| dc.description.abstract | Why species are where they are is a central question in ecology. Distributions can be determined by a combination of individual requirements, environmental variables, and intra and interspecific interactions. Yet, it is difficult to identify and monitor these factors across large spatial and temporal scales, particularly for marine species with complex life history traits (e.g., migration) and ontogenetic shifts. In this thesis, together with my coworkers, I used unoccupied aerial systems (UAS, hereafter, aerial drones) to explore the factors that shape the distributions of loggerhead sea turtles (Caretta caretta) in western Greece (Mediterranean Sea), and that of marine megafauna in Cabo Verde (Northeast Atlantic). Firstly, Chapter 2 shows that different life stages (juvenile, subadults and adults) were associated with different foraging habitats. We identified one vegetated habitat type supporting a mixed-age group that was characterised by a high density of turtles (> 40 turtles/km2). We then used predictive models to find that this specific habitat was not particularly common, nor were those habitats associated with specific life stages. Nonetheless, along the Ionian and Adriatic coastline, there were other suitable habitats for mixed-age classes supporting low-density aggregations, which could potentially act as refuges of all life stages. For Chapter 3, we compared aerial drone data and published datasets (tracking, stranding and observations) to find that non-breeding and breeding distributions of sea turtles were more likely to be inside marine protected areas (MPAs) than outside. These distributions also largely overlapped with the distributions of other marine megafauna (pinnipeds, cetaceans, elasmobranchs, 76% overlap), demonstrating that sea turtles can serve as potential umbrella species to protect other marine megafauna in this region. In Chapter 4, we then questioned whether the mass arrival of sea turtles during their nesting season affected the in-water community structures of marine megafauna in Cabo Verde, a biodiversity hotspot. Communities near high-density nesting beaches became structured when the abundance of adult sea turtles was at its highest. Sharks positioned themselves in between the shoreline and adult turtles, potentially to take advantage of emerging hatchlings as prey. Near low nesting density beaches, no such structuring was detected, indicating a likely density-dependent threshold where the community structure changes because of a seasonal influx of sea turtles. The by-product of this high use of drone footage for this work highlighted that large-scale monitoring is limited by the time needed to process media. Hence, in Chapter 5 alongside software developers, I evaluated and compared the use of a novel, semi-automated detection software (Dronespot) against manual detection methods in terms of the number of turtles detected, the speed of analysis, and the accuracy of detection. We found that Dronespot accelerated data analysis with increased speed and accuracy, as well as detecting three times more turtles than with manual detection. Overall, these results demonstrate the potential for Dronespot in the data analysis of drone media, thus providing a novel tool for citizen science projects. In conclusion, this thesis highlights the use of aerial drones for evaluating wildlife populations and communities across large spatiotemporal scales. This thesis also demonstrates how aerial drones can be used to evaluate conservation strategies and to involve citizen scientists in analysing large quantities of data for monitoring projects. | en_US |
