Linking leaf and tree water use with an individual-tree model.

We tested the ability of a model to scale gas exchange from leaf level to whole-tree level by: (1) measuring leaf gas exchange in the canopy of 10 trees in a tall Eucalyptus delegatensis RT Baker forest in NSW, Australia; (2) monitoring sap flow of the same 10 trees during the measurement week; and (3) using an individual-tree-based model (MAESTRA) to link the two sets of measurements. Photosynthesis and stomatal conductance components of the model were parameterized with the leaf gas exchange data, and canopy structure was parameterized with crown heights, dimensions and leaf areas of each of the measurement trees and up to 45 neighboring trees. Transpiration of the measurement trees was predicted by the model and compared with sap flow data. Leaf gas exchange parameters were similar for all 10 trees, with the exception of two smaller trees that had relatively low stomatal conductances. We hypothesize that these trees may have experienced water stress as a result of competition from large neighboring trees. The model performed well, and in most cases, was able to replicate the time course of tree transpiration. Maximum rates of transpiration were higher than measured rates for some trees and lower than measured rates for others, which may have been a result of inaccuracy in estimating tree leaf area. There was a small lag (about 15-30 minutes) between sap flow and modeled transpiration for some trees in the morning, likely associated with use of water stored in stems. The model also captured patterns of variation in sap flow among trees. Overall, the study confirms the ability of models to estimate forest canopy transpiration from leaf-level measurements.