Collective escape dynamics of fission-fusion groups in the wild

Abstract

Collective movement is a fundamental process affecting the survival and reproductive success of group-living animals. Many of the hypothesized benefits of grouping such as predation evasion and foraging efficiency may require the individuals to move in a coordinated way. While moving in groups, animals are not only responding to the environment but also interacting with each other. These interactions give rise to emergent collective movement and behavioural patterns. Such visually spectacular patterns may arise from simple local interactions between the individuals. Individuals also respond to their environment (habitat structure) and predators. These external factors (habitat and predation) may further affect individuals' social interactions and hence ultimately affect the collective behavioural patterns.

Most studies on the emergent properties of collective behaviour are conducted in controlled conditions. However, in natural settings, habitats are heterogeneous in resource distribution, availability of hiding places and substrate for movement. Empirical studies have rarely investigated such fine-scale interactions (e.g. alignment, attraction among individuals) in their natural habitat. One reason for the shortage of such studies is the difficulty of data collection. Recent advances in techniques of aerial imagery allow us to observe and record such fine-scale data. For my PhD project, I studied the collective behaviour of blackbuck herds in their natural habitat. More specifically, I investigated the collective response of blackbuck herds to predation-like events. By analysing fine-scale movement patterns and multiple interactions among group members simultaneously, I aimed to understand the role of social interactions in shaping blackbuck herd's collective response when faced with predation-like threats.

First, we overcome the difficulty of observing fine-scale interactions in animal groups (in their natural habitat) using Unmanned Aerial Vehicles (UAVs). We recorded blackbuck herding behaviour at high spatio-temporal resolutions (30 frames per second). Using this method, we were able to record blackbuck herd’s collective escape behaviour in the context of predation using controlled-simulated threats. Tracking animals in the videos recorded in natural habitat is extremely difficult due to varying background and light conditions and clutter in the background. Relatively basic image processing methods and default tools do not perform satisfactorily in such a scenario. Hence, we developed a machine learning method (using Convolutional Neural Networks) and GUI tool to extract all the individuals' spatial locations and movement trajectories from the videos recorded in natural field conditions.

Once we were able to obtain the movement trajectories from the videos, we then analyse these trajectories and interactions between individuals to explore - how the information about predatory risk spreads through a group in natural conditions. Broadly, our results suggest that transient leader-follower relationships emerge in these groups while performing a high-speed coordinated movement. Also, males and females respond differently to the threat scenario: adult females are more likely to be the response initiators, whereas adult breeding males are more likely to influence the group movement during the escape response. Our results indicate that in fission-fusion groups, associations are likely to last for short time scales and spatial positions of the individuals only affect their response-time (vigilance behaviour) but not their influence on the group.

This study throws light on how fission-fusion groups operate and function, and we see a distinct set of decision rules emerging in such groups. It also provides a framework to study other types of animal societies which may aid in further exploring the ecological conditions that may favour the evolution of different sets of decision rules.