Uncovering the critical soil moisture thresholds of plant water stress for European ecosystems

Understanding the critical soil moisture (SM) threshold (θcrit) of plant water stress and land surface energy partitioning is a basis to evaluate drought impacts and improve models for predicting future ecosystem condition and climate. Quantifying the θcrit across biomes and climates is challenging because observations of surface energy fluxes and SM remain sparse. Here, we used the latest database of eddy covariance measurements to estimate θcrit across Europe by evaluating evaporative fraction (EF)-SM relationships and investigating the covariance between vapor pressure deficit (VPD) and gross primary production (GPP) during SM dry-down periods. We found that the θcrit and soil matric potential threshold in Europe are 16.5% and −0.7 MPa, respectively. Surface energy partitioning characteristics varied among different vegetation types; EF in savannas had the highest sensitivities to SM in water-limited stage, and the lowest in forests. The sign of the covariance between daily VPD and GPP consistently changed from positive to negative during dry-down across all sites when EF shifted from relatively high to low values. This sign of the covariance changed after longer period of SM decline in forests than in grasslands and savannas. Estimated θcrit from the VPD–GPP covariance method match well with the EF–SM method, showing this covariance method can be used to detect the θcrit. We further found that soil texture dominates the spatial variability of θcrit while shortwave radiation and VPD are the major drivers in determining the spatial pattern of EF sensitivities. Our results highlight for the first time that the sign change of the covariance between daily VPD and GPP can be used as an indicator of how ecosystems transition from energy to SM limitation. We also characterized the corresponding θcrit and its drivers across diverse ecosystems in Europe, an essential variable to improve the representation of water stress in land surface models.
Global Change Biology, 28 (6)
ISSN:1354-1013
ISSN:1365-2486