Fission-fusion events in social groups: the challenge of identifying group membership from continuous GPS data

Katja S. Della Libera, Minerva Schools at KGI, San Francisco, CA, Ariana Strandburg-Peshkin, Biology Department, University of Konstanz, Konstanz, Germany; Department for the Ecology of Animal Societies, Max Planck Institute of Animal Behavior, Konstanz, Germany, Simon C. Griffith, Department of Biological Sciences, Macquarie University, Sydney, Australia and Stephan T. Leu, School of Animal and Veterinary Sciences, University of Adelaide, Adelaide, Australia

Background/Question/Methods
Animals form social groups when the benefits of group living outweigh the costs. The cost-benefit ratio can be group size dependent which led to the development of the concept of optimal group size. However, group size does not have to be static: fission-fusion events, when groups split into smaller groups or merge into larger groups, can drive changes in the size and composition of groups over time.The mechanisms that link individual movement to fission-fusion dynamics at the group level are not well understood.
With the rise of GPS devices, tracking of individuals with high precision over longer time scales has become possible. This new methodology brings with it new questions about how best to define social groups, merging, and splitting events based on individual movement trajectories.
Here we developed a method based on a modification of DBSCAN for identifying group membership over time and extracting fisson-fusion events, and applied it to GPS data of 51 female sheep(Ovis aries) tracked over four 3-day periods in a 6 km2 paddock in the semi-arid region of Australia. Using this method, we then investigate drivers of group size and splitting / merging events.

Results/Conclusions
We found that group size and overall connectivity between individuals (density) increased overnight and during the middle of the day, which are periods when animals typically rest. This is consistent with anti-predator behavior and the increased benefits of larger groups in relation to predator detection and safety in numbers. The net benefit is further increased at these time periods because intra- and inter-group competition for food is absent when animals either rest, sleep or ruminate. Analyzing the spatial and temporal occurrence of fission and fusion events revealed that both types of events were most frequently observed early in the morning and late at night. This suggests that the start and end of the daily activity is a time of frequent social restructure. In contrast to this clear temporal pattern, spatially we did not detect a general pattern. In particular, splitting and merging events could occur anywhere in the pasture and did not only occur in the vicinity of food and water resources. This suggests that fission and fusion events were not predominantly driven by resource competition. We discuss our findings in a socio-ecological context further improving our understanding of the drivers of fission-fusion events in social mammals, such as sheep.