Benthic communities of DeepInsight Hill, Mohn’s Ridge (Arctic Ocean)

The DeepInsight Hill area on Mohn’s Ridge exhibits high biological heterogeneity associated with geological complexity ranging from fault scarps to sedimentary plains. This study applies the DeepSee object detection model alongside manual video review to map and quantify benthic megafauna across multi-year ROV surveys (2023-2025), characterizing community structure and bathymetric patterns across depth gradients from 980 to 2100 m. Three major community-defining taxa were identified: Rossellidae (glass sponges), Tetractinellida (demosponges), and Ophiuroidea (brittle stars), which showed clear bathymetric and geomorphological partitioning. Ecologically significant features were documented, including structure-forming sponge aggregations, ophiuroid beds, sedimentary cave structures, and probable breeding grounds for Arctic skate (Amblyraja hyperborea) and glacial eelpout (Lycodes frigidus). The DeepSee model successfully detected community-defining taxa at high spatial resolution, reducing annotation workload substantially compared to manual methods. Network and correlation analyses revealed distinct, structured species associations. When integrated with spatial data, these fine-scale patterns reveal how community composition shifts across diverse topographic features, from summits to steep flanks and sedimentary plains. By uncovering localized patterns that are masked at coarser resolutions, this work demonstrates that meter-scale mapping is critical for accurately assessing heterogenous benthic megafaunal communities on mid-ocean ridges.