Divergent spatial responses of plant and ecosystem water-use efficiency to climate and vegetation gradients in the Chinese Loess Plateau

Water-use efficiency (WUE) is an important indicator of carbon–water interactions and is defined as the ratio of vegetation productivity to water loss. However, how WUE varies along climate and vegetation gradients at the regional scale remains elusive. In this study, we investigated the spatial patterns of plant-canopy WUE (PWUE, i.e., ratio of gross primary productivity to plant transpiration) and ecosystem WUE (EWUE, i.e., ratio of gross primary productivity to evapotranspiration) in the Chinese Loess Plateau (CLP), which has seen large changes in the biosphere-atmosphere carbon and water cycles due to large-scale revegetation with the CLP. Spatial responses of PWUE and EWUE variations to the mean annual precipitation (MAP), mean annual air temperature (MAT), and normalized difference vegetation index (NDVI) gradients were examined based on remote-sensing and geostatistical model-based datasets. Results showed that mean EWUE estimated from two approaches was 1.26 ± 0.28 and 1.37 ± 0.68 g C kg⁻¹ H₂O, respectively, lower than the mean PWUE (3.16 ± 0.71 g C kg⁻¹ H₂O) across the CLP. EWUE and PWUE estimates showed similar spatial distributions, generally with higher values in the areas with more water available. Precipitation sensitivities of EWUE and PWUE appeared to be positive except the very cold regions, and gradually decreased with increasing MAT in the forest-steppe and forest zones. Spatial variation in EWUE is intrinsically affected by both of PWUE and ecosystem water allocation (i.e., ratio of transpiration to evapotranspiration), and NDVI sensitivity of EWUE is dominant by ecosystem water allocation, leading to postive sensitivity of EWUE to NDVI for most MAP range in the CLP. PWUE variation depends on the geographic patterns of vegetation communities determined by precipitation pattern, leading to relatively lower and stable NDVI sensitivity than EWUE along the MAP gradient in the CLP. Our study revealed the divergent spatial responses of WUE to climate and vegetation gradients at the plant-canopy and ecosystem levels, which could enhance our understanding on the regional-scale carbon-water relationships across multiple organismic levels, and provide essential information for the WUE upscaling and modeling efforts.