Self-organized collective escape in bird flocks

Bird flocks show fascinating patterns of collective motion, particularly when escaping a predator. Little is known, however, about how these patterns come to be. This thesis aimed to fill this gap by analysing empirical data of bird flocks under attack by a robotic-predator and studying birds' collective escape in computer simulations. This approach was based on self-organization: the process with which patterns at the level of a group emerge from local interactions among individuals. We first studied the collective escape of pigeon flocks, discovering that group members avoid a predator more as the predator gets closer, even when they do not mind its position. We further investigated what patterns of collective escape arise in pigeons and studied how they emerge by self-organization. Second, we focused on the most common pattern of collective escape in bird flocks, the collective turn. We built an agent-based model in which flocks turn to escape a predator and used it to investigate how different turning types and specifics of coordination relate to a predator's confusion. Third, we studied the species demonstrating the most complex patterns of collective escape, the European starling. We identified that more than one pattern of collective escape may simultaneously co-occur in a single flock and developed a 3-dimensional agent-based model to study the emergence of this phenomenon. Finally, we presented the new framework in which our agent-based models have been developed, emphasizing on its contribution to the modelling of collective behaviour and towards a deeper understanding of collective escape.