Reciprocal exchange obscures phylogenetic patterns of introgression

Hybridization and introgression are major evolutionary forces in eukaryotes that have shaped the evolutionary trajectories of crops, livestock, and modern humans. Yet current theoretical and statistical frameworks provide only a partial view of the dynamics and genomic consequences of introgression. An unexplored dimension is the possibility that introgression between two lineages can occur simultaneously in both directions at a single genomic locus--a process we term reciprocal introgression. Although this type of introgression is plausible, especially in cases of bidirectional introgression across the genome, it remains largely undetectable under widely used statistical approaches, hindering evaluation of its prevalence in natural systems. Moreover, unrecognized reciprocal introgression may bias summary statistics commonly used to quantify gene flow, obscuring our ability to resolve introgression events. To address this, here, we develop theoretical expectations of the impact of reciprocal introgression on phylogenetic and site-pattern-based results. To test these expectations, we implement coalescent simulations to model genomes experiencing reciprocal introgression and apply sliding-window introgression statistics to these data. We find that reciprocally introgressed loci produce phylogenetic patterns opposite to those expected under unidirectional introgression. These mixed signals can dilute evidence of introgression or even generate false signals of introgression events that never occurred. Taken together, our results highlight the need for updated statistical approaches and pave the way toward a more complete understanding of the impacts of introgression on gene histories and organismal evolution.