| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Unit of Forest Nutrition and Water Resources, Dep. of Ecology, Technical Univ. of Munich, Am Hochanger 13, D-85354 Freising, Germany
b Jr., Dep. of Forest Science, 321 Richardson Hall, Oregon State Univ., Corvallis, OR 97331
c 3342A Kauhana Pl., Honolulu, HI 96816
d Faculty of Forestry, Univ. of Istanbul, Soil Science and Ecology Dep., 80895 Bahcekoy, Istanbul, Turkey
e Dep. of Forest Resources and Environmental Sciences, Korea Univ., Seoul 136-701, Korea
* Corresponding author (kermit.cromack{at}orst.edu)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: DF, Douglas-fir • MDD, minimum detectable difference • RA/DF, a mixed stand of red alder and Douglas-fir • RA, red alder
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In 1979, a long-term study with randomized, replicated plots was established in a 9-yr-old coastal Oregon DF plantation with admixed volunteer RA. The following year, intensive soil investigations were performed (Cromack et al., 1999). In 1999, we resampled a portion of the same plots having either pure DF or RA/DF to address the following questions under controlled experimental conditions: (i) Did soil C and N pools change within the last 19 yr? (ii) How much soil C and N was accumulated by the admixture of RA? (iii) Did the admixture of RA influence resin N availability and DF foliar N concentration? (iv) How do temporal and spatial methods compare in the same forest soil experiment?
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1 m deep, over saprolite of weakly cemented sandstone. Soil series include Slickrock (medial over loamy, ferrihydritic over isotic, mesic Alic Fulvudand) on the steeper slopes and Bohannon (fine-loamy, isotic, mesic Andic Haplumbrept) on the ridges (Corliss, 1973; Cromack et al., 1999). These coastal sandstone soils have Andic soil properties, with low bulk densities and high organic C concentrations (Homann et al., 1995; Cromack et al., 1999). After clearcutting and broadcast burning a 130-yr-old DF and western hemlock [Tsuga heterophylla (Raf.) Sarg.] stand, the site was replanted in 1971 with DF seedlings. Volunteer RA was partially controlled by herbicide treatment. In 1979, a total of eighteen 0.3-ha square (55 by 55 m) plots were randomly selected on the 22-ha research site to study the effects of vegetation control on tree growth and survival. Fifteen of the 18 plots were thinned in 1980 to retain 745 DF ha-1 and 0 to 190 RA ha-1, with three unthinned controls. Six thinned plots had DF and no RA, with three each having 745 DF ha-1 and either 54, 94, or 190 RA ha-1. Three pure DF plots were fertilized with 22 g N m-2 as urea in 1986 (Miller et al., 1999), which left three unfertilized DF plots available as a treatment comparison in 1999. The experiment is described in detail, together with silvicultural results through 1996, by Miller and Obermeyer (1996) and Miller et al. (1999). An assumption being made in our study is that initial site conditions for the different treatments were identical, so that treatments could be randomly assigned. In designing the silvicultural study, one plot having no alder was randomly assigned to retain 54 RA ha-1, and necessitated a treatment switch with another plot to try to accommodate a completely randomized design (Miller and Obermeyer, 1996). In June 1980, the forest floor and mineral soil were sampled on all 18 plots (Cromack et al., 1999). Nine randomly located subsamples within each plot were combined into a composite sample. Forest floor was removed using a 25 by 25 cm sampling frame, and mineral soil was sampled with two integrated 7.5-cm diam. cores per each 15-cm depth interval with a sliding hammer bulk density sampler (Blake and Hartge, 1986). Total soil N was measured after using the micro-Kjeldahl digestion method (Bremner, 1996), measuring NH4–N with a Technicon Autoanalyzer (Technicon Industrial Systems, Tarrytown, NY). Total soil C was measured by dry combustion with a LECO WR-12 C analyzer (LECO Corp., St. Joseph, MI) (Nelson and Sommers, 1996).
In June, 1999, six of the 18 plots were selected for resampling. For the pure DF treatment, we selected the three unfertilized DF plots. Of nine thinned plots having both RA and DF (RA/DF), we selected the three RA/DF plots that had been thinned to retain the highest density of RA (190 ha-1). After thinning in 1980, the pure DF plots had a mean density of 745 DF ha-1, while the three RA/DF plots were thinned to 745 DF ha-1 and 190 RA ha-1, or 935 trees ha-1. Initial tree density was 25.5% greater on the RA/DF plots in 1980. We used the original 1980 soil data for the six specific plots utilized for resampling in 1999, and not the entire 1980 data set for all 18 plots as a basis for our comparisons. Red alder comprised about 25% of the total basal area on the three mixed plots having 190 RA ha-1 (Miller et al., 1999). By 1996, about 27% of the 745 DF trees ha-1 on the mixed plots had died, with >90% of the RA remaining; survival of trees on the DF plots was about 94% (Miller et al., 1999). Thus, by 1996 the pure DF plots had a mean density of 700 trees ha-1, and the RA/DF plots had a mean density of 543 DF trees ha-1 and 171 RA trees ha-1, or 714 trees ha-1. In 1996, the difference in tree density had decreased to about 2.1%, showing treatment effects on stand dynamics (Oliver and Larson, 1996). Our study was designed to minimize a priori soils effects, except that the RA was not planted, but only thinned, and could have responded to such effects when colonizing the site initially.
On each of the six plots, five forest floor and mineral soil subsamples were taken and analyzed individually. The five sample points in each plot were selected systematically along one of the two 77-m plot diagonal transects, which were first selected at random in each plot. For the organic forest floor material, we used the same 25 by 25 cm sampling frame. Mineral soil was sampled to a 45-cm depth with a steel soil corer, which allowed us to take complete soil cores in polyethylene liners of 6-cm diam. In the laboratory, the 1999 soil cores were cut into 15 cm lengths to obtain the same sampling depth intervals as was done in 1980, when two integrated 7.5 cm cores per each 15-cm depth were used. We determined bulk density by weighing the whole sample and drying subsamples at 45°C (forest floor) and at 105°C (mineral soils). After determination of bulk density, soils were sieved with a 2-mm sieve, homogenized, and dried to constant weight at 40°C. Subsamples were ground and analyzed for total soil C and N by dry combustion with a LECO CHN analyzer (LECO Corp., St. Joseph, MI) (Bremner, 1996; Nelson and Sommers, 1996). Mean within plot values for forest floor and soil C and N, and soil bulk density, were used as a basis for mass C and N calculations.
Soil C and N pools were calculated only for the fine-earth soil fraction (<2 mm), using the same adjustments made by Cromack et al. (1999) for the coarse soil fractions (>2 mm), including gravel and rock fragments, which further assumes that cobble-size rock on the plots occupies 7.5% of the soil volume in the top 45-cm depth. In some cases in 1999, the lower part of the core was missing. To have comparable soil mass estimates for both sampling dates, we used the bulk density calculated with the measured thickness to extrapolate soil mass for a 15 cm layer. Since recent work has shown micro-Kjeldahl digestion to underestimate total N concentrations relative to LECO combustion, we increased the original 1980 forest floor N concentrations by 5% (Prietzel et al., 1997) and by 6% for mineral soil N concentrations for the six specific 1980 plot data sets used for our current comparisons. The 6% correction for mineral soil N is a mean value, based upon the 4% correction estimated by Johnson and Todd (1998) and an 8% correction from Knoepp and Swank (1997).
We used ion-exchange resin bags as a field index method to assess N availability in the soil (Binkley et al., 1986). Individual resin bags were placed beneath the forest floor (approximately 5 cm from the surface) at 1-m intervals in a randomly chosen diagonal transect through each plot, with a total of five bags per transect. The resin bags included 14 mL of anion (Sybron IONAC ASB-IPOH, Sybron International, Milwaukee, WI) and 14 mL of cation (Sybron IONAC c-251 H+) exchange resins in separate compartments in a nylon stocking. After an incubation period of 14 wk, the resin bags were retrieved, brushed to remove adhering soil, air-dried, and extracted with 100 mL of 2 M KCl. The extracts were analyzed colorimetrically for NH4–N and NO3–N (Alpkem Continuous Flow Autoanalyzer, O-I Analytical, College Station, TX) (Mulvaney, 1996). Results were expressed as milligrams of N per bag (NH4, NO3, or total N) on an air-dry weight basis.
Douglas-fir tree foliar N was investigated using a neighborhood design. We selected 20 DF trees in the upper part of the plantation and estimated the proportion of RA by measuring the basal area of all trees within an 8-m circle surrounding the sample tree. Needle samples were taken from mid-crown using a shotgun, and only mature needles from the previous year were used. After drying the needles at 45°C to a constant weight, we measured the concentration of total N using a LECO CHN-Analyzer (Bremner, 1996). Results were expressed as milligrams of N per gram of dry weight.
Statistical analyses were performed using SPSS 9.01 software (SPSS Inc., Chicago, IL). The method of paired observations was used to compare C, N, and soil bulk density results temporally for the three RA/DF plots and for the three DF plots, using paired t-tests (Steel et al., 1997), while the ANOVA procedure was used for the comparison of the two different treatments (k = 2), RA/DF and DF (Steel et al., 1997; Zar, 1999). Linear regression with Pearson correlation coefficients was used to analyze the influence of RA on DF foliar N (Steel et al., 1997). We used a log transformation of resin means to test for treatment effects in total mineral N content of resin bags, and, separately for differences in NH4–N and NO3–N content. The 95% confidence interval was calculated for the mean differences in soil variables studied for each depth (Steel et al., 1997). Results were considered significant for P < 0.05. Trends were reported for P < 0.10, since Steel et al. (1997) suggest using 
= 0.10 for field experiments having smaller numbers of replicates for each treatment.
Minimum detectable differences, detectable 90% of the time (Type II error, ß = 0.10) at P < 0.05 were calculated for the forest floor and total soil C and N pools to a 45-cm depth for temporal changes in each treatment (RA/DF and DF), using the variance of differences in plot pairs from a paired t-test (Zar, 1999). For comparisons between the two spatial treatments (DF and RA/DF), MDD values were calculated based upon the common variance derived from a two sample t-test (Zar, 1999). The MDD values calculated for both temporal and spatial changes are more conservative than 95% confidence intervals, since both Type I ( = 0.05) and Type II (ß = 0.10) errors are minimized, using a two-tailed t = 0.05 + tß = 0.10, with appropriate degrees of freedom, for each MDD value (Zar, 1999).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
Calculation of MDD for temporal treatment changes (Table 2) shows that the MDD for total C to a 45-cm depth in the RA/DF plots, 5737 g C m-2, is 2.6 times the mean temporal difference, 2232 g m-2. Temporal comparisons, with paired t-tests, use the variance of paired plot differences (Steel et al., 1997), and thus reduce MDDs for temporal differences compared with MDDs between the two spatial treatments (Zar, 1999). The MDD for total N under RA/DF is 148 g N m-2, which is 22% less than the mean increase in N, 190 g N m-2. If one assumes that the total soil C increase under RA/DF represents a real gain of 117 g C m-2 yr-1, then approximately 30 more years of soil C accretion at this annual rate would be necessary to detect a significant gain in soil C 90% of the time at P < 0.05. A less conservative estimate, using the 95% confidence interval, suggests an additional time interval of about 15 yr. Calculation of MDD estimates for differences between RA/DF and DF treatments in 1999 shows MDDs for total soil C and N of 9226 g C m-2 and 442 g N m-2, respectively, which are much larger than those for within treatment comparisons (Table 2).
The admixture of RA increased ion-exchange resin total N mineralization significantly (P < 0.05), and is shown in Fig. 1c . Total resin mineral N content was the same magnitude for most of the individual resin bags from the RA/DF and DF plots. However, about one-third of the RA/DF resin bags had 10 to 50 times higher mineral N content compared with the average N content of the DF bags. This spatial pattern with a very high resin N availability at certain hot spots on the RA/DF plots also explains the high standard errors. The mean resin NH4–N content in RA/DF plots, 1.78 mg bag-1 was significantly larger (P < 0.05) than the mean of 0.11 mg bag-1 in the DF plots (Fig. 1a). The mean resin NO3–N content under RA/DF, 1.44 mg bag-1, also was greater than the mean NO3–N content, 0.22 mg bag-1, under DF, but the difference was only a trend (P < 0.10), because of the high variability in NO3–N content (Fig. 1b). The proportion of NO3–N in total N was 66% for DF and 45% for RA/DF.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Even if treatments are significantly different, the exact rates fluctuate within a wide range. For example, to determine a 20% change of the soil N pool on the RA/DF plots with a 95% certainty, about five replicate plots would be necessary, using a two-stage set of calculations to estimate the required sample size, and the variance from the original set of three RA/DF plots in 1980 (Steel et al., 1997). About 12 yr would have been required to detect a 20% difference with five plots, if the N-accretion rate and initial site conditions were the same as in our experimental plantation. This contrasts with the 19 yr it took to observe a 32% increase in soil N in the mixed plots (Table 1), with three sample plots, and at P < 0.02. These examples show the challenges of investigating soil pools in this variable landscape, even with well designed experiments. In one such experiment, a significant change in soil C mass was not evident from sugar cane (Saccharum officinarum L.) (C4) conversion to eucalypt forest (C3), but 12C/13C stable isotope analysis for soil C showed that the trees had replaced a significant portion of the original soil C from sugar cane (Binkley and Resh, 1999). Fractionation of soil to augment whole soil analyses, to obtain isotopic changes in light and heavy density fractions for 13C or 15N isotopes, has been done in other studies (Connin et al., 1997; Swanston and Myrold, 1997; Compton and Boone, 2000) to help interpret mechanisms contributing to changes in soil C and N. Thus, a combination of techniques, in addition to having an appropriate experimental design, may provide better insights into soil C and N quality and quantity changes under a variety of experimental conditions.
At the Waldport site, the admixture of RA/DF had a 13% greater N accumulation than the difference between RA/DF and DF. How leaching loss differences between these two treatments could have affected the treatment differences is unknown. Significant leaching losses have been reported for mixed RA/DF plantations (Van Miegroet and Cole, 1984; Binkley et al., 1992). Binkley et al. (1992) reported NO3–N losses of 5 g m-2 yr-1, but minimal denitrification, while another study indicated a potential for greater denitrification under RA with increased NO3 availability (Griffiths et al., 1998). Our resin data show a six-fold greater trend (P < 0.10) in NO3 availability in the RA/DF plots, thus increasing the potential for NO3 leaching. If NO3 leaching losses were to approach the 5 g N m-2 estimated by Binkley et al. (1992), then net soil N accumulation rates would be about 33% lower than total N inputs to the RA/DF plots.
Combining the results obtained from the temporal and spatial approaches, a reasonable estimate for N accumulation at the Waldport site is about 9.3 g m-2 yr-1, which is an average of these two estimates. The plausibility of this estimate is confirmed by the fact that 58% of the N accumulation was found in the mineral soil surface to a 15-cm depth, where higher concentrations of tree fine roots and RA nodules should occur. The N accumulation rate at the Waldport site is at the higher end, but well within the range reported in the literature. Nitrogen-fixation rates of mixed RA/DF stands mostly were between 5 and 10 g m-2 yr-1 (Binkley et al., 1994).
The higher N accumulation rates at the Waldport site presumably result from the favorable site conditions there, at least for RA. Although the factors affecting RA N-fixation are not well known at present, environmental ones, such as adequate soil moisture and adequate nutrition, which promote growth, seem to favor N-fixation. High fixation rates have also been documented for fertile sites with high leaching losses (Binkley et al., 1994). The high retention of fixed N, even on fertile sites, appears to be because of a combination of abiotic and biotic retention mechanisms, with abiotic retention relatively more important on N-rich sites (Johnson et al., 2000). Leaf litter decomposition rate of an N-fixing species such as RA decreases after the first year, probably because of the increased N concentration (Edmonds, 1980; Cole et al., 1995; Berg, 2000), thus contributing to an increased mass of N-enriched forest floor and soil organic matter. Incorporation of both above and belowground litter components into soil by soil animals and microorganisms contributes to increased formation of stabilized soil organic matter, and incorporation of various forms of N into humus (Sollins et al., 1996; Drinkwater et al., 1998; Zang et al., 2000; Berg et al., 2001).
Nitrogen fixation not only leads to soil N accumulation, but also accelerates N cycling. Decomposition of N-rich litter from N-fixers increases availability of this nutrient. It has been documented repeatedly that the inclusion of RA in DF forests increases N mineralization and nitrification several-fold (Binkley et al., 1992; Hart et al., 1997). These increases were higher on some N-rich sites. Other factors, in addition to N, may limit decomposition rates on poor sites, such as availability of Ca and Mn (Berg, 2000), and P (Waring and Running, 1998). The nearly ten-fold increase in resin N at Waldport we observed supports previous work there showing high soil N mineralization potential (Cromack et al., 1999). The increase was caused by hot spots, where the small-scale combination of high N supply, an adapted microbial community, and favorable soil conditions led to very high mineralization rates on about 30% of the subplots. Even on the pure DF plots, nitrate accounted for 66% of resin mineralized N. This shows that nitrification plays an important role at the Waldport site. The inherent high N availability of this site presumably is the reason that the inclusion of RA had little influence on the nitrate/ammonium ratio. Other nutritional factors, such as the availability of P, S, base cations, and micronutrients, may also explain why the higher N availability in the mixed stand had no obvious effect on DF tree foliar N concentration, though there may have been effects on DF leaf area. In agreement with our study, recent work at the H.J. Andrews Experimental Forest in the Oregon Cascades found that RA had no significant effect on DF foliar N concentration (Rothe and Binkley, 2001).
From fertilization experiments, it is known that where high N availability exists, the addition of extra N may have minor effects or result in a relatively low uptake of the added N (Kenk and Fischer, 1988; Fisher and Binkley, 2000). At Waldport, N-fertilization (22 g N m-2) of three other pure DF plots resulted in no measurable increase in tree growth, presumably because of the high N availability at this site prior to N fertilization (Cromack et al., 1999; Miller et al., 1999). Future studies are needed to elucidate the effects of other soil nutrients, such as P, S, base cations, and micronutrients, together with research on soil faunal and soil microbial components at sites such as Waldport, on long-term ecosystem productivity, and on soil C and N changes (Coleman and Crossley, 1996; Fisher and Binkley, 2000). Red alder can influence availability and cycling rates of other nutrients, such as P and cations, in addition to N (Binkley et al., 1992; Giardina et al., 1995; Rothe and Binkley, 2001), but it is also competitive with DF (Miller et al., 1999).
There was much greater mortality and lower total stem volume production by DF in the higher density RA/DF plots at Waldport (Miller et al., 1999). Initial differences in mean tree density between the DF and RA/DF plots would have influenced stand development rate (Oliver and Larson, 1996), and the much greater DF tree mortality on the RA/DF plots would have increased detrital C and N inputs to the forest floor and mineral soil relative to the DF plots, especially for C. Thus, our finding that N storage is enhanced more than C storage, even with the likely greater total detrital inputs of C, indicates that the N accretion rate is relatively greater than the C accretion rate at this stage of ecosystem development, with only N accretion being statistically significant (P < 0.02) for all soil depths combined relative to C (Table 2). The lower proportion of RA in the mixed plots and its interaction with DF detrital inputs may create differential effects on C and N accretion, which could be studied with a variety of methods, such as soil density fractionation, use of stable isotope methods, or soil enzyme research, for example.
This study has shown the value of having an experimental design affording an opportunity for both temporal and spatial sampling comparisons. In the absence of a real time series, the comparison of a nonfixing with an N-fixing species has been used as a space for time substitution to identify the effect of N-fixers on soils, as discussed by Fisher (1995) and Binkley and Giardina (1998). This method is based on the assumption that initial site conditions for different treatments were identical, which may not always be the case. Temporal comparisons provided much lower MDD estimates for both soil C and N accretion than those calculated for between treatment spatial effects. In this study, the variance of paired plot differences over time was lower than the variance among spatial treatment plots (Steel et al., 1997), thus decreasing temporal MDD values. Use of MDDs is a valuable statistical tool for evaluating spatial and temporal effects in this study, and has provided useful comparisons in evaluating possible pretreatment and posttreatment effects in other forest soil C and N research (Homann et al., 2001).
This study has shown that substantial amounts of N can accrete in coastal Oregon forests having mixed RA and DF, in agreement with previous work (Binkley et al., 1992). The forest floor in the RA/DF plots showed a trend toward an increased C mass, and continuation of the current annual rate of soil C input would result in a significant C accumulation during an additional 15 to 30 yr. Resin mineral N results showed increased soil N availability (NH 4–N + NO3–N), with a trend toward increased resin NO3–N, in RA/DF plots. This suggests a potentially more important role for NO3–N leaching losses in these coastal Oregon forests, especially in mixed alder and conifer forests. Red alder will continue to function as an integral component of these forest ecosystems, and the management implications for both pure and mixed forests merit further research.
|
ACKNOWLEDGMENTS |
|---|
Received for publication February 27, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. Binkley, J. Kaye, M. Barry, and M. G. Ryan First-Rotation Changes in Soil Carbon and Nitrogen in a Eucalyptus Plantation in Hawaii Soil Sci. Soc. Am. J., September 1, 2004; 68(5): 1713 - 1719. [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 | |||
