Throughout the world, large populations directly depend on the food, fuel and fiber produced by agroecosystems. These agricultural systems must be resilient to both increasing population pressures and changing environmental conditions. In many cases these systems are rainfed and, in these instances, increased rainfall variability and the elevated risk of extreme events associated with global climate change have the potential to adversely affect agricultural productivity. Adaptive management strategies are therefore necessary to support the long-term sustainability of these systems. An important consideration in developing management practices is the structure of the landscape, or spatial arrangement of land use practices. Landscape structure affects the resilience of agricultural systems in a fundamental way. Hydrological function and the consequent transport of soil are influenced by the configuration of land use practices. Location-specific management practices are thus required for different portions of any given landscape, and while this is understood, there is limited research supporting spatially explicit allocations of specific management practices. To understand and quantify the importance of landscape structure, this research models the effect of spatial arrangements of management practices. Different spatial arrangements are tested using the Unit Stream Power-based Erosion Deposition (USPED) model to assess potential for soil conservation. Multiple landscape configurations are simulated by placing regionally appropriate crop types in management zones based on physical characteristics of the watershed. For example, certain policy-related documents suggest that different management practices are more appropriate for specific portions of an area based on physical characteristics of the watershed. Four management scenarios were designed using variables such as topographic position, slope, and flow accumulation as the basis for assigning agricultural practices. In turn, the effectiveness of each of the scenarios is measured by a series of simulations to assess the sensitivity of the landscape to different management approaches as expressed by soil transport. Scenarios are compared against benchmark landscape configurations, including in situ practices and a randomly allocated configuration. The results of these model scenarios suggest soil transport is sensitive to the spatial arrangement of management practices. Management configurations designed to limit soil transport in areas of large upstream contributing areas show measurable reductions in potential soil transport. Counterintuitively, configurations based on watershed position and slope resulted in higher levels of soil transport than the in situ observations. The presented approach represents a useful model for understanding best management practices, exploring potential optimal land-management configurations, and serves as a framework for broader scale analyses.