Forecast-ready models to support fisheries' adaptation to global variability and change

Abstract

Ocean and climate drivers affect the distribution and abundance of marine life on a global scale. Marine ecological forecasting seeks to predict how living marine resources respond to physical variability and change, enabling proactive decision-making to support climate adaptation. However, the skill of ecological forecasts is constrained by the skill of underlying models of both ocean state and species-environment relationships. As a test of the skill of data-driven forecasts for fisheries, we developed predictive models of catch-per-unit-effort (CPUE) of tuna and billfish across the south-west Pacific Ocean, using a 12-year time series of catch data and a large ensemble climate reanalysis. Descriptors of water column structure, particularly temperature at depth and upper ocean heat content, emerged as useful predictors of CPUE across species. Enhancing forecast skill over sub-seasonal to multi-year timescales in any system is likely to require the inclusion of sub-surface ocean data and explicit consideration of regional physical dynamics.