Page  65 ï~~2007 THE MICHIGAN BOTANIST 65 CHANGES IN HOOSIER PRAIRIE OAK SAVANNA DURING 27 YEARS OF PRESCRIBED FIRE MANAGEMENT John A. Bacone Paul E. Rothrock Division of Nature Preserves Department of Earth and Environmental Science Indiana Department of Natural Resources Taylor University 402 West Washington Street 236 W. Reade Avenue Indianapolis, Indiana 46204 USA Upland, Indiana 46989 USA Gerould Wilhelm Thomas W. Post Conservation Design Forum Division of Nature Preserves 375 W First Street Jasper-Pulaski Fish and Wildlife Area Elmhurst, Illinois 60126 USA 5822 N. Fish and Wildlife Lane Medaryville, Indiana 47957 USA ABSTRACT Hoosier Prairie Nature Preserve, located in northwestern Indiana, protects a rare and unusually high quality example of a Midwestern (USA) oak savanna. Although many savannas in the region have been lost to agriculture, urbanization, and fire suppression, parts of the 178 hectare preserve never lost its historic community structure. A program of prescribed burning was initiated in 1978 and monitoring of two 10 x 10 m sample areas (Blocks) began in 1979. Throughout the 27 year observation period each Block retained a distinct understory composition even though subjected to similar fire regimes. At the same time, each underwent notable and individualistic changes from their original species composition (SSI = 64-69%) but continued to support a remarkably high diversity of xeric and mesic forbs. Native species richness ranged from 34-58 species per Block and as high as 16.8 species per 0.67 m2 quadrat. Additionally, according to the Floristic Assessment metrics, the species quality of contemporary Hoosier Prairie is comparable (mean C = 4.9) to that expected in regional pre-settlement oak savanna communities. KEYWORDS: savanna, fire management, succession, species composition, floristic quality INTRODUCTION In the pre-settlement Midwestern United States, oak savanna formed a transition zone between the eastern deciduous forest and tall grass prairie biomes (Nuzzo 1986). Although definitions of Midwestern oak savanna vary (Nuzzo 1986; Heikens and Robertson 1994), these savannas are generally described by their physiognomy: a community characterized by scattered, open-grown, somewhat even age trees that mostly belong to the genus Quercus (Curtis 1959). This leaves open the question of how much canopy cover may be associated with the savanna community. Curtis (1959), working in Wisconsin, suggested a range of at least one tree per acre but less than 50% canopy cover, boundaries affirmed by Anderson and Bowles (1999). In contrast, D.M. Anderson (1982), in classifying Corresponding author: Paul E. Rothrock, Randall Environmental Center, Taylor University, 236 W Reade Avenue, Upland, IN 46989-1001. Fax: 765-998-4976, [email protected]

Page  66 ï~~66 THE MICHIGAN BOTANIST Vol. 46 vegetation communities for Ohio, placed more emphasis on the prairie-like nature of the savanna understory. As a result his definition of oak savanna encompasses a canopy cover from 10 to near 100 per cent (Anderson 1982). Regardless of the precise definition or range of canopy cover, the open canopy structure and prairie-like understory of Midwest savannas are a result of periodic drought and fires (Anderson and Bowles 1999). During the presettlement period and through the 19th century, the frequency of fire in Midwestern savannas was remarkably high. Guyette and Cutter (1991), for example, estimated that the fire-free interval was only 4.3 years in their study of Ozark post oak savanna. With fire suppression during the 20th century (Taylor 1990), many savannas converted into closed forest as woody cover from oak sprouts and grubs increased (Cottam 1949, Anderson and Brown 1983, Robertson and Heikens 1994) and more shade tolerant and fire intolerant species invaded (Anderson and Schwegman 1991; McClain et al. 1993). The conversion to closed woodland was particularly rapid in deep-soil savannas, a community type now completely lost from the upper Midwest region (Leach and Ross 1995). Hoosier Prairie Nature Preserve, a part of the Indiana Nature Preserve system since 1977, includes a dry-mesic sand savanna (Bacone 1977) with widely spaced black oak trees (Quercus velutina Lam.). Given its location in northwestern Indiana one would consider Hoosier Prairie to be part of the so-called prairie peninsula (Transeau 1935). Because of its proximity to Lake Michigan, it should be noted that the site was lake bottom during a portion of the post-Wisconsinan glacial period. Thus, topographic relief is minimal and the vegetational history of the site is relatively short at approximately 8000 years or less (Hartke et al. 1975). Because of this combination of factors, Hoosier Prairie may represent a particularly uncommon type of remnant oak savanna. At the time of purchase and its designation as an Indiana state nature preserve, portions of Hoosier Prairie retained a high quality prairie understory while other degraded sections had dense woody understory (Bacone 1977). Although the use of prescribed fires was an unfamiliar and even controversial management tool for nature preserves in Indiana at the time, immediate fire management was recommended (Bacone 1977). In the intervening period, a more substantial literature regarding the benefits and community dynamics in response to fire has developed. These studies have demonstrated that fire increases cover of prairie herbaceous species (e.g., Tester 1989; Abella et al. 2004) especially after a program of multiple prescribed burns (King 2000; Heikens et al. 1994). Although a 2-year burn cycle may produce the highest species richness (Tester 1989), the effects of prescribed burn are also dependent upon the season of the year and fuel load. Dornbush (2004) reported a shift from xeric toward more mesic and late flowering species when burns were limited to spring and Howe (1994, 1995) observed an increase in C3 over C4 species following cycles of mid-summer fire. In terms of fuel load, biomass accumulation between burns needs to be sufficient to achieve the anticipated reduction in sapling and shrub cover (Anderson and Brown 1986; Rebertus and Burns 1997). A thirteen year program of annual burning (White 1983), while successful in increasing the herbaceous species richness, failed to return tree density to historic savanna condition due to low mortality in individuals exceeding 25 cm dbh. In fact, once a site has lost its historic com

Page  67 ï~~2007 THE MICHIGAN BOTANIST 67 munity structure due to fire suppression, its restoration can prove to be slow and uneven (Anderson et al. 2000; Abella et al. 2004). At Hoosier Prairie, a prescribed burn program was initiated in winter of 1978 that sought to simulate a presettlement fire regime. The adopted regimen recognized that regional Amerindians used frequent dormant season burns to drive game, improve pasture, and increase yields of food plants (Anderson 1990). At the same time, it ignored the possible role of less frequent summer fires set by dry lightning (Howe 1994,1995; Loope and Anderson 1998), in part because their potential importance was unknown at the time. Over the past 27 years, as many as 13 burns were administered on some sections of Hoosier Prairie Nature Preserve. The program included late fall as well as spring burns at intervals that ranged from seven to 72 months apart. In 1978 and 1979, plots were established with the intent of following the long term changes at Hoosier Prairie and monitoring the effects of the fire management program. The objective of this paper is to report the changes observed over the past 27 years in understory vegetation including fluctuations in species composition and floristic quality. METHODS Site Description Hoosier Prairie is a 178 hectare site located on the west side of the town of Griffith in Lake County, Indiana (Figure 1). The property is bounded on the west by Kennedy Avenue and more or less bisected by the east-west road, Main Street. The topography has only minor relief of less than 2 m. As a result, small changes in elevation result in rapid transitions from the xeric savanna into wet prairies and marshes. The soils of xeric areas are porous Brems fine sand, Watseka loamy fine sand, and Maumee Loamy fine sand (NRCS 2006). Data Collection and Analysis In 1978 and 1979, 2 permanent plots were established, one south of Main Street (designated South Block; N 41.521710 / W 87.453150) and the other to the north of road (North Block; N 41.523610 / W 87.454350), separated by a distance of approximately 225 m. These Blocks were randomly placed within areas deemed as having the highest quality oak savanna condition. Each 10 x 10 m plot consisted of 25 contiguous sample areas of 2 x 2 m. Within each sample area the species present were recorded from a 0.67 m2 area demarked by the more or less blind toss of a circular hoop. Nomenclature for species follows Swink and Wilhelm (1994); vouchers for Hoosier Prairie are available at MOR. In the South Block the first sampling was attempted in September 1978 but the data set is incomplete. Thus, these data provide some useful qualitative information about the site before fire management but could not be used in quantitative analyses. Full data collection on both Blocks began in 1979, the year following the first prescribed burns, and was performed annually from 1979 to 1994 during late July / early August. Two additional samplings were completed in early August 2004 and 2006. No data are available for the North Block in 1980. In the South Block, prescribed burns were administered during the spring of 1984, 1985, 1987, 1988, 1990, 1992, 1994, 1997, 2000, 2006 and early winter of 1978, 1979, and 1982. The North Block was burned during the spring of 1978, 1984, 1988, 1990, and 1998 and early winter of 1979, 1981, 1985, 1990, 1993, and 2001. Changes in community structure were examined by Nonmetric Multidimensional Scaling (NMS) using PC-ORD5 software (McCune and Mefford 2006). Sorenson (Bray-Curtis) served as the distance measure. A random starting configuration was employed in the first run of 50 real data and 50 randomized data iterations. Scree plots recommended that two dimensions be analyzed in the final runs. The final runs had 200 iterations and in each case an r2 greater than 0.84. The resulting Kruskal stress was very low (<1) for the combined Block analysis and low (7.5) to moderate (14.3) for the South and North Blocks respectively. To further quantify the degree of community change over time

Page  68 ï~~68 THE MICHIGAN BOTANIST Vol. 46 68 THE MICHIGAN BOTANIST Vol. 46 Hoosier Prairie Nature Preserve Lake County 0 225 450 900 1,350 1,00 Metr IDNR Nature Preserves June 6, 2007 N, S Survey Plots E Hoosier Prairie FIGURE 1. Site map of Hoosier Prairie Nature Preserve, Lake County, Indiana. Note its proximity to the town of Griffith and land dominated by industrial activity.

Page  69 ï~~2007 THE MICHIGAN BOTANIST 69 Sorenson's Similarity Indices (SSI) were calculated for each Block with 1979 as the base year. In order to assess whether the management program affected any long term changes in the floristic quality on the site, several metrics used in Floristic Quality Assessment (FQA; Swink and Wilhelm 1994) were applied. FQA draws upon the concept of conservatism, an estimation of the fidelity of any particular plant species to an undisturbed, pre-settlement community such as an oak savanna. Each species in the Chicago region has an assigned coefficient of conservatism (Swink and Wilhelm 1994) where C ranges from 0 for ruderal species to 10 for species limited to communities with little human disturbance. Based upon the species observed and their richness (n), mean coefficients of conservatism (mean C) and floristic quality indices (FQI) can be calculated, where FQI = mean C Vn. In this study, we calculated these metrics at both an overall Block level as well as at the quadrat level. FQA metrics generally are calculated both with and without non-native species in order to evaluate the impact of non-native species upon community structure. In these sites non-native species were few and of low frequency and, therefore, only FQA metrics for native species are reported. RESULTS Changes in Community Composition According to NMS, the understories of the North and South Block throughout the observation period were markedly different from each other (Figure 2A). The North Block (Figure 3) is dominated by the herbs Carex pensylvanica, Pteridium aquilinum, Helianthus divaricatus, Maianthemum canadense, Potentilla simplex, and Arenaria lateriflora (Table 1). Dominant shrubs include Vaccinium angustifolium and Rosa carolina. Other characteristics species include Amphicarpaea bracteata, Rubus flagellaris, Rhus glabra, and Anemone quinquefolia. The South Block (Figure 4), on the other hand, has a higher frequency of shrubs including Vaccinium angustifolium and Comptonia perigrina and a high frequency of the herbs Carex pensylvanica, Comandra umbellata, Parthenium integrifolium and Potentilla simplex (Table 1). Other species prevalent in the South Block include Helianthus mollis, Aster dumosus, and Andropogon scoparius. Although A. scoparius and seven additional native grass species were observed within the sample plots, none were dominant species. The strong inter-Block differences in species composition (Figure 2A) were maintained throughout the 27 year observation period and had a much greater magnitude than within-Block changes (Figure 2B-C). As a result, the overall NMS plot of Figure 2A is not able to depict fine grained, year to year intra-Block changes. Nonetheless there were important underlying changes in community composition at the Block level. One measure of change, namely similarity (SSI) based upon the presence or absence of individual species, indicates that the amount of change was almost equivalent within each Block (Table 2): in 2006, the North Block had an SSI = 64% and the South Block an SSI = 69% when compared to the base year of 1979. NMS (Figure 2B-C) and SSI (Table 2) of each Block alone revealed more specific historical change. During the first three years of observation (1979-1981), both Blocks exhibited a directional change (Figure 2B-C). This directionality was also evident in coefficients of similarity (Table 2), especially for the South Block. Over the period of 1981-1983, the South Block quickly fell to the range of SSI = 68-77% while the North Block declined to an SSI = 83-89%. The stronger changes in the South Block can primarily be attributed to a decline

Page  70 ï~~70 THE MICHIGAN BOTANIST Vol. 46 70 THE MICHIGAN BOTANIST Vol. 46 A Combined North and South Block 1.0North Block North Block N 0.5 0.0 -0.5 -1.0 2 - South Block 0 B N1 t0 c: 1.0 -0.5 0.0 0.5 1.( Axis 1 North Block 1979 1991 1981 " 1982 * 0 1987 94 1982 1988198 4 1983 1992' 198501986 1990 * 01994 * * 0 2004 1993 1989 2006 -3 -2 -1 0 1 Axis 1 South Block 0 * 2006 2004 1991 1994. 1992*1993 *1981 19880 19900 e1987 1985 * * 1982 98 1989 1979 1983 1986 1980 * C (N -2 -1 0 Axis I 1 2 FIGURE 2. Nonmetric Multidimensional Scaling for two observation Blocks at Hoosier Prairie, 1979-2006. A. Combined data for the North and South Block. B. Analysis for North Block alone. C. Analysis for South Block alone.

Page  71 ï~~2007 THE MICHIGAN BOTANIST 71 2007 THE MICHIGAN BOTANIST 71 FIGURE 3. View of the North Block in August, 2006. in the frequency of the old field species Solidago altissima (from 44 to 0%) and the shrub Gaylussacia baccata (from over 50% to mostly less than 20%) and an increase in that of Carex pensylvanica (from 50 to over 80%). In addition, several species, including Aster azureus, Krigia biflora, and Panicum virgatum, were not observed within the South Block until 1981-1982. From the period of 1981 to 1994 further changes in the community composition of the two Blocks lacked any obvious pattern. On the other hand, the most recent sample dates, 2004 and 2006, detected a shift in composition. In the South Block several grass species, Andropogon gerardii and Sorghastrum nutans, as well as several forbs declined in frequency, while during this same period the frequency of Panicum implicatum, Aster umbellatus, Cuscuta coryli, and the annual Erechtites hieracifolium increased. On the North Block during the period from 1994 to 2006, there was an increase in the frequency of Desmodium glutinosum (from 12 to 76%) as well as Quercus velutina saplings (from 36 to 68%) and Comptonia peregrina (from 32 to 52%). By 2004 Panicum virgatum and Tradescantia ohiensis, which in some years before 1994 were present in more than half of the quadrats, were completely absent. Pteridium aquilinum and Helianthus divaricatus also exhibited meaningful decreases in frequency. Floristic Quality Assessment Based upon the concepts of conservatism and species richness, FQA can provide insight into whether there are changes in community quality through time. Overall the richness of native species (Figure 5A) in the South Block ranged from 41 to 58 species. This was, on average, higher than that observed in the

Page  72 ï~~72 THE MICHIGAN BOTANIST Vol. 46 TABLE 1. List of the 60 most common understory species in two 10 x 10 m blocks at Hoosier Prairie Nature Preserve: species, mean frequency, and frequency range (over the period 1979-2006). The North Block had a total of 75 species and the South Block had 96 species. NORTH BLOCK SOUTH BLOCK Frequency Frequency Range Range Mean (1979- Mean (1979 -Species Frequency 2006) Species Frequency 2006) Carex pensylvanica Pteridium aquilinum Helianthus divaricatus Maianthemum canadense Vaccinium angustifolium Rosa carolina Potentilla simplex Arenaria lateriflora Smilacina racemosa Rubus flagellaris Solidago juncea Tradescantia ohiensis Rhus glabra Amphicarpaea bracteata Anemone quinquefolia Coreopsis tripteris Coreopsis palmata Comptonia peregrina Fragaria virginiana Quercus velutina Aster azureus Panicum virgatum Parthenium integrifolium Rubus hispidus Krigia biflora Baptisia leucantha Gaylussacia baccata Poa pratensis Viola sagittata Solidago speciosa Salix humilis Desmodium canadense Euphorbia corollata Apocynum androsaemifolium Rudbeckia hirta Aster umbellatus Galium triflorum Andropogon gerardii Muhlenbergia mexicana Poa compressa Desmodium glutinosum Solidago altissima Asclepias syriaca Polygonatum canaliculatum Sorghastrum nutans Viola fimbriatula Achillea millefolium Cirsium discolor 98.4% 88-100% Vaccinium angustifolium 85.8% 60-96% 88.5 86.4 78.7 76.0 74.4 66.1 65.3 55.2 53.9 48.3 39.5 36.8 35.2 32.0 28.8 25.9 24.0 23.7 22.4 22.1 22.1 19.7 19.7 19.5 14.1 11.7 11.2 11.2 8.3 8.3 7.5 7.2 6.4 6.1 5.3 5.3 4.8 3.5 3.2 2.9 2.9 2.1 2.1 2.1 2.1 1.9 1.9 56-100 60-96 56-96 52-92 64-92 44-80 0-88 24-80 4-88 12-92 0-60 4-56 12-68 4-56 12-48 20-32 4-52 4-36 8-40 0-36 0-64 12-28 0-76 0-40 4-24 4-20 0-60 0-48 0-24 0-16 0-24 0-20 0-12 0-24 0-12 0-96 0-20 0-12 0-20 0-76 0-16 0-32 0-16 0-24 0-20 0-4 0-8 Comptonia peregrina 81.3 Carex pensylvanica 81.0 Comandra umbellata 67.3 Parthenium integrifolium 64.5 Potentilla simplex 64.3 Tradescantia ohiensis 57.8 Rubus setosus 55.8 Pteridium aquilinum 50.5 Salix humilis 45.8 Solidago juncea 45.5 Rubus hispidus 42.5 Helianthus mollis 36.8 Aster umbellatus 35.3 Fragaria virginiana 34.3 Andropogon scoparius 31.0 Sorghastrum nutans 28.3 Aronia prunifolia 28.0 Viola sagittata 25.5 Solidago gymnospermoides 24.3 Gaylussacia baccata 23.8 Panicum virgatum 21.0 Smilacina racemosa 17.8 Krigia biflora 13.8 Aster dumosus 13.5 Euphorbia corollata 12.5 Dryopteris thelypteris 11.3 Lespedeza capitata 11.3 Andropogon gerardii 10.8 Phlox pilosa 10.8 Populus tremuloides 10.3 Aster azureus 10.0 Coreopsis tripteris 10.0 Spiraea tomentosa 9.6 Baptisia leucantha 8.5 Panicum villosissimum 8.3 Arenaria lateriflora 8.0 Quercus velutina 7.8 Houstonia caerulea 7.5 Spartina pectinata 6.8 Quercus alba 6.7 Maianthemum canadense 6.5 Lycopus uniflorus 6.0 Liatris spicata 5.3 Solidago altissima 5.3 Rudbeckia hirta 5.0 Calamagrostis canadensis 4.5 Poa pratensis 4.5 60-96 64-92 12-96 48-84 36-80 4-72 16-68 28-72 12-72 4-72 32-60 4-68 0-64 4-56 0-76 8-60 16-40 0-44 0-32 12-52 0-80 4-28 0-32 0-32 0-24 0-20 0-28 0-80 0-20 0-28 0-28 0-20 0-24 1-20 0-20 0-40 0-16 0-60 0-20 0-20 0-16 0-12 0-20 0-44 0-28 0-40 0-24

Page  73 ï~~2007 THE MICHIGAN BOTANIST 73 TABLE 1. (Continued). NORTH BLOCK SOUTH BLOCK Frequency Frequency Range Range Mean (1979- Mean (1979 -Species Frequency 2006) Species Frequency 2006) Liatris aspera 1.9% 0-12% Eryngium yuccifolium 4.3% 0-12% Liatris spicata 1.9 0-12 Lechea tenuifolia 3.8 0-16 Lespedeza capitata 1.3 0-4 Tephrosia virginiana 3.8 0-8 Panicum depauperatum 1.3 0-4 Desmodium canadense 3.5 0-20 Rhamnusfrangula 1.3 0-20 Helianthus rigidus 2.8 0-28 Prenanthes racemosa 1.1 0-4 Panicum implicatum 2.8 0-48 Prunus serotina 1.1 0-8 Apocynum Comandra umbellata 0.8 0-4 androsaemifolium 2.5 0-16 Solidago graminifolia var. Cuscuta coryli 2.5 0-60 nuttallii 0.8 0-8 Helianthus divaricatus 2.5 0-16 Sonchus uliginosus 0.8 0-4 Polygala sanguinea 2.5 0-12 Agrostis alba 0.5 0-4 Pycnanthemum virginianum 2.5 0-8 Andropogon scoparius 0.5 0-4 Salix petiolaris 2.3 0-16 North Block (which ranged from 34 to 48 species per sample date). During the observation period, according to regression analysis, the richness of native species increased (r2 = 0.63, p = 0.0001). The South Block, by contrast, had more variation in species richness (r2 = 0.15, p = 0.12). It experienced a slight increase in species richness through 1994, followed by an unexpected low value in 2004. However, in 2006 species richness returned to the higher levels observed in 1993-4. We attribute this fluctuation to the almost 5-year lapse time in prescribed burn (from 2000 to 2005) followed by a post-burn rejuvenation (in 2006). Throughout the observation period the two Blocks were remarkably free of non-native species. Six species were present in the North Block and three in the South Block. The most frequent non-native species in both locations was Poa pratensis. During most years its frequency was well below 20% in each Block and in 2006 it was only encountered in one quadrat. A second exotic species, Rhamnusfrangula, is the object of ongoing management to prevent invasion. In 2006 it was sighted in two quadrats within each Block. Because non-native species have a very limited presence in the monitored areas, the remaining FQA analysis results are based upon native species only. As already described above, the overall species richness of the more shaded North Block was lower than that of the South Block. The same trend applied when species richness is analyzed at the quadrat level (Figure 5A). In peak year of 1994, the Hoosier Prairie savanna supported a mean of 15.3 (North Block) to 16.8 (South Block) species per quadrat. The years with fewest species per quadrat occurred in 1979 (North Block) and 1981 (South Block). Regression analysis suggests that quadrat-level species richness in the two Blocks has not changed meaningfully over the 27 year period. As with species richness, mean C values (both overall as well as at the quadrat level; Figure 5B) were higher in the South Block. Over the 26 years,

Page  74 ï~~74 THE MICHIGAN BOTANIST Vol. 46 74 THE MICHIGAN BOTANIST Vol. 46 FIGURE 4. View of the South Block in August, 2006. mean C (overall) for the South Block ranged narrowly between 5.7 and 6.3. Those of the North Block ranged from 4.8 to 5.6 with the lowest mean occurring in 2004. As one would expect, FQI values for the South Block were higher than those for the North (Figure 5C). Across the two Blocks, FQI values ranged from 31 to 47 with those in the North Block increasing with time (r2 = 0.31, p = 0.02). DISCUSSION When dedicated as a state nature preserve in 1977, Hoosier Prairie supported a variety of savanna, prairie, and wet habitat of varying degrees of natural quality (Bacone 1977). The two Blocks used in this study were established in areas where the understory supported a diverse mosaic of xeric to mesic species. The changes in species composition were traced annually over approximately 16 years and then twice after a decade long gap. This more fine grain record is in contrast to typical long term studies of community change in prairie or savanna communities where observations years were few and widely spaced (White 1983; Bowles et al. 2003; Dornbush 2004) or where data were only collected at

Page  75 ï~~2007 THE MICHIGAN BOTANIST 75 TABLE 2. Sorenson's Similarity Index (SSI) within each Block for selected years compared to 1979, the first year of monitoring. The first prescribed burns occurred during spring (North Block) and winter (South Block) of 1978 and again in the winter of 1979. Year North Block South Block 1981 cp. 1979 0.89 0.77 1982 cp. 1979 0.83 0.71 1983 cp. 1979 0.85 0.68 1984 cp. 1979 0.87 0.72 1985 cp. 1979 0.83 0.70 1986 cp. 1979 0.78 0.70 1988 cp. 1979 0.75 0.75 1990 cp. 1979 0.77 0.72 1992 cp. 1979 0.72 0.74 1994 cp. 1979 0.72 0.71 2004 cp. 1979 0.70 0.70 2006 cp. 1979 0.64 0.69 the terminus of the experiment (Tester 1994). One truly longitudinal study on southern Illinois barrens (Anderson et al. 2000) included ten observation years; however, during most of the study, fire was being excluded from the site. Our comparisons on Hoosier Prairie of two Blocks indicated that each sample area had a unique community structure at the outset and that their uniqueness persisted throughout the 27 year period of fire management. The uniqueness of each Block was undoubtedly the result of both long term differences in local site history (Sousa 1984; Lesica and Cooper 1999) as well as on-going differences in microhabitat (Zedler and Loucks 1969; Beatty 1984), which can be more pronounced in oak savanna ecosystems (Leach and Givnish 1999). In terms of readily observable differences between the two Blocks, the North Block (Figure 3) was more shaded and perhaps slightly higher topographically than the South Block (Figure 4). In addition to the strong and persistent differences between the two Blocks, we observed gradual intra-Block change over time. In the South Block, the early cycles of prescribed fire were associated with the decline of the old field species Solidago altissima and the shrub Gaylussacia baccata (Matlack et al. 1993) and the gradual appearance of several herbaceous species. The North Block, after experiencing relatively small amounts of change in community composition during the first 16 years, ended in 2006 with as much change in community composition (as measured by SSI) as the South Block. And so, not only does the Preserve harbor a diverse understory mosaic as represented by the two sample Blocks, but over time these exhibited individualistic behavior even when exposed to similar management regimes. Although this study documents long term changes during a program of prescribed burn, it has several limitations. Pre-burn data (i.e., 1978 or earlier), that would allow tracing the initial community changes with fire, are limited or lacking. Furthermore, one might expect that factors aside from fire, including drought and herbivory (Inouye et al 1994; Ritchie and Tilman 1995; Ritchie et al. 1998), contributed to the fluctuation in community structure during our monitoring period. For example, the classic study by Weaver et al. (1940) demonstrated

Page  76 ï~~76 THE MICHIGAN BOTANIST Vol. 46 76 THE MICHIGAN BOTANIST Vol. 46 A so 6 40 50 0 B,0 C 5 Z 0 040 0 > 0 (U Z 20 North Block South Block 0 0* overall I - overall per quadrat per quadrat 0 0 O 0 SW W SW S SW W S W WW W SS SS S S S S S S sw w sw s sw w s w w w 011s ss ".* co 0 4 3 2 50 C 40 LL. 30 20 1980 1985 1990 1995 2000 2005 Year 1980 1985 1990 1995 2000 2005 Year FIGURE 5. Changes in Floristic Assessment for two observation Blocks at Hoosier Prairie, 1979-2006. A. Number of native species observed. B. Mean C for native species. C. Floristic Quality Index (FQI) for native species. Closed circles are overall values for an entire Block. Open circles are per quadrat averages within each Block. Figure B indicates years with prescribed burn: w = early winter burn and s = late winter to early spring burn. No data were collected in the interval 1995-2003.

Page  77 ï~~2007 THE MICHIGAN BOTANIST 77 that in periods of drought there may be a differential response among guilds such as C4 grass, annuals, and C3 forbs. These fluctuations may be particularly pronounced in microsites where herbaceous vegetative must compete with mature trees for moisture (Anderson et al. 2001). Increasingly Hoosier Prairie is an island in an urban/suburban landscape (Figure 1). In spite of this reality, it has been possible to simulate, by means of prescribed bums, one of its significant historical rhythms. The sample areas in Hoosier Prairie have continuously supported high species richness, one comparable on a quadrat basis to that reported by Leach and Givnish (1999) for Wisconsin oak savannas. In fact, as emphasized by Leach and Givnish (1999), oak savannas should be more widely recognized for their rich diversity and cover of forbs rather than grasses, an essential floristic difference from tall grass prairies that holds true for Hoosier Prairie Nature Preserve (see cover photo). During the past 27 years of fire management the sample Blocks maintained high species richness; each 100 m2 Block contained in excess of 30 plant species. The North Block experienced an increase in overall species richness and, on a per quadrat basis, the richness of both Blocks usually only had small fluctuations. Of some concern, however, was the decline in 2004 of overall species richness in the South Block. We attributed this temporary decline in richness to an almost 6 year bumrn hiatus in this portion of the savanna (Anderson et al. 2000; Bowles et al 2003). Unfortunately, data are not available to detail the decline over that period. However, it was clear from the 2006 sampling that the diversity had indeed returned. During 2006 our qualitative field observation was that recently burned quadrats (i.e., South Block) contained numerous seedlings, a condition not apparent in 2004. At the same time, seedlings were absent in quadrats where 4 years had lapsed since burn (i.e., North Block). Based upon measures of mean C, fire management was successful in maintaining an assemblage of species strongly characteristic of what is considered pre-settlement condition. Each Block typically had mean C values above 5.0. These values fall at or near the expected upper limit for an undisturbed presettlement community (Swink and Wilhelm 1994). What will be interesting going forward is whether the Preserve can continue to achieve these remarkable quality measures in decades to come (Kellman 1996). As an island in an urban/suburban setting it does not support the historical diversity of herbivores and carnivores, it has small sized meta-populations for many plant species, and, of course, a different physical environment stemming from surrounding changes in land use (Saunders et al. 1991). As an island it is particularly dependent upon the continued cooperation of local and regional citizens to protect it against undo human disturbance and to support the on-going efforts to manage its historical oak savanna condition. ACKNOWLEDGEMENTS The authors would like to thank all the folks who assisted with plot sampling over the term of the study, including Jim Keith, Irene Herlocker, Ron Campbell, Ken Klick, Laura Rericha, Deb Petro, Drew Daily, Linda Wetstein, and John Ervin. Also, the authors would like to thank the staff from the Division of Nature Preserves and from the Division of Forestry's Fire Headquarters for carrying out the prescribed burning program over the years, Alice Heikens for her helpful insights into the oak sa

Page  78 ï~~78 THE MICHIGAN BOTANIST Vol. 46 vanna ecosystem, and Mary Damm for her critique of an earlier draft of this paper. Ron Hellmich supplied the site map of the Preserve. LITERATURE CITED Abella, S.R., J.F. Jaeger, and L.G. Brewer. (2004). Fifteen years of plant community dynamics during a northwestern Ohio oak savanna restoration. The Michigan Botanist 43:117-127. Anderson, D.M. (1982). Plant Communities of Ohio: A Preliminary Classification and Description. Division of Natural Areas and Preserves. Ohio Department of Natural Resources. Columbus, Ohio. 183p. Anderson, L.J., M.S Brumbaugh, and R.B. Jackson. (2001). Water and tree-understory interactions: A natural experiment in a savanna with oak wilt. Ecology 82: 33-49. Anderson, R.C. (1990). The historic role of fire in the North American grassland, pp. 8-18, in S. L. Collins and L.L. Wallace (eds.). Fire in North American Tallgrass Prairies. University of Oklahoma Press, Norman. Anderson, R.C. and M.L. Bowles. (1999). Deep-soil savannas and barrens of the Midwestern United States, pp. 155-170, in R.C. Anderson, J.S. Fralish, and J.M. Baskin (eds.). Savannas, Barrens, and Rock Outcrop Plant Communities of North America. Cambridge University Press. Anderson, R.C. and L.E. Brown. (1983). Comparative effects of fire on trees in a Midwestern savannah and an adjacent forest. Bulletin of the Torrey Botanical Club 110:87-90. Anderson, R.C. and L.E. Brown. (1986). Stability and instability in plant communities following fire. American Journal of Botany 73:364-368. Anderson, R.C. and J.E. Schwegman. (1991). Twenty years of vegetational change on a southern Illinois barren. Natural Areas Journal 11:100-107. Anderson, R.C., J.E. Schwegman, and M.R. Anderson. (2000). Micro-scale restoration: A 25-year history of a southern Illinois barrens. Restoration Ecology 8:296-306. Bacone, J.A. (1977). The Hoosier Prairie: A discussion of natural communities, disturbances, management problems, and recommendations. Internal report from Division of Nature Preserves. 18p. Beatty, S. (1984). Influence of microtopography and canopy species on spatial patterns of forest understory plants. Ecology 65:1406-1419. Bowles, M.L., M.D. Jones, and J.L. McBride. (2003). Twenty-year changes in burned and unburned sand prairie remnants in northwestern Illinois and implications for management. American Midland Naturalist 149:35-45 Cottam, G. (1949). The phytosociology of an oak woods in southwest Wisconsin. Ecology 30:271-287. Curtis, J.T. (1959). The Vegetation of Wisconsin: An Ordination of Plant Communities. University of Wisconsin Press, Madison. 657p. Dornbush, M.E. (2004). Plant community change following fifty-years of management at Kalsow Prairie Preserve, Iowa, U.S.A. American Midland Naturalist 151:241-250. 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 93:93-99. Hartke, E.J., J.R. Hill, and M. Reshkin. (1975). Environmental geology of Lake and Porter Counties, Indiana-an aid to planning. Indiana Geological Survey, Special Report 11. 80 p. Heikens, A.L. and P.A. Robertson. (1994). Barrens of the Midwest: A review of the literature. Castanea 59:184-194. 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. Howe, H.F. (1994). Response of early- and late-flowering plants to fire season in experimental prairies. Ecological Applications 4:121-133. Howe, H.F. (1995). Succession and fire season in experimental prairie planting. Ecology 76: 1917-1925. Kellman, M. (1996). Redefining roles: Plant community reorganization and species preservation in fragmented systems. Global Ecology and Biogeography Letters 5:111-116. King, R. (2000. Effects of single burn events on degraded oak savanna. Ecological Restoration 18:228-233. Inouye, R.S., T.D. Allison, and N.C. Johnson. (1994). Old-field succession on a Minnesota sand

Page  79 ï~~2007 THE MICHIGAN BOTANIST 79 plain: Effects of deer and other factors on invasion by trees. Bulletin of the Torrey Botanical Club 121: 266-276. Leach, M.K. and T.J. Givnish. (1999). Gradients in the composition, structure, and diversity of remnant oak savannas in southern Wisconsin. Ecological Monographs 69:353-374. Leach, M.K. and L. Ross. (1995). Oak Ecosystem Recovery Plan: A Call to Action. United States Environmental Protection Agency, Great Lakes National Program Office, Chicago, Illinois. 11ip. Lesica, P. and S.V. Cooper. (1999). Succession and disturbance in sandhills vegetation: Constructing models for managing biological diversity. Conservation Biology 13:293-302. Loope, W.L. and J.B. Anderson. (1998). Human versus lightning ignition of presettlement surfaces fires in coastal pine forests of the upper Great Lakes. American Midland Naturalist 140: 206-218. Matlack, G.R., D.J. Gibson, and R.E. Good. (1993). Clonal propagation, local disturbance, and the structure of vegetation: Ericaceous shrubs in the Pine Barrens of New Jersey. Biological Conservation 63:1-8. McClain, W.E., M.A. Jenkins, S.E. Jenkins, J.E. Ebinger. (1993). Changes in the woody vegetation of a bur oak savanna remnant in central Illinois. Natural Areas Journal 13:108-114. McCune, B. and M.J. Mefford. (2006). PC-ORD. Multivariate analysis of ecological data. Version 5. MjM Software, Gleneden Beach, OR. National Resource Conservation Service (NRCS). (2006). Web Soil Survey 1.1. http://websoilsurvey.nrcs.usda.gov/app/WebSoilSurvey.aspx Nuzzo, V.A. (1986). Extent and status of Midwest oak savanna: Presettlement and 1985. Natural Areas Journal 6:6-36. Rebertus, A.J. and B.R. Burns (1997). The importance of gap processes in the development and maintenance of oak savannas and dry forest. Journal of Ecology 85:635-645. Ritchie, M.E. and D.Tilman. (1995). Responses of legumes to herbivores and nutrients during succession on a nitrogen-poor soil. Ecology 76: 2648-2655. Ritchie, M.E., D. Tilman, and J.M.H. Knops. (1998). Herbivore effects on plant and nitrogen dynamics in oak savanna. Ecology 79: 165-177. Robertson, P.A. and A.L. Heikens. (1994). Fire frequency in oak-hickory forests of southern Illinois. Castanea 59:286-291. Saunders, D.A., R.J. Hobbs and C.R. Margules. (1991). Biological consequences of ecosystem fragmentation: A review. Conservation Biology 5:18-32. Sousa, W.P. (1984). The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15:353-391. Swink, F. & G. Wilhelm. (1994). Plants of the Chicago Region. 4th edition. Indiana Academy of Science, Indianapolis, Indiana. 921p. Taylor, R.S. (1990). Reconstruction of Twentieth Century Fire Histories in Black Oak Savannas of the Indiana Dunes National Lakeshore. Masters Thesis, University of Wisconsin - Madison. Tester, J.R. (1989). Effects of fire frequency on oak savanna in east-central Minnesota. Bulletin of the Torrey Botanical Club 116:134-144. Transeau, E. (1935). The prairie peninsula. Ecology 16:423-437. Weaver, J.E., J.H. Robertson, and R.L. Fowler. (1940). Changes in true-prairie vegetation during drought as determined by list quadrats. Ecology 21: 357-362. White, AS. (1983). The effects of thirteen years of annual prescribed burning on a Quercus ellipsoidalis community in Minnesota. Ecology 64:1081-1085. Zedler, J. and O.L. Loucks. (1969). Differential burning response of Poa pratensis fields and Andropogon scoparius prairies in central Wisconsin. American Midland Naturalist 81:341-352.