Behavioural ecology of the flatback turtle in the era of big data

Jenna L Hounslow

Sea turtles have faced a long history of many anthropogenic threats and as a result are highly conservation dependent. Despite spending most of their lives at sea, our understanding of sea turtles is profoundly biased towards their reproductive life-history stages and associated nesting, inter-nesting, and post-nesting habitat. The foraging life-history stage of sea turtles is considerably less well understood, despite this period and the habitats used being crucial for population level processes, such as growth, reproductive success, and survival. Understanding the behavioural ecology of sea turtles during the foraging life-history stage is therefore a high research priority. Further, out of the seven extant species of sea turtle, this paucity is most conspicuous for the Australian endemic flatback turtle Natator depressus. The knowledge gap surrounding the foraging life-history stage for sea turtles is largely owing to both logistical and technological limitations for studying the behaviour of free-ranging aquatic animals. Telemetry systems have enabled us to track movements at broad scales but offer fundamentally limited insight into precise behaviours. Over recent decades, biologging tools integrating multiple channels of remotely sensed, high-resolution data have transformed our capacity to study the behaviour of cryptic animals in the wild. But while our ability to collect vast amounts of complex multivariate data has given us astounding insight into behaviour for a variety of species, we are now faced with emergent challenges related to managing, processing, exploring, visualising, and analysing ‘big data’. The overarching aim of this thesis was to contribute to our understanding of the behavioural ecology of the flatback turtle during its’ foraging life-history stage, with a key focus on developing quantitative tools for monitoring in-water behaviour. To do this, I use a combination of contemporary biologging tools, sophisticated data modelling and artificial intelligence techniques to elucidate behaviour at a foraging site. First, I demonstrate the extraordinary observational powers afforded by biologging tools coupled with animal-borne video cameras, by describing, quantifying, and visualising subsurface behaviours newly discovered for sea turtles. I reinforce the potential for biologging technologies to increase the behavioural repertoire of cryptic species and to allow us to consider unexpected hypotheses. Second, using a data-driven approach, I discuss the challenges for objectively and reliably inferring diverse subsurface behaviours from complex, multivariate big data derived from biologgers. By demonstrating a viable alternative statistical technique, I provide a detailed description of in-water behaviour for flatback turtles and determine that turtles altered their diving behaviour according to their environment (e.g., extreme tides, wide ranging sub-tropical water temperatures), to thermoregulate and avoid predators while optimising foraging. Following this, using biologging tools coupled with animal-borne video cameras, I develop supervised machine learning algorithms to automatically detect precise behaviours from biologging data. I confirm formerly elusive characteristics of key foraging and resting behaviours. Subsequently, I investigate behaviour-specific spatiotemporal patterns of habitat use, predict suitable areas that support key behaviours and examine the influence of environmental variables. Taken together, I found minimal spatial segregation between foraging and resting sites. I argue that the extreme tide at the nearshore macrotidal study site prevents spatial delineation of areas used for foraging and resting, owing to the energetic cost of compensating for tidal current displacement, therefore impacting behaviour in both space and time. I then contextualise exposure to local threats and the level of protection offered to flatback turtles at the study site and recommend that dynamic management might confer better conservation outcomes. This will be particularly important in a region facing increasing development. Finally, I discuss the implications of using biologging tools to study animal movement and behavioural ecology. Collectively, this thesis offers a variety of novel and accessible methods that are broadly applicable to a range of taxa and study systems. Overall, this thesis plays an important role in refining our analysis of information rich, big data derived from biologging tools, and contributes to the continual advancement of a contemporary scientific discipline experiencing rapid uptake.

Behavioural ecology of the flatback turtle in the era of big data

Files and links (1)

Abstract

Details

UN Sustainable Development Goals (SDGs)

Metrics