Water-Limited Evapotranspiration in Two Contrasting and Heterogeneous Mediterranean Ecosystems of Sardinia

Two micrometeorological towers were installed at two contrasting sites characterized by different precipitation regimes. The first tower is located in Orroli, an area with a mean annual precipitation of approximately 600 mm, selected as a case study within the AQUEDUCT European project. This site is characterized by a heterogeneous landscape composed of wild olive trees interspersed with C3 herbaceous vegetation. Vegetation develops on shallow soils overlying a partially fractured basaltic bedrock, with soil depths ranging between 15 and 40 cm. Tree cover accounts for about 25% of the tower footprint.

The second tower is located in a mountainous forested area dominated by Quercus ilex, characterized by steeper slopes, frequent rocky outcrops, and higher annual precipitation, averaging about 800 mm. In this site, tree cover represents approximately 67% of the tower footprint.

At both locations, land surface energy fluxes and CO₂ exchanges were measured using the eddy covariance technique. Soil moisture was monitored using water content reflectometers, while leaf area index (LAI) was periodically estimated to capture vegetation dynamics. In addition, the tree transpiration component was quantified using sap flow sensors, allowing the separation of vegetation contributions to evapotranspiration.

Results indicate that the Orroli site is strongly influenced by rainfall seasonality. Vegetation species at this site rely on water stored within the fractured rocky substrate to maintain physiological activity during dry periods. Pronounced seasonal patterns were observed in both CO₂ fluxes and evapotranspiration (ET), with higher values during periods when both herbaceous and woody vegetation are active, and lower values following the senescence of the grass component.

In contrast, the Marganai forest site exhibited relatively stable ET rates throughout the year, highlighting the high efficiency of tree species in accessing deep water reserves. ET of the site is similar at the Orroli site during periods of active grass growth, latent heat fluxes became greater in the Marganai forest once the herbaceous layer senesced. The relationship between ET and potential ET versus soil moisture suggest that Quercus ilex in Marganai appears largely independent of surface soil moisture, emphasize the contribution of the rock water reservoir.  This contribution is also present at the Orroli site, and the water balance analysis shows that it plays a key role in sustaining grass vegetation during the late spring period. Overall, the results suggest the existence of a rainfall threshold of approximately 700 mm per year, below which precipitation becomes a limiting factor for tree cover development.