In an increasingly modified world, understanding and predicting the consequences of landscape alteration on biodiversity is a challenge for ecologists. To this end, metacommunity theory has developed to better understand the complexity of local and regional interactions that occur across larger landscapes. While metacommunity ecology has now provided several alternative models of species coexistence at different spatial scales, predictions regarding the consequences of landscape alteration have been done exclusively for the competition-colonization trade off model (CC). In this paper we investigate the effects of landscape perturbation on source-sink metacommunities. We show that habitat destruction perturbs the equilibria among species competitive effects within the metacommunity, driving both direct extinctions and an indirect extinction debt. As in CC models, we found a time lag for extinction following habitat destruction that varied in length depending upon the relative importance of direct and indirect effects. However, in contrast to CC models, we found that the less competitive species are more affected by habitat destruction. The best competitors can sometimes even be positively affected by habitat destruction, which corresponds well with the results of field studies. Our results are complementary to those results found in CC models of metacommunity dynamics. From a conservation perspective, our results illustrate that landscape alteration jeopardizes species coexistence in patchy landscapes through complex indirect effects and delayed extinctions patterns.