- Objectives
- Build and evaluate an agent-based, adaptive fishing fleet model as an analytic tool to support management
- Background
- POSEIDON is a coupled human-ecological model that combines an agent-based, adaptive fishing fleet model with existing fishery models or simple biological data, to simulate vessel behavior and fishery outcomes based on policies, market influences, and environmental factors.
- POSEIDON provides a powerful platform for policy evaluation and decision support, with a strong focus on the spatial and human dimensions of fisheries management.
- POSEIDON was originally developed by a multidisciplinary team from the University of Oxford, Ocean Conservancy, George Mason University, the University of California, Santa Barbara, and Arizona State University, as part of an effort to advance innovation in fisheries management.
- The model has been calibrated and validated to the U.S. West Coast groundfish fishery. It is now being adapted to explore MSC certification for Indonesia’s deep-water snapper fishery (in partnership with The Nature Conservancy, Indonesia).
- Relevance for management
- The model will be used to explore timely research questions, including FAD management, understanding the spatial dynamics of the fishery, as well as some of the social and economic issues which effect management.
- Duration
- 36 months (end year 2024)
- Workplan and status
- A post-doctoral researcher will be based at the IATTC’s office in La Jolla, and will be charged with 1) scoping model application and designing a use cases that are supportive of IATTC policy evaluation processes, 2) understanding and accessing relevant datasets from IATTC, and 3) conducting statistical analyses of data to support model development.
- This researcher will work closely with the modeling team based at the University of Oxford and Ocean Conservancy to drive model design, calibration and validation of the tool and its outputs, as well as evaluation of model results.
- External collaborators
- University of Oxford
- Ocean Conservancy
- Deliverables
- A computer algorithm with which to run simulations to explore management options.
- A project report and publications in peer-reviewed journals.
- Updated date: 01 May 2023
- Progress summary for the reporting period
- Following the developing of an initial version of the POSEIDON operating model, the POSEIDON team developed a joint research plan in 2021 to continue developing the simulation tool in support of IATTC priorities. Following that plan, the POSEIDON team has recently completed a series of model development milestones. The POSEIDON model was expanded to include several feature expansions and updates, as follows. To represent a complete picture of the purse seine fishery, the POSEIDON team revised the fleet behavioral model to incorporate dolphin-setting vessels and improving the realism of unassociated sets in the simulation. They also augmented the model with an age-structured population dynamics model for Yellowfin, Bigeye and Skipjack tuna. In consultation with key international FAD researchers, changes were implemented to improve FAD aggregation dynamics. Further, an additional module was added to the model to represent value chain dynamics, such that the model can support evaluation of economic impacts of changes in the fishery. A joint diagnostics plan was developed to outline the standards that the tool must meet to match the IATTC’s standard of accuracy and scientific rigor. Thus, a model selection process was performed to identify the best performing model across a set of different FAD dynamic and trip planning algorithms.
- The revised model was used to perform a full calibration on 2017 observer and other supporting data. The results were compared to a series of diagnostics, co-developed by IATTC staff and the POSEIDON team, to measure the performance and skill of the model to capture important elements of the fishery including spatial and non-spatial catch, actions, and other trip planning indicators such as trip length.
- Overall, the POSEIDON model was able to reproduce catch, effort, and overall trip dynamics with low error. The spatial results were more error prone but overall were able to capture large scale patterns in fishing effort as well as the heterogeneity of actions from class 6 fishing vessels.
- The POSEIDON team is currently working to address comments and clarifications requested by IATTC to better understand the elements of the calibration process as well as suggestions made to improve the spatial “fit” of the calibrated model.
- Last, the model dynamics and infrastructure are being tailored to the management needs requested by IATTC staff by 1) improving the usability of the model by developing and R interface and 2) refining the spatial model validation process to be more flexible so that IATTC staff can better understand model skill for a range of spatial resolutions.
- Challenges and key lessons learnt
- The greatest challenge has been identifying a proper set of diagnostics to evaluate the model performance as agent-based models are not typically used in a fisheries management capacity. The co-development of these diagnostics with POSEIDON and IATTC staff was a significant undertaking but resulted in a tangible and novel set of diagnostics to judge the model. We expect these metrics to evolve over time as both teams learn more about the management needs and model capabilities and constraints.
- Another challenge was to identify the secondary drivers of the spatial fit in the southern region of the eastern Pacific Ocean.
- Presentations:
-EPO POSEIDON model diagnostics. IATTC staff. Jan 2023
-Development of an Agent-Based Bio-Economic Model for Tropical Tunas (POSEIDON). ICCAT SCRS. 2023
- Benefits of an Agent-Based Bio-Economic Model for the Indian ocean Tropical Tunas. 3rd IOTC Ad Hoc Working Group on Fads. 2023
- Modeling fish aggregating device drift in the Eastern Pacific Ocean using estimated ocean currents. 5th IATTC Ad Hoc Working Group on Fads. 2021
- POSEIDON Model of Eastern Pacific Tropical Tunas can inform management issues. 5th IATTC Ad Hoc Working Group on Fads. 2021
- Exploring FAD Management in the Eastern Pacific Ocean using an Agent-Based Bio-Economic Model: POSEIDON. World Fisheries Congress. 2021 - Comments
- Given the positive outcomes of the initial model diagnostics there has been some interest in applying the POSEIDON EPO-tuna model to other tropical tuna fisheries. The POSEIDON team is currently performing a data gap analysis to implement a similar model in the Atlantic Ocean with the goal of developing a joint project with several research institutions.