Critical Review
Modeling the Past Atmospheric Deposition of Mercury Using Natural Archives H A R A L D B I E S T E R , * ,† R I C H A R D B I N D L E R , ‡ ANTONIO MARTINEZ-CORTIZAS,§ AND DANIEL R. ENGSTROM| Institute of Environmental Geochemistry, University of Heidelberg, INF 236, 69120 Heidelberg, Germany, Department of Ecology and Environmental Sciences, Umeå University, 907 42 Umeå, Sweden, Department of Soil Science and Agricultural Chemistry, Faculty of Biology, University of Santiago, E-15782 Santiago de Compostela, Spain, and St. Croix Watershed Research Station, Science Museum of Minnesota, St. Croix, Minnesota 55047
Historical records of mercury (Hg) accumulation in lake sediments and peat bogs are often used to estimate human impacts on the biogeochemical cycling of mercury. On the basis of studies of lake sediments, modern atmospheric mercury deposition rates are estimated to have increased by a factor of 3-5 compared to background values: i.e., from about 3-3.5 µg Hg m-2 yr-1 to 10-20 µg Hg m-2 yr-1. However, recent studies of the historical mercury record in peat bogs suggest significantly higher increases (9-400 fold, median 40×), i.e., from about 0.6-1.7 µg Hg m-2 yr -1 to 8-184 µg Hg m-2 yr -1. We compared published data of background and modern mercury accumulation rates derived from globally distributed lake sediments and peat bogs and discuss reasons for the differences observed in absolute values and in the relative increase in the industrial age. Direct measurements of modern wet mercury deposition rates in remote areas are presently about 1-4 µg m-2 yr -1, but were possibly as high as 20 µg Hg m-2 yr -1 during the 1980s. These values are closer to the estimates of past deposition determined from lake sediments, which suggests that modern mercury accumulation rates derived from peat bogs tend to overestimate deposition. We suggest that smearing of 210Pb in the uppermost peat sections contributes to an underestimation of peat ages, which is the most important reason for the overestimation of mercury accumulation rates in many bogs. The lower background mercury accumulation rates in peat as compared to lake sediments we believe is the result of nonquantitative retention and loss of mercury during peat diagenesis. As many processes controlling timeresolved mercury accumulation in mires are still poorly understood, lake sediments appear to be the more reliable archive for estimating historical mercury accumulation rates.
Introduction Many advances have been made in recent years in the understanding of mercury biogeochemistry in the contem* Corresponding author e-mail:
[email protected]. † University of Heidelberg. ‡ Umeå University. § University of Santiago. | St. Croix Watershed Research Station. 10.1021/es0704232 CCC: $37.00 Published on Web 06/16/2007
2007 American Chemical Society
porary environment. For mercury such studies include environmental manipulations (e.g., METAALICUS project, (1)), the effects of the polar sunrise (2, 3), and the influence of dissolved organic carbon and sulfate on methylation processes (4-6). While most studies emphasize present-day contamination and biogeochemistry, another area of research has focused on estimating the natural contribution of mercury prior to human impacts, which is based on the analysis of natural environmental archives such as lake sediments (7-24), peat (25-41), and also glacial ice (42, 43). We exclude ice from further discussion because of its limited geographic distribution and the small number of published records. Data on metal deposition from monitoring programs are temporally very limited with reliable records spanning only the past decade or two at best. Thus, natural archives provide the only link between current and past mercury loading to terrestrial and aquatic environments, and they give us insights into timescales and spatial patterns not available in contemporary monitoring programs. There is broad consensus that natural archives provide a means to reconstruct atmospheric deposition trends at local, regional, and even global scales (23, 44), and that mercury deposition rates are related to, for example, levels of fish contamination (45-47). In addition, estimates of the natural background deposition rate of mercury and certain other elementssas well as subsequent human impactssin lake sediment and peat also provide us with insights on how natural environmental factors have influenced deposition and accumulation. Such environmental factors include climate changes (warmer-colder or wetter-drier) or volcanic activity (29, 43). Both lake-sediment and peat records show an increase in mercury accumulation which parallels increasing industrialization during the past two centuries. Whereas studies of lake sediments suggest an increase in mercury atmospheric deposition rates that are, in recent decades, about 3-5 times above natural background (pre-industrial) rates in the northern hemisphere, studies of peat suggest an increase in deposition of 30 to as much as 500-fold in recent versus pre-industrial times. The disparity in past atmospheric mercury deposition rates between the archives is problematic for understanding anthropogenic perturbation of the global mercury cycle. It affects not only our perception of the magnitude of human impact, but also our understanding of emission sources, atmospheric processes, and rates of exchange between terrestrial, ocean, and atmospheric pools (44, 48, 49). Two studies of mercury accumulation rates in lakes and a peat VOL. 41, NO. 14, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4851
deposit in Greenland serve as an initial example. Greenland is a remote area with no significant local anthropogenic mercury sources, thus the increase in modern atmospheric deposition of mercury in Greenland archives should reflect long-range transport from anthropogenic sources in the northern hemisphere as well as inter-hemispheric mixing. Based on three lakes in western Greenland, Bindler et al. (19) reported background mercury accumulation rates in the range of 1-2 µg m-2 yr-1 and a maximum of modern mercury accumulation of 5-10 µg m-2 yr-1, which corresponds generally to a 3-5 fold increase during the industrial period. Shotyk et al. (37) analyzed a single core from a peat deposit in southern Greenland and reported lower background values of 0.3-0.5 µg m-2 yr-1, but a maximum modern mercury accumulation rate of 164 µg m-2 yr-1, which corresponds to an increase of ∼300-500 times over background values. This striking difference in the two archives could hardly be explained by local differences in mercury deposition, and it raises questions about the reliability of these archives to reflect true atmospheric mercury fluxes and the magnitude by which they have changed. In this paper we review mercury data from a range of peat and lake sediment studies that have reconstructed accumulation rates and subsequently modeled past atmospheric deposition rates of mercury. We discuss some potential reasons for the differences observed between the two archives. In addition to the mercury studies, we include anthropogenic lead for comparative purposes.
Comparison of Lake Sediment and Peat Properties of the Archives. Lake sediments and peat constitute two substantially different media in terms of composition, biogeochemistry, and hydrology. Lake sediments represent relatively closed systems once the sediment has been buried below the active surface layer, which in most lakes constitutes only the unconsolidated, uppermost few centimeters. This active layer, where bioturbation and redox processes can redistribute some sediment or elements, typically comprises sediments accumulated over the past one or a few decades. Hg in lake sediments is likely bound to reduced sulfur groups of the organic matter or precipitated as metacinnabar (HgS). The solubility of organo-Hg-sulfides or of metacinnabar is very low, so that diffusion of Hg within sediments is low. There might be some reflux of formed Hg(0) during dissolution of Fe-oxides or decomposition of organic matter at the sediment-water interface but once recorded in the anaerobic zone diffusion appears to be negligible. There is strong empirical and theoretical evidence for the stability of inorganic mercury in lake sediments including (i) the temporal coherence of mercury increases among multiple cores and lakes (11), (ii) the preservation of distinct peaks in mercury profiles from cores collected many years apart in the same lake (50), (iii) the absence of mercury redistribution in experimental core incubations (51), and (iv) the strong solid-phase partitioning of mercury from porewaters of intact cores (log Kd > 5) (51). Because lake sediments represent an integrated record reflecting both changes in direct atmospheric deposition of mercury and other metals to the lake surface and changes in transport from surrounding catchment soils, it is difficult to model actual rates of depositionsas opposed to relative changes in depositionsof mercury (or other metals) over time. The rate of atmospheric metal deposition can be modeled using lake sediments, but this requires timeconsuming mass-balance studies of multiple lakes (11, 23, 52) or a focusing correction of mercury flux data from multiple lakes based on 210Pb inventories (24, 31), which thus far have only been applied to the time-scale covered by radiometric lead (210Pb) dating. From the perspective of making environmental reconstructions, peat cores from ombrotrophic 4852
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 14, 2007
bogs benefit from the fact that they are supplied with nutrients and metals only via the atmosphere. Consequently, peat potentially offers a less-complicated medium to model longterm changes in atmospheric inputs of mercury (53) and lead to the environment, and many presume that the peat archive corresponds to a record of absolute atmospheric deposition rates. However, peat initially represents a more open system than lake sediment because it is composed of a build-up of organic matter that is supported by water, where the uppermost sectionsthe acrotelmsis exposed to aeration for decades or centuries before permanent incorporation below the water table. In addition, organic matter is continually lost from the peatsfirst rapidly in the acrotelm, where 50% to as much as 80% of the mass can be lost before burial in the catotelmsand thereafter much more slowly in the catotelm, where a further 10% of the original mass can be lost (54). Historic Mercury Accumulation Rates. Most available data of historic and modern mercury accumulation rates (HgAR) are derived from archives located in the northern hemisphere, whereas only a small number of studies are available from the southern hemisphere (30, 31, 55). Figure 1 provides an overview of mercury accumulation rates derived from lake sediments and peat bogs from different locations worldwide. One might argue that data from different sites are not readily comparable because of differences in distances to local or regional pollution sources. However, most sites, and all of those discussed in this review, are located in remote areas far from local mercury sources. Because of mercury’s long atmospheric residence time, these remote sites are thought to represent mercury emissions on a hemispheric, if not global, scale (44). Although we can rule out the effects of local emission sources, there remains the issue of geographic variability in mercury deposition rates (largely a function of differences in precipitation) as well as the amplification (or diminution) of the atmospheric signal introduced by mercury transport and cycling within the individual lake or peatland. However, assuming site-specific factors remain constant over time, the relative change in mercury accumulation from pre-industrial to modern times (mercury flux ratio) should provide a robust and comparable measure of the magnitude of change in atmospheric mercury deposition (56). As the increase of mercury deposition rates in modern times (from ca. 1860 to present) is in many cases of particular interest, we depict pre-industrial and modern mercury accumulation rates separately. The data show that modern mercury accumulation rates derived from bogs are systematically higher than those from lake sediments. Moreover, the maximum mercury accumulation rates in bogs show much greater variation as compared to those in lake sediments. In the past 150 years the median peak HgAR in peat bogs is about 40 µg m-2 yr-1 (8-184 µg m-2 yr-1) whereas those in lake sediments are only ∼15 µg m-2 yr-1 (5-68 µg m-2 yr-1), which is a factor of 2.5 lower (most lake sediment data in Figure 1 are not corrected or normalized for catchment inputs or sediment focusing). Far higher rates of modern HgAR have been reported for lakes in urban and agricultural areas (100-200 µg m-2 yr-1); however, such sites are clearly impacted by local emission sources as well as large mercury inputs from soil erosion (21) and are not comparable to the remote sites discussed here. The highest peat mercury accumulation rates in the industrial period are from higher latitude sites, where they exceed those found in mid-latitude bogs by a factor of ∼3-4. The highest reported modern mercury accumulation rates are from a bog in Denmark (184 µg m-2 yr-1) and a fen in southern Greenland (164 µg m-2 yr -1), which Shotyk et al. (37) explain are the result of longrange mercury transport from mid-latitude sources.
FIGURE 1. Comparison of background and the modern maximum mercury accumulation rates as estimated from lake sediments and peat bogs (only lake sediment values from Alaska, Nova Scotia, and Minnesota have been corrected for the influence of sediment focusing or catchment size). The mean background and modern maximum accumulation rate for lake sediments (6.9 and 24 µg m-2 yr-1, respectively) and peat (1.4 and 59 µg m-2 yr-1, respectively) are marked by a dashed line. For lake sediments accumulation rates corrected for the influence of, e.g., sediment focusing are also indicated by a gray dashed line (3.5 and 12 µg m-2 yr-1, respectively). Data are from refs 11, 14, 15, 17, 19, 22, 23, 27-29, 31-33, 35-37, 39, 40, 80, 83, and 84. The finding that peat bogs show significantly higher maximum HgAR than lake sediments is surprising, because lake sediments would be expected to show higher accumulation rates due to catchment inputs and sediment focusing. Such amplification of the atmospheric mercury flux by lake sediments would explain the differences in background accumulation rates. The median background HgAR in peat bogs is ∼1 µg m-2 yr-1 (0.4- 4 µg m-2 yr-1) as compared to ∼5 (1.4-18 µg m-2 yr-1) in lake sediments. Based on these data the median mercury flux ratio (increase from background to modern HgAR) is 32 (9- ∼500) in peat but only 3.6 (2-6.3) in lake sediments, which represents an order of magnitude difference between the two archives. This large disparity in flux ratios raises the question as to which archive gives the more realistic estimates of past and modern atmospheric mercury fluxes. First of all, it is inconsistent that lake sediments should have higher background mercury accumulations than peat, but that the situation should be reversed in more recent times (following industrialization). Possible reasons for this inconsistency could be changes in peat decomposition patterns, as discussed by Biester et al. (30, 32), or problems
with accurate dating of the uppermost peat sections (discussed below). An important first indication of the reliability of the two geochemical archives is a comparison of accumulation rates with direct measurements of atmospheric mercury deposition. Recent wet deposition rates for mercury in remote areas geographically and climatically comparable to those hosting mires, such as the northeastern United States, show values (∼4-8 µg m-2 yr -1; 57) that are lower by a factor of 3-6 (and as much as 18) than recent mercury accumulation rates derived from mires. These wet deposition fluxes are more similar to those determined in lakes. However, Lamborg et al. (31) did find good agreement in Nova Scotia between wet deposition of mercury and recent rates of mercury accumulation in both lake sediment and peat, but only when the peat was dated by Polytrichum increment-counting, and not 210Pb. Although these results imply that 210Pb-dating of peat is problematic (see below), it is possible that the modern mercury fluxes reported for some peat cores may include additional mercury inputs from dry deposition to the mire surface. However, the significance of this flux is difficult to evaluate, not only because we know little about dry deposition in nonforested landscapes, but because a large portion of VOL. 41, NO. 14, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4853
terrestrial mercury inputs may be revolatilized back to the atmosphere (58). Another important factor for the evaluation of the two archives is the chronology of mercury accumulation, especially the timing of the maxima associated with peak emissions in the late 20th century. Such a comparison might be questionable for archives located comparatively close to anthropogenic sources where atmospheric mercury deposition might vary considerably over short distances, but it should work for remote sites that reflect mercury emission trends at regional or continental scales. Dating of the maximum mercury accumulation in bogs is different from that derived from lake sediments. Lake-sediment records generally indicate a peak in mercury deposition during the 1970s to 1990s (e.g., 14, 17, 20, 23, 24, 31), consistent with mercury emission inventories for North America and Europe (14, 59, 60). In contrast several peat studies suggest a peak in deposition 10-20 years earlier (e.g., 37). Dating of Peat and Estimation of Mercury Accumulation Rates. The basis for calculating metal accumulation rates in geochemical archives is accurate dating. For the upper recent sections of peat and sediment cores there are two preferred methods for continuous dating: these are 210Pb and bombpulse carbon-14, which span approximately the past 150 and 50 years, respectively. Dating using bomb-pulse C-14 is based on matching the recent stratigraphic record of C-14 in peat with the known changes in atmospheric C-14 caused by weapons testing (61). Problems with this approach are differential uptake of carbon (62) and redistribution and recycling of carbon within the plants and substrate (63), which can result in adjacent peat sections containing the same amount of modern carbon and thus the same age. Peat layers can also be dated by biological chronometers, i.e., increment dating based on Polytrichum, which Lamborg et al. (31) applied to recent peat (
