Restoration of degraded drylands in northwestern China commonly relies on planting drought-resistant shrubs, yet the biophysical mechanisms by which such plantings alter land-atmosphere energy exchange remain unclear. Here, we explicitly couple the energy and water cycles by decomposing energy consumption into transpirationand evaporation-driven fractions (EFt and EFe) and by quantifying the controls on each. We do so using eddycovariance observations (2019-2021) over a shrub plantation in the desert steppe of Yanchi County, Ningxia, combined with in-situ and remote-sensing drivers and an improved Shuttleworth-Wallace model that partitions evapotranspiration (ET) into evaporation (E) and transpiration (T). We quantified net radiation (Rn), sensible heat (H), latent heat (LE), and soil heat flux (G), and linked them to canopy conductance (Gc), soil moisture (SM), vapor-pressure deficit (VPD), and albedo. Annually, H dominated the energy budget (58.3% of Rn), whereas LE dominated during the growing season. Transpiration accounted for 71.8% of ET, indicating strong vegetation control of water loss. Gc was highly sensitive to SM and VPD and governed LE partitioning via T/ET, rendering EFt comparatively stable while EFe varied with moisture availability and atmospheric demand. A structuralequation model explained 67% of the variability in evaporative fraction and identified SM as the principal positive driver, with additional pathways through Gc and albedo. Together, these results show how water availability and canopy structure jointly regulate energy partitioning in planted shrubs, providing actionable guidance for restoration and management in drylands.