Satellite observations reveal ecosystem resistance and resilience to short-term water stress driven by dominant vegetation along a rainfall gradient in Australia

Climate change is projected to intensify water stress in many ecosystems and poses threats to their stability, which can be quantified through ecosystem resistance and resilience. Relevant studies mostly focused on multi-year or annual droughts, and in spatially homogeneous or species-specific ecosystems. However, resilience and resistance within complex ecosystems, where different plants exhibit different adaptations and recovery behaviours, are less understood. Using productivity data from satellite-derived GOSIF (Global Orbiting Carbon Observatory-2 Solar-Induced Fluorescence) and flux towers, we examined vegetation responses to short-term (<1 year) water stress events from 2000 to 2018 along the North Australia Tropical Transect, which spans a 1600 mm rainfall gradient and transitions from seasonal mesic to non-seasonal arid ecosystems. We define resistance as productivity maintained during stress relative to a multi-year average baseline, and resilience as the extent to which productivity recovered one year after stress relative to the same baseline. Our results show that ecosystem resistance to water stress was lowest in semi-arid regions but higher in both arid and mesic regions, while ecosystem resilience showed the opposite pattern. These spatial patterns occurred regardless of seasonality and were mainly associated with dominant vegetation type. Woody savanna-dominated mesic regions exhibited highest resistance (0.82 ± 0.13, p < 0.001) and lowest resilience (0.26 ± 0.19, p < 0.001), shrublands in arid areas had intermediate values of both resistance (0.81 ± 0.14, p < 0.001) and resilience (0.27 ± 0.22, p < 0.001), while the grasslands in semi-arid regions had low resistance (0.78 ± 0.15, p < 0.001) and high resilience (0.38 ± 0.24, p < 0.001). The highest likelihood (>75.0 %) of full recovery (i.e., exceeding baseline after one year) occurred during the wet season in mesic regions, likely due to energy limitation, while arid regions showed a lower likelihood (57.0 %). This study provides a remote sensing framework for quantifying ecosystem resistance and resilience under water stress.