PS 42-146 - The effects of forest microclimate on future species distribution: assessing buffering and decoupling scenarios

Abstract: Human-driven global changes such as land use and climate warming are triggering biodiversity erosion. Among many tools and techniques developed to forecast future species distribution and ranges, Species Distribution Models (SDMs) emerge as the most adopted tool in ecological conservation. Nevertheless, their implementation usually consists in free-air macroclimate variables with coarse spatial and temporal resolution, neglecting the fine-scale habitat conditions experienced by the organisms during their life span. Hence, improving the reliability of these ecological models is crucial for biologists, land managers, and policymakers.

The goal of this study is to quantify predicted changes in the community under future sub-canopy microclimate conditions by comparing two microclimate scenarios and relating these changes to the common free-air modeling approach.

Using Bayesian Additive Regression Tree algorithms, we built dynamic SDMs for different species of birds, plants, insects, and mammals in a temperate rainforest landscape of the Pacific Northwest (HJ Andrew Experimental Forest, Oregon, US). We developed current (2010-2022) and future (2041-2070) predictions for both sub-canopy microclimate and free-air macroclimate conditions and compared the distribution of species. Current microclimate conditions were derived from below-canopy hourly data loggers and free-air macroclimate from a mechanistic downscaling of global reanalysis data at 30 arcsecs (CHELSA). We derived future free-air macroclimate for scenario SSP 5-8.5 using an ensemble of climate models. We implemented microclimate buffering scenario by adding the temperature increase at the free-air level from CHELSA. The decoupling scenario was calculated by applying a correction factor that accounts for the degree of local decoupling between sub-canopy microclimate and free-air macroclimate temperatures. We derived the local decoupling factor from monthly timeseries by computing the slope parameter between microclimate and macroclimate temperatures, being the response and explanatory variables in an ordinary least-square regression.

Current scenarios for birds and insects show a great improvement in accuracy for models trained with microclimate compared to those trained with macroclimate data. Temperature differences in future microclimate scenarios (buffering v. decoupling) reach 2°C at the bottom of the valley and in old growth stands during summer, suggesting a fundamental role of forest cover temperature buffering that will be likely reflected in species distribution shifts.

In light of the improvement of data availability on microclimate conditions at the landscape scale, this study suggests how developing fine-scale distribution predictions is crucial to improve biodiversity conservation plans. Long-term microclimate data and species occurrences allow for testing differences in modeling techniques and frameworks.