Topographic water redistribution, rather thanclimate, dictates eco-hydrological recovery: Flux- and runoff-based validation in the Upper Minjiang River, eastern Qinghai-Tibet Plateau

Study region The upper Minjiang River, a 22,954 km² alpine-gorge headwater basin on the eastern margin of the Qinghai-Tibet Plateau and a critical water conservation area for the Yangtze River. Study focus Large-scale greening is widely reported across mountain headwaters, yet optical indices demonstrate only that a basin has become greener—not that its water cycle has been physically restructured. This study integrates a 39-year (1986–2024) Landsat-derived Remote Sensing Ecological Index (RSEI) series, GeoDetector-based spatial attribution, MODIS evapotranspiration (ET) fluxes, and 33 years of in-situ discharge records to determine what dictates the spatial pattern of ecological recovery and whether that recovery has genuinely enhanced watershed hydrological regulation. New hydrological insights for the region Eco-hydrological quality declined to a historical nadir in 1997 and then recovered rapidly through 2024 (Sen's slope = 0.0027 yr⁻¹, p < 0.01), while no background hydro-climatic variable exhibited a significant trend (all p > 0.05)—indicating that macro-climatic amelioration alone cannot explain the reversal. Elevation dominated the spatial attribution (q = 0.14), roughly sevenfold that of precipitation (q = 0.02), and their pronounced non-linear interaction (DEM∩P, q = 0.194) revealed a “topographic amplifier” whereby rugged terrain intercepts, stratifies, and laterally redistributes an otherwise uniform climatic input, thereby dictating where recovery materializes. Recovery proved hydrologically heterogeneous: arid valleys (< 2000 m) greened most strongly but at the steepest water cost (ET trend = 2.40 mm yr⁻¹); the subalpine belt regenerated most water-efficiently (1.33 mm yr⁻¹); and the nival zone showed rising ET al.ongside declining RSEI, signaling warming-driven rather than restorative wetting. Basin-wide ET rose significantly (trend = 1.458 mm yr⁻¹) and the inter-annual runoff coefficient fell from 0.79 to 0.74, yet the dry-season baseflow contribution increased from 20.16% to 25.02%—two expressions of a single “ecological sponge” effect. Restoration in such basins should therefore be hydrologically constrained rather than area-maximizing.