Three ways of assessing metapopulation structure in the butterfly <i>Plebejus argus</i>

1. Three independent methods were used to investigate population structure in the butterfly Plebejus argus . First, migration and dispersal ability were measured by mark–release–recapture in seven adjacent habitat patches, and by release of butterflies in unoccupied habitat. Secondly, colonization of newly created habitat was observed over 7 years. Finally, genetic differentiation of local populations within a metapopulation was investigated. Sampled local populations included parts of the mark–release–recapture study area. 2. Plebejus argus is relatively sedentary: the maximum movement detected was 395 m, and only 2% of individuals moved further than 100 m between recaptures on different days. None the less, adjacent local populations in the mark–release–recapture study area were linked by occasional migration, with ≈ 1.4% of individuals moving between patches separated by 13–200 m. 3. Despite low mobility, observed colonizations occurred rapidly over distances of 1 km. Because P. argus occurs at high population densities, 1.4% migration can generate enough migrants to colonize newly suitable habitat quickly at this spatial scale. 4. Mark–release–recapture data were used to predict that there would be limited genetic differentiation through drift between local populations at this spatial scale. The prediction was supported by allele frequency data for the same local populations. 5. Genetic differentiation often indicates higher levels of migration than are revealed by the movements of marked individuals. This study shows that when experimental releases and extensive marking are undertaken in areas that are large relative to most movements, indirect measures of gene flow and direct measures of dispersal can concur. 6. Evidence from the three different approaches was complementary, indicating that P. argus occurs as metapopulations within the study area.