Highly resolved spatiotemporal simulations for exploring mixed fishery dynamics

Highlights

• We developed a highly spatially resolved discrete-event simulation model of mixed fishery interactions with multiple populations.

• The simulation framework incorporates i) delay-difference population dynamics, ii) population movement using Gaussian Markov Fields to simulate patchy, heterogeneously distribution populations, and iii) individual-based fishery dynamics for multiple fleet characteristics using an explore exploit strategy.

• The framework allows users to explore assumptions in modelling observational data and examine dynamics at a fine spatiotemporal scale.

• We simulate fifty years of multiple fisheries exploiting four separate populations with different demographics and find nuanced spatiotemporal patterns in exploitation.

• A simulated spatial closure shows that when aggregating point data on fishing activity to a grid, the scale of aggregation matters for the ability to achieve fishery-conservation objectives.

• The simulation framework is available as an R package and we suggest multiple potential uses of the package (e.g. survey design evaluation, testing index standardisation, in-year fishery and biological modelling) for researchers.

Abstract

To understand how data resolution impacts inference on mixed fisheries interactions we developed a highly resolved spatiotemporal discrete-event simulation model MixFishSim incorporating: i) delay-difference population dynamics, ii) population movement using Gaussian Random Fields to simulate patchy, heterogeneously distributed and moving fish populations, and iii) fishery dynamics for multiple fleet characteristics based on population targeting under an explore-exploit strategy. We applied MixFishSim to infer community structure when using data generated from: commercial catch, a fixed-site sampling survey design and the true (simulated) underlying populations. In doing so we thereby establish the potential limitations of fishery-dependent data in providing a robust characterisation of spatiotemporal distributions. Different spatial patterns were evident and the effectiveness of a simulated spatial closure was reduced when data were aggregated across larger spatial areas. The simulated area closure showed that aggregation across time periods has less of a negative impact on the closure success than aggregation over space. While not as effective as when based on the true population, closures based on high catch rates observed in commercial data were still able to reduce fishing on a protected species. Our framework allows users to explore the assumptions in modelling observational data and evaluate the underlying dynamics of such approaches at fine spatial and temporal resolutions. From our application we conclude that commercial data, while containing bias, provides a useful tool for managing catches in mixed fisheries if applied at the correct spatiotemporal scale.