Considerable research has been performed in applying run-time reconfigurable component models to wireless sensor networks. The capability to dynamically deploy or update software components allows the changing requirements of sensor network applications to be effectively managed, while concrete interface definitions promote re-use. Realizing distributed reconfiguration in wireless sensor networks is complicated by the inherently asynchronous and unreliable nature of sensor network environments. In such an environment, traditional, centralized approaches to achieving distributed reconfiguration are impractical. This paper introduces a graph-based approach to specifying the reconfiguration of software resources that may be distributed across multiple sensor networks. This approach requires application developers to specify only high-level reconfiguration graphs, which are then optimized and enacted in a hierarchical and autonomic manner. We demonstrate and evaluate our approach using a case-study scenario.