Environ. Sci. Technol. 2004, 38, 4921-4927
Redox Chemistry in Minnesota Streams during Episodes of Increased Methylmercury Discharge S T E V E N J . B A L O G H , * ,† YABING H. NOLLET,† AND EDWARD B. SWAIN‡ Metropolitan Council Environmental Services, 2400 Childs Road, St. Paul, Minnesota 55106-6724, and Minnesota Pollution Control Agency, 520 Lafayette Road, St. Paul, Minnesota 55155-4194
Mercury (Hg) and methylmercury (MeHg) are flushed from watersheds during hydrological events, contaminating downstream surface waters and resident fish populations. We monitored total mercury (THg), MeHg, and ancillary water chemistry parameters in two streams (Cedar Creek and Trott Brook) in east-central Minnesota on a weekly or semiweekly basis from April through October 2003. Heavy precipitation in late June resulted in discrete episodes of high concentrations (>1.2 ng/L) of MeHg in both streams in early July. The MeHg/THg ratio increased from 0.15 to 0.36 in Cedar Creek and from 0.13 to 0.46 in Trott Brook during the event. The high MeHg concentrations were accompanied by low dissolved oxygen concentrations and increased concentrations of dissolved organic carbon, Mn, Fe, and orthophosphate. A prolonged absence of precipitation during August and early September brought stream levels back to baseflow values, and MeHg concentrations decreased to less than 0.1 ng/L. These results suggest that warmweather, high-discharge events are the primary route of export of MeHg from these watersheds, and baseflow contributes much less MeHg to downstream waters. The redox water chemistry during the events sampled here suggests that MeHg in these streams is discharged from wetland areas where anoxic/anaerobic conditions prevail.
Introduction Mercury (Hg) is a highly toxic pollutant found in terrestrial and aquatic systems throughout the world. It contaminates surface waters and their fisheries, and the tainted fish represent a health threat to human and other consumers. Methylmercury (MeHg) bioaccumulates with increasing trophic level in aquatic systems and accounts for 95% of the total Hg (THg) in top-predatory fish (1). Some MeHg enters surface waters from atmospheric deposition, but MeHg production in upstream areas within the watershed and in-situ production are often more important (2). Wetlands have been identified as being particularly significant sources of MeHg to surface waters (3-9). The saturated, anoxic, organic soils in wetlands are favorable environments for the production of MeHg by sulfate-reducing bacteria. Areas of discharging groundwater may provide both anaerobic con* Corresponding author telephone: (651)602-8367; fax: (651)6028215; e-mail:
[email protected]. † Metropolitan Council Environmental Services. ‡ Minnesota Pollution Control Agency. 10.1021/es049696c CCC: $27.50 Published on Web 08/20/2004
2004 American Chemical Society
ditions and a steady stream of nutrients to promote the bacterially mediated production of MeHg (7, 10). The broader chemical composition of wetland waters is also largely determined by the prevailing redox conditions. In flooded wetland soils, microorganisms quickly exhaust available oxygen supplies and then utilize, sequentially, nitrate, Mn(IV), Fe(III), and sulfate as alternative electron acceptors (11). Manganese and iron oxyhydroxide particles dissolve under the reducing conditions, releasing adsorbed or coprecipitated orthophosphate and organic carbon. Runoff from anoxic wetland soils would be expected to reflect these reducing conditions, with relatively lower concentrations of dissolved oxygen (DO), nitrate, and possibly, sulfate (depending on the redox potential) and higher concentrations of soluble reduced Fe and Mn, soluble orthophosphate, and dissolved organic carbon (DOC). Peat-based wetlands are major sources of DOC to surface waters (12). This DOC has its origin primarily in the slow microbial degradation of waterlogged plant tissue and includes macromolecular fulvic and humic acids as well as smaller organic acids. These materials are responsible for the acidic nature of peatland waters (13) and can form strong bonds with both Hg(II) (14) and MeHg (15). Numerous studies have found positive correlations between THg or MeHg, or both, and DOC or TOC concentrations in streams draining catchments containing peatlands (5, 7, 16-22). Lee and Iverfeldt (17) attributed the covariation of MeHg and THg with water color (a proxy for DOC) to the flushing of these materials from organic soils during runoff events. In the present study, we characterized the concentrations of THg, MeHg, DOC, DO, soluble reactive phosphorus (SRP), soluble Fe and Mn, and other parameters in two streams in east-central Minnesota from April through October 2003. We sought to characterize the relationships between THg and MeHg concentrations and the redox conditions in the streams. Our hypothesis was that peatlands are the major source of MeHg to these streams and that elevated MeHg concentrations would be accompanied by elevated DOC, decreased DO, and other indicators of reduced water chemistry.
Experimental Section Environmental Setting. Cedar Creek and Trott Brook are fourth-order tributaries to the Rum River lying within the Anoka Sand Plain, an extensive (4400 km2) landform of sandy lacustrine and fluvial glacial deposits located in east-central Minnesota, north and northwest of the Minneapolis-St. Paul (Twin Cities) metropolitan area (Figure 1). Red-brown sandy till from the Superior lobe of the Laurentide Ice Sheet and gray, calcareous silty till from the Grantsburg (Des Moines) sublobe lie beneath the outwash and lacustrine sand deposits (23). Surficial deposits in the Cedar Creek watershed are mostly fine lacustrine sand and peat (Table S1, Supporting Information). A large area of mixed glacial till rises above the outwash sand plain and peat deposits in the Trott Brook basin. The topography of the area is mostly flat but ice-block depressions and “till-islands” provide some relief (23). Organic peat and loamy hydric soils occupy the numerous depressional areas, many of which lie at or below the local water table. Wetlands make up 9% and 10%, respectively, of the Trott Brook and Cedar Creek watersheds (Table S2, Supporting Information). Cultivated lands and pasture comprise approximately 40-50% of the land use in each of the two watersheds. The Cedar Creek basin has a greater land area in forest and is slightly less developed than the VOL. 38, NO. 19, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4921
FIGURE 1. Maps showing the locations of Trott Brook and Cedar Creek in east-central Minnesota. The main map shows the location of the Rum River watershed (cross-hatched) within the headwater Mississippi River watershed. Box A (upper right) shows the location of the Trott Brook and Cedar Creek watersheds (cross-hatched) within the Rum River watershed. Box B (lower right) shows the location of the sampling sites (4). Trott Brook watershed. The surficial drainage areas of the two streams are essentially the same: 194 km2 for Cedar Creek and 193 km2 for Trott Brook. Climate in the study area is subhumid continental with dry, cold winters and warm, moist summers. The mean annual precipitation at Cedar, MN (Figure 1), is 78 cm, and the mean total precipitation for the period of April-October is 63 cm (1971-2000; 24). In 2003, the total precipitation falling between April and October was near normal (60 cm; 24). Precipitation amounts in April, May, and June were higher than normal levels by 2, 6, and 7 cm, respectively, and amounts in July, August, September, and October were lower by 2, 11, 2, and 4 cm, respectively. Sampling and Analysis. Sampling was carried out near the mouths of both streams on a weekly or semi-weekly basis between April 23 and October 28, 2003. The sampling location on Trott Brook was approximately 100 m downstream of the confluence of Ford Brook with Trott Brook and 1.4 km upstream of the confluence of Trott Brook with the Rum River (Figure 1). The sampling location on Cedar Creek was 3 km upstream of that stream’s confluence with the Rum River. All samples were collected between 7:00 a.m. and 8:00 a.m. Samples for THg and MeHg analyses were collected in 500-mL acid-cleaned Teflon bottles using strict “clean hands/ dirty hands” protocols (25, 26). Samples were returned to the analytical laboratory within 2 h of collection. Subsamples (250 mL) were filtered immediately upon return using acidcleaned disposable filtration units (0.2-µm cellulose nitrate membrane). 4922
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004
Subsamples (50 mL) of both the filtered and unfiltered samples were placed in 60-mL Teflon vials and distilled to remove matrix interferences (26, 27). Methylmercury was determined in the distilled samples using aqueous phase ethylation/gas chromatography with Tenax preconcentration and atomic fluorescence detection (28, 29). All samples were analyzed in duplicate; the relative standard error (standard error/mean; RSE) for these duplicate determinations was typically between 5 and 15%. Analytical accuracy was checked by daily analysis of DORM-2 Dogfish Muscle or DOLT-3 Dogfish Liver certified reference materials or both. The mean concentration determined for DORM-2 was 4.77 mg/kg (n ) 27; CV ) 6%), and that for DOLT-3 was 1.54 mg/kg (n ) 24; CV ) 7%). The certified value for DORM-2 is 4.47 ( 0.32 mg/kg, and the informational (uncertified) value for DOLT-3 is 1.7 mg/kg. Spiked samples gave a mean recovery of 101% (n ) 33; CV ) 8%). Concentrations of MeHg in filter blanks were less than our method detection limit of 0.03 ng/L. Samples for THg analysis were digested using an acidpermanganate digestion (30), directly in the collection bottle. Determination of THg was by cold vapor atomic fluorescence with single gold trap amalgamation (31). All samples were analyzed in duplicate; the RSE was typically less than 5%. Daily analysis of NIST 1641d Mercury in Water certified reference material gave a mean value of 1.58 mg/kg (n ) 24; CV ) 1.6%). The certified value is 1.590 ( 0.018 mg/kg. The mean matrix spike recovery was 100% (n ) 18; CV ) 5%). Concentrations of THg in filter blanks were less than our
FIGURE 2. Concentrations of THg and MeHg in Trott Brook (a and b, respectively) and Cedar Creek (c and d, respectively). Symbols distinguish unfiltered ([) and filtered (]) samples. The solid lines show “1/alkalinity”, a proxy for discharge in these streams. method detection limit (0.05 ng/L, based on three times the standard deviation of the bubbler blank values). Samples for DOC analysis were collected in 250-mL amber glass bottles with Teflon-lined lids. Prior to use, these bottles were heated to 450 °C to eliminate carbon contamination. Samples for all other analytes were collected in acid-cleaned 2-L brown polypropylene bottles. Aliquots of these samples were filtered in the lab within 3 h of collection. We assumed that the concentrations of redox-sensitive elements did not change during the time between sample collection and filtration. Total suspended solids (TSS), sulfate, nitrate, iron, manganese, and alkalinity were determined in unfiltered samples. Orthophosphate (SRP), iron, manganese, and potassium were determined in filtered samples. Iron, manganese, and potassium were determined by ICP-MS. All other analyses were by standard methods (32). Measurements of DO concentration and temperature were made in the field using a handheld meter with a galvanic oxygen sensor. A single-point calibration of the oxygen sensor was made immediately prior to each field measurement. A glass electrode was used to measure pH in the water samples brought back to the lab. Stream stage height data for Trott and Ford Brooks were recorded using capacitance-based water level monitors. The monitoring location on Ford Brook was 0.46 km upstream of its confluence with Trott Brook. The location of the Trott Brook stage monitor was 1.8 km upstream of that stream’s confluence with Ford Brook. Analysis of the combined stage data showed a strong inverse correlation between stage and alkalinity (r 2 ) 0.84; n ) 33). We therefore assume that discharge in Trott Brook varied inversely with alkalinity. No stage or discharge data were collected on Cedar Creek. Instead, it was assumed here as well that discharge varied inversely with alkalinity. These relationships are not quantitative; however, we feel they give an adequate representation of the relative discharge levels in these streams over the course of our study.
Results and Discussion A particularly heavy precipitation event on June 25-26 brought over 12 cm of rain to the study area, and streamflow in both Cedar Creek and Trott Brook increased sharply (Figure 2). Concentrations of unfiltered THg and MeHg were higher on June 27 as streamflow peaked at both sites (Figure 2; Tables S3 and S4 in Supporting Information). Concentrations of THg then decreased on July 1, before increasing again on July 4. This second peak in THg concentrations corresponded
with a peak in MeHg concentrations, which increased continuously through July 7, before falling back. Filtered concentrations of both MeHg and THg were unusually high during this latter peak as the additional THg entering the streams was mostly “dissolved” MeHg (i.e., MeHg that passed a 0.2-µm filter). As a result, both THg (>65%) and MeHg (>82%) were predominantly in the filtered phase in both streams at this time. The high MeHg concentrations observed in both streams during this July runoff event indicate that runoff flow paths at that time accessed sites of MeHg production/storage. Higher temperatures and generally moist soil conditions during the 2 weeks prior to the event may have facilitated the increased net production of MeHg that was then exported from the watershed in early July. An extended period of higher maximum daily ambient air temperatures (>24 °C) began on June 13 and continued through June 25. Streamwater temperatures were less than 16 °C prior to June 10 and then rose to above 20 °C on June 17 and June 24. Higher temperatures are known to favor Hg methylation in lakes (33, 34), and higher methylation rates were observed during the summer months in a stream draining a headwater wetland in Ontario (3). Lee et al. (21, 22) suggested that the higher temperatures and wetter conditions they observed during July and August might have promoted the net production of MeHg and its discharge to streams draining wetlands in northern Sweden. It is clear, however, that higher temperatures alone were not sufficient to lead to higher MeHg concentrations in Trott Brook and Cedar Creek. Measured streamwater temperatures remained above 18 °C from July 1 through August 26, and maximum ambient air temperatures exceeded 22 °C for all but one day between June 29 and August 29, yet MeHg concentrations in both streams were much lower after the peak of early July (Figure 2b,d). The lack of precipitation obviously precluded any major flushing of MeHg produced during this time, and it is impossible for us to say to what extent MeHg production continued under these warm but drier conditions. Concentrations of MeHg did remain relatively high (>0.5 ng/L) despite increased streamflow on July 17 after additional rain (3-6 cm) fell in the area, suggesting that hydrologically accessible stores of MeHg remained or were replenished after the early July event. There was, however, no significant increase in MeHg concentrations in either stream after the minor runoff events in September and October (see below). It may be that these later events were not sufficiently large to remove MeHg from source areas VOL. 38, NO. 19, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4923
FIGURE 3. Concentrations of unfiltered MeHg, DO, SRP, filtered Mn, and filtered Fe in Trott Brook (a-e, respectively) and Cedar Creek (f-j, respectively). Discharge (relative units) is represented by the dashed lines in panels a and f. or that the source areas had been depleted and not replenished by this time. The major MeHg export event in early July was accompanied by changes in other water chemistry parameters that give some indication of the source areas of the MeHg found in Trott Brook and Cedar Creek. The peak concentration of unfiltered MeHg in Trott Brook on July 7 (Figure 3a) was associated with a minimum in the DO concentration (Figure 3b) and maximums in SRP (Figure 3c) and filtered Mn concentrations (Figure 3d). A peak in the concentration of filtered Fe (Figure 3e) was recorded a few days later, on July 10. The concentration of DOC in Trott Brook also peaked on July 7 (Figure 4b), and streamwater pH (Figure 4c), sulfate (Figure 4d), and nitrate (data not shown) concentrations at that time were the lowest we recorded. The sulfate concentration remained at or below 1 mg/L (our detection limit) for almost the entire duration of the event, increasing to 5 mg/L only at the end, on August 4. Concentrations of filtered K were also elevated during this event (Figure 4e). The event lasted from June 27 until August 4, when the MeHg concentration returned to a value (0.20 ng/L) similar to the pre-event value (0.24 ng/L on June 24). During the event, filtered MeHg concentrations showed a strong negative correlation with DO (r 2 ) 0.92; Figure 5b) and strong positive correlations with DOC (r 2 ) 0.87; Figure 5a), SRP (r 2 ) 0.71; Figure 5c), and filtered K (r 2 ) 0.80; Figure 5d). The increase 4924
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004
in the concentration of filtered Fe early in the event lagged the increase in MeHg concentration (Figure 5e). Later in the event, MeHg and filtered Fe concentrations returned to preevent levels in a relatively linear manner. Filtered Mn concentrations also showed this same type of lag as compared to the MeHg concentrations (data not shown). These results indicate that the high MeHg concentrations observed in Trott Brook in early July were associated with streamwater chemistry exhibiting a relatively reduced character (i.e., low DO, sulfate, and nitrate concentrations and elevated dissolved orthophosphate, Mn, and Fe concentrations). Concentrations of DOC were elevated during the event and strongly correlated with MeHg concentrations. The MeHg was almost entirely in the filtered phase, apparently associated with DOC materials. Wetlands are major source areas of DOC, suggesting that, during the event, both DOC and MeHg in Trott Brook originated in wetland areas of the watershed. The reduced nature of the streamwater at this time is consistent with this conclusion. In addition, the biodegradation of organic material that is the source of DOC in peatlands also releases K (35), which may account for our observations of increased filtered K concentrations during the event. The lowered pH during the event is also associated with the increased DOC concentrations (13). Other studies have found that the MeHg/THg concentration ratio in waters draining wetlands is elevated as compared
FIGURE 4. Concentrations of unfiltered MeHg, DOC, pH, sulfate, and filtered K in Trott Brook (a-e, respectively) and Cedar Creek (f-j, respectively). Discharge (relative units) is represented by the dashed lines in panels a and f. to other catchment types (3, 6, 36). The MeHg/THg concentration ratio is generally higher in peat pore waters as compared to other surface waters, and elevated MeHg/THg concentration ratios in surface waters may indicate contributions from peatland runoff. In Trott Brook, during the July episode of high MeHg concentrations, the MeHg/THg ratio (unfiltered concentrations) increased to 0.44 on July 4 and 0.46 on July 7, well above the values recorded (range: 0.08-0.17) before or after the event. These observations are thus consistent with the hypothesis that peatlands were the primary source of MeHg in Trott Brook at this time. The results for Cedar Creek were similar to those for Trott Brook. The July peak in MeHg concentrations in Cedar Creek was also accompanied by substantial variations in the concentrations of other water chemistry parameters. The concentration of DO decreased sharply on June 27 as streamflow peaked and then was a minimum (1.55 mg/L) on July 1 (Figure 3g). This preceded the maximum unfiltered MeHg concentration (1.25 ng/L) on July 7 (Figure 3f). The concentration of filtered Mn (Figure 3i) peaked on July 7 as well, whereas the SRP (Figure 3h) and filtered Fe (Figure 3j) concentrations lagged, both peaking on July 10. The DOC concentration in Cedar Creek increased sharply at the onset of higher streamflow, from 9.4 mg/L on June 24 to 18.1 mg/L on June 27 (Figure 4g). It increased further to peak at 21.4 mg/L on July 4 and remained high (20.4 mg/L) on July 7. A
sharp decrease in streamwater pH (Figure 4h) accompanied the increase in DOC concentration on July 27. The sulfate concentration decreased to less than 1 mg/L on July 1, increased slightly to 2 and 3 mg/L on July 4 and July 7, respectively, and then increased to 7 mg/L by July 10 where it stabilized for the remainder of the event (Figure 4i). Only a minor increase in filtered K concentrations was observed on July 27, with a peak value (1.69 mg/L) on July 4 (Figure 4j). Nitrate concentrations in Cedar Creek were lowest on July 1 (0.06 mg/L) and 4 (0.07 mg/L). The July event in Cedar Creek lasted slightly longer than that observed in Trott Brook, until August 12 when the unfiltered MeHg concentration again equaled the pre-event value (0.15 ng/L). During this event, filtered concentrations of MeHg showed a strong positive correlation with DOC concentrations (Figure 5f), similar to our observation in Trott Brook, indicating that both the MeHg and the DOC may have originated in wetland areas. The decrease in DO concentration in Cedar Creek during the July event preceded the increase in MeHg concentration (Figure 5g). Nonsynchronicity with the MeHg peak was also observed for SRP (Figure 5h) and filtered Fe (Figure 5j), both of which peaked on July 10, after the MeHg peak on July 7. The concentration of filtered K in Cedar Creek peaked on July 4, prior to the MeHg peak (Figure 5i). Concentrations of filtered Mn peaked on July 7 along with the MeHg VOL. 38, NO. 19, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4925
FIGURE 5. Bivariate plots of unfiltered MeHg concentrations vs DOC, DO, SRP, filtered K, and filtered Fe for Trott Brook (a-e, respectively) and Cedar Creek (f-j, respectively). Arrows indicate the temporal progression of the samples. concentration (data not shown). Some of the discrepancy in the timing of the various peak concentrations is due to our noncontinuous sampling, which shows only snapshots of the actual conditions. It is difficult to explain, based on these data alone, the dynamic variations in the various water chemistry components. The mixing of waters originating in various source areas and traveling via various flow paths is largely responsible for the ultimate character of the streamwater we sampled. The later appearance of the filtered Fe peak relative to the DOC peak may indicate, for instance, that water from deeper, mineral soil flow paths is mixing with surficial flow through organic horizons during the later stages of the event. Other studies have concluded that Fe and DOC in runoff from peatlands originate in mineral soils and organic deposits, respectively (37, 38). In addition, reactive redox-sensitive species may undergo redox transformations during transport from their source areas. A more detailed examination of the water chemistry in wetland source areas and the changes in chemical speciation along flow paths is needed to help clarify the observed streamwater chemistry patterns. While it does appear that the streamwater chemistry and hydrological flow paths in Cedar Creek at this time were more complicated than those in Trott Brook, it also seems clear that higher MeHg concentrations in Cedar Creek were associated with reduced conditions in the streamwater, similar to Trott Brook. The unfiltered MeHg/THg concentration ratio in Cedar Creek was elevated during the July peak in MeHg concentrations. It reached 0.34 on July 4 and 0.36 on July 7, whereas the range before and after the event was 0.05-0.15. This also points to peatland areas as the source of MeHg in Cedar Creek at this time. Concentrations of MeHg in both streams were much lower and varied little both before and after the July runoff event, but THg concentrations were sometimes elevated during the runoff events that occurred. Unfiltered THg concentrations in Trott Brook ranged from 1.41 to 2.60 ng/L between April 23 and June 24 (Figure 2a), and unfiltered MeHg concentrations ranged from 0.13 to 0.24 ng/L (Figure 2b). In Cedar Creek (Figure 2c), concentrations of unfiltered THg were higher on May 6 as runoff brought higher TSS loadings to the stream (data not shown). Filtered THg concentrations peaked 4926
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004
a week later on May 13. Concentrations of MeHg in Cedar Creek (Figure 2d) remained low (
