Abstract
The genetic structure of populations is often shaped by processes and events that introduce asymmetries to gene flow between geographic locations. Existing population genetic methods like EEMS and FEEMS summarize gene flow using effective equilibrium migration rates that are assumed to be symmetric, providing representations of underlying structure that have proven useful, but that fail short of representing signatures of aysmmetric gene flow. More elaborate methods capable of inferring asymmetric gene flow exist, but are computationally intensive, restricting analyses to small sets of sub-populations (demes). Here, we introduce FRAME (Fine-Resolution Asymmetric Migration Estimation), a method that significantly reduces the computational burden of estimating asymmetric gene flow. This efficiency allows FRAME to infer asymmetric migration patterns for scenarios with a large number of demes, enabling fine-scale analyses. Under the assumed model and its equilibrium condition, FRAME's estimates can be interpreted as backward migration rates; in non-equilibrium conditions, FRAME detects and summarizes asymmetries in gene flow with effective equilibrium migration rates, which allows a richer representation of genetic structure. We assess the method using a variety of simulated histories of gene flow and apply the method to datasets from North American gray wolves, poplar trees, white-crowned mannakin, and human archaeogenetic samples. The applications demonstrate FRAME's scalability and flexibility in detecting complex asymmetric migration signals and population structures.