1994 Proceedings
North American Conference on Savannas and Barrens

The ParkLands Foundation is a non-profit private foundation whose mission is to purchase natural lands, mostly in McLean County, Illinois, and make them available for public use and enjoyment. The largest of the ParkLands holding, the 700 acre Merwin Preserve, is bisected by the Mackinaw River, and occurs on morainal uplands adjacent to the river. The presettlement vegetation of the Preserve was oak savanna and open oak woodlands, with tallgrass prairies occurring on the more level landscapes to the north and south. Most of the savanna areas in the Merwin Preserve were destroyed by agricultural activities, or fire suppression, which allowed trees to fill in the spaces between the scattered savanna trees. ParkLands best examples of remnant savanna occurs on the north side of the Mackinaw River. We have been trying to restore the original character of the vegetation on some of these sites (Nelson et al. 1992).

STUDY AREA

Three study sites were established within the savanna restoration area. One site is characterized by a high abundance of shooting stars (Dodecatheon meadia), which is absent from the other two sites. This site had fewer and larger trees and a more open canopy than the other sites. It occupies a level area immediately above and north of the slope adjacent to the Mackinaw River. The second area occurs immediately to the north of the first site. It has a more closed canopy, and more, but smaller trees. At the time the restoration began, this site had a large number of small (5-8 cm dbh) ironwood trees (Ostrya virginiana) forming a second canopy layer. The third site occupies the upper 10 meters of a steep (40-45o ) south facing slope adjacent to the Mackinaw River. The three sites will be referred to as savanna, woodland, and slope. All sites had some trees removed. In the woodland, the second canopy of ironwood trees and some of the smaller oaks and hickories were removed. All sites were burned annually in late February or March beginning in 1989.

METHODS

The sites were sampled on May 12 and 18 and June 22 and 27 in 1994. In each site, the presence of herbaceous vegetation was recorded in 25 m2 quadrats which were located using a stratified random sampling procedure along sampling lines traversing the length of each site. The length of the sampling lines were 90 meters in the savanna and slope sites and the quadrats were located at about 3.5 meter intervals along the lines. In the woodland, the transect was 75 meters in length, and quadrats were located at about three meter intervals. Frequency and relative frequency of herbaceous species were calculated to determine leading species and to calculate Shannon-Wiener diversity index (Shannon and Weaver 1949) between sites.

Trees (woody stems > 9.0 cm DBH) were sampled on June 15 in belt transects that were 10 meters in width and centered on the transect lines. Canopy coverage was determined for each site by estimating the percent tree canopy directly above the m2 quadrats used to sample herbaceous vegetation. An importance value based on the sum of relative density plus relative dominance divided by two was calculated for the trees.

RESULTS AND DISCUSSION

Based on importance values, the three leading tree species in the savanna were Quercus alba (40.9), Q. velutina (15.8), and Carya ovata (14.0). For the woodland, Q. imbricaria (43.8), Q. alba (20.1), and Carya tomentosa (19.3) were the three leading dominants and on the slope Q. rubra (25.6), Carya ovata (22.5), and Q. macrocarpa (19.8) were the three leading species. Tree density was highest in the woodland and nearly twice that occurring in the savanna. Tree basal areas were similar among the tree sites, with the savanna having the highest tree basal area (Table 1). The trees in the savanna, some of which display an open growth form, tended to be larger than those in the other two sites, with a mean tree basal area of 944 cm2, compared with 455 cm2 and 730 cm2 for the woodland and slope sites, respectively. Percent canopy was also highest for the savanna and lowest for the woodland (Table 1).

The May sample of the savanna had the highest number of herbaceous species sampled. The woodland had the highest number of species sampled in June, and the open slope had the fewest species sampled for May and June. For both samples, the savanna had the highest Shannon-Wiener diversity index (Table 2).

For the savanna site, the two leading herbaceous species in May were Tradescantia virginiana and Agrimonia rostellata and in June, Juncus tenuis and Dicanthelium acuminatum were the two leading species (Table 1). The high abundance of J. tenuis, commonly called path rush, is indicative of the relative open nature of the habitat for herbaceous species in the savanna. None of these species were restricted to the savanna site. The woodland had Sanicula gregaria and Viola sororia as the two leading species in the May and the June samples. For the slope site, Agrimonia rostellata and Taenidia intergerrima were the two leading species for both of the samples.

One of the characteristic features of the savanna was the presence of species of forbs which only occurred in the savanna and are associated with prairies or other open habitats. These species included Dodecatheon meadii, Rudbeckia hirta, Pycnanthemum pilosum, Solidago nemoralis, Hypoxis hirsuta, Krigia biflora, Veronicastrum virginicum, and Ruellia humilis ( Mohlenbrock 1986; Curtis 1959; Mead 1846). It is of interest that none of the grasses occurring in the savanna are associated with prairies and most of them are C-3 grasses associated with open woodlands such as Elymus hystrix, Bromus pubescens, Elymus villosus, and Festuca ovata. The site also had a fairly large number of members of genus Carex (7) and several (Carex pennsylvanica, C. blanda, and C. hirsutella) had frequencies greater than 50% in at least one of the two samples.

The savanna also displayed more variation in species composition and abundance between the May and June samples that the two other sites. Several species including Tradescantia virginiana, Camassia scilloides, Dodecatheon meadii, and Ranunculus fascicularis were common in May but were not generally included in the June sample.

Our data show that the savanna we studied mostly contains a mixture of forest grasses and woodland and prairie forbs. The species which are unique to the savanna habitat for the most part are species of prairie forbs. Many other species occur in all three habitats supporting the concept of continuous changes in species composition from savanna to forest and savanna to grassland.

Other authors have historically treated savannas as part of a continuum from prairie to forest, at least within a transition zone. The Cross Timbers of Oklahoma were described by Dyksterhuis (1948) as an "oak savanna" with grassland understory, maintained by fire. He concluded that this timbered area, disconnected from eastern deciduous forests, was located in a grassland climate, and in the absence of fire and abundant grazing, was becoming closed woodland.

While we refer to our most open site as a savanna, it does not meet the criterion established by Curtis (1959), that a savanna should have less than 50% tree canopy cover (Table 1). It is however, the most open of three sites studied and falls within the range of canopy openings described as savanna by the Illinois Natural Areas Inventory (White 1979).

ACKNOWLEDGMENTS

ParkLands Foundation permitted the conduct of research on the Merwin Preserve. In addition to the authors, Michelle Woodby assisted with data collection and analysis. Many dedicated ParkLands Foundation volunteers worked to restore savanna habitat, including selective removal of woody plants and burning.

LITERATURE CITED

Curtis, J. T. 1959. The Vegetation of Wisconsin: An ordination of plant communities. University of Wisconsin Press, Madison, WI.

Dyksterhuis, E. J. 1948. The Vegetation of the Western Cross Timbers. Ecological Monographs 18: 326-376.

Mead, S. B. 1846. Catalogue of Plants Growing spontaneously in the State of Illinois, the principal part near Augusta, Hancock County. The Prairie Farmer 6: 35-36, 60, 93, 119-122.

Mohlenbrock, R. H. 1986. Guide to the Vascular Flora of Illinois. Southern Illinois University Press, Carbondale, IL.

Nelson, A., K. Bastien and R. Anderson. 1992. Reconstructing our Vegetational History. Parklands Foundation Newsletter, August-September.

Shannon, C. E. and W. Weaver. 1949. The Mathematical Theory of Communication. University of Illinois Press, Urbana, IL.

White, J. 1978. Illinois Natural Areas Inventory Technical Report. Illinois Natural Areas Inventory. Urbana, IL.

Table 1. Community characteristics and frequency for species with > 50% frequency in May or June, 1994, in savanna, woodland or open slope, ParkLands Foundation Merwin Preserve. Underline indicates highest %F in all samples.

Table 2.  Shannon-Weiner Index of Diversity for ParkLands Foundations Savanna Restoration Project, Savanna, Woods, Open Slope, in May and June, 1994.