Patterns of oak regeneration in a Midwestern savanna restoration experiment

Savannas’ defining feature is a scattered tree layer. However, woody plants have encroached savannas throughout the world, altering tree densities and potentially modifying regeneration dynamics. We used a replicated large-scale restoration experiment with Midwestern oak savannas (USA) to understand spatial patterns of regeneration by the dominant overstory species, Quercus alba. Q. alba is not only of interest as a savannaforming species, but also a species of concern due to widespread low rates of regeneration. In this experiment, four sites received a restoration treatment, whereby all encroaching mesophytic trees and large shrubs were mechanically removed, and four sites operated as encroached controls. Within each site, we monitored naturally occurring Q. alba seedlings in plots under overstory trees and in plots in inter-canopy gaps for 4 years, including 1 year before and 3 years after treatment. Seedlings were 5 more abundant below overstory Q. alba trees, which we attribute to seedrain. There was no effect of restoration on seedling densities. Year-to-year survival and survival of seedlings tracked over the course of the study were greater in sites undergoing restoration. Four-year survival rates were 31% in treatment sites and 10% in control sites. Across all sites, seedlings survived equally as well under canopy trees as in inter-canopy gaps. Mean seedling height, basal diameter, and number of leaves were greater in sites undergoing restoration, relative to control sites, and in inter-canopy gaps, relative to under canopy trees. Furthermore, there was an interaction between treatment and plot type, whereby mean seedling attributes were greatest in inter-canopy gaps of treatment sites. Seedlings tracked over the course of this study displayed greater survival, D (change in) height, D basal diameter, and D number of leaves in treatment sites, relative to control sites, and greater survival, D basal diameter, and D number of leaves in inter-canopy gaps, relative to under canopy trees. Individual seedling survival,D basal diameter, andD number of leaves were greatest in inter-canopy gaps of treatment sites. Our study suggests that woody encroachment is limiting regeneration of Q. alba in savannas; however, this may be ameliorated by removal of encroaching woody vegetation. Thus, the act of restoring these savannas may hold promise for promoting regeneration of Q. alba. Finally, the restored savannas in this study appear inherently unstable, as seedlings performed best in inter-canopy gaps. Further work with prescribed fire and/or grazing may elucidate stable tree–herbaceous understory coexistence in Midwestern oak savannas. # 2007 Elsevier B.V. All rights reserved.