Evaluation of species distribution forecasts: a potential predictive tool for reducing incidental catch in pelagic fisheries

Nontarget catch restrictions are becoming common in fisheries management. We test a potential tool for reducing nontargeted catch that combines species' distribution models and ocean forecast models. We evaluated our approach for Atlantic herring (Clupea harengus), Atlantic mackerel (Scomber scombrus), alewife (Alosa pseudoharengus), and blueback herring (Alosa aestivalis). Catch of the latter two species is capped in commercial fisheries of the former two species. Ocean forecasts were derived from a data-assimilative ocean forecast model that predicts conditions 0-2 days into the future. Observed oceanographic conditions were derived from CTD casts and observed fish presence-absence was derived from fishery-independent bottom trawl collections. Species distribution models were used to predict presence-absence based on observed and forecasted oceanographic conditions, and predictions for both were very similar. Thus, most of the error in predicted distributions was generated by the species distribution models, not the oceanographic forecast model. Understanding how predictions based on forecasted conditions compare with predictions from observed conditions is key to developing an incidental catch forecast tool to help industry reduce nontarget catches.