Objective Increasing ecological risks have emerged in the process of urbanization. Scientific assessment and management of these ecological risks are prerequisites for establishing harmonious human – land relations and improving ecosystem health. Landscape ecological risk (LER), a critical subset of ecological risk assessment, focuses on the coupling of ecological processes and spatial patterns, emphasizing the changes in regional ecological risk under the combined influence of human behavior and natural activities. A large amount of research has concentrated on risk evolution, the identification of influencing factors or drivers, and the simulation and prediction of risks in natural areas such as watersheds, coastal zones, wetlands, and ecological reserves. However, relatively little attention has been given to urbanized areas characterized by high rates of population growth and land development. The formation mechanism, spatio-temporal heterogeneous driving effect and scale response law of LER in highly urbanized areas still need to be systematically explored.

Methods Utilizing multi-source data spanning the period from 1995 to 2020, including Landsat TM/ETM imagery, digital elevation model (DEM) and NASA datasets, this research focuses on exploring the characteristics of spatio-temporal variability of LER and corresponding drivers in Suzhou at the optimal scale, utilizing the LER evaluation model and the geographically and temporally weighted regression (GTWR) framework.

Results 1) The landscape pattern of Suzhou City is scale-dependent, with significant variations in the landscape pattern index for each land type at the optimal scale. In this research, the optimal granularity and magnitude are identified as 550 m and 1 250 m. 2) The overall LER demonstrates an improving trend. Low-risk and lower-risk zones are predominant, and the spatial heterogeneity of risk zones across all levels appears to be increasing. The LERI exhibits a pattern of increase followed by a decrease, characterized by phased recovery, and has gradually improved since 2010. The proportion of low-risk areas remains consistently high, widely distributed, and stable. In contrast, the proportions of medium- and high-risk areas show continuous growth, with small fluctuations in their respective increases, indicating significant aggregation. 3) LER is primarily influenced by socio-economic factors, exhibiting notable differences in the degree of spatial variability and the characteristics of the coefficients associated with these driving factors. The analysis reveals a positive correlation between natural and socio-economic factors and changes in LER, while proximity factors demonstrate a negative correlation. The spatial intensity of each driver varies significantly. The natural environment factor displays the most stable pattern, forming a concentric distribution centered around high-value areas such as the southern region of the Taihu Lake, southwestern Kunshan, and northern Zhangjiagang. In contrast, socio-economic factors, particularly population and GDP, contribute to an outward expansion of influence from the city center. Additionally, the neighborhood factor has undergone significant changes in its pattern since 2010, with these changes being more pronounced than those observed prior to that year. The alterations in the proximity factor are also more evident after 2010, with the influence thereof increasingly concentrated in city centers and sub-centers, rather than being dispersed over a broader area.

Conclusion This research advances the scale-integrated assessment of LER in urban environments by determining optimal granularity (550 m) and magnitude (1 250 m). It reveals that the precision of landscape analysis is enhanced when the scale effect is integrated with the assessment of ecological risk in urban landscapes. The optimal granularity and magnitude identified in this research are 550 m and 1250 m, respectively. In general, the LER of Suzhou City is favorable and is characterized by an overall improvement and a staged progression of “slowing down – exacerbation – recovery”. According to the distribution characteristics of LER evolution in Suzhou, low-risk and lower-risk areas consistently dominate the landscape. However, in recent years, there has been a notable shift from low-risk areas to medium-risk areas, and subsequently from medium-risk areas to higher-risk areas. It is imperative that high-risk areas be subjected to stringent control to prevent the indiscriminate expansion of construction land and to reinforce the efficient and intensive utilization of existing land resources. The research results can provide support for deepening the LER theoretical system at the optimal scale and high urbanization level, and offer scientific references for ecological risk management and control in the same type and post-urbanization areas.