HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Périé, C. Articles by Caron, J. Search for Related Content PubMed Articles by Périé, C. Articles by Caron, J. Agricola Articles by Périé, C. Articles by Caron, J. Related Collections Forest Soils Spatial Variability Ecosystem Management

Forest, Range & Wildland Soils

Use of Spectral Analysis to Detect Changes in Spatial Variability of Forest Floor Properties

^a Direction de la recherche forestière, Ministère des Ressources naturelles, de la faune 3700 rue Einstein, Sainte-Foy, QC G1P 3W8, Canada
^b Centre de recherche en biologie forestière, Faculté de foresterie et de géomatique, Pavillon Abitibi-Price, Université Laval, Sainte-Foy, QC G1K 7P4, Canada
^c Département des sols et de génie agroalimentaire, Pavillon Paul-Comtois, FSAA, Université Laval, Sainte-Foy, QC G1K 7P4, Canada

^* Corresponding author (Catherine.perie{at}mrnfp.gouv.qc.ca).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Understanding how silvicultural interventions affect soil spatial variability will improve our ability to predict forest ecosystem function in response to different degrees of management intensity. We demonstrate the use of spectral analysis, a geostatistical technique, to understand how management interventions affect soil spatial variability. The technique was applied to determine whether an intensive vegetation control treatment modifies the spatial patterns of soil microclimate and soil quality indicators: microbial biomass C (SMB-C) and N (SMB-N) as well as net N mineralization rate. A secondary objective was to investigate the contribution of soil microclimate factors to the explanation of spatial patterns of microbial biomass (C and N). Measurements were performed on transects laid out in two plots: a plot that was undisturbed since clearcut and replanted 11-yr earlier with white pine (Pinus strobus L.) and another plot which was subject to annual herbicide application during 4 yr following reforestation with white pine. Forest floor temperature (TEMP) and water content (WC), net N mineralization rate and SMB-C and SMB-N concentrations were measured every 25 cm along each transect. Spatial patterns were assessed using spectral analysis. In the two plots, microclimatic variables and SMB-N presented complex spatial patterns with several scales of spatial dependency, whereas SMB-C and net N mineralization did not demonstrate a spatial pattern at this scale of observation. In the herbicide-treated plot, the spatial pattern of SMB-N was influenced by the plantation grid. Herbicide applications markedly decreased spatial variability of forest floor properties. In some frequencies, SMB-N was markedly positively correlated with forest floor layer WC but not with forest floor TEMP.

Abbreviations: SMB-C, soil microbial biomass C • SMB-N, soil microbial biomass N • TEMP, temperature • WC, water content

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
HIGH INHERENT SPATIAL VARIABILITY of forest soils often limits our ability to characterize soil properties, and to quantify soil processes, for example, the activity of soil microbes (Parkin, 1993). Soil microbial biomass C, SMB-N, and net N mineralization rate are potential indicators of soil quality (Gregorich et al., 1994) that exhibit spatial and temporal patterns (Parkin, 1993; Robertson et al., 1988). In the context of monitoring impacts of management on forest ecosystem function, it is important to increase our understanding of the significance of these patterns at different spatial scales.

High variability also hinders the definition of predictive relationships among physicochemical factors, which are known to contribute to microbially mediated processes. If our goal is to understand and ultimately predict and model soil processes, then we must have a quantitative understanding of the factors that influence these processes. Evaluating spatial and temporal variability can contribute to identification of key driving factors controlling microbial processes in soil, because the variance associated with a given process is as much a property of the process as is the mean. Quantification and identification of these factors may thus be a potentially useful tool for constructing relationships between soil variables.

Spatial variability is manifested at many scales: microscale, plot, stand, landscape, and regional levels. Different processes operate at these scales to create a pattern of nested variability (Robertson and Gross, 1994). Moreover, the scale-dependency of both ecosystem processes and community attributes means that the phenomenon of interest may display spatial patterns only over certain distances or sets of distances, and not over others. For example, Decker et al. (1999) observed that activity within a soil enzyme assemblage exhibited strong spatial patterning that could occur at a scale of several dozens of centimeters and then reoccur at a scale of several hundreds of meters. In a geostatistical analysis of agricultural field properties, Stoyan et al. (2000) demonstrated that extremely high small-scale variability in soil respiration occurs at a scale smaller than a few square centimeters, which was a pattern that was imposed on a broader-scale spatial dependency in these data and associated with the effects of nearby individual plants. In the present study, we focus on the forest stand scale since this is the most appropriate scale to study the impacts of silvicultural treatments affecting soil biological processes (Barg and Edmonds, 1999; Lovell and Jarvis, 1998; Lovell et al., 1995).

The primary objective of this study was to investigate how an intensive silvicultural treatment such as an herbicide application (applied for 4 yr) can affect the spatial pattern of forest floor microclimatic and biological parameters. By eliminating hardwood and herbaceous vegetation, the herbicide treatment created a much more uniform type of litterfall (almost 100% pine needles; Munson, unpublished data, 1995–1999) and therefore forest floor substrate, versus a diverse mix of herbaceous, hardwood and pine litter under untreated conditions. A secondary objective was to relate the spatial variability of biological parameters to microclimatic conditions (TEMP and WC) in the forest floor.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Study Area, Experimental Design, and Sampling Method
The experimental site is situated within the Petawawa Research Forest, which is located on the north shore of Cartier Lake (45°57'50'' N, 77°34'45'' E; 170 m above sea level) in central Ontario (Canada). The site is in the Middle Ottawa Section of the Great Lakes-St. Lawrence Forest region (Rowe, 1984). The underlying bedrock is Precambrian granite and gneiss that is overlain by a deep, well-drained loam to sandy-loam soil, which is classified as an Orthic Humoferric Podzol (= Typic Haplorthod, Anonymous, 1998). The regional climate is moist-humid (Hills, 1959), with annual precipitation totaling 800 mm. Mean daily maximum temperatures are recorded in July (25.4°C) and minimum temperatures in January (–18.4°C). The experimental site is located in a 10-ha clearcut, which was logged for aspen, birch, spruce, and pine during the summer of 1985. In April 1986, each experimental plot (20 x 40 m), was planted with 100 3-yr-old bare-root seedlings of eastern white pine (Pinus strobus L.) and 100 3-yr-old bare-root seedlings of white spruce [Picea glauca (Moench) Voss], at 2 by 2 m spacing (each species planted in one half of the plot: 20 x 20 m). The initial experimental design (Munson et al., 1993) was a 2³ factorial, randomized complete block. The factors were scarification, fertilization, and herbicide application, each with two levels (treatment applied vs. treatment not applied). In this experiment, only two plots of the original experimental design were used. The control plot was undisturbed since plantation establishment in 1986, while the "treated plot" received glyphosate herbicide [N- (phosphonomethyl) glycine] applied at 2.0 kg active ingredient ha^–1 in midsummer for four consecutive years (1987–1990) to suppress competing vegetation.

The experimental unit was a transect of 25 m in length. In both plots, the transect was situated mid way between two rows (or former rows) of planted pines at 2 by 2 m spacing. Distance to the nearest tree varied much more in the control plot, which was dominated by poplar trees that regenerated naturally, with only a very few surviving white pine trees in the former rows. One hundred sampling locations were distributed at 0.25-m intervals along the transects.

Nitrogen mineralization rates were evaluated using the buried polyethylene bag technique (see details on the method used in Périé and Munson, 2000), in which forest floor material was incubated in situ for 8 wk (June–August 1997). In June, when the polyethylene bags were installed, a second set of cores was sampled to assess initial levels of inorganic N (NH₄⁺ plus NO₃^–). Soils were extracted with 2 M KCl, filtered, and concentrations of inorganic N in extracts were determined by flow-injection analysis (Lachat Instruments Inc., Zelleweger Analytics, Milwaukee, WI). In August, forest floor TEMP (0.04-m depth) was measured at each sampling point using an electronic temperature probe, and forest floor samples (cores of 10 cm in diameter) consisting of all organic matter above the mineral soil surface were collected. Forest floor depth was measured at the same time, and used to calculate bulk density. Samples were transported to the laboratory in a cooler with cold packs and stored at 5°C before being processed in the laboratory within 48 h of sampling. Water content was determined gravimetrically after oven-drying the subsamples (65°C for 48 h). Fresh forest floor pH was measured according to McLean (1982). Quantification of organic matter in the forest floor was measured gravimetrically by loss on ignition (Gallardo et al., 1987). The organic matter was converted to organic C (C_org) by a conversion factor of 0.56 (Nelson and Sommers, 1982). Total N was analyzed using a Tetacor 1030 Macro-Kjeldhal Analyzer. Soil microbial biomass-C and SMB-N were determined on forest floor subsamples by CHCl₃ fumigation-extraction (Brookes et al., 1985). Values of extractable-C and N were calculated as the difference between C and N extracted with 0.5 M K₂SO₄ from chloroform-fumigated (for 24-h) and from non-fumigated samples. Analyses of the soluble organic N in the extracts were made with a Tecator 1030 Macro-Kjeldahl Analyzer (Hoganas, Sweden). A k_EN factor of 0.45 (Jenkinson, 1988) was used to convert the extractable-N flush to NMB. All data are expressed on an oven-dry basis (65°C) and are converted into a mass per hectare of soil using measured bulk density values for each measurement site.

Statistical Analyses
Probability levels were set a priori at = 0.05.

Classical Analyses
Descriptive statistics and Pearson product-moment correlations (probability level was adjusted for multiple comparisons using Bonferroni adjustment) were performed using SYSTAT software version 10 (Systat Software Inc., Richmond, CA).

Time Series Analyses
Regionalized variable theory (Davis, 1986; Matheron, 1963) assumes that observations of a spatial series are usually not independent from each other and the spatial variability of any variable can be expressed as the sum of three major components:

[1]

where Z(x) is a spatial variable at location x; m(x) is a deterministic function describing the systematic variability of Z(x) at x; '(x) denotes the random locally varying spatially dependent residuals from m(x); and '' is a spatially independent residual having a mean 0 and a variance ² (Burrough, 1991). Observations obtained close to one another are more likely to be similar than observations at greater distances from each other. Typically, the spatial structure underlying the observations is a component of random variation, which can be used to model local variation in the data using variograms and spatial interpolation by kriging. However, regional variation may only be captured by fitting a broad-scale smoothing function to the data, such as a trend surface. A spatial series of observations is just as likely to include values that recur at a given distance or set of distances, and which could be ascribed to periodicity in the data. The detection of these recurrent, deterministic features of the data require yet another analytical tool. However, periodicity or periodicities in spatial data sets can be assessed using the same procedures as those developed for time-series.

Partitioning of the variation in regionalized variables into components, according to the length of the intervals at which the variation occurs, can be used to detect the periodicity of either temporal or spatial phenomena (Davis, 1986). This type of analysis, known as spectral analysis, can decompose all temporal or spatial processes, no matter how complex, into sinusoidal components of varying frequencies and amplitudes. The resulting power spectrum is often best described in terms of Fourier harmonic analysis (Shumway, 1988). Fourier harmonic analysis yields a periodogram, which is a plot of the variance (k) versus the frequency of the K^th harmonic. Periodograms usually are expressed as functions of frequencies instead of periods (frequency = 1/period). The amplitude of the periodogram for each frequency combination measures the amount of variance of the regionalized variable explained by the given frequency, which is estimated by least-squares fitting to the data of a Fourier series with the given frequency. The minimum and maximum frequencies that can be separated out are directly determined by the length of the transect (the fundamental cycling interval) and the sampling interval (the Nyquist frequency); the corresponding periods are equal respectively to the length of the transect and to twice the sampling interval. Non-stationarity of the variable may seriously distort their respective spectra (Kendall and Ord, 1990). For this reason, and according to Worrall and Burt (1998), all spatial series were detrended before computation of the periodograms. Detrending was accomplished by removing linear trends from the data using linear regression. Each value of k has two degrees of freedom (Shumway, 1988), and it is common practice to average adjacent values of k to obtain estimates with higher degrees of freedom, and thus create a smoothed periodogram called the power spectrum. For a forest floor indicator or soil microclimate factor x_j[j = 1,2,...,N] measured at N discrete equally spaced intervals across a finite distance L, the power spectrum S_x²(f_k) can be estimated by

[2]

where f_k = K/N, K = 0,1,2,...,N/2 and m is a smoothing coefficient determining the degree of adjacent averaging of the periodogram (Kachanoski et al., 1985). The estimator S_x²(f_k) gives the average spectral variance over the frequency interval (bandwidth) B, centered at f_k:

[3]

As the bandwidth B gets larger, the estimated power spectrum gets smoother. It is possible to compare two different spectra over the same frequency band with an F-test. The ratio of power spectra for a given bandwidth then is exactly equivalent to a ratio of variances, and therefore, an F-test. The hypothesis that the two spectra are equal at is rejected for (Shumway, 1988):

[4]

with

[5]

Bivariate Spectral Analysis
This analysis establishes inter-relationships between two spatial series and uses the relationships to forecast the trend of a target series (y; output signal) from observations made on a primary series x; input signal). The smoothed cross-spectral estimate for two observed variables x and y is given by:

[6]

where * = complex conjugate and f_k = K/N, K = 0,1,2,...,N/2 cycles/measurement interval. The cross-spectral estimate (Eq. [6]) is a complex function, which can be represented by:

[7]

where C_xy(f_x) is the cospectrum and Q_xy(f_x) is the quadrature spectrum, which are the respective in-phase and out-of-phase covariances of x and y. The correspondence between the two series (x and y), as a function of frequency, was described using the smoothed squared-coherency spectrum R_xy(f_x)

[8]

The squared-coherency spectrum is analogous to the R² in a simple linear regression analysis and measures the influence of noise in the system. A large noise spectrum relative to the signal will result in low coherency.

Phase spectra between two variables can be calculated (Jenkins and Watts, 1968)

[9]

where P_xy(f_x) is the phase spectrum. A simple interpretation of the phase of linear filter is that it represents the time delays as a function of frequency in the same way as spectrum measures the variance as a function of frequency.

Confidence intervals for the power spectra and the coherency spectra are calculated from equations given by Shumway (1988). All estimates, including confidence intervals, depend on the choice of bandwith B or more specifically, on the choice of the smoothing coefficient m. Small values of m help in locating periodic (deterministic) components, while larger values of m yield smoother curves for stochastic model fitting.

In the present experiment, 100 locations (N = 100) were used in spectral estimation. Shumway (1988) recommends that a value of m equal to 1/20 of N be used in the analysis, that is, m = 5, in our study. The corresponding bandwidth of 0.44 cycle m^–1 gave reasonably smooth spectra without decreasing the resolution of what are considered to be significant peaks. Bartlett's Kolmogorov Smirnov (BKS) test (Fuller, 1995) was used to compare the normalized cumulative periodogram of each series against the cumulative distribution function of a normal random variate (0,1), under the null hypothesis that the series are white noise.

Time series analyses were performed using the procedure called Proc Spectra in SAS version 8.0 (SAS Institute, Cary, NC).

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Descriptive statistics for the two transects are summarized in Table 1. Mean values are consistent with previous studies conducted at this site (Périé and Munson, 2000). Figure 1 presents the values of forest floor WC, TEMP, net N mineralization rate, SMB-C, and SMB-N as a function of distance along transects. Although the data for all variables behave in a stochastic manner, values seemed to exhibit a periodic pattern such as peaks for WC occurring nearly every meter, and this is as much in the control plot as in the treated plot. In the treated plot (herbicide application treatment), values of forest floor WC, net N mineralization, and SMB-N are lower than those of the control plot (50, 80, and 60% for WC, net N mineralization rate, and SMB-N, respectively; Table 1). On the contrary, forest floor TEMP was higher (on average 1°C) in the treated plot where competing vegetation was suppressed, compared with the control plot. Temperature seemed to vary with approximately the same magnitude for both treatments, but for a high peak at the beginning of the transect (Fig. 1).

View larger version (22K): [in this window] [in a new window]   Fig. 1. Observed values of forest floor gravimetric water content (WC), temperature (TEMP), and microbial biomass N (SMB-N) for two different conditions: control plot which is an undisturbed plot since pine plantation and treated plot which is a plot where competing vegetation was suppressed for 4 yr by herbicide application.

Simple correlations between microclimatic parameters (WC and TEMP) and SMB-N are presented in Table 2. Nitrogen mineralization was not correlated with soil microclimatic parameters, perhaps because of high variability (CV = 296 and 164% for control plot and treated plot, respectively). This high variability is probably due to the coexistence of mineralization and immobilization phenomena along each transect. Soil microbial biomass-N is slightly positively correlated with WC in the control plot and is slightly negatively correlated with TEMP in the treated plot; however, only 7% of the variability of SMB-N can be explained by WC (control plot) or by TEMP (treated plot). In a number of studies (Menyailo and Huwe, 1999; Sarathcandra et al., 1989; Sato and Seto, 1999; Verburg et al., 1999) soil microclimate influenced SMB-N levels. The presented correlation results should be interpreted with caution because results may be influenced by the spatial dependence of the samples, which could invalidate the statistical analysis (Legendre and Fortin, 1989).

View larger version (23K): [in this window] [in a new window]   Fig. 2. Bartlett's Kolmogorov Smirnov (BKS) statistic for white noise test for forest floor water content (WC), temperature (TEMP), net N mineralization rate (N min), and microbial biomass N (SMB-N).

Herbicide application in the treated plot appears to both markedly alter the composition and structure of understory vegetation (Burgess et al., 1995) as well as the mosaic of patches and microsites in soils, thereby reducing the discontinuity between patches. Ross-Davis and Frego (2002) observed that plantations result in a more uniform environment in terms of available substrates and microclimatic conditions than do naturally regenerated clearcut forests, and as a result, plantations exhibit correlated reduced floor bryophyte diversity. There is general agreement in literature that habitat heterogeneity is related positively to diversity (Fox, 1981). For example, Reynolds et al. (1997) noted that plant diversity contributes to soil heterogeneity and in turn, soil heterogeneity feeds back to contribute to plant species richness. In the present study, the composition of the litter of the treated plot (dominantly conifer needles) is markedly different (Périé and Munson, 2000) from that of the control plot (mainly deciduous leaves from trembling aspen, other shrubs, and forbs). This difference may affect decomposition rates, mineral nutrient dynamics, and also the structure of soil food webs (Chapman et al., 1988). If a community is primarily determined by habitat features (Söderström et al., 2001) then changes in substrate quality and microclimate conditions are likely to alter community composition. These modifications could potentially reduce the diversity of the soil organisms (Beare et al., 1995).

When the variable is clearly regionalized (non random), the presence of specific peaks in the time series provides information on the spatial or temporal structure of the underlying processes. Nielsen et al. (1983), Shumway et al. (1989), Kachanoski et al. (1985), and Timlin et al. (1998) have used such techniques to analyze spatial variability pattern in uniformly managed fields and identify at which scale they occurred. Periodograms are used herein to identify spatial pattern corresponding to vegetation patterns and altered by silvicultural treatments. Hence, periodograms are compared one relative to another, in a similar way to Perfect and Caron (2002), to identify at which spatial scale silvicultural treatments alter specific forest floor properties. Figure 3 illustrates the power spectrum (illustrating the repartition of the variance as a function of the scale of sampling, expressed in frequency) for forest floor WC, forest floor TEMP, and finally for forest floor SMB-N measured in the control plot (a) and in the treated plot (b). The WC spectrum observed in the control plot presents four peaks but only the first two are significant. This indicates that the variance was mainly observed at some specific scale. In the control plot, WC shows a reasonable peak at a frequency of 0.01 to 0.05 cycle/0.25 m. This frequency range corresponds to tendency for peaking every 5 to 25 m, the largest variance corresponding to a period of about 6 m. The second peak, at a frequency of 0.10 to 0.14 cycle/0.25 m may indicate the presence of additional cycling with a period of about 1.6 to 2.4 m, the largest variance corresponding to a period of 1.7 m.

View larger version (51K): [in this window] [in a new window]   Fig. 3. Power spectra for forest floor water content (WC), temperature (TEMP), and microbial biomass N (SMB-N) along transect situated in two different plots: control plot (a) which is an undisturbed plot since pine plantation and treated plot (b) which is a plot where competing vegetation was suppressed for 4 yr by herbicide application. Shading represents significant peaks.

View this table: [in this window] [in a new window]   Table 4. Frequency of the significant peaks ( = 0.05) for organic layer water content (WC), organic layer temperature (TEMP), and organic layer microbial biomass N (SMB-N). Power spectra were measured both in the control plot and in treated plot. Common frequencies between WC and SMB-N and between TEMP and SMB-N are also indicated. The number of frequencies covered by each peak is indicated in parentheses.

Impacts of herbicide application on the spatial pattern of forest floor WC are not the same as those on spatial pattern of TEMP. The spatial pattern of temperature seems to be only affected by the quantity of light that reaches the forest floor, which depends on crown closure (and hence is related to large period around 5–6 m) and not on the aboveground composition. In contrast, the spatial pattern of the WC seems to depend on aboveground and canopy structure, since in the treated plot, the spatial pattern corresponds to the plantation pattern (period around 2 m).

The plot of the power spectrum for SMB-N in the control plot reveals a dominant component of variance of SMB-N at low frequencies (Fig. 3a). Two peaks are significant in this range of frequencies (0.01–0.04 cycle/0.25 m and 0.06–0.09 cycle/0.25 m). In the treated plot, SMB-N power spectrum shows significant components of SMB-N variance at periods varying between 2 and 8 m (0.03 < f < 0.10) with the maximum variance at a period of 4.85 m. This cycling of SMB-N at a period of approximately 5 m can be seen in Fig. 1. However, two smaller significant peaks are also present and are located in the high frequencies, centered at a period of 1 and 0.7 m respectively (Fig. 3b).

The analysis performed above clearly provided much more information that the summary statistics for Table 1, from which no information on the spatial structure can be inferred, but on the overall variance across all scales using the SD. With the frequency domain approach, the information on the spatial structure of variability was particularly relevant, indicating that changes about the mean were created by the patterns of WC and biomass, at a scale corresponding to plantation grid, an information otherwise extremely difficult to obtain from the examination of the variable scatter plot and the use of summary statistics.

Inasmuch as spectra of Fig. 3 suggest the existence of cycling in some variables, the cospectra and squared coherency spectra may allow the linear relationships existing between SMB-N and microclimate variables (WC and TEMP) at different common frequencies (Table 4) to be elucidated. While spectra partition the variance as a function of spatial scale, cospectra and coherency partition the covariance and the correlation between two variables as a function of spatial scale. Hence, for example, coherency reflects the contribution of WC (or TEMP) to the total variability of SMB-N at that scale.

In the control plot, the squared coherency is >80% at low frequencies (f < 0.03, corresponding to cycles greater than 8 m; Fig. 4a ) for the WC series and SMB-N series, indicating that SMB-N was strongly correlated with WC changes at that scale. The cospectrum showed strong positive covariances (related to correlation) at low frequencies and near zero or negative covariances at frequencies > 0.03. This shows that SMB-N is nearly always positively correlated to WC over all significant frequencies. The more WC increases, the more SMB-N increases. Because quadrature spectrum (which represents the out of phase covariance) was smaller than cospectrum (which represents the in phase covariance), there is no spatial delay between the two variables (Fig. 4b).

View larger version (33K): [in this window] [in a new window]   Fig. 4. Cospectrum, quadrature spectra (dotted line) and coherency spectra for microbial biomass N vs. forest floor water content measured in two plots: control plot (a) which is an undisturbed plot since pine plantation and treated plot (b) which is a plot where competing vegetation was suppressed for 4 yr by herbicide application. Shading represents significant common peaks.

View larger version (34K): [in this window] [in a new window]   Fig. 5. Cospectrum, quandrature spectrum (dotted line) and coherency spectra for microbial biomass N vs. forest floor temperature measured in two plots: control plot (a) which is an undisturbed plot since pine plantation and treated plot (b) which is a plot where competing vegetation was suppressed for 4 yr by herbicide application. Shading represents significant common peaks.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

	ACKNOWLEDGMENTS

 
We thank Mathieu Côté, André Beaumont, Gina Racine, and Alain Brousseau for assistance in the field and in the laboratory. Thanks to Craig Robinson and Steve D'Eon at the Petawawa Research Forest for logistical support. Two anonymous reviewers provided excellent suggestions for improving the manuscript. Funding for the research was provided by a Natural Sciences and Engineering Research Council Research Grant to Alison Munson.

Received for publication September 28, 2004.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 

• Anonymous. 1998. The Canadian System of soil classification. 2nd ed. NRC Research Press, Ottawa, Canada.
• Barg, A.K., and R.L. Edmonds. 1999. Influence of partial cutting on site microclimate, soil nitrogen dynamics, and microbial biomass in Douglas-fir stands in western Washington. Can. J. For. Res. 29:705–713.[CrossRef]
• Beare, M.H., D.C. Coleman, D.A. Crossley, P.F. Hendrix, and E.P. Odum. 1995. A hierarchical approach to evaluating the significance of soil biodiversity to biogeochemical cycling. Plant Soil 170:5–22.[CrossRef]
• Boucher, J.F., S. Wetzel, and D.A. Munson. 1998. Leaf level response of planted eastern white pine (Pinus strobus L.) seven years after intensive silvicultural treatments. For. Ecol. Manage. 107:291–307.[CrossRef]
• Brookes, P.C., A. Landman, G. Pruden, and D.S. Jenkinson. 1985. Chloroform fumigation and the release of soil nitrogen: A rapid direct extraction method to measure microbial biomass nitrogen in soil. Soil Biol. Biochem. 17:837–842.[CrossRef]
• Burgess, D., J.A. Baldock, S. Wetzel, and D.G. Brand. 1995. Scarification, fertilization, and herbicide treatment effects on planted conifers and soil fertility. Plant Soil 168–169:513–522.
• Burrough, P.A. 1991. Sampling designs for quantifying map unit composition. p. 89–125. In M.J. Mausbach and L.P. Wilding (ed.) Spatial variability of soils and landforms. SSSA Spec. Publ. 28, SSSA, Madison, WI.
• Chapman, K., J.B. Wittaker, and O.W. Heal. 1988. Metabolic and faunal activity in litters of tree mixtures compared with pure stands. Agric. Ecosyst. Environ. 24:33–40.
• Davis, J.C. 1986. Statistics and data analysis in geology. 2nd ed. New York.
• Decker, K.L.M., R.E.J. Boerner, and S.J. Morris. 1999. Scale-dependant patterns of soil enzyme activity in a forested landscape. Can. J. For. Res. 29:232–241.[CrossRef]
• Fox, B.J. 1981. Niche parameters and species richness. Ecology 62:1415–1425.
• Fuller, W.A. 1995. Introduction to statistical time series, 2nd ed. John Wiley & Sons, New York.
• Gallardo, J.F., J. Saavedra, T. Martin-Patino, and A. Millen. 1987. Soil organic matter determination. Commun. Soil Sci. Plant Anal. 18:699–707.
• Gregorich, E.G., M.R. Carter, D.A. Angers, C.M. Monreal, and B.H. Ellert. 1994. Towards a minimum data set to assess soil organic matter quality in agricultural soils. Can. J. Soil Sci. 74:367–385.
• Hills, G.A. 1959. A ready reference to the description of the land of Ontario and its productivity. Ontario Dep. Lands and Forests, Toronto.
• Jenkins, G.M., and D.G. Watts. 1968. Spectral analysis and its application. Holden Day, San Francisco.
• Jenkinson, D.S. 1988. Determination of microbial biomass and nitrogen in soil. p. 368–386. In J.R. Wilson (ed.) Advances in nitrogen cycling in agricultural ecosystems. Commonwealth Agricultural Bureau International, Wallingford.
• Kachanoski, R.G., D.E. Rolston, and J.E. DeJong. 1985. Spatial and spectral relationships of soil properties and microtopography: I. Density and thickness of A horizon. Soil Sci. Soc. Am. J. 49:804–812.[Abstract/Free Full Text]
• Kendall, M., and J.K. Ord. 1990. Time series. 3rd ed. Griffin, London and Oxford Univ. Press, New York.
• Legendre, P., and M.-J. Fortin. 1989. Spatial pattern and ecological analysis. Vegetatio 80:107–138.
• Lovell, R.D., and S.C. Jarvis. 1998. Soil microbial biomass and activity in soil from different grassland management treatments stored under uncontrolled conditions. Soil Biol. Biochem. 30:2077–2085.[CrossRef]
• Lovell, R.D., S.C. Jarvis, and R.D. Bardgett. 1995. Soil microbial biomass and activity in long-term grassland: Effects of management changes. Soil Biol. Biochem. 27:969–975.[CrossRef]
• Matheron, G. 1963. Principles of geostatistics. Econ. Geology 58:1246–1268.
• McLean, E.O. 1982. Soil pH and lime requirement. p. 199–223. In A.L. Page et al. (ed.) Methods of soil analysis. Part 2. 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI.
• Menyailo, O., and B. Huwe. 1999. Denitrification and C, N mineralization as function of temperature and moisture potential in organic and mineral horizons of an acid spruce forest soil. J. Plant Nutr. Soil Sci. 162:527–531.[CrossRef]
• Munson, A.D., H.A. Margolis, and D.G. Brand. 1993. Intensive silvicultural treatment: Impacts on soil fertility and planted conifer response. Soil Sci. Soc. Am. J. 57:246–255.[Abstract/Free Full Text]
• Nelson, D.W., and L.E. Sommers. 1982. Total carbon, organic carbon, and organic matter. p. 539–579. In A.L. Page et al. (ed.) Methods of soil analysis. Part 2. 2nd ed. Agron. Monogr. No. 9. ASA and SSSA, Madison, WI.
• Nielsen, D.R., P.M. Tillotson, and S.R. Vierra. 1983. Analyzing field measured soil water properties. Agric. Water Manage. 6:93–109.
• Parkin, T.B. 1993. Spatial variability of microbial processes in soil—A review. J. Environ. Qual. 22:409–417.[Abstract/Free Full Text]
• Perfect, E., and J. Caron. 2002. Spectral analysis of tillage-induced differences in soil spatial variability. Soil Sci. Soc. Am. J. 66:1587–1595.[Abstract/Free Full Text]
• Périé, C., and A.D. Munson. 2000. Ten-year responses of soil quality and conifer growth to silvicultural treatments. Soil Sci. Soc. Am. J. 64:1815–1826.[Abstract/Free Full Text]
• Reynolds, H.L., B.A. Hungate, F.S. Chapin, III, and C.M. D'Antonio. 1997. Soil heterogeneity and plant competition in an annual grassland. Ecology 78:2076–2090.[ISI]
• Robertson, G.P., and K.L. Gross. 1994. Assessing the heterogeneity of belowground resources: Quantifying pattern and scale, p. 237–253. In M.M. Caldwell and R.W. Pearcy (ed.) Exploitation of environmental heterogeneity by plants. Ecophysiological processes—above- and belowground. Academic Press, New York.
• Robertson, G.P., M.A. Huston, F.C. Evans, and J.M. Tiedje. 1988. Spatial variability in a successional plant community: Patterns of nitrogen availability. Ecology 69:1517–1524.[CrossRef][ISI]
• Ross-Davis, A.L., and K.A. Frego. 2002. Comparison of plantations and naturally regenerated clearcuts in the Acadian forest: Forest floor bryophyte community and habitat features. Can. J. Bot. 80:21–33.[CrossRef]
• Rowe, J.S. 1984. Forestland classification: Limitations of the use of vegetation. p. 132–148. In J.G. Bockheim (ed.) Forestland classification: Experiences, problems, perspectives. Department of Soil Science, Madison, WI.
• Sarathcandra, S.U., K.W. Perrot, and R.A. Littler. 1989. Soil microbial biomass: Influence of simulated temperature changes on size, activity and nutrient content. Soil Biol. Biochem. 21:987–993.[CrossRef]
• SAS Institute. 1992. SAS/ETS software: Application guide 1. SAS institute, Cary, NC.
• Sato, A., and M. Seto. 1999. Relationship between rate of carbon dioxide evolution, microbial biomass carbon, and amount of dissolved organic carbon as affected by temperature and water content of a forest and an arable soil. Commun. Soil Sci. Plant Anal. 30:2593–2605.
• Shumway, R.H. 1988. Spectral analysis and filtering. p. 47–116. In R.A.J.D.W. Wichern (ed.) Applied statistical time series analysis. Prentice Hall, Englewood Cliffs, NJ.
• Shumway, R.H., J.W. Biggar, F. Morkoc, M. Bazza, and D.R. Nielsen. 1989. Time and frequency domain of field observations. Soil Sci. 147:286–298.
• Söderström, B., B. Svensson, K. Vessby, and A. Glimskär. 2001. Plants, insects and birds in semi-natural pastures in relation to local habitat and landscape factors. Biodivers. Conserv. 10:1839–1863.[CrossRef]
• Stoyan, H., H. De-Polli, S. Böhm, G.P. Robertson, and E.A. Paul. 2000. Spatial heterogeneity of soil respiration and related properties at the plant scale. Plant Soil 222:203–214.[CrossRef]
• Timlin, D.J., Y. Pachepsky, V.A. Snyder, and R.B. Bryant. 1998. Spatial and temporal variability of corn grain yield on a hill slope. Soil Sci. Soc. Am. J. 62:764–773.[Abstract/Free Full Text]
• Verburg, P.S.J., D. VanDam, M.M. Hefting, and A. Tietema. 1999. Microbial transformations of C and N in a boreal forest floor as affected by temperature. Plant Soil 208:187–197.[CrossRef]
• Worrall, F., and T.P. Burt. 1998. Decomposition of river water nitrate time-series—Comparing agricultural and urban signals. Sci. Total Environ. 210/211:153–162.

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Périé, C. Articles by Caron, J. Search for Related Content PubMed Articles by Périé, C. Articles by Caron, J. Agricola Articles by Périé, C. Articles by Caron, J. Related Collections Forest Soils Spatial Variability Ecosystem Management