Quantifying isolation-by-resistance and connectivity in dendritic ecological networks

A central theme in landscape ecology is the translation of individual movements within a population by deconstructing/interpreting the components of its topographical environment. Most such endeavors rely heavily on the concept of landscape resistance - a composite of an arbitrary number of features/covariates that, when identified/compiled, yield a surface inversely related to net movement. However, the statistical methodologies underlying this compilation have limited applicability when applied to dendritic ecological networks (DENs), including riverscapes. Herein we provide an analytical framework (ResistNet) that more appropriately annotates DEN segments by first aligning individual genetic distances with environmental covariates within a graph structure, then employing a genetic algorithm to optimise a composite model. We evaluated the efficacy of our method by first testing it in silico across an array of sampling designs, spatial trajectories, and levels of complexity, then applying it in an empirical case study involving 13,218 ddRAD loci from N=762 Speckled Dace (Leuciscidae: Rhinichthys osculus), sampled across N=78 Colorado River localities. By doing so, we underscored the utility of ResistNet within a large-scale conservation study, as well as identified prerequisites for its appropriate application. Our contemporary framework not only allows an interpretation of meta-population/meta-community structure across DENs, but also highlights several innovative applications. These are: (a) Expanding an ongoing study design, and thus its hypotheses, into yet unsampled temporal and/or spatial arenas, and; (b) Promoting multi-species management through comparative analyses that extend across species and/or drainages.