Identifying hubs and assessing the impact of large-scale phenomenon from movement data using a novel statistical tool

Background/Question/Methods

Understanding movement patterns of individuals is critical in multiple disciplines but paradoxically most of the methods to analyze movement data focus on locations, not individuals. Here we describe a novel statistical method that clusters individuals with similar movement patterns, enabling the identification of sites that are used by multiple groups (i.e., hubs), in sharp contrast with most of the existing network analysis methods, where locations can only belong to a single group. These hubs are likely to play a critical role in several contexts. For example, for endangered species, these hubs may be important to ensure population connectivity. For disease spread, these hubs could be strategically targeted for interventions to avoid the spatial spread of an epidemic. By focusing on individuals, the proposed method also enables a more direct evaluation of how large-scale phenomenon impacts movement patterns.

Results/Conclusions

Using simulated data, we show that our method reliably recovers the underlying groups of individuals, regardless of the presence or absence of mixed-membership locations, whereas the more traditional clustering algorithms used in network analysis struggle to determine the relationship between these locations. We then analyze long-term data from an endangered raptor, the snail kite, and show how this bird species has radically changed landscape use through time. We find that older individuals rely predominantly on southern lakes in Florida while younger individuals consistently use northern lakes. Furthermore, despite heavy reliance on southern lakes, our analysis reveals that older individuals exhibit a shift towards greater utilization of northern lakes in recent years. We attribute this change in landscape use to the exotic snail invasion of several northern lakes beginning in 2005, which substantially enhanced resources for snail kites. Finally, our analysis suggests substantial differences in hub sites before and after invasion, highlighting the challenges of identifying hub sites for conservation planning in the presence of global change. We believe that the proposed method is likely to be an important addition to the tool kit of researchers interested in understanding movement patterns and the effect of global change in these patterns.