Population genetics theory predicts loss in genetic variability because of drift and inbreeding in isolated plant populations; however, it has been argued that long-distance pollination and seed dispersal may be able to maintain gene flow, even in highly fragmented landscapes. We tested how historical effective population size, historical migration and contemporary landscape structure, such as forest cover, patch isolation and matrix resistance, affect genetic variability and differentiation of seedlings in a tropical palm (Euterpe edulis) in a human-modified rainforest. We sampled 16 sites within five landscapes in the Brazilian Atlantic forest and assessed genetic variability and differentiation using eight microsatellite loci. Using a model selection approach, none of the covariates explained the variation observed in inbreeding coefficients among populations. The variation in genetic diversity among sites was best explained by historical effective population size. Allelic richness was best explained by historical effective population size and matrix resistance, whereas genetic differentiation was explained by matrix resistance. Coalescence analysis revealed high historical migration between sites within landscapes and constant historical population sizes, showing that the genetic differentiation is most likely due to recent changes caused by habitat loss and fragmentation. Overall, recent landscape changes have a greater influence on among-population genetic variation than historical gene flow process. As immediate restoration actions in landscapes with low forest amount, the development of more permeable matrices to allow the movement of pollinators and seed dispersers may be an effective strategy to maintain microevolutionary processes.