Identification of bundles and driving factors of ecosystem services at multiple scales in the eastern China region

Intensive human activities have caused serious damage to the natural environment and threatened ecosystem services, while the relationships between ecosystem services at different spatial scales vary to some extent. Therefore, identifying different patterns, evolutionary trends, and influencing factors of ecosystem services at different scales is more helpful in providing scientific decision-making for sustainable ecosystem management. Taking eastern China as the research area, with the help of a Gaussian mixture model, redundancy analysis and geographically weighted regression models, the spatiotemporal pattern evolution of ecosystem services from multiple scale perspectives were conducted, and the interrelationships between ecosystem services and the driving factors were explored. It expanded the knowledge of ecosystem services, the relationships between ecosystem services, and how the socio-ecological drivers of ecosystems vary at different spatial scales, and enabled more precise manipulation of socio-ecological drivers through different driving models. The average value of ecosystem services in the southern region was 0.672, which was better than that in the northern region, which was 0.353. Ecosystem services were lower in regions with a wide distribution of arable land and high population density and economic development. The relationship between ecosystem services was more manifested as synergistic relationship. As the scale expanded, the synergies and trade-offs between most ESs were also stronger and the average increase was 18.42 %. Especially, the relationships between some services developed from synergistic relationships at small scales to trade-offs at the county level. From the perspective of ecosystem service bundle, the overall ecosystem environment of the research area has been greatly improved, with enhanced supply function and weakened vulnerability. The area of ecosystem provisioning bundles increased by an average of 10.31 % as the scale increased, while the ecosystem regulation bundle decreased by 7.33 %. By comparison, the non-spatially driven model had better accuracy at the scale of 1 km2 and 100 km2, while the spatially driven model had higher accuracy at the county scale. Besides, normalized difference vegetation index occupied an absolute dominant position, and the natural elements acted more at the micro-scale, while the socio-economic elements were more influential at the macro-scale. These results provided a basis for subsequent research on multi-scale ecosystem control and ecological compensation strategies, and showed that natural and social factors were scale-dependent, that it was necessary to study ecosystem services, and emphasized the importance of developing appropriate policies at particular scales.