| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a The Schiermeier Olentangy River Wetland Research Park, School of Natural Resources, The Ohio State University, 352 West Dodridge Street, Columbus, OH 43202
b Present address: School of Civil Engineering and Environmental Science, The University of Oklahoma, 202 West Boyd Street, Norman, OK 73019
* Corresponding author (anderson.1093{at}osu.edu)
Received for publication May 4, 2005.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: EM, emergent • ORWRP, Olentangy River Wetland Research Park • OW, open water
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
When terrestrial soils become flooded, there are several biogeochemical transformations that can occur over different time intervals. After only a few days of flooding, oxygen in the soil column becomes depleted and microbial activity will be dominated by facultative and strict anaerobes (Mitsch and Gosselink, 2000). Soil colors will become darker as reduced Fe and Mn are transported out of the soil column during flooded conditions. Soils with chroma values of 
2 are used to indicate the presence of hydric soil conditions (Tiner, 1999). Flooding also influences soil P availability due to its release into the water column (Sanyal and De Datta, 1991). Longer-term changes in soil condition are influenced by the buildup of soil organic matter at the surface caused by the reduced rate of decomposition. The accumulation of soil organic matter has been identified as an indication of soil maturity in created wetlands because of the time required for it to develop (Craft, 2001; Nair et al., 2001). Several important biogeochemical processes associated with wetlands (e.g., denitrification) are dependent on adequate soil carbon being present (Mitsch and Gosselink, 2000).
While several studies have evaluated temporal changes in the soil organic matter of created wetlands (Bishel-Machung et al., 1996; Nair et al., 2001; Anderson and Cowell, 2004), few have evaluated the spatial patterns that occur over time. This is partially because of the explicit sampling design that is required to evaluate spatial dynamics. Spatial patterns associated with natural wetland soil characteristics such as P enrichment in the Everglades (DeBusk et al., 2001) and P-sorption capabilities in North Carolina floodplains (Bruland and Richardson, 2004) have been studied. Changes in how soil properties were distributed within a created wetland were observed after 2 yr in response to flooding at the Des Plaines River wetlands near Chicago, Illinois (Fennessy and Mitsch, 2001). They found that spatial variability of soil organic C and exchangeable nutrient concentrations as measured by the range of autocorrelation influence declined after 2 yr of flooding.
At the Olentangy River Wetland Research Park in Columbus, OH, two 1-ha experimental wetlands were constructed in 1993. One was planted with native macrophytes and the other was not. Extensive soil surveys were conducted at the two wetlands in August 1993 (after excavation but before flooding), September 1995 (18 mo after flooding), and May 2004 (10 yr and 2 mo after flooding) to evaluate changes in response to flooding. Over the last 10 yr, several investigations have identified the rapid development of a sediment layer, and significant increases in soil organic C, Ca, Fe, P, and total C (Nairn, 1996; Liptak, 2000; Harter and Mitsch, 2003). Initial accumulations were attributed to the autochthonous production of dense algae mats (Wu and Mitsch, 1998) and allochthonous import of sediment (Harter and Mitsch, 2003). After the third year, both wetlands had developed significant cover by macrophyte communities, which are now considered the primary contributor to soil organic matter. This study represents the first to examine changes in soil condition at the two wetlands since its creation along with changes in spatial variability. Detailed descriptions of the hydrologic, biogeochemical, and ecological patterns of these experimental wetlands are given by Metzker and Mitsch (1997), Mitsch et al. (1998)(2005a, 2005b, 2005c), Kang et al. (1998), Koreny et al. (1999), Nairn and Mitsch (2000), Spieles and Mitsch (2000a)(2000b, 2003), Ahn and Mitsch (2002), Anderson et al. (2002), Selbo and Snow (2004), and Zhang and Mitsch (2005).
The first objective of this study was to compare soil data collected in 1993, 1995, and 2004 to evaluate changes at the soil surface that have occurred as a result of the created riverine-wetland conditions. Given the high productivity and flooded conditions, we hypothesized that the wetland soil surface has substantially increased in its concentration of organic matter and nutrients associated with organic matter (organic C, N, and P, and exchangeable cations). Our second objective was to compare the spatial patterns of soil organic matter concentrations in samples collected in 1993, 1995, and 2004. Starting with the antecedent soil conditions (1993), we expected to see an increase in the concentration and spatial variability of organic matter.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Sampling methods described below are specific to the 2004 sampling period, but were designed to be consistent with methods used in 1993 and 1995 (Nairn, 1996). Water depths were lowered to minimize standing water at each grid point and facilitate soil extraction. At each grid point, soils were collected 0.5 m east of the field marker. Soils were collected using a 10-cm-diameter steel soil-corer, carefully removed, and split into 0- to 8- and 8- to 16-cm sections using a sharp knife. The 0- to 8-cm section was then halved length wise and stored in separate water-tight freezer bags. Because most 8- to 16-cm sections were typically dense clay, this section was split into quarters and two of the quarter-sections were placed in separate plastic freezer bags. Soil remnants were replaced into the sample hole. For each sample, the hue, value, and chroma were determined using a Munsell Color Chart and other visual characteristics were noted. Because of the dense clay consistency of the antecedent soil surface, the development and boundary of the accreted sediment layer was usually apparent. When it was, the depth was measured to the nearest 0.5 cm. Each sample section was placed in a plastic freezer bag and kept in an ice-packed cooler until being returned to the laboratory where samples were refrigerated at 4°C until laboratory analysis.
Physical and Chemical Soil Analyses
One section of each soil sample was weighed and placed in a drying oven at 105°C for 5 d or until constant mass occurred. Soil sections were reweighed to determine soil moisture content and bulk density. The second section of each soil sample was kept in its field-moist, natural condition and completely homogenized by hand. A 30-g subsample of each sample was air-dried at room temperature for 1 d and placed in a drying oven at 60°C overnight. Each soil was then ground using a pestle and mortar, and passed through a 2-mm sieve. Duplicate subsamples (approximately 10 g each) were placed in a crucible, weighed, and ignited in a muffle furnace at 550°C for 1 h. The post-combustion material was reweighed and the duplicates averaged to estimate the percent organic matter of the soil.
Among the samples analyzed for organic matter, a subsample was used to characterize soils for various chemical parameters. For each year, samples (at 0- to 8- and 8- to 16-cm depths) were collected and analyzed from the same grid points. Samples were selected to analyze chemical conditions over an even spatial distribution and to be proportionate among the cover zones. A total of 168 samples (56 per year based on two samples [0–8 and 8–16 cm] collected at 28 grid points) were analyzed for available P by the Bray-P1 extraction (Kuo, 1996), exchangeable K, Ca, and Mg by 1 M ammonium acetate extraction (Warncke and Brown, 1998), and pH (Thomas, 1996). A total of 108 of these samples (36 per year based on two samples [0–8 and 8–16 cm] collected at 18 grid points) were further analyzed for total C by combustion (International Organization for Standardization, 1995; AOAC International, 2002) and total P by digestion with HClO4 and HNO3 followed by inductively coupled plasma emission spectrometry (Sommers and Nelson, 1972).
Temporal and Geostatistical Statistical Analyses
Because several of the soil parameters had unequal variances and could not be transformed to fit a normal distribution, mean comparison of each soil parameter in 1993, 1995, and 2004 was conducted using nonparametric Friedman Two-Way Analysis of Variance of repeated measure with post hoc comparison of years conducted using Wilcoxon Signed Ranks Test. The Friedman test was used to strictly evaluate changes over time at each repeatedly sampled grid point (no intra-annual comparisons were considered); therefore, the potential ramifications of using non-independent data were minimized. All tests were conducted using Systat v.10.2 (Systat Software, 2002). For each statistical test, differences were considered significant at p < 0.05 and highly significant at p < 0.01 with a Bonferroni adjustment for individual Wilcoxon tests.
Changes in the spatial pattern of soil organic matter were evaluated for both wetlands using data collected in 1993, 1995, and 2004. GS+ Software (Version 7.0) (Gamma Design Software, 2004) was used to assess for autocorrelation based on the semivariance of paired groups of data points within each wetland (each wetland was analyzed separately). Isotropic variograms were used to detect semivariance and were constructed using 20-m interval classes over 100-m distances for the 1993 and 1995 data, and 10-m interval classes over 70-m distances for the 2004 data. H-scatterplots were used to detect for outliers or aberrant data that may have had excessive influence on the model parameters (Isaaks and Srivastava, 1989). Variograms consist of a graphical output in which the semivariance is measured at increasingly further interval distances. When autocorrelation occurs, the level of variance between interval classes increases and eventually reaches an asymptote and levels off, representing the extent of autocorrelation (Isaaks and Srivastava, 1989). Characteristics of the variogram graph include (i) the nugget variance (C0), which is the experimental variance unaccounted for by the spatial model; (ii) the sill (C0 + C), which is the total variance as measured at the asymptote of the variogram; and (iii) the range (A0), which is the spatial distance in which autocorrelation is detected.
The spatial structure detected through the variogram was used to conduct kriging analyses, which is an unbiased procedure that uses the modeled spatial relationship to interpolate values between data points (Isaaks and Srivastava, 1989). The interpolated data from each kriging analysis were used to calculate frequency distributions of soil organic matter in Wetlands 1 and 2 for 1993, 1995, and 2004. The interpolated data were also used with the GS+ software to illustrate kriging maps for comparisons between wetlands and years.
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
2, compared to 51% in 1995 (78% at the 0- to 8-cm depth and 24% at the 8- to 16-cm depth), and none in 1993.
|
Subsurface soil conditions (8- to 16-cm depth) had fewer significant changes between years (Fig. 3). No substantial changes between years were detected for percent organic matter (Fig. 3a), exchangeable Ca (Fig. 3e), and exchangeable Mg (Fig. 3g). Percent total C (Fig. 3b) and exchangeable K (Fig. 3f) increased since 1993 but only significantly after 10 yr. Other soil attributes such as available P (Fig. 3d) and soil pH (Fig. 3h) showed temporal changes that were similar to those observed at the 0- to 8-cm depth.
Spatial Characteristics and Changes of Soil Organic Matter
Variogram characteristics using soil organic matter data from 1993, 1995, and 2004 were evaluated for spatial structure (Table 1). For the 2004 data, two outliers from Wetland 1 and two from Wetland 2 were identified through a review of an h-scatterplot and removed because of their excessive influence on the model parameters. Strong autocorrelation was detected for Wetlands 1 and 2 in 1993 and 2004 with no spatial structure detected in 1995. Both wetlands had similar changes between 1993 and 2004, with an overall increase in variance (represented by the sill value, C0 + C) and substantial decreases in the range at which autocorrelation was detected (A0) (Table 1). The range of autocorrelation detected in Wetlands 1 and 2 decreased by 90 and 60%, respectively (Table 1). The proportion of variance explained by autocorrelation was moderate to high in both wetlands, increasing from 0.59 to 0.91 in Wetland 1 and decreasing from 0.99 to 0.57 in Wetland 2.
|
Review of the frequency distributions (Fig. 4) and maps (Fig. 5) of spatially interpolated data showed that soil organic matter at the 0- to 8-cm depth exhibited a fairly even distribution in 1993 and 1995 compared to conditions in 2004. Soil organic matter contents in 1993 and 1995 were similar between wetlands, with Wetland 1 having a slightly broader range in 1993 than Wetland 2 (Fig. 4). Both wetlands showed an increase in the percent organic matter over the 10-yr period and developed concentration ranges that appeared comparable. Evaluating the 2004 field data, soil organic matter ranged from 4.2 to 15.5% in Wetland 1 and 4.5 to 19.1% in Wetland 2. Kriging analysis tends to suppress the range of interpolated values; therefore, extreme high and low measurements tend to be less represented by this procedure. Nevertheless, it was apparent that a wide range of soil organic matter conditions existed for both wetlands in 2004. Based on 0.5% organic matter intervals, the frequency distribution did not exceed 20% of the total at any point for either wetland (Fig. 4). Wetland 2 showed a distribution that was slightly higher with a narrower range compared to Wetland 1. Evaluating the kriging maps (Fig. 5), it is apparent that the amount and distribution of soil organic matter has increased with time. In 2004, both wetlands tended to have the greatest concentrations of organic matter along the wetland periphery (Fig. 5c) in areas associated with the EM zones. Conditions in Wetland 1 were patchier and exhibited a broader range than Wetland 2. It should be mentioned that part of the differences between years may be attributed to the higher intensity of sampling conducted in 2004 (n = 127) compared to 1993 (n = 43) and 1995 (n = 43). However, given the relatively homogeneous conditions observed in 1993 and 1995, it is unlikely that more intensive sampling during those years would have revealed substantial differences in the kriging maps or frequency distributions. The wetlands also showed differences in organic matter concentrations from inflow to outflow. Evaluating the Wetland 2 kriging map, its greatest levels of soil organic matter tended to be in the northern half of the wetland (closer to the inflow) while this trend was less apparent in Wetland 1.
|
|
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Available P concentrations in the created wetland soils were influenced by the release of P in response to flooded conditions. The highest concentrations of available P were observed at the surface and subsurface soils in 1995 and likely reflected the initial soil response to flooded conditions. It has been shown that flooding previously terrestrial soils can cause created wetlands to be a source rather than a sink for P (Newman and Pietro, 2001). Upon submergence, P can be released through the reduction of Fe(III), and after subsequent pH increases, further desorption of P from clays, Al oxides, and Fe oxide surfaces may also occur (Sanyal and De Datta, 1991). In the case of the ORWRP wetlands, available P actually increased 11 to 16% during the first 18 mo after flooding (Nairn, 1996). Since 1995, available P concentrations declined at both surface and subsurface depths and were substantially less than pre-wetland (1993) levels.
In contrast to available P, total P concentrations at the soil surface increased continually. Total P accumulation in created wetlands has been shown to be influenced by a combination of sorption, organic accumulation, sedimentation, and precipitation (Craft, 1997). At the ORWRP wetlands, high algal photosynthesis has elicited significant coprecipitation of CaCO3 and P, particularly in the open water zones where algal productivity is the greatest (Liptak, 2000). This was especially the case between 1993 and 1995 when high algal productivity occurred throughout both wetlands. During this 18-mo period, the average concentration of total P increased approximately 100 µg g–1 (or 60.2 µg g–1 yr–1); however, since 1995 the concentration of total P has only increased by another 120 µg g–1 (or 14.1 µg g–1 yr–1). This can be attributed to the colonization of macrophytes in the EM zone and the overall suppression of algal productivity over most of the wetland area. Also, sedimentation rates have reduced after 10 yr of flooding compared to the initial years of wetland formation (Nairn, 1996; Harter and Mitsch, 2003).
Temporal changes in exchangeable cation concentrations were likely influenced by several factors over different time scales, including short-term mobilization in response to flooding (1993–1995), sedimentation and sorption (1993–2004), and the higher exchange capacity associated with organic accumulation (1993–2004). Concentration of exchangeable Ca at the soil surface was greatest in 1995 and reflects the high deposition of CaCO3 that occurred during this time period (Liptak, 2000). The trend in concentration between years is similar to that exhibited by available P which is not surprising because of the shared biogeochemical processes that influenced both of them. Between 1993 and 2004, the steady increase in concentration of exchangeable K in both surface and subsurface soils was likely influenced by increased soil organic matter accumulation. Exchangeable Mg showed significant but relatively small changes between years which is indicative of its long residence time and conservative nature in response to solubility equilibria (Hem, 1989). Soils also exhibited a continual significant increase in pH between 1993 and 2004. A convergence of soil pH to neutral is the typical response that mineral soils have when they are flooded (Ponnamperuma, 1972) and this appears to be happening at the ORWRP.
Spatial Characteristics and Changes of Soil Organic Matter
The spatial changes seen between 1993, 1995, and 2004 illustrated that soil conditions have become increasingly variable. There are several factors related to wetland morphology that likely contributed to the strong spatial structure detected in 2004 (Johnston et al., 2001). Location within the two cover zones would have had a much greater influence on soil development leading up to the 2004 sampling compared to conditions in 1993 and 1995. Longer-term differences in water level and the frequency of inundation can influence macrophyte colonization (Grace and Wetzel, 1981) and productivity (Newman et al., 1998). Standing water will also influence soil conditions such as oxygen availability. Based on the 2004 soil organic matter maps (Fig. 5), it was apparent that the greatest concentrations of organic matter were detected along the wetland periphery (EM zones) with lesser concentrations in the central portions of the wetland (OW zones), even though the OW zones have remained inundated longer since 1993. Other investigators have also found that created wetland soils supporting emergent vegetation accumulated more organic matter than deep areas that were devoid of vegetation (Shaffer and Ernst, 1999).
Sediment accumulation and depth played a key role in organic matter concentrations at both wetlands. Net sediment accumulation has been similar between wetlands with the greatest accumulation occurring in the OW zones (unpublished data). However, as indicated in the results of this study, the concentration of organic matter was greatest in the EM zones suggesting that these areas are heavily influenced by the annual deposition of autochthonous organic matter. A review of annual vegetation community maps (Mitsch and Zhang, 2004) has shown that some of the areas with the greatest concentrations of organic matter in both wetlands have been dominated by cattail (Typha spp.) (the most productive macrophyte community) over several years. It is uncertain how much movement of organic matter occurs from the EM to the OW zones, but based on the refined quality of the organic matter in the OW zones, much of it appears to be allochthonous material or decomposed algal material, and therefore organic matter accumulation within the two cover zones may be two separate processes.
The 2004 organic matter concentration maps (Fig. 5) also illustrated a trend of decreasing soil organic matter from inflow to outflow in Wetland 2 that may have contributed to the overall spatial structure. Larger macrophyte production has been detected in the northern half of the wetland during previous years (Mitsch et al., 2004) and it is hypothesized that this was elicited by greater nutrient availability closer to the wetland inflow than the outflow. However, there are other circumstances associated with Wetland 2 that may also explain this condition. In 2003, most of the southern section of Wetland 2 was denuded of vegetation by apparent muskrat (Ondatra zibethicus) activity (Mitsch and Zhang, 2004). Along with the physical upheaval of the soil, large sections of Wetland 2 were left without any recruitment of detritial matter in the year leading up to the 2004 soil sampling and this may have reduced soil organic matter concentrations in this area compared to the northern sections.
On a smaller scale, specific topographic conditions also influenced the reported soil organic matter concentrations. In both wetlands, areas sampled near the crest of the OW subbasins tended to have the least amount of sediment accretion and consequently the lowest organic matter concentrations. It appeared that the unconsolidated sediment that tends to accumulate in the OW zones may be susceptible to sloughing down into the deeper portions of the subbasin or being transported elsewhere during high flow. All of these factors contributed to the patchy spatial structure in the two experimental wetlands.
Spatial trends in the antecedent soil organic matter (represented by the 1993 data) showed moderate to strong autocorrelation that had a much broader range of influence compared to 2004 conditions. This was expected from exposed subsurface soils that were developed by broad influencing factors such as climate and geology, rather than the new surface soil conditions incurred at the two 10-yr old wetlands. The lack of a detected spatial structure in 1995 may be attributed to the soil sampling design, specifically the distance between sampling points (typically 20 m apart) which may have been too far apart to detect spatial structure. However, conditions at the wetlands during this time may also have contributed to an overall lack of structure. After flooding the wetlands in 1993, the productivity and deposition of autochthonous metaphyton was identified as the major source of organic matter in the newly forming soil surface (Nairn, 1996; Wu and Mitsch, 1998). Aerial photography from summer 1994 (Mitsch and Zhang, 2004) indicated that algae coverage throughout both wetlands was extensive and may have provided a uniform influence on surface organic matter concentrations. Because of this, it is reasonable to expect that spatial structure was minimal after the initial 2 yr of flooding. After 2 yr of flooding at the Des Plaines wetlands in Illinois, Fennessy and Mitsch (2001) found that the range of autocorrelation decreased for most soil attributes. The most substantial changes were seen in exchangeable P and K; however, a modest decrease in organic C was also detected (196 m in 1988 to 175 m in 1990). They also found that after 2 yr, the combined variance (nugget and sill) had decreased in most soil attributes after flooding. However, after 10 yr at the ORWRP wetlands, we found that variance had increased substantially for surface soil organic matter concentrations.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Related articles in JEQ:
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]
|
|
|
|
 |
|
 L. M. Malecki-Brown, J. R. White, and K. R. Reddy Soil Biogeochemical Characteristics Influenced by Alum Application in a Municipal Wastewater Treatment Wetland J. Environ. Qual., October 24, 2007; 36(6): 1904 - 1913. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. E. Hernandez and W. J. Mitsch Denitrification Potential and Organic Matter as Affected by Vegetation Community, Wetland Age, and Plant Introduction in Created Wetlands J. Environ. Qual., January 9, 2007; 36(1): 333 - 342. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
