A river runs through it: Causal graphs capture rivers’ complex control on the genetic structure of populations

Earth’s physiographic features shape the genetic evolution of organisms. Understanding the conditions under which such features act as barriers to gene flow requires quantifying and articulating the features of both the barrier and the organism(s). Many such physiographic features, however, have known interdependencies that are not expressed through common multivariate statistics. Here, we evaluate the use of directed acyclic (causal) graphs and structural equation modeling (SEM) to articulate and test these relationships. We chose the longstanding and contested Riverine Barrier Hypothesis as a test-case using 28 river-spanning population genomic datasets of plants and animals associated with 25 rivers across the contiguous United States; data were paired with seasonality, river width, and river discharge data for those rivers. SEMs revealed insights that could not be captured by traditional non-structured multivariate statistics. Discharge had the greatest direct effect on low-dispersing species. However, discharge has negative, indirect effects on other river features making its total effect on population differentiation negligible. River width was important for low dispersers, but surprisingly, narrower rivers were associated with higher Fst—this may be due to the association of higher topography with narrower (e.g., headland) parts of rivers. Or, wide lowland rivers may be more dynamic and facilitate dispersal more than highland rivers. Therefore, topography or landscape history and not wetted river area may determine barrier efficacy. This proof of concept shows the utility of causal graphs and SEM at articulating and testing complex relationships between Earth’s physiographic features and the organisms that evolve with them.