dissolved organic matter accumulation, reactivity, and ...

WETLANDS, Vol. 28, No. 3, September 2008, pp. 747–759 ’ 2008, The Society of Wetland Scientists

DISSOLVED ORGANIC MATTER ACCUMULATION, REACTIVITY, AND REDOX STATE IN GROUND WATER OF A RECHARGE WETLAND Natalie Mladenov1,2, Philippa Huntsman-Mapila3, Piotr Wolski3, Wellington R. L. Masamba3, and Diane M. McKnight1 1 INSTAAR, University of Colorado 450 UCB Boulder, Colorado, USA 80309 E-mail: [email protected] 2

Departamento de Ecologı´a, Universidad de Granada 18071 Granada, Spain

3

Harry Oppenheimer Okavango Research Centre, University of Botswana Private Bag 285 Maun, Botswana

Abstract: Ground water beneath the seasonal swamp of the Okavango Delta, a recharge wetland in northwestern Botswana, is known to be a sink for solutes. In this study, measurements of organic carbon and inorganic ion concentrations, as well as UV-visible and fluorescence spectroscopy, were used to examine dissolved organic matter (DOM) storage and redox state of fulvic acids in ground water beneath an island and riparian woodland. Increasing dissolved organic carbon (DOC) concentrations along the ground-water flowpath suggests an accumulation of DOM in ground water, especially beneath island centers. However, the increase in DOC concentration was relatively less than the increase in chloride and sulfate concentrations, indicating non-conservative behavior of DOM in ground water beneath wetland islands. In combination with a decrease in fulvic acid content and specific UV absorbance, this result suggests that preferential sorption or destabilization of more aromatic organic compounds may be occurring under conditions of high pH and salinity. Finally, the increase in reduced fluorescence components (semiquinone- and hydroquinone-like components) along the ground-water flowpath strongly supports the transition to reduced fulvic acids in ground water of island centers. The reactivity and potential electron-shuttling function of fulvic acids may play an important role in the dissolution of metal oxides and associated DOM-iron-arsenic interactions in ground water of this recharge wetland. Key Words: EEM, fluorescence index, humic substances, Okavango Delta, PARAFAC, SUVA

INTRODUCTION

involved in strong metal binding (McKnight et al. 1992) and, in laboratory experiments, their role as electron shuttles between iron (Fe)-reducing bacteria and Fe-oxides has been shown to enhance Fe (III) reduction (Lovley et al. 1996). The concentration of DOC in ground water is generally lower than in many surface waters, reflecting the chemical and biotic processing of DOM in the subsurface (Thurman 1985). However, dissolved organic carbon (DOC) concentrations in ground water can be much higher if recharged by wetlands. In the Okavango Delta, a large wetland in northwestern Botswana, high DOC concentrations (from 13 to 25 mg C L21) were measured in ground water of the seasonal swamp region (Figure 1) adjacent to channels and floodplains (Mladenov 2004, Bauer-Gottwein et al. 2007). Mladenov et al.

In aquatic ecosystems, dissolved organic matter (DOM) represents the major pool of organic carbon and is a substrate for heterotrophic microorganisms. DOM originating from plant/soil and microbial sources can be transported to ground water by infiltration. DOM can also be produced within the ground-water system as extracellular microbial products. DOM can be removed via bacterial degradation, as some organic compounds are substrates that support microbial growth. Some DOM fractions can adsorb to clay and oxide surfaces, whereas these and other fractions can also influence metal cycling. Fulvic acids are humic substances that are soluble across the pH range and have high electron accepting capacity (Scott et al. 1998). Fulvic acids, in particular, are known to be 747

748

WETLANDS, Volume 28, No. 3, 2008

Figure 1. Features of the study site, including the Boro Channel, study floodplain, surface water sampling points, island and woodland transects. C1 to C10 (approximately 1 m to 240 m from edge of water) and W1 to W8 (approximately 1 m to 200 m from edge of water) correspond to the locations of island and woodland piezometers, respectively. Inset shows Okavango Delta and Botswana, and the study site is indicated with an arrow.

(2007b) hypothesized that high ground-water DOC concentrations were directly related to the high DOC concentrations in surface waters (from 8 to 30 mg C L21 during flood periods) and high infiltration rates. Mladenov et al. (2007b) modeled temporal variations in DOC concentrations in a seasonal floodplain of the Okavango Delta. The model results indicated that approximately 28% of the DOM brought to the floodplain during a high flood season infiltrated into the subsurface. Because infiltrating water follows lateral ground-water flowpaths driven by evapotranspiration (McCarthy and Ellery 1994), it was hypothesized that infiltration of large amounts of DOM would result in substantial DOM storage in ground water beneath islands and uplands adjacent to surface water (Mladenov et al. 2007b). Chemical speciation modeling of ground water beneath three islands in the seasonal swamp of the Okavango Delta (Bauer-Gottwein et al. 2007)

showed that density-driven flow transports solutes downward into deeper ground-water layers beneath these islands. However, both the onset of densitydriven flow and the accumulation of solutes could be delayed if ground-water humic substances concentrations are high enough to induce CO2 degassing and/or mineral precipitation. The results of BauerGottwein et al. (2007) suggested that humic substances concentrations were not high enough for CO2 degassing to represent a major mass loss mechanism, even though ground-water DOC concentrations were found to be very high (on the order of 1,000–4,000 mg C L21) in ground water beneath island centers. Although humic or fulvic acid content was not directly measured, Bauer-Gottwein et al. (2007) concluded that the influence of humic substances on the geochemistry of ground water beneath islands was minimal, based on the results of geochemical modeling.

Mladenov et al., CHARACTERIZATION OF DOM IN WETLAND GROUND WATER The conclusions of Bauer-Gottwein et al. (2007) diverge from the expected high reactivity of humics in ground water of the seasonal swamp, based on the known infiltration of surface water DOM with a high degree of aromaticity and high N and S content (Mladenov et al. 2007a). Isolation of humic substances by XAD-8 resin also revealed high fulvic acid content (approximately 70% of DOC) in surface waters in the seasonal swamp (Mladenov et al. 2005). Because surface water recharges ground water in the seasonal swamp of the Okavango Delta, these findings suggest that humic substances exert an important influence on biogeochemical processes, such as nutrient immobilization (Wolski et al. 2005) and metal cycling processes (Huntsman-Mapila et al. 2006) in ground water of the Okavango Delta region. Recent advances in fluorescence spectroscopy have made it possible to elicit information regarding the sources (McKnight et al. 2001, Mladenov et al. 2005) and chemical character of DOM (Cory and McKnight 2005, Mladenov et al. 2007a), as well as the redox state of fulvic acids (Klapper et al. 2002, Fulton et al. 2004, Cory and McKnight 2005, Miller et al. 2006). Parallel factor analysis (PARAFAC) has been applied to resolve the dominant fluorescent components present in three-dimensional excitation emission matrices (EEMs) of DOM from diverse natural waters (Stedmon et al. 2003, Cory and McKnight 2005). These fluorescent components include oxidized and reduced quinone-like fluorophores (Cory and McKnight 2005). The ratio of reduced quinones to total quinones, known as the redox index (RI) (Miller et al. 2006), is a useful tool to identify the redox state of fulvic acids in ground water, and has been used to examine transformations of humic substances along flowpaths in a wetland hyporheic zone (Miller et al. 2006). In order to explore relationships between hydrologic controls and the sources, chemical character, and redox state of fulvic acids in wetland-recharged ground water, this study investigates chemical properties of DOM in ground water of the seasonal swamp in the Okavango Delta. Also, this study employs direct measurements of ground-water DOC concentrations and fulvic acid content to evaluate the hypothesis posed by Bauer-Gottwein et al. (2007) that humics concentrations are not high enough to induce geochemical reactions that influence density-driven flow of solutes in island ground water. Finally, by examining basic water chemistry and UV absorbance and fluorescence properties of DOM along a hydrologic flowpath from surface water to ground water in the island interior (Figure 1), we evaluate the potential for both biogeochemical processing and substantial storage

749

of DOM in ground water supplied by DOM-rich surface water. METHODS Site Description and Field Sampling Ground water was sampled at two locations adjacent to surface water sources. The first transect (‘‘island transect’’) terminates on the southeastern end of an island that receives regular annual flooding on its southwest side due to its proximity to the Boro River channel (Figure 1). This island is one of the larger islands in the Delta (approximately 400 m 3 1,500 m) and becomes completely surrounded by water only during high flood years (as in 2001). The island transect is referred to as the ‘‘AB transect’’ in Wolski and Savenije (2006) and Wolski et al. (2005) and as ‘‘ORC Island’’ by BauerGottwein et al. (2007). The second location (‘‘woodland transect’’) crosses an area of dryland adjacent to a seasonal floodplain, which is hydraulically connected to the Boro River channel (Figure 1). The annual flood occurs during the dry season when floodwaters originating in the Angolan Highlands inundate the Okavango Delta. During the annual flood, surface water levels in the Boro River increase between 0.8 and 1.5 m; and surface water infiltrates and flows laterally toward island centers. This lateral ground-water flow is driven by evapotranspiration at the island centers and has been documented by McCarthy and Ellery (1994) and modeled by Gieske (1997) and Wolski and Savenije (2006). The island has a relatively wide (100 m) vegetated fringe composed of riparian woodland species. Ground-water piezometers sampled in this region (C1 to C7) are referred to as being in the ‘‘island fringe.’’ During the annual flood, the water table of the island fringe is active and can rise 3 m (Wolski and Savenije 2006). The island’s center is covered by grassland and salt crusts and is underlain by a relatively thick (more than 7 m) lens of deposits of predominantly clayey texture. Such lenses are typical of Okavango Delta islands and have been described by McCarthy and Ellery (1994). The lenses are composed of quartz parent sand and ground-water precipitates comprising amorphous silica and carbonates. Ground-water piezometers sampled in this region (piezometers C8 (at 140 m) to C10 (at 240 m)) are referred to as being in the ‘‘island center.’’ During the annual flood, the water table of the island center remains fairly constant, rising only between 0.10 m and 0.25 m about 1–5 months after the surface water peak (Wolski and Savenije 2006).

750 The dryland body adjacent to the floodplain at the woodland transect is covered by a mixture of dryland and riparian species, while further inland, a typical dryland forest occurs. The floodplain is covered by dryland grasses in dry periods, and by aquatic sedges and grasses while inundated during the annual flood. The substratum penetrated by piezometers (W1 to W8; at 1 m to 200 m) is uniformly sandy, apart from a small clayey lens in the central part of the floodplain. No ground water was sampled from the central part of the floodplain. The floodplain is annually flooded, and the duration and level of the floods vary between years. Annually the ground-water table can rise up to 2.5 m (Wolski and Savenije 2006). Ground-water samples were collected from a network of piezometers located along the island and woodland transects (Figure 1) on June 12, 26, and 28, 2001; on July 10, 12, and 25, 2001; on June 26, 2002; on September 21, 2005; on December 2, 2005; and on May 25, 2007. At locations W1-W8 and C1-C10 nested piezometers were installed, comprising two or three pipes filtered at different depths, usually at 2, 4, and 6 m below ground. Only the two deepest piezometers were sampled for this study (Figure 2). Piezometers were pumped dry five times prior to sample collection. All samples were filtered with GF/ C glass fiber filters with 1.2 mm nominal pore size. Samples collected for organic chemistry and UV-vis and fluorescence spectroscopy were acidified with concentrated HCl to a pH of about 2. Conductivity and temperature were measured in situ using a YSI30 Salinity-Conductivity-Temperature meter. Dissolved oxygen (DO) was measured with a YSI-55 DO meter and pH was measured with a Fisher Scientific Accumet AP60 portable pH meter.


Figure 2. Profiles of piezometer depth, pH, alkalinity (alkalinity expressed as HCO3 is circled), DO, and temperature in the ground water of island (left panels) and woodland (right panels) transects. In order to maintain figure clarity, all distances and alkalinity y-axis are shown using a log scale. Mean values and standard deviations are shown. Surface water results are shown with a dashed line.

Alkalinity, Anions, and Total Fe. Alkalinity was measured on September 21, 2005 and December 2, 2005 samples in the lab with a Mettler Toledo DL50 auto Titrator. Anions (chloride and sulfate) were measured on the samples collected on September 21 and December 2, 2005 using a Dionex Series 4500I Liquid Ion Chromatograph with a 50 ppb detection limit. Fe was measured within one month of collection on May 25, 2007 samples on a Finnigan Element 2 ICP-MS.

sample collection and were replicated within runs and over time. The standard deviation of replicates was , 5% for all samples. It has been shown that high inorganic carbon (IC) concentrations can lead to an overestimation of DOC concentrations when the TOC analysis technique uses oxidation to drive off the IC (Potter and Wimsatt 2005). In order to accurately measure DOC concentrations in piezometers C8–C10, samples were diluted (1:100) with MilliQ water and re-acidified to pH 2 to drive off high inorganic carbon content. CO2 degassing was observed during acidification.

DOC Concentrations. DOC concentrations were measured on samples collected in 2001, 2002, and 2005 using a Shimadzu TOC-5050 Total Organic Carbon Analyzer, within one to four months of

Absorbance. Absorbance at 280 nm was measured on samples collected in 2001, 2002, and 2005 using an Agilent 8453 UV-VIS spectrophotometer with ChemStation software and a 1 cm path length cell.

Laboratory Analyses

Mladenov et al., CHARACTERIZATION OF DOM IN WETLAND GROUND WATER Each sample was measured three times and the standard deviation of replicates was , 5% for all samples. Specific UV absorbance (SUVA) was calculated as the absorbance (measured at 280 nm) normalized to the DOC concentration and reported in units of m21 mg21 L. UV absorbance at 280 nm or 254 nm have both been commonly used to calculate SUVA (Chin et al. 1994, Weishaar et al. 2003). Because nitrate can absorb at the lower wavelength, SUVA at 280 nm was chosen for this study. For comparison, the SUVA of two end member aquatic fulvic acids, Suwannee River reference fulvic acid (SRFA, a standard of the International Humic Substances Society) and Lake Fryxell fulvic acid (LFFA, McKnight et al. 2001), was measured to indicate plant/soil source and microbial DOC sources, respectively. Fulvic Acid Isolation. Hydrophobic organic acids were isolated from approximately 150 mL of filtered and acidified water samples collected in 2002 and 2005 using small volume (10 mL) columns filled with XAD-8 resin, following the method of Thurman and Malcolm (1981). For consistency with other studies (McKnight et al. 1997, Klapper et al. 2002, Hood et al. 2003), these hydrophobic organic acids will be referred to as fulvic acids (FA). Fluorescence. Fluorescence spectroscopy can provide insights into the chemical properties of DOM by generation of EEMs, determination of the fluorescence index (FI), and PARAFAC modeling. EEMs are a 3-dimensional representation of fluorescence intensities scanned over a range of excitation/emissions (Ex/Em) wavelengths. Prominent peaks in fulvic acids have been found at Ex/Em wavelengths of approximately 240/450 nm (referred to as region A) and 320/450 nm (referred to as region C) (Coble 1995). EEMs were generated for the 2005 whole water samples (measured using a JYHoriba/Spex Fluoromax-3 spectrophotometer) and were scanned over an excitation range of 240 to 450 nm at 10 nm increments and an emission range of 350 to 550 nm at 2-nm increments with DataMax data acquisition software. To minimize quenching of the fluorescence signal due to metal complexation by iron, all samples were acidified to a pH of about 2. MilliQ water blanks were subtracted to remove Raman scattering and all samples were normalized to the Raman area to account for lamp decay over time. All samples were corrected for the inner-filter effect (Mobed et al. 1996) using the correction specified in McKnight et al (2001). High DOC samples (with concentrations . 20 mg C L21) were also diluted with MilliQ water so that absorbance (measured at 300 nm) was below 0.02. Corrections

751

and generation of EEMs was performed using MATLAB. Peaks in region A were identified and the intensities of those peaks were normalized to fulvic acid concentration for each sample, if available. To obtain the FI (McKnight et al. 2001), twodimensional spectra excited at 370 nm were generated and corrected for Raman scattering by blank subtraction. Two-dimensional spectra were collected on a Fluoromax-3 spectrofluorometer in 2005 and on a Fluoromax-2 spectrofluorometer in 2001 and 2002 and the spectra run on the two spectrofluorometers were corrected for instrument specific response (Cory 2005). FI values (dimensionless) were calculated from the ratio of intensities emitted at 470:520 nm (Cory 2005) with a confidence interval of 0.01. Among samples, collected over time from the same system and analyzed with the same instrument with appropriate corrections, changes in FI of 0.05 have been found to indicate shifts in dominant DOM source (Hood et al. 2003, Mladenov et al. 2005). Recently, PARAFAC has been used to decompose EEMs into different classes of fluorophores, referred to as components (Stedmon et al. 2003, Cory and McKnight 2005). Using a dataset of 379 DOM samples from diverse aquatic environments, Cory and McKnight (2005) developed a PARAFAC model that identified 13 individual components responsible for fluorescence and showed that quinone-like fluorophores accounted for about 50% of the fluorescence of every sample analyzed. In our study, EEMs of 23 ground-water samples collected in 2005 were fit to the 13-component model of Cory and McKnight (2005), and the relative amount (percent of total) of each component was measured. Model fit was considered suitable if intensities in the residual EEM, generated by subtracting the PARAFAC-modeled EEM from the measured EEM, were within 10% of measured EEM intensities. To assess redox state of aquatic fulvic acids, a redox index, as defined in Miller et al. (2006) was used. The RI is a ratio of reduced quinone components to total quinone components (sum of reduced and oxidized components) identified by the PARAFAC model (Miller et al. 2006). Of the 13 components identified by PARAFAC, the three quinone-like components (Q1, Q2, and Q3) represent oxidized components, and the semiquinoneand hydroquinone-like components (SQ1, SQ2, SQ3, and HQ) represent reduced components (Cory and McKnight 2005). Components found to be associated with microbial DOM sources (C2, C3, C6, C7, C8, C9, C12, and C13; Cory et al. 2007) were also examined.

752


Figure 3. Profiles of conductivity, chloride, sulfate, and total dissolved iron (Fe) in the ground water of island (left panels) and woodland (right panels) transects. In order to maintain figure clarity, distances and the y-axes of all chemical species are shown using a log scale. Mean values and standard deviations are shown. Surface water results are shown with a dashed line.

Given that the handling and transport of groundwater samples would have resulted in some atmospheric exposure, samples for fluorescence analysis were filtered and acidified immediately upon collection but not processed in an anoxic environment. Studies have found that reduced quinones can remain stable without precautions taken to limit oxygen exposure (Scott et al. 1998, Klapper et al. 2004). Even in waters with high iron concentration, the fluorescence spectra remained relatively unaltered in the presence of oxygen (Mladenov, unpublished). RESULTS Changes in Subsurface Chemistry along a Flowpath Island Transect. Along the island fringe (1–100 m), ground-water pH, alkalinity, DO concentrations, conductivity, and chloride and sulfate concentrations remained constant (Figures 2 and 3). Similarly, DOC concentrations were fairly constant along the island fringe, ranging from 11.4 to 16.1 mg C L21 (Figure 4). FA content also remained fairly constant

Figure 4. Profiles of DOC concentration, fulvic acid (FA) content, specific UV absorbance at 280 nm (SUVA), fluorescence index (FI), and redox index (RI) in the ground water of island (left panels) and woodland (right panels) transects. In order to maintain figure clarity, distances and DOC concentrations are shown using a log scale. Mean values and standard deviations are shown. RI data is presented only for samples collected on September 21 and December 2, 2005. Surface water means are shown with a dashed line and standard deviations are reported in Table 1.

in island fringe ground water (at about 70%), similar to that of water in the adjacent Boro channel and significantly higher than the estimate of 50% used in Bauer-Gottwein et al. (2007). Only temperature showed a gradual increase along the flowpath from surface water toward the island interior (Figure 2). At the island center (100–240 m), where the island vegetation structure is dominated by salt-tolerant grasses, distinct increases in ground-water pH, alkalinity, conductivity, chloride, sulfate, iron (Fe), and DOC concentration were observed (Figures 2– 4). Conductivity ranged from 5,030 to 18,860 mS cm21, the pH was basic, and total Fe concentrations ranged from 4.0 to 7.8 ppm. Chlo-

Mladenov et al., CHARACTERIZATION OF DOM IN WETLAND GROUND WATER

753

Table 1. Mean values of measured parameters (dissolved organic carbon (DOC), conductivity, fulvic acid (FA) content, specific UV absorbance (SUVA), fluorescence index (FI), and redox index (RI)) for surface water (SW) vs. adjacent ground-water (GW) samples at island and woodland transects for all sampling periods. Standard deviations and number of samples (in parentheses) are shown. Island Transect SW 21

DOC (mg C L ) Conductivity (mS cm21) FA content (%) SUVA (L mg21)a FIb RI

13.3 101 68 2.30 1.50 0.44

6 6 6 6 6

Fringe GW

Woodland Transect Center GW

2.6 (8) 13.8 6 3.1 (48) 170 6 76 (5)* 26 (4) 250 6 114 (19) 15016 6 5882 (5)** 4 (5) 70 6 4 (23) 58.1 6 5.7 (2) 0.10 (5) 2.32 6 0.40 (37) 1.59 6 0.37 (5)** 0.01 (8) 1.48 6 0.10 (18) 1.20 6 0.02 (4)** 0.41 6 0.01 (9) 0.58 6 0.01 (3)**

SW 16.6 178 65 2.70 1.45 0.44

6 6 6 6 6

GW

4.5 (9) 16.3 6 2.6 (17) 64 (5) 202 6 138 (15) 1 (5) 70 6 4 (5) 0.20 (6) 2.59 6 0.50 (15) 0.06 (9) 1.31 6 0.10 (12)* 0.46 6 0.004 (8)

*Two-sample student t-test (unequal variances) indicates significant difference from SW, with p , 0.01. **Two-sample student t-test (unequal variances) indicates significant difference from SW, with p , 0.001. a For comparison, SUVA of the terrestrial (SR) and microbial (LF) end-members are 2.5 and 1.3, respectively. b For comparison, FI of the terrestrial (SR) and microbial (LF) end-members are 1.24 and 1.74, respectively.

ride, generally considered a conservative ion, increased in concentration by about 3–4 orders of magnitude (from 0.3 ppm at the channel to 470 ppm at 240 m), while DOC concentrations increased by only one order of magnitude (from almost 10 mg C L21 at the channel to . 170 mg C L21 at 240 m). Significant differences were not found (using Student t-tests) in any of the parameters listed in Table 1 between ground-water samples collected from the island fringe and adjacent channel surface water. In contrast, ground-water samples collected from the island center had significantly higher DOC concentrations and conductivity than either island fringe ground water or channel surface water (Table 1). Woodland Transect. DOC concentrations ranged from 14.2 to 21.3 mg C L21 and FI ranged from 1.23 to 1.37 in ground water of the woodland transect (Figure 4). Student t-tests indicated that all measurements shown in Table 1, except FI, were not significantly different between woodland ground water and woodland surface water. Temperature and alkalinity were the only variables that showed an increase with distance toward the woodland interior (Figure 2). Changes in DOM Spectroscopic and Redox Properties along a Flowpath Along the island fringe, SUVA, FI, and RI values were constant (Figure 4). SUVA and FI decreased and RI increased in island center ground water. Student t-tests indicated that these differences between island center and island fringe ground water were significant (p , 0.001 for all). High RI values (. 0.5), indicative of a greater relative

abundance of reduced fulvic acids, observed at the island center transect were similar to values found in reduced shallow ground water of an alpine wetland (Miller et al. 2006). Student t-tests indicated that FI was significantly lower in island fringe and woodland ground water than in adjacent surface water. Lower FI signifies greater contributions from plant/ soil-derived DOM than from microbially derived DOM. Ground water of the island fringe had higher fluorescence intensities (normalized to fulvic acid concentration) than surface water (representative EEMs are shown in Figures 5A and 5B). EEMs from the seven piezometers at the island fringe transect contained a distinct shoulder in region C and peak in region A. The 200 m and 240 m EEMs (representative EEM is shown in Figure 5C) had lower region A intensities than EEMs of island fringe ground water and displayed a much broader shoulder in region C. Also the region A peaks of island center ground water were more red-shifted (to higher emission wavelengths) than in island fringe ground water. Along the woodland transect, the shape of the EEMs (representative EEMs are shown in Figures 5D through 5F) resembled those from the island fringe transect with a distinct shoulder in region C and peak in region A present in all samples. PARAFAC modeled EEMs matched measured EEMs with a residual of , 10% in all cases. In island fringe samples and in all woodland transect samples, the amounts of each component present in the EEM changed very little or not at all with distance along the flowpath. To illustrate the differences between the large number of samples more clearly, mean values of each component (expressed as percentage of total components) are

754

Figure 5. Representative EEMs of A) island surface water, B) island fringe ground water at 50 m, C) island center ground water at 240 m, D) woodland surface water, and woodland ground water at E) 50 m and F) 200 m collected on September 21 and December 2, 2005. Positions of the region A peak are shown below each EEM. Normalized intensities of the region A peak (in parentheses) are shown only for samples for which fulvic acid concentrations were known. Approximate locations of region A peak and region C shoulder are labeled with capital letters A and C, respectively, in Graph A.

shown for island fringe ground water, island center ground water, woodland ground water, and the corresponding channel and floodplain surface water samples and are grouped according to their molecular association (Table 2). Of the 13 components, C2, C3, C6, C7, C8, C9, and C13 are components that have been associated with microbial sources (Cory et al. 2007). Using the entire dataset of island and woodland transect samples, the sum of microbial fluorescent components was significantly related with FI (Figure 6A). Two of the quinone-like components (C2 and C11), the hydroquinone (C4), and the unknown component (C6), were the most prevalent in the island fringe and woodland samples. At the island center, reduced quinone content (specifically the hydroquinone C4 and the terrestrial semi-quinone C5) was higher than in island fringe ground water, whereas the content of oxidized quinones (C2, C11, and C12), tryptophan-like component (C8), and most microbially associated components were lower (Table 2). In all ground water and surface water samples, the total oxidized and reduced quinone content were significantly inversely related (R2 5 0.97, p , 0.01, Figure 6B). Differences in Subsurface Chemistry between Island and Woodland Transects While most ground-water properties were similar at the woodland and island fringe transects, student


Figure 6. Relationship between A) fluorescence index and total microbial components (as defined in Table 2) and B) between total oxidized (C2, C11, and C12) and total reduced (C4, C5, C7, and C9) quinone-like components. Dataset includes all ground-water samples (n 5 23) collected September 21 and December 2, 2005. Regression lines, equations and level of significance are shown. **p , 0.01.

t-tests indicated that DOC concentrations of woodland ground water were significantly higher (p 5 0.011, n 5 12) and SUVA and FI values were significantly lower than those measured at the island fringe transect (p , 0.001 for both). The only significant difference between shallow (approximately 3.3 to 4.5 m below surface) and deep (approximately 5 to 6 m below surface) piezometers was observed in ground-water conductivity measured in both the island and woodland transects (p 5 0.008, n 5 11), with higher conductivity occurring in shallower piezometers (Figure 3). DISCUSSION The critical role that islands play in maintaining the Delta as a freshwater system by acting as sinks for inorganic solutes (including dissolved inorganic carbon (DIC)) has been documented (Gieske 1997, McCarthy et al. 2006, Ramberg and Wolski 2007). The storage of organic matter in ground water may also have an important role in terms of influencing the biogeochemistry of the Delta. Our results, showing substantially higher DOC concentrations, conductivity, and alkalinity in island and woodland ground water than in adjacent surface water, confirm an enrichment of both dissolved organic and inorganic ions. These results are consistent with the known ground-water flowpath toward island interiors, maintained by surface water recharge (Wolski and Savenije 2006). In our study, DOC concentrations measured in island center ground water were an order of magnitude higher than those measured in island fringe ground water. Yet our island center measurements were an order of magnitude lower than those measured in the same piezometers (‘‘ORC island transect’’) in a previous

Mladenov et al., CHARACTERIZATION OF DOM IN WETLAND GROUND WATER

755

Table 2. Distribution of PARAFAC components in whole waters of island and woodland surface water (SW) and ground water (GW) shown as percent contribution of each component to the total modeled EEM. Standard deviations and number of samples (n) are also shown. Molecular association

PARAFAC componenta

Island Transect SW (n 5 1)

Woodland Transect

Fringe GW (n 5 9) Center GW (n 5 3)

SW (n 5 1)

GW (n 5 8)

Quinone C2 (Q2, M) C11 (Q1, T) C12 (Q3, M) Hydroquinone and semi-quinone C4 (HQ, both) C5 (SQ, T) C7 (SQ, M) C9 (SQ, M) Amino-acid C8 (Trp, M) C13 (Tyr, M) Unknown C1 (T) C3 (M) C6 (M) C10 (T)

20.4 13.1 8.2

21.8 6 0.3 13.4 6 0.1 9.3 6 0.5

16.5 6 0.3 10.1 6 0.3 5.6 6 0.4

20.6 13.0 8.4

20.7 6 0.8 13.0 6 0.1 7.7 6 0.4

21.2 5.3 4.4 2.4

18.2 4.3 5.6 2.8

29.9 9.2 3.1 2.6

1.0 0.5 0.4 0.2

21.3 5.4 3.9 1.9

22.9 5.8 4.0 1.9

6 6 6 6

0.4 0.2 0.3 0.3

6 6 6 6

6 6 6 6

0.2 0.1 0.4 0.1

0.5 3.0

1.5 6 0.5 2.4 6 0.6

0.0 2.2 6 0.1

0.8 3.3

0.1 6 0.2 2.8 6 0.5

7.0 3.9 8.2 2.5

6.9 4.9 7.0 1.9

5.8 2.4 9.9 2.7

7.0 3.4 8.6 2.5

7.0 3.0 8.5 2.4

6 6 6 6

0.3 0.2 0.3 0.4

6 6 6 6

0.1 0.3 0.2 0.2

6 6 6 6

01 0.1 0.3 0.1

a Components are labeled and identified according to Cory et al. (2007). M 5 components associated with microbially derived organic matter; T 5 components associated with terrestrially derived organic matter; Q 5 quinone; HQ 5 hydroquinone; SQ 5 semi-quinone; Trp 5 tryptophan; Tyr 5 tyrosine.

study (Bauer-Gottwein et al. 2007). This may be due, in part, to analytical differences between the two studies. Island center ground water is known to have high salt and inorganic C concentrations (McCarthy and Ellery 1994, Wolski et al. 2005, Wolski and Savenije 2005, Bauer-Gottwein et al. 2007, Wolski and Ramberg 2008), which present inherent analytical challenges in the measurement of DOC concentrations. For example, if inorganic C is not completely removed, measurements of DOC concentration can be overestimated (Potter and Wimsatt 2005). Also, different sample preservation techniques may influence DOC concentration measurements. Nevertheless, accumulation of DOM in ground water of the island center suggests that these zones serve as sinks for OM and is consistent with the model results of Mladenov et al. (2007b), which showed substantial infiltration of DOM (between 24% and 62% of total DOM removal in 2001–2002). The potential for a large carbon sink beneath other recharge wetlands, such as the Hadejia-Nguru wetlands in Nigeria (Goes 1999), the tree islands of the Everglades, Florida, USA (Gann and Childers 2006), and the River Murray floodplains in Australia (Holland et al. 2006), could have important implications for regional and global C budgets. The ultimate fate of organic C stored beneath wetland islands, however, merits further research.

Another important finding of this study is that the chemical and spectroscopic properties of DOM from surface water and the adjacent ground water were very similar, whereas the properties of DOM beneath island centers were clearly distinct from those of the island fringe, woodland transect, and adjacent surface waters. These patterns may be related to both hydrologic and biogeochemical processes. During the annual flood, ground-water table elevations in the island fringe and woodland can rise over 1 m, while ground-water table fluctuations in the island center are fairly low (between 0.10 m and 0.25 m), reflecting the evapotranspirative uptake of water along the flowpath (Wolski and Savenije 2006). More active ground-water recharge may explain, in part, the greater similarities in chemical and spectroscopic properties of surface water and adjacent (island fringe and woodland) ground water. In particular, the similar FI and SUVA values of surface water and island fringe and woodland groundwater suggests that ground-water DOM originates in DOM-rich channels and floodplains of the Okavango Delta. These similarities also reflect a dynamic hyporheic connection between surface water and ground water in the island fringe. Similar patterns between FI and SUVA at the woodland transect suggest that a dynamic hydrologic connection between surface water and adjacent ground water is also present at this site.

756 In comparison, the slow, 1–5 month long response to recharge by the annual flood in ground water beneath island centers (Wolski and Savenije 2006) results in long water travel times and may promote biogeochemical transformation of DOM in the subsurface. At the island transect, the conductivity and concentrations of conservative ions (chloride and sulfate) increased by about 500 fold along the ground-water flowpath from the channel surface water to the ground water at 240 m. In contrast, DOC concentrations increased only 10-fold over the same distance. This non-conservative behavior of DOM suggests that DOM evapoconcentration in the subsurface is offset by DOM removal processes such as coagulation and settling, sorption, and possibly microbial uptake (preferential) along the flowpath. High salt concentrations, such as those measured at this site, have been shown to result in destabilization of humic substances and subsequent coagulation in estuary waters (Sholkovitz 1976). Also, conditions of high pH (. 8.5) and high calcium concentration have been shown to induce a swelling/condensation transition of DOM to particulate organic matter (POM) microgels that can result in POM settling (Chin et al. 1998). Both of the former processes, previously observed in marine systems, warrant consideration in this ground-water setting. Furthermore, there may be preferential losses of aromatic DOM via sorption to sediments that would explain the decrease in SUVA from the island fringe (mean of 2.3 L mg C21) to the island center (less than 1 L mg C21 at 240 m). McKnight et al. (2002) observed a 50% reduction in SUVA in an alpine stream when abundant iron (Fe) oxyhydroxides were present on the streambed. This was attributed to surface complexation of strongly binding aromatic compounds with Fe oxides (McKnight et al. 2002). Additionally, nitrogen (N) and sulfur (S) groups in fulvic acid are known to be involved in strong metal binding (McKnight et al. 1992). In our study, a decrease in SUVA along the flowpath by this mechanism is consistent with the known high N and S content of infiltrating surface water (Mladenov et al. 2007) and the presence of Fe in Okavango sediments (Huntsman-Mapila et al. 2006). A corresponding shift to lower FA content, from 70%–80% in island fringe ground water to 50%–60% in island center ground water, may further reflect preferential sorption of the more hydrophobic organic acids, resulting in an increase in the non-humic fraction of DOM. Although low SUVA values have been associated with the presence of microbially derived DOM (Hood et al. 2003), the low FI values of island center groundwater suggest that the correspondingly

WETLANDS, Volume 28, No. 3, 2008 low SUVA is not likely to be related to increased microbial DOM sources. In other ground-water systems in which microbial DOM sources dominate, high FI values (approaching 1.90) have been reported (McKnight et al. 2001), but this is not the case in either of the ground-water transects of this study. In fact, the decrease in FI in island center ground water can be interpreted as indicating a loss of microbial precursor material in ground-water DOM, a finding supported by the highly significant relationship between FI and the sum of microbial fluorescent components (Figure 6A). Additionally, amounts of fluorescent components associated with microbial sources (Cory et al. 2007; Table 2), including component C8 that represents tryptophan-like fluorescence known to be associated with bacteria (Cammack et al. 2004), were lower in island center samples than in island fringe samples. These results further suggest that microbially derived moieties were also preferentially removed along the flowpath. Taken together with the removal of reactive fulvic acids (by sorption to sediments or coagulation in the saline ground-water environment), the loss of microbially derived fluorescent components along the flowpath means that a highly altered DOM, deficient in both aromatic moieties and microbial-type fluorophores, is transported to ground water beneath island centers. The preferential removal of reactive fulvic acids along the ground-water flowpath is likely responsible for the non-conservative behavior of DOM and lower DOC concentrations in island center ground water. These new findings of lower DOC concentrations than those measured by Bauer-Gottwein et al. (2007) and the accompanied lower fulvic acid content may help to resolve the contradictory findings (e.g., steady state composition of ground water was found to be sodium chloride dominated rather than sodium bicarbonate dominated) obtained when humics substances were included in model simulations (Bauer-Gottwein et al. 2007). Therefore, our findings support the model results of Bauer-Gottwein et al. (2007) that dissolved humic substances concentrations in island centers are not high enough to trigger CO2 degassing and delay the onset of density-driven flow. However, taking into account the potential sorption and coagulation of humics that may occur along the ground-water flowpath, the net influence of humic substances on the geochemistry of islands is likely to be substantial. Differences in redox state between island center and island fringe ground water also demonstrate the importance of humic substances in ground-water biogeochemistry. The presence of reduced fulvic acids (quantified using the RI) was more pro-

Mladenov et al., CHARACTERIZATION OF DOM IN WETLAND GROUND WATER nounced at island centers, where the highest DOM and Fe accumulation occurs, than at the island fringe or woodland sites, where lower DOC and Fe concentrations were also measured. Additionally, the highly significant correlation between reduced and oxidized quinone-like fluorescent components in ground water of the island and woodland transects (Figure 6B) indicates that the increase in reduced components is directly related to the loss of oxidized components and not other fluorophores. Further, the lower intensity and red-shifting (to higher emissions wavelengths) of the region A peak in island center ground water indicates more reducing conditions and is consistent with other studies (Klapper et al. 2002, Fulton et al. 2004) that attributed lower peak intensities to microbial reduction of fulvic acids. The redox state of fulvic acids in island center ground water is significant when considering the solubility of metals in the subsurface. The solubility and reactivity of Fe and manganese (Mn) has been linked to the electron-shuttling role of fulvic acids (Lovley et al. 1996, Klapper et al. 2002, Nevin and Lovley 2002, Fulton et al. 2005), and this role has been attributed specifically to quinone moieties (Cory and McKnight 2005). Quinones can shuttle electrons to facilitate metal reduction if Fereducing or other metal-reducing bacteria are present and if DOM (substrate, electron donor) and metals (electron acceptors) are present in sufficient concentration (Nevin and Lovley 2002, Klapper et al. 2002). In island center ground water, the dominance of reduced (over oxidized) quinone moieties in combination with high total Fe concentrations (reaching 8 ppm) suggests that an electron shuttling cascade may be underway that can promote metal dissolution in the subsurface. In ground water near the Okavango Delta, Huntsman-Mapila et al. (2006) found a positive correlation between specific UV absorbance and high dissolved arsenic concentrations and invoked a hypothesis of arsenic liberation through iron dissolution. Given the importance of this finding and its potential relationship to DOM cycling, a better understanding of DOM-redox-metal interactions is needed specifically for this system. Additionally, given the preferential removal of reactive fulvic acids along the flowpath in this study, the potential role of competitive sorption by DOM in promoting arsenic liberation should be evaluated. Finally, our results show that ground water beneath between bare island centers (lacking vegetation other than salt tolerant grasses) contains not only accumulated inorganic ions (McCarthy et al. 1993, McCarthy and Ellery 1995, Ramberg and Wolski 2007) but also DOM containing reduced fulvic acids. Whether the appearance of bare island

757

centers and salt crusts elsewhere in the Okavango Delta corresponds to similar ground-water organic geochemistry is a question for future research. The changes in chemistry and spectroscopic properties of DOM along the woodland transect flowpath were similar to those observed at the island fringe, but from this study alone it is not possible to determine whether greater accumulation (as occurs at the island center) also occurs with greater distance inland in woodland areas adjacent to seasonal floodplains or whether sustained recharge of DOM by a permanent water supply is needed to facilitate this condition.

CONCLUSION The biogeochemical significance of islands in global wetlands is just beginning to be understood. Specifically, the chemical character of ground-water DOM may have an important influence on the biogeochemistry of ground water beneath wetland islands. Our findings provide chemical evidence for the non-conservative behavior of DOM in the subsurface and indicate that ground water beneath island centers has undergone a greater degree of biogeochemical processing. The potential removal of reactive humic substances with distance along a flowpath may explain why Bauer-Gottwein et al. (2007) concluded that humic substances are not found in high enough concentrations to drive degassing of CO2 in island center ground water. However, the interactions of humic substances in ground water may be extremely important in terms of biogeochemical processes, such as metal-DOM interactions, electron shuttling, sorption, and/or coagulation. Therefore, we conclude that humic substances probably exert a significant influence on the geochemistry of ground water beneath islands. The significant differences we observed in chemical and spectroscopic properties between ground water of the island fringe and island center provide evidence for surface water sources of ground-water DOM, accumulation of DOM in the subsurface, and the occurrence of important redox processes in the ground water beneath island centers of the Okavango Delta. The reducing conditions in the ground water of island interiors may be linked to microbial reduction of metals using the DOM as substrate and fulvic acids as electron shuttles. Our findings have important implications for the Okavango Delta and other net recharge wetlands in regards to estimating carbon budgets, understanding redox processes, and evaluating the influence of DOM and humic substances on ground-water geochemistry.

758

WETLANDS, Volume 28, No. 3, 2008 ACKNOWLEDGMENTS

We are grateful to I. Mosie, B. Mogojwa, K. Mohembo for field expertise and assistance, K. Mohembo, F. Luiszer, M. P. Miller, and M. Norris for assistance with chemical analyses, R. D. McGrath and L. Ries for assistance with sample transport, and M. P. Miller and anonymous reviewers for helpful comments on the manuscript. LITERATURE CITED Bauer-Gottwein, P., T. Langer, H. Prommer, P. Wolski, and W. Kinzelbach. 2007. Okavango Delta islands: interactions between density-driven flow and geochemical reactions under evapo-concentration. Journal of Hydrology 335:389–405. Cammack, W. K. L., J. Kalff, Y. T. Prairie, and E. M. Smith. 2004. Fluorescent dissolved organic matter in lakes: relationships with heterotrophic metabolism. Limnology and Oceanography 49:2034–45. Chin, Y., G. Aiken, and E. O’Loughlin. 1994. Molecular weight, polydispersivity, and spectroscopic properties of aquatic humic substances. Environmental Science and Technology 28: 1853–58. Chin, W., M. V. Orellana, and P. Verdugo. 1998. Spontaneous assembly of marine dissolved organic matter into polymer gels. Nature 39:568–72. Coble, P. G. 1995. Characterization of marine and terrestrial DOM in seawater using excitation-emission matrix spectroscopy. Marine Chemistry 51:325–46. Cory, R. M. 2005. Redox and photochemical reactivity of dissolved organic matter in surface waters. Ph.D. Dissertation. University of Colorado, Boulder, CO, USA. Cory, R. M. and D. M. McKnight. 2005. Fluorescence spectroscopy reveals ubiquitous presence of oxidized and reduced quinones in dissolved organic matter. Environmental Science and Technology 39:8142–49. Cory, R. M., D. M. McKnight, Y. Chin, P. Miller, and C. L. Jaros. 2007. Chemical characteristics of fulvic acids from Arctic surface waters: microbial contributions and photochemical transformations. Journal of Geophysical Research, 112, G04S51, doi:10.1029/2006JG000343. Fulton, J. R., D. M. McKnight, C. M. Foreman, R. M. Cory, C. Stedmon, and E. Blunt. 2004. Changes in fulvic acid redox state through the oxycline of a permanently ice-covered Antarctic lake. Aquatic Sciences 66:27–46. Gann, T. T. and D. L. Childers. 2006. Relationships between hydrology and soils describe vegetation patterns in seasonally flooded tree islands of the southern Everglades, Florida. Plant and Soil 279:271–86. Gieske, A. 1997. Modelling outflow from the Jao/Boro river system in the Okavango delta, Botswana. Journal of Hydrology 193:214–39. Hood, E., D. M. McKnight, and M. W. Williams. 2003. Sources and chemical character of dissolved organic carbon across an alpine/subalpine ecotone, Green Lakes Valley, Colorado Front Range, United States. Water Resources Research, 39, 1188, doi:10.1029/2002WR001738. Huntsman-Mapila, P., T. Mapila, M. Letshwenyo, P. Wolski, and C. Hemond. 2006. Characterization of arsenic occurrence in the water and sediments of the Okavango Delta, NW Botswana. Applied Geochemistry 21:1376–91. Klapper, L., D. M. McKnight, J. R. Fulton, E. L. Blunt-Harris, K. P. Nevin, D. R. Lovley, and P. G. Hatcher. 2002. Fulvic acid oxidation state detection using fluorescence spectroscopy. Environmental Science and Technology 36:3170–75. Lovley, D. R., J. D. Coatest, E. L. Blunt-Harris, E. J. P. Phillipst, and J. C. Woodward. 1996. Humic substances as electron acceptors for microbial respiration. Nature 382:441–48.

McCarthy, T. S. 2006. Groundwater in the wetlands of the Okavango Delta, Botswana, and its contribution to the structure and function of the ecosystem. Journal of Hydrology 320:1–19. McCarthy, T. S. and W. N. Ellery. 1994. The effect of vegetation on soil and ground-water chemistry and hydrology of islands in the seasonal swamps of the Okavango-Fan, Botswana. Journal of Hydrology 154:169–93. McCarthy, T. S. and W. N. Ellery. 1995. Sedimentation on the distal reaches of the Okavango Fan, Botswana, and its bearing on silcrete and calcrete (ganister) formation. Journal of Sedimentary Research 65:77–90. McCarthy, T. S., W. N. Ellery, and K. Ellery. 1993. Vegetationinduced, subsurface precipitation of carbonate as an aggradational process in the permanent swamps of the Okavango (delta) fan, Botswana. Chemical Geology 107:111–31. McKnight, D. M., K. E. Bencala, G. W. Zeiiweger, G. R. Alken, G. L. Feder, and K. A. Thorn. 1992. Sorption of dissolved organic carbon by hydrous aluminum and iron oxides occurring at the confluence of Deer Creek with the Snake River, Summit County, Colorado. Environmental Science and Technology 26:1388–96. McKnight, D. M., E. W. Boyer, P. K. Westerhoff, P. T. Doran, T. Kulbe, and D. T. Andersen. 2001. Spectrofluorometric characterization of dissolved organic matter for indication of precursor organic material and aromaticity. Limnology and Oceanography 46:38–48. McKnight, D. M., R. Harnish, R. L. Wershaw, J. S. Baron, and S. Schiff. 1997. Chemical characteristics of particulate, colloidal, and dissolved organic material in Loch Vale Watershed, Rocky Mountain National Park. Biogeochemistry 36: 99–124. McKnight, D. M., G. M. Hornberger, K. E. Bencala, and E. W. Boyer. 2002. In-stream sorption of fulvic acid in an acidic stream: a stream-scale transport experiment. Water Resources Research 38(1), doi:10.1029/2001WR000269. Miller, M. P., D. M. McKnight, R. M. Cory, M. W. Williams, and R. L. Runkel. 2006. Hyporheic exchange and humic redox reactions in an alpine stream/wetland ecosystem, Colorado Front Range. Environmental Science and Technology 40:5943–49. Mladenov, N. 2004. Evaluation of the effects of hydrologic change in the Okavango Delta of Botswana: analyses of aquatic organic matter transport and ecosystem economics. Ph.D. Dissertation. University of Colorado, Boulder, CO, USA. Mladenov, N., D. M. McKnight, S. A. Macko, L. Ramberg, R. M. Cory, and M. Norris. 2007a. Chemical characterization of DOM in channels of a seasonal wetland. Aquatic Sciences 69:456–71. Mladenov, N., D. M. McKnight, P. Wolski, and M. MurrayHudson. 2007b. Simulation of DOM fluxes in a seasonal floodplain of the Okavango Delta, Botswana. Ecological Modelling 205:181–95. Mladenov, N., D. M. McKnight, P. Wolski, and L. Ramberg. 2005. Effects of annual flooding on dissolved organic carbon dynamics within a pristine wetland, the Okavango Delta, Botswana. Wetlands 25:622–38. Mobed, J. J., S. L. Hemmingsen, J. L. Autry, and L. B. McGown. 1996. Fluorescence characterization of IHSS humic substances: total luminescence spectra with absorbance correction. Environmental Science and Technology 30:3061–65. Nevin, K. P. and D. R. Lovley. 2002. Mechanisms for Fe(III) oxide reduction in sedimentary environments. Geomicrobiology Journal 19:141–59. Potter, B. B. and J. C. Wimsatt. 2005. Method 415.3. Measurement of total organic carbon, dissolved organic carbon and specific UV absorbance at 254 nm in source water and drinking water. U.S. Environmental Protection Agency, Washington, DC, USA. Ramberg, L. and P. Wolski. 2007. Growing islands and sinking solutes: processes maintaining the endorheic Okavango Delta as a freshwater system. Plant Ecology 196:215–31.

Mladenov et al., CHARACTERIZATION OF DOM IN WETLAND GROUND WATER Scott, D. T., D. M. McKnight, E. L. Blunt-Harris, S. E. Kolesar, and D. R. Lovley. 1998. Quinone moieties act as electron acceptors in the reduction of humic substances by humicsreducing microorganisms. Environmental Science and Technology 32:2984–89. Sholkovitz, E. R. 1976. Flocculation of dissolved organic and inorganic matter during the mixing of river water and seawater. Geochimica and Cosmochimica Acta 40:831–45. Stedmon, C. A., S. Markager, and R. Bro. 2003. Tracing dissolved organic matter in aquatic environments using a new approach to fluorescence spectroscopy. Marine Chemistry 82:239–54. Thurman, E. M. 1985. Organic Geochemistry of Natural Waters. Martinus Nijhoff/Junk, Boston, MA, USA. Thurman, E. M. and R. L. Malcolm. 1981. Preparative isolation of aquatic humic substances. Environmental Science and Technology 15:463–66.

759

Weishaar, J. L., G. R. Aiken, B. A. Bergamaschi, M. S. Fram, R. Fujii, and K. Mopper. 2003. Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon. Environmental Science and Technology 37:4702–08. Wolski, P., M. Murray-Hudson, P. Fernkvist, A. Liden, P. Huntsman-Mapila, and L. Ramberg. 2005. Islands in the Okavango Delta as sinks of water-borne nutrients. Botswana Notes and Records 37:253–63. Wolski, P. and H. Savenije. 2006. Dynamics of surface and groundwater interactions in the floodplain system of the Okavango Delta, Botswana. Journal of Hydrology 320:283– 301. Manuscript received 29 July 2007; accepted 25 April 2008.