A meta-analysis of connectivity metrics: Comparing binary and complex landscapes

Background/Question/Methods

Connectivity is a key ecological phenomenon that mediates population extinction and persistence, influences the structure of metapopulations and metacommunities, and modifies the strength of ecosystem-level flows. At landscape scales, species distributions are often modeled using a metapopulation framework in which patches are distinguished from non-patch habitat, and isolation measures based on patch geometry are expected to be important predictive variables. Although hundreds of studies have utilized a binary (patch/non-patch) landscape perspective, these measures are often poor predictors of population parameters. Recent syntheses of isolation effects on animal populations and communities underscore the limitations of the binary landscape paradigm and argue for the need to incorporate data on matrix composition into models of patch population parameters. We provide the first empirical evaluation of the binary landscape paradigm by conducting a meta-analysis to determine whether connectivity metrics that include terms for matrix composition explain more variance in patch occupancy and abundance than metrics based solely on patch geometry. We assembled a unique dataset of over 200 effect sizes scored on several factors (taxon, dispersal type, latitude, landscape size) that allowed us to characterize the generality of results across taxa, ecological groups, and geographic contexts. Results/Conclusions Our results indicate that metrics including matrix composition tend to have greater effect sizes than metrics based on analyses of binary landscapes, and this result holds across taxa and space. Connectivity metrics based only on a binary view of landscapes obscure the important ecological dynamics underlying the distribution and abundance of organisms, whereas metrics including matrix composition provide more information at little additional cost relative to distance-based metrics.