Animal aggregations are common in nature and directly influence important biological processes such as resource acquisition, predator avoidance, and reproduction. Yet our ability to study the behavior of animal groups with precision has been limited by the challenge and expense of tracking multiple individuals at once. As a result, the majority of knowledge about collective behaviors – particularly in large, or far-ranging species – has been derived from detailed observations of relatively few individuals (i.e. field observations or marked individuals), or theoretical simulations lacking validation in the field. While these approaches have formed the foundation of animal behavior research, they frequently suffer from an inability to generalize individual-level observations to population-level processes.
Recent advancements in high-resolution remote sensing now present exciting opportunities to overcome many of these limitations by obtaining a comprehensive view of individual behaviors and the full socio-environmental context in which they are embedded. By adapting high-resolution remote sensing technologies to study large groups of animals in the wild, I aim to examine the role of environmental and social forces in driving behavioral shifts in ungulate herds across multiple scales of space and biological organization (i.e. individuals to populations).
To accomplish this, I first establish the historical context for these new techniques by conducting a literature review of modern and historic methods for studying collective animal behavior in the field. This review outlines the strengths and limitations of such methodologies while identifying opportunities for advancement afforded by recent and upcoming technological innovations. I then demonstrate the utility of this technology by developing a new method to study the distribution and behavior of tens of thousands of animals (white-bearded wildebeest; Connochaetes taurinus) identified in high-resolution (< 50 cm) satellite imagery. This non-invasive method is directly scalable from individuals to populations and reliably predicts three behavioral states (82% accuracy overall) from a single metric of group structure (i.e. coordinated orientation). Such an advancement represents a step-change in our ability to study social processes under natural conditions and forecasted advancements in both resolution and automated image processing are likely to expand applications for this technique in the near future.
Finally, I combine these new remote sensing techniques with traditional animal tracking methods (e.g. GPS telemetry and transect surveys) to evaluate the factors that drive habitat selection by a reintroduced, free-ranging ungulate (tule elk; Cervus canadensis nannodes) in a cattle-dominated ecosystem in northern California. These analyses confirmed that reintroduced elk largely avoided areas managed for and used by cattle across all seasons. Further, the use of remotely sensed data revealed that this pattern likely resulted from differential habitat preferences rather than outright avoidance behavior. As a result, the potential for conflict between cattle and reintroduced elk is expected to be minimal in this ecosystem, and managers may continue to rely on manipulation of resources (e.g. providing artificial water sources) to limit interactions between the two species. By presenting this updated approach to the study of animal behavior, I aim to demonstrate the value of remotely sensed data for providing both basic and applied insights into the behavioural ecology of large animal systems.
