Raven Hartman recently published her MSc thesis entitled, “Hierarchically embedded scales of movement shape the social networks of vampire bats” in the journal Proceedings of the Royal Society B.
There are many studies of social networks in bats. Before the popularity of biologgers, almost all of these studies created social networks from co-roosting associations. That is, the links between bats in the network represent the probability you would see those two bats roosting in the same tree (or bat house, or crevice, or cluster in a cave ceiling). However, these bat social networks are missing social dynamics at a smaller scale. In reality, not all the bats staying within the same hollow tree would be equally associated and likely to interact. We are also missing social dynamics at a larger scale, because the only bats that can interact within the same tree are the ones that live in the same forest. Social structure therefore often emerges from what we call “hierarchically embedded scales of movement“. That phrase means that movement at one scale is constrained within a larger scale (e.g. among branches, trees, forests). If you study movement at just one of these scales, you don’t get the full picture of the social network and what shapes it.
To explore this in vampire bats, Raven first compiled existing data. She took field observations of roost switching by wild bats observed by Jerry Wilkinson from 1978-1983. She then took our observations of clustering and cluster switching observed in captive bats. Using these data, she created an agent-based model in which virtual bats moved between trees at the rates observed in wild bats, and within each tree, they moved and groomed at the rates we observed in our captive bats.
To isolate the role of movement, she removed all spatial and social preferences. She then asked a series of questions. For instance: Do bats that move more also become more socially connected? What if they switch trees a lot but they rarely switch clusters? Or what if they switch grooming partners but rarely switch trees? What scales of movement are most important for being socially connected?
She found that, given how often real vampire bats moved, within-roost movements caused more variation in social connectedness than variation in roost switching. This results shows that co-roosting networks do not fully capture bat social structure. She also looked at the effects of correlations between different movement types.
Raven also used this paper to show how all the complexity of individual variation in movement occurring at nested spatial scales can make it difficult to identify and measure “preferred social relationships”. These preferred relationships are what give networks the interesting structure and properties that they have, but animal networks can be shaped more by observational biases than by true social preferences. To identify preferred relationships, one typically makes a null model of how often each pair of bats would be seen together under random movement or association (to compare against the observed data). But even “random” movements and associations are constrained by time and space in quite nuanced ways. Raven’s model illustrated the importance of temporal and spatial structure in creating sampling biases. Next, Raven is building on this agent-based model to compare the effectiveness of different social networking strategies that virtual bats will use to form and maintain cooperative relationships. She is also heading to Panama this summer to study variation in movement rates and sociality in real vampire bats.
Carter Lab 