Theoretical and empirical studies suggest that geographical isolation and extinction-recolonization dynamics are two factors causing strong genetic structure in metapopulations, but their consequences in species with high dispersal abilities have not been tested at large scales. Here, we investigated the effect of population age structure and isolation by distance in the patterns of genetic diversity in a wind-pollinated, zoochorous tree (Olea europaea subsp. guanchica) sporadically affected by volcanic events across the Canarian archipelago. Genetic variation was assessed at six nuclear microsatellites (nDNA) and six chloroplast fragments (cpDNA) in nine subpopulations sampled on four oceanic islands. Subpopulations occurring on more recent substrates were more differentiated than those on older substrates, but within-subpopulation genetic diversity was not significantly different between age groups for any type of marker. Isolation-by-distance differentiation was observed for nDNA but not for cpDNA, in agreement with other metapopulation studies. Contrary to the general trend for island systems, between-island differentiation was extremely low, and lower than differentiation between subpopulations on the same island. The pollen-to-seed ratio was close to one, two orders of magnitude lower than the average estimated for other wind-pollinated, animal-dispersed plants.Our results showed that population turnover and geographical isolation increased genetic differentiation relative to an island model at equilibrium, but overall genetic structure was unexpectedly weak for a species distributed among islands. This empirical study shows that extensive gene flow, particularly mediated by seeds, can ameliorate population subdivision resulting from extinction-recolonization dynamics and isolation by distance.K E Y W O R D S : cpDNA, extinction-recolonization, genetic structure, island colonization, pollen-to-seed ratio, seed dispersal.Patterns of population genetic structure are shaped by synergistic effects of spatial, historical, and ecological factors. Empirical studies have largely focused on the spatial distribution of allelic frequencies, dispersal syndromes, or mating systems to explain the levels of differentiation observed in neutral markers across the range of a species (Loveless and Hamrick 1984;Petit et al. 2005). Species, however, are dynamic entities, and historical factors may also significantly contribute to patterns of genetic differentiation among populations. Thus, natural populations may experience catastrophic events, promoting local extinctions and subsequent opportunities for recolonization (Wade and Goodnight 1998). As a result, population turnover significantly modifies the constant population sizes assumed under Wright's island model (Wright 1951), and hence metapopulation dynamics provide an