The structure of borders in a small world

TL;DR

This paper investigates the multi-scale structure of human mobility networks to identify effective borders that influence various spatial phenomena, revealing that large-scale boundaries often differ from administrative borders and are crucial for understanding mobility-driven processes.

Contribution

It introduces a multi-scale proxy network for human mobility and uncovers effective borders that differ from administrative boundaries, enhancing understanding of spatial coherence in mobility patterns.

Findings

Large-scale boundaries define spatially coherent regions.

Effective borders only partially align with administrative borders.

Long-range traffic reduces spatial coherence.

Abstract

Geographic borders are not only essential for the effective functioning of government, the distribution of administrative responsibilities and the allocation of public resources, they also influence the interregional flow of information, cross-border trade operations, the diffusion of innovation and technology, and the spatial spread of infectious diseases. However, as growing interactions and mobility across long distances, cultural, and political borders continue to amplify the small world effect and effectively decrease the relative importance of local interactions, it is difficult to assess the location and structure of effective borders that may play the most significant role in mobility-driven processes. The paradigm of spatially coherent communities may no longer be a plausible one, and it is unclear what structures emerge from the interplay of interactions and activities across spatial scales. Here we analyse a multi-scale proxy network for human mobility that incorporates travel across a few to a few thousand kilometres. We determine an effective system of geographically continuous borders implicitly encoded in multi-scale mobility patterns. We find that effective large scale boundaries define spatially coherent subdivisions and only partially coincide with administrative borders. We find that spatial coherence is partially lost if only long range traffic is taken into account and show that prevalent models for multi-scale mobility networks cannot account for the observed patterns. These results will allow for new types of quantitative, comparative analyses of multi-scale interaction networks in general and may provide insight into a multitude of spatiotemporal phenomena generated by human activity.