The influence of landscape on genetic structure of a threatened reptile: the eastern massasauga rattlesnake

Thumbnail Image
DiLeo, Michelle Francis
population genetics , cdpop , landscape genetics , Sistrurus catenatus , assignment test , circuitscape
Understanding the impacts of both natural and anthropogenic landscape features on genetic diversity, population structure and connectivity has important implications for conservation of species living in fragmented environments. Here, I combine population genetic data, detailed land cover information, and computer simulations to explore how landscape shapes genetic structure across two regional populations of the threatened eastern massasauga rattlesnake (Sistrurus catenatus catenatus) in Ontario, Canada: one along the eastern shores of Georgian Bay and the other largely confined to the northern half of the Bruce Peninsula. First I used spatial Bayesian assignment to quantify the genetic population structure within each regional population. I found marked subpopulation structure within eastern Georgian Bay with differentiation of island and mainland snakes, a north-south split within the mainland coinciding with the town of Parry Sound, and evidence of further subdivision within the cluster of snakes north of Parry Sound. In contrast I found no population subdivision within the mainland of the Bruce Peninsula, but genetic distinction of mainland and island snakes. Next, I identified the landscape features that shape spatial genetic structure within regional populations. In eastern Georgian Bay I found local variation in the effect of landscape on populations. North of Parry Sound I found no effect of landscape on inter-individual genetic differentiation, but a strong pattern of isolation-by-distance. In contrast I found that both open water and roads restrict gene flow of snakes south of Parry Sound. I found no evidence of isolation-by-distance or that landscape shape genetic structure within the Bruce Peninsula. Finally I used individual-based, spatially explicit simulations to identify the lag-time associated with the detection of contemporary landscape feature effects on genetic structure of massasaugas, and explore the consequences of using spatially correlated land cover elements in landscape genetic analyses. I found that the genetic consequences of roads could be detected within 2-12 generations when population sizes were small or juvenile dispersal was low. However, I also found that roads could be spuriously identified as impediments to gene flow when spatially correlated features such as water are included in genetic models.
External DOI