Abstract
Mapping the seabed along the Norwegian coast is costly and time consuming. Hence, finding a modeling method to separate rocky seabed from other substrate types will provide digital maps that may be used to develop cost-effective sampling designs to predict species and habitat distribution. Our approach was to use geophysical data that were quantitative and objectively defined, generalized additive models (GAMs), and Akaike information criterion (AIC) to develop statistical models and select among them. We found that slope, terrain curvature, wave exposure, and depth predicted rocky seabed occurrence with a high degree of certainty.
Acknowledgements
The project was a cooperation between the SUGAR KELP project (funded by the Research Council of Norway and the Norwegian Pollution Control Authority) and the DYNAMOD project (funded by the Norwegian Institute for Water Research). We wish to thank the National Marine Mapping and Monitoring Program (funded by the Ministry of Environment, the Ministry of Fisheries and Coastal Affairs and the Ministry of Defense) and the SPIMOD project (financed by Norwegian Institute for Water Research) for data supplies. Thanks to Pål Erik Isachsen (Norwegian Meteorological Institute) and the two reviewers for valuable comments and to Martin Isæus (AquaBiota Water Research, Sweden) for adjustments of the wave exposure model.