Evapotranspiration partitioning in dryland ecosystems: a global meta-analysis of in situ studies

Abstract In drylands, evapotranspiration ( ET ) is the dominant ecohydrologic process. For this reason, partitioning of ET —determining the relative importance of interception (I), soil evaporation ( E ), and plant transpiration ( T )—is critical, but remains a challenge. Recently, however, advances in measurement technologies and data availability have triggered an upsurge in in situ studies focused on quantifying T/ET . We carried out a meta-analysis of 38 datasets drawn from 31 studies done in drylands worldwide. This analysis showed that over the growing season, E and T are roughly equivalent for most natural ecosystems, whereas T/ET is higher in irrigated agro-ecosystems. With respect to factors controlling variations in T/ET , we found (1) no clear correlation for annual precipitation, soil texture, or ecosystem type; (2) leaf area index is a more significant controlling factor than fractional cover; and (3) T/ET varies most during dynamic wetting–drying episodes. We also found that controlling factors are different for E and T . Because these two processes differ in temporal dynamics, the factors controlling ET partitioning vary with temporal scale. Further, when interception and shallow groundwater are substantial, including these factors is essential for accurate T/ET quantification. The isotopic approach, especially using laser spectroscopy, is now indispensable for such studies. However, issues related to sampling protocols and quality assurance still must be resolved. We propose three promising areas for future studies in drylands: (1) isotopic sampling of vadose-zone water vapor using laser spectroscopy; (2) improved definition/identification of diffusive pathways; and (3) robust upscaling from incongruent hydrometric and isotopic measurements.