Determination of zero-flow for the thermal dissipation method of sap flow measurements in Mediterranean climates

Sap flow measurements are essential for estimating whole plant- and ecosystem water use. Yet, several challenges and pitfalls exist in the application and analysis of sap flow data, such as the determination of zero-flow (ΔT_max). Here, we used different methods to estimate ΔT_max and their effect on the determination of sap flow of Quercus suber L. trees measured with the thermal dissipation method under varying environmental conditions. For the estimation of ΔT_max, we used data from two different sap flow sensor types: UP (SFS2, Type M-M12, UP GmbH, Ibbenbüren, Germany) and Dynamax (TDP30, Dynamax, Texas, USA). We determined ΔT_max daily at i) pre-dawn, ii) using maximum and average 3- and 11-day moving windows, iii) linear and iv) double regression as well as v) based on vapor pressure deficit thresholds (VPD <0.1 kPa/<0.25 kPa) approaches. Data was analyzed for several seasonal conditions: moist pre-drought, dry-down, drought peak, and recovery. Overall, the double regression yielded the highest sap flow estimates independent of the season. The higher differences were observed between the methods during pre-drought and recovery. The moving window and regression methods led to a sensor-specific variability in sap flow: with highest values for UP during recovery and for Dynamax during pre-drought. During dry-down and drought peak, all methods yielded comparable results. The large discrepancies during periods of pre-drought and recovery are most likely due to sapwood baseline shifts of trees and the lack of nights with low VPD in Mediterranean summers. Application of short moving windows reduced the variability of sap flow estimates and between sensor types. Hence, we recommend the use of a short average moving window as the most robust method to determine ΔT_max in Mediterranean ecosystems, where environmental conditions are highly variable.