Seminar: Xin Yu

Droughts can impact terrestrial ecosystems concurrently but also lagged in time as legacy effects. Although drought legacy effects on plants have been thoroughly shown using tree radial growth and greenness, understanding of legacy effects on ecosystem carbon cycles remains limited. Using the global network of long-term eddy-covariance measurements since the late 1990s, we develop a machine-learning-based approach to quantify drought legacy effects on ecosystem carbon fluxes. We find sizeable and widespread drought legacy effects on gross primary productivity (GPP) at 57 out of 59 sites experiencing strong droughts. We find drought legacy effects on GPP diverge globally, with about as many sites experiencing positive (30 sites) and negative (27 sites) legacy effects. Grasslands tend to exhibit positive legacy effects on GPP whereas other ecosystems such as forests show a mixture of positive, negative, or no legacy effects. Climatological aridity and drought duration are the most relevant factors explaining spatial variability of drought legacy effects on GPP among forest sites. Further study is examining the prevalence and magnitude of drought legacy effects on ecosystem respiration, which will be combined with effects on GPP for better understanding and projecting the impacts of drought on terrestrial carbon cycling.