Abstract:     A resilient innovation ecosystem can provide greater room for high-quality development. Therefore, it is of great significance to investigate how to improve the resilience of the regional innovation ecosystem. Based on the 2010-2019 Chinese co-application inventions and geographic information data, this study used the complex network methods and panel models to investigate how community dynamics affect the resilience of the regional innovation ecosystem. The results of this study showed that, under the impact of knowledge source risks, community evolution has an inverted U-shaped impact on the resilience of regional ecosystems. With the evolution of the community structure of “tight inside and loose outside” (close connection within the community and sparse connection between communities), the resilience of the regional ecosystem first increased and then decreased. It showed that when the “tight inside and loose outside” of communities increases from a low level to a moderate level, it is beneficial to reduce information transmission costs and transaction costs, increase social capital within the community, and thus have a positive impact on the resilience of the regional innovation ecosystem; but when the community structure dynamics is changed from a moderate level to a higher level, it will lead to “lock-in dilemma”, which will have a negative effect on the resilience of the regional innovation ecosystem. Therefore, the moderate level of community structure is more conducive to the improvement of the resilience of regional innovation ecosystems. Knowledge diversity positively mediates the relationship between community evolution and regional ecosystem resilience. The evolution of community structure further affects the resilience of regional ecosystems by directly affecting the knowledge diversity within the community. As the degree of “tight inside and loose outside” of the community increases, the knowledge diversity within the community will first increase and then decrease. We explain this phenomenon by investigating the learning mechanism within communities. Innovation community theory points out that the learning mechanism of the community has undergone a transition from “learning by localization” to “learning by specialization”. After entering the stage of specialized learning, innovation entities in the community will pay more attention to specific fields with strong technical similarities, which limits the development of knowledge diversification. Knowledge diversity is the guarantee that the innovation ecosystem has more alternative technological paths in the face of external shocks. Therefore, the change in the level of knowledge diversity from high to low will have an inverted U-shaped impact on the resilience of regional innovation systems. On the basis of existing research, this paper has investigated the mechanism by which community evolution affects the resilience of regional innovation ecosystems, and will further complement the research on the evolution mechanism of innovation ecosystems.The community structure of regions with higher levels of innovation performance or acting as knowledge fountains has a stronger marginal effect on regional innovation ecosystem resilience. Regional innovation ecosystems have different characteristics. The results of heterogeneity analysis showed that within the scope of the positive marginal effect of the inverted U-shaped in baseline regression, regions with a higher level of innovation performance or acting as knowledge fountains demonstrate higher marginal effects. This result explained the role of innovation capability and knowledge base in enhancing the resilience of innovation ecosystems and responded to existing research on the operating efficiency of regional innovation ecosystems. For policymakers, we propose that in the context of Chinese innovation ecosystem management, they should focus on how to get close to the “best balance point” in the evolution of the innovation community structure. Possible actions should be taken to monitor and manage the moderate turnover rate of community members in the regional innovation ecosystem.

Key words: innovation community, regional innovation ecosystem, knowledge source, resilience