1994 Proceedings
North American Conference on Savannas and Barrens

The University of Wisconsin-Madison Arboretum is known for its restorations of the many natural communities that once covered Wisconsin. In general, each restored community is modeled upon the description of that community in The Vegetation of Wisconsin (Curtis1959). Each occupies a rather discreet area of the arboretum, with sharp boundaries between it and adjacent communities some of which may occur naturally only in a different part of the state. These restorations continue to be extremely valuable for teaching and research, and have frequently provided new insights into the ecology of the natural communities they depict.

However, the plan for the "Grady Tract" calls for a somewhat different approach. In this unit, separated from older parts of the Arboretum by the Beltline Highway, there will be a 60-hectare mosaic of related communities--the prairies, savannas, and oak woodlands of southern Wisconsin--all managed by prescribed burning and having gradual transitions from one to another. This approach is similar to that of Stritch (1990) in southern Illinois. The present vegetation of the area includes restored prairie, restored savanna, old field, plantings of conifers and oaks, and oak forest. One area of oak forest known as the Grady Knolls Forest is the site of this study in which we explore the effects of introducing the presettlement process, fire, into the community with little or no additional manipulation. We assume that whatever tree density and composition result probably were matched somewhere in the presettlement landscape. Our object is 1) to let fire change the woods into a community more closely resembling one which was part of the presettlement fire-maintained prairie/savanna/woodland continuum, and 2) to gain insight into the fire/vegetation interactions by recording changes that occur in tree, shrub and ground layer species.

STUDY AREA

As described in a previous paper (Kline and McClintock 1993) Grady Knolls Forest is a 10 ha oak stand on an end moraine of sandy soils and 2-10% slopes, mostly southerly. The forest is bounded on the north by an old field invaded by sumac, on the east by developed private property, on the south by Greene Prairie, and on the west by a restored oak savanna. A small restored savanna extends into the east portion of the forest. The history of the site, described in more detail by Kline and McClintock (1993), includes presettlement fires, a period of grazing ending in 1940, wildfires in 1946 and 1954, and infestation by oak wilt which began over 20 years ago. When this study began, there were widely spaced large oak trees surrounded by younger oaks and vigorously invading black cherries (Prunus serotina). The most important oak was a hybrid of black oak (Quercus velutina) and Hill's oak (Q. ellipsoidalis), but there were some white oaks (Q. alba) and bur oaks (Q. macrocarpa) as well. Two exotic shrubs, honeysuckle (Lonicera X bella) and buckthorn (Rhamnus cathartica), formed a nearly continuous tall shrub layer, and the ground layer was extremely sparse.

METHODS

Extent and Timing Of Prescribed Burns

The first burns took place in mid-April 1990 and 1991. The burns were patchy. We estimated that 50% of the area burned in 1990 and 60-70% in 1991. The 1994 burn was carried out later in spring (May 9), in the hope that honeysuckle and buckthorn might be more severely impacted when leafed out. The plan called for burning the woods in two sections, but because of weather and time constraints, only the south section was burned. The day of the burn was clear, with temperature at 63o F, relative humidity 25%, and the wind speed 15 mph. The fire was a backing fire that moved slowly through the woods with flame heights of approximately 0.6 m. The burn was less patchy than the first two burns. We estimated that approximately 90% of the area was burned well.

Field Sampling

Trees and saplings were recorded within 20 permanently marked 200 m2 plots. Sampling was carried out in late July each year from 1989 to 1992, and again in 1994. Within each plot we identified and measured each tree (DBH > 10 cm) and identified and counted saplings (DBH > 2.5 cm, but <10 cm). We also made a visual estimate of honeysuckle and buckthorn cover and noted whether the plot had burned well.

We sampled the ground layer vegetation using 1.0 m2 quadrats. One quadrat was placed in each of two adjacent corners of the 200 m2 plot for a total of 40 quadrats. The quadrats were offset one meter toward the inside to ensure sampling of undisturbed vegetation. Herb and shrub presence, number of tree seedlings, estimate cover of honeysuckle and buckthorn, and completeness of burn were recorded for each quadrat.

Data Analysis

Data for 1994 were analyzed for the entire site and also separately for the burned and unburned portions. One of the 200 m2 tree plots on the burned section showed no evidence of burning; it was combined with the 8 plots north of the firebreak leaving 11 plots considered to be burned. Four of the 1.0 m2 quadrats south of the firebreak appeared to have been missed by the fire; these were combined with those north of the firebreak making a total of 20 unburned and 20 burned quadrats.

RESULTS

Sapling Response

The number of saplings of each of the four major canopy species decreased between 1992 and 1994, thus continuing the trend of previous years (Table 1). The two most important species, black oak and black cherry, had the greatest losses. For black oak the reduction was 37.9% (Table 2), from 165 to 103 saplings/ha over the two year period. Black cherry saplings were reduced 23.9 %, from 168 to 128 stems/hectare. The number of saplings of black oak and black cherry decreased in both the burned and unburned plots, but the greatest reduction (45.8%) was in black oak saplings in the burned plots.

Tree Response

Although there were 15.9% more black oak trees in 1994 compared with 1992, the average tree size was less and basal area per hectare was essentially unchanged. In contrast, black cherry, with no increase in number of trees but a slight increase to average tree size, had a 12.7% increase in basal area per hectare. White oak and bur oak, as well as black cherry, made small gains in relative dominance at the expense of black oak.

When the data are separated into burned and unburned categories (Table 2), the increase in number of black oak trees is seen to have actually occurred only in the burned plots, where the increase was 34.8%. The unburned plots experienced a loss of only 4.8% over the two year period. In contrast, black cherry had a loss of 6.4% in the burned plots and a gain of 10.0% in the unburned. For black oak, basal area per hectare was essentially unchanged whether the plot was burned or unburned, while basal area for black cherry increased in both areas, but substantially more in the unburned plots.

When the trees were separated by species into size classes, black cherry, white oak and bur oak showed the reverse-J distribution one would expect for young, invading populations. Except for two large white oaks, all the trees of these species in the sample were less than 7 dm2 (12 inches DBH). For black oak, trees less than 8 dm showed a similar distribution. In addition, a substantial number of large, somewhat open-grown black oak stems remained. In 1994, these trees with individual basal areas ranging from 10 dm2 to 33 dm2 (14-26 inches DBH), accounted for 23.5% of the trees and 50.6% of the total basal area for black oak. In 1989, at the start of the experiment, large black oaks accounted for 22.4% of the trees and 44.6% of the total basal area for black oak.

Honeysuckle And Buckthorn Response

Mean estimate of percent cover for honeysuckle and buckthorn in the unburned plots was 44% in 1992 and changed very little in subsequent years. Mean per cent cover in the plots burned in 1994 was 38 % in 1992 and 18 % in 1994. These exotic shrubs were noticeably taller in the unburned area, often tall enough to obstruct the view through the woods. In the burned area much of the re-growth was less than 1.0 m.

Frequency of honeysuckle seedlings in the 40 small quadrats increased slightly, from 27.5% in 1992 to 32.5% in 1994, with all the increase occurring in the 20 unburned quadrats. Frequency of buckthorn seedlings decreased slightly from 65% to 55%, with all of the decrease occurring in the 20 burned quadrats.

Ground Layer Response

From 1992 to 1994, the number of species per 1.0 m2 quadrat increased from 6.0 to 7.6 in the burned quadrats, while it decreased from 5.3 to 4.9 in the unburned (Table 3). Prairie species that germinated in response to the 1990 and 1992 burns continued to disappear. Of the 18 species that were once found in the quadrats only 5 remained in 1994, and only one occurred in as many as 2 quadrats.

Woodland species, including herbs, shrubs and vines, increased from 4.2 to 4.7 species appearing in the burned plots and from 4.0 to 4.2 in the unburned. New species appearing in the burned plots included Phytolacca americana and Polygonum virginianum. There were substantial increases in Rhus glabra and Ribes sp., while Smilacina racemosa and Carex pensylvanica were substantially reduced. New species appearing in the unburned quadrats were Desmodium glutinosum and Prunus americana. Berry-fruited species are prominent among those increasing or newly appearing, continuing the trend notes in Kline and McClintock (in press).

The average number of weed species per quadrat increased from 1.4 to 2.6 in the burned and decreased from 1.1 to 0.6 in the unburned plots. The change largely was due to major increases of Solanum nigrum and Erectites hieracifolium.

DISCUSSION

Management Implications

The late spring (May 9, 1994) burn was somewhat more effective in reducing the cover of honeysuckle and buckthorn than were the two previous earlier burns. Cover was reduced by approximately one-half compared to the one-third reduction of each of the first burns. However, the greater patchiness of the first burns, which may or may not have been due to the seasonal timing, should be taken into account. A complete burn would be expected to reduce the cover more regardless of time of year. Factors that might have contributed to the more complete burn include the lower relative humidity and higher temperature on that day.

The dramatic reduction of two of the few early flowering herbs (Smilacina racemosa and Carex pensylvanica) by the late spring is cause for concern. It is also noteworthy that the two new species in the burn plots were both tall, late flowering species that tend to benefit from some disturbance.

The substantial decrease in black oak saplings (Table 2) can be partly explained by the advancement of many of them into the tree category, but the advancement of so many at the same time is difficult to explain in terms of the late burn. This phenomenon was not observed after the other two burns. It is possible that the burn reduced competition from other saplings and the saplings remaining were able to capitalize on the summer's unusually favorable growing season.

Change And Resistance To Change

It is apparent after five years that some changes are rapid - some even evident after the first burn - and others that might be expected are either not taking place, or are taking place at a pace so slow that they are difficult to detect. The forest structure is still that of a savanna being rapidly invaded by young trees. Few of the old trees have died, and the young trees maintain the size distribution they had at the start of the project, while increasing slightly in number. Basal area of all trees combined has changed so little that light levels beneath the trees are similar to initial measurement. Changes in relative dominance of the major species proceeds, but at a slow pace. In Minnesota, White (1983) notes that 13 years of annual prescribed burns of a dry oak woods had failed to change the woods into a savanna.

More rapid changes have been the reduction in density of saplings, particularly of black oak after the first and most recent burns, and the immediate reduction of honeysuckle and buckthorn cover after each burn. These changes have increased the light level at the forest floor and affected the ground layer. Ground layer responses have been rapid. The greatest increase in species diversity of the ground layer took place the year of the first burn, and the increase continues at a slower rate.

Whether these changes taking place at various rates have succeeded or will eventually succeed in changing Grady Knolls Forest into a community that might have been similar to the presettlement scene of Dane County is a difficult question to answer. However it is likely that various rates of change for different functional groups within a community have been the typical response of a community to changes in fire regime and/or climate.

LITERATURE CITED

Curtis, J. T. 1959. The Vegetation of Wisconsin: An Ordination of Plant Communities. University of Wisconsin Press. Madison, WI.

Kline, V. M. and T. McClintock. 1993 Effect of burning on a dry oak forest infested with woody exotics. In: R. G. Wickett, P.D. Lewis, A. Woodcliffe and P. Pratt, eds. Proceedings of the Thirteenth Annual Prairie Conference. Windsor, Ontario.

Kline, V. M. and T. McClintock. The ground layer of an oak forest in transition under prescribed burning. In: Proceedings of the Midwest Savanna Conference. Chicago, Illinois. In press.

Stritch, L. R. 1990. Landscape-scale restoration of barrens-woodland within the oak-hickory forest mosaic. Restoration and Management Notes 8:2:73-77.

White, A. 1983. The effect of thirteen years of annual prescribed burning on a Quercus ellipsoidalis community in Minnesota. Ecology 64:1081-1085.

Table 1. Change in sapling and tree number, basal area and relative dominance from 1989 to 1994.

*Species basal area as a percent of total basal area.

Table 2. Percent change in black oak and black cherry density from 1992 to 1994 for burned and unburned plots.

Table 3.  Mean number of species/m2 quadrat in burned and unburned areas.