Wetland restoration is a pressing conservation priority, but there are few replicated field studies that provide a scientific foundation for these activities. We conducted a 3-year, replicated field experiment to examine the effectiveness of initial site preparation techniques (combinations of solarization, herbicide, tilling, and thermal weed control) in restoring native plant biodiversity to an agricultural field in a former wetland prairie in Oregon, USA. Post-treatment, plots were sown with a typical restoration mix of native graminoids and forbs. Treatments were compared to three high-quality managed reference wetlands and the adjacent agricultural field. Site preparation treatments varied in their effectiveness in suppressing extant vegetation and eliminating the residual seed bank. After 1 year, the solarization and fall herbicide application treatments were the most effective at reducing exotic cover. However, after 3 years, plant community composition converged in all treatments due to a loss of annual species and increasing dominance of native perennial bunchgrasses. Plant community composition became more similar to the reference wetlands each year, but diversity and richness diverged, apparently due to a trade-off between the cover of the dominant native bunchgrasses and diversity. Successional theory offers insights into how priority effects and competitive inhibition may influence community trajectories, and offers a useful model for restoring plant communities with high native diversity and dominance. Finding ways to mitigate the tradeoff between native plant cover and diversity by actively managing successional trajectories is an important challenge in wetland restoration that deserves further investigation.