On the potential strength and consequences for nonrandom gene flow caused by local adaptation in flowering time

Gene flow is generally considered a random process, that is the loci under consideration have no effect on dispersal success. Edelaar and Bolnick (Trends Ecol Evol, 27, 2012 659) recently argued that nonrandom gene flow could exert a significant evolutionary force. It can, for instance, ameliorate the maladaptive effects of immigration into locally adapted populations. I examined the potential strength for nonrandom gene flow for flowering time genes, a trait frequently found to be locally adapted. The idea is that plants that successfully export pollen into a locally adapted resident population will be a genetically biased subset of their natal population - they will have resident-like flowering times. Reciprocally, recipients will be more migrant-like than the resident population average. I quantified the potential for biased pollen exchange among three populations along a flowering time cline in Brassica rapa from southern California. A two-generation line cross experiment demonstrated genetic variance in flowering time, both within and among populations. Calculations based on the variation in individual flowering schedules showed that resident plants with the most migrant-like flowering times could expect to have up to 10 times more of the their flowers pollinated by immigrant pollen than the least migrant-like. Further, the mean flowering time of the pollen exporters that have access to resident mates differs by up to 4 weeks from the mean in the exporters' natal population. The data from these three populations suggest that the bias in gene flow for flowering time cuts the impact on the resident population by as much as half. This implies that when selection is divergent between populations, migrants with the highest mating success tend to be resident-like in their flowering times, and so, fewer maladaptive alleles will be introduced into the locally adapting gene pool.