Managing forests with prescribed fire: Implications for a cavity-dwelling bat species

Justin G. Boyles^a, Doug P. Aubrey^b,*

^a Department of Ecology and Organismal Biology, Indiana State University, Terre Haute, 47809, USA
^b USDA Forest Service, Savannah River, P.O. Box 700, New Ellenton, SC 29809, USA
Received 1 June 2005; received in revised form 27 September 2005; accepted 30 September 2005

Abstract

Prescribed burning is used as a restoration and management technique in many deciduous forests of eastern North America. The effects of fire have been studied on habitat selection of many vertebrate species, but no studies have reported the effect of fire on bat roosting habitat. Fire initially leads to an influx of dead and dying trees, an increase of light availability, and a decrease of canopy and sub-canopy tree density. These characteristics are beneficial to many forest-dwelling vertebrates including cavity-roosting bats. We evaluated evening bat (Nycticeius humeralis) roost-site selection at the stand-scale in order to determine roosting preferences as they relate to prescribed burning. Standard radiotelemetry techniques were used to locate evening bat roost trees. Canopy light penetration and overstory tree density were measured in both burned and unburned forests. Sixty-three trees used as roosts by both male and female evening bats were located during both the summer and winter and all 63 roosts were located in the burned portion of the study area. Canopy light penetration was higher and canopy tree density was lower in the burned forest than unburned forest. An increase in light availability may release bats from one of the constraints suggested for many forest-dwelling bat species in roost tree selection—sun-exposure. This should increase the abundance of trees with characteristics suitable for roosting and may allow bats to roost throughout the interior of the forest as opposed to only on forest edges, thereby allowing bats to roost closer to foraging grounds and possibly lessening predation rates. Lower tree density may allow for ease of flight within the forest as well as more efficient locating of roost trees. In addition, there were a significantly higher proportion of dead trees, which evening bats commonly use as roost trees, in burned forests compared to unburned forests. Prescribed burning appears to initially lead to creation or restoration of favorable cavity-dwelling bat habitat and its continual implementation perpetuates an open sub-canopy. Therefore, we suggest that prescribed burning may be a suitable tool for management of roosting habitat for cavity-roosting bats.

# 2005 Elsevier B.V. All rights reserved.

Keywords: Fire ecology; Forest restoration; Nycticeius humeralis; Oak-hickory forest; Prescribed burning; Roost selection

CLICK HERE TO DOWNLOAD AS PDF

1. Introduction

Fire is critical in regulating and maintaining many forest ecosystems (Huddle and Pallardy, 1996; van Lear, 2002). In particular, much of the western portion of North America’s eastern deciduous forest is thought to have been shaped and maintained through fires set by Native Americans prior to European settlement (Cottam, 1949; Ladd, 1991; Pyne, 1982). Prescribed burning has been shown to alter many characteristics of forest habitat potentially affecting forest-dwelling bats including tree mortality, which increases available trees commonly used by wildlife (Huddle and Pallardy, 1996; Arthur et al., 1998; Hartman and Heumann, 2003; Aubrey, 2004); pathogen susceptibility, which expedites cavity formation (Burns, 1955; Paulsell, 1957; Smith and Sutherland, 2001); and canopy light penetration (Anderson and Brown, 1986; McCarty, 1998; Aubrey, 2004), which is known to affect roost suitability (Kurta et al., 1993; Brigham et al., 1997).

* Corresponding author. Tel.: +1 803 725 1758; fax: +1 803 725 0311.
E-mail address: daubrey@fs.fed.us (D.P. Aubrey).

0378-1127/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.foreco.2005.09.024

Many forests are managed for timber harvest by the use of mechanical thinning or clear-cutting. Prescribed fire differs from these management techniques because fire is generally used to maintain or restore natural forest ecosystems and reduce understory competition, while other silvicultural methods are predominately used to maximize harvest yield and quality. When first implemented, prescribed burning decreases overstory tree density and basal area (Anderson and Brown, 1986; Peterson and Reich, 2001). Following this initial fire with frequent  burns generally reduces fuel accumulation and subsequent burns are therefore relatively cool and somewhat nonuniform (Ladd, 1991). However many fire-sensitive seedlings and saplings are generally eliminated or prevented from regenerating (Lorimer, 1985; Ladd, 1991; Moser et al., 1996; Brose and van Lear, 1997; Barnes and van Lear, 1998). Fire-tolerant trees remain as overstory dominants and canopy recruitment coincides with periods of decreased burn frequency or intensity (Crow et al.,1994). For example, oaks possess thick bark which insulates them from heat associated with burning and are capable of rapid compartmentalization when damaged by fire which inhibits fungal infection and makes mature individuals relatively tolerant of fire (Lorimer, 1985; Abrams, 1985; Crow, 1988; Stearns, 1991; Smith and Sutherland, 1999; Peterson and Reich, 2001). Much of the western portion of the eastern deciduous forest is believed to have resembled the structure and composition of a woodland (i.e. moderate to low canopy coverage) more than a forest (i.e. high canopy coverage) prior to European settlement, which is largely attributed to fire (Cottam, 1949; Curtis, 1959; Nuzzo, 1986; Ladd, 1991). Therefore, managing forests with prescribed fire reintroduces a historic disturbance process that other silvicultural techniques lack, and may provide a more heterogeneous habitat for wildlife that is similar to forested areas prior to wide-spread fire suppression by European settlers. Furthermore, prescribed burning has become increasingly common over the past few decades, as land managers have seen the effects of fire suppression. Specifically, fire has been most commonly reintroduced to systems where conservation of biodiversity and restoration of historic ecosystem processes are paramount.

Numerous studies have focused on vertebrates in forests maintained by prescribed burning (e.g. reptiles, McLeod and Gates, 1998; small mammals, Simon et al., 2002; amphibians, Schurbon and Fauth, 2003; birds, Blake, 2005). However, to our knowledge, there are no studies reporting habitat selection of tree-dwelling bats in burned forests, although the need for such studies has been suggested (Menzel et al., 2001b; Carter et al., 2002). Furthermore, the effects of forest management on roosting habitat of bats are not clear and it has been suggested that wildfires and prescribed burning may have detrimental (Chambers et al., 2002) or beneficial (Carter et al., 2002) impacts on bats. For some bat species, such as the eastern red bat (Lasiurus borealis), fire may pose a direct threat to survival as early spring burns remove litter where occasional winter roosting occurs and can potentially scorch hibernating individuals (Moorman et al., 1999; Rodrigue et al., 2001). Bats roosting in snags (standing dead trees) are also susceptible to direct disturbance from fire if roost trees ignite (Rodrigue et al., 2001), but this is probably not a major cause of mortality unless the roost burns quickly and the bat is in deep torpor (Carter et al., 2002). Availability of suitable roosts likely influences habitat selection (Kunz, 1982) and roosting sites are thought to be of critical importance to conservation of many bat species (Fenton, 1997). Fire generally removes understory saplings and may lead to the creation of suitable roost trees for cavity-roosting bats. Fire may also remove standing snags if fuel loading is high. However, fuel should not accumulate directly under a snag when burning is frequent so snags should persist under these circumstances. As fire increases the heterogeneity of forest structure and composition (Fule et al., 2004), there should be an increase in the diversity of available roosting habitat.

Previous studies have focused on activity (Humes et al., 1999; Patriquin and Barclay, 2003; Tibbels and Kurta, 2003; Mazurek and Zielinski, 2004), demographic parameters (Miller, 2003), and roosting preference (Campbell et al., 1996; Menzel et al., 2002; Elmore et al., 2004) of bats in heavily managed forests, but have not examined fire-based management. With the exception of a few anecdotal reports of L. borealis being driven from their leaf litter  ibernacula during winter burns (Moorman et al., 1999; Rodrigue et al., 2001), there is little empirical data about how fire affects forest dwelling bats.

The objective of this study was to determine evening bat (Nycticeius humeralis) roost-site selection at the stand-scale in a forest heavily managed by fire and to understand those characteristics of the forest that may influence roost tree selection. The evening bat is a locally abundant cavity-dwelling bat species found throughout much of the southeastern United States, but it may be declining in parts of its range (Whitaker et al., 2002; Whitaker and Gummer, 2003). Evening bats are known to roost in large numbers in man-made structures (Watkins, 1969; Watkins and Shump, 1981; Bain and Humphrey, 1986; Wilkinson, 1992) and tree cavities (Wilk-inson, 1992; Bowles et al., 1996; Menzel et al., 1999, 2001a; Boyles et al., 2003; Boyles and Robbins, in press). Evening bats roost mainly in cavities in trees of various stages of decay (Bowles et al., 1996; Boyles and Robbins, in press), but relatively little is known about the formation or microclimate of the roosts. During the summer, evening bats roost mainly in large dead trees, but during the winter, live trees are commonly used (Boyles and Robbins, in press). They have a short wingspan and high wing loading so it has been predicted that they are not highly maneuverable relative to other bat species (Norberg and Rayner, 1987). Due to relatively inefficient flight, this species may avoid long foraging trips (Norberg and Rayner, 1987) and it has been suggested that evening bats forage close to their roosting areas  Duchamp et al., 2004). Evening bats feed heavily on coleopterans, homopterans, and hemipterans (Whitaker, 2004). As with other tree roosting bat species, they spend over half of its time each day in roosts (Brigham et al., 1997), so conserving roosts is important in managing this species. Evening bats are commonly referred to as migratory in middle latitudes (Jones et al., 1967; Humphrey and Cope, 1968; Watkins, 1969; Wilkinson, 1992; Sparks et al., 1999; Geluso et al., 2004), but it appears that the population discussed herein is largely non-migratory (Boyles et al., 2003; Boyles and Robbins, in press) so roost trees refer to trees used throughout the year.

It was predicted that evening bats would prefer roosting in burned forests because of an increase in the density of dead and dying trees, an increase in light penetration and an overall decrease in overstory and understory tree  ensity. These characteristics should benefit forest-dwelling bats and promote use of burned forests as roosting habitat. This study compliments previous work (Boyles and Robbins, in press), which reports the characteristics of roost trees and the surrounding habitat used by this population of evening bats during both the summer and winter. The results presented herein focus on roost selection at the forest stand-level.

2. Methods

2.1. Study area

We conducted this research on the Drury Conservation Area (DCA) in Taney County, Missouri (UTM 40.47.000N, 4.93.000E). DCA is a 1200 ha area located in extreme southwestern Missouri in the Ozark Mountains Region and is bordered on two sides by Bull Shoals Lake. It is actively managed by the Missouri Department of Conservation, which has implemented prescribed burning on approximately 55% of the potential area available for roosting habitat in an attempt to restore historic glades and oak-hickory woodlands and reduce red cedar (Juniperus virginiana) concentrations. Burning was initiated in 1999 after nearly 50 years of fire suppression and the area was then burned on a biennial schedule. All burning was conducted in March or April. The initial 1999 burn was probably more intense than subsequent burns because of accumulated fuels; therefore more overstory tree mortality may have occurred during or as a result of the initial burn (Aubrey, 2004).

Approximately 60% of DCA is dominated by oak-hickory forest with the remainder of the area being glades, wildlife food plots, ponds, and riparian areas (Missouri Department of Conservation, 1991). Elevation ranges from 185 to 335 m on DCA. Several gravel roads facilitate access to the interior of much of the forest and one large gravel road serves as the firebreak between burned and unburned portions of the forest. The canopy of the area consists almost entirely of deciduous trees from the white and red oak groups (Quercus spp.), hickories (Carya spp.), elms (Ulmus spp.), and ashes (Fraxinus spp.).

2.2. Location of roosting sites

Bats were captured from March 2003 to March 2004 using mist nets (Avinet, Dryden, NY, USA) of various lengths (6, 9, 12, or 18 m) placed across ponds or forest roads. The majority of netting sites were on the gravel road that acts as a firebreak between the burned and unburned forest, but one pond and two roads in the burned forest and one pond and one creek bed in the unburned forest were also netted. Approximately 55% of the available roosting area for evening bats was located in an area treated with prescribed fire. Thus, if evening bats were roosting at random, we would expect 55% of roost trees to be located in burned forest and 45% of roost trees to be located in unburned forest. Twenty-three evening bats were fitted with radio- transmitters (0.52 g; Model LB-2N, Holohil, Carp, Canada) by clipping fur to the skin in the interscapular region and affixing the transmitter with surgical adhesive (Skin Bond, Smith and Nephew Inc., Largo, FL, USA). The range of weights of bats fitted with transmitters was 7.5-13.5 g; therefore the transmitter represented 3.9-6.9% of the individual’s body mass.

Each bat’s roost tree was located every day following attachment of the transmitter and tracking was discontinued when the transmitter expired, was shed by the bat, or the bat remained outside of the study area for more than 5 days. This study was conducted year-round so roost trees include those used during both summer and winter by males and females in all reproductive classes (pregnant, lactating, post-lactating, and non-reproductive). All animal handling methods follow guidelines of the American Society of Mammalogists (Animal Care and Use Committee, 1998).

2.3. Canopy light penetration sampling

Leaf area index (LAI) was used as a measure of canopy light penetration to determine if three biennial prescribed burns had resulted in a more open canopy. LAI is an estimate of the ratio of overstory leaf area relative to ground area. A LAI of 0 indicates complete canopy light penetration whereas a LAI of 12 indicates no canopy light penetration (Hyer and Goetz, 2004). Here we define canopy light penetration (CLP) as: 12 LAI.

Two blocks each containing three transects in burned and unburned forests were monitored throughout the 2003 growing season. Unburned and burned forests were adjacent to one another with a gravel road acting as a fire buffer between them. LAI was obtained indirectly at each transect using an AccuPAR PAR-80 light interception device (Decagon Devices Inc., Pullman, WA, USA). Individual light measurements were collected 1.2 m above ground-level at five randomly spaced points along each transect. These five measurements were then averaged to calculate one LAI value, and therefore, one CLP value for each transect per sample period. Measurements were collected monthly throughout the 2003 growing season (April through September). The majority of canopy trees were deciduous species; thus, CLP was high in both burned and unburned forests during winter and was not measured from October through March.

2.4. Tree density

Canopy tree density was estimated in 0.05 ha circular plots centered on 25 randomly selected trees in each of the burn treatments. A random number generator was used to select UTM coordinates for the trees. Coordinates were located using a GPS unit (eTrex, Garmin International Inc., Olathe, Kansas) and selected the tree nearest that point to serve as the center of the plot. All trees in the plot greater than 10 cm diameter at breast height (dbh) were considered overstory trees and were counted and classified as either live or dead.

2.5. Statistical analysis

A binomial probability distribution was used to determine if bats roosted at random or at a proportion different from what would be expected given the amount of burned and unburned forest available. The effect of prescribed fire on CLP was assessed using a multi-factorial analysis of variance (ANOVA). The experiment was a nested block design with repeated measures. Month (April-September) and habitat (burned forest and unburned forest) were treated as fixed factors. Block (n = 2) and transect (n = 12) were treated as random factors, with transect nested under treatment. Mean separations were performed using a post hoc Tukey test. Tree density was analyzed using a t-test. In addition to absolute densities, the arcsine of the proportion of dead trees compared to total trees was calculated and analyzed using a t-test (Zar, 1984). All statistical analyses were conducted in Minitab 14. Alpha is 0.05 for all analyses.

3. Results

3.1. Roost tree location

Fig. 1. Seasonal trends (mean S.E.) in canopy light penetration (CLP) calculated as 12 leaf area index in burned and unburned forest on Drury Sixty-three roost trees were used by 23 (11 females and 12 males) evening bats from 9 March 2003 to 31 March 2004. All 63 roost trees were located in the portion of the area that was subjected to prescribed burning, although many of the bats were captured on the road that serves as the break between burned and unburned forest or within 200 m of that road. Bats roosted in the burned forest exclusively and significantly more often than expected if they selected burned or unburned forest randomly (P < 0.001). In addition, nearly all the trees used as roosts were located more than 50 m from any forest edge (62 of 63). The one exception was a large white oak Quercus alba (L.) used by a male evening bat in October that was located at the forest edge less than 50 m from where that individual was captured. This tree was not used the first day after capture; it was used on the second and fourth days, so it appears that this tree was actively selected and was not used in response to stress caused by handling.

3.2. Canopy light penetration

Both the main effects of treatment (P < 0.001) and month (P < 0.001) as well as their interaction (P < 0.001) signifi-cantly affected CLP. Maximum leaf expansion occurred in May for the burned forest and July in the unburned forest. Averaged over the 6-month sample period, canopy light penetration was significantly greater in burned forest as compared to unburned forest (P < 0.001). CLP was high in both the burned and unburned forests in April (Fig. 1). Pairwise comparisons suggest that there was no difference in canopy light penetration between burned and unburned forests in April or May but differences were significant throughout the rest of the growing season (June-September, P < 0.05 in all comparisons). As expected, measurements collected prior to leaf expansion (April) suggest that there was no difference in CLP between the burned and unburned forests; therefore, there was likely no difference in light availability between the treatments during the winter.

3.3. Tree density

Mean overstory tree density per hectare was significantly higher in unburned forest (612.0 25.0) relative to burned Conservation Area, Taney County, Missouri during the 2003 growing season. CLP, as defined herein, is measured on a scale of 0-12, with 12 indicating complete light penetration and 0 indicating no light penetration. All measurements were collected mid-month. Asterisks indicate significant differences (at alpha = 0.05) between burned and unburned forests in each month.

forest (513.6 24.2; t = 2.83, P = 0.007) and live tree density was significantly higher in unburned forest (576.0 24.6) than burned forest (468.0 24.1; t = 3.13, P = 0.003). The density of dead trees was not significantly different between the burned (45.6 7.3) and unburned forests (36.0 4.3; t = 1.14, P = 0.263). However, the proportion of dead trees compared to total trees was significantly higher in burned forest (0.092 0.014) relative to unburned forest (0.060 0.007; t = 2.08, P = 0.045).

4. Discussion

In our study area, evening bats showed a strong preference for forests managed with prescribed fires. Although there are no data on roost tree selection by evening bats before prescribed burning was initiated, it is likely that the forest characteristics were similar between what are now burned and unburned forests (Aubrey, 2004). All roost trees used by both males and females throughout the year were in the burned portion of the study area. For at least a few years after the initial burn, prescribed fire enhances roosting habitat for evening bats in several ways. First, burning increases the abundance of dead trees in some forests (Burns, 1955; Paulsell, 1957; Huddle and Pallardy, 1996; Peterson and Reich, 2001; Fule et al., 2004) and therefore increases the number of tree cavities formed by decay. We did not directly test the effect of fire on overstory tree mortality; however, dead trees were found in higher densities and at significantly higher proportions in the burned forest. Initial burns with high fuel accumulation can result in the formation of dead trees (Paulsell, 1957; Scowcroft, 1966; Anderson and Brown, 1986; White, 1986; Peterson and Reich, 2001) and leads to a forest with a large number of trees in various stages of decay. Ambient temperatures vary widely during the winter in southwestern Missouri so a large number of trees in varying stages of decay may provide sufficient options for evening bats to meet thermoregulatory needs simply by roost-switching. It has also been suggested that fire-scars allow entry for fungal pathogens, which can lead to heartrot and possibly mortality, which should promote cavity formation (Smith and Sutherland, 2001; Parsons et al., 2003). This population of evening bats is known to roost in trees in all decay stages (Boyles and Robbins, in press) and frequent burning should increase options for roost tree selection.

Second, CLP is significantly higher in burned forest than the unburned forest during the growing season. Although evening bats are not known to preferentially roost on forest edges, many tree-dwelling bats are thought to select roost trees that receive high levels of sun-exposure during the summer months (Kurta et al., 1993); therefore, trees selected as roosts are commonly taller than the canopy (Vonhof and Barclay, 1996; Britzke et al., 2003), near forest edges (Grindal, 1999) or areas with an open canopy (Menzel et al., 2001a). High light exposure is known to improve roost suitability (Kurta et al., 1993; Brigham et al., 1997) so high light penetration into the interior forest will allow bats to roost in trees that would not receive adequate sunexposure to facilitate thermoregulation in unburned forests. The ability to roost in interior forest trees that would not receive adequate sun-exposure in unburned forest may yield many benefits for evening bats. For example, the opportunity to use interior forest trees as roosts increases the availability of suitable roost trees in the habitat relative to populations that are forced to roost near forest edges or openings. This increase in suitable roost trees may allow bats to roost closer to their preferred foraging area and therefore reduce commuting distance and energy expenditure. Predation by meso-carnivores (Sparks et al., 2003) should also be lessened because the roosting sites are not concentrated in small forest patches or along forest edges, which may be easily found by and accessible to predators. An abundance of suitable roost trees may also facilitate frequent roost switching, which in turn may lessen ecto-parasite loads (Lewis, 1995).

Finally, prescribed burning may improve foraging habitat, thereby encouraging bats to roost in the vicinity. High fire frequency of low to moderate intensity prevents the regeneration of a sapling layer if preceded by a high intensity fire that removes this layer (Peterson and Reich, 2001). Exclusion of this stratum will reduce obstructions and make navigation easier in burned forest compared to unburned forest (Boyles and Robbins, in press). Evening bats forage in open areas and wooded habitats (Duchamp et al., 2004) and it has been noted that activity is higher for some bat communities in thinned forest compared to unthinned forest (Humes et al., 1999). It has been predicted that evening bats are not highly maneuverable and have relatively inefficient flight (Norberg and Rayner, 1987). This may prohibit evening bats from roosting in unburned areas because of dense understory that would make straight flight difficult. Other studies have also noted a preference for roosting in areas with a more open understory (Castleberry et al., 2005). Previous work has suggested that evening bats forage close to their roosting areas (Duchamp et al., 2004); therefore roost selection in burned forest may simply be an artifact of selection of burned habitat for foraging. In addition, the major food source of evening bats, coleopterans (Whitaker, 2004), have been found at both higher abundances and greater species richness in burned forests than unburned forests (Saint-Germain et al., 2004). Although it is difficult to distinguish if roosting habitat or foraging habitat is more important in observed habitat selection, anecdotally there does appear to be a close association between the two in this population of evening bats.

It is unlikely that any one of these reasons alone can adequately explain the preferential habitat selection seen in evening bats. Combinations of any or all of these broad forest characteristics may contribute to suitable habitat for this species and each may be of fluctuating importance throughout the year. For example, the difference in light penetration between burned and unburned forest is probably only important during the summer months. Because of the deciduous nature of the majority of the trees in southwestern Missouri, the difference in light penetration between burned and unburned forest will be lessened during the winter. Therefore, it is likely that evening bats roosted exclusively on the burned portion of the area during the winter because of prey availability, lessened clutter, or simple roost fidelity. However, it should be noted that many tree-dwelling bat species in the eastern United States only roost in trees during the summer months, so this may be inconsequential in reference to those species.

5. Conclusions

Prescribed burning has the potential to provide habitat for evening bats and possibly other species and should be taken into consideration when constructing management plans for these species. Fire may initially increase both the quantity and quality of roosting habitat for evening bats by creating an influx of dead and dying trees as well as facilitating disease and decay in live trees. This study deals only with the first 6 years following the initial burn, and long-term effects of burning may not be as beneficial to evening bats. For example, frequent burning may lead to a decline in roost trees by felling suitable snags and creating a canopy of fire-tolerant species that are not generally killed by fire (Curtis, 1959; Pyne, 1982; Crow, 1988; Ladd, 1991). However, anecdotal evidence suggests that snags persist through these low intensity burns, possibly because leaf litter fails to accumulate underneath a snag (personal observation). The same type of anecdotal evidence exists in southeastern long-leaf pine systems where biennial fire leaves snags intact because of reduced fuel accumulation near the base of the tree (S. Castleberry, University of Georgia, personal communication). To our knowledge, no studies have addressed the fate or residence time for snags suitable for roosting in a frequently burned system. Fire frequency should not remain a static part of management strategy. If biennial burns are continued, eventually the only trees left in the canopy will be fire-tolerant. It is  necessary to withhold fire from areas to allow a sufficient fuel source to accumulate, which will allow for highintensity fires that can cause canopy tree mortality and create snags. However, benefits of managing with prescribed fire may surpass those of mechanical thinning techniques because of the random and delayed mortality imposed on canopy trees (Loomis, 1974). Furthermore, prescribed burning is a more cost-effective management tool than mechanical methods (Wade and Lunsford, 1988; Dubois et al., 1999).

This study demonstrates the preference of evening bats to periodically burned forest, but fire can also be beneficial to other bat species that utilize trees as roosts. For example, the Indiana bat (Myotis sodalis), listed as federally endangered in the United States, is known to use a large number of dead and dying trees as roosts because these trees often offer the exfoliating bark that this species uses as a roost-site (Kurta et al., 1993). Through mortality and damage to bark, fire will increase the abundance of trees with exfoliating bark (Burns, 1955; Paulsell, 1957; Smith and Sutherland, 1999) and thereby increase available roosting habitat for species such as the Indiana bat. Other bat species could benefit from fire as a management tool and it should be investigated in other regions with other bat species.

Fire does have the potential to negatively affect forest dwelling bat habitat. For species that preferentially roost in trees in late stages of decay, frequent fire may destroy roost trees before they become suitable for roosting. Frequent fire may also cause the loss of winter habitat for litter-roosting species such as L. borealis (Moorman et al., 1999; Boyles et al., 2003) by continually removing accumulated leaf litter from the forest floor. Long-term studies are necessary to understand the patterns and processes of snag residence time in a periodically burned forest. Resource managers should consider the different habitat needs and life history traits of both tree-dwelling and litter roosting bats when creating management plans. Furthermore, forests respond differently according to the season when burning occurs. For example, summer burns have been shown to cause increased overstory tree mortality relative to burns conducted in other seasons (van Lear and Waldrop, 1991; Brose and van Lear, 1997). Fires in this season would therefore benefit tree-dwelling species as well as allow leaf litter to accumulate in fall benefiting litter-roosting species. More research examining the effect of burn frequency and season of burning on a variety of taxa is necessary to properly manage forests for biodiversity.

Acknowledgements

Funding for this project was provided in part by the Missouri Department of Conservation, the City of Springfield, MO, Dickerson Park Zoo, and Southwest Missouri State University. We would like to thank L. Robbins and D.A. Wait for assisting in obtaining funding and providing equipment for this project. The Missouri Department of Conservation and the Bull Shoals Field Station provided vehicles, equipment, and access to their property. J. Timpone, M. Milam, B. Mormann, P. Brown, and many others assisted in capturing and tracking bats and conducting vegetation analysis. M. Milam, M. McKnight, M. Coleman, J. Orrock, and anonymous reviewers provided many helpful comments on earlier versions of this manuscript.

References

Abrams, M.D., 1985. Fire history of oak gallery forests in northeast Kansas tallgrass prairie. Am. Midl. Nat. 114, 188-191.

Anderson, R.C., Brown, L.E., 1986. Stability and instability in plant communities following fire. Am. J. Bot. 73, 364-368.

Animal Care and Use Committee, 1998. Guidelines for the capture, handling, and care of mammals as approved by The American Society of Mammalogists. J. Mamm. 79, 1416-1431.

Arthur, M.A., Paratley, R.D., Blakenship, B.A., 1998. Single and repeated fires affect survival and regeneration of woody and herbaceous species in an oakpine forest. J. Torrey Bot. Soc. 125, 225-236.

Aubrey, D.A., 2004. Savanna restoration through prescribed fire: demographic and physiological responses of oak and hickory seedlings and saplings to a changing light environment. Master’s Thesis, Southwest Missouri State University.

Bain, J.R., Humphrey, S.R., 1986. Social organization and biased primary sex ratio of the evening bat Nycticeius humeralis. Fla. Sci. 49, 22-31.

Barnes, T.A., van Lear, D.H., 1998. Prescribed fire effects on advanced regeneration in mixed hardwood stands. South. J. Appl. For. 22, 138-142.

Blake, J.G., 2005. Effects of prescribed burning on distribution and abundance of birds in a closed-canopy oak-dominated forest, Missouri, USA. Biol. Conserv. 121, 519-531.

Bowles, J.B., Howell, D., Van Zee, J.W., Wilson, G.M., 1996. Use of alternate roost trees by the evening bat, Nycticeius humeralis, in Iowa. In: Genoways, H.H., Baker, R.J. (Eds.), Contributions in Mammalogy: A Memorial Volume Honoring Dr. J. Knox Jones, Jr. Museum of Texas Tech University, Lubbock, Texas, pp. 217-224.

Boyles, J.G., Timpone, J.C., Robbins, L.W., 2003. Late-winter observations of red bats, Lasiurus borealis, and evening bats, Nycticeius humeralis, in Missouri. Bat Res. News 44, 59-61.

Boyles, J.G., Robbins, L.W. Characteristics of summer and winter roost trees used by evening bats (Nycticeius humeralis) in southwestern Missouri. Am. Midl. Nat., in press.

Brigham, R.M., Vonhof, M.J., Barclay, R.M.R., Gwilliam, J.C., 1997. Roosting behavior and roost-site preferences of forest-dwelling California bats (Myotis californicus). J. Mamm. 78, 1231-1239.

Britzke, E.R., Harvey, M.J., Loeb, S.C., 2003. Indiana bat, Myotis sodalis, maternity roosts in the southern United States. Southeast. Nat. 2, 235 242.

Brose, P.H., van Lear, D.H., 1997. Effects of seasonal prescribed fires on density of hardwood advanced regeneration in oak-dominated shelterwood stands.

In: Proceedings of the 25th Annual Hardwood Symposium, ‘‘25 Years of Hardwood Silviculture: A Look Back and a Look Ahead’’, 7-10 May, 1997, Cashiers, NC. National Hardwood Lumber Association, Memphis, TN, pp.
139-146.

Burns, P.Y., 1955. Fire scars and decay in Missouri oaks. University of Missouri College of Agriculture, Agricultural Experiment Station Bulletin 642.

Campbell, L.A., Hallett, J.G., O’Connell, M.A., 1996. Conservation of bats in managed forests: use of roosts by Lasionycteris noctivagans. J. Mamm. 77, 976-984.

Carter, T.C., Ford, W.M., Menzel, M.A., 2002. Fire and bats in the Southeast and Mid-Atlantic: more questions than answers? In: Ford, W.M., Russell, K.R., Moorman, C.E. (Eds.), The Role of Fire for Nongame Wildlife Management and Community Restoration: Traditional Uses and New Directions. United States Department of Agriculture, General Technical Report NE-288. pp. 139-143.

Castleberry, S., Miles, A., Miller, D., Conner, L., 2005. Habitat ecology of bats at multiple scales in natural longleaf and intensively managed loblolly pine forests in Georgia. In: Abstract from 90th Annual Ecological Society of America Meeting, Montreal, Canada.

Chambers, C.L., Alm, V., Sider, M.S., Rabe, M.J., 2002. Use of artificial roosts by forest-dwelling bats in northern Arizona. Wildl. Soc. Bull. 30, 1085-1091.

Cottam, G., 1949. The phytosociology of an oak woods in south-western Wisconsin. Ecology 30, 271-287.

Crow, T.R., 1988. Reproductive mode and mechanisms for self-replacement of northern red oak (Quercus rubra): a review. Forest Sci. 34, 19-40.

Crow, T.R., Johnson, W.C., Adkisson, C.S., 1994. Fire and recruitment of Quercus in a postagricultural field. Am. Midl. Nat. 131, 84-97.

Curtis, J.T., 1959. The vegetation of Wisconsin. University of Wisconsin Press, Madison.

Dubois, M.R., Straka, T.J., Crim, S.D., Robinson, L.J., 1999. Costs and cost trends for forestry practices in the South. Forest Landowner 50, 28 32.

Duchamp, J.E., Sparks, D.W., Whitaker Jr., J.O., 2004. Foraging-habitat selection by bats at an urban-rural interface: comparison between a successful and less successful species. Can. J. Zool. 82, 1157-1164.

Elmore, L.W., Miller, D.A., Vilella, F.J., 2004. Selection of diurnal roosts by red bats (Lasiurus borealis) in an intensively managed pine forest in Mississippi. Forest Ecol. Manage. 199, 11-20.

Fenton, M.B., 1997. Science and the conservation of bats. J. Mamm. 78, 1-14.

Fule, P.Z., Cocke, A.E., Heinlein, T.A., Covington, W.W., 2004. Effects of an intense prescribed forest fire: is it ecological restoration. Restor. Ecol. 12, 220-230.

Geluso, K.N., Benedict, R.A., Kock, F.L., 2004. Seasonal activity and reproduction in bats of east-central Nebraska. Trans. Nebraska Acad. Sci. 29, 33 44.

Grindal, S.D., 1999. Habitat use by bats, Myotis spp., in western Newfoundland. Can. Field-Nat. 113, 258-263.

Hartman, G.W., Heumann, B.H., 2003. Prescribed fire effects in the Ozarks of Missouri: the Chilton Creek project 1996-2001. In: Proceedings of the Second International Wildland Fire Ecology and Fire Management Congress, 16-22 November, 2003, Orlando, FL.

Huddle, J.A., Pallardy, S.G., 1996. Effects of long-term annual and prescribed burning on tree survival and growth in a Missouri Ozark oak-hickory forest. Forest Ecol. Manage. 82, 1-9.

Humes, M.L., Hayes, J.P., Collopy, M.W., 1999. Bat activity in thinned, unthinned, and old-growth forests in western Oregon. J. Wildl. Manage. 63, 553-561.

Humphrey, S.R., Cope, J.B., 1968. Records of migration of the evening bat Nycticeius humeralis. J. Mamm. 49, 329.

Hyer, E.J., Goetz, S.J., 2004. Comparison and sensitivity analysis of instruments and radiometric methods for LAI estimation: assessments from a boreal forest site. Agric. Forest Meteorol. 122, 157-174.

Jones Jr., J.K., Fleharty, E.D., Dunnigan, P.B., 1967. The distributional status of bats in Kansas. Misc. Publ. Mus. Nat. Hist. Univ. Kansas 46, 1-33.

Kunz, T.H., 1982. Roosting ecology of bats. In: Kunz, T.H. (Ed.), Ecology of Bats. Plenum Press, New York, pp. 1-55.

Kurta, A., King, D., Teramino, J.A., Stribley, J.M., Williams, K.J., 1993. Summer roosts of the endangered Indiana bat (Myotis sodalis) on the northern edge of its range. Am. Midl. Nat. 129, 132-138.

Ladd, D., 1991. Reexamination of the role of fire in Missouri oak woodlands. In: Buger, G.V., Ebinger, J.E., Wilhelm, G.S. (Eds.), Proceedings of the Oak Woods Management Workshop, Eastern Illinois University, Charleston, pp.
67-80.

Lewis, S.E., 1995. Roost fidelity of bats—a review. J. Mamm. 76, 481-496.

Loomis, R.M., 1974. Predicting the losses in sawtimber volume and quality from fires in oak hickory forests. USDA Forest Service Research Paper NC- 104.

Lorimer, C.G., 1985. The role of fire in the perpetuation of oak forests. In: Johnson, J.E. (Ed.), Challenges in Oak Management and Utilization. Cooperative Extension Service, University of Wisconsin, Madison, pp. 8-25.

Mazurek, M.J., Zielinski, W.J., 2004. Individual legacy trees influence vertebrate wildlife diversity in commercial forests. Forest Ecol. Manage. 193, 321-334.

McCarty, K., 1998. Landscape-scale restoration in Missouri savannas and woodlands. Restor. Manage. Notes 16, 22-32.

McLeod, R.F., Gates, J.E., 1998. Response of herpetofaunal communities to forest cutting and burning at Chesapeake farms, Maryland. Am. Midl. Nat.  139, 164-177.

Menzel, M.A., Krishon, D.M., Carter, T.C., Laerm, J., 1999. Notes on tree roost characteristics of the northern yellow bat (Lasiurus intermedius), the Seminole bat (L. seminolus), the evening bat (Nycticeius humeralis), and the eastern pipistrelle (Pipistrellus subflavus) Fla. Sci. 62, 185-193.

Menzel, M.A., Carter, T.C., Ford, W.M., Chapman, B.R., 2001a. Tree roost characteristics of subadult and female adult evening bats (Nycticeius humeralis) in the upper coastal plain of South Carolina. Am. Midl. Nat. 145, 112-119.

Menzel, M.A., Menzel, J.M., Carter, T.C., Ford, W.M., Edwards, J.W., 2001b. Review of the forest habitat relationships of the Indiana bat (Myotis sodalis).United States Department of Agriculture General Technical Report NE 284.

Menzel, M.A., Owen, S.F., Ford, W.M., Edwards, J.W., Wood, P.B., Chapman, B.R., Miller, K.V., 2002. Roost tree selection by northern long-eared bat (Myotis septentrionalis) maternity colonies in an industrial forest of the
central Appalachian mountains. Forest Ecol. Manage. 155, 107 114.

Miller, D.A., 2003. Species diversity, reproduction, and sex ratios of bats in managed pine forest landscapes of Mississippi. Southeast. Nat. 2, 59 72. Missouri Department of Conservation, 1991. Natural features inventory of Drury-Mincy Wildlife Area.

Moorman, C.E., Russell, K.R., Menzel, M.A., Lohr, S.M., Ellenberger, J.E., Van Lear, D.H., 1999. Bats roosting in deciduous leaf litter. Bat Res. News 40,  74-75.

Moser, W.K., Ducey, M.J., Ashton, P.M.S., 1996. Effects of fire intensity on competitive dynamics between red and black oaks and mountain laurel. North. J. Appl. For. 13, 119-123.

Norberg, U.M., Rayner, J.M.V., 1987. Ecological morphology and flight in bats (Mammalia; Chiroptera): wing adaptations, flight performance, foraging strategy and echolocation. Philos. Trans. R. Soc. London B: Biol. Sci. 316, 335-427.

Nuzzo, V., 1986. Extent and status of midwest oak savanna: presettlement and 1985. Nat. Areas J. 6, 6-36.

Parsons, S., Lewis, K.J., Psyllakis, J.M., 2003. Relationships between roosting habitat of bats and decay of aspen in the sub-boreal forests of British Columbia. Forest Ecol. Manage. 177, 559-570.

Patriquin, K.J., Barclay, R.M.R., 2003. Foraging by bats in cleared, thinned and unharvested boreal forest. J. Appl. Ecol. 40, 646-657.

Paulsell, L.K., 1957. Effects of burning on Ozark hardwood timberlands. University of Missouri College of Agriculture, Agricultural Experimental Station Bulletin 640.

Peterson, D.W., Reich, P.B., 2001. Prescribed fire in oak savanna: fire frequency effects on stand structure and dynamics. Ecol. Appl. 11, 914 927.

Pyne, S.J., 1982. Fire in America. Princeton University Press, Princeton, NJ. Rodrigue, J.L., Schuler, T.M., Menzel, M.A., 2001. Observations of bat activity during prescribed burning in West Virginia. Bat Res. News 42, 48-49.

Saint-Germain, M., Drapeau, P., Hebert, C., 2004. Comparison of coleoptera assemblages from a recently burned and unburned black spruce forests of northeastern North America. Biol. Conserv. 118, 583-592.

Schurbon, J.M., Fauth, J.E., 2003. Effects of prescribed burning on amphibian diversity in a southeastern U.S. National Forest. Conserv. Biol. 17, 1338 1349.

Scowcroft, P.G., 1966. The effects of fire on hardwood forests of the Missouri Ozarks. M.S. Thesis, University of Missouri Columbia, MO, pp. 126.

Simon, N.P.P., Stratton, C.B., Forbes, G.J., Schwab, F.E., 2002. Similarity of small mammal abundance in post-fire and clearcut forests. Forest Ecol. Manage. 165, 163-172.

Smith, K.T., Sutherland, E.K., 1999. Fire-scar formation and compartmentalization in oak. Can. J. Forest Res. 29, 166-171.

Smith, K.T., Sutherland, E.K., 2001. Terminology and biology of fire scars in selected central hardwoods. Tree-Ring Res. 57, 141-147.

Sparks, D.W., Choate, J.R., Winn, R.J., 1999. Observations on reproduction in three species of bats. Prairie Nat. 31, 245-248.

Sparks, D.W., Simmons, M.T., Gummer, C.L., Duchamp, J.E., 2003. Disturbance of roosting bats by woodpeckers and raccoons. Northeast. Nat. 10, 105-108.

Stearns, F., 1991. Oak woods: an overview. In: Buger, G.V., Ebinger, J.E., Wilhelm, G.S. (Eds.), Proceedings of the Oak Woods Management Workshop, Eastern Illinois University, Charleston, pp. 1-7.

Tibbels, A.E., Kurta, A., 2003. Bat activity is low in thinned and unthinned stands of red pine. Can. J. Forest Res. 33, 2436-2442.

van Lear, D.H., 2002. Upland oak ecology and management. In: Spetich, M.A. (Ed.), Upland Oak Ecology  Symposium: History, Current Conditions, and Sustainability.

van Lear, D.H., Waldrop, T.A., 1991. Prescribed burning for regeneration. In: Duryea, M.L., Dougherty, P.M. (Eds.), Forest Regeneration Manual. Kluwer Academic Publishers, The Netherlands, pp. 235-250.

Vonhof, M.J., Barclay, R.M.R., 1996. Roost-site selection and roosting ecology of forest-dwelling bats in southern British Columbia. Can. J. Zool. 74, 1797-1805.

Wade, D.D., Lunsford, J.D. 1988. A guide for prescribed fire in southern forests. Revised Edition. United States Department of Agriculture, Forest Service, TechnicalPublicationR8-TP 11,Southern Region. Atlanta,Georgia, pp. 56.
Watkins, L.C., 1969. Observations on the distribution and natural history of the evening bat (Nycticeius humeralis) in northwestern Missouri and adjacent Iowa. Trans. Kansas Acad. Sci. 72, 330-336.

Watkins, L.C., Shump Jr., K.A., 1981. Behavior of the evening bat Nycticeius humeralis at a nursery roost. Am. Midl. Nat. 105, 258-268.

Whitaker Jr., J.O., 2004. Prey selection in a temperate zone insectivorous bat community. J. Mamm. 85, 460-469.

Whitaker Jr., J.O., Gummer, S.L., 2003. Current status of the evening, Nycticeius humeralis, in Indiana. Proc. Indiana Acad. Sci. 112, 55- 60.

Whitaker Jr., J.O., Brack Jr., V., Cope, J.B., 2002. Are bats in Indiana declining? Proc. Indiana Acad. Sci. 111, 95-106.

White, A.S., 1986. Prescribed burning for oak savanna restoration in central Minnesota. USDA Forest Service, North Central Forest Experiment Station; Research Paper NC-266.

Wilkinson, G.S., 1992. Information transfer at evening bat colonies. Anim. Behav. 44, 501-518.

Zar, J.H., 1984. Biostatistical Analysis, 2nd ed. Prentice-Hall, Englewood Cliffs, NJ.

Sparks, Jeffrey C.¹, Masters, Ronald E.^1*, Engle, David M.²,
Palmer, Michael W.³ & Bukenhofer, George A.⁴

¹Department of Forestry, ²Department of Agronomy, ³Department of Botany, Oklahoma State University, Stillwater,
OK 74078, USA; ⁴U.S. Forest Service, Ouachita National Forest, Heavener, OK 74937, USA;
^*Corresponding author; Tel.+1 405 744 6432; Fax +1 405 744 9693; E-mail: rmaster@okway.okstate.edu

Abstract. We compared the effects of late dormant-season and late growing-season prescribed fires on herbaceous species in restored shortleaf pine- (Pinus echinata) grassland communities in the Ouachita Highlands of western Arkansas.Herbaceous species richness, diversity, and total forb and legume abundance increased following fire. Late growing season burns reduced distribution and abundance of panicums (primarily Panicum boscii, P. dichotomum, and P. lineari folium) while late dormant-season burns increased Panicum distribution and abundance. Density of legumes (such as Stylosanthes biflora) increased following frequent or annual dormant-season fires. However, season of fire influenced the distribution and abundance of fewer than 10 % of the species. Fire plays an essential role in pine-grassland communities by creating and maintaining open canopy conditions that perpetuate understory herbaceous plant communities.

Keywords: Arkansas; Fire ecology; Fire frequency; Fire season; Ouachita Mountains; Restoration ecology.

NomenclatFure: Smith (1988).

CLICK HERE TO DOWNLOAD AS PDF

Introduction

Fire played an important role in shaping formerly abundant pine- (Pinus spp.) grassland communities in the southeastern United States (Buckner 1989; Platt et al. 1988; Waldrop et al. 1992; Masters et al. 1995). Historical accounts before settlement describe these pine-grassland communities as open ‘park-like’ pine stands with a distinct grass-dominated herbaceous layer and recurrent woody layer, depending on fire frequency (James 1823; Featherstonhaugh 1844; Komarek 1965; Nuttall 1980; Waldrop et al. 1992; Masters et al. 1995). The accumulation of herbaceous material provided adequate fuels for frequent fires of aboriginal and lightning origin which maintained the open structure of these pine-grassland communities (Komarek 1965; Buckner 1989; Foti & Glenn 1991; Waldrop et al. 1992; Masterset al. 1995).

Similar to other forest communities of the World, settlement in the southeastern United States (18th to mid-19th century) altered these landscapes by removing or changing much of the natural vegetation, resulting in fragmented and dissected landscapes (Cottam 1949; Stearns 1949; Curtis 1956; Forman & Godron 1986; Kreiter 1995). The frequency and scale of fires in the region declined after settlement because of aboriginal displacement, fragmentation of habitats causing artificial fire breaks, and fire suppression by settlers (Pyne 1982). This decline in fire frequency caused once open pine-grassland communities to become much more densely forested. Dense forests minimize light reaching the forest floor, thus reducing the herbaceous plant community, understory forage, and habitat quality for many species of wildlife (Lewis & Harshbarger 1976; Masters 1991a; Wilson et al. 1995). The endangered redockaded woodpecker (Picoides borealis), an endemic of southeastern pine forests, is one example of a species that has declined, in part, as a result of increased forest density in the southeastern United States.

The U.S. Forest Service has begun to reconstruct or restore shortleaf pine- (Pinus echinata) grassland communities to benefit both plant and wildlife species dependent on these systems. In the Ouachita National Forest of western Arkansas, the Forest Service uses a program known as Wildlife Stand Improvement (WSI) that consists of thinning midstory and codominant pine and hardwood trees to near pre-settlement basal areas. Currently, WSI treated stands are burned during the dormant season on 3-yr intervals to maintain open structure. However, recent studies in the Ouachitas suggest that the historical fire regime was one of predominantly late growing-season fires and to a lesser extent dormant season burns (Foti & Glenn 1991; Masters et al. 1995). To effectively restore this system, knowledge of the effects of both growing-season and dormant-season prescribed burns is necessary (Masters et al. 1995, 1996).

Numerous studies have compared the effects of growing-season and dormant-season fires on vegetation in Coastal Plain regions of Florida, Louisiana, and South Carolina (Grelen 1975; Hughes 1975; Lewis & Harsh-barger 1976; Platt et al. 1988; Waldrop et al. 1992; Glitzenstein et al. 1995). Masters (1991a, b) and Masters et al. (1993) described the effects of dormant season burns of varying frequency on vegetation under a variety of overstory conditions in interior highlands. Masters et al. (1996) described the effects of WSI and dormant-season burns on restored pine-bluestem communities. However, no information is available on the effects of growing-season burns in the Ouachita Mountains. Our main objective was to compare the effects of growing-season and dormant-season burns on herbaceous vegetation richness, diversity, and abundance in WSI-treated stands.

Study area

Our study focused on stands under active management for the endangered red-cockaded woodpecker within the 40 000-ha Pine-bluestem Ecosystem Renewal Area, on the Poteau Ranger District of the Ouachita National Forest (ONF) in Scott County Arkansas. The ONF lies in the 2 280 000 ha Ouachita Mixed Forest  Meadow Province and comprises 648 000 ha throughout the Ouachita Mountains in Arkansas and Oklahoma (Neal & Montague 1991; Bailey 1995). The Ouachita mountains are east-west trending, strongly dissected and range in elevation from 150 – 790 m (Fenneman 1938: 669). South-facing slopes tend to be dominated by shortleaf pine and more mesic north-facing slopes tend to be dominated by oaks (Quercus spp.), hickories (Carya spp.) and other hardwoods (Johnson 1986; Foti & Glenn 1991). Ouachita Mountain soils developed from sandstone and shales and are thin and drought prone. A semi-humid to humid climate prevails with hot summers and mild winters (Smith 1989).

Pinus echinata was the dominant overstory species in all stands (Fig. 1). Codominant and intermediate over story species included Quercus stellata, Q. marilandica, alba, Q. rubra, Q. velutina, Carya texana and C. tomentosa. Tree heights in our study stands ranged from 15 – 23 m ( x = 18.3 m; S.D. = 3.1). Canopy cover ranged  from 68 – 93 % (x = 84.1% ; S.D. = 7.5). Woody sprouts  (≤ 3 m tall) dominated the understory of these stands. The dominant understory woody species and vines included Toxicodendron radicans, Vaccinium pallidum, Quercus stellata, Carya tomentosa, Rubus spp., Parthenocissus quinquefolia, Ceanothus americanus, Vitis rotundifolia, Quercus alba and Pinus echinata (Sparks 1996).

Methods

Experimental design

Our experimental design encompassed two studies (Study 1 and Study 2) and was completely randomized. In these studies we used 12 stands (13.8 to 26.7 ha) that had been previously subjected to WSI and prescribed fire at 3-yr intervals (≥ 3 prescribed fire cycles). Overstory pine density and basal area was similar across all stands (Sparks 1996). Study 1 consisted of three treatments with four replications of each treatment (n = 12). Study 2 used the control and dormant-season fire stands from Study 1 (n = 8). Treatments are as follows:

Study 1

(1) No-burn control (CON1; n = 4);

(2) Late growing-season burn, September 1994 (GS1; n = 4);

(3) Late dormant-season burn March-April 1995 (DS1; n = 4);

Study 2

(4) Late growing-season burn, October 1995 (GS2; n = 2);

(5) Late dormant-season burn, March 1996 (DS2; n = 2);

(6) Frequent dormant-season fire, March-April 1995 and March 1996

(FDS; n = 2);

(7) Infrequent dormant-season fire, burned March-April 1995, no-
burn 1996 (IFDS; n = 2).

Study 1 and Study 2 dormant-season and growing season fire treatments differed in that prescribed burns were applied after three vs. four growing seasons, respectively, following previous dormant-season fire. Study 2 used the dormant-season fire treatments from Study 1 to determine the effects of fire frequency on the herbaceous community. In both studies, late growing-season fires were performed because of poor burning conditions (primarily fuel moisture, presence of live vegetation and high relative humidities) earlier in the season.

Vegetation sampling

We sampled herbaceous vegetation during a two week period in late July 1994 (Study 1 pre-treatment), July 1995 (Study 1 post-treatment; Study 2 pre-treatment), and July 1996 (Study 2 post-treatment). In each stand, we established 30, 1 m × 1 m permanent plots (after Oosting 1956) at 30-m intervals on two to four randomly spaced lines perpendicular to the contour (after Masters 1991a, b). To avoid bias from surrounding stands, we did not sample within 50 m of any edge (Mueller-Dombois & Ellenberg 1974: 123). For each herbaceous species, we recorded percent frequency of occurrence and stem density within plots. Percent cover for vascular plant groups and objects such as rocks, tree boles, and logs was also estimated. Voucher specimens were collected, verified and deposited in the Oklahoma State University Herbarium.

Data analysis

We calculated species richness and diversity (Shannon-Weaver H’) after Ludwig & Reynolds (1988) at the sample (m²) and stand scales. In both studies, we summarized herbaceous species by mean density and percent frequency of occurrence for each year and treatment. All plant species were classified according to plant growth form (e.g., forb, legume, grass, etc.) and season of growth (cool vs. warm). Season of growth was determined by flowering dates described by the Great Plains Flora Association (1986) with cool-season species flowering from November to mid May, and warm season species flowering from mid May through October. To account for pre-treatment differences, we determined the percent change [(post-treatment – pre-treatment / pre-treatment) × 100] in density and frequency of occurrence caused by treatments. All variables were tested for homogeneity of variance using Levene’s test (Snedecor & Cochran 1980). These tests indicated homogeneity of variances, so we tested for treatment differences in percent change using a one-way GLM in which treatment was the factor of interest (Anon. 1985).

In Study 1, we used orthogonal contrasts (burn vs noburn and growing-season fire vs. dormant-season fire) and separated treatment means (P ≤ 0.05) with the protected least significant difference test (Steel & Torrie 1980; Conover & Iman 1981).

We performed Detrended Correspondence Analysis (DCA) using CANOCO (ter Braak 1988), to analyze the species composition data. We checked the results for instability caused by a bug in the program (Oksanen & Minchin 1997). DCA is a multivariate indirect gradient analysis that uses variation in species abundance data to display species and stand locations in a two-dimensional ordination space (ter Braak 1986). DCA axes are in units of constant beta-diversity, where one unit is equal to one standard deviation of species turnover (Hill & Gauch 1980). In DCA, changes in location of a stand over time indicate corresponding changes in real or relative species composition of the stand (Wyant et al. 1991). DCA was used to analyze importance values relative density + relative frequency) to determine changes in stand composition from pre-treatment to post-treatment (after Mueller-Dombois & Ellenberg 1974; Smith 1990). We square-root transformed species abundances before analysis.

Results

Response to fire and fire season

We observed more than 150 herbaceous species during these two studies. Fewer than 10% of these species were influenced (P ≤ 0.05) by season of fire. Late dormant-season fires produced a greater frequency of occurrence of Panicum dichotomum (Study 1: F = 26.9; P = 0.0006, Study 2: F = 29.7, P = 0.0320) and Scleria triglomerata (Study 1: F = 15.3; P = 0.0035, Study 2: F = 19.9, P = 0.0467) than late growing-season fires. Density of Panicum dichotomum (F = 54.5; P = 0.0001) and Scleria triglomerata (F = 5.6; P = 0.0416) was less after late growing-season fires than after late dormant season fires in Study 1.

Although few species were influenced by season of fire, differences (P ≤ 0.05) in density and frequency of major plant categories were apparent (Tables 1 and 2). Late dormant-season fires increased panicum density (primarily Panicum boscii, P. dichotomum, and P. linearifolium) while late growing-season fires greatly reduced total panicum density (Tables 1 and 2). Panicum frequency also declined after late growing-season fires in Study 1 (Table 1). Grasses showed a tendency to decrease in percent cover following fire (Table 3), and a tendency for further decline in density following late growing-season fires (Table 2).

Table 1. Study 1, herbaceous stem density (stems/m²) and percent frequency of occurrence response to season of fire in restored pine-grassland communities on the Ouachita National Forest, Arkansas, summer 1994 and 1995.¹

Table 2. Study 2, herbaceous stem density (stems/m²) and percent frequency of occurrence response to season of fire in restored pine-grassland communities on the Ouachita National Forest, summer 1995 and 1996.¹

Table 3. Post-treatment percent cover in restored red-cockaded woodpecker clusters on the Ouachita National Forest in 1995.¹

Regardless of season, fire increased density of legumes, however legume frequency did not increase with burning (Tables 1 and 2). Legume species such as Stylosanthes biflora increased in density (F = 16.9; P = 0.0026) after fire, while other legumes such as Desmodium ciliare (F = 6.58; P = 0.0334) and Lespedeza procumbens (F = 8.37; P = 0.0179) increased in frequency of occurrence after fire. Fire also increased density and frequency of occurrence of numerous forbs such as Coreopsis tinctoria, Polygala alba, and Erechtites hieraciifolia, resulting in an increase in total forb density in Study 1 (Table 1). We found that forbs after late dormant-season fires occurred more frequently than after late growing-season fires and generally increased with fire, although it was not biologically significant (Table 1). Cover of herbaceous vegetation was similar for all treatments, but stands burned during the late dormant season had more bare ground and exposed rock (Table 3). Warm season species had lower densities in response to late growing-season burns than late dormant-season burns (Tables 1 and 2).

Response to frequent fire

Panicum frequency increased with frequent late dormant-season burns (Table 4). However, density of Chasmanthium sessiliflorum declined (F = 35.6; P = 0.0270) after frequent late dormant-season fire. Legume density was greater after frequent late dormant-season fires (Table 4). Density of Lespedeza procumbens (F = 124.0, P = 0.0080) and Stylosanthes biflora (F = 124.9; P = 0.0079) was greater after frequent late dormant-season fires. Helianthus hirsutus (F = 33.7; P = 0.0284) frequency of occurrence was also greater after frequent fire. Stand species richness in frequently burned stands remained stable, but declined in control stands (Fig. 2).

Fig. 3. a. Sample (m²) species richness by study and treatment; b. Net-change and standard errors in sample species richness by study and treatment. Means followed by different letters are different (P ≤ 0.05, Least Significant Difference). CON1 = Study 1, no-burn control; GS1 = Study 1, growing-season burn; DS1 = Study 1, dormant-season burn; GS2 = Study 2, growing-season burn; DS2 = Study 2, dormant-season burn; FDS = Study 2, frequent dormant-season burn; and IFDS = Study 2, infrequent dormant-season burn.

Table 4. Study 2, herbaceous stem density (stems/m²) and percent frequency of occurrence response to frequent fire in restored pine-grassland communities on the Ouachita National Forest, summer 1995 and 1996.¹

Community response to fire

Fire dramatically influenced community composition in restored pine-grassland stands. Species diversity when compared to unburned stands was greater (P ≤ 0.05) after both late growing-season and late dormant season prescribed fires. Stand species richness increased after both late growing-season and late dormant-season fires, while declining in unburned stands (Fig. 2b). Furthermore, post-treatment stand species richness after late dormant-season and late growing-season fires was greater than the unburned controls in Study 1 (Fig. 2a). Sample (m²) species richness increased dramatically after late dormant-season fires with net change in stand species richness being greatest after late dormant-season fires (Fig. 3b).

Detrended Correspondence Analysis illustrated the nature of change in these stands over time and in response to fire (Fig. 4). Axis 1 indicated that year-to-year variation may be the most important factor in determining species composition of these stands (Fig. 4). Axis 2 indicated that geographical location of stands also determines species composition. Axis 3 was tentatively interpreted as an indicator of stand openness with species assemblages more characteristic of prairies being grouped together versus those more characteristic of closed forest being grouped together. Axis 4 may be interpreted as a treatment axis (Fig. 4). We included Axis 4 because treatment effect was our primary interest.

Control stands shifted to the right on Axis 1 and upward on Axis 4 indicating a year and treatment effect, while late dormant-season fire stands shifted right on Axis 1 and down on Axis 4 also indicating a year and treatment effect (Fig. 4). Late growing-season fire stands shifted directly to the right on Axis 1 indicating that year effects had an overriding influence on treatment (Fig. 4). The shift in stands after treatment indicates a similar change in species composition among the treatments. Axes 1 through 4 had eigenvalues of 0.161, 0.081, 0.060, and 0.040 respectively. Together all axes account for 26.7% of the total variation in species data. With an eigenvalue of only 0.042, and the fact that the apparent ‘treatment axis’ is the 4th axis, it is obvious that the effects of treatment, while highly significant, is minor compared with year-to-year effects and site location effects.

Discussion

Treatment response

Burned stands had higher stand species richness and diversity than no-burn controls (Fig. 2). These results are similar to many studies that indicate an initial in crease in species diversity and richness following fire (Trabaud & Lepart 1980; Armour et al. 1984; Thanos et al. 1996). In Study 2, stands in the 2nd growing season since late dormant-season fire (DSC), declined in species richness, indicating that the initial increase in stand richness after fire is short lived (Figs. 2 and 3) and probably influenced by environmental conditions during a given year (Fig. 4). The majority of individual species in both studies did not respond favorably to any one treatment, but were common in all treatments. We believe this is because species present (e.g. Andropogon spp. and various legume species) in restored pine-grassland communities are well adapted to fire, and community changes are small and of short duration. Waldrop et al. (1992) noted that the pine-grassland ecosystem once common throughout the southeastern U.S. was fire derived and fire maintained. Herbaceous species in these restored pine-grassland communities were likely present in pre-settlement communities that developed under a periodically frequent fire regime during both the dormant and growing seasons (Masters et al. 1995).

Fire does not drastically alter species composition in stands with a recent history of fire. Pre-fire composition is a major factor in determining post-fire composition (Armour et al. 1984; Stickney 1986; Rego et al. 1991). Adjacent forests without WSI treatment have dense midstories minimizing light from reaching the forest floor, so species richness and abundance of herbaceous species is much less than in WSI-treated areas (Masters et al. 1996). We also suggest that post-fire species richness and composition is influenced by fire intensity, which is related to litter consumption and reduction in the stature of woody species (Masters et al. 1993; Sparks 1996).

Stand structure

Prescribed fire plays a major role in determining the vegetation structure and composition in restored pinegrassland communities (Wilson et al. 1995). Understories of stands treated with WSI are characteristically dominated by woody sprouts (> 50 000 stems/ha) that restrict light from reaching the forest floor. Late dormant-season fires in these stands on average produce greater fireline intensity than growing-season fires (1300 Kw/m versus < 300 Kw/m), and are more effective at maintaining an open forest structure by reducing the stature of woody sprouts (Sparks 1996).

The effect of a disturbance such as fire on any community or ecosystem depends on the intensity, scale, and frequency (Sousa 1984; Perry 1994; Sparks 1996). Late dormant-season fires in these stands act as more intense disturbances than late growing-season fires, by more effectively reducing stature of the woody community and reducing the litter layer. Increased light penetration due to the reduced stature of the woody understory and reduction of litter after fire provides an opportunity for new herbaceous species to become established, thereby significantly increasing species richness and diversity (Sousa 1984; Masters 1991a, b; Masters et al. 1993). But, fire in either season increases light and allows species already present to prosper, thus the increase in density and percent frequency of occurrence of forbs after fire.

Species composition

Herbaceous species actively growing at the time of a fire in grassland systems are more susceptible to injury than species that are dormant or in early stages of development (Towne & Owensby 1984). Fires during the dormant season reduce cool-season species while favoring many warm-season species (Owensby & Anderson 1967; Hover & Bragg 1981; Towne & Owensby 1984; Hulbert 1988; Howe 1994a). In contrast, growing-season fires reduce warm season species while favoring cool-season species (Hover & Bragg 1981; Ewing & Engle 1988; Biondini et al. 1989; Howe 1994a). Our results in Study 2 showed an increase in density of warm-season species after burning, Study 1 also showed an increase, but not significantly. Neither study showed an increase for cool-season species when late dormant- and late growing-season burns were compared. Growing-season burns may have increased coolseason species had our growing-season fires been conducted earlier in the growing season and before coolseason species initiated new growth. It is important to note that in both studies we attempted to burn earlier in the growing season, but burning conditions (primarily fuel moisture, presence of live vegetation and high relative humidities) were not conducive to fire until later in the growing season.

Several studies have noted that growing-season fires when compared to dormant-season fires and unburned areas increase diversity and richness by increasing the number of annuals and promoting cool-season grasses and forbs (Biondini et al. 1989; Howe 1994b). Platt et al. (1988) noted that growing-season fires produced more flowering stems than fires in other seasons. Many warm season grasses such as wiregrass (Aristida stricta) and little bluestem (Schizachrium scoparium) flower profusely after growing-season fires (Lewis 1964; Robbins & Myers 1992). Hodgkins (1958) noted that composites and legumes increase in response to growing-season fires. Our results indicate an aggressive response from legumes and forbs (Tables 1 and 2), and a larger increase in species richness after dormant-season fires (Fig. 2 and 3). Other studies have found similar results (Grelen & Lewis 1981; Landers 1981; White et al. 1991). Legumes in particular are adapted to fire and benefited because of a hard seed coat and subsequent persistence in the soil seed bank (White et al. 1991; Arianoutsou & Thanos 1996; Thanos et al. 1996).

In Study 2, legume density (primarily Amphicarpa bracteata, Clitoria mariana, Lespedeza repens, and Stylosanthes biflora) was 2 × greater after two frequent late dormant-season fires, similar to Masters et al. (1993), who found legume biomass > 4 × greater after 5 years of annual burning compared to unburned controls. White et al. (1991) found that 43 years of annual winter fires increased legumes by > 25 × over periodic summer and winter burns or annual summer burns. The lack of an increase in legume frequency in our study may be due to relatively high frequency for legumes (80 – 88 %), especially since we observed that most of these species had a tendency towards aggregation. Stem densities within aggregations or sample plots can increase without increasing frequency (Mueller-Dombois & Ellenberg 1974). Further, initial high legume frequency may have been related to the previous fire history within these stands.

Conclusions

Because pine-grassland communities developed under a fire regime that included both dormant and growing season fire, both seasons of fire should be used as management tools in a restoration context. Fire in either season increased species richness, diversity, and total abundance of forbs and legumes, while herbaceous species abundance and richness declined in no-burn controls. Fire reduces woody structure, which influences herbaceous plant composition in restored pine-grassland ecosystems. Increased light and presence of bare ground after fire provide the opportunity for many herbaceous species to become established. Change in species composition and abundance is linked to change in stand structure. Late dormant-season fires are more effective than late growing-season fires at reducing woody sprouts in the understory and at providing bare ground for colonization. As a result, herbaceous species abundance and richness was greater after late dormant season fires.

Acknowledgements. We thank M. E. Payton for assistance with experimental design and statistical analysis, J. Kulbeth, S. Farley, S. Crockett, and D. Gay, for assistance in collecting data and W. Montague for assistance with experimental burns. This project was funded by the U.S. Department of Agriculture, Forest Service, Department of Forestry at Oklahoma State University and was a cooperative effort with Oklahoma Agricultural Experiment Station. This article was published with authorization by the director of the Oklahoma Agricultural Experiment Station.

References

Anon. 1985. SAS users’ guide: statistics, version 5 edition. SAS Institute Inc., Cary, SC.

Arianoutsou, M. & Thanos, C.A. 1996. Legumes in the fireprone Mediterranean regions: an example from Greece. Int. J. Wildland Fire 6: 77-87.

Armour, C.D., Bunting, S.C. & Neuenschwander, L.F. 1984. Fire intensity effects on the understory in ponderosa pine forests. J. Range Manage. 37: 44-49.

Bailey, R.G. 1995. Description of the ecoregions of the United States. 2nd ed. USDA, For. Serv., Misc. Publ. 1391 (revised), Washington, D.C.

Biondini, M., Steuter A.A. & Grygiel, G.E. 1989. Seasonal fire effects on diversity patterns, spatial distribution and community structure of forbs in northern mixed prairie, USA. Vegetatio 85: 21-31.

Buckner, E. 1989. Evolution of forest types in the Southeast. In: Waldrop, T.A. (ed.) Proc. Pine-Hardwood Mixtures: A Symposium on Management and Ecology of the Type, pp. 27-33. USDA, Forest Service, Southeastern Forest Ex-
perimental Station, General Technical Report SE-58. Asheville, NC.

Conover, W.J. & Iman, R.L. 1981. Rank transformations as a bridge between parametric and nonparametric statistics. Am. Stat. 35: 124-129.

Cottam, G. 1949. The phytosociology of an oak woods in southwestern Wisconsin. Ecology 30: 271-287.

Curtis, T.J. 1956. The modification of mid-latitude grasslands and forests by man. In: Thomas, W.L. (ed.) Man’s role in changing the face of the earth, pp. 721-736. University of Chicago Press, Chicago, IL.

Ewing, A.L. & Engle, D.M. 1988. Effects of late summer fire on tallgrass prairie microclimate and community composition. Am. Midl. Nat. 120: 212-223.

Featherstonhaugh, G.W. 1844. Excursion through the slave states. Harper Bros., New York, NY.

Fenneman, N.M. 1938. Physiography of eastern United States.McGraw-Hill Book Co., New York, NY.

Forman, R.T.T. & Godron, M. 1986. Landscape ecology. John Wiley & Sons, New York, NY.

Foti, T.L. & Glenn, S.M. 1991. The Ouachita Mountain landscape at the time of settlement. In: Henderson, D. &

Hedrick, L.D. (eds.) Restoration of old growth forests in the interior highlands of Arkansas and Oklahoma: Proc. of the Conference, pp. 49-68. Morrilton, AR.

Glitzenstein, J.S., Platt, W.J. & Streng, D.R. 1995. Effects of fire regime and habitat on tree dynamics in North Florida longleaf pine savannas. Ecol. Monogr. 65: 441-476.

Great Plains Flora Association. 1986. Flora of the Great Plains. Univ. Press of Kansas, Lawrence, KS.

Grelen, H.E. 1975. Vegetation response to twelve years of seasonal burning on a Louisiana longleaf pine site. USDA, For. Serv., South. For. Exp. Stn., Res. Note, SO-192, New Orleans, LA.

Grelen, H.E. & Lewis, C.E. 1981. Value of range data on fire effects to the wildlife manager. In: Wood, G.W. (ed.) Prescribed fire and wildlife in southern forests, pp. 155-161. The Belle W. Baruch Forest Science Institute, Clemson University, Clemson, SC.

Hill, M.O. & Gauch Jr., H.G. 1980. Detrended correspond ence analysis, an improved ordination technique. Vegetation 42: 47-58.

Hodgkins, E.J. 1958. Effects of fire on undergrowth vegetation in upland southern pine forests. Ecology 39: 36-46.
Hover, E.I. & Bragg, T.B. 1981. Effect of season of burning and mowing on an eastern Nebraska Stipa-Andropogon prairie. Am. Midl. Nat. 105: 13-18.

Howe, H.F. 1994a. Response of early- and late-flowering plants to fire season in experimental prairies. Ecol. Appl. 4: 121-133.

Howe, H.F. 1994b. Managing species diversity in tallgrass prairie: assumptions and implications. Conserv. Biol. 8: 691-704.

Hughes, R.H. 1975. The native vegetation in South Florida related to month of burning. USDA, For. Serv., Southeast. For. Exp. Stn., Res. Note, SE-222, Asheville, NC. Hulbert, L.C. 1988. The causes of fire effects in tallgrass prairie. Ecology 69: 46-58.

James, E. 1823. Account of an expedition from Pittsburgh to the Rocky Mountains, performed in the years 1819 and 1820. H.C. Carey, Philadelphia, PA.

Johnson, F.L. 1986. Woody vegetation of southeastern Leflore County, Oklahoma, in relation to topography. Proc. Okla. Acad. Sci. 66: 1-6.

Komarek, E.V. 1965. Fire ecology – grasslands and man. Proc. Tall Timbers Fire Ecol. Conf. 4: 169-220.

Kreiter, S.D. 1995. Dynamics and spatial pattern of a virgin old-growth hardwood-pine forest in the Ouachita Mountains, Oklahoma, from 1896 to 1994. M.Sc. Thesis, Oklahoma State University, Stillwater, OK.

Landers, J.L. 1981. The role of fire in bobwhite quail management. In: Wood, G.W. (ed.) Prescribed fire and wildlife in southern forests, pp. 73-80. The Belle W. Baruch Forest Science Institute, Clemson University, Clemson, SC.

Lewis, C.E. 1964. Forage response to month of burning. USDA Forest Service,Southeastern Forest Experimental Station, Research Note SE-35, New Orleans, LA.

Lewis, C.E. & Harshbarger, T.J. 1976. Shrub and herbaceous vegetation after 20 years of prescribed burning in South Carolina Coastal Plain. J. Range Manage. 29: 13-18.

Ludwig, J.A. & Reynolds, J.F. 1988. Statistical ecology. John Wiley & Sons, New York, NY.

Masters, R.E. 1991a. Effects of timber harvest and prescribed fire on wildlife habitat and use in the Ouachita Mountains of eastern Oklahoma. Ph.D. Thesis, Oklahoma State University, Stillwater, OK. kansas Press, Fayetteville, AR.

Masters, R.E. 1991b. Effects of fire and timber harvest on vegetation and cervid use on oak-pine sites in Oklahoma Ouachita Mountains. In: Nodvin, S.C. & Waldrop, T.A. (eds.) Fire and the environment: ecological and cultural perspectives, pp. 168-176. USDA, Forest Service, General Technical Report, SE-69, Asheville, NC.

Masters, R.E., Lochmiller, R.L. & Engle, D.M. 1993. Effects of timber harvest and prescribed fire on white-tailed deer forage production. Wildl. Soc. Bull. 21: 401-411.

Masters, R.E., Skeen, J.E. & Whitehead, J. 1995. Preliminary fire history of McCurtain County Wilderness Area and
implications for red-cockaded woodpecker management. In: Kulhavy, D., Hooper, B. & Costa, R. (eds.) Red cockaded woodpecker: species recovery, ecology and management, pp. 290-302. Center for Applied Studies, Stephen F. Austin University, Nacogdoches, TX.

Masters, R.E., Wilson C.W., Bukenhofer, G.A. & Payton, M.E. 1996. Effects of pine-grassland restoration for red-
cockaded woodpeckers on white-tailed deer forage production. Wildl. Soc. Bull. 25: 77-84.

Mueller-Dombois, D. & Ellenberg, H. 1974. Aims and methods of vegetational ecology. John Wiley & Sons, New York, NY.

Neal, J.C. & Montague, W.G. 1991. Past and present distribution of the red-cockaded woodpecker Picoides borealis and its habitat in the Ouachita Mountains, Arkansas. Proc. Ark. Acad. Sci. 45: 71-75.

Nuttall, T. 1980. A journal of travels into the Arkansas Territory during the year of 1819. University of Oklahoma Press, Norman, OK.

Oksanen, J. & Minchin, P.R. 1997. Instability of ordination results under changes in input data order: explanations and remedies. J. Veg. Sci. 8: 447-454.

Oosting, H.J. 1956. The study of plant communities. W.H. Freeman & Co., San Francisco, CA.

Owensby, C.E. & Anderson, K. L. 1967. Yield responses to time of burning in the Kansas Flint Hills. J. Range Manage. 20: 12-16.

Perry, D.A. 1994. Forest ecosystems. John Hopkins Univ. Press, Baltimore, MD.

Platt, W.J., Evans, G.W. & Davis, M.M. 1988. Effects of fire season on flowering forbs and shrubs in longleaf pine forests. Oecologia (Berl.) 76: 353-363.

Pyne, S.J. 1982. Fire in America: a culture history of wildland and rural fire. Princeton University Press, Princeton, NJ.

Rego, F.C., Bunting, S.C. & da Silva, J.M. 1991. Changes in understory vegetation following prescribed fire in mari-
time pine forests. For. Ecol. Manage. 41: 21-31.

Robbins, L.E. & Myers, R.L. 1992. Seasonal effects of prescribed burning in Florida: a review. Tall Timbers Research, Inc., Miscellaneous Publication Number 8. Tallahassee, FL.

Smith, E.B. 1988. An atlas and annotated list of the vascular plants of Arkansas, 2nd ed. University of Arkansas, Fayettville, AR.

Smith, R.L. 1990. Ecology and field biology. Harper & Row, New York, NY.

Smith, R.M. 1989. The atlas of Arkansas. University of Ar

Snedecor, G.W. & Cochran, W.G. 1980. Statistical methods. 7th ed. Iowa State University Press, Ames, IA.

Sousa, W.P. 1984. The role of disturbance in natural communities. Annu. Rev. Ecol. Syst. 15: 353-391.

Sparks, J.C. 1996. Effects of growing-season and dormant season prescribed fire on vegetation in red-cockaded woodpecker clusters. M.Sc. Thesis, Oklahoma State University, Stillwater, OK.

Stearns, F.W. 1949. Ninety years change in a northern hard wood forest in Wisconsin. Ecology 30: 350-358.

Steel, R.G.D. & Torrie, J.H. 1980. Principles and procedures of statistics, a biometrical approach. 2nd ed. McGraw-
Hill Book, New York, NY.

Stickney, P.F. 1986. First decade plant succession following the Sundance Forest Fire, northern Idaho. USDA, Forest

Service, Interior Forest & Range Experimental Station, General Technical Report,INT-197, Ogden, UT. ter Braak, C.J.F. 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67: 1167-1179.

ter Braak, C.J.F. 1988. CANOCO – a FORTRAN program for Canonical Community Ordination. Microcomputer Power, Ithaca, NY.

Thanos, C.A., Daskalakou, E.N. & Nikolaidou, S. 1996. Early post-fire regeneration of a Pinus halepensis forest on Mount Parnis, Greece. J. Veg. Sci. 7: 273-280.

Towne, G. & Owensby, C. 1984. Long-term effects of annual burning at different dates in ungrazed Kansas tallgrass prairie. J. Range Manage. 37: 392-397.

Trabaud, L. & Lepart, J. 1980. Diversity and stability in garrigue ecosystems after fire. Vegetatio 43: 49-57.

Waldrop, T.A., White, D.L. & Jones, S.M. 1992. Fire regimes for pine – grassland communities in the southeastern United States. For. Ecol. Manage. 47: 195-210.

White, D.L., Waldrop, T.A. & Jones, S.M. 1991. Forty years of prescribed burning on the Santee fire plots: effects on understory vegetation. In: Nodvin, S.C. & Waldrop, T.A. (eds.) Fire and the environment: ecological and cultural perspectives, pp. 51-59. USDA, Forest Service, South eastern Forest Experimental Station., General Technical Report, SE-69, Asheville, NC.

Wilson, C.W., Masters, R.E. & Bukenhofer, G.A. 1995. Breeding bird response to pine-grassland community restoration for red-cockaded woodpeckers. J. Wildl. Manage. 59: 56- 67.

Wyant, J.G., Ralph, J.A. & Bechold, W.A. 1991. Physiographic position, disturbance and species composition in North Carolina coastal plain forests. For. Ecol. Manage. 41: 1- 19.

Received 11 March 1997;
Revision received 4 August 1997;
Accepted 13 August 1997;
Final revision received 9 January 1998.