We report a likelihood-based method that estimates both dispersal and natural selection using the rate of change of the shape of a cline when selection and migration are not constant through time. We have investigated the case of local adaptation of the mosquito Culex pipiens to organophosphate insecticides in the Montpellier area in France. We have analyzed the modification of the clinal patterns at two resistance loci during the period from breeding to overwintering. We show that mosquitoes migrate extensively from breeding to overwintering sites at a rate that is markedly different from previous estimates made during the breeding season only. This migration is also strongly asymmetrical, which can be explained by different geographical distributions of breeding and overwintering sites, by variation in mosquito density along the transect, or by behavioral biases. We found that the starting time of overwintering is likely to vary between northern and southern populations and that substantial fitness costs are associated with resistance alleles at the two loci during overwintering. These results illustrate how demography and adaptive microevolution can be studied using selected markers. The method provides a framework to use population genetics and statistical models to reveal ecological and evolutionary processes.
Keywords: asymmetrical gene flow; cline; fitness cost; insecticide resistance; selection; variable gene flow.