A new, multi-scaled graph visualization approach: an example within the playa wetland network of the Great Plains

Abstract

We employed a sliding-window approach at multiple scales (window sizes and dispersal distances) to calculate seven standard graph-theoretical metrics within a subset of a large, freshwater wetland network. In contrast to most graph analyses, which quantify connectivity at a single (global) scale or at a patch-level scale, a multi-scaled, sliding-window approach provides an assessment that bridges these two approaches to examine patch clusters. As a case study we focused on a subset of a habitat patch network in a ~20,000 km2 area encompassing 2,782 playa wetlands in the panhandle of Texas. Playas are seasonal wetlands of the southern Great Plains of North America that form a network of regional habitat resources for wildlife. The large size of this network meant that global metrics failed to capture localized properties, so we used contour mapping to visualize continuous surfaces as functions of playa density, linkage density, and other topological traits at different window sizes and dispersal distances. This technique revealed spatial patterns in the components (i.e., the network properties of regions of the landscape at a given dispersal scale), with the spatial scale of habitat clustering varying with the size of the sliding window and dispersal distance. Using a tool familiar to landscape ecology (sliding-window methodology) in a novel way (to examine ecological networks at multiple scales), our approach provides a way to represent ecologically determined local-scale graph properties and illustrates how a multi-scaled approach is useful in examining habitat connectivity to investigate graph properties.