Streams are patchy environments characterized by high variation in flow and connectivity. Heterogeneous conditions among stream reaches results in a similarly patchy competitive landscape for individual organisms. This can lead to pronounced genetic differentiation and varied success among local populations within a species inhabiting a particular river basin. Stream ecologists have hypothesized that, when basin-wide populations decline due to anthropogenic stressors, conservation practitioners can leverage heterogeneity in resilience to bolster or re-introduce individuals to struggling locations. There have been few opportunities to test this hypothesis due to limited availability of long-term, regionally complete data sets. However, the U.S. Fish and Wildlife Service (USFWS), in response to the listing of the leopard darter (Percina pantherina), began monitoring several endemic darter species across the whole of the Little River basin in the early 1990’s. Since 1998, this monitoring program has maintained a consistent and robust occupancy- and depletion-sampling protocol. In this study, we took advantage of this unique data set to test for the degree of spatial heterogeneity in the population dynamics of the leopard darter. To do so, we constructed a custom Bayesian model to integrate data from diverse sources in a spatiotemporal, state-space framework. Results suggest that leopard darters experience strong cyclical fluctuations in abundance throughout their range. We also show that some populations are declining while others are robust to environmental change. These results were coupled with simulations to determine the feasibility of re-introducing leopard darters to drainages from which they have been extirpated. Our approach demonstrates the high value of spatially broad data, as well as the promising potential of heterogeneous landscapes to support conservation of stream organisms.