<p>The balance of regional ecosystems has been increasingly disrupted by climate change and human activities. The Sanjiang Plain, an important commercial grain production base and a concentrated area of black-soil wetlands in China’s high-latitude region, has suffered a decline in regional ecosystem service functions as a result of large-scale agricultural development. However, the driving mechanisms and vulnerability of the ecological security pattern in the Sanjiang Plain remain poorly understood. This study integrates the Remote Sensing Ecological Index (RSEI) and the Ecological Service Function Index (ESI), uses the Optimal Geographical Detector (OPGD) to examine the Ecological network spatio-temporal evolution and driving mechanisms of the Sanjiang Plain, and employs circuit theory to delineate ecological corridors and identify ecological choke points and barriers. Finally, FANMOD software and robustness analysis were applied to evaluate the vulnerability of the ecological network. The results indicate: (1) Between 2000 and 2020, the coupling index of RSEI and ESI in the study area increased by 0.022. The OPGD results indicated that, during the study period, the interaction between the Normalized Difference Vegetation Index (NDVI) and the Normalized Difference Built-up Index (NDBSI) had the strongest explanatory power for ecological evolution, followed by Wetness (WET) and Land Surface Temperature (LST).(2) The number of ecological source areas increased from 35 to 44, with the total area expanding by 6,033 km² and covering most nature reserves. The number of corridors increased from 60 to 83, with the total length extending by 425.81&#xa0;km. The number of junction points also rose from 47 to 80. (3) Among all node types in the study area, the discrete pattern was the most widespread, with an average occurrence probability of 18.89%, followed by the single-gap pattern at 15.14%. (4) The results of the centralization analysis are consistent with those of the modularity analysis, indicating that connections among the various sources in the Sanjiang Plain tend to span long distances and form relatively few clusters. The network robustness assessment suggests that the network has weak resistance to node attacks. Therefore, while enhancing the influence of core nodes, it is also necessary to strengthen the connections between edge nodes and other nodes. This study provides new insights into the construction of ecological networks in black soil agricultural planting areas and offers a scientific basis for ecosystem conservation and regional sustainable development planning in large-scale agricultural expansion areas.</p>