Monitoring populations in the wild – their community assembly, population dynamics and seasonal movements – requires considerable time and effort. Remote techniques such as passive acoustic monitoring present new opportunities, but also new challenges. Soundscape analysis and machine learning approaches allow estimates of species richness, daily activity patterns, and relative species abundance. However, the volume (terabytes) and velocity (continuous) of the data involved require new approaches to visualization, inference, and hypothesis testing. Here, we present a preliminary study by OIST on Okinawa island, based on an island-wide acoustic monitoring network. By collecting data from a network of 24 field sites across an urban-rural gradient, we aim to describe the behavior of wild populations at the interface of nature and humanity, as well as to produce a dataset suitable for the needs of other researchers. Our preliminary results provide a proof of concept for automated species identification on Okinawa, and reveal patterns of biogenic vs. anthropogenic noise across the landscape. Future applications of this approach include monitoring endangered or invasive species, and connecting diversity across trophic levels to produce a holistic picture of ecosystem health and function.