Validation and improvement of species distribution models for structure-forming invertebrates in the eastern Bering Sea with an independent survey

Species distribution modeling is a useful tool for informing ecosystems management. However, validation of model predictions through independent surveys is rarely attempted in marine environments, which are challenging to study and often contain sensitive habitats. We conducted an underwater camera survey of the eastern Bering Sea slope and outer shelf as an independent test of species distribution modeling of deep-sea corals, sponges and sea whips based on bottom trawl survey data. We also refined model predictions by combining species distribution models based on both bottom trawl and underwater camera survey data. The camera survey also was conducted to determine density and size of the taxa. The trawl model predictions generally were confirmed by the camera observations (area under the receiver-operator curve [AUC] values of 0.63 to 0.73). Combining bottom trawl and camera survey model predictions improved predictive ability (AUC values of 0.74 to 0.90 for camera observations). Corals were distributed in Pribilof Canyon and the slope area to the northwest of the canyon, and colony densities averaged 0.005 ind. m(-2) and ranged from 0 to 0.28 ind. m(-2). The low densities were consistent with the absence of hard substrates for coral attachment in most areas of the eastern Bering Sea. Sponge and sea whip density averaged 0.11 ind. m(-2), with sponge density ranging from 0 to 13.1 and sea whip density ranging from 0 to 8.4 ind. m(-2). Invertebrate heights were generally small, with most taxonomic groups < 20 cm in average height. This type of study is vital to providing the best scientific advice for spatial management of structure-forming invertebrates, so that decisions concerning the protection of these vulnerable communities can be implemented with a clear basis for priorities.