Evapotranspiration is by far the dominant component of the water budget in drylands. After accounting for intercepted water, evapotranspiration has two major components: evaporation from soil and transpiration from plants. Understanding the partitioning of evapotranspiration into these components is critical because it affects many ecological, hydrological, and biogeochemical processes. However, the partitioning of evapotranspiration between soil evaporation and plant transpiration remains highly uncertain. This partitioning has been hypothesized to present non-linear, threshold type, behavior in response to changes in canopy cover, but fundamental relationships between amount of tree cover and the partitioning of evapotranspiration between soil evaporation and transpiration are lacking. In this study we quantify the partitioning on evapotranspiration into its components for a gradient of increasing canopy. Our objectives included developing a systematic approach based on field observations to quantify the variation in key properties associated with evapotranspiration fluxes as a function of canopy cover and to quantify its partitioning between soil evaporation and plant transpiration as a function of woody-plant canopy cover. We simulated different levels of canopy cover by establishing regular 10 x 10 arrangements of containers with either bare soil or a mesquite tree (Prosopis chilensis) approximately 2.5 m tall. A pulse of moisture was added to each container and water loss from individual containers was monitored. Total evapotranspiration was calculated as the integration of water loses from individual containers. Containers with bare soil were sources of evaporation, while containers with trees were sources of evapotranspiration, discriminated by the installation of sap flow sensors in the trees.
Our results quantify how evapotranspiration varies in response to changes in vegetation cover. In particular, they illustrate how changes in the amount of mesquite-dominated cover result in a quasi-linear response in the ratio of either evaporation or transpiration to total evapotranspiration. These responses suggest that the partitioning of evapotranspiration is likely to be even more non-linear for thicker, denser vegetation structures, such as the evergreen conifer-dominated systems. Our experiments, which used the unique logistical capabilities of the Biosphere 2 glasshouse facility, can have important implications for assessing ecosystem responses to current and predicted changes in ecological and hydrological dynamics of arid and semiarid ecosystems and highlight systematic and interactive effects of canopy cover and along gradients of vegetation cover in hydrological variables that influence ecosystem processes and properties.