Abstract:
Marine ecosystems are rapidly changing under increasing anthropogenic pressures, creating a growing need for efficient and scalable monitoring approaches. Unoccupied Aerial Systems or 'drones' have expanded our capacity to acquire high-resolution, spatially explicit imagery of coastal ecosystems, while advances in photogrammetry and computer vision are transforming visual data analysis. However, drone-based remote sensing approaches for ecological research in optically challenging coastal environments remain comparatively underdeveloped. This thesis investigates how drone-based remote sensing, photogrammetry, and computer vision can be integrated into novel workflows to quantify ecological patterns and spatial heterogeneity across central Red Sea coral reef ecosystems. Chapter 1 systematically evaluated the accuracy of Structure-from-Motion- and spectrally derived bathymetry in shallow water (< 2 m deep) across a range of ground-truth availability and hardware constraints, providing practical guidance for bathymetry retrieval on remote or offshore coral reefs. Chapter 2 focused on how sympatric stingrays partition coastal habitats. By integrating standardized drone surveys with AI-assisted image analysis, we found that bluespotted ribbontail rays (Taeniura lymma) selected shallow (< 0.4 m deep) lagoonal habitats, while larger-bodied whiprays (predominantly Himantura uarnak) occupied macroalgal reef flat habitats. Active tracking surveys revealed corresponding differences in foraging behaviour. Chapter 3 examined the reef-flat-scale distribution and bleaching responses of yellow and pink Stylophora colour morphs. Combining AI-based instance segmentation with photogrammetric reprojection enabled the mapping of thousands of colonies, revealing strong spatial structuring of intraspecific colour morphs along the reef exposure gradient. Repeated drone surveys further uncovered markedly higher short-term mortality in the pink morph following the 2024 bleaching event. Collectively, these studies demonstrate the utility of drone-based remote sensing for tropical seascape ecology. When combined with fit-for-purpose analytical workflows, drone observations can resolve seafloor topography, species-habitat associations, fine-scale behavioural variation, coral population structure, and disturbance responses across areas that would be difficult to characterise using conventional field observations. Rather than replacing in-water investigations, drone-based remote sensing offers a complementary means of linking organism-level observations to their broader seascape context. The increasing automation of image analysis provides new opportunities for regularly repeated, large-scale monitoring of rapidly changing tropical coastal ecosystems.