Experiments that have manipulated species richness with random draws of species from a larger species pool have usually found that invasibility declines as richness increases. These results have usually been attributed to niche complementarity, and interpreted to mean that communities will become less resistant to invaders as species go locally extinct. However, it is not clear how relevant these studies are to real-world situations where species extinctions are non-random, and where species diversity declines due to increased rarity (i.e. reduced evenness) without having local extinctions. We experimentally varied species richness from 1 to 4, and evenness from 0.44 to 0.97 with two different extinction scenarios in two-year old plantings using seedling transplants in western Iowa. In both scenarios, evenness was varied by changing the level of dominance of the tall grass Andropogon gerardii. In one scenario, which simulated a loss of short species from Andropogon communities, we directly tested for complementarity in light capture due to having species in mixtures with dissimilar heights. We contrasted this scenario with a second set of mixtures that contained all tall species. In both cases, we controlled for factors such as rooting depth and planting density. Mean invader biomass was higher in monocultures (5.4 g m−2 week−1) than in 4-species mixtures (3.2 g m−2 week−1). Reduced evenness did not affect invader biomass in mixtures with dissimilar heights. However, the amount of invader biomass decreased by 60% as evenness increased across mixtures with all tall species. This difference was most pronounced early in the growing season when high evenness plots had greater light capture than low evenness plots. These results suggest that the effect of reduced species diversity on invasibility are 1) not related to complementarity through height dissimilarity, and 2) variable depending on the phenological traits of the species that are becoming rare or going locally extinct.