Skip common site navigation and headers
United States Environmental Protection Agency
Great Lakes Ecosystems
Begin Hierarchical Links EPA Home > Great Lakes EcosystemsUpland Ecosystems > 1993 Oak Savanna Conferences > Ken McCarty
Aquatic Ecosystems
EPA Region 5 Critical Ecosystems
Ecosystem Funding
Great Lakes Basin Ecosystem
Great Lakes Biological Diversity
Green Landscaping
Rivers and Streams
Upland Ecosystem



1993 Proceedings of the Midwest Oak Savanna Conferences1993 Proceedings of the Midwest Oak Savanna Conferences



Ken McCarty
Division of State Parks
Missouri Department of Natural Resources 
Jefferson City, MO 65102-0176 
Tel: (314) 751-2479  


The Missouri Department of Natural Resources began restoring native savanna and woodland ecosystems in state parks 10 years ago. These Projects now exceed 10,000 acres and span a wide spectrum of savanna community types with many different disturbance histories. This paper presents several case histories to illustrate typical responses to restoration in both high and low quality remnants. It's chief purpose is to show that restoration can be successful even in the severely degraded remnants that are left to work with today. From these case histories, we can distill some basic principles related to restoring moderately-to well-drained, upland savanna and woodland ecosystem. We have learned that it is relatively easy to recover the characteristic open structure and greatly increase the diversity and abundance of the typical savanna herbaceous species. Most important is to restore fire and light to these fire and light dependent communities, and the surviving plants will do the rest. However, the degree to which the full spectrum of native understory species recover largely depends on the proportion of their propagules that still survive. Our experience indicates that much attention needs to be paid to reinstating natural processes such as fire in our remanat savannas and woodlands before such propagules disappear. Once lost to shade and midstory sapling development that most contemporary management practices foster, the species diversity and richness once native to these systems is difficult -- perhaps impossible -- to recover.


Savanna and fire-maintained woodlands were once common in Missouri (Nelson, 1985; Gremaud, 1992). Nuzzo (1986) mapped a wide belt through western Missouri where post oak savannas were especially prevalent on cherty, well drained western Ozark soils; and through northern and western Missouri where prairies, woodlands, and groves intermingled. Elsewhere, there were also pine and oak/pine dominated woodlands and savannas in the eastern and southern Ozarks; post oak flatwoods covering poorly drained ridges and divides through the central Ozarks; deep soil wet savannas in the southeast lowlands and along many river floodplains; shallow soil barrens in the northeast, central, and southern Ozarks; and fire-influenced north-slope woodlands statewide (Nelson, 1985).

There are many historical descriptions of these savanna and woodland communities, with particularly good ones being found in Marbut (1914); Houck (1908); Brackenridge, in Thwaites (1904); Schoolcraft (1821); and in many county histories. Schroeder (1981) systematically surveyed the General Land Office records for Missouri, and gives an excellent characterization of those lands throughout Missouri which were not prairie, but obviously not typical forest, either, at the time of historic settlement. Ladd (1991), and Nigh (1992), summarize several early descriptions and make the point that savanna and woodland natural communities originally existed throughout Missouri. For upland, and even some bottomland landscapes, they were the rule; not the exception. They existed in ecosystem-rich collages according to how physiography, soils, and geology mediated moisture and fire patterns over the past several thousand years. In fact, most of Missouri's contemporary upland forests are residual woodland, savanna, and prairie ecosystems. Their surviving plants can be found almost everywhere on undeveloped lands and roadsides.

These former woodlands and savannas were once exceptionally diverse. Judging from contemporary remnants, they were among our most species rich natural environments; and most of this richness was linked to their understory layer. For example, there are 342 native vascular plant taxa in only 160 acres of The Nature Conservancy's Bennett Springs Savanna. Twenty are sedges; 41 are grasses; and 243 are forbs. Although trees are the dominant visual feature, with nearly touching canopies, only 24 species are present. Likewise, the Turkey Pen Hollow Savanna restoration site contains 232 native species beneath its 20 tree species. Regarding wildlife, Evans and Kirkman (1981) found 64 bird species to occur in grass and shrub savanna, compared to 39 species in the mature oak hickory forests that most are changing to. Several of today's extirpated or endangered species, such as the red-cockaded woodpecker and Bachman's sparrow, are obligates of these ecosystems.

While the trees are still with us, or have redeveloped, the rest of the savanna complement is vanishing or absent. However, vast acreages across Missouri are yet restorable. Unlike our prairies, wetlands, and riparian forests, which have been essentially or wholly converted to other purposes; savannas and woodlands are the one major Missouri natural environment where we may sustain substantial portions of our native biological diversity.

Many savanna restoration projects are already underway in Missouri. The Missouri Department of Natural Resources, Missouri Department of Conservation, The Nature Conservancy, and the U.S. Forest Service, are all involved on their respective lands. Most effort has gone towards the drier savannas and associated glades of the central and western Ozarks, particularly those with site indexes lower than 50. More recent work is expanding into other types and regions; and focuses particularly on landscape scale projects that cover several hundred to a few thousand acres, with a better representation of natural community types. These are showing us the potential and process for recovery.

Iffrig and Nelson (1983) described early restoration efforts in Missouri state parks, that began with 2000 acres of prescribed fire and clearing in five parks. That program now covers over 16,000 acres in 23 state parks, and a whole range of savanna and woodland natural communities. Growth continues towards a goal of 62,000 acres. Drawing primarily from these experiences, this paper explains how we approach the restoration process; examines our current techniques; and describes some typical results. It deals primarily with what is happening in restorations on Department of Natural Resources lands, but this is similar to what others are doing in Missouri, as well. It attempts to convey what we consider to be the key points for restoring our savanna and woodland ecosystems.


Though altered in form, large savanna and woodland expanses still exist in Missouri. They survive under essentially the same climate (Delcourt and Delcourt, 1986), and in the same physiographic context, as they have for the past 3500 years. Most of their plant species, though today considerably less abundant, still exist. What is missing are the repetitive disturbances -- fire, grazing, and in some instances flooding or water saturation -- that originally maintained their structure, diversity, and species richness.

These natural disturbance processes must be replaced to preserve these native landscapes. However, in most cases, we must first mitigate for several decades of historic overgrazing, timber exploitation, and altered fire regimes.

Landscape preservation has two aspects: restoration, and maintenance (Fig. 1). Restoration strives to reassemble functioning ecosystems from relict pieces of the originals, and create a favorable environment for perpetuating their inherent diversity. Maintenance perpetuates intact or restored natural environments. Species diversity and richness -- which was characteristically high in savannas and woodlands -- are references for reestablishing disturbance regimes. 


The most important steps in restoring Missouri's savanna and woodland ecosystems are to: A) reopen their midstories, and B): reinstate fire regimes. The former provides light to the ground layer, which is critical for the many thousands of individuals of the several hundred plant species that potentially exist there. With it comes the necessary environment for the surviving propagules to grow and reproduce. The latter, fire, nurtures and sustains these reviving plants as they become competitive with the understory trees and sprouts. The restoration-regime fires must do several things: create and maintain an open midstory; hold the oak sprouts in check while the rest of the plant community redevelops; stimulate expansion of the numerous grass and sedge species that will eventually carry the fires of a natural regime; stimulate reproduction in herbaceous plants; remove surface litter and provide germination sites; and generally stimulate redevelopment of the forb community. These play several ecological roles, including nitrogen fixation (legumes and nitrogen fixing forbs like Ceonothus are abundant in fire-influenced woodland communities) and provide allelopathic agents to counter competing woody plants, as Godman, Yawney, and Tubbs, 1990, reported for asters and goldenrods, (also abundant in savannas and woodlands) such as sugar maples that often invade fire protected relicts. 

A.  Restoring Community Structure: 
At the ground level savannas require light, and its relative abundance will largely determine the herbaceous response. Well drained, shallow soil sites, and those with gently rolling topography that carried fire well, characteristically had open canopies. Their understories were grassy, with strong prairie affinities. Canopies were more closed on moister, deeper soil and deeply dissected, or poorly drained sites where fire wasusually less intense or frequent. There understories tended towards forb dominance, and to more shade-tolerant woodland grasses and sedges. Typical upland Missouri landscapes contained several savanna and fire-influenced woodland community types across their undulating topographies. Along with them the more classic multi-layered forest ecosystems occured lower on steep north facing slopes, in coves and alongside rivers, and in other fire-secluded areas. Recreating the proper structure requires recognizing this variation, and replicating tree densities for which the physical setting and relict understory plant community is naturally suited. 

The procedure for providing light varies, depending on the type and quality of the site being restored. If the tree and brush intrusion is fairly recent, a restoration-oriented prescribed fire program works well. This is often all that's needed, because many of Missouri's former savannas remained relatively open through the end of the open range era (early 1960's) and beyond. But where the understory trees are dense and three to four inches in diameter, moderate prescribed fires do not easily remove enough of them to reopen the midstory. Less light has resulted in lower species density and less aboveground growth, even after ten years and as many as seven prescribed fires (Fig. 2). 

If dense midstories need reducing, two techniques are typically used. Mechanical thinning is one expensive but effective approach. Hiring crews to thin, treat stumps, and burn the slash has run as high as $1200 per acre for heavily tree-infested sites. Contracting to fell cedar trees (the common invader of shallow-soil barrens and glade complexes) has cost us from $50 to $175 per acre. Where there are markets, sales of firewood, and of cedar and oak posts, poles, and timber, have also worked. In all cases, more light results in abundant oak sprouts. Treating freshly cut stumps with an herbicide such as Garlon is almost essential to avoid having to shape early fire regimes more for sprout control than for restoration of herbaceous plants.

Most DNR restorations where thinning is applied deal primarily with removing midstory tree layers of fairly recent origin. Shade-tolerant, fire sensitive species are removed, as well as all but a replacement guild of the young of canopy species (if any are present, which frequently is not the case). In the drier sites, the original overstory trees often still survive. In the moister ones, the difference in tree species composition between overstory and midstory is a guide, as is the relict herbaceous plant community. In nearly every case, removing most of the 30 to 50 year old understory trees has solved the structure problems. 

The advantage to mechanical thinning is that it immediately recreates the savanna's open structure, preventing further possible loss of plant species. The alternative to mechanical thinning is a hot initial fire, taking advantage of heavy fuels that usually exist in long-unburned sites. These sites are most often set after the trees have budded in spring, under unstable atmospheric conditions, and at temperatures, humidities, and wind speeds approaching the limits of a typical management fire prescription. They work best in steep terrain, where landform promotes convection columns that magnify the effects of hotter burning conditions. Such burns are effective at removing midstory tree layers, over much larger areas then the cost of mechanical thinning would allow. But be careful, because this requires highly trained managers and adequate safeguards. There also is a risk in creating too hot a fire, which would damage mature trees that otherwise survived two to three centuries of natural and anthropogenic fire. 

B. Reinstating Fire:
Where a fairly open savanna structure is already present, or achieved, a prescribed fire program becomes the principle restoration tool. This fire is applied in a restoration mode, different from it's presettlement counterpart. It has two major components: frequency and timing.

Most savanna managers burn frequently in the early stages, primarily to control oak sprouts and prevent litter buildup while the herbaceous plants recover. Sixty percent of the DNR's restorations are burned each year, with some areas having been burned six times the past six years. Our average restoration site is burned on a one to three year cycle, with those having the poorest herbaceous ground cover receiving the most frequent fire; except where serious erosion is a concern or fuel buildup is slow. Repeat annual fires tend to be light, sometimes spotty because of low fuel loads; but they are effective against the oak sprouts and stimulate a good herbaceous response. On dry sites, where fuels simply do not accumulate fast enough to support annual fires, the return interval is expanded appropriately. Ultimately, the fire frequency decision must balance fuel loading and potential fire intensity against advances in the inevitable oak sprout growth rate during the early restoration stage. Later, sprout vigor tends to be reduced, and fire frequencies are based on annual assessments of other plant community responses. Theoretically, the fire frequency as the herbaceous community becomes more established, will more closely approximate that of the site in presettlement times. Future research andexperimentation will help establish a range of parameters.

Fire timing varies according to the restoration stage and site quality. Most early DNR restorations are burned sometime between late winter and mid spring. The dormant season burns often tend to leave some of the wet or frozen litter layers intact, and thus protect the soil of degraded remnants. These top-kill woody plants, but do not scorch the still dormant herbs and grasses. They usually are controllable because of the stable atmospheric conditions under which they typically occur, but can still burn hotly with strong winds and afternoon humidities in the 15 to 25 percent range in winter or 30 to 35 percent in early spring. Standing dead trees and snags are often too wet to smoulder or catch fire. 

Mid- and late-spring burns are well known for their ability to top-kill woody plants, and stimulate, especially, the warm season grasses. Green-up occurs rapidly after the fire, so soil erosion is less often perceived as a concern. If burns occur as the buds are swelling or new leaves unfolding, fire also places an extra drain on carbohydrate reserves in competing woody plants. These fires can be especially hot, and effective at removing midstories, because of the fuel and weather conditions of a typical, unstable mid-spring burning day. They do impact early spring flora, much of which is producing green tissue, often even flowering at that time.

Fall burns are another option in the restoration sequence. They work well where light leafy fuels are the only fire carrier, because the cover is still fluffy and intact through the warm, dry, Indian summer period. Fall burns provide germination sites for fall germinating seeds, and green fall and winter forage from the cool-season grasses, sedges, and perennial plant rosettes. They presumably provide an advantage to the forb community, different from that given the grasses by spring fires; and give greater access to the fall seed crop by wildlife. Based on the frequency of historical references to fall fires, and the typicla fall burn-conducive fuel and weather conditions, it is likely that the function of savanna and woodland ecosystems is tied in many ways to fall fires. They become a significant part of our management prescriptions in later restoration and maintenance stages. Approximately 35 percent of DNR's burns take place from October 15 to December 15. A few are being done in late August through September.


As fire and light are restored to these fire and light dependent communities, they progress through fairly predictable stages. Phase I is dominated by brush. It is the normal starting point for areas suffering the typical Missouri pattern of several decades of open range grazing followed by periods of fire exclusion. In Phase II, the brush remains, but with it come increases in certain forbs and grasses, and a fairly robust herbaceous understory dominated by the less conservative species. Phase III sees the development of a dense and species rich herbaceous community in which grass, sedge, legume, composite, and other species predominate. With them will be many the more conservative species characteristic of mature communities. The following paragraphs generally describe these phases, according to our observations from DNR restorations. Nomenclature follows Yatskievych and Turner, 1990.

Saplings and basal sprouts of the oaks and hickories typically dominate Phase I, usually with numerous clumps of fragrant sumac (Rhus aromatica). Grape vines (Vitis sp.), dewberry (Rubus sp.) and, on acidic sites, lowbush blueberry (Vaccinium pallidum) are also often common. These are extremely aggressive in the absence of competition, and become all the more so as light levels increase. Herbaceous species will likely number between one hundred to two hundred, but they will be widely spaced and low in coverage; much less than that of the woody species. The sedge Carex albicans var. muhlenbergii, Muhly grass (Muhlenbergia sobolifera), and Lespedeza procumbens are three of the more common herbaceous dominants in this phase for upland areas; Chasmanthium latifolium, Diarrhena americana, Desmodium glutinosum, D. marilandicum, Amphicarpaea bracteata and others in more mesic settings. Sites that suffered extreme and prolonged grazing pressure will have a preponderance of buckbrush (Symphoricarpos orbiculatus). 

Phase II is marked by a three to eight fold increase in aboveground understory biomass over that of untreated or Phase I restorations; with consistent gains in the proportion of that biomass occupied by forbs (Fig. 3). Early in Phase II, a few of the more easily spreading herbaceous species form fairly large colonies. Notable for this are Lespedeza procumbens and Muhlenbergia sobolifera in drier sites, and Chasmanthium latifolium in more mesic ones. Helianthus hirsutus, Verbesina helianthoides, Desmodium paniculatum, Solidago ulmifolia, Aster turbinellus, A. patens, A. anomalus, and several other forbs will typically become abundant. As recovery advances, the other herbaceous plants gain greater dominances and distributions. Common among these are the grasses Agrostis perennans, Bromus pubescens, Danthonia spicata, Dichanthelium dicotomum, D. sphaerocarpon,D. boscii, D. linearifolium, Elymus virginicus, Schizachyrium scoparium, and others; sedge species including Carex brevior, C. cephalophora, C. complanata, C. muhlenbergii, and others; and, again, a multitude of forbs (particularly the legumes and composites). Most species become fairly robust and generally distributed across the site, and there are normally a few scattered, but expanding, concentrations of the more conservative species that are typical of mature or recovered systems. Areas that did not suffer midstory closure typically kept a fairly heavy grazing history, and so are usually dominated instead by species such as Chamaecrista fasciculata, Dichanthelium acuminatum, Danthonia spicata, Erichtites hieracifolia, Erigeron strigosus, Eupatorium serotinum, Rudbeckia hirta, Schizachyrium scoparium, Tridens flavus, Vernonia baldwinii, and other adventives and grazing increasers.

Phase III is marked by a dense understory layer that still includes robust tree sprouts and shrubs, with a strong grass and forb community. In the years immediately following fires, the grasses and forbs comprise half or more of the aboveground understory plant production; and little bare ground is visible from eye level. Species density averages fairly high, with 8 to 10 species or more per quarter meter square plot; and the conservative species rise in importance value. More importantly, the concentrations of these conservative species expand and spread during this phase. The driest, best drained, and most open sites will have many prairie dominants like Andropogon gerardii, Sorghastrum nutans, and Schizachyrium scoparium; but the grassy matrix will include strong representation of as many as 60 other grass and sedge species. This rich grass and sedge community is one of the trademarks of savanna ecosystems. So are the rich assortments of forbs, that include prairie species like Ceonothus americanus, Amorpha canescens, Tephrosia virginiana, Eryingium yuccifolium, Coreopsis palmata, Parthenium integrifolium, Baptisia bracteata, Liatris aspera, and several others that actually reach some of their peak abundances in savanna systems. The understory forbs also contain species that are more characteristically savanna, such as Aster anomalus, A. linariifolius, A. cordifolius, Aureolaria grandiflora, Solidago petiolaris, Silene stellata, Lespedeza hirta, L. violacea, Monarda bradburiana, Echinacea purpurea, Asclepias purpurascens, and Penstemon pallidus. Where less light is available, or when canopy closure is 90 percent or greater, the typical prairie species disappear in favor of more shade tolerant savanna and woodland species. These are more closed and mesic sites, so they tend towards a forb-weighted species composition that includes numerous sedges and woodland grasses. They are productive, species rich communities in their own right that also respond positively when fire returns as a recurrent natural influence.

Presumably, the increases in understory production, species number, diversity coefficients, and coefficient of conservatism will become smaller through the years. For sites that never lost most of their original plant propagules, this signals the theoretical end to restoration, and beginning of maintenance. In contrast, consistant low species numbers and conservatism coefficients (relative to similar good quality relicts) signals historic damage and perhaps the need to begin reintroducing lost species.


In restoration work, how completely the original community recovers, and how quickly, depends entirely on what propagules survive or can be reintroduced. Spontaneous generation does not work; and even if most of the species still exist, recovery from decades of damage will not happen overnight. The trend to progress from low quality relicts through phases of increasing diversity, understory coverage and biomass, and conservative species abundance has been repeatedly accomplished; but the good ones take several years, and the poor ones a decade or more to even secure Phase II. In general, the denser and more diverse the herbaceous community to start with, the faster and more completely restoration occurs. More time will show us the full range and extent of species recovery that is possible for savanna and woodland ecosystems through restoration. 

Work to date indicates, however, that large acreages in Missouri still have the potential to recover substantial proportions of their native character and diversity. In no case should we presume that their original diversity is lost, because thousands of acres of restorations are showing otherwise. As little as ten years of effort has much improved species diversity, and community diversity and variability, even on highly degraded sites (Fig. 4); and the best have recaptured rich expressions of their native floristic potential. 

This leads us to the fact that our best biodiversity investments are those where the native elements still occur in their original ecosystem context. The gains in reviving living systems are more certain and less costly than recreating extirpated ones; and a surer bet than assuming that whatever replaces the originals in today's highly altered, fragmented environments will be equally as diverse and beneficial. This is essentially the decision we are dealing with in determining whether to restore our savanna and woodland ecosystems, or letthem go. Restoration is teaching us that this opportunity may not always exist, because the longer these relicts go without light and fire, or suffer more overgrazing or exploitation, the longer, more costly, and problematic their recovery becomes. 


Brackenridge, H.M. 1816. Journal of a voyage up the river Missouri; second edition. Reprinted in: Thwaites, R.G. 1904. Early Western Travels 1748-1846. Vol. VI. A.H. Clark, Cleveland, OH.

Delcourt, H.R., P.A. Delcourt, G.R. Wilkins, and E.N. Smith. 1986. Vegetational history of the cedar glades region of Tennessee, Kentucky, and Missouri during the past 30,000 years. Assoc. Southeastern Biologists Bulletin 33(4):128-137.

Evans, K.E., and R.A. Kirkman. 1981. Guide to bird habitats of the Ozark Plateau. USDA Forest Service General Tech. Report NC-51. pp. 214-225.

Godman, R.M., H.W. Yawney, and C.H. Tubbs. 1990. Acer saccharum Marsh, sugar maple. In: R.M Burns and B.H. Hankala, tech. cord., Silvics of North America, Vol. 2, Hardwoods. Agriculture Handbook 654. USDA Forest Service, Washington, D.C. pp. 78-91.

Gremaud, G. 1992. The forested natural communities of Missouri. In: Towards a Vision for Missouri's Public Forests. A.R.P Journet and H.G. Spratt, eds. Southeast Missouri State University, Cape Girardeau, MO. pp. 35-40.

Houck, L. 1908. History of Missouri. Lakeside Press, Chicago, IL.

Iffrig, G., and P.W. Nelson. 1983. Savanna stewardship in Missouri state parks. Missouri Prairie Journal 5(1-2):3-12.

Ladd, D. 1988. Flora of Missouri, with coefficients of conservatism. Missouri Field Office, The Nature Conservancy. St. Louis, MO.

Ladd, D. 1991. Reexamination of the role of fire in Missouri oak woodlands. In: Proceedings of the Midwest Oak Management Workshop. Eastern Illinois University, Charleston, IL. pp. 67-80.

Marbut, C.F. 1914. Soil reconnaissance of the Ozark region of Missouri and Arkansas. Bureau of Soils, United States Department of Agriculture, Washington, DC.

Nelson, P.W. 1985. The Terrestrial Natural Communities of Missouri. Missouri Natural Areas Committee, Jefferson City, MO.

Nigh, T.A. 1992. The forests prior to European settlement. In: Towards a vision for Missouri's public forests. A.R.P. Journet and H.G. Spratt, eds. Southeast Missouri State University, Cape Girardeau, MO. pp. 6-13.

Nuzzo, V.A. 1986. Extent and status of Midwest oak savanna: presettlement and 1985. Natural Areas Journal 6(2):6-36.

Schoolcraft, H.R. 1821. Journal of a tour into the interior of Missouri and Arkansas in 1818 and 1819. London. Reprinted in 1955 by Press-Argus Printers, Van Buren, AK.

Schroeder, W. A. 1981. Presettlement Prairie of Missouri. Natural History Series No. 2, Missouri Department of Conservation, Jefferson City, Mo. 

Yatskievych, G., and J. Turner. Catalogue of the Flora of Missouri. Missouri Botanical Garden, St. Louis, MO.


Figure 1

Figure 1.  Restoration and Preservation Protocol for Native Woodland and Savanna Landscapes Currently Practiced by the Missouri Department of Natural Resources

Figure 2

Figure 2. Examples of Differences Made by Midstory Tree Thinning in Two Ten Year Old Savanna Restoration Projects

Figure 3

Figure 3. An Example of How Understory Biomass Differs Between 5 Adjacent Savanna Restoration

Figure 4

Figure 4. Restoration in a Highly Repressed Former Savanna


Begin Site Footer

EPA Home | Privacy and Security Notice | Contact Us