Reduced resilience as an early warning signal of forest mortality

Abstract

Climate-induced forest mortality is being widely observed across the globe. Predicting forest mortality remains challenging because the physiological mechanisms causing mortality are not fully understood and empirical relations between climatology and mortality are subject to change. Here, we show that the temporal loss of resilience, a phenomenon often detected as a system approaches a tipping point, can be used as an early warning signal (EWS) to predict the likelihood of forest mortality directly from remotely sensed vegetation dynamics. We tested the proposed approach on data from Californian forests and found that the EWS can often be detected before reduced greenness, between 6 to 19 months before mortality. The EWS shows a species-specific relation with mortality, and is able to capture its spatio-temporal variations. These findings highlight the potential for such an EWS to predict forest mortality in the near-term. Predicting mortality in forests is challenging because its underlying causes are spatially varied and not well known. Reduced resilience detected from remotely sensed time series of vegetation dynamics can serve as an effective early warning signal to indicate the potential for forest mortality.