2006 SOUTHEASTERN NATURALIST 5(1):93–102 Comparison of Sandhills and Mixed Pine-hardwood Communities at Fort Benning, Georgia BEVERLY COLLINS1,*, REBECCA SHARITZ1, KATHRYN MADDEN1, AND JOHN DILUSTRO1 Abstract - Fall Line sandhills vegetation occurs on dry, sandy ridgetops and supports a suite of rare or uncommon plant species (TES). We surveyed nine sandhills sites and 32 “matrix” mixed pine-hardwood stands at Fort Benning to characterize canopy and groundlayer vegetation patterns and determine the extent of sandhills vegetation, including characteristic dominant species and TES, over the upland landscape. The relative abundance of Pinus palustris (longleaf pine), P. taeda (loblolly pine), and P. echinata (shortleaf pine) and sandhills oaks contributed to canopy composition differences among sites. The sandhills communities support a unique set of groundlayer species, including state-listed Chrysoma pauciflosculosa. Although there is some species overlap, especially in overstory composition, characteristic sandhills vegetation is not widely distributed in mixed pine-hardwood stands at Fort Benning and conservation might best be achieved by maintaining existing sites. Introduction Fire-adapted longleaf pine-scrub oak communities of the Coastal Plain– Piedmont Fall Line Sandhills region from Georgia to North Carolina (Keys 1995) support a characteristic flora that includes a suite of rare or uncommon plant species (Table 1). These communities occur on dry, sandy ridgetops (Christensen 1988, Jacqmain et al. 1999), which are remnants of ancient dunes in the Fall Line region. More fertile, less droughty sites and slopes in this region grade into upland pine and hardwood forests (Skeen et al. 1993, Ware et al. 1993). Land use, including agriculture and fire exclusion since European settlement, likely has modified sandhills vegetation (Christensen 1988), including the distributions of rare and uncommon species. Currently, many upland forests on federal and state lands in the Fall Line region are managed through periodic harvests and prescribed fires to promote habitat for the endangered Picoides borealis Viellot (Red-cockaded Woodpecker). Understanding how characteristic sandhills communities are arrayed over landscape and land management gradients can contribute to an ecological understanding of these communities and inform land-management decisions. We surveyed and compared sandhills communities that contain threatened sandhills species (Table 1) and a representative set of upland forests that range within the gradient from sandhills scrub oak-pine to pinehardwood stands (Dilustro et al. 2002) to characterize canopy and groundlayer vegetation patterns and determine the extent of sandhills vegetation over the upland landscape at Fort Benning. 1Savannah River Ecology Laboratory, PO Drawer E, Aiken, SC 29802. *Corresponding author - Collins@srel.edu. 94 Southeastern Naturalist Vol. 5, No. 1 Methods Fort Benning is located in the Sandhills region (Keys et al. 1995) on the Coastal Plain-Piedmont Fall Line in west-central Georgia and eastern Alabama. The installation, which now covers 73,533 ha, was acquired in phases from 1918 through 1942. Prior land-use was mainly farming and grazing, with some remnant forested land. The climate is temperate, with maximum summer temperatures averaging 32 °C (Johnson 1983). Precipitation averages 132 cm per year; 50% falls from April to September (Johnson 1983). The terrain of the base is characteristic Fall Line topography, with level ridge tops and gentle slopes. Upland soils on our study sites are primarily ultisols and entisols and include Troup sandy loams, Lakeland sands, Siley loamy sand, and Nankin sandy loams (Johnson 1983). We selected state (GA) or federally listed threatened or endangered plant species (TES) that typically occupy sandhills habitats (Table 1) and may extend into sandy woodlands and dry upland woodlands (Radford et al. 1968). We selected nine sites at Fort Benning with one or more of the TE species (Fig. 1), using GIS-data provided by the The Nature Conservancy. In 2003, these sites were surveyed for vegetation (summer months) and soil texture (winter months). TES plant populations were delineated and transects, spaced 30 m apart (100 m apart for extremely large populations), were established. Transects reach just outside the population boundary. At 30-m intervals along each transect, a 1- x 1-m plot was established to estimate percent cover of herbs, bare ground, and woody debris to the nearest one percent. Canopy species were quantified at each point using point-quarter method. Canopy and groundlayer vegetation on the TES sites were compared to a set of 32 upland forest stands (Fig. 1), each 400 m square, that are part of a larger study of the ecological effects of land-use and management practices on upland forests at Fort Benning. These stands represent the range of “matrix” upland forest canopy composition at Fort Benning. They include Table 1. Some plant species of concern in the Fall Line Sandhills region of Georgia. * denote threatened or endangered species found at the Fort Benning sites. Conservation Scientific name Common name statusA Astragalus michauxii (Kuntze) F.J. Herm. Sandhill milk-vetch G3; GA-S2 Baptisia lanceolata (Walt.) Ell. Lance-leaf wild indigo G4; GA-S4 Carphephorus bellidifolius (Michx.) Torr. & Gray Sandhill chaffhead G4; GA-S1 Chrysoma pauciflosculosa (Michx.) Greene * Woody goldenrod G4,G5; GA-S3 Liatris secunda Ell. Sandhill gay-feather G4,G5; GA-S1 Nolina georgiana Michx. Georgia beargrass G3, G5; GA-SNR Phaseolus polystachios (L.) B.S.P. * Sandhills bean G4,T3; GA-S2 Stylisma pickeringii (Torr. Ex M.A. Curtis) Gray * Pickering’s morning-glory G4,T3; GA-S2 Warea cuneifolia (Muhl. ex Nutt.) Nutt. * Sandhill-cress G4; GA-S3 AState (GA-S) or Global (G, T) rankings: SNR = species not ranked; S2, G2 = Imperiled; S3, G3 = Very rare and local throughout its range or found locally in a restricted range; S4, G4 = Apparently secure, although it may be quite rare in parts of its range; G5 = Secure globally; T = Threatened. Table source: Georgia Department of Natural Resources 2004. 2006 B. Collins, R. Sharitz, K. Madden, and J. Dilustro 95 mixed pine-hardwoods (6 stands) and pine stands dominated by longleaf (7 stands), loblolly (15 stands), or shortleaf pine (4 stands). Half of the sites are in land compartments open to mechanized training (tracked vehicles, heavier military use). The remaining sites are in compartments limited to dismounted infantry training (foot traffic, lighter military use). All sites are in active forest management and were prescribe-burned in early 2000. Half were burned again in early 2002 (2-yr stands) as part of the experimental land-use treatments for the larger study. In each 400- x 400-m stand, we established a 100- x 100-m vegetation sampling plot gridded at 20-m intervals to yield 25 sampling points in each stand. In 2003, trees were surveyed at each of the 25 points using the point quarter method. Groundlayer vegetation, defined as vegetation < 1.4 m tall, was surveyed by line-intercept along a 6-m transect at each sampling point. Vegetation that intercepted the vertical plane of the transect was included. Proportional representation of each species in the canopy of a site or group of sites (see below) was calculated from diameter at breast height (dbh) Figure 1. Location of TES (triangles) and mixed pine-hardwood sites (circles) at Ft. Benning. Sites are exaggerated relative to map scale and overlapping sites are hidden. Dark triangles indicate two unique TES sites; outlined circles indicate five mixed pinehardwood sites closest in canopy composition to the TES sites (see results). 96 Southeastern Naturalist Vol. 5, No. 1 values. Groundlayer data for the TES and mixed pine-hardwood forest sites were combined by converting to percents and standardizing values for each species within each plot to 100%. Nonmetric multidimensional scaling (NMDS; SAS v. 9), based on Lance-Williams (Bray-Curtis) dissimilarities, was used to explore patterns in canopy and groundlayer vegetation among the stands. Canopy-tree ordination results were used to distinguish overstory groups. Analysis of dissimilarity (ANOSIM; SAS v. 9), based on Lance- Williams dissimilarities, was used to test differences in canopy and groundlayer vegetation composition among overstory groups (Clarke’s R: dissimilarity between groups > dissimilarity within groups; Clarke 1993). Stepwise analysis of dissimilarity (ANOSTEP; P. Minchin, Southern Illinois University, Edwardsville, IL, unpubl. data), based on Bray-Curtis dissimilarities, was used to identify the suite of tree species that contributed most to compositional differences among overstory groups. Indicator species analysis or ISA (Dufrêne and Legendre 1997) was used to identify the groundlayer species that best discriminated groups. Significance of the test statistic, IVmax , was tested by random permutation with 1000 trials (PC-ORD v. 4). Results Canopy composition NMDS revealed that canopy composition of two of the nine sites where sandhills TES are known to occur is distinct from that of the other seven sites (Figs. 1 and 2; TES sites noted by triangles, distinct sites by darker triangles). These seven sites appear to have similar composition to five of the 32 mixedpine stands (noted by outlined circles in Figs. 1 and 2). The remaining mixed pine stands form a large third “group” of sites (Fig. 2). The five mixed pinehardwood sites that are most similar in composition to the TES sites are not the closest to them on the landscape (Fig. 1); further, there was no relationship between physical distance and compositional similarity of the TES and mixed pine sites (regression results not reported). Weighted by abundances, community similarity was low (dissimilarity averaged 77%) between the two unique TES sites and the group that included the five other TES sites; similarity was somewhat higher (dissimilarity averaged 67%) between the latter five sites and the mixed pine-hardwood group. Stepwise analysis of dissimilarity (ANOSTEP, P. Minchin, unpubl. data) revealed abundances of Pinus palustris, P. taeda, and Quercus laevis contributed to differences among the three groups. Pinus palustris dominated the group of seven TES sites and five mixed pine-hardwood stands (Table 2). The two TES sites that were separated in the ordination were similar to the other seven TES sites in having Quercus laevis as an important canopy species, but they had low representation of Pinus palustris and higher representation of Quercus hemisphaerica and Q. margaretta. Pinus taeda was the more important pine in these two sites, and it was the dominant canopy species in the large group of mixed pine-hardwood stands (Table 2). Trees were significantly smaller (P = 0.0003) in the TES sites (mean dbh = 18.7 cm for the two unique sites and 23.7 cm for the remaining five sites) 2006 B. Collins, R. Sharitz, K. Madden, and J. Dilustro 97 Figure 2. NMDS of canopy composition among nine stands with sandhills TES (triangles) and 32 mixed pine stands (solid circles) at Fort Benning. Dark triangles = two “unique” TES sites; outlined circles = five mixed pine sites most similar to TES sites. Table 2. Canopy composition (species percentages) of the two TES sites with unique composition (2 TES), the other seven TES sites, the five mixed pine-hardwood sites that were grouped with the TES sites (5 Mixed), and all other mixed pine-hardwood sites (27 Mixed). Bold text indicate TE species that comprise ≥ 10% of the canopy vegetation; - indicate TE species not present within the grouped stands; only species that comprise ≥ 1% of the canopy in at least one group of stands are shown. 2 TES 7 TES 5 Mixed 27 Mixed Carya spp. - 1.1 0.2 3.6 Cornus florida L. - 0.4 0.7 1.2 Cratagus spp. - 0.1 - 0.1 Diospyrus virginiana L. - 0.1 - 0.1 Liquidambar styraciflua L. 2.4 0.3 0.6 5.7 Liriodendron tulipifera L. - - - 1.4 Pinus echinata P. Mill. 9.7 0.6 3.9 14.7 P. palustris P. Mill. 6.8 64.0 75.7 5.9 P. taeda L. 15.3 1.6 13.3 47.6 Quercus alba L. - - - 1.5 Q. hemisphaerica Willd. 17.3 0.2 - - Q. falcata Michx. - 1.2 1.5 6.6 Q. incana Bartr. - 0.6 - 0.4 Q. laevis Walt. 33.7 24.1 0.1 0.7 Q. margaretta Ashe 11.3 2.9 1.4 2.0 Q. marilandica Muenchh. 1.8 0.4 0.3 0.4 Q. nigra L. - - 0.3 5.2 Q. stellata Wagenh. - - 1.2 0.1 98 Southeastern Naturalist Vol. 5, No. 1 than in the five most similar (mean = 35.3 cm), or the remaining (mean = 37.2 cm), mixed pine-hardwood stands. Soils in the two unique TES sites averaged 4% clay and 94% sand content; the remaining TES sites averaged 8% clay and 89% sand. The five mixed pine-hardwood sites that were most similar to the TES sites included both sandy (75–90% sand) and clayey (11– 32% clay) soil categories, and both lighter and heavier military-use categories; however, all were in the group of sites that were prescribe-burned in both 2000 and 2002. In addition, these five sites are in land-management compartments that had burned (prescribed fire or wildfire) an average of 4.5 times over the preceding 20 years (1980–2000; data from the Fort Benning historical fire coverage). The remaining mixed pine-hardwood sites are in compartments that had burned an average of 3.8 times; the TES sites, with the exception of the two that grouped by themselves, had burned 4 times over the last 20 years. The two unique TES sites are in compartments that had only burned once between 1980 and 2000. Groundlayer composition NMDS revealed that groundlayer composition differed between the TES sites and the mixed pine-hardwood stands (Fig. 3). Vitis sp., Andropogon sp. (primarily A. virginicus L.), Myrica cerifera, and Vaccinium arboreum were the most abundant plants in the groundlayer of mixed pine-hardwood sites, while Rhus copallinun and Vaccinium species (V. stamineum, V. myrsinites, V. arboreum) were most abundant in the TES sites (Table 3). MDS scores among the TES sites primarily reflected differences in abundance of the Vaccinium species and increasing abundance, from lower right to upper left Figure 3. NMDS of groundlayer species (herbs and woody seedlings and sprouts) in nine sites (two sites overlap) with sandhills TES (triangles) and 32 mixed pine forests (circles). 2006 B. Collins, R. Sharitz, K. Madden, and J. Dilustro 99 in Fig. 3, of one of the TE species, Chrysoma pauciflosculosa (woody goldenrod). Indicator species analysis revealed a suite of taxa that discriminated between the TES and mixed pine-hardwood sites (Table 3). These included dominant taxa (Myrica cerifera, Pteridium aquilinum, Vitis sp.) and minor species (< 1% of vegetation) that had little overlap between the TES and mixed pine-hardwood sites (Table 3). We asked if the five mixed pine-hardwood stands that were similar to the TES sites in canopy composition (Fig. 2) were also more similar to them in groundlayer composition than were the other mixed pine stands. Average dissimilarity did not differ (P > 0.05); mean dissimilarity was 0.57 between the five sites and the TES sites compared to 0.58 between the remaining mixed pine-hardwood stands and the TES sites. The five mixed pine-hardwood sites also had more dominant species in common with the other mixed pine-hardwood sites than with the TES sites (Table 3). Only Pteridium aquilinum was more abundant in the five mixed pine-hardwood and TES sites than in the other mixed pine-hardwood sites (Table 3). Table 3. Groundlayer composition (species percentages) of the nine TES sites, the five mixed pine-hardwood sites that were grouped with the TES sites (5 Mixed), and all other mixed pinehardwood sites (27 Mixed) and indicator values (IV) of the taxa for the two ordination groups (TES sites and mixed pine-hardwood sites, Fig. 3). Bold text indicate TE species that comprise ≥ 4% of the vegetation or have significant IV (P < 0.01); - indicates TE species not present within a group of stands; only species that comprise ≥ 2% of the vegetation in at least one group of stands or have highly significant IV (P ≤0.001) are shown. Abundance (%) TES 5 Mixed 27 Other IV P > 2% of vegetation Andropogon sp. 1.4 10.7 5.9 65.4 0.4660 Callicarpa americana L. 3.6 2.0 1.3 10.6 1.0000 Collinsonia serotina Walt. 1.0 2.4 0.3 9.4 0.7165 Chrysoma pauciflosculosa (Michx.) Greene 3.4 - - 44.4 0.0012 Elephantopus tomentosus L. 1.1 2.3 1.9 75.0 0.0003 Gelsemium sempervirens St. Hil. 1.8 0.8 2.5 66.3 0.2631 Myrica cerifera Small - 4.8 5.0 78.1 0.0007 Pityopsis graminifolia Nutt. 1.5 1.8 3.2 55.1 0.6606 Pteridium aquilinum Kuhn 3.6 5.6 2.3 77.4 0.0011 Rhus copallinum L. 4.6 3.9 3.7 88.5 0.1146 Rubus sp. 2.9 2.3 3.4 99.5 0.0001 Sorghastrum secundum Nash 3.1 - 0.2 100 0.0001 Tephrosia virginiana Pers. 1.3 3.3 0.5 44.3 0.0598 Toxicodendron radicans Kuntze 3.3 1.8 2.2 88.6 0.0003 Vaccinium arboreum Marsh. 4.1 5.6 5.2 88.3 0.0763 Vaccinium elliottii Chapman 2.2 2.2 2.9 65.6 0.0051 Vaccinium myrsinites Lam. 4.3 1.1 0.7 33.3 0.3667 Vaccinium stamineum L. 5.1 2.9 3.8 88.6 0.0509 Vitis sp. 2.0 4.4 9.9 90.6 0.0001 < 2% vegetation Amphicarpa bracteata Fern. - 1.2 1.0 75 0.0005 Aristida sp. 0.9 0.6 - 66.6 0.0001 Gaylussacia dumosa Torr.& Gray 1.0 - - 77.8 0.0001 Minuartia caroliniana Mattf. 0.9 - - 66.7 0.0001 100 Southeastern Naturalist Vol. 5, No. 1 We also examined patterns in the distribution of tree regeneration (seedlings and sprouts < 1.4-m height) in the groundlayer. NMDS revealed the same pattern as shown by the total groundlayer (herbs and tree seedlings/sprouts combined, Fig. 3). Sandhills scrub oaks, Quercus laevis (two sites), Q. hemispherica (two sites) and Q. margaretta (one site), were the most abundant seedlings/sprouts in the TES sites. In the mixed pine-hardwood stands, Liquidambar styraciflua was the most abundant species in 17 sites, Quercus stellata in three sites, and Ilex glabra (L.) Gray and Pinus taeda in two sites. Discussion The relative abundance and species present of pines (Pinus taeda, P. palustris, P. echinata) and oaks (Quercus laevis, Q. hemisphaerica, Q. margaretta) associated with dry or sandhill communities (Christensen 1988, Jacqmain et al. 1999) contributed to canopy composition differences among sites with sandhills TE species and 32 “matrix” mixed pine-hardwood stands at Fort Benning. Pinus palustris (longleaf pine) and Quercus laevis (turkey oak), were dominants in seven of the nine sites with sandhills TES, while P. taeda (loblolly pine), Q. laevis, Q, hemisphaerica, and Q. margaretta were abundant in the remaining two TES sites. Longleaf shared dominance with loblolly pine in five mixed pine-hardwood stands, while loblolly and P. echinata (shortleaf pine) were most abundant in the remaining 27 stands. Thus, the five mixed pine-hardwood sites are similar to seven of the TES sandhills sites with respect to the pine component of the canopy, but they lack the dominant oaks that re-sprout after fire and characterize dry, firemaintained sandhills scrub communities (Jacqmain et al. 1999). The two distinct TES sites lack a strong longleaf component, but all TES sites are characterized by sandhills oaks. The sandhills sites also are distinct from the mixed pine-hardwood sites in structure; trees are smaller (smaller average dbh). The two distinct TES sites had the smallest trees (dbh averaged 18.7 cm compared to 23.7 for the remaining TES sites). These sites also had sandier soil (average 94% compared to 89%), but less open canopy (60% compared to 38%), and had burned less frequently over the past 20 years (once compared to 4 times) than the other TES sites. Thus, the two distinct TES sites are among the sandiest, least productive, and least frequently burned (over the last 20 years) sites we surveyed on Fort Benning. Composition of the forest groundlayer (herbs and woody species < 1 m) and regeneration (woody seedlings and sprouts < 1 m) overlapped little between the entire group of 32 mixed pine-hardwood sites and the TES sites. Vaccinium was abundant in all sites, but species presence and abundance varied between the mixed pine-hardwood and TES sites, and among the TES sites. In addition to Vaccinium spp., one of the TE species, Chrysoma pauciflosculosa, was a dominant in the TES sites, while Andropogon sp. (primarily A. virginicus) and Vitis sp., both of which are found in sandhill communities, pine plantations, and upland hardwood forests (Workman and McLeod 1990), were most abundant in the mixed pine-hardwood stands. The 2006 B. Collins, R. Sharitz, K. Madden, and J. Dilustro 101 most common dominant seedling/sprouting species in the mixed pine-hardwood sites, Liquidambar styraciflua (most abundant species in 17 sites), is associated with old fields, pine plantations, and upland hardwood forests, but may be less common in sandhills communities (Workman and McLeod 1990) and was not abundant in the TES sites. Sandhills oaks, Quercus laevis, Q. hemispherica, and Q. margaretta, were dominants of seedling/sprout regeneration in the TES sites. These results suggest the groundlayer and regenerating canopy trees show more “site fidelity” than do the canopy trees and provide further evidence that the TES sites are on a different successional trajectory from the mixed pine-hardwood sites. Some of the differences in groundlayer composition between the TES and mixed pine-hardwood sites, including the greater abundance of sandhills oaks in the regeneration pool of the TES sites, may due to soil texture. Dilustro et al. (2002, 2005) have shown linkages among soil texture, soil moisture cycles, and groundlayer composition at Fort Benning. Soil moisture and physical properties, including texture, have been shown to be related to the distribution of sand ridge vegetation and sandhills oaks (Q. laevis, Q. margaretta) relative to intermediate and wet coastal plain sites (Jacqmain et al. 1999). Composition of the groundlayers of sandhills and mixed pine-hardwood forest at Fort Benning also may be related to historical land use, including the fire regime, or the interaction of land use and soil characteristics. Species, such as Pityopsis graminifolia, that show increased presence or biomass after fire (Anderson and Menges 1997) were common throughout the mixed pine-hardwood and TES sites. In addition, some forbs, such as species of Eupatorium (primarily E. capillifolium (Lam.) Small and E. compositifolium Walt.), Tephrosia virginiana, and T. spicata (Walt.) Torr. & Gray, have a positive relationship with the frequency of recent fires (1980–2000) in the mixed pine-hardwood sites (B. Collins, unpubl. data), and were common in the sandhills TES sites. Overall, our results indicate communities with the TE species differ from the “matrix” mixed pine-hardwood stands surveyed in the relative abundance of particular species in the groundlayer and the abundance of sandhills oaks in the canopy and regeneration pool. The mixed pine-hardwood stands, including those with abundant longleaf pine, do not appear to be regenerating toward typical sandhills vegetation while the TES sites, including those with abundant loblolly pine, have distinct “sandhills” groundlayer composition that includes Vaccinium species and abundant sprouts and seedlings of sandhills oaks. These results suggest that conservation of sandhills groundlayer communities, including TES and regenerating canopy dominants, may best be achieved by maintaining existing sites at Fort Benning. Acknowledgments This research was supported by Department of Defense Strategic Environmental Research and Development Program (SERDP) SI-1302, SERDP Ecosystem Management Project CD-1114E-00, and Financial Assistance Award Number DE-FC09- 96SR18546 between the US Department of Energy and the University of Georgia. For their help with this research, we thank: Don Imm, Lisa Duncan, James Mason, Marlon 102 Southeastern Naturalist Vol. 5, No. 1 Smith, Erin Kough, Thomas Lee, Mary Harper, Jason Baxley, Aaron Campbell, Linda Lee, Leslie Zorn, Aubree McArthur, Aaron Knowles, and the natural resources staff at Fort Benning. Hal Balbach provided thoughtful comments on the manuscript. Literature Cited Anderson, R.C., and E.S. Menges. 1997. Effects of fire on sandhill herbs: Nutrients, mycorrhizae, and biomass allocation. American Journal of Botany 84:938–948. Christensen, N.L. 1988. Vegetation of the southeastern coastal plain. Pp. 317–364, In M.G. Barbour and W.D. Billings (Eds.). North American Terrestrial Vegetation. Cambridge University Press, Cambridge, UK. 434 pp. Clarke, K.R. 1993. Nonparametric multivariate analyses of changes in community structure. Australian Journal of Ecology 18:117–143. Dilustro J.J., B. Collins, L. Duncan, and R. Sharitz. 2002. Soil texture, land use intensity, and vegetation of Ft. Benning upland forest sites. Journal of the Torrey Botanical Society 129(4):80–297. Dilustro, J.J., B. Collins, L. Duncan, and C. Crawford. 2005. Moisture and soil texture effects on soil CO2 efflux components in southeastern mixed pine forests. Forest Ecology and Management 204:87–97. Dufrêne, M., and P. Legendre. 1997. Species assemblages and indicator species: The need for a flexible asymmetrical approach. Ecological Monographs 67:345–366. Georgia Department of Natural Resources. 2004. Natural Heritage Program. Special concern plant species in Georgia. Available at: http:// georgiawildlife.dnr.state.ga.us. Jacqmain, E.I., R.H. Jones, and R.J. Mitchell. 1999. Influences of frequent coolseason burning across a soil moisture gradient on oak community structure in longleaf pine ecosystems. American Midland Naturalist 141:85–100. Johnson, J.H. 1983. Soil survey of Muscogee County, Georgia. US Department of Agriculture, Soil Conservation Service, Columbus, GA. 130 pp. Keys, Jr., J.A., C.A. Carpenter, S.L. Hooks, F.G. Koeneg, W.H. McNab, W.E. Russell, and M.L. Smith. 1995. Ecological units of the eastern United States— first approximation. Technical Publication R8-TP 21. Map (scale 1:3,500,000). US Department of Agriculture, Forest Service, Atlanta, GA. Radford, A.E., H.E. Ahles, and C.R. Bell. 1968. Manual of the vascular flora of the Carolinas. The University of North Carolina Press, Chapel Hill, NC. 1183 pp. Skeen, J.N., P.D. Doerr, and D.H. van Lear. 1993. Oak-hickory-pine forests. Pp. 1– 34, In W.H. Martin, S.G. Boyce, and A.C. Echternacht (Eds.). Biodiversity of the Southeastern United States Upland Terrestrial Communities. John Wiley and Sons, Inc., New York, NY. 502 pp. Ware, S., C. Frost, and P. Doerr. 1993. Southern mixed hardwood forest: The former longleaf pine forest. Pp 447–493, In W.H. Martin, S.C. Boyce and A.C. Ecternacht (Eds.). Biodiversity in the Southeastern United States: Lowland Terrestrial Communities. John Wiley and Sons, New York, NY. 502 pp. Workman, S.W., and K.M. McLeod. 1990. Vegetation of the Savannah River Site: Major community types. Savannah River Site National Environmental Research Park Program SRO-NERP-19. 137 pp.