Uncertainty in sap flow-based transpiration due to xylem properties

Abstract:

Transpiration, the evaporative loss of water from plants through their stomata, is a key component of the terrestrial water balance, influencing streamflow as well as regional convective systems. From a plant physiological perspective, transpiration is both a means of avoiding destructive leaf temperatures through evaporative cooling and a consequence of water loss through stomatal uptake of carbon dioxide. Despite its hydrologic and ecological significance, transpiration remains a notoriously challenging process to measure in heterogeneous landscapes. Sap flow methods, which estimate transpiration by tracking the velocity of a heat pulse emitted into the tree sap stream, have proven effective for relating transpiration dynamics to climatic variables. To scale sap flow-based transpiration from the measured domain (often <5 cm of tree cross-sectional area) to the whole-tree level, researchers generally assume constancy of scale factors (e.g., wood thermal diffusivity (k), radial and azimuthal distributions of sap velocity, and conducting sapwood area (A_s)) through time, across space, and within species. For the widely used heat-ratio sap flow method (HRM), we assessed the sensitivity of transpiration estimates to uncertainty in k (a function of wood moisture content and density) and A_s. A sensitivity analysis informed by distributions of wood moisture content, wood density and A_s sampled across a gradient of water availability indicates that uncertainty in these variables can impart substantial error when scaling sap flow measurements to the whole tree. For species with variable wood properties, the application of the HRM assuming a spatially constant k or A_s may systematically over- or underestimate whole-tree transpiration rates, resulting in compounded error in ecosystem-scale estimates of transpiration.