finally, an effort reduction scenario assumes that fishing activity from within MPAs is not redistributed and is instead removed completely from the system, such that:

\[E_{(j,k)} = 0, \quad \text{if } \text{MPA}_{(j,k)} = 1\]

Figure 3. Illustration of the spatial effort distribution of the OTB_DEF_100-119 metier under Baseline conditions and the three effort scenarios: Boundary, Proportional, and Reduction. Baseline effort is derived by taking the average number of annual fishing days for 2016 - 2020 from the FDI STECF database (STECF et al., 2024). Effort values are represented by a log10 scale. Gridlines represent the OSMOSE model spatial resolution, 1/9th ICES Statistical Rectangle (ISR). Black lines represent the borders of MPAs designated for the restriction of demersal trawling activity. Figure taken from Binch et al., (2025).

4 Running OSMOSE

Unlike many other ecosystem modelling tools, OSMOSE does not have a graphical user interface (GUI). Instead, it is run through the command line in Linux using a Java executable file. While this may seem less intuitive at first, the process is fairly straightforward if you follow the steps below.

Step 1 – Download the model files

• Access the version of the OSMOSE model used in this tutorial via Zenodo: https://doi.org/10.5281/zenodo.16781983

• This download contains the Java executable I-Osmose.jar file along with all required parameter files.

Step 2 – Prepare your working directory

• Place the I-Osmose.jar file and the parameter folder (e.g. paramNS/) in a directory on your computer

• Note the full path to this directory (you will need it when running the model).

Step 3 – Open the Linux command line

• Navigate to the folder containing the OSMOSE executable and parameter files.

• Alternatively, you can call the full path from anywhere in the command line.

Step 4 – Run the model

Execute the model using the following command to start running the OSMOSE model:

java -DCONFIG_DIR=/PATH_TO_DIRECTORY/paramNS/ -jar I-Osmose.jar

5 Outputs and Analysis

All files required for processing outputs can be found using the following link: https://github.com/LoganBinch/NS_OSMOSE_SEAwise.git

The output data from the OSMOSE model run comes in the form of .bin files. First they must be converted to to .db file format using the create_db.py python script.

In the following example we only look at outputs related to the school.bin file. This file represents data at each timestep in the final year of our model run, containing data on the species, length, weight, age, latitude, and longitude, of each super-individual. We look here specifically at a total of 6 indicators, three representing the biological state at species group level and three representing the whole ecosystem. However, it should be noted that the outputs given by OSMOSE support the calculation of numerous different indicators.

Table 3. Classification of species groups based on species specific ecotype and feeding guild characteristics, derived from the OSPAR indicator database (OSPAR, 2017). The Protected, Endangered, and Threatened (PET) species group is considered separately due to their unique vulnerability to fishing pressures. For species codes, see Table 2. Table taken from Binch et al., (2025).

It should be noted that in all outputs we calculate and represent in this tutorial are percentage changes relative to baseline conditions, unless otherwise stated.

5.1 Biological Species Indicators

The biological indicators were selected from Lynam et al. (2023), and calculated for each species group. The first indicator we look at is biomass (BIO) calculated using:

\[\text{BIO} = \sum_{s=1}^{S} \sum_{i=1}^{n} B_{i,s}\]

This can be calculated for each grid cell enabling us to obtain a spatial representation of relative biomass distribution (Figure 4).

Figure 4. Relative spatial biomass distribution aggregated for all species in the final year of OSMOSE model run, representing the three effort scenarios: Boundary, Proportional, and Reduction, relative to baseline. Black lines represent the borders of MPAs where demersal trawling is prohibited under the effort scenarios. Figure taken from Binch et al., (2025).

Next we look at the proportion of mature biomass (PropM), this represents the cumulative biomass of all mature individuals as a proportion of the total biomass such that:

\[\text{PropM} = \frac{ \sum_{s=1}^{S} \sum_{i=1}^{n} B_{i,s} \cdot I(L_{i,s} \ge L_s^{\text{mat}})}{ \sum_{s=1}^{S} \sum_{i=1}^{n} B_{i,s}}\]

Finally, we calculate the typical length (TyL) which represents the geometric mean length of species groups (ICES, 2014; OSPAR, 2017) calculated by:

\[\text{TyL} = \exp \left( \frac{\sum_{s=1}^{S} \sum_{i=1}^{n} B_{i,s} \cdot \log(L_{i,s})}{\sum_{s=1}^{S} \sum_{i=1}^{n} B_{i,s}} \right)\]

Figure 5. Biological indicators calculated using the final year of OSMOSE model output, comparing the three effort scenarios: Boundary, Proportional, and Reduction relative to baseline conditions, for species groups described in table 3. Total biomass (BIO – orange), proportion mature biomass (PropM – yellow), and typical length (TyL – blue), calculated both inside (striped bars) and outside (solid) of MPA designations. Figure taken from Binch et al., (2025).

5.2 Ecosystem Indicators

The first ecosystem indicator we look at is the Large Fish Indicator (LFI40) that is calculated using the following equation:

\[\text{LFI}_{40} =\frac{ \sum_{i \in D} B_i \cdot I(L_i \ge 40)}{ \sum_{i \in D} B_i } \]

This represents the proportion of biomass from demersal ecotype individuals, greater than or equal to 40cm in length, relative to the total demersal biomass. Next, we calculate the Mean Mature Trophic Level (MMTL) such that

\[ \text{MMTL} = \frac{ \sum_{s=1}^{S} \sum_{i=1}^{n} TL_{i,s} \cdot B_{i,s} \cdot I(L_{i,s} \ge L_s^{\text{mat}}) }{ \sum_{s=1}^{S} \sum_{i=1}^{n} B_{i,s} \cdot I(L_{i,s} \ge L_s^{\text{mat}}) } \]

This represents the the mean trophic level of mature individuals. Finally we calculate the Size-Spectra Slope (SSS), an indicator used to represent the relationship between the individual weight and biomass biomass. A relatively steeper (more negative) value represents a relative increase in large individuals compared to small ones.

\[ \text{SSS} = \frac{ \sum_{i=1}^{n} \left( \log(W_i) - \overline{\log(W)} \right) \cdot \left( \log(A_i) - \overline{\log(A)} \right) }{ \sum_{i=1}^{n} \left( \log(W_i) - \overline{\log(W)} \right)^2 } \]

Figure 6. Ecological indicators calculated using the final year of OSMOSE model output, comparing the three effort scenarios: Boundary, Proportional, and Reduction relative to baseline conditions. Large Fish Index >40cm (LFI40 – orange), Mean Mature Trophic Level (MMTL – yellow), and Size-Spectra Slope (SSS – blue). Figure taken from Binch et al., (2025).

Accompanying worksheets and guides can be found on the SEAwise website under the online course page.

6 References

Bastardie, F., Spedicato, M., Bitetto, I., Romagnoni, G., Zupa, W., Letschert, J., Püts, M., Damalas, D., Kavadas, S., Maina, I., Tsagarakis, K., Poos, J.J., Papantoniou, G., Depestele, J., Batts, L., Bluemel, J., Astarloa Diaz, A., van de Wolfshaar, K., Binch, L., Rindorf, A., 2023. Predicting effect of changes in ’fishable’areas on fish and fisheries. https://doi.org/10.11583/DTU.24331198

Binch, L., Poos, J.J., van de Wolfshaar, K., 2025. Fishing effort displacement drives ecosystem impacts within and beyond marine protected areas. Ecol Modell 510, 111336. https://doi.org/https://doi.org/10.1016/j.ecolmodel.2025.111336

Butenschön, M., Clark, J., Aldridge, J.N., Icarus Allen, J., Artioli, Y., Blackford, J., Bruggeman, J., Cazenave, P., Ciavatta, S., Kay, S., Lessin, G., van Leeuwen, S., van der Molen, J., de Mora, L., Polimene, L., Sailley, S., Stephens, N., Torres, R., 2016. ERSEM 15.06: a generic model for marine biogeochemistry and the ecosystem dynamics of the lower trophic levels. Geosci Model Dev 9, 1293–1339. https://doi.org/10.5194/gmd-9-1293-2016

ICES, 2014. Report of the Working Group on the Ecosystem Effects of Fishing Activities (WGECO). https://doi.org/10.17895/ices.pub.5676

Lynam, C., Thorpe, R., van de Wolfshaar, K.E., Puts, M., Corrales, X., Romagnoni, G., Tsagarakis, K., Potier, M., Papantoniou, G., Halouani, G., others, 2023. SEAwise Report on the effects of fishing on food webs and community diversity aimed at populating the MSFD Descriptor 4 and based on food web and end-to-end modelling.

OSPAR, 2017. Intermediate Assessment 2017 Available at: https://oap.ospar.org/en/ospar-assessments/intermediate-assessment-2017

Scheffer, M., Baveco, J., DeAngelis, D., … K.R., 1995. Super-individuals a simple solution for modelling large populations on an individual basis. Ecol Modell 80, 161–170. https://doi.org/10.1016/0304-3800(94)00055-M

Shin, Y., Shannon, L.J., Cury, P.M., 2004. African Journal of Marine Science Simulations of fishing effects on the southern Benguela fish community using an individual-based model: learning from a comparison with ECOSIM. Afr J Mar Sci 26, 95–114. https://doi.org/10.2989/18142320409504052

Shin, Y.J., Cury, P., 2001b. Exploring fish community dynamics through size-dependent trophic interactions using a spatialized individual-based model. Aquat Living Resour 14, 65–80. https://doi.org/10.1016/S0990-7440(01)01106-8

STECF, European Commission Joint Research Centre, Scientific Technical Economic Committee for Fisheries, Motova-Surmava, A., Zanzi, A., Hekim, Z., 2024. Fisheries Dependent Information FDI (STECF 23-10). Publications Office of the European Union. https://doi.org/doi/10.2760/676073

Travers, M., Shin, Y.-J., Jennings, S., Machu, E., Huggett, J.A., Field, J.G., Cury, P.M., 2009. Two-way coupling versus one-way forcing of plankton and fish models to predict ecosystem changes in the Benguela. Ecol Modell 220, 3089–3099. https://doi.org/10.1016/j.ecolmodel.2009.08.016

van de Wolfshaar, K.E., Daewel, U., Hjøllo, S.S., Troost, T.A., Kreus, M., Pätsch, J., Ji, R., Maar, M., 2021. Sensitivity of the fish community to different prey fields and importance of spatial-seasonal patterns. Mar Ecol Prog Ser 680, 79–95. https://doi.org/10.3354/MEPS13885