From composite resilience assessment to indicator-based structural fragility diagnosis in the Lower Yellow River Basin

Composite resilience indices summarize observed conditions in river-basin social-ecological systems, but they rarely reveal whether administrative units are structurally linked through similar responses to shared pressures. Here, we extend composite resilience assessment into an indicator-based diagnosis of structural fragility in the Shandong section of the Lower Yellow River Basin, China. A harmonized county-level dataset covering 25 counties and six benchmark years from 2000 to 2022 was used to construct a 34-indicator resilience assessment across social, economic, ecological, and water-resource domains. We first identified recurrent dominant drivers using Geodetector and estimated domain-specific driver coefficients. These coefficients were combined with county-level driver values to construct response attributes, response-similarity networks, and random and targeted node-removal stress tests. Composite resilience increased overall from 2000 to 2022, but its spatial hierarchy persisted: higher values remained concentrated in central and eastern counties, whereas several western counties repeatedly fell into the lower-value group. The response-similarity networks showed that state-based rankings did not necessarily correspond to structural embeddedness. Counties with moderate or lower composite scores could be more strongly embedded in response networks than some higher-scoring counties. At the node level, residual structural cohesion identified recurrent cross-domain dependence more clearly than functional retention. Four county nodes repeatedly appeared in the residual-cohesion upper tail across multiple networks, revealing cross-domain structural relevance that was not visible from composite scores alone. Progressive node-removal tests confirmed this structural sensitivity. Random removal produced gradual degradation, whereas targeted removal of structurally central counties caused earlier and sharper losses in normalized global efficiency and structural cohesion. These findings show that composite resilience scores alone are insufficient for diagnosing structural fragility in regulated and resource-constrained basins. The framework separates counties with weak observed conditions from counties that disproportionately maintain response-network performance, offering a clearer basis for basin management priorities and cross-county coordination.