Climate change regulates autumn leaf senescence date (LSD), exhibiting a strong phenological control of plant carbon uptake. Unlike the delaying effect of daily mean temperature (
Europe.
1950–2015.
Plant phenology.
We used partial correlation analysis, multiple linear regression and ridge regression to explore the impacts of DTR on LSD. To quantify the importance of potential drivers of LSD, we trained random forest models and applied the SHapley Additive exPlanations method to isolate the marginal contributions of each predictor on LSD. For LSD modelling and projection, we first evaluated two temperature‐driven LSD models [i.e., cooling‐degree‐day (CDD, without DTR effect) and day–night‐temperature CDD (DNCDD, with DTR effect)], then applied them to predict future LSDs.
We found that observational increases in
Our findings highlight the importance of DTR in controlling LSD variations with an advancing‐dominant effect and call for the improvement of phenology modelling incorporating the DTR effect. Given that DTR showed a globally narrowing trend over the last several decades, more efforts are needed to understand the potential ecological impacts of warming asymmetry and vegetation response to climate change.