High-resolution predictive mapping reveals novel benthic habitats in the southern Baltic Sea.

Marine benthic habitats are essential for biodiversity and ecosystem functioning, yet their complexity and inaccessibility challenge effective mapping and management. This study presents the first high-resolution predictive benthic habitat maps at the EUNIS Level 5 for the Polish Exclusive Economic Zone (EEZ) in the southern Baltic Sea, covering over 7000 km2 of direct high-resolution measurements. Using advanced hydroacoustic technologies-multibeam echosounders and side-scan sonars-combined with 1425 sediment grab samples and underwater video ground-truthing, we applied a Random Forest classifier trained on bathymetric derivatives via Object-Based Image Analysis (OBIA). The model achieved high accuracy: 96.94 % overall and a kappa coefficient of 0.94 for substrate classification, and 97.88 % overall with a kappa of 0.97 for dominant species. We identified 47 distinct benthic habitats, including 24 novel EUNIS Level 5 unique to the southern Baltic, requiring the proposal of three new EUNIS Level 4 categories to maintain taxonomic hierarchy. These findings enhance understanding of Baltic Sea biodiversity and provide a scalable framework for benthic habitat mapping, supporting ecosystem-based management under increasing anthropogenic pressures.