Quantifying Bufo boreas connectivity in Yellowstone National Park with landscape genetics

TLDR

Ecology seeks to understand how processes limit connectivity and distribution, and spatially quantifying these components can explain why suitable habitats remain unoccupied, though stochasticity and statistical assumptions complicate such estimates. The study applies Random Forests to landscape genetic data to assess how ecological process, scale, and hierarchical organization affect Bufo boreas connectivity in Yellowstone National Park. Random Forests were used to model the influence of ecological components on Bufo boreas connectivity across multiple scales and hierarchical levels. Habitat permeability, topographic morphology, and temperature–moisture regime were significant drivers of Bufo boreas connectivity, with fine‑scale permeability and multi‑scale morphology and climate explaining 56 % of variation, and hierarchical analysis further revealed that within‑cluster connectivity was better explained (74 %) than between‑cluster connectivity (45 %), demonstrating the value of cross‑scale and hierarchical approaches and the general applicability of the developed Random Forest methods.

Abstract

A major objective of ecology is to understand how ecological processes limit population connectivity and species' distributions. By spatially quantifying ecological components driving functional connectivity, we can understand why some locally suitable habitats are unoccupied, resulting in observed discontinuities in distribution. However, estimating connectivity may be difficult due to population stochasticity and violations of assumptions of parametric statistics. To address these issues, we present a novel application of Random Forests to landscape genetic data. We address the effects of three key ecological components on Bufo boreas connectivity in Yellowstone National Park: ecological process, scale, and hierarchical organization. Habitat permeability, topographic morphology, and temperature-moisture regime are all significant ecological processes associated with B. boreas connectivity. Connectivity was influenced by growing-season precipitation, 1988 Yellowstone fires, cover, temperature, impervious surfaces (roads and development), and topographic complexity (56% variation explained). We found that habitat permeability generally operates on fine scales, while topographic morphology and temperature-moisture regime operate across multiple scales, thus demonstrating the importance of cross-scale analysis for ecological interpretation. In a hierarchical analysis, we were able to explain more variation within genetic clusters as identified using Structure (a Bayesian algorithm) (74%; dispersal cover, growing-season precipitation, impervious surfaces) as opposed to between genetic clusters (45%; ridgelines, hot, dry slopes, length of hot season, and annual precipitation). Finally, the analytical methods we developed are powerful and can be applied to any species or system with appropriate landscape and genetic data.