Adaptive introgression in subdivided populations leaves distinct genomic footprints under global and local selection: a case study in Ciona intestinalis

Abstract

Adaptive introgression, the transfer of beneficial alleles between species through hybridization, is increasingly recognized as an important source of adaptation to novel environments. However, the theory of adaptive introgression has primarily been developed for a single panmictic population, whereas many empirical examples involve spatially subdivided populations. In such systems, a low-frequency adaptive allele may either represent the leading edge of an adaptive travelling wave or occur in a subpopulation where the introgressed allele is locally selected against. We developed a simple prediction based on the relationship between adaptive allele frequency and introgressed tract length. Simulations predict a positive correlation during an ongoing sweep of an unconditionally beneficial allele, but no correlation under local adaptation. Tract erosion provides a recombination clock for time since introgression, which also determines the distance travelled by the adaptive wave and hence allele frequency in the former, whereas allele frequency is determined by local selection in the latter. We tested this prediction in a well-characterized case of adaptive introgression, in the sea squirt Ciona intestinalis, using haplotagging to reconstruct phased genomes and local ancestry. We found no relationship between adaptive allele frequency and introgressed tract length, providing additional support for a model of local adaptation rather than ongoing global selective sweep.