Disturbance often has the net effect of promoting invasive plant establishment, but the precise nature of the relationship between disturbance and invasion can depend on community context. We used simulated gopher mounds in bare and monoculture plots of three California native grasses (Bromus carinatus var. maritimus [Piper] C. L. Hitch., Festuca rubra L., and Calamagrostis nutkaensis [J. Presl] Steudel) to test the effects of mounds on seedling establishment and survival of the European perennial grass Holcus lanatus L. Soil disturbance treatments were crossed with manipulations of the plant canopy (shade cages in bare and Bromus plots vs. pinning back grass leaves in Festuca and Calamagrostis plots) to separate some of the positive and negative effects of natural mounds. Mean PAR measured in February at the soil surface varied from 11 to 969 µm/m²/s. As predicted, shade structures that decreased light availability but also increased soil moisture generally increased Holcus seedling establishment and survival in bare and Bromus plots. In contrast, Holcus seedling establishment increased in response to increased light availability and soil disturbance in Festuca and Calamagrostis plots, even where water availability was lower. Thus, the relative importance of light and water availability across plot types appeared to determine the effect of disturbance on invasive plant establishment. Ultimately, Holcus survival was low on mounds in bare plots and on unshaded mounds in Bromus plots, and similarly low numbers of Holcus seedlings survived across all treatment combinations in Festuca and Calamagrostis plots. Therefore, our results do not support the hypothesis that gopher mounds explain the invasion of Holcus in native-dominated coastal prairie sites.