The devil’s in the details when using correlative and mechanistic species distribution models to inform multispecies and ecosystem models

The ability to predict spatial distributions of marine species has advanced rapidly in recent years. Benefits include extrapolating habitat maps when data are sparse, thus improving our ability to quantify shifts in species ranges seasonally and over past decades, understanding the ocean conditions that drive these shifts, and projecting spatial redistribution due to climate change. Efforts have begun to couple these spatial distributions to complex multispecies and end-to-end-ecosystem models. The resulting integration of distributions, stock size, trophic interactions, and fishing effects allows for more realistic dynamics than what either model type is likely to attain alone. Coupling species distribution models to multispecies and ecosystem models is intuitively attractive, yet presents us with multiple decisions and approximations. Here we identify best practices for the detailed decisions (“the devil’s in the details”) required for such coupling. We summarize lessons learned from five case studies in the California Current and Peru Current ecosystems, the Celtic Sea, and the Gulf of Alaska, with a focus on examples regarding small pelagic fish.