Disentangling temporal and population variability in plant root water uptake from stable isotopic analysis: a labeling study

Abstract. Isotopic labeling techniques have the potential to minimize the uncertainty of plant root water uptake (RWU) profiles estimated through multi-source (statistical) modeling, by artificially enhancing soil water isotopic gradient. Furthermore, physical models can account for hydrodynamic constraints to RWU if simultaneous soil and plant water status data is available. In this study, a population of tall fescue (Festuca arundinacae cv Soni) was grown in a macro-rhizotron setup under semi-controlled conditions to monitor such variables for a 34-hours long period following the oxygen stable isotopic (18O) labeling of deep soil water. Aboveground variables included tiller and leaf water oxygen isotopic compositions as well as leaf water potential (ψleaf), relative humidity, and transpiration rate. Belowground profiles of root length density (RLD), soil water content and isotopic composition were also sampled. While there were strong correlations between hydraulic variables as well as between isotopic variables, the experimental results underlined the discrepancy between variations of hydraulic and isotopic variables. In order to dissect the problem, we reproduced both types of observations with a one-dimensional physical model of water flow in the soil-plant domain, for 60 different realistic RLD profiles. While simulated ψleaf followed clear temporal variations with little differences across plants as if they were “on board of the same rollercoaster”, simulated δtiller values within the plant population were rather heterogeneous (“swarm-like”) with relatively little temporal variation and a strong sensitivity to rooting depth. The physical model thus suggested that the discrepancy between isotopic and hydraulic observations was logical, as the variability captured by the former was spatial and may not correlate with the temporal dynamics of the latter. For comparison purposes a Bayesian statistical model was also used to simulate RWU. While they predicted relatively similar cumulative RWU profiles, the physical model could differentiate spatial from temporal dynamics of the isotopic signature, and supported that the local increase of soil water content and formation of a peak of labelled water observed overnight were due to hydraulic lift.