Characterization of a North Carolina Lower Coastal Plain Wet Pine Savanna.

Abstract

Savannas in the southeastern United States formerly occurred from East Texas to southern Virginia. Logging, conversion to agriculture, and fire suppression have reduced their area and extent, and there is an urgent need to catalogue and classify the remaining savannas for both restoration and resource management purposes. The Big Savannah was a unique wet savanna on a geographically-limited Inceptisol in North Carolina that was studied and described by B.W. Wells during the 1920s. Unfortunately, the site underwent agricultural conversion and the unique community it supported was believed to have been eliminated from the North Carolina landscape. Subsequent savanna classifications have either not included this type of community or placed it with savannas occurring on Ultisols. During the late 1990s, vegetation reminiscent of that described for the Big Savannah was discovered 8 km north of the original site in two power line corridors. Shelingoski (2005) used the Carolina Vegetation Survey protocol to survey 13 100 m2 quadrats at Wells Savannah, the name given to the newly-discovered savanna. During summer 2005, we surveyed an additional 56 100 m2 quadrats on the Taylor Tract, private property adjacent to Wells Savannah. We compiled a data set from existing quadrat data inventoried from other lower coastal plain savannas to compare to the natural community at Wells Savannah and the Taylor Tract and to determine the relationship of this community to those found on other savannas in the region. We combined 12 of the Wells Savannah quadrats, 26 of the Taylor Tract quadrats, 6 quadrats from Holly Shelter Game Lands, and 27 quadrats from the Carolina Vegetation Survey (CVS) database. Quadrats from Holly Shelter Game Land and the CVS database were selected because of their similarities to the Wells Savannah and Taylor Tract quadrats, in terms of vegetation and topography, as well as geographical proximity. We used a variety of multivariate techniques to analyze the data. Multi-response permutation procedure (MRPP) results suggested significant vegetation and soil differences between the Wells Savannah/ Taylor Tract and surrounding savannas. Non-metric multidimensional scaling (NMDS), indicator species analysis (ISA), linear discriminant analysis (LDA), and classification tree modeling (CART) also clearly separated Wells Savannah and the Taylor Tract quadrats from other regional wet savannas. Species identified as indicators of Wells Savannah and the Taylor Tract tended to be either obligate or facultative wetland species, while the other wet savannas in the outer coastal plain of North Carolina contained a number of species, such as Aristida stricta, typically associated with areas experiencing shorter hydroperiods. NMDS ordination based on vegetation data separated Wells Savannah and the Taylor Tract from the other wet savannas, and discriminant analysis using soil characteristics correctly identified quadrats as either from Wells Savannah⁄ Taylor Tract or other wet savannas 76% of the time. Important distinguishing soil characteristics separating Wells Savannah and the Taylor Tract from the other wet savannas are the combination of higher clay percentage, higher calcium (ppm), and higher phosphorus. Recognition of a separate community type for Wells Savannah and the Taylor Tract is warranted, based on differences in vegetation and soil characteristics. Further exploration of the area is necessary in order to identify other remaining sites similar to the Big Savannah that may warrant preservation.