HighlightHigh-frequency deficit irrigation to maintain sustained low soil moisture results in wet upper soil layers which increase bulk root water potential and diminish ABA levels.

Abstract. Transpiration is a crucial component in the hydrological cycle and a key parameter in many disciplines like agriculture, forestry, ecology and hydrology. Sap flow measurements are one of the most widely used approaches to estimate whole-plant transpiration in woody species; this is due to their applicability in different environments and in a variety of species as well as the fact that continuous high temporal resolution measurements of this parameter are possible. Several techniques have been developed and tested under different climatic conditions and using different wood properties. However, the scientific literature also identifies considerable sources of error when using sap flow measurements that need to be accounted for, including probe misalignment, wounding, thermal diffusivity and stem water content. This study aims to explore probe misalignment as a function of time in order to improve measurements during long-term field campaigns (>3 months). The heat ratio method (HRM) was chosen because it can assess low and reverse flows. Sensors were installed in four Pinus halepensis trees for 20 months. The pines were located in a coastal valley in south-eastern Spain (39∘57′45′′ N 1∘8′31′′ W) that is characterised by a Mediterranean climate. We conclude that even small geometrical misalignments in the probe placement can create a significant error in sap flow estimations. Additionally, we propose that new statistical information should be recorded during the measurement period which can subsequently be used as a quality control of the sensor output. The relative standard deviation and slope against time of the averaged v1v2 were used as quality indicators. We conclude that no general time limit can be set regarding the longevity of the sensors, and this threshold should rather be determined from individual performance over time.

No abstract