Page  3 ï~~2000 THE MICHIGAN BOTANIST FIRE HISTORY OF THE BUR OAK SAVANNAS OF SHEGUIANDAH TOWNSHIP, MANITOULIN ISLAND, ONTARIO Judith Jones Winter Spider Eco-Consulting, R.R. #1 Sheguiandah, Ontario POP 1WO Canada ABSTRACT The fire history of the bur oak savannas of Manitoulin Island, a globally rare vegetation community, was examined in Sheguiandah Township from the notes of two early land surveys (1864 and 1866) done before and after a catastrophic and widespread fire in 1865. Current locations of bur oak savanna were also surveyed and documented. Most current savannas match up with sites that were hardwood forest, at least in part, before the 1865 fire. All current savannas burned in 1865. A few current savannas already existed and some had burned, prior to the 1865 fire. No correlation was found between burn intensity and savanna creation. The fire of 1865 is shown to have maintained some savannas and probably to have created most others; therefore, fire is suggested as a management tool to maintain the openness of this vegetation type. Key Words: alvar, vegetation history, early land surveys, Manitoulin Island, savanna creation INTRODUCTION The bur oak savannas of Manitoulin Island, Ontario, Canada (45Â~050'N, 81 Â~55'W) are a globally and provincially rare vegetation community (Reschke et al., 1999; Bakowsky, 1997) tentatively ranked G1?S1S2 (Bakowsky, pers. comm. 2000). They feature scattered trees of Quercus macrocarpa Michx. (bur oak) over-topping a grass- or shrub-dominated groundlayer (Figure 1). The community differs from typical oak savannas, found further south and west, in having extremely shallow soils (0-20 cm) over horizontal limestone or dolostone bedrock, and in suffering seasonal drought stress during July and August. Typical bur oak savannas on Manitoulin Island also differ from those elsewhere in having an alvar-like species composition, and many in fact consider these savannas to be a type of alvar (Bakowsky, pers. comm., 2000; Brownell, 2000). Shrubs such as Viburnum rafinesquianum Schultes, Amelanchier spicata (Lam.) K. Koch, Rhus aromatica Aiton, and Symphoricarpos albus (L.) S. F. Blake are typical, while the herbaceous layer is usually dominated by Danthonia spicata (L.) P. Beauv., with Cerastium arvense L., Geum triflorum Pursh, Ranunculusfascicularis Muhl., and Senecio pauperculus Michx. common. Uncommon species of the Sheguiandah savannas which show the vegetation's prairie affinities include Astragalus neglectus (Torr. & A. Gray) Sheldon, Penstemon hirsutus (L.) Willd., Potentilla arguta Pursh, and Coreopsis lanceolata L. In addition, the moth Zale calycanthata is known in Canada only from bur oak sites on Manitoulin Island (J.K. Morton, pers. comm., 2000).

Page  4 ï~~THE MICHIGAN BOTANIST Vol. 39 THE MICHIGAN BOTANIST Vol. 39 FIGURE 1. An example of bur oak savanna in Sheguiandah Township. At this site, common juniper is the dominant understory shrub. At other sites this stratum may be dominated by other species, especially Viburnum rafinesquianum or Amelanchier spicata. All of these oak savanna sites are presently used as livestock pasture, or have been used as such in the recent past. Grazing introduces weed species and soil disturbance, and causes a slow degradation of the diversity of native vascular plant species in the community (Reschke et al. 1999; Jones, unpublished data). With the recent recognition of the rarity of this vegetation type, protection and management of these bur oak savannas will be issues of concern in the near future. Fire is known to be involved in the maintenance of bur oak savannas elsewhere (Tester, 1996); therefore, the fire history of the Manitoulin bur oak savannas was investigated to see whether fire was involved in their origin and whether controlled burning should be looked into as a possible management tool. Sheguiandah Township, on Manitoulin Island (Figure 2), is uniquely placed for such a study. The township was originally surveyed in 1863-64 (McPhillips, 1864) to lay out the location of lots and roads before the island was opened to settlers. In 1865 a catastrophic fire burned most of the township, destroying all but eighteen of the survey markers; consequently, the township had to be surveyed again. A second survey was done in 1866 (O'Keefe, 1866). As the surveyors cleared the lines they were laying out, they noted facets of the land-usually dominant tree species, sometimes with the addition of soil or landform characteristics-along accurately measured sections of the surveyed lines. The information was intended to help settlers choose property site-unseen and to help in recognition of the property upon arrival. Consequently, sufficient

Page  5 ï~~2000 THE MICHIGAN BOTANIST 2000 THE MICHIGAN BOTANIST Manitoulin Island, Ontario FIGURE 2. Location of Sheguiandah Township on Manitoulin Island, and location of Manitoulin Island (inset) in the Great Lakes region. information exists from this survey grid to construct vegetation maps of Sheguiandah Township before and after the catastrophic fire. METHODS Current occurrences of bur oak savanna were field surveyed in Sheguiandah Township in 1997 (Figure 3). The vegetation of the township before the fire (and prior to settlement) was mapped from notes of the survey of 1863-64 (Figure 4). The post-fire vegetation was mapped based on the notes of the survey of 1866 (Figure 5). This type of mapping process has some inherent error since the surveyors did not see the interior of each lot. In this study it was assumed that the recorded vegetation extended to the back of each lot, unless the surveyor indicated otherwise or topography such as a cliff precluded it. Double or heavy lines on the maps show the actual paths the surveyors walked. Accuracy of interpretation is highest where double lines intersect, and lowest in the interior of the rectangles their paths form. Accuracy is higher in open areas, such as those created by burning, because the surveyors' line of sight extended farther. Comparisons were made to see how many current bur oak sites burned in 1865 and what type of vegetation they were prior to the fire. In addition, the diaries and letters of the surveyors were examined for any further clues about oak savanna origins. The intensity of burning at each site was interpreted from the notes. O'Keefe frequently uses the term "clear burning" to indicate a lot that had burned so completely that homesteaders could consider it cleared. For the purpose of this study, clear burning was assumed to indicate high intensity burning because it shows much of the timber was consumed by the fire. The surveyor describes many other sites simply as "burnt," "burnt woods," "windfall, burnt," etc. These comments were interpreted to mean an intermediate (and typical) level of burning. Lot descriptions that have no direct mention of burning, but where O'Keefe notes the corner posts or bearing trees burned down, are interpreted to be areas of lighter burning (see dots on Figure 4). Finally, cross sections ("cookies") were taken from stumps of 5 bur oak trees recently cut by landowners for firewood at one site. Tree ages were determined from growth ring-counts to try to estimate the dates of savanna establishment.

Page  6 ï~~6 THE MICHIGAN BOTANIST Vol. 39 I1N 1000 m anitowaning Bay FIGURE 3. Locations of bur oak savanna study sites field surveyed in 1997.

Page  7 ï~~2000 THE MICHIGAN BOTANIST 2000THE MICHIGAN BOTANIST Bass Lake Sheguiandah Bay ~1 1000 m Bay.Hardwoods 1 Windfall Burned FIGURE 4. Pre-fire distribution of hardwoods, windfall and burning based on the survey of 1863-64 (McPhillips, 1864) compared with current bur oak savanna study sites. Rectangles shown are hundred acre lots. Double lines are road allowances walked by the surveyors.

Page  8 ï~~THE MICHIGAN BOTANIST Vol. 39 THE MICHIGAN BOTANIST Vol. 39 Bass Lake Sheguiandah Bay Turtle LEI Pine Lake 1000 m " "... T i Bay Hardwoods. 1864 Windfall Burned * Post and/or bearing tree burned FIGURE 5. Location of windfall and burning in 1866 (O'Keefe, 1866), superimposed on 1863-64 hardwood distribution and current bur oak savanna study sites. Dots indicate corners where the surveyor specifically notes that posts or bearing trees burned. The lack of a dot does not indication a lack of burning since the surveyor did not always indicate if he found the old post.

Page  9 ï~~2000 THE MICHIGAN BOTANIST RESULTS Table 1 gives a comparison of the descriptions of current oak savanna locations before and after the fire of 1865 (italics are direct quotes). Table 2 summarizes the comparison. Comparisons of the current locations of savannas to pre-fire vegetation show that most bur oak savannas were hardwood stands, at least in part, before the fire of 1865. Seventeen of 27 sites were listed as hardwoods, at least in part, and an additional 2 sites included maple or basswood (hardwood species) in the lot description. Three sites were windfall only, and while there is no indication of species here, the first surveyor McPhillips (1864), notes in his cover letter to his superior that "Some windfalls were originally hardwood but the young timber growing up is now poplar, birch, tamarack and spruce." On the other hand, before the fire some 4000 acres of the township were described as windfall, and it is unlikely that all of it was formerly hardwoods. A further 4 sites (nos. 4, 5, 6 and 12) had no indication of hardwoods at all but were listed as swamps or mixed woods. Bearing tree notations and lists of dominant species show "hardwoods" were composed of maple, basswood, red oak, ironwood, "oak" (not identified to species), and elm in various combinations. The occasional cedar or balsam fir was also sometimes present, but usually when these species had a significant presence the surveyors recorded a species list such as "cedar balsam maple." Pine is rarely mentioned in connection with the term hardwoods, although it is frequently listed with other coniferous species. Thus, "hardwoods" indicates a forest composed almost entirely of deciduous trees. It is possible bur oak was present, but it is never mentioned directly. Some bur oak savanna existed prior to the fire of 1865. One site was listed in the first survey as "stunted oaks," while a second site, listed as "a few trees of hardwood," was also probably a savanna. In addition, in his cover letter McPhillips describes a rocky area at the top of a steep slope (a cliff) with shallow soil and "a girdle of small stunted hardwoods running the same direction as the edge of the rocks." This perfectly describes at least two of the bur oak savannas that exist today which exist as narrow bands along the edge of cliffs. As expected, all locations of current bur oak savannas burned to some degree in the 1865 fire. However, current savanna occurrence does not correlate with any particular level of burning. Ages determined from growth-ring counts (made in 1996) on bur oaks varied widely, ranging from 71-115 years. These are minimum ages since the counts were done above the base of the tree. The ages put establishment of the oldest tree sampled somewhere around 16 years after the fire. More interestingly, increasing diameter of trees did not correlate with age, showing that very local stresses which can inhibit growth occur in these shallowsoil sites. Therefore, size does not make it immediately apparent which trees are the oldest.

Page  10 ï~~10 THE MICHIGAN BOTANIST Vol. 39 TABLE 1. 1864 and 1866 descriptions of current bur oak savanna locations (direct quotes are in italics). Site # and Location 1863-64 description 1866 description Interpretation Sitel Lot 11, Con. 12 Hardwoods-no mention of windfall; Site 2 Lots 4+5, Con. 11 Lots 26+27, Con. B Site 3 Lot 2, Con. 11 Hardwoods in rear Lot 5 and junction Lots 26-27 Swamp in some parts. Swamp-hardwoods at western edge Windfall, hardwoods Light burning assumed from new windfall and clear burning on adjacent lot. Windfall. Hardwood Light burning: posts still standing Lots burned. 26-27. Burnt windfall, swamp Area is on plateau at edge of steep slope. Burnt windfall, swamp Currently rocky area in large, open pasture Burnt windfall, swamp Currently rocky area in large, open pasture Site 4 Lot 2, Con. 10-11 Swamp Site 5 Lot 3, Con. 10 Site 6 Lots 22-23, Con. A-B Site 7 Lots 21-22, Con. B Site 8 Lots 3-4, Con. 9 Swamp Tamarack, poplar stunted pine; cedar, black ash, balsam Hardwoods on Lot 22; Hardwoods in part; rocky outcrops Site 9 Lots 18-20, Con. B Hardwoods on Lot 20 Scattered windfall, partly burnt Windfall, burnt; some hardwoods remained Burnt woods, limestone rock, mixed forest. Burnt woods, windfall; part maple, basswood, ironwood, birch Good hardwood Good hardwood land Scattered windfall Does not appear to have been hardwoods. Site 10 Lots 4-5, Con. 7 Site 11 Lot 4-5, Con. 6 Site 12 Lot 11, Cons. 8 -Site 13 Lot 11, Con. 7 Lots 16-17, Con. A Site 14 Lots 14-15, Cons. 6-7 Site 15 Lots 16-17, Cons. 6-7 Site 16 Lot 11, Cons. 5 -Lots 11, 13, 14, 15 Con A Good Hardwood, Low land near lake Good Hardwood Swamp; cedar, poplar, birch, balsam Good hardwood, rolling Good land, woods* land, limestone covered Bearing trees were with a few inches of maple. Earth; burned windfall Second surveyor indicates more hardwoods here than first Light burning: posts all burned Light burning: posts burned, bearing trees lived Light burning, posts burned. Current site is small. Light burning: posts burned, bearing trees lived Intense burning Intense burning Large, heterogeneous area. Burn intensity varies A few trees of hardwood (prob. savanna). Windfall in part. Windfall, burned; Small brush 6 Most was windfall; Hardwoods in part; Some small brush (may mean past burning) Clear burning, good land*. Clear burning, good land* Woods and scattered windfall; clear burning; not much surface, little soil (Continued)

Page  11 ï~~2000 THE MICHIGAN BOTANIST 11 TABLE 1. Continued Site # and Location 1863-64 description 1866 description Interpretation Site 17 Lot 14 Cons. 5-6 Hardwood and mixed Clear burning, good Current site is small forest land Site 18 Lot 15-16, Con. 5 Stunted cedar small Clear burning Current site is small Site 19 Lotsl12-13 Con. 5 Lots 11-13 Con. 4 Lot 11, Con. 3 Site 20 Lots 8-9, Cons A-B pine, balsam; (stunted oaks =savanna? adjacent) Windfall and small brush (shows past burning?) Dry cedar, maple, small pine. Bearing trees were maple Clear burning, burnt Current site is large windfall area Clear burning Site 21 Lots 12-13, Con. High hardwood on 3 Lots 12-14, Con. 2 half of site Site 22 Lots 15-16 Con. 3 Lots 14-16, Con. 2 Site 23 Lots 16-17, Con. 2 Site 24 Lot 17, Con. 2 Site 25 Lots 20-21, Con. 8 Site 26 Lots 20-21, Con. 10 Site 27 Lot 24, Cons. 10-11 Strip of hardwood in mixed forest; small hardwood timber, rocky land (prob. savanna); hardwoods-east side Hardwoods; maple basswood, cedar, balsam Hardwoods; maple, basswood, cedar, balsam Windfall Mixed forest: basswoodindicates some hardwood present Hardwoods; maple, basswood and poplar Scattered windfall. Good hardwood land left in part. Some clear burning Strip of hardwood; rocky land; windfall. Posts and bearing trees burnt; good hardwoods remained at east side Hardwoods; maple, basswood, cedar, balsam, windfall, rocky Hardwoods; maple, basswood, cedar, balsam; all woods Clear burning All very bad windfall; cedar, birch, basswood Very bad windfall; maple, basswood and poplar Current site is large area. Burn intensity varies Current site is large area. Burn intensity varies Light burning: posts burned, bearing trees lived Light burning: posts burned, bearing trees lived Strip of hardwood at edge of field Light burning: posts and bearing trees burned Light burning: posts and bearing trees burned. Strip of hardwoods along edge of cliff. *Usually meant either usable or marketable timber or cleared land desirable for homesteading. DISCUSSION Some lots were hardwoods only in part prior to the fire. Yet, it is clear that the presence of hardwood is what is important. Non-hardwood vegetation existed extensively throughout the township, yet savanna formed at only 4 locations where no hardwood was listed (sites 4, 5, 6, and 12). Furthermore, bur oak is an occasional component of some hardwood forests on Manitoulin Island today

Page  12 ï~~12 THE MICHIGAN BOTANIST Vol. 39 TABLE 2. A summary of the changes in vegetation at current bur oak savanna sites, before and after the fire of 1865. 1864 description; of 27 current bur oak savanna sites: * 19 were hardwoods in part (3 had trees scattered, small or in a band) * 3 were windfall* * 3 were swamp* * 1 was stunted oaks * 1 was mixed forest of cedar, tamarack, poplar, and balsam 1866 description; of 27 current bur oak savanna sites: * 27 burned. Intensity varied. * 15 had some windfall after burning * 10 received light intensity burning * 7 received intermediate intensity burning * 7 received high intensity "clear burning" * 3 are large areas where burn intensity varied. *some of these originally may have been patches of hardwoods. See below and discussion. (personal observation), and this was probably also true in 1865. Bur oak has long been known as a very fire-resistant tree (for example Fowells, 1965), so it is likely some bur oaks from hardwood stands survived the fire. Of the 4 non-hardwood locations, three sites (numbers 4, 5, and 12) were listed as swamp, and today are only small patches of oak savanna. Two factors show these "swamp" sites could have been small patches of hardwood vegetation within the swamp, or at least something other than swamp. First, McPhillips's (1864) cover letter to his superior mentions "hardwood and swamp in such small patches that I could not mark them on the plan." Second, these three sites today are small islands of bur oak vegetation in (or at the edge of) large, open pastures, indicating small patches of different conditions exist within the area. Only site 6 lacks a reasonable explanation for its savanna, but it is possible that hardwoods may have occurred in the interior of the site where there is little survey information. There are many places in the township where hardwood stands burned that do not currently support savanna vegetation. There could be numerous reasons for this, but so far these sites have not been studied. In addition, many of the best working farms in the area occupy some of these locations, so in some cases it is impossible to know whether savanna vegetation occurred there at one time. It is obvious that burning occurred prior to the major fire of 1865. McPhillips discussed soils in his cover letter, and remarked "Being so often run over by the fire the good soil is burnt off." As well, he describes several locations, including some where savanna occurs today such as site 15, as burnt (Figure 4). Thus there is at least an indication that fire has occurred more than once in some places, maintaining savannas at those locations. There is some correlation with clear burning in the three larger sites in the southern part of the township. In general, most clear burnt areas were windfall prior to the fire of 1865 (compare Figures 4 and 6), and again, it is possible that some of these large windfalls had been hardwood forests prior to some earlier fire. In the fire of 1865 hardwoods rarely clear burned; most became areas of

Page  13 ï~~2000 THE MICHIGAN BOTANIST 13 2000 THE MICHIGAN BOTANIST 13 Bay 1000 m Manitowaning Bay Hardwoods, 1863-4 Clear burning FIGURE 6. Location of clear burning superimposed on 1863-64 hardwood distribution and current bur oak savanna study sites. Note that most clear burnt lots were windfall in 1863-64 (see Figure 2).

Page  14 ï~~14 THE MICHIGAN BOTANIST Vol. 39 windfall (compare the northern part of the study area in Figures 4 and 5 for example). This gives some support to the likelihood that the windfalls and clear burns in the three large southern sites were probably hardwoods at one time. Regardless of what species they once were, the large windfalls present in 1863-4 probably were responsible for the particularly devastating and widespread nature of the fire in 1865. It is possible more bur oak savannas existed, in either survey, than just the few well-described sites which are clearly this type of vegetation. Either surveyor might have encountered bur oak vegetation (especially bur oak woodland) and simply labeled it "hardwood." Fortunately, the few well-described sites give at least a positive indication that savannas existed, even if it is uncertain how many there were. Yet, most of the places labeled hardwood were probably not bur oak dominated because oak is never used as a bearing tree for those lots. Furthermore, the description of hardwoods is sometimes qualified as "good hardwood land," which probably would not have described a bur oak savanna. Good land to a homesteader either meant rich soil (savanna soil is generally shallow and rocky) or supporting useable or marketable timber (the savannas have small, almost stunted trees even now after 130 years of growth). Therefore, it is assumed that most of the current bur oak savannas did not exist in 1863-4 and were created in the single event of the 1865 fire. None of the bur oak savannas in Sheguiandah Township have burned since the fire of 1865, and nearly all areas are currently used as livestock pasture and have been grazed more or less continuously since settlement times, according to local oral history. It is probable that grazing keeps these bur oak savannas in an open state, since adjacent bur oak areas outside livestock fences have grown into a closed woodland, dominated by bur oak but with a ground flora and composition more like a hardwood forest (Jones, 1996; Jones, unpublished data). Grazing also prevents fire by keeping fuel build-up low. Two lightning strikes observed in bur oak savannas resulted in burns of only a few meters, and the fire was not carried across the grassland (personal observation). It appears that many of these savannas were created from a single fire, and that (in the absence of fire or grazing) these bur oak savannas can grow into woodland in roughly 130 years. Substrate, growing conditions, and seed stocks suitable for creating oak savannas exist on Manitoulin Island, but their origins appear to be left to the randomness of fire. If fire occurs (at least once in approximately 130 years), existing savannas may be maintained. If fire does not occur, the savannas may grow into hardwood forests only to be re-created again some day in another major fire. Therefore, it is suggested that burning be tested as a management tool for the maintenance of this rare and interesting vegetation type. ACKNOWLEDGEMENTS This work was funded by The Nature Conservancy's Great Lakes Program as part of the International Alvar Conservation Initiative. Unpublished data on floristics and cover vegetation of these study sites is on file with the Natural Heritage Information Centre of the Ontario Ministry of Natural Resources, Peterborough, Ontario. Thanks to Wasyl Bakowsky and John Riley for comments on the text.

Page  15 ï~~2000 THE MICHIGAN BOTANIST 15 LITERATURE CITED Brownell, V., 2000. Significant alvar natural heritage areas in the Ontario Great Lakes Region. Federation of Ontario Naturalists, Don Mills, Ontario. Fowells, H.A., 1965. Silvics of Forest Trees of the United States. Agriculture Handbook No. 271, U.S. Forest Service, U.S. Department of Agriculture, Washington, D.C. Jones, J.G., 1996. Report on the fire history of the bur oak alvars, Sheguiandah Township, Manitoulin Island. Unpublished report for the International Alvar Conservation Initiative, The Nature Conservancy, Great Lakes Program Office, Chicago, Illinois. Reschke, C., Reid, R., Jones, J., Feeney, T. and Potter, H., 1999. Conserving Great Lakes Alvars: final report of the International Alvar Conservation Initiative. The Nature Conservancy, Great Lakes Program Office, Chicago, Illinois. McPhillips, G., 1864. Report, field notes and diary of the survey of the Township of Sheguiandah on the Manitoulin Island. Field book number 1735. Microfilm. Ontario Provincial Archives, Peterborough. O'Keefe, D.G., 1866. Field notes of the Township of Sheguiandah (resurvey) Manitoulin Island. Field book number 1733. Microfilm. Ontario Provincial Archives, Peterborough. Tester, John R., 1996. Effects of fire frequency on plant species in oak savannas in eastcentral Minnesota. Bulletin Torrey Botanical Club 123(4): 304-308. Personal communications: Wasyl Bakowsky, Community Ecologist, Ontario Ministry of Natural Resources, Natural Heritage Information Centre, Peterborough, Ontario K9J 4Y5 John K. Morton, Professor, Department of Biology, University of Waterloo, Waterloo, Ontario N2L 3G1 ANNOUNCEMENT: MICHIGAN FLORA There have been statements made by bookdealers and stores that Michigan Flora is out of print. Not true! All three volumes remain in print, even though Cranbrook Institute of Science no longer handles sales of the volumes. At present, the sole source is the University of Michigan Herbarium. Please access the Herbarium website for complete ordering information: herbarium.lsa.umich.edu. All prices include shipping and handling; each hardbound volume is $18 within the USA, $21 per volume for non-USA orders. Check or money order or purchase order, in US dollars; no credit cards.