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

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Báez-Cazull, S. E. Articles by Voytek, M. A. Search for Related Content PubMed PubMed Citation Articles by Báez-Cazull, S. E. Articles by Voytek, M. A. GeoRef GeoRef Citation Agricola Articles by Báez-Cazull, S. E. Articles by Voytek, M. A. Related Collections Wetlands and Aquatic Processes Redox Processes Spatial Variability

TECHNICAL REPORTS

Ground Water Quality

Determination of Dominant Biogeochemical Processes in a Contaminated Aquifer-Wetland System Using Multivariate Statistical Analysis

^a Texas A&M Univ., 3115 TAMU College Station, Texas 77843
^b USGS, 431 National Center, Reston, Virginia 20192
^c USGS, 430 National Center, Reston, Virginia 20192

^* Corresponding author (mcguire{at}geo.tamu.edu).

Received for publication April 4, 2007.

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION Materials and Methods Results and Discussion Conclusions REFERENCES

 
Determining the processes governing aqueous biogeochemistry in a wetland hydrologically linked to an underlying contaminated aquifer is challenging due to the complex exchange between the systems and their distinct responses to changes in precipitation, recharge, and biological activities. To evaluate temporal and spatial processes in the wetland-aquifer system, water samples were collected using cm-scale multi-chambered passive diffusion samplers (peepers) to span the wetland-aquifer interface over a period of 3 yr. Samples were analyzed for major cations and anions, methane, and a suite of organic acids resulting in a large dataset of over 8000 points, which was evaluated using multivariate statistics. Principal component analysis (PCA) was chosen with the purpose of exploring the sources of variation in the dataset to expose related variables and provide insight into the biogeochemical processes that control the water chemistry of the system. Factor scores computed from PCA were mapped by date and depth. Patterns observed suggest that (i) fermentation is the process controlling the greatest variability in the dataset and it peaks in May; (ii) iron and sulfate reduction were the dominant terminal electron-accepting processes in the system and were associated with fermentation but had more complex seasonal variability than fermentation; (iii) methanogenesis was also important and associated with bacterial utilization of minerals as a source of electron acceptors (e.g., barite BaSO₄); and (iv) seasonal hydrological patterns (wet and dry periods) control the availability of electron acceptors through the reoxidation of reduced iron-sulfur species enhancing iron and sulfate reduction.

Abbreviations: PCA, principal component analysis • TEAP, terminal electron accepting process • USGS, U.S. Geological Survey • DOC, dissolved organic carbon

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION Materials and Methods Results and Discussion Conclusions REFERENCES

 
BIOGEOCHEMICAL dynamics in sediments are controlled by linked biological, physical, and chemical processes including hydrologic fluctuations, seasonal changes in temperature and biological activity, and rock-water interactions (Aller et al., 1998; Dahm et al., 1998; Donahoe and Liu, 1998; Eser and Rosen, 1999; Kemp et al., 1992; Koretsky et al., 2003; Neubauer et al., 2005). These factors interact to create different hydrochemical facies that change temporally and spatially (Thyne et al., 2004). Variability in type and concentration of organic matter, availability of electron acceptors, chemical composition of the lithology, biological activity, and hydrogeologic conditions of the system will determine fluctuations in the redox reactions occurring in situ (Koretsky et al., 2003) including aerobic respiration, nitrate reduction, iron/manganese reduction, sulfate reduction, and methanogenesis. These reactions occur in sequential order according to thermodynamic energy yields (Champ et al., 1979; Froelich et al., 1979; Megonigal et al., 2004). However, in dynamic environments such as wetlands and anaerobic aquifers, these processes are linked and can coexist reflecting the complexity of the system. In biogeochemical studies, these processes are often interpreted from the analysis of geochemical parameters such as redox couples (e.g., H₂S and SO₄^2–). One caveat with this approach is that many geochemical parameters are affected by multiple hydro-bio-geo-chemical processes. Thus to interpret processes, it is important to determine the correlations between parameters to identify the contributing process(es) for each parameter. These correlations can then give insights into the dominant biogeochemical processes in space and time within a given system.

In a wetland-aquifer system the distribution of redox processes responds to solute transport processes, biological activities, and solid-phase composition, which all can vary on small spatial scales (Hunt et al., 1997; Brune et al., 2000; Kappler et al., 2005). Hydrologic fluctuations, such as aquifer recharge, provide a mechanism for the delivery of electron acceptors or electron donors to a discrete zone. The availability of these transported solutes (e.g., electron donors and acceptors) can enhance microbial activity and in turn result in microbial community shifts. Biological activity is also impacted by seasonal changes in vegetation growth, which exerts an important control on redox processes. For example, in root zones oxygen is introduced increasing the availability of electron acceptors (Neubauer et al., 2005). In addition, vegetation is a primary source of organic carbon to the wetland sediments. Heterogeneities in the solid-phase composition (ex. distribution of Fe-oxide minerals) provide another source of heterogeneity in the distribution of redox processes. Superimposed on the spatial variability in redox processes are temporal fluctuations driven by seasonal dynamics and recharge events (Chapelle et al., 1995; Eser and Rosen, 1999; McGuire et al., 2000; Seybold et al., 2002). Therefore, gathering the necessary data to describe the controls on biogeochemical cycling at appropriate spatial and temporal scales generally results in large, complex datasets that are difficult to interpret (Hunt et al., 1997). Often expected relationships cannot be easily discerned in geochemical datasets due to the effects of multiple processes on a single geochemical indicator. Multivariate statistical analyses can provide details on the correlations between parameters often revealing obscured relationships between hydro-bio-geochemical parameters to potentially give new insights into the processes controlling the variability in the dataset.

Since 1901, when first introduced by Spearman, principal component analysis (PCA) has routinely been applied in a wide variety of fields in the natural and social sciences. Its primary function is to explore complex data sets of many dimensions, collapsing many variables, based on their correlations, to a few factors that explain the observed variability and thereby reveal underlying data structure and highlight relationships between the variables. Although the analysis does not provide a mechanism or demonstrate causality, it does provide a quantitative measure of relatedness of variables to one another that can be suggestive of the underlying processes controlling the variability in the dataset (Thyne et al., 2004; van Helvoort et al., 2005; Kumar et al., 2006; Mathes and Rasmussen, 2006). Recent studies that have used PCA analysis to interpret geochemical datasets include: delineation of groundwater contamination potential (Xie et al., 2005; Mathes and Rasmussen, 2006), identification of hydrogeochemical processes (Liu et al., 2003; Thyne et al., 2004; Kumar et al., 2006), characterization of fluvial deposits (van Helvoort et al., 2005), identification of contaminated aquifer zones (Suk and Lee, 1999; McGuire et al., 2005), and hydrologic effects on trace metal geochemistry (Farnham et al., 2003; van Griethuysen et al., 2005). Generally, variability in water-chemistry parameters has been associated with linked biogeochemical processes such as microbiological reactions, abiotic reactions, macro-biological seasonal activities (e.g., vegetation growth in a wetland), groundwater contamination, hydrologic fluxes, mineral dissolution, anthropogenic influence, recharge area (precipitation), climate, and topography (Thyne et al., 2004). Knowledge of the extent to which these factors are important and the scale at which they occur is essential to understanding elemental cycling in natural systems, and multivariate statistical analysis is the preferred technique for discerning these controls. In addition, mapping of factor scores obtained from PCA is useful for describing the spatial and temporal components of the dataset (Suk and Lee, 1999; McGuire et al., 2005; Mathes and Rasmussen, 2006).

The system discussed in this article consists of a seasonally submerged wetland hydrologically connected to a regional aquifer impacted by municipal solid-waste landfill leachate located in Norman, Oklahoma, USA. Wetlands (both natural and constructed) have been studied for their "water-quality function," specifically their ability to retain and transform nutrients and metals (Hunt et al., 1997). Understanding the dominant biogeochemical processes in wetlands is critical to assess the water-quality function. The most biogeochemically active zone of a wetland is near the surface, where the system is susceptible to changes in temperature, precipitation, infiltration, and nutrient loading (Hunt et al., 1997). In addition, other interfaces with depth have been identified as important zones of biogeochemical cycling in a wetland-aquifer system (Báez-Cazull et al., 2007). From the hydrochemical viewpoint, a wetland functions both as an interface zone for water exchange between groundwater and surface water, and as a biochemical filter through which water quality changes greatly (Dahm et al., 1998). Water from infiltration, migration, and the waste itself produces leachate as moisture passes through the waste in the landfill. Leachate can contain an undesirable mix of toxic organics and inorganic chemicals. Wetlands are complex sediment matrices that are organic-rich and provide chemical and biological conditions conducive to the removal of contaminants from the leachate plume as it moves into and through the system. The complex linkages between the wetland and the underlying leachate-contaminated aquifer make determining the dominant biogeochemical processes a challenge. A previous study on the spatial vertical variations in the system (Báez-Cazull et al., 2007) found sharp geochemical gradients at small (centimeter) scale associated with lithological, hydrological, and chemical interfaces. Subsequent field studies to tease out these relationships have resulted in a large spatial and temporal dataset containing centimeter-scale depth profiles of water-chemistry parameters at a single location over 3 yr. Complex and variable solute transport pathways, heterogeneous distribution of solid phase minerals (acting as electron acceptors, nutrients, etc.), microbial growth and decay, and seasonal variations in temperature, recharge, and macrobiology are all linked controls on the biogeochemical cycling of the system (Cozzarelli et al., 1999; Kostka et al., 2002; Koretsky et al., 2003; Weiss et al., 2004; Breit et al., 2005; Scholl et al., 2005; Báez-Cazull et al., 2007).

This article describes the dominant biogeochemical processes operating in a linked wetland-aquifer system over space (centimeter-scale depth profiles) and time (study duration of 3 yr). Multivariate statistics were used in conjunction with traditional graphical and modeling techniques to explore the more than 8000-point dataset to expose statistically significant relationships among geochemical parameters and evaluate changing processes. The use of PCA for exploring parameter correlations and the mapping of factor scores improved the understanding and interpretation of processes controlling the surface water and groundwater composition affected by seasonal changes and hydrologic fluctuations. This study demonstrates the value of using multivariate statistics to reveal previously obscured biogeochemical relationships and enhance understanding of complex system dynamics.

	Materials and Methods

TOP NOTES ABSTRACT INTRODUCTION Materials and Methods Results and Discussion Conclusions REFERENCES

 
Site Description
The wetland (slough) is situated in the Canadian River alluvial plain in central Oklahoma (Fig. 1 ), and formed in a previous location of the main channel of the Canadian River (Schlottmann, 2001). The geologic setting is characterized by moderately permeable alluvial and terrace deposits with a shallow water table that overlies a Permian shale and mudstone confining unit known as the Hennessey Group (Scholl and Christenson, 1998). The wetland acts as a surface expression of the local water table with some lateral flow during the wet season due to minor inputs upstream. It is fed by groundwater discharge, runoff, and precipitation. Water levels vary seasonally, ranging from approximately 1 m deep in the spring to dry in the summer (Fig. 2 ). Upper sediments have been variably saturated during the summer months. The area surrounding the wetland is densely vegetated, dominated by Phragmites, Leersia, and Phalaris and with at least three species of phreatophytes (willow, cottonwood, and tamarisk) (Christenson et al., 1999). The growing season is from mid-April through October. The wetland is approximately 700 m long and 15 to 25 m wide and is downgradient from the Norman municipal landfill (50–100 m from the edge).

View larger version (55K): [in this window] [in a new window]   Fig. 1. Map of the Norman Landfill site in Oklahoma, U.S., showing the sample location and water table elevations indicated by contour lines collected in 1998 by Scholl and Christenson (1998). Star symbol marks approximate location of USGS (U.S. Geological Survey) well SI 102. Coordinates projected in UTM, zone 14 N, datum NAD 1983.

View larger version (28K): [in this window] [in a new window]   Fig. 2. (a) Water levels recorded for the 3 yr period when the samples were collected. (b) Average monthly precipitation data collected from March through September during 2003, 2004, and 2005. Data obtained from USGS Water Resources of Oklahoma (2007).

View larger version (49K): [in this window] [in a new window]   Fig. 3. Cross-sectional view of the Norman Landfill along section A-A' (as shown in Fig. 1) and larger cross-sectional view of the wetland's conceptual model. The leachate plume and recharge zones are drawn on the basis of chemical measurements made in the aquifer between 1997 and 2002 (from Scholl et al., 2005).

In Situ Measurement
Surface and pore-water samples were collected from the wetland adjacent to the Norman Landfill site in May 2003 (P1-May-03 and P2-May-03), March 2004 (P1-Mar-04, and P2-Mar-04), April 2005 (Pa-Apr-05, Pi-Apr-05), May 2005 (P-May-05), and September 2005 (P-Sept-05) (see Fig. 4 ). Water samples were collected by means of multi-chambered equilibrium dialysis frames ("peepers") (Hesslein, 1976). Custom peepers were designed to span a vertical profile of 75 cm capturing the sediment–water interface as well as interfaces across the wetland-aquifer transition zone with a total of 75 horizontal ports with apertures and spacing of 0.5 cm covered by a Millipore membrane* (Billerica, MA, USA) with a pore size of 0.45 µm.

View larger version (17K): [in this window] [in a new window]   Fig. 4. Location of peeper deployment (all near USGS well SI 102, see Fig. 1). Exact peeper location in May 2005 was not surveyed but was deployed near P-Sept-05. Coordinates projected in UTM, zone 14 N, datum NAD 1983, vertical datum GEOID 99.

Geochemical Analysis
Geochemical parameters collected during the 3-yr sampling period represent indicators of biotic and abiotic redox processes, as well as indicators of landfill leachate measured with the purpose of understanding biogeochemical processes in the wetland-aquifer system. Dissolved oxygen, pH, conductivity, temperature, and redox potential were measured using a 600 XLM YSI Hydrodata multiparameter meter (Yellow Springs, OH, USA). Alkalinity was measured in the field using electrometric titration. Fe²⁺ and H₂S were also measured in the field using the phenanthroline and methylene blue photometric methods (APHA, AWWA, and WEF, 2005). Cations (Na⁺, K⁺, Ca²⁺, Mg²⁺, NH₄⁺) were preserved in 1% metal-grade hydrochloric acid, except for ammonium, which was flash frozen with dry ice, and analyzed by capillary electrophoresis (CE) (Agilent Technologies, Germany) (Báez-Cazull et al., 2007). Some elements were also measured by inductively coupled plasma–mass spectrometer (ICP–MS); these included Mn, Ba, Fe, Na, K, Ca, and Mg collected in 2005. The ICP–MS does not distinguish between species; therefore, cations analyzed by ICP–MS are expressed as elements. Anions (SO₄^2–, Cl^–, and NO₃^–) were preserved in 0.5% formaldehyde and analyzed by CE. Organic acids such as propionate, acetate, butyrate, oxalate, and lactate were flash frozen with dry ice and analyzed in the laboratory by CE. Water samples for measurement of CH₄ concentrations were collected in Glasspak syringes and stored in 25 mL serum bottles flushed with N₂ gas and containing tri-sodium phosphate (TSP) to inhibit microbial degradation (Baedecker and Cozzarelli, 1992). These samples were analyzed by gas chromatography. Water samples for DOC analyses were filtered through a 0.45-µm filter, collected in a baked glass bottle, and acidified with H₃PO₄ to a pH of 2. Dissolved organic C was measured by high-temperature combustion techniques following the method of Qian and Mopper (1996). Formate and benzoate were determined on samples collected raw and kept frozen until analyzed by ion chromatography.

Data Analysis
The data were analyzed with multivariate factor analysis using SAS software JMP (SAS Institute, 1999). The geochemical data arranged in R-mode factor analysis included: acetate, propionate, methane, ammonium, sulfate, pH, Na, K, Ca, Mg, Fe²⁺, Cl^–, HCO₃^–, H₂S, butyrate, Mn, Ba, Fe³⁺(Total iron– Fe²⁺), and DOC. Because the data collected for 2003 and 2004 did not include DOC, CH₄, Ba, Fe³⁺ and Mn, these parameters were excluded from the first PCA, which includes all sampling periods. Nitrate, formate, and benzoate were not included in any of the analyses because over 90% of the data were below detection limits; a practice adopted by other studies (Güler et al., 2002; Farnham et al., 2003; van Helvoort et al., 2005). The complete dataset was tested for normality using the natural logarithm, square root, inverse, and power data transformations. For each of the transformations, the data were tested for normality by evaluating boxplots, normal quantile plots (Q-Q plots of expected normal value vs. observed value), and Kolmogorov-Smirnov (K-S) tests for goodness of fit. The K-S tests resulted in p values less than 0.001 for all the parameters except for alkalinity, which fit a normal distribution with the natural log transformation. For all the other parameters, the null hypothesis for the K-S test (data exhibiting a normal distribution) had to be rejected. As a result of the non-normal distribution, conservative nonparametric statistical tests were chosen. To avoid problems with the differences in scale among the chemical parameters, the data were standardized using z-scores. A nonparametric Spearman's Rho correlation analysis was performed on the standardized data. A pairwise correlation matrix was selected to exclude cases with missing data. A factor analysis was performed on the correlated variables using the standardized dataset.

Factor Analysis
A PCA was performed to determine the factors that best explain the variation in the dataset. Factors were rotated using varimax orthogonal rotation to maximize the relationship between the variables and some of the factors. Factor scores were computed from the selected factors for each case and the variability was plotted by location (depth below surface) and season. Centimeter-scale data were divided into five intervals selected according to previous knowledge of important interfaces and sedimentary layers in the profile (see Fig. 3) (Báez-Cazull et al., 2007). For example, an interval comprised of data from the uppermost 9 cm of sediment (0–9 cm) was grouped to represent processes occurring in the portion of the organic-rich silt layer that contained the greatest concentration of particulate organic matter. Similarly, data from the surface water were grouped together as a separate interval. Grouping depth data into smaller intervals based on site characteristics allowed for a simple graphical output to contextualize the factor analysis. Bars shown in the variability plots indicate the range of the factor scores for that depth interval and each of the dots indicates the factor score for a particular depth from within the group.

	Results and Discussion

TOP NOTES ABSTRACT INTRODUCTION Materials and Methods Results and Discussion Conclusions REFERENCES

 
Correlation Matrix
Spearman's rank order correlation test was chosen as a non-parametric method to estimate the degree to which two variables vary together. In this analysis, the assumption is that correlated variables are affected by a common cause. The resulting correlation matrix is plotted in Fig. 5 and only includes correlations that had a significant p < 0.0001. Because large sample sets can result in low p-values, suggesting a significant correlation when the actual correlation is low, only values having a rho value larger than 0.5 were considered significant.

View larger version (35K): [in this window] [in a new window]   Fig. 5. Spearman's Rho correlation analysis for correlations with a probability less than 0.0001 for all the sampling periods.

Multivariate Factor Analysis
The first factor analysis was performed on 13 variables (acetate, propionate, pH, Na, K, Ca, Mg, Fe²⁺, Cl^–, HCO₃^–, SO₄^–2, H₂S, and NH₄⁺). Principal component analysis was the method chosen for extraction of factors explaining most of the variability and it was performed on correlated variables. Table 1 includes the eigenvalues, which are a measure of the variability of a factor, the percent of the total variance attributed to each factor, and the cumulative percent variance. According to the Kaiser criterion (i.e., those factors that have an eigenvalue larger than unity (Kaiser, 1960)), four factors were determined significant and they account for 75% of the variability in the dataset.

Factor 2 (F2) explained 20% of the variability and is highly correlated with H₂S and pH, and moderately correlated to Fe²⁺ and Cl^– in opposite dimensional space. Sulfide and Fe²⁺ are products of sulfate and iron reduction in anoxic sediments and are attributed to microbial reduction. Sulfate is considered chemically to be metastable at standard Earth-surface temperatures (Nealson, 1997): thus, its reduction to H₂S under these conditions suggests a microbial pathway. At circum-neutral pH, microbial iron reduction is more important in the reduction of iron (III) than abiotic reduction (Lovley et al., 1991). Therefore, F2 was interpreted as anaerobic respiration, specifically, sulfate and iron reduction. The opposite relationship of H₂S and Fe²⁺ suggests that the processes may be occurring in separate locations or, if simultaneous, the products may be precipitating into iron sulfides due to sulfide's high reactivity with Fe²⁺. The Fe²⁺ and Cl^– association may indicate that iron reduction may be happening elsewhere in the system and the Fe²⁺ produced is simply transported to this location along with other landfill-leachate components because these constituents are both present in high concentrations in the Norman Landfill leachate (Cozzarelli et al., 2000). Because of the multitude of processes affecting the Fe³⁺/Fe²⁺ redox couple, determining iron cycling in the system is complex.

Factor 3 (F3) accounts for 17% of the variability and contains the variables Na, Mg, Ca, and K with negative loadings. Na and Mg had the highest loadings on F3, whereas Ca and K exhibit moderate correlations on this factor. This factor was interpreted as a mineral dissolution factor. Other studies have interpreted these parameters as a weathering process (Puckett and Bricker, 1992; Schot and van der Waal, 1992; Wayland et al., 2003) from the dissolution of calcite, dolomite, and potassium feldspar minerals. At the Norman Landfill research site, the alluvial sediments contain potassium feldspars, calcite, dolomite, and clay minerals and the silts have distinct increased abundances of clay and calcite minerals (Breit et al., 2005).

Factor 4 (F4) accounts for 10% of the variability and is highly correlated with SO₄^2– and moderately correlated with Cl^–. Two distinct processes with opposite loadings are interpreted from this factor, plume advection and sulfide oxidation. Chloride, often used as a proxy for leachate transport in this and other systems (Röling et al., 2001; Grossman et al., 2002; van Breukelen and Griffioen, 2004), has a positive loading in this factor, which is interpreted as plume advection process, where contaminated aquifer water containing a high concentration of Cl^– is transported upward into the wetland sediments. Sulfate has a strong negative loading on F4 and is interpreted to be a result of oxidizing conditions. During wet conditions, sulfate is primarily observed in the surface water and has a low concentration in the sediments where microbial respiration depletes it (Báez-Cazull et al., 2007). However, when the water table drops, iron sulfide minerals can be reoxidized providing a new source of SO₄^2– to the system.

Interpretation of Factor Scores
Factor scores were assigned to each case (depths). Scores on each factor were plotted by sampling date and depth to determine the variability associated with the factors (Fig. 6 ) and aid in the interpretation of factors.

View larger version (31K): [in this window] [in a new window]   Fig. 6. Variability chart of factor scores by sampling date and depth. Factor scores were computed from the principal component analysis (PCA) results in Tables 1–2. The bars indicate the range in the factor scores data, the connecting lines indicate the mean for each depth interval, and the dashed line represents the true mean (0). Depths are reported in cm below the sediment–water interface (0 cm).

Figure 6 shows the F2 scores loaded on the positive side to indicate sulfate reduction and on the negative side to indicate iron reduction. The plot of factor scores by depth and date indicate that sulfate reduction is higher in May (Fig. 6) than other times in the year. High factor scores for F2 found in the surface water for May are a product of either sulfate reduction occurring in the surface water or fluxes of sulfide produced in the first 9 cm in the sediments. Though a lesser signal, in March 2004 sulfate reduction is also observed in the upper sediments. Another location where sulfate reduction seems to be occurring throughout different time periods (May, March, and April) is at the depth of 40 to 60 cm. In September 2005, sulfide is not observed, suggesting either decreased sulfate reduction, rapid oxidation of sulfide in this drier month, or precipitation of iron sulfides with the increased levels of Fe²⁺ found in the upper sediments in September.

Iron reduction is most strongly indicated in April 2005, peaking at the depths of 10 to 39 cm below the sediment–water interface. In May 2003, iron reduction and sulfate reduction are indicated in the first 39 cm. The bars in Fig. 6 show an equal distribution of factor scores on the positive side as on the negative side, indicating an equal importance for iron reduction and sulfate reduction in the first 39 cm. This finding indicates spatially simultaneous terminal electron-accepting processes (TEAPs) occurring this month, supporting prior observations (Báez-Cazull et al., 2007). Due to the elevated concentrations of organic acids (data not shown), it is possible that both reactions can be microbially competitive given the availability of electron donors. In addition, our samples integrate conditions at a scale (cm) greater than the scale (µm) of documented heterogeneities in microbial processes that result in microzones of competing anaerobic respirations (Kappler et al., 2005). In March 2004, the same situation occurs at a depth of 10 to 39 cm where the two processes have an equal significance in the distribution of factor scores, indicating simultaneous processes. In May 2005, the Fe²⁺ suggested by F2 found at depths of 21 to 39 cm and 61 to 79 cm may be an indication of iron reduction and/or leachate plume advection. Similarly, in September 2005, iron reduction is suggested at 0- to 20-cm depth although the gradual increase with depth below 39 cm of factor scores on F2 suggests that transport of the leachate plume is the dominant control deeper in the profile.

Mineral dissolution interpreted from a negative factor score in F3 was not significant in March 2004 and May 2005 (Fig. 6). April 2005 and May 2003 exhibit greater mineral dissolution. Mineral dissolution in April is strongly observed throughout the sediment profile except for the 40- to 60-cm depth. In May 2003 mineral dissolution is greater at the surface water and the upper sediments, and a decrease is also observed at the 40- to 60-cm interval. In September 2005, mineral dissolution is observed in the upper and lower sediments. Note that the trend observed in F3 is the same trend as in F2. This suggests that mineral dissolution varies with the anaerobic respiration processes. The exception is the surface water in May 2003 in which mineral dissolution is not associated with anaerobic respiration. In May 2003, mineral dissolution is higher in the surface water and upper sediments and decreases with depth. The parameter likely contributing most to this factor is the high calcium concentrations (1000 ppm) in the surface water and upper sediments. May 2003 exhibited different conditions in the surface water than the other months, which contributed to different water chemistry. The water was stratified with respect to redox potential and dissolved oxygen and a dense layer of senescing aquatic grass vegetation covered the wetland. A previous study (Báez-Cazull et al., 2007) suggests that the dying vegetation in this month contributed to the dissolution of minerals in the surface water.

In F4, the oxidation process plotted as a negative factor score was significant in the upper 39 cm in September 2005. Oxygenation is expected in this region due to a lower water table. The other months did not exhibit an oxidation component. The plume advection component, indicated by Cl^– concentrations, was significant at depths, which suggested a diffusion-gradient component increasing with depth in all sampling periods except for March 2004.

Principal Component Analysis Results Eliminating May 2003
In May 2003, data suggest that a unique incidence of elevated fermentation, as indicated by elevated concentrations of acetate, ammonium, and propionate, occurred at aprox. 30 cm depth and was likely due to localized decomposition of organic matter (Báez-Cazull et al., 2007). To evaluate the effect this event had on the overall variability of the dataset, the May 2003 data were removed and a separate PCA performed on the other months to explore whether or not this event was obscuring significant processes that might be dominant in other periods. Generally, dominant processes remained the same but the relative importance of these processes changed. Most notably, fermentation no longer accounted for the greatest variability but rather was associated with other processes (sulfate reduction) in the second factor and other processes that were once grouped together (oxidation and plume advection) were teased apart. This discrimination allowed for a better evaluation of the linkages between dominant processes.

Five factors that explain 74% of the variability were selected for the factor rotation according to the Kaiser criterion (Kaiser, 1960). The factor loadings for each parameter are shown in Table 3 . Five processes were attributed to the grouped parameters and are discussed below. Figure 7 presents the variability of factor scores for each factor by date and depth, obtained from the rotated factor model.

View larger version (36K): [in this window] [in a new window]   Fig. 7. Variability chart of factor scores by sampling date and depth excluding May 2003. Factor scores were computed from the PCA results shown in Table 3. The bars indicate the range in the factor scores data, the connecting lines indicate the mean for each depth interval, and the dashed line represents the true mean (0). Depths are reported in cm below the sediment–water interface (0 cm).

The source of potassium in April and May 2005 was explored. Cation exchange was evaluated as a possible process for the changes in K concentration. The exchange capacity of the soil was determined at depths of 10 and 30 cm using the equation of Breeuwsma (Amini et al., 2005) where cation exchange capacity (cmol_C/kg) = 0.32 (%clay) + 3.1 (1.72 x %carbon). The clay and organic carbon content of the sediments was determined from cores described in a previous study (Báez-Cazull et al., 2007). At 10 cm the clay content was less (2.8%) than at 30 cm (3.9%) but the organic carbon content was higher at 10 cm (2.7%) than at 30 cm (0.3%). The exchange capacity at depths of 10 and 30 cm was determined using PHREEQC-2 (Parkhurst, 1995) with the solution at these depths for April and May 2005. The results indicate that the soil K exchange composition is 2.1 x 10^–3 moles at 10 cm for April 2005 and 2.6 x 10^–3 moles in May 2005. At 30 cm, the exchange composition is 1.8 x 10^–4 moles and 4.2 x 10^–4 moles for April and May 2005, respectively. This indicates that in the upper sediments, K will have a greater affinity to the particles than in the deeper sediments due to higher concentrations in the upper sediments. However, cation exchange capacity could not be a major control because K exchange capacity was lower than for other cations that were found throughout the sediment profile, such as Ca, Mg, Na, and NH₄⁺, and high concentrations of these cations are often found in the system. This suggests that another process is controlling the K cycle to explain the shifts in concentrations in the sediments. In the first PCA (see Fig. 6) it was noted that the anaerobic respiration factor (F2) mimicked the mineral dissolution factor (F3), suggesting that mineral dissolution is linked to anaerobic respiration.

The second factor, Factor II (FII), includes high negative factor loadings for acetate, sulfide, and propionate (Table 3) and explains 16% of the variability. These associations suggest that the underlying processes controlling this factor are fermentation and sulfate reduction. In the first PCA (including May 2003 data) the factor interpreted as fermentation was the first factor, F1 (explaining 28% of the variability). However, the strength of the signal in the May 2003 dataset obscures the importance of this process in time and space, as well as the relationship between fermentation and sulfate reduction. Once the May 2003 data were removed, it became evident that fermentation and sulfate reduction were dominant in May 2005 in the first 9 cm and at the transition zone at 21 to 39 cm. It is also observed in April 2005 at 40 to 60 cm and in March 2004 in the transition zone (21 to 39 cm) but not in September 2005.

The third factor, Factor III (FIII), with a variability of 13%, includes sulfate and magnesium with high negative loadings and was interpreted as an indication of oxidation. This factor was dominant in September 2005 in the first 20 cm, just as in F4 in the first PCA.

The fourth factor, Factor IV (FIV), with a 12% variability, contains the variables Cl^–, Na, and alkalinity and was interpreted as an indication of plume advection (loaded negatively). With the exception of March 2004, the variability in FIV scores plotted with depth and season shows that the trend for this factor increases with depth, suggesting a diffusion gradient of plume components. This pattern is consistent with previously determined hydrologic flow paths between the aquifer and wetland sediments that have documented landfill-leachate water moving upward into the wetland sediments (Lorah et al., unpublished data, 2007). Toxicology studies have also demonstrated leachate-plume discharge to the slough resulting in elevated developmental toxicity levels (Bruner et al., 1998).

The fifth factor, Factor V (FV), has a variability of 8% and includes positive loadings for Fe²⁺ and NH₄⁺, and a negative loading for Ca. This factor was interpreted as an indicator of iron reduction and organic matter degradation either in the direct vicinity of the profile or elsewhere in the system. This factor is dominant in September 2005 below 10 cm and continues throughout the sediments. This indication of iron reduction at depth may be the result of in situ processes (including respiration and mineral precipitation/dissolution) or may reflect the transport of Fe²⁺ upward into the wetland sediments via advection from the aquifer. It is also observed in the middle sediments (silt layer-transition zone) in May 2005 and observed in the upper sediments in March 2004. This indicates that the activity of iron reduction changes in space, with active iron reduction having likely occurred in the upper sediments in the spring season and throughout the profile in September when oxidizing conditions enhance the availability of iron (III) minerals. However, this factor (FV), and indeed iron reduction itself, are difficult to interpret because the indicators are affected by multiple processes, including transport and phase changes between solid and dissolved forms.

Principal Component Analysis Including Additional Data from 2005
To further explore the relationships among processes illuminated by these first two PCAs, a third PCA was performed on a subset of the data (April, May, and September 2005) because additional geochemical parameters were only collected in 2005 and indicators including CH₄, DOC, Ba, Mn, and Fe³⁺ could be included. A Spearman Rho's correlation analysis was evaluated for 2005 and compared to correlations on the entire dataset (figure not shown). Unlike the Spearman's correlation for the whole dataset (Fig. 5), the correlation analysis for 2005 data yields the highest positive correlation for Cl^– and Na with a rho = 0.76. All the other positive correlations remain similar (Fig. 5) and include Fe³⁺ and Ca; Mn and Ca; Fe³⁺ and Mn; CH₄ and K; and DOC and NH₄⁺. The correlation matrix for the whole dataset showed significant negative correlations only between Ba and SO₄^2– (Fig. 5), although the correlation matrix for the 2005 dataset showed additional significant negative correlations for Na and H₂S (rho = –0.52), and Cl^– and K (rho = –0.50). The negative correlation between sulfate and barium is an indication of barite utilization as described previously. The negative correlation between K and Cl^– indicates that the potassium observed in the system is not controlled by the same mechanism as Cl^– in the system, which is attributed to plume advection.

To better understand the relationships from the PCA, the six factors with eigenvalues higher than 1 (Kaiser, 1960) and explaining 76% of the variability were rotated using varimax orthogonal rotation. The factor loadings obtained from the rotation are shown in Table 4 .

View larger version (36K): [in this window] [in a new window]   Fig. 8. Variability chart of factor scores by sampling date and depth of six factors obtained from the PCA including only data for 2005 (Table 4). The bars indicate the range in the factor scores data, the connecting lines indicate the mean for each depth interval, and the dashed line represents the true mean (0). Depths are reported in cm below the sediment–water interface (0 cm).

Factor C (FC) includes the variables CH₄, K, and Ba as positive loadings and pH as a negative loading and explains 12.2% of the variability. The variables expressed in this factor were not collected for the sampling periods 2003 and 2004 and demonstrate important redox processes that could not be determined in previous years but were probably present. The processes interpreted for FC are methanogenesis and bacterial mineral utilization, such as utilization of barite (BaSO₄) as a source of sulfate by sulfate-reducing bacteria. In addition, K associated with this factor had 72% communality and exhibited a strong Spearman's correlation with methane. This is an indication that the concentration of K in the system is associated with reducing conditions, such as a methanogenic environment, with K concentrations increasing with an increase of methane in the system. A plausible hypothesis for the release of K is the dissolution of silicate minerals by bacteria seeking electron acceptors (Fe³⁺) or nutrients (e.g. K, P).

It has been demonstrated in other studies that microorganisms are capable of utilizing crystalline minerals to obtain electron acceptors for their metabolic activities and can release mineral components to solution. One example is sulfate bacteria harvesting sulfate from minerals such as barite in dissolved sulfate-limited environments (Bolze et al., 1974; Ulrich et al., 2003). Ba strongly associated in this factor supports the hypothesis that in methanogenic conditions, sulfate reducers can utilize sulfate from barite as an electron acceptor. Iron reducers have also been documented to utilize mineral forms as a source of oxidized iron for their metabolic activities. Some studies even recognize the importance of this pathway in the release of K through the dissolution of minerals. Clay minerals such as illites and smectites are potassium-bearing phyllosilicates that have been studied for their implications on potassium availability as a result of microbial activity. At the Norman Landfill site, illites and smectites are abundant, comprising most of the clay minerals (Breit et al., 2005). In other studies, these minerals have shown significant reactivity with iron reducers and have been documented to have an impact on soil and physicochemical properties of sediment (Kostka et al., 1999; Stucki and Kostka, 2006). Stucki and Shen (1993) determined that potassium fixation occurs when microbial iron reduction causes the smectite minerals to collapse and trap ions such as NH₄⁺ and K⁺. Conversely, they observed a release of K⁺ to solution during microbial iron reduction of illites. In their study they attribute the release of K⁺ to exchange with Fe²⁺ produced from the reaction (Stucki and Shen, 1993). There have also been other studies that report dissolution of potassium-bearing minerals such as jarosite by iron reducers in which K⁺ and Fe²⁺ are released to solution at neutral pH (Jones et al., 2006).

The finding that in this system K is linked to reducing conditions has implications for studies relating to nutrient availability in redox-sensitive systems (Chen et al., 1997; Babu et al., 2006). Because K⁺ has a lower exchange capacity affinity and will remain in solution, it could be used as an indicator of mineral utilization by bacteria and/or methanogenesis in systems lacking vegetation, such as aquifers. A previous study by McGuire et al. (1999) observed important correlations between K⁺ and high levels of H₂, indicating methanogenic processes in a contaminated aquifer. Further studies would be needed to support this hypothesis.

The variability of FC by depth and season is shown in Fig. 8. Positive factor scores in FC indicate greater methanogenesis and mineral utilization and were observed in both April and May 2005 at a depth of 10 to 20 cm. In September 2005, this factor was not significant, which was likely due to the oxidizing conditions in this month. This suggests that the system presents more reducing conditions in May, which is a warm and wet month.

The fourth factor, Factor D (FD), explains 11.3% of the variability and contains the variables NH₄⁺, Ca, and HCO₃^– (Table 4). This factor was interpreted as organic matter degradation (negative factor scores). The high levels of ammonium together with higher alkalinities suggest significant organic matter degradation (Bally et al., 2004). This process is observed throughout May 2005 except in the 40- to 60-cm zone (Fig. 8). This is expected because May presents the most reducing conditions, indicating enhanced microbial activities. The trend in FD is similar to the trend in FC, suggesting that the concentrations of NH₄⁺ and alkalinity in this factor may also be related to the methanogenic and mineral utilization activity discussed for FC. For example, organic degradation fuels the progression of TEAPs and provides substrates for methanogenesis. The one exception is in September 2005 below 10-cm depth where this factor (FD) does not appear to be related to the methanogenic activities observed in FC. However, the higher NH₄⁺ and alkalinities may be attributed to organic degradation by iron reduction combined with transport of ammonium and alkalinity from the contaminated aquifer, which has high ammonium concentrations and high alkalinity (Christenson and Cozzarelli, 1999; Cozzarelli et al., 2000).

Factor E (FE), explaining 9.99% of the variability, was interpreted as a plume advection process because it contains geochemical parameters that are associated with landfill-contaminated water. Chloride, which has been used as a conservative tracer for landfill contamination, is highly correlated in this factor together with its common counter ion, Na. High alkalinity is also characteristic from the aquifer plume (Cozzarelli et al., 2000) and it has a moderate negative loading on this factor. In September and April 2005, the general trend for this process increases with depth. In May 2005, the trend is markedly disrupted at 40 to 60 cm, suggesting that the water contained in this sandy layer has a source other than direct flow from the contaminated aquifer. This finding helps to understand why this zone consistently presents an opposite process for all the factors in April and May. It is suspected that the composition of the water in this zone is affected by the water level or recharge conditions because the pattern is different for September. This is consistent with hydrologic patterns discerned by other studies (Lorah et al., unpublished data, 2007; Scholl et al., 2005).

The sixth factor, Factor F (FF), only explains 5.66% of the variability. It includes Fe³⁺ and total Mn and was interpreted as available electron acceptors. These electron acceptors are dominant in April 2005 in the 21- to 60-cm depth. April 2005 had the highest recharge and the lowest microbial activity, and therefore it is not surprising to observe higher concentrations of electron acceptors during this month. Moreover, the location where these are found (the sandy layer) confirms the disconnection of this layer from the deeper sediments and suggests that this oxidized layer may be connected to the surface.

	Conclusions

TOP NOTES ABSTRACT INTRODUCTION Materials and Methods Results and Discussion Conclusions REFERENCES

 
Biogeochemical cycling is complex and is controlled by linked geochemical, biological, and hydrogeological processes that vary in space and time. Establishing significant relationships between linked geochemical parameters could not be accomplished by graphical and numerical techniques alone. Multivariate statistics were used to illuminate relationships between (bio)geochemical parameters to determine dominant biogeochemical processes occurring in a wetland-aquifer system impacted by landfill leachate. For example, determination of iron redox cycling and its spatial and temporal controls cannot be teased out from plots of the Fe²⁺ concentrations in the system alone because Fe²⁺ concentrations are affected by many processes in the system, including transport and solid phase precipitation/exchange reactions. Factor analysis determines correlations among geochemical parameters allowing for interpretation of processes. Results show that Fe²⁺, at various points in time and space, associates with indicators of transport in the aquifer (e.g., Cl^–), with indicators of anaerobic respiration, and with indicators of mineral precipitation/dissolution reactions. Spatial patterns of geochemical associations as calculated using PCA (factor scores) provide a robust means of interpreting Fe²⁺ concentrations in the system throughout time. The greatest variability in the dataset was explained by fermentation and it is thus considered to be the primary process governing biogeochemical cycling in the system. Sulfate and iron reduction were found to be the dominant TEAPs, followed by methanogenesis and sulfate reduction via utilization of barite (BaSO₄). Seasonal hydrological patterns (wet and dry periods) are an important control on the availability of electron acceptors through the reoxidation of reduced iron-sulfur species, enhancing iron and sulfate reduction.

The combined use of PCA and spatial representation of factors was a powerful technique to explore the linked biogeochemical variability in a large dataset containing spatial and temporal data and to expose underlying relationships as well as evaluate the importance of various processes in the system. Though exploratory in nature, this application of PCA provides a mechanism to formulate hypotheses regarding process that may not have been observed previously that can then be tested with further experimentation.

	ACKNOWLEDGMENTS

 
We would like to thank the staff of the USGS in Oklahoma City, Oklahoma, in particular Scott Christenson, for field and analytical support, and the staff from the USGS in Reston, Virginia; Jeanne Jaeschke for DOC, organic acid, and methane analyses, and Michael Lowit and Gregory Noe for their insightful comments and review of the manuscript. This project was funded by the National Science Foundation, Biocomplexity in the Environment grant EAR-0418488, and the Center of Environmental and Rural Health at TAMU grant #: P30ES09106. This research was also supported by the USGS Toxic Substances Hydrology Program and the USGS National Research Program. *The use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

	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.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION Materials and Methods Results and Discussion Conclusions REFERENCES

 

• Aller, R.C., P.O.J. Hall, P.D. Rude, and J.Y. Aller. 1998. Biogeochemical heterogeneity and suboxic diagenesis in hemipelagic sediments of the Panama Basin. Deep-sea Res. I 45:133–165.[CrossRef]
• Amini, M., K.C. Abbaspour, H. Khademi, N. Fathianpour, M. Afyuni, and R. Schulin. 2005. Neural network models to predict cation exchange capacity in arid regions of Iran. Eur. J. Soil Sci. 56:551–559.[CrossRef]
• APHA, AWWA, and WEF. 2005. Standard methods for the examination of water & wastewater. p. 3–77, 4–174. In A.D. Eaton et al. (ed.) 21st ed. American Public Health Association, American Water Works Association, Water Environment Federation, Washington, DC.
• Azcue, J.M., F. Rosa, and G. Lawson. 1996. An improved dialysis sampler for the in situ collection of larger volumes of sediment pore waters. Environ. Technol. 17:95–100.
• Babu, Y.J., D.R. Nayak, and T.K. Adhya. 2006. Potassium application reduces methane emission from a flooded field planted rice. Biol. Fertil. Soils 42:532–541.[CrossRef]
• Baedecker, M.J., and I.M. Cozzarelli. 1992. The determination and fate of unstable constituents in contaminated ground water. p. 425–461. In S. Lesage and R. Jackson (ed.) Groundwater quality and analysis at hazardous waste sites. Marcel Dekker, Inc., New York.
• Báez-Cazull, S.E., J.T. McGuire, I.M. Cozzarelli, A. Raymond, and L. Welsh. 2007. Centimeter-scale characterization of biogeochemical gradients at a wetland-aquifer interface using capillary electrophoresis. Appl. Geochem. (in press) doi:10.1016/j.apgeochem.2007.06.003.
• Bally, G., V. Mesnage, J. Deloffre, O. Clarisse, R. Lafite, and J.P. Dupont. 2004. Chemical characterization of porewaters in an intertidal mudflat of the Seine estuary: Relationship to erosion-deposition cycles. Mar. Pollut. Bull. 49:163–173.[CrossRef][Web of Science][Medline]
• Bolze, C.E., P.G. Malone, and M.J. Smith. 1974. Microbial mobilization of barite. Chem. Geol. 13:141–143.[CrossRef][Web of Science]
• Boschker, H.T.S., S.C. Nold, P. Wellsbury, D. Bos, W. de Graaf, R. Pel, R.J. Parkes, and T.E. Cappenberg. 1998. Direct linking of microbial populations to specific biogeochemical processes by C-13-labelling of biomarkers. Nature 392:801–805.[CrossRef]
• Breit, G.N., M.L.W. Tuttle, I.M. Cozzarelli, S.C. Christenson, J.B. Jaeschke, D.L. Fey, and C.J. Berry. 2005. Results of chemical and isotopic analyses of sediment and water from alluvium of the Canadian River near a closed municipal landfill, Norman, Oklahoma. USGS Open-File Rep. 2005-1091, p. 1–37. USGS, Washington, DC.
• Brune, A., P. Frenzel, and H. Cypionka. 2000. Life at the oxic-anoxic interface: Microbial activities and adaptations. FEMS Microbiol. Rev. 24:691–710.[Web of Science][Medline]
• Bruner, M.A., M. Rao, J.N. Dumont, M. Hull, T. Jones, and J.A. Bantle. 1998. Ground and surface water developmental toxicity at a municipal landfill: Description and weather-related variation. Ecotoxicol. Environ. Saf. 39:215–226.[CrossRef][Web of Science][Medline]
• Capone, D.G., and R.P. Kiene. 1988. Comparison of microbial dynamics in marine and freshwater sediments: Contrasts in anaerobic carbon catabolism. Limnol. Oceanogr. 33:725–749.
• Champ, D.R., J. Gulens, and R.E. Jackson. 1979. Oxidation-reduction sequences in groundwater flow systems. Can. J. Earth Sci. 127:85–108.
• Chapelle, F.H., and P.B. McMahon. 1991. Geochemistry of dissolved inorganic carbon in a coastal-plain aquifer: I. Sulfate from confining beds as an oxidant in microbial CO₂ production. J. Hydrol. 127:85–108.[CrossRef]
• Chapelle, F.H., P.B. McMahon, N.M. Dubrovsky, R.F. Fujii, E.T. Oaksford, and D.A. Vroblesky. 1995. Deducing the distribution of terminal electron-accepting processes in hydrologically diverse groundwater systems. Water Resour. Res. 31:359–371.[CrossRef]
• Chen, J., J. Xuan, C. Du, and J. Xie. 1997. Effect of potassium nutrition of rice on rhizosphere redox status. Plant Soil 188:131–137.[CrossRef][Web of Science]
• Christenson, S., and I.M. Cozzarelli. 1999. Geochemical and microbiological processes in ground water and surface water affected by municipal landfill leachate. In D.W. Morganwalp and H.T. Buxton (ed.) USGS Toxic Substances Hydrology Program, Proc. of the technical meeting, Charleston, South Carolina, 8–12 Mar. 1999: USGS, Subsurface Contamination from Point Sources: Water-Resources Investigations Rep. 99-4018, 3, p. 499–500. USGS, Washington, DC.
• Christenson, S., M.A. Scholl, J.L. Schlottmann, and C.J. Becker. 1999. Ground-water and surface-water hydrology of the Norman Landfill Research Site. In D.W. Morganwalp and H.T. Buxton (ed.) USGS Toxic Substances Hydrology Program, Proc. of the technical meeting, Charleston, South Carolina, 8–12 Mar. 1999: USGS, Subsurface Contamination from Point Sources: Water-Resources Investigations Rep. 99–4018C, 3, p. 501–507. USGS, Washington, DC.
• Cozzarelli, I.M., J.M. Suflita, G.A. Ulrich, S.H. Harris, M.A. Schroll, J.L. Schlottmann, and S. Christenson. 2000. Geochemical and microbiological methods for evaluating anaerobic processes in an aquifer contaminated by landfill leachate. Environ. Sci. Technol. 34:4025–4033.
• Cozzarelli, I.M., J.M. Suflita, G.A. Ulrich, S.H. Harris, M.A. Schroll, J.L. Schlottmann, and J.B. Jaeschke. 1999. Biogeochemical processes in a contaminant plume downgradient from a landfill, Norman, Oklahoma. In D.W. Morganwalp and H.T. Buxton (ed.) USGS Toxic Substances Hydrology Program, Proc. of the technical meeting, Charleston, South Carolina, 8–12 Mar. 1999: USGS, Subsurface Contamination from Point Sources: Water-Resources Investigations Rep. 99-4018C, 3, p. 521–530. USGS, Washington, DC.
• Dahm, C.N., N.B. Grimm, P. Marmonier, H.M. Valett, and P. Vervier. 1998. Nutrient dynamics at the interface between surface waters and groundwaters. Freshwater Biol. 40:427–451.[CrossRef]
• Donahoe, R.J., and C. Liu. 1998. Pore water geochemistry near the sediment-water interface of a zoned, freshwater wetland in the southeastern United States. Environ. Geol. 33:143–153.[CrossRef]
• Duddleston, K.N., M.A. Kinney, R.P. Kiene, and M.E. Hines. 2002. Anaerobic microbial biogeochemistry in a northern bog: Acetate as a dominant metabolic end product. Global Biogeochemical Cycles 16:11-1–11-9.
• Eser, P., and M.R. Rosen. 1999. The influence of groundwater hydrology and stratigraphy on the hydrochemistry of Stump Bay, South Taupo Wetland, New Zealand. J. Hydrol. 220:27–47.[CrossRef]
• Farnham, I.M., K.H. Johannesson, A.K. Singh, V.F. Hodge, and K.J. Stetzenbach. 2003. Factor analytical approaches for evaluating groundwater trace element chemistry data. Anal. Chim. Acta 490:123–138.[CrossRef][Web of Science]
• Froelich, P.N., G.P. Klinkhammer, M.L. Bender, N.A. Luedtke, G.R. Heath, D. Cullen, P. Dauphin, D. Hammond, B. Hartman, and V. Maynard. 1979. Early oxidation of organic matter in pelagic sediments of the eastern equatorial Atlantic: Suboxic diagenesis. Geochim. Cosmochim. Acta 43:1075–1090.[CrossRef][Web of Science]
• Grossman, E.L., L.A. Cifuentes, and I.M. Cozzarelli. 2002. Anaerobic methane oxidation in a landfill-leachate plume. Environ. Sci. Technol. 36:2436–2442.[Medline]
• Güler, C., G. Thyne, J.E. McCray, and A.K. Turner. 2002. Evaluation of graphical and multivariate statistical methods for classification of water chemistry data. Hydrogeol. J. 10:455–474.[CrossRef]
• Hesslein, R.H. 1976. An in situ sampler for close interval pore water studies. Limnol. Oceanogr. 21:912–914.
• Hines, M.E., G.T. Banta, A.E. Giblin, J.E. Hobbie, and J.B. Tugel. 1994. Acetate concentrations and oxidation in salt-marsh sediments. Limnol. Oceanogr. 39:140–148.
• Ho, T.Y., M.I. Scranton, G.T. Taylor, R. Varela, R.C. Thunell, and F. Muller-Karger. 2002. Acetate cycling in the water column of the Cariaco Basin: Seasonal and vertical variability and implication for carbon cycling. Limnol. Oceanogr. 47:1119–1128.
• Hunt, R.J., D.P. Krabbenhoft, and M.P. Anderson. 1997. Assessing hydrogeochemical heterogeneity in natural and constructed wetlands. Biogeochemistry 39:271–293.
• Jacobs, P.H. 2002. A new rechargeable dialysis pore water sampler for monitoring sub-aqueous in-situ sediment caps. Water Res. 36:3121–3129.[Medline]
• Jones, E.J.P., T.-L. Nadeau, M.A. Voytek, and E.R. Landa. 2006. Role of microbial iron reduction in the dissolution of iron hydroxysulfate minerals. J. Geophys. Res. 111:G01012.
• Kaiser, H.F. 1960. The application of electronic computers to factor analysis. Educ. Psychol. Meas. 20:141–151.[CrossRef][Web of Science]
• Kappler, A., D. Emerson, K. Edwards, J.P. Amend, J.A. Gralnick, P. Grathwohl, T. Hoehler, and K.L. Straub. 2005. Microbial activity in biogeochemical gradients: New aspects of research. Geobiology 3:229–233.
• Kemp, W.M., P.A. Sampou, J. Garber, J. Tuttle, and W.R. Boynton. 1992. Seasonal depletion of oxygen from bottom waters of Chesapeake Bay: Roles of benthic and planktonic respiration and physical exchange processes. Mar. Ecol. Prog. Ser. 85:137–152.[CrossRef]
• Kleikemper, J., M.H. Schroth, W.V. Sigler, M. Schmucki, S.M. Bernasconi, and J. Zeyer. 2002. Activity and diversity of sulfate-reducing bacteria in a petroleum hydrocarbon-contaminated aquifer. Appl. Environ. Microbiol. 68:1516–1523.[Abstract/Free Full Text]
• Koretsky, C.M., C.M. Moore, K.L. Lowe, C. Meile, T.J. DiChristina, and P. Van Cappellen. 2003. Seasonal oscillation of microbial iron and sulfate reduction in saltmarsh sediments (Sapelo Island, GA, USA). Biogeochemistry 64:179–203.
• Kostka, J.E., A. Roychoudhury, and P. Van Cappellen. 2002. Rates and controls of anaerobic microbial respiration across spatial and temporal gradients in saltmarsh sediments. Biogeochemistry 60:49–76.
• Kostka, J.E., J. Wu, K.H. Nealson, and J.W. Stucki. 1999. The impact of structural Fe(III) reduction by bacteria on the surface chemistry of smectite clay minerals. Geochim. Cosmochim. Acta 63:3705–3713.[CrossRef][Web of Science]
• Kumar, M., A. Ramanathan, M.S. Rao, and B. Kumar. 2006. Identification and evaluation of hydrogeochemical processes in the groundwater environment of Delhi, India. Environ. Geol. 50:1025–1039.[CrossRef]
• Liu, C.-W., K.-H. Lin, and Y.-M. Kuo. 2003. Application of factor analysis in the assessment of groundwater quality in a blackfoot disease area in Taiwan. Sci. Total Environ. 313:77–89.[CrossRef][Medline]
• Lovley, D.R., E.J.P. Phillips, and D.J. Lonergan. 1991. Enzymatic versus nonenzymatic mechanisms for Fe(III) reduction in aquatic sediments. Environ. Sci. Technol. 25:1062–1067.
• Mathes, S.E., and T.C. Rasmussen. 2006. Combining multivariate statistical analysis with geographic information systems mapping: A tool for delineating groundwater contamination. Hydrogeol. J. 14:1493–1507.[CrossRef]
• McGuire, J.T., D.T. Long, and D.W. Hyndman. 2005. Analysis of recharge-induced geochemical change in a contaminated aquifer. Ground Water 43:518–530.[CrossRef][Web of Science][Medline]
• McGuire, J.T., E.W. Smith, D.T. Long, D.W. Hyndman, S.K. Haack, M.J. Klug, and M.A. Velbel. 2000. Temporal variations in parameters reflecting terminal-electron-accepting processes in an aquifer contaminated with waste fuel and chlorinated solvents. Chem. Geol. 169:471–485.[CrossRef][Web of Science]
• McGuire, J.T., E.W. Smith, D.T. Long, D.W. Hyndman, S.K. Haack, J.J. Kolak, M.J. Klug, M.A. Velbel, and L.J. Forney. 1999. Temporal variations in biogeochemical processes that influence ground water redox zonation. In D.W. Morganwalp and H.T. Buxton (ed.) USGS Survey Toxic Substances Hydrology Program, Proc. of the technical meeting, Charleston, South Carolina, 8–12 Mar. 1999: USGS, Subsurface Contamination from Point Sources: Water-Resources Investigations Rep. 99-4018C, 3, p. 641–652. USGS, Washington, DC.
• Megonigal, J.P., M.E. Hines, and P.T. Visscher. 2004. Anaerobic metabolism: Linkages to trace gases and aerobic processes. Elsevier-Pergamon, Oxford, UK.
• Nealson, K.H. 1997. Sediment bacteria: Who's there, what are they doing, and what's new? Annu. Rev. Earth Planet. Sci. 25:403–434.[CrossRef][Web of Science][Medline]
• Neubauer, S.C., K. Givler, S. Valentine, and J.P. Megonigal. 2005. Seasonal patterns and plant-mediated controls of subsurface wetland biogeochemistry. Ecology 86:3334–3344.[CrossRef][Web of Science]
• Parkhurst, D.L. 1995. PHREEQC: User's guide to PHREEQC: A computer program for speciation, reaction-path, advective-transport, and inverse geochemical calculations. USGS Water Resource, Investigations Rep. 95-4227, p. 1–143. USGS, Washington, DC.
• Puckett, L.J., and O.P. Bricker. 1992. Factors controlling the major ion chemistry of streams in the Blue Ridge and Valley and Ridge physiographic provinces of Virginia and Maryland. Hydrol. Processes 6:79–98.[CrossRef]
• Qian, J., and K. Mopper. 1996. Automated high performance, high temperature combustion total organic carbon analyzer. Anal. Chem. 68:3090–3097.
• Röling, W.F.M., B.M. van Breukelen, M. Braster, B. Lin, and H.W. van Verseveld. 2001. Relationships between microbial community structure and hydrochemistry in a landfill leachate-polluted aquifer. Appl. Environ. Microbiol. 67:4619–4629.[Abstract/Free Full Text]
• SAS Institute. 1999. JMP Version 4. SAS Inst., Cary, NC.
• Schlottmann, J.L. 2001. Water chemistry near the closed Norman Landfill, Cleveland County, Oklahoma, 1995. USGS Water-Resources Investigations Rep. 00-4238, p. 1–41. USGS, Washington, DC.
• Scholl, M.A., and S. Christenson. 1998. Spatial variation in hydraulic conductivity determined by slug tests in the Canadian River alluvium near the Norman Landfill, Norman, Oklahoma. USGS Water-Resources. Investigations Rep. 97-4292, p. 1–28. USGS, Washington, DC.
• Scholl, M.A., S.C. Christenson, I.M. Cozzarelli, D. Ferree, and J. Jaeschke. 2005. Recharge processes in an alluvial aquifer riparian zone, Norman Landfill, Norman, Oklahoma, 1998–2000. USGS Scientific Investigations. Rep. 2004-5238, p. 1–53. USGS, Washington, DC.
• Schot, P.P., and J. van der Waal. 1992. Human impact on regional groundwater composition through intervention in natural flow patterns and changes in land use. J. Hydrol. 134:297–313.[CrossRef]
• Seybold, C.A., W. Mersie, J.Y. Huang, and C. McNamee. 2002. Soil redox, pH, temperature, and water-table patterns of a freshwater tidal wetland. Wetlands 22:149–158.[CrossRef]
• Shannon, R.D., and J.R. White. 1996. The effects of spatial and temporal variations in acetate and sulfate on methane cycling in two Michigan peatlands. Limnol. Oceanogr. 41:435–443.
• Stucki, J.W., and J.E. Kostka. 2006. Microbial reduction of iron in smectite. C. R. Geosci. 338:468–475.[CrossRef]
• Stucki, J.W., and S. Shen. 1993. Effects of iron oxidation state on potassium fixation. Univ. of Illinois, Urbana. Available at http://frec.cropsci.uiuc.edu/1993/report12 (verified 13 Nov. 2007).
• Suk, H., and K.-K. Lee. 1999. Characterization of a ground water hydrochemical system through multivariate analysis: Clustering into ground water zones. Ground Water 37:358–366.[CrossRef][Web of Science]
• Thyne, G., C. Güler, and E. Poeter. 2004. Sequential analysis of hydrochemical data for watershed characterization. Ground Water 42:711–723.[CrossRef][Web of Science][Medline]
• Ulrich, G.A., G. Breit, and I.M. Cozzarelli. 2003. Sources of sulfate supporting anaerobic metabolism in a contaminated aquifer. Environ. Sci. Technol. 37:1093–1099.[Medline]
• USGS Water Resources of Oklahoma, 2007. Daily data of the Canadian River tributary at Norman, OK. Available at http://waterdata.usgs.gov/ok/nwis/inventory/?site_no=07229053&amp (verified 16 Oct. 2007).
• van Breukelen, B.M., and J. Griffioen. 2004. Biogeochemical processes at the fringe of a landfill leachate plume: Potential for dissolved organic carbon, Fe(II), Mn(II), NH₄, and CH₄ oxidation. J. Contam. Hydrol. 73:181–205.[CrossRef][Web of Science][Medline]
• van Griethuysen, C., M. Luitwieler, J. Joziasse, and A.A. Koelmans. 2005. Temporal variation of trace metal geochemistry in floodplain lake sediment subject to dynamic hydrological conditions. Environ. Pollut. 137:281–294.[CrossRef][Medline]
• van Helvoort, P.-J., P. Filzmoser, and P.F.M. van Gaans. 2005. Sequential factor analysis as a new approach to multivariate analysis of heterogeneous geochemical datasets: An application to a bulk chemical characterization of fluvial deposits (Rhine-Meuse delta, The Netherlands). Appl. Geochem. 20:2233–2251.[CrossRef]
• Wayland, K.G., D.T. Long, D.W. Hyndman, B.C. Pijanowski, S.M. Woodhams, and S.K. Haack. 2003. Identifying relationships between baseflow geochemistry and land use with synoptic sampling and r-mode factor analysis. J. Environ. Qual. 32:180–190.[Abstract/Free Full Text]
• Webster, I.T., P.R. Teasdale, and N.J. Grigg. 1998. Theoretical and experimental analysis of peeper equilibration dynamics. Environ. Sci. Technol. 32:1727–1733.
• Weiss, J.V., D. Emerson, and J.P. Megonigal. 2004. Geochemical control of microbial Fe(III) reduction potential in wetlands: Comparison of the rhizosphere to non-rhizosphere soil. FEMS Microbiol. Ecol. 48:89–100.[CrossRef]
• Wellsbury, P., and R.J. Parkes. 1995. Acetate bioavailability and turnover in an estuarine sediment. FEMS Microbiol. Ecol. 17:85–94.[CrossRef]
• Xie, Z., L. Sun, P. Zhang, S. Zhao, X. Yin, X. Liu, and B. Cheng. 2005. Preliminary geochemical evidence of groundwater contamination in coral islands of Xi-Sha, South China Sea. Appl. Geochem. 20:1848–1856.[CrossRef]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Báez-Cazull, S. E. Articles by Voytek, M. A. Search for Related Content PubMed PubMed Citation Articles by Báez-Cazull, S. E. Articles by Voytek, M. A. GeoRef GeoRef Citation Agricola Articles by Báez-Cazull, S. E. Articles by Voytek, M. A. Related Collections Wetlands and Aquatic Processes Redox Processes Spatial Variability