Experimental evolution, where fast replicating organisms are evolved in controlled environments for thousands of generations, has shown that microorganisms are able to evolve at an amazing speed: in virtually all experimental frameworks that use bacteria or viruses, important phenotypic innovations have emerged in only a few tens of generations [1], and ecological diversifications are commonly observed [2]. Experimental evolution, by providing a variety of data from genetic mutations to ecological interactions, is an excellent tool to study multilevel evolution. Unfortunately, those experiments remain a long and costly process. As an alternative, computational models of In Silico Experimental Evolution (ISEE), where artificial organisms are evolved in a computer for thousands of generations [3], have already explored a lot of theoretical questions [4,5,6]. However, these models usually include only two or three scales (typically the genome, the phenotype and the environment), strongly limiting their possibility to mimic in vivo experiments, since evolution of real microorganisms implies the interaction of a wide range of biological structures and levels. We developed a multiscale framework of ISEE. In this model, bacterial-like organisms own a genome encoding a genetic regulation network and a metabolic network, and evolve on a virtual medium for tens of thousands of generations. By up-taking nutrients and releasing by-products, organisms modify their environment , possibly leading to complex ecosystem evolution. Thus, our individual based model evolves complex genotype-to-phenotype mappings and fitness landscapes. This model allows us to study a large variety of questions raised by experimental evolution, e.g. the evolution of the genome and the genetic regulation network, the evolution of ecological interactions, and so on [3]. A more complete description of the model is available in [7] as well as on the EvoEvo project website (http://www.evoevo.eu). The Long Term Experimental Evolution (LTEE), the longest bacterial experimental evolution experiment to date [8] has revealed an ecological diversification based on a niche construction associated to a negative frequency-dependent interaction [9]. By performing ISEE experiments with our model, we studied the environmental conditions in which such a diversification could occur. More precisely , we let initial random viable populations evolve during 500,000 time steps (∼40,000 generations) in three different environments: