Phase‐Asymmetric Thermal Sensitivity Amplifies Respiration Hysteresis in Heatwaves

ABSTRACT Understanding how extreme climate events reshape the temperature response of ecosystem respiration (Re) is critical for predicting carbon–climate feedbacks. Here, we combine observational data and land surface model simulations to investigate Re responses to heatwaves, focusing on temperature sensitivity ( Q 10 ) and thermal hysteresis. We develop a quantitative framework capturing hysteresis via temperature thresholds, their seasonal timing, and phase‐specific Q 10 indices ( Q 10, I , warming phase; Q 10, II , cooling phase). Flux‐tower observations from the 2003 and 2018 heatwaves showed widespread amplification of thermal hysteresis in comparison with years without heatwaves. This amplification was primarily driven by asymmetric shifts in thermal sensitivity, especially a dominant decline in Q 10, II . The response of ecosystem respiration after summer is the most affected in years that have experienced strong heatwaves at most sites. Q 10, II retains strong associations with the timing of optimal temperature ( r = 0.55), whereas Q 10, I becomes largely decoupled. Principal component analysis further confirms this divergence, with the two phase‐specific sensitivities varying along orthogonal axes under heatwaves. We further analyzed individual sites to see how much each Q 10 phase contributed to the overall change in temperature sensitivity. Across sites, the temperature sensitivity of respiration in the post‐heatwave phase accounts for an average of 71% of the total phase‐specific Q 10 change induced by heatwaves. This reinforces the central role of Q 10, II in amplifying thermal hysteresis. Structural equation modeling shows that the heatwave‐amplified thermal hysteresis in Re is fully mediated by Q 10, II decline, shaped by both thermal thresholds and their seasonal timing. These findings reveal a phase‐asymmetric temperature sensitivity of Re that underpins thermal hysteresis, suggesting that short‐term heatwaves could trigger prolonged carbon losses even under post‐peak warming scenarios.