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


     

Published online 28 September 2007
Published in Soil Sci Soc Am J 71:1748-1757 (2007)
DOI: 10.2136/sssaj2006.0007
© 2007 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 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 Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Shukla, M. K.
Right arrow Articles by VanLeeuwen, D.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Shukla, M. K.
Right arrow Articles by VanLeeuwen, D.
Agricola
Right arrow Articles by Shukla, M. K.
Right arrow Articles by VanLeeuwen, D.
Related Collections
Right arrow Soil Organic Matter
Right arrow Spatial Variability
Right arrow Other Soil Types

SOIL & WATER MANAGEMENT & CONSERVATION

Spatial Variability of Aggregate-Associated Carbon and Nitrogen Contents in the Reclaimed Minesoils of Eastern Ohio

Manoj K. Shuklaa,*, Rattan Lalb and Dawn VanLeeuwenc

a Dep. of Plant and Environmental Sciences, New Mexico State Univ., Las Cruces, NM 88003
b Carbon Management and Sequestration Center, Ohio State Univ., Columbus, OH
c Agricultural Biometric Service, New Mexico State Univ., Las Cruces, NM 88003

* Corresponding author (shuklamk{at}nmsu.edu).


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Associations of organic C and N with primary (sand, silt, and clay) and secondary (aggregates) particles result in their storage and retention in the soil. The objectives of this study were to assess (i) the CV of C and N contents in the primary and secondary particles and (ii) the spatial variability of water-stability of aggregates (WSA), geometric mean diameter (GMD) and mean weight diameter (MWD) of aggregates, and C and N contents in macro- (>2 mm), meso- (2–0.25 mm), and microaggregate (0.25–0.05 mm) fractions. Forty-five soil samples were collected at the 0- to 15-cm depth of a 20- by 20-m grid at an unmined (UMG) reference site and sites reclaimed in 1978 (R78G), 1982 (R82GT), and 1987 (R87G). The R82GT was seeded to grass in 1982 and planted with trees in 1987. Other sites were under continuous grass cover since reclamation. Soil samples (<2 mm, four samples per site) were fractionated into primary particles (sand, silt, and clay), and aggregate samples (8–5 mm, 45 samples per site) into macro-, meso-, and microaggregate fractions; C and N contents were determined for each fraction. The CV for C and N contents was low in all three primary particles. Carbon contents increased with decreasing size of primary particles and clay-associated C contents also increased with amount of time since reclamation. Geostatistical analysis showed that dispersion variance for WSA, GMD and MWD, and C and N contents (45 samples per site) was higher in R78G than other reclaimed sites. Interpolation using kriging displayed heterogeneity of properties across experimental sites. The relative nugget for most aggregate-associated properties was lowest for R78G and became stronger with the amount of time since reclamation. The range for aggregate-fraction-associated C contents was similar across reclaimed sites and spatial dependence became stronger with increasing amount of time since reclamation. The unmined site displayed large variability; however, low variability for the relatively newly reclaimed site (R87G) indicated that reclamation initially reduced the variability but as time increased, variability also increased, with simultaneous improvement in soil quality.

Abbreviations: GMD, geometric mean diameter • MWD, mean weight diameter • WSA, water stability of aggregates


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil properties and their spatial distribution on a large scale are available in soil survey reports for most counties in the USA. On the other hand, designing site-specific management strategies for practicing precision farming, or applying simulation modeling, requires soil data at a much finer scale, which are obtained mostly by on-site detailed sampling across the mapping units and land use or management practices (Gaston et al., 2001). Soil variability is usually associated with spatial, temporal, or management-related factors and each of these sources can partially or fully contribute to the variability of a soil property under investigation (van Es et al., 1999).

For an arbitrary separation distance, a stationary random function has the same probability distribution for all locations. It is generally assumed that first- and second-order stationarity exists within the domain. It is also assumed that the process can be adequately characterized by a constant mean and a covariance function. A constant mean indicates that measured attributes are independent of their spatial location and covariance is only a function of the spatial separation (Schabenberger and Pierce, 2002). Second-order stationarity also implies that the variogram exists (Isaaks and Srivastava, 1989). Geostatistics is a useful tool for analyzing spatial variability, interpolating between point observations, and ascertaining the interpolated values with a specified error using a minimum number of observations (Burrough, 1991). Several attempts have been made to characterize the variability of soil properties, and spatial dependence is reported for scales ranging from a few meters (Gajem et al., 1981; Trangmar et al., 1987) to several meters (Trangmar et al., 1987; Cambardella et al., 1994; Sun et al., 2003; Shukla et al., 2004b) to several kilometers (Ovalles and Collins, 1988; Lin et al., 2005). Most of these studies have been conducted on agricultural fields, and some studies have attempted to characterize the variability and spatial dependence in reclaimed minesoils (Keck et al., 1993; Schafer, 1979; Gerke et al., 2001; Mummey et al., 2002).

Surface mining for coal removal was an important economic activity in Ohio and elsewhere in the northern Appalachian region. Restoration of disturbed minesoil is equally important because it improves soil quality and increases the organic C pool (Pederson et al., 1980; Barnhisel and Gray, 2000; Shukla et al., 2004a). Increases in mean organic C stock with time have been reported in numerous studies (Paustian et al., 1997; West and Post, 2002); however, only a few accounts exist for reclaimed minesoils (Akala and Lal, 2001; Shukla et al., 2004b; Shukla and Lal, 2005).

Organic C in soil is associated with primary mineral particles (i.e., sand, silt, and clay) and secondary particles (i.e., micro-, meso-, and macroaggregates). Such associations, which physically and chemically protect organic C, are known as the controlling factors of C storage and retention in soil. In many soils, most of the C is stored in the clay size fraction (Anderson et al., 1981; Christensen, 1996). Organic C associated with primary particles differs in chemical composition, and recalcitrant fractions are mostly associated with clay particles (Christensen, 1996). Even within the clay fraction, organic C is reported to vary between different types of soils (Kogel-Knabner, 1997), probably due to the differences in mineralogy and humification processes (Laird et al., 2001).

For agricultural soils, variability of soil C due to changes in aggregate size or mineralogy have been reported in some studies (Tisdall and Oades, 1982; Denef et al., 2004); however, studies on the spatial variability of soil C in different aggregate fractions from agricultural soils are rare. There is a general lack of information on minesoils, and studies on the spatial variability of C associated with soil aggregate fractions from reclaimed minesoils are nonexistent. Thus, the objectives of this study were: (i) to quantify the statistical variability of C and N in primary and secondary particles; (ii) to assess field-scale spatial variability of C and N in secondary particles in reclaimed minesoils, and (iii) to assess the influence of duration since reclamation on increases in C in secondary particles. Our hypothesis for this research was that the spatial dependence of aggregate-fraction-associated C increases with the amount of time since reclamation.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Experimental Sites
The experimental sites were reclaimed in conformity with the 1972 Ohio Mineland Reclamation Act (or 1977 Surface Mining Control and Reclamation Act), which made mandatory the removal of topsoil from the mined land in a separate layer, preservation from wind and water erosion, and restoration or spread as the last step of reclamation. All experimental sites are maintained by the American Electric Power Co. of Ohio. This study included the analysis of soil data from three sites reclaimed in 1978 (R78G) and 1987 (R87G), both under continuous grass (G) cover, and in 1982 (R82GT) but under grass and tree (GT) cover. The R82GT site was seeded to grass immediately after reclamation in 1982 and trees were planted in 1987. An unmined site (UMG) under continuous grass cover was selected as a reference. Site R78G is located in Muskingum County, R87G in Noble County, R82GT in Guernsey County, and UMG in Morgan County, Ohio. All these sites are located within a radius of 3.5 km (Fig. 1 ). The entire study area was in the process of being mined when the soil survey was being published. Therefore, this area was not surveyed and mapped as Udorthents, meaning mine spoil (J. Mizik, personal communication, 2007). The exact soil classification is unknown for the study area; however, the entire region including the unmined area around the experimental sites prior to mining was identified as Gilpin–Upshur–Lowell–Guernsey (fine, mixed, superactive, mesic Typic Hapludalfs). Post-mining reclassification is not available.


Figure 1
View larger version (33K):
[in this window]
[in a new window]

  		Fig. 1. Location of the unmined reference (UMG, grass-covered) and three sites reclaimed in 1978 (R78G, grass), 1982 (R82GT, grass and trees), and 1987 (R87G, grass) in Ohio (modified from cartograph by worldAtlas.com).

 
Soil Sampling and Analysis
A total of 45 bulk soil samples were collected from each of the experimental sites on a regular grid of 20 by 20 m in April to May 2004 at the 0- to 15-cm depth (Fig. 2 ). There were three transects parallel to the x axis with a spacing of 20 m, and transects were oriented north to south in UMG, R78G, R82GT, and west to east in R87G. Soil samples were collected mostly 1 to 2 m north or south of the center of each grid as well as close to the center of the grid in a systematic unaligned sampling (see de Gruijter, 2002, Fig. 1.4–2). All soil samples were air dried and passed through nested sieves, which were arranged in the order 8, 5, and 2 mm. The portions retained on the 5-mm sieve and passed through the 2-mm sieve were stored separately. A subsample of 50 g of the sieved (<2-mm) soil was used for particle size analysis by the hydrometer method (Gee and Bauder, 1986).


Figure 2
View larger version (5K):
[in this window]
[in a new window]

  		Fig. 2. Map of the study site (not to the scale) showing sample locations and grids (20 by 20 m). Note: samples were collected 2 m north or south of the center of each grid. In R82GT, trees were planted on each row; therefore, each grid has two trees on the north and south side of the grid.

 
An automatic fractionator was used to separate soil fractions <2 mm into primary particles (sand, silt, and clay) without removing the organic C and Fe oxides (Rutledge et al., 1967). Since textural differences were low and fractionation is time consuming, only 16 soil samples were selected (four from each site) for fractionation into primary particles. All 16 samples were initially centrifuged at 1500 rpm for 10 min. Subsequently, 30 mL of 0.25 mol L–1 Na2CO3 was added as a dispersant and the dispersion was completed by ultrasonic treatment for 5 min at 100 to 120 W. The dispersed samples were passed through nested sieves and the fraction >53 µm (sand) was retained, oven dried, and weighed. The fraction <53 µm was used for sedimentation analysis to separate silt (2–53 µm) and clay (<2 µm) fractions (Laird and Dowdy, 1994). X-ray diffraction patterns were obtained with oriented clay slides to determine the clay mineralogy.

About 50 g of macroaggregates retained on the 5-mm sieve were separated by wet sieving. The wet-sieving device was comprised of six sets of sieves suspended from a bar, with each set consisting of three nested sieves arranged in a descending sequence of the opening diameter (2, 0.25, and 0.05 mm). A shaft and crank system connected to an electric motor moved the bar at approximately one oscillation every 2 s. The nested sets of sieves moved through a vertical distance of about 3 cm. Macroaggregates were kept on the top sieve (2 mm) and were allowed to prewet by capillary rise for about 30 min. Wet sieving was performed for 30 min and the material retained on each sieve was air dried and stored separately (Yoder, 1936). The water-stable aggregation (WSA), mean weight diameter (MWD), and geometric mean diameter (GMD) of the aggregates were determined by the method suggested by Yoder (1936) and Youker and McGuinness (1957). The portions retained on 2-, 0.25-, and 0.05-mm sieves were classified as macroaggregate, mesoaggregate, and microaggregate fractions, respectively.

About 1 g of each (<0.25-mm size) primary particle and aggregate fraction sample was used to determine total C and total N contents by dry combustion using a C-N analyzer (Elementar GmbH, Hanau, Germany). A dilute HCl solution was poured on the crushed aggregates from each site and no effervescence was detected.

Statistical Analysis
The variability relative to the mean of soil properties can be expressed as a CV. Wilding (1985) proposed a ranking of variability in soil properties that occur within landscape units of a few hectares or less. He ranked a CV <15% as the least, 15% < CV < 35% as moderate, and CV >35% as the most variable. Mean separation was computed for C and N contents from primary particles using the Bonferroni multiple t-test at a level of P ≤ 0.05 (SAS Institute, 1989). Regression analysis was performed between the amount of time since reclamation as the independent variable and aggregate properties as dependent variables using SigmaPlot (Version 8.0, Systat Software, San Jose, CA).

As long as the random field has a constant mean and the semivariogram model is correct, least squares fitting of semivariogram model parameters will yield a satisfactory result with a sufficient number of point pairs and it may not be necessary that the random field satisfy particular distributional assumptions. Consequently, semivariograms can be, and often are, fit using methods that do not rely on normality assumptions (Schabenberger and Pierce, 2002). Since skewed data distributions tend to mask the spatial structure (Haws et al., 2004), all measured soil physical and chemical property data were checked for normality by using a Kolmogorov–Smirnov test (SAS Institute, 1989). The spatial structure for each measured variable was studied using the following experimental semivariogram:

Formula 1[1]
where {gamma}(h) is the semivariance for lag interval h, z(xi) and z(xi + h) are variables at locations xi and xi + h, and N(h) is the number of pairs separated by distance h. Using VARIOWIN (Pannatier, 1996), experimental variograms of each soil property were obtained by minimizing the sum of squared deviations between experimental and theoretical semivariograms. Theoretical variograms were obtained by fitting spherical, exponential, and Gaussian models. The spherical model was selected because it produced the least sum of squared deviations between experimental and theoretical semivariograms for most attributes. The theoretical semivariograms for each property were obtained by (i) assuming a constant nugget equal to zero, (ii) fitting a pure nugget effect (range = 0), and (iii) including the nugget effect and range as fitting parameters. The following spherical model with nugget effect was finally selected as the best-fitting model:

Formula 2[2]
where C0 is nugget, h is lag distance, and a is the range of spatial dependence to reach the sill (C0 + C1). Continuous maps of individual attributes were generated by point kriging without drift, which estimates the values of the points at the grid nodes (Isaaks and Srivastava, 1989). To have a uniform comparison index among variables with different measurement units, the nondimensional number obtained as the ratio of the nugget to the sill "relative nugget effect" suggested by Cambardella et al. (1994) was used as an indicator of spatial dependence for those properties for which range values were similar.


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Particle Size Distribution
Increases in aggregate stability and C are directly related to the silt and clay contents of the soil (Kemper and Koch, 1966; Attou et al., 1998; Boix-Fayos et al., 2001). The accretion in soil C and aggregation are functions of time, climate, vegetation, parent material, and post-mining management (Schulze and Stitt, 1995; Merrill et al., 1998). Therefore, a similar silt and clay content for all experimental sites, which are located within a similar climate zone and under similar vegetation, is essential for quantifying the temporal changes in aggregate stability and organic C accretion. In our study, average sand content varied from 16% for R87G to 22% for R78G, silt content from 57% for R78 to 62% for UMG, and clay content from 20% for R82GT to 21% for UMG. Overall, the variability of sand, silt, and clay contents was mostly low for clay (CV = 6–8%), and low to moderate for silt (CV = 6–16%) and sand content (CV = 8–17%) (data not presented). Soil texture varied slightly along the divide between clay loam and silt loam according to the USDA textural triangle for all sites. The x-ray diffraction patterns were similar for the experimental sites and showed that mineralogy was fairly similar among reclaimed and reference sites. Among the principal silicate clay minerals of importance in soils, illite, vermiculite, smectite, and kaolinite were present in all sites; in addition, traces of geothite were also observed in all three reclaimed sites.

Carbon and Nitrogen Contents Associated with Primary Particles
Some studies have reported that C and N contents vary within the size range of each primary particle, especially clay (Anderson et al., 1981; Tiessen and Stewart, 1983). We found, however, that the variability of C and N contents associated with each primary particle was low, with CV ranging from 1 to 12% for C content and 4 to 16% for N content at each site (Table 1 ).


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

  		Table 1. The CV of C and N contents in primary particles (sand, silt, and clay) from three sites reclaimed in 1978 (R78G, grass-covered), 1982 (R82GT, grass and trees), and 1987 (R87G, grass) and an unmined reference (UMG, grass) (n = 4 for each site).

 
In general, changes in land use pattern and duration affect soil chemical properties (Haering et al., 1993; Schjonning et al., 1994; Haynes and Naidu, 1998; Shukla and Lal, 2005). Despite similarity in particle size distribution and mineralogy, there were differences in C and N contents associated with primary particles among some sites. The C and N contents within each primary particle from R82GT were equal to or greater than that from UMG (Table 2 ). In general, C and N contents associated with primary particles ranked in the order clay > silt > sand for all sites except R87G. Accordingly, C/N ratios associated with primary particles were lower for clay than sand or silt fractions.


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

  		Table 2. Mean C and N contents in sand, silt, and clay fractions in the three sites reclaimed in 1978 (R78G, grass-covered), 1982 (R82GT, grass and trees), and in 1987 (R87G, grass), and an unmined reference (UMG, grass) at 0- to 15-cm depth.

 
The larger C content associated with clay than with silt and sand fractions has been reported by Balesdent and Mariotti (1996) to be the refractory pools of long turnover time and was also in accord with those reported by Christensen (1996) and Anderson et al. (1981). One year after reclamation of a drastically disturbed soil, 65 to 98% of the total C was reported to be the recalcitrant fraction (Shukla and Lal, 2005). There were only three time increments and the power of the t-test was not very high; still, linear regression between C contents associated with primary particles as the dependent variable and the amount of time since reclamation as the independent variable showed that clay-associated C content increased with increasing amount of time since reclamation (R2 = 0.81). Higher clay-associated C content for the R78G site than other reclaimed sites also showed that the refractory pools increased with increasing amount of time since reclamation. No clear-cut relationship was observed for the primary-particle-associated N contents and amount of time since reclamation.

Macroaggregate Stability and Diameters
The primary estimates of central tendency (mean and median) remained similar (within 3%) for WSA, GMD, and MWD. The coefficient of skewness was also small (or close to zero) and varied between –1.9 and 0.17 for WSA, –0.22 and 0.17 for GMD, and –0.45 and –0.15 for MWD within experimental sites. The Kolmogorov–Smirnov test indicated that for most WSA, GMD, and MWD values, observed distributions did not differ significantly from a normal distribution (Table 3 ). Low CVs of soil texture, mineralogy, and reclamation procedure were also consistent with a normal distribution. The variability of WSA, GMD, and MWD ranged from low (CV <15%) in R78G, R82GT, and R87G, to moderate (CV = 15–28%) in UMG (Table 4 ). The low variability of macroaggregate properties in reclaimed sites may be due to the low soil disturbance after reclamation, but the CVs were also affected by the means, which were lower for the UMG site. Similar observations for WSA, GMD, and MWD were also reported by Shukla et al. (2004a) for reclaimed minesoils of southeastern regions of Ohio.


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

  		Table 3. Normality test using the Kolmogorov–Smirnov test statistics for aggregate water stability (WSA), aggregate geometric mean diameter (GMD) and mean weight diameter (MWD), and C and N contents of macro- (>2 mm), meso- (2–0.25 mm) and micro- (0.25–0.53 mm) aggregate fractions obtained by wet sieving from three mined sites reclaimed in 1978 (R78G, grass-covered), 1982 (R82GT, grass and trees), and in 1987 (R87G, grass), and an unmined reference (UMG, grass) at 0- to 15-cm depth (P < 0.01) (n = 45 for each site).

 

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

  		Table 4. The mean ± SD and CV for for aggregate water stability (WSA), aggregate geometric mean diameter (GMD) and mean weight diameter (MWD), and C and N contents of macro- (>2 mm), meso- (2–0.25 mm) and micro- (0.25–0.53 mm) aggregate fractions obtained by wet sieving from three mined sites reclaimed in 1978 (R78G, grass-covered), 1982 (R82GT, grass and trees), and in 1987 (R87G, grass), and an unmined reference (UMG, grass) at 0- to 15-cm depth (n = 45 for each site).

 
The WSA, GMD, and MWD means were higher for the three reclaimed sites than for the UMG site (Table 4). Low WSA and MWD in the UMG site were expected, as the reference site was neither mined nor reclaimed and was characterized by shallow and variable depth of soil. The grass cover was patchy and biomass returns to the soil were low. The UMG site was located close to an area reclaimed in 2003 and was probably subjected to traffic. Contrary to our expectations, linear regression analysis of WSA, GMD, or MWD as the dependent variable and time as the independent variable showed that macroaggregate stability and geometric mean and mean weight diameters were not directly related to the amount of time since reclamation (P > 0.05). During soil sampling, we noticed that Site R78G remained wet for a considerably longer period than R82GT and R87G for some high rainfall events. Site R78G also remained frozen for a longer duration than other sites. Thus, no differences among reclaimed sites were the result of the slaking of aggregates in the R78G site caused by prolonged wetting. In addition, C contents associated with primary particles and aggregate fractions were also lower in R78G than R82GT.

Carbon and Nitrogen Contents Associated with Aggregate Fractions
Mean and median values were fairly similar (<10%) for the C and N contents associated with macro- (>2 mm), meso- (2–0.25 mm), and micro- (0.25–0.053 mm) aggregate fractions. The coefficient of skewness for C and N contents in aggregate fractions, however, ranged from 0.16 to 1.80. According to the Kolmogorov–Smirnov test, C contents associated with each aggregate fraction followed a normal distribution but N contents did not (Table 3). The logarithmic transformation of N contents from all three aggregate fractions showed that data for only the microaggregate fraction became normally distributed. No improvement in test statistics was seen for the other two fractions. For ease of comparison, we log-transformed all the N content data associated with the microaggregate fraction for all sites.

In the macroaggregate fraction, the CV for C content was high at each site (Table 4). Among the reclaimed sites, the CV for aggregate-associated C content was largest for R78G. The CV for N content ranged from moderate to high among R78G, R82GT, and UMG (Table 4). Variability of N content was moderate for the youngest site, R87G. Overall, CVs for C and N contents in each aggregate fraction were highest for UMG because of the scanty grass cover and variable topsoil depth, which ranged from 10 to 50 cm. Among the reclaimed sites, the CVs for C and N contents among aggregate fractions mostly increased with increasing amount of time since reclamation.

Land use and management systems have a strong influence on the WSA as well as soil C contents (Rasmussen et al., 1998; Akala and Lal, 2001; VandenBygaart et al., 2002; Shukla et al., 2004b). While the exact reasons for normal distributions and variability as shown above are unknown, a normal distribution for most primary and secondary aggregate properties was expected because of the low variability of extrinsic (management), and intrinsic (soil type, climate, and parent material) factors. The moderate to high variability of C contents in the aggregate fraction could be due to the rapid microbial turnover of labile organic C and N in the early spring, as reported by Cambardella et al. (1994). One may argue that the seasonal trends in microbial biomass can be large, but total biomass C could be small compared with the total C. There is no doubt, however, that such a process can create areas of high and low microbial activity and, along with the variability in root distribution and other soil properties, can influence the WSA and C contents associated with aggregate fractions as well as their spatial variability.

The CV alone cannot distinguish between the extrinsic and intrinsic sources of variation. Classical statistical analysis assumes deviations about the mean to be randomly distributed and ignores the spatial dependence of attributes (Burgess and Webster, 1980; Brus and de Guijter, 1997). The sampling design, a random or unaligned grid, has no bearing on the randomness of an attribute of interest and a spatial attribute is not considered random because of the random sampling (Schabenberger and Pierce, 2002). Measurements of WSA, GMD, MWD, and C and N contents in aggregate fractions were made on a 20- by 20-m grid, and a switch from classical sampling theory to a model-based approach (or geostatistics) (Brus and de Guijter, 1997) allowed investigating the spatial dependence and interdependence of the measured attributes.

Spatial Dependence of Aggregate Properties
The experimental semivariograms for WSA, GMD, MWD, and C and N contents of macro-, meso-, and microaggregate fractions were obtained to a lag of 20 m and a cut distance of 200 m to avoid the error that can result from greater lag distances (Journel and Huijbregts, 1978; Table 5 ). Since most of the measured soil data closely resembled a normal distribution, logarithmic transformation was performed only on the N contents associated with all three aggregate fractions. There was no anisotropy seen in the directional semivariograms for any of the measured soil properties. Therefore, only isotropic models were fit. The isotropic behavior was in accord with the low differences in management practices, soil mineralogy, and soil-forming factors in the study area, although it could also be due to the low number (only three) of transects parallel to the x axis. Measured soil attributes exhibited differences in their spatial patterns at each site. Spatial patterns for some attributes also differed among sites in both magnitude and space (Fig. 3 and 4 ). The spherical model provided good estimates of isotropic semivariogram parameters, as indicated by the small sum of squared deviations between experimental and theoretical semivariograms (Table 5). For GMD from the UMG site, an exponential model produced a slightly lower (0.024) sum of squared deviations (SS) than a spherical model (0.029). The exponential model (SS = 0.01) also produced a better fit than the spherical model (SS = 0.02) for C content associated with the macroaggregate fraction from R87G. These improvements are minor, however, and parameters for the spherical model are reported for ease of comparison across reclaimed sites. For fitting theoretical semivariogram models with an angular tolerance of 90°, the number of sample pairs per lag ranged from 256 (20-m lag distance) to 162 (120-m lag distance).


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

  		Table 5. Parameters for best-fit variogram model for aggregate water stability (WSA), aggregate geometric mean diameter (GMD) and mean weight diameter (MWD), and C and N contents of macro- (>2 mm), meso- (2–0.25 mm) and micro- (0.25–0.53 mm) aggregate fractions obtained after wet sieving 5- to 8-mm aggregates from three mined sites reclaimed in 1978 (R78G, grass-covered), 1982 (R82GT, grass and trees), and in 1987 (R87G, grass), and an unmined reference (UMG, grass) at 0- to 15-cm depth. (Note: Spherical model was the best fit model for most of the soil properties and sites.)

 

Figure 3
View larger version (34K):
[in this window]
[in a new window]

  		Fig. 3. Kriged maps of water-stable aggregation (WSA, %), geometric mean diameter (GMD, mm), mean weight diameter (MWD, mm), and N and C contents associated with macro-, meso-, and microaggregate fractions in the unmined reference (UMG, grass-covered) and three mined sites reclaimed in 1978 (R78G, grass), 1982 (R82GT, grass and trees), and 1987 (R87G, grass).

 

Figure 4
View larger version (8K):
[in this window]
[in a new window]

  		Fig. 4. Experimental (symbol) and fitted (line) semivariograms of C contents associated with macroaggregate fractions in the unmined reference (UMG, grass-covered) and three sites reclaimed in 1978 (R78G, grass), 1982 (R82GT, grass and trees), and 1987 (R87G, grass) (note: y axis of the figures is at different scales).

 
The semivariance usually increases with distance between sample locations, or lag distance, to a constant value or sill (total variance) at a given distance, known as the range of spatial dependence. Semivariogram ranges depend on the spatial interaction of soil processes affecting each property at the sampling scale used (Trangmar et al., 1987). Theoretically, the semivariance at h = 0 is equal to zero but the experimental semivariogram frequently exhibits a discontinuity known as the nugget variance. Nugget variance represents the random variability at the sampling scale and reflects the relationship between the size of the sample and the intersampling distance (Russo and Bresler, 1981). All variogram models of WSA, GMD, MWD, and C and N contents in macro-, meso-, and microaggregate fractions from each site showed a positive nugget effect (Table 5).

Overall, nugget variances ranged from low (or 0.01, 13% of the sill) to high (or 19.44, 68% of the sill); however, they were moderate for most of the attributes (<35%). Thus, for most properties, little variation was present at distances shorter than the first lag (20 m) of the semivariogram. Pure nugget effect was obtained for GMD and MWD in the R87G site, C content in meso- and microaggregate fractions in the UMG site, and N content in meso- and microaggregate fractions in the R82GT site and suggested that spatial dependence may occur at distances smaller than the measurement scale (<20 m). For GMD and MWD in the R87G site and N contents associated with meso- and microaggregate fractions in the R82GT site, the greatest degrees of irregularity were also observed (Fig. 3). High nugget variance seemed to smooth out mean WSA, GMD, or MWD among reclaimed sites.

Stochastic determination of variations in a soil property using a semivariogram function always implies that the random field has similar properties in different parts of the domain. If an experimental semivariogram reaches a sill value asymptotically, it implies that the random field is second-order stationary (Schabenberger and Pierce, 2002). Most variogram models of WSA, GMD, MWD, and C and N contents in macro-, meso-, and microaggregate fractions showed a positive sill, and therefore without further testing we assumed that the random fields were second-order stationary (Table 5). The stabilization of variance also means that there is no drift in the data, and the mean of the variable is constant. Sill values of the theoretical semivariograms for all measured attributes were similar to the sample variance, indicating a general absence of trends (Trangmar et al., 1987). A high sill value is an indication of higher variability of the regionalized variable being examined. Among the three reclaimed sites, the sill value for WSA was highest in the R87G site. The higher dispersion variance for WSA in the R87G site than other reclaimed sites was largely due to the variability of the grass cover and the shorter time since reclamation. For MWD, GMD, and C and N contents in aggregate fractions, however, dispersion variance was higher in R78G than in other reclaimed sites. Higher dispersion variance in the R78G site than other reclaimed sites is consistent with the low level of soil disturbance (i.e., no tillage).

The range of influence is considered to be the distance beyond which observations are not spatially correlated. This value is important for finding the minimum sampling distance for ensuring independence. Range defines the initial search neighborhood for kriging and is scale dependent. A definite and positive range and no pure nugget (range = 0) for experimental variograms showed that most attributes were not completely random at the scale of measurement. Among three reclaimed sites, range values did not show any consistent trends for the various soil properties among sites. The only exception was C content associated with all three aggregate fractions, and range values were similar across reclaimed minesoils. Thus, the amount of time since reclamation had no definite influence on the range of most soil properties.

The range for most measured properties varied from 28 to 176 m (Table 5). We did not expect such a high variation in the range because of the low variability of intrinsic (mineralogy, climate) and extrinsic (management) factors across reclaimed sites. The range values were always larger than the minimum sampling distance, and the soil sampling scheme used in this study was adequate for most attributes (Trangmar et al., 1987). Neglecting three very high range values (>100 m) in Table 5, two from the UMG site and three from the R78G site, a mean range value of 43.5 m was obtained, which was much higher than those (1–10 m) reported by Schafer (1979) for some soil attributes from reclaimed sites in southwestern Montana. In contrast to this study, Keck et al. (1993) reported that soil properties were not spatially dependent in reconstructed minesoils of Montana. The important difference between their study and this study is the size of the grid for soil sampling. Montana reconstructed minesoil was sampled at a 100-m interval, which was five times the sampling interval in this study (20 by 20 m). In addition, unlike this study, sites sampled in Montana were a mosaic of individual reclaimed sites, which together made up the larger reclaimed areas. These sites were also reclaimed in different years. This discrete nature of individual reconstructed minesoils within the larger area violated the underlying assumptions of kriging (Keck et al., 1993). Our study demonstrates that sampling individual fields at smaller separation distances can reveal the inherent spatial dependency of soil properties and kriging can be used to estimate the value at an unsampled location along with a kriging variance.

Very high nugget values tend to mask the differences in measured values in space by decreasing the degree of spatial structure (least continuity) (Schabenberger and Pierce, 2002); therefore, relative nugget effect has been used to classify the spatial dependence of soil properties (Cambardella et al., 1994). The lower the relative nugget effect, the stronger is the spatial dependence. This approach has been used as well as criticized in the literature because the basis for classifying the level of spatial dependence is essentially the relative nugget effect, while the effects of range are not taken into account. We used this approach to classify the spatial dependence for all the measured properties (Table 6 ) but will discuss those aggregate properties for which the range values were similar.


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

  		Table 6. Relative nugget effect for aggregate water stability (WSA), aggregate geometric mean diameter (GMD) and mean weight diameter (MWD), and C and N contents of macro- (>2 mm), meso- (2–0.25 mm) and micro- (0.25–0.53 mm) aggregate fractions obtained after wet sieving 5- to 8-mm aggregates from three mined sites reclaimed in 1978 (R78G, grass-covered), 1982 (R82GT, grass and trees), and in 1987 (R87G, grass), and an unmined reference (UMG, grass) at 0- to 15-cm depth.

 
In the R87G site, the range was similar for all measured soil properties (45.7 ± 10.5 m) except N contents from all three aggregate fractions (Table 5). Relative nugget effect ranged from high to moderate but was lowest for most of the soil properties from Site R78G (Table 6). The lowest relative nugget values supported our hypothesis that spatial dependence became stronger with an increasing amount of time since reclamation. In Site R82GT, the range values were similar (64.8 ± 11.5 m) among GMD, MWD, and C contents associated with all three aggregate fractions, and spatial dependence was mostly moderate. In the R87G site, range values were similar (47.2 ± 7.0 m) for WSA and C and N contents associated with all three aggregate fractions, and spatial dependence ranged from high to moderate. Since range values for all these measured soil properties were widely different, no attempts were made to characterize the spatial dependence using relative nugget effect across reclaimed mine sites.

Range values for C contents associated with all three aggregate fractions were similar (51.4 ± 7.2 m) across the three reclaimed mine sites and clearly showed a weaker relative nugget effect with increasing amount of time since reclamation. A high coefficient of determination (R2 = 0.66) and a significant linear relationship (P < 0.001) was obtained between the amount of time since reclamation and the relative nugget effect for aggregate-fraction-associated C contents (pooling all data), which supported our hypothesis that spatial dependence increases with time since reclamation. Range values for WSA, GMD, MWD, and N content associated with different fractions were different among various reclaimed sites. Therefore, we believe that the relative nugget effect may not provide enough evidence to characterize the spatial dependence of these properties across reclaimed sites. Similar observations can be made while comparing the relative nugget effect for MWD and N content associated with all three fractions among reclaimed and unmined reference sites.

In spite of some clear-cut relationships between the spatial dependence of aggregate-fraction-associated C content and the amount of time since reclamation, repeated measures in each of the sites at the measurement scale as well as at additional sites are needed for assessing the exact influence of time on the spatial dependence of C contents. Cambardella et al. (1994) attributed high spatial dependence to the variability of intrinsic factors of soil formation and mineralogy. Although mineralogy was similar within sites, soil-forming processes are variable. In addition, small differences in topsoil depth and compaction during reclamation caused some heterogeneity, especially in the years following reclamation, which grew with time because no efforts were made to reduce the heterogeneity by altering the management options.

Overall, the study showed that the unmined reference site had large variability and reclamation reduced the variability initially at the scale of measurement (human influence); however, as time since reclamation increased, the soil variability increased further (intrinsic and extrinsic factors). Results of the study also elucidate that spatial dependence of aggregate-associated properties is not a function of land use and soil C accretion rate, which was highest in all three aggregate fractions from the R82GT site. The success of these sites to provide practical information and advance scientific knowledge of variability of minesoil properties at the local scale is attributed to: (i) reclamation in accord with the SMCRA act, (ii) no disturbance in the site maintenance for as much as 26 yr, (iii) low variability of intrinsic factors among sites, and (iv) availability of an undisturbed reference site for a comparative assessment of temporal changes in soil quality. Since soil properties exhibit variability at multiple scales, future efforts can be made at quantifying the variability for scales greater than the local scale.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
For primary particles, an inverse relationship was obtained between particle size and C content, and clay-associated C content increased with increasing amount of time since reclamation. Despite similar particle size distribution, mineralogy, and climate, significant differences were obtained for C and N contents among some sites. The site reclaimed in 1982 and under grass and tree cover had the highest mean C and N contents in all three primary particles, as well as C contents in all three aggregate fractions. Among reclaimed sites, the statistical variability of C and N contents associated with aggregate fractions increased with increasing time since reclamation. Overall, spatial dependence ranged from moderate to high for most soil properties in each site. The relative nugget effect demonstrated that C contents associated with all three aggregate fractions increased with increasing time since reclamation. The higher WSA and C contents associated with all three aggregate fractions for reclaimed sites than the reference site suggest that reclaimed sites are on a different successional trajectory than the unmined site. The improved aggregation and C accretion of reclaimed sites also demonstrated the usefulness of reclamation with topsoil application on soil quality enhancement.


   ACKNOWLEDGMENTS
 
We thank New Mexico State University Agriculture Experiment Station, Las Cruces, for support. Special thanks to Brian Cox for help during the field work. We are thankful to the Department of Energy–NETL for sponsoring the research. All the views and opinions expressed in this report are that of the authors and DOE is not responsible for their accuracy or completeness. We also thank American Electrical Power for allowing us to collect soil samples from their property.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.

Received for publication January 6, 2006.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
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 Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Shukla, M. K.
Right arrow Articles by VanLeeuwen, D.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Shukla, M. K.
Right arrow Articles by VanLeeuwen, D.
Agricola
Right arrow Articles by Shukla, M. K.
Right arrow Articles by VanLeeuwen, D.
Related Collections
Right arrow Soil Organic Matter
Right arrow Spatial Variability
Right arrow Other Soil Types

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