Evaluating the spatial transferability of habitat suitability models: implications for conservation and management

Habitat models often rely on data with incomplete or patchy survey coverage, making it necessary to project observed habitat relationships into unsampled areas. Model transferability is often assumed, rarely tested, and may result in uninformed management recommendations if not validated. In the eastern Gulf of Maine, data from the federal trawl survey has historically been used to identify suitable habitat for juvenile Atlantic cod (Gadus morhua), despite mainly sampling in offshore waters. This study evaluated the spatial transferability of a habitat suitability index model by projecting offshore habitat relationships inshore and comparing them to models built using two inshore surveys. Due to rare warm-water exposure in offshore waters, the federal survey model was unable to identify suitable habitat in the warmer, shallower areas that the inshore surveys identified, suggesting that offshore models are not transferable inshore for juvenile cod in the eastern Gulf of Maine. The results of this study have implications for fisheries worldwide and suggest that incorrectly assuming model transferability could pose a significant hurdle for effective conservation and management.