Statement of Leonard Levin, Ph.D.
EPRI Palo Alto, California
to U.S. Senate Subcommittee on Clear Air, Wetlands, Private Property, and Nuclear Safety
Committee on Environment and Public Works,
Washington, D.C.
October 1, 1998


I am Leonard Levin, manager of the research program on air toxics and mercury at EPRI (founded as the Electric Power Research Institute). We are a nonprofit collaborative science and technology consortium with headquarters in Palo Alto, California. Members of EPRI represent about 87% of the U. S. regulated electric power industry with significant international participation. EPRI has a twenty-five year record of providing highly respected and objective science and technology to address important energy and environmental questions.

My own background is primarily in atmospheric sciences. My doctorate is from the Institute for Fluid Dynamics and Applied Mathematics at the University of Maryland, with other degrees from MIT and the University of Washington. I have over 25 years' experience in environmental sciences.

EPRI has sponsored research on mercury since the early 1980s, and in the past few years has joined in cooperative mercury research programs with the U.S. Environmental Protection Agency, the Department of Energy, the National Park Service, and other federal, state, and international agencies. The total EPRI research budget on mercury to date is about $20 million, at a level of about $2 million per year. My purpose today is to discuss what science currently can tell us about mercury in the environment, and more importantly to highlight the key areas where we know our understanding is less than sufficient.


The issue of mercury in the environment is complex. Mercury is not only emitted from many currently-operating industrial sources, but was used industrially at a much greater rate earlier in the century. As an natural element in the earth's crust, mercury cannot be created or destroyed. It may persist in various compartments of the natural environment. Industrial activity has liberated a great deal of mercury from the earth's crust and mobilized it into many other compartments, including the atmosphere, the biosphere, and the human environment. Recent concern has focused on the potential health risk to U.S. consumers of freshwater fish that might contain mercury, and in particular on those who consume fish at a relatively high rate.

Foodfish taken from both freshwater and marine environments exhibit mercury concentrations that extend over a wide range. Some of these fish, particularly larger, older fish such as shark and swordfish that feed on other aquatic species, may exhibit mercury levels that can raise health concerns. These levels of concern are set by federal and state agencies based on studies of accidental mercury poisonings at high doses that are extrapolated to estimate the effects of low doses on U.S. populations. Some of these poisonings involved fish as the route of exposure to mercury, but others did not.

Based on acute mercury poisonings that occurred in Japan and Iraq, it is known that high levels of mercury may cause measurable deficits in the mental and physical development of young children exposed during gestation. New scientific studies are now underway to clarify what mercury levels in children can produce adverse outcomes. Initial results from those studies indicate that mercury health effects on childhood development may be significantly less severe than previously believed.


Prior to enactment of the 1990 Clean Air Act Amendments, EPRI research teams carried out the first accurate measurements of mercury emissions from operating power plants. These and other data were provided to the U.S. Environmental Protection Agency by EPRI and other utility groups, and by the U.S. Department of Energy. As a result of such joint research efforts, our understanding of mercury in the U.S. environment has improved significantly in the last 10 years.

Nonetheless, many significant questions remain before a definitive quantitative assessment of the issue may be provided. Among these, three stand out:

What portion of all mercury emissions is contributed by U.S. industrial emissions?

How might mercury levels in fish reflect changes in the input of mercury to the atmosphere?

At what levels of exposure might mercury pose a health threat


A Note on Mercury Health Risk

Human health risk assessments are carried out to evaluate the likelihood that adverse effects may result from exposure to mercury. These assessments can be visualized as moving along two parallel pathways. One pathway is human exposure analysis: how

much mercury gets into the environment and reaches the human organism? The parallel analysis pathway is health hazard assessment at what levels of exposure does mercury pose a health threat?

When these two assessments are combined, they address how much mercury humans are exposed to, and whether that level of mercury represents a health threat. That result is the human health risk assessment.

A Note on Data vs. Computer Models

Because of the complex cycling of mercury through the environment, it is difficult to determine which sources contribute how much of the mercury found in fish. There are no current methods that allow us to, for instance, release a unique, benign material along with mercury from a particular industrial plant that would help us "tag" the mercury from that facility as it moves through the environment. For that reason, we must rely on computer models of mercury to assess its fate from source emissions, through atmospheric transport and deposition, to how much eventually makes its way into fish.

It is important to distinguish, therefore, between data, or observed and measured occurrences of mercury in the environment, and model results, which are computer outputs from the models used. As one example, we have surprisingly few data points on how much mercury deposits from the atmosphere to the surface, where people live, but there are many model results that portray what those numbers might look like. When these model results are compared to the sparse data, the model results tend to be rather uncertain, by a factor of two or more, either over- or underpredicting the observations. The conclusions drawn from these estimates concerning management of mercury should, therefore, be tempered by the uncertainties in the estimates on which the conclusions are based.

Emission sources

EPA and EPRI are in essential agreement that the amount of mercury currently emitted by electric utility generation is about 50 tons per year nationally. Essentially all of the utility-emitted mercury is attributable to coal-fired plants. Most of these emissions come from power plants that are equipped with very tall stacks. The available evidence suggests that such tall stack sources disperse mercury over great distances, with small fractions being deposited locally. Extensive measurements done by EPRI and DOE make utility plants the best-characterized sources of mercury.

By contrast, other man-made sources of mercury are less well-characterized with respect to amounts and properties of emissions. Very few measurements exist on some other categories, such as chloralkali plants. In addition, these sources tend to emit mercury closer to the ground surface and, in some cases, in amounts greater than power plants.

Recently, evidence has come to light that indicates motor vehicle fuels may be a source of mercury emissions and that source is therefore under-represented in our database.

These measurements, primarily by the University of Michigan, are only indirect to date, with more data expected in the near future.

The linkage between these sources and deposition is poorly understood, even for very large sources of mercury. This is largely because the chemical state of the mercury leaving a source is critical in determining how far the mercury will travel before it is removed from the atmosphere. Data on whether mercury from a particular source is ionized or in its "elemental" form are basically lacking for most sources.

"Legacy" emission sources remain a very large area of uncertainty. These include both natural releases from geological formations, and previously emitted mercury issuing from soil that is due to activities (such as gold mining) extending back for hundreds of years. Recent progress has been made by U.S. and Canadian teams in measuring these "legacy" quantities directly. These studies indicate that U.S. "legacy" releases may equal all current industrial emissions of mercury combined.

Mercury environmental fate and transport

Mercury emitted to the atmosphere from point sources can be either an elemental form, essentially mercury metal, or an ionic form that is more easily combined into chemical compounds. The ionic mercury is more water-soluble and can be deposited more easily close to its source by precipitation. The elemental mercury tends to enter long-range and global circulation, contributing to regional and global background levels. EPRI model results indicate that between about 70 percent and 95 percent of mercury exiting a coal utility stack will travel hundreds or thousands of kilometers before depositing to the earth's surface.

Other data, more indirect, indicate that about 2200 to 3300 tons of mercury are released to the atmosphere globally each year from today's human activities; all current U.S. manmade releases make up only about 7 to 10 percent of this total. Recent indications are that up to 1/4 of the global total may emanate from mainland Asia, for which data are essentially lacking.

Direct measurements of mercury depositing to U.S. territory are still sparse, and not representative yet of the entire nation. Model results are useful to indicate where potential "hot spots" may be located, but we often do not have verifying measurements from those areas. We also do not have any direct data quantitatively linking sources of atmospheric mercury with measured concentrations in soils, waterways, or aquatic species.

New field studies of mercury in the U.S. environment are just reaching fruition. Major work in Florida, funded by state and U.S. agencies, EPRI, and Florida utilities, is approaching completion, showing that a great deal of Florida mercury appears to originate globally, arriving in Florida with air masses traveling from Africa or southern Europe. A very large field study in the Lake Superior region began in mid-1998, and will continue for at least two years, involving researchers funded by a number of states, Federal agencies, EPRI, U.S. and Canadian utilities, and Canadian national and provincial agencies. These studies may begin to provide the extensive field data sets

needed to understand how mercury released from particular sources behaves in the environment.

Mercury in terrestrial and aquatic systems

Mercury in most parts of the environment is at extremely dilute levels, measured typically in trillionths of an ounce for each pint of water (or, nanograms per liter). Mercury entering U.S. waterways by atmospheric deposition or by runoff tends to wind up mostly in lake bottom sediments, with a small portion (perhaps one-one thousandth) eventually moving into fish tissue.

Some of the mercury that reaches lakes and wetlands is converted by bacterial action into an organic form called "methylmercury." Methylmercury is taken up by fish, and can be measured in some fish at concentrations that are very much higher than those in the water in which the fish live. As predatory fish eat other fish that contain some mercury, the concentrations of mercury in the predatory fish can increase many-fold, often reaching levels millions of times as high as the concentration in the surrounding water. High levels of mercury have been observed in predatory fish caught from remote lakes far removed from any point sources of mercury.

Despite our understanding in principle of how mercury enters fish caught for food, measurement data do not reveal the origins of the mercury found in these fish. We have no way of knowing how much originates from currently operating U.S. sources versus how much is recycled within the lake from the mercury "reservoir" in bottom sediments. This mercury in sediments may have originated decades or centuries earlier, or be due to releases from sources outside of the U.S.

Most importantly, we do not have a good understanding of the means by which open ocean fish, such as tuna or shark, take up and concentrate mercury. Indeed, one of the best-documented cases of elevated mercury levels in U.S. residents arose in a Wisconsin family that ate a diet rich in both locally-caught and imported fish. Subsequent studies showed that this family received its mercury dose primarily from supermarket-purchased Chilean seabass. The domestic fish consumed were found to have mercury levels of no concern.

Human exposure and health effects

A number of studies shows that nearly all the human exposure to methylmercury occurs via fish consumption. [There are two primary exceptions to this. One is accidental releases, usually in industrial processes, and usually for short periods. The second is mercury used in tooth filling amalgams; exposures from amalgams to developing fetuses are still under scrutiny.] This exposure may subject consumers to the organic form of mercury found in fish tissue. Such mercury typically resides in the human body for several weeks.

To date, mercury health risk estimates have primarily relied on data from a 1970 acute poisoning incident in Iraq that involved severe, rapid exposure from consumption of contaminated grain, and some deaths. These data are the basis for the current EPA

"Reference Dose" or health effects threshold level. This Reference Dose refers to a level of exposure that is without expected risk over a lifetime. A larger Reference Dose implies a less severe health effect from the substance, since it allows more mercury intake per day. The most sensitive individuals are children, who, even before birth, may suffer developmental effects from mercury entering their bloodstream from the mother. During pro- and post-natal development, mercury can act as a toxin to development of the central nervous system.

Recent research shows the current Reference Dose may be unnecessarily strict. By setting this threshold at a level well below that truly necessary to protect human health, unnecessarily stringent protective measures may be inadvertently required by regulatory agencies. In addition, there are proven health benefits from eating fish, not only for adult cardiovascular health, but for childhood neurological development. In addition, efforts to restrict mercury exposure may lead consumers to reduce their consumption of fish, even though the available evidence indicates that fish consumption has significant health benefits, even for children.

This re-examination of the Reference Dose is based on two new studies of children exposed to mercury via fish consumption (by themselves or their mothers, during children's gestation). These studies, in the Seychelle Islands in the Indian Ocean, and the Farce Islands in the North Atlantic Ocean, are of populations with diets that are highly dependent on marine life. The new studies are more relevant to U.S. populations that consume fish than are the data from the acute grain poisoning incident that took place in Iraq.

Results from these studies are still being analyzed. The initial findings from the Seychelles study indicate that no significant mercury effect was found over a wide range of pre-natal exposures to children. The Farces study has reported finding evidence of a neurological effect at the highest mercury levels. However, the biological significance of these findings remains unclear. Further analyses and refinements are expected in the results of these studies over the next two or three years.

Two independent analyses of the Seychelles results have suggested that the current EPA Reference Dose may be too severe by a factor of 3 to 5; that is, consumers can be exposed to mercury levels 3 to 5 times as great as the EPA level without harmful effect to children. This implies in turn a wider availability of fish from U.S. waters that can be considered safe for consumption.

Mercury Management Options

We do not yet have enough data to draw conclusions about which sources, or source types, contribute the most to mercury found in U.S. fish. Analyses indicate that background air emissions of mercury in the continental U.S., from natural plus "legacy" source areas, may be as large a source as all current industrial sources combined. In addition, global contributions to mercury are much larger than all U.S. sources combined. As a result, any potential changes in U.S. industrial emissions may leave the overall source term basically unchanged.

As indirect evidence of this, there are no data showing any overall lowering of mercury levels in fish from remote lakes over the last 35 years, despite an 85% drop in U.S. industrial mercury use in that time. Indeed, despite a relatively uniform national pattern of mercury deposition from the atmosphere, there are stark differences from lake to lake in the levels of mercury in fish, even the same fish species. For example, the State Health Department in Maine has noted that two adjacent lakes in Acadia National Park were found to have such different fish mercury levels that one lake carried a "no consumption" fish advisory, while the other lake was open to consumption of ail fish caught.

Additional Research

Our understanding of mercury has significantly advanced in the last decade, but a great deal remains to be done. It is important to remember that all aspects of the issue have critical uncertainties. It would be unwise, for example, to consider mercury health effects alone as the remaining uncertainty on the issue. Significant research is underway on other critical areas of uncertainty as well. Some of the major studies to be completed include:

_National studies of background mercury sources and fish consumption;

_ Refined methods to evaluate the links between source types and fish occurrences of mercury;

_ Completion and analyses of the new health effects studies in children by independent investigators


At the beginning of this discussion, I proposed three key questions that need to be examined. The first question suggested that we needed better information on mercury emissions from undocumented sources, particularly background sources. That information is now emerging, and appears to represent a large contribution to the national total. The second question asked how responsive fish mercury levels in the U.S. were likely to be to changes in atmospheric input. Indirect evidence shows that these levels are likely to respond very slowly to emissions changes. Finally, I raised the question of how severe the health effects from mercury are likely to be at U.S. exposure levels. Initial findings indicate we may have to revise our understanding of these health effects. In the case of one study, indications are that the effects of mercury might be quite a bit less than we once thought.

Thus, our understanding of the sources and behavior of mercury in the environment, and of its potential health effects, is entering a new phase. Over the next two to three years, results from current studies will appear in the scientific literature and allow a more informed examination of whether there is a basis in health risks for managing mercury sources. Informed decisionmaking can then be based on the most relevant scientific information.