Ecosystem classification is fundamental for its identification of spatiotemporal changes and type-oriented management. Based on the dominant service function and human intervention intensity, this study developed an ecosystem classification system comprising 9 first-level and 25 second-level categories, identified their spatial distribution, and further analyzed the spatiotemporal changes between 2000 and 2020, by integrating multi-source data of land use, climate, topography, vegetation, soil, and settlement distribution. The results showed that urban/township ecosystem in China expanded 1.1 times, with 64.51% of the increased area converted from the cultivated land ecosystem. Cultivated land ecosystem was shrunk by 0.88×10⁴ km², its second-level ecosystems of paddyland and dry farmland decreased by 0.60×10⁴ km² and 2.09×10⁴ km², respectively, and oasis farmland expanded by 1.81×10⁴ km². Due to implementation of the "Grain for Green" project, mixed farmland/pastureland and farmland/forestland ecosystems decreased by 2.88×10⁴ km² and 0.92×10⁴ km², respectively, while the forestland ecosystem increased by 1.61×10⁴ km². The water/wetland ecosystem increased by 0.31×10⁴ km², of which 70% was from the expansion of marshland ecosystem. With the climate becoming warmer and wetter, the arid desert and frozen land ecosystems were shrunk by 14.98×10⁴ km² and 0.92×10⁴ km², respectively, while the pastureland ecosystem was expanded by 9.97×10⁴ km². Associated with these changes, the landscape connectivity declined while the fragmentation and diversity, as well as the resistance of ecological processes increased. The ecological status in China improved significantly from 2000 to 2020, with 92.06% of the ecosystems showing an increase in the normalized difference vegetation index (NDVI) at the mean increase rate of 0.74%/a. Of the identified ecosystems, the mixed farmland/pastureland and farmland/forestland ecosystems showed the most evident increase in the NDVI, increased by 1.26%/a and 0.85%/a, respectively. The classification and mapping results highlight the differences in ecosystem structure, ecological environment risks and productivity, providing scientific support for macro-scale ecosystem management.