Rhizosphere water content drives hydraulic redistribution: implications of pore-scale heterogeneity to modeling diurnal transpiration in water-limited ecosystems

Trees typically survive prolonged droughts by absorbing water from deeper layers. Where soils are shallow, roots may be extract water from the underlying fractured bedrocks. In dry seasons, surface-soil moisture dynamics reflect hydraulic redistribution (HR). HR is usually estimated based on the gradient of mean, or bulk, soil water potential among layers in the rooting zone (HRB). This approach neglects the potential effect of spatial heterogeneity of water content at the millimeter scale between the rhizosphere and bulk soil. We proposed to account for the rhizosphere water balance, estimating HR to the rhizosphere (HRR) of the dry surface soil from the underlying fractured rock. The model was evaluated using a 15-year dataset collected in Sardinia. When the typical approach, based on moisture gradients among bulk soil layers, was used for estimating HRB, tree transpiration was underpredicted in all seasons, especially in spring and summer. Forcing the model with measured tree transpiration, HRB decreased during spring and summer, while the contribution of the underlying rock layer to tree transpiration was threefold that estimated using HRR-based model. The average water content of the bulk surface soil layer was very low, reaching 0.06 in the driest summers while showing little diurnal dynamics; however, concentrating water in roughly estimated rhizosphere volume, produced rhizosphere water content appreciably higher (≈0.16), and much more dynamic. Predicted HRR dominated evapotranspiration (60% - 65%) in dry springs and summers reaching 80% of tree transpiration. Most importantly, the proposed rhizosphere-HR model correctly predicts the diurnal dynamics of tree transpiration year-round, and the grass transpiration in its active spring period. Eco-hydrological models operating at sub-daily scale should consider partitioning the soil to rhizosphere volume, thus allowing both diagnostic and prognostic estimates of diurnal biosphere-atmosphere mass and energy exchanges.