Climate change stressors are driving ecosystems to shift away from their current distributions throughout Alaska. Such distributional shifts are forecasted to have local as well as global consequences, affecting wildlife habitat, forest carbon sequestration potential, and surface reflectance properties. Many studies focus on remotely sensed data of vegetation classifications and how those classifications map into novel climate space. These studies often lack extensive field datasets that could provide information about how landscape complexities can induce unexpected resilience between woody plant volume and climate. It is critical to understand where these landscape complexities may buffer against climate changes, but it is difficult to assess without long-term time series data. The National Park Service has collected an unprecedented amount of species-specific plot-level data across interior and northern Alaskan national parks with comparable methodologies over the past 20 years. We leveraged this large spatial dataset, using a space-for-time substitution, to determine where woody plant volume shows a typical or atypical relationship to climate. To do this, we used a novel Bayesian clustering model that categorized plots into two clusters, based on the estimated relationship between climate and plant volume.
Results/Conclusions
Our methodology gives us the ability to rely on the species distribution modeling concept while accounting for unforeseen variability caused by the interaction of buffering landscape characteristics and changing climate. To assess our model, we also compared our clustering results with inferred vegetation changes based on aerial photo data collected over the last 50 years. Our Bayesian clustering model was able to account for species interactions and geomorphological characteristics that affect plant volume’s relationship with climate covariates across a broad landscape. We were able to determine which segments of our study area may have landscape properties that buffer against climate changes. Our results illuminate a complex array of factors contributing to shifting species distributions across interior and northern Alaska and highlight sites, currently unoccupied by woody plants, that may be sensitive to change. Aerial photo data confirmed that some sites expected by species distribution models to be sensitive to climate changes did not have atypical relationships, i.e., were not sensitive, because of geomorphological constraints on the landscape. In general, these results have implications for assessing predictions of high latitude vegetation relationships to changing climate. Forecasting models are improved by incorporating information from field data that shows where the relationship between vegetation volume and climate may be atypical because of species-level effects or geomorphological constraints.