Background/Aims:
Northern Australia’s coastal ecosystems support a diverse range of culturally and ecologically important marine species, including green turtles, dugongs, and reef fish. Yet, marine habitats, particularly in remote and turbid regions, remain poorly mapped and underrepresented in national environmental assessments.
Methods:
Visual classifications of towed video transect data were used in a Support Vector Machine Learning Model to predict habitat across 379 km2 of remotely sensed satellite imagery, encompassing two green turtle foraging grounds within jointly managed parks: Trepang Bay (Cobourg Marine Park) and Field Island (Kakadu National Park).
Results:
Habitat types such as seagrass and macroalgae—critical resources for herbivorous fish, green turtles, and other grazers—comprised 30% of habitat cover at Trepang Bay and 18% at Field Island. Model accuracy was high at both sites (0.63 and 0.75), demonstrating the robustness of the method for use in remote environments.
Conclusion:
These maps directly support ecosystem-based management, environmental approvals, and impact assessments. To ensure long-term value and cultural relevance, the project was co-produced with Indigenous ranger groups, including training in field methods and data management. Outputs include an accessible science communication platform featuring interactive habitat maps, videos, and plain-language summaries. This work strengthens Indigenous Sea Country management and provides a scalable model for filling spatial data gaps to inform national frameworks like Biologically Important Areas under the EPBC Act.