In the context of global warming, this study examines the spatiotemporal evolution of ecosystem services (ES) in vegetation-dominated landscapes of the Loess Plateau and the responses of these services to climate change and human activities. Ecological parameter models, random forest models, and spatiotemporal interaction analyses were applied to quantify ES trends across three vegetation-dominated landscape configurations over the past 30 years and under future climate change scenarios. Spatial regression models identified key drivers. Results showed that Forest-Dominated Landscape Configurations (FDLC) outperform others in Sediment Delivery Ratio, Habitat Quality, and Carbon Storage, whereas Cropland-Dominated Landscape Configurations (CDLC) perform better in Seasonal Water Yield and Nutrient Delivery Ratio. FDLC and CDLC were more sensitive to driving factors than Grassland-Dominated Landscape Configurations (GDLC), with the spatial lag model providing the best fit. By 2030, FDLC demonstrates a higher capacity for addressing climate change. These results clarify configuration-specific ES functions and sensitivities and support targeted landscape management strategies by prioritizing FDLC for climate mitigation and habitat quality while optimizing CDLC and GDLC to balance water supply and nutrient retention.