Global amphibian climate refugia for amphibians

Methods

MaxEnt species distribution models for 4,035 species projected under four SSPs across four time periods (2021-2100) using CMIP6 models MIROC6 and CNRM-CM6-1; refugia calculated as the mean of current and all future projected suitabilities, thresholded at the maximum training sensitivity plus specificity value; results extracted to geographic regions, Udvardy biomes, and the global protected area network; minimum protected area subset identified by optimization.

Results: Refugia area was reduced relative to current suitable area for 2,864-2,959 species dependent on SSP; 136-139 species face complete refugia loss. Caudata and Gymnophiona showed the greatest declines (-29.8 to -37.5%). Only 539 protected areas cover projected refugia for nearly all species.

Patterns of refugia loss varied significantly across Udvardy biomes and geographic regions, with tropical and island-locked families disproportionately affected. Main conclusions: A targeted subset of the existing protected area network is sufficient to conserve projected refugia for most modeled amphibian species, though climate change will eliminate refugia entirely for 136-139 species irrespective of protection. Species losing all refugia are disproportionately island, high-elevation, and range-restricted endemics with limited dispersal capability, for whom no PA-based solution exists under either climate model.

Biogeographic position, dispersal limitation, and life-history traits predict family-level vulnerability. These findings provide a spatially explicit global framework for amphibian refugia conservation.