Response and adaptation of global terrestrial vegetation production to extreme precipitation

Abstract

Extreme precipitation events (EPEs) are expected to become increasingly frequent and severe in the future. As a critical meteorological factor, extreme precipitation (EPre) strongly regulates ecosystem carbon sequestration, highlighting the need to understand its effects on global terrestrial ecosystems. However, quantitative assessments of how ecosystem carbon sequestration adapts to EPre at the global scale and over the long term are still lacking. To address this issue, we define EPEs using threshold-based methods and extreme climate indices. Using observational data from 29 FLUXNET stations, we employ partial correlation analysis and Random Forest model to systematically quantify the response, adaptation, and recovery of gross primary productivity (GPP) to EPEs. The results demonstrate that: (1) A total of 499 EPEs persisting for more than five days comprised 95% of all identified EPEs. The duration of these EPEs ranged from 5 to 28 days in evergreen needleleaf forest, mixed forest, and deciduous broadleaf forest ecosystems, while woody savannas ecosystems exhibited significantly longer durations. (2) EPre significantly reduced ecosystem carbon sequestration. In most ecosystems (excluding evergreen broadleaf forests and woody savannas), the mean partial correlation coefficient between EPre and GPP was less than −0.66 (p < 0.05). Furthermore, EPre accounted for more than 41% of the observed decline in GPP. (3) Most ecosystems (deciduous broadleaf forests, evergreen broadleaf forests, evergreen needleleaf forests, mixed forests and woody savannas) typically recovered their carbon sequestration capacity within 12 days. The duration of EPEs and vapor pressure deficit during the recovery period were identified as the primary drivers influencing the recovery of GPP. Our findings refine the understanding of diverse ecosystem responses to extreme precipitation and offer a framework for enhancing the simulation of such events in dynamic vegetation models.