Effect of natural and human factors on hydrological connectivity in the arid region: Application of water-ecological network and XGBoost model

Study area: Altay Prefecture, Xinjiang, China Research focus: This study constructed a node–corridor water–ecological network that integrated river networks and wetlands to elucidate its structural and functional connectivity. An XGBoost model was employed to investigate the mechanisms by which natural and anthropogenic factors influenced hydrological connectivity. New hydrological insights for the region: High structural connectivity occurred south of Beitun City (>0.48). Functional connectivity was strongest in snowmelt-recharged mountains (0.79–1.0), and the eastern Ulungu Lake agricultural area exhibited the highest comprehensive index (0.56–0.70), in contrast to the extremely low values in the Gobi Desert. NDVI significantly promoted connectivity (r = 0.727), while elevation (r = –0.514), temperature (r = –0.564), and evapotranspiration (r = –0.266) inhibited it (all p < 0.001). The nonlinear enhancement of connectivity by human factors exhibited a critical threshold, and when GDP density exceeded 400,000 yuan/km² or population density surpassed 10 people/km², the marginal effect of ecological connectivity was reversed by high-intensity human activities. This study innovatively quantified arid-zone water transport loss using cost paths within a constructed water–ecological network, providing a new perspective and quantitative tool for understanding regional hydrology.