Abstract: Landscape pattern optimization is an effective way to build ecological networks, which can substantially improve the regional ecological security. This study, with Meilan district in Haikou City as the research area, objectively selected the ecological sources by the way of granularity inverse method and principal components analysis from the point of increasing the overall connectivity. The ecological network was built by combining the minimal cumulative resistance model. Through the spatial network analysis and hydrological analysis, this paper mainly discussed how to determine the scale, shape, and form of the ecological nodes. It’s an appropriate reference for the selection of ecological sources in Meilan district when the ecologicallandscape component granularity is 550 m. There were 38 ecological sources in Meilan district. The landscape types of 23 plaques needed to be converted. The scale and layout of ecological sources were different due to the anthropogenic influences. The difference between dominant and recessiveecological resistance indicated that there were five ecological frangible parts in Meilan district. The causing factors for such differences should be given full consideration when adaptive management strategies being taken during the construction of ecological networks. The mostcrucial one was to reserve ecological lands. The ecological network of Meilan district was made of 90 ecological corridors and 89 ecological nodes. Together with ecological sources, they improved the connectivity of ecosystem. Ecological nodes formed 79 ecological regions with a total area of 729.15 hm², within which 10 regions should be turned into aquatic systems and 69 regions into forests. We put forward a specific scheme for the construction of ecological nodes, which mayimprove the practical significance of landscape pattern optimization.

Key words: soil reservoir, winter wheat, dried soil layer, spring corn, rain-fed area of the Loess Tableland