Watershed-level brook trout genetic structuring: Evaluation and application of riverscape genetics models

Stream connectivity promotes resilience and population viability of aquatic organisms. Landscape genetic approaches, traditionally applied to terrestrial systems, may reveal important watershed-level dynamics that influence connectivity. Additional validation would improve understanding of how the models perform when gene flow is constrained to dendritic networks. The objectives of this study were to use simulations to assess the utility of landscape genetics analyses in dendritic stream networks and investigate riverscape variables influencing gene flow among brook trout Salvelinus fontinalis populations in headwater streams. We used an individual-based simulation program to simulate different dispersal scenarios and used combinations of landscape genetic models, model selection methods, and genetic metrics to determine the best combination for riverscape systems. We also sampled brook trout from 76 headwater streams in two watersheds (c. 1,000 km(2)) in Connecticut, U.S.A. to assess connectivity and riverscape influences on genetic structuring. Gravity models with Bayesian information criterion (BIC) selection were the most accurate (>85%) landscape genetic models that consistently identified the correct simulated gene flow barrier. However, all models performed poorly when unidirectional barriers were simulated without a distance-based dispersal limitation. Excluding this scenario, model accuracy for the gravity models using BIC selection was >90% across multiple genetic metrics, validating the application of landscape genetic models to riverscape systems. We found highly variable levels of brook trout genetic connectivity (F-ST range 0.01-0.19) at the watershed level (5-15 river km). Gravity models identified increases in upstream impervious surfaces and decreases in riparian tree canopy cover as riverscape variables associated with increases in genetic differentiation in one watershed, while the other watershed was consistent with an isolation by distance pattern. In this study we used demogenetic (i.e. combined demographic and genetic) simulations to demonstrate the utility of landscape genetics techniques in dendritic river networks. Our empirical genetic study documented gene flow among headwater populations of brook trout at the watershed level and also suggested connectivity can be limited by watershed development. Incorporating the heterogeneity of riverscapes into connectivity-focused conservation planning is essential to the development of effective restoration actions, and landscape genetics approaches can be useful tools to identify watershed-level connectivity in stream systems.