Analysis of the heterogeneity of landscape risk evolution and driving factors based on a combined GeoDa and Geodetector model

•Scale response indicates that the township scale is the best study scale.•Areas with higher landscape ecological risk are more likely to gather spatially.•Coupling DPSIR model and Geodetector to explore the influence of driving forces.•GDP per capita has the greatest impact on landscape ecological risk.•Research results can inform regional ecological protection policies. Liaoning Province, one of China’s top 13 grain-producing areas, has recently implemented various ecological construction and protection plans. To ensure regional environmental protection, national food security, and the development of scientific ecological protection strategies, it is crucial to understand the spatial and temporal evolutionary patterns and driving forces of the landscape’s ecological risk. This study used spatial statistics and spatial autocorrelation to examine the scale response features of landscape risk in Liaoning Province. The best study scale was then selected and the GeoDa software platform was used to examine the pattern of landscape risk evolution in Liaoning Province from 2000 to 2020. Finally, the main spatial drivers of ecological risk in Liaoning Province were analyzed hierarchically according to the magnitude of the explanatory power of the q-statistic by combining the driving force-pressure-state-impact-response model with Geodetector. Several experiments were conducted to determine the optimal taxonomy and classification number for the discretization of each driver. The results showed that (1) the Liaoning River had a high-risk value and thus required careful consideration to maintain sustainable and healthy growth of the area and conserve ecological environment of the southeastern coastline of Liaoning Province. (2) The township scale was regarded as the ideal scale because all nine scales of landscape risk indices in Liaoning Province showed positive spatial connections and the greatest significant difference with it. (3) Clustering and significance analysis determined that townships in Liaoning Province with higher landscape risk indices were more likely to cluster spatially and had less spatial variance. (4) The two factors with the greatest impact on landscape risk in Liaoning Province were temperature and gross domestic product per capita. Landscape risk was the result of the combined influence of these multiple factors. Hence, the carrying capacity of the environment itself cannot be ignored while pursuing economic growth. Our findings offer theore