Facilitating natural succession in a heavily invaded ecosystem.

• We altered key successional factors to promote restoration. • Rodeo® outperformed mulching + Oust® in decreasing the cover of Phalaris arundinacea. • Propagule supply limited the establishment of woody but not herbaceous species. • Bare root tree seedlings responded similarly to the two herbicide treatments. • Deer preferred cottonwood and silver maple and avoided swamp white oak. Invasive plants have major impacts on forest ecosystems. Where invasive species alter successional pathways, a site may reach an alternate stable state in which compositional change is unlikely without management intervention. We conducted a restoration experiment designed to alter the successional factors of site availability, species availability, and species performance to disrupt the stable state of invader dominance. In a field dominated by invasive reed canarygrass (Phalaris arundinacea L.), we compared two plot preparation methods (late-season site mulching plus Oust® application, vs. late-season Rodeo® application) to increase site availability and reduce Phalaris performance, and planted native trees and shrubs as seeds and as bare root seedlings at three densities to improve native species availability. We monitored Phalaris growth, other herbaceous species establishment, tree survival, and deer browsing (another influence on species performance) for two growing seasons. Although the mulching plus Oust treatment was initially more successful in terms of its ability to increase site availability (as measured by litter percent cover and depth), this pattern reversed in the second year of the experiment. Throughout the experiment, the Rodeo treatment was superior in terms of its ability to decrease Phalaris performance (as measured by cover and height), and in the second year it also improved the performance of herbaceous plants naturally establishing in the plots (as measured by cover). Volunteer herbaceous plant cover averaged 60% in Rodeo-treated plots by the end of the second growing season. The performance of planted, seeded, and naturally-occurring woody plants was similar across the two herbicide treatments. Overall survival of bare root seedlings was 41%; survival was greater than 70% for both swamp white oak and Dutch elm disease-resistant American elm. Deer preferentially browsed on silver maple seedlings, while demonstrating avoidance of American elm, cottonwood, sandbar willow and swamp white oak. Our results suggest that to promote succession in Phalaris -dominated sites, decreasing Phalaris performance is more important than immediately removing thatch. In terms of species availability, we conclude that propagule limitation was occurring for woody plants but not for herbaceous species; natural tree recruitment was very low, so our alteration of tree species availability was critical in terms of promoting succession to floodplain forest. The restoration approach of stimulating natural succession can be applied to other ecosystems, providing a link between successional theory and management practice. [ABSTRACT FROM AUTHOR]