Abstract:In this study, we examined interactions between landscape structure, land management and soil total nitrogen (TN; top 30 cm) and total phosphorus (TP; top 30 cm) within the densely populated rural landscapes of the Yangtze Plain, based on an intensive field study in a regionally-representative sample cells in Yixing County. The rural landscapes were mapped at a high spatial resolution and classified to ecologically-distinct features (ecotopes) by the direct interpretation and field validation of features in 2002 IKONOS imagery using a standardized mapping procedure. Then the samples of soils and sediments were collected for analysis at random points within ecotope selected by a cluster distance weighting stratified sampling design. The five most extensive land use/land cover categories in the Yangtze Plain were paddy, aquaculture, built structures with impervious cover, rainfed annual cropland and fallow water, which occupied 829% of the total area, contained 82.6% of the total nitrogen and 80.8% of the total phosphorus stored in the region′s rural landscapes, respectively. Regional scaling analysis indicated that 85.24×103 km2 of Yangtze Plain rural landscapes sequestered 29.87 Tg of the total nitrogen and 19.79 Tg of the total phosphorus, of which 51% and 45% were in paddy land respectively (15.26 Tg of TN and 9.13 Tg of TP), covering 45.5% of the region′s total surface area (38.93×103 km2). Regional estimates of landscape class areas, soil total nitrogen and soil total phosphorus were made by upscaling data obtained from the sample cells by using a multivariate regional optimization procedure combined with bootstrapping and Monte Carlo uncertainty analysis.Then the role of land use/land cover in determining local total nitrogen and total phosphorus patterns and the regional consequences of these patterns were revealed across a densely populated rural region of China′s Yangtze Plain. This fine-scale approach to investigating land management practices and their impacts on total nitrogen, total phosphorus and other ecosystem properties offers significant advantages of accuracy rate over conventional lower-resolution (30-1000m) methods for land cover.