Comparison of forest under-canopy and regional warming over 35 years (1986 - 2021)

Climate warming is expected to outpace species' adaptive capacities. Though forests buffer understory microclimates by moderating temperature and humidity, it is unknown if forest understories will warm at similar rates to open environments under climate warming to affect potential forest refugia and climate resilience. We found that forests decoupled under-canopy temperatures from regional increases in maximum daily temperatures over 35 years, resulting in a 40% reduction in the rate of warming under forest canopies (0.41 ± 0.07 °C decade^-1) relative to open sites (0.70 ± 0.07 °C decade^-1). This decoupling effect delayed extinction of simulated populations by 130-160 years, suggesting that current terrestrial biodiversity predictions may substantially over-estimate temperature-driven extinction risks. Thermal decoupling therefore provides an additional climate-related incentive for forest conservation and restoration.We used 35 years of daily temperature data collected from meteorological stations west of the Cascade Mountain range crest in Oregon, USA, including a globally unique set of closed-canopy stations in the H.J. Andrews Experimental Forest (HJA). We selected 13 closed-canopy and 13 open-canopy stations along an elevational gradient from 75 m to 1609 m that met our standards for surface substrate (i.e., not concrete) and low canopy cover change over the years 1985 - 2021(range < 35%; Fig. S1). Because elevation is known to affect temperatures and warming rates (33), we then paired a subset (n = 18) of the 26 stations based on elevation, constraining open- and closed-canopy pairs to be less than 130 m apart in elevation. To test the sensitivity of our results to our paired station assumptions, we replicated our analysis 1000 times using data constructed by resampling the full 26 stations, randomly holding out eight stations (four open- and four closed-canopy) in each iteration.