Behaviour 2019
Applying community algorithms to signal sequence networks
Heath R. Petkau, Samantha W. Krause, Peter C. Mower, Liam R. Mitchell, David M. Logue. University of Lethbridge, Lethbridge, AB, Canada

Signal sequence analysis can provide insights into signal function and development. One recently developed approach treats signal sequences as networks, in which nodes represent signal types and edges represent first-order transitions between signal types. For the present study, we applied community clustering algorithms to male Adelaide’s warbler (Setophaga adelaidae) song sequence networks. These algorithms identify clusters of nodes that are highly connected to one another, but weakly connected to nodes outside of the cluster. Our goals were to characterize the community structure of song type sequences during pre-dawn and post-dawn singing and test a prediction of the hypothesis that song types used during pre-dawn singing tend to cluster together when sung after dawn. Song sequences included communities, the structure of which differed between pre-dawn and post-dawn songs. We did not find support for the idea that common pre-dawn songs cluster together when sung after dawn. We conclude that community algorithms are a promising tool for the analysis of signal sequence networks.