Xylem hydraulic properties in subtropical coniferous trees influence radial patterns of sap flow: implications for whole tree transpiration estimates using sap flow sensors

A high spatial resolution dataset of sap flux density in subtropical conifers is used to assess the minimum number and location of sap flow sensors required to monitor tree transpiration accurately. Tree transpiration is commonly estimated by methods based on in situ sap flux density (SFD) measurements, where the upscaling of SFD from point measurements to the individual tree has been identified as the main source of error. The literature indicates that the variation in SFD with radial position across a tree stem section can exhibit a wide range of patterns. Adequate capture of the SFD profile may require a large number of point measurements, which is likely to be prohibited. Thus, it is of value to develop protocols, which rationalize the number of point measurements, while retaining a satisfactory precision in the tree SFD estimates. This study investigates cross-sectional SFD variability within a tree and successively for six individual trees within a stand of Pinus elliottii var. elliottii x caribaea var. hondurensis (PEE x PCH). The stand is part of a plantation in subtropical coastal Australia. SFD is estimated using the Heat Field Deformation method simultaneously for four cardinal directions with measurements at six depths from the cambium. This yields a reference value of single tree SFD based on the twenty-four point measurements. Large variability of SFD is observed with measurement depth, cardinal direction and selected tree. We suggest that this is linked to the occurrence of successive narrow early and latewood rings with contrasting-specific hydraulic conductivities and wood water contents. Thus, an accurate placement of sensors within each ring is difficult to achieve in the field with the sensor footprint covering several rings of both early and latewood. Based on the reference dataset, we identified both an "ideal" setup and an "optimal" setup in terms of cost effectiveness and accuracy. Our study shows the need of using a systematic protocol to optimize the number of sensors to be used as a trade-off between precision and cost. It includes a preliminary assessment of the SFD variability at a high spatial resolution, and only then based on this, an appropriate placement of sensors for the long-term monitoring.