Chapter V - Presettlement Piedmont Savanna Ecology

Rather than an unbroken forest, the Southern Piedmont could have been described in the years of first European contact, as a patchwork of adjacent, dissimilar communities with often indistinguishable boundaries between them. A transect of a several miles could cross many variations of the landscape, exhibiting various oak-hickory-pine forest scenarios, with the occasional appearance of open grassy savannas, all resulting from complex patterns of physiography and landscape disturbances. Understanding presettlement Piedmont savanna ecology is essential, not only in explaining the existence of savanna communities, but also to serve as a model for any related reconstruction of such an environment. In order to understand the condition of this landscape, examples from the nearest documented savanna and savanna-like landscapes will be compared with remnants found within the Piedmont and historical descriptions.

Historical descriptions of presettlement savannas throughout the southern Piedmont were made by explorers including John Lederer in Virginia and the Carolinas in the 1670s, John Lawson in the Carolinas in 1701, William Byrd on the dividing line between North Carolina and Virginia in 1728, Bishop Spangenberg in North Carolina in 1752, William Bartram in South Carolina, Georgia and Alabama in the 1770s, and Benjamin Hawkins in Georgia and Alabama in the 1790s. Figure 5.1 shows a rough approximation of the areas these explorers traversed. As original accounts, their observations will be used throughout this and the next chapter to help explain savanna composition. To the best of my knowledge, all observations cited were made in the southeastern Piedmont region. Consideration will be given to the unique perspectives they offer due to their varied backgrounds as explorer, surveyor, religious leader, farmer, naturalist, Indian agent, and as gentlemen of their times. Except perhaps for Bartram, the naturalist of the group, these men were likely to have viewed the landscapes they encountered in terms of commodity (Cronon 21). Their accounts might be biased in that they may remark only on the unusual, discuss only areas traversed by major trails, and they may differ in their methods and knowledge of the natural elements encountered (Nuzzo 9). Nevertheless, the descriptive evidence they offer of this extinct landscape does provide valuable comparative information, even if some ecological details are distorted in their descriptions.

Figure 5.1 Approximation of areas traversed by Piedmont explorers.

 

Savanna Distribution and Function

Savanna studies in the Midwest look at soil type, existing vegetation and historical information to locate the range of presettlement oak savannas (Nuzzo 10). In this thesis, I will look at similar elements; because of limited information, some comparisons will be made to savanna landscapes in the Midwest.
 

Physiography

The Piedmont (shown in Figure 2.1), which takes it name from the Italian Piemonte, or foothills (Godfrey 11), is the oldest, most highly eroded of the provinces of the Appalachian orogeny. It lies between the Atlantic Coastal Plain to the Southeast and the Appalachian mountain chain to the northwest, and between them extends from New Jersey to Alabama. Toward its southern end, the lower elevations along the fall line, which divides it from the Coastal Plain, are approximately 600 feet, and higher elevations near the Appalachians are approximately 1500 feet. It has a rolling topography of broad, convex ridges and is irregularly and frequently dissected by generally southeast running drainages in relatively narrow floodplains (Skeen, Doerr and VanLear 2).

The topography of the savanna landscapes in historical descriptions is a reflection of the general lay of the Piedmont uplands: gently rolling, flat, or sometimes steep; and level lowlands along streams. Of the various configurations in which they were described by early explorers, upland settings, with both wet and dry characteristics, seem to be most common, perhaps in part because they were the location of travel routes, but also because the savannas of relatively dry uplands were most removed from firebreaks and were frequently burned (Cowell 59). Because grassland maintenance is directly related to topography, flat to rolling Piedmont lands would be the most ideal for savannas, since rougher, more dissected topography favors development of woody vegetation over grassland (Wright and Bailey 82). Savanna landscapes were also described in lowlands usually along rivers and streams and often associated directly with Indian settlements. Historic citations illustrate the variety of settings. Near the Roanoke river in Virginia:

The Countrey here, though high, is level, and for the most part a rich soyl, as I judged by the growth of the Trees; yet where it is inhabited by Indians, it lies open in spacious Plains, and is blessed with very healthful Air,... (Lederer 24)

An upland southwest of the Yadkin river in North Carolina:

We travell'd, this day, about 25 Miles, over pleasant Savanna Ground, high, and dry, having very few trees upon it, and those standing at a great distance. The Land was very good, and free from Grubs or Underwood. (Lawson 51)

The country above the fall line up to the headwaters of the Flint and Ocmulgee rivers in Georgia:

... open, flat, land, the soil stiff, the trees post and black oak, all small. The land is generally rich, well watered, and lies well, as a waving [hilly] country, for cultivation. (Hawkins 20)

Near the fall line west of the Oconee river:

...a pleasant territory, presenting varying scenes of gentle swelling hills and levels, affording sublime forests, contrasted by expansive illumined green fields, native meadows and Cane breaks... (Bartram 307)

Uplands, especially hilltops, could present more xeric type savanna settings. A xeric setting north of Augusta, Georgia:

...vast forests, expansive plains and detached groves; then chains of hills whose dry, barren summits present detached piles of rocks. (Bartram 263)

Savanna landscapes with wetland attributes could also be found in the Piedmont, perhaps similar in nature to locations where stream headwaters emerged from glades, as Hawkins saw in Alabama (47), or as Lederer describes in Virginia or North Carolina:

To heighten the beauty of these parts, the first Springs of most of those great Rivers which run into the Atlantick Ocean, or Cheseapeack Bay, do here break out, and in various branches interlace the flowry Meads, whose luxuriant herbage invites numerous herds of Red Deer (for their unusual largeness improperly termed Elks by ignorant people) to feed. (Lederer 34,35)

In the next century, Byrd noted another type of open upland wetland, the "high-land pond" extending for about two miles as they traveled in the area west of Danville, North Carolina. He describes a:

...dead levil, without the least declevity" that was moist in many places and produced an abundance of grass. the woodsmen called these flat areas, "highland ponds" and they are exellent areas to forage worn out horses. (Byrd 212)

Such upland wetlands might be represented in the remnant community of Link Bog (Radford 111-112) at the base of the Brushy mountains in Iredell County, North Carolina. There are also many illustrations of lowland savanna settings particularly within the lands of the Creek Indians in Georgia and Alabama. This, the view from the banks of the Oconee river near Milledgeville:

This flourishing grove was an appendage of the high forest we had passed through, and projected into an extensive, green, open, level plain, consisting of old Indian fields and plantations, being the rich low lands of the river, and stretching along its banks upwards to a very great distance, charmingly diversified and decorated with detached groves and clumps of various trees and shrubs, and indented on its verge by advancing and retreating promontories of the high land. (Bartram 307)

Though there are no historical maps accurately showing the distribution of savannas in the Piedmont, a demonstration of its pattern in the area of Winston-Salem, North Carolina may potentially exist in a survey map drawn for the Moravian pioneer Community of Bethania in 1766. Shown in Figure 5.2, the white areas between stream drainages could very well represent existing grassland or "prairie" while the shaded areas along creeks may represent forested slopes and bottoms. Support for this idea is illustrated by Keever citing Jethro Rumple's 1881 Rowan County in which he gives the account of a recently deceased respected citizen, who would have seen the North Carolina "prairie" landscape at the end of the 1700s. The "old timer" says he remembered open prairies on the uplands between creeks covered with grass and wild pea vines where deer mingled with their cattle (Keever 39-40).

Comparison can also be made to a map of prairie vegetation in Missouri (Figure 5.3) where the light color representing prairies occupies the spaces between the dark colors representing wooded river valleys. Though at a different scale, there seems to be a remarkable similarity in the landscape patterns of the two maps. Further exploration of this concept might reveal models for savanna landscape patterns.

Hydrology and Climate In the Hypsithermal period, climate expansion may have caused an eastward migration of prairie species into the Piedmont. The more recent Piedmont climate is cooler and wetter, however, and the scattering of prairie-type remnants across the Southeast does not argue strongly for climate as a control over their development (Deselm and Murdock 89,90,92). Annual precipitation over the region ranges from 32-48 inches (Skeen, Doerr and VanLear 3). In contrast to that, eastern prairie regions from Texas to Indiana, where forest-to-prairie transitions occur, have a lower annual precipitation of 29-39 inches. There are, however, other factors that influence grassland development in the Piedmont. Generally, all North American grasslands have a wet period followed by a period of drought conditions (Risser et al. 10). In the Piedmont, precipitation is relatively evenly distributed throughout the year without significant deficit periods (Skeen, Doerr and VanLear 3). The Ultisol and Alfisol soil types of the region, however, are characterized as "dry part of the time" during the "warm season of the year" (USDA SCS Principal Kinds of Soils) reflecting the fact that a dry soil season does occur. Soil moisture depletion in the summer occurs to a depth of 66 inches under forest cover and to 30 inches under old-field herbaceous cover, while summer rainfall seldom infiltrates below 30 inches. Partly responsible for drought conditions, grasses can prevent up to .4 inches of rainfall from reaching the soil, precipitation which is then lost to evaporation (Wright and Bailey 25). Other factors responsible for moisture depletion are high summer evaporation and evapotranspiration rates. Recharge of groundwater occurs in the winter months (Skeen, Doerr and VanLear 5). Other factors that help maintain open areas result from climate and microclimate. They include frost damage, soil extrusion by freezing, and moisture stress from higher maximum temperatures and greater wind speeds (Deselm and Murdock 89). Seasonal differences in temperature in the southern Piedmont range from a summer average high of 89( to 99( F, to a winteraverage low of 30( to 40( F with an average of 240 frost free days (Skeen, Doerr and VanLear 3). These temperatures are not unlike those found in prairie regions of the southern central United States.

In the savanna microclimate, sunshine, wind speed and rainfall can either increase or decrease as a result of canopy structure. As canopy height and density increase, wind speed and light intensity decrease near the ground, soil temperature decreases during the growing season, and differences between air temperature within and above the canopy decrease (Risser et al. 149-150).

Figure 5.2 "Map of Bethania Tract, 1766" (Fries 374).

Figure 5.3 Landscape patterns of Missouri prairies and woodlands (Schroeder 20).

 
Soils

The distribution of Piedmont savannas cannot be determined solely by remnants or by incomplete historical information. As a more permanent, more quantifiable entity, soils and soil relationships can be examined for evidence of presettlement distribution. Basic soils, droughty soils, and temporarily saturated soils emerge as being particularly important conditions.

The most widespread soil order in the southern Piedmont is Ultisols while significant areas of Alfisols also occur (Skeen, Doerr and VanLear 4). Both soil orders have clay subsurface horizons which are usually moist but may dry out in the warm season of the year, while Ultisols are more acidic and Alfisols are more basic (USDA SCS Principal Kinds of Soils). Soils of the central United States grasslands are usually deep, fertile, neutral to basic, and high in organic matter. Most true grasslands are found on soil orders of Mollisols and Aridisols (Risser et al. 4) though there are many other orders in the prairie region. Within the south, mollisols are associated with the Kentucky barrens and the very small areas of serpentine barrens in the Piedmont. Notably, all Southern grassland communities, except for balds and shale barrens, include Alfisols among the orders on which they occur. These grasslands shown in Figure 2.1, include the Black Belt of Alabama, the barrens of Kentucky and Tennessee, the coastal prairies of Louisiana and Texas, and the serpentine barrens of the Piedmont (Deselm and Murdock 90-92). This suggests a relationship involving the basic soil characteristics of Alfisols (which develop over gabbro, diabase-basalt and serpentine [Radford 23]), to grassland development in areas of the Piedmont. Figure 5.4 shows the distribution of Alfisols in the Piedmont. Many of the savanna or "prairie" landscapes described by Lederer, Lawson, Byrd, Spangenberg, Foote, and others in the Carolinas and Virginia seem to have a physiographic distribution related to these Alfisols.

Another aspect of soils and grassland development is the texture and behavior of the soil. The Blackland Prairie of Texas and the Black Belt Prairie of Alabama occur in areas with enough rainfall to support forest; however, due to controlling conditions of fine-textured soil, seasonal drought conditions favor grassland development. In the Crosstimbers oak savanna in Texas, post and blackjack oak growth is found on coarser- textured soils while nearby fine-textured soils support fewer trees and more grasses (Risser et al. 60,64). In the Piedmont, fine textured Iredell soils are credited as one of the major factors in the development of the former "Piedmont Prairie" of North Carolina (Horan 9). The map Soil Associations and Land Use Potential Of Georgia Soils points out that Iredell and other fine textured soils of the Piedmont are often of basic rock origin. Montmorillonite is the major clay mineral in Alfisols and accounts for their high shrink-swell capacity and plastic properties (Skeen, Doerr and VanLear 4). Being virtually impermeable when wet, they can cause a perched water table on concave topography, and during dry periods become extremely dry (Waldrop Establishment 5), even developing wide cracks and breaking roots (Horan 9). Soil freezing is also more severe in these fine- textured soils (Wright and Bailey 29). High shrink-swell soils can also be acidic, such as those developed over acid rock found along the Carolina slate belt, shown on the General Soil Map Of North Carolina. Distribution of these plastic soils, shown in Figure 5.5, may be another clue to the distribution of presettlement savannas in the Piedmont.

Most upland soils in the Piedmont are Ultisols including the widespread Cecil, Madison, Musella, and Pacolet series (USDA, Principal). Although these soils are generally conducive to oak-hickory forest development (Skeen, Doerr and VanLear 4), presettlement influences, especially that of fire, would have kept these woodlands open. (Deselm and Murdock 89). Bartram and Hawkins offer many descriptions of savanna-like scenes as they traveled through areas shown to be Ultisols on the USDA map Principal Kinds of Soils: Orders, Suborders, and Great Groups. It may be reasonable to presume, however, that upland areas of Alfisols, particularly with shrink-swell soils, were more likely to attain the openness approaching that of the prairie than areas of Ultisols. Whatever the degree of openness, soils were only one factor. Savanna openness would be more influenced by drought and/or flooding and the disturbances of grazing/browsing and particularly fire, especially on areas of deeper soils (Deselm and Murdock 93).

Figure 5.4 Distribution of Alfisols in the Piedmont (USDA, Soils of the Southern US).

Figure 5.5 Distribution of high shrink-swell soils in the Piedmont based on information available from statewide soil maps. Upland Piedmont soils included in associations described as moderate to extremely plastic include Iredell, Enon, Wilkes, Mecklenburg, Helena, Whitestore, Creedmore, Cecil, Appling, Lignum, Winnsboro, Davidson, and Lloyd (Alabama; Georgia; South Carolina; USDA General).

Disturbance Mechanisms

Of presettlement Piedmont disturbances, such as extreme climatic events, Indian clearing, and grazing, fire was the most important in facilitating grassland development in an area where there is more than enough precipitation to grow forests in only a few decades. The often large-scale landscape patches of savanna openings would have required large disturbances for their development. Smaller-scale forest canopy openings of gap-phase regeneration just don't add up to produce the prairie-like landscapes described by Foote in North Carolina. Large-scale catastrophic disturbances, such as the blow-downs from hurricane Hugo, or major ice storms, followed by intense summer fires, could have created ideal conditions for near pure pine establishment (Skeen, Doerr and VanLear 14) or open savanna, if pine seed sources were unavailable (Joern and Keeler 158; Keever 40-41). Of the many disturbance types, fire was undoubtedly the most influential.
 

Fire

Indian-Caused Fires

There is much debate and many questions to answer on causes and effects of fire in the presettlement Piedmont. Different fires caused for different reasons would have had different ecological effects (Bryant, McComb, and Fralish 187), so it is important to understand what factors caused fires, to what extent, and with what result. Unfortunately, there is only indirect evidence of the likely frequency and intensity of historic Piedmont fires. Lake sediments which could provide such data are not available because of Piedmont physiography. In the absence of clear evidence, it is difficult to attribute the apparent fire disturbance regime to either "natural" or anthropogenic causes (Cowell 61, 138). They are examined here, however, being the only two causes to choose from for what was certainly a common and influential disturbance.

That both causes were present was indicated by William Bartram in 1773 as he wrote of fire that "It happens almost everyday throughout the year, in some part or the other, by Indians for the purpose of rousing game, as also by lightning" (Wright and Bailey 363). There have been cases made for both causes, though it seems historical evidence favors Indians as the dominant, though certainly not sole, fire causing agent. Byrd, Lawson, and Spangenberg all mention Indian burning though they do not mention lightning fires. When Byrd travels under skies smoky from wildfires, he attributes it to "the firing of the woods by the Indians" caused by their abandoned campfires (Byrd 218). Lawson refers to driving game as the reason for Indians burning the woods (Lawson 215). Spangenberg echoes European sentiments about fire as he describes a tract of mostly pine and some hardwood as "ruined" because Indians burning it to drive game had eliminated all the smaller trees (Fries 49). For the Indians there were many reasons for burning besides driving game to a place for an easy kill. They burned to encourage new tender growth for attracting grazers and browsers to hunting areas. They burned to clear forest patches for agriculture by killing trees: with the dead trees standing, their roots intact to hold the soil and provide long term fertilizer, annual dormant season burns of the agricultural fields released additional mineral nutrients for fertilizer. They burned the woods to clear underbrush and open them for ease of travel and line of sight for defense against attack; to reduce catastrophic fire hazard from accumulated litter; and to get rid of biting insects such as fleas, chiggers, lice, ticks, mosquitoes, and spiders. They also burned the woods accidentally from unwatched cooking fires (Silver 49-62).

Depending on the reason, there were different times of year and different areas and sizes of land that were burned, but, in general, annual burning was the rule for Indians. Burning activity for agriculture, defense, ease of travel, fire protection, and insect pest reduction was concentrated in the area of settlements, perhaps within an area of a few miles, and would have occurred at various times of year depending on the purpose and available fuel buildup. Burning for agricultural improvement and most other burns would have occurred in the dormant season, while burning to reduce insect populations might have occurred in the summer. For improving game forage and ease of travel, and driving game to the kill, burning would have been conducted on a larger landscape scale. For the most part, as suggested by explorers' accounts, late fall and winter were the most common burning time. That was the primary hunting season and enough dry fuel had accumulated by that time to carry a fire. According to legend, in the area of Hart County, Georgia, fires were set in the late fall when the wind was from the southwest burning an area across the center of the county to the Tugaloo River (Kaufhold 2). To attract game to fresh forage, however, burns could have been conducted at other times of the year. Accidental fire might occur at any time, but most likely when fuels were dry and plentiful and the need for camp fires great as in the fall of the year. Where burning was not frequent enough to reduce fuel loads, and not infrequent enough to allow mature shady forests to develop, fires might have a very destructive effect on forest landscapes. In an area where no tree of "any bigness" could be seen due to "fire or catapillars," Byrd succinctly discusses the effects of fire and why it "makes such a Havock in these lonely parts".

The woods are not there burnt every year, as they generally are amongst the Inhabitants. But the dead Leaves and Trash of many years are heapt up together, which being at length kindled by the Indians that happen to pass that way, furnish fewel for a conflagration that carries all before it." (Byrd 228)

Large areas acting as uninhabited wilderness buffers between chiefdoms may have had very infrequent fires and better-developed forests. The area of Jackson and Madison Counties, north of Athens, Georgia, was sparsely populated and only infrequently used as a hunting ground between the Creek and Cherokee nations. Also, most Native inhabitants had removed to more westward locations by the time Bartram saw the area. Given the lack of settlement, the area was probably predominantly mature, mixed hardwood and pine forest (Pluckhan 127-129) similar to the "high forests" of the "great ridge" area to its south (Bartram59). Ten to twenty years or more might be the fire frequency for such a predominantly fire-tolerant oak-pine forest (Cowell 59). It is also interesting to note that as Bartram traveled westward into more heavily populated Creek territory, his descriptions of the landscape include a great deal more of "expansive savannas".

Lightning Caused Fires

Komarek points out that lightning storms are more frequent in the Southeast than any other part of the United States (Barden, management 5). Cecil Frost, at the 1995 Cullowhee Conference, showed estimates of 15,000 lighting strikes per storm producing 400 strikes per minute. He said it has been estimated that there are 1,100 lightning fires a year in Florida, based on reports of fires sparked on the dry areas at the edges of storms. The usual accompaniment of rain often reduces the intensity of these fires; however, fires can smolder for days till enough drying occurs, then flare up again.

Historic records indicate an average wildfire interval of 2-3 (Cowell 58) or 3-5 (Hughes 155) years in the Coastal Plain, where fire compartments are large and lightning fires frequent. At the 1997 Native Grass Conference in Asheville, North Carolina however, Lawrence Barden, arguing for more anthropogenic fire causation, pointed out that Piedmont uplands, frequently dissected by drainages, have fire compartments too small to allow lightning fires to frequently burn large areas as they do in the Coastal Plain.

Lightning fires depend on atmospheric conditions for their ignition source and fuel moisture conditions for their ignition and spread. April, before trees leaf out, is the driest month of the year, making fire danger high. The cool dry air masses typical of the spring, however, are not conducive to generating thunder storms. Fall conditions are similar to those of the spring. In the summer, as lightning storm frequencies increase, leafing of vegetation causes moisture levels to rise, and fire danger to decrease. In the savanna setting, however, the drying effects of an open canopy continue into the summer season, often combining with the droughty soils of such sites to keep fire danger high as thunderstorm activity increases. This would suggest that lightning caused fires were more frequent in savanna settings than forests, and were potentially responsible for their maintenance (Barden, management 5-6).

Because there is only indirect evidence of the likely frequency and intensity of historic Piedmont fire, and even less direct evidence of the causes of such fires, it is difficult to attribute definitively the apparent fire disturbance regime to either non-human or anthropogenic causes. (Cowell 138).

Fire and Vegetation

Fire is very effective at changing or maintaining vegetation composition. A "conflagration that carries all before it" (Byrd 228) can happen when conditions are right and fuel is abundant, having collected over the years. Some Piedmont areas burned frequently, some not for many decades. Ten to twenty year frequencies may have been common in parts of the Piedmont (Cowell 58, 59) causing a "Smoky Bear" nightmare scene as observed by Byrd, turning many acres into dense, woody thickets after several years' growth (Byrd 212, 228; Nuzzo 8). The more frequent fires, potentially set by Indians, with lower fuel loads, would have been less intense, mainly consuming the accumulated understory litter, and causing saplings to be pruned back to their roots. It is likely that frequent fires of this kind were responsible for maintaining the openness of the presettlement Piedmont savanna over most soil types, while perpetuating prairie-like composition over shrink-swell soils. With such a variety of fire related vegetation types across the Piedmont landscape, it would be difficult to argue that the presettlement vegetation structure was a climax community (Cowell 144).

The average surface temperature of the less intense ground fires might be similar to the 216(-498( C observed in prairie fires. These temperatures are more than enough to kill mesophytic plants, which are usually killed at only 122(-131( F. Many species, however, are adapted to survive such extreme conditions and can dominate an environment where fire is a common phenomenon. Such species might sport protective armor, employ special seeding characteristics, or have the ability to resprout vigorously. Trunks of trees, including many pines and oaks with bark half an inch thick, will incur very little damage in ground fires. On the other hand, trees with thin bark, such as the beech, would not survive in a fire prone environment. Seed characteristics such as the ability to sprout in bare mineral soil, windblown dispersal, and the ability to tolerate heat, allow many species to prosper in fire environments. Grass seeds can tolerate up to 241( F for 5 minutes. Also, grass fire temperatures have little direct effect on buried seeds (Wright and Bailey 11, 16, 18). The ability to resprout from the roots gives a competitive edge to many herbaceous species, especially grasses, as well as some adapted woody species such as oaks. In a Wisconsin brush prairie, oak grubs (oaks that repeatedly resprout from well developed roots) have persisted for over 100 years with annual fires (Nuzzo 8). Midwestern trees such as ash, elm, oak, and sumac can resprout from the roots and are not easily removed by fire (Joern and Keeler 62). Hickory, like many oaks, also sprouts from the roots after burning (Bryant, McComb and Fralish 187). Shortleaf pine, due to its sprouting ability, tolerates more frequent fires than does loblolly (Wright and Bailey 365) and can dominate with 5 year fire intervals, while loblolly prefers 10 years. In the Piedmont, 3-6 year fire intervals prevent most hardwood establishment in pine plantings. Thus fires, especially in the canopy, can promote pure pine stands (Cowell 58).

Burning also encourages an herbaceous ground cover under canopies and into open spaces (Skeen, Doerr and VanLear 6). This can be seen in the modern Coastal Plain woodlands, where many grasses and leguminous herbs increase after burning. These include grass genera such as Andropogon, Panicum (Wright and Bailey 372, 376), and Arundinaria (Hughes 150), and leguminous genera such as Cassia, Clitoria, Galactia, and Lespedeza. Also in the Coastal Plain, occasional fires prune and stimulate fruiting of huckleberries, blackberries and dewberries, and annual winter fires increase cover of blackberry and flameleaf sumac (Wright and Bailey 376, 377).

Fire frequency and seasonal timing are very important in determining the composition of grassland communities exposed to fire. For instance, increasing fire frequency in the spring from triennial to annual increases total plant production, increases warm season grasses and forbs, and decreases cool season annual production. After several years, however, annual fires can result in decreased production as nitrogen losses add up (Risser et al. 428). Spring burning is most commonly used in tallgrass prairie management because it increases productivity in warm season plants, controls exotic cool season species, and does not greatly influence species composition over the short term. Late spring burns done consistently, however, can reduce species diversity by continually selecting warm over cool season species (Joern and Keeler 65). According to True Prairie Ecosystem (Risser et al. 413-432), early spring burning increases stands of little bluestem and late spring burning increases big bluestem and Indian grass. Several studies show that spring burning increases the amount of flowering, especially in big bluestem, and in most species, while mowing and removing clippings also increases it, but only half as much. Seed production is higher mainly the first two years after a burn, and seeds have a better germination rate (413, 414). Spring fires can also increase soil water content for the growing season, due in part to the reduced moisture needs of cool season grasses. For grazing, late spring (more than early or mid-spring) burns maximize forage when done once every 3 - 4 years (418, 431, 432). Probably most common in the presettlement Piedmont, dormant season fires in the fall and winter remove litter, allow the soil surface to heat up sooner, and favor cool season species. Though grass cover increases after winter fires (414), productivity of big and little bluestem and Indian grass is reduced. Continued over many years, such fires would shift species composition to cool season species (Joern and Keeler 65).

Late summer burns can set back warm season species. When perennial plants have translocated food from underground organs to leaves, as they do in the growing season, they are more susceptible to fire damage (Risser et al. 411). Unfortunately, not much research has been done on the effects of summer burning even though it was certainly common in prairie areas (Joern and Keeler 65).

Fire and Soil

Fire affects the structure, organic material, nutrients, microbes, and moisture in the soil. In grasslands, the removal of excessive litter accumulations by burning (particularly spring burning) during normal to wet years raises soil temperatures, increasing bacterial activity, nutrient cycling and plant growth. In dry years, however, fires that remove litter function to reduce insulation, increase drought stress and allow erosion from flash floods. Severe fires in forests consume all litter and humus and can destroy up to 65% of the top inch of soil organic matter. This has the effect of accelerating oxidation of organic matter in the soil profile, reduces infiltration and storage capacity, and is detrimental to soil nutrient properties. Under normal fire conditions, the effect on percolation is negligible (Wright and Bailey 27, 29, 35). Upland sites in the Piedmont with thin soils are and were most susceptible to the effects of damaging fires, as noted by early explorers (Silver 61; Lawson 56; Bartram 263; Fries 50).

Though forests are more productive of organic carbon forms, organic matter content in temperate grasslands averages twice the amount found in temperate forests. This building of organic matter in the soil can be attributed to rapid root growth and turnover where grasses dominate due to frequent fires. Though fire consumes above-ground material and generates carbon, most carbon is contributed to the soil by roots. (Joern and Keeler 164, 168, 169; Risser et al. 431). For example, where there is a cover of grasses under pines in the Coastal Plain, frequent burning can increase organic matter content in the upper 4 to 6 inches of soil by up to 60%. A study conducted in Kansas showed that six consecutive years of burning did not reduce organic matter in the soil (Wright and Bailey 28).

Ninety percent of nitrogen in standing grass fuels, and less in woody fuels, is volatilized through burning. Total nitrogen loss can be reduced by burning over moist soil on humid days. Despite the large losses of organic material and nutrients that are volatilized during a fire, large quantities of nitrogen, phosphorus, potassium, calcium, magnesium, sodium, and to some extent sulfur are made readily soluble. This is most important where conditions such as low temperatures or drought make release of these nutrients through decomposition a very slow process. If drought or some other condition inhibits plant growth after burning, however, much of the free nitrate produced, if not consumed by plants, will be quickly lost. In warm, moist environments nutrients are well supplied by decomposition (Wright and Bailey 32,33).

Beneath the soil surface the temperature increase from a grass fire is negligible below 1/4 inch and has little direct effect on soil microbes, organic matter, or buried seeds. Indirectly however, harsh conditions resulting from fires can cause reduction in soil organisms' population because of food source depletion, moisture decrease, and temperature and pH increase (Wright and Bailey 11, 31).

Fire and Animals

Animals are a product of their habitat, which is often a product of fire, therefore they are generally well adapted to it. Species composition will change when fires convert one type of habitat to another. The mosaic of seral and climax communities, often referred to as the "edge effect," that may result from fires over time is the preferred habitat of a high number and diversity of wildlife (Wright and Bailey 49-51). During a fire event various animals are affected differently. Headfires a few feet above the ground burn up to ten times hotter and faster than backfires, though at the soil surface backfires are nearly as intense and burn more thoroughly. Hot, fast-moving headfires could catch and kill larger animals, passing over smaller ground dwellers, while larger animals could escape slow moving backfires that might kill small animals (Joern and Keeler 61). Small, often underground-dwelling mammals usually survive fires that leave unburned patches and are of a low intensity. Following a fire, their populations decrease due to the lack of cover (Wright and Bailey 50).

When 40 percent or more cover for escape is left intact, deer populations increase dramatically after burns. Succulent shoots of sprouting plant material provide easily available and abundant food. The grasses and forbs that increase after burning are generally only important as forage during spring and summer. Browse is the most important food source for deer and should increase where fire resistant grubs resprout prolifically after burning. Increased deer or other prey populations will attract more predators (Wright and Bailey 64-69).

Birds, being very reliant on vegetation structure, are greatly affected by fires. (Risser et al. 424). The absence of fire or other disturbances in forests will generally result in decreased niche diversity and capacity. Forests in preclimax stages generally have the highest bird diversity (Wright and Bailey 52). Turkeys, for instance, prefer park-like or savanna-like habitats where they can both roost and feed. In burned areas they can find new green shoots and a good supply of insects for their poults. Bobwhite, quail, and mourning doves are also attracted to freshly burned areas and will nest in or near them. Quail visit edges of burns while they are still smoking and fill their crops in a matter of minutes on the abundant exposed dead insects and seeds.

Burning in marsh communities should be coordinated with the migration patterns of waterfowl and with the special requirements of other wetland species (Wright and Bailey 58-64). When burning affects stream and wetland habitat, increased sediment loads and loss of shady cover have a negative short term impact, though not a toxic one. Long term effects such as increased flow due to reduced soil moisture and new beaver-created habitat are beneficial. Beavers prefer the regenerating saplings of burned areas over the large trunks of mature forest trees (Wright and Bailey 70-71).
 

Grazing and Browsing

Herbivores range from animals such as bison to rabbits to grasshoppers to nematodes. Herbivores and North American grasslands, including those of the Piedmont, have evolved as interacting partners for thousands of years. Grazers and browsers are adapted to the growing habits of plants and plants have adapted strategies and morphology to survive under herbivory (Risser et al. 334). Small mammals' main influence on vegetation may be seed dispersal rather than grazing; however rodents, such as rabbits, are important in maintaining a tussocky character in grasslands, especially on edges (Deselm and Murdock 104). Eastern grasslands support a vast number of insects which are the most significant consumers of plant material in grassland. An invertebrate, the crayfish, on the Black Belt Prairie of Mississippi, was at one time a significant agricultural pest. Found at a density of several thousand per acre, they could destroy a crop in a few nights (Deselm and Murdock 105, 110).

Grazing can have beneficial effects on prairie-type environments. Midwestern studies showing that moderate clipping under moist conditions increases above-ground production in prairie grasses suggest the benefit of moderate grazing. Annual clipping in the middle of July showed the greatest increase in forage production in Oklahoma. Early growing season clipping is the most detrimental, while in the summer clipping seedstalks will result in increased leaf production (Risser et al. 353). This seems to differ in effect from burning, where late spring burns maximize forage for grazing when done once every 3 - 4 years and where summer fires should apparently be more detrimental to warm season plants (Risser et al. 411, 418, 432).

As in overgrazing, the lower and more frequent the cutting height of prairie grasses, the greater the reduction in root mass and the greater the depletion of food reserves (Risser et al. 353). In defoliated plants, unable to send sufficient food to roots, root growth will slow or cease under overgrazing (Joern and Keeler 180). As a result, most prairie dominants decrease under close, frequent grazing while other prairie species become more conspicuous. They in turn succumb to be replaced by broadleaf weeds and annuals as close grazing continues. Switchgrass is shown to be most susceptible to the effects of clipping while big bluestem and Indian grass may hold out the longest under intense grazing or mowing (Joern and Keeler 352-355).

Grazing also affects soil, as it tends to lower soil water content. Soil trampling reduces soil porosity, particularly on clayey and organic soils. Rates of diffusion of water, oxygen and nutrients to root surfaces lessen and cause a reduction in plant growth (Joern and Keeler 179,180). Compaction from large herbivores especially affects infiltration on fine-textured soils, though effects of compaction may disappear after freeze-thaw cycles (Risser et al. 347-349).

The co-evolution of grazers and prairie species found in the Piedmont savanna has a slightly different history than that of the Great Plains. On the Plains, large grazers have exerted influences on grassland species for thousands of years. In southeastern grasslands that influence was removed from the end of the Pleistocene, until about the 1600s when buffalo began to migrate into the Southeast. In the seventeenth century, when buffalo migrations were greatest and herds of thousands, or at least hundreds, were observed, Indian populations had been seriously decimated, some as much as 80% by disease. The abandoned grasslands the Indians provided may have acted as links in a migration corridor for buffalo into the Southeast (Rostlund 403, 407) just as they did later for pioneering Europeans. As latecomers, it is unclear how much they contributed to the local ecology. Perhaps, as in the Midwest, their grazing helped preserve some savannas (Nuzzo 10). Unfortunately, there is little information about what southeastern bison ate, though it is reasonable to assume that the buffalo, as in the great plains, was mainly a grazing animal that browsed only occasionally (Burt and Grossenheider 224; Rostlund 406). They also tended to graze an area heavily before moving on (Joern and Keeler 184). Herds were commonly described as "feeding," "pasturing," or "grazing" in open savannas, prairies, meadows and old fields of the Southeast (Rostlund 406). The Piedmont was described as having "large and extensive plains and savannas full of deer and buffalo." (Landrum 1) Of the eight genera of grasses that bison feed on in the plains, six genera, including over 50 species, are native to the Southeast (Rostlund 406). It is quite possible that their unmistakable trails, which are mentioned in many accounts, would have wandered through the tall grass of almost any Piedmont savanna.

Deer, which are primarily browsers, can have a marked effect on vegetation. Deer are denizens of early successional and edge habitats and are one of the most adaptable mammals in the world (Skeen, Doerr and VanLear 16). They prefer edge environments because they are high in forage and refuge (Silver 54). Their diet is composed of tender shoots, twigs, and leaves, herbaceous plants in season, fungi, mast and fruits (Skeen, Doerr and VanLear 16; Burt and Grossenheider 218; Silver 25). Woody plants comprise more than 85% of their diet, even in early successional habitats such as clear cuts, where they eat over 65% twigs and shoots of woody species (Skeen, Doerr and VanLear 16). As very populous herbivores they may have had an important role in controlling woody growth in the Piedmont savanna, although there is no documentation of that.

Elk, drawn from the mountains by the abundant grasslands of the Piedmont, were not very common and probably had little effect on savanna functions. The fact that elk and bison prefer open environments where they can outrun predators and fend them off with horns and hooves (Silver 26) is further evidence of the extent of Piedmont savannas.
 

Vegetation Structure and Species Composition

For the Midwestern oak savanna, general consensus on composition is an oak-dominated community, having between 10 and 80 percent canopy cover, with or without a shrub layer, with an herbaceous, predominantly grassy ground layer of both prairie and forest species and appearing as either open or scrub savanna. It is also considered a fire-dependent community, or, in its absence, reliant on extremely droughty or otherwise severe conditions (Nuzzo 10). Adding a few pines to the mix might make a useful description of Piedmont savanna composition.
 

Canopy

Currently, the forest type of the Piedmont is described as oak-hickory-pine by Küchler (Doerr and VanLear 1). Studies indicate that most of the presettlement forest was dominated by hardwoods, mainly of an oak-hickory type, which reached their best development on the widespread deep, sandy loams overlaying the clays of the Cecil, Lloyd, and Davidson series. A survey by Plummer in 1957, examining original land survey records of over half a million acres of Georgia Piedmont, found that forest vegetation at the time of settlement was oak-pine-hickory at a ratio of 53:23:8 (Skeen, Doerr and VanLear 6-7). Old surveys of the Black Belt Prairie in Alabama showed a similar ratio of 52% oak, 9% pine and 5% each of hickory, gum, and ash (Risser et al. 64). Examination of remnant old growth, and studying the patterns following disturbances, show that oaks found on favorable sites were mainly white, northern red, and black; on intermediate sites were mainly black, southern red, and post; and on marginal sites were mainly post and blackjack (Skeen, Doerr and VanLear 14).

On the 1773 Map of the Ceded Lands by Phillip Yonge, an area encompassing much of the east Georgia Piedmont is described:

The lands in general consist of Oak and Hickory in many places intermixed with black walnut, chestnut, and tupelo especially in the vallies, level lands and cane brakes, the hilly lands consisting of Oak and Hickory with some few pines.

Benjamin Hawkins, traveling through the west Georgia and Alabama Piedmont in an area form the Chattahoochee to the Tallappoosa and Coosa rivers, noted the tree composition among his descriptions. While informative, his descriptions of species composition are not necessarily specific enough to be used as a model:

The soil is stiff, with course gravel, and in some places, stone. The trees are post oak, white and black oak, pine, hickory and chestnut, all of them small. (Hawkins 19)
... open, flat, land, the soil stiff, the trees post and black oak, all small. (19)
The growth of timber is oak, hickory, and the short leaf pine; pea-vine on the hill sides and in the bottoms, and a tall, broad leaf, rich grass, on the richest land. (20)
The waving land to its source is stiff. The growth is post oak, pine and hard shelled hickory. (26)
The timber is post oak, hickory, and pine, all small. (28)
...post oak, hickory, and pine land...(29)
The upland is generally stiff, rich and fit for culture. Post oak, black oak, pine and hickory, all small are the growth. (44)
...there are sharp, stoney hills, the growth is pine, and the branches all have reed. (45)
...pine, oak, hard shelled hickory, and on the ridges chestnut... (46)
...oak, hickory and pine, gravelly, trees all small, chestnuts on the ridges...(47)
...large reedy glades in flat land; red, post, and black oak, all small...(47)
...gravelly with oak, hickory, and pine...(47)
...pine, oak and small hickory...(49)
...red oak and small hickory... (49)
...oak, hickory and a few poplar... (54)
...broken land, chestnut, pine, post oak, hickory, and red oak...(54)
...post, red oak, pine and hickory...(54)
...broken land, hickory, pine, chesnut, cane along the creek ...(55)

In his 1992 doctoral thesis, Charles Mark Cowell examined historic survey records, which note species of marker trees, to reconstruct the structure of the presettlement forest. The study area, between the Oconee and Ocmulgee rivers, is representative of much of the Piedmont in its typical rolling convex topography and eroded Ultisols, mostly of the Cecil series. As seen in Figure 5.6, the 1804-1806 surveys indicate an extensive change in composition frequency from the presettlement forest to today's secondary forest.

common name	species name	% presettlement	% change
pine	Pinus spp.	26.8	-24.6
post oak	Quercus stellata	17.5	-15.2
black oak	Quercus velutina	10.9	-9.7
red oak	Quercus rubra	10.5	-1.7
hickory	Carya spp.	10.1	7.5
white oak	Quercus alba	7.3	14.1
s. red oak	Quercus falcata	2.7	3.4
dogwood	Cornus florida	2.7	5
chestnut	Castanea dentata	1.7	-1.5
poplar	Liriodendron tulipifera	1.5	4.6
black gum	Nyssa sylvatica	1.3	1.8
blackjack oak	Quercus marilandica	1.1	-1.2
total		94.1

Figure 5.6 Forest-wide presettlement species composition and change to the present species composition (Cowell 32,122).

Of the most increased species, white oak has risen 14.1%, hickory 7.5%, dogwood 5.0%, and tulip poplar 4.6%. Of the most decreased is pine at -24.6%, post oak at -15.2%, and black oak at -9.7%. This reflects an increase in fire intolerant species and a decrease in fire tolerant species across the forest spectrum. The differences in the two periods show a recent suppression of large scale disturbance by fire (121-122).

Cowell's study also describes floodplain, lower slope, mid-slope, upper slope and flat upland composition as shown in Appendix A (123-127). Floodplain, now dominated by sweet gum, box elder, birch and ash, was once dominated, though not by great margins, by white oak, hickory, pine, post oak and black oak. The remaining 50% of tree species found on the floodplain included northern red oak, ash, tulip poplar, dogwood, maple, black gum, sweet gum, southern red oak, sassafras, beech, chestnut, persimmon, buckeye, mulberry, willow, holly, basswood, sourwood, cucumber, chinquapin, bay, crab, hawthorn, evergreen, cherry, blackjack oak, elm, ironwood, sugar, chestnut oak, wahoo, water oak, and birch.

The lower slope community, now dominated by hickory, white oak, dogwood, tulip poplar, and red oak, was dominated by pine, post oak, hickory, white oak, black oak, and northern red oak. The remaining 20 % of species included ash, elm, chestnut, blackjack oak, buckeye, persimmon, sugar, sassafras, cucumber, chinquapin, birch, cedar, walnut, alder, basswood, mulberry, water oak, wahoo, southern red oak, sourwood, ironwood, black gum, maple, beech, sweet gum, and dogwood.

The mid-slope community, now dominated by white oak, hickory, and dogwood, was dominated by pine, post oak, red oak, black oak, and hickory. The remaining 21% of species included white, oak, chestnut, blackjack oak, sassafras, elm, persimmon, bay, willow oak, walnut, chinquapin, basswood, cherry, ironwood, mulberry, cucumber, water oak, southern red oak, ash, beech, sourwood, black gum maple, tulip poplar, sweet gum, hop hornbeam, and dogwood.

The upper slope community, now dominated by white oak, hickory, and northern red oak, was dominated by pine, post oak, black oak, northern red oak, and hickory. The remaining 19% of species included white oak, chestnut, blackjack oak, sassafras, persimmon, elm, ash, mulberry, sugar, chinquapin, basswood, wahoo, cherry, cucumber, water oak, black gum, hop hornbeam, beech, dogwood, sweet gum, tulip poplar, southern red oak, maple, and sourwood.

The flat upland community, now dominated by white oak, hickory, southern red oak, and northern red oak, was dominated by pine, post oak, black oak, northern red oak, and hickory. The remaining 15% of species included white oak, southern red oak, chestnut, blackjack oak, sassafras, black gum, persimmon, hawthorn, cherry, wahoo, willow oak, walnut, sweet gum, ash, holly, water oak, dogwood, maple, and tulip poplar.

Interestingly, both upland and floodplain species were found along major streams and rivers, while smaller creeks were dominated by mainly upland species. This reflects disturbances such as upland fire encroachment and Indian agricultural practices (138). Shortleaf pine was probably predominant among pine species in savanna landscapes of the upper Piedmont, as it is more tolerant of frequent fire than loblolly (58), while longleaf pine may have been the dominant among pine species in the lower Piedmont, especially in Piedmont longleaf pine flatwoods (Schafale and Weakley 80).

With large scale landscape disturbances widespread in the presettlement Piedmont, climax forest communities were probably not always of the typical oak-hickory-pine composition. Different varieties of major disturbances, followed by still different variations of subsequent disturbances, might have initiated, and then maintained, different types of stands (Cowell 3). For example, as a result of canopy fires, pines may have formed pure stands in some areas (Cowell 58). On the other hand, Spangenberg in his account transcribed by Fries, and Lawson described the lack of pines for roughly two days' journey in the area of Iredell and Rowan Counties, North Carolina. A trial analysis of corner trees from old land grants also showed only a few pines, though there was a preponderance of oaks and hickories in the area (Keever 40-41). Keever believes the lack of pines to be a result of Indian burning and buffalo grazing. Perhaps in such large open areas, as in the prairie, where forest edges and seed sources were distant, pines were not able to invade (Joern and Keeler 158). Also, pines may not have been able to regenerate fast enough between very frequent fires, leaving only oaks able to resprout. Periodic burns have been shown to stimulate hardwood regeneration of southern red, post and blackjack oaks (Cowell 59).

While Cowell's study is a suitable model for forest structure for much of the Piedmont, a different composition of tree species may have occurred over areas with either more intense Indian population influences and/or more stressful soil conditions. Blackjack oak, which is a minor component in Cowell's study area, may have been more prominent on other Piedmont areas associated with more severe conditions resulting from shrink-swell soils or topography. At Mineral Springs Barren, a Nature Conservancy site near Charlotte, North Carolina, blackjack oaks dominate the canopy with short leaf pines, post oaks, red oaks, southern red oaks, and white oaks. This community is characteristic of xeric hard pan forest, and is believed to be a remnant of formerly extensive prairies and blackjack oak savannas in the Rock Hill, South Carolina area. Where canopies of this forest type have been kept open, a diverse cover of herbaceous flora occurs, with a number of species having prairie affinities (Barden, Management 2-4; Schafale and Weakley 78). A similar situation occurs in the post oak savanna being restored in Saluda County, South Carolina, where post oaks dominate an open canopy (Waldrop, Establishment 5). This site also has a well-developed herbaceous cover of grasses and forbs over shrink-swell soils. On Burkes Mountain, a serpentine barren near Augusta, Georgia, blackjack oaks are dominant with shortleaf and, formerly, longleaf pines in this blackjack oak savanna (Radford 49-50).

Longleaf pine, now virtually gone from the Piedmont, also deserves mention. Longleaf pine forests once covered large portions of the lower Piedmont along its southeastern side. Occurring over various hapludults, they may have had a flat-woods-like structure, with wire grass dominating the herbaceous layer (Schafale and Weakley 80, 81). Bartram mentions the longleaf pine as one of the savanna tree species encountered along the fall line (305).
 

Ground Layer

Southeastern grasslands are often described as prairie due to their floristic similarity to prairies of the central United States (Deselm and Murdock 89). The many species with prairie affinities, and others indigenous to the Southeast, composed the ground layer of vegetation, often described by early Piedmont explorers. Most accounts are very general in referring to the ground layer in terms of meadows, fields, plains, savannas, barrens, lawns, cleared ground or simply as open. Some later historians have used "prairie" to describe the same effects. A typical commodity-based description is given by Spangenberg describing a tract encompassing the area around Winston-Salem:

...much beautiful meadow land, good pasturage for cattle, cane along the creeks... ...also a good deal of barren land...

... one can hardly find 600 acres in NC without having some "barren land" in it. (Fries 59)

...the more barren grassy hills...(268.)

...very extensive grassy savannas...

...pleasant grassy open plains... (306)

...tall, broad leaf, rich grass, on the richest land. (Hawkins 20)

...well spread with fine bladed Grass...(Lawson 38)

Bartram, as a naturalist, was the only explorer to describe the actual species that composed the ground layer. He recognized that the lists he compiled represent only a few of the "vast variety and abundance" of herbacea. With a translation to modern nomenclature from Harper's The Travels Of William Bartram: Naturalist's Edition, the sun-loving species from those lists are shown in Appendix B.

Various terminology used by different explorers, other than Bartram, only adds confusion and highlights the problem of historical accounts as ecological models. For instance, what are the "Strawberry Vines" that Lawson (38) sees growing in and about abandoned Indian fields? Other plants noted by explorers, usually in association with old Indian fields, include "vetch, ground nuts, or wild pea vine". Silver believes these may have been escaped bean plants, as the Moravian surveyor Reuter described: "Indian beans resemble garden Beans, though they are small. They grow abundantly in the woods, especially on good soil." (Silver 50) Bartam noted two leguminous vines, which today are natives of forest settings, and may have been cultivated by Indians in their clearings. Wild bean, Phaseolus polystachios, has legumes, or pods up to 8 cm long (Radford, Alhes, Bell 639) and ground nut, Apios americana, was a staple food source for Indians and Pilgrims (Porcher 103). There are also other non-agricultural native legumes present in the herb layer that have pea-like flowers, so it is difficult to identify the "wild pea vine" with certainty.

There is no way known to define accurately presettlement herbaceous layer composition. Decades of soil disturbance, exotic species invasion, and canopy closing from fire suppression have obliterated all but a few tiny ground layer communities found on high shrink-swell soils, basic soils, or topographically exposed sites. These communities include the xeric hardpan forest, the diabase glade, and the ultramaphic outcrop barren. Lists of key herbaceous species found in these communities, many with prairie affinities, are provided in Appendix C. Appendix C also includes ground layer species of the Longleaf pine-wire grass communities, also once part of the southern Piedmont; because of its proximity as the nearest studied prairie, species of the Alabama Black Belt Prairie are listed as well. Over more typical Piedmont soils, such as the Cecil series, no relict communities were identified in this study, though roadsides, ditches, and meadows may provide refuge for savanna species. A list of such species found in the Clemson, South Carolina area (see Appendix D) compiled for use in the thesis application represents the diversity of these potential savanna survivors. Many of the species with prairie affinities found in the xeric hardpan, diabase glade and serpentine barren communities do not appear on the extensive list of roadside/meadow species found in the Clemson area. It is unclear whether this is a reflection of the altered distribution of species due to fire suppression or a reflection of the soil preferences of these species. The great variety of ground layer species on these lists reflects the potential diversity of the savanna ground plane where probably about 40-70 species per acre, as in the tallgrass prairie, would be typical (Morrison, Design 22).

Though fire and moisture had the greatest influence on herbaceous composition, variations in species, common on most soil types, would occur under specific conditions such as xeric hardpan barrens and upland wet depressions associated with high shrink-swell soils. Link Bog, a remnant grassland community (Radford 111-112) in Iredell County, North Carolina, is one of six such bogs located along the headwaters of Rocky Creek at the foothills of the Brushy Mountains. It represents a formerly more common ecosystem, perhaps like the "highland ponds" seen by Byrd in the 1700s (Byrd 212). Still found in the bog, though it is growing over, are fire tolerant carnivorous species: Sarracenia flava, S. purpurea and Drosera rotundifolia. They are virtually unknown elsewhere in the Piedmont. Seasonally wet upland depressions over similar soils occur in other places in the Piedmont, such as in the post oak savanna in Saluda County (Waldrop Establishment) and may have once also supported distinct species. Camassia Flat (Radford 189-190) in York County, South Carolina is another wet site over montmorillonitic soils that is now a woodland, but was probably a savanna or prairie in presettlement times. It still supports some herbs with prairie affinities which are very rare east of the Appalachians, or rare in South Carolina, including Camassia scilloides, wild hyacinth; Ranunculus fascicularis, prairie buttercup; Scutellaria parvula, a skullcap; as well as other wet meadow species such as Allium bivalve, A. canadense, Cardamine bulbosa, Melanthium virginicum, Viola septemloba, and Zephyranthes atamasco. Mineral Springs Barren in Union County, North Carolina, a xeric site also on montmorillonitic soils, still supports many prairie species including the endangered Helianthus schweinitzii, and the rare Lotus helleri, Aster georgianus and Gnaphalium helleri (Barden, Management 8). Other Piedmont savanna areas to consider as having distinct communities of herbaceous species include the former longleaf pine-wire grass communities near the Coastal Plain; and Burkes Mountain, serpentine barren near Augusta, Georgia where Georgia basil, broomstraw, panic grasses, and legumes such as rattlebox, pencil flower, white baptisia, tephrosia, lespedeza, and butterfly pea can be found, as well as the rare Clematis albicoma, Polygonatum tenue, and Manfreda virginica (Radford 49-50).

The following information on prairie plant characteristics is summarized from Marguerite McCrary Haywood's 1989 thesis (32-40). Of the prairie herbaceous species there are two general categories: grasses and forbs. Grasses dominate the prairie community, composing 90% of the vegetation though composing only a quarter or so of species diversity. This gives grasslands a strong visual continuity. The great majority of prairie and thus Piedmont savanna grasses are warm season perennials. Resprouting year after year from the roots, warm season grasses begin growing when minimum daily temperatures are in the low 60s. There are also cool season grasses to be found that grow in the spring and fall, and in the winter in the south, when minimum daily temperatures are in the low 40s. Grasses with their vertically arranged, long, narrow blades maximize photosynthesis, while minimizing leaf transpiration, making them well adapted to exposed environments. Forbs, by contrast, have a broader, more horizontal leaf pattern. Forbs, only 10% of the total vegetation, make up the majority of the diversity. Forbs and grasses have variations in root depth, plant height, form, and seasonal requirements. This means that during the different yearly, seasonal, and daily conditions, at least some species will be able to capitalize on existing conditions. This also means that many plants of many varieties can occupy a small space while they utilize different levels of soil, canopy light, and other resources through the year. It is this sort of diversity that gives prairie communities a great deal of range and stability and protection from weedy invaders. Distribution of the herbaceous layer is primarily influenced by microclimate, which involves soil type, soil pH, soil depth, soil moisture, fertility, topography, slope orientation, and, in the case of the savanna, light availability. Of these conditions, moisture is the most important and soil pH the least. Along a moisture gradient species composition varies from more spaced, shorter, and less diverse, to denser, taller, and more diverse.
 

Cane

Canebrakes, sometimes of immense size, were ubiquitous in the wet soils of the Piedmont. Because they very often intermingled and formed edges with Piedmont savannas, it is important to consider them as a component, though not actually part of the savanna. Though cane (Arundinaria gigantea) is a grass, and constitutes a graminoid stratum, its height and density obstruct the open characteristic of savannas.

In the presettlement Piedmont the abundant canebrakes provided both cover and forage for deer and buffalo, especially in the winter (Silver 25, 26). Explorers, especially those who traveled along rivers, noted it often. As a member of a survey party, Byrd frequently mentioned canes along the creeks and rivers they had to cross. Along one creek, he described cane growth of 12-16 feet high, some as "thick as a mans wrist" (Byrd 192). Spangenberg in North Carolina, and Hawkins and especially Bartram in Georgia and Alabama all refer to the abundant cane, often as it relates to adjacent grasslands:

...rich low land cover with canes, adjacent land is higher also rich where indians plant, grass grows freely, about half the land is barren, but has some trees on it. (Fries 53)

...creeks are "margined with cane or reed, on narrow strips or coves, of rich flats." (Hawkins 19)

... the face of the country is chiefly a plain of high forest, savannas, and cane swamps..." (Bartram55)

We then passed over large rich savannas or natural meadows, wide spreading cane swamps, and frequently old Indian settlements, now deserted and overgrown with forests. (57)

A large cane swamp and meadows, forming an immense plain, lie S.E. from it; in this swamp I believe the head branches of the great Ogeeche river take their rise." (58)

...the low lands of the Oconee... ...the cane swamps of immense extent, and the oak forests, on the level lands, are incredibly fertile;...(60)

There are extensive cane brakes or cane meadows spread abroad round about, which afford the most acceptable and nourishing food for cattle. (304)

... on the gradual descents of the ridges and their bottoms bordering on creeks, and very extensive grassy savannas and cane meadows always in view on one hand or the other... ... here being pleasant grassy open plains to spread our beds upon, environed with extensive cane meadows, ... (306)

...a pleasant territory, presenting varying scenes of gentle swelling hills and levels, affording sublime forests, contrasted by expansive illumined green fields, native meadows and Cane breaks... (307)

...low swelling hills and plains supporting grand forests, vast Cane meadows, savannas, and verdant lawns. (308)

...with hills and dales, savannas, and vast cane meadows... ... diversified with expansive groves, savannas and Cane meadows... (309)

...low hills affording high forests, with expansive savannas, Cane meadows and lawns between,... (312)

plains, detached forests and groves... ... vast Cane meadows, and lastly a chain of grassy savannas. (317)

vast level plain country of expansive savannas, groves, Canes swamps and open Pine forests (318)

The reasons for the development and disappearance of such large and widespread brakes are related mainly to fire. Ralph Hughes, in his 1966 article "Fire Ecology of Canebrakes" (149-158), explains the relationship of Arundinaria to fire, pointing out that fire every 3-4 years is of great benefit to cane stands. Cane has a heavy underground stem or rhizome protecting its food reserves from fire and grazers. Fire reduces canopy competition, and promotes new and immediate growth. Following dormant season fires, new cane shoots grow at a phenomenal rate of as much as 1.5 inches per day in late May and June. Such quick cover would also reduce competition. Unfortunately, canebrakes reach maturity and lose dominance over other plants after only 10 years of fire suppression, causing them to thin out and die. Even when appropriate conditions are restored regeneration is very slow, since stands low in vigor cannot reclaim bare areas due to the plant's sporadic flowering, scant seed production, and slow seedling growth.

 
Fauna

Piedmont savanna landscapes might have been characterized as high in edges and ecotones. The edge, or ecotone, results where two communities come together to create habitat supporting greater number and variety of creatures than could either community alone. For instance, where forest and grassland abut, species will use one community for feeding and the other for shelter. The barred owl and the great horned owl do not nest in grasslands but use them for foraging where they are adjacent to woodland (Deselm and Murdock 103-105).

It seems reasonable that the majority, if not all, of the major native animal species found in Piedmont edge environments today were present to some extent in the presettlement savanna. There are, however, a few animals that were unique to the Piedmont savanna that are no longer found in the region. As mentioned in the section on fire and animals, large herds of buffalo, and a scattering of elk, were once found in the Piedmont savannas, as were wolves (Silver 26). It has been mentioned by (Kaufhold 3) that prairie chickens may have been present, presumably because the Cherokee had a name for them. However no mention of Piedmont prairie chickens has been found elsewhere. Today prairie species are generally not found in the Southeast, though many southeastern species are adapted to grassland habitats (Deselm and Murdock 103).

Some southeastern bird species are known to frequent maintained areas such as small grass-strip airports, large lawns of school campuses, and golf courses, while a few others are mainly associated with coastal grasslands. As birds associated with modern southeastern grassland habitat, they might have once been found in the Piedmont savanna, although there is no record to show this. These birds include the horned lark, brown-headed cowbird, cliff swallow, dickcissel, loggerhead shrike, eastern meadowlark, savanna sparrow, and upland sandpiper (Deselm and Murdock 104).

Burrowing mammals such as voles, moles and groundhogs provide soil structural diversity for invertebrates and create burrow habitat for other ground-dwelling species. They also enhance grassland structure by preparing seed beds, contributing to the aeration of the soil, increasing porosity, improving local drainage, and helping incorporate humus and nutrients into the soil. Rodents, especially rabbits, are important in maintaining a tussocky character in grasslands, especially on edges. Prairie deer mouse, meadow vole, and eastern mole are prairie species that are found in southeastern grasslands (Deselm and Murdock 104-105).

In the modern Piedmont, edge communities of grasslands or other human-caused disturbances have increased deer and red fox distribution and may be responsible for the recent eastward migration of the coyote (Deselm and Murdock 104). Such a dynamic resembles the relationship between Indian clearing and southeastern buffalo migration. This comparison points out that humans have been and continue to be very much involved in the "natural" processes of the Piedmont.

Many of their accounts refer to the appearance resulting from grasses being dominant, as with Bartram and Hawkins in Georgia and South Carolina and Lawson in North Carolina:

...a fine expanse of level grassy plains...

...and lastly a chain of grassy savannas. (Bartram 316-317)

back to top