New applications are motivating and informing the design of sensor/actuator networks, and, more broadly, research in cyber-physical systems (CPS). Our knowledge of many physical systems is uncertain, so that sensing and actuation must be mediated by inference of the structure and parameters of physical-system models. One CPS application domain of growing interest is ecological systems, motivated by the need to understand plant survival and growth as a function of genetics, environment, and climate change. For this effort to be successful, we must be able to infer coupled, data-driven predictive models of plant growth dynamics in response to climate drivers that allow incorporation of uncertainty. We are developing an architecture and implementation for precise fine-scale control of irrigation in an array of geographically-distributed outdoor gardens on an elevational gradient of over 1500 m, allowing design of experiments that combine control of temperature and water availability. This paper describes a system architecture and implementation for this class of cyber-eco systems, including sensor/actuator node design, site-level networking, data assimilation, inference, and distributed control. Among its innovations are a modular, parallel-processing node hardware design allowing real-time processing and heterogeneous nodes, energy-aware hardware/software design, and a networking protocol that builds in trade-offs between energy conservation and latency. Throughout, we emphasize the changes in system architecture required as missions evolve from sensing-only to sensing, inference, and control. We also describe our developmental implementation of the architecture and its planned deployment. Future extensions will likely add negative control of precipitation using active rain-out shelters and additional plant-level control of air or soil temperature.