Until recently, a key assumption in biology has been that DNA mutations are the only source of heritable phenotypic diversification and therefore that adaptation is impossible in the absence of DNA sequence variants. However, this view is being increasingly challenged by the observation that changes in chromatin states, which are pivotal for the control of genome activity in eukaryotes, can sometimes become locked in and inherited across multiple sexual generations, independently of any change in the genome sequence. This system of inheritance, called epigenetic, is best documented in plants and often involves differential DNA methylation of repeat sequences, notably transposable elements. However, there is still a lack of systematic studies on the stability of epigenetic variants and their phenotypic consequences. Thus, the ecological or evolutionary impact of this type of variation, which could be different from that of DNA sequence variants by potentially providing a more rapid and reversible route to adaptation, remains unclear. I will present our efforts at addressing some of these issues using a population of near-isogenic, epigenetic Recombinant Inbred Lines (epiRILs) in Arabidospsis.