Fish respond when the mercury rises

Approximately 40 years ago, researchers in Sweden and Canada made the startling discovery that fish from relatively pristine northern lakes contained unusually high levels of mercury (Hg) in their muscle tissue. Up until that point, mercury-contaminated fisheries were known only from locations receiving end-of-pipe discharges of mercury-laden wastes. In this case, however, lakes were remote from any such point-sources but still had fish mercury levels high enough to cause human-health problems, principally neurological damage, if consumed in sufficient quantities (1). It would be another decade before the development of clean sampling and analytical techniques would allow researchers to link the exceedingly low levels of mercury present in lake water to their biomagnification in the aquatic food chain and back to their ultimate source in industrial atmospheric emissions (2, 3). Since that time, mercury scientists have determined that thousands of lakes worldwide (the vast majority of those tested) are similarly contaminated; they have unraveled the microbial processes by which elemental mercury is converted to its bioaccumulative form, methylmercury (4); and they have determined through the analysis of natural geologic archives (cores of lake sediments and peat bogs) that human activities are responsible for approximately two-thirds of the mercury now present in the global atmosphere (5). In this issue of PNAS, Harris et al. (6) add a key piece to this giant biogeochemical puzzle: the experimental determination that mercury levels in fish respond rapidly and directly to changes in atmospheric mercury deposition.

Most mercury researchers have assumed for some time that fish mercury levels were tied to rates of atmospheric deposition. Evidence that this is so comes from (i) observed reductions in fish mercury with cessation of point-source mercury discharges, (ii) the correlation along spatial gradients of mercury deposition rates with mercury levels in fish, and (iii) the observed declines in fish mercury in regions experiencing reductions in mercury emissions and deposition (7). Each of these observations has its weaknesses; sites of point-source mercury discharges, for the most part, are massively contaminated and therefore might be poor analogues for sites experiencing modest changes in atmospheric deposition. The other two approaches are correlative only and may be underlain by other factors that control mercury cycling and uptake.

Although it would seem only logical that the solution to the mercury problem would be to greatly reduce atmospheric emissions and that mercury levels in fish (and fish-eating wildlife) would fall accordingly, the connection among mercury emissions, deposition, and biological uptake in aquatic systems is exceedingly complex, with multiple factors controlling the levels of mercury found in predatory game fish at the top of the food chain (Fig. 1). For example, a recent study of lakes in Voyageurs National Park along the U.S.–Canadian border of Minnesota found that mercury levels in northern pike (Esox lucius) varied by an order of magnitude, despite a regionally uniform rate of atmospheric mercury deposition, largely because of differences in the methylating efficiency of the lakes and their watersheds (8). Furthermore, there are large stores of mercury present in soils and lake sediments from past atmospheric deposition, some natural and some anthropogenic, that might sustain high levels of fish mercury long after a reduction in atmospheric emissions. Given the complex linkage between mercury deposition rates and mercury levels in fish and the economic costs associated with mercury control technology, there are those who would argue for a go-slow regulatory approach to emission reductions, such as that recently adopted by the U.S. Environmental Protection Agency.

Fig. 1.

Mercury cycling in a lake and its watershed. Mercury emissions are transported long distances, primarily as gaseous elemental mercury [Hg (0)], oxidized in the atmosphere to reactive gaseous mercury [Hg(II)], and deposited in precipitation and by surface contact (dry deposition). Anaerobic bacteria convert a small portion of the incoming Hg(II) to methylmercury (MeHg), which is then bioconcentrated in the aquatic food chain (by a factor of ≥10⁶). Various biotic and abiotic reactions interconvert the different forms of Hg, affecting uptake, burial, and evasion back to the atmosphere.

The power of the study by Harris et al. (6) in addressing this debate is that it is based on a whole-ecosystem manipulation of mercury deposition rates in which changes in mercury levels in a small boreal lake and its watershed were followed through the addition of highly enriched stable isotopes of mercury, one each to the lake surface, its upland catchment, and a lake-fringing wetland. By operating at the ecosystem scale, the research team was able to overcome natural complexities in the transport and cycling of mercury, something that laboratory and small-scale field experiments cannot do. Moreover, the use of isotopic spikes allowed the researchers to distinguish between movements of the newly added mercury and that of the large stores of preexisting (ambient) mercury, all at environmentally relevant levels.

Now, after 3 years of experimental additions, the rewards of this effort are clearly evident. Among its key findings are the rapid movement of the lake spike (²⁰²Hg) through the various aquatic compartments. Within days of the first biweekly addition, methylated ²⁰²Hg appeared in the lake's bottom waters; in 1 month, it was in the zooplankton and benthos; and within 2 months, it was measurable in several fish species. Concentrations rose linearly with each successive year and were 30–40% higher at the end of the third year. In contrast, it was not until year 3 that trace amounts of the upland spike (²⁰⁰Hg) began to show up in the fish, whereas no wetland spike (¹⁹⁸Hg) was ever detected in the lake. It appears that new mercury inputs are held tightly by catchment soils and vegetation and bleed out only slowly along with the ambient mercury. Indeed, the pools of ambient mercury in upland soils and wetland peat were still orders of magnitude larger than the isotope spikes after 3 years of addition. Such results imply that mercury levels in lakes will respond rapidly to reductions in mercury deposition directly to their surfaces and much more slowly to changing inputs to their watersheds. Thus, as the authors predict, lakes should show a two-step response to declines in atmospheric mercury, rapid (years) at first and gradual (decades) thereafter, and lakes with small watersheds (relative to their surface areas) should be most responsive.

One of the few disappointments of the experiment was the lack of response

Mercury levels in fish respond rapidly and directly to changes in atmospheric mercury.

to the wetland mercury spike. Previous studies of mercury cycling, many by coauthors of the current PNAS report, have demonstrated that wetlands, with their water-saturated partially anaerobic soils, are important sites for mercury methylation and that lakes with extensive wetlands in their watersheds have higher fish-mercury levels than lakes that do not (8). In fact, Harris et al. (6) expected a rapid response from the wetland addition based on a pilot study at a nearby lake, where mercury spiked to a shoreline wetland was quickly methylated and exported to the lake (9). The reason that the full-scale experiment did not conform to expectations appears to be related to differences in hydrology. The water table in the pilot wetland was near the peat surface and well connected to its lake, whereas the full-scale wetland was higher and drier and did not have much of an outflow. Hence, after 3 years of addition, most of the full-scale mercury spike was still tied up in the vegetation and had not reached the water table, where it might be methylated and exported. What this points out is the importance of hydrological connectivity: sites with high methylating potential need to be intimately tied to a lake to have much of an effect. In the study by Harris et al., most of the methylmercury was produced within the lake (in the sediments or anoxic bottom waters), whereas other lakes with extensive and well connected wetlands receive much of their methylmercury from their watersheds.

Such is the complexity of mercury cycling, where side-by-side lakes can show greatly different responses to the same (or changing) inputs of atmospheric mercury. Little wonder that some might question the link between mercury emissions and mercury levels in fish. Yet, with the report of Harris et al. (6), we now have the “smoking gun”: mercury levels in fish respond directly and rapidly to changing mercury inputs from atmospheric deposition. That we are observing decreases in fish mercury in areas where mercury deposition has declined in recent years (10, 11) is both confirmatory and encouraging. Mercury is a ubiquitous and serious pollutant, and through the course of our industrial activities, we have painted the globe with it. Some of the slowly cycling pools at the Earth's surface, the deep oceans for example, are still increasing and are far from steady-state with respect to current emissions (5). It will take centuries or longer for current levels of anthropogenic mercury to be permanently sequestered in deep geologic stores. Yet here is evidence that lakes will respond to reductions in atmospheric deposition, rapidly at first and more slowly over the long haul. Such results would indicate that emission reductions are both prudent and warranted.