The long-term fate of populations experiencing disequilibrium conditions with their environment will ultimately depend on how local colonization and extinction dynamics respond to abiotic conditions (e.g. temperature and rainfall), dispersal limitation and biotic interactions (e.g. competition, facilitation or interactions with natural enemies). Understanding how these factors influence distributional dynamics under climate change is a major knowledge gap, particularly for small ranged and dispersal-limited plant species, which are at higher risk of extinction. Epiphytes are hypothesized to be particularly vulnerable to climate change and we know little about what drives their distribution and how they will respond to climate change. To address this issue, we leveraged a 10-year dataset on the occupancy dynamics of the endemic orchid Lepanthes rupestris to identify the drivers of local colonization and extinction dynamics and assess the long-term fate of this population under multiple climate change scenarios. We compared 290 dynamic occupancy models in their ability to predict the colonization and extinction dynamics of a L. rupestris metapopulation. The model set predicted colonization–extinction dynamics as a function of asymmetric patch connectivity, moss area, elevation, temperature (minimum, maximum and variability) and/or rainfall. The best model predicted that local colonization increases with increasing asymmetric patch connectivity but decreases as minimum temperature and maximum temperature variability increase. The best model also predicted that local extinction increases with increasing variability in maximum temperature. Negative effects were more severe in smaller patches. Synthesis. Overall, our results demonstrate the role of asymmetric connectivity, climate and interactions with moss area as drivers of colonization and extinction dynamics. Moreover, our results suggest that asymmetrically dispersed epiphytes may struggle to persist under climate change because their limited connectivity may not be enough to counterbalance the negative effects of increasing mean or variability in temperature.