Sensitivity of regional evapotranspiration partitioning to variation in woody plant cover: Insights from experimental dryland tree mosaics

Aim: Movement of water from the land surface to the atmosphere (evapotranspiration, ET) is the dominant output flux in the global terrestrial surface water budget. The partitioning of ET between soil evaporation (E) and plant transpiration (T) couples important ecological, hydrological and atmospheric processes. ET partitioning has been hypothesized to vary as a function of woody plant cover, yet a relationship between ET partitioning and woody cover has not been quantified empirically. Land surface models assume unit increase in T per unit increase in vegetation cover (woody cover), following a proportional linear relationship. Recent assessments have questioned the validity of this assumption for heterogeneous canopies, but we lack experimental data across an explicitly defined gradient of woody cover to characterize this relationship. Location: North American monsoon region. Methods: In a controlled dryland environment experimental facility, we manipulated woody cover and documented the response of ET and its component fluxes. We incorporated the resulting functions into a widely used coupled land-atmosphere model (WRF-Noah) to document the implications of modifying specific model parameters that assume (1:1) proportionality. Results: As total ET increased with woody cover, T/ET deviated below 1:1 proportionality. Using our experimentally determined relationship for ET partitioning and woody cover in the model, we observed reductions in ET of as much as 40% during the monsoon season and annual increases of almost 200% in regional E. Main conclusions: Our results highlight a limitation of modelled ET that affects regional to global patterns of water flux, with implications for a number of earth surface processes. A better understanding of how changing woody cover influences patch-scale ecohydrological processes is needed, particularly under current changes in woody cover associated with deforestation, afforestation and drought-induced mortality. More specifically, improved representation of E and T fluxes will improve understanding and modelling of large-scale ecological, hydrological and atmospheric processes.