Discrimination between regional and point-source atmospheric Hg pollution using sediment records from drainage lakes, Maine, USA

S. A. Norton* (*University of Maine)(Norton@Maine.Edu), C. T. Hess*, J. A. Cangelosi*, M. J. Norris*, E. R. Perry*, J. S. Kahl*, and D. L. Courtemanch (Dept. of Envir. Protection, Maine)

ABSTRACT

The only operating chlor-alkali plant in the northeastern United States began operation in Maine in 1967. In 1990, self-reported emissions were approximately 460 kg/year and have presumably declined since that time. We are assessing ²¹⁰Pb-dated sediment records in 8 lakes and 2 bogs within 20 km of the point source to determine whether or not the local deposition field is significantly enhanced above the regional background during the period of high local emissions of Hg. Pre-industrial accumulation rates in three lakes range from 0.7 to 2 ng Hg/cm²/yr. Maximum total Hg accumulation rates are 8 to 18 ng/cm²/yr. The portion attributed to anthropogenic atmospheric (Hg_A) deposition started increasing about 1900. Maximum values of Hg_A are 4 to 8 ng/cm²/yr. Fluxes of sediment and Hg are lower in the shallowest of the three lakes. Hg_A increases to the present in two lakes but reached quasi-steady state 50 years ago in the third. Our results show that absolute values of and variations in Hg accumulation rates are influenced by (drainage basin)/(lake basin) area ratios, lake bathymetry (focusing), percentage of wetlands, and land use. We are unable to definitively assign any portion of the Hg_A flux to a specific point source.

INTRODUCTION

Mercury pollution in surface waters and sediments and related high concentrations of Hg in fish have been identified problems for the northern tier of states from Minnesota to Maine for a decade (Engstrom and Swain, 1997; Mower et al., 1997; Norton et al., 1997). Atmospheric deposition of Hg in Maine increased from background about 1900, reached a maximum between about 1970 and 1980, and has subsequently declined. These trends are recorded in dated lake sediment and bog cores (Norton et al., 1997). Recent surveys of concentrations of Hg in surface sediments and fish in Maine lakes (Mower et al., 1997) indicate that lakes and fish near several point source emitters of Hg may be impacted by local deposition from these atmospheric emissions of Hg. We hypothesized that elevated atmospheric deposition of Hg occurred in a pattern related to these sources. We collected sediment cores for ²¹⁰Pb dating and chemical analysis from eight lakes and two bogs spatially around and at different distances from these point sources. The modeled atmospheric deposition signal from the point source was sufficiently strong such that comparison between the Hg flux to the coring site and the history of atmospheric deposition could be compared. We report here on sediment records from three headwater lakes with relatively small and undeveloped forested catchments, except for shore-side summer cottages. They are located five to eight km south of the point source.

Our research has been supported by the Maine Department of Environmental Protection and a grant from the U.S. Geological Survey (Grant 1434-HQ-96-GR-02674) to the University of Maine.

METHODS

Lake description: The physical characteristics of the lakes are given below.

Coring and pre-analysis processing: Cores were obtained in 1999 with a 10 cm diameter stationary piston corer suspended from a floating platform or through ice (Williams Pond). Sediment was sectioned in the field into

	Maximum depth, m	Lake area, ha	Catchment (excluding lake), ha	S^210Pbu, Bq/cm^2	SHgA ng/cm^2
Jacob Buck Pond	16	86	745	0.88	236
Swetts Pond	8	57	992	0.37	146
Williams Pond	16	51	509	1.23	505

WhirlPakÔ plastic bags. We sectioned at 0.5 cm intervals from 0 to 10 cm, 1.0 cm intervals from 10 to 30 cm, and in 2 cm intervals below 30 cm. Sediment was stored near 0^oC and dark until processed. Sediment was dried at 35^oC to constant weight for determination of % H₂O. This fraction was used for determination of the Hg concentration. A dried aliquot was dried at 110^oC and ashed at 550^oC for four hours to determine loss on ignition (LOI).

Dating of sediment: ²¹⁰Pb gamma-ray activity was determined using the 46.52 keV emission. We used a Canberra germanium well detector (1 by 4 cm) with 22.5% efficiency for ⁶⁰Co. Data were processed using GammaTrac software (Oxford Instruments). Dried sediment in capped 1 by 4 cm polyethylene vial was counted for 43,200 to 259,200 seconds. Data were analyzed by Compton continuum subtraction of the peaks. Calibration of the detector was done using U.S. EPA National Exposure Research Laboratory aqueous standards (²¹⁰Pb, ²⁴¹Am, ²²⁶Ra, ¹³⁷Cs, and ⁶⁰Co) in the same geometry as the sediment samples. The ²¹⁰Pb_u (unsupported) activity was estimated by subtracting the constant background ²¹⁰Pb activity, deep in the core, from total ²¹⁰Pb_t. The (integrated) S²¹⁰Pb_u (Bq²¹⁰Pb_u/cm²/core), necessary for dating, also assesses sediment focusing. Calculation of age of interval mid-points was based on the CRS model of Appleby and Oldfield (1983). We used linear interpolation between interval mid-point ages to determine ages of interval boundaries and thus the years represented by an interval.

Determination of Hg: We digested 0.25 to 0.50 g aliquots of each interval using a microwave-assisted acid digestion technique. Mercury was brought into solution with HNO₃ (Swetts Pond) or HNO₃ and HCL, with closed vessel heating by microwave, followed by oxidation with permanganate/persulfate solutions, and then reduced with hydroxylamine hydrochloride. The digestate was filtered and diluted to 100 ml with deionized water. Concentrations of Hg were determined by cold vapor atomic absorption. Ten ml of the digestate were combined with a solution of stannous chloride and with blank acid matrix to convert the mercury in solution to vapor. Argon gas carries the mercury to the detector. The signal produced by the UV lamp is converted to a signal peak, which was measured against a 5- point calibration curve. Based on the original weight of the sample a ng Hg/gram sediment value is obtained. Precision and accuracy are checked with standard reference material (SRM) analyses, blanks, duplicates, replicate analyses, and standard checks during a run. Two SRMs are digested per core (minimum), a blank is prepared every 14 samples, and 1 of every 9 samples is duplicated. Check standards are run once for every 10 analyses. Concentrations are +/-10%.

Accumulation rates for Hg: The net accumulation rate for total Hg (ng Hg/cm²/yr) equals:

Hg_T = [(mass of sediment/interval/cm²)(concentration of Hg in interval)]/(years/interval) (1)

This total flux is composed of three components. (1) The natural background flux of Hg, Hg_B. Commonly, variations in % organic matter cause variations in Hg concentration. However, LOI does not vary appreciably in background portions of the three cores and therefore we have made no correction for this effect. We use pre-1880 sediment for this estimate. (2) Variations in the gross sedimentation rate caused by human activities in the watershed cause variations in the flux of Hg, Hg_V. This variation is estimated from pre-1880 sediment using the ratio (sedimentation rate for any sediment interval [g/m²/yr])/(pre-1880 sedimentation rate [g/m²/yr]). (3) Variations in deposition of anthropogenic Hg directly to the lake and from leaching of Hg from the watershed to the lake, Hg_A. Thus:

Total Hg (ng Hg/cm²/yr) = Hg_B + Hg_V + Hg_A (2)

RESULTS AND DISCUSSION

We illustrate our approach with Williams Pond data. In Williams Pond, the concentration of Hg increases markedly just prior to 1900 (Figure 1) to a broad peak between 1920-1950 and then declines to a value about 50% higher than the pre-industrial value. The flux of total Hg increased slightly through the 19^th century, possibly due to minor erosion caused by forestry practices, and then climbed steeply to a peak in the 1940s, and to a higher peak between 1980-1999 (Figure 1). The uppermost interval may be a transient value. Concurrent with this increase there has been a sharp increase in the sediment flux (Figure 1) which we assume carries with it a Hg burden (concentration) (Hg_V) at least equal to the pre-pollution sediment. The residual = (Hg_T – (Hg_B + Hg_V)) is termed the anthropogenic component (Hg_A) of atmospheric deposition (Figure 1). The atmospheric pollution signal starts in sediment dated at about 1890. The maximum Hg_A, 8 ng/cm²/yr, is less than half of the Hg_T but 8 times the background flux, Hg_B. The Hg_A flux is approximately twice that observed in ombrotrophic Caribou Bog (20 km north of the point source) (Benoit, 1999) and Big Heath (60 km southeast) (Norton et al., 1997). The Hg_A fluxes for the three lakes, located within 4 km of each other, are different (Figure 2). The initial increase in Hg_A occurs at approximately 1900 for all three lakes. However, Hg_A increases to the present at two lakes whereas Williams peaks by about 1945.

At least three processes control the sediment signal. First, the different response in the sediment record may be a consequence of the time to steady state delivery of increased Hg deposition from the catchment. For example, with a 5% retention of Hg by the watershed, a typical value in the literature, it requires about 40-50 years to reach a steady state export of Hg to the lake. Greater retention of Hg by the watershed results in a longer period to steady state export of Hg to the lake. Williams has abundant fen wetlands; the other two lakes do not. Watershed slopes are steeper and have thinner till for Jacob Buck and Williams than for Swetts. Second, the magnitude of the flux from the catchment to the lake is also controlled by the watershed/lake area ratio (Table 1). The ratio is Jacob Buck<Williams<Swetts. Third, focusing of sediment controls sedimentation fluxes. Jacob Buck and Williams are twice as deep as Swetts, and their Hg_A values are approximately twice that of Williams. Similarly, the integrated Hg_A and unsupported ²¹⁰Pb_u (atmospheric) in the two deep lakes are greater than that of Williams (Table 1).

The chronology and magnitude of changes for Hg_A for these three lakes are not substantially different from what we have observed for other lakes in Maine (Evans, 1998; Smith, 1998). For those lakes, Hg_B ranges from 0.1 to 6 ng/cm²/yr, and the peak Hg_A ranges from 1 to 20 ng/cm²/yr. Consequently, we can not unequivocally attribute the accumulation of Hg to a point source rather than being representative of the regional picture. The natural differences among lake sediment records appear to be sufficiently large as to obscure the influence of the point source, if that source is significant.

The atmospheric deposition of Hg has declined at least 50% since the 1970s in Maine, based on records from ombrotrophic peat cores (Norton et al. 1997; Benoit, 1999). Further, G. C. Evans et al. (this meeting) have inferred a substantial decline in atmospheric deposition of Hg based on high elevation forest soils in Maine. In spite of this decline, the three lakes do not unequivocally point to a decline. Because of net retention of Hg since the onset of significant air pollution, the leaching of Hg from the increased soil reservoir is elevated.

REFERENCES

Appleby, P. G., Oldfield, F. (1983), Hydrobiologia. 103. 29-35.

Benoit, A. (1999), Unpub. Honors Thesis, University of Maine, 28 p. plus Appendices.

Engstrom, D. R., Swain, E. B. (1997), Envir. Sci. Technol. 31. 960-967.

Evans, G. C. (1998), Unpub. M.Sci. Thesis. University of Maine. 106 p.

Mower, B., DiFranco, J., Bacon, L., Courtemanch, D., Schmidt, V., Hopeck, J. (1997), Fish tissue contamination in Maine lakes. Dept. Envir. Prot., State of Maine. 63p.

Norton, S.A. Evans, G. C., Kahl, J. S. (1997), Water, Air, Soil Poll. 100. 271-286.

Smith, A. M. (1998), Unpub. M.Sc. Thesis, University of Maine. 69p.