Fire history and the establishment of oaks and maples in second-growth forests

Todd F. Hutchinson, Robert P. Long, Robert D. Ford, and Elaine Kennedy Sutherland


Abstract: We used dendrochronology to examine the influence of past fires on oak and maple establishment. Six study units were located in southern Ohio, where organized fire control began in 1923. After stand thinning in 2000, we collected basal cross sections from cut stumps of oak (n = 137) and maple (n = 204). The fire history of each unit was developed from the oaks, and both oak and maple establishment were examined in relation to fire history. Twenty-six fires were documented from 1870 to1933; thereafter, only two fires were identified. Weibull median fire return intervals ranged from 9.1 to 11.3 years for the period ending 1935; mean fire occurrence probabilities  (years/fires) for the same period ranged from 11.6 to 30.7 years. Among units, stand initiation began ca. 1845 to 1900, and virtually no oak recruitment was recorded after 1925. Most maples established after the cessation of fires. In several units, the last significant fire was followed immediately by a large pulse of maple establishment and the cessation of oak recruitment, indicating a direct relationship between fire cessation and a shift from oak to maple establishment.

Résumé : Nous avons eu recours à la dendrochronologie pour étudier l’influence du feu dans le passé sur l’établissement  du chêne et de l’érable. Six unités expérimentales ont été localisées dans le sud de l’Ohio où la lutte organisée contre les  feux a débuté en 1923. Après que des peuplements eurent été éclaircis en 2000, nous avons collecté des sections radiales  sur des souches de chêne (n = 137) et d’érable (n = 204). Dans chaque unité, l’historique des feux a été établi à partir des  chênes et l’établissement du chêne et de l’érable a été étudié en lien avec l’historique des feux. Vingt-six feux ont été documentés de 1870 à 1933; par la suite, seulement deux feux ont été identifiés. L’intervalle médian de Weibull entre les feux variait de 9,1 à 11,3 ans pour la période se terminant en 1935; la probabilité moyenne d’occurrence de feux (années/ feux) pendant la même période variait de 11,6 à 30,7 ans. Parmi les unités, l’origine des peuplements remonte aux environs de 1845 à 1900 et pratiquement aucun chêne n’a été recruté après 1925. La plupart des érables se sont établis après que les feux eurent cessé. Dans plusieurs unités, le dernier feu important a immédiatement été suivi d’une importante vague d’établissement de l’érable et de l’arrêt du recrutement du chêne, indiquant qu’il y a une relation directe entre la cessation des feux et le changement marqué par l’établissement de l’érable au lieu du chêne.

[Traduit par la Rédaction]

CLICK HERE TO DOWNLOAD AS PDF


Introduction

Across much of the eastern United States, red maple (Acer rubrum L.), sugar maple (Acer saccharum Marsh.), and other mesophytic and (or) shade-tolerant species have become abundant in historically oak-dominated landscapes, threatening the continued dominance of oak (Lorimer 1984; Abrams 1992). Fire-control policies instituted ca. 1910 to 1930 often are considered a primary cause of these successional trends (e.g., Lorimer 1993).

Oaks are considered to be better adapted than maples to a regime of periodic fires primarily because of their relatively thick and, thus, fire-resistant bark; their ability to compartmentalize wounds caused by fire; and the capacity of established seedlings to continue to sprout after being top-killed  repeatedly (Smith and Sutherland 1999; Johnson et al. 2002;  Van Lear and Brose 2002). Periodic anthropogenic fire is widely considered to have promoted and sustained eastern  oak ecosystems throughout their postglacial history  (Abrams 2002). However, specific knowledge of past fire  regimes, which can be obtained by analysis of fire-scarred  trees, is limited to relatively few areas. Several studies  show that fire was frequent in oak ecosystems prior to  (Cutter and Guyette 1994; Guyette et al. 2003; Shumway  et al. 2001) and after Euro-American settlement (Sutherland 1997; Schuler and McClain 2003; Guyette and Stambaugh 2004; Soucy et al. 2005) until fire control was instituted.


Received 8 August 2007. Accepted 8 November 2007. Published on the NRC Research Press Web site at cjfr.nrc.ca on 26 April 2008.
T.F. Hutchinson,1 R.P. Long, and R.D. Ford. USDA Forest Service, Northern Research Station, 359 Main Road, Delaware, OH 43015, USA.
E.K. Sutherland. USDA Forest Service, Rocky Mountain Research Station, 800 Block East Beckwith, P.O. Box 8089, Missoula, MT 59807, USA.

1Corresponding author (e-mail: thutchinson@fs.fed.us).


Much more is known about long-term patterns of tree establishment in old-growth oak forests. These studies often suggest a strong influence of fire cessation on tree recruitment (e.g., Abrams and Downs 1990; Abrams and Copenheaver 1999; Aldrich et al. 2005). Oak recruitment is known to have occurred for up to several hundred years but decreased or ceased around the time that fire control began. Several of these studies also document greatly increased recruitment of maples and other nonoak species during the same period (e.g., Abrams and Downs 1990; Abrams and Copenheaver 1999).

Shumway et al. (2001) were the first to document both the fire history and patterns of establishment for oak and other species in an old-growth oak-dominated stand in western Maryland. The authors showed that fire and oak recruitment were frequent from the early 1600s through the early 1900s. Both the cessation of oak recruitment and the increased recruitment of red maple and black birch (Betula lenta L.) coincided with reduced fire frequency ca. 1930. Soucy et al. (2005) showed that oak-hickory stands in the Arkansas Ozarks originated following harvesting or fire ca. 1900 and that fires were frequent through the 1930s. As fires became much less frequent after ca. 1940, oak recruitment ceased, and other shade-tolerant, nonoak species, such as flowering dogwood (Cornus florida L.), red maple, and blackgum (Nyssa sylvatica Marsh.), became established (Soucy et al. 2005).

The unglaciated ‘‘hill country’’ of southeastern Ohio was dominated by oak forests ca. 1800, just prior to Euro-American settlement (Beatley 1959; Gordon 1969; Dyer 2001). After nearly all of the forests in the region were cut over in the 19th century, many stands regenerated to oak dominance (Goebel and Hix 1997; Dyer 2001; Yaussy et al. 2003), and oak remains abundant in the region today (Griffith et al. 1993). However, as with most areas in the eastern United States that historically were oak dominated, the continued abundance of oak is threatened by increasing densities of maples and other species (e.g., blackgum and beech (Fagus grandifolia Ehrh.)) and poor oak regeneration. Dendrochronology fire histories indicate that fires occurred frequently in the region from ca. 1870 to 1935 (Sutherland 1997; McEwan et al. 2007b). It is hypothesized that this fire regime sustained oak dominance as second-growth forests developed (McEwan et al. 2007b) and that fire control directly facilitated the establishment of the now-abundant maples and other competitors (Sutherland et al. 2003).

To better understand how past fires were related to tree establishment, we conducted a dendrochronology study at the Ohio Hills site of the national Fire and Fire Surrogate Study (FFS). Our study was carried out on three replicate sites, each containing two separate units (*20 ha each) that here thinned. The second-growth forests were dominated by oak in the overstory, but maples and other shade-tolerant species were abundant in the midstory and understory. We collected basal cross sections from cut stumps of both oaks and maples to document stand fire histories and tree establishment. We hypothesized that, within a stand, temporal patterns of oak and maple recruitment would be closely related to the occurrence of past fires. Specifically, we hypothesized that (i) fires were frequent prior to the initiation of fire control, (ii) oak establishment occurred primarily before the initiation of fire control, and (iii) maples established primarily during the fire-control era. By studying units within three spatially separated replicate sites that likely had different fire histories, we also hoped to better understand how variability in past fire regimes affected oak and maple establishment. To our knowledge, this is the first study that directly examines the relationship between specific historic fire events and oak and maple recruitment events. A better understanding of how past fire regimes affected the pattern and pace of recruitment during stand development could provide new insights for the use of prescribed fire to manage oak forests.


Methods

Study area and site descriptions

The study area is in southern Ohio within the Southern Unglaciated Allegheny Plateau (McNab and Avers 1994). The topography is highly dissected, consisting of sharp ridges, steep slopes, and narrow valleys. The bedrock geology is predominantly sandstones and shales that produce well-drained and acidic soils.

The Ohio Hills FFS study site has three replicates (hereafter sites): one each in the Raccoon Ecological Management Area (REMA), Zaleski State Forest, and Tar HollowState Forest. The REMA site (39812’34@N, 82823’07@W) is in Vinton County and within the Vinton Furnace Experimental Forest; owned by Forestland Group, LLC, and comanaged with the USDA Forest Service Northern ResearchStation. The Zaleski site (39821’22@N, 82821’59@W), also in Vinton County, is 19 km north of the REMA site. The Tar Hollow site (39819’ 47@N, 82846’11@W) is in Ross County, 35 km west of the Zaleski site. Soils at both REMA and Zaleski are predominantly Steinsburg and Gilpin series silt loams (Typic Hapludalfs); Tar Hollow soils are predominantly Shelocta-Brownsville complex sandy loams (Typic Hapludalfs and Typic Dystrochrepts, respectively) (Boerner et al. 2007). Both state forests are managed by the Ohio Department of Natural Resources’ (ODNR) Division of Forestry.

Human land use has had a major effect on these forests. Both the REMA and Zaleski sites are located near charcoal iron furnaces that operated in the 1800s. REMA is <2 km from Vinton Furnace and Zaleski is <4 km from Hope Furnace; these were in operation from 1853 to 1883 and from 1854 to 1874, respectively (Stout 1933). Forests at both sites presumably were harvested at least once to provide charcoal for iron smelting. The Tar Hollow site was not affected by the iron industry since it was more than 25 km from the nearest furnace. Land deeds show direct human occupation in small parcels within the Tar Hollow study site until the mid-1930s (ODNR, Division of Forestry, District Office, Chillicothe, Ohio).

The forests generally were similar in structure and composition across the three sites. Prethinning data collected in 2000 showed that mean stand basal area at REMA was 28 m2/ha; white oak (Quercus alba L.) accounted for 21% of the basal area; black oak (Quercus velutina Lam.), 17%; chestnut oak (Quercus montana Willd.), 15%; and scarlet oak (Quercus coccinea Muenchh.), 12% (D.A. Yaussy, USDA Forest Service, Delaware, Ohio, unpublished data).

Mean basal area at Zaleski was 27 m2/ha and was dominated by chestnut oak (31%), followed by white oak (22%), red maple (13%), and black oak (13%). At Tar Hollow, mean basal area was 32 m2/ha, and the dominant species were chestnut oak (33%), white oak (20%), and black oak (16%). Oak site indices (base age 50 years) are variable across the landscape because of the dissected topography, ranging from about 17 m (55 ft) on upper south-facing slopes to 24 m (80 ft) on lower north-facing slopes (D.A. Yaussy, USDA Forest Service, Delaware, Ohio, personal communication). On all sites, the sapling layer (1.4 m tall to 9.9 cm diameter at breast height (DBH)) and the midstory (trees 10-25 cm DBH) were dominated by shade-tolerant trees, the most abundant of which were red maple, sugar maple, blackgum, and beech (Albrecht and McCarthy 2006). Shade-tolerant trees >25 cm DBH occurred at low densities on all sites.

The mean annual temperature and precipitation are 11.3 8C and 1024 mm, respectively. Precipitation is distributed fairly evenly throughout the year with no months averaging <60 mm. Today, most fires occur during the early spring (March and April) and fall (October and November), when vegetation is predominantly dormant; spring dormant season fires are the most frequent (Haines et al. 1975; Sutherland et al. 2003), and nearly all fires are anthropogenic in origin.

At each FFS site, four treatment units (19-26 ha) were established: an untreated control (control), mechanical thinning (thin), prescribed fire (burn), and a combination of thinning and fire (thin+burn). Our dendrochronology study was conducted on the two thinned units (thin and thin+burn) at each site. Midstory thinning occurred from November 2000 to April 2001 and favored the retention of dominant and codominant oaks. However, to meet commercial thinning objectives, some dominant and codominant oaks were harvested. Across sites, stand density (trees ‡10 cm DBH) was reduced by 32% from a mean of 400 to 269 trees/ha, and tree basal area was reduced by 30% from a mean of 29 to 20 m2/ha.

Sampling design and field methods
At the REMA site, the two thinned units (hereafter, REMA 2 and REMA 3) were separated by a triangular wedge of untreated forest that ranged in width from several meters to 275 m. The Zaleski thinned units (Zaleski 2 and Zaleski 3) were contiguous, and the boundary between units was an intermittent stream drainage. The Tar Hollow thinned units (Tar Hollow 2 and Tar Hollow 3) also were contiguous, but the boundary did not follow a major topographic feature. Despite the contiguous units at Zaleski and Tar Hollow, we treated the units separately for summary and  nalyses, because 50% (14 of 28) of the fires that we document were recorded only in a single unit.

Ten 0.1 ha plots were established in each unit to monitor vegetation and soils for the FFS study. Plot corners were georeferenced with global positioning system (GPS) technology. The plots were distributed across the landscape to represent a continuous range of soil moisture conditions from dry (upper south-facing slopes) to mesic (lower north-facing slopes). In 2000, all overstory trees (‡10 cm DBH) were tallied by species and DBH on each plot prior to treatments. We focused our collection of oak and maple basal cross sections on the plots to utilize the tree data and georeferenced locations.

Full basal cross sections were cut from stumps with a chainsaw from December 2000 to May 2001, soon after thinning operations had been completed in each unit. All oak stumps in a plot were examined for the presence of wounds (discoloration, seams, staining, and wound wood ribs) that might indicate a fire event (Smith and  Sutherland 1999). We attempted to locate at least two oak stumps within or adjacent to each plot. We collected 137 oak cross sections across all units and recorded the upslope position on each. Samples included 64 white oak, 40 chestnut oak, 22 black oak, 7 scarlet oak, and 4 northern red oak (Quercus rubra L). All samples were cut at a height of about 5-10 cm aboveground. The mean basal diameter of the oak samples was 47.4 cm and ranged from 23.1 to 98.1 cm. We mapped the approximate location of each sample based on its position within or adjacent to a georeferenced vegetation plot.

To determine the temporal pattern of maple establishment, our objective was to collect basal cross sections from three stumps in and (or) adjacent to each plot. Our goal was to collect three maples for every two oaks, because maples were approximately 1.5 times as abundant as oaks in the midstory, where the thinning treatment was focused. Our first priority was to obtain larger maples to document the time when establishment began, but we sampled across a range of maple stump diameters. We collected 30-39 maples per unit for a total of 204 (142 red maple and 62 sugar maple). Nearly all sugar maple samples were collected in three units: Tar Hollow 2 (n = 27), Tar Hollow 3 (n = 20), and REMA 3 (n = 15). The mean basal diameter of maple samples was 27.5 cm and ranged from 10.1 to 58.4 cm.

Laboratory methods
Cross sections were planed and sanded to enhance ring boundaries and facilitate dating. Each oak sample was crossdated using skeleton plots (Stokes and Smiley 1968) against a previously established master chronology for the region (Sutherland 1997). Maple cross sections ‡70 years old also were skeleton plotted and cross-dated. Younger maple samples were ring-counted along two to four radii and crossdated by identification of key stress years. Several factors contributed to make the exact pith dates for maples somewhat less precise than that for the oaks. Firstly, the crossdating was less clear for maples than for the oaks, i.e., key stress years were not as consistent among the maples. Secondly, a small proportion of the maples had decay or incipient decay in or near the pith, which obscured the ring boundaries. Thirdly, some maples had rings that were locally absent in a portion of their circumference, a common phenomenon documented by Lorimer et al. (1999) for suppressed sugar maple trees. Several samples that we felt could not be reliably dated (primarily small suppressedstems) were omitted from further analyses.

We required at least three scarred samples per unit per year to classify a wound event as a fire scar and, consequently, a year as a fire year. If only two wounds were present in a unit in a given year but the adjacent unit showed evidence of a fire in the same year (three or more samples scarred), we recorded a fire for the unit with only two scars (this occurred once). These criteria were applied to limit the likelihood that wounds caused by other factors (e.g., logging, falling trees and branches, or animals) were recorded as fire scars (see McEwan et al. 2007a). The seasonality of each fire event was determined by examining where the wounds intersected the annual growth ring. For dormant-season scars (located between annual growth rings) it was not possible to determine whether the scar occurred in the fall after the previous growing season or in the late  Hutchinson et al. winter – early spring prior to the upcoming growing season (Sutherland 1997). Because fires in our region are most frequent in the early spring dormant season (March-April), we assigned dormant-season wounds to the calendar year of the upcoming growing season. For example, a dormant-season wound located between the 1922 and 1923 annual growth rings was recorded as a 1923 wound.

To better examine how the relative intensity and (or) extent of fires may have affected tree establishment, we defined fires as ‘‘significant’’ if (i) ‡33.3% of the samples exhibited wounds and (ii) at least five samples had wounds. Since little is known about the relationship between fire intensity and scarring in oaks (see Smith and Sutherland 1999; Guyette and Stambaugh 2004; McEwan et al. 2007a), this definition provides a relative measure of the intensity and extent of the fires within this study. We mapped the location of all samples (oaks and maples) in all fire years at REMA to visualize the spatial pattern of fire scars and tree establishment across the landscape. We used the ArcView version 3.2a geographic information system to map samples based on our field maps that showed locations within or adjacent to a georeferenced vegetation plot. We selected eight of the nine fire years at the REMA study site to illustrate spatial patterns of fire scarring and tree establishment.

Ring widths were measured on the oaks so that growth dynamics and potential release events could be determined. Oak cross sections were scanned and measured using WINDENDRO (Regent Instruments Inc., Ste-Foy, Que.). Two radii approximately 1808 apart were measured in each cross section. Radii were located to minimize the influence of wounds and associated wound wood on growth measurements. Ring-width measurements and crossating were verified with the program COFECHA 2.1 (Holmes 1983; Grissino-Mayer 2001b). The program ARSTAN (Cook and Kairiukstis 1990) from the Lamont Doherty Earth Observatory’s Tree-Ring Laboratory, was used to detrend measurements with a negative exponential curve or linear regression line. This standardization procedure removes the growth trend associated with age and produces dimensionless indices that can be averaged to create a master chronology for a site (Fritts 1976). A master chronology was created for each of the six sampled sites.

Potential releases associated with disturbance events were identified in each master chronology with the JOLTS program (Holmes 1999) from the International Tree-Ring Data Bank Dendrochronology Program Library. Major releases were those where there was a >100% increase in growth expressed as the mean chronology ring-width index over a 15 year period compared with the mean chronology ringwidth index in the preceding 15 year period. Minor releases were those where growth increased 50% over a 10 year period compared with a previous 10 year period (Lorimer and Frelich 1989; Soucy et al. 2005). We report release events only when there were at least 10 trees present in the master chronology. Fire-return interval analyses Data on fire history derived from oak cross sections were analyzed using the FHX2 program (Grissino-Mayer 2001a, 2004). Because fires were so infrequent after 1935, it was not possible to statistically compare fire frequency between pre- and post-fire suppression periods. Instead, for each study unit, we calculated fire intervals from the first fire to 1935 (before and during the early fire-control period) and compared these with fire intervals from the first fire to 2000. Both mean fire intervals (MFI) and Weibull median fire intervals (WMFI) were calculated. The latter is considered a better estimator of central tendency for the typically nonnormal fire-interval distributions (Grissino-Mayer and Swetnam 1997; Grissino-Mayer et al. 2004).

For each unit, we also calculated the mean fire occurrence probability (MFOP) for two periods (stand origination to 1935 and to 2000). Defined by Guyette et al. (2006), MFOP is the number of years divided by the number of fires in a chronological period. Guyette et al. (2006) calculated the MFOP to account for the fire-free period prior to the first recorded fire. For each site, we defined the stand origination as the first year in which at least four samples were present that could potentially record a fire.

Results

Twenty-eight fires were recorded, of which 26 occurred from 1870 to 1933 (Table 1). Twelve fires scarred five or more samples and were classified as significant fires (‡33.3% of the samples were scarred); these fires occurred from 1877 to 1923. Most wounds attributed to fire were recorded on small-diameter trees; for all fires, the basal diameter of oaks at the time of wounding was 12.7 ± 0.7 cm (mean ± SE). In most of the fires (n = 24), all wounds were located between annual growth rings, indicating occurrence in the dormant season (September to early April). In fire years, 147 of the 213 total wounds (69%) were on the uphill portion of the stem (3008 clockwise to 608), based on the uphill position recorded on the sample in the field. Of the 204 maple samples, only one exhibited a fire scar (a wound in a fire year); that maple, from Zaleski 2, had a wound in 1965.

Fire histories of the study units

REMA 2 had the greatest number of fires (n = 7) and significant fires (n = 5) (Table 1); fires were documented from 1877 to 1933. The 1917 significant fire had both dormant and earlywood scars, suggesting an early growing season fire. In the 1933 fire, most wounds were present in the late earlywood; in that fire, all seven scarred trees were young and small, having established in 1923 or 1924 and averaged only 5.1 cm in basal diameter (Table 1). We recorded six fires at REMA 3, three of which were significant, from 1878  1923 (Table 1; Fig. 1). As with REMA 2, the wounds in the 1917 fire indicate an early growing season fire; in the 1906 fire, all three wounds intersected the earlywood.

Zaleski 2 had evidence of six fires; five occurred from 1870 to 1928, and a sixth occurred in 1965 (Table 1). Although only the 1923 fire was classified as significant, it wounded 83% (15 of 18) of the samples; none of the other fires wounded more than three samples. No fires were recorded at Zaleski 2 during a 25 year period from stand origination (1844 to 1869). At Zaleski 3, the chronology was shorter, dating from 1880 to 2000. Although only three fires were recorded, both the 1917 and 1923 fires were significant. Again, the 1923 fire wounded a high percentage (61.9%) of the samples. All fire scars in both Zaleski units were in the dormant season.

Table 1. Summary data for the 28 fires documented on the six study units.


Diameter scarred (cm)

Study site, sample size,and chronologyaFire yearFire seasonbScarred tree (%)No. of scarred treesTotal no of treesMeanRange
REMA 2
n = 29
1855-2000
1877
1885
1895
1900
1917
1923
1933
D
D
D
D
D and E
D
E
50.0
66.7
25.0
50.0
50.0
50.0
20.7
6
8
3
6
6
9
7
12
12
12
12
12
18
29
8.6
13.0
14.4
18.1
22.0
13.1
5.1
7.0-10.7
9.5-17.2
13.5-15.2
15.4-23.1
18.0-28.8
3.3-25.8
2.7-7.5
REMA 3
n = 22
1858-2000
1878
1885
1895
1906
1917
1923
D
D
D
E
D and E
D
23.1
38.5
57.1
23.5
33.3
18.2
3
5
8
3
7
4
13
13
14
17
21
22
7.4
11.0
14.2
18.4
11.7
16.8
4.6-10.5
6.7-14.5
5.5-19.5
12.1-21.5
3.9-21.7
8.6-25.0
Zaleski 2
n = 24
1844-2000
1870
1897
1917
1923
1928
1965
D
D
D
D
D
D
75.0
33.3
16.7
83.3
12.5
12.5
3
3
3
15
3
3
4
9
18
18
24
24
4.7
5.0
13.3
16.0
11.1
22.5
4.2-5.4
1.2-8.3
6.3-18.1
7.5-29.8
4.0-15.6
16.8-33.7
Zaleski 3
n = 25
1880-2000
1914
1917
1923
D
D
D
22.2
35.0
61.9
4
7
13
18
20
21
5.2
8.6
10.3
3.8-6.8
4.9-13.6
6.3-17.2
Tar Hollow 2
n = 22
1844-2000
1883
1900
1926
D
D
D
71.4
26.7
22.7
5
4
5
7
15
22
7.9
4.4
13.4
5.7-13.7
1.5-8.0
8.0-24.0
Tar Hollow 3
N = 17
1899-2000
1900
1912
1984
D
D
D
50.0
25.0
17.6
2
3
3
4
12
17
6.4
7.2
40.7
3.0-6.8
3.7-10.7
36.3-47.0

Note: All data are from the oak samples. Years in bold type indicate fires that scarred five or more trees and ‡33.3% of samples.
aSample size is the total number of oak samples. Chronology period begins with the first year when four or more samples were present to record fires.
bD, dormant season; E, earlywood; D and E, both dormant and earlywood wounds were present in the samples.


We recorded three fires at Tar Hollow 2 from 1883 to 1926 (Table 1). No fires were documented in the 39 years from stand origin (1844) to the 1883 fire; that fire was the only significant fire, wounding five of seven trees. Tar Hollow 3 had the shortest chronology (1899-2000) of all units; fires were documented in 1900, 1912, and 1984. Only two trees were scarred in 1900, but this is included as a fire because of its concordance with the four trees scarred in 1900 in Tar Hollow 2. Only three trees were scarred in the 1912 and 1984 fires at Tar Hollow 3. As with Zaleski, all fire scars were located between annual growth rings, indicating dormant-season fires.

Fire-return intervals
In the period before active fire control and ending in 1935, composite mean fire intervals (MFI) only could be calculated at three of the six units (REMA 2, REMA 3, and Zaleski 2). At these units, MFI ranged from 9.0 to 14.5 years (Table 2). Likewise, the composite Weibull median fire interval (WMFI) ranged from 9.1 years at REMA 2 to 11.3 years at Zaleski 2. For the same pre-1936 period, the mean fire occurrence probabilities (MFOP; Guyette et al. 2006), which also take into account the period of time prior to the first fire, ranged from 11.6 and 12.2 years at REMA 2 and REMA 3, respectively, to 30.7 years at Tar Hollow 2. For the five units originating in 1880 or before (all but Tar Hollow 3), there was a period of at least 20 years from stand origination to the first recorded fire. As only two fires were documented from 1936 to 2000, fire-interval calculations that end in 2000 are longer (Table 2). The WMFI ranged from 12.7 and 13.9 years at REMA 2 and REMA 3, respectively, to 35.4 years at Tar Hollow 2.

Fire and the establishment of oaks and maples
At REMA 2, all oak samples established prior to 1924, and nearly every maple recruited after the 1923 fire (Fig. 2a).

Fig. 1. Fire history diagram for REMA 3. The broken horizontal lines represent the growth years for the 22 oak samples. The solid triangles are wounds that were in a recorded fire year; vertical bars are wounds present in years not recorded as fire years. The six fire years are indicated by the vertical lines located above the timeline.

 


 

Study sitePeriod ending in1935Period ending in 2000
MFIWMFI (87.5%-12.5%)MFOP (years/fires)MFIaWMFI (87.5%-12.5%)aMFOP (years/fires)
REMA 29.39.1 (4.8-14.1)11.6 (81/7)17.612.7 (2.8-36.1)20.9 (146/7)
REMA 39.09.2 (6.7-11.3)12.2 (73/6)20.313.9 (2.6-42.4)23.8 (143/6)
Zaleski 214.511.3 (2.8-28.8)18.4 (92/5)21.718.2 (5.5-40.3)26.2 (157/6)
Zaleski 318.7 (56/3)28.714.0 (1.5-63.2)40.3 (121/3)
Tar Hollow 230.7 (92/3)39.035.4 (13.3-68.2)52.3 (157/3)
Tar Hollow 318.5 (37/2)33.327.3 (7.6-64.2)34.0 (102/3)

Note: A dash indicates that there were an insufficient number of fire events to calculate the interval.
aFire interval calculations in these columns are based on a final incomplete interval ending in 2000.

After an initial period of oak establishment (1852 to 1865), presumably after harvesting for the charcoal iron industry, there was a 51 year period (1866 to 1916) when no oak recruitment was recorded. Thereafter, two pulses of oak establishment were documented immediately after the significant fires of 1917 and 1923. No maples predated the 1917 fire, and several maples established between the 1917 and 1923 fires. In 1923, immediately after the last significant fire, 15 maples recruited. Thereafter, 17 maples established from 1924 to 1938.

At REMA 3, initial oak establishment occurred from 1849 to 1860. As with REMA 2, no establishment was recorded after 1924 (Fig. 2b). After 1860, there were no large pulses of oak establishment, but there were 7 years from 1885 to 1924 in which pith dates were recorded for one or two oaks. The oldest maple dated to 1921, and a pulse of 11 stems established in 1923, immediately after the last fire. An additional 12 maples established from 1924 to 1949.

The temporal patterns of fire and establishment at the Zaleski units were similar to those of REMA, remarkably so for the initiation of maple establishment and the corresponding cessation of oak recruitment. At Zaleski 2, four oaks established in the 1840s (Fig. 2c). There was a period of oak recruitment from 1872 to 1880, with a pulse of eight stems in 1879 and 1880, following the 1879 fire. After 1880, we record virtually no oak recruitment for 42 years (1881 to 1922). Oaks then established in 1923 and 1924, immediately after the 1923 fire which scarred 15 of 18 oaks. Maple establishment began in 1922 (n = 4), just before the 1923 fire; four others dated to 1923. Thereafter, 22 maples established from 1924 to 1965, with a maximum of three stems in a single year.

 

Fig. 2. (af) Temporal establishment of oaks and maples for the six study units. Fires are indicated by vertical lines above the timeline; significant fires, those with ‡33.3% of samples wounded, are indicated by vertical arrows.

 

At Zaleski 3, we recorded 5 oaks that established before 1900 (primarily ca. 1880); then, 10 trees established in 1902 (Fig. 2d). The 1902 pulse of oak recruitment was not associated with a fire. As in unit 2, there was another period of oak establishment (n = 5) in 1923 and 1924, immediately following the 1923 significant fire; thereafter, we recorded only a single oak that established in 1954. As in unit 2, maple recruitment initiated in 1922 (n = 4), and eight trees established in 1923, directly after the significant fire. From 1926 to 1928, 11 maples established, and 11 others had pith dates from 1937 to 1959.

Tar Hollow 2 exhibited an early period of oak establishment (n = 8) from 1835 to 1851, six of the eight trees had pith dates of 1842 and 1843 (Fig. 2e). We recorded no oak establishment from 1852 to 1885; 15 trees established from 1886 to 1919. There was a small pulse (n = 3) of oak recruitment in 1900 after the fire of that year. Unlike REMA and Zaleski, maple establishment began nearly 40 years earlier at Tar Hollow 2. Maples (both red and sugar) recruited for 70 years (1881 to 1951) in a fairly continuous manner but did not exhibit the large pulses recorded at REMA and Zaleski.

The oldest oak recorded in Tar Hollow 3 dated to 1851, but no other samples predated 1894 (Fig. 2f). We record fires in 1900 and 1912. The 1900 fire scarred two of the four samples, and there was a pulse of oak recruitment (n = 6) that year. Thereafter, six oaks had pith dates from 1902 to 1924; no more than one tree was recorded in any single year. Maple recruitment at Tar Hollow 3 spanned from 1897 to 1963. As in Tar Hollow 2, there were no large establishment events. At both Tar Hollow units, despite different patterns of fire and an earlier initiation of maple establishment compared with REMA and Zaleski, oak establishment ceased at the same time at all sites (ca. 1920 to 1925).

Spatial distribution of fire scars and tree establishment at REMA

Fires that occurred in 1885, 1895, 1917, and 1923 were recorded in units 2 and 3 (Fig. 3). The 1885 and 1917 fires were classified as significant in both units. By contrast, the presence of fire-scarred trees was limited to one unit in the other five fire years (1877, 1878, 1900, 1906 [not shown], and 1933). Presumably, these fires did not burn across the intermittent stream drainage separating the two units. Similarly, the stream drainage in the center of unit 2, running southwest, appears to have limited fire spread in several years when trees were scarred only northwest (1895 and 1900) or southeast (1933) of the drainage.

For all fire years, scarred oaks were located near oaks that were not scarred. The trees most prone to exhibiting fire scars were in the northern portion of unit 2, near the top of the ridge; two or more of the seven trees that had established there before the first fire in 1877 were scarred in all unit 2 fires prior to 1933.

Maple establishment is first shown in both units on the 1923 fire map (Fig. 3g); these trees established from 1917 to 1922 and survived the 1923 fire. In unit 3, most of the maples that established before 1923 were in areas where fire scars were not recorded on oaks in 1923, suggesting that those small trees were in unburned patches. In unit 2, all three maples predating 1923 are within 5-20 m of an oak with a 1923 fire scar; however, all three oaks with fire scars were small in 1923, each having established immediately after the 1917 fire. By the time of the 1933 fire (Fig. 3h), maples had established across most of the landscape. All of the maples in the 1933 fire map were relatively near scarred oaks and thus escaped that fire; none of these samples had 1933 wounds. However, the low-intensity and perhaps patchy nature of the 1933 spring growing season fire is suggested by the fact that only small oaks (mean basal diameter 5.1 cm) that established after the 1923 fire were scarred.

Radial growth and releases

The master chronologies show growth that is typical of trees from forest interior sites and show only several sustained release events (Fig. 4). Growth releases were identified at only two of the six units; however, none of these releases coincided with a fire. Zaleski 2 had a major release beginning in 1896, perhaps coinciding with a harvest based on its magnitude. No oak recruitment was associated with this release. A moderate release also was identified at this site for 1906. REMA 3 had a major release in 1864  (Fig. 4), although considerable variability in early growth associated with the small sample size may partially account for this release event. No growth releases were identified at Zaleski 3, REMA 2, or at the Tar Hollow units. At REMA 3, some oak recruitment preceded the 1864 release event, but there is no evidence that these were related (Figs. 2 and 4). These analyses, based on the standardized mean ring width chronologies, indicate that fires were of insufficient intensity to cause standwide mortality and the release of surviving oaks.


Discussion

Historic fire regime

Generally, fires were frequent from ca. 1870 to 1935 as stands developed but were uncommon thereafter, reflecting the regional postsettlement history of anthropogenic fire and its suppression. A 1920-1922 forest survey of 10 southern Ohio counties reported that 25% of all forested land showed visible evidence of having burned at least once within the previous decade (ODNR, Ohio Division of Forestry, Columbus, Ohio). Data from the same survey indicated that 5% to 7% of forested land burned annually (Ohio Experiment Station 1922). Organized fire control was instituted in 1923, and its infrastructure and effectiveness developed rapidly.  By 1935, 19 fire lookout towers had been erected in 8 southern Ohio counties, and 447 fire wardens were employed (Leete 1938). From 1926 to 1935, the mean annual forest acreage burned had been reduced to 0.8% (Leete 1938); from 1950 to 2000, it was further reduced to only 0.1% per year (ODNR, Division of Forestry, Columbus, Ohio).

Our study adds to the growing body of dendrochronological evidence that fire was frequent in the central hardwood region prior to organized fire control; examples include oak and oak-pine community types in the Missouri and Arkansas Ozarks (Cutter and Guyette 1994; Guyette et al. 2002; Guyette and Spetich 2003; Soucy et al. 2005); pine-oak communities in the southern Appalachians (Brose and Waldrop 2006); post oak (Quercus stellata Wang.) barrens in Indiana (Guyette et al. 2003) and Tennessee (Guyette and Stambaugh 2004); and oak forests in southern Ohio (Sutherland 1997; McEwan et al. 2007b), Maryland (Shumway et al. 2001), and West Virginia (Schuler and McClain 2003). The fire-return intervals that we calculated for the REMA and Zaleski sites, ranging from 9.1 years prior to 1936 to 18.2 years overall (WMFI), are within the range reported in those studies (2-24 years), despite our more conservative criteria for classifying fire years. However, the 35 year firereturn interval at Tar Hollow 2 exceeds the range in the other studies.

In the central hardwoods region, dissected topography is known to have limited the spread of fires historically (Guyette et al. 2002). In our study, mapped fire-scarred trees suggest that even relatively small intermittent stream drainages limited fire spread in some years, resulting in some fires that were recorded on, and presumably burned, only a portion of the 20 ha units. By contrast, several of the significant fires spanned two units, scarring trees as far as 900 m apart.

Fig. 3. (ah) Spatial distribution of oaks and maples at REMA in eight fire years. Solid circles indicate oaks scarred by a fire in the year associated with the map, and open circles show oaks that were not scarred in that year. The shaded triangles indicate the location of maples that had established by the time of the 1923 (Fig. 3g) and 1933 (Fig. 3h) fire years. (No maples were documented to have established at the time of the 1917 fire or before.)

 

Fig. 4. Master tree ring chronologies (ARSTAN) showing fire events (vertical arrows) and the point when a minimum of 10 trees were averaged (vertical line) into the mean chronology. Major (M) and moderate releases (m) were only noted in the Zaleski 2 and REMA 3 chronologies.

As other dendrochronological fire-history studies in the region have shown (e.g., Sutherland 1997; Shumway et al. 2001; McEwan et al. 2007b), the great majority of fires occurred in the dormant season (September to early April). Only fires in 1906, 1917, and 1933 at REMA had wounds located in the earlywood. In southern Ohio, radial growth (earlywood production) in oak begins in middle to late April, during bud-swelling and leaf unfolding (Phipps 1961). Oak cross sections collected in early May clearly show earlywood production, whereas samples from mid June show latewood production (R.W. McEwan, Department of Forestry, University of Kentucky, Lexington, Ky., unpublished data). Thus, we estimate that fires exhibiting both dormant and earlywood scars likely occurred in mid-April at the onset of radial growth (e.g., the 1917 REMA fire). The single fire (REMA 2 in 1933) that exhibited wounds intersecting the late portion of the earlywood probably occurred in May.

Sutherland (1997) and McEwan et al. (2007a) showed that historic fire occurrence in this region was not related strongly to monthly climatic conditions. Similarly, we recorded fires in both wet and dry periods. However, for the years in which fires occurred at more than one study site (1900, 1917, and 1923), all exhibited two or more months of drought conditions (Palmer drought severity index more negative than -1.5) the previous fall (1900, 1917, 1923) or also in the spring of the recorded fire year (1900).

The intensity of fires as these stands developed is difficult to determine with certainty because research is lacking that directly relates fire intensity to scarring in oak. However, several studies that have examined patterns of scarring in oak following prescribed fires provide some insight. Smith and Sutherland (1999) found that 14 of 18 small oak trees (4-23 cm DBH) had at least one fire scar after two low-intensity prescribed fires (flame lengths generally <50 cm with no overstory tree mortality). Guyette and Stambaugh (2004) showed that 35% to 65% of mostly small post oak trees (10-25 cm basal diameter) were scarred during three separate prescribed fires in an oak community in Tennessee. These fires burned 72%-93% of the area and reduced stand density (mostly small-diameter trees) by 35%. However, in both studies, only trees with visible bark char were selected for sampling. In our study, we found that, on average, 40% of the oak samples, most of which were small at the time (5 to 25 cm basal diameter) were scarred in the historic fires. These scarring percentages suggest that the fires would have been similar in intensity to the prescribed fires reported by Smith and Sutherland (1999) and Guyette and Stambaugh (2004). Although pulses of oak establishment immediately after some historic fires in our study suggest abundant resprouting after top kill, there is no evidence of high-severity stand-replacement fires even in these relatively young, regenerating stands.

McEwan et al. (2007a) reported that during 15 separate prescribed fires that were similar in intensity to those in Smith and Sutherland (1999), the scarring rate was much lower (12.6%) in white oak. However, because the sample trees in that study were much larger (most were >20 cm DBH), it is difficult to compare those scarring percentages with the historic scarring of small trees in our study.

Five years after a prescribed fire, Wendel and Smith (1986) found that 66% of overstory trees (all species, >12.7 cm DBH) exhibited fire scars visible on the exterior of the stem as exposed wood with callous tissue. The fire in their study was higher in intensity, reducing stand basal area by nearly 20%. The high scarring percentage of larger trees in their study suggests a higher intensity fire than was typical of the historic fires in our study.

Fire, land use, and tree establishment
At REMA and Zaleski, periods of oak and maple establishment were related to specific fire events as these stands developed. The establishment and subsequent survival of maples generally began immediately after the cessation of significant fires, i.e., fires that wounded at least one-third of the oak samples. The final oak establishment event also occurred directly after the last significant fire at three of the four REMA and Zaleski units. Because maples seldom were recorded as witness trees in upland forests just before Euro American settlement (Beatley 1959; Dyer 2001), these results lend support to the hypothesis that organized fire control facilitated the invasion of maples into the uplands from the more fire-protected lowlands (Abrams 1998).

The temporal patterns of fire history and maple establishment were similar at all four units at REMA and Zaleski. All units had periodic fires from ca. 1870 to 1925, and all units burned in both 1917 and 1923; in each of those years, fires were significant in three units. The initiation of maple establishment was similar in that none was documented in any units before the 1917 fires. Limited establishment was documented between the 1917 and 1923 fires, and large pulses occurred immediately after the 1923 fires followed by continuous establishment into the 1960s. The large pulses of maple establishment after the 1923 fires suggest resprouting from previously established individuals. Red maple, which accounted for 89% of the maple samples at REMA and Zaleski, has thin bark and is highly susceptible to top kill by fire (Harmon 1984; Regelbrugge and Smith 1994; Hutchinson et al. 2005); however, it also sprouts prolifically after topkill (Albrecht and McCarthy 2006; Blankenship and Arthur 2006). Maples probably began recruiting into these stands earlier than we document, perhaps much earlier, but presumably were being killed or top-killed until the cessation of fires.

The limited establishment and survival of maples before the 1923 fires (1917 to 1922) in all units may have resulted from several factors. Firstly, wildfires usually burn in a mosaic pattern, particularly in dissected landscapes, resulting in variable fire intensities and including unburned patches. Established maples may have escaped the 1923 fires in unburned patches. We also speculate that the initiation of maple establishment at REMA and Zaleski may have been facilitated by reduced anthropogenic land use, particularly by livestock in open-range woodland livestock grazing (Green 1907). The human population of Vinton County declined steadily with the demise of the iron furnace industry from a maximum of 17 223 in 1880 to 10 287 by 1930 (Vinton County Ohio Genealogy 2005). During the same period, farmland in the county decreased from 93283 to 61559 ha, and most of these lands reverted to forest (Bromley 1934a). Woodland grazing also likely decreased during this period, which would have favored the recruitment of trees, including maples. Brose and Waldrop (2006) showed that the cessation of livestock grazing in the Great Smoky Mountains National Park contributed to increased tree recruitment there in the 1920s and 1930s.

Oak also exhibited pulses of establishment immediately after some fires, suggesting resprouting from previously established stems. However, some fires were not followed by pulses of oak establishment. The final period of oak establishment occurred immediately after the 1923 fires; thereafter, we recorded virtually no additional oak stems. These data suggest that the large increase in maple establishment after the cessation of fires contributed via competition to the lack of subsequent oak recruitment. The absence of fire  after 1923, probably coupled with reduced woodland grazing, also likely facilitated the development of higher stand densities. The resulting closed-canopy conditions that developed would have greatly limited the ability of the relatively shade-intolerant oaks to establish from seed (Beck 1970) but not the shade-tolerant red maple, which can persist for long periods beneath a canopy (Tift and Fajvan 1999). Thus, further oak recruitment from seed, followed by growth and survival, probably was limited by a combination of shading and competition from both overstory trees and understory maples after fires ceased (e.g., Aldrich et al. 2005)

The history of fire and tree establishment at Tar Hollow differed from that of REMA and Zaleski in several aspects. First, at Tar Hollow, there were fewer historical fires (n = 5) and only one was significant. In all, we recorded only 19 historic fire scars at Tar Hollow compared with 48 at Zaleski and 75 at REMA.

The temporal pattern of maple establishment also differed, beginning nearly 40 years earlier (1881) and not exhibiting the distinct pulses after fire cessation that occurred at REMA and Zaleski. Tar Hollow also differed in that there was a long period of fairly continuous oak establishment (ca. 1890 to 1925), that coincided with the continuous recruitment of maples. These differences in fire and regeneration among sites may have resulted from different human land use.

Timber harvesting in the 1800s at Tar Hollow was not associated with charcoal production as at REMA and Zaleski and, thus, may have differed in intensity and extent. Perhaps more important is the evidence of greater and more varied human land use at the Tar Hollow site. In the 1930s, several Land Utilization Project (LUP) areas were established in which the State of Ohio purchased submarginal farmlands and then resettled the occupants (Bromley 1934a, 1934b). The REMA and Zaleski study sites were located near LUP areas while the Tar Hollow site was within the Ross-Hocking LUP. Land titles and appraisals that included detailed ownership and land-use maps from the time of purchase (ca. 1935) indicate that the Tar Hollow site consisted of a number of small parcels of mixed-ownership (ODNR, Division of Forestry, Regional Office, Chillicothe, Ohio). The maps indicate a patchy mixture of cover types: the most abundant was ‘‘forest land (including woodland pasture),’’ but it also included some areas of ‘‘grazing land (grazing or open pasture)’’ and a smaller portion as ‘‘crop land (including orchard and hay meadows).’’ The more varied land uses at Tar Hollow likely would have created a more patchy distribution of disturbances (fire, grazing, and harvesting). In particular, the patchy ownership and land use might have limited fire spread, resulting in the lower observed occurrence of fire. In turn, fewer fires probably facilitated the recruitment and survival of maples beginning much earlier at this site. At Tar Hollow, woodland grazing may have been more prevalent for a longer period with direct human occupation into the mid-1930s, potentially limiting the large pulses of maple establishment that occurred at REMA and Zaleski. Although fires generally were less frequent and wounded fewer trees at Tar Hollow, these units also developed into oak-dominated forests.

Other potential factors affecting tree establishment In addition to fire and human land use (particularly woodland grazing), the regeneration of oak forests was surely influenced by dramatic changes in wildlife populations. The decline and extirpation of the acorn-consuming white-tailed deer (Odocoileus virginianus (Zimmermann)), wild turkeys (Meleagris gallopavo L.), and passenger pigeons (Ectopistes migratorius (L.)) occurred from the mid-1800s to the early 1900s, as these stands were developing (Chapman 1938). Although these declines could have benefited oak regeneration from seed, the consumption of acorns and grazing of seedlings by domestic livestock probably were widespread during this period. No deer were present in Ohio from 1904 to 1922 when a restocking program was initiated (Chapman 1938). By 1938, it was estimated that only 2000 deer were in Ohio (Chapman 1938); the current estimate is 650 000 (ODNR, Division of Wildlife, Columbus, Ohio). Thus, excessive deer browsing clearly would not have contributed to the cessation of oak recruitment ca. 1925. In fact, a lack of deer browsing, the cessation of fire, and a decrease in livestock grazing may have facilitated the dramatic increase in maple establishment during the 1920s and 1930s.

It also is unlikely that American chestnut (Castanea dentata (Marsh.) Borkh.) mortality, caused by the chestnut blight fungus, facilitated the initial recruitment of maple in these stands. Even at REMA and Zaleski where maple recruitment began later (ca. 1920), it predated the arrival of chestnut blight, which caused mortality in Vinton County primarily from 1928 to 1936 (Beatley 1959). Also, chestnut accounted for only 4%-6% of witness trees ca. 1800 (Beatley 1959).

Selective harvesting was common after the mid-1930s on the Zaleski and Tar Hollow State Forests and at the REMA, then owned by D.B. Frampton and Co. (Beatley 1959). Previous work on sites near our REMA sites showed some growth releases suggestive of selective harvesting (Hutchinson et al. 2003). However, our data indicate that, in the absence of periodic fire, canopy disturbances after ca. 1925 did not facilitate oak recruitment. Similarly, during the fire control era, small openings in closed-canopy stands would have favored the growth of maples, which can respond with rapid growth even after long periods of suppression (Tift and Fajvan 1999).

Implications for oak regeneration today
Our results show the past importance of periodic fire in sustaining oak establishment and in limiting maple recruitment as stands developed. However, in many oak forests, there is now an abundance of maples in the midstory that are large enough to be fire resistant. Also, many forests that remain dominated by overstory oaks may be too dense to support oak regeneration even if the maple midstory could be removed with fire. Research has shown that simply returning low-intensity prescribed fires to fully stocked stands does not open the canopy sufficiently to improve the competitive status of oak regeneration (Hutchinson et al. 2005; Blankenship and Arthur 2006). Similar to the past importance of fire in early stand development, oak regeneration has improved when prescribed fire was applied to openstructured stands that developed after partial harvest (Kruger and Reich 1997; Brose and Van Lear 1998; Iverson et al. 2008). However, other studies have shown that the timing and intensity of the mechanical treatments and fire are critical to their success (Franklin et al. 2003; Albrecht and McCarthy 2006). Although fire was important in sustaining oak forests in the past, the legacy of prolonged fire exclusion necessitates research to refine oak regeneration prescriptions that incorporate canopy disturbances, fire, and other tools (Brose et al. 2006).

Acknowledgements

We thank David Hosack, Kristy Tucker, Brad Tucker, Bill Borovicka, Tim Fox, Joan Jolliff, and Zachary Traylor for field and laboratory assistance. We thank Patrick Brose, James Rentch, Ryan McEwan, Marty Jones, and two anonymous reviewers for providing many valuable suggestions on previous drafts of the manuscript. We thank the Ohio Department of Natural Resources, Division of Forestry, for supporting this research on Tar Hollow and Zaleski State Forests; we thank Bob Boyles and Michael Bowden of the Division of Forestry for assistance with historical documents. We also thank Forestland Group, LLC, for supporting this research on the Vinton Furnace Experimental Forest. This is publication No. 172 of the Fire and Fire Surrogate Network Project funded by the Join Fire Sciences Program.

References

Abrams, M.D. 1992. Fire and the development of oak forests. Bioscience, 42: 346-353. doi:10.2307/1311781.

Abrams, M.D. 1998. The red maple paradox: what explains the widespread expansion of red maple in eastern forest? Bioscience, 48: 355-364. doi:10.2307/1313374.

Abrams, M.D. 2002. The postglacial history of oak forests in eastern North America. In Oak forest ecosystems. Ecology and management for wildlife. Edited by W.J. McShea and W.M. Healy. Johns Hopkins University Press, Baltimore, Md. pp. 34-45.

Abrams, M.D., and Copenheaver, C.A. 1999. Temporal variation in species recruitment and dendroecology of an old-growth white oak forest in the Virginia Piedmont, USA. For. Ecol. Manage. 124: 275-284. doi:10.1016/S0378-1127(99)00071-7.

Abrams, M.D., and Downs, J.A. 1990. Successional replacement of old-growth white oak by mixed mesophytic hardwoods in southwestern Pennsylvania. Can. J. For. Res. 20: 1864-1870. doi:10. 1139/x90-250.

Albrecht, M.A., and McCarthy, B.C. 2006. Effects of prescribed fire and thinning on tree recruitment patterns in central hardwood forests. For. Ecol. Manage. 226: 88-103. doi:10.1016/j. foreco.2005.12.061.

Aldrich, P.R., Parker, G.R., Romero-Severson, J., and Michler, C.H. 2005. Confirmation of oak recruitment failure in Indiana old-growth forest: 75 years of data. For. Sci. 51: 406-416.

Beatley, J.C. 1959. The primeval forests of a periglacial area in the Allegheny Plateau (Vinton and Jackson Counties, Ohio). Bull. Ohio Biol. Surv. 1: 1-166.

Beck, D.E. 1970. Effect of competition on survival and height growth of northern red oak seedlings. USDA For. Serv. Southeast. For. Exp. Stn. Res. Pap. SE-5.

Blankenship, B.A., and Arthur, M.A. 2006. Stand structure over 9 years in burned and fire-excluded oak stands on the Cumberland Plateau, Kentucky. For. Ecol. Manage. 225: 134-145. doi:10.1016/j.foreco.2005.12.032.

Boerner, R.E.J., Brinkman, J.A., and Yaussy, D.A. 2007. Ecosystem restoration treatments affect soil physical and chemical properties in Appalachian mixed-oak forests. In Proceedings, 15th Can. J. For. Res. Vol. 38, 2008 Central Hardwood Forest Conference, 27 February – 1 March 2006, Knoxville, Tenn. [CD-ROM.] Edited by D.S. Buckley and W.S. Clatterbuck. USDA For. Serv. Gen. Tech. Rep. SRS-101. pp. 107-116.

Bromley, W.C. 1934a. Preliminary report on the Zaleski rehabilitation and forest project of Vinton County Ohio. Ohio Division of Forestry, Columbus, Ohio.

Bromley, W.C. 1934b. Preliminary report on the submarginal land and state forest project of southeastern Vinton County Ohio. Ohio Division of Forestry, Columbus, Ohio.

Brose, P.H., and Van Lear, D.H. 1998. Responses of hardwood advance regeneration to seasonal prescribed fires in oak-dominated shelterwood stands. Can. J. For. Res. 28: 331-339. doi:10.1139/ cjfr-28-3-331.

Brose, P.H., and Waldrop, T.A. 2006. Fire and the origin of table mountain pine – pitch pine communities in the southern Appalachian Mountains, USA. Can. J. For. Res. 36: 710-718. doi:10. 1139/x05-281.

Brose, P.H., Schuler, T.M., and Ward, J.S. 2006. Responses of oak and other hardwood regeneration to prescribed fire: what we know as of 2005. In Fire in Eastern Oak Forests: Delivering Science to Land Managers. Proceedings of a Conference, 15-17 November 2005, Columbus, Ohio. Edited by M.B. Dickinson. USDA For. Serv. Gen. Tech. Rep. NRS-P-1. pp. 123-135.

Chapman, F.B. 1938. The development and utilization of the wildlife resources of unglaciated Ohio. Ph.D. dissertation, Ohio State University, Columbus, Ohio.

Cook, E.R., and Kairiukstis, L.A. 1990. Methods of dendrochronology. Kluwer Academic Publications, Hingham, Mass.

Cutter, B.E., and Guyette, R.P. 1994. Fire frequency on an oakhickory ridgetop in the Missouri Ozarks. Am. Midl. Nat. 132: 393-398. doi:10.2307/2426595.

Dyer, J.M. 2001. Using witness trees to assess forest change in southeastern Ohio. Can. J. For. Res. 31: 1708-1718. doi:10. 1139/cjfr-31-10-1708.

Franklin, S.B., Robertson, P.A., and Fralish, J.S. 2003. Prescribed burning effects on upland Quercus forest structure and function. For. Ecol. Manage. 184: 315-335. doi:10.1016/S0378-1127(03) 00153-1.

Fritts, H.C. 1976. Tree rings and climate. Academic Press, London. Goebel, P.C., and Hix, D.M. 1997. Changes in the composition and structure of mixed-oak, second-growth forest ecosystems during the understory reinitiation stage of stand development. Ecoscience, 4: 327-340.

Gordon, R.B. 1969. The natural vegetation of Ohio in pioneer days. Ohio State University, Columbus, Ohio. Bull. Ohio Biol. Surv. New Ser. 3. pp. 1-109.

Green, W.J. 1907. First report on forest conditions in Ohio. Bull. Ohio Agric. Exp. Stn. 188: 37-51.

Griffith, D.M., DiGiovanni, D.M., Witzel, T.L., and Wharton, E.H. 1993. Forest statistics for Ohio, 1991. USDA For. Serv. Northeast. For. Range Exp. Stn. Res. Bull. NE-128.

Grissino-Mayer, H.D. 2001a. FHX2—software for analyzing temporal and spatial patterns in fire regimes from tree rings. TreeRing Res. 57: 115-124.

Grissino-Mayer, H.D. 2001b. Evaluating crossdating accuracy: a manual and tutorial for the computer program COFECHA. Tree-Ring Res. 57: 205-221.

Grissino-Mayer, H.D. 2004. FHX2 user’s manual: software for analyzing temporal and spatial patterns in fire regimes from tree rings. University of Tennessee, Knoxville, Tenn.

Grissino-Mayer, H.D., and Swetnam, T.W. 1997. Multi-century history of wildfire in the ponderosa pine forests of El Malpais National Monument. Bull. N.M. Bur. Mines Mineral Resour. 156: 163-172.

Grissino-Mayer, H.D., Romme, W.H., Floyd, M.L., and Hanna, D.D. 2004. Climatic and human influences on fire regimes of the southern San Juan Mountains, Colorado, USA. Ecology, 85: 1708-1724. doi:10.1890/02-0425.

Guyette, R.P., and Spetich, M.A. 2003. Fire history of oak-pine forests in the Lower Boston Mountains, Arkansas, USA. For. Ecol. Manage. 180: 463-474. doi:10.1016/S0378-1127(02) 00613-8.

Guyette, R.P., and Stambaugh, M.C. 2004. Post-oak fire scars as a function of diameter, growth, and tree age. For. Ecol. Manage. 198: 183-192. doi:10.1016/j.foreco.2004.04.016.

Guyette, R.P., Muzika, R.M., and Dey, D.C. 2002. Dynamics of an anthropogenic fire regime. Ecosystems (NY., Print), 5: 472-486. doi:10.1007/s10021-002-0115-7.

Guyette, R.P., Dey, D.C., and Stambaugh, M.C. 2003. Fire and human history of a barren-forest mosaic in southern Indiana. Am. Midl. Nat. 149: 21-34. doi:10.1674/0003-0031(2003) 149[0021:FAHHOA]2.0.CO;2.

Guyette, R.P., Spetich, M.A., and Stambaugh, M.C. 2006. Historic fire regime dynamics and forcing factors in the Boston Mountains, Arkansas, USA. For. Ecol. Manage. 234: 293-304. doi:10.1016/j.foreco.2006.07.016.

Haines, D.A., Johnson, V.J., and Main, W.A. 1975. Wildfire atlas of the northeastern and north central states. USDA For. Serv. Gen. Tech. Rep. NC-16.

Harmon, M.E. 1984. Survival of trees after low-intensity surface fires in Great Smoky Mountains National Park. Ecology, 65: 796-802. doi:10.2307/1938052.

Holmes, R.L. 1983. Computer-assisted quality control in tree-ring dating and measurement. Tree-Ring Bull. 43: 69-78.

Holmes, R.L. 1999. Program JOLTS: finding growth surges or suppression in trees. Laboratory of Tree-Ring Research, University of Arizona, Tucson, Ariz.

Hutchinson, T.F., Rubino, D., McCarthy, B.C., and Sutherland, E.K. 2003. History of forests and land use. In Characteristics of mixed-oak forest ecosystems in southern Ohio prior to the reintroduction of fire. Edited by E.K. Sutherland and T.F. Hutchinson. USDA For. Serv. Gen. Tech. Rep. NE-299. pp 17-27.

Hutchinson, T.F., Sutherland, E.K., and Yaussy, D.A. 2005. Effects of repeated fires on the structure, composition, and regeneration of mixed-oak forests in Ohio. For. Ecol. Manage. 218: 210-228. doi:10.1016/j.foreco.2005.07.011.

Iverson, L.R., Hutchinson, T.F., Prasad, A., and Peters, M.P. 2008. Thinning, fire, and oak regeneration across a heterogeneous landscape in the eastern U.S.: seven-year results. For. Ecol. Manage. In press.  oi:10.1016/j.foreco.2007.09.088.

Johnson, P.S., Shifley, S.R., and Rogers, R. 2002. The ecology and silviculture of oaks. CAB International, New York.
Kruger, E.L., and Reich, P.B. 1997. Responses of hardwood regeneration to fire in mesic forest openings. I. Post-fire communitydynamics. Can. J. For. Res. 27: 1822-1831. doi:10.1139/cjfr-27- 11-1822.

Leete, B.E. 1938. Forest fires in Ohio 1923 to 1935. Bull. Ohio Agric. Exp. Stn. 598.

Lorimer, C.G. 1993. Causes of the oak regeneration problem. In Oak Regeneration: Serious Problems, Practical Recommendations, Symposium Proceedings, 8-10 September 1992, Knoxville, Tenn. Edited by D. Loftis and C.E. McGee. USDA For. Serv. Gen. Tech. Rep. GTR-SE-84. pp. 14-39.

Lorimer, C.G. 1984. Development of the red maple understory in northeastern oak forests. For. Sci. 30: 3-22.

Lorimer, C.G., and Frelich, L.E. 1989. A methodology for estimating canopy disturbance frequency and intensity in dense temperate forests. Can. J. For. Res. 19: 651-663. doi:10.1139/X89-102.

Lorimer, C.G., Dahir, S.E., and Singer, M.T. 1999. Frequency of partial and missing rings in Acer saccharum in relation to canopy position and growth rate. Plant Ecol. 143: 189-202. doi:10.1023/A:1009847819158.

McEwan, R.W., Hutchinson, T.F., Ford, R.D., and McCarthy, B.C.
2007a. An experimental evaluation of fire history reconstruction using dendrochronology in white oak (Quercus alba). Can. J. For. Res. 37: 806-816. doi:10.1139/X06-294.

McEwan, R.W., Hutchinson, T.F., Long, R.P., Ford, R.D., and McCarthy, B.C. 2007b. Temporal and spatial patterns of fire occurrence during the establishment of mixed-oak forests in eastern North America. J. Veg. Sci. 18: 655-664. doi:10.1658/1100 9233(2007)18[655:TASPIF]2.0.CO;2.

McNab, W.H., and Avers, P.E. 1994. Ecological subregions of the United States: section descriptions. USDA For. Serv. Wash. Off. Admin. Publ. WO-WSA-5. Ohio Experiment Station. 1922. 42nd annual report. Bull. Ohio Ag-
ric. Exp. Stn. 373.

Phipps, R.L. 1961. Analysis of five years of dendrometer data obtained within three deciduous forest communities of Neotoma. Neotoma Ecological and Bioclimatic Laboratory, Ohio Agricultural Experiment Station, Wooster, Ohio. Res. Circ. 105. pp. 3-34.

Regelbrugge, J.C., and Smith, D.W. 1994. Postfire tree mortality in relation to wildfire severity in mixed oak forests in the Blue Ridge of Virginia. North. J. Appl. For. 11: 90-97.

Schuler, T.M., and McClain, W.R. 2003. Fire history of a Ridge and Valley forest. USDA For. Serv. Northeast. Res. Stn. Res. Pap. NE-724.

Shumway, D.L., Abrams, M.D., and Ruffner, C.M. 2001. A 400 year history of fire and oak recruitment in an old-growth oak forest in western Maryland, USA. Can. J. For. Res. 31: 1437 1443. doi:10.1139/cjfr-31-8-1437.

Smith, K.T., and Sutherland, E.K. 1999. Fire-scar formation and compartmentalization in oak. Can. J. For. Res. 29: 166-171. doi:10.1139/cjfr-29-2-166.

Soucy, R.D., Heitzman, E., and Spetich, M.A. 2005. The establishment and development of oak forests in the Ozark Mountains of Arkansas. Can. J. For. Res. 35: 1790-1797. doi:10.1139/x05 104.

Stokes, M.A., and Smiley, T.L. 1968. An introduction to tree-ring dating. University of Chicago Press, Chicago, Ill.

Stout, W. 1933. The charcoal iron industry of the Hanging Rock Iron District—its influence on the development of the Ohio Valley. Ohio Archaeol. Hist. Q. 42: 72-104.

Sutherland, E.K. 1997. The history of fire in a southern Ohio second-growth mixed-oak forest. In Proceedings of the 11th Central Hardwoods Forest Conference, 23-26 March 1997, Columbia, Mo. Edited by S.G. Pallardy, R.A. Cecich, H.E. Garrett, and P.S. Johnson. USDA For. Serv. Gen. Tech. Rep. NC-188. pp. 172-183.

Sutherland, E.K., Hutchinson, T.F., and Yaussy, D.A. 2003. Introduction, study area description, and experimental design. In Characteristics of mixed-oak forest ecosystems in southern Ohio prior to the reintroduction of fire. Edited by E.K. Sutherland and T.F. Hutchinson. USDA For. Serv. Gen. Tech. Rep. NE-299, pp. 1-16.

Tift, B.D., and Fajvan, M.A. 1999. Red maple dynamics in Appalachian hardwood stands in West Virginia. Can. J. For. Res. 29: 157-165. doi:10.1139/cjfr-29-2-157.

Van Lear, D.H., and Brose, P.H. 2002. Fire and oak management.  In Oak forest ecosystems. ecology and management for wildlife. Edited by W.J. McShea and W.H. Healy. Johns Hopkins University Press, Baltimore, Md. pp. 269-279.

Vinton County Ohio Genealogy. 2005. Vinton County Ohio population changes 1850-1990 [online]. Available from www.rootsweb.com/~ohvinton/vintpop.gif.

Wendel, G.W., and Smith, H.C. 1986. Effects of prescribed fire in a central Appalachian oak-hickory stand. USDA For. Serv. Northeast. For. Range Exp. Stn. Res. Pap. NE-RP-594.

Yaussy, D.A., Hutchinson, T.F., and Sutherland, E.K. 2003. Structure, composition, and condition of overstory trees. In Characteristics of mixed-oak forest ecosystems in southern Ohio prior to the reintroduction of fire. Edited by E.K. Sutherland and T.F. Hutchinson. USDA For. Serv. Gen. Tech. Rep. NE-299. pp. 99-111.

Dynamics of an Anthropogenic  Fire Regime

R. P. Guyette,1* R. M. Muzika,1 and D. C. Dey2

1Department of Forestry, 203 Natural Resources Building, University of Missouri, Columbia, Missouri 65211, USA
2USDA North Central Research Station, Columbia, Missouri 65211 USA

 


ABSTRACT

Human interaction with fire and vegetation occurs at many levels of human population density and cultural development, from subsistence cultures to highly technological societies. The dynamics of these interactions with respect to wildland fire are often difficult to understand and identify at short temporal scales. Dendrochronological fire histories from the Missouri Ozarks, coupled with human population data, offer a quantitative means of examining historic (1680-1990) changes in the anthropogenic fire regime. A temporal analysis of fire scar dates over the last 3 centuries indicates that the percent of sites burned and fire intervals of anthropogenic fires are conditioned by the following four limiting factors: (a) anthropogenic ignition, (b) surface fuel production, (c) fuel fragmentation, and (d) cultural behavior. During an ignition-dependent stage (fewer than 0.64 humans/km2), the percent of sites burned is logarithmically related to human population (r2 0.67). During a fuel-limited stage, where population ensity exceeds a threshold of 0.64 humans/km2, the percent of sites burned is independent of population increases and is limited by fuel production. During a fuel-fragmentation stage, regional trade allows population densities to increase above 3.4 humans/km2, and the percent of sites burned becomes inversely related to population (r2 0.18) as decreases in fuel continuity limit the propagation of surface fires. During a culture-dependent stage, increases in the value of timber over forage greatly reduce the mean fire interval and the percent of sites burned. Examples of the dynamics of these four stages are presented from the Current River watershed of the Missouri Ozarks.

Key words: human population density; Ozarks; Missouri; disturbance; dendrochronology; fire regimes.

CLICK HERE TO DOWNLOAD AS PDF


INTRODUCTION

Anthropogenic, or human-caused, fire has influenced ecosystem processes for millennia, and at broader scales has been an important determinant of landscape character and global carbon cycles (Pyne 1995; Delcourt and Delcourt 1997; Bird and Cali 1998). Despite progress in recent years in understanding how the spatial and temporal dynamics of wildfire shape natural ecosystems (Swetnam 1993; Agee 1993; Whelan 1995), we know little about the ways in which changes in human population and culture alter the use of anthropogenic


Received 24 January 2001; accepted 9 January 2002.
*Corresponding author; e-mail: guyetter@missouri.edu


fire and consequently how that dynamic affects ecosystem pfrocesses and attributes.

Anthropogenic fire occurs throughout a range of human-related activities and settings, from widely scattered  populations living in subsistence cultures to densely settled urban societies; therefore, anthropogenic fire is both temporally and spatially dynamic. Within a region, a temporal dynamic in anthropogenic fire is driven by changes in population and culture that occur over decades and centuries. In North America, dramatic temporal changes in the frequency of anthropogenic fire over the past 400 years have resulted from the displacement of aboriginal cultures and populations by high-density European populations and industrial economies (Pyne 1982; Williams 1989). Human activities strongly influence fire frequencies in such areas as Patagonia (Veblen and others 1999) and the Amazon basin (Kauffman and Uhl 1990); and human populations account for variation in landscape-level fire  egimes in North America (Turner and Romme 1994; Guyette and Dey 2000; Dey and Guyette 2000). In the  Midwest, historical narratives suggesting the use of fire by aboriginal people have been summarized by Ladd (1991), and archacological evidence supports the presence of human populations in Missouri for many thousands of years (Marriott 1974; Price and others 1987; O’Brien and Wood 1998).

Many fire ecologists and historians have defined the concept of fire regime (Pyne 1982; Agee 1993; Whelan 1995; Pyne and others 1996). According to Pyne and others (1996), “A fire regime is intended to characterize the feature of historic, natural fires that have been typical for a particular ecosystem or set of ecosystems.” In many regions, such as the Ozark physiographic province in midwestern North America, human presence on the landscape (O’Brien and Wood 1998) predates contemporary forest-vegetation associations and climatic conditions (Delcourt and Delcourt 1987, 1991); therefore, any characterization of the “natural” fire regime should incorporate all organisms in the ecosystem and all sources of fire, including humans and their ignitions. For the purposes of this analysis, we define an anthropogenic fire regime as patterns of wildland fire shaped by the dynamic interactions of vegetation (fuels) and human populations (ignitions). This definition is functional and consistent with the standard dictionary definition of a regime as “a system of rules”, or in our case, factors that govern the frequency and intensity of fire. Although fire regimes are often defined for specific ecosystems that have a particular combination of vegetation (fuels), climate, and a certain frequency of ignition, the perception that an anthropogenic fire regime is a static combination of these factors may be unrealistic because of the extreme variability in time and space of fluctuating human populations and cultures.

The goal of the research presented here is to demonstrate that change in human population density and culture has been one of the major factors influencing the frequency and effects of wildland fire over the last 320 years in southeastern Missouri. The objectives of this paper are to (a) describe the historic changes in an anthropogenic fire regime; and (b) elucidate the limiting factors that affect these changes through time, including the relationships among human population density, culture, and fuels as determinants of fire regimes. To fulfill these objectives, we used dendrochronological fire histories coupled with data on human population density from AD 1680 to 1998 in the Current River watershed, a topographically highly dissected section of the Ozark Highlands in Missouri, USA.


A Temporal Anthropogenic Framework

The oak-pine forests of the Ozark Highlands offer an appropriate setting to document changes in human-fire interactions over hundreds of years. Schroeder and Buck (1970) estimated that less than one lightning fire per 4000 km2 occurs annually in the Missouri Ozarks. Missouri State Fire Protection records between 1970 and 1989, summarized by Westin (1992), indicate that an average of 108 fires per year per 4000 km2 occurred in the region of the Current River watershed. Of these, less than 1% were of lightning ignition origin; therefore, the overwhelming majority of these fires were caused by humans. The lack of “natural” fires in the Current River region, a temporally variable human population, and variable topography make this region ideal for studying change in anthropogenic fire regimes and its effects on ecosystems. Fire scars from primarily low-intensity surface fires offer high-resolution spatial-temporal data on fire frequency and extent and can be tree-ring-dated on survivor trees and woody remnants (Guyette and McGinnes 1982; Guyette and Cutter 1991; Guyette and Dey 1997a; Jenkins and others 1997). Finally, a highly dissected topography inhibits the propagation of fire, thus making mean fire intervals and the percent of sites burned sensitive to the number and distribution of anthropogenic ignitions. The foregoing evidence suggests a close association between humans and fire in the Ozarks. We hypothesize that identified changes in the percent of sites burned can be attributed to changes in human population density and cultural behavior. Consequently, we propose that a temporal framework can be used to describe these changes. In this analysis, we examine whether fire frequency is a function of ignitions (anthropogenic), fuels, and cultural behavior based on economics and technology.

Study Area

The study area is in the upper Current River watershed (including the Jack Fork River), is heavily forested (more than 80%), and measures about 4316 km2 in area (Figure 1). The study area is located near the western edge of the eastern deciduous forest and is dissected by steep ridges and numerous streams. The slope of the sample sites averages 18° and ranges from 10° to 32°. Elevations in the study area range from 140 to 414 m a.s.l. The climate of the study area is humid and continental. Precipitation ranges from 60 to 152 cm and averages 115 cm per year. Spring is the wettest season, followed by fall, summer, and winter. During the fall, winter, and spring of most years, dry warm weather during only a few days may be sufficient to dry surface fuels and permit the spread of surface fires. Fires during the growing season are rare but do occur during very hot and dry summers. Natural ignition is rare despite an abundance of thunderstorms (50-70 thunderstorm days per year) (Baldwin 1973).

Figure 1. Study area, with fire history sites, historic tree species associations, mean fire intervals, topography, archeological sites, and historic sites. The location of the fire history sites and their mean fire intervals (1700-1850) are indicated by the mapped numbers in black and white. Mean fire intervals in white text represent sites where the interval decreased by more than 10 years during the ignition-dependent stage; mean fire intervals in black indicate a decrease of 0-10 years. The percent of canopy closure defines forests (more than 75%), woodlands (25%-75%), and savannas (less than 25%). Mesic mixed oak-riparian forest is more than 25% mesic species or more than 25% riparian species; mixed oak forest is more than 75% oak and hickory; oak-pine woodland/forest is oak with more than 25% and less than 65% pine; pine savanna/woodland is more than 65% pine; post oak savanna is more than 75% post oak; post oak-black oak savanna is more than 75% post and black oak. Tree species associations are based on General Land Office Survey Notes as interpreted by Porter (1998), Batek and others (1999), and Hughes and Nigh (2000). Archaeological and historic sites are based on data from Price and others (1987), Lynott (1989), and Stevens (1991).


METHODS

Fire History Development

Site locations within the study area were chosen based on the presence of fire-scarred wood, and the majority of fire-scarred wood was found in steep terrain. Thus, the sites are not necessarily truly representative of the area as a whole. Twenty-three of the sites were located in areas dominated historically by oak (Quercus species) forests; four other sites occurred in forests historically dominated by shortleaf pine (Pinus echinata) (Batek and others 1999). Oak has become a dominant species in the overstory of all the fire history sites.

Cross sections of more than 257 shortleaf pine remnants (cut stumps, natural snags, downed trees) were cut between ground level and 30 cm. Wedges were cut from the scar face of live trees. Fire scars were identified by callus tissue, traumatic resin canals, and cambial injury. All samples had charcoal present on the scarred exterior. Scars were dated to the 1st year of cambial injury. Ring-width series from each sample were measured and plotted by year. Ring-width plots were used for visual crossdating (Stokes and Smiley 1968; Guyette and Cutter 1991). The COFECHA computer program (Grissino-Mayer and others 1996) was used to assist in ensuring the accuracy of both relative and absolute dating of the samples by correlation analysis. Absolute dating of the pine remnants was accomplished by cross-dating with a ring-width chronology (Guyette 1996) based on live shortleaf pine growing in Shannon County, Missouri, within the study area. Over 2500 scars were identified and dated from 27 sites. Site-level data  were averaged into a regional composite of the percent of sites burned. An 11-year moving average was applied to percent of sites burned to reduce variability in the time series due to changes in annual climate and to enhance long-term trends in the data. An 11-year moving average was chosen because it preserved variability on a decadal scale, it is symmetrical, and it corresponds to the decadal census data. Correlation coefficients between the percent of sites burned and human population density were calculated without the moving average and were adjusted for autocorrelation in the time series of the percent of sites burned. A representative subset of the data from a site in the study watershed (Figure ) is graphed using FHX2 software (Grissino-Mayer 1995) and illustrates the structure of the data and the changing fire return interval over nearly 4 centuries.

Figure 2. Fire scar data plotted by tree sample and calendar year for a 1-km2 area near Blue Spring along the Current River in Missouri. Each horizontal line represents the tree-ring record of a fire-scarred shortleaf pine tree, stump, or natural snag. Dendrochronologically dated fire scars are represented by vertical lines. A composite fire scar chronology (all fire scar dates) is plotted at the bottom of the graph.


Topographic Roughness

Irregularities in the landscape, or “topographic roughness”, can contribute to the fire behavior in an area of highly dissected topography. The rate of spread of a low-intensity surface fire may decrease because fire burns slower down steep slopes; because fuel continuity is broken by creeks, rivers, and rocky outcrops; or because fuel moisture content increases on northern aspects. Indexes of topographic roughness were used to reflect topographic inhibition of the propagation of fire across the landscape. These indexes were developed by comparing surface area measurements made with two different-sized scales. A circle 5000 m in diameter is marked on a digital elevation map. The surface area of the earth circumscribed by this circle is calculated from pixels that are 30 m on a side. Their slope and a trigonometric conversion are used to estimate the area of the uneven land surface. The pixels are summed to estimate the surface area of the landscape enclosed by the circle. This measure is then divided by the planimetric surface area (the large scale in this case) of a circle that is 5000 m in diameter. This ratio of the actual surface area to the planimetric surface area is the Index of Topographic Roughness. A correlation analysis was used to document changes in the influence of topographic roughness on mean fire intervals by stage of the anthropogenic fire regime.

Human Populations

Population since 1820 was derived for the study area from United States Census data (Shannon County, Missouri) and population estimates and maps in Stevens (1991) and Rafferty (1982). Historic Native American population density was estimated from many sources (Table 1 and references therein). The population density of each group in the Current River watershed was calculated by dividing their historical population estimates by the area of a circle whose radius was the distance between their population center and the watershed. Changes in the location and territory of populations documented in the historical literature were also used to estimate population trends of groups. Although this method does not take into account the great variation of population density within a tribal territory, it does provide an estimate of changes in population density through time that have affected  he population of the Current River watershed. These population density estimates are consistent with aboriginal population densities (0.07-6 humans per km2) given for the Great Lakes region and eastern North America (Kroeber 1934; Dobyns 1983; Ramenofsky 1987; Thornton 1987) and with mapped estimates by Paullin (1932) for the period after 1790. The relative spatial distribution of population within the watershed has remained consistent over time and reflects spatial patterns in past fire occurrence and the recent population of towns (Guyette and Dey 2000). Linear interpolation was used to estimate annual population from decadal census data.


DISCUSSION

The Dynamics and Sequence of the Fire Regime 

A historic and sequential interpretation of fire, human population, and culture (Figure 3) is essential to an understanding of the events that have occurred in the study area over the last 350 years. Although fire historyresearch often ends with the identification of fire intervals and ecological relationships, such findings represent only a first step in identifying the processes and variables underlying changes in ecosystem processes. When population and fire data identical to those in Figure 3 are plotted with axes that are independent of time (Figure 4a), the resulting pattern reflects critical changes in the relationships between humans and the environment (fire). The changing slope and direction of the plotted data reflect four temporal stages that are built upon the interaction among ignitions, fuels, and topography as a function of human population density.

The temporal progression of limiting factors defining these stages in the anthropogenic fire regime involves (a) human ignitions , (b) fuel availability , (c) fire propagation and fuel continuity , and (d) cultural values.  Consequently, these factors define four stages: ignition-dependent, fuel-limited, fuel-fragmentation, and culture-dependent (identified in Figure 4a). All stages are linked to population density in the way that population affects ignitions and fire frequency (the percent of sites burned), as well as the ways in which that relationship is modified by culture. Descriptions of these stages, based on limiting factors and interpretations, follow.

Ignition-dependent (Stage 1). The anthropogenic fire regime is population-dependent during this stage and directly related to human population. Human population density is one of the most important factors in an anthropogenic fire regime, especially in early stages or at low levels of human population density. Consequently, ignition sources are the most limiting factor influencing the percent of sites burned during this period. During this stage (before 1850), low population densities (fewer than 0.64 humans/km2) limited the frequency and distribution of anthropogenic sources of ignition in the Current River watershed. Correlation between percent of sites burned and human population density (nontransformed data) is strongest during this stage (Table 2). Verification of this relationship is demonstrated by a positive nonlinear relationship between the percent of sites burned and human population density. During the ignition-dependent stage, percent of sites burned is related (r2 0.67,  0.01) to the natural logarithm of human population density at low levels (fewer than 0.64 humans/km2) by:

F%    0.32   0.069(ln[P])       (1)

where F% is the percent of sites burned and P is the human population density.

Topographic roughness inhibits the spatial extent and spread of fire during this stage. The degree of topographic roughness and the frequency of fire at each site are inversely correlated (Table 2). Both endogenous and exogenous factors play roles during this stage, since population levels at the study  sites were controlled by local human reproduction, human migration, technological development, and introduced disease (Table 2). This stage is the longest in duration over the past 3 centuries and includes notable fire years (see Appendix).

Table 1. Population Densities (Humans per km2) Estimated and Measured by Decade and Cultural Group


DateAllQuapawOsageCherokeeDelawareShawneeEuropean
16500.0470.0421a0.0049f
16600.0470.04210.0049
16700.0500.04210.0082g
16800.0520.04210.0099
16900.0530.04210.0115
17000.0270.0139b0.0132
17100.0270.01260.0148
17200.0380.01130.02640.0001
17300.0380.00840.02970.0007
17400.0420.00840.03310.0014
17500.0450.0063c0.03640.0028
17600.0480.0049d0.03970.0041
17700.0610.00440.0430h0.0046l0.0002n0.0033r0.0055
17800.13400.0039.03970.01850.03200.03200.0082
17900.1670.00340.0364i0.4630.03200.03840.0110
18000.22100.00290.03470.09270.03200.04490.0137
18100.20200.00290.03310.06480.03200.05130.0179
18200.6620.00240.0198j0.0289m0.3012p0.0926s0.2175u
18300.3710.0009e0.00660.02310.0033q0.09260.2447
18400.2940.0001k0.01540.00660.2719
18500.6420.6420v
18600.8730.8736
18700.8990.8998
18801.3231.3237
18903.4233.4231
19004.3264.3268
19104.4024.4022
19204.5644.5645
19304.1914.1910
19404.5514.5515
19503.6383.6389
19602.7262.7264
19702.8862.8853
19803.0333.0334
19903.0733.0738

The sum of the densities is given in the column labeled “All.” Superscripts refer to source data and type by period.
Not all figures are significant.

aPopulation estimate (Baird 1980)
b,c,dPopulation reductions by disease (Ramenofsky 1987; Baird 1980) eMigration out of the watershed (Baird 1980)
fTerritory just west of Current River (Bailey 1973; Wiegers 1985)
gAcquisition of horse (Wiegers 1985; Waldman 1985)
hMaximum expansion of territory (Bailey 1973; Wiegers 1985)
iPopulation estimates (Marriott 1974; Banks 1978)

jTerritorial reduction (Wiegers 1985) and movement west by treaty (Marriott 1974; Banks 1978) kLast Osage removed from Missouri (Wiegers 1985)
lPopulation estimates, trends, movement into Missouri (Gilbert 1996; Stevens 1991) mMigration to Arkansas (Pitcaithley 1978)
nMovement into east Missouri (Stevens 1991)
oPopulation estimates (Marriott 1974)

pMigration across Missouri River, population estimate, and encampment on the Jacks Fork of the Current River (Weslager 1978) qMovement out of Current River watershed (Stevens 1991; Weslager 1978)
rMovement into east Missouri (Stevens 1991)
sMovement out of southeast Missouri (Howard 1981) tSpanish census data (Gerlach 1986)
uEarly settlement (Stevens 1991)
vShannon County census data (Stevens 1991)

Figure 3. (a) The percent of 26 fire history sites in the Current River watershed that were burned each year (dotted line) and an 11-year moving average of the percent of sites burned annually (solid line) plotted by calendar year. The text above the graph identifies the corresponding cultural periods. (b) A record of the human population density and the population densities of individual cultural groups (see Table 1) in the Current River watershed and the Ozark Highlands (log scale) by calendar year.


Fuel-limited (Stage 2). The anthropogenic fire regime is limited by fuel availability in this stage; within limits, population has no effect. Fire reduces surface fuels in Missouri by more than 50% for up to 2.5 years (Scowcroft 1965). During the fuel limited stage (1850-90), human population exceeded 0.64 humans/km2 in the study area, and the percent of sites burned became independent of in reases in human population density. The percent of sites burned during this stage is limited by the availability and production of surface fuels. Frequent fire created open woodlands with grassy herbaceous understories; therefore, biomass accumulation was limited by the short fire return interval. But by 1850 a threshold in population density (0.64 humans per km2) was reached in terms of burning and fuels wherein the environment became ignition-saturated and the fire regime was fuel limited. The great number of ignitions during this stage nullified the effects of topographic roughness in inhibiting the propagation of fire; hence, topographic roughness became less imporant as a factor controlling fire (Table 2), as illustrated by the lack of significant correlations in stages 2, 3, and 4 of the fire regime. During this stage, endogenous factors that limit fire frequency change from human ignitions to fuel production. This is the shortest of the four stages; it is less than 25% as long as the previous stage. Within the fuel limited stage, a broad range of human population densities can maintain a consistently short fire interval; however, permanent fuel elimination does not characterize this stage.

Figure 4. (a) Scatterplot illustrating the change in the relationship between the percent of sites burned annually and human population density through time. These relationships help to identify the stages in an anthropogenic fire regime. Data points in the scatterplot are labeled with decadal calendar dates to identify the temporal sequence. The scatterplot data are the same data represented by the solid lines in Figure 3a and b. (b) The regional and exogenous population density (state of Missouri) plotted with the percent sites burned during the 20th century contrasts with the endogenous population trends of the Current River watershed.


Table 2. Attributes and Interactions of a Four-phased Anthropogenic Fire Regime


 

Interactions and AttributesStage 1: Ignition-dependent (1680-1850)Stage 2: Fuel-limited (1851-90)Stage 3: Fuel-fragmentation (1891-1940)Stage 4: Culture-dependent (1940-90)
Key Factors Limiting FireNumber of human ignitionsAmount and production of fuelDiscontinuity of fuels,propagation of fireCultural attitude, fire suppression, value of wood over forage
Origin and Examples of
Factors that Influence
Development
Endogenous: human population
change
Exogenous: human migration,
introduced
Endogenous: fuel
production, ignition
saturation
Endogenous: fuel continuity,
human incursions
Exogenous: regional
economics, resource
exploitation
Exogenous: cultural value
of forests, value of
timber versus forage
Population (humans km          2)<0.03-0.640.64-3.43.4-4.64.6-2.7
Average and Range of
Mean
10 (2.3-45)3.5 (1.5-6.8)5.8 (1.7-19)>20 (6.8-50)
Mean and Range (PercentSites Burned Annually)b 12 (0-60)38 (10-55)29 (0-51)10 (0-32)
Correlation (Population,Percent Sites Burned)r =0.63, P <0.01r =0.05, P> 0.05r =0.30, P >0.05r =0.52, P <0.01
Culture (price ratio: (timber/livestock)Insufficient data2.14.37.2
Correlation: (Topography and Mean Fire Interval)c(1580–1700) r= 0.71
(1701–1850)r=        0.43 P< 0.05
r = 0.14, P>  0.05r =0.15, P> 0.05Insufficient data
Ca in Woodb  (  g/g)678755700650
Pine Abundanced Culture and Land Use51 stems per ha Aboriginal American, hunting,and gatheringEuro-American settlement and agriculture17 stems per ha Euro-American, logging,grazing, and agricultureAmerican, forestry, recreation, and agriculture

aGuyette and Cutter 1997; Batek and others 1999
bGuyette and Cutter 1997
cGuyette and Dey 2000
dGuyette and Dey 1997b

Table 3. Agricultural Statistics Indicating Increases in Land Uses that Fragmented the Continuity of the Surface Fuel Environment Between Stage 2 and Stage 3a


 

DecadeHogs per km2Cattle per km2Improved landa(%)
18500.870.381.3
18602.51.82.2
18704.11.72.2
18808.62.26.3
18907.85.211.6
19008.42.614.2
19108.17.317.5
19206.68.419.7
19304.07.417.2
19406.27.326.7

a“Improved land” includes pasture fenced land, orchards, crops, and fallow fields (Jacobson and Primm 1997).

Fuel-fragmentation (Stage 3). During the fuel fragmentation stage (1890-1940), regional trade, railroads, and markets allowed population densities to increase above 3.4 humans/km2. A peak in human population density within the watershed (about 4.6 km2) was reached in the 1920s. Increases in human population density led to increases in the number of artificial fuel breaks caused by livestock grazing, road building, and the conversion of field and forest to crop land and pasture (Table 3). This development led in turn to a reduction in the percent of sites burned because it inhibited the propagation of low-intensity surface fires, the mode of fire in the previous two stages. The percent of sites burned became inversely related to population density (r2 0.18, P 0.05) because the continuity of wildland fuels became fragmented by agricultural and rural development. A decrease in the percent of sites burned coincided with increasing population density between 1890 and 1940. Thus, the percent of sites burned, once a function of the ability of fires to spread and increase in size, was limited by the decreased propagation of surface fires across the ndscape due to fragmentation of the fuel environment. The fragmentation stage can be followed in subsequent stages by the elimination of wildland  uels, as might occur through urban or agricultural development.

Culture-dependent (Stage 4). During this stage, cultural values, practices, and technology influence human nvironmental interactions. Cultural attitudes toward the benefits, costs, and dangers of wildland fire are second only to gross population density in importance to anthropogenic fire regimes. During the culture-dependent stage (1940- 96), increases in the value of timber (to a growing exogenous human population) (Figure 4b) relative to that of pasture resulted in attitudes and cultural constructs that reduced the frequency of fire. The price ratio of wood products (Anonymous 1965; Gregory 1972) to livestock (data provided by the Missouri Agricultural Statistical Service) was highly correlated with the percent of sites burned and probably reflects changes in attitudes about the value of forests versus livestock range. The low forage value of forest lands for livestock and increasing demand for wood products were important factors that inspired education on the economic destructiveness of wildland fire, the desire to protect forested lands, and the institution of fire suppression over the last 60 years in the Ozark Highlands.

Attributes of the Stages

Throughout all four of the stages of the anthropogenic fire regime, the factors that limit fire change along the temporal framework. Both causal and resulting factors exert varying influences on the character of the fire regime, depending on the stage. We have identified several components of the anthropogenic fire regime that figure prominantly in determining the stage or represent a substantial response in any or all stages. Although the implications of changing fire regimes are numerous, we  will focus here on a few interactions, effects, and attributes that are useful in elucidating the complexity of the human-fire interaction.

Endogenous and Exogenous Factors. A classification of important endogenous and exogenous factors by stage (Table 2) shows that endogenous factors predominate in early stages, and that exogenous factors become more important in later stages. For the purposes of this perspective, we consider humans to be an integral part of the Current River ecosystem; therefore, anthropogenic factors within the ecosystem are endogenous, and anthropogenic factors outside the ecosystem are exogenous. Significant endogenous factors during early stages include indigenous human population growth and technological innovation. Introduced diseases, however, were a major exogenous factor during the ignition-dependent stage. Endogenous factors pertinent in the early stages of development include fuel abundance, fuel type, and vegetative change and succession. The fuel-limited stage is the only stage where there are no important contributions by exogenous factors. Exogenous factors resulting from human population changes, predominantly migration and population growth, contribute to the fuel-fragmentation and culture dependent stages. Throughout the stages, the change from endogenous to exogenous influence is accompanied by increases in human population, transportation, and communication.

Vegetation-Fire Interactions. There are several tree species that provide particularly useful evidence of the effects of the stages of the fire regime on vegetation because of their longevity and fire sensitivity. Using data compiled from land survey records (circa 1830), the mean fire interval at 23 sites (Batek and others 1999) in the Current River watershed was positively and significantly correlated with the abundance of Pinus echinata and neg atively correlated with the abundance of Quercusvelutina (black oak). These relationships imply that fire frequency may have been a factor influencing vegetation during stage 1 (Table 2). Reductions in the mean fire interval during stages 2 and 3 (1820- 1940) to near annual burning in some areas (Guyette and Cutter 1997), combined with the logging of pine, may have inhibited pine regeneration, reduced the abundance of advanced (prelogging) reproduction, and decreased the amount of mature pine (Record 1910). Currently, pine abundance is only 34% of its historic levels in some areas of the Current River watershed (Guyette and Dey 1997a).

The past and present distribution of Juniperus virginiana (eastern redcedar), a fire-sensitive species, may reflect the intensity and frequency of fires during the ignition-dependent stage. Circa 1840, surveyors noted redcedar on sites with long fire intervals but made no mention of redcedar on sites with a short fire interval (Batek and others 1999). Old (200 years) redcedar persist to the present day on sites that had longer fire intervals during the ignitionependent stage, as documented by the fire scar record. In contrast, there are no old and few young eastern redcedar on sites that had short fire intervals during the ignition-dependent stage. Abundance of Quercus stellata (post oak), a fire-tolerant and shade-intolerant species, as determined from surveyor notes, is significantly correlated with mean fire intervals during the ignition-dependent stage. The ignition-dependent stage has the greatest spatial and  temporal variability in the percent of sites burned and mean fire intervals (Table 2); it is the disturbance regime under which recent vegetation associations have developed for centuries.

Figure 5. Scatterplots illustrating the similarity in pattern between the percent sites burned and heartwood calcium concentration (Juniperus virginana) as related to human population density. Labels on the graph data points represent bidecadal calendar years.


Stage-related Calcium Dynamics. Wildland fire is a chemical process that can cause sudden changes in the nutrient status of a site. Calcium (Ca) is a necessary plant macronutrient whose bioavailability has been shown to increase after fires (DeBano and others 1977; Zinke 1977; Agee 1993) and over longer periods by increasing the rate of Ca cycling in organic debris that is frequently burned (Alban 1977; Binkley and others 1992). An association between trends in Ca availability and mean fire intervals over a 340-year period was inferred from Ca in growth increments of eastern redcedar heartwood growing in the Current River watershed (Guyette and Cutter 1997). Although eastern redcedar is a fire-sensitive species, old individuals of this species persist where they have some protection from the lethal effects of fire—for example, on rhyolite glades, where there is extensive barren rock surface and fire intensity is low. In addition, eastern redcedar has many anatomical and ecological characteristics uniquely suited for dendrochemical studies (Cutter and Guyette 1993) and has been used to reconstruct many changes in environmental chemistry (Guyette and others 1989, 1991, 1992; Guyette and Cutter 1994).

The association between heartwood Ca and human population density (Figure 5) suggests that the stages of the fire regime may have different effects on nutrient cycling. The Ca chronology of dated redcedar heartwood increments was highly correlated with the 20-year grouped means of the percent of sites burned (r 0.81) and trees scarred(r 0.77). During the ignition-dependent stage,Ca is variable and increases steadily with small increases in human population density. Ca concentrations in the heartwood are highest during the fuel-limited stage, the period with the lowest mean fire interval and the greatest percent of sites burned (Table 2). During this stage, the availability of Ca increases owing to the release of Ca in organic litter by the rapid abiotic decomposition mechanism of frequent low-intensity fires. As fire frequency declines in subsequent stages, Ca concentrations decrease in wood.

Human Population Density and Topographic Interactions. Certain stages in an anthropogenic fire regime are the result of complex interactions between human population density and topographic roughness. At low population densities, the percent of sites burned increases with population density (Fig ure 4a). Topographic  controls on the frequency of fire become less important as population density and the frequency and distribution of anthropogenic ignitions increases (Table 2). The forcing factor of the ignition-dependent and fuel-limited stages is the interaction of human population density and topographic roughness. A topographically smooth landscape, such as a large plain or plateau (often prairies), might require only a few humans to reach and maintain a fuel-limited stage. On the other hand, many topographically rough landscapes, such as forests in the Ozark Highlands, require a relatively high human population density to reach and maintain a fuel-limited stage.

Culture and Fire Stages. In the later stages that we have identified, factors controlling the regime change from environmental to cultural. Fuel and local human population density, possibly irrespective of ignition motivation, govern the fire regime in the early stages. Subsequently, during the fuel fragmentation stage, cultural artifacts—such as roads and agricultural development—begin to affect the spread, frequency, and size of fires. Artificial fire breaks in an ignition-saturated environment are now replacing the natural role of topographic roughness as an inhibitor of the spread of fire. Economic values, coupled with the technology of fire suppression, become dominant during the cultural-dependent stage and account for the lowest frequency of fire across all stages.

Human Risk and Fire Stages. The culture-dependent stage may be unstable. During this stage, the cultural separation of ignitions and fuels continues to have increasingly serious implications for human societies that live in environments with highly volatile fuels, particularly those with growing human populations. The accumulation of fuel, ignition potential, and increasing human occupation of the landscape act together to increase the potential risk to human life and property. The most important factors affecting the dynamics of the stages and their associated risk to human life and property are conceptualized using the dynamic interactions of humans and fuels (Figure 6a, b, c). Using a theoretically extended temporal framework, we developed a conceptual model to examine the overall risk to humans (Figure 6d).

Figure 6. Theorized time series of factors influencing the dynamics of the stages of the fire regime (a, b, c). The
variables associated with each of the stages include I: human population density, Fa: total surface fuel accumulation, and Fc: relative fuel continuity. The conceptual model of risk (d) to humans, based on calculations from the formula presented, represents the contribution of these three variables.


The square of human population density is the dominant variable in this model. The exponential nature of this variable is derived from the dual effect of human population density on fire regimes. One significant effect is that increases in human population density increase ignition potential; the other effect is that higher human population density across the landscape increases the likelihood that wildland fire will damage human life and property. Fuel fragmentation is the only physical factor in this conceptual model of risk that reduces the risk of wildland fire for humans. Fuel fragmentation, however, may be outweighed by the exponential effects of human population and the accumulation of fuels.

SYNTHESIS

Dendrochronological histories allow the quantitative analysis of the role of humans in an ecosystem and show that this role is part of an ecological dynamic controlled by changes in factors that limit the occurrence of fire. Temporal trends in the percent of sites burned and human population density, coupled with historic information on human cultures, allow us to define four stages within an anthropogenic fire regime in the Current River watershed in Missouri.

The stability of any stage in the fire regime is dependent on exogenous human-related factors—for example, war, migration, and introduced disease—as well as changes in endogenous factors, such as fuel accumulation. Some stages may persist and appear to be somewhat stable. In North America, for example, the culture-dependent stage may be a prolonged endpoint under continued fire suppression. The apparent stability of this stage, however, is subject to stochastic phenomena—for example, extreme climate events, which could result in catastrophic or frequent fires. The separation of ignitions and fuels (fire suppression) that occurs during a culture-dependent stage creates an inherently unstable condition.

Fire histories reveal many examples of the influence of exogenous human-related factors on the initiation and termination of anthropogenic fire stages. In Patagonia, the link between humans and fire frequency suggests a sequence of human-fire interactions wherein climatic influences are significant in determining fire frequency on a short-term basis, but the role of humans is significant when examined over decades and centuries (Veblen and others 1999). Bird and Cali (1998) have presented evidence of fire from sediments that demonstrates the influence of humans on fire regimes in sub Saharan Africa over a time scale of hundreds of thousands of years. Charcoal accumulation rates from sediments in the southern Appalachian Mountains also suggest a strong correspondence between human population density (based on the number of archaeological sites) and fire (based on charcoal accumulation) (Delcourt and Delcourt 1997).

The extension of these stages to other ecosystems has limitations. Topographic roughness mitigates the spread of fire and is an integral component of the stages we have identified in our landscape. The effect of human ignitions could be mitigated by topographic roughness or the frequency of water bodies, nonvegetated lands, or other natural fire breaks. Guyette and Cutter (1991) described the fire histories of an oak savanna in Missouri where continuous fine fuels contributed to a consistent mean fire interval, with no population effect evident before Euro-American  settlement. Few ignitions are needed in topographically uniform areas (for example, plains or plateaus) for fires to propagate; therefore, fire frequency tends to be independent of human population density. For example, Abrams (1985) found no indication of distinct stages in the fire history of an oak gallery forest in northeastern Kansas. Sufficient topographic inhibition of the spread of fire is necessary for an ignitiondependent stage in an  anthropogenic fire regime.

We think that the four stages identified by our research could be used as a temporal model to describe and characterize other ecosystems. For example, Amazonia may typify an area in the ignition-dependent stage because population density is closely linked with fire frequency (Laurance 1998). In Australia, many ecosystems may have attributes of a fuel-limited stage (Pyne 1991). Moreover, tropical savanna and brush lands burn frequently, and these fires often result from human ignitions (Andreae 1991). More rigorous analysis of the dynamic interactions among human population density, fuels, and culture could thus enhance our understanding of change and process in ecosystems.

ACKNOWLEDGMENTS

We thank the Missouri Ozark Forest Ecosystem Project, the Missouri Conservation Department, the National Park Service, and the US. Forest Service North Central Station for their support. We are grateful to the Ozark National Scenic Riverways and Charles Putnam for field support. We thank John Krstansky, Tim Nigh, and Mike Stambaugh
for their contributions to the concept of topographic roughness and their expertise in mapping.

REFERENCES

Abrams MD. 1985. Fire history of oak gallery forest in a north eastern Kansas tallgrass prairie. Am Midl Nat 114:188-91.

Agee JK. 1993. Fire ecology of the Pacific Northwest. Washington (DC): Island Press. 493 p.

Alban DH. 1977. Influence on soil properties of prescribed burning under mature red pine. USDA Forest Service ResearchPaper NC-139. North Central Forest Experiment Station, St. Paul, Minn.

Andreae MO. 1991. Biomass burning: its history, use, and distribution and its impact on environmental quality and globalclimate. In: Levin JS, editor. Global biomass burning: atmosphere, climate, and biosphere. Cambridge (MA): MIT Press. p3-21.

Anonymous. 1965. Timber trends in the United States. Forest Resource Report No. 17. Washington (DC): US Government Printing Office. 235 p.

Bailey GA. 1973. Changes in Osage social organization 1673-1906. Anthropological Papers, No. 5. Eugene (OR): University of Oregon.

Baird WD. 1980. The Quapaw Indians: a history of the downstream people. Norman (OK): University of Oklahoma Press. 290 p.

Baldwin JL. 1973. Climates of the United States. Washington (DC): US Department of Commerce, National Oceanic and Atmospheric Administration, Environment Data Service.113 p.

Banks A. 1978. Indians of the upper Current River. Eminence (MO): Alan Banks. 87 p.

Batek MJ, Rebertus AJ, Schroeder WA, Haithcoat TL, Compas E, Guyette RP. 1999. Reconstruction of early nineteenth century vegetation and fire regimes in the Missouri Ozarks. J Biogeogr 26:397-412.

Binkley D, Richter D, David MB, Caldwell B. 1992. Soil chemistry in a loblolly/longleaf pine forest with interval burning.Ecol Appl 2:157-64.

Bird MI, Cali JA. 1998. A million-year record of fire in sub Saharan Africa. Nature 394:767-76.

Cook ER, Meko DM, Stahle DW, Cleaveland MK. 1999. Drought reconstructions for the continental United States. J Clim 12: 1145-62.

Cutter BE, Guyette RP. 1993. Factors affecting species choice for dendrochemistry studies. J Environ Qual 22:611-9.

Cutter BE, Guyette RP. 1994. Fire history of an oak-hickory ridge top in the Missouri Ozarks. Am Midl Nat 132:393-8. Cwynar LC. 1977. The recent fire history of Barron Township, Algonquin Park. Can J Bot 55:1524-8.

DeBano LF, Dunn PH, Conrad CE. 1977. Fire’s effect on physical and chemical properties of chaparral soils. In: Mooney H, Conrad C, Tech. Coord. Proceedings of the symposium on the environmental consequences of fire and fuel management in Mediterranean ecosystems. Forest Service General Technical Report WO-3. Washington, D.C. US Department of Agriculture. p 65-74.

Delcourt HR, Delcourt PA. 1991. Late-quaternary vegetation history of the interior highlands of Missouri, Arkansas, and Oklahoma. In: Henderson D, Hedrick LD, editors. Restoration of old growth forests in the interior highlands of Arkansas and Oklahoma. Proceedings of the Conference Winrock International. Morrilton (AR): Winrock  International Institute for Agricultural Development. 15-30.

Delcourt PA, Delcourt HR. 1987. Long-term forest dynamics of the temperate zone. New York: Springer-Verlag. 439 p.

Delcourt HR, Delcourt PA. 1997. Pre-Columbian Native American use of fire on southern Appalachian landscapes. Conserv Biol 11:1010-014.

Dey DC, Guyette RP. 2000. Anthropogenic fire history and red oak forests in south-central Ontario. For Chron 76:339-47.

Dobyns HF. 1983. Their number become thinned. Knoxville (TN): University of Tennessee Press. 378 p.

Foley WE. 1999. A history of Missouri; vol. 1 Columbia (MO): University of Missouri Press. 249 p.

Gerlach RL. 1986. Settlement patterns in Missouri. Columbia (MO): University of Missouri Press. 88 p.

Gilbert J. 1996. The Trail of Tears across Missouri. Columbia (MO): University of Missouri Press. 122 p.

Gregory GR. 1972. Forest resource economics. New York: Wiley. 548 p.

Grissino-Mayer HD. 1995. Fire and climate reconstructions at El Malpais National Monument, New Mexico [dissertation]. Tucson (AZ): University of Arizona. 407 p.

Grissino-Mayer HD, Holmes RL, Fritts HC. 1996. International Tree-Ring Data Bank Program Library user’s manual. Version 2.0. Tucson (AZ): Laboratory of Tree-Ring Research. 106 p.

Guyette RP. 1996. Tree-ring data, Ontario and Missouri. International Tree-Ring Data Bank. IGBP PAGES/World Data Center-A for Paleoclimatology Data Contribution Series No. 92-014. Boulder (CO): National Oceanic and Atmospheric Admin./National Geophysical Data Center Paleoclimatology Program.

Guyette RP, Cutter BE. 1994. Barium and manganese trends in tree-rings as monitors of sulfate deposition. J Water Air Soil Poll 73:213-23.

Guyette RP, Cutter BE. 1997. Fire history, population, and calcium cycling in the Current River watershed. In: Pallardy S, Cecich R, Garrett H, Johnson P, editors. Proceedings of the 11th Central Hardwood Forest conference. Forest Service General Technical Report NC-188. St. Paul (MN): US Department of Agriculture. p 355-73.

Guyette RP, Cutter BE. 1991. Tree-ring analysis of fire history of a post oak savanna in the Missouri Ozarks. Nat Areas J 11:  93-9.

Guyette RP, Cutter BE, Henderson GS. 1989. Long-term changes in molybdenum and sulfur concentrations in redcedar treerings. J Environ Qual 18:385-9.

Guyette RP, Cutter BE, Henderson GS. 1991. Long-term correlations between mining activity and levels of lead and cadmium in tree-rings of redcedar. J Environ Qual 20:146-50.

Guyette RP, Cutter BE, Henderson GS. 1992. Reconstructing soil pH from manganese concentrations in tree-rings. For Sci 38: 727-37.

Guyette RP, Dey DC. 1995a. A dendrochronological fire history of Opeongo Lookout in Algonquin Park, Ontario. Forest. Research. Report No. 134. Sault Ste. Marie (Ont.): Ontario Forest Research. Institute. 4 p.

Guyette RP, Dey DC. 1997a. A fire history of Huckleberry Hollow. Forest Research. Paper No. 1. Jefferson City (MO): Missouri Department of Conservation. 10 p.

Guyette RP, Dey DC. 1997b. Historic shortleaf pine (Pinus echinata Mill.) abundance and fire frequency in a mixed oak-pine forest. In: Brookshire B, Shifley S, editors. Proceedings of the Missouri Ozark Forest ecosystem project symposium. USDA Forest Service General Technical Report NC-193. St. Paul (MN): US Department of Agriculture. p 136-49.

Guyette RP, Dey DC. 2000. Humans, topography, and wildland fire: the ingredients for long-term patterns in ecosystems. In: Yaussy D, editor. Proceedings: workshop on fire, people, and the Central Hardwoods landscape. USDA Forest Service General Technical Report NE-274. US Department of Agriculture. Newton Square (PA): p 28-35.

Guyette RP, Dey DC. 1995b. A presettlement fire history in an oak-pine forest near Basin Lake, Algonquin Park, Ontario. Forest. Research. Report No. 132. Sault Ste. Marie (Ont.): Ontano Forest Research Institute. 11 p.

Guyette RP, McGinnes EA. 1982. Fire history of an Ozark glade. Trans Missouri Acad Sci 16:85-93.

Heinselman ML. 1981. Fire intensity and frequency as factors in the distribution and structure of northern ecosystems. In: Mooney and others, Tech. Coord. Proceedings on fire regimes and ecosystem properties, Honolulu, HI. 11-15, December 1978. Forest Service General Technical Report WO-26. Washington (DC): US Department of Agriculture. p 5-57.

Howard JH. 1981. Shawnee: the ceremonialism of a native Indian tribe and its cultural background. Athens (OH): Ohio University Press. 454 p.

Hughes LG, Nigh TA. 2000. Historic vegetation of Missouri,lower Ozarks Pilot Area. Jefferson City (MO): Missouri Department of Conservation.

Jacobson RB, Primm AT. 1997. Historical land-use changes and potential effects on stream disturbance in the Ozark Plateaus, Missouri. Water-Supply Paper 2484. Denver (CO): US Geological Survey. 85 p.

Jenkins SE, Guyette RP, Rebertus AJ. 1997. Vegetation diversity and fire history on Turkey Mountain. In: Pallardy S, Cecich R, Garrett H, Johnson P, editors. Proceedings of the 11th Central Hardwood Forest conference. Forest Service General Technical Report NC-188. St. Paul (MN): US Department of Agriculture. p 184-201.

Kauffman JB, Uhl C. 1990. Interactions of anthropogenic activities, fire, and rain forests in the Amazon basin. Ecol Stud: Anal and Synth 84:117-34.

Kroeber AL. 1934. Native American population. Am Anthropol 36:1-25.

Ladd D. 1991. Reexamination of the role of fire in Missouri oak woodland. In: Burgur G, Ebinger J, Wilhem G, editors. Proceedings of the oak woods management workshop. Charleston (IL): Eastern Illiniois University p 67-80.

Laurance WF. 1998. A crisis in the making: responses of Amazonian forests to land use and climate change. Trends Ecol Evol 13:411-15.

Lynott MJ. 1989. An archeological evaluation of the Gooseneck and Owls Bend sites: Ozark National Scenic Riverways, southeast Missouri. Conducted for the US Department of the Interior, National Park Service, Midwest Archeological Center. Lincoln (NE) Center of Archaeological Research, Southwest Missouri State University

Marriott A. 1974. Osage Indians II. New York: Garland. 270 p. O’Brien MJ, Wood WR. 1998. The prehistory of Missouri. Columbia (MO): University of Missouri Press. 417 p.

Paullin CO. 1932. Atlas of the historical geography of the United States. Carnegie Institute of Washington and the American Geographical Society of New York. Washington (DC) 166 p.

Pitcaithley DT. 1978. Let the river be: a history of the Ozark’s  Buffalo River. Santa Fe (NM): Southwest Cultural Resources Center, National Park Service. 133 p.

Porter SR. 1998. Modeling historic woody vegetation in the lower Ozarks of Missouri [thesis]. Columbia (MO): University of Missouri. 134 p.

Price JE, Price CR, Saucier R. 1987. Archeological investigations in the Ozark National Scenic Riverways, 1984-1986. Conducted for the US Department of the Interior, National Park Service, Midwest Archeological Center. Lincoln (NE): Center of Archaeological Research, Southwest Missouri State University, Springfield (MO) 169 p.

Pyne SJ. 1991. Burning bush. New York: Holt. 520 p.

Pyne SJ. 1982. Fire in America. Princeton (NJ): Princeton University Press. 654 p.

Pyne SJ. 1995. World fire: the culture of fire on earth. New York: Holt. 379 p.

Pyne SJ, Andrews PL, Laven RD. 1996. Introduction to wildland fire. New York: Wiley. 769 p.

Rafferty MD. 1982. Historical atlas of Missouri. Norman (OK): University of Oklahoma Press. 113 p.

Ramenofsky AF. 1987. Vectors of death: the archaeology of European contact. Albuquerque (NM): University of New Mexico Press. 300 p.

Record SJ. 1910. Forest conditions of the Ozark region of Missouri. Bulletin No. 89. Columbia (MO): Missouri Agriculture Experiment Station.

Schroeder MJ, Buck CC. 1970. Fire weather. Agricultural Handbook 360. Washington (DC): US Department of Agriculture, Forest Service. 229 p.

Scowcroft PG. 1965. The effects of fire on the hardwood forest of the Missouri Ozarks [thesis]. Columbia (MO): University of Missouri. 126 p.

Shumway DL, Abrams MC, Ruffner CM. 2001. A 400-year history of fire and oak recruitment in an old-growth oak forest in western Maryland, U.S.A. Can J For Res 31:1437-43.

Stevens DL. 1991. A homeland and a hinterland: the Current and Jacks Fork riverways. Historic Resource Study, Ozark National Riverways. Van Buren (MO): National Park Service. 248 p.

Stokes MA, Smiley TL. 1968. Introduction to tree-ring dating. Chicago (IL): University of Chicago Press. 78 p.

Swetnam TW. 1993. Fire history and climate change in GiantSequoia groves. Science 262:886-9.

Thornton R. 1987. American Indian holocaust and survival.Norman (OK): University of Oklahoma Press. 292 p.

Turner MG, Romme WH. 1994. Landscape dynamics in crown fire ecosystems. Landscape Ecol 9:59-77.

Veblen TT, Kitzberger T, Villalba R, Donnegan J. 1999. Fire history in northern Patagonia: the roles of humans and climatic variation. Ecol Monogr 69:47-67.

Waldman C. 1985. Atlas of the North American Indian. New York: Facts On File. 276 p.

Weslager CA. 1978. The Delaware Indian westward migration. Wallingford (PA): Middle Atlantic Press. 266 p.

Westin S. 1992. Wildfire in Missouri. Jefferson City (MO): Missouri Department of Conservation. 161 p.

Whelan RJ. 1995. The ecology of fire. New York: Cambridge University Press. 346 p.

Wiegers RP. 1985. Osage culture change inferred from contact and trade with the Caddo and the Pawnee [dissertation]. Columbia (MO): University of Missouri. 229 p.

Williams M. 1989. Americans and their forests: a historical geography. New York: Cambridge University Press. 599 p.

Wolferman KC. 1997. The Osage in Missouri. Columbia (MO): University of Missouri Press. 119 p.

Zinke PJ. 1977. Mineral cycling in fire-type ecosystems. In: Proceedings of the symposium on the environmental Consequences of fire and fuel management in Mediterranean ecosystems. Forest Service General Technical Report WO-3. Washington (DC) US Department of Agriculture. 498 p.

Appendix. The Fires of 1780

Dendrochronological evidence suggests the occurrence of extensive, and possibly intensive, fires in at least three areas of eastern North America during 1780, a year of drought and increasing human turmoil. In the Algonquin Highlands of southern Ontario, sites over an area of 2000 km2 show evidence of fire in 1780 (Cwynar 1977; Dey and Guyette 2000a). At two of these Algonquin sites, 66% and  43% of the trees were scarred, indicating that surface fires must have been intense (Guyette and Dey 1995a, 1995b). In Missouri, 43% of the study sites in the Current River watershed were burned and 28% of the sample trees were scarred along the North Fork of the White River; evidence of fire in the same year was also found near the Gasconade River (Cutter and Guyette 1994). Along the breaks of the Arkansas River in the Boston Mountains, trees at three sites separated by 15 km had scars formed in 1780 (R. P. Guyette and M. Spetich unpublished). In addition, fires occurred circa 1780 in Minnesota (Heinselman 1981) and Maryland (Shumway and others 2001). The extent and severity of these fires was probably the result of a drought in 1780 (Cook and others 1999) coupled with concurrent human activities. Human conflict frequently results in wildland fire. Eastern Native American tribes were forced west and north into areas not yet populated by Euro-Americans circa 1780 and were often in conflict with the indigenous tribes they encountered such as the Osage in the Ozarks (Stevens 1991; Wolferman 1997). Spain declared war on England in 1779. By May 1780, the English, Menominee, and Winnebago were attacking the Spanish colony of St. Louis, Missouri (Foley 1999). There was a considerable struggle among the Americans, French, Osage, Cherokee, Spanish, and English for the control of trade in Missouri. Thus, a combination of human activities and drought probably resulted in one of the worst fire years of the 1700s.