Habitat networks are important to the long-term survival of many species. However, due to frequent human disturbance, the habitat networks in most developing urban regions of China desperately need rebuilding or optimization. Identification of potential conservation areas is usually based on land use and land cover data from a single year; multiple years of data are rarely used. Additionally, in majority of the studies, only one network building method is applied to optimize networks; for example the least-cost path method, the Spatial Links Tool or the LARCH model. Furthermore, studies which combine spatio-temporal analysis and different network building methods to optimize networks are rare. The aim of this study was to determine:how spatio-temporal analysis and different network building methods affect network optimization and how combining spatio-temporal analysis and different network building methods can contribute to the methodology of network optimization. In this case study, the Su-Xi-Chang area of the Yangtze River Delta Region was taken as the study area and Egretta garzetta was selected as a representative species of wild animal. Egretta garzetta distribution was identified and recorded in a grid using ArcGIS software along with land use/land cover data from 2010 and 2000. Potential corridors were identified using the least-cost path method and a 10 km-radial-line path method, based on land use/land cover data from 2010 and 2000, respectively. The potential corridors and habitats identified using these two methods and years were overlaid. Habitats that existed in 2000 but not in 2010, and were linked by both types of potential corridors were identified. These were named Rebuild Potential Habitats (RP habitats). Habitats that did not exist in either 2000 or 2010, but were at the junctions of potential corridors were also identified; these were named Newly-added Potential Habitats (NP habitats). The landscape connectivity indices of RP habitats, NP habitats and their comprehensive values were calculated and arranged. The habitats that had higher comprehensive values were selected from the perspective of a Set Covering Problem. These selected habitats along with habitats from 2010 formed the optimized habitat network. 17 RP habitats and 18 NP habitats were identified. The identified potential habitats and the existing habitats from 2010 formed network scenarios Ⅱ, Ⅲ and Ⅳ, respectively. The RP habitats had areas of approximately 10-50 hectares. The land-use type for most RP habitats was arboreal forest. The land-use types for most NP habitats were arboreal forest, lakes, ponds, and rivers. These have suitable conditions for Egretta garzetta. There were 14 RP habitats and 12 NP habitats left after selection. The selected potential habitats and the existing habitats in 2010 formed network scenario Ⅴ. Comparison of three network structure indices (α, β, and γ) for network scenarios Ⅱ, Ⅲ, Ⅳ, and Ⅴ showed that scenario network Ⅴ offered the maximum economic and ecological benefits from a limited land area. The network structure connectivity in 2010 could emulate that of 2000, even if RP habitats were rebuilt in the same pattern as 2010. This indicates that spatio-temporal changes had an obvious effect on ecological processes and patterns. Network structure is not necessarily optimized even if all RP habitats and NP habitats meet the conditions of the different network construction methods. This suggested that the chosen ecological process model or network optimization criteria have important influences on the results. The method developed in this study was helpful in analyzing the relationships among spatio-temporal patterns, changes in network structure and ecological processes and patterns. Our analysis also highlights on the methodology of network structure optimization.