Remotely-sensed reduced ecosystem resilience reveals large variations of drought-mortality relations driven by fine-scale ecosystem heterogeneity in tropical dry forests

Quantifying the determinants of forest mortality responses to drought is critical for assessing forest vulnerability to climate change, particularly in carbon-dense tropical forests. Despite enhanced ecophysiological understanding in drought-driven tree mortality, inference of the mortality sensitivity to drought at ecosystem scale and its variation has, however, been challenging largely because mortality observations are sparse in both space and time. Drawing on the theory of critical transition in complex systems, we integrated in situ observations and ecosystem resilience derived by optical remote sensing time series to generate 30m resolution annual biomass mortality maps that cover the 2015 ENSO drought for tropical dry forests in Costa Rica. Results revealed substantial fine-scale heterogeneity in forest mortality under extreme drought. Subsequent spatial analysis showed that deciduousness and elevation were major determinants of the spatial heterogeneity in mortality sensitivity to drought. The regression slopes of forest mortality to water stress, i.e. drought sensitivity, were higher for evergreen and low elevation forest patches. Fine-scale heterogeneity in ecosystem composition and micro-environment, especially those associated with plant water use, strongly regulate ecosystem resistance to drought in tropical forests. Our analyses highlight the potential of high-resolution remote sensing to advance quantitative understanding in forest mortality.