Marine megafauna includes marine mammals, sea turtles, sharks and rays that are increasingly threatened worldwide. These species occur at low densities over vast areas so an efficient and large-scale monitoring method is urgently needed to identify and protect their critical habitats. Drones are emerging low-cost and low-carbon technologies for monitoring megafauna from the air. However, existing drone surveys fail to achieve both large spatial coverage and high image resolution. The higher the drone altitude, the larger spatial coverage but at the expense of image resolution. Yet, high image resolution is key to derive accurate species identifications and individual body measurements. To break this compromise between spatial coverage and image resolution, the overall objective of SMART-WING is to design an adaptive survey methodology based on an intelligent wing able to switch between high and low altitude modes. We will embed an artificial intelligence (AI) detection model in a vertical take-off and landing (VTOL) wing characterized by a flexible long-range flight. The wing will be pre-programmed in high-altitude mode (160 m for ~2 cm / pixel ground resolution) for large-scale survey coverage. In the event of megafauna detection, it will descend to lower altitude (40 m) and dynamically track the animal, transitioning from horizontal survey to vertical hovering to acquire high-resolution images (~0.5 cm / pixel), so representing a 4-time gain in ground resolution. The wing will then resume its initial high-altitude survey. We hypothesise that this novel method will provide an increased image resolution without compromising spatial coverage. We will implement SMART-WING in the vast lagoon of Mayotte hosting exceptional but poorly known populations of marine mammals, sea turtles, sharks and rays. SMART-WING will provide novel ecological information on megafauna populations at low financial and carbon costs toward the identification of key areas for their protection. The project has 3 scientific objectives: 1) Develop a smart wing prototype. We will develop an affordable, long-endurance, eco-responsible and easy-to-operate VTOL wing able to fly both horizontally and vertically. The wing will be equipped with a mini-computer embedding a lightweight AI model to detect megafauna species in real time. We will implement a re-routing algorithm on the wing’s autopilot in order to adapt the trajectory to real-time detections. The wing prototype will be tested in coastal habitats of the Mediterranean Sea and then in Mayotte to gradually improve its operational performance. 2) Monitor species-specific abundances and body characteristics. We will deploy the wing to monitor megafauna across thousands of hectares in Mayotte’s coastal habitats during 2 seasons and 2 successive years. We will analyse the collected images with a full-size AI model, quantifying performance gain when switching from the high-altitude and to the low-altitude flight mode. We will then apply this AI model to high-resolution images to estimate species-specific abundances as well as individual sizes and body conditions. 3) Simulate marine protected area (MPAs) scenarios to safeguard megafauna. We will build statistical models to relate species-specific abundances and relevant environmental predictors in order to provide abundance predictions in unsurveyed areas of Mayotte. We will then use systematic conservation planning algorithms to derive scenarios of strict MPA establishment for safeguarding megafauna while accounting for the loss of fishery catches. SMART-WING will pave the way toward the development of a new generation of drones able to dynamically adjust their trajectory in real time for the optimised survey of rare and vulnerable species. It will foster cross-domain interactions between scientists in aerial robotics (I3S and ISEN), artificial intelligence (LIRMM), marine ecology (CUFR, MARBEC) and environmental management (OFB-PMM).