Physiological fidelity of a satellite-derived forest resilience indicator in the Amazon

Reliable resilience indicators are urgently needed to monitor global forest health under increasing climate stress. Lag-1 temporal autocorrelation (TAC) of satellite vegetation indices captures forest recovery speed and is widely used as a resilience metric, but its mechanistic underpinnings and methodological reliability have been questioned. Here we use Landsat observations and field data from the Amazon to investigate how satellite-derived TAC relates to in situ measurements of the hydraulic safety margin (HSM), a key physiological trait for plants' capacity to accommodate drier conditions. We find a strong direct correlation between HSM and the resilience proxy (1-TAC) (R² up to 0.57, P = 0.0116), providing evidence that forest recovery speed is constrained by drought resistance. A model incorporating both HSM and cumulative water deficit explains up to 83% of the variance in resilience, indicating that hydraulic strategies and local climate act as a joint, primary axis of forest resilience. Importantly, we show that this relationship is contingent on the selected vegetation index, the temporal frequency of satellite observations and the rolling window size. Taken together, our findings establish the physiological fidelity of TAC as a forest resilience metric and offer methodological guidance.