Abstract: Based on the 16 high-resolution autumn remote sensing images during 1973-2013, the classification characteristics of wetland landscape and spatial distribution pattern of vegetation and their succession trends in Poyang Lake were studied through  interpretation technology. Combined with water level fluctuations corresponding to the day of each remote sensing image obtained, the change of wetland landscape types at Poyang Lake wetland and its dynamic response to water regime were explored by multivariate statistical method. The results showed that: 1) there were 9 types of extracted wetland landscapes, viz, deepwater, moderatedepth water, shallow water, Carex, reed, swamp, sparsegrassland, mudflat and bareland. The total area of the 9 types of extracted wetland landscapes did not change significantly, which was always maintained at about 3000 km² during 1973-2013; 2) the areas of swamp and sparse-grassland both revealed an increasing trend with a higher increase for swamp, and the area of Carex declined while that of reed increased; 3) there was a significant negative correlation between the water level and the mudflat area. The water level fluctuation had no significant correlation with the areas of the bare-land and wetland vegetation. The results are helpful to explore the eco-hydrology process and its trend at Poyang Lake wetland, and provide a scientific reference for the maintenance of stable ecosystem functions of the seasonal freshwater lake wetland.

Key words: mean trophic level index, non-metric multidimentional scaling, multivariate statistical analysis, cluster analysis, community dynamics, index of relative importance.