Quantifying water use resilience from sap flow data to better understand post-drought effects on tree functioning

Drought impacts on vegetation function have been widely assessed globally but our understanding of the global patterns of drought recovery and its mechanistic underpinnings is comparatively less understood. The quantification of vegetation resilience to drought has been mostly based on the analysis of time series of remotely-sensed vegetation indices, tree-ring data or ecosystem-level fluxes. While useful, these approaches have not provided a mechanistic link between resilience patterns and post-drought effects on plant hydraulics because they lack sufficient temporal and spatial resolution. Resilience quantified from tree-level sap flow can provide these mechanistic insights on post-drought effects but its estimation using classical resilience metrics requires defining a reference sap flow, unaffected by drought. Here, we compare two different approaches to estimate tree water use resilience to soil drought for >500 drought events using global, tree-level sap flow data in the SAPFLUXNET database (Poyatos et al. 2021). For both approaches, soil droughts were defined using the same criteria based on soil relative extractable water (REW), ensuring a minimum intensity and duration (10 days) and the presence of well-defined pre- and post-drought periods with REW values sustainedly above the threshold. In the first approach, we apply classical resilience metrics obtained from the comparison of pre- and post-drought sap flow. We show that water use resilience is related to soil drought characteristics such as intensity and duration and also to atmospheric vapour pressure deficit. In the second approach, we present a model-based resilience framework in which actual sap flow during and after a drought is compared against a reference sap flow modelled using a random forest regression with hydrometeorological drivers, but excluding potential drought legacy effects. This latter approach is tested in two Mediterranean forests in SAPFLUXNET where additional data on tree water status (leaf water potentials or tree water deficit derived from automatic dendrometry) are also available. We show that the model-based resilience framework applied to sap flow data is a promising avenue to better understand the global patterns of drought recovery and the underlying hydraulic mechanisms.

Poyatos, R. et al.: Global transpiration data from sap flow measurements: the SAPFLUXNET database, Earth System Science Data, 13, 2607–2649, https://doi.org/10.5194/essd-13-2607-2021, 2021.