Data-driven modeling of leading-following behavior in Bechstein’s bats

Abstract

Leading-following behaviour in Bechstein's bats transfers information about suitable roost sites from experienced to inexperienced individuals, and thus ensures communal roosting. We analyze 9 empirical data sets about individualized leading-following (L/F) events, to infer rules that likely determine the formation of L/F pairs. To test these rules, we propose ve models that dier regarding the empirical information taken into account to form L/F pairs: activity of a bat in exploring possible roosts, tendency to lead and to follow. The comparison with empirical data was done by constructing social networks from the observed L/F events, on which centralities were calculated to quantify the importance of individuals in these L/F networks. The centralities from the empirical network are then compared for statistical dierences with the model-generated centralities obtained from 10 5 model realizations. We nd that two models perform well in comparison with the empirical data: One model assumes an individual tendency to lead, but chooses followers at random. The other model assumes an individual tendency to follow and chooses leaders according to their overall activity. We note that neither individual preferences for specic individuals, nor other inuences such as kinship or reciprocity, are taken into account to reproduce the empirical ndings. * Corresponding author, fschweitzer@ethz.ch 1/25 P. Mavrodiev, D. Fleischmann, G.Kerth, F. Schweitzer:Data-driven modeling of leading-following behavior in Bechstein's bats Submitted for publication coordination problems in animal systems (Kerth and Reckardt, 2003;Kerth et al., 2006). First, the frequent switching of communal day roosts implies that group coordination and collective decision-making regarding communal roosting is vital for individuals getting grouping benets.Second, coordination must be achieved in the presence of dierent individual preferences and limited information about the suitability of roosts , which has been recognized as particularly challenging for collective decision-making (Conradt, 2011)As Bechstein's bats forage solitarily , they have only limited information of the nightly exploration behaviour of others. Moreover, some individuals are more active in exploring their habitat and thus become better informed about the location of suitable roosts than their less active colony mates (Kerth and Reckardt, 2003;Fleischmann et al., 2013). Information asymmetry and heterogeneous roosting preferences in the colony are, thus, invariable challenges in coordinating roosting behaviour and ensuring grouping benets.All these challenges not only make it dicult for the colony members to achieve group coordination, they also make it hard for us to infer which inuence, on the individual level, is relevant for the outcome. The information transfer from informed to naive individuals does not explain how such individuals nd each other. Recruitment plays an important role (Richner and Heeb, 1996;Kerth and Reckardt, 2003), but the rules by which individuals form leading-follo...

Quantifying individual influence in leading-following behavior of Bechstein’s bats

“…Null models are recognized as useful tools to test the viability of these recruitment rules in the presence of inherently non-independent behavioral data (Farine, 2017). We investigate a variety of such null models about recruitment behavior in Bechstein's bats in a subsequent paper (Mavrodiev et al, 2019).…”

An agent-based algorithm resembles behaviour of tree-dwelling bats under fission–fusion dynamics

“…This decision-making process observed in roost switching in tree-dwelling bats suggests a computational algorithm, a set of rules that defines a sequence of instructions to rule cooperation in a multi-agent system. Modern computational approaches to the study of social behaviour of animals are increasingly used and this emerging field provides many similar agent-based models today [39][40][41][42] even in social behaviour of bats [43][44][45][46] . The first agent-based model that attempted to simulate roost switching in tree-dwelling bats using biologically relevant data was run by Ruczyński and Bartoń 38 .…”