1993 Proceedings of the Midwest Oak Savanna Conferences
THE GROUND LAYER OF AN OAK FOREST IN TRANSITION UNDER PRESCRIBED BURNING
Changes in the ground layer of a dry oak forest after two, consecutive-year, prescribed burns which substantially reduced a dense cover of honeysuckle (Lonicera x bella) and buckthorn (Rhamnus cathartica) are positive. The frequency of each ground layer species, based on presence in m2 quadrats, was recorded each year for four years, beginning the year before the first burn. The species data are presented in five categories: honeysuckle and buckthorn seedlings, herbaceous weeds, prairie species, native woodland shrubs and vines, and native woodland herbs. Compared with a forest in which the two shrubs were removed by cutting and herbicide treatment but were not burned, frequencies of honeysuckle and buckthorn seedlings were lower in the burned forest, but were still high enough to threaten heavy reinfestation in the future. The burns appeared to facilitate weed invasion, but frequencies generally remained low or were tending to decrease, and the species did not include any known to cause long-term serious problems in forests. No prairie species were present before the first burn; 15 appeared that year, but few persisted, probably due to lack of sufficient light. Woodland shrubs and vines, and woodland herbs showed the greatest positive response over the four years and were clearly the most important groups in the ground layer. All of the shrubs and vines and 29% of the herbs were berry-fruited. We identified a group of 14 "light generalist" ground layer species that are likely to grow well under the present canopy conditions and also under the more open conditions that may develop under continued fire management. Eight of the 14 are already present, including 7 that are berry-fruited. We also developed a list of species suitable for the site which may be "light specialists" having narrow ranges of light tolerance. Uncommon now, these species may have been more widespread in the presettlement savannas.
Before European settlement much of southern Wisconsin was covered with a fire-maintained continuum of vegetation ranging from open prairie with only an occasional oak tree to oak forests with a nearly continuous oak canopy. Intermediate segments along the continuum have been variously categorized as "savanna", "oak woodland", etc., but the boundaries were not discreet and their locations changed over time in response to changes in climate or fire.
The long term management plan for the University of Wisconsin-Madison Arboretum calls for a 60-ha block in the "Grady Tract" to become a continuum of prairie, oak savanna and oak forest and their intermediates. The entire block will be managed by prescribed burning. The present vegetation of the area includes restored prairie, restored savanna, old field, conifer plantings, and oak forest. The site selected for this study is Grady Knolls Forest, an oak stand located within the block on an end moraine, with sandy soils and 2-10% slopes, generally southerly. As part of the implementation of the plan, prescribed burns were carried out in the forest in 1990 and 1991. The rationale is to reintroduce fire as a process that once shaped the landscape, and to assume that whatever tree density develops over time under burn management is representative of some part of the pre-European settlement landscape.
We established permanent plots to monitor the effects of burning on tree density and reproduction, ground layer composition, and cover of honeysuckle (Lonicera X bella) and buckthorn (Rhamnus cathartica), two nonnative shrubs heavily infesting the site. In an earlier paper (Kline and McClintock, 1993) we reported that the prescribed burns in 1990 and 1991 followed by a year of no burn were accompanied by: (1) a decrease in tree density and basal area per hectare during the two burn years that was balanced by increases during the no-burn year, resulting in little net change, (2) a change in the size distribution (due to greater top-kill of small trees) of the black oak hybrid (Quercus velutina x Q. ellipsoidalis) that dominates the site, and a substantial decrease in its density, (3) a large reduction in number of saplings of black oak, white oak (Quercus alba) and black cherry (Prunus serotina), (4) a decrease in cover of honeysuckle and buckthorn in each of the burn years, followed by a slight increase in the year of no burn, and (5) increased diversity of the ground layer, especially the patches where burning removed all leaf litter and substantially reduced the cover of buckthorn and honeysuckle. In this paper we examine the ground layer data in greater detail, considering the site as a whole (all quadrats combined), recognizing that patchiness is a usual characteristic of prescribed burns in oak forests (Stritch 1990).
When the experiment began, the ground was visibly bare beneath the densest honeysuckle and buckthorn. We were concerned that if the burns were successful in reducing the cover of these large shrubs, as we hoped, there might not be enough native herbs and shrubs to provide good ground cover, and that with the increased light levels the bare ground might be reinvaded by seedlings of the exotic shrubs, and perhaps by weedy herbaceous species.
Many southern Wisconsin oak forests, especially those near urban areas and having a history of soil disturbance, are also infested with honeysuckle and buckthorn. A dense infestation will effectively exclude a large number of ground layer species, which have not developed adaptations that would allow them to co-exist with these newcomers. This loss is often in addition to loss of species through grazing (Rogers 1959). The ground layer of such a forests may have very low diversity even after the exotic shrubs are removed (Kline, et al 1982), especially if there is no seed source in the near vicinity. In the study cited, honeysuckle and buckthorn in two heavily infested oak forests were eliminated, using cutting followed by treatment of the cut stumps with herbicide. A year after treatment, the most disturbed stand, which also lacked a nearby seed source, had only 2.0 native species/m2 compared with 1.2 species/m2 before treatment. Only one native herbaceous species, enchanter's nightshade (Circaea quadrisulcata), had a frequency greater than 25%.
The forest in which our study is located is bounded by an old field planted with western conifers on the north, a developed neighborhood of wooded lots on the east, a 45 yr-old planted prairie on the south and a small restored savanna on the west (Fig.1). Glass and Howell (1993), in a study carried out within the forest near the west boundary, found that 17 nonnative herbaceous species and 23 native herbaceous species were in either the soil bank or the seed rain or both. This was also true of honeysuckle and buckthorn. To discourage reinvasion of the exotic shrubs and invasion of weeds, a robust ground layer of native species will be important. Species that can thrive under present conditions and also in whatever more open community develops under fire management would be particularly important in ensuring continued cover during the transition.
Bray (1955) showed that species composition beneath individual savanna trees varies along the light gradient from trunk to canopy edge. Light available at ground level is also dependent upon tree density, so that ground layer composition must have changed along the oak density continuum, with some species succeeding best under the sunny conditions of low density stands and others succeeding best in the shade of high density stands. Tree density at any particular location would have changed over time in response to short and long term changes in climate and the associated changes in the frequency and intensity of fire. Chance fire events may have caused rapid local changes in tree density as well. Light specialists with very narrow ranges of light tolerance would have been at a disadvantage in the ground layer in such periods and may have been extirpated locally. Such specialists are likely to have developed good long-range dispersal of seeds, to allow them to take advantage of scattered locations in the landscape where appropriate light conditions developed over time. Fenner (1985) suggests that colonization of widely separated patches of a particular habitat and colonization of oceanic islands may present similar problems. Light specialists would have been very vulnerable to the rapid changes taking place in the landscape since European settlement.
Species with a broad range of light tolerance would have had some advantage under changing conditions, and we hypothesize that such species were well represented in the pre-European settlement landscape, and persisted more successfully under post-settlement conditions. Light generalists would have been able to persist in spite of changes in canopy, so may not have been as dependent on long-distance dispersal. It is likely that vegetative spread was very important in this group, since there would have been substantial areas with continuous suitable habitat available to them in space and time. Such species could be important in providing a robust ground layer that would prevent soil erosion, provide fuel for prescribed burns and discourage invasion of exotics as the oak forest in our experiment changes in response to the new prescribed burn regime.
We used our field data, in conjunction with species frequency and presence data of Curtis (1959), to develop a general procedure to identify a group of generalist species that would be adapted to this site. We also developed a list of possible light specialists that might be successfully introduced after first determining their light requirements. Introduction of these species would probably be more successful if tree density changes are slow and if potential habitat is available elsewhere in the planned 60-hectare mosaic of fire-managed communities.
Location of sample quadrats was facilitated by prior placement of permanent markers at the corners of a grid of 0.25-ha squares (Fig. 2) that eventually will be established throughout the Arboretum as a critical component of the Arboretum’s geographic information system (GIS). Twenty 200-square-meter plots, each in the southwest corner of a grid square, were used to sample trees and visually estimate honeysuckle and buckthorn cover. We used square meter quadrats to sample the ground layer. One quadrat was placed in each of two adjacent corners of each of the 200-square-meter plots, a total of 40 quadrats. The quadrats were offset one meter toward the inside of the large plotto ensure sampling of undisturbed vegetation. Herb and shrub presence, number of tree seedlings and estimated cover of honeysuckle and buckthorn were recorded for each quadrat.
Sampling took place in early July in each of the four years, starting in 1989. Prescribed burns were carried out April 16, 1990 and April 17, 1991. Prior to the first burn, honeysuckle and buckthorn were cut and treated in the strip along the east property line to facilitate movement of the burn crew. This section was burned April 18, 1991. For all burns temperature was approximately 100C (500F); relative humidity was 50%; and flame heights were 15-30 cm. The burns were patchy. An estimated 50% of the area burned in 1990 and 60-70% in 1991.
Honeysuckle and buckthorn seedlings
On the burned site, honeysuckle seedling frequency decreased substantially after the first burn and then leveled off, while on the unburned sites honeysuckle seedling frequency showed increases each year. Buckthorn seedling frequency increased on all sites, reaching values much higher than the starting values. This suggests that a substantial invasion of buckthorn seedlings can be anticipated when dense buckthorn and honeysuckle are removed, whether the removal is accomplished with or without fire. However, future monitoring may show that repeated prescribed burns result in substantial mortality of the small seedlings.
Table 2 shows species frequencies for each of the 4 years, arranged according to the year of first appearance. It appears that fire facilitates weed invasion. There were 6 species present at the start; 4 additional species appeared after the first burn, and 8 after the second burn. There were large peaks of black nightshade (Solanum nigrum) after the first burn and of dandelion (Taraxacum sp.) and fireweed (Erechtites hieraciifolia) after the second, although these peaks were not sustained. In 1992 only 2 species appeared in more than one quadrat.
Distribution of weeds was very patchy on the site, and more were recorded in the quadrats closest to the south and west boundaries, where the prevailing southwest winds enter.
The patchy distribution of the prairie species is even more pronounced than that of the weeds. Both groups are concentrated in the same general area, down wind from adjacent prairie and oak barren restorations. It seems likely that prairie seeds were carried by wind from this nearby source, although Glass and Howell (1993) found few prairie species in the seed rain.
Native vines and shrubs
Native herbaceous woodland species
The group includes 5 species with hooked seeds ("sticktights") and 4 with berries. For these species and the woody shrubs and vines, dependent upon small mammals and birds (Fenner 1985), it may actually have been advantageous that some of the honeysuckle and buckthorn shrubs survived the burn and were available for cover and perches.
Three of the species are short-lived annuals or biennials. One, hog peanut (Amphicarpa bracteata), was found only the year of the first burn; the other two dropped in frequency in the final, no-burn year.
At least 9 of the 11 perennials are known to spread vegetatively. Enchanter's nightshade (Circaea quadrisulcata), a species with hooked seeds and considerable ability to spread vegetatively, reached the highest frequency of the study (80.0%) in 1992.
Comparison of the importance/abundance of the five species groups
Figure 3 shows yearly changes in the number of species for each of the species groups. Number of species is a measure of diversity, so the graph might be interpreted as showing the change in diversity within each group over time; or, for each year, it shows the contribution each group made to the species diversity of the community. The prairie group is seen to make its largest contribution to species diversity the year of the first burn, followed by two years of decline. Weed diversity rises in near-linear fashion for the two burn years, then declines. Like the prairie species, the woodland herbaceous species have the greatest increase in diversity the year of the first burn, but then they continue to increase.
We also used a measure of importance or "abundance" that consists of the sum of the frequencies of all the species within each group (Fig. 4). This graph gives a better idea of the contribution each group makes to the structure and function of the community. Each group shows small increases after the first burn, and the four lines are fairly close together. After the second burn (1991), the woodland group diverges up sharply, while the prairie group diverges downward. The prairie and weed groups show nearly parallel decreases from 1991 to 1992. Overall the herbaceous woodland species and the shrubs show a trend of increasing importance over time.
The results provided an opportunity to let the present ground layer indicate the site potential for other ground layer species. This is an approach that has been proposed for forest management, where ground layer species can help make site-sensitive decisions about sustainable forest types (Kotar et al, 1988). In the case of the study site we have introduced a process, and we do not have a definite compositional or structural endpoint in mind. We may want to favor or add appropriate species to the ground layer, and we need to identify a group of transitional species with broad light tolerances that are likely to thrive under present conditions and will be able to maintain good ground cover even if conditions change under fire management.
We first compared a list of all native ground layer species (including herbs, shrubs and vines) recorded in our study quadrats in 1992 with lists of prevalent ground layer species for southern Wisconsin upland forests as listed by Curtis (1959), and found that the ground layer that matched best was that of the southern dry forest, although there were similarities with the southern dry mesic forest ground layer as well. The southern dry forest was also the best match for the canopy in this case, but this would not necessarily be true. On disturbed sites, ground layer vegetation can be a better indicator of environmental conditions and original forest type than the existing canopy (Kline 1976, Kotar et al, 1988).
We then listed all the prevalents of southern dry forests having quadrat frequencies of 10% or higher according to Curtis (1959). These are the species most likely to be found in an undisturbed dry oak forest (high prevalence values) that in addition are likely to be fairly common there (high frequency values)--potentially very successful species for our site. The frequency values given by Curtis are often ignored in compiling lists of species for restorations, but species with high frequency values are the ones most likely to do well throughout a community. They are likely to be generalists with fairly broad tolerances of environmental factors. Ten of the species with high frequencies also occurred on the study site.
Since we were interested in species that would also do well in whatever more open community might develop over time, we also compared the list of study site species to the lists of prevalents for savanna types. (Curtis profiles only forest, savanna and prairie along the tree density gradient.) We found that the best match was with the oak opening, with oak barrens a close second.
Six species found on the study site had frequencies >10% in southern dry forests and also were prevalents in the oak opening, where their frequencies ranged from 7.6% to 21.8% (Table 6a). Two species had the next highest frequencies (9.4%, 7.4%) in the forest and frequencies >12% in either oak opening or oak barrens (Table 6a). These were added, to give a total of 8 generalist species already present on the site. Seven of these species have berries or other small fleshy fruits.
There were 6 species not represented in the sample that had frequencies >10% in the forest and were prevalents in the oak opening, where their frequencies were also >10% (Table 6b). These clearly were members of the "guild" of generalist species with light tolerances broad enough to succeed in the present forest and in whatever community is likely to develop under prescribed burning. There were no species with more than 10% frequency in one of the two communities and between 7% and 10% in the other, in contrast to those in Table 6a.
Choosing common species with broad tolerances is an approach that seems compatible with our goal of a vigorous, functioning ground layer that can persist under the conditions produced by prescribed burns. These species should be easy to establish and will help to quickly increase soil protection.
After several years of prescribed burns we expect that rates of canopy change, and therefore changes in light levels at the ground, will moderate. At that point consideration will be given to introducing species that may be light specialists, in particular those that may be limited to a particular light level somewhere between full sun and deep shade. To identify species that might fit this "light specialist" category, we considered the modal species for dry and dry mesic forests (Curtis 1959) that have very low presence values. These are species that are uncommonly found in the very type in which they do best, a paradox. We used our own experience and standard references listing species habitats to eliminate those which may be specialists in microclimate or require soil disturbance, and select those that might be limited to light levels found only in particular edge or gap locations (Table 7). They may have done very well in some intermediate savanna or woodland community type before settlement when such communities were widespread and provided the right light conditions in many locations within seed dispersal distance.
Using a similar argument, we considered modal species with prevalence values high enough to make them prevalent species for SD or SDM but with low frequency values, and selected those that may have had low frequencies because they required specific light conditions that were uncommon in the forests (Table 7).
Of the 27 possible light specialists (Table 7), 26% have seeds with pappus adapted for wind dispersal; 26% have berries and other fruits or seeds that are attractive to birds and small mammals; and 26% have hooked seeds or fruits. Comparable figures for the 14 species with broad light tolerances are 0%, 73% and 7%. We had predicted that long-range dispersal would be important for specialists. Since long-range dispersal can be provided by wind, or by vertebrate transport either internally or externally (Willson, 1992), the potential is there for both specialists and generalists. However, special appendages for wind transport are found only among the possible specialists. Wind dispersal distances depend in part on the height of the plant in relation to adjacent plants, and Willson found that wind-dispersed species in fields are generally taller than those not wind-dispersed. This is true of the species with appendages for wind dispersal on the list of possible specialists. As tree density increases, wind becomes less effective at moving seeds of ground layer plants. It is possible that wind-dispersed light specialists require a fairly open community not only to provide particular light levels, but also to allow wind movement of seeds. Our prediction that vegetative spread would be characteristic of generalist species is supported by the data, which show that 79% of the species spread vegetatively. We were unable to find information about vegetative spread for many of the specialists, but it appears to be an uncommon character for the group.
The most important families in the group of possible light specialists are Asteraceae (5 species), Poaceae (4 species) and Ranunculaceae (4 species). The generalist group has a single representative of the Asteraceae and none in the other two families.
The list will be useful for trial plantings to determine which species actually do show a limited range of light tolerance, and for which of them the forest provides suitable conditions after several years of prescribed burns.
Fires that maintained the prairies and savannas of the presettlement landscape in southern Wisconsin undoubtedly entered adjacent oak forests, burning through the ground layer repeatedly. Thus it is not surprising that the native ground layer species of the oak forest in this study are adapted to survive fire and to take advantage of the reduced competition from buckthorn and honeysuckle that the fire provided. The overall effects of the prescribed burns on the ground layer were positive. Woodland species, including native herbs, woody vines and shrubs, have increased in diversity, and have become clear dominants of the ground layer. Several species now have frequencies comparable to those in the presettlement dry forests.
The woodland species already present before burning and the more recent arrivals include a high percentage having fruits that are berries (100% of the shrubs, 29% of the herbs) or have hooks to attach to fur or feathers (36% of the herbs). Many of the berry-fruited native species known to occur in forests in this area were found at the site as soon as the exotic shrubs were reduced in cover. Honeysuckle and buckthorn also have berries and they were extremely successful in producing seedlings. It is interesting to speculate that these two species, that completely dominated the site, may have provided conditions under which only species with similar reproductive strategies could maintain a foothold.
Although honeysuckle and buckthorn cover was substantially reduced, tree canopy changed very little and the light levels even after two burns were probably too low for most of the 15 prairie species that germinated. The high mortality plus the trend of decreasing frequencies of all the survivors should discourage planting of these or additional prairie species until conditions are more favorable. The prairie plants observed in flower at the south edge of the forest may be occupying the only suitable habitat the forest presently provides. In a sense, the seeds that managed to get to other parts of the site were sacrificed when the fire triggered germination but failed to produce conditions conducive to survival. Like the woodland herbs, most of the prairie species appeared to lack mechanisms for delayed germination and responded the year of the first burn. On the negative side, the two prescribed burns failed to control reproduction of honeysuckle and buckthorn. Although the frequency of honeysuckle seedlings was reduced somewhat and buckthorn seedlings increased less than in a forest in which the mature shrubs were removed by means other than fire, seedling frequencies for these species were more than adequate to restock the forest with continuous exotic shrubs. It is possible that these seedlings will be less able than the native shrubs and herbs to resprout after fire, and that future prescribed burns will keep them in check enough to allow the native species
to dominate. It is probably not possible to achieve complete elimination of honeysuckle and buckthorn as long as seed sources are widespread in surrounding areas.
The germination of weedy herbaceous species in response to the burns is of less concern. There were no strong trends of increasing frequency, and none of the species present the last year is known to compete aggressively with native ground layer species. If we are successful in encouraging appropriate native species, we expect weedy herbaceous species to become less common even though seeds continue to arrive at the site.
The long term success of this restoration will depend to a large extent on the development and maintenance of a ground layer of native species capable of competing under the changing conditions resulting from prescribed burning. Such a ground layer will provide competition for weedy herbs and seedlings of exotic shrubs, and will also provide fuel for prescribed burns. Oak leaves now provide the major fuel. If oak canopy is reduced there will be fewer oak leaves and ground layer input will be especially important. Fortunately, as the canopy is thinned, the increased light will allow more vigorous growth, providing there are species present that can respond to the increased light.
To maintain a robust, resilient ground layer as tree density changes in response to prescribed burns, we expect the key species will be the 14 we have identified as site-appropriate light generalists, whose range of light tolerance extends from dry forest to savanna and whose ranges of tolerance for soil and other environmental factors are broad enough to allow them to succeed throughout the site. Eight of the generalists, including all seven of those producing berries, are already present and should increase naturally not only by seed but also by spreading vegetatively. Establishment of the remaining six will be the first planting priority.
Light specialists were probably also important in the presettlement ground layer. Species with narrow ranges of tolerance for light intensity and/or pattern would have had much more widespread habitat available in the presettlement landscape than at present. Those that were once quite common may have become limited to a small number of forest edges or gaps meeting their requirements, as Packard (1990) has suggested for species formerly common in Illinois savannas that have become rare with diminishing habitat. We have selected 27 possible light specialists from among species that Curtis (1959) and his students found over 40 years ago to be uncommon in dry and dry mesic forests, yet more common in those forests than anywhere else. None have appeared spontaneously. We expect to determine by test planting whether some or all of these species are light specialists that can succeed under the conditions resulting from a sequence of prescribed burns of the study forest. For long-term success it may be important to have enough habitat available in the total 60-ha savanna/prairie/forest project area to allow them to move to suitable habitat as light conditions change. Restoration of adequate scale may be an important factor for populations of these species to be viable.
The generalist and possible specialist groups differ from each other in type of seed dispersal, ability to spread vegetatively, vulnerability to post-settlement changes, present distribution and family representation.
Bray, J.R. 1955. The savanna vegetation of Wisconsin and an application of the concepts order and complexity to the field of ecology. Ph.D. dissertation, University of Wisconsin.
Curtis, J.T. 1959. The vegetation of Wisconsin. University of Wisconsin Press.
Fenner, M. 1985. Seed ecology. Chapman and Hall, New York.
Glass, S. and E.A. Howell 1993. Mechanisms responsible for the revegetation of a former oak barrens after removal of the canopy. (These Proceedings).
Jordano, P. 1992. Fruits and frugivory. In: Seeds: the ecology of regeneration in plant communities. M. Fenner, editor. C.A.B. International.
Kline, V.M. 1976. Dynamics of the vegetation of a small watershed. Ph.D. dissertation, University of Wisconsin.
Kline, V.M., G. Cottam and T. Samingan. 1982. Response of oak woods understory plants to eradication of dense growth of an exotic invader, Lonicera X bella. Paper presented at Ecological Society of America meeting, State College, Pennsylvania.
Kline, V.M. and T. McClintock. 1993. Effect of burning on a dry oak woods infested with woody exotics. Proceedings of the thirteenth North American Prairie Conference, Windsor, Ontario. (In press)
Kotar, J., J.A. Kovach and C.T. Locey. 1988. Field guide to forest habitat types of northern Wisconsin. Department of Forestry, University of Wisconsin-Madison and Wisconsin Department of Natural Resources.
Packard, S. 1991. Rediscovering the tallgrass savanna of Illinois. In: Proceedings of the Oak Woods Management Workshop, G.V.Burger, J.E.Ebinger and G.S.Wilhelm, Eds. Eastern Illinois University, Charleston, Illinois.
Stritch, L.R. 1990. Landscape-scale restoration of barrens-woodland within the oak-hickory forest mosaic. Restoration and Management Notes 8:2 73-77.
Willson, M.F. 1992. The ecology of seed dispersal. In: Seeds: the ecology of regeneration in plant communities. M. Fenner, editor. C.A.B. International.