Abstract
Predators are generally much bigger, faster and stronger than their prey. Despite this, the vast majority of attacks are unsuccessful. A general explanation for this paradox has long eluded ecologists. As a result, the mechanisms that underlie rates of interactions between predators and their prey – one of the most important quantities in ecological theory – remain poorly understood. This is in part because the outcome of predator-prey interactions appear to depend on animal behaviours that are complex and idiosyncratic. However, recent research from the field of neuroethology has revealed that such complexity often emerges from surprisingly simple behavioural rules that couple sensory inputs to motor outputs. This suggests that by identifying these rules, and modelling their outcomes, we may be able to resolve the paradox of prey evasion, and more generally develop tractable theory to predict the outcome of predator-prey interactions. However, such a theory is currently limited by a paucity of fine-scale behavioural data of predator-prey interactions in natural contexts. In the proposed research, I will overcome this limitation by improving upon an innovative underwater field observatory that I recently developed to continuously film attack and evasion behaviours. This behavioural observatory will allow me to study predator-prey interactions at spatial and temporal resolution that have never before been possible in a field setting. Using a novel machine learning pipeline to process these videos, I will reconstruct both the visual sensory inputs and behavioural outputs of predators and prey during interactions and use this dataset to test theory-driven hypotheses about the rules that underlie these behaviours. By identifying these “rules of life” and using mathematical models to explore their consequences, I will resolve the paradox of how prey evade much bigger, faster, stronger predators. More broadly, this work will generate a fundamentally new and tractable theory of predator-prey interactions.
Netherlands