2006 SOUTHEASTERN NATURALIST 5(1):113–126 Effects of Prescribed Fire on the Vegetation of a Savanna-Glade Complex in Northern Arkansas SEÁN E. JENKINS1,* AND MICHAEL A. JENKINS2 Abstract - In Spring of 1995 and 1997, 170 ha of a savanna-glade complex in the Ozark Mountains of Arkansas were burned. These prescribed burns included 5 of 18 permanent plots established across the site in 1993. We surveyed the ground flora, seedling, sapling, and overstory vegetation of these plots before and after burning. The burns greatly impacted the sapling layer, where density decreased from 2540/ha preburn to 610/ha after the second burn. Ground flora richness was unchanged following burning, but evenness and diversity increased. Species richness, evenness, and diversity also increased on the unburned plots. We observed large increases in the cover of several glade and savanna species after burning. We observed similar trends, but of lesser magnitude, on unburned plots. Introduction Prescribed fire is an essential tool in the management of oak savanna, glades, and other community types throughout eastern North America (Abrams 1992, Cole et al. 1990, Rouse 1986, Stritch 1990). Oak savannas and glades in the Ozark Highlands were historically maintained by anthropogenic fire and had a diverse ground flora (Baskin and Baskin 2000, Guyette and Cutter 1991, Jenkins 1997, Ladd 1991). However, the vast majority of these communities have been lost since European settlement (Christensen et al. 1996, Nuzzo 1986). Fire suppression has resulted in compositional and structural changes in the remaining remnants, with a dramatic shift from open landscape mosaics to closed-canopy, even-age stands of oaks and fireintolerant species (Beilmann and Brenner 1951, Dey 2003, Ebinger 1997, Guyette and McGinnes 1982, Kucera and Martin 1957, Lorimer 1993, Robertson et al. 1997) and a corresponding loss of ground flora diversity (McClain and Anderson 1990, Taft 1997). In the Ozark Highlands, savannas have been overgrown with Quercus marilandica Muench. (blackjack oak), Carya texana Buckl. (black hickory), and fire-sensitive woody species such as Amelanchier arborea (Michx. f.) Fern. (shadbush). According to Reiter (1991), glades have been invaded and fragmented by Juniperus viriginiana L. (eastern red cedar), Q. marilandica, Ulmus alata Michx. (winged elm), and Rhus copallinum L. (winged sumac). To maintain and restore remaining remnants of savanna and glades, the National Park Service and other public and private agencies have initiated prescribed burning programs. While burning has been extensive, the impacts of prescribed fire have been difficult 1Department of Biological Sciences, Western Illinois University, Macomb, IL 61455. 2National Park Service, Inventory and Monitoring Program, Great Smoky Mountains National Park, Gatlinburg, TN 37738. *Corresponding author - SEJenkins@ wiu.edu. 114 Southeastern Naturalist Vol. 5, No. 1 to assess due to the limited pre and post-burn data available from sites that have had multiple burns. On April 3–4, 1995, and again on March 19–20, 1997, the National Park Service burned a 170-ha (420-acre) section of a large savanna-glade complex at Cook Hollow, Buffalo National River, AR. Five permanent plots from an 18-plot array established and sampled in 1993 as part of a long-term study of landscape-scale fire effects (Jenkins and Rebertus 1994) were burned both years. The original burn plan called for the area containing all 18 plots to be burned. However, due to staffing limitations, the burn unit was reduced after the 18 plots were established. While this smaller burn reduced the number of plots burned, it provided unburned reference plots for comparison. Following the two burns, we resampled the woody vegetation (trees, shrubs, and vines) on the five burned plots. During the summer of 1999, we resampled the ground flora on all 18 plots and the woody vegetation on the 13 unburned plots. The prescribed burns at Cook Hollow were conducted to accomplish two management goals: (1) restore overstory structure by reducing woody stem density and basal area, and (2) increase the cover of prairie and glade species while increasing the overall diversity of ground flora species. In this paper, we use data from the burned plots to determine whether the two burns met these management objectives. In addition, we examine structural and compositional changes on the burned and unburned plots to determine if factors other than fire may be impacting the savanna-glade complex. Study Area Cook Hollow is part of the Lower Buffalo Wilderness Area of the Buffalo National River and is located at the confluence of the Buffalo River with the White River in north-central Arkansas. The area is part of the White River Hills Subsection of the Ozark Highlands Section within the Eastern Broadleaf Forest Province (Keys et al. 1995) and is underlain by strata of Ordovician limestone and sandstone, thus it has a rugged topography and frequent rock outcroppings. Approximately 324 hectares (800 acres) of oak woodland-savanna-glade complex cover the south aspect of Turkey Mountain and the adjacent north aspect of Granite Mountain (Fig. 1). The flora of the area is diverse. Jenkins and Rebertus (1994) reported 240 species from the eighteen 500-m2 plots in Cook Hollow on Turkey and Granite Mountain. In addition, Logan (1992) reported 193 species on 42 limestone glades located in the uplands along the Buffalo National River, AR. Carya texana, Quercus stellata Wangenh. (post oak), Quercus marilandica, and Quercus prinoides Willd. (dwarf chinkapin oak) are dominant overstory species across the glade complex at Cook Hollow (Jenkins and Rebertus 1994), and they dominated the overstory on burned and unburned plots sampled in this study. During the first survey in 1993, the 2006 S.E. Jenkins and M.A. Jenkins 115 combined importance value (IV; [relative basal area + relative density]/2) of these 4 species was 69 on the burned and 55 on the unburned plots (Jenkins and Jenkins 1999). Juniperus virginiana was a major overstory species on the unburned plots (IV = 21.8), but it was mostly found in the pre-burn sapling layer of the burned plots. Pinus echinata P. Mill. (shortleaf pine) was a relatively important species on the burned plots (IV = 14.1), but it was a minor component of the unburned plots (IV < 1). Carya texana and Q. marilandica dominated the pre-burn sapling layers of both burned and unburned plots (Jenkins and Jenkins 1999). Methods In 1993, eighteen permanent plots were established and sampled to monitor the long-term effects of fire at Cook Hollow (Rebertus and Jenkins 1994; Fig. 1). In August of 1995 and 1998, following two prescribed burns, the woody vegetation (trees, shrubs, and vines) was resurveyed on five of the plots within the burn unit (1, 2, 3, 10, and 18; Fig. 1). During the summer of 1999, we sampled the ground flora (herbaceous plants and woody/semi woody shrubs < 1 m tall) on all 18 plots and the woody vegetation on the 13 unburned plots. Because we had just sampled woody vegetation on the burned plots in 1998, we did not resample these plots in 1999. Figure 1. Locations of plots in Cook Hollow, Buffalo National River, AR. Plots 1, 2, 3, 10, and 18 are within the 170-ha (420-acre) unit burned in 1995 and 1997. 116 Southeastern Naturalist Vol. 5, No. 1 Field sampling We sampled woody vegetation with a nested plot design. The dbh of all trees ( ≥ 5 cm dbh) was measured within each 500-m2 plot. Saplings (2.5– 4.99 cm dbh) were tallied by species within two 100-m2 subplots nested within each 500-m2 plot. During the 1993 survey, large seedlings (woody stems > 0.5 m tall but < 2.5 cm dbh) were tallied in two 10-m2 circular subplots, and small seedlings (woody stems < 0.5 m tall) were tallied within a 1-m2 quadrat nested within each 10-m2 subplot. During the 1995 and successive surveys, both large seedlings and small seedlings were tallied in four 10-m2 subplots to reduce the large variability we observed in the preliminary analysis of seedling data (Jenkins and Jenkins 1999). We sampled the ground flora in June and in early September of 1993 and 1999. Some pre-vernal species may have already become senescent by June, but most species were detected during sampling. The canopy cover of all herbaceous, semi-woody, and woody ground flora species (< 1m tall) within the plot was estimated in 20% cover-increment classes (1 = 0–20%, 2 = 21– 40%, 3 = 41–60%, 4 = 61–80%, and 5 = 80–100%) within each 500-m2 plot. Species nomenclature follows Kartesz (1999). Data analysis We calculated overstory density (stems/ha), overstory basal area (m2/ha), and sapling density (stems/ha) for each species on each plot for each remeasurement. The percent of total sapling density comprised of dead stems (hereafter “percent dead saplings”) was also calculated for each plot. We used one-way repeated measures ANOVA to compare mean overstory density, overstory basal area, sapling density, and percent dead saplings between years (1993, 1995, and 1999) for the 5 burn plots. When ANOVA revealed a significant effect, we used the Tukey multiple comparison test (α = 0.05) for post-hoc comparisons between sample years. Paired t-tests were used to compare overstory density, overstory basal area, sapling density, and percent dead saplings between years (1993 and 1999) for the 13 unburned plots. Percent dead sapling data were arcsine transformed prior to statistical analysis to improve normality (Zar 1996). Even though the burned and unburned plots were located on similar landscape positions and had similar species composition, they are not true paired replicates. Therefore, we did not make direct quantitative comparisons between burned and unburned plots. Due to the large discrepancy in sampling area used for large seedlings between the pre-burn and post-burn surveys, statistical comparisons were only made between the post-burn surveys. Because of a severe growing season drought, we observed no small seedlings during the 1998 resurvey of the burned plots that could be identified to species. Consequently, only the densities of large seedlings occurring in the two post-burn surveys (1995 and 1998) were compared on the 5 burned plots. All large-seedling densities were converted to a per-ha basis prior to analysis with paired t-tests. We calculated species richness (S), evenness (E), and Shannon-Weiner diversity (H’) for ground flora species on each plot during each survey. We 2006 S.E. Jenkins and M.A. Jenkins 117 used paired t-tests to compare differences in ground flora species S, E, and H’ between the 1993 and 1999 surveys of burned and unburned plots. The scale values of herbaceous vegetation were converted to class midpoints to calculate S, E, and H’. Mid-point cover values were also used to examine changes in the dominance of individual species. The covers of individual prairie grass species, Andropogon gerardii Vitman. (big bluestem), Schizachyrium scoparium (Michx.) Nash (little bluestem), Sorghastrum nutans (L.) Nash (Indian grass), Sporobolus compositus (Poir.) Merr. var. compositus (composite dropseed), Sporobolus clandestinus (Biehler) A.S. Hitchc. (rough dropseed), and Bouteloua curtipendula (Michx.) Torr. (sideoats grama), were combined for comparisons. Paired t-tests were used to compare herbaceous cover mid-points between the 1993 and 1999 surveys of burned and unburned plots. Prior to statistical testing, all cover data were arcsine transformed to improve normality (Zar 1996). In comparisons of ground flora cover, individual species or groups of species often have coverages that exceed 100%. When this occurred, all cover values in comparison groups were mathematically converted to a 100-point scale prior to transformation. Results Overstory trees The mean overstory basal area on the burn plots did not differ significantly between surveys (P = 0.759; 16.3 ± 5.7 m2/ha in 1993, 16.8 ± 6.0 m2/ ha in 1995, and 16.4 ± 5.5 m2/ha in 1999). Total basal area remained relatively constant across the three surveys since most fire-killed trees were of small diameter (data not shown). Mean overstory density was lower following the burns, but these decreases were statistically insignificant (P = 0.358; 1164 ± 275 stems/ha in 1993, 992 ± 273 stems/ha in 1995, and 1048 ± 207 stems/ha in 1999). On the unburned plots, basal area increased (P = 0.118) between 1993 (15.8 ± 2.1 m2/ha) and 1999 (16.5 ± 2.1 m2/ha). Tree density on the unburned plots changed little between 1993 and 1999 (P = 0.554; 1246 ± 116 m2/ha in 1993 and 1274 ± 101 m2/ha in 1999). However, the basal area of three species increased slightly; Carya texana (P = 0.056), Juniperus virginiana (P = 0.049), and Quercus stellata (P = 0.044). Saplings The results of the two post-burn surveys show that the fires had a dramatic impact on the sapling layer (Fig. 2). The density of saplings decreased after each successive burn from an initial density of 2540 stems/ ha in 1993 to 610 stems/ha after the second burn (Fig. 2A). We observed a significantly greater percent of dead saplings following the two burns than in the preburn survey (P < 0.05; Fig. 2B). The mean percent of dead saplings increased from 3% in 1993 to 73% in 1995 after the first burn and 69% after the second burn. Carya texana, Quercus marilandica, Q. stellata, and Sassafras albidum (Nutt.) Nees (sassafras) exhibited the 118 Southeastern Naturalist Vol. 5, No. 1 greatest overall reductions in density, although the reductions were not statistically significant (p > 0.05; Fig. 2A). After the first fire, the percentage of dead saplings for all these species was greatest on two upper slope plots and a lower slope plot dominated by Pinus echinata (data not shown). The mean density of saplings on the unburned plots did not change significantly between surveys (P = 0.602; Fig. 3). In addition, we observed no significant changes in the density of individual species between 1993 and 1999. The mean percent dead saplings did not differ between surveys for all pooled species or any individual species (data not shown). On individual plots located on more drought-prone glades, densities were reduced by up to Figure 2. (A) Sapling density (mean stems/ha ± 1 SE) and (B) percent dead saplings (mean percent ± 1 SE) on 5 burned plots sampled pre-burn in 1993 and post-burn in 1995 and 1998. Mean values from each sample year for each species were compared using a one-way repeated measures ANOVA with Tukey multiple comparison tests. Sample years within a species not superscripted with the same letter are significantly different (P < 0.05). 2006 S.E. Jenkins and M.A. Jenkins 119 72% following three successive years of growing season drought (data not shown). These glade plots have shallow soil over bedrock, and most of the saplings that died were Quercus marilandica, Q. stellata, and Juniperus virginiana. This loss of density was counterbalanced by density increases on plots located in more mesic landscape positions. Seedlings Following the first burn in 1995, overall mean density of large tree seedlings remained stable (4100 stems/ha in 1993 and 4050 stems/ha in 1995; Table 1). However, these results must be interpreted with caution since sample area was increased in the post-burn surveys and statistical analysis was only performed on post-burn data. The density of Sassafras albidum was greater following the first fire. Of the oak species, Quercus Figure 3. Sapling density (mean stems/ha ± 1 SE) on 13 unburned plots sampled in 1993 and 1999. Mean values from each sample year for each species were compared using a one-way repeated measures ANOVA with Tukey multiple comparison tests. Means did not differ significantly (P > 0.05) between years for any comparison group. Table 1. Density per ha (mean ± 1 SE) of selected large seedling (stems > 0.5 m tall; < 2.5-cm dbh) on the 5 burned plots. Data were collected pre-burn in 1993 and post-burn in 1995 and 1998. Mean densities between 1995 and 1998 were compared with paired t-tests. Densities from 1993 are presented for qualitative comparison; no statistical analyses were conducted on these data. Species 1993 1995 1998 P-value Carya texana 200 ± 100 500 ± 300 750 ± 100 0.326 Quercus marilandica 200 ± 200 550 ± 400 400 ± 200 0.727 Quercus stellata 1100 ± 1000 200 ± 100 250 ± 250 0.815 Sassafras albidum 1000 ± 800 2250 ± 1000 2050 ± 1000 0.528 Total Density 4000 ± 900 4050 ± 1400 4600 ± 900 0.504 120 Southeastern Naturalist Vol. 5, No. 1 stellata seedling density was lower after the first burn. The second burn did not significantly increase total large seedling density or the density of any individual species. Field observations showed that resprouts were less numerous and not as vigorous after the second burn, especially those of woody vine and shrub species including Vaccinium stamineum L. (deerberry), Acacia angustissima (P. Mill.) Kuntze (prairie acacia), Passiflora lutea L. (passion flower), Rhus aromatica Ait. (aromatic sumac), Rosa carolina L. (prairie rose), Smilax bona-nox L. (green brier), Vaccinium pallidum Ait. (lowbush blueberry), and Vitis spp. (grape vine). Although present in the larger plots, these woody shrub and vine species were not encountered in the permanent nested circular seedling plots due to patchy distributions and/or lower abundances after the second burn. Ground flora vegetation The results of the pre- and post-burn surveys suggest that the prescribed burns have influenced the diversity and composition of the ground flora (Table 2, Fig. 4A). Species richness (S) did not differ significantly between surveys (P = 0.840), but species evenness (E) and diversity (H’) increased significantly (P = 0.02 and P = 0.04, respectively). However, we noted no significant change in the cover of any individual mesic-site species after burning (P > 0.1; data not shown). The combined cover of prairie grass species increased greatly (171%) after the two burns (Fig. 4A). In addition, the cover of Dichanthelium species (panic grasses) increased 82%, and the cover of two glade species, Lespedeza virginica (L.) Britt.-slender bush clover and Solidago radula Nutt. (western rough goldenrod), more than doubled (Fig. 4). Triodanis perfoliata (L.) Nieuwl. (Venus looking glass— a light-seeded annual) exhibited a large increase in cover due to the seedbed created by the burns. Solidago nemoralis Ait. (gray goldenrod), a common species in abandoned fields and open woods, disappeared from all 5 plots following the two burns. We observed significant increases in the values of S, E, and H’ for the unburned plots (Table 2). We observed a decrease in the cover of 7 mesic and old-field species between the two surveys (Fig. 4B). These species likely increased in cover as a result of the large rainfall in 1993, but decreased in cover in response to the severe droughts of 1996, 1998, and 1999. Conversely, the cover of 5 dry-site and glade species increased between surveys (Fig. 4B). Table 2. Changes in ground flora species richness, evenness, and Shannon-Weiner diversity on 5 burned and 13 unburned plots in 1993 and 1999. Prescribed burns were conducted in 1995 and 1997. Mean values were compared between years with paired t-tests. Richness Evenness Diversity 1993 1999 P-value 1993 1999 P-value 1993 1999 P-value Burned 46.0 45.4 0.840 0.659 0.757 0.026 2.49 2.83 0.042 Unburned 50.5 55.4 0.035 0.689 0.784 < 0.001 2.69 3.13 < 0.001 2006 S.E. Jenkins and M.A. Jenkins 121 Discussion Burning was initiated at Cook Hollow to reduce overstory density and basal area and increase ground flora diversity and extent. Although the prescribed burns have not altered overstory basal area and density, the fires have drastically reduced the density of both C. texana and Q. marilandica Figure 4. Percent cover (mean ± 1 SE) of Prairie grasses, Dichanthelium species, and selected individual species on (A) the 5 burned plots in 1993 (preburn) and 1999 (postburn) and (B) the 13 unburned plots in 1993 and 1999. Species on unburned plots are grouped according to common habitat (mesic site and old-field species vs. dry site and glade species). Mean species cover was compared in burned and unburned plots between 1993 and 1999 with paired t-tests. *P < 0.1, **P < 0.05, ***P < 0.01. Species that exhibited large or significant changes in cover are included. 122 Southeastern Naturalist Vol. 5, No. 1 saplings. Reduced recruitment of saplings and small trees into the larger size classes as a result of repeated burning will lead to a gradual long-term reduction in both tree basal area and density (Harmon 1984, Taft 1997). After thirteen years of burning, White (1983) observed a reduction in the density and basal area of stems less then 25 cm dbh in a Quercus ellipsoidalis E.J. Hill (northern pin oak)-dominated former upland savanna in Minnesota. Peterson and Reich (2001) found that burning every 2 to 3 years reduced sapling density as well as canopy ingrowth, while less frequent fires lead to increased sapling density in savannas. In our study, reductions in sapling density were not uniform across the burned plots. We observed the greatest reductions on upper slopes, which experienced greater preheating of fuels, and in pine stands, which have more pyrogenic liter. Both of these factors led to greater fire intensity in these two areas (Jenkins and Jenkins 1999, Jenkins et al. 1997). On the unburned plots, we observed little change in either the overstory or sapling layers between 1993 and 1999. As with the burned plots, our results were not uniform across all unburned plots. We observed a decrease in the density of Quercus marilandica, Q. stellata, and Juniperus virginiana stems on unburned plots located on exposed glades. These glade sites have shallow soils that amplify the impacts of drought on tree survival (Baskin and Baskin 2000). However, these decreases in stem density were counterbalanced by density increases on plots located in more mesic landscape positions. Gradual increases in the basal area and density of Carya texana, Juniperus virginiana, and Quercus stellata on these sites may have a pronounced negative effect on ground flora diversity due to increased shading by trees and saplings (Taft 1997). In particular, invasion by J. virginiana has been observed on other glades (Erickson et.al. 1942, Heikens and Robertson 1995, Jeffries1985, Kucera and Martin 1957, McClain and Ebinger 2002, Nelson and Ladd 1983) and is often of concern to managers (Jenkins and Rebertus 1994, Reiter 1991). While largely absent in the overstory of the burned plots, Juniperus virginiana was a major component across the 13 unburned plots, where it is more of a management concern. In 2000, the entire glade complex was burned and current plans call for continued burning of this larger unit. This expanded burning should eventually reduce the density of J. virginiana across the savanna-glade complex. A combination of overstory thinning and prescribed burning could be used to more rapidly decrease overstory shading and increase ground flora cover and diversity (Neilson et al. 2003). Our results show that the diversity of ground flora species and the cover of prairie grasses and other savanna/glade species have increased on the burned plots, meeting the management goals of the National Park Service. Other studies have also shown increased cover of prairie grass species following fire in oak savannas (DeSelm and Clebsch 1991, Heikens et al. 1994, Tester 1996). However, Taft (2003) observed a decrease in the cover of C4 prairie grasses (e.g., Schizachyrium scoparium and Sorghastrum nutans) following two burns of a dry sandstone barrens. The significant 2006 S.E. Jenkins and M.A. Jenkins 123 increases in E and H’ we observed may have resulted from post-fire reduction in the cover of more mesic species that have invaded with fire suppression. Similar decreases in the cover of mesic non-savanna species have been observed in other oak savannas in the Midwest following both single (Hruska and Ebinger 1995) and multiple burns (Tester 1996). Since the ground flora was not sampled until two years after the second burn, it is impossible to know the immediate effects of the fire on ground flora diversity at the site. For instance, Nuzzo et al. (1996) observed an increase in herbaceous species richness in an Illinois sand savanna after a prescribed fire, but richness declined following a fire-free year. In a long term study of post-fire effects on a southern Illinois barren, Taft (1997) found that some annual and biennial species increased dramatically following a dormant-season burn, but then declined to preburn abundances two years after the fire. Heikens et al. (1994) found that the abundance of herbaceous species did not significantly increase following a spring burn on shale and chert savannas in southern Illinois. The increase in species richness, evenness, and diversity on the unburned plots was unexpected. However, since there were 6 years separating the surveys, such fluctuations in diversity and composition may have occurred as a result of factors other than fire including periodic droughts, spatiotemporal variability in colluvial processes, and annual variability in freeze-thaw cycles. Periodic disturbance, low soil fertility, and resource heterogeneity have all been suggested as potential mechanisms that promote diversity and prevent dominance by a single species or group of species (Huston 1994, Tilman 1982). We conducted our first survey in 1993, one of the wettest years on record in much of the central United States. Therefore, an abundance of moisture may have given a short-lived competitive edge to the more dominant species. These species likely increased in cover as a result of the large rainfall in 1993, but decreased in cover in response to the severe droughts of 1996, 1998, and 1999. As cover of these species decreased, evenness may have increased, causing a corresponding increase in species diversity. In addition, prairie grasses may have increased in cover during droughts years as growing space was relinquished through the dieback of more mesic woodland species. Cyclic changes in the cover of prairie grasses, such as Schizachyrium scoparium, regardless of the burning regime, have been noted (DeSelm and Clebsch 1991). However, total percent increases in the cover of prairie grasses and several other glade/savanna species were much greater on burned plots than on unburned plots, suggesting that fire may have increased the cover of these species beyond the baseline increase that occurred on the unburned plots. While our results show that the two spring burns were effective in meeting management goals, changing overstory structure will be a slow process under this low-intensity burning regime. While spring has traditionally been the dominant season for prescribed burning in the East, a more mixed seasonal burning regime with more variable fire intensities may better reproduce past 124 Southeastern Naturalist Vol. 5, No. 1 conditions that maintained savannas and glades. Historically, burns in Midwest savanna and glades occurred throughout the growing season with many fires occurring during October and November (McClain and Elzinga 1994). Varying the season, return interval, and size of burns will better maintain tree recruitment and favor a variety of ground flora reproductive strategies (Jenkins and Rebertus 1994). This will, in turn, enhance and maintain avian and general faunal diversity (Brawn et al 2001, Madden et al. 1999). The results of our study highlight the importance of including unburned reference plots in fire effects monitoring. If data were not available from unburned plots, observed increases in species diversity and cover would have been attributed solely to fire. However, our results suggest that, in addition to fire, changes in ground flora diversity and composition are influenced by a combination of poorly understood factors including precipitation cycles, light regimes, topography, edaphic characteristics, resource heterogeneity, and disturbance. Understanding these factors may not be critical in long-term monitoring studies that seek only to quantify change across a given area. However, their understanding is critical to understanding the related mechanisms responsible for this change. Acknowledgments We thank Robert Pulchow, Thara Baker-Alley, Joseph Alley, and Mike Morris for their field assistance. We also thank Chris Webster and two anonymous reviewers for helpful comments on an earlier version of this manuscript. This study was funded through grants from the National Park Service, Buffalo National River, Harrison, AR. Literature Cited Abrams, M.D. 1992. Fire and the development of oak forests. Bioscience 42:346–353. Baskin, J.M., and C.C. Baskin. 2000. Vegetation of limestone and dolomite glades in the Ozarks and Midwest regions of the United States. Annals of the Missouri Botanical Garden 87:286–294. Beilmann, A.P., and L.G. Brenner. 1951. The recent intrusion of forest in the Ozarks. Annals of the Missouri Botanical Garden 38:261–282. Brawn J.D., S.K. Robinson, and F.R. Thompson. 2001. The role of disturbance in the ecology and conservation of birds. Annual Review of Ecology and Systematics 32:251–276. Christensen, N.L., A.M. Batuska, J.H. Brown, S. Carpenter, C. D’Antonio, R. Francis, J.F. Franklin, J.A. MacMahon, R.F. Noss, D.J. Parsons, C.H. Peterson, M.G. Turner, and R.G. Woodmansee. 1996. The report of the Ecological Society of America committee on the scientific basis for ecosystem management. Ecological Applications 6:665–691. Cole, K.L., P.K. Benjamin, and K.F. Klick. 1990. The effects of prescribed burning on oak woods and prairies in the Indiana Dunes. Restoration Management Notes 8:37–38. DeSelm, H.R., and E.E.C. Clebsch. 1991. Response types to prescribed fire in oak forest understory. Pp. 22–24, In S.C. Nodvin and T.A. Waldrop (Eds.). Fire and Environment: Ecological and Cultural Perspectives. Proceedings of an International Symposium, TR-SE 69, US FS SFES, Knoxville, TN. 2006 S.E. Jenkins and M.A. Jenkins 125 Dey, D. 2003. Fire history and postsettlement disturbance. Pp. 46–59, In W.J. McShea and W.M. Healy (Eds.). Oak Forest Ecology: Ecology and Management for Wildlife. John Hopkins University Press, Baltimore, MD. 432 pp. Ebinger, J.E. 1997. Forest communities of the Midwestern United States. Pp. 1–23, In M.W. Schwartz (Ed.). Conservation of highly fragmented landscapes. Chapman and Hall Inc., International Thomson Publishing, New York, NY. 436 pp. Erickson, R.O., L.G. Brenner, and J. Wraight. 1942. Dolomitic glades of east-central Missouri. Annals Missouri Botanical Garden 29:89–101. Guyette, R.P., and B.E. Cutter. 1991. Tree ring analysis of fire history of a post-oak savanna in the Missouri Ozarks. Natural Areas Journal 11:93–99. Guyette, R.P., and E.A. McGinnes. 1982. Fire history of an Ozark glade in Missouri. Transactions of the Missouri Academy of Science 16:85–93. Harmon, M.E. 1984. Survival of trees after low-intensity surface fires in Great Smoky Mountains National Park. Ecology 65:796–802. Heikens, A.L., and P.A. Robertson. 1995. Classification of barrens and other natural xeric forest openings in southern Illinois. Bulletin of the Torrey Botanical Club 122:203–214. Heikens, A.L., K.A. West, and P.A. Robertson. 1994. Short-term response of chert and shale barrens vegetation to fire in southwestern Illinois. Castanea 59:274–285. Hruska, M.C., and J.E. Ebinger. 1995. Monitoring a savanna restoration in east-central Illinois. Transactions of the Illinois State Academy of Science 88:109–117. Huston, M. 1994. Biological Diversity: The Coexistence of Species in Changing Landscapes. Cambridge University Press, Cambridge, UK. 681 pp. Jeffries, D.L. 1985. Analysis of vegetation and soils of glades on Calico Rock Sandstone in northern Arkansas. Bulletin of the Torrey Botanical Club 112:70–73 Jenkins, M.A., and S.E. Jenkins. 1999. Savanna and glade vegetation of Turkey Mountain, Arkansas: Effects of a single prescribed burn. Pp. 127–134, In C. Warwick (Ed.). Proceedings of the 15th North American Prairie Conference. Natural Areas Association, Bend, OR. Jenkins, S.E. 1997. Spatial demography of trees in an oak savanna and adjacent dry chert woodland in the Missouri Ozarks. Ph.D. Dissertation. University of Missouri- Columbia. Columbia, MO. 116 pp. Jenkins, S.E., and A.J. Rebertus. 1994. Spatial demography of an oak savanna in the Ozarks. Pp. 107–112, In R.C. Anderson and J.S. Fralish (Eds.). Proceedings of the North American Conference on Savannas and Barrens. Illinois State University, Bloomington-Normal, IL. Jenkins, S.E., R.P. Guyette, and A.J. Rebertus. 1997. Vegetation-site relations and fire history of a savanna-glade-woodland mosaic in the Ozark Highlands. Pp. 184–199, In S.G. Pallardy, R.A. Cecich, H.G. Garrett, and P.S. Johnson (Eds.). Proceedings of the 11th Central Hardwood Conference, University of Missouri, Columbia, MO. US FS NCES GTR-NC 188. Kartesz, J.T. 1999. A Synonymized Checklist and Atlas with Biological Attributes for the Vascular Flora of the United States, Canada, and Greenland. First Edition. In J.T. Kartesz and C.A. Meacham (Eds.). Synthesis of the North American Flora, Version 1.0. North Carolina Botanical Garden, Chapel Hill, NC. Keys, J., Carpenter, C., Hooks, S., Koenig, F., McNab, W.H., Russell, W., and M.L. Smith. 1995. Ecological Units of the Eastern United States: First Approximation. US Department of Agriculture, Forest Service, Atlanta, GA. Kucera, C.L., and S.C. Martin. 1957. Vegetation and soil relationships in the glades region of the southwestern Missouri Ozarks. Ecology 38:285–291. Ladd, D. 1991. Re-examination of the role of fire in the Missouri oak woodlands. Pp. 67–80, In G.V. Burger, J.E. Ebinger, and G.S. Wilhelm (Eds.). Proceedings of the Oak Woods Management Workshop, Eastern Illinois University, Charleston, IL. 126 Southeastern Naturalist Vol. 5, No. 1 Logan, J.M. 1992. The glades of the Buffalo National River, Arkansas. M.Sc. Thesis. Iowa State University, Ames, IA. 103 pp. Lorimer, G.C. 1993. Causes of the oak-regeneration problem. Pp.14–39, In D.L. Loftis and C.E. McGee (Eds.). Oak Regeneration Symposium Proceedings. USDA Forest Service, Asheville, NC. General Technical Report SE-84. Madden, E.M., A.J. Hansen, and R.K. Murphy. 1999. Influence of prescribed-fire history on habitat and abundance of passerine birds in northern mixed-grass prairie. Canadian Field-Naturalist 113:627–640. McClain, W.E., and E.A. Anderson. 1990. Loss of hill prairie through woody stem invasion at Pere Marquette State Park, Jersey County, Illinois. Natural Areas Journal 10:69–75. McClain, W.E., and J.E. Ebinger. 2002. A comparison of the vegetation of three limestone glades in Calhoun County, Illinois. Southeastern Naturalist 1:179–188. McClain, W.E., and S.L. Elzinga. 1994. The occurrence of prairie and forest fires in Illinois and other Midwestern states, 1670 to 1854. Erigenia 13:79–90. Nelson, P.W., and D. Ladd. 1983. Preliminary report on the identification, distribution, and classification of Missouri glades. Pp. 59–76. In C.L. Kucera (Ed.). Procceedings of the Seventh North American Prairie Conference, Southwestern Missouri State University, Springfield, MO. Nielsen, S., C. Kirschbaum, and A. Haney. 2003. Restoration of midwest oak barrens: Structural manipulation or process-only? Conservation Ecology 7(2):10. Available online at: www.consecol.org/vol7/iss2/Art10. Nuzzo, V.A. 1986. Extent and status of Midwest oak savanna: Presettlement and 1985. Natural Areas Journal 6:6–36. Nuzzo, V.A., W.E. McClain, and T. Strole. 1996. Fire impact on groundlayer flora in a sand forest. The American Midland Naturalist 136:207–221. Peterson, D.W., and P.B. Reich. 2001. Prescribed fire in oak savanna: Fire frequency effects on stand structure and dynamics. Ecological Application 11:14–927. Reiter, S.R. 1991. Woody invasion onto glades of the Ozark National Scenic Riverway, Missouri. M.Sc. Thesis. Iowa State University, Ames, IA. 80 pp. Robertson, K.R., R.C. Anderson, and M.W. Schwartz. 1997. The tallgrass prairie mosaic. Pp. 55–87, In M.W. Schwartz (Ed.). Conservation in Highly Fragmented Landscapes. Chapman and Hall, Inc., New York, NY. 436 pp. Rouse, C. 1986. Fire effects in northeastern forests: Oak. USDA Forest Service, NCFES, St Paul, MN. GTR NC-105. 7 pp. Stritch, L.R. 1990. Landscape-scale restoration of barrens-woodlands within the oak-hickory forest mosaic. Restoration Management Notes 8:73–77. Taft, J.B. 1997. Savanna and open-woodland communities. Pp. 24–54, In M.W. Schwartz (Ed.). Conservation in Highly Fragmented Landscapes. Chapman and Hall, Inc., New York, NY. 436 pp. Taft, J.B. 2003. Fire effects on community structure, composition, and diversity in a dry sandstone barrens. Journal of the Torrey Botanical Society 130:170–192. Tester, J.R. 1996. Effects of fire frequency on plant species in oak savanna in eastcentral Minnesota. Bulletin of the Torrey Botanical Club 123:304–308. Tilman, D. 1982. Resource Competition and Community Structure. Princeton University Press, Princeton, NJ. 296 pp. White, A.S. 1983. The effects of thirteen years of annual prescribed burning on a Quercus ellipsoidalis community in Minnesota. Ecology 64:1081–1085 Zar, J.H. 1996. Biostatistical Analysis. Prentice Hall, Upper Saddle River, NJ. 662 pp.