Compared to their historical extent, oak savannas at the forest-prairie transition in the Midwestern United States are greatly reduced, in part, due to tree encroachment that produced closed woodlands. Thinning can reduce tree densities, but restoring the herbaceous understory, which often is sparse after years in a closed woodland, remains challenging. Re-seeding can be used for herbaceous understory restoration, but its efficacy is likely to show large spatial variability in savannas where isolated trees create pronounced resource heterogeneity. Further, efficiency of re-seeding depends upon how seeding rate translates into adult plant density and cover. In an recently-thinned blackjack oak (Quercus marilandica) savanna restoration in southeast Kansas I am asking 1) what is the relationship between seed addition rate and the resulting population density and cover of native grasses, 2) how does this relationship differ between oak canopy edges and openings between canopies and 3) how does this relationship differ for two native grasses, switch grass (Panicum virgatum) and indiangrass (Sorghastrum nutans). In spring 2009 I initiated a factorial experiment involving seed addition (0, 50, 250, 1000 seeds per 0.25 m2 plot) X habitat positions (tree canopy edge vs. between canopies) X grass species.
Preliminary analyses showed that in July, four months after seed sowing, the functional group composition (woody plants vs. forbs vs. legumes vs. grasses) of the understory plant community did not vary significantly with any of the treatments. Further, adding grass seed at any of the levels did not reduce bare ground in plots in July. Tiller densities in October increased significantly and linearly with seed addition rate (F1,114 = 39.11, p <0.001). There was a marginally significant grass species X seed addition interaction effect (F1,114 = 3.71, p = 0.0566) with tiller density increasing more rapidly with seed addition rate for Sorghastrum nutans than for Panicum virgatum. The relationship between seed addition rate and tiller density did not depend on position relative to tree canopies. Absence of effects of seed addition rate on percent bare ground and on functional group composition may result from the very small size of the juvenile grass plants that established. At these early life-stages, density-dependent processes do not counteract the beneficial effects of high seed addition rates on tiller density.