The longleaf pine (Pinus palustris) ecosystem is one of the most botanically diverse systems in the world outside of the tropics. Although less than 3% of historical longleaf pine ecosystems remain, many degraded stands in the southeastern United States are suitable for restoration. The existence of representative longleaf savannas plays an important role in this restoration. These reference stands are useful tools for measuring restoration progress, as they are typically chosen based on a set of biotic and abiotic features that define the ecosystem’s natural range of variation while maintaining biodiversity. However, reference sites may change over time and may no longer be useful benchmarks for gauging restoration success. Understanding how reference sites change over time is important if they are to be reliably used as standards for restoration. We compiled plant survey data collected by the Carolina Vegetation Survey (CVS) in the 1980s from sites considered to be representative of longleaf pine savannas across locations in North Carolina, South Carolina, and Georgia. The CVS is a multi-institutional research program designed to document the composition and status of natural vegetation communities for inventory and assessment of conservation status. These reference sites were resampled in 2009, and results were compared to initial CVS surveys to examine the extent to which species composition had changed over time.
Our results indicate that species composition of the reference sites was similar across the three locations (NC, SC, GA) when initially sampled by CVS. However, the species composition of these sites was significantly different in 2009. In addition, change was not consistent over the landscape, as species compositions from 2009 were significantly different across the three locations. Multiple factors likely play a cumulative role in changing species composition, and the combination and intensity of these factors vary by site. Compositional changes in reference sites may result from disturbance events, such as windthrow, flooding or forest fire, and represent the natural range of variation. Other sources of change can shift the species composition outside of this range and be difficult to detect at a local level, such as regional fragmentation and dispersal limitation. Our results emphasize the importance of reevaluating reference sites through time when applying them as a standard for restoration. In addition, determining the mechanisms that drive changes in species composition can provide insights into landscape-scale factors that will also influence restoration success.