Jump to main content.

1994 North American Conference on Savannas and Barrens

Midwest Oak Ecosystems Recovery Plan DRAFT September 1994


    1. Table 1:  Midwestern endemic rare plants identified from analysis of Appendices A & B
    2. Table 2:  Other rare/endangered plants reported from savanna/woodland communities in the Midwest
      1. Table 2a
      2. Table 2b
    3. Table 3:  Rare Animals of Midwestern Oak Savannas and Woodlands
      1. Table 3a
      2. Table 3b

Appendix A:  Attachments
Appendix B:  Comprehensive Midwest List of State-listed plants that are reported from savannas
Appendix C:  Geographic Affinities of State-listed Savanna Plant Species
Appendix D:  Status of Rare Animals associated with Midwestern Oak Savannas and Woodlands
Appendix E:  Research Questions for Restoration of Oak Savannas and Woodlands

A savanna is any area where scattered trees and/or shrubs and other large persistent plants occur over a continuous and permanent groundlayer visually dominated by herbs, usually graminoids.

Subject Editors:
John Bacone
Sonja Eichelis
Don Faber-Langendoen
Alan Haney
Doug Ladd
George Malanson
Noel Pavlovic
Larry Stritch
Coordinating Editors:
Paul Botts
Alan Haney
Karen Holland
Steve Packard


This second draft toward a Midwest Oak Ecosystems Ecosystem Recovery Plan is the product of technical working sessions held in conjunction with the Midwest Oak Savanna Conference in Chicago in February 1993. Scientists, natural resource managers, and practitioners met to pool their knowledge of fire-dependent oak ecosystems, and focus public and private resources on their restoration and management. With the assistance of Working Session and Conference organizing sponsors, The Nature Conservancy, Northeastern Illinois University, the U.S. Environmental Protection Agency, and the University of Wisconsin-Stevens Point, the 107 experts from public agencies, non-governmental organizations, and academic and research institutions contributed freely of their experiences and expertise prior to and during the Working Sessions. Their comments and input are the basis of this second draft of the Recovery Plan.

In the months after the conference, the Recovery Plan was revised by section editors. They are John Bacone, Sonja Eichelis, Don Faber-Langendoen, Alan Haney, Doug Ladd, George Malanson, Noel Pavlovic, and Larry Stritch.

More than 40 recommendations for oak ecosystem research surfaced during the Working Sessions. Many suggestions were overlapping or were kernels of ideas requiring expansion and development. All research recommendations were turned over to the Interagency Committee on Ecosystem Management (ICEM) Research Workgroup and discussed, catalogued and edited for the purpose of developing a concise, specific, set of research needs. These needs are outlined in this document. ICEM is a working group of staff representing a diverse group of federal and state agencies. The Research Workgroup assists agencies in developing research projects that are relevant to resource managers and practitioners. Workgroup participants include Sybill Amelon, Fran Harty, Rich Henderson, Ken McCarty, Noel Pavlovic, Lisa Potter-Thomas, Larry Stritch, Ron Sundell, and Richard Whitman. Paul Botts, Bill Franz, and Karen Holland helped to facilitate the discussions.

Because the editors and the Research Workgroup worked independently of each other, a comprehensive edit to smooth out different writing styles, eliminate redundancies, and review all submitted materials to make sure nothing was overlooked was needed. Paul Botts, Alan Haney, Karen Holland, and Steve Packard took the section editors' and Research Workgroup's work and edited it into this postconference draft of the Oak Ecosystems Recovery Plan.

Once again, this is a draft. More work needs to be done. This October 15 and 16, at the Conference on Savannas and Barrens to be held at Illinois State University in Normal are invited to review this draft. Based on comments received during this review, a third draft will be developed. This Plan will be a focus of a September 1995, conference on midwestern oak ecosystems in Missouri.

I. Definition

The scope of this plan was articulated originally as "the fire-dependent oak ecosystems of the Midwest." This is the part of the original natural landscape that had been so poorly served by traditional conservation efforts that for decades had focused on forests, wetlands, and prairie.

In referring to such a forgotten portion of the landscape as "savanna," we followed Curtis (1959), who used the term savanna generically to refer to fire-dependent wooded grasslands. However, most state conservation programs substantially increased the canopy coverage limits from those which Curtis used. This extension was a natural one for conservation purposes, since it provided a framework for conservation of a portion of the landscape continuum that hitherto had been essentially ignored, so far as natural areas protection and management was concerned.

As interest in this area has increased, use of the term "savanna" to describe the more heavily wooded parts of the continuum has led to unproductive debate and confusion. Therefore, in this draft, the word "woodland" is used as a technical term indicating that part of the continuum lying between open savanna and forest.

Thus, following Eiten (1986) and others, and for the purposes of this Recovery Plan, we define savanna (both wooded and shrub grassland) as follows:

A savanna is any area where scattered trees and/or shrubs and other large persistent plants occur over a continuous and permanent groundlayer visually dominated by herbs, usually graminoids.

No particular vegetation type is implied in the name, except to exclude weedy or exotic vegetation. A number of world-wide or nation-wide classifications have added some refinements to this broad definition (UNESCO 1973, Driscoll, 1984).

In this recovery plan, then, "savanna" will be used in the popular sense as a physiognomic category encompassing what also might be called "wooded grassland" or "shrub grassland". "Barrens" is frequently used interchangeably with the latter, and is treated here as a type or condition of savanna in which a combination of poor soil and frequent, intense fire prevents tree forms, even of arborescent species such as oaks. Some generalization about common use to nomenclature may help:

For the purposes of this Recovery Plan, we define woodland as follows:

A woodland is an area where moderately dense trees and shrubs (30% to 70% canopy) occur over a continuous and permanent herbaceous groundlayer.

The woodland portion of the continuum has been little studied in its natural state, that is, under a regime of regular fire. Thus all discussion must be ever more tentative than for the open savannas.


The term "savanna" has been in use since at least the 16th century. Excellent reviews are available from Bourliere and Hadley (1983) and Eiten (1986, 1992). Bourliere and Hadley point out that the popular usage of the term originally referred to a grassland without woody vegetation. Over time, and particularly among naturalists and botanists in the 19th and 20th centuries, the term was applied to areas with a continuous, usually tall, grass layer with scattered trees.

As Eiten (1986) states, viewpoints about the definition of savanna fall into two major groups: a) that savanna is a physiognomic term referring to vegetation with scattered trees and/or shrubs (or stemless palms), over a continuous and permanent ground layer visually dominated by herbs, usually graminoids, and is applicable to any vegetation type from the equator to the poles; or b) that it is a large-scale vegetation type found only in the tropics-subtropics, intermediate between rainforest and desert. When used as a tropical large-scale vegetation type, "savanna" may include many physiognomic forms, including, as Bourliere and Hadley (1982) and Eiten (1986) note, "savanna grassland," "low tree and shrub savanna," "savanna woodland," and even "savanna forest." Those interested in the many complexities of the application of the term savanna in both senses should refer to Eiten's article. Its application in the tropics has been troublesome to the point that some tropical classifications, even in Africa, have dropped the term (UNESCO 1973, White 1983).

Many midwestern naturalists and ecologists have accepted the first viewpoint, namely that savanna is primarily a physiognomic term. They have chosen to apply it to Midwestern vegetation (Dyksterhuis 1957). Common terms used in the Midwest for savanna-like systems are "oak openings," "plains" (this could include prairie and savanna) and "groves." Curtis (1959) defined savannas as areas with a dominant grass layer and tree canopy cover up to 50%. Others have used similar definitions, but specified canopy covers between 10 and 25% to 10 and 90%. State Natural Heritage Programs have often operationally used savanna in a broad sense to include groundlayers dominated by grasses and where tree cover ranges from 10-50% (Curtis 1959), 10-80% (INAI 1978, Nelson 1985), and 10-100% (Anderson 1982).

If the term savanna is to be used broadly, it should be subdivided into separate categories as is done in the tropics. More usefully, given the original meaning of the term savanna as a grassland, it seems appropriate to call a closed savanna something else, namely woodland, as is now also done in the tropics. This was proposed by Penfound (1962), who distinguished forest, woodland and savanna types in Oklahoma. Such an approach is used by the Minnesota Department of Natural Resources Natural Heritage Program (1993), which defines savannas and woodlands as follows: A savanna has more of a scattered tree canopy with 5-50% (70%) total cover; the matrix surrounding the trees is greater than 30% open grassland or primary communities, and the tall brush cover is generally sparse. A woodland contains a broken to scattered tree canopy, with 30 to 70% (10%) total cover; the matrix surrounding the trees is less than 30% open grassland or primary communities, and the tall brush cover is sparse to dense.

Other naturalists and ecologists agree that a distinction is needed for systems different from either prairie or forest, but suggest "barrens" as a more historically accurate term in the Midwest (Hutchison et al 1986). However, in Missouri "savanna" and "barrens" were used interchangeably (Ladd 1991). The term "barrens" was extensively used in the pre-European settlement survey records in at least several Midwestern states and occasionally by midwestern scientists (Vestal 1936). In a popular sense, barrens most often indicates a site with open, often scrubby vegetation and droughty characteristics, though in Missouri it could also refer to a mosaic of open prairie, open glade, semi-open savanna and open woodland (T. Nigh, pers. comm). It may or may not contain a dominant graminoid layer and can even have a primarily bedrock groundlayer. Unlike savanna, the term barrens has less often been technically defined. In addition, some argue that shrub oak savannas need to be recognized, with shrub cover of 10-25%. This is an active area of research in the Midwest.

From the above discussion it would appear that the terms barrens and savanna are more-or-less equivalent, though not identical. Thus, if the term savanna is used, barrens per se should not be treated merely as a subtype. It is important to emphasize that definitions of savannas should use the joint contribution of trees, shrubs and grasses to total cover, and that the definition of savanna types should proceed beyond structure and define the floristic composition.


At one time, prior to European settlement, oak savanna and woodland ecosystems occupied a significant portion of the Midwest, probably from 11,000,000 to 13,000,000 hectares. Klopateck et al (1977) estimates that 17-22% of Midwest oak savannas still exist; however, most are highly degraded as a result of timber harvesting methods, over-grazing, agricultural use, fragmentation, and especially, fire suppression.

From a regional perspective, oak ecosystems were probably more stable between 8,000 and 2,000 years ago when the prairie peninsula expanded eastward through the Midwest under hotter and drier conditions than now exist. It is believed they existed in a dynamic equilibrium relative to the vagaries of fire from both lightening and Native American causes. Although oak ecosystem maintenance requires fire, dry climatic conditions favored oak ecosystem species over those inclined to invade in the absence of fire.

As climate moderated, local and regional stability probably decreased as less fire-tolerant species became more prone to invade oak savannas and woodlands. This instability has been exacerbated by post-European settlement anthropogenic actions, especially elimination of bison and elk, introduction of exotic species, and fire suppression.

The prairie-forest border that extends throughout the Midwest has been recognized for some time (Anderson 1983). Over the past 10,000 years, as glacial activity receded northward, the vegetation responded to varying climatic and topographic conditions in a complex fashion.

Efforts to reconstruct the distribution of prairie are useful in indicating how the savanna and woodlands must also have varied (Webb, Cushing and Wright 1983). Prairies moved eastward and westward in response to the changing climate. This climate acted on the vegetation indirectly through fire regimes and topography (Grimm 1983).

Tree composition of the prairie-forest border varied as tree species migrated differentially based on dispersal characteristics and responses to changing climatic conditions (Grimm 1983, Jacobson and Grimm 1986). Many efforts have been made to map pre-European settlement vegetation patterns in the prairie-forest border. These efforts include mapping savanna distribution across entire states (Finley 1976, Marschner 1974).

These maps, which use the early government land survey descriptions, indicate a widespread distribution of oak savanna and woodland vegetation. They are also a snapshot in time, so care must be taken not to infer long-term ecological processes strictly based on spatial distribution. A major implication of all such maps, however, is that oak savannas and woodlands were widespread on the landscape (savanna and woodland are rarely distinguished on state-wide maps; but see Anderson and Anderson 1975, Rodgers and Anderson 1979, Leitner et al 1991, for county applications of these categories). Descriptions in the journals of early European settlers verified the widespread distribution of oak ecosystems.

European settlement either radically altered or eliminated the ecological processes that maintained healthy oak savanna and woodland ecosystems. In fact, processes have been detrimentally affected to a far greater degree than for most other natural systems (forests, marshes, bogs), such that we have few healthy examples with which to understand the past. What we now document and describe are systems functioning with altered ecological processes. Some of these ecosystems may have the potential to function as did their pre-European settlement analogs; others may not.

The prairie-forest border was never fixed on the landscape and some ecosystems no doubt would have changed regardless of human influence. The effects of human influence versus natural dynamic changes, however, cannot always be easily teased apart.

Top of page


"Beyond question, an oak savanna with an intact groundlayer is the rarest plant community in Wisconsin today."
(Curtis, 1959)

Nuzzo (1986) estimated that only 0.02% of pre-European settlement oak savanna areas still supported good savannas. In 1985, 113 sites totaling 2,607 hectares of relatively high-quality oak savanna communities of various types were located. Two-thirds were on sandy, rocky, or similarly droughty sites. Nuzzo found no intact, high-quality, deep-soil mesic savannas. Continued work throughout the Midwest confirms this dismal appraisal, although a few additional high-quality fragments have been located, and many degraded remnants are responding well to restoration efforts.

Oak savannas and woodlands on drought-prone sites are much more numerous, contain most of their potential plant diversity (often badly degraded), and respond more quickly to restoration efforts. In contrast, mesic savannas are less common and much more degraded. Because they are so overgrown with trees and shrubs, many can scarcely be recognized. Under pre-European settlement fire regimes mesic savannas may have contained a greater species richness than oak savannas on drier sites. However, a rapid increase in or invasion of a few aggressive native or non-native species following fire suppression has resulted in greater loss of species richness of native plants and a corresponding loss of animal species (Apfelbaum and Haney 1991).

Curtis (1959) suggested that on mesic sites where oaks had been replaced by shade tolerant mesophytes, fires were capable of eliminating all trees, resulting in the development of mesic prairies. On more xeric sites, according to Curtis, fires did not eliminate oaks, and woodlands became savannas. Infrequent fire, therefore, would tend to eliminate mesic savannas while maintaining xeric savannas.

Mesic savannas are even less common now because they occurred on better agricultural soils, were more easily converted to mesic forests, and tended to occur in areas that were more intensively developed. One of the rarest but formerly most widespread mesic savanna types is the tallgrass savanna found on level or gently rolling rich soils. Bur oak is the prevailing tree species.

In the absence of fire, all mesic savannas are quickly overwhelmed by an increase in woody plant species, including exotics such as European and smooth buckthorns. Native herb species are soon out-competed by woody species resulting in the reduction or outright loss of the herbaceous layer (Apfelbaum and Haney 1991). The increased shade from the woody species contributes to the reduction in herbaceous species. With deepening shade and increased moisture, leaf litter decomposes, leaving the soil surface exposed to erosion, and making fire nearly impossible. This syndrome is also common in other mesic savannas, but usually develops more slowly.

The most extensive mesic savanna complex still existing occurs on Walpole Island, Ontario. Continued burning by Native Americans maintains savanna communities together with associated wetlands and prairies.

How abundant oak ecosystems were at the time of pre-European settlement depends on how these ecosystems are defined and interpreted. Curtis (1959) tended to view savannas as transition communities between grasslands and forests. He argued that the understory of oak openings were dominated by prairie forbs and grasses, along with a few forest herbs. His data indicated only six species that reached peak dominance in oak openings.

Packard (1988) points out that most of Curtis's work occurred in the 1940s, after 20 years of aggressive fire suppression and nearly 100 years after European settlement of the region. Many plant species, Packard argues, which were abundant in the pre-European settlement savanna landscape, were already uncommon by the time Curtis documented the vegetation of Wisconsin.

Top of page


Ecologists and conservationists began describing what was left of oak savannas in the 1940's. In the Upper Midwest, much of the early work was done at the University of Wisconsin by John T. Curtis and his students, J. Roger Bray and R.P. McIntosh. Since much of their work guided other savanna researchers, it is important to review their contribution to savanna characterization.

In general, Curtis emphasized both the continuous nature of floristic change among stands along gradients, with no discrete associations being evident, and the dynamic nature of the vegetation. Vegetation composition does not reflect a tight connection to site factors, but ultimately depends on historic factors. The continuum was a reflection of both abiotic gradients and historic factors.


Concerning structural differences, Curtis (1959, p 331) cited Clements (1928), who stated, "The transition [from savanna] to forest or woodland is usually gradual, and it is impossible to draw a sharp line between the two." Curtis arbitrarily divided the physiognomic continuum into prairie (trees < 1 acre), savanna (trees > 1 acre and up to 50% canopy cover), and forest (canopy cover > 50%). He also noted that tree density and spatial distribution varied, with extreme aggregation being the rule.

Curtis considered the prairie-savanna distinction problematic. He acknowledged that the limit of savanna at 1 tree per acre was highly artificial and may be biologically misleading, since in his words, "Savanna lands were recorded by the land surveyors where the trees were an average of more than 200 feet apart and where many witness trees were as far apart as a quarter of a mile. Such very open savannas would be classed as prairies in this study, but their physiognomy as seen by a traveler and their behavior with respect to fire would be very different from those of a true prairie. As a matter of fact, all of these savannas are related to brush prairie rather than true prairie." Curtis, along with Bray (1960) also mentioned that, despite the presence of a tree layer, savanna was dominated by grasses.

Thus in Curtis' mind there was, in addition to the prairie continuum along a moisture gradient, a prairie continuum along a structural gradient, which he distinguished as "true prairie" and "brush prairie". However, he made no summary of floristic composition based on these categories, simply stating that the true prairie was most common in southwest Wisconsin, and brush prairie more common in south central and central Wisconsin. He did note that because mesic prairies were almost completely utilized by farming, most sites were along railroad rights of way and old cemeteries.


In questions of classification, this plan follows the Midwest Regional Community Classification (see Attachment 2 of Appendix A) developed by the Nature Conservancy in cooperation with state and federal conservation agencies (Faber-Langendoen, in preparation).

Following Eiten (1986) and others, savanna (both wooded and shrub grassland) is defined as follows: "A savanna is any area where scattered trees and/or shrubs and other large persistent plants occur over a continuous and permanent groundlayer visually dominated by herbs, usually graminoids."

Woodlands, by contrast have an open to partially closed canopy where shrubs, forbs and other non-graminoid plants may dominate or codominate with the graminoids. The groundlayer is no longer the clear dominant. In this way the range of physiognomic classes for all vegetation can be defined as follows:

Additional categories can be recognized for SHRUBLAND (25-100%, including shrub thickets) and SHRUB GRASSLAND (10-25%). An outline of the limits of savanna are presented in Table 1.

This classification is offered for several reasons:

Several additional classification systems were offered for consideration as a part of this plan (see Appendix A). Two of them, by Rich Henderson and Dennis Nyberg, deserve particular consideration as this plan is refined. Henderson distinguished between two forest types, while Nyberg pegged definitions to easily measured parameters. A chart comparing a variety of definition systems as to canopy cover and certain other characteristics can also be found in Appendix A.


Ecosystem classification systems provide a framework for describing the patterns of ecosystems, including biodiversity. They facilitate communication between scientists and managers. Even the very effort to develop a classification system stimulates discussion and focuses attention on characteristics or attributes that help our understanding of ecosystems. How do they differ, and why? What are the commonalities, and the discontinuities? Describing and classifying ecosystems also calls attention to those that are most rare. This helps guide conservation practices.

Any savanna and woodland classification method must use ecologically meaningful and defensible criteria that allows for their recognition, identification, and determination of restorability. Classification typically results in identification of discrete and definable units, despite the continuous variation present within a unit. We alert the reader to other problems inherent in classification. It is far too easy to design a classification that fixes on a point in time in evaluation and administration of programs of restoration. Private landowners and agencies and others will need to appreciate the dynamic nature of savanna and woodlands to enable restoration to take an appropriate course. We may visit an area and decide, because trees are abundant, that the site is a woodland when a single fire can convert it to a savanna or even a barrens. It is difficult to make the choice to reduce tree cover even though the system historically functioned with much less tree cover. Decisions regarding whether the classification permits or emphasizes historic, current or future (site potential) conditions need to be weighed.


Given the fragmentary condition of the natural vegetation in Wisconsin at the time Curtis studied it (the picture was no different in Illinois, Iowa or northern Missouri), any assessment of original biodiversity is greatly hindered. Curtis attempted to draw together the existing information for Wisconsin using the continuum approach. He usually summarized the vegetation patterns by their position along the moisture gradient, as inferred from the vegetational continuum index, e.g. southern dry-mesic forest, dry prairie. He then determined which species were most prevalent or constant (modal) in each type. Types varied as to how many modal species they contained, with the suggestion that types with few modal species were less distinctive than other vegetation types.

A great deal of debate has emerged over how distinctive savannas were, given that some of the savanna types, particularly oak openings and pine barrens, appeared to be the least distinctive of all community types with the number of modal species found there being only 8 and 7 respectively. Some have argued that savannas were simply an ecotone between prairie and forest, and therefore less distinctive; others wondered whether conservation efforts were all that critical given the lack of distinctiveness and the dependence on disturbances.

Two major cautionary points are in order for any assessment of savanna, for example with respect to Curtis' work in Wisconsin. First, we have already seen how fragmentary were the areas available for study. Second, because of the lack of information, Curtis actually combined several types of oak savannas together: "Because of these incompleteness in both north and south, it has been deemed best to pool the information for all stands of each of the four major types [Pine Barrens, Oak Barrens, Cedar Glades, and Oak Openings] and to present their average composition as a basis for further discussion (1959, p. 328)." Thus Curtis lumped together all dry-mesic and mesic oak openings (and potentially woodlands). Such lumping could have obscured any species that showed strong preferences for certain segments of the oak savanna continuum (Packard 1993). A reanalysis of the Curtis data set is underway (Faber-Langendoen and Drake in prep), and a significant number of new modal species are added to the various oak savanna and woodland community types that sort out along the physiognomic and moisture gradient, suggesting that the lack of distinctive species in oak savanna types was partially an artifact of lumping.

Even with an improved species list for the various oak ecosystems, we are unable to easily understand how to relate this information to restoration activities because few extensive stands remain in which to interpret how stand composition (floristics) was organized spatially and temporally. Nevertheless, floristic lists provide a beginning point for determining what belongs in the system, even if we are unsure about the organization.

Conservation of biodiversity requires the consideration of all levels of diversity: a (community species diversity at a site), including genes, species, communities, landscapes and ecological and evolutionary processes. We focus here on the genetic and the species levels.


The Midwest has relatively low species diversity (see endemic plants in Table 2, for example), in part because its moderate topography does not foster speciation. But the genetic allelic variability of species with populations spanning great ranges (east-west, north-south, wet-dry, sun-shade, etc., see Appendix C) is thought to be great. Although many savanna-inhabiting species are near the peripheries of their ranges, their contributions to the diversity within the species' gene pool should not be discounted. Peripheral populations may be genetically unique and should therefore be protected.

Conserving genetic diversity within savanna inhabiting species depends on identifying populations which encompass the full range of morphological and genetic variation within each species' gene pool. The genetic structure of species populations can be quite different. For some species, much of the genetic diversity is present within each local population. In these cases, local populations should be kept as large as possible. Other species have their genetic diversity dispersed among many populations. In these cases, many individual local populations should be protected. Although knowledge about the genetic variation within savanna-inhabiting species is sparse and difficult to obtain, the following guidelines should improve the likelihood that genetic diversity is maintained during conservation planning and management.


Among the species that are associated with oak savanna landscapes, there are many that are rare nationally, regionally, or at the state level. For example, more than 800 species of vascular plants have been recorded from savanna remnants. Of these, 28 species are listed as threatened or rare. Many species are representative of savanna/woodland communities. Before discussing in detail the importance of savanna for conservation of biodiversity, we need to examine the organisms reported from savanna/woodland communities. Appendices B, C, & D list known rare or endangered Midwestern savanna woodland plants and animals.

In combination, the plant and animal lists illustrate that there is a component of Midwest savanna/woodland diversity that is endemic. Conserving species populations depends on identifying sites or concentrations of viable populations of plants and animals that depend on functional oak savanna landscapes.


Appendix B is a comprehensive list of state listed vascular plants that are reported from savanna/woodland in at least one state. The common names and federal, global, and state status and presence are given for each species. An additional list of 55 coastal plain disjunct species that are associated with depressions in the savanna woodland matrix are also presented. Although these are not representative of savannas and woodlands per se, their conservation in many cases is intimately tied to savanna conservation.

Appendix C is a similar list that is sorted by geographic affinities. From the 211 species recorded from savanna/woodland, only 9 (4.3%) can be considered endemic to Midwestern savannas and woodlands (Table 1). No doubt additional species (Table 2) are quite rare and may be most abundant in these partially shaded savanna woodland communities, but are not limited to Midwestern savanna/woodland communities (Pavlovic and Swisher in prep). What is surprising is that a larger number of species reported from Midwestern savannas and woodlands range into similar open habitats in the Southeast (59 species) and the South (53 species). On the basis of endemic plants the Midwestern savanna woodland complex cannot be singled out for endemism as a single rationale for conservation; however, since many of these rare species represent peripheral populations and may be genetically and evolutionarily distinct, their preservation is warranted.


Appendix D presents a summary of the animals reported for Midwest savannas and woodlands. Little is known about many of these organisms. It is likely that many may now depend on savanna/woodland remnants throughout the Midwest. Reptiles and amphibians of wetlands adjacent to savannas may be dependent on the maintenance of open savanna for thermoregulation of adults and developing eggs during times of high water levels (Resetar pers. comm.). It is quite clear that many of the invertebrates such as the Karner blue butterfly are highly dependent on the conservation of oak savannas.

Table 1: Midwestern endemic rare plants identified from analysis of Appendices A & B
Table 2: Other rare/endangered plants reported from savanna/woodland communities in the Midwest
Table 3: Rare Animals of Midwestern Oak Savannas and Woodlands

V. Management


An outline approach similar to a U.S. Fish and Wildlife Service recovery plan is used here to identify the steps necessary to conserve savanna/woodland biodiversity. Careful consideration must be given to the fundamental goal of the savanna/woodland restoration because we cannot always manage for historical states of a system or region. In addition, the consequences and costs of restoring and not restoring should be clarified.



Achieve highest level of protection possible for species and their sites.

Manage savanna/woodland sites

Natural Processes

Restore priority damaged sites

Law enforcement

Promote ecosystem level management and rare species biology research. Only by integrating species biology and management research can we predict the impacts of management on these species and influence their viability. The goal is to focus on the processes rather than costly single species triage.

Foster development of buffer zones to minimize fragmentation by encouraging appropriate surrounding land use. Need to consider surrounding land use and how it might be modified to minimize the insularity of the landscape. It might be possible that some portion of the landscape may be maintained in a less than natural state.

Encourage federal, state, municipal, and private cooperation in rare species management and protection.

Educate public about oak savanna/woodland conservation and ecosystem conservation in general. The value of biodiversity is poorly recognized by the public. Without public support conservation of oak savanna/woodland biodiversity will fail. Education should incorporate all media and all education, race, and age levels.


Virtually all savannas are degraded to some extent and will require various degrees of restoration. Some of the potentially best savannas are so degraded that the composition and structure of the canopy is typically a poor criteria for their recognition. Therefore, they may not be recognized as savannas and may be lost for lack of proper management.

Sustainable restoration of woodlands and savannas for biodiversity attributes is a new field, with much of the theory and technology being invented as current efforts progress and new efforts are initiated. Derivation of restoration and management technologies will not be a linear process, but will proceed by fits, starts, and setbacks as the science grows and develops. Innovation and open-mindedness will be primary requirements for success.

Community structure refers to horizontal and vertical placement of biotic elements in the community, and has typically been measured for larger components of the system, such as trees. The primary objective of structural restoration goals are to recreate/emulate the pre-European settlement structure of the community, based on the best information available. Structural parameters must be linked to the response of a broad spectrum of the biota to be meaningful.

Compositional restoration goals focus on the re-establishment of the floristic and faunistic components of the community believed to have occurred historically on a given site. Management for a predetermined structural or species abundance level risks disenfranchising other important components of the system in blind pursuit of criteria that may have less ecological validity.

Restoration goals related to environmental criteria focus on re-creating or emulating environmental factors, including prevailing process regimes, to which the organisms comprising the system are genetically attuned. In practice, successful restoration and management of woodland and savanna systems will include elements of all three concepts, with structural considerations largely driven by compositional and environmental parameters.

It is imperative that restoration and management take a landscape approach. Larger areas, encompassing wetlands, prairie, barrens, woodlands, and forests should be incorporated into single management units as much as possible. This often will necessitate cooperation of multiple owners and managers. Using broadscale prescribed fire, across a varied landscape, has the greatest potential to restore the natural mosaic and minimize the fragmentation and frail human judgment that necessarily accompanies single community approaches to restoration.

As much as possible, restoration should allow fire to cull species and plants from the landscape. Exotic plants, however, may require special attention, although care is needed that attempts to eradicate them not endanger native species.

Generally, mechanical removal of exotics, species not tolerant to fire, shade tolerant, and small diameter trees will most closely simulate what fire would have eliminated. Large diameter oaks, especially bur and black, are remarkably fire resistant. Large oaks were often the only trees that survived in savannas that burned regularly.

Many scientists have expressed concern that extreme caution be exercised with plant introductions. Because of widespread disturbance and degradation, it is impossible to ascertain the pre-European settlement diversity and composition of degraded savannas. It is risky to assume that one might introduce all species potentially found on a site, and let survival establish a community composition identical to the pre-European settlement community. Many native species have the potential to become aggressive under some conditions, and could disrupt delicate ecological balances. Where possible, most scientists prefer to put primary emphasis on restoring natural landscape processes, and allowing species to establish their place in the landscape. At a minimum, species introduced to a community should come from local sources, and very similar communities.

Several factors must be considered when evaluating species reintroductions as part of management or restoration work. Savanna/woodland management should be based on a holistic, systems-oriented approach, and not predicated on single species management, or maximizing the prevalence of a small group of organisms. While rare species may be important components of some restoration work, rare species should not distract the focus from conservation of a synecologically healthy, biologically diverse system complex. Common species that are conservative to savanna/woodland systems should be a major component of restoration and management work.


Of the many processes influencing woodland and savanna systems, fire is perhaps the most visible example of a process totally impacted by patterns of post-European settlement anthropogenics. Fire was a regular component of oak systems. Post-European settlement fire suppression has resulted in a proliferation of woody vegetation in contemporary woodland and savanna remnants, with a corresponding reduction of graminoid ground cover matrices and their associated forb diversity. Fire suppression has also resulted in shifts in canopy replacement dynamics.

Any management or restoration scenario for oak woodland or savanna must include provisions for prescribed fire. Although fire seasonality, timing, frequency, and fuel bed characteristics are largely unknown, fire was a common denominator linking oak systems with related components of the pre-European settlement landscape. Fire plays an essential role in nutrient cycling, biotic composition, and structural attributes of woodlands and savannas.

Fire promotes survival of savanna communities by reducing shade, removing duff, stimulating germination, and maintaining habitat structure for certain animal species. In general, diversifying the fire regime for a site will increase biological diversity. Fire diversification includes varying the proportion, location, and size of area burned on any single burn, varying the frequencies that areas are burned, varying the mosaic of burns on a particular site, and varying the intensities of burns by conducting fire management in different weather and seasonal conditions.

Most accept that emulating pre-European settlement fire events in terms of area, extent, seasonality, frequency, and fire behavior will maximize the ecological health of woodland and savanna systems, and facilitate retention of the conservative elements of their biodiversity. Management should include monitoring the site-specific effects of burns and modifications of ongoing protocols in response to the results. No single prescription can apply to all sites.

Fire regimes in the savanna region varied spatially and temporally. Sites on south and west slopes, on the west side of rivers and wetlands, and those with adjacent prairies apparently burned more regularly and probably with greater intensity. Those near major travel routes, villages, and popular encampments of Native Americans apparently burned more often. Dry sites burned more often and with greater intensity than mesic sites. All sites burned more often and with greater intensity during periods of hotter and drier climatic shifts.

Fire frequency and intensity (the fire regime), therefore, was specific to each site and was a major factor in shaping the species composition of the communities that developed.

Fall burning emulates the aboriginal fire cycle throughout much of the Midwest, and allows for fall seeding of degraded areas and the opportunity for seed to stratify over winter. Spring burning reduces the erosion potential of poorly vegetated restoration sites, and allows prescribed fire to be used in areas where the fuel is primarily leaf litter from the previous season. The frequency of burning will vary with the individual area and its current condition. There are many viewpoints regarding fire timing, frequency, and necessity of special protective measures for invertebrates. Some practitioners advocate annual burning, while others favor burning on two to four year rotations or less frequently. Some managers advocate intense fires in the initial stages of site rehabilitation, while others favor cooler, less intense burns. Similarly, there are valid arguments for utilizing spring, winter, fall, or even summer burns.

Each site must be evaluated on an individual basis, in the context of what is known about its presettlement biology and process cycles. Care must also be taken, especially in early stages of site restoration, to insure that desired fire behavior and ecological objectives are achieved. For example, fuel bed characteristics may be so altered that fire behavior is markedly different from fire behavior under similar conditions in an intact site.

Herbivory, sometimes including grazing by large mammals, is a component of woodland and savanna management. Little research has been conducted on diversity-related effects of large mammal grazing in oak woodlands and savannas. In many regions, there is a pre- European settlement antecedent for such grazing and browsing; many of our woodland and savanna systems were utilized by deer, bison, elk, and pronghorn.

Post-European settlement use of woodland and savanna systems more often has been intensive grazing by cattle, or total exclusion of grazing activity in areas identified as having conservation values. The role of low intensity, or short duration, high intensity grazing needs to be explored in these systems. There may well be a cohort of organisms dependant upon the presence of grazers and their associated processes. Mowing has been suggested as an alternative to both grazing and fire management in woodland systems, but does not fully emulate the effects of either process, and the biological compatibility of mowing as a substitute for burning or grazing remains unknown.

If utilized, grazing by livestock should be managed to reflect the grazing patterns of the native elk and bison herds with sufficient time for the recovery of herbaceous vegetation. The general consensus is that grazing should not be allowed in areas where the seed bank contains a significant presence of disturbance-adapted exotics. In some cases, deer are overabundant and need to be controlled at an appropriate level.

Other Vegetation Management
Particularly in the early phases of site restoration, mechanical clearing may be required to achieve management goals, such as the establishment of a graminoid matrix or to facilitate prescribed fire. Removing selected canopy trees and excessively dense native shrubs may be necessary to ensure the survival of the desired savanna community. Often, only the younger invading species need be cut. Opening up the savanna too much can cause vigorous sprouting of oaks and hickories and eliminate the micro-habitat variation provided by scattered trees, unless provision is made for ongoing fire management.

Whenever heavy slash is created by cutting activities, caution must be used in evaluating the fuel potential of the slash and choosing burn conditions that will minimize impacts from burning it. One alternative to cutting and leaving fallen slash throughout an area is to pile and burn it as it is cut. This localizes the effects of the burning and reduces problems associated with fallen slash during the growing season. Experience on several sites has demonstrated that the sites of pile burning have good recovery potential.

Too rapid a reduction in the canopy can lead to severe encroachment of weedy species, so a gradual clearing sequence, undertaken over several seasons, may be more desirable. In areas where good community diversity still exists, the use of manual tools or smaller scale power equipment (e.g., chain saws or weed eaters) is recommended. Hand tools may be good choices for smaller sites. Larger and lower quality areas may be cleared by heavy machinery. Mowers, shredders, and bulldozers are a cost-effective way to remove brush from the large areas. In such instances, the deleterious effects of heavy equipment must be weighed against their increased efficiency. Soil disruptions induced by bulldozers may be acceptable in some areas but not in others.

Regardless the means utilized for the physical removal of brush, in most cases it must be followed with prescribedfire. Opinions vary on the role of herbicides in conjunction with tree and brush cutting, but the general consensus appears to favor their use as stump treatments in initial phases of restoration work.

The benefits of brush removal can be maximized by clearing at the appropriate stage in plant growth cycles, such as before seed dispersal or when root metabolite levels are lowest. Girdling certain species may be more efficacious than the cut-and-herbicide method. Removal of unripe seed heads may be appropriate to reduce undesired species on small or sensitive sites.

Plant Reintroduction
Depending on the condition of the site and its previous land use history, reintroduction of certain species may be desirable. In general, species should not be translocated into high-quality remnants. It is preferable, where possible, to allow native, on-site germ material to recoalesce as conditions favor their establishment, but this is not possible in very degraded areas.

In degraded areas, when brush is removed, a fuel matrix may need to be reestablished, comprised of grasses, sedges, and forbs. Some perennial woodland grasses are neglected components of site restorations. In northeastern Illinois these include long-awned wood grass (Brachyelytrum erectum), broad-leaved panic grass (Panicum latifolium), and leafy satin grass (Muhlenbergia mexicana). Sod-forming woodland sedges, such as Pennsylvania sedge (Carex pensylvanica), may be key components of this matrix. These species provide fuel for the fire regime the woodland system is dependent upon, as well as facilitating water infiltration and pedogenesis.

In very open savanna, on the bared soil resulting from clearing and initial restoration efforts, the planting of an temporary fuel matrix mixed with appropriate native grasses and forbs has been suggested. This would facilitate rapid stabilization of erosion-prone soils and create a temporary fuel matrix, and might facilitate establishment of slower growing native species. A better understanding of how successional restoration can be applied in savannas and woodlands is needed.

Techniques for compositional restoration of plant species include seeding directly into an established matrix (successional restoration), direct seeding into bare ground, establishment of rootstocks and planting of transplants. Savanna composition, especially in the early stages of recovery, will fluctuate dramatically. Certain conservative species develop more slowly than "first wave" species, such as the rye grasses (Elymus) and many savanna composites. These "first wave" species will experience rapid expansion in the early stages of recovery, and eventually decline in importance as the site matures.

A regimen of prescribed fire and weed control will help ensure the integrity and diversity of the restored savanna community. Weedy species, both exotic and native, such as buckthorn (Rhamnus spp.), honeysuckle (Lonicera spp.), garlic mustard (Alliaria petiolaris), Canada and nodding thistles (Cirsium arvense and Carduus nutans), and burdock (Arctium minus) can move into newly cleared sites, temporarily threatening restoration efforts. This problem may result from an inability to establish an adequate and appropriate native plant matrix that can effectively out-compete the aggressive weeds. Perseverance and maintaining or emulating the pre-European settlement process regimes and environmental conditions to which the native biota are genetically acclimated will maximize restoration potential.

The success of efforts to restore savanna plant species is heavily dependant on the timing of seeding and the availability of viable seed sources of local genotypes. If reseeding is to be employed, it should accompany structural restoration. If desirable savanna species are not present on site or are introduced shortly after structural restoration is completed, the cleared area will fill in with whatever species are able to colonize the disturbed area; these are often weedy exotics. Fall seeding after a burn allows natural stratification through frost and thaw, but also risks unacceptable levels of seed predation in some species. Raking seed in after a burn results in a higher recruitment rate than simply scattering seeds, but it is not always practical in large areas.

If adequate seed sources are not available, efforts should be taken to establish seed production gardens to ensure increased propagation of local genotypes. Whenever local sources are available, the purchase of seed from commercial sources should be avoided, thus eliminating questions of genotype and source location. Any seed collection efforts should be undertaken ethically, and so as not to compromise the integrity or sustainability of donor sites. Efforts should be made to insure that this protocol is adhered to by any commercial sources employed in the restoration. As representative examples, some suggested species to include in restoration planting mixes are provided in Appendices 6.1 to 6.3.

The identification of appropriate blends of seeds to achieve viable and sustainable savanna matrices is crucial to restoration efforts. These must be derived on a local basis, based on careful study of contemporary examples and analysis of historical accounts. Successful restoration of these matrices will help to maintain soil structure, and minimize invasion of unwanted pioneer and weedy species. Multiple savanna matrices may need to be identified, for the various microhabitats found in most woodlands. Seeding with the appropriate matrices for savanna restoration should lead to the evolution of a diverse savanna community.

The introduction of certain slow-growing or physically small species may require staged plantings, to increase their survival. Seeds of smaller plant species or of species that develop slowly can be seeded directly into appropriate matrices which will prevent frost heave of delicate seedlings.

Animal Reintroductions
Many conservative animals will require a higher quality habitat than afforded by the initial phases of restoration. Introduction of animals requires habitat of sufficient size and quality for the species. As with plant reintroductions, animal reintroductions should be undertaken only if there is supporting evidence for the species presence as a native component of the local biota. Restoration techniques for animal species include habitat improvement to augment naturally occurring populations, release of captive reared stock, and transplants from areas of surplus population to areas of low or non-existent populations.

Site Selection Considerations
The following size considerations play an important role in site selection. Woodland and savanna remnants of 30-40 acres or smaller may have biodiversity values, especially for plant communities and as reservoirs of locally adapted genotypes. Small sites are subject to degradation from catastrophic disturbances, like tornados, which can occur on a scale exceeding the size of the unit. To protect many animals of woodlands and savannas, especially birds, reserves should be 300-400 acres. To accommodate catastrophic disturbance as well as larger animal populations and area-dependent meta-populations, savanna and woodland reserves should be several thousand acres or larger. In all cases, derivation of conservation goals should be made within a regional or landscape context.

Small nodes of fair quality savanna and woodland remnants are found within larger degraded woodlands. Small prairie remnants often are found near degraded woodlands. In designing woodland and savanna reserves, it is beneficial to select large tracts including these nodes, or to connect nodes through suitable corridors and linkages. Many species that once occupied savannas and woodlands are now restricted to areas at the edges of fields and in rights-of-way. By connecting such nodes, restoring site integrity and process regimes, notably reintroduction of a fire regime, and allowing time, organisms have an opportunity to move across the landscape and reconstitute the communities of which they were part. When connections are not feasible, clustering of preserves may serve to mitigate some effects of fragmentation, especially for highly vagile or dispersable taxa.

Preservation strategies for private lands include working with state/local governments to create incentives that establish reduced property tax rates for preserving natural areas of significant biodiversity. Changes in local zoning ordinances may be required in order to allow burning and other management activities.

In many cases, the most logical sites for restoration are degraded savannas on public lands acquired for conservation purposes. Savannas and open oak woodlands are prime recreational land. The mix of sun and shade, the handsome vistas, the close-up beauty of the grasses and flowers, the abundance of wildlife, including remarkable numbers of brightly colored birds and butterflies, would be quickly appreciated by the public. Unfortunately, most former savanna land in public ownership is now managed as if it were a prairie or forest.

In a large-scale restoration, public hunting may be a component of site restoration, and would serve to engender additional support. Some of potentially restorable sites are extensive. New ones with particular savanna restoration and game-habitat potential could be sought. Restoration in such areas might consist principally of spring burning, accompanied by introduction of native grassland plants, perhaps with particular emphasis on those that provide good food and cover for wildlife. Savannas represents optimal habitat for a number of the most popular game animals (deer, turkey, rabbit, squirrel, quail, dove, and others.)

Contemporary oak systems reflect the impacts of a century or more of post-European settlement disturbances. Restoration will in many cases be a slow process. While it is natural to want quick results, restoration and rehabilitative management must be undertaken with the knowledge that centuries of abuse can not be reversed overnight.

In order to assess the success of restoration and to obtain data to drive future management activity, some type of ecological monitoring is requisite. Monitoring protocols should be repeatable, efficient, and provide information of direct relevance in formulating management decisions. Use of fixed transects with nested plots to estimate vegetative cover is the most widely employed approach. To convey assurances that monitoring results actually reflect site improvement or degradation, ecological monitoring protocols should reflect inputs from as large a spectrum of the biota as possible, in some type of unified array. Abiotic criteria, such as water quality and quantity, are often a useful adjunct to organismally-based monitoring systems. Sophisticated monitoring may not be possible on all sites. It should occur on several sites throughout the savanna to represent savanna types. Simple monitoring techniques, such as taking photos annually at permanent photo points, can be low-tech, low-cost methods of achieving crude levels of monitoring.

Monitoring protocols that focus on species conservatism satisfy many of the requirements demanded of a pragmatic, management-based ecological monitoring program. One such system, devised by Gerould Wilhelm, is summarized in Swink and Wilhelm (1979).

Baseline assessments are an important tool for ecological monitoring, and serve as a reference point against which to assess the effectiveness of management treatments. Baseline data should be taken wherever possible, but restoration of degraded or critically threatened sites should not be held hostage to obtaining baseline data.



Presently, numerous public and private agencies, organizations and institutions are conducting research and prescribing and implementing management to restore savanna and woodland natural communities. Often managers and researchers are not aware of databases of ongoing or recently completed research. One important challenge is to bring the vast amount of information contained in these databases together in a format that is accessible.

The following information is needed by managers:

Information sources and database systems of use include:

Many diverse delivery systems can be used for distributing information:

Systems need to be developed or improved for relaying information among decison-makers. Systems include:


Efforts toward preservation of biodiversity can be strengthened through public information/education on what savanna communities are, how rare they are, why they are important, and why they must be managed in order to be preserved. The development of information/outreach programs should also benefit efforts to gain support for preservation and restoration of savannas on private lands. Additional emphasis may be placed on acquisition of critical areas and/or protecting savanna communities to prevent further degradation.

Techniques and ideas for improving education include the following:

There is a definite need for education of public and private sectors in savanna restoration. The general public will need to be supportive of restoration of savannas for their ultimate success. Also, restoration managers need to be trained in proper management. A continuing education outreach to these managers will be needed to bridge the gap to science and technology.


Goal 1. Obtain a broad consensus for implementation of the Ecosystems Recovery Plan.

Goal 2. Identify, evaluate, and integrate acquired oak savanna/woodland acreages.

Goal 3. Coordinate and integrate agencies, states, and people into functionally specific groups.

Goal 4. Develop and assess restoration and protection technology.

Goal 5. Collect and disperse management strategies and other information.

Goal 6. Educate and involve a diverse constituency.

Goal 7. Obtain and coordinate additional land and funding sources.


The Research Working Group of the Midwest Environmental Roundtable's Interagency Cooperation on Ecosystem Management (ICEM) met on February 24, 1994 to consider recommended research questions that served efforts to restore oak savannas and woodlands. Working Group members were representatives from participating state and federal agencies. Individuals were selected, in part, because of their experience with savanna and woodland management.

These individuals were:

The Working Group prioritized a list of research questions for restoration of Oak Savanna and Woodlands (Appendix E). Research questions were submitted from three different sources: a 'pre-conference draft', which was created for the attendees of the conference working sessions, the results of the conference working sessions and, finally, suggestions submitted by Recovery Plan Section Editors.

Most research questions were left unedited for the purposes of this exercise. Modification of some questions was made for 1) clarification, 2) minimizing redundancy, 3) separation of compound/multiple research inquiries into separate statements. Some questions were deleted because the Group felt a valid scientific hypothesis could not easily be derived from the original submitted statement. A limited number of additional questions were added by the Group after noting gaps in research left by submitted questions. The Group did not substantially change the majority of originally submitted questions.

Submitted research questions were first categorized along a two dimensional matrix. The horizontal dimension consisted of descending hierarchical ecological compartments: Landscape, Ecosystem, Communities, Populations, Species, Genes. The vertical dimension consisted of questions which seem to naturally fall into the following categories: Ecological Structure, Ecological Function, Restoration, Management, Classification, Abiotic, Fire. Cross tabulations down rows and columns thus allowed for summary of suggested research topics.

The Working Groups found that absolute categorization of Research Questions was often difficult and sometimes arbitrary. Often questions fit into several categories cutting across functional, disciplinary and ecological units. In other cases, the categorization of the question was dependent on the Groups immediate interpretation of what the author's meaning in framing the original question. Particularly difficult was separation of population, species and genetics type questions; between ecosystems and communities units, and between restoration and management categories.

The nature of submitted research questions by ecological unit can likely be explained to some degree by the academic backgrounds of the scientist presenting the inquires. From this premise, it follows that pre- and post-conference participants tended to orient more towards organismal biology, with relatively fewer geneticists, and ecosystem and landscape ecologists represented. For convenience, we are using the term 'organismal' to represent the community, population and species array in our ecological unit matrix. The Working Group commented that similar questions might have been framed in different disciplinary terms or approaches if participants represented wider biological backgrounds. Thus, the paucity of landscape, ecosystem and genetics type questions did not reflect a lack of importance of these subjects relative to restoration of savannas and woodlands. To the contrary, the lack of questions in these areas highlights the greater need for more research along these disciplinary lines and the need to attack more researchers with related training and interest. This is especially true of genetics and systems (ecosystem and landscape ecology) type research. If this survey reflects the true balance of current research in the area of savanna restoration (we suspect that it does), the Working Group recommends increasing emphasis in these disciplinary approaches. We can not adequately address the long-term restoration of savannas only in the context of organismal biology, while ignoring the physico-chemical, environmental, genetic, and landscape nature, influence and interactions on these complex ecosystems.

The submitted questions emphasize the importance of our understanding of organisms occurring in woodlands and oak savannas, especially of the demography, species composition, distribution, interaction, co-action, dispersal characteristics, interdependence. Research questions commonly emphasized spatial-temporal distributional questions at a variety of scales: microhabitat-ecosystem-landscape-biomes; seasonal-successional-geological. It is imperative that we understand the immediate and long-term implications of management actions. This understanding should go beyond the targeted individual organisms or communities and extend to contiguous or nested communities (e.g. ponds, streams, soil arthropods, migrating species). While an autoecological understanding or targeted species is pre-requisite to the organism's recovery, managing that species at the exclusion of the residual ecosystem/landscape is myopic and ill advised. Spatial limitations and possibilities must be clearly understood both within and between patches, at a fine and coarse grain level. This understanding needs to be extended beyond the biological and include biogeochemical fluxes, fire, edaphic, atmospheric, hydrologic influences.

Structural, functional and management oriented topics were most common, when categorized along topical line as outlined in the vertical axis of the above cross-tabulation. In order of frequency, restoration, fire, abiotic/edaphic, and classification type questions followed in the number of research questions submitted. The high frequency of structural and functional question is not surprising since most research questions can easily be categorize as one or a combination of these two biological approaches. It might be argued that this was an important artifact and elimination of these categories might have further clarified the relative emphasis in residual categories. Likewise, we could argue that none of the categories were exclusive and many problem statements would have overlapping elements. For instance, the subject ' restoration of savanna/woodlands by use of fire' would simultaneously have structural, functional, restoration, management, abiotic/edaphic and, of course, fire components. The subject would also cut across all ecological units considered in our matrix, depending on the framing of the research problem statement or hypothesis.

The Work Group noted the few questions relating to classification issues and abiotic factors were present. The latter omission is not so much related to its lack of importance, but rather the lack of chemists, physical scientists, geologists, soil scientists present at the referenced meetings versus the stronger representation by organismal biologists. Again, the omission of abiotic components of the subject ecosystems highlights the great need for geophysico-chemical researches and the need to attract trained scientist to these inquiries relating to these units of the ecosystem. Classification questions were also relatively less frequently submitted. On the contrary, good science follows from systematic understanding of the role, function and relative position of the subject matter in relation to other natural features and interactive elements. More likely the paucity of questions may be due to the fact that: classification questions can be reduced to a few well balanced questions, classification questions are often not perceived by the non-scientist as stereotypic 'research questions', even scientist do not always see the management/restoration implication of community taxonomic questions. The true explanation for the lack of questions is secondary to the need to have classification questions resolved. These questions are central to the target restoration goal and the management process in achieving these goals. In other words, we can not hope to achieve the goal, if we do not clearly understand and communicate the restoration endpoint and how to get there.

Savanna Ecosystem Research

Poorly represented in the research questions compiled for the Savanna Ecosystem Recovery Plan are research investigations focusing on ecosystem ecology. For example, little is known about how fire influences biogeochemical cycling in savannas and how fire and nutrients interact to influence community successional rates. This type of research would be relevant to the question of whether savanna sustain long term harvest of oak biomass and what influence does this have on savanna dynamics. Understanding the pathways, pools and fluxes of nutrients such as nitrogen and phosphorous and other chemicals would be important in addressing the impacts of nutrient enrichment and toxic chemicals from air pollution on savanna dynamics. More research is needed in the pattern and process of savanna populations and communities and landscape dynamics.


No projects incorporate modeling in the recovery of savanna ecosystems. For example, successional models could be developed incorporating fire effects on vegetation and animals to address concerns of species viability and landscape dynamics. In this regard landscape models could be developed to investigate the role of nature preserve design on ecosystem dynamics and immigration and emigration on rare species.


Few genetics projects were identified in the draft ecosystem recovery plan; however, there are several avenues of genetics research that would be important in restoring savanna ecosystems. Little is known about the impacts of past management and the increasingly fragmented landscapes on the genetics and viability of savanna organisms. Are primarily outcrossing or outbreeding populations of plants and animals showing effects of inbreeding in remnant populations? If so how could these negative effects be ameliorated? In rehabilitating damaged savanna ecosystems, how much attention must be paid to genetic diversity in developing stock for transplanting or translocating? Rare species and largely outcrossing species would be likely to be the primary focus of research. If many of the species are peripheral in their range in the Midwest, how important are these populations in preserving species wide genetic diversity? What is the likelihood of outbreeding depression when reintroducing mixed stock into savanna rehabilitation? The model tool box of the population geneticist should play an integral role in guiding savanna ecosystem restoration through active research.

Non-indigenous and pathogenic organisms

Exotic plants, animals and pathogens are playing a significant role in altering the biodiversity of North America. Once exotics become naturalized they profoundly affect the landscape as witnessed by the gypsy moth and chestnut blights. Exotics may disrupt ecosystems functioning. For instance, grass carp and zebra mussels have fundamental change system dynamics of respectively impacted aquatic habitats. Non-native species introductions have changed community structure and species compositions as well. Introduction of fire and leafy spurge have significantly shifted species composition, feral animals such as goats, pigs have altered community structure. Often native species are outcompeted or crowded out by newly introduced species (e.g. purple loosestrife, starlings), sometimes native gene pools are depleted by hybridization with exotic species (Viburnum opulus vs. V. trilobum, Celastrus orbculatus vs. C. scandens).

Moreover, once an exotic becomes naturalized and spreads throughout a system, its removal is currently very difficult. Oak dominated savannas, prairie groves, and woodlands require periodic disturbance to perpetuate an oak canopy. Because these communities are disturbance dependent they may be more susceptible to invasion by exotic species. For example, exotic species of shrubs and vines are now being identified as serious competitors to oak re-vegetation in the Appalachians.


High Priority Questions

Ecosystem Dynamics -- Organizing Forces

Landscape Ecology -- Spatial pattern

Restoration and Management

Composite of side-effects question and #36.



There are many government agencies, non-governmental organizations, businesses and individuals working to preserve, protect, and restore oak savanna and woodlands in the Midwest. It is important therefore to identify and integrate the current roles and priorities of the various agencies, organizations, and other parties, to encourage the most efficient use of scarce resources.


The following organizational roles have been identified as necessary for the protection and restoration of the Midwest oak savanna and woodland ecosystems.


The descriptions of oak savanna-related work underway or proposed by federal and state agencies were submitted by staff at those agencies. This should not be taken as a comprehensive, exhaustive list of agency action on behalf of oak savanna restoration; undoubtedly, other efforts underway or in the planning stages did not make it into this draft of the Recovery Plan. Nor should discussion of future directions contemplated by federal and state agencies be read as official proposals endorsed by the heads of those agencies; inspired in part by the 1993 Oak Savanna conference and the Ecosystems Recovery Plan, many agencies are in the process of evaluating their present and future directions in oak savanna management and restoration.

Federal Agencies

 The U.S. Department of the Army

USDA Forest Service
Forest Service staffers have been active participants in the efforts to develop the Ecosystems Recovery Plan. The Forest Service has entered into a Memorandum of Understanding with the Forest Preserve District of Cook County Illinois, the state of Illinois, and The Nature Conservancy to fund and support work including demonstration projects at important oak savanna sites in the Cook County Forest Preserves. The Forest Service conducts management and research projects at oak savanna and woodland sites in national forests across the region.

The North Central Forest Experiment Station unit at Columbia, Missouri, is studying the regeneration and management of upland central hardwood forests. While most of the research concerns forests other than oak savanna, some related activities do occur. A paper prepared by Paul Johnson and Jay Law for the Oak savanna Conference provides guidelines for managers based on the stocking chart concept. This information has now been developed into a manuscript for the Northern Journal of Applied Forestry.

The Columbia unit has funded a small cooperative study at the University of Missouri on inventory and classification of savannas. The unit has also supported a small study related to Bachman's sparrow, principally a glade species which is apparently considered endangered by the state of Missouri. The Station has developed a proposal to initiate new research related to upland forest glades, barrens, and savannas. However, initiation of this research is contingent on obtaining new funding.

U. S. Fish and Wildlife Service
Fish and Wildlife staffers have been active supporters of and participants in the regional effort which has led to the development of the Oak Ecosystems Recovery Plan.

Six national wildlife refuges in the Region are known to have remnant oak savannah habitat, with restoration efforts either underway or in the planning stages at four of them. The biggest hurdle that we have is one of insufficient resources (dollars and staff) to meet our needs.

U.S. Environmental Protection Agency
The USEPA was a co-sponsor of the Midwest Oak Savanna Conference in Chicago in February, and has been a key partner in the followup efforts including further development of the Recovery Plan. The USEPA is also currently developing a plan to use satellite imaging to map oak savanna and woodland communities across the region.

The USEPA is co-sponsoring the next two planned regional followup conferences on oak savanna restoration and management: with the Illinois Department of Conservation scheduled for October 1994, and with the Missouri Department of Natural Resources tentatively scheduled for fall 1995.



County, municipal, and regional agencies around the Midwest are stewards of prime restorable savanna sites. Though constrained by limited resources, local governments can play a crucial role in oak savanna and woodlands restoration, often with the support and/or active participation of state or federal agencies, and/or conservation organizations.

A comprehensive list of savanna restoration and management projects led by local governments is beyond the scope of this document. The following examples illustrate the potential and the limitations of action by local governments.


Many high-quality oak savanna remnants are privately held (do we have a figure?). Oak savanna and woodland restoration is a rapidly evolving science, changing as new research and field results are made public. Restoration techniques, especially controlled burning, can be difficult to properly target and safely manage. Landowners should seek expert assistance before attempting landscape-scale restoration.

A number of resources are available to assist landowners in evaluating their land for possible restoration, and for managing or protecting high-quality remnants.


Expertise and other characteristics:

Current role:

Actions taken to protect the oak savanna ecosystem:

Possible future actions:


Expertise and other characteristics:

Current Role:

Actions taken to protect the oak savanna ecosystem:

Possible future actions:

Top of page


The roles of the government agencies, conservation organizations, businesses and private individuals have been traditionally defined through the existing regulatory functions and missions currently in place. One of our challenges will be to determine how the functions of the various agencies and groups can be more clearly defined to more efficiently address savanna protection and management. Another challenge will be to articulate how agencies and organizations might expand their roles. To maximize efficiency, agencies and groups should continue to explore ways to work together towards a common goal of providing for adequate protection for a viable oak savanna ecosystem.

Top of page


Abrams, M. 1986. Historical development of gallery forests in northeast Kansas. Vegetatio 65: 29-37.

Anderson, D.M. 1982. Plant communities of Ohio: a preliminary classification and draft. Division of Natural Areas and Preserves. Ohio Department of Natural Resources. Columbus Ohio. Unpublished draft.

Anderson, R.C. 1983. The eastern prairie-forest transition-an overview. Pp. 86-92 in R. Brewer (editor). Proceedings of the Eighth North American Prairie Conference, 1982. Kalamazoo, MI.

Anderson, R.C. and M.R. Anderson. 1975. The presettlement vegetation of Williamson County, Illinois. Castanea 40: 345-363.

Apfelbaum, S. I., and A. Haney. 1991. Management of degraded oak savanna remnants in the Upper Midwest: Preliminary results from three years of study. Pg. 81-89, In: John Ebinger (ed.) Proc. of the Oak Woods Management Workshop, Peoria, IL.

Bray, J.R. and J.T. Curtis. 1957. An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27:325-349.

Curtis, J. T. 1952. The Vegetation of Wisconsin. The University of Wisconsin Press, Madison. 687 pp.

Curtis, J.T. 1959. The vegetation of Wisconsin. University of Wisconsin Press. Madison, WI. 657 p.

Driscoll, R.S., D.L. Merkel, D.L. Radloff, D.E. Snyder and J.S. Hagihara. 1984. An ecological land classification framework for the United States. USDA Forest Service Misc. Publ. No. 1439. Washington, D.C. 56p.

Dyksterhuis, E.J. 1957. The savannah concept and its use. Ecology 38: 435-442.

Eiten, G. 1986. The use of the term "savanna." Tropical Ecology 27: 10-23.

Eiten, G. 1992. How names are used for vegetation. Journal of Vegetation Science 3:419-424

Finley, R.W. 1976. Original Vegetation Cover of Wisconsin: compiled from U.S. General Land Office Notes. North Central Forest Experiment Station, U.S.D.A. Forest Service, St. Paul, MN.

Gleason, H. A. and A. Cronquist. 1991. Manual of vascular plants of the Northeastern United States and adjacent Canada. Bronx, New York: The New York Botanical Garden. 910pp.

Graumlich, L.J. and M.B. Davis. 1993. Holocene variation in spatial scales of vegetation pattern in the Upper Great Lakes. Ecology 74: 826-839.

Grimm, E.C. 1983. Chronology and dynamics of vegetation change in the prairie-woodland region of southern Minnesota, U.S.A. New Phytologist 93:311-350.

Haney, A., and S. I. Apfelbaum. In press. Characterization of Midwestern oak savannas. In Proc. of Midwest Oak Savanna Conference, Feb. 20, 1993. Northeastern Illinois University.

Haney, A., S. I. Apfelbaum, and L. Ebbecke. In press. Oak barrens in the upper Midwest. Castanea.

Herkert, J. R., ed. Endangered and Threatened Species of Illinois: Status and Distribution, Volume 1 -Plants. Illinois Endangered Species Protection Board: Springfield, Illinois. 158pp.

Higgins, K. F., A. D. Kruse, and J. L. Piehl. 1989. Prescribed burning guidelines in the northern great plains. U. S. Fish and Wildlife Service EC 760.

Hutchison, M.D., S. Olson, and T. Vogt. 1986. A survey of the barrens region in Massac and Pope Counties, Illinois. Unpublished report to the Natural Land Institute, Belknap, Illinois. 63 p.

Indiana Department of Natural Resources. 1990. Indiana's Rare Plants and Animals: A Checklist of Endangered and Threatened Species. 36pp.

Jacobson, Jr., G.L. and E.C. Grimm. 1986. A numerical analysis of holocene forest and prairie vegetation in central Minnesota. Ecology 67: 958-966.

Ladd, D. 1991. Reexamination of the role of fire in Missouri oak woodlands. Pp-67-80 In Burger, G.V., J.E. Ebinger, and G.S. Wilhelm (editors). Proceedings of the Oak Woods Management Workshop. Eastern Illinois University, Charleston, IL.

Leitner, L.A. C.P. Dunn, G.R. Guntespergen, F. Stearns, and D.M. Sharpe. 1991. Effects of site, landscape features, and fire regime on vegetation pattern in presettlement southern Wisconsin. Landscape Ecology 5:203-217.

McClain, W.E., M.A. Jenkins, S.E. Jenkins, and J.E. Ebinger. 1993. Changes in the Woody Vegetation of a Bur Oak Savanna Remnant in Central Illinois. Natural Areas Journal 13: 108-123.

Marschner, F.J. 1974. The Original Vegetation of Minnesota. North Central Forest Experiment Station, U.S.D.A. Forest Service, St. Paul, MN.

McCance, R. M. and J. F. Burns. 1984. Ohio Endangered and Threatened Vascular Plants: Abstracts of State-Listed Taxa. Division of Natural Areas and Preserves, Department of Natural Resources: Columbus, Ohio. 635pp.

Minnesota Department of Natural Resources Natural Heritage Program. 1993. Minnesota's Native Vegetation: A Key to Natural Communities.

Missouri Department of Conservation. 1985. Checklist of Rare and Endangered Species of Missouri. 17pp.

Nekola, J. C. 1990. Rare Iowa plants notes from the R. V. Drexler Herbarium. Journal of Iowa Academy of Science 97:55-73.

Nelson, P.W. 1985. The terrestrial natural communities of Missouri. Missouri Natural Areas Committee. Jefferson City, Missouri. 197 p.

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

Packard, S. 1988. Just a few oddball species: restoration and rediscovery of the tallgrass savanna. Restoration Management Notes 6:13-22.

Packard, S. 1993. Restoring Oak Ecosystems. Restoration and Management Notes 11: 5-16.

Penfound, W.T. 1962. The savanna concept in Oklahoma. Ecology 43: 774-775.

Penskar, M. 1993. Preliminary list of Michigan savanna rare plants.

Rodgers, C.S. and R.C. Anderson. 1979. Presettlement vegetation of two prairie peninsula counties. Botanical Gazette 140:232-240.

Roosa, D. M. and L. J. Eilers. 1979. Endangered and threatened Iowa vascular plants. Special Report No. 5, State Preserves Advisory Board, State Conservation Commision: Des Moines, IA. 93pp.

Schroeder, M. J., and C. C. Buck. 1970. Fire weather. USDA Forest Service Agriculture Handbook 360.

Seastedt, T.R. and A.K. Knapp. 1993. Consequences of nonequilibrium resource availability across multiple time scales: the transient maxima hypothesis. The American Naturalist 141:621-633.

Smith, W. 1988. Vascular Plants. In: Coffin, B. and L. Prannmuller, eds. Minnesota's Endangered Flora and Fauna. University of Minnesota Press: Minneapolis, MN. pp. 33-217.

Steyermark, J. A. 1963. Flora of Missouri. Iowa State University Press: Ames, IA. 1725pp.

Stritch, L. 1990. Landscape-scale restoration of barrens-woodland within the oak-hickory forest mosaic. Restoration and Management Notes 8: 73-77.

UNESCO. 1973. International classification and mapping of vegetation, Series 6. Ecology and conservation. United Nations Education, Scientific and Cultural Organization. Paris, 93 p.

Vestal, A. 1936. Barrens vegetation in Illinois. Transactions of the Illinois State Academy of Science 11: 122-126.

Wade, D. D., and J. D. Lunsford. 1988. A guide for prescribed fire in southern forests. USDA Forest Service, Southern Region Technical Publication R8-TP 11.

Webb, III, T., E.J. Cushing, and H.E. Wright, Jr. Holocene Changes in the Vegetation of the Midwest. Pp. 142-165 in H.E. Wright, Jr. (editor). Late Quaternary Environments in the United States. Volume 2, The Holocene. University of Minnesota Press, Minneapolis, MN.

White. F. 1983. The Vegetation of Africa. A Descriptive Memoir to Accompany the UNESCO/AETFAT Vegetation Map of Africa. UNESCO, Paris.

Wisconsin Natural Heritage Program. 1990. Wisconsin Rare Vascular Plant Working List. Department of Natural Resources: Madison, WI. 6pp.

Wright, H. A. and A. W. Bailey. 1982. Fire Ecology. John Wiley & Sons, N. Y.

Top of page

Local Navigation

Jump to main content.