Abstract: Habitat fragmentation and loss have become the major threats to local biodiversity protection and regional ecosystem integrity. Developing urban ecological networks and improving connectivity between habitats could facilitate energy and matter fluxes, species dispersal, genetic exchange and many other ecological processes, and further relieve the ecological environmental crisis in cities. With rapid urbanization in the Three Gorges Reservoir region, natural area fragmentation and habitat loss are inevitable. Taking Kaizhou New Town as an example, our study built the urban ecological networks by leastcost path modeling and GIS technology in order to facilitate the establishment of native tree species. The best threshold for urban ecological networks was obtained by the verification from graph theory functioning index. Finally, the priority of habitat protection and restoration was determined after evaluating the importance grades of patches/connectivities. We set a native tree, Pterocarya stenoptera as a surrogate species and used the habitat suitability assessment values to build the cost surface for leastcost model by Linkage Mapper Toolist. The observed maximum seed dispersal distance (300 m) of P. stenoptera was used as the distance threshold for network analysis. 75 target patches were selected as habitat sources to develop networks. Under the 300 m distance threshold, 98 links were simulated for the network, and were grouped into six grades according to their importance values. We found the links with higher importance values almost all existed along the bank between the Pengxi River and Hanfeng Lake. Among these, the most important link was between Hanfeng Lake Park and Kaizhou Planning Institute; the second important link was located on both sides of the highway bridge over the Pengxi River. The integral index of connectivity of the whole landscape would be reduced by 19.1% if the two links were removed. Additionally, we found that the importance of the link was not significantly correlated with the important values of the two patches that the link connected. Link importance may also relate to the shape and spatial location of patches. Based on these results, we proposed some network improvement strategies. This study combined leastcost path analysis and a graphtheoretical algorithm and provided a scientific, reliable reference for planning and construction of urban ecological networks.