| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Univ. of Florida, Institute of Food and Agricultural Sciences, Soil and Water Science Dep., McCarty Hall/P.O. Box 11090, Gainesville, FL 32611
b Duke Univ. Wetland Center, Nicholas School of the Environment and Earth Sciences, Box 90333, Durham, NC 27708-0328
* Corresponding author (GBruland{at}ifas.ufl.edu)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CWs, created wetlands • Db, bulk density • HGM, hydrogeomorphic • IDW, inverse distance weighting • NCDOT, North Carolina Department of Transportation • NWs, natural wetlands • RWs, restored wetlands • SOM, soil organic matter
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
In contrast to NWs, soils of CWs and RWs appear to exhibit much lower spatial variability (Stolt et al., 2000). Several factors contribute to the homogeneity of CW and RW soils. First, wetland creation and restoration often involve the use of heavy machinery to remove topsoil and excavate into subsoil. In this process, soil surfaces are extensively cut and scraped, leaving flat, compacted surfaces with little relief (Clewell and Lea, 1990; Stolt et al., 2000). Second, grading and site preparation activities tend to mix soils in both horizontal and vertical directions, disrupting soil zonation and horizons. Third, the use of uniform fill material or upland topsoil also leads to homogeneous soil conditions. Fourth, RWs are often located on former agricultural land. Long-term agricultural activity homogenizes the topsoil, which is most evident for attributes such as organic matter, nitrogen, and cation exchange capacity (Robertson et al., 1993; Whisenant et al., 1995; Paz-Gonzales et al., 2000). Over time, the combined action of physical, chemical, and biological processes can be expected to generate spatial heterogeneity in CW and RW soils comparable to that of NWs. However, these processes occur over time scales of decades to millennia rather than 5-yr jurisdictional monitoring periods (Mitsch and Wilson, 1996). Comparing the spatial distribution of soils in CWs or RWs to paired NWs is important because a number of researchers have argued that environmental variability (which includes soils, topography, microclimate, etc.) and species richness are positively correlated (Williams, 1964; Jeltsch et al., 1998; Ettema and Wardle, 2002). If the soils of CWs and RWs are more homogeneous than those of NWs, CWs and RWs could be expected to have lower diversity of soil biota and vegetation. Differences in the spatial characteristics of the soils of CWs, RWs, and NWs may indicate disparate controls on populations and processes (Levin, 1992) and imply unsuccessful mitigation in the short-term.
Thus, the objective of this study was to compare the spatial variability of soil properties of CWs or RWs to those of paired NWs across a range of HGM subclasses. We tested two hypotheses: (i) spatial variability of soil properties in riverine wetlands would be structured along gradients running perpendicular to streams, while spatial variability of soil properties in nonriverine wetlands would be structured in patches related to local factors (i.e., microtopography, vegetation); and (ii) soil properties of CWs or RWs would exhibit less spatial variability than soil properties of NWs as prior land-use and mitigation activities tend to homogenize soil properties hypothesized that soil properties of CWs or RWs would exhibit less spatial variability than soil properties of NWs as prior land-use (i.e., clearing, ditching, and agriculture) and mitigation activities (i.e., topsoil removal, excavation, and grading) tend to homogenize wetland soil properties.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
2); mainstem riverine (stream order > 2); nonriverine mineral soil flat; and nonriverine organic soil flat (Brinson, 1993; Cole et al., 1997). Paired plots were used to compare differences in spatial variability of soil properties of CWs or RWs and those of NWs. Both CW–NW and RW–NW pairs were located in similar hydrogeomorphic settings, and in areas of related or identically mapped soil series. All four of the CWs and RWs were between 3 and 6 yr since construction.
|
|
Both plots at the third site, ABC, were categorized as nonriverine mineral soil flats as the majority of the site was part of a Coastal Plain interstream divide and all of the site consisted of mineral soil substrate (Table 1). The agricultural field was restored by constructing impervious ditch plugs, backfilling ditches, recreating surface microtopography with hummocks and hollows, and planting tree seedlings. The NW plot was located in part of the forested interstream area that was not converted to agriculture.
Due to their highly organic soils and location in an interstream divide, both the RW and NW plots at the Dismal Swamp site were classified as nonriverine organic soil flats (Table 1). Unlike the mineral soils of the other three sites, the Scuppernong soils (loamy, mixed, dysic, thermic Terric Haplosaprists) can have organic contents of up to 90%. The site was originally an Atlantic white cedar [Chamaecyparis thyoides (L.) B.S.P.] swamp that had been cleared, ditched, and drained to facilitate silvicultural and agricultural activities. Restoration activities consisted of filling ditches, recontouring soil to establish surface microtopography, and planting tree seedlings.
Soil Sampling
At each site, we identified flat areas in the CW, RW, and NW zones with no obvious elevation gradients. Microtopography in these areas was generally within ±0.5 m from the mean elevation. We established 32 by 32 m plots (0.1-ha) in the CWs or RWs and in their paired NWs. This plot size was chosen because it was the largest square plot that could be fit onto all sites. To allow for a robust geostatistical analysis of the data, we established sampling designs with four transects, separated by 8 m (Fig. 2)
. At the two riverine sites (Rowel Branch and Grimesland), transects ran perpendicular to the direction of water flow, while at the nonriverine sites (ABC and Dismal Swamp), transects were oriented in a North–South direction. Each transect consisted of four centroids, allowing us to establish clusters of samples separated by a range of distances.
|
Cores were collected from the upper 20 cm of the soil in 5-cm diameter plastic sleeves with a piston corer from 9 July to 15 July 2002. Cores were stored on ice until being transported back to the Duke Wetland Center Laboratory. Upon arrival, the cores were extruded from the plastic sleeves and split in half vertically with a sharp knife. Half of the core was oven dried at 105°C for 24 h to determine the Db. Once dried, this half of the core was ground and passed through a 2-mm sieve to remove rock fragments and organic matter. The fine-earth fraction was used to determine the percent SOM by loss on ignition (Campbell et al., 2002) and the soil texture (% clay, % silt, % sand) by the pipette method (Sheldrick and Wang, 1993). Only the percent sand data are reported here as all three parameters were highly correlated. The other half of the core was kept at field moisture and passed through a 2-mm sieve. The wet sieved soil was analyzed for pH (Hendershot et al., 1993).
Statistical and Geostatistical Analyses
As our sampling design included cores that were located in close proximity to each other, the majority of the cores collected at each plot could not be considered independent. Therefore, while we calculated means and standard deviations for each of the four soil properties measured at each plot, we did not compare means with t tests or analysis of variance. Frequency distributions were also produced for each soil property at each plot. The distributions of the values from the CW, RW, and NW plots were plotted on a common scale for comparison purposes. The Shapiro–Wilk test (Shapiro and Wilk, 1965) was used to determine if the frequency distributions had significant departures from normality with Statistica (Version 5.5, Statsoft, Tulsa, OK). Cochran's test (Cochran, 1947) was also calculated with Statistica to test for homogeneity of variances for soil properties from the CW or RW plots and NW plots at each site. Soil properties that exhibited non-normal distributions were log-transformed to better conform to the normality assumption of the Cochran's test (McGuinness, 2002). Variances were considered significantly different when P < 0.05.
The relationships between the soil properties and the x and y coordinates of their measurement location within the sampling plots were determined with the trend surface model:
 | [1] |
The Moran's I analysis (Moran, 1950; Legendre and Fortin, 1989; Cressie, 1993) was used to quantify the degree of spatial autocorrelation that existed among soil cores taken from each plot. This analysis calculates an index of autocorrelation (the Moran's I) among groups of paired samples separated by increasing distances. Computation of this index is achieved by division of the spatial covariation in the data by the total variation. The resulting Moran's I values are in the range from approximately –1 to 1. Positive Moran's I values indicate positive spatial autocorrelation. In other words, similar values (either high or low) are spatially clustered. Negative Moran's I values indicate negative spatial autocorrelation, or that neighboring values are dissimilar. Fluctuating positive and negative spatial autocorrelation is characterized by a checkerboard pattern of the observed feature or process. Moran's I values of 0 indicate no spatial autocorrelation, or spatial randomness.
Moran's I values were calculated for eight equal distance classes from 0 to 16 m with the program Moran (D. Urban, Durham, NC, pers. comm., 2004). For all plots, the average number of sample pairs per distance class was 29, which is very close to the standard of 30 pairs recommended by Legendre and Fortin (1989). Moran's I correlograms were calculated for each soil property measured at each plot. As stationarity and normality are requirements of the Moran's I analysis, all soil properties from each plot were detrended with linear or polynomial regression and the residuals were used for the Moran's I analysis. For soil properties that did not exhibit significant trends in space, we used the raw data rather than the residuals. After detrending and removing outliers all soil properties met the stationarity assumption and exhibited normal distributions according to the Shapiro–Wilk test (Shapiro and Wilk, 1965).
As the sample plots at each site were chosen for their homogeneous topography, we expected that isotropic semivariograms would be acceptable for modeling patterns of spatial variability. We intended to map the distribution of soil properties across the study plots using isotropic semivariogram models in conjunction with ordinary kriging. Ordinary kriging is an interpolation method that estimates values at unmeasured locations from the weighted averages of values from nearby observed locations (Isaaks and Srivastava, 1989). These weighted averages are determined by the semivariogram model parameters. However, the semivariograms had poor fits with the actual semivariance data for the majority of the plots due to violations of the stationarity assumption, relatively small sample sizes, and erratic behavior of the data. Thus, we decided to use inverse distance weighting (IDW) interpolation (Isaaks and Srivastava, 1989). The advantage of IDW is that the weights for each observation are not dependent on the semivariogram parameters. Instead, they are inversely proportional to a power of its distance from the location being estimated. Exponents between 1 and 3 are typically used for IDW, with 2 being the most common (Gotway et al., 1996). Tests with different IDW exponents indicated that 2 was optimal as predicted values generated with this exponent showed the best fit with actual data in cross validation tests. GS+ software (Version 5, GammaDesign, Plainwell, MI) was used to generate the IDW maps and perform the cross validations.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
|
The frequency distributions for SOM from Rowel Branch, Grimesland, and Dismal Swamp overlapped, while there was no overlap in the frequency distributions from ABC (Fig. 3). The distribution of SOM from the Rowel Branch RW exhibited a significant deviation from normality. Mean SOM values were lower in CW and RW plots than in their NW pairs across all sites (Table 2). The Dismal Swamp NW plot had an extremely high mean SOM content of 80.8%. Despite being lower than the Dismal Swamp NW plot, mean SOM at the Dismal Swamp RW plot (44.9%) was higher than the mean SOM contents of the other three NW plots. The Grimesland CW plot had the lowest mean SOM content (2.4%). The Cochran's test indicated that the SOM variances were significantly lower in CW or RW plots than in NW plots for Grimesland and ABC, while the opposite was true at Rowel Branch and Dismal Swamp (Table 3).
A number of previous studies have also reported considerably lower SOM in CWs and RWs than in NWs (Bishel-Machung et al., 1996; Galatowitsch and Van der Valk, 1996; Shaffer and Ernst, 1999; Campbell et al., 2002; Bruland et al., 2003). The lower SOM in the CW or RW than in the NW plots may be due to increased decomposition during prior land-uses or to the removal of organic-rich topsoil during creation/restoration. Low levels of SOM in CWs and RWs may hinder the development of vegetative and microbial communities that are critical to wetland function.
There was considerable overlap in the frequency distributions of pH for the CWs or RWs and NWs at Rowel Branch and Grimesland (Fig. 3). In contrast, there was much less overlap for Dismal Swamp, and no overlap for ABC. The distributions of pH in the NW plot at Rowel Branch, both plots at Grimesland, and both plots at Dismal Swamp were significantly skewed. Soil pH values were higher in the CW or RW plots than in their NW pairs for all sites but Rowel Branch site. The Cochran's test indicated that pH variances were significantly lower in CW or RW plots than in NW plots for Rowel Branch and Dismal Swamp, were equivalent at Grimesland, and were significantly higher in the RW than in the NW plot at ABC (Table 3).
While frequency distributions of sand content for Rowel Branch showed considerable overlap, there was only minor overlap for sand content at Grimesland and ABC (Fig. 3). Sand content at Dismal Swamp was not measured due to the high organic content. Across all sites at which it was measured mean sand content was higher in the CW or RW plots than in their NW pairs. Furthermore, at the Rowel Branch site, mean sand content in the RW plot was more than double that of the NW plot. According to the Cochran's test, the variance of sand content was significantly higher in the RW than in the NW plot at Rowel Branch, and was significantly lower in the RW than in the NW plot at ABC.
Bishel-Machung et al. (1996) and Stolt et al. (2000) also found higher sand content in CWs and RWs than in NWs. Creation and restoration activities involve removal of fine-textured surface soils during excavation, loss of fine-textured materials by erosion during and after construction, and use of coarse-textured soils as fill. Soils comprised primarily of sand-sized particles, such as those found in three of the CWs or RWs in this study, generally have lower water and nutrient retention capacities as well as higher permeability and porosity than soils comprised of finer silts and clays that were characteristic of the NWs (Stolt et al., 2000). These textural differences may affect microbial diversity. For example, considerably lower microbial diversity was found in sites with sandy loam vs. organic soils (Øvreås and Torsvik, 1998) suggesting that replacing the organic-rich soils of NWs with coarse-textured soils of CWs and RWs may not result in the replacement of microbial diversity or community structure of the original soil.
While numerous studies have compared mean values for soils of CWs, RWs, and NWs, few have compared frequency distributions or variances of CW, RWs, and NWs. Our study shows that there were a number of cases for which the frequency distributions for soil properties of the CW or RW plots and NW plots did not overlap, such as Db at ABC and Dismal Swamp, SOM at ABC, and pH at ABC. This indicated that there were major differences in the edaphic conditions at these paired sites. However, for other cases, such as Db and pH at Rowel Branch, there was considerable overlap between the two frequency distributions. For pH at Grimesland, the two frequency distributions were almost identical. In addition, certain soil properties such as Db, SOM, and especially pH, exhibited highly skewed frequency distributions. This suggested that prior land-use and mitigation activities have the potential to increase, decrease, or have no effect on patterns of spatial variability in the soil properties of CWs and RWs.
Trend Surface and Moran's I Analyses
It is important to point out that the sampling design used in this study may not have provided sufficient analytical power for the trend surface or Moran's I analyses to detect differences in spatial patterns between the different site pairs or across the HGM subclasses. Alternatively, the patterns detected by the trend surface analysis may have been statistically significant but not ecologically meaningful (Legendre and Fortin, 1989). Despite these caveats, the trend surface and Moran's I analyses revealed a number of significant patterns in the spatial variability of the soil properties across the plots.
The results of the trend surface analysis revealed that linear gradients were present in seven cases, more complex nonlinear trends were present in eight cases, and no significant trends were present in the other 15 cases (Table 4). Significant trends were encountered in both CWs or RWs (seven cases) and NWs (eight cases). Nonlinear trends were present in CWs or RWs for three cases while they were present in NWs for five cases. Trends were present in riverine plots for eight cases and in nonriverine plots for seven cases.
|
At the ABC site (nonriverine mineral soil flat), there were significant trends in the SOM in the RW and NW plots, as well as with pH in the RW plot. Soil organic matter decreased monotonically in the y direction in the RW plot, while it increased monotonically in the y direction at the NW plot. At the Dismal Swamp site (nonriverine organic soil flat) there were significant nonlinear trends in Db and SOM in both the RW and NW plots. Overall, there were trends in soil properties in the CW or RW plots and NW plots at each one of the paired sites, although trends were most pronounced at the Rowel Branch and Dismal Swamp sites.
The trend surface analysis revealed that despite our efforts to locate plots in areas with homogenous topography and to avoid any obvious elevation gradients, subtle linear and nonlinear gradients were present in soil properties of each plot. The differences in trend patterns between riverine CWs or RWs and NWs may be explained by the fact that overbank flooding in NWs is a process that occurs during individual flood events, over seasons, and across decades. It may be unreasonable to expect trends in soil properties of CWs and RWs will immediately approximate those of NWs. The lack of consistency in trend patterns suggested that it may be difficult to generalize about spatial distributions of soil properties across riverine NWs as individual sites may have unique trends due to subtle differences in topography and hydrodynamics. Research has shown that soil properties of riverine sites not only exhibit large-scale spatial structure corresponding to elevation gradients running perpendicular to the stream, but also small-scale structure corresponding to local microtopography (Darke and Walbridge, 2000; Gallardo, 2003).
Contrary to our hypothesis, nonriverine wetland sites also displayed significant linear trends in basic soil properties. Higher-order trend surface models indicative of more complex spatial structure driven by microtopography or vegetation did not appear to be more appropriate for the nonriverine sites than for the riverine sites. The coarse-scale trends in the RW and NW plots from the nonriverine Dismal Swamp site appeared to be more similar to those of the RW and NW plots from riverine Rowel Branch site, while the trends in the RW and NW plots at ABC appeared to be more similar to those of the CW and NW at the riverine Grimesland site. Thus, it also may be difficult to generalize about coarse-scale spatial trends in riverine vs. nonriverine sites, as no consistent differences were observed in the plots sampled in this study. These results also hinted at the site-specific nature of spatial variability, in which unique geologic, hydrologic, vegetative, and land-use histories may interact to create unique patterns of spatial variability.
After removing the coarse-scale trends from the data, we used the Moran's I analysis to detect any residual fine-scale spatial structure in the CW, RW, and NW plots across the HGM subclasses. We present two examples that illustrated the range of outcomes from the Moran's I analysis (Fig. 4) . The first example is from the Moran's I correlograms for pH in the RW and NW at Rowel Branch (Fig. 4a). In the RW plot, pH was randomly distributed, with no significant positive or negative spatial autocorrelation at any of the lag classes. Conversely, at the NW plot, pH values with lag distances less than 4 m displayed significant positive autocorrelation (Fig. 4b). In addition, pH values separated by 4–6 m displayed significant negative autocorrelation. pH values separated by distances greater than 6 m displayed no further autocorrelation. These results indicated a significant fine-scale variability in the distribution of soil pH at the NW plot that was absent in the RW plot. Similar results of significant autocorrelation in the NW plot but not in the paired CW or RW plot were observed for Db at Grimesland and ABC, and for pH at ABC and Dismal Swamp.
|
Instead of displaying all 30 correlograms, we stratified the data by soil property and summed the significantly autocorrelated Moran's I values by CW (or RW) or NW (Fig. 5) . For Db, SOM, and pH there were a total of 32 possible significant Moran's I values (4 plots x 8 lag classes = 32). The number of lag classes exhibiting significant Moran's I values was divided by the total number of lag classes to determine the percent of total possible significant Moran's I values for each soil property (Roberts and Jones, 2000). The percent occurrence of Moran's I values was lower in CW and RW plots than in NW plots for three of the four soil properties measured. Specifically, autocorrelation for Db in CW and RW plots was approximately three times lower than that of the NW plots, autocorrelation for SOM in CW and RW plots was almost half that of the NW plots, and autocorrelation for pH in CW and RW plots was ten times lower than that of the NW plots. The trend was reversed for sand content, which had slightly greater autocorrelation for sand in the CW and RW than in the NW plots. Moran's I values were significantly autocorrelated for 16.4% of all lag distances when summed across all riverine plots, while they were significantly autocorrelated for 15.2% of all lag distances when summed across all nonriverine plots.
|
Spatial Distributions of Soil Properties
Two-dimensional maps produced by IDW provided an additional perspective from which to compare the spatial distributions of the soil properties among the plots. Two selected examples are presented here. In the first example, sand content at the ABC site had a homogeneous distribution in RW plot compared with the much more heterogeneous distribution of sand content in the NW plot (Fig. 6)
. There was still variability in the sand content in the RW plot, but it was expressed across a much finer range than that of the NW. The uniform distribution of sand content may contribute to more uniform drainage and decomposition rates across the plot. In the NW plot, on the other hand, areas with higher sand content may experience rapid drainage and high decomposition while areas with low sand content may experience slow drainage and low decomposition. These results were similar to other studies that have shown that land-use activities such as agriculture, grazing, and surface mining tend to homogenize soil properties (Marriott et al., 1997; Boerner et al., 1998; Paz-Gonzalez et al., 2000). Thus, the more homogeneous soils of the ABC RW may have experienced a decline in soil diversity that is associated with the disappearance of niches with high moisture, SOM, or fine-textured material (Paz-Gonzalez et al., 2000).
|
|
Although the presence of spatial heterogeneity in the soil properties of CW/RW plots was unexpected, it was not implausible. There are a variety of factors that may have contributed to this heterogeneity: (i) the action of physical processes such as sedimentation and erosion are beginning to influence the distribution of soil properties across the plots; and (ii) the action of biological processes such as root growth, litterfall, and bioturbation have affected soil structure and chemistry in localized areas. Furthermore, it is possible that prior land-use and mitigation activities had no effect on spatial structure, or that they may have even increased it. The lack of spatial structure in the CW, RW, and NW plots may be result of variance at scales that were not captured by the sampling design, erratic data that hindered geostatistical extrapolation, or that fact that soil properties did not exhibit significant spatial structure.
Two landscape-scale studies of spatial variability in natural and disturbed ecosystems have concluded that human activities simplify landscape structure (Krummel et al., 1987; Turner and Ruscher, 1988), while another study by Mladenoff et al. (1993) concluded that human disturbances have the potential to either increase or decrease spatial heterogeneity depending on the parameter and spatial scale being examined. In the case of the sites sampled in this study, human activities may have reduced variability in certain soil properties at some sites while increasing variability in certain soil properties at other sites. For CW and RWs, mitigation should attempt to recreate or restore not only the mean values for basic soil properties, but also the appropriate variability associated with those mean values. While reducing the variability in edaphic conditions at CWs and RWs in comparison to NWs is not ideal, neither is increasing variability to levels that greatly exceed those of NWs. The ability to mimic the complexities of natural ecosystems in our human-modified ecosystems may be the next great challenge of ecological restoration (Mladenoff et al., 1993).
To our knowledge, this is the first study to document the differences in the variability of soil properties from CWs and RWs and paired NWs. While patterns of variability were often considerably different in CWs and RWs vs. NWs these differences were not consistent across riverine vs. nonriverine wetlands or within HGM subclasses. These results suggested that our two hypotheses were oversimplifications of the complex patterns of spatial variability in the soil properties of CW, RW, and paired NWs in the Southeastern Coastal Plain. Despite the inconsistent results of this study, we recommend that sampling schemes in wetlands be designed to capture spatial variability of soil properties to extrapolate soil properties across sites with greater accuracy. The additional sampling and labwork required for spatially explicit research is both worthwhile and necessary to further our understanding of edaphic dynamics in these ecosystems. Results from this and future spatial studies will be critical in providing insights on how to better reproduce patterns of natural variation in CWs and RWs.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The Moran's I analysis indicated that prior land-use and mitigation activities may erase fine-scale spatial structure in CW and RW plots. This analysis also indicated that land-use and mitigation activities randomized rather than homogenized fine-scale spatial structure in soil properties. This may occur as excavation, grading, and earth-moving activities mix soil patches horizontally and soil horizons vertically. Fine-scale spatial randomness is most likely better for wetland function than fine-scale homogeneity, as randomly distributed soil properties would allow for more vegetative diversity and a wider range of edaphic conditions.
As hypothesized, certain soil properties exhibited lower spatial variability in CWs and RWs than in NW plots. This implied that prior land-use and mitigation activities homogenize wetlands soils. Replacing the heterogeneously distributed soils of NWs with more homogeneously distributed soils of CWs and RWs may not result in functional equivalency. Contrary to our hypothesis, a number of soil properties, such as SOM, actually exhibited greater spatial variability in CW and RW plots than in their NW counterparts. The presence of spatial variability in soil properties from the CWs and RWs may be due to the action of physical and biogeochemical processes or to the fact that prior land-use and mitigation activities may actually increase the variability of certain soil properties. Other soil properties did not exhibit spatial structure in CW, RW, or NW plots. The lack of spatial structure in CW or RW plots and NW plots may have been a result of variability at scales not captured by our sampling design, erratic data, or the fact that soil properties did not exhibit spatial structure.
These results indicated that patterns of spatial variability of basic soil properties in CWs, RWs, and NWs were inconsistent, although they appeared to be influenced by a variety of factors including HGM setting, prior land-use, and mitigation activities. Further research is needed to quantify and understand patterns of spatial variability in soil properties in CWs, RWs, and NWs. A better understanding of these patterns will help us to incorporate variability into CW–RW design and to promote conditions that will allow for appropriate variability to develop, ultimately leading to improvements in creation and restoration of functional wetlands.
|
ACKNOWLEDGMENTS |
|---|
Received for publication November 25, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Grunwald and K. R. Reddy Spatial Behavior of Phosphorus and Nitrogen in a Subtropical Wetland Soil Sci. Soc. Am. J., June 18, 2008; 72(4): 1174 - 1183. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. M. Tarquis, N. R. A. Bird, A. P. Whitmore, M. C. Cartagena, and Y. Pachepsky Multiscale Entropy-based Analysis of Soil Transect Data Vadose Zone J., May 27, 2008; 7(2): 563 - 569. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. L. Bruland, S. Grunwald, T. Z. Osborne, K. R. Reddy, and S. Newman Spatial Distribution of Soil Properties in Water Conservation Area 3 of the Everglades Soil Sci. Soc. Am. J., August 3, 2006; 70(5): 1662 - 1676. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME |
