1993 Proceedings of the Midwest Oak Savanna Conferences
CHARACTERIZATION OF MIDWESTERN OAK SAVANNAS
Steven I. Apfelbaum
Oak savannas occur along the prairie border from south-central Texas to Manitoba (Fig. 1). In the upper Midwest, oak dominated savannas develop on both xeric and mesic sites; six distinct community types can be identified, all fire dependent. These communities are distributed along geographic and physiographic gradients, and each also varies along a successional fire gradient. This paper provides an interpretation of these community types and their successional derivatives which develop in the absence of fire.
In previous papers, we discussed Midwest oak savanna structure and dynamics (Haney and Apfelbaum 1990, Apfelbaum and Haney 1991). Nuzzo (1986) pointed out that high quality examples of this once prominent biome were extremely rare. Noss and Scott (in press) listed the oak savanna and prairie complex in the Midwest as critically imperiled. Specifically, nearly all savannas on mesic sites have been destroyed by land-use changes or so altered by successional change and invasion of exotic species, that most are no longer recognizable as savannas (Haney and Apfelbaum 1990). Trees and shrubs also have increased several fold in savannas on dry and mesic sites where fires have long been excluded, but dry savannas are more easily restored by fire (White 1986) and have fewer exotic invasives (Apfelbaum and Haney 1991).
Lack of understanding of the fire successional gradient has lead to much confusion regarding definition of savannas (Henderson and Long 1984, EPA 1993). Community classification systems have largely ignored the fire gradient in savannas, suggesting that oak communities are either barrens, savannas, woodlands, or forests. In the upper Midwest, these are variations of similar communities along the fire gradient. For example, barrens, or brush prairies, were once wide-spread in the Midwest. Most are now oak or oak-pine woodlands or forests, as a result of fire exclusion. Curtis (1959) labeled barrens with jack pine (Pin us banksiana) pine barrens, although virtually all now have varying amounts of oak and, in the continued absence of fire, most develop into oak-dominated forests. Consequently, most barrens in Wisconsin are now closed oak forests with varying amounts of jack pine (Haney and Apfelbaum, in press). Barrens had as few as four trees per acre when the region was first surveyed in the mid-i 9th Century, and were forested 100 years later (Brown and Curtis 1952).
Remnants of over 100 oak savannas were located from northern Indiana through eastern Iowa and the upper peninsula of Michigan. Perrnanent transects were placed systematically in communities representing the full range of oak savanna types and conditions found in the region. In most cases, initial data were collected prior to any restoration effort. Many of the savannas had not been burned for 70 or more years. A few, in contrast, had been burned shortly before sampling, and some had combinations of grazing and fire disturbance. Restoration efforts, including a combination of brush removal and fire, have been initiated in some savannas. Where possible, transects were placed in restored areas as well as control areas to determine community responses to fire or other restoration treatments. Many of the oak savannas were resurveyed following restoration treatments. Where transects were resurveyed, data were treated as a separate community from the initial data. This information gave us an opportunity to determine the response of different community types along the disturbance gradient, providing further insight into the dynamics of oak savannas.
Transects were laid out in 50 m. units along compass lines with each end point marked with a metal pipe. In larger communities, as many as 25 or more 50 m. transects were established in one to as many as 10 sampling lines. Tree cover was estimated for each 50 m. transect by line intercept. Trees were defined as stems -> 5cm. (2 inches) dbh. Intercept of each species was recorded separately and summed to calculate the percent coverfor each sample unit. Small tree and shrub species cover was estimated along the same 50 m. intercept. Small trees and shrubs were defined as living stems -< 5 cm. dbh, but more than 1 m. tall. Cover of herbaceous and woody species less than 1 m. tall, were estimated in 1 m2 circular quadrats centered at each 10 m. point beginning at the zero point of each 50 m. line. In many savannas, herb layer oak regeneration in each circular 1 m2 was plotted. Young oaks were defined according to whether they were seedlings (plants originating directly from seed without resprouting), or grubs. Grubs were identified as young grubs, having a root crown diameter < 2 cm., or old grubs, those having a root crown -> 2 cm. in diameter.
All trees within 1 m. of each 50 m. line were tallied by diameter and noted as to whether they were alive or dead. Ages of selected trees were determined by increment cores. In selecting trees to be cored, we attempted to sample the full range of diameters and species. Shrub and small tree stems 1 m. to the right side of each 50 m. line were tallied; alive and dead stems were tallied separately. Basal area of alive and dead woody stems was determined at each 25 and 50 m. point along each transect using a 10 factor prism. A thorough search was made of each community for species not sampled in transects. These were recorded as present. Species cover data were summarized by mean and standard deviation, and ordinated by detrended correspondence analysis (DCA) and cluster analysis (Hill and Gauch, 1980; Gauch and Whittaker, 1981) to determine overall community similarities and relationships along environmental and disturbance gradients.
Oak savannas of the Midwest fall into two types, those on dry sites, and those on mesic sites (Fig. 2). The former are characterized by black oak (Quercus velutina), Pennsylvania sedge (Carex pensylvanica), Hills oak (Q. el/ipsoidalis), hazilnut (Corylus americana or C. cornuta), post oak (Q. stellata), or blackjack oak (Q. marilandica). Mesic savannas are chaacterized by bur oak (Q. macrocarpa), white oak (Q. a/ba), northern red oak (Q. rubra), and swamp white oak (Q. bicolor). Degraded mesic savannas often have dense understories of prickly ash (Xanthoxy/um americana) or buckthom (Rhamnus fran gula or R. cathartica). Both types of savannas generally have a great diversity of herbaceous species with Pennsylvania sedge the most ubiquitous and a dominant, especially on most dry sites. The major oak savanna associations in the Midwest are summarized in Table 1. This classification is based on ordination of 150 communities with 830 species from 100 sites. Data provide a basis for interpreting savanna communities along successional gradients as well as along geographic and physiographic gradients.
Eastern Sand Savanna
Historically, intense fire in the eastern sand savanna produced a barrens community with most trees reduced to grubs in a prairie-like matrix of grass and forbs (Table 2). In experiments, the Indiana Department of Natural Resources used a single hot fire to convert some blocks of dry oak forest at Jasper-Pulaski Fish and Wildlife Area to barrens communities. In contrast, moderate to light fires reduced tree and shrub cover only 10 to 30 percent. In the eastern sand savannas, tree cover varies from virtually zero, after intense fire, to nearly 100 percent in the absence of fire for 50 or more yeas. In the long-term absence of fire, black cherry (Prunus serotina) increases along with back oak. Under the heavier tree canopy, shrubs also increase and litter accumulates, resulting in further loss of forbs and grasses. Light or intermediate fire, however, removes the smaller trees and shrubs, and opens the canopy with a corresponding increases in forb and grass cover, and in species. Often as much as 30% increase in plant species richness develops the first year following moderate to hot fire. Nearly all shrubs and small diameter trees are top-killed. Species that vigorously resprout, such sassafras (Sassafras aibidum) or blackberry (Rub us sp.), sometimes increase after one fire, but are reduced by recurring fire. Because black oak is moderately tolerant to fire, higher fire frequency and intensity favors black oak over other tree species in the eastern sand savanna.
Aronia spp. often forms a dense
shrub layer, especially on the mesic end of the hydrologic gradient.
Northern Sand Savanna
The northern sand savanna historically was a barrens when subjected to intense or frequent fire (Curtis 1959, Haney and Apfelbaum, in press). In many areas, from northern Illinois throughout the range of the northern sand savanna, hazel or blueberry formed shrub-dominated communities with an herbaceous component similar to the oak and pine dominated communities. These shrub dominated communities were fire maintained (Bowles, in press). Oaks in these barrens were reduced by fire to shrub-layer grubs. Fire suppression in northern barrens during the past 70 years has resulted in their developing into closed forests with a canopy of Hill's oak (black and Hill's oak in the transition area in the Central Sands of Wisconsin) and jack pine, with a scattered understory of serviceberry and hazelnut. In openings, black cherry and aspen also occur, particularly on more mesic sites. These closed forests have lost most of the associated forb and grass cover that were typical of the open barrens. Tree productivity is limited, probably by nitrogen (Tilman 1984). As a result of the nutrient limitation in the coarse, acid soils, accumulation of litter and standing crop biomass quickly curtails productivity, often resulting in high populations of small, stunted trees. Fire releases nutrients and stimulates productivity (VogI 1965, 1964), and leads to a remarkable increase in herbaceous species. At Moquah Barrens, in Chequamegon National Forest, clearing followed by prescribed fire has resulted in good responses of the rare temate grape fern (Botrychium ternatum), and dwarf bilberry (Vaccinium caespitosum), host plant for the rare Naboror's blue butterfly (Lycacides idas nabokovi) (Vora 1993).
Southern Oak Savanna
On deeper soils, white oak occurs while chinkapin oak (Q. muehlenbergii) occurs on alkaline sites. In the absence of fire, these savannas are invaded by winged elm (Uhnus alata), Eastern red cedar (Juniperus virginiana), black (Acer nigrum) and sugar maple (A. saccharum), hickory (Caiya spp.), and other species, with the corresponding demise of the herbaceous layer. In addition to thin soils over parent material on steep slopes, this savanna formation is associated with older leached clay-loam soils, usually with pans or impermeable layers. Some of the southern oak savanna occurs on sandy soils where post and blackjack oak are scattered in sand prairies. Overall, the southern savannas are confined to sites with limited water-holding ability.
Southern oak savannas support many species that do not extend into the eastern or northern savannas (Table 2). Post and blackjack oak are virtually unknown in eastern sand savannas whereas they often are dominant in the southern savanna. Many southern oak savannas look much like the eastern and northern sand sayannas, and also were sometimes converted by fire to barrens. Fire is essential to maintain southern oak savannas against the invasion of juniper, maple, and other species. Grazing presumably was important as well. Southern oak savannas frequently are interspersed with prairie openings, particularly on south facing slopes where exposure, fire frequency and intensity was somewhat greater.
Mesic Loam Savanna
With fire exclusion, most mesic loam savannas soon develop nearly 100 percent tree cover. Older oaks, particularly white and red oaks, and on occasion, bur oak in overgrown savannas, have wide crowns that indicate the once open to semi-open nature of these savannas. Maples, hickory, ironwood, boxelder (Aces negundo), green ash (Fraxinus pensylvanica) red elm, honeysuckle, and buckthorn can form such dense shade that fire may be difficult to reintroduce, especially in the spring. Fall fires can be used in many cases to begin to open the understory, but in extreme cases, woody plants have to be cut to provide sufficient fuel and drying to support fire. In time, spring fires can be used to continue to restore the rich diversity of the herb layer.
Oak savannas of the upper Midwest can be broadly divided into two types, those associated with dry sites and those of mesic sites. The former were more subject to frequent intense fire and often were reduced to barrens. Barrens, primarily of fire-stunted Hill's oak and young jack pine, were once common in the upper Midwest, although shrub and grass dominated communities further south also were called barrens. In the upper Midwest, barrens represent one extreme of the fire gradient in the oak savanna complex. The three types of dry site oak savannas in the Midwest are separated primarily along a geographic gradient with corresponding differences in species. The southern oak savanna is characterized by blackjack and post oak. This savanna occurs on shallow soils, or soils with impermeable layers, or sands. The eastern sand savanna is characterized by black oak. The northern sand savannas are characterized by jack pine and Hill's oak. Jack pine is frequently more common than oak, although oak grubs, in the absence of fire, result in oak forests with varying amounts of jack pine.
The three types of mesic site savannas occur primarily on different soils and, therefore, are distributed along edaphic/hydrologic gradients. They have an especially high vascular plant richness. The clay-loam savanna, also referred to as the tall-grass savanna (Packard 1983), occurs on moderate to poorly drained, rich black soils characterized primarily by bur oak and a high diversity of herbaceous plants. Flood plain savannas occur on floodplains or in swales and are characterized by bur oak or swamp white oak. Mesic loam savannas vary from mesic to moderately dry sites, but occur consistently in silt loam soils, often of loess origin. The mesic loam savanna is characterized by white oak, northern red oak, bur oak, black oak, Hill's oak, and occasionally chinkapin and swamp white oak.
Mesic savannas are especially vulnerable to rapid invasion by shade tolerant species in the absence of fire. European buckthorn and garlic mustard have become particular problems in many unmanaged savannas. Native tree and shrub species such as sugar and black maple, red elm, black cherry, ironwood, basswood, boxelder, green ash, and prickly ash, also are prone to invade or increase in mesic savannas in the absence of fire. As a consequence, mesic savannas are typically more degraded and require more effort to restore than dry site savannas which usually can be restored with fire, often with a few successive burns. Some of the most degraded mesic savannas have lost most native species, their soil seed-banks, and even topsoil, and require long-term restoration efforts to recover. Those with altered hydrology or that experienced severe erosion are the most difficult to restore.
We are indebted to dozens of colleagues who helped collect field data. The Indiana Department of Natural Resources, Division of Nature Preserves provided financial assistance for field work in Indiana. Nongame wildlife funds were provided by the Illinois Department of Conservation. The Cook and Dupage County Forest Preserve Districts, U.S. Forest Service, U.S. Army, Wisconsin and Illinois Chapters of The Nature Conservancy, and numerous private owners provided support and assistance to this project. This work was undertaken with encouragement and direct assistance from Marlin Bowles, Tom Post, and John Bacone. Lauren Ebbecke and Marie Brown did much of the data processing, and also assisted in the field. The computer software used for detrended correspondence analysis was adapted for our data by Noel Pavlovic of the U.S. Park Service.
Apfelbaum, S.L, and A. Haney. 1991. Management of degraded oak savanna remnants in the Upper Midwest: Preliminary results from three years of study. Pp.81-89, in John Ebinger, ed., Proceedings of the Oak Woods Management Workshop, Peoria, IL.
Apfelbaum, S.L and A. Haney. 1987. Unpublished report to Illinois Non-game Fund.
Anderson, R.C. and J. E. Schwegman. 1990. Twenty years of vegetation change on a southern Illinois barren. Natural Area Journal 11(2): 100-107.
Anon. 1983. Think sheet on savannas in Indiana. Indiana Natural Heritage Program.
Bacone, J.A., et al. 1979. Presettlement vegetation of the Indiana Dunes National Lakeshore. Unpub. rept. to Science Division, Indiana Dunes National Lakeshore.
Bowles, M. In press. Toward development of a landscape continuum model for savanna: correlations between historic and modern descriptions, species lists, and soil survey. Proceedings of Oak Savanna Conference, March 3, 1993, Northeastern Illinois University.
Bray, J.R. 1960. The composition of savanna vegetation in Wisconsin. Ecology 41: 721-732.
Brown, R.T., and J. T. Curtis. 1952. The upland conifer-hardwood forest in northern Wisconsin. Ecological Monographs 22: 217-234.
Curtis, J.T. 1959. The Vegetation of Wisconsin. The University of Wisconsin Press, Madison. 657 pp.
Finley, D. and J. Potzger. 1952. Characteristics of the original vegetation in some prairie counties in Indiana. Butler Univ. Botanical Studies 10:114-118.
Gauch, H.G., Jr. and R. H. Whittaker. 1981. Hierarchical classification of community data. Journal of Ecology 60:537-557.
Grimm, E.C. 1984. Fire and other factors controlling the Big Woods vegetation of Minnesota in the mid-l9th Century. Ecological Monographs 54(3): 291-311.
Haney, A., and S.I. Apfelbaum. 1990. Structure and dynamics of Midwest oak savannas. Pp.19-30 in J. M. Sweeney, ed. Management of Dynamic Ecosystems. North Cent Sect., The Wildlife Soc., West Lafayette, IN.
Haney, A., and S.I. Apfelbaum. Unpublished ms. Oak barrens of the Upper Midwest.
Henderson, N. R., and J. N. Long. 1984. A comparison of stand structure and fire history in two black oakwoodlands in northwestern Indiana. Botanical Gaz. 145(2): 222-228.
Hill, M.O., and H. G. Gauch, Jr. 1980. Detrended correspondence analysis; an improved ordination technique. Vegetatlo 42: 47-58.
Howell, D.L. and C. L. Kucera. 1956. Composition of presettlement forests in three counties in Missouri: Bull. Torrey Botanical Club 83: 207-217.
Matthiae, P.E., and F. W. Stearns. 1978. Ecological analysis of Moouah Barrens Wildlife Management Area. Univ. of Wisconsin-Milwaukee. Unpub. report.
Noss, R. and M. Scott. (in press) Ecosystem protection and restoration: The core of ecosystem management. In Boyce, M. and A Haney (eds.) Ecosystem Management: Application for Sustainable Forests and Wildlife Resources. Yale University Press.
Nuzzo, V. 1986. Extent and status of midwest oak savannas: presettlement and 1985. Natural Areas Journal 6: 6-36.
Ovington, J.D. 1963. Plant biomass and productivity of prairie, savanna, oakwood, and maize field ecosystems in central Minnesota. Ecology 44 (1): 52-63.
Packard, S. 1988. Just a few oddball species: restoration and the rediscovery of the tallgrass savanna. Restoration Management Notes. 6:13-22.
Stout, A.B. 1944. The Bur oak openings in southern Wisconsin. Trans. Wisconsin Acad. Sci. 36:141-161.
Stritch, L.R. 1990. Landscape-scale restoration of barrens-woodland within the oak-hickory forest mosaic. Restoration and Management Notes 8: 73-77.
Tilman, D. 1984. Plant dominance along an experimental nutrient gradient. Ecology 65:1445-1453.
U.S. EPA. In press. Proceedings of Workshop on Oak Savannas. February 18-19, 1993. Chicago, IL.
Vestal, A.G. 1956. Barrens vegetation in Illinois. Trans. Illinois State Acad. of Science 29: 79-80.
Vogl, R.J. 1965. Effects of spring burning on yields of brush prairie savanna. Journal of Range Management 18(5): 202-205.
Vogl, R.J. 1964. The effects of fire on the vegetational composition of bracken grasslands. Transactions of the Wisconsin Academy of Science, Arts and Letters. 53: 67-82.
Vora, R. S. 1993. Moquah Barrens: Pine barrens restoration experiment initated in Chequamegon National Forest. Restoration and Mangement Notes 11:37-44.
White, A.S. 1986. Prescribed burning for oak savanna restoration in central Minnesota. U.S.D.A. Forest Service Res. Pap. NC266, North Central Forest Experiment Station, St. Paul, MN. 12 p.
FIGURE 1. Approximate distribution of oak savanna, barrens, and prairie complexes in eastern United States. Adapted from Nuzzo (1986).