Regensburg 2022 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
BP: Fachverband Biologische Physik
BP 23: Evolution
BP 23.3: Talk
Thursday, September 8, 2022, 10:30–10:45, H13
Data-driven modeling of social interactions in bats across time scales — •Frank Schweitzer1, Pavlin Mavrodiev1, and Gerald Kerth2 — 1Chair of Systems Design, ETH Zürich, Switzerland — 2Applied Zoology and Nature Conservation, University of Greifswald, Germany
The study of bat's social and foraging behavior is of great relevance to forecast the outbreak and distribution of virus induced deseases. To analyze this behavior we use a large-scale data set from two colonies of Bechstein's bats over five years. From this data, we reconstruct the social interactions of bats at three different time scales: (a) At the scale of minutes: social influence and information transfer. This leads to the formation of leader-follower pairs, where an informed individual leads an uninformed one to a roost box. (b) At the time scale of days: fission-fusion dynamics. This leads to the formation and dissolution of roosting groups of different size, composed of different individuals. (c) At the time scale of months: Emergence of social structures. This leads to the formation of communities within a colony. While the analysis of (a) requires statistical data analysis and hypothesis testing, for (b) we employ agent-based models, and for (c) social network analysis. The combination of these approaches allows us to bridge time scales in social behavior, which cannot be observed together. With our models we are able to develop the bigger picture of how social interactions feed back to long-term social structures.