Community persistence, or the ability of a community to maintain species composition and diversity through time, is a component of stability that is important to restoration. We ran a biodiversity-ecosystem functioning experiment for three years, and then stopped weeding it for 5-6 years, which allowed us to test whether increased plant species diversity and dissimilarity in height would lead to increased community persistence in the face of high invasion pressure by non-native species. Our approach was unique in that the experiment varied richness (one or four species) and evenness (three levels plus monocultures of the dominant species) using two separate dissimilarity types (having all tall species or having tall and short species combined) in six spatiotemporal blocks. Persistence was quantified as to how well positive productivity-diversity relationships, proportion of planted native species, and species richness remained unchanged over time. Thus, high persistence values indicate low levels of invasion and local extinction. We found that the positive relationship between diversity measures and productivity persisted after cessation of weeding. The proportion of planted species was 32% higher in mixture than in monoculture plots, indicating that monocultures were more heavily invaded by non-native species. Reduced evenness did not affect persistence measures in plots with dissimilar heights, but measures declined linearly with decreased evenness in plots with all tall species. Our results suggest that (1) persistence-diversity relationships are likely to vary with the traits of species becoming rare and going extinct, and (2) it is important to restore higher species diversity in restoration projects to favor the long-term persistence of planted species.