We propose an original approach to model the effect of the initial spatial distribution of a reintroduced metapopulation (one-patch release versus multi-patch release) on local adaptation. Genetic and demographic processes are considered to investigate how the different patterns of adaptation resulting from initial conditions affect metapopulation viability. In agreement with classical interpretations in the fields of population biology and evolutionary genetics, we observe an influence of the degree of environmental correlation among patches on demographic processes and an influence of inter-patch connection on genetic processes. However, our results uncover some further effects of the environment, owing to positive feedback interactions among demographic and genetic processes. When considering the one-patch release, demographic stochasticity engenders a delay in the colonization of initially empty patches. This delay enhances the genetic asymmetry among patches (in terms of local adaptation), which in turn increases demographic asymmetry. In contrast, the multi-patch release produces similar levels of adaptation among patches. Metapopulation dynamics is strongly influenced by these differences, with contrasting effects under different environments. The pattern of adaptation produced by the one-patch release strategy is optimal under a regime of recurrent slight perturbations or environmental stochasticity, while the multi-release strategy is optimal in the presence of rare and severe perturbations.