PNNL

Abstract

Forest ecosystems depend on their ability to withstand and recover from natural disturbances. Forest resilience is expected to decline with climate warming, but evidence is scant. Here we integrate satellite-based estimates of vegetation indices with machine learning to examine forest resilience from 2000-2020. All biomes show a decline in resistance to disturbance except boreal ecosystems, which exhibit a small increase in resilience. These patterns emerge in both managed and unmanaged forests, pointing to a common climatic driver. About 23% of global forests have already reached a tipping point beyond which resilience is extremely low. Together these results reveal a widespread decline in the capacity of forests to withstand disturbances.