Objective  Ecological spatial networks play an important role in safeguarding ecological processes, maintaining ecological security, and enhancing ecosystem services. Currently, most of the research on ecological spatial networks focuses on potential networks, with relatively little attention paid to the study of existing ecological spatial networks. Based on long-term land use data, the extraction of existing ecological spatial networks and analysis of their structural evolution characteristics provides an important basis for optimizing network structure, constructing regional ecological security patterns, and enhancing regional ecosystem service functions.

  Method  Taking the Inner Mongolia sand-prevention belt as the research area, this study first used spatial morphological analysis method to analyze the spatial distribution characteristics of multiple landscape components in 1990, 1995, 2000, 2005, 2010, 2015, and 2020. Then, using the core area as the node and bridges as edges, the existing ecological spatial network of the research area was constructed. The structural evolution characteristics of the existing ecological spatial network were studied using various topological structure indicators, such as degree, degree distribution, diameter, average path length, and clustering coefficient in complex network analysis.

  Result  The landscape components in the research area were mainly core areas and connecting bridges, which accounted for 32.13% and 28.72% of the research area, respectively, followed by edges, islands and rings. The core area had been decreasing continuously, while the number of islands had been gradually increasing. The average degree of the ecological spatial network in the research area fluctuated between 20.01 and 39.98 over the past 30 years, and the power-law characteristics of degree distribution were obvious, with the power-law index showing an upward trend. The network diameter decreased from 24 in 1990 to 17 in 2020, and the average path length increased from 4.20 in 1990 to 5.14 in 2020. The average number of connected subgraphs was 61 over the past 30 years, showing an increasing trend, while the average size of connected subgraphs was 33.35, showing a decreasing trend. The network had a strong degree-degree correlation, and there was no obvious quantitative relationship between degree and clustering coefficient.

  Conclusion  The trend of various indicators over the past 30 years indicates that the ecological spatial network in the research area is a scale-free, non-hierarchical, assortative small-world network. The network has strong heterogeneity but shows a decreasing trend, with local connections being dense and community structure being obvious. However, long-distance connections across communities are decreasing, and the network is gradually splitting into multiple isolated subgraphs, resulting in a decrease in overall connectivity. Optimization of the landscape pattern in the research area can be achieved through measures such as building cross-regional ecological corridors and increasing connections between different communities.