Stream fishes are highly diverse, yet highly imperiled by human alterations of stream environments. Many species are poorly characterized with regard to the size and structure of populations and patterns of dispersal between populations, which complicates assessment of how human activities, both harmful and beneficial, will affect persistence. I used genetic tools to further this understanding in three case-study fish species of the southeastern United States: Roanoke logperch (Percina rex) of the greater Roanoke River basin and redline (Etheostoma rufilineatum) and greenside darters (E. blennioides) of the upper Tennessee River basin.

I found that endangered P. rex persists in seven isolated populations. Within populations, individuals exhibit extensive dispersal and gene flow, which maintains connectivity throughout entire watersheds. Most populations exhibit small contemporary effective population sizes and occupy few stream channels, and thereby face an elevated risk of extinction. Genetic estimates of divergence indicate that fragmentation was recent, coincident with the construction of major dams throughout the species' range. Close evolutionary relationships between most populations suggest that a translocation strategy could decrease extinction risks. I developed a framework to help guide the process of balancing small-population versus translocation risks when formulating conservation strategies. When the framework was applied to populations of P. rex, straightforward management prescriptions emerged. The framework also may prove useful for other fragmented species.

Unlike P. rex, E. rufilineatum and E. blennioides are relatively abundant where they occur. However, both species were strongly affected by fragmentation due to hydroelectric dams and reservoirs. Populations in small streams flowing directly into a reservoir had lower genetic diversity than populations in larger, more fluvially connected streams. Furthermore, indices of watershed urbanization (e.g., percent impervious surface, road density) were negatively correlated with genetic diversity and with a genetic index of population stability. This suggests that darters occupying isolated streams and/or urbanizing watersheds experience smaller, more variable population sizes than darters elsewhere. Monitoring of such genetic responses could provide a useful early indicator of ecosystem stress and a useful complement to other biomonitoring techniques.