Testing early warning metrics for drought-induced tree physiological stress and mortality

Abstract

Climate change-driven drought stress has triggered numerous large-scale tree mortality events in recent decades. Advances in mechanistic understanding and prediction are greatly limited by an inability to detect in situ where trees are likely to die in order to take timely measurements and actions. Thus, algorithms of early warning and detection of drought-induced tree stress and mortality could have major scientific and societal benefits. Here, we leverage two consecutive droughts in the southwestern United States to develop and test a set of early warning metrics. Using Landsat satellite data, we constructed early warning metrics from the first drought event. We then tested these metrics' ability to predict spatial patterns in tree physiological stress and mortality from the second drought. To test the broader applicability of these metrics, we also examined a separate drought in the Amazon rainforest. The early warning metrics successfully explained subsequent tree mortality in the second drought in the southwestern US, as well as mortality in the independent drought in tropical forests. The metrics also strongly correlated with spatial patterns in tree hydraulic stress underlying mortality, which provides a strong link between tree physiological stress and remote sensing during the severe drought and indicates that the loss of hydraulic function during drought likely mediated subsequent mortality. Thus, early warning metrics provide a critical foundation for elucidating the physiological mechanisms underpinning tree mortality in mature forests and guiding management responses to these climate-induced disturbances.