| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Plant, Soil, and Environmental Sciences
b Dep. of Geological Sciences, Univ. of Maine, Orono, ME 04469
c USDA Forest Service, Northeastern Experiment Station, Durham, NH 03824
* Corresponding author (ivanjf{at}maine.edu)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: ANP, Acadia National Park • BBWM, Bear Brook Watershed in Maine • BRN-REF, reference watershed for the wildfire • BRN-TRT, burned watershed • GLM, general linear model • MWD, moderately well drained • NIT-REF, reference watershed for the (NH4)2SO4 treatment • NIT-TRT, (NH4)2SO4–treated watershed • SOM, soil organic matter • VPD, very poorly drained • WPW, Weymouth Point Watershed • WTH-REF, reference watershed for whole-tree harvesting • WTH-TRT, whole-tree harvested watershed
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
22% of the world's terrestrial C pool (Bruce et al., 1999). Significant changes in the C storage of these soils may alter the global C cycle. Carbon stored in SOM represents the net balance between litter inputs and heterotrophic respiration in terrestrial ecosystems. The factors that control organic matter levels in soils include climate, topography, parent material, biological activity, vegetation, and time (Jenny, 1941). Terrestrial C balances may also be influenced by perturbations, such as N deposition, fire, and harvesting. Elevated emissions of nitrogen oxides and ammonia are primarily because of combustion of fossil fuels, manufacture and use of fertilizers, livestock waste, and burning of biomass (Galloway et al., 1995). Subsequent increases in N deposition have been concentrated in mid-latitude regions where fertilizer use and industrial emissions are the highest (Peterson and Mellilo, 1985; NADP/NTN, 1998). Increased N deposition has been estimated to increase global terrestrial C uptake at a rate of 0.1 to 2.3 Pg yr-1 (Peterson and Mellilo, 1985; Schindler and Bayley, 1993; Townsend et al., 1996; Holland, 1997), assuming that terrestrial productivity is limited by N (Townsend et al., 1996). Nadelhoffer et al. (1999b) demonstrated that soils are the dominant sink for N deposition in Maine and other northern temperate forests. Although soils were shown to assimilate nearly 15 times more N deposition than wood, soils sequestered slightly less C due to lower soil C/N ratios. However, soil acidification associated with N deposition may reduce decomposition rates (Martikainen et al., 1989; Nohrstedt et al., 1989; Prescott, 1995), thereby increasing the soil C reservoir. Clearly, the effect of N deposition on the soil C reservoir remains complex and subject to debate.
The potential for a warmer climate with altered precipitation patterns and lightning frequency, in response to or exacerbated by increased greenhouse gas emissions, may change wildfire frequency. Flannigan and Van Wagner (1991) predicted a 50% increase in fire frequency in the boreal and subboreal forests in Canada if the concentration of atmospheric CO2 doubled. Increased fire occurrence may act as a positive feedback to climate change by reducing terrestrial C and N reservoirs and increasing atmospheric CO2 and NOx concentrations. Wildfire could be an important factor in determining the long-term sequestration of C and N in forest soils. Despite the current low fire frequency in Maine forests, the long-term effects of wildfire on soil C and N pools must be better understood, given future climatic uncertainties.
As the demand for forest products increases, there is also a concomitant concern about the potential effects of forest harvesting on soil C storage. In a review of the literature, Johnson (1992) found no change in average soil C content with forest harvesting, although individual sites showed net losses or gains depending on residue management. However, Fan et al. (1998) attributed modeled terrestrial uptake of C in North American forests to the regrowth of abandoned farmland and previously logged forests. In Maine, 90% of the land area is forested, of which 43% is owned by large industrial forest companies who provide 25% of the nation's paper (Seymour and Lemin, 1989; Gadzik et al., 1998). Thus forest harvesting and forest policy in Maine has the potential to significantly influence soil C storage.
The objective of this study was to evaluate soil C and N pools at three forested watersheds in Maine that represented perturbations due to experimental N enrichment, wildfire, and whole-tree harvesting.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The BBWM is located in eastern Maine (44°52'N, 44°52'W), 50 km from the Gulf of Maine, on the upper 265 to 475 m of the southeast slope of Lead Mountain. The average slope from the top of the watershed to the weirs is 31%. The West Bear watershed has been treated bimonthly with ammonium sulfate [(NH4)2SO4] since November 1989 as part of a whole-watershed manipulation experiment designed to investigate the effects of atmospheric deposition of N and S. Granular (NH4)2SO4 has been aerially applied at 28.8 kg S ha-1 yr-1 and 25.2 kg N ha-1 yr-1. The treated watershed, West Bear (NIT-TRT, 10.2 ha), is adjacent to the reference watershed, East Bear (NIT–REF, 10.7 ha). The upper reaches of the watersheds are predominately pure red spruce (Picea rubens Sarg.) stands, and the lower 60% of the watersheds are mixed northern hardwood stands, dominated by American beech (Fagus grandifolia Ehrh.), sugar maple (Acer saccharum Marsh), and red maple (Acer rubrum L.). Hardwood stands are 40 to 50 yr old, reflecting previous harvesting practices, while softwood stands on the upper slope are 80 to 100 yr old. Soils are predominantly Typic and Lithic Haplorthods formed from dense basal till. Additional site characteristics and biogeochemical data on soils and soil solutions from this study site may be found in Fernandez et al. (1999), Kahl et al. (1999), and Norton et al. (1999b).
During the dry season of 1947, wildfires consumed thousands of hectares throughout northern New England. Approximately 6956 ha burned on Mount Desert Island, located on the coast of eastern Maine (44°23'N, 69°15'W), which included >4000 ha of ANP. The Canon Brook watershed (BRN-TRT) is located on the eastern slope of Cadillac Mountain at a 152- to 457-m elevation in an area of ANP that burned in 1947. The southern tributary of the Canon Brook stream is located in the BRN-TRT watershed. Pioneer species such as paper birch (Betula papyrifera Marsh) and striped maple (Acer pennsylvanicum L.) currently dominate the BRN-TRT watershed. Prior to European settlement in the late 1700s, fires burned infrequently (approximately every 500 yr) in coastal Maine forests (Patterson et al., 1983). However, large fires occurred on Cadillac Mountain in 1889 and in 1896 or 1889 (Moore and Taylor, 1927). The Hadlock Brook watershed (BRN-REF) lies on the southeastern slope of Sargent Mountain at an elevation of 152 to 396 m, and is used as a reference watershed. Hadlock Brook drains an area that escaped the 1947 fire and red spruce is the dominant canopy species. Both watersheds have similar soils, predominantly coarse-loamy, mixed, frigid, Aquic Haplorthods. The watersheds are 4.5 km apart and sampling sites had slopes of 20 to 30%.
The WPW is located on commercial spruce-fir forest land in northern Maine (49°57'N, 69°19'W) at an elevation of 287 to 315 m. In 1981, a whole-tree harvest was conducted, removing 90% of the 232 Mg ha-1 of available biomass from a 48-ha watershed (WTH-TRT) (Smith et al., 1986). An adjacent watershed (73 ha) was not harvested and served as the reference watershed (WTH-REF). Vegetation on WTH-REF consists of a two-aged red spruce and balsam fir [Abies balsamea (L.) Mill.] forest that developed from the 1913 to 1919 spruce budworm [Choristneura fumiferana (Clem.)] epidemic. Regeneration on the WTH-TRT watershed is predominantly red spruce and balsam fir, which were 2 to 3 m in height at the time of sampling. Soils in the WPW are coarse-loamy, mixed, frigid Aquic Haplorthods and Aeric Haplaquepts of the Chesuncook catena formed from dense basal till. Drainage class differs significantly across the gently sloping landscape because of classic pit and mound topography, ranging from moderately well drained (MWD), which accounts for 25% of the total WPW area, to very poorly drained (VPD), which accounts for 34% of the total WPW area.
Soil Sampling
Soil sampling depth increments included the forest floor, the upper 5 cm of the B horizon, and the 5- to 25-cm increment of the B horizon. A horizons were not present. The E horizon was thin, discontinuous, and typically low in C or N content in the soils at our sites. The chemical characterization of E horizons also often reflects admixtures of overlying O or underlying B horizon material incorporated during sampling. For these reasons, E horizons were sampled, but chemical analyses were not conducted on these samples. Forest floors were quantitatively sampled at each site with a 0.71 by 0.71 m frame (Fernandez et al., 1993). Mineral soil was also quantitatively sampled at BBWM, but grab samples were collected at ANP and WPW.
Sampling at BBWM focused on capturing potential contrasts between NIT-TRT and NIT-REF watersheds within two dominant forest types after 8 yr of N additions. The contrast of forest stand types at BBWM also allowed potential differences between hardwood and softwood forest types to be evaluated. Within each watershed, three soil pits were excavated in hardwood stands and three in softwood stands within the Tunbridge (coarse-loamy, isotic, frigid Typic Haplorthods) soil series. Soil samples were collected June through August of 1998, and watershed and forest types were equally represented during each sampling month.
Sampling at ANP was intended to capture potential differences in BRN-TRT and BRN-REF watersheds 50 yr after a wildfire. Soil sampling at ANP was limited to two stands to minimize destructive sampling within the national park as per our sampling permit. One stand was selected in each of the BRN-TRT and BRN-REF watersheds at similar elevations (275 m). A transect was established in each stand with 12 equidistant sampling points spaced at 4-m intervals. At each sampling point, a 15 by 15 cm soil sample was excavated from the Dixfield (coarse-loamy, isotic, frigid Aquic Haplothods) soil series. Three adjacent sampling points were bulked and homogenized in the field to create four samples per transect. Sampling at BRN-TRT was conducted in September 1998 prior to leaf fall, and BRN-REF was sampled in October 1998.
Sampling at WPW focused on capturing potential contrasts between WTH-TRT and WTH-REF watersheds within two soil drainage classes 17 yr after the harvest. Sixteen study plots (10 by 10 m) were selected from the 27 plots described in the experimental design of Briggs et al. (1999). The selected study plots were stratified by two soil drainage classes, MWD and VPD. For this study, soil pits were excavated adjacent to the four existing plots in each of MWD and VPD soil drainage classes in each watershed. Samples were collected in June and July of 1998, and watersheds and drainage classes were equally represented during each sampling month.
Mineral soils from all sites were sieved (6 mm) and homogenized in the field, and subsamples were taken to the laboratory for analysis. However, WPW-VPD soils were too wet to sieve in the field and were sieved (2 mm) in the laboratory after they were air dried. Forest-floor samples were collected in their entirety and taken to the laboratory for analysis.
Laboratory Analysis
Soils were air dried, then sieved through either 2-mm (mineral) or 6-mm (organic) mesh sieves to isolate the respective fine earth and coarse fractions. Coarse fragments were removed from the coarse organic samples (>6 mm). Percent air-dry moisture was measured for fine earth soils and coarse organic soils (>6 mm) (Robarge and Fernandez, 1986). Fine earth soils were measured for pH using 0.01 M CaCl2 (Hendershot et al., 1993), and SOM by loss-on-ignition at 450°C for 12 h. Total C and N were measured using a LECO CN 2000 (St. Joseph, MI) analyzer employing the Dumas method of combustion at 1350°C in a pure O2 environment. Coarse organic soils were ground and homogenized prior to C and N analysis.
Computations
Quantitative soil sampling allowed for direct computation of soil mass per unit area (Fernandez et al., 1993) for the O horizon at all sites and for all sampling increments at BBWM. To calculate C, N, and SOM concentrations for the entire forest floor, fine (<6 mm) and coarse (>6 mm) O horizon data were mass weighted. The concentration of SOM in the coarse fraction of the forest floor was assumed to be 100%.
Statistical Analysis
All analyses were carried out using the Statistical Analysis System (SAS Institute, 1988) with an alpha level of 0.05. Because the data did not meet the assumptions of normality and equality of variance, a rank transformation was used (Conover, 1971; Zar, 1984). A general linear model (GLM) was applied to the ranked data for each site to analyze the differences among main effects (watersheds, forest types, and drainage class) and interactions. Layout for the statistical analyses are shown in Table 1.
|
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Nitrogen Enrichment
After 8 yr of experimental whole-watershed (NH4)2SO4 additions at BBWM, N concentrations were significantly higher in the upper 5 cm of the B horizon of NIT-TRT than NIT-REF (Table 2). Soil C/N ratios in the upper 5 cm of the B horizon of NIT-TRT reflected higher N concentrations and were significantly lower than NIT-REF. Because of numerically lower C and significantly higher N concentrations in the forest floor of NIT-TRT, forest-floor C/N ratios in NIT-TRT were also significantly lower than NIT-REF. After 3 yr of treatment at BBWM, Wang and Fernandez (1999) found significantly lower forest-floor C concentrations in mixedwood stands of NIT-TRT compared with NIT-REF, although no significant differences were detected in softwood and hardwood stands. We used power analyses (Zar, 1984) with our data to determine that future studies of similar design would require 106 and 536 samples per watershed for forest-floor C and N concentrations, respectively, to detect significant differences (power = 0.80) between NIT-TRT and NIT-REF watersheds.
|
82% of the cumulative N additions from 1989 to 1997 (Kahl et al., 1999). Nitrogen enrichment in the NIT-TRT watershed produced higher foliar N concentrations (White et al., 1999), higher soil solution NO-3 concentrations (Fernandez et al., 1999), and higher stream N export (Kahl et al., 1993, 1999; Norton et al., 1999a). Higher concentrations of N in surface soil horizons of NIT-TRT than NIT-REF are consistent with anticipated increases in N because of surface applications of the (NH4)2SO4 treatment and litter inputs with higher N concentrations. Nitrogen pools in the upper 5 cm of the B horizon were significantly greater (22%) in NIT-TRT than NIT-REF, reflecting higher N concentrations in NIT-TRT (Table 2). However, forest-floor SOM, C, and N pools were significantly lower in NIT-TRT than NIT-REF. These differences were driven by a significantly lower forest-floor mass in NIT-TRT than NIT-REF. It is possible that N enrichment in NIT-TRT may have increased forest-floor decomposition rates compared with NIT-REF, reducing NIT-TRT forest-floor mass (Gill and Lavender, 1983; Hunt, 1988; Fenn, 1991; McNulty et al., 1991; Conn and Day, 1996; Downs et al., 1996). Higher C turnover in the NIT-TRT forest floor leading to N-pool depletion could have contributed to some illuvial accumulation of N in the underlying 5 cm of mineral soil. Because C and SOM concentrations were not statistically different between watersheds by depth, it is also probable that these forest-floor differences may simply reflect antecedent conditions.
Forest Type Effects
We also evaluated the potential effects of stand composition on soil C and N at BBWM. The distribution of C and N concentrations and pools in hardwood and softwood stands by depth are shown in Table 3. Significantly lower forest-floor C and N pools were found in soils supporting hardwoods compared with softwoods when comparing the main effects in the data (n = 6). Forest-floor C and N concentrations were not significantly different between softwood and hardwood stands. Power analyses suggest that future studies would require 14 and 86 samples per forest type to detect significant differences (power = 0.80) in forest-floor C and N concentrations, respectively, between these forest types based on these data. Lower forest-floor C and N pools in hardwood stands compared with softwood stands were driven by significantly lower forest-floor masses. Carbon and N concentrations in the upper 5 cm of the B horizon were significantly lower in hardwood soils compared with softwood soils, although C and N pools were not different between forest types. Carbon and N pools in the 5- to 25-cm increment of the B horizon were significantly lower in hardwood stands than in softwood stands. Unlike the forest floor, differences in the 5- to 25-cm increment were a consequence of significantly lower C and N concentrations in hardwood stands than softwood stands. Significant interactions among watersheds and forest types were not found; however, Table 3 shows trends in soil C and N data consistent with the main effect response of reduced C and N pools and narrower C/N in hardwood soils at all depth increments compared with softwoods. This is also consistent with evidence of higher rates of N mineralization and C turnover in NIT-TRT soils compared with NIT-REF soils as shown by Wang and Fernandez (1999).
|
50-yr-old hardwood stands may have also exacerbated differences between hardwood and softwood soils, particularly in the forest floor and uppermost mineral soils (0–5 cm). Nevertheless, these findings highlight the importance of considering forest type when quantifying forest-soil C and N pools, even though forest type differences often reflect a combination of ecological and management factors.
Wildfire
Fifty years after the wildfire in ANP, C and N concentrations and C/N ratios in the forest floor of BRN-TRT were significantly lower than BRN-REF (Table 2). Consequently, C, N, and SOM pools in the forest floor of BRN-TRT were significantly lower than BRN-REF. Forest-floor C and N pools in BRN-TRT were 52 and 23% lower than forest-floor C and N pools in BRN-REF, respectively. Because this wildfire occurred during a particularly dry period and because wildfires generally have a low frequency in coastal Maine forests (Moore and Taylor, 1927; Patterson et al., 1983), the wildfire in this study was an intense burn. Thus fire may have directly reduced forest-floor C and N concentrations and pools in BRN-TRT in gaseous or particulate forms via volatilization or ash convection (Boerner, 1982; Raison et al., 1985). Fire may have also induced additional C and N losses by increasing soil erodibility (Díaz-Fierros et al., 1987; McNabb and Swanson, 1990; Andreu et al., 1996) and decomposition rates (Schoch and Binkley, 1986; Fernández et al., 1997).
Despite the plausible reduction in forest-floor C and N at the time of the fire, the concomitant change from a softwood to hardwood forest in BRN-TRT probably contributed to lower forest-floor C and N concentrations and pools in BRN-TRT 50 yr after the wildfire.
At BBWM, forest-floor C and N pools were similarly lower in hardwood stands than softwood stands; however, this was a result of lower forest-floor masses in hardwood stands. At ANP, lower forest-floor C and N pools in BRN-TRT reflected lower C and N concentrations rather than lower forest-floor masses. The interplay between changes in forest-floor mass and composition with different forest types deserves further study. Nevertheless, it is probable that the change in forest type and thus litter quality increased forest-floor decomposition rates at BRN-TRT compared with BRN-REF. Furthermore, antecedent conditions may have also contributed to lower C and N pools in BRN-TRT. Given the variability in the ANP data, future studies at this site would require 22 samples per watershed to detect significant differences (power = 0.80) between forest-floor masses in these watersheds.
In contrast to the forest floor, the upper 5 cm of the B horizon in BRN-TRT had higher C and N concentrations and lower C/N ratios than BRN-REF (Table 2). Elevated C and N concentrations in the upper mineral soil may be a direct result of the fire. Charcoal and partially burned organic matter may have been incorporated into the mineral soil, increasing mineral soil C and N concentrations (Johnson, 1992). In addition, an increased presence of N-fixing species after the wildfire may have increased mineral soil C and N concentrations (Johnson, 1992). Downward movement of finely divided particulate matter can also contribute to C and N enrichment in the upper mineral soil after fires (Dyrness and Norum, 1983). However, the shift from softwood to hardwood forest types probably contributed to higher C and N concentrations and lower C/N ratios in the upper 5 cm of the B horizon in BRN-TRT 50 yr after the wildfire. Thus presumably faster rates of decomposition in the hardwood forests of BRN-TRT compared with BRN-REF may have redistributed C and N from the forest floor to the upper 5 cm of the B horizon in BRN-TRT soils. Contrary to our results at ANP, C and N concentrations were significantly lower in the upper 5 cm of B horizon soils in hardwood stands compared with softwood soils at BBWM (Table 3). Therefore, differences at ANP may reflect additional influences of the 1947 wildfire beyond contrasting vegetation and litter quality between watersheds, including differences in antecedent conditions prior to the fire. It is probable that all of these factors played a role in the results.
Whole-Tree Harvest
Seventeen years after the whole-tree harvest at WPW, forest-floor C concentrations were significantly lower in WTH-TRT than WTH-REF (Table 2). The WTH-TRT forest-floor mass and forest-floor C and N pools were significantly lower than WTH-REF. Forest-floor C and N pools in WTH-TRT were 36% and 38% of the forest-floor C and N pools, respectively, in WTH-REF. Because conifer forest types were present at WTH-TRT before and after the harvest, differences are not attributed to forest-type effects. However, whole-tree harvests commonly result in increased rates of decomposition because of increased soil temperature and moisture (Ovington, 1968; Witkamp, 1971; Marks and Bormann, 1972; Edwards and Ross-Todd, 1983; Mroz et al., 1985), and in initial decreases in leaf and wood litter inputs because of overstory removal. Thus accelerated decomposition rates and reductions in litter production probably resulted in lower forest-floor C concentrations and masses in WTH-TRT. Lower biological nutrient demands and increased decomposition rates in the WTH-TRT watershed may also explain increases in stream water N concentrations immediately following the whole-tree harvest in 1981 to 1984 (Hornbeck et al., 1990).
In northern hardwood forests, Covington (1981) found that the mass of SOM in the forest floor decreased by more than 50% (a decrease of 30.7 Mg ha-1) in the 15 yr following a whole-tree harvest. During the next 50 yr in that study, SOM content in the forest floor increased by 28.0 Mg ha-1 and by Year 64 was within 5% of an equilibrium value of 56.0 Mg ha-1. Aber et al. (1978) used a modeling approach that predicted declines in forest-floor organic matter for 15 to 30 yr after harvesting, and estimated that 60 to 80 yr may be required for SOM levels to recover. Thus at 17 yr after harvest, WPW may have shifted from the degradation phase to the aggradation phase for SOM.
Drainage Class Effects
Contrasting soil drainage classes (MWD and VPD) at WPW allowed us to evaluate the potential effects of soil drainage class on C and N pools at this site. One might expect that anaerobic conditions in WPW-VPD soils would increase soil C and N by reducing heterotrophic oxidation of SOM compared with more aerobic conditions in WPW-MWD soils. However, C and N concentrations and pools were not different between drainage classes. In addition, significant interactions between watershed and drainage class were not found. Given the variability in these data, power analyses indicated that future studies at this site would require 93 and 42 forest-floor samples, 85 and 128 upper B horizon (0–5 cm) samples, and 49 and 74 lower B horizon (5–25 cm) samples per drainage class to detect significant differences (power = 0.80) between C and N concentrations, respectively, in these highly variable WPW-VPD and WPW-MWD soils.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
Received for publication April 17, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
naas, C.J. Koopmans, P. Schleppi, A. Tietema, and R.F. Wright. 1999b. Nitrogen deposition makes a minor contribution to carbon sequestration in temperate forests. Nature 398:145–148.This article has been cited by other articles:
|
|
 |
|
  P. S. Homann, S. M. Remillard, M. E. Harmon, and B. T. Bormann Carbon Storage in Coarse and Fine Fractions of Pacific Northwest Old-Growth Forest Soils Soil Sci. Soc. Am. J., November 1, 2004; 68(6): 2023 - 2030. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
