In this study, the Poyang Lake Eco-economic Zone was selected for investigation, and changes in wetland landscape pattern in the area from 1995 to 2015 were explored by interpreting remote sensing images of Poyang Lake Eco-economic Zone in 1995, 2005 and 2015. At the same time, the driving forces of change were studied using geographic detectors. The objectives were to provide a theoretical basis and data to support wetland protection in the Poyang Lake Eco-economic Zone, and to gain experience for freshwater wetland protection in other regions of China. The interpretation and classification of the remote sensing images of Poyang Lake Eco-economic Zone in 1995, 2005 and 2015 was conducted by the eCognition object-oriented nearest neighbor method using the best feature combination. The dynamic degree, centroid model and landscape pattern index were used to analyze the dynamics of wetland change over the past 20 years. The results demonstrate that wetland image extraction by the eCognition object-oriented classification method was reliable, and the land types of Poyang Lake Eco-economic Zone consisted of two groups, wetland and non-wetland, and nine types: river, lake, swamp, paddy field, reservoir and pond, woodland, bare land, construction land and grass land. From 1995 to 2015, the total wetland area in Poyang Lake Eco-economic Zone decreased from 1 257 500 hm2 in 1995 to 972 500 hm2 in 2015, with an overall change rate of -22.7%. The area of natural wetland decreased the most, with a total decrease of 191 100 hm2. The wetland landscape index of Poyang Lake Eco-economic Zone showed that fragmentation of each patch increased, and landscape diversity and uniformity displayed an inverted "V" trend during 1995-2015. The centroid transition of the primary wetland types showed an obvious northward trend as wetlands in the south shrunk. Driving force analysis of wetland change using geographic detectors showed that population density, average annual temperature and slope direction were the primary drivers of wetland change, and the explanatory power of population density to the spatial differentiation of wetland change were higher than that of average annual temperature and slope direction. |