Measurements and simulation of leaf xylem water potential and root water uptake in heterogeneous soil water contents

The relationship between leaf water potential, transpiration rate and soil water potential is complex, particularly when the soil water potential in the root zone is not uniform, which is the rule rather than the exception in soils. Our objectives were: 1) to measure the effect of heterogeneous soil water potentials on the relation between leaf water potential and transpiration rate and 2) to test whether root water uptake models could predict this relation. To this end, we combined the root pressure chamber technique, which allows measuring the suction in the leaves of transpiring plants, with two models of root water uptake, a simple one where soil and roots are presented as resistances in series and a more detailed 3D root architecture model. The experiments were carried out with lupines grown in sandy soil, for which the root architecture and root hydraulic properties had been previously estimated. The soil was partitioned in two layers separated by a coarse sand layer that allowed the roots to grow through but limited the water redistribution between the layers. Three scenarios (wet-wet, dry-wet, dry-dry) were tested. The results showed that the relation between transpiration and leaf water potential was linear in all scenarios. As the upper soil layer severely dried, the conductance of the soil-plant system decreased by ca. 1.65 times compared to the conductance of the plant-soil system in a uniform wet soil. As both layers dried, the conductivity was 8.26 times lower compared to the uniform-wet case. The combination of the experiment and modelling showed that a simple model is capable to reproduce the relation between transpiration, leaf water potential and soil water potential (despite an offset in the leaf water potential). Both simplified and the 3D root architecture models were capable of reproducing the measured changes in hydraulic conductance of the plantsoil system due to the soil drying. However, both models overestimated the measured leaf water potential by 0.1 MPa, probably because of a gradient in osmotic potential between the xylem and the soil. The simulations predicted the occurrence of hydraulic lift, even at day time conditions, although the hydraulic lift was relatively more important at low transpiration rates. The simulation suggested that a root architecture model is needed to estimate the variations of water uptake along the individual roots and this might be crucial to properly model hydraulic lift.