Assessment modelling approaches for stocks with spawning components, seasonal and spatial dynamics, and limited resources for data collection.

Abstract

The true spatiotemporal structure of a fish population is often more complex than represented in assessments because movement between spawning components is disregarded and data at the necessary scale are unavailable. This can generate poor advice. We explore the impacts of modelling choices and their associated risks given limited data and lack of biological knowledge on spawning component structure and connectivity. Pseudo-data for an age structured fish population were simulated with two spawning components that experience various levels of connectivity and that might overlap during a certain period but segregate during reproduction. A variety of implicit spatiotemporal and simpler models were fitted to the pseudo-datasets, mimicking different situations of data availability. To reproduce the true stock characteristics, the spatiotemporal models required total catch data disaggregated by spawning component; however, catch-at-age was not as important nor were disaggregated biomass indices to reproduce true dynamics. Even with just 5% connectivity between spawning components, both the spatiotemporal models and simpler alternatives generally overestimated stock biomass. Although bias was smallest when considering one unit population, spawning components might still need to be considered for management and conservation. In such case, the spatiotemporal model was less influenced by ignored connectivity patterns compared to a model focussing on one spawning component only.