Little is known on the potential of ecological disturbance to cause genetic and epigenetic changes in plant populations. We take advantage of a long-term field experiment initiated in 1986 to study the demography of the shrub Lavandula latifolia, and compare genetic and epigenetic characteristics of plants in two adjacent subplots, one experimentally disturbed and one left undisturbed, 20 years after disturbance. Experimental setup was comparable to an unreplicated ‘Before-After-Control-Impact’ (BACI) design where a single pair of perturbed and control areas were compared. When sampled in 2005, plants in the two subplots had roughly similar ages, but they had established in contrasting environments: dense conspecific population (‘Undisturbed’ subpopulation) versus open area with all conspecifics removed (‘Disturbed’ subpopulation). Plants were characterized genetically and epigenetically using amplified fragment length polymorphism (AFLP) and two classes of methylation-sensitive AFLP (MSAP) markers. Subpopulations were similar in genetic diversity but differed in epigenetic diversity and multilocus genetic and epigenetic characteristics. Epigenetic divergence between subpopulations was statistically unrelated to genetic divergence. Bayesian clustering revealed an abrupt linear boundary between subpopulations closely coincident with the arbitrary demarcation line between subplots drawn 20 years back, which supports that genetic and epigenetic divergence between subpopulations was caused by artificial disturbance. There was significant fine-scale spatial structuring of MSAP markers in both subpopulations, which in the Undisturbed one was indistinguishable from that of AFLP markers. Genetic differences between subpopulations could be explained by divergent selection alone, while the concerted action of divergent selection and disturbance-driven appearance of new methylation variants in the Disturbed subpopulation is proposed to explain epigenetic differences. This study provides the first empirical evidence to date suggesting that relatively mild disturbances could leave genetic and epigenetic signatures on the next adult generation of long-lived plants.