A dynamic reference model: a framework for assessing biodiversity restoration goals in a fire-dependent ecosystem

The use of reference models as templates of historical or natural conditions to assess restoration progress is inherently logical; however, difficulties occur in application because of the need to incorporate temporal variation in ecosystems caused by disturbance and succession, as well as seasonal, interannual, or decadal variability. The landscape-scale restoration of the globally threatened and fire-dependent longleaf pine ecosystem in the southeastern United States is an example in which restoration efforts are even more complicated by the limited availability of extant reference sites. This study uses the dynamic reference conceptual framework to assess the direction and rate of recovery with respect to biodiversity restoration goals using a 15-year vegetation data set from an experimental restoration treatment in fire-excluded, hardwood-encroached longleaf pine sandhills. We compared ground-cover vegetation response to midstory hardwood removal through herbicide application, mechanical removal, and fire only. Nonmetric multidimensional scaling ordinations and proportional similarity analyses suggest that, while vegetation changed in all treatments over time, no differences in species composition or hardwood density in the ground cover were attributable to hardwood reduction treatments after 15 years with frequent prescribed fire. Furthermore, the results of this study indicate that considerable variability is associated with reference sites over time. Sites identified in 1994 as attainable restoration targets had become a moving target themselves, changing in magnitude consistent with alterations in restoration plots attributable to treatment effects and shaped by the modest increase in fire frequency imposed since 1998. In a broad restoration context, this study demonstrates a conceptual framework to better understand and integrate the range of spatial and temporal variation associated with the best available reference sites. It also illustrates a practical tool for statistically defining reference sites and for measuring restoration success in continually changing conditions that should be widely applicable to other ecosystems and restoration goals.