Applying landscape genetics to evaluate threats affecting endangered Atlantic salmon populations

Landscape genetics affords a potential analysis framework to evaluate the effect of contemporary land use on endangered species at a population level. However, historical patterns of population connectivity need to be accounted for prior to testing for the contemporary effect of threats. The goals of this study were to (1) optimally describe historical patterns in population connectivity for a diadromous fish species before (2) evaluating whether residual genetic variation was correlated with ecological changes arising from several types of land use. Using endangered Atlantic salmon populations as a case study, we evaluated whether historical patterns in population connectivity were more likely to result from dispersal limitation (isolation by distance) relative to habitat choice and reproductive success (isolation by environment). Second, we used Reciprocal Causal Modeling to identify the types of land use contributing to three threat indices, and subsequently Multiple Regression on Distance Matrices to evaluate the relative severity of each. These analyses suggest that straying Atlantic salmon avoid watersheds with reduced water quality (resulting from acidification and abandoned mines) and higher road density, yet are not responding to watershed fragmentation (from road-river crossings and dams) at a population level. This study is among the first to explicitly compare alternate behavioural hypotheses leading to dispersal patterns for diadromous fishes and to quantitatively assess freshwater threats for Atlantic salmon at a population level using landscape genetics.