Legacy effects of drought on gross primary productivity from eddy-covariance measurements

The frequency and severity of droughts are expected to increase in the wake of climate change. Drought events not only cause direct impacts on the ecosystem carbon balance but also result in legacy effects during the following years. These legacies result from, for example, drought damage to the xylem or the crown which causes impaired growth, or from higher vulnerability to pests and diseases. To understand how droughts might affect the carbon cycle in the future, it is important to consider both direct and legacy effects. Such effects likely affect interannual variability in C fluxes but are challenging to detect in observations, and poorly represented in models. Therefore, the patterns and mechanisms inducing the legacy effects of drought on ecosystem carbon balance are necessarily needed to improve.

In this study, we analyze gross primary productivity (GPP) from eddy-covariance measurements in Germany to detect legacy effects from recent droughts. We follow a data-driven modeling approach using a random forest model trained in different sets of drought and non-drought periods. This approach allows quantifying legacy effects as deviations of observed GPP from modeled GPP in legacy years, which indicates a change in the vegetation response to hydro-climatic conditions as compared with the training period.