A genomic assessment of population structure and gene flow in an aquatic salamander identifies the roles of spatial scale, barriers, and river architecture

Population structure and gene flow of species in lotic environments can be constrained by river network architecture, species life history and heterogeneous local barriers. Identifying the factors that influence population structure and gene flow, especially in species limited to movement within a river network, is vital for understanding the evolutionary and demographic history of a species. We explored population structure and gene flow for a fully aquatic salamander, the common mudpuppy (Necturus maculosus), in Kentucky (USA) using genomic data. We examined population structure using both parametric and nonparametric methods, and we tested for a history of lineage divergence among identified genetic clusters. We quantified the partitioning of genetic variation at different hierarchical levels, and we tested for signatures of isolation by distance. Additionally, we used coalescent-based model selection to identify a best-fit model of gene flow between our three sampled basins. We found the greatest support for population structure between the Kentucky River basin and the combined Licking and Kinniconick basins, with further subdivision within both the Kentucky and Licking River basins. However, we found no evidence for a history of lineage divergence among these structured units. The movement of N. maculosus is constrained by the lotic network architecture, which likely drives the evolution of this hierarchical population structure, with increasing differentiation between sites nested in river basins, and even greater differentiation between basins. However, we also found evidence for population structure not explained by river architecture, with an isolated population embedded within the Kentucky River basin. This study demonstrates the heterogeneity in population structure that can evolve in aquatic species occupying lotic systems and illustrates the potential for genomic data to disentangle these complex patterns.