Impacts of management and antecedent site condition on restoration outcomes in a sand prairie

Landscape context and site history, including antecedent site conditions, may constrain restoration potential despite the efforts of restoration practitioners. However, few experimental studies have investigated the relative importance of antecedent site conditions and the intensity of on‐site management in driving restoration outcomes. We established small‐scale prairie restoration experiments within the Lost Mound Unit of the Upper Mississippi River National Wildlife and Fish Refuge in Illinois, U.S.A. We investigated the effectiveness of two restoration treatments, herbicide application and seeding of native plants, on removal of invasive crown‐vetch (Securigera varia) and recovery of sand prairie plant communities. We replicated treatment plots across 15 locations with three levels of antecedent condition and fire treatment (burned, undegraded; burned, degraded; and unburned, degraded) to determine whether antecedent condition constrained the effectiveness of on‐site restoration. Two years after initial herbicide application crown‐vetch cover was significantly reduced relative to untreated controls. This effect was more pronounced in plots treated twice with herbicide. However, removal of crown‐vetch facilitated invasion by Kentucky bluegrass (Poa pratensis). Addition of native prairie seed had little effect on restoration outcomes, regardless of herbicide application. Native community recovery was greater in plots restored in less degraded locations. Herbicide application tended to increase native species cover, but importantly, this effect was significant only in the least degraded locations. Intensive restoration management conducted in degraded landscapes can result in undesirable outcomes such as secondary species invasion. Reestablishment of native species following restoration is more likely where the surrounding remnant communities are intact.