Influences of frequent cool-season burning across a soil moisture gradient on oak community structure in longleaf pine ecosystems.
|Abstract:||Fire and soil moisture gradients are thought to influence oak community structure in longleaf pine (Pinus palustris Mill.) woodlands. To explore these influences, we measured oak density and species composition within a longleaf pine-wiregrass (Aristida stricta Michaux) dominated landscape subjected to frequent cool-season fires for 70 yr. At 64 locations within the Joseph W. Jones Ecological Research Center in southern Georgia, oak trees, saplings and regeneration (seedlings plus sprouts) were counted in nested plots, and 20 additional measurements were taken to assess physical and chemical properties of soils, disturbance, density, of other vegetation and topography. Principal components analysis of soil data revealed a soil moisture and soil chemistry gradient. Trees and saplings were sparse (mean of 37 and 81 per ha) and most common in dry and dry-mesic sites, while regeneration was abundant (mean of 110,100 stems per ha) and well-distributed across the soil moisture gradient. Fifty percent or more (depending on vegetation stratum analyzed) of cumulative species-environment relation was accounted for in the first two axes of a canonical correspondence analysis with axes representing gradients in soil moisture and several chemical and physical properties including mineralizable N, extractable Ca and Mg and soil texture. Three oak species were concentrated in the dry end of the soil moisture gradient, four in the moist end and one was common across much of the gradient. Species distributions probably reflect physiological tolerances of soil moisture extremes plus variation in fire regime caused by differences in soil moisture. Forest fragmentation and prescribed cool season burning may have increased oak densities in this landscape. If so, then this legacy of past management should be considered if managers wish to change fire regimes to better mimic natural disturbance patterns.|
Longleaf pine (Environmental aspects)
Forest fires (Environmental aspects)
Plants, Effect of soil moisture on (Environmental aspects)
Forest ecology (Research)
Jacqmain, Elizabeth I.
Jones, Robert H.
Mitchell, Robert J.
|Publication:||Name: The American Midland Naturalist Publisher: University of Notre Dame, Department of Biological Sciences Audience: Academic Format: Magazine/Journal Subject: Biological sciences; Earth sciences Copyright: COPYRIGHT 1999 University of Notre Dame, Department of Biological Sciences ISSN: 0003-0031|
|Issue:||Date: Jan, 1999 Source Volume: 141 Source Issue: 1|
Oaks (Quercus) are an important component of fire-dependent longleaf pine (Pinus palustris Mill.) ecosystems that once dominated the Coastal Plain of the southeastern United States. However, the structure and dynamics of oak populations within these ecosystems, and the environmental factors that control oak distribution are only known superficially (Christensen, 1988; Ware et al., 1993). A better knowledge of oak ecology could help in current efforts to conserve and restore longleaf pine ecosystems. Previous reports have emphasized the importance of soil moisture, fire regimes and their interaction for determining patterns of oak distribution in longleaf pine woodlands (Christensen, 1988; Myers, 1990; Peet and Allard, 1993; Stout and Marion, 1993; Ware et al., 1993). Three general oak habitats that reflect different soil moisture regimes have been identified: dry sandhills, intermediate sites and low wetlands (Marks and Harcombe, 1981; Christensen, 1988; Peet and Allard, 1993; Stout and Marion, 1993; Ware et al., 1993). Within each of these habitats, fire is believed to control oak density and species composition directly through differential mortality and indirectly through effects on soil resources (Ware et al., 1993).
Beyond the general agreement that soil moisture and fire control oak distributions, ecologists have not yet uncovered the details of how and why oaks occupy various landscape positions within longleaf pine woodlands. Actual measurements of key environmental conditions are scarce (Christensen, 1988; Ware et at., 1993), and effects of long-term burning regimes are difficult to predict because most fire studies have: (1) examined short term effects only (Rebertus et al., 1989; Olson and Platt, 1995); (2) included temporally variable fire regimes (Boyer, 1990); (3) examined extreme habitats only (e.g., dry sandhills or moist flatwoods; Glitzenstein et al., 1995); or (4) investigated only a single habitat type (Myers, 1985; Waldrop et al., 1992). Even data on oak community structure are uncommon, especially for mesic landscape positions (Jones et al., 1984; Glitzenstein et al., 1995).
We measured oak species composition and size class distribution within a landscape dominated by longleaf pine ecosystems and subjected to more than 70 yr of dormant season burning prescribed at 2-4 yr intervals. Our objectives were to: (1) document oak community structure across a soil moisture gradient from dry to moist; and (2) identify environmental gradients important in explaining oak community structure. Our underlying goal was to facilitate development of management strategies for restoration and maintenance of longleaf pine ecosystems.
The climate of the region is humid-temperate characterized by long hot summers and cool winters. The coldest month, January, has a mean temperature of 9 C and the warmest month, July, has a mean temperature of 28 C. The growing season extends from March to November. Annual precipitation ranges from 81 to 182 cm with the greatest proportion occurring between April and September (Lynch et al., 1986).
A unique feature of the study area is that it has not been farmed extensively due to its relatively poor soils (Lynch et al., 1986). The Jones Center was formerly a hunting plantation and has been managed for over 70 yr to optimize habitat for bobwhite quail (Colinus virginianus L.) and other game species. Management has included burning every 2-4 yr, usually from late February through April. Approximately 6880 ha (about 60%) of the property is upland longleaf pine forest that contains relatively undisturbed ground cover vegetation representing a broad range of soil types and topographic positions (Lynch et al., 1986).
Plot selection. - Aerial photographs (1:20,000) and soil maps (USDA, 1986) were used to stratify the study area into longleaf pine-wiregrass (Aristida stricta Michaux) units selected to represent a broad soil moisture gradient. Fourteen areas, located from 0.29 to 15 km apart, with little or no evidence of agricultural activity, ranging in landform from dry sand ridges (sandhills) to moist depressional pond margins were chosen. Soil series included Bigbee, Kershaw, Lakeland, Norfolk, Orangeburg, Pelham, Troupe and Wagram (USDA, 1986). A systematic grid of potential sample points (starting at a random point) was located at each study unit. From a total pool of 114 potential points, 64 were randomly selected for study. Points were discarded (and new ones randomly chosen) if they had evidence of soil disturbances, such as old roads and old fields, that might have substantially altered vegetation or burning patterns. All sites, regardless of landscape position, had evidence of fires occurring within the previous 4 yr including charcoal and burn scars on trees. The distribution of points across soil moisture classes was not uniform because undisturbed dry and wet sites (sand ridges and pond margins) are rarer than undisturbed mesic sites at the Jones Center (Lynch et al., 1986).
Oak sampling. - At each sample point, we recorded oak species and density for three strata: trees, saplings and seedlings/sprouts (hereafter referred to as regeneration). Trees were defined as stems [greater than or equal to] 9.5 cm at dbh (diameter at 1.4 m), saplings as stems 2.5 cm [less than or equal to] dbh [less than or equal to] 9.4 cm; and regeneration as dbh [less than] 2.5 cm. All stems within the regeneration size class were counted as individuals, even if they originated from the same rootstock. However, notes were taken on the number of stems per rootstock to better interpret the potential for future production of saplings and trees. Trees were counted in rectangular 20 by 60 m (1200 [m.sup.2]) plots. Sapling density was recorded in rectangular 20 by 40 m plots (800 [m.sup.2]) nested within the center of the tree quadrat. Regeneration was counted in square 20 by 20 m (400 [m.sup.2]) plots nested within the sapling plots. Relatively large plots were used to capture the infrequent and widely scattered trees anti saplings and sometimes scattered patches of regeneration. Taxonomic nomenclature follows Godfrey (1988).
Soil measurements and analysis. - Soil moisture was measured by time domain reflectometry in May and June, 1993 (Topp et al., 1980). Moisture was measured at 0-30 cm and 0-90 cm depths at two of the plot corners (four per plot) and averaged by depth and date for each plot. Depth to soil mottling and to any layer with restricted permeability (clay or rock) was estimated using a bucket auger, 3.5 m long, near the center of the plot. If no layer of restricted permeability was encountered depth was recorded as 3.5 m.
For chemical analysis, soil from 15 cm deep (below litter layer) was collected from 15 locations systematically located within the plots. The samples were composited and passed through a 2 mm sieve. Soil pH was measured in a 1:2 soil:deionized [H.sub.2]O solution. Potential N mineralization was estimated as the increase in [[NH.sub.4].sup.+] and [N[O.sub.3].sup.-] in a 1:1 soil:2M KCl solution (measured by Lachat Autoanalyzer) after a thirty day incubation at constant temperature and moisture (Wood et al., 1992). Available phosphorus (P[O.sub.4]) was estimated from replicate 5 g samples using the dilute acid-fluoride (AF) extraction technique described by Olsen and Sommers (1982; in Walbridge, 1991). Samples analyzed for microbial P were treated with 0.5 mL liquid CH[Cl.sub.3] and incubated for 18 h before AF-extraction. Microbial P was estimated as the net increase in AF-extractable P[O.sub.4]. Soil cations (Ca, Mg, K) were measured by double-acid extraction (Isaac and Kerben, 1971) and a Perkin Elmer Atomic Absorption Spectrophotometer.
To measure horizon thickness and soil texture, five additional soil samples (5 cm diam by 1 m deep) were collected from the plots using a customized steel pipe inserted with a plastic tube that was driven into the ground, after which the tube was removed, capped and returned to the laboratory. For each core, horizons were delineated and their thicknesses measured. The five samples were composited into one bulk sample per horizon per plot. Soil texture analysis followed standard wet and dry-sieving techniques (Day, 1965). Soil that passed through sieve sizes of [greater than or equal to] 2 mm, [greater than or equal to] 0.500, [greater than or equal to] 0.250, [greater than or equal to] 0.045 and [less than] 0.045 were classified as coarse fragments, coarse medium and fine sands, and silt plus clay, respectively. Particle size fractions were calculated as a profile weighted average percentage based on mean horizon thicknesses from the five soil cores (Day, 1965). Soil variables are summarized in Table 1.
[TABULAR DATA FOR TABLE 1 OMITTED]
Non-soil variables. - Each plot was assigned to a landform class adapted from Jones et al. (1984; see also Palik and Pederson, 1996; Goebel et al., 1997). Classes were sand ridges (xeric sites with deep sands and slightly rolling topography), slopes (areas with 1-4% slopes and therefore some lateral flow of soil moisture), high flat (flat sites with soils that are well drained year-round), low flat (flat sites with waterlogged soil during some winter and spring months) and pond margin (transition areas between uplands and wetlands). Average overstory pine basal area was estimated from three variable-plot prism samples (basal area factor of 10, English units; Grosenbaugh, 1952) taken in the center and two corners of each regeneration measurement (400 [m.sup.2]) plot. Pine basal area is negatively correlated with canopy openness and positively correlated with pine litter production which may influence fire behavior (Ware et al., 1993). Percent wiregrass cover, which can influence fire behavior (Ware et al., 1993; Stout and Marion, 1993), was estimated using a scale adapted from Peet and Allard (1993) as follows: 1 = trace, 2 = [less than] 1%, 3 = 1-2%, 4 = 2-5%, 5 = 5-10%, 6 = 10-25%, 7 = 25-50%, 8 = 50-75%, 9 = 75-95% and 10 = [greater than] 95%. Cover estimates were made using 1 X 2 m subplots, located at the center and each corner of the regeneration plot. Distance to the nearest firebreak, which included roads, plow lines, food plots and water bodies, was measured from the plot center. The average number of gopher tortoise (Gopherus polyphemus Daudin) mounds (an indication of soil disturbance and grazing) was estimated from the same subplots used to measure wiregrass cover. Each gopher tortoise plot was 10 m in diameter and included active, inactive and abandoned burrows.
Statistical analyses. - Physical and chemical soil variables (Table 1) were analyzed by principal components analysis separately ([PCA.sub.physical] and [PCA.sub.chemical]) and in combination ([PCA.sub.soil]) to assess the strengths of soil gradients independently of species distributions across the 64 plots. PCAs were performed using correlation matrices (Palmer, 1993). Soil chemical data were log-transformed (except pH) and soil texture data were arc-sine transformed before analysis (Palmer, 1993).
Trends in vegetation and environmental data were analyzed by canonical correspondence analysis (CCA) without downweighting of rare species. Ordination axes were constrained to be linear combinations of environmental variables (ter Braak, 1986). Landform classes were coded as separate dummy variables (following ter Braak, 1990). We analyzed the vegetation data by strata (trees, saplings and regeneration) and as a combined data set. In the analysis of separate strata, data were not transformed and plots not containing any species in the strata were not included. For the analysis of the combined data set, we used "pseudospecies" (sensu Jones et al., 1984). For example, Quercus laevis trees, saplings and regeneration would each represent an individual species variable. This technique allows for interpretation of structural, as well as compositional variations among plots (Jones et al., 1984). Species densities were log transformed in the combined CCA to prevent the high density of the regeneration layer from overwhelming the ordination.
These analyses were exploratory and intended to develop hypotheses concerning factors controlling oak distributions. Consequently, we did not employ the forward selection option (a step-wise procedure) available in the CCA. Forward selection may result in the omission of environmental variables that, although perhaps not statistically significant, may still be ecologically important (James and McCulloch, 1990). CCA was used instead of detrended CCA based on arguments that the former captures the structure of species data better than latter, which is subject to distortion (Palmer, 1993). We used CANOCO version 3.1 for all CCA (ter Braak, 1990).
Soil gradient. - The combined soil data PCA ([PCA.sub.soil]) accounted for 30.3% of the variation in the first axis and 17.9% for the second. The two axes were broadly interpretable as soil moisture (Axis 1) and soil chemistry (Axis 2). Pond margins and low flats occurred on the right (moist) side of the diagram, sand ridges occurred on the left and slopes and high flats were in the middle [ILLUSTRATION FOR FIGURE 1 OMITTED]. Two factors particularly important in defining Axis 1 (long arrows parallel to Axis 1 in [ILLUSTRATION FOR FIGURE 1 OMITTED]) were TDR-determined soil moisture and depth to soil mottling. TDR soil moisture is directly correlated with water availability for plants, and mottling is indirectly so because mottles form where water is abundant. Other physical factors correlated to Axis 1 included percent coarse fragments, medium sand, depth to a restrictive horizon, fine sand and silt plus clay [ILLUSTRATION FOR FIGURE 1 OMITTED]. Of the chemical factors measured, three were correlated with Axis 1; potential mineralizable N and available P increased toward the dry end of the soil moisture gradient, and microbial P increased toward the moist end.
The chemically driven second axis explained 18% of the soil data variation among plots. The primary variables related to Axis 2 were calcium, magnesium and percent coarse sand; high fiats had particularly large values for these variables [ILLUSTRATION FOR FIGURE 1 OMITTED]. Soil K increased in the same direction as Ca and Mg but was less important (shorter arrow in [ILLUSTRATION FOR FIGURE 1 OMITTED]) in defining the second PCA axis.
Separate analyses of soil physical properties ([PCA.sub.physical]) and chemistry ([PCA.sub.chemical]) each accounted for more variation among plots in the first two ordination axes than did [PCA.sub.soil] (63.7% for physical and 57.2% for chemical versus 48.2% for physical + chemical). [PCA.sub.physical] had the clearest one-dimensional environmental gradient as it accounted for the greatest amount of cumulative variance in the first axis (42.0% versus 32.2 and 30.3% for [PCA.sub.chemical] and [PCA.sub.soil]). The largest factor loadings for Axis 1 scores in [PCA.sub.physical] were soil moisture (0.83) and depth to mottling (-0.83); absolute values of all other loadings were [less than or equal to] 0.68.
Stand structure across the soil gradient. - To see how forest structure relates to the soil moisture gradient, we plotted density of oak trees, saplings and regeneration for each stand, with stands arrayed in order of their occurrence along the most readily visible moisture gradient; i.e., Axis 1 of [PCA.sub.physical]) [ILLUSTRATION FOR FIGURE 2 OMITTED]. Trees occurred throughout much of the moisture gradient with a concentration toward the dry end. Saplings followed the same pattern with an even greater affinity for the dry end of the gradient. However only 48% of the sample plots had oak trees, saplings or both. In contrast, oak regeneration was well distributed across the entire soil moisture gradient and occurred in nearly 100% of the plots [ILLUSTRATION FOR FIGURE 2 OMITTED]. Regeneration was far more abundant than trees or saplings; the overall mean numbers of stems per ha were 37 trees, 81 saplings and 110,100 regeneration.
Distribution of oak species. - Eight oak species were found within the study plots including: Quercus falcata Michx., Q. geminata Small, Q. hemisphaerica Bartr., Q. incana Bartr., Q. laevis Walt., Q. margaretta Ashe, Q. nigra L. and Q. virginiana Mill. Additional species observed in study units but not in the study plots were: Quercus arkansana Sarg., Q. laurifolia Michx., Q. minima (Sarg.) Small, Q. phellos L. and Q. stellata Wang. The tree stratum was dominated by Quercus laevis and Q. margaretta which together comprised 76% of the 287 trees sampled in all plots combined. Quercus incana was 11.7% of the tree-sized stems sampled. Saplings were dominated by the same three species (Q. laevis 14.0%, Q. margaretta 38.8% and Q. incana 43.6% of 414 saplings counted). The dominant species in the regeneration stratum were Quercus margaretta and Q. incana (respectively 43.7, and 43.1% of 281,853 individual stems counted).
The CCA of the species-strata data with regeneration, saplings and trees counted as pseudospecies, showed the same soil moisture and soil chemical gradients revealed during PCA of soil data (i.e., [PCA.sub.soil]). Axis 1 was primarily a soil moisture gradient [ILLUSTRATION FOR FIGURE 3 OMITTED] and accounted for 21.2% of the species inertia (Table 2). Inertia is a measure of how well the CCA model captured trends in community structure and an indication of the variability within the data (ter Braak, 1990). Axis 2 was correlated with both soil moisture and soil chemical variables and accounted for an additional 7.8% of species inertia. Toward the wet end of the gradient (upper left of [ILLUSTRATION FOR FIGURE 3 OMITTED]) soil moisture, coarse fragments and basal area of pine increased and topographic landforms were mostly low flats and pond margins. Toward the dry end (right side of diagram), potential mineralizable N, depth to soil mottling and to a restrictive layer, percent fine sand and numbers of gopher tortoise mounds increased; predominant landforms there were slopes and sand ridges. Values for Mg, Ca and percent coarse sand increased toward the lower left of the diagram where high flats dominated. The peak abundance of Quercus virginiana trees and regeneration (no saplings found of this species) was at the moist end of the gradient; Q. geminata regeneration and saplings (no tree-sized stems found) peaked at the dry extreme. Greatest densities of Quercus falcata occurred in the lower left area where high flats were prevalent. The remaining species were concentrated toward the middle of the ordination diagram. For some species, most notably Quercus nigra, peak abundance of trees, saplings and regeneration occurred at widely different positions along the gradients [ILLUSTRATION FOR FIGURE 3 OMITTED].
When oak species distributions were analyzed separately by tree, sapling and regeneration strata, the CCA model fit the data better. Percent of species inertia and cumulative percent of the species-environment relationship increased when the CCAs were performed for individual strata (Table 2). In each of the CCAs for individual strata, the first ordination axis could be interpreted as a moisture gradient (data not shown).
For a clearer picture of how individual species were distributed across the moisture gradient, we calculated average density of each strata for species in each landform. Relative densities were computed (mean divided by maximum) and plotted in order of wettest to driest landform. For regeneration no species had peak density in pond margins. Quercus falcata, Q. hemisphaerica, Q. nigra and Q. virginiana were most abundant in low flats, the second wettest landform after pond margins [ILLUSTRATION FOR FIGURE 4 OMITTED]. Three species Q. geminata, Q. laevis, and Q. margaretta, had peak densities on sand ridges, the driest sites. Quercus incana was distributed relatively evenly across land forms, except for its virtual absence from pond margins. Plots of the tree and sapling layers (not shown) revealed similar patterns.
Oak size structure. - Oak saplings and trees were concentrated toward the dry end of a soil moisture gradient. This is a common feature for frequently burned longleaf pine woodlands in the southeastern United States. Stand structure within the intermediate to moist end of our gradient resembles the pine savanna and subxeric longleaf pine woodland community types described by others (Harcombe et al., 1993; Peet and Allard, 1993). Savannas and subxeric woodlands occur on mesic to moist sites, anti are maintained by frequent fires at regular intervals (Miyanishi and Kellman, 1986; Waldrop et al., 1992; Harcombe et al., 1993). The structurally diverse forests at the dry end of our gradient fit descriptions of sandhill vegetation (Christensen, 1988; Rebertus et al., 1989; Harcombe et al., 1993). If just saplings and tree-sized stems are considered, our data fit Ware et al.'s model (1993) for frequently burned landscapes (see level A of their [ILLUSTRATION FOR FIGURE 5 OMITTED]).
Several mechanisms may account for the unequal distribution of sizes across the moisture gradient. First, the two most abundant oaks on dry sites (Quercus laevis, and Q. margaretta) grow very slowly (especially in dry sites) and become mature at relatively small sizes. Thus, if sustained oak height growth occurred across the soil moisture gradient (e.g., during a long fire-free period), stems may remain in the sapling size class longer and experience less self thinning in the dry end of the gradient than in the moist end. This argument is weakened, however, by the fact that two relatively slow-growing species of small-stature (Quercus margaretta and Q. incana) are abundant in the regeneration layer across all of the soil moisture gradient except pond margins. A second, and perhaps complementary mechanism that explains our results is variability in fire regimes across the soil moisture gradient. Spatial and temporal variability of fire may be greatest in dry sites due to patchily distributed fuels (Glitzenstein et al., 1995). With greater variation in fire frequency, longer fire-free periods are possible, which may allow seedlings or sprouts enough time to grow into a fire-resistant size (Rebertus et al., 1993). Some experimental evidence supports this model; Rebertus et al. (1993) found that Quercus laevis saplings were more abundant when fire frequency was irregular than when it was regular during a 15 yr burning study in Florida. Fire may be patchy in time and space at the wettest end of our soil moisture gradient (pond margins) due to presence of patchy surface water during the burning season in some years. The scarcity of saplings and trees in pond margins may reflect competition from large oak and pine trees, or intolerance of flooding in some of the oak species.
Detailed long-term fire histories are lacking for our research plots, and thus we are unable to rigorously test hypotheses concerning fire effects on oak community structure. However, data on ecosystem structure across the gradient support the idea that fires have been more variable in dry sites. Overstory pine basal area increased along the moisture gradient (from 4.7 [m.sup.2]/ha in sand ridges to 8.6 [m.sup.2]/ha in pond margins with a peak of 14.8 [m.sup.2]/ha in low flats; see also [ILLUSTRATION FOR FIGURE 3 OMITTED]). Very likely, the increased basal area was associated with increased pine leaf litter production. Longleaf pine needles are extremely flammable and therefore capable of increasing fire-induced top-kill of oaks (Williams and Black, 1981; Rebertus, 1988; Rebertus et al., 1989, 1993). In addition, many plants in the longleaf pine-wiregrass ecosystem produce highly flammable litter, and are likely to increase in abundance along productivity/moisture gradients (Platt et al., 1989). Our assessment of wiregrass abundance, however, shows no increase from dry to moist sites (mean cover class of 6.6 in sand ridges and 5.9 in pond margins). Additional studies on fuels and fire behavior are needed to confirm our hypothesis that fires vary in time and space across the soil moisture gradient.
Oak regeneration was surprisingly abundant. Mean density was 110,100 per ha which is an order of magnitude greater than other coastal plain forests periodically burned for 15 or more yr (Boyer, 1993; Waldrop et al., 1992; Rebertus et al., 1993). Although our densities are somewhat misleading because all stems were counted even if they were from the same rootstock, our field notes indicated that rootstocks were also abundant (at least 35,000 per ha and perhaps greater). Such densities could be typical for frequently burned longleaf pine ecosystems in the region of our study, We suspect, however, that densities have increased during the past seven decades as has been demonstrated in other long-term burning studies. Waldrop et al. (1992) found that hardwood stem densities (all stems regardless of rootstock numbers) increased from 16,000 to 47,000 per ha over a 30 yr period of periodic burning in Pinus taeda forests in South Carolina. Boyer (1993) found increases in stem density (over 69% oaks) from 13,250 to 33,000 per ha during 18 yr of cool season burning of sandy uplands in southwest Alabama.
If oak densities have indeed increased over time, there are several plausible reasons why. First, forest fragmentation may have protected some microsites from burning, allowing oaks to mature and increase seed inputs into the more frequently burned portions of the landscape. We have observed many stringers of mature oaks at the Jones Center; most are associated with road sides, edges of agricultural fields and edges of wildlife food plots. We also observed abundant large Quercus incana at one slope site where burning was apparently patchy. Protection from fire not only allows oaks to mature, it also stimulates seed production. For example, mast production by Q. laevis and Q. margaretta in the Wade Tract, an old-growth longleaf pine woodland in southern Georgia, occurs only in areas not burned in the last eight years (Rebertus et al., 1993). To test the idea that landscape fragmentation is increasing oak populations, studies are needed to compare acorn production and dispersion for oaks in fire-protected versus unprotected settings. Furthermore, the ability of oak seedlings to become established in the frequently burned landscapes at the Jones Center must be demonstrated.
In addition to fragmentation, the regular timing of fires and their low intensity may have facilitated increases in oak regeneration. The 2-4 yr gaps between fires can allow seedlings to become established. Once established, oak rootstocks are rarely killed by fire, although they may be top-killed (Streng et al., 1993; Olson and Platt, 1995). Furthermore, since fires at the Jones Center are set only in the cool season when fuel conditions are somewhat moist (i.e., to prevent spread of fires to unwanted areas), some soil patches may not burn, and those that do may carry a relatively low intensity flame. It is possible that if the Jones Center had used more frequent warm season burning, the regeneration layer would be less dense than we observed. Waldrop et al. (1992) found that annual warm season burning reduced woody plant densities in South Carolina, and Olson and Platt (1995) found that warm season burning blocked recruitment of new woody plant stems. However, season of fire effects on oak densities are not fully understood (Streng et al., 1993), and annual warm season burns are not possible at the Jones Center because more than 1 yr of fuel buildup is needed before a fire can be carried across a site.
Environmental gradient and oak species distributions. - As expected, soil moisture was a key factor affecting the distribution of oak species within our study area. From the moist to dry ends of the soil moisture gradient identified by CCA, species composition changed from plots containing Quercus virginiana, Q. nigra, and Q. falcata to plots dominated by Q. laevis and Q. margaretta, with some Q. geminata and Q. hemisphaerica [ILLUSTRATION FOR FIGURE 3 OMITTED]. When we plotted species densities for each landform, we found the same trends except that Q. hemisphaerica was clearly a moist site species, and Q. incana had a broad ecological amplitude [ILLUSTRATION FOR FIGURE 4 OMITTED].
Community structure was more closely related to soil moisture than to soil chemical properties (i.e., compare lengths of arrows in Fig. 3). The importance of water relative to nutrients for determining oak distribution is clear in the literature. For example, a broad array of physiological and morphological traits to deal with water deficits or flooding have been identified. Quercus laevis and Q. margaretta maintain high water conductance under deficit conditions (Mavity, 1992), Q. margaretta can limit water loss through stomatal conductance (Long, 1993), Q. geminata has highly sclerophyllous leaves and Q. laevis has relatively high levels of net photosynthetic gain under deficit conditions (Vaitkus and McLeod, 1995). Oaks adapted to dry soils have larger rout systems (greater root/shoot ratios) than mesic or wetland oaks, presumably for accessing and storing water (Long and Jones, 1996). Oaks from mesic and hydric environments have smaller root/shoot ratios and more rapid stem height growth after germination (Long and Jones, 1996). Although none of the species in our study are known for tolerating extensive flooding (Hook, 1984), seasonal inundation may have favored Q. virginiana and Q. nigra on low flats and pond margins.
A soil chemical gradient, although secondary in importance to soil moisture, was also detected. Mineralizable N, and exchangeable Ca and Mg in particular were correlated with the distribution of oak species [ILLUSTRATION FOR FIGURE 3 OMITTED]. Quercus falcata had a positive correlation with Ca, Mg, and perhaps with K. Quercus virginiana was positively correlated with Ca and Mg, particularly on coarser soils. These two species were concentrated in the moist end of the soil moisture gradient (Figs. 3, 4) which suggests that nutrients have stronger impacts on oak distribution in moist than in dry sites.
The increase in mineralizable N toward the dry end of the soil moisture gradient was surprising. While only potential mineralizable N was quantified in this study, the pattern of N increase in dry sites has been confirmed with in situ incubations in separate investigations of the same study landscape. Based on predictions by Christensen (1981) and Ware et at (1993) concerning the effects of fire on soil nutrients, we expected N to increase, as other nutrients did, with increasing soil moisture across the environmental gradient.
Perhaps oaks are not just responding to N availability among sites; they may in fact be a major regulator of N flux. Large oak stems were more common in dry sites [ILLUSTRATION FOR FIGURE 2 OMITTED] where mineralizable N was greatest [ILLUSTRATION FOR FIGURE 3 OMITTED]. In ponderosa pine (Pinus ponderosa Laws) forests, Klemmedson (1987, 1991) found that N mineralization was linearly related to the basal area of gambel oak (Quercus gambelii Nutt.), a deciduous species ecologically and genetically similar to Q. margaretta (Muller, 1952). Wood et al. (1992) found less N mineralized in pure stands of loblolly pine than in the more complex communities containing herbaceous and deciduous hardwoods. The conventional ecological perspective has been that species respond to variation in nutrient availability across sites. However, a greater appreciation is being built for the function of specific species, or groups, in regulating N availability, rather than simply responding to site differences (Wedin and Tilman, 1990).
Although the environmental gradients we examined were correlated with definable patterns in species distributions, only 39.3% of the potentially explainable community structure (as measured by species inertia) was captured by the first four CCA ordination axes when all strata were included in analysis (Table 2). When each strata was examined separately, however, more of the variation in the species data was explained. Sapling species distributions had the greatest amount of community structure revealed in the first four axes (96%), followed by trees (78%) and regeneration (67%). Since each strata responds to the soil moisture gradient differently [ILLUSTRATION FOR FIGURE 2 OMITTED], CCA may have a more difficult time capturing trends when all are combined into a single analysis.
Differences in species distribution across the soil moisture gradient may be partly explained by differences in seedling morphology and the interaction of morphology with fire regime. Several studies have shown that thin bark and intense fires can kill oaks (Williams and Black, 1981; Waldrop et al., 1992; Rebertus et al., 1995). In our study, the two live oak species, Quercus geminata, and Q. virginiana have particularly thin bark as juveniles and are extremely vulnerable to fire during early stages of establishment (Rebertus, 1988; Guerin, 1993). Their distributions were centered on the driest and wettest sites, respectively. It is likely, therefore, that these two species were restricted to locations that could provide safe sites for establishment and juvenile growth. We have already discussed our hypothesis that dry sites burn extremely heterogeneously, frequently leaving unburned patches, and that fires can be patchy in wet sites too if burning occurs when parts of the landscape are flooded (which we have observed at the Jones Center).
In the sapling layer, we hypothesize that Quercus incana may be the most resistant (i.e., able to survive top-kill) to fire. Quercus incana was about the only species to occur as saplings outside of the sand ridges. In the regeneration layer, both Q. incana and Q. margaretta were extremely resilient to 70 yr of cool season burning (i.e., were able to establish and vigorously resprout after fire) with Q. margaretta most common in dry sites and Q. incana in intermediate sites. The occurrence of Q. incana across the soil moisture gradient in this study (except in pond margins) shows that it is not a xeric site specialist as implied by Christensen (1988) and Ware et al. (1993), but is in fact present across a fairly broad range of soil moisture conditions (Peet and Allard, 1993).
Management implications. - Whether oak densities at our study sites have been stimulated or not by a long-term cool season burning regime, fragmentation of the landscape or their interaction, the fact remains that oak regeneration was extremely abundant. Without diligent use of periodic fire, overstory structure in the Pinus palustris forests at the Jones Center would quickly change from mostly pine to mixed oak-pine with a dense midstory of oaks. Furthermore, efforts to use ecosystem management may have unintended outcomes. For example, if prescribed fires are changed to match the natural disturbance regimes of pre-European longleaf pine-wiregrass woodlands (i.e., 1-10 yr between fires for a given patch with appropriate heterogeneity in return intervals; Robbins and Myers, 1992), many oaks may escape fires in time and threaten the maintenance of pine dominance of the overstory. Understanding the controls on oak population dynamics is essential to ecosystem management of these fire maintained systems.
Acknowledgments. - Funding for this project was provided by the Joseph W. Jones Ecological Research Center We thank Jimmy Adkinson, Robert Smith, Greg Houseal and Preston Parker for technical assistance and Brian Palik and Larry Landers tot help with study design and analysis.
BOYER, W. D. 1990. Growing-season burns for control of hardwoods in longleaf pines stands. USDA Forest Service Res. Pap. SO-256.
-----. 1993. Season of burn and hardwood development in young longleaf pine stands, p. 511-515. In: J. C. Brissette (ed.). Proceedings of the seventh biennial silvicultural research conference. USDA Forest, Service Gen. Tech. Rep. SO-93. 7 p.
CHRISTENSEN, N. L. 1981. Fire regimes in southeastern ecosystems, p. 112-136. In: H. A. Mooney, T. M. Bonnicksen, N. L. Christensen, J. E. Lotan and W. A. Reiners (eds.). Fire regimes and ecosystem properties. USDA Forest Service Gen. Tech. Rep. WO-26.
-----. 1988. Vegetation of the Coastal Plain of the southeastern United States, p. 317-363. In: M. G. Barbour and W. D. Billings (eds.). North American terrestrial vegetation. Cambridge University Press, New York.
DAY, P. R. 1965. Particle fractionation and particle-size analysis, p. 545-567. In: C. A. Black et al. (eds.). Methods of soil analysis, Part I. Agronomy 9:545-567.
GLITZENSTEIN, J. S, W. J. PLATT AND D. R. STRENG. 1995. Effects of fire regime and habitat on tree dynamics in north Florida longleaf pine savannas. Ecol. Monogr., 65:441-446.
GODFREY, R. K. 1988. Trees, shrubs and woody vines of northern Florida and adjacent Georgia and Alabama. University of Georgia Press, Athens, Georgia. 734 p.
GOEBEL, P. C., B. J. PALIK, L. K. KIRKMAN AND L. WEST. 1997. Landscape ecosystem types of Ichauway. Joseph W. Jones Ecol. Res. Center, Newton, Georgia. 99 p.
GROSENBAUGH, L. R. 1952. Plotless timber sampling - new, fast, easy. J. Forestry, 50:32-37.
GUERIN, D. N. 1993. Oak dome clonal structure and fire ecology, in a Florida longleaf pine dominated community, Bull. Torrey Bot. Club, 120:107-114.
HARCOMBE, P. A., J. S. GLITZENSTEIN, R. G. KNOX, S. L. ORZELL AND E. L. BRIDGES. 1993. Vegetation of the longleaf pine region of the west gulf coastal plain, p. 83-125. In: Proceedings of the 18th Tall Timbers Fire Ecology Conference. Tall Timbers Research Station, Tallahassee, Florida.
HOOK, D. D. 1984. Waterlogging tolerance of lowland tree species of the South. South. J. Appl. For., 8:136-149.
ISAAC, R. A. AND J. D. KERBEN. 1971. Atomic absorption and flame photometry: techniques and uses in soil, plant, and water analysis, p. 17-38. In: L. M. Walsh (ed.). Instrumental methods for analysis of soil and plant tissue. Soil Science Society of America, Madison, Wisconsin.
JAMES, F. C. AND C. E. MCCULLOCH. 1990. Multivariate analysis in ecology and systematics: Panacea or Pandora's Box? Annu. Rev. Ecol. Syst., 21:129-166.
JONES, S. M., D. H. VAN LEAR AND S. K. COX. 1984. A vegetation-classification of forest sites within the upper Coastal Plain of South Carolina. Bull. Torrey Bot. Club, 111:349-360.
JONGMAN, R. H. G., C. J. F. TER BRAAK AND O. F. R. VAN TONGEREN (eds.). Data analysis in community and landscape ecology. Cambridge Univ. Press, Cambridge, U.K.
KLEMMEDSON, J. O. 1987. Influence of oak in pine forests of central Arizona on selected nutrients of forest floor and soil, Soil Sci. Soc. Am, J., 51:1623-1628.
-----. 1991. Oak influence on nutrient availability in pine forests of central Arizona. Soil Sci. Soc. Am. J., 55:248-253.
LONG, T. J. 1993. Biomass partitioning and water relations in seedlings of Quercus species native to the southeastern United States. M.S. Thesis. Auburn University, Alabama. 84 p.
----- AND R. H, JONES. 1996. Seedling growth strategies and seed size effects in fourteen oak species native to different soil moisture habitats. Trees Struct. Funct., 11:1-8.
LYNCH, J. M., A. K. GHOLSON JR. AND W. W. BAKER. 1986. Natural features of Ichauway Plantation, Georgia. J.W. Jones Ecological Research Center, Newton, Georgia. 308 p.
MARKS, P. L. AND P. A. MARCOMBE. 1981. Forest vegetation of the big thicket, southeast Texas. Ecol. Monogr., 51:287-305.
MAVITY, E. M. 1992. Physiological ecology of four species of Quercus on the sandhills of Georgia. M.S. Thesis. University of Georgia, Athens, Georgia. 65 p.
MIYANISHI, K. AND M. KELLMAN. 1986. The role of fire in recruitment of two neotropical savanna shrubs, Miconia albicans and Clidemia sericea. Biotropica, 18:224-230.
MULLER, C. H. 1952. Ecological control of hybridization in Quercus: a factor in the mechanism of evolution. Evolution, 6:147-161.
MYERS, R. L. 1985. Fire and the dynamics relationship between Florida sandhill and sand pine scrub vegetation. Bull. Torrey Bot. Club, 112:241-252.
MYERS, R. L. 1990. Scrub and high pine, p. 150-193. In: R. L. Myers and J. J. Ewel (eds.). Ecosystems of Florida. University of Central Florida Press, Orlando, Florida.
OLSEN, S. R. AND L. E. SOMMERS. 1982. Phosphorus, p. 430-430. In: A. L. Page, R. H. Miller, and D. R. Keeney (eds.). Methods of soil analysis. Part 2. Chemical and microbiological properties, 2nd ed. American Society of Agronomy and Soil Science Society of America, Madison, Wisconsin.
OLSON, M. S. AND W. J. PLATT. 1995. Effects of habitat and growing season fires on resprouting of shrubs in longleaf pine savannas, Vegetatio, 119:101-118.
PALIK, B. J. AND N. PEDERSON. 1996. Overstory mortality and canopy disturbances in longleaf pine ecosystems. Can. J. For. Res., 26:2035-2047.
PALMER, M. W. 1993. Putting things in even better order: the advantages of canonical correspondence analysis. Ecology, 74:2215-2230.
PEET, R. K. AND D. J. ALLARD. 1993. Longleaf pine vegetation of the Southern Atlantic and Eastern Gulf Coast regions: a preliminary classification, p. 45-81. In: Proceedings of the 18th Tall Timbers Fire Ecology Conference. Tall Timbers Research Station, Tallahassee, Florida.
PLATT, W. J., J. S. GLITZENSTEIN AND D. R. STRENG. 1989. Evaluating pyrogenicity and its effects on vegetation in longleaf pine savannas, p. 143-161. In: Proceedings of the 17th Tall Timbers Fire Ecology Conference. Tall Timbers Research Station, Tallahassee, Florida.
REBERTUS, A. J. 1988. The effects of fire on forest community composition, structure, and pattern in Florida sandhills. Ph.D. Dissertation. Louisiana State University, Louisiana. 127 p.
-----, G. B. WILLIAMSON AND E. B. MOSER. 1989. Longleaf pine pyrogenicity and turkey oak mortality in Florida xeric sandhills. Ecology, 70:60-70.
-----, ----- and W. J. Platt. 1993. Impact of temporal variation in fire regime on savanna oaks and pines, p. 215-225 In: Proceedings of the 18th Tall Timbers Fire Ecology Conference. Tall Timbers Research Station, Tallahassee, Florida.
ROBBINS, L. E. AND R. L. MYERS. 1992. Seasonal effects of prescribed burning in Florida: a review. Tall Timbers Research, Inc. Miscellaneous Publication No. 8.
STOUT, I. J. AND W. R. MARION. 1993. Pine Flatwoods and Xeric Pine Forests of the southern (lower) coastal plain, p. 373-446. In: W. H. Martin, S. G. Boyce and A. G. Echternacht (eds.). Biodiversity of the southeastern United States: lowland terrestrial communities. John Wiley, New York.
STRENG, D. R. AND P. A. HARCOMBE. 1982. Why don't East Texas savannas grow up to forest? Am. Midl. Nat., 108:278-294.
-----, J. F. GLITZENSTEIN AND W. J. PLATT. 1993. Evaluating effects of season of burn in longleaf pine forests: a critical review and some results front an ongoing long-term study, p. 227-263. In: Proceedings of the 18th Tall Timbers Fire Ecology Conference. Tall Timbers Research Station, Tallahassee, Florida.
TER BRAAK, C. J. F. 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology, 67:1167-1179.
-----. 1990. Update notes: CANOCO, 3.1. Agricultural Mathematics Group, Wageningen, The Netherlands. 35 p.
TOPP, G. C., J. L. DAVIS AND A. P. ANNAN. 1980. Electromagnetic determination of soil water content: measurement in coaxial transmission lines. Water Res. Res., 16:579-582.
USDA, SOIL CONSERVATION SERVICE. 1986. Soil surveys of Baker and Mitchell counties, Georgia. 121 p.
VAITKUS, M. R. AND K. W. MCLEOD. 1995. Photosynthesis and water-use efficiency of two sandhill oaks following additions of water and nutrients. Bull. Torrey Bot. Club, 122:30-39.
WALBRIDGE, M. R. 1991. Phosphorus availability in acid organic soils of the lower North Carolina Coastal Plain. Ecology, 72:2083-2100.
WALDROP, T. A., D. L. WHITE AND S. M. JONES. 1992. Fire regimes for pine-grassland communities in the southeastern United States. For. Ecol. Manage., 47:195-210.
WARE, S., C. FROST AND P. D. DOERR. 1993. Southern Mixed Hardwood Forest: the former Longleaf Pine Forest, p. 447-493. In: W. H. Martin, S. G. Boyce and A. C. Echternacht (eds.). Biodiversity of the southeastern United States: lowland terrestrial communities. John Wiley, New York.
WEDIN, D. A. AND D. TILMAN. 1990. Species effects on nitrogen cycling: a test with perennial grasses. Oecologia, 84:433-441.
WOOD, C. W., R. J. MITCHELL, B. R. ZUTTER AND C. L. LIN. 1992. Loblolly pine plant community effects on soil carbon and nitrogen. Soil Sci., 154:410-419.
WILLIAMS, G. B. AND E. M. BLACK. 1981. High temperatures of forest fires tinder pines as a selective advantage over oaks. Nature, 293:643-644.
TABLE 2. Canonical correspondence analysis of 64 plots along environmental gradients in longleaf pine woodlands of the Gulf Coastal Plain of Georgia Data set Axis 1 Axis 2 Axis 3 Axis 4 Eigenvalues All strata 0.40 0.15 0.11 0.09 Trees 0.73 0.57 0.37 0.17 Saplings 0.60 0.39 0.11 0.04 Regeneration 0.51 0.33 0.17 0.08 Cumulative percent of species inertia(*) All strata 21.2 29.0 34.6 39.3 Trees 31.2 55.4 71.2 78.4 Saplings 50.5 83.9 92.9 96.2 Regeneration 31.3 51.1 61.7 66.8 Species-environment correlation All strata 0.96 0.81 0.80 0.79 Trees 0.96 0.93 0.93 0.92 Saplings 0.99 0.98 0.97 0.98 Regeneration 0.99 0.89 0.81 0.80 Cumulative percent of species-environment relationship All strata 36.7 50.3 60.0 68.2 Trees 35.9 64.0 82.1 90.5 Saplings 51.8 86.0 95.3 98.7 Regeneration 43.0 70.1 84.6 91.6 * Inertia is the variation in the species data that can be potentially explained by ordination, it is calculated as the sum of eigenvalues for ordination unconstrained by environmental variables (ter Braak, 1990)
|Gale Copyright:||Copyright 1999 Gale, Cengage Learning. All rights reserved.|