Resumen
Identifying bottom-up (e.g., physical and biogeochemical) drivers for fish catch is essential for sustainable fishing and successful adaptation to climate change through reliable prediction of future fisheries. Previous studies have suggested the potential linkage of fish catch to bottom-up drivers such as ocean temperature or satellite-retrieved chlorophyll concentration across different global ecosystems. Robust estimation of bottom-up effects on global fisheries is, however, still challenging due to the lack of long-term observations of fisheries-relevant biotic variables on a global scale. Here, by using novel long-term biological and biogeochemical data reconstructed from a recently developed data assimilative Earth system model, we newly identified dominant drivers for fish catch in globally distributed coastal ecosystems. A machine learning analysis with the inclusion of reconstructed zooplankton production and dissolved oxygen concentration into the fish catch predictors provides an extended view of the links between environmental forcing and fish catch. Furthermore, the relative importance of each driver and their thresholds for high and low fish catch are analyzed, providing further insight into mechanistic principles of fish catch in individual coastal ecosystems. The results presented herein suggest the potential predictive use of their relationships and the need for continuous observational effort for global ocean biogeochemistry.