Abstract Prediction of species distributions in an altered climate requires knowledge on how global‐ and local‐scale factors interact to limit their current distributions. Such knowledge can be gained through studies of spatial population dynamics at climatic range margins. Here, using a butterfly (Pyrgus armoricanus) as model species, we first predicted based on species distribution modelling that its climatically suitable habitats currently extend north of its realized range. Projecting the model into scenarios of future climate, we showed that the distribution of climatically suitable habitats may shift northward by an additional 400 km in the future. Second, we used a 13‐year monitoring dataset including the majority of all habitat patches at the species northern range margin to assess the synergetic impact of temperature fluctuations and spatial distribution of habitat, microclimatic conditions and habitat quality, on abundance and colonization–extinction dynamics. The fluctuation in abundance between years was almost entirely determined by the variation in temperature during the species larval development. In contrast, colonization and extinction dynamics were better explained by patch area, between‐patch connectivity and host plant density. This suggests that the response of the species to future climate change may be limited by future land use and how its host plants respond to climate change. It is, thus, probable that dispersal limitation will prevent P. armoricanus from reaching its potential future distribution. We argue that models of range dynamics should consider the factors influencing metapopulation dynamics, especially at the range edges, and not only broad‐scale climate. It includes factors acting at the scale of habitat patches such as habitat quality and microclimate and landscape‐scale factors such as the spatial configuration of potentially suitable patches. Knowledge of population dynamics under various environmental conditions, and the incorporation of realistic scenarios of future land use, appears essential to provide predictions useful for actions mitigating the negative effects of climate change.