Kartläggning av viktiga livsmiljöer för hajar och rockor i Västerhavet

Distribution maps of essential fish habitats are crucial in coastal and marine management, at local, regional and national levels. Maps are used within fish management, habitat conservation and marine spatial planning, as well as for handling of permits for construction or water operations. Habitat modelling, where statistical methods are used to link the occurrence or abundance of species to various environmental variables, is a method that makes it possible to produce distribution maps of suitable habitats and abundances. In this study, data from trawl surveys at the Swedish west coast during 2005–2022 were used to map suitable habitats and abundance for two shark species, dogfish (Squalus acanthias) and lesser spotted dogfish (Scyliorhinus canicula) and two ray species, starry ray (Amblyraja radiata) and thornback ray (Raja clavata). For each of the studied species the analyses resulted in an annual map based on all available data and two half-year maps, based on data from the first and last six months of the year. We used a modelling technique referred to as delta modelling (also called hurdle models), where you first map the presence of suitable habitats using presence-absence data and then predict the abundance within these areas using abundance models. For both occurrence and abundance models, ensemble modelling was applied, where a number of conceptually different statistical methods were combined to obtain more reliable distribution maps. The presence-absence models, which were used to map the occurrence of suitable habitats, were consistently strong with AUC values ranging between 0.87– 0.92 in the model based on all data and 0.85–0.95 in the half-year maps, meaning that they could distinguish between suitable and less suitable habitats. A majority of the fitted abundance models explained a relatively large proportion of variation, with a range of 15–49 %. The produced distribution maps (where abundances outside suitable habitat were set to zero) explained 1–15 % of the variation. Maps produced in this study can be used within fish management, marine spatial planning, green infrastructure and area protection mainly at a county to national level.