Searching for Refuge: A Framework for Identifying Site Factors Conferring Resistance to Climate‐driven Vegetation Change

Aim Climate change is occurring at accelerated rates in high latitude regions such as Alaska, causing alterations in woody plant growth and associated ecosystem patterns and processes. Our aim is to assess the magnitude and speed that climate-induced changes in woody plant distribution and volume may be reduced and/or slowed by relatively static landscape features like physical characteristics (e.g. depth to gravel, mineral cover percent and slope degree) and/or edaphic properties (e.g. soil organic matter, soil pH and site wetness rating) that resist climate-vegetation responses. Location We leveraged a large field data set collected across a network of Alaskan national parks, which allows for comprehensive spatial data analysis over a uniquely large spatial extent. Methods To learn about the conditions that may either impede or accelerate vegetation changes in northern Alaska, we used a Bayesian hierarchical model to identify which landscape features may decelerate change or offer refuge for plant species. Our model quantifies the contribution of fast ('dynamic') versus slow ('static') changing variables to predict plant volume and categorize landscape types into either robust or nonrobust to climate changes. Results We found that two landscape features, low soil wetness and low soil organic matter comprising 63.1% of sites in the data set, were the most likely landscape features to inhibit vegetation expansion. We also found that fewer numbers of sites have the potential to offer refuge to existing plant species (5.43% on average) because few sites had high soil wetness as a landscape feature. Main conclusions Our analyses highlight the importance of incorporating static covariates representing landscape resistance to vegetation change for improving realism in forecasts of vegetation change in Alaska.