JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 9 January 2007
Published in J Environ Qual 36:91-100 (2007)
DOI: 10.2134/jeq2006.0233
© 2007 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow An erratum has been published
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Saito, L.
Right arrow Articles by Szameitat, P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Saito, L.
Right arrow Articles by Szameitat, P.
Agricola
Right arrow Articles by Saito, L.
Right arrow Articles by Szameitat, P.
Related Collections
Right arrow Fire
Right arrow Isotopes

TECHNICAL REPORTS

Landscape and Watershed Processes

Fire Effects on Stable Isotopes in a Sierran Forested Watershed

Laurel Saitoa,*, Wally W. Millera, Dale W. Johnsona, Robert G. Quallsa, Louis Provencherb, Erin Carrolla and Peter Szameitatc

a Dep. of Natural Resources and Environmental Science, Univ. of Nevada Reno, Mail Stop 186, 1000 Valley Road, Reno, NV 89512
b The Nature Conservancy, One E. First St., Suite 1007, Reno, NV 89501
c 1475 Folsom Street, #362, Boulder, CO 80302

* Corresponding author (lsaito{at}cabnr.unr.edu)

Received for publication June 16, 2006.

   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
This study tested the hypothesis that stable C and N isotope values in surface soil and litter would be increased by fire due to volatilization of lighter isotopes. The hypothesis was tested by: (1) performing experimental laboratory burns of organic and mineral soil materials from a watershed at combinations of temperature ranging 100 to 600°C and duration ranging from 1 to 60 min; (2) testing field samples of upland soils before, shortly after, and 1 yr following a wildfire in the same watershed; and (3) testing field soil samples from a down-gradient ash/sediment depositional area in a riparian zone following a runoff event after the wildfire. Muffle furnace results indicated the most effective temperature range for using stable isotopes for tracing fire impacts is 200 to 400°C because lower burn temperatures may not produce strong isotopic shifts, and at temperatures ≥600°C, N and C content of residual material is too low. Analyses of field soil samples were inconclusive: there was a slightly significant effect of the wildfire on {delta}15N values in upland watershed analyses 1 yr postburn, while riparian zone analyses results indicated that {delta}13C values significantly decreased ~0.71{per thousand} over a 9 mo post-fire period (p = 0.015), and ash/sediment layer {delta}13C values were ~0.65{per thousand} higher than those in the A horizon. The lack of field confirmation may have been due to overall wildfire burn temperatures being <200°C and/or microbial recovery and vegetative growth in the field. Thus, the muffle furnace experiment supported the hypothesis, but it is as yet unconfirmed by actual wildfire field data.

Abbreviations: GLM, generalized linear model • MSE, mean squared error • SS, sum of squares • UNR, University of Nevada Reno


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
SURFACE RUNOFF can be a source of dissolved and suspended nutrient transport to nearby lakes and streams. For example, recent research detected high concentrations of biologically available N and P in surface runoff from natural Sierran watershed ecosystems (Miller et al., 2005). This is particularly important in pristine areas of the Sierra Nevada and elsewhere, where ultra-oligotrophic lakes are now trending toward mesotrophic status, and where direct links between watershed processes, surface runoff, nutrient content, and discharge water quality have not been well established (Reuter and Miller, 2000).

Fire (prescribed and wildfire) is known to have a major effect on long-term N budgets in Mediterranean-type forest ecosystems (Grier, 1975; Raison et al., 1985; Trabaud, 1994; Johnson et al., 1998; Baird et al., 1999; Neary et al., 1999), and wildfire is particularly important because of its greater severity, usually resulting in complete combustion of O horizons and often some loss of soil organic matter as well. Simple calculations indicate that even infrequent fire can have large impacts on the long-term N budgets of more mesic systems (Johnson et al., 2004). Wildfire typically burns components of forest ecosystems that have low C/N ratios (e.g., forest floor and foliage), leaving large woody tissues largely unburned in most cases (Auclair, 1985; Johnson et al., 1998, 2004). Thus, wildfire apparently causes disproportionately large losses of N compared to C.

Higher water yields and soil erosion are often observed after fires, especially in watersheds where burn events consume large amounts of vegetation and organic matter and are soon followed by large rainfall events (Gresswell, 1999). These large precipitation events can also be associated with increased nutrient inputs to water bodies (Minshall et al., 1997; Gresswell, 1999). If these nutrients are taken up by aquatic organisms, the stable isotope values of organisms in the aquatic food web may reflect changes in soil stable isotope values due to fire. Thus, it may be possible to detect the influence of fire-affected areas on the base isotope values of aquatic food webs if the effects of fire on mineral and organic soil residues can be traced with stable isotopes.

Stable isotope analysis involves the analysis of ratios of naturally occurring heavy and light stable isotopes for various elements. These ratios are measured for a given sample and are then compared with the ratio of a primary standard of the same element. The ratios (R), are calculated as the ratio of the heaviest isotope (higher mass number) to the lighter one, and are measured using an isotope ratio mass spectrometer. The standard notation used is (Peterson and Fry, 1987):

Formula 1[1]
where {delta} is reported in units of per mil (parts per thousand), Rsa = isotope ratio of a sample; Rstd = isotope ratio of a standard.

Many studies of fire effects on stable isotopes address {delta}15N as observed in plants after fire. Högberg (1997) notes that fire consumes the relatively 15N-depleted O horizons, causing plants invading after the fire to seek deeper, more 15N-enriched sources, which results in a net increase in foliar {delta}15N after a fire. Similarly, Grogan et al. (2000) found that all plant and soil samples were enriched in 15N after fire in a California Bishop pine (Pinus muricata D. Don) forest. On the other hand, Cook (2001) found greater foliar {delta}15N in rainforests and fire-excluded mesic savannahs than in fire-prone savannahs in Australia. He speculated that this pattern was attributable to post-fire nitrification and nitrate uptake by plants. Post-fire nitrification has been previously noted to cause a strong increase in soil NO3–, and because nitrification discriminates against 15N, this may in turn lead to a 15N-enriched residual total N pool over time (Herman and Rundel, 1989; Högberg, 1997).

Studies of fire effects on {delta}13C in soils have primarily focused on changes in soil C isotope ratios because of vegetation changes due to fire (Aranibar et al., 2003; Michelsen et al., 2004; Sala et al., 2005). The distinctive {delta}13C values of plants that use the Calvin cycle (C3), Hatch-Slack cycle (C4), and Crassulacean acid metabolism (CAM) photosynthetic pathways (Lajtha and Marshall, 1994) can be reflected in soil profiles after fire, with C3 plants often dominating soils with lack of fire, and C4 plants dominating soils exposed to frequent fires (Biggs et al., 2002; Aranibar et al., 2003).

Of more interest to the current study are changes in {delta}13C and {delta}15N values due to fire on surface soils that are likely to be washed off during runoff events to nearby streams, thereby affecting the aquatic food web. Aranibar et al. (2003) examined surface soils to a depth of 5 cm on plots with no fire vs. 3- and 1-yr fire frequencies in South Africa and found that surface soil {delta}15N did not exhibit any clear patterns along fire gradients, while soil {delta}13C was significantly higher in two of the recently burned sites as compared to the respective controls. Similarly, Roscoe et al. (2000) found that soil {delta}13C values in the surface litter at sites in Brazil were higher (i.e., had less 13C) as fire frequency increased from no fire to 10 fires in 21 yr. A recent study by Fernandez et al. (2004) using a 13C-labeled soil that was heated for 10 min at 385°C resulted in significant loss of 13C from the soil material and organic matter after heating.

The hypothesis of this study was that {delta}13C and {delta}15N values in soil and organic matter that is most likely to be washed off the watershed during storm runoff will be increased by fire due to volatilization of the lighter isotopes. This hypothesis was tested by: (1) performing experimental laboratory burns of O and A horizon materials from a South Lake Tahoe watershed; (2) analyzing field samples of upland soils before, shortly after, and 1 yr following a wildfire in the same watershed; and (3) analyzing field samples from a down-gradient ash/sediment depositional area in a riparian zone following a major runoff event 2.5 wk after the wildfire.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Site Description
The general study site location (Fig. 1) was 2 km southeast of South Shore Lake Tahoe, near Stateline, Nevada (57°30' N; 119°55' W). The dominant overstory consisted of late-development white fir (Abies concolor Gord. & Glend.), Jeffrey pine (Pinus jeffreyi Grev. & Balf.), and lesser amounts of sugar pine (Pinus lambertianna Dougl.[Strobus L. Mold.]). The understory was dominated by Sierra chinquapin (Castanopsis sempervirens Dudl. [Castanea s. Kell]), currant (Ribes spp.), and minor amounts of snowbush (Ceanothus velutinis Dougl. Ex Hook.) and bitterbrush (Purshia tridentata (Pursh) DC.). Soils have developed on granitic parent material belonging to the Cagwin-Rock Outcrop complex and are classified as mixed Typic Cryopsamments on predominantly north-facing slopes of 10 to 40% gradients.


Figure 1
View larger version (102K):
[in this window]
[in a new window]

  		Fig. 1. Location map of Gondola wildfire and sampling locations. Triangles ("riparian" and "forest") indicate sites where samples were collected for muffle furnace experiment. Squares indicate locations of upland samples collected before, shortly after, and 1 yr following the wildfire. Circles indicate locations of samples collected from the runoff event after the wildfire. The hatched area indicates the extent of the wildfire.

 
Sixteen 0.04-ha study plots were randomly selected in 2001 for a study originally designed and sampled to examine the effects of mechanical harvest and prescribed fire on forest health, nutrient cycling, and discharge water quality. On 3 July 2002, a wildfire occurred under the Heavenly Gondola in South Lake Tahoe. Approximately 280 ha of forested watershed were burned, a portion of which consumed approximately 7 of the 16 study plots (Sites 10 through 16), and partially burned two more (Sites 8 and 9; Fig. 1). The fire consumed >95% of all forest floor material, but had no significant effect on soil C concentrations, indicating that soil temperatures were not sufficiently elevated as to cause soil organic matter combustion (Murphy et al., 2006).

Two and a half weeks after the fire (18 July 2002), a 15.2 mm (0.6 inches) precipitation event caused massive erosion and a subsequent flow of uphill soil and ash materials into a riparian zone below the fire, covering an area of approximately 0.8 ha. The down-gradient riparian zone consisted of a well-vegetated quaternary alluvium deposit on slopes of 0 to 5%. The area contained an ephemeral drainage to Edgewood Creek with buried channel characteristics (i.e., sediment accumulation from previous events had buried stumps and shrub branches to some extent).

Muffle Furnace Experiment
Controlled burns were performed on soil and forest floor samples collected from the Gondola study site in a muffle furnace at varying intensities using a procedure similar to Blank et al. (1996) to evaluate the potential effects of fire on {delta}13C and {delta}15N values. One pair of samples was taken from the nondepositional area of the riparian zone down-gradient of the actual wildfire, and a second pair was collected from an adjacent unburned forested portion of the upland watershed (Fig. 1). For both pairs, a rectangular area of approximately 1350 cm2 was cut through the organic mat down to the mineral surface and retained for analysis. The underlying mineral surface was then sampled to a depth of 3 cm, providing O and A horizon material for each location.

Samples were first allowed to dry at ambient room temperature, then oven-dried at 60°C for 72 h. O horizon samples were ground through a Wiley mill and mineral portions were sieved through a 2-mm sieve to establish homogeneity. Following homogenization, all materials were again dried for 24 h at 60°C. Three replicate 1.5-gram subsamples of each horizon were weighed and burned in 25-mL crucibles in a preheated Thermolyne 1500 muffle furnace at four different temperatures (100, 200, 400, and 600°C) over four burn durations (1, 5, 15, and 60 min). The selected temperatures and durations were determined from studies of typical ground temperatures sustained in forest fires (Blank et al., 1996; Neary et al., 1999). Three samples were left in the room during the experiment as unburned controls. Burned samples were weighed before and after burning to determine weight loss due to volatilization. Burned and unburned samples were prepared for stable isotope and elemental analysis using about 2 mg of each sample from the O horizon, and 20 to 40 mg of each sample from the A horizon.

Upland Watershed Study
Preburn mineral soil was sampled from five random locations within each 0.04-ha plot from the control (i.e., unburned Sites 1 through 7) and burn treatment areas (Sites 8 through 16). Because this was a wildfire and therefore unanticipated, it was not possible to obtain soil temperature data; however, the presence of black ash and scorched needles that remained on trees suggested that this was a rather low intensity wildfire and thus soil temperatures should not have reached high enough levels for substantial soil organic matter combustion in most areas. Postburn mineral soil samples were collected approximately 1 m away from preburn sample locations. Soil was collected from four depths corresponding roughly to the designated A (0 to 10 cm; 10 to 30 cm), AC (30 to 60 cm), and C horizons (60 to 100 cm). All soil samples were homogenized according to study plot and depth. Composite samples were oven-dried at 55°C, passed through a 2-mm standard testing sieve, and prepared for stable isotope analysis. About 5 to 35 mg of each sample was analyzed.

Riparian Zone Study
Ash and sediment deposits and underlying A horizon soil samples were collected from 17 point locations within the general ash/sediment depositional area (Fig. 1) on 8 Oct. 2002, and again on 26 June 2003. As was the case with soils in the upland areas, the presence of black ash suggested that most of this material was not subjected to excessively high temperatures. Samples were collected using a standard core sampler for bulk density determination (Blake and Hartge, 1986). Approximately 5 to 10 mg of ash/sediment samples and 20 to 30 mg of A horizon samples were used for stable isotope analysis.

Stable Isotope and Carbon/Nitrogen Analysis
Stable C and N isotope analyses of all soil samples were performed at the University of California, Davis Stable Isotope Facility with a continuous flow isotope ratio mass spectrometer (IRMS; 20–20 mass spectrometer, PDZEuropa, Northwich, UK) after sample combustion in an on-line elemental analyzer (PDZEuropa ANCA-GSL). Gases were separated on a 0.5 m x 6 mm Carbosieve G chromatography column (Supelco, Bellefonte, PA, USA) before introduction to the IRMS. Sample isotope ratios were compared to those of standard gases injected directly into the IRMS. Final isotope values were adjusted to bring the mean of standard samples distributed at intervals in each analytical run to the correct values of working standards. The working standards are a mixture of ammonium sulfate and sucrose with {delta}15N v air 1.33{per thousand}, and {delta}13C v Pee Dee Belemnite-23.83{per thousand}. These working standards are periodically calibrated against international isotope standards (IAEA N1, N3; IAEA CH7, NBS22; D. Harris, personal communication, 2006). Analytical errors varied from 0.03 to 0.41{per thousand} for {delta}13C and from 0.03 to 0.41{per thousand} for {delta}15N. Mean standard errors of replicate measurements varied from 0.21 to 0.34{per thousand} for {delta}13C and from 0.19 to 1.24{per thousand} for {delta}15N.

Carbon/nitrogen ratios on residues of soil samples after the muffle furnace experiment were also calculated from C and N content in stable isotope samples. The mean standard error of replicate C/N ratios was 1.63 (n = 4).

Data Analysis
Statistical analyses of the results from the muffle furnace experiment, upland watershed study, and riparian zone study were performed using SAS (SAS Institute, 2005) and Statistica for Windows (StatSoft, 2001). {delta}13C or {delta}15N values for samples with too little C or N content (i.e., <100 µg C or <10 µg N, respectively) were deemed unreliable and were not included in analyses.

Muffle Furnace Experiment
Sample replicates within each temperature and duration were averaged for analyses, with site and soil horizon used as the replicates through blocking for all analyses of variance (ANOVA) (Steel and Torrie, 1980; Hurlbert, 1984). The main blocking factor was site with soil horizon nested within site because both soil horizons were sampled from the same locations.

The effects of site (riparian and forest), soil layer (O and A horizons), muffle furnace temperature (20, 100, 200, and 400°C), and duration of burn (1, 5, 15, and 60 min) on {delta}13C and {delta}15N values and C/N ratios were tested using a randomized complete block 4 x 4 factorial design (Steel and Torrie, 1980). The ANOVA model had mixed effects because blocking effects were random, whereas both temperature and duration treatments were fixed effects. Site and horizon were the blocking factors of the ANOVA, with horizon nested within site, whereas the 4 x 4 factorial design corresponded to a three-way analysis (Steel and Torrie, 1980). Pooled mean squared error (MSE) terms from interactions between blocking and treatment effects served as the error term for treatment and site effects, while the error term for the site effect was obtained from horizon nested within site. {delta}13C values and C/N ratios for the 400°C, 60 min treatment were not reliable for the forest A horizon because of too little C content, so a mean substitution was used with the average of measurements from all other forest A horizon treatment combinations to replace the one missing value. The 600°C treatment produced no reliable {delta}13C or {delta}15N values for many treatment combinations, so it was not possible to fully cross the design with the duration factor. Thus, contributions of the 600°C burn for durations of 1 and 5 min were analyzed with all treatments at other temperatures (i.e., transforming the blocked two-factor design into a blocked one-factor design) using a simple two-way ANOVA for randomized complete block designs (Steel and Torrie, 1980). In both designs, the 20°C factor represented the control (no burning). Different control samples were randomly assigned to each level of the duration factor to ensure that the designs were fully crossed. To test the critical assumption of homogeneous variance for the ANOVA tests (Steel and Torrie, 1980; Sokal and Rohlf, 1981), the ratio of the minimum to maximum variances for all treatment combinations were examined and were all determined to be <2 (see Results section). Furthermore, tests for normality using the Kolmogorov-Smirnov approach yielded nonsignificance for all treatments. Thus, no transformations of data were performed.

This was not a ‘loss-on-ignition’ experiment in which the organic matter content of surface soils can be determined (Nelson and Sommers, 1996); only the amount of material volatilized was determined. The percentage of C volatilized in each sample during the muffle furnace experiment was calculated using the following equation:

Formula 2[2]
where MC,i is the mass in grams of C in the initial unburned material and MC,r is the mass of C in the burned residual material. MC,i and MC,r were calculated with Eq. [3] and [4], respectively:

Formula 3[3]

Formula 4[4]
where Mi is the total material in a particular sample before burning, Mr is the total material remaining after the sample was burned, fC,control is the mean proportion of C in the three replicate control (i.e., unburned) samples for the corresponding soil type, and fC is the proportion of C in the burned material calculated as the amount of C measured during stable isotope analysis divided by the weight of the isotope sample analyzed. The same approach was used to calculate the percentage of N volatilized. The calculated % volatilizations were analyzed against the difference of the unburned in {delta}13C or {delta}15N values minus the corresponding value of the burned sample.

Upland Watershed Study
The t tests were performed on the control (i.e., unburned) vs. wildfire {delta}13C and {delta}15N values to determine if there were differences between the two types of plots on each date. To examine the effects of the wildfire on {delta}13C and {delta}15N values, each postburn date was analyzed using a generalized linear models (GLM) procedure in SAS, with additional analyses that included a repeated measures test for time, and t tests between the prewildfire data and the immediate and 1 yr postburn data.

Riparian Zone Study
Of the 17 point locations sampled from the riparian zone, only one site (Sample Point 1) was outside the depositional area of the ash flow. This site served as the control and no ash/sediment material was collected because it was located out of the depositional area. On the October 2002 visit, the ash/sediment at Site 6 was of insufficient quantity for analysis, and no ash/sediment was found at Site 3 on the June 2003 sampling. Furthermore, insufficient N content in the A horizon sample from Site 2 in June 2003 rendered that sample's {delta}15N value unreliable, but the {delta}13C value was included in analysis of the results.

A correlation analysis was performed using SAS to determine if there was a correlation between {delta}13C and {delta}15N values. Because there was evidence of non-normality, the nonparametric Spearman rho (for analysis based on differences between ranked observations; Sokal and Rohlf, 1981) and Kendall tau (for analysis based on differences between sums of ranked observations; Sokal and Rohlf, 1981) correlations coefficients were used. A mixed procedure (Proc MIXED) with a repeated measures analysis and the heterogeneous compound symmetry option was used to determine if date (8 Oct. 2002 or 26 June 2003), point location (Points 1through 17), and/or sample type (O or A horizon) had an effect on the {delta}13C and {delta}15N values.


   RESULTS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Muffle Furnace Experiment
The {delta}13C and {delta}15N values and C/N ratios were not significantly different between sites, but values were different (p < 0.001) for soil horizon nested within site for all three measures (Table 1). In particular, the O horizons had lower {delta}13C and {delta}15N values than the A horizons, but the lower value for the {delta}13C values was mainly attributed to the average riparian O horizon measurements (Table 2).


View this table:
[in this window]
[in a new window]

  		Table 1. ANOVA results for the randomized complete block factorial design for muffle furnace experiment. Site and soil horizon are random effects with horizon nested within site (i.e., the nested term is the error for the test of site). Temperature and duration are fixed effects.

 

View this table:
[in this window]
[in a new window]

  		Table 2. Means (± 95% confidence intervals) for riparian vs. forest, and O vs. A horizon {delta}13C, {delta}15N, and C/N ratios for muffle furnace experiment. n is the number of observations compared, which includes controls and all treatments with measurable {delta}13C or {delta}15N values. Because soil horizon was nested in site, the confidence intervals for horizon comparisons should not be used to detect a significant effect between sites.

 
The separate effects of temperature (Table 1; Fig. 2) and duration (Table 1; Fig. 3) were significant for both {delta}15N and C/N ratios (p < 0.001 to p < 0.05). All soil layers responded to the 400°C temperature treatments, with higher {delta}13C and {delta}15N values (Fig. 2) than at all other temperatures. The overall effect of the burn treatments was much greater for {delta}15N than for {delta}13C. For example, the increase in stable isotope values between 200 and 400°C (Fig. 2) was much greater for {delta}15N (>1{per thousand}) than {delta}13C (~0.4{per thousand}), which indicated that a higher percentage of the lighter N (14N) was lost relative to the heavier N (15N). However, total C loss was higher than N loss, as C/N ratios were also lower at the 400°C treatments than at all other temperature treatments (p < 0.001). {delta}15N values significantly increased as burn duration increased except between the 1- and 5-min durations (p < 0.001). Only the 60-min duration resulted in significantly lower C/N ratios from the 1-min duration as indicated by the overall test of duration (other contrasts were not significant and the 1 and 15 min comparison was not a preplanned contrast), but a decreasing trend with increasing duration was clearly evident (Fig. 3).


Figure 2
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 2. Average (with 95% confidence intervals) (a) {delta}13C values, (b) {delta}15N values, and (c) C/N ratios from muffle furnace experiment for muffle burn temperatures using factorial data from all durations. The error term for the test of temperature was the mean expected squared error from the factorial component of the ANOVA, which used sites and soil layers as replicates. Three pre-planned contrasts were performed to compare means while respecting the overall experiment-wide alpha error rate of 0.05: 20 vs. 100, 200, and 400°C; 100 vs. 200°C, and 200 vs. 400°C. Different letters under the lower confidence interval apply only to preplanned contrasts and indicate that means are significantly different at alpha = 0.05. Different letters were also assigned to the two most different means if the overall test was significant, but preplanned contrasts were all nonsignificant.

 

Figure 3
View larger version (12K):
[in this window]
[in a new window]

  		Fig. 3. Average (with 95% confidence intervals) (a) {delta}13C values, (b) {delta}15N values, and (c) C/N ratios from muffle furnace experiment for muffle burn duration using factorial data from all temperatures. The error term for the test of duration was the mean expected squared error from the factorial component of the ANOVA, which used sites and soil layers as replicates. Three preplanned contrasts were performed to compare means while respecting the overall experiment-wide alpha error rate of 0.05: 1 vs. 5 min, 5 vs. 15 min, and 15 vs. 60 min. Different letters under the lower confidence interval apply only to preplanned contrasts and indicate that means are significantly different at alpha = 0.05. Different letters were also assigned to the two most different means if the overall test was significant, but preplanned contrasts were all nonsignificant. Letters are not shown for {delta}15N as the overall effects were not significant.

 
Interactions between temperature and duration were not significant for {delta}13C values (Table 1), but they were significant for {delta}15N values and C/N ratios at all interactions at 400°C (Table 1; Fig. 4). Although interactions are significant and therefore main effects are less relevant (Steel and Torrie, 1980), main effects for {delta}15N and C/N are shown because the interactions in Fig. 4 masked the effect of duration for these variables due to averaging the values. Furthermore, longer durations resulted in significantly higher {delta}15N values at 400°C. Duration effects on C/N ratios at 400°C were similar for the 1- and 5-min durations, but much lower for the 15- and 60-min durations. Although measurements at 600°C were not included in the fully crossed factorial analysis, the two-way ANOVA with results for durations of 1 and 5 min at 600°C indicated that the patterns observed in the three-way ANOVA for the 400°C treatments at 5 and 15 min were respectively similar at 600°C for 1 and 5 min (Table 3; degrees of freedom [DF] for all tests = 17; error DF = 51; {delta}13C: MSEnum = 0.14, MSEden = 0.16, p = 0.623; {delta}15N: MSEnum = 1.94, MSEden = 0.08, p < 0.001; C/N: MSEnum = 109.26, MSEden = 17.66, p < 0.001).


Figure 4
View larger version (24K):
[in this window]
[in a new window]

  		Fig. 4. Average (with 95% confidence intervals) (a) {delta}15N values and (b) C/N ratios from muffle furnace experiment for the interaction of muffle burn temperature and duration. The error term for the test of the interaction was the mean expected squared error from the factorial component of the ANOVA, which used sites and soil layers as replicates. Eight preplanned contrasts were performed to compare interaction means while respecting the overall experiment-wide alpha error rate of 0.05: 20 vs. 100, 200, and 400°C at each of 1, 5, 15, and 60 min (four separate contrasts, one per duration); and 200 vs. 400°C at each of 1, 5, 15, and 60 min (four separate contrasts, one per duration). Different letters under the lower confidence interval apply only to preplanned contrasts and indicate that means are different among temperatures whereas different numbers adjacent to letters (e.g., "a1" vs. "a2") indicate different means among duration between 200 vs. 400°C.

 

View this table:
[in this window]
[in a new window]

  		Table 3. Means and standard deviations (n = 4) for {delta}13C, {delta}15N, and C/N ratio for 400°C at 5 and 15 min and 600°C at 1 and 5 min.

 
Examination of the changes in {delta}13C and {delta}15N values of the burned vs. unburned material as a function of the percentage volatilization of C or N indicated that regression slopes were significantly different than zero (p < 0.05; Table 4) for changes in both {delta}13C and {delta}15N values in the O and A horizons. Slopes were negative for changes in {delta}15N in both O and A horizons and for changes in {delta}13C in the A horizon, indicating a loss of the lighter isotope with increasing percentage volatilization. There was a slight positive slope for {delta}13C in the O horizon that only ranged about 1{per thousand}. A horizon results showed stronger trends in the difference in {delta}13C and {delta}15N than O horizon results, as indicated by the r2 values (Table 4).


View this table:
[in this window]
[in a new window]

  		Table 4. Regression analysis results of comparisons between changes in {delta}13C and {delta}15N between burned and unburned O or A horizon material as a function of the percentage volatilization of C or N. n is the number of observations included in the analysis.

 
Upland Watershed Study
The t tests performed on the stable isotope values of the samples from control vs. burn sites for each date were significant for {delta}15N at {alpha} = 0.10 on all three dates (preburn, p = 0.072; immediate postburn, p = 0.087; 1 yr postburn, p = 0.046), but not for {delta}13C on any of the dates. Thus, the {delta}15N values for the unburned area were statistically significantly different at {alpha} = 0.10 from the area that was subsequently burned regardless of whether or not the wildfire had occurred (Fig. 5).


Figure 5
View larger version (11K):
[in this window]
[in a new window]

  		Fig. 5. Comparison of average stable (a) carbon and (b) nitrogen isotopes values for fire zone samples before the burn, immediately after the burn, and 1 yr after the burn.

 
The Proc GLM analysis revealed that there was only an effect of the burn on {delta}15N values 1 yr postburn in the positive direction (p = 0.033; {alpha} = 0.05 = 0.10 adjusted for two comparisons done on same dataset). The repeated measures test for time and paired t tests between the preburn data and the immediate and 1 yr postburn data did not indicate any significant differences for {delta}13C or {delta}15N values, except for immediate postburn {delta}13C control values (p = 0.100 for paired t test).

Riparian Zone Study
Results indicated that the {delta}13C values were relatively constant throughout the study (Fig. 6), and did not differ between ash/sediment and A horizon samples. {delta}15N values, however, appeared to have a difference between ash/sediment and A horizon samples that broadened from the October 2002 to the June 2003 sampling, but this difference was not significant because of the large error bars.


Figure 6
View larger version (12K):
[in this window]
[in a new window]

  		Fig. 6. Comparison of (a) average stable carbon isotope values and (b) average stable nitrogen isotope values for riparian zone samples gathered on 8 Oct. 2002 and 26 June 2003 at site of runoff from Gondola fire. Error bars indicate standard deviations.

 
The results of the correlation analysis indicate that there are correlations between {delta}15N and {delta}13C according to the Kendall tau approach only (p = 0.045; N = 79). The results of the mixed procedure indicated that: (1) the {delta}13C values decreased about 0.71{per thousand} from 8 Oct. 2002 to 26 June 2003 (p = 0.015; {alpha} = 0.05 = 0.10 adjusted for two comparisons; 90% confidence interval ± 0.55{per thousand}), and (2) the {delta}13C values of the ash/sediment layer were ~0.65{per thousand} higher than those in the A horizon (p = 0.022; {alpha} = 0.05, adjusted for two comparisons; 90% confidence interval ± 0.54{per thousand}). The first result indicates a possible temporal effect, but because of the length of time between sampling, this interpretation is only speculative. The second result suggests a possible fire effect on {delta}13C values. Surprisingly, no significant relationship was found between the {delta}15N values and date, point location, and/or soil horizon despite the appearance of a reasonable difference by date and/or soil horizon (Fig. 6b). However, there was high variability in the {delta}15N values, so the differences were not statistically significant.


   DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
While overall {delta}13C values were lower at 100°C compared with control values (Fig. 2), closer examination of the soil types revealed that the {delta}13C values in the riparian A horizon increased at all durations for this temperature. The higher percentage of C loss as compared with N loss between 200 and 400°C (Fig. 2) is consistent with the temperature dependence of nutrient loss to the atmosphere, where N loss due to volatilization begins at 200°C, with over 50% of N in the O horizon volatilized at temperatures >500°C (Neary et al., 1999). The corresponding pattern in higher {delta}13C and {delta}15N values, especially at 400°C or higher in our study indicates that the lighter isotope is preferentially volatilizing (Fig. 2; Table 3).

Furthermore, higher durations resulted in significantly higher {delta}15N values. Consideration of the 1- and 5-min duration at 600°C supported this pattern and indicated that the increase in {delta}15N values between 1 and 5 min for the 600°C treatments were similar to {delta}15N increases for the 400°C treatments between 5 and 15 min (Table 3). This could mean that a burn duration of 1 min at 600°C is approximately equivalent to a burn duration of 5 min at 400°C in terms of volatilization of the 15N isotope.

While {delta}13C and {delta}15N in the A horizon showed trends of higher values in the burned residual material vs. the unburned material as a function of the percentage C or N volatilization, respectively (Table 4), the scale of isotopic change was much greater for {delta}15N (~0.03{per thousand} per %N volatilized vs. ~0.01{per thousand} per %C volatilized). An explanation of the differences in tendency of the C and N isotopes to be fractionated during combustion may lie in the potential for low molecular weight intermediates to be produced (Tillman et al., 1981) and then fractionated by further volatilization. Most total N and C in plant litter and soil organic matter is in the form of rigid nonvolatile macromolecules such as cellulose, proteins, or humic acids (Paul and Clark, 1996). The initial stage of combustion of this material is solid particle pyrolysis in which low molecular weight volatile species (e.g., alcohols) are given off and convected into the oxidizing portion of the flame (Tillman et al., 1981). A low molecular weight compound is more likely to be subject to fractionation caused by the presence of a heavy isotope because the proportional difference in isotopic mass would be expected to be of greater importance in smaller molecules. It is possible that the residual N that was not volatilized during the muffle furnace experiment included low molecular weight species such as amides or amines (Chiavarai and Galleti, 1992) produced from solid particle pyrolysis that had been subject to fractionation during a second stage of volatilization (Tillman et al., 1981). If so, this N could be characterized by a high isotopic ratio after fire, not only in soil, but also in runoff nutrients that could be taken up by vegetation and the aquatic food web.

A second process that might account for fractionation of {delta}13C and {delta}15N is the preferential combustion of compounds that are already isotopically enriched or depleted. For example, lignin {delta}13C values are typically depleted by about 3 to 6{per thousand} relative to whole plant tissue {delta}13C values due to processes believed to occur during biosynthesis of lignin (Benner et al., 1987; Schmidt and Gleixner, 1998). A selective combustion of lignin-poor tissues might result in apparent isotope fractionation. In addition, charring may involve selective preservation of aromatic rings originating from lignin (Tillman et al., 1981), which could result in less 13C in residue left behind in aromatic-rich char. This might explain the tendency of the forest O horizon material to have a slightly depleted {delta}13C value. However, the small changes in {delta}13C make it unlikely that it would be a useful tracer for the effects of fire.

Results from the upland watershed and riparian zone studies showed no significant patterns in {delta}13C, {delta}15N, or C/N ratios. This is likely due to relatively low soil temperatures during the fire. Although temperature was not measured because this was an unforeseen wildfire, the lack of soil C loss (Murphy et al., 2006) indicates that soil temperatures were probably low. Soil temperatures in wildfires are highly variable, with typical maximum ground temperatures ranging between 200 and 300°C, although instantaneous temperatures greater than 1500°C in the soil surface can occur (Neary et al., 1999). While the fire may have burned quite hot and rapid in the tree crowns, it may not have been so hot at the actual ground surface, especially since the fuel loading in the burned area was naturally less than in the unburned area. In addition, the detection of the differences in {delta}15N values in the more controlled muffle furnace experiments may have been overwhelmed in the field by plot-specific differences in isotope values caused by differences in vegetation, microbial activity, or degree of decomposition, all of which are known to result in localized differences in isotope values (e.g., Quideau et al., 2003).

Examination of mean values of the riparian zone stable isotope values indicated an apparent change in the {delta}15N values between sampling dates and soil horizons (i.e., ash/sediment vs. O horizon), suggesting a temporal effect. However, the differences were not statistically significant for {delta}15N because of the high variability of the values. Because of restricted access, initial sampling of the riparian zone occurred over 2 mo after the wildfire, and the subsequent sampling date was 7 mo later following the ensuing winter. Thus, it is difficult to determine if the observed temporal changes in field values could be specifically attributed to the effects of fire or are simply due to annual variability of variables such as rainfall or snowpack duration.

Overall, these preliminary findings are intriguing because if it is possible that certain soil horizons and burn intensities do exhibit significant shifts in {delta}15N values, stable N isotope analysis may well be a useful means of tracing impacts of fire on water quality and the aquatic food web, especially if runoff events deposit recently burned soils into nearby streams. In this study, stable isotope analyses of the aquatic food web were not possible because of the ephemeral nature of the downstream water body and the lack of prefire aquatic food web stable isotope data. However, Spencer et al. (2003) found higher {delta}15N and lower {delta}13C values in fish and macroinvertebrates following a large wildfire in Montana. Thus, stable isotope analysis could also be useful for monitoring the recovery of aquatic ecosystems from these impacts as the food web values return to their preburn values. Moreover, there are indications that both burn duration and temperature (independently and interactively) affect the degree of change in isotope values, which could further emphasize the utility of this approach since sediment yields from more severely burned sites in which the entire O horizon is consumed (Robichaud et al., 2000) tend to be much greater than yields from sites with low severity burns (Benavides-Solorio and MacDonald, 2001). Burn duration and temperature also may affect which soil types and which horizons will exhibit changes in isotopic value. That is, soil moisture retention, soil texture, parent material, and organic matter will vary by soil and horizon and could well affect responses (Neary et al., 1999).

{delta}13C and {delta}15N will only have value as tracers if fire results in larger isotopic changes than the existing variability at the base of the aquatic food web. It is difficult to quantify ‘natural variability’ of {delta}13C and {delta}15N values in fresh water because the source of dissolved CO2 varies and will affect localized {delta}13C values (Peterson and Fry, 1987), and {delta}15N values can depend on source signatures in precipitation and soils (Heaton, 1986). Furthermore, the incorporation of the isotopes in the aquatic food web can be affected by factors such as water velocity (Finlay et al., 1999), denitrification (Heaton, 1986), and nitrogen fixation (Peterson and Fry, 1987). This can lead to high variability of {delta}13C and {delta}15N values in primary producers. For example, variability of both {delta}13C and {delta}15N values for periphyton on the Truckee River in Nevada range up to more than 3{per thousand} at some sites (L. Saito, unpublished data, 2005). In this study, the maximum increases in {delta}15N values for all soil types occurred at 400°C for the 60-min duration and were on the order of 2 to 3{per thousand}. The greatest increase in {delta}13C (1.50{per thousand}) also occurred at this temperature and duration for the riparian A horizon material. Results indicated that burn temperatures of 200°C or less may not produce strong enough shifts in surface soil stable isotope values for tracing the impacts of fire, but at temperatures that are too high (i.e., 600°C or more), stable isotopes may not be useful because the N and C content of residual material is too low. Thus, stable isotopes are most likely to detect changes occurring when temperatures range between 200 and 400°C, although experiments in the 400 to 600°C range would be useful to determine the threshold above which isotopes are no longer useful indicators. Further investigation is needed to determine if greater changes in {delta}13C and {delta}15N values may occur at discrete temperatures and durations that were not explored in this study.


   ACKNOWLEDGMENTS
 
This study was supported by the Nevada Agricultural Experiment Station, College of Agriculture, Biotechnology, and Natural Resources, University of Nevada Reno, publication number 52055531, and the US Forest Service, Lake Tahoe Basin Management Unit. We greatly appreciate the assistance of Dr. Robert Blank in providing laboratory space, equipment, and assistance during the project, and the UNR Statistical Consulting Unit for providing statistical assistance. The authors thank three anonymous reviewers for their thorough and careful reviews of the manuscript.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
This study was supported by the Nevada Agricultural Experiment Station, College of Agriculture, Biotechnology, and Natural Resources, University of Nevada, Reno, publication number 52055531, and the US Forest Service, Lake Tahoe Basin Management Unit.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow An erratum has been published
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Saito, L.
Right arrow Articles by Szameitat, P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Saito, L.
Right arrow Articles by Szameitat, P.
Agricola
Right arrow Articles by Saito, L.
Right arrow Articles by Szameitat, P.
Related Collections
Right arrow Fire
Right arrow Isotopes

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
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome