Evapotranspiration and soil water relationships in a range of disturbed and undisturbed ecosystems in the semi-arid Inner Mongolia, China

Evapotranspiration (ET) is a key component of water balance and is closely linked to ecosystem productivity. In arid regions, large proportion of precipitation (PPT) is returned to the atmosphere through ET, with only a small amount available to plants. Our objective was to examine the variability in ET-soil water relationship based on a set of ecosystems that are representative for semi-arid Inner Mongolia and its main land use practices. This study used Eddy covariance (EC) data of water vapor (i.e. ET, mm), PPT (mm), soil volumetric water content (VWC, %), root biomass density and soil properties from three paired sites in semi-arid Inner Mongolia: cropland (Cropland-D) versus undisturbed grassland (Steppe-D), grazed grassland (Grazed Steppe-X) versus fenced grassland (Fenced Steppe-X) and poplar plantation (Poplar-K) versus undisturbed shrubland (Shrubland-K). The paired sites experienced similar climate conditions and were equipped with the same monitoring systems. The ET/PPT ratio was significantly lower at Cropland-D and Grazed Steppe-X in comparison to the undisturbed grasslands, Steppe-D and Fenced Steppe-X. These differences are in part explained by the lower VWC in the upper soil layers associated with compaction of surface soil in heavily grazed and fallow fields. In contrast, the ET/PPT ratio was much higher at the poplar plantation compared to the undisturbed shrubland because poplar roots tap groundwater. The VWC of different soil layers responded differently to rainfall events across the six study sites. Except for Poplar-K, ET was significantly constrained by VWC at the other five sites, although the correlation coefficients varied among soil layers. The relative contribution of soil water to ET correlated with the density of root biomass in the soil (R-2 = 0.67, P < 0.01). The soil water storage in the upper 50 cm of soil contributed 59, 43, 64 and 23% of total water loss as ET at Steppe-D, Cropland-D, Shrubland-K and Poplar-K, respectively. Our across-site analysis indicates that the site level of soil water for ET differs between land use and land cover type due to altered root distribution and/or soil physical properties. As a result, we recommend that ecosystem models designed to predict the response of a wide variety of vegetation to climatic variation in arid regions include more detail in defining soil layers and interactions between evaporation, infiltration and root distribution patterns.