Redox Processes Controlling Manganese Fate and Transport in a

Jan 4, 2002 - Institute of Arctic and Alpine Research, University of Colorado at Boulder, Campus Box 450, Boulder, Colorado 80309-0450,. School of Nat...
0 downloads 0 Views 115KB Size
Environ. Sci. Technol. 2002, 36, 453-459

Redox Processes Controlling Manganese Fate and Transport in a Mountain Stream D U R E L L E T . S C O T T , * ,† DIANE M. MCKNIGHT,† BETTINA M. VOELKER,‡ AND DUANE C. HRNCIR§ Institute of Arctic and Alpine Research, University of Colorado at Boulder, Campus Box 450, Boulder, Colorado 80309-0450, School of Natural Sciences and Mathematics, Mesa State University, 1100 North Avenue, Grand Junction, Colorado 81501, and Department of Civil and Environmental Engineering, MIT 48-415, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

The biogeochemical processes controlling the speciation and transport of manganese in a Colorado mountain stream were studied using a conservative tracer approach combined with laboratory experiments. The study stream, Lake Fork Creek, receives manganese-rich inflows from a wetland contaminated by acid mine drainage. A conservative tracer experiment was conducted on a 1300-m reach of the stream. Results indicate that manganese was progressively removed from the stream, resulting in a loss of 64 ( 17 µmol day-1 m-1. Laboratory experiments using streamwater, mine effluent, and rocks from the stream showed the importance of surface-catalyzed oxidation and photoreduction on the speciation of manganese. The field and modeling results indicate that light generally promotes oxidation and removal of manganese from the stream, presumably through a photosynthetically enhanced oxidation process. Differences in Mn speciation within the stream suggest that reductive processes affect Mn speciation within the water column. These results identify the rapid oxidation and precipitation of MnOx as a dominant process within this freshwater stream.

Introduction The transport and biogeochemical cycling of metals in streams is controlled by a range of processes that may directly affect nutrient bioavailability and metal toxicity. Manganese is a redox-sensitive metal whose dissolved concentrations have been observed to undergo diel variations in streams (1). Because both toxic metals and nutrients may coprecipitate with or sorb to manganese oxides (MnOx), processes controlling the precipitation and dissolution of Mn could be important to stream ecosystems with abundant MnOx on the streambed (2). The effects of light on manganese cycling and transport in freshwater streams are not well understood (3). Light may lead to an increase in manganous ion (Mn2+), the reduced * Corresponding author present address: Landcare Research, Private Bag 11 052, Palmerston North, New Zealand; phone: +64 6 356 7154; fax: +64 6 355 9230; e-mail: [email protected]. † Univeristy of Colorado at Boulder. ‡ Massachusetts Institute of Technology. § Mesa State University. 10.1021/es010951s CCC: $22.00 Published on Web 01/04/2002

 2002 American Chemical Society

form of manganese, through a photoreductive mechanism. Previous studies have shown that MnOx in seawater is photoreduced in the presence of dissolved organic carbon (DOC) (4). Mn2+ in coastal seawater exhibits a diel pattern partially due to a photoreductive mechanism, with Mn2+ concentration reaching its maximum in the late afternoon (5). Laboratory studies have shown that the nature of the oxide is important in the overall rate of photoreduction and the type of photoreductive mechanism (6, 7). The photoreductive mechanisms thought to be involved in Mn photoreduction are either a direct ligand-to-metal chargetransfer reaction (LMCT) or an indirect mechanism involving DOC photoproducts such as hydrogen peroxide (H2O2) (5, 7, 8) (Table 1, eqs 1 and 2). Light may also serve to decrease the concentrations of manganese in freshwater, by increasing the rate of oxidation in response to photosynthesis by algae (Table 1, eq 3). Richardson et al. have shown that dissolved Mn concentrations decrease in the presence of light and an active photosynthesizing algal community (9). Because the rate of oxidation increases as a function of [OH-]2, enhanced oxidation occurs when the microenvironment around the algae increases in pH due to photosynthesis (10). To determine the relative importance of light on oxidative and reductive processes pertaining to manganese, it is necessary to examine manganese speciation and transport over the course of a full diel cycle. Previous studies attributed the diel variations in Mn concentration in acid mine drainage streams to redox processes occurring in sediments (1). Results from another field study highlighted the importance of microbially enhanced oxidation within the hyporheic zone (11). Previous diel studies examining the diel variations of Mn have only examined the changing concentrations in Mn. Our study examines the flux of manganese within the system by quantifying the flow rate within the stream and modeling the transport of manganese within the stream. We examined the flux of manganese over a 32-h period in Lake Fork Creek, a mountain stream that receives acid mine drainage (AMD) from an abandoned mine. To gain a better understanding of the possible reactions involved in manganese transport, conservative tracer experiments were carried out to quantify the hydrology of the system. This approach allows for a detailed accounting of lateral inflow from the riparian zone, inflow from groundwater, and hyporheic zone exchange processes (12). Laboratory experiments were also performed to study photoreductive dissolution of MnOx, the influence of DOC and DOC photoproducts, and surface-catalyzed oxidation under conditions relevant to our field system. The flux of manganese and reactive transport processes were then quantified using a one-dimensional solute transport model.

Experimental Methods Study Location. The study site is located on Lake Fork Creek in Lake County, approximately 8 km west of Leadville, CO (Figure 1). The stream is part of the upper Arkansas River watershed, an area containing numerous abandoned mines that discharge AMD into the Arkansas River. The experimental stream reach is downstream of Turquoise Reservoir and received a constant discharge over the course of the study period (August 10-11, 1999) (0.418 m3 s-1). In the beginning of the stream reach, an adit (Dinero Tunnel) discharges AMDcontaminated water into a large wetland, which then drains into the stream (Figure 1). Several small inflows are located along the upper portion of the stream reach along with one primary inflow upstream of site 1. Downstream of the wetland, the rocks within the stream are coated with VOL. 36, NO. 3, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

453

TABLE 1. Dominant Reductive and Oxidative Reactions Involving Mna (1) MnO2 + HA f HAreduced + Mn2+ + O2 where HA ) humic or fulvic acid; rate of this reaction is enhanced in the presence of light indirect (2) MnO2 + H2O2 + 2H+ f Mn2+ + O2 + photoreduction 2H2O where H2O2 is photoproduced oxidation (3) Mn2+ + 2H2O f MnO2 + 4H+; rate of this of Mn2+ reaction in enhanced by sorption of Mn2+ and biological processes

reductive dissolution

a MnO represents colloidal Mn within the water column and oxides 2 on the stream bottom.

FIGURE 1. Map of the Lake Fork watershed and experimental reach. amorphous MnOx. Upstream of the wetland, algae are abundant in the stream but decrease in the study reach below the wetland. The injection site for our conservative tracer was located approximately 10 m upstream of the wetland. Two sampling stations were located downstream of the wetland inflows, at approximately 560 m (site 1) and 1255 m (site 2) downstream of the injection site. Samples were also collected at a small pool adjacent to the stream (at 560 m) open to the sun with a large surface area to volume ratio and a very low flow rate, functioning as a dead zone (12). The average depth of the pool was 0.1 m. Conservative Tracer Experiment. During the course of the experiment (August 11-12, 1999), a conservative tracer (NaCl) was injected for 32 h upstream of both sampling sites to quantify the hydrological characteristics of the system. Sodium chloride was injected at a concentration of approximately 230 g L-1 NaCl and a rate of 230 mL min-1 into the stream for 32 h, and intensive sampling was performed during the arrival and departure of the tracer at each of the sites. The concentration of injectate varied slightly during the middle of the experiment, leading to changes in chloride downstream. The changing upstream boundary was incorporated into the model. Diel sampling of Mn in Lake Fork was also conducted on August 11-12, 2000. Measurements included conductivity, pH, major cations and anions, DOC, and hydrogen peroxide (H2O2). Samples for pH and conductivity analysis were collected and transferred into 125-mL polyethylene bottles. Samples for anions analysis were collected and filtered through a 0.45-µm Gelman filter capsule and then transferred into a 125-mL polyethylene bottle. Chloride and sulfate were measured using a Dionex ion chromatograph. Samples for filterable manganese analysis were filtered through a 0.45-µm Gelman filter capsule and then transferred into an acid-rinsed bottle. Both the total and filterable manganese samples were acidified with 0.5 mL of Ultrex nitric acid. Total and filterable manganese for all of the samples were analyzed using a Perkin-Elmer 100 Analyst flame atomic absorption spectrometer. Manganous ion was also measured during the experiment on the filtered samples. Samples for manganous ion were taken from the anion sample bottle in the field and 454

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 3, 2002

measured onsite within 1 h using a chemiluminescent approach (13). Briefly, a subsample was pipetted into a reaction vessel with 0.1 N NaOH. The reaction vessel was placed into a TD 20/20 luminometer, and the chemiluminescent reagent containing TCNQ was automatically injected into the reaction chamber. The resulting signal was recorded and integrated and was calibrated using standard additions. DOC samples were collected and filtered through precombusted 0.45-µm glass fiber filters and stored in precombusted amber borosilicate glass bottles. DOC samples were stored in the dark at 5 °C and were run on a Dohrman DC190 carbon analyzer within 5 days. For hydrogen peroxide measurements, 20 mL of sample filtered through the 0.45µm Gelman capsule was immediately transferred into amber polyethylene bottles containing 2 mL of the H2O2 reagent (horseradish peroxidase and p-hydroxyphenylacetic acid) (14). The H2O2 samples were analyzed using a fluorometer within 36 h; the fluorescence signal was shown to be stable over this time period in field tests. UV-a (einstein m-2) was recorded from 320 to 400 nm in the field using a hand-held meter. Laboratory Experiments. Laboratory experiments were performed using filtered streamwater collected above and below the wetland and using wetland inflow water. Rocks stained with MnOx were collected in the field and air-dried. The initial laboratory experiment was designed to assess the rate of removal of dissolved Mn, presumably by surfacecatalyzed oxidation and precipitation, in the presence and absence of streambed rocks. Streamwater from above the wetland was mixed with wetland inflow water at a ratio of 10:1 in a mixed reactor that was kept in the dark. Subsamples were collected every 10 min, filtered through a 0.1 µM Gelman filter, and analyzed for filterable Mn and Zn by atomic adsorption spectroscopy as described above. After 60 min, two small rocks (average cross section 6.5 cm2) were added to the reactor. Because the rocks had been air-dried prior to being placed into the reactor, the initial activity of the microbial flora on the rock surface would not be expected to be as great as in the field, limiting the importance of bacterial-catalyzed oxidation in this experiment. The rate of photoreductive dissolution of the MnOx surface coatings on streambed rocks was also assessed in a series of three laboratory experiments. An Oriel solar simulator was used to irradiate samples at 600 W in a jacket-cooled stirred beaker. In the first experiment, two rocks were placed into the beaker with 220 mL of filtered streamwater from above the wetland, irradiated for 60 min, and compared to a dark control. In the second experiment, the light and dark beakers were amended with 10 mg L-1 Suwannee River fulvic acid (SRFA) in order to assess the importance of fulvic acid. To assess the importance of Mn reduction by photoproduced H2O2, two rocks were placed in a beaker with 220 mL of filtered upstream streamwater spiked with 13 µM H2O2 and kept in the dark. In all of these experiments, subsamples were collected, filtered through a 0.1 µM Gelman filter, and measured for filterable Mn and H2O2. An experiment was also performed in the field alongside the stream channel using 500-mL acid-rinsed jars. Three small rocks removed from the stream (each rock was approximately 5 cm in diameter) were placed into each of the 12 jars, of which half were painted black and then white to prevent the transmittance of light. In the first set of 4 jars (2 dark and 2 light), 400 mL of deionized (DI) water was added. In the second set, 400 mL of DI water with 10 mg L-1 SRFA was added. In the third set, 400 mL of DI water with 10 mg L-1 phthalic acid was added as a model organic ligand that has been shown to enhance photoreactivity of iron oxides (15). The jars were exposed to the sun from 10:00 h to 14:00 h, and subsamples were taken at the beginning and end of the

exposure period. Modeling Approach. The data from the tracer experiment were analyzed to quantify rates of Mn removal from the streamwater as influenced by diel processes. The chloride data were used to determine the hydrological characteristics of the stream using a one-dimensional transport model with storage and inflow (OTIS) (16). Since the primary assumption is that chloride is conservative in this system, the concentrations of chloride in the main stream channel (C, in µM) and in the storage (hyporheic) zone (Cs, in µM) as a function of distance along the stream reach (x) and time (t) are given by the following equations:

qL Q ∂C 1 ∂ ∂C ∂C )+ AD + (CL - C) + ∂t A ∂x A ∂x ∂x A R(Cs - C) - λ(C) (1)

(

)

dCs A ) R (C - Cs) dt As

FIGURE 2. Measured UV-a (320-400 nm) on August 10-11, 1999.

(2)

where Q is discharge (m3 s-1), A is the cross-sectional area of the stream (m2), CL is the lateral inflow concentration (µM), D is the dispersion coefficient (m2 s-1), qL is the lateral inflow rate (m3 s-1 m-1), R is the exchange coefficient (s-1), and As is the storage zone cross-sectional area (m2). The lateral inflow concentration CL was set to the background levels for chloride, and both A and Q were measured in the field and confirmed from a gauging station 500 m upstream of the injection site. Using the chloride injectate flux as the upstream boundary condition, values of D, qL, R, and As which best fit the chloride observations at 560 and 1255 m were determined by applying the inverse modeling software UCODE (17). The observations in the shoulder and tail of the breakthrough curve were given higher weights, for these areas have higher sensitivities to the transient storage parameters (18). UCODE then determines the optimum parameter values by minimizing the least-squares objective function with respect to the parameter values (17). Equations 1 and 2 were also used to model the conservative transport of Mn. Site 1 was used as the upstream continuous boundary condition. The hydrological parameters determined from the chloride data were then applied from 560 to 1255 m. Using these estimated parameters, total manganese, filterable manganese, and manganous ion at site 2 were predicted assuming conservative transport from site 1. The first-order decay approach is based on the dependence of the rate of surface-catalyzed reactions on the concentration of manganous ion. A value of the first-order decay coefficient, λ (s-1), was then estimated for each of the manganese species for the entire reach using inverse modeling.

Results Field Experiments. Over the course of the tracer experiment on August 10-11, 1999, several water quality parameters, including pH, sulfate, and DOC, did not vary to a great extent. The pH of the stream was approximately 6.9 ( 0.4. Sulfate decreased during the nighttime and increased during the day (see Figure S-1, Supporting Information). The pH of the primary wetland inflow was 3.55 and had a Mn concentration of 0.24 mM. Dissolved organic carbon was approximately 292 ( 17 µM at all of the sites. The background chloride concentration was 4.0 µM. The measured UV varied greatly over the course of the day but was most variable in the afternoon of the first day due to clouds. On the second day, a large storm passed through the area in the mid-morning, reducing UV light significantly (Figure 2). Hydrogen peroxide concentrations varied between 0 and 0.4 µM and were highest at approximately 14:00 h (see Figure S-1, Supporting Information).

FIGURE 3. Filterable manganese, total manganese, and manganous ion at site 1, site 2, and pool site (560 m). SD ( 0.6 µM for Mnt and Mnf; SD ( 1.1 µM for Mn2+ approximated from 3 separate replicates. Filterable and total manganese concentrations varied over the course of the experiment (Figure 3). Similar variations in Mn concentrations were also observed within the stream during the diel sampling in August 2000 (see Figure S-2, Supporting Information). At site 1, total manganese (Mnt) is slightly higher than filterable manganese (Mnf) and manganous ion (Mn2+) throughout the experiment, with the exception of the morning hours on the second day. The concentrations of Mnt, Mnf, and Mn2+ at the pool site generally follow the manganese pattern at site 1. The observed pattern at site 2 follows the same pattern as site 1, although the concentrations of manganese are lower. Mnf at site 2 VOL. 36, NO. 3, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

455

FIGURE 4. Filterable manganese and zinc concentrations as a function of time in a dark, stirred reactor with AMD-amended streamwater. Rocks were added at 60 min. SD ( 0.6 µM.

FIGURE 5. Changes in filterable manganese concentrations as a function of irradiation time and FA amendment (0 mg of C L-1; 10 mg of C L-1) in Lake Fork streamwater (from upstream of the AMD inflow). SD ( 0.6 µM. does not have the same dip in the second morning as at site 1. Batch Experiments. The downstream and temporal patterns of manganese concentrations in Lake Fork Creek suggest the rapid removal of Mn by oxidation and precipitation. To assess the importance of surface-catalyzed Mn oxidation within the stream and the effects of this process on concentrations of toxic metals, rates of removal from solution were measured in batch experiments conducted in the dark (Figure 4). During the initial 60 min in the absence of rocks, the concentration of manganese and zinc was constant, at 79 and 11.3 µM, respectively. Once rocks collected from the field were added to the beaker, rapid losses of filterable manganese and zinc were observed, with a linear trendline indicating a constant loss rate from 60 to 240 min. For manganese, the slope had a value of 45.1 nM min-1. For zinc, the slope was 16.4 nM min-1. To confirm that the MnOx coating the rocks from the streambed at Lake Fork is photoreactive, rocks from the field were irradiated with a solar simulator. Results indicated that irradiation led to an overall increase in Mnf as compared to the dark controls (Figure 5). In the reactors with added fulvic acid, a larger increase was measured in the dark than for the unamended irradiated sample. The fulvic acid-amended and irradiated reactor had the largest increase in filterable manganese. To investigate the importance of photoproduced H2O2 as a reductant of MnOx, a reactor that contained rocks and streamwater was placed in the dark, and the reactor was then spiked with H2O2. Over the course of the experiment, H2O2 decreased while Mnf increased (Figure 6). Approximately 4 mol of H2O2 was consumed for every mole of Mn produced. 456

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 3, 2002

FIGURE 6. Changes in manganese and H2O2 concentrations in the dark with MnOx-coated rocks in Lake Fork streamwater. The solution was spiked with 13 µM H2O2.

FIGURE 7. Increase in filterable manganese over the course of 4 h under natural sunlight in deionized water with rocks from the stream. Concentrations of fulvic acid and phthalic acid were 10 mg of C L-1. Initial Mnf concentrations were approximately 0.5 µM before exposing the beakers to sunlight. Field Photoreduction. The influence of natural sunlight was measured in the field using rocks from the stream (Figure 7). Over the course of 4 h, the transparent reactors for all of the treatments had higher concentrations of manganese as compared to the dark controls. The unamended light reactor had the lowest increase in filterable Mn among the transparent reactors, followed by the reactors amended with fulvic and phthalic acids. Simulation of Manganese Transport. Chloride was used as a conservative tracer in the solute transport modeling to determine the hydrological characteristics of reach 1 (0-560 m) and reach 2 (560-1255 m) (Table 2, Figure 8). The inverse modeling results were unable to support transient storage in reach 1 due to the lack of resolution in the shoulder and tail of the breakthrough curve. However, inverse modeling showed in reach 2 that transient storage was an important component to solute transport. The optimized parameter values were then used to simulate conservative transport of manganese downstream from site 1 to site 2 during the experiment (Figure 9). The manganese conservative transport simulation yielded higher concentrations than the observed measurements for Mnt, Mnf, and Mn2+ (Figure 9). This result confirms that manganese was lost from the system. From site 1 to site 2, the average loss of Mnt calculated by taking the difference between conservative transport and the observations, multiplying by the discharge, and dividing by the reach length was 64 ( 17 µmol day-1 m-1. This value represents the average deposition rate of Mn on the streambed in that reach under the summer conditions. An average first-order decay coefficient λ (s-1) for the main water column was determined for Mnt, Mnf, and Mn2+ between sites 1 and 2 using all of the observations at site 2

TABLE 2. Hydraulic and Transport Parameters for Reach 1 and Reach 2a reach

D (m2 s-1)

As (m2)

r (s-1)

qL (m3 s-1 m-1)

A (m2)

1 2

1.2 (0.70-2.0) 1.2 (0.70-2.0)

0 0.69 (0.38-1.2)

0 1.5 × 10-4 (3.2-6.9 × 10-4)

1.55 × 10-4 (1.26-1.90 × 10-4) 0

1.67 (1.60-1.75) 1.73 (1.61-1.87)

a

The parentheses represent the upper and lower 95% confidence intervals.

FIGURE 8. Chloride conservative simulation and observed values at 560 and 1255 m on August 10-11, 1999. to estimate λ (Figures 9 and 10). The simulations including the estimated first-order decay for Mnt, Mnf, and Mn2+ provide a good representation of the data. A close comparison of the data and simulations reveals a pattern of positive and negative residuals with respect to time (Figure 11), implying that Mn decay is not constant over the course of 32 h. In general, Mnt and Mnf exhibit opposite behavior during the daytime hours, but both residuals are positive during the nighttime hours, indicating that λ is overestimated. To further assess the differing decay rates for manganese, two time periods were used as observations to estimate the best λ value for each period: nighttime (19:00-06:00) and daytime (6:00-19:00) (Figure 10). These λ values were compared to the λ value estimated using all of the observations in the calibration. For Mnt, Mnf, and Mn2+, the estimated λ values were all lower during the nighttime hours as compared to the daytime hours.

Discussion Our study shows that manganese concentrations in Lake Fork varied considerably over the course of a 24-h period (Figure 3, Figure S-2). The overall pattern of manganese variation in the stream is controlled by the changes in Mn input due to hydrologic and biogeochemical processes occurring in the wetland, upstream of site 1 (19). A previous study of the seasonal and diel changes in the adit-wetland system showed that the flow from the adit is generally constant during the day and that variations in the mass flow of Mn draining from the wetland to the stream during the day could be caused by plant uptake, photochemical processes in the small pools in the wetland, or hydrologic variation (19). Photosynthetic activity by wetland plants and

FIGURE 9. Conservative and first-order decay simulations for manganese (total, filterable, and manganous ion) from site 1 to site 2. algal mats may contribute to the dynamic pattern of Mn observed in the stream by increasing the pH in microenvironments and enhancing the rate of Mn oxidation in the wetland (9). The laboratory experiments using rocks from the stream provide insight into some of the in-stream processes, which may modulate Mn concentrations below the Mn input from the wetland. The rapid and sustained decrease in Mn concentrations in the dark experiment with added rocks (Figure 4) indicates the importance of surface-catalyzed oxidation in the stream. The wide and shallow nature of the streambed enhances contact of the streamwater with the rock surfaces and would enhance the importance of any surface reactions in the stream. The experiments in which the rocks were exposed to light showed that direct photochemical reduction of manganese oxides could also be important and was enhanced by the presence of dissolved organic compounds (Figures 5 and 7). The rate of reductive dissolution of MnOx in the dark is also substantial and VOL. 36, NO. 3, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

457

FIGURE 10. First-order decay coefficients λ for manganese (total, filterable, and manganous ion) with their associated 95% confidence intervals estimated using inverse modeling under three scenarios: (i) estimating λ with all of the observations (A), (ii) estimating λ with only the nighttime observations (N), and (iii) estimating λ with only the daytime observations (D).

FIGURE 11. Mnt and Mnf residuals (Mnobserved - Mnsimulated with decay) over the course of the experiment. Positive residuals indicate model over predicts decay while negative residuals indicate model underpredicts decay. enhanced by dissolved organic matter (Figures 5 and 7) (20). Because of the low H2O2 concentrations in Lake Fork, the production of H2O2 from DOC photolysis is unlikely to significantly influence Mn concentrations in the stream. The estimated values for the first-order decay coefficient (λ) decay values determined in the reactive solute transport simulations using the field data attempt to quantify the combined influences of oxidative, dark reductive, and photoreductive processes within the stream. The positive λ values show that oxidative processes are dominant within the stream, resulting in a net removal of Mn from the water column. From the laboratory experiments, it is apparent that several different redox reactions can affect estimated values of λ. At night, surface-catalyzed and possibly biotic oxidation of Mn contributes to the overall removal of manganese while dark reductive dissolution of MnOx releases manganese to the system. Upon exposure to sunlight, the estimated λ value should also be affected by photoreduction. However, the estimated λ values indicate that the net rate of removal of Mn increases during the day, suggesting that photosynthetically enhanced oxidation may have a greater effect on the Mn concentrations in the stream reach than photoreduction. Although the streambed does not support an abundant algal community, some periphyton on the surface of the rocks may be responsible for the enhanced oxidation during the day. Even though oxidation is the dominant redox process, comparisons of Mnt and Mnf from sites 1 and 2 highlight the importance of reductive processes in the speciation of manganese within the water column (Figure 3). On the second morning of the experiment, Mnf and Mn2+, but not Mnt, 458

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 3, 2002

decrease steeply at site 1, whereas Mnt and Mnf are approximately equal at site 2. While the decrease in Mn2+ and Mnf may be due to processes occurring in the wetland, this observation suggests that rapid reduction of colloidal and particulate Mn(IV) transported in the water column between site 1 and site 2 increases dissolved Mn2+. Light may affect the dynamics of Mn oxidation and reduction in both the wetland and the stream. For example, the dip in Mnf and Mn2+ coincides with a period of low light resulting from a passing storm. However, the effects of light are complicated, and no clear day-night pattern in the ratio of Mnf or Mn2+ to Mnt can be observed. These results highlight the important redox reactions involving manganese in natural stream systems. Photosynthetically enhanced and surface-catalyzed oxidation appears to dominate the transport and cycling of manganese in this stream. The overall loss of manganese, as represented by first-order decay, was fast. Small day-night differences in the rates of decay, as indicated by the time dependence of the residuals in the transport simulation (Figure 11), suggest that the net rate of oxidation may vary over the course of 24 h. While laboratory experiments (Figure 5) indicate that photoreductive dissolution of MnOx on the streambed may alter the net rate of removal of manganese in the stream reach, the overall effect of light was to enhance the removal of manganese during the day. Differences in Mn speciation (Mnt and Mnf) from the beginning to the end of the reach suggest that rapid, possibly photochemically enhanced, reduction may affect the solution phase speciation of Mn. Improved resolution of the speciation of manganese is necessary to further our understanding of the dynamic behavior of Mn in streams. The oxidation of manganese within the stream may also affect the cycling and transport of other metals such as zinc, as shown in the laboratory removal experiment (Figure 4).

Acknowledgments This work would not have been completed without the assistance of many field volunteers, including Lisa Klapper, Michael Gooseff, Arne Bomblies, Natalie Mladenov, Eric August, Chris Jaros, and Dev Niyogi. Specifically, we would like to thank Eric August and Elmira Kujundizic for their dedication and enthusiasm. We would also like to thank Terry Plowman and Howard Taylor for assistance in the anion analyses. This project was funded by EPA Grant 98-NCERQA83.

Supporting Information Available Two figures showing diel sampling. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Brick, C. M.; Moore, J. N. Environ. Sci. Technol. 1996, 30, 19531960. (2) Crerar, D. A.; Cormick, R. K.; Barnes, H. L. In Geology and Geochemistry of Manganese; Varentsov, I. M., Grasselly, Gy., Eds.; Schweizerbart: Stuttgart, 1976. (3) Morgan, J. in Metal Ions in Biological Systems; Sigel, A., Sigel, H., Eds.; Marcel Dekker: New York, 1999; Vol. 37. (4) Sunda, W. G.; Huntsman, S. A.; Harvey, G. R. Nature 1983, 301, 234-236. (5) Sunda, W. G.; Huntsman, S. A. Limnol. Oceanogr. 1990, 35, 325338. (6) Sunda, W. G.; Huntsman, S. A. Mar. Chem. 1994, 46, 133-152. (7) Waite, T. D.; Wrigley, I. C.; Szymczak, R. Environ. Sci. Technol. 1988, 22, 778-785. (8) Bertino, D. J.; Zepp, R. G. Environ. Sci. Technol. 1991, 25, 12671273. (9) Richardson, L. L.; Auilar, C.; Nealson, K. H. Limnol. Oceanogr. 1988, 33, 352-363.

(10) Stumm, W.; Morgan, J. J. Aquatic Chemistry, 3rd ed.; John Wiley & Sons: New York, 1996. (11) Harvey, J. W.; Fuller, C. C. Water Resour. Res. 1998, 34, 623636. (12) Bencala, K.; Walters, R. Water Resour. Res. 1983, 19, 718. (13) Chapin, T. P.; Johnson, K. S.; Coale, K. H. Anal. Chim. Acta 1991, 249, 469-478. (14) Kok, G. L.; Thompson, K.; Lazrus, A. L. Anal. Chim. Acta 1986, 58, 1192-1194. (15) Hrncir, D. C.; McKnight, D. Environ. Sci. Technol. 1998, 32, 2137-2141. (16) Runkel, R. L. Water Resour. Invest. (U.S. Geol. Surv.) 1998, No. 98-4018; http://co.water.usgs.gov/otis/.

(17) Poeter, E.; Hill, M. Water Resour. Invest. (U.S. Geol. Surv.) 1998, No. 98-4080. (18) Wagner, B. J.; Harvey, J. W. Water Resour. Res. 1997, 33, 1731. (19) August, E. M.S. Thesis. University of Colorado at Boulder, 2001. (20) Stone, A. T.; Morgan, J. J. Environ. Sci. Technol. 1984, 18, 617624.

Received for review May 8, 2001. Revised manuscript received September 18, 2001. Accepted October 2, 2001. ES010951S

VOL. 36, NO. 3, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

459