1994 Proceedings
North American Conference on Savannas and Barrens

Flatwoods in the Midwestern United States are typically open, post oak-dominated woodlands of level uplands and stream terraces. They are characterized by soils with nearly impervious subsoil horizons and surface horizons that are seasonally saturated and seasonally dry. Flatwoods occur locally throughout the southeastern United States (Braun 1950; Shelford 1963). In the Midwest, flatwoods are reported from Ohio (Braun 1936), Indiana (Aldrich and Homoya 1984; Dolan and Menges 1989), Illinois (Telford 1926), and Missouri (Nelson 1985). The goals of this study were to describe edaphic factors that unify flatwoods and identify which of these edaphic factors best explain variation in flatwoods vegetation.

STUDY AREA

Six sites, all located within the Southern (Illinoisan) Till Plain Natural Division of Illinois (Schwegman et al. 1973), were chosen for this study based on their size (>10 ha) and natural quality (Grades A or B, according to White [1978]). One site (Lake Sara Flatwoods) had been managed with nearly annual fire for about 20 years prior to sampling. No other sites had a recent salient fire history. Post oak flatwoods formerly were common throughout the Illinoisantillplain south of the Illinois River. Telford's (1926) post oak forest type occupied a region of about 2,250,000 ha and was about 63% forested, the remaining area mostly prairie. The upland forested regions included flatwoods with upland oak-hickory woodland and barrens where the tillplain is more dissected (Englemann 1863). Only about 257 ha of southern flatwoods remain in Illinois in a condition perceived as high natural quality (White 1978).

METHODS

Vegetation was sampled in a stratified-random sampling regime using 0.05-ha circular plots for trees, 0.005-ha circular plots for shrubs and saplings, and 25 0.25-m2 ground-cover quadrats along a line transect oriented randomly through the plot center. A total of 50 tree and shrub/sapling plots and 1,130 ground cover quadrats were sampled among the six study sites. Diversity was measured with the Shannon-Weiner index of diversity (H') (Shannon and Weaver 1949) and, for ground-cover species, species density (the average number of species sampled in the 0.25-m2 ground cover frequency/cover plots). Coefficients of change between woody strata (i.e., similarity between tree and sapling strata) at each site were compared following Loucks (1970) using absolute density for the quantitative modification. Soil samples were taken from the A, E, and B horizons at the tree plot centers yielding a total of 150 soil samples. Tree, shrub/sapling, ground cover and soils data from the 50 tree plots were ordinated using principal components analysis. We examined the relationship of vegetation to site characteristics (primarily soils) using redundancy analysis (RA) CANOCO (Ter Braak 1988). A more detailed account of data analysis methods is described separately (Taft et al., in review).

RESULTS

Soils at all but two of 50 plots were characterized by the presence of argillic horizons (claypans) (Figure 1). Flatwoods remnants in this study are separable into two groups based on parent material origin and an ordination of 49 soil variables (Taft et al., in review). One group is composed of plots from sites with soils developed entirely in loess (Posen, Williams Creek, and Lake Sara Flatwoods). These sites are characterized by silt-loam soils with greater available water-holding capacity (AWC) compared with the second ordination group (t-test, N = 50, P < 0.001). The second group includes plots from two sites, both near the Kaskaskia River, with sandy loam and sandy-clay loam soils developed primarily in lacustrine deposits (Recker and Chip-O-Will woods). Soils at Jackson Slough Woods developed primarily in alluvium and possibly lacustrine deposits and show the most variability in PCA space (Taft et al., in review).

Density of woody stems (trees and shrub/saplings combined) is partially explained by variance in AWC (R2 = 0.13, P < 0.02) and depth to B horizon (R2 = 0.11, P < 0.03). In contrast, importance of Quercus stellata, by far the most important tree species in these study sites with 92 % frequency and 57.4 % of the total importance value, is inversely correlated to the depth to B horizon (R2 = -0.27, P < 0.001). The dominance of Q. stellata is particularly strong in the mid-to-larger size classes (Fig. 2). Recruitment of Q. stellata in the smallest size class (6-10 cm DBH), however, varied between sites from 5 or fewer trees/ha at four sites (Lake Sara Flatwoods, Posen Woods, Williams Creek Woods and Jackson Slough Woods) to 35 trees/ha at Chip-O-Will Woods and 52.5 trees/ha at Recker Woods.

The ordination (RA) relating trees to soils showed a strong association between calcium in the A horizon and Axis I, which had a strong negative loading of Q. stellata, positive loadings for Quercus velutina and Quercus alba, and accounted for 58 % of the variation in tree basal areas. Plots with relatively greater importance of Q. velutina (ranking third in importance with 7.5 % of the IV) and Q. alba tend to be those highest in calcium in the A horizon and with the greatest depth to B horizon. The plots supporting the greatest importance of Q. alba included three of the four plots lacking Q. stellata.

Importance of Quercus marilandica (ranking second in importance with 8.7% of the IV) is positively correlated with magnesium content in the E and B horizons (R2 = 0.14, P<0.008; R2 = 0.26, P<0.001), percent sand (particularly in the B horizon, R2 = 0.24, P<0.001), and negatively correlated with the Ca:Mg ratio, particularly in the E horizon (R2 = 0.15, P<0.006). The significance of these relationships was also tested with Chi-square yielding a significant association between Q. marilandica and % sand in the B horizon (X2 = 14.3, P<0.001) and Q. marilandica and the B horizon Ca:Mg (X2 = 4.4, P<0.05).

Total shrub/sapling density is related partially to water-holding capacity in the surface horizons; this stratum is negatively correlated to percent sand in the A horizon (R2 = 0.12, P<0.013) and positively correlated to depth to the B horizon (R2 = 0.09, P<0.038). Quantitative similarity between tree and sapling strata varied widely among sites with Recker and Chip-O-Will woods, the two sites with soils developed primarily in lacustrine deposits, with 49 % and 50 % similarity, respectively. The remaining sites were characterized by between-strata similarity from 12 % to only 21 %.

The ground cover is dominated at the unburned sites by Parthenocissus quinquefolius (top ranking species overall with 15 % of the IV). Following in importance is a floristic combination of upland species (e.g., Helianthus divaricatus, Carex artitecta) with grasses and sedges of lowland and transitional habitats (e.g,. Agrostis perennans, Cinna arundinacea, Carex festucacea, C. hirsutella, C. caroliniana, and Eleocharis verrucosa). Ground cover species diversity (H') is inversely related to tree diversity and total woody stem density (Table 1). The ordination of herbaceous strata data distinctly separated the 10 Lake Sara plots from the tight cluster of the remaining 40 plots along PCA Axis I (explaining 61 % of the variation in ground cover importance values) (Taft et al., in review). Significant differences in measures of ground cover diversity and percent bare ground between Lake Sara and the other sites account for this separation. Comparing differences between Lake Sara and the other sites, diversity (H'), richness, and average density in quadrats are all significantly greater at P<0.005. Ground cover species density at Lake Sara (x = 7.48, sd = 1.6) is over four times that in the remaining sites (x = 1.85, sd = 0.69). Percent bare ground is significantly less at Lake Sara (x = 32 % [sd = 14] compared with the other sites (x = 76 % [sd = 9.8]) at P<0.005. In contrast, Posen Woods, with soils sharing the most similarity in PCA space to Lake Sara Flatwoods and the greatest total stem density, ranks with the least diverse ground cover in this study.

DISCUSSION

The edaphic variability as it relates to effective soil depth (depth to B horizon) and AWC, as in other studies (Fralish 1994; Adams and Anderson 1980), has imposed habitat patches in post oak flatwoods of varying size and severity. However, in flatwoods, where the drought tolerant Q. marilandica can occur a mere few meters from Q. palustris (a wetland species), the moisture gradients ordinarily influenced by slope and aspect are compressed to edaphic characteristics in an elevational plane with only slight and local microdepressions. Floristic composition and, to an extent, diversity depend in part on selection through habitat sorting of species in the matrix environments into these patches of varying AWC, calcium, and magnesium content. However, since many vascular plant species in flatwoods are not currently present in the matrix environments, long-distance dispersal and probably historic vegetation patterns established during the last hypsithermal interval (King 1981) when Quercus spp. showed marked increase (Prentice et al. 1991) are likely contributing factors.

As in the forests studied by Adams and Anderson (1980), abiotic factors affecting diversity of tree species and total stem density are related to effective soil depth (depth to B horizon) and water-holding capacity (Table 1). As with other studies (Risser and Rise 1971), tree species diversity (H') was inversely related to importance of Q. stellata (R2 = 0.21, P< 0.001). Although Q. stellata and Q. marilandica are viewed as more drought tolerant than, for example, Q. alba and Q. velutina (Reich and Hinckly 1980), edaphic factors independent of soil moisture availability in serpentine soils are also known to impose selective forces for these species (Hull and Wood 1984). However, the results from the RA indicating a negative correlation between post oak importance and nutrient content, particularly calcium in the A horizon, differ from other studies where Q. stellata was an important species (Johnson and Risser 1972; Reich and Hinckly 1980). Reduced importance of Q. stellata with increasing soil depth suggests that claypans, particularly when near the surface, simulate conditions of upland, shallow-soil habitats that support high importance of Q. stellata (e.g., Fralish et al. 1991).

Quercus marilandica reaches its greatest importance in this study at Recker (41 % IV) and Chip-O-Will (30 % IV) woods where magnesium and sand content, at all horizons, are significantly higher (T-test, N = 50, P < 0.001) than plot data from the other sites and Ca:Mg ratio is significantly lower (T-test, N = 50, P<0.001). The enhanced Mg content is apparently related to the separate parent material origin of the Hurst soils. The sand is likely of aeolian origin following deposition in Lake Kaskaskia, a late Pleistocene slackwater lake in the Kaskaskia River basin (Willman and Frye 1970; Lineback 1979). These materials were probably derived from Silurian-aged dolomite in the Lake Michigan basin (Willman 1973), while the other sites with soils developed solely in loess have mineral contributions from other sources. Cations like calcium and magnesium readily are leached out of upper soil horizons, particularly in coarse-textured soils (Brady 1974). The abundance of Mg in the sandy E horizon characteristic of Recker and Chip-O-Will woods may be due to secondary precipitates seasonally abundant during dry periods that simulate conditions characteristic of more arid environments (Tisdale and Nelson 1975). Although Q. marilandica is typical of xeric sites (Bruner 1931; Adams and Anderson 1980; Fralish et al. 1991), Johnson and Risser (1972) and Reich and Hinckley (1980) found it to be a species selectively adapted to nutrient poor conditions over drought tolerance. Dry environments have been shown to have reduced cation movement (Brown 1953) and may exacerbate the effects of low nutrient habitats on vegetation (Reich and Hinckley 1980).

Soils with Ca:Mg ratios less than 1.0 have been shown to be limiting for plant growth (Loew and May 1901) and may limit competitive forces on Q. marilandica in flatwoods. Though the Ca:Mg ratios at Recker and Chip-O-Will woods in this study, with E and B horizon means of 0.42 (sd 0.17) and 0.32 (sd 0.09) are not as low as can be found in some serpentine soils (Walker et al. 1955), selective tolerance of low Ca:Mg ratios and possibly magnesium toxicity for certain species (Proctor 1970), may be an important factor that locally promotes the importance of Q. marilandica at these two sites. Patterns of distribution of trees in and adjacent to a Maryland serpentine area show strong selection for Q. marilandica and Q. stellata on the serpentine soils (Hull and Wood 1984). Low tree diversity at Chip-O-Will woods (5 species) may also be partially explained by edaphic exclusion.

Post oak flatwoods occurred within a prairie/forest patchwork. Recent concerns over the use of fire management in midwestern savanna restoration efforts have cited the loss of certain species and biomass in woody substrata (Mendelson et al. 1992). Comparative data from the present study suggests that, although fire absence may yield an increase in diversity in the shrub/sapling stratum (except at Recker and Chip-O-Will woods where these changes appear to be edaphically retarded), there is a loss of long-term stability and ground cover diversity. The greatest diversity in post oak flatwoods is in the ground cover. The pattern of decreasing ground cover diversity with increasing tree and total stem density and diversity found in this study offers a measure of the cost of the continued absence of fire. Comparative ground cover species diversity data between the edaphically similar Posen Woods and Lake Sara Flatwoods are particularly suggestive of the loss of diversity associated with woody encroachment. In the increasingly fragmented landscape of the midwest, "natural" fires are essentially non-existent. Efforts to conserve biodiversity appear to be dependent on management practices, such as prescribed fire, that maintain or enhance existing levels of diversity as a buffer against loss of small population sizes. Prescribed fire also appears to be necessary to ensure long-term vegetational stability in flatwoods remnants, particularly sites with silty loam soils.

LITERATURE CITED

Adams, D. E., and R. C. Anderson. 1980. Species response to a moisture gradient in central Illinois forests. American Journal of Botany 67: 381-392.

Aldrich, J. R. and M. A. Homoya. 1984. Natural barrens and post oak flatwoods in Posey and Spencer Counties, Indiana. Proceedings of the Indiana Academy of Science 93: 291-301.

Brady, N. C. 1974. The Nature and Properties of Soils. 7th edition. MacMillan Publishing Co., Inc. NY. 639 pp.

Braun, E. L. 1936. Forests of the Illinois till plain of southwestern Ohio. Ecological Monographs 6: 89-149.

• 1950. Deciduous Forests of Eastern North America. Hafner Publishing Company, New York. 596 pp.

Brown, D. A. 1953. Cation exchange in soils through the moisture range saturation to wilting point. Soil Science Society of America Proceedings 17: 92-96.

Bruner, W. E. 1931. The vegetation of Oklahoma. Ecological Monographs 1: 99-188.

Dolan, R. W. and E. S. Menges. 1989. Vegetation and environment in adjacent post oak (Quercus stellata) flatwoods and barrens in Indiana. American Midland Naturalist 122: 329-338.

Engelmann, H. A. 1863. Remarks upon the causes producing the different characters of vegetation known as prairies, flats, and barrens in southern Illinois with special reference to observations made in Perry and Jackson Counties. American Journal of Science and Arts 36: 384-396.

Fralish, J. S., F. B. Crooks, J. L. Chambers, and F. M. Harty. 1991. Comparison of presettlement, second-growth and old-growth forest on six site types in the Illinois Shawnee Hills. American Midland Naturalist 125: 294-309.

• 1994. The effect of site environment on forest productivity in the Illinois Shawnee Hills. Ecological Applications 4: 134-143.

Hull, J. C. and S. G. Wood. 1984. Water relations of oak species on and adjacent to a Maryland serpentine soil. American Midland Naturalist 112: 224-234.

Johnson, F. L., and P. G. Risser. 1972. Some vegetation-environment relationships in the upland forests of Oklahoma. Journal of Ecology 60: 655-663.

King, J. E. 1981. Late quaternary vegetational history of Illinois. Ecological Monographs 51(1): 43-62.

Lineback, J. A. 1979. Quaternary deposits of Illinois (map). Illinois State Geological Survey. Urbana.

Loew, O., and D. W. May. 1901. The relation of lime and magnesia to plant growth. I. Liming of soils from a physiological standpoint. II. Experimental study of the relation of lime and magnesia to plant growth. United States Department of Agriculture, Plant Industries Bulletin 1. 53 pp.

Loucks, O. L. 1970. Evolution of diversity, efficiency, and community stability. American Zoologist 10: 17-25.

Mendelson, J., S. P. Aultz, and J. D. Mendelson. 1992. Carving up the woods. Savanna restoration in northeastern Illinois. Restoration and Management Notes 10(2): 127-131.

Nelson, P. W. 1985. The terrestrial natural communities of Missouri. Missouri Department of Natural Resources and Missouri Department of Conservation, Jefferson City. 197 pp.

Prentice, I. C., P. J. Bartlein, T. Webb III. 1991. Vegetation and climate change in eastern North American since the last glacial maximum. Ecology 72: 2038-2056.

Proctor, J. 1970. Magnesium as a toxic element. Nature 227: 742-743.

Reich, P. B., and T. M. Hinckley. 1980. Water relations, soil fertility, and plant nutrient composition of a pigmy oak ecosystem. Ecology 61: 400-416.

Risser, P. G., and E. L. Rise. 1971. Diversity in tree species in Oklahoma upland forests. Ecology 52: 876-880.

Schwegman, J. E., G. B. Fell, M. D. Hutchinson, G. Paulson, W. M. Shephard, and J. White. 1973. Comprehensive plan for the Illinois Nature Preserve system. Part 2. The natural divisions of Illinois. Illinois Nature Preserves Commission, Rockford, IL. 32 pp.

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

Shelford, V. E. 1963. The Ecology of North America. University of Illinois Press, Urbana. 610 pp.

Taft, J. B., M. W. Schwartz, and L. R. Phillippe. Vegetation, biogeography, and physiographic ecology of flatwoods on the Illinoian tillplain. In review.

Telford, C. J. 1926. Third report on a forest survey of Illinois. Illinois Natural History Survey Bulletin 16: 1-102.

Ter Braak, C.J.F. 1988. CANOCO - a FORTRAN program for canonical community ordination by correspondence analysis, principal components analysis and redundancy analysis (version 2.1). Agricultural Mathematics Group. Wageningen, The Netherlands.

Tisdale, L. S., and W. L. Nelson. 1975. Soil Fertility and Fertilizers. Third edition. Macmillan Publishing Co., Inc. NY. 694 pp.

Walker, R. B., H. M. Walker, and P. R. Ashworth. 1955. Calcium-magnesium nutrition with special reference to serpentine soils. Plant Physiology 30: 214-221.

White, J. 1978. Illinois Natural Areas Technical Report Volume 1. Survey Methods and Results. Urbana. Illinois Natural Features Inventory.

Willman, H. B. 1973. Rock stratigraphy of the Silurian System in northeastern and northwestern Illinois. Illinois State Geological Survey Circular 479. Urbana, Illinois.

Willman, H. B., and J. C. Frye. 1970. Pleistocene stratigraphy of Illinois. Illinois State Geological Survey Bulletin 94. Urbana, Illinois

Table 1. Correlation coefficients (R2) for selected linear regressions for data from six flatwoods on the Illinoisan tillplain of Illinois. Significance levels are indicated * = P > 0.05, ** = P > 0.01, *** = P > 0.001.