Human activities alter processes that control local biodiversity, causing changes in the abundance and identity of species in many ecosystems. However, restoring biodiversity to a previous state is rarely as simple as reintroducing lost species or restoring processes to their pre-disturbance state. Theory suggests that established species can impede shifts in species composition via a variety of mechanisms, including direct interference (e.g. territoriality), preempting resources, or habitat alteration. Here we use a long-term experimental manipulation of a desert rodent community to examine differences in the recolonization dynamics of a dominant competitor (kangaroo rats of the genus Dipodomys) when patches were already occupied by an existing rodent community relative to when patches were empty. Recovery of kangaroo rat populations was slow on plots with an established community of other rodent species, taking approximately two years. In contrast, recovery of kangaroo rat populations was rapid on empty plots with no established residents (approximately 3 months). We found little evidence that the delay in kangaroo rat colonization was due to direct interference from competitors, or could be explained by differences in habitat, implicating resource preemption by the established community as the most likely mechanism. These results demonstrate that the presence of an established alternate community inhibits recolonization by new species, even those that should be dominant in the community. This has important implications for understanding how biodiversity may change in the future, and what processes may slow or prevent this change.Significance statementEcological communities are changing due to human activities altering the processes governing local biodiversity. However restoring these processes often fails to restore the previous biodiversity state, implying that additional mechanisms contribute to community dynamics. Here we use an experimental manipulation of a desert rodent community—in which dominant competitors (kangaroo rats) were removed and then reintroduced years later—to show that the presence of previously-established species alters the dynamics of the dominant competitor’s recovery. Kangaroo rat populations took two years to recover on patches where inferior competitors were already established, compared to three months on uninhabited patches. This suggests that priority effects and initial conditions are critical to consider when predicting community response to disturbance, or in ecological restoration projects.