Acid Deposition: Long-Term Trends (1986)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

1 Summary and Synthesis Over the past decade or so, the phenomenon of acid rain, or more properly, acid deposition, has evolved in the United States from a scientific curiosity to an issue of considerable public concern and controversy. The issues raised by its possible adverse effects are not confined to specific localized areas, but are regional, national, and even international in scope. A wide variety of effects have been attributed to acid deposition, its gaseous precursors, and certain products of their chemical reactions including ozone. Possible environmental con- sequences include adverse effects on human health, acidi- fication of surface waters with subsequent decreases in fish populations, the acidification of soils, reduced forest productivity, erosion and corrosion of engineering materials, degradation of cultural resources, and impaired visibility over much of the United States and Canada. To evaluate these possibilities, scientific hypotheses have been formulated linking postulated or observed effects to acid deposition and/or its precursors. For example, acidification of lakes is thought to be the result of the deposition of acidifying substances, either directly as deposition to the water surface or indirectly by interaction with soils in the watershed to enhance transport of hydrogen and aluminum ions to surface waters Fish are adversely affected by acidification and the increased concentrations of aluminum that frequently accompany it. Alternatively, other hypotheses involving both natural acidification processes and/or other factors related to human activity have been proposed to explain the same effects. For example, land use practices such as timber harvesting, agriculture, and residential development are known to affect surface water chemistry and in specific circumstances might be more important for 1

2 water quality than acid deposition. Similarly, changes in fish populations may be influenced by changes in stocking policies, the introduction of competing fish species, commercial or sport fishing, and pollution from pesticides or other commercial chemicals. It appears that an alternative explanation based on other human activities or natural phenomena can be or has been proposed for every proposed link between the depo- sition of airborne chemicals and an adverse environmental effect. In many instances several alternatives are plausible. This study was organized to investigate spatial pat- - trends in acid deposition and its terns and temporal gaseous~precursors in eastern North America and patterns and trends in environmental parameters that might result from acid deposition. The Committee on Monitoring and Assessment of Trends in Acid Deposition was asked not only to review previous efforts in this regard, but also to extend the analyses if there were approriate data. To meet these requirements we had to perform new analyses and make extensive checks on the quality of original data. We assumed that if a mechanism existed linking acid deposition to an environmental effect, it should be possible to demonstrate that acid deposition is associated spatially and/or temporally with the effect. If such an association cannot be established, then either a cause- and-effect relationship does not exist or our understand- ing of the mechanisms and rate-governing factors is not adequate. Despite imperfect knowledge of the relation- ships between emissions and deposition and rates of responses of ecosystems, careful evaluation of data on phenomena that are linked to acid deposition by plausible mechanisms should provide additional insight into the relationships among emissions, deposition, and effects. Some of the questions asked about the spatial relations of acid deposition and related phenomena were the follow- ing: Is the deposition of acidic sulfates and nitrates highest in areas where densities of emissions of sulfur and nitrogen oxides are highest? Are patterns of ecosystem changes attributed to acid deposition also . found in regions of low deposition? With regard to temporal associations, we asked: How has the chemical composition of precipitation changed with time? How acidic was precipitation or dry deposition 30 years ago? 50 years ago? 100 years ago? Does the timing of postulated changes in aquatic and terrestrial ecosystems coincide with changes in patterns of sulfur and nitrogen oxide emissions?

3 Simultaneous examination of multiple patterns of spatial distributions and temporal trends provides a more robust test for the existence of linkages between emis- sions, deposition, and environmental effects than infer- ences based on data for pairs of phenomena. In conducting the study, it quickly became apparent that answers to questions about acid deposition have proved elusive because historical data from which to judge trends and hypothesize environmental responses are scarce. The earliest records of the direct measurement of levels of acidity, sulfate, and nitrate in deposition in North America date from as early as 1910 (McIntyre and Young 1923, Harper 1942, Hidy et al. 1984), one year after the invention of the pH scale for measurements of the acidity of aqueous solutions. Although they are informative, data obtained before 1950 are sketchy and of limited value because they pertain only to a few locations over short periods of time and in many cases are not reliable. Only since the late 1970s have extensive deposition monitoring networks been established to gather quality- assured data in a systematic way. In some cases however, longer and more extensive records do exist for systems thought to be affected by acid deposition or its airborne precursors. These records include long-term data on visibility, chemical composition of waters in lakes and streams, chemical and biological composition of lake sediments, fish popula- tions, Growth patterns in trees as evidenced in ring widths, distributions of lichens, erosion of tombstones, and chemical composition of glacial ice cores, ground- water, and soils. Unfortunately, investigations in a number of these areas conducted before the early 1970s were not designed to study acid deposition per se, and hence the original records usually do not include all the information required for a definitive intepretation of temporal changes. For example, the interpretation of earlier lake and stream chemical data is difficult because the records frequently fail to include adequate documentation of the sampling procedures and chemical analytical methods employed. Also often missing are quantitative descriptions of the variability in the data that may have been introduced by analogous changes in climate or weather or by human activities such as changes in land use patterns. On the other hand, the establishment of spatial rela- tionships among current values for emissions, deposition, and environmental responses appears to be more straight-

4 forward because information about factors that might bias the analyses is generally more readily available. To determine which types of phenomena to study and which data were appropriate and feasible to include in the study, we developed certain criteria, the most important of which were the following: 1. A documented or postulated relationship, either direct or indirect, to acid deposition. 2. Availability of published data or original data with sufficient documentation to permit peer review. 3. Availability of data representative of broad geographical regions and/or temporal data with unambiguous dating. Consequently, we selected the following for inclusion in the study: 1. Emissions of sulfur and nitrogen oxides (Chapter 2). The major contributors to atmospheric deposition of sulfur and nitrogen compounds in North America are anthropogenic emissions of sulfur and nitrogen oxides. Estimates of emissions have been compiled in this report based on data on the production and use of fossil fuels, estimates of their sulfur content, and emission factors for nitrogen oxides released during combustion. 2. Precipitation chemistry (Chapter 5). Quality- assured data on precipitation chemistry for a broad region in eastern North America have been available since about 1978. These data permit spatial analyses of the chemistry of wet deposition in the region, but the time period of this record is not sufficiently long to estab- lish statistically significant temporal trends. Time series of longer duration, dating from the early to the middle 1960s, are available at a few sites, however, and we have performed trend analyses on some of these data. 3. Atmospheric sulfates and visibility (Chapter 4). A direct effect of sulfur dioxide emissions is the production of atmospheric sulfate aerosols that reduce visibility. Historical and spatial data on atmospheric sulfate and visibility are available from a number of stations. 4. Surface water chemistry (Chapter 7). One of the most studied effects of the deposition of acidic chemical species is the acidification of surface waters. Many data are available for analysis. We selected what we judged to be key data sets and included in our analysis _

5 those that were amenable to rigorous assessment of their reliability. In all cases checks for internal consistency of the original historical data had to be Der formed befor e data were incorporated into the analyses. , ~ 5. Sediment chemistry and abundance of diatom tax a - (Chapter 9). Changes in watershed and lake chemistry are recorded in lake sediments, which provide historical data of the longest time series. Dating and chemical analysis of successive intervals of sediment cores provide chrono- logical information on changes in water chemistry. Assemblages of diatoms and chrysophytes in sediments can be analyzed to reconstruct historical lake water acidity. 6. Fish populations (Chapter 8). Fish populations are hypothesized to decline in acidified lakes. Some records are available relating fluctuations in popula- tions of selected fish species to historical records of lake and stream chemistry. 7. Tree rings (Chapter 6). The decline of forest trees is perhaps the most controversial phenomenon that some researchers have attributed to acid deposition. Some tree ring data are available for red spruce populations in the higher elevation forests of the northern Appalachian Mountains. It became apparent during the evaluation processes that we would not be able to rely exclusively on the published literature if we were to meet our goal of examining spatial patterns or temporal trends of multiple phenomena. In some cases extensive evaluations of ~ _ _, _ spatial distributions or temporal trends had not been performed for even a single phenomenon and thus our evaluation depended, at least in part, on our own analysis of unpublished data. Owing to limitations in available data, the effects of oxidants and other air pollutants are not considered in this report. The committee's findings and conclusions are listed below. In subsequent sections of this chapter we describe the rationale employed in drawing these conclusions. First we present our methodology for assessing the likelihood of a cause-and-effect rela- tionship based on the criteria of mechanism and consistency of data. We then briefly discuss the specific mechanisms that may link acid deposition to related phenomena and some factors that complicate our analysis. We follow this section with a detailed analysis of the degree of consistency in spatial patterns and temporal trends among acid deposition, emissions, and .

6 the environmental changes often attributed to acid deposition. F INDINGS AND CONCLUS IONS When trends and patterns are found that establish temporal and spatial consistency in cases for which plausible mechanisms link acid deposition to other phenomena, cause-and-effect relationships can be postulated with some degree of confidence. Previous attempts to evaluate temporal trends have been limited because of large uncertainties inherent in historical data bases. Our premise was that through careful selection of a number of types of data and a number of quality-assured data bases, a more robust analysis for consistency and associations among the data might be possible. We believe that the results of our analyses, presented in later sections of this chapter and in the following chapters of this report, demonstrate the validity of this approach, and that we can formulate the following major findings and conclusions: 1. Through statistical analysis of regional spatial patterns, we find a strong association among the following five parameters: (a) emission densities of sulfur dioxide (SO2), tb) concentrations of sulfate aerosol, (c) ranges of visibility, (d) sulfate concen- trations in wet precipitation, and (e) sulfate fluxes in U.S. Geological Survey Bench-Mark streams. From this result and because of the existence of plausible mechanisms linking the phenomena, we conclude that in eastern North America a causal relationship exists between anthropogenic sources of emissions of SO2 and the presence of sulfate aerosol, reduced visibility, and wet deposition of sulfate. Our analysis also indicates that for Bench-Mark streams in watersheds showing no evidence of dominating internal sources of sulfate there is a cause-and-effect relationship between SO2 emissions and stream sulfate fluxes. Magnitudes of sulfur emissions and deposition of sulfur oxides are highest in a region spanning the midwestern and northeastern United States. 2. Based on data on fossil fuel production and consumption, we conclude that acid precursors, par- ticularly S02, have been emitted in substantial quantities in the atmosphere over eastern North America

since the early l900s. In particular, SO2 emissions in the northeastern quadrant of the United States have fluctuated near current amounts since the 1920s. These conclusions are supported by limited data on long-term trends in visibility and the presence in lake sediments of chemicals emitted during combustion of coal and other fossil fuels. 3. Substantial differences in temporal trends in SO2 emissions among regions of the United States have emerged since about 1970. Before 1970, temporal trends in SO2 emissions in the various regions were congruent, although the amounts of emissions were of different magnitudes. From data on SO2 emissions, reduction in visibility, and sulfate in Bench-Mark streams since about 1970, we conclude that the southeastern United States has experienced the greatest rates of increase in parameters related to acid deposition. The midwestern United States has experienced rates of increase somewhat lower than the Southeast. In the northeastern United States the trend has been one of modest decreases. 4. The record of the chemistry of Bench-Mark streams suggests that changes in stream sulfate flux determine changes in stream water alkalinity and base cation concentrations in drainage basins that have acid soils and low-alkalinity waters. Increases in stream sulfate flux are associated with decreases in alkalinity and/or increases in amounts of base cation in surface waters. The change in alkalinity per unit change in sulfate depends on site-specific characteristics. Changes in sulfate observed in Bench-Mark streams are consistent with changes in SO2 emissions on a regional basis. Analysis of a sulfur mass balance for 626 lakes in the northeastern United States and southeastern Canada demonstrates that the sulfate output from lakes in general is proportional to sulfate inputs in wet deposition. The ratio of output to input decreases with distance from major source regions, suggesting that dry deposition is an important contributor to sulfate flux inputs, especially near major source regions. 5. Data on alkalinity of some lakes in New York, New Hampshire, and Wisconsin suggest that changes in alkalinity greater in magnitude than about 100 peq/L can occur over time periods of about 50 years. Changes of this magnitude are too large to be caused by acid deposition alone and may result from other human activities or natural causes. We have not attempted to identify the exact nature of the causes of these large changes.

8 6. Analysis of diatom and chrysophyte stratigraphy for sediments in 10 low-alkalinity Adirondack Mountain lakes studied indicates that 6 of them became increasingly acidic between 1930 and 1970. Because the trend in acidification is consistent with both the presence of other substances in sediments that indicate fossil fuel combustion and current lake acidification models, and because the observed acidification cannot be explained by known disturbances of the watersheds or by other natural processes, acid deposition is the most probable causal agent. These findings are supported by fish population data for 9 of the 10 lakes in the Adirondacks for which concurrent data exist. Diatom data from lakes in New England indicate slight or no decrease in pH. Data for southeastern Canada are insufficient to examine trends in acidification. 7. Based on comparisons of historical data on alka- linity and pH recorded in the 1920s, 1930s, and 1940s with recent data for several hundred lakes in Wisconsin, New Hampshire, and New York, we find that many lakes have decreased in pH and alkalinity and many have increased in pH and alkalinity. On average, lakes sampled in Wisconsin have increased in alkalinity and pH. The New Hampshire lakes on average show no overall change in alkalinity and a small increase in pH. Interpretation of changes in the New York lakes is sensitive to assumptions about the application of calorimetric techniques in the historical survey and the selection of recent data bases. Depending on the assumptions, New York lakes on average either experienced no changes in alkalinity and pH or have decreased in alkalinity and pH. In the judgment of the committee, the weight of the evidence indicates that the atmospheric deposition of sulfate has caused some lakes in the Adirondack Mountains to decrease in alkalinity. We base this conclusion on three types of evidence: (a) Sulfate concentrations in wet-only deposition in the region of the Adirondack Mountains and sulfate concentra- tions in Adirondack lakes are relatively high in comparison with those in other areas in the northeastern United States and southeastern Canada. We have demon- strated that increasing sulfate in surface waters is associated with decreasing alkalinity in low-alkalinity surface waters. (See Conclusion 4.) (b) Diatom-inferred pH and other supporting evidence provide a strong indication of acidification from acid deposition in low-alkalinity lakes. (See Conclusion 6.) (c) We calculated alkalinity changes in New York lakes four

9 different ways to account for different assumptions. Three of the results indicate, on average, a decrease in alkalinity (median values of -28, -44, and -69 peq/L), and one result shows no overall change (median value of +1 peq/L). Because of ambiguities regarding the assumptions employed in the historical New York survey, we cannot currently determine which of these results is most accurate, and hence we cannot quantify the number of New York lakes that have been affected by acid deposition. 8. Emissions of oxides of nitrogen (NOX) are estimated to have increased steadily since the early 1900s, with an accelerated rate of increase in the Southeast since about 1950. Reliable data do not exist to determine historical trends of nitrate concentrations in the atmosphere, precipitation, or surface waters. 9. Although high-quality data to assess trends in fish populations as a function of surface water acidity are sparse, the data that are available indicate that fish populations decline concurrently with acidification. The strongest evidence in support of this finding comes from some Adirondack Mountain lakes. These data demon- strate declines in acid-sensitive fish species populations over the past 20 to 40 years in lakes thought to have been acidified over the same time period. (See Conclusion 6.) The number of cases studied is too small to permit any projections of the total number of fish populations that may have been affected by acidification. 10. Geographically widespread reductions in tree ring width and increased mortality of red spruce in high- elevation forests of the eastern United States began in the early 1960s and have continued to the present. The changes occurred about the same time as important climatic anomalies and in areas subject to comparatively high rates of acid deposition. The roles of competition, climatic and biotic stresses, and acid deposition and other pollutants cannot be adequately evaluated with currently available data. METHODS It is important to establish the requirements for inferring that a relationship between data sets implies causality. Two variables can be considered to be associated if their values are paired in some related way across a population, and they are unassociated if a

10 special pairing does not exist. To establish that an association exists, it is necessary only to show that the variables do not appear to be paired at random. Thus, as we will see in Chapter 5, sulfate and nitrate in wet deposition are associated across eastern North America; regions with high wet sulfate deposition also tend to have high wet nitrate deposition, and vice versa. Association between variables is necessary but not sufficient to infer the existence of a causal relation- ship. Mosteller and Tukey (1977) list three criteria-- consistency, responsiveness, and mechanism--at least two of which are usually needed to support causation. Con- sistency implies that (all other things being equal) the relationship between the variables is consistent across populations in direction, or perhaps even in amount. If a relationship between the variables holds in each data set, then the relationship is consistent. Responsiveness involves experimentation. If we can manipulate a system by changing one variable, does the other variable also change appropriately? Mechanism means a step-by-step path from the "cause" to the "effect," with the ability to establish linkage at each step. Observation of a correlation between two variables can establish consistency, but it cannot establish either responsiveness or the mechanism of possible causation. Thus, correlation is not adequate to prove a cause-and- effect relationship. Continuing the earlier example, sulfate and nitrate in wet deposition have a clear, con- sistent relationship, but neither experiment nor mechanism implicates changes in one as causing changes in the other. In fact, we know that they both arise from a common source, the high-temperature combustion of fossil fuels (National Research Council 1983). In this report, we use the criteria of mechanism and consistency for suggesting cause-and-effect relationships. Some controlled studies of responsiveness are discussed, but field experiments on most of the variables are gen- erally not considered practicable. In the next section of this chapter, we describe a number of conceptual linkages among the various types of data. These linkages are in fact mechanisms, stepping from one variable, e.g., emissions of SC>, to another variable, e.g., visibility, with causation natural at each step. Ideas about mech- anisms then motivated analyses to determine whether consistent relationships exist among the variables over space and time.

11 MECHANISMS In this section, we summarize the conceptual mechanisms that could link emissions, atmospheric deposition, and environmental responses; the analyses of trends and spatial patterns are discussed in the following section. The general conceptual relationships are shown schemati- cally in Figure 1.1 and are described further in the following chapters. The figure does not depict all the possible environmental interactions of sulfur and nitrogen oxides or the many possibilities for their ultimate fates. It does, however, indicate relationships among the phenom- ena examined in this report: emissions, visibility, chemistry of precipitation, chemistry of lake and stream waters, fish populations, forests, and chemical and biological stratigraphy of lake sediments. Dry deposition is shown in this general diagram and is discussed in various chapters of this report, but lack of data precluded any detailed quantitative analysis of its temporal trends. ATMOSPHERE dispersion + transformation modulated by climate -2 deposition: wet and dry emissions ~ ambient SO4 Box+ NOx ~ and visibility forests ~ i: :''~''~'~'': __ combustion : ~:~ i:: ~::~: ~~:~ ~~ ~~ -~:~ ~:~ ~~-~;;~;=~_ of fossil fug s : BIOSPHERE ~~ ~ ; watershed) ~~° fish >by : :: :~: :: :~:: : :~ ~ ~ ~~ i::: ~ ~ i: ~ ~~:.~ .~: ~ ~ i: i: ~ ~ ::: ~~:~ ~ :-: ~ :~:: ~ ~ ~ ~ : FIGURE 1.1 Acid deposition: affected ecosystems. diagram of sources and

12 Mechanisms of Linkage Among Atmospheric Processes Emissions In eastern North America more than 90 percent of the emissions of sulfur and nitrogen oxides into the atmos- phere are the result of combustion of fossil fuels (Robinson 1984). These oxides (SO2 and ~Ox) are transformed in the atmosphere into sulfate (SOi~) and nitrate (NOR), respectively, by complex series of catalytic and photochemically initiated reactions (National Research Council 1983). The transformation processes and rates depend on factors such as sunlight, temperature, humidity, clouds, and the presence of the various catalysts. The sulfates, nitrates, and other reaction products are subsequently removed from the atmosphere by both wet- and dry- deposition processes. Visibility Sulfate in the atmosphere takes the form of fine particles that scatter light and reduce visibility. While other types of airborne particles including moisture, carbonaceous particles, and road dust cause visibility reductions, sulfate aerosol is believed to account for a large fraction of the observed effect in the eastern United States, and in some areas visibility reduction may be the most directly observable effect resulting from emissions of sulfur dioxide. (See Chapter 4.) Although the exact portion of visibility reduction attributable to particulate sulfate is uncertain and probably varies from region to region, investigators have estimated that, on average, about 50 percent of the reduction is caused by particulate sulfate (often dissolved in fine water droplets) in the eastern United States, with a somewhat lesser contribution in urban than in nonurban areas. Short- and long-term variability in climatic factors such as temperature, relative humidity, and frequency of air stagnation events may affect visibility and can complicate the association between trends in visibility and trends in SO2 emissions.

13 Deposition Historically, concern about acid deposition has focused on the acidity (pH) of precipitation as the important factor in assessing potential environmental effects. The terms acid rain and acid deposition reflect this focus. More recently, as the understanding of the phenomena has improved, attention has shifted to the deposition of sulfur and nitrogen compounds (as well as other chemicals) from the atmosphere. Concentrations of sulfate and nitrate ions (SO4 and Not , respectively) in wet and dry deposition are related to sulfur and nitrogen oxide emissions, while the "acidity" of wet deposition is the result of concentra- tions of both acidic compounds (i.e., strong acids (nitric and sulfuric acids) as well as weak organic acids) and alkaline materials incorporated in deposition. Thus, in establishing the relationship between emissions and environmental effects, knowledge of the fate of sulfur and nitrogen compounds is critical, and in this report we have focused attention on the emissions, deposition, and fates of sulfur and nitrogen oxides. Sulfur and nitrogen oxides and their transformation products are deposited from the atmosphere onto the Earth's surface as acids or acidifying substances. The processes that remove the constituents from the atmos- phere are complex (National Research Council 1983). How quickly a given substance is removed from the atmosphere depends on its physical form, meteorological events, and the characteristics of the surface. Average residence times in the atmosphere of SC2, NOk, and the important reaction products (SO42, Nod , etc.) range from hours to weeks. During this time they may travel either short distances (0 to 100 km) or longer distances (100 to 1000 km) from their sources before they are removed from the atmosphere. Average distances of transport are generally ~ _ a few tens to a few hundred kilometers. Wet deposition is only one of two major pathways of atmospheric deposition. The other pathway, dry depo- s~tzon, is an Important process for the removal of gases and airborne fine and coarse particles. Wet and dry deposition combined provide the bases for interactions of the compounds involved in acid deposition with the ter- restrial and aquatic ecosystems. Data on dry deposition, however, are sparse, and our analyses of temporal trends and spatial distributions of acid deposition are based primarily on wet-only deposition. (See Chapter 5.)

14 Mechanisms of Linkage Between Acid Deposition and Surface Water Chemistry Most of the water in lakes and streams has flowed through the forest canopy, soils, and geologic materials of the watershed before entering the lake or stream. Because terrestrial watersheds are chemically reactive media, any model linking acid deposition to surface water acidification must account for the nature of the inter- actions among the soils and soil solutions in watersheds. Similarly, the analysis of trends must be sensitive to the uncertainties inherent in quantifying the processes. (See Chapter 7.) The "mobile anion" hypothesis implicates sulfate as the major determinant in lake acidification (Seip 1980). Sulfate in soil solution moves through soils as mobile (negatively charged) anions. When this occurs, an equivalent amount of (positively charged) cations must also be transported. This equivalence is necessary to satisfy the requirement for a charge balance in all aqueous solutions. The common cations in acidic soils are aluminum iA13+), hydrogen tH+), calcium (Ca2+), magnesium (Mg +), potassium (K ), and sodium (Na+). As the concentration of sulfate in the soil solution and soil leachate increases, the concentrations of dissolved cations must also increase. In general, lakes and streams in watersheds having hydrologic flow paths through highly acidic soils are expected to match an increase in sulfate concentration with a higher proportion of acid cations (H+ and A1+3), thereby reducing surface water alkalinity. Less acid soils will tend to counter an increased sulfate flux with a higher proportion of base cations (Ca2+, Mg2+, or Na+). Sulfates are not mobile in all soils. Watersheds may retain or release sulfate by adsorption and reduction- oxidation (redox) reactions that are often associated with biological fixation. These reactants may cause increases in acidity (oxidation) or decreases in acidity (reduction) (Wagner et al. 1982). Some watersheds may contain weatherable sulfur-bearing minerals in the soil, subsoil, and bedrock that can act as sulfur sources. The fate of nitrate is also complicated because it is biologically active. During the growing season, nitrate introduced into the terrestrial environment is likely to be converted to organic nitrogen in plant material or microbial biomass by a number of processes. As with sulfur, most of these processes either produce acid

15 (oxidation) or consume acid (reduction). Nitrate entering surface waters through either runoff or direct deposition may be rapidly consumed by aquatic biota with the result that excess concentrations of nitrate are eliminated. early spring, nitric acid released from the watershed can be the major cause of a share, but temporary decrease in pH in lakes and streams. This nitrate may be released from the snowpack, the soil, or both. Organic soils of geologic terrain that are resistant to weathering are naturally acidic. Waters moving through these soils, if not already acidic, may acidify, and receiving lakes naturally tend to become more acidic with time. This process is usually slow (i.e., on the order of centuries to millennia for changes of the order of 1 pH unit). If the hydrologic pathways to surface waters through humic soils are direct, acidic surface waters dominated by organic acids may result. Further- more, changes in land use practices can cause changes in inputs and outputs of acids in a watershed with subse- quent changes in surface water chemistry. The effects of land use practices on the pH of surface waters are not well known, but lakes may become more or less acidic, depending on site-specific conditions. Differences in the geologies of watersheds and in the histories of land use practices within watersheds can cause surface waters in a region to respond differently even while they are receiving equivalent amounts of acidity from atmospheric inputs. Other factors, such as changes in the volume of water, can cause seasonal and interannual changes in alkalinity. Thus there are likely to be a number of contributing causes for trends in surface water alkalinity. However, for a very brief period in the , _ . . . . . . Mechanisms of Linkage Between Acid Deposition and Lake Sediments The chemical composition of lake sediments reflects the net results of all chemical inputs and outputs over time. Changes in atmospheric chemical inputs to soils can modify leaching rates of certain chemical species from the terrestrial environment that serve as inputs to surface waters. ~ Additionally, many constituents deposited from the atmosphere, both solids and liquids, may be incorporated directly into accumulating sediment. Thus, lake sediments provide, directly or indirectly, a natural

16 stratigraphic record of the changes in the chemical composition of surface waters and may reflect changes in atmospheric deposition. In addition, the sedimentary record may preserve information on the populations of biological organisms that can be related to water quality at the time when sediment was deposited. (See Chapter 9.) However, in the case of both the chemical and the biological record, older sediment from shallower water may, with their respective characteristics, be incor- porated into younger accumulating sedimentary sequences Thus changes in atmospheric deposition of metals to the lake on watershed may not be directly recorded by the sediments. Sediment Chemistry Sulfur deposited from the atmosphere may be added to sediment in a variety of ways, including incorporation into sedimentina organic matter. adsorption onto the . ~ — A J surface of particles, chemical migration into sediment as sulfate, reduction of sulfate to sulfide, and precipita- tion from solution as sulfide. Thus, it is not possible to draw conclusions about the precise chronology of the deposition of sulfur compounds from analysis of concen- tration as a function of depth in a sediment core. No data exist to suggest that the nitrogen content of sediments reflects the deposition rates of nitrogen compounds. However, the industrial processes from which SO2 and NOx are emitted also produce emissions of a number of heavy metals (for example, lead, zinc, copper, and vanadium). Concentrations of these metals typically are low in lake sediments in unpolluted regions and increase in more densely populated regions. Net accumulation in sediment of certain metals (for example, zinc, manganese, and calcium) depends in part on pH. Because these metals are more soluble at lower pH values, they may be prefer- entially lost (redissolved into the lake water) from sediment as the pH of the lake water decreases. This loss caused by decreasing pH may be reflected in the sediment record and may provide information on the period of time over which acidification has occurred. However, leaching of sediments by acidified lake water may affect the chemistry of sediments deposited up to a few tens of years prior to acidification. Acidic components (H , SO4 , etc.) can diffuse centimeters into the sediment

17 and desorb or dissolve metal that diffuses out of the sediment. Other materials deposited in sediment that are indicative of fossil fuel combustion include polycyclic aromatic hydrocarbons (PAHs) and components of fly ash. Although these trace substances are not so direct a measure of acid deposition as are the measurement of sulfur and nitrogen oxides, their presence in elevated concentrations in lake sediments is indicative of emissions from the same sources. Diatoms and Chrysophytes in Sediments and pH of Lake Water Diatoms and chrysophytes are single-celled algae that are abundant in most lakes. The relative abundance of specific taxa can be related to the lake water pH because most species are sensitive to the pH of the water in which they live. For a given region (such as the Adirondack Mountains of the northeastern United States), a calibration data set can be established by correlating the current chemistry (especially the pH) of lake waters to the populations of different diatoms species in the surface (recent) sediment. Because diatoms are well preserved in sediments, an analysis of their distribution and abundance in deeper (older) sediments may be used to reconstruct a quantitative estimate of historical trends in pH. Mechanisms of Linkage Between Acid Deposition and Biologic Effects Forests From a wide variety of theoretical and experimental studies, a considerable amount of information exists on the effects that acid deposition and other airborne chemicals might have on trees in the field. Researchers have investigated the possibility of adverse effects on foliar integrity, foliar leaching, root growth, soil properties, microbial activity in the soil, resistance to pests and pathogens, germination of seeds, and establish- ment of seedlings and have shown a variety of positive and negative effects. Effects of acid deposition combined with natural abiotic stresses and with gaseous pollutants such as ozone are also possible. There are many ways in

18 which acid deposition might affect forests, but to date it is not clear from the existing field evidence in North America that any of the mechanisms have caused changes in the forest or in the growth of trees. Thus, although there are many possible pathways that might link acid deposition to changes in tree growth, no proposed linkages between deposition and effects on trees in North America are adequately supported by the mechanistic data currently available. (See Chapter 6.) Recent field studies have shown that red spruce have died in unexpected numbers in high-elevation forests of the Appalachians. Because of the interception of highly acidic cloud waters at these elevations, the deposition of acidic substances and other airborne chemicals is high in comparison with that in the rest of the forests of eastern North America. The lack of any obvious natural cause and the high rates of acid deposition make it logical to examine the timing of this phenomenon to determine the extent to which it corresponds with trends in deposition of acidic substances. It is well known that climatic factors can induce decline diseases of forest trees (complex diseases related to a multiplicity of factors). Examining trends in climate in relation to trends in tree ring widths and mortality of forest trees is critical in the context of this report. Fish Populations One of the most widely publicized of the biotic responses attributed to acid deposition has been the decline of fish populations in certain freshwater lakes and streams in portions of the eastern United States. Declining populations have also been noted in the Scandinavian countries and Canada. Many species of fish cannot tolerate surface waters in which the pH has dropped below about 5, and surface waters attaining this level of acidity, from whatever the cause, could experience fish population losses. Simultaneously, trivalent aluminum ion (All ) can reach toxic concentrations in sufficiently acidified surface waters. Factors unrelated to acidity such as chances in stocking policy, beaver activity, or intensified sport fishing may affect fish populations. Even when fish declines have been directly linked to increasing acidity of the aqueous environment, linking fish loss to acid

19 deposition requires the further step of linking the increasing acidity to acid deposition. Relationships Among Mechanisms The several mechanisms described above provide the basis for exploring further whether various biological and geochemical processes are linked to emissions of SO2 and NOx. Thus, for example, a sustained increase in SO2 and Ned emissions on a wide geographic basis may result in a number of responses, including reduced visibility, increased concentrations of sulfate and nitrate in precipitation and surface waters, and increased deposition in lake sediments of other pollutants asso- ciated with these same emissions sources. A premise of this study is that simultaneous examination of trends and spatial patterns in all the relevant phenomena may help to establish further evidence of linkages despite the complicating factors that may frustrate interpretation of data about a single phenomenon. Complicating Factors The mechanisms described above suggest that certain relationships exist among the data variables. In a perfect world, perhaps we could establish the relation- ships with rigor. In the practical world, however, various factors complicate the analysis, raising the possibility that the relationships may not be discernible in existing data. We discuss some of those complicating factors here. 1. Atmospheric processes that transport, chemically convert, and remove acidic or acidifying substances from the atmosphere are subject to meteorological and climato- logical variability. The distance and direction of the flow of gaseous emissions and their transformation products from their sources depend on meteorological conditions at the time of emission and in the subsequent few days. The conditions can be expected to vary seasonally and interannually. On longer time scales, mild winters with small amounts of snow may lessen the usually sharp rise in acidity observed in lakes and streams during spring snowmelt in areas receiving acid deposition. In addition, long-term climatological

20 changes in patterns of atmospheric circulation may result in changes in patterns of acid deposition. Climatic variations also complicate attempts to assess the effects of acid deposition since ecosystems may be responding to changes in both climate and deposition. Separating an "acid deposition" signal from a "climatic" signal is a major problem in evaluating trends in effects of acid deposition in natural systems. 2. The relative importance to atmospheric deposition of distant and nearby sources is uncertain. Sulfur compounds emitted from a stack may be deposited near the stack, a few tens or hundreds of kilometers away, or even at distances of thousands of kilometers (National Research Council 1983). On the other hand, the amount deposited generally decreases with distance from the source. Thus, variables related to acid deposition at any given location reflect both nearby and distant sources, but in propor- tions that are not precisely known and that may change over time and space. This complication makes it more difficult to establish associations between emissions and variables thought to be related to atmospheric deposition since we do not have a precise way to determine the relative contributions of local and distant sources. 3. The effect observed may not be synchronous with the onset of deposition. For example, the response times of lakes receiving acid deposition may vary widely, depending on characteristics of the lake and watershed and on the rate of deposition (National Research Council 1984). This consideration is important in the evaluation of trends because it suggests, for example, that lakes with different watersheds may respond to changes in deposition on different time scales. 4. Frequently, data are not available for all the relevant variables in a hypothetical relationship. Studies of mass balance of sulfur and nitrogen are vulnerable to this complication, since there are no extensive data on dry deposition (Chapter 7). The flux of sulfate into a lake can be approximated by data on wet-only deposition, but the contribution of dry deposition to the watershed can vary significantly from year to year and place to place. Measurements have shown that dry deposition is an important fraction of total deposition in areas close to point sources. It is suspected that in some ecosystems, particularly in forested areas where the forest canopy may serve as an efficient collector of dry sulfate, net dry sulfate deposition may be equal to or greater than wet-deposited

21 sulfate. Furthermore, depending on the type of soil, sulfate may be adsorbed by the soil or additional sulfate may be added to the soil solution by weathering of ~ Biological processes also play a role in sulfur cycling in watersheds. The lack of data for some of the important Processes affecting sulfur mass internal mineral sources. _ , , , balances complicates the interpretation of results and introduces uncertainties. 5. _ _ _ _ _ Analytical methods have changed over time. To obtain historical data that can be compared with more recent data, it often is necessary to adjust the data for systematic bias resulting in changes in the analytical techniques. In the case of the chemical analysis of lake waters, scientists have applied different assumptions to derive different adjustment factors. Such disparity may cause confusion in interpreting historical trends. SPATIAL PATTERNS AND TEMPORAL TRENDS Regions It was helpful in our analyses to define six specific geographical regions in eastern North America (Figure 1.2). The regions were established as contiguous groups of states and provinces that in our judgment have experi- enced similar temporal trends in SO2 emissions over the past 50 years. Regions A (the Canadian provinces) and F (the western United States) were not included in all the analyses. We have analyzed emissions and environmental responses in each region independently. Thus, our analyses do not specifically account for interregional dependences as these might be affected, for example, by long-range trans- port from one region to another. The approach is valid for our purposes since we are interested in describing trends averaged over broad geographical areas rather than focusing on specific source-receptor relationships. In general, comparisons of trends cannot be used to deduce relative contributions to deposition at one location by a number of sources at diverse, distant locations. The results of the analyses--that trends in environmental indices appear to respond to trends in sulfur emissions on a regional basis--do not imply that long-range trans- port between the regions is necessarily unimportant. Clearly, interregional transport will be important, especially near the boundaries of the regions.

22 V"% . or ~ ~ ~ al Region A (Eastern Canada; Southern Quebec B (Northeastern U.S.): Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont C (Southeastern U.S.): Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia, West Virginia D (Midwestern U.S.): Illinois, Indiana, Michigan, Missouri, Ohio E (North central U.S.): Iowa, Minnesota, Wisconsin F (Western U.S.): All states in the contiguous U.S. not included in Regions B to E I` ~ ~~ ,, r ''---l---- c ;';:: ye Lamp, .. Eastern Ontario, New Brunswick, FIGURE 1.2 Designated regions for analysis of temporal trends and regional patterns.

23 Spatial Patterns Spatial associations among different types of data can be established by testing for similarities in the geo- graphical distribution of density, concentration, flux, or the magnitude of environmental indices. Because historical data are too sparse for developing spatial patterns, we relied on data from the recent past, for which more extensive quality-assured data sets are available. To test for possible spatial associations, we first compared the magnitudes of various parameters on a regional scale. For example, Figure 1.3 shows that the areas of highest SO2 emission density are, in general, areas where the concentrations of sulfate and hydrogen ions in wet deposition are highest. In a similar way, regional contours of range of visibility and concentra- tion of atmospheric sulfate can be compared with emission density patterns. Although such visual presentations may generally suggest spatial associations among phenomena, we also applied Friedman's test (Lehmann 1975) to data from all regions except Region A--eastern Canada. (See Figure 1.2.) This test allowed a more quantitative and objective appraisal of spatial associations among the following variables: emission density of SO2 (1980), concentration of sulfate aerosol (1972), concentration of sulfate in wet-only deposition (1980), flux of sulfate in U.S. Geological Survey Bench-Mark streams (1980), and visibility (lg74-1976). Each variable was ranked from 1 to 5 across the five regions. Rankings were based on the relative magnitude of each variable, expressed as a density, concentration, and flux parameter or, in the case of visibility, as visual range; except for visibility which is assigned a ranking was next rankings ences In ,, _ _ _,, _ _ . an Inverse relationship, a ranking of to the region where the magnitude was greatest, of 4 assigned to the region where the magnitude greatest, etc. The results of the regional are shown in Table 1.1. In cases when differ- variables between regions are too small to be distinguished, equal rankings are assigned to the similar regions (for example, 4.5 for sulfate in aerosol in Regions B and D). The null hypothesis tested in the analysis is that there are no relationships among the variables. The p value obtained with this array of rankings is O.OO9, which implies that if there were in fact no true 5 was

24 association among the variables, an apparent association to this degree or greater would occur by chance only 0.9 percent of the time. The result indicates of a strong spatial association among SO2 emissions, atmospheric sulfate aerosol, sulfate in precipitation, sulfate in Bench-Mark streams, and visibility, corroborating conclusions reached by visual comparisons of the relationships displayed in contoured maps. On the basis of the rankings of the five variables given in Table 1.1, it appears that, overall, Region D experiences the highest amounts of SO2 emissions and related effects, Region F experiences the lowest amounts, and Region E experiences the next-to-lowest amounts. Regions B and C are generally more affected than Region E but less affected than Region D. Nazi / (a) ~3 \_~!, ,` V- -2 -1 I g m yr FIGURE 1.3 Comparison of spatial patterns between (a) SO2 emission density (g m~2yr~l); (b) concentration of H+ (pmoles/L) in wet deposition; (c) concentration of SO42- (pmoles/L) in wet deposition; (d) concentration of sulfate aerosol (pg/m3); and (e) visible range (in miles).

25 (b) 1,,,.~; .'0 .' . .. . , , :_',-\ . . . '' )'~' (c) ,~ IN · WIb ~ - .' 1, · \2. i, .,, : 10. , , . ~ · _ -,, <5l,, 'I-\ ~o,, 1 FIGURE 1.3 (continued) .

26 (d) c: ( I '- 1 '°° ;W _~ 3 _— > 11 ,ug/m cam , ~ 41~ ~0 P: Based on pi 36 4)/: I/ photometry data ~ ~ ~ Ail 13~ \\ 14: Based on nephelometry data 457, ~ ~ \ 9 cat 1U ~' - _ \ \ *: Based on uncertain extrapolation of ~4 TV ~$ ~ \ \ visibility frequency distribution 1~< ~~ s t\~ 25~; 1\ ~ 1 15 FIGURE 1.3 (continued). In a different type of analysis, we examined the spatial association between sulfate in wet precipitation and sulfate in 626 lakes in the northeastern United States and southeastern Canada (Chapter 7). We tested the assumption that sulfate in wet deposition was the

27 TABLE 1.1 Regional Rankings by Magnitude of SO2 Emissions Densities and Related Variables so2 Region Emissions Sulfate Sulfate Sulfate Aerosola Visibilityb Precipitation Streams B C D 5 E 2 F 1 3 4.5 2 4 4 4.5 5 2 3 3 s 2 MOTE: p value of Fr~edman's test, 0.009. aThe difference in concentration of sulfate aerosol in Regions B and D is too small to be distinguished. In order to avoid possible bias toward maximum association, we assigned a 5 to Region B and a 4 to Region D, because this ranking (i.e., 5, 3, 4, 2, 1) shows a lesser association than would the alternative ranking (i.e., 4, 3, 5, 2, 11. b Rankings for visibility are inversely related; i.e., the greater the value, the worse the . . . . VlS1 31 1~. Only source of sulfate to the lake. Sulfate output from the lake was calculated from the product of the sulfate concentration in the lake and the net precipitation (i.e., precipitation minus evaporation). These simplifying assumptions were necessary because additional data required for a more complete analysis of sulfate mass balance, such as inputs through dry deposition or chemical weathering of sulfate minerals in the watersheds, were not available. The lakes were grouped into six subregions: Connecticut-Massachusetts-Rhode Island (CT-MA-RI), Maine-Vermont-New Hampshire (ME-VT-NH), New York (NY), Quebec (QU), Labrador (LB), and Newfoundland (NF). Data for each of the subregions were plotted on a log-log scale (Figure 1.4). The figure shows each of the regression lines having equal slopes of about one, but with differ- ent intercepts, ml'. On the natural scale, these data correspond to a linear model, y = mix, where y (output flux of sulfate from lakes) is proportional to x (flux of sulfate into lakes from wet deposition) by the subregional multiplying factor mi (the antilogarithm of mi'). The values of mi for the six subregions are given in Table 1.2. Regions where lakes demonstrate the greatest excess of sulfate output (mi > 1), suggesting that wet deposition accounts for only part of the sulfate input, are those nearest to large emission sources (CT-MA-RI, NY, ME-NH-VT); those with values of mi near or less than

28 4 _ 3 _ x 2— o 1 _ O _— 4.5 0.0 _ ~ 4 _ .- _ X 2— o O ---I . ~ ~ . 1 _ ~ —~ t' ~.5 0.0 o CT-MA-A I m' = 0.691 /1 ~ ; ~ ' log x axis 0.5 1.0 1.5 NY mj = 0.280 ~ ~ . log x axis L ~ 0.5 1.0 1.5 ~ ~ it. ~ O ,:, -0.5 0.0 0.5 1.0 o .'~ m',=~.515 LB LOG SULFATE INPUT (9 m~2 yr~1 ) 4 _ 3 - x 2 _ o 1 _ O ~ -1 _ -2 _ 3 i I -0.5 0.0 0.5 1.0 ME-VT-NH mj = 0.257 . .' . log x axis 1 1 1 3 2 o -1 _0.5 0.0 -1 _ -2 I -3 _ 1.5 -0.5 QU mj = 0.098 . log x axis 0.5 1.0 1.5 ·.~ . \ a-' m j = -0.004 NF loo x axis 1 1 1 0.0 0.5 1.0 1.5 LOG SULFATE INPUT (9 m~2 yr~1 ~ FIGURE 1.4 Regional relationships between the log of sulfate flux into lakes from wet deposition and the log of sulfate flux out of lakes from outflow. mi' denotes the log y axis intercept. 1.0 are more distant (QU, NF, LB). This pattern of association is consistent with the hypothesis that dry deposition is a major contributor to the sulfate mass balance near emission sources. The result is also consistent with hypotheses linking emissions, sulfate in deposition, and sulfate in surface waters.

29 TABLE 1.2 Regional Multiplying Factors, mi Region mi Standard Error CT-MA-RI 2.00 0.16 ME-VT-NH 1.29 0.14 NY 1.32 0.09 QU 1.10 0.06 LB 0.60 0.02 NF 1.00 0.09 NOTE: mi is the antilogarithm of mj'. See Figure 1.4. Data on fish population records and tree ring widths were not available for conducting detailed spatial analyses. However, declines in fish populations in selected lakes and streams have been documented in watersheds extending from Pennsylvania to Nova Scotia. Surface waters in these areas tend to have low alkalinity and receive relatively high inputs of atmospheric deposition of sulfate and nitrate. Although data are sparse, well-documented fish declines have occurred in a few locations in lakes of the Adirondack Mountains and in rivers of Nova Scotia, waters that have experienced increases in acidity thought to be caused by acid deposition. Although the mechanisms of cause and effect between acid deposition and forest decline have not been established, the areas experiencing the most noticeable declines of red spruce--the high-elevation areas of New York, Vermont, and New Hampshire--receive large amounts of acid deposition as well as other pollutants. Only limited data exist to evaluate spatial patterns relating nitrogen oxide emissions and nitrate in the environment. Both stationary and mobile sources emit NOx, which is formed mainly by the oxidation of nitrogen in the air during combustion. The formation of NOx depends strongly on conditions in the combustion zone, such as temperature. These uncertainties limited our ability to estimate NOx emissions with a high degree of confidence. As a result, we can mare only the general statement that areas of the country with the highest NOx emissions are generally those in which nitrate concentra- tions in precipitation are greatest. In summary, statistical tests of spatially differen- tiated data demonstrate regional associations among sulfur dioxide emissions, sulfates in aerosol, visibility, sul- fates in precipitation, and sulfate fluxes in Bench-Mark

30 stream waters. Spatial analysis of sulfate mass balance in 626 lakes further strengthens the postulated relation- ship between sulfate deposition and surface water concen- trations of sulfate. The analysis is also consistent with the ideas that dry deposition of sulfate decreases with distance from regions of highest sulfur emissions and that in some cases sources of sulfur within the watershed also contribute to the sulfate input of lakes. These findings do not resolve questions about specific source-receptor relationships. However, they demonstrate strong regional association in spatial patterns among density of SO2 emissions, atmospheric concentrations of sulfate, range of visibility, concentration of sulfate in wet precipitation, and fluxes of sulfate in Bench-Mark streams. Temporal Trends The Record Since Industrialization (about 1880) Only a few pieces of evidence exist for estimating the acidity of atmospheric deposition of 100 years ago. Hidy et al. (1984) have reviewed much of the early literature on the chemical composition of precipitation in the United States. Records of sulfate concentrations are available from a few locations over periods ranging from about 1910 to 1950. These data, however, are not of sufficient quality to permit a definitive determination of long-term trends or comparisons with more recent data. (See Chapter 5.) Hence we examined a larger pool of historical data that may be associated with acid deposition to determine whether a consistent and clearer picture of long-term trends emerges. Examples of the data examined are shown in Figure 1.5 and include (a) estimates of coal consumption in the eastern United States (Regions B. C, D, and E) and in Region B alone (Chapter 2); (b) annual emissions of SO2 for the eastern United States and for Region B alone (Chapter 2); (c) light extinction (an inverse measure of visibility) at the Blue Hill Observatory in Massachusetts (Chapter 4); (d) lead, vanadium, and PAHs in sediments in Big Choose Lake in New York (Chapter 9); (e) diatom-inferred pH (Chapter 9) and fish populations for the same lake (Chapter 8); (f) actual and estimated annual tree ring widths (expressed as indices) for red spruce in the Adirondack Mountains (Chapter 6); and (g)

31 600 500 400 CC i: he o 14 10 8 6 4 2 o (a) Eastern U.S. Coal Consumption Regions (B. C, D, E) ,~ \ V Coal Consumption for Region B / - o 1880 1 900 1920 1940 1960 1980 YEAR Eastern U.S. SO2 Emissions ~ / Regions | ~' (B. C, D, E) | ~1 1880 1900 1920 1940 1960 1980 (b) SO2 Emissions Region B - YEAR FIGURE 1.5 100-year trends in indices that may be related to acid deposition .

32 .1 6 ~ Blue H ill, Massachusetts 'A .14 _ by _ _ o ~ .12 _ At X _ us ~ .10 1 I ~ |JLong Term Record j \J of Light Extinction of 1 1 1 1 (c) 1 1 1 1 1 1940 1960 1980 1900 1 920 6 — 300 5 - 250 _ 4 3 2 — 200 — '> 150 · D I — 100 50 o FIGURE 1 . 5 (cons inued ) YEAR Big Moose Lake Sediment Conc. Al / \ / / Pbf j ,~~_,', . _ .'' lv i _~'j O -_ 1 ~ 1880 1900 1920 (d) PA | ! f il< , i . 1.4 !1 - l l / . _ 1.2 1.0 N _ 1F C7) L/1 - 0.8 > 0.6 1 1 1 1 1 1 1 1 _ 0.4 1940 1960 1980 YEAR .

33 (e) r' (f) 5.0 4.5 1 .4 X UJ Z 1.2 A CC ink' 1.0 0.8 0.6 (9) 29 27 25 23 F 21 19 17 15 13 Big Moose Lake Adirondacks 6.0 ~ Reconstructed pH / smallmouth bass lost lake whitefish lost —longnose sucker lost ~ lake trout lost 1 1 1 1 1 1 1 780 V 1 880 1 900 1 920 6.0 1 1 1 1 1 4.5 1 940 1 960 1 980 YEAR Red Spruce at Lake Arnold Ad irondacks telex. 1 150 m ) 1 ,~ I 11 1 11 6: rat :1 1~1 ~ ~ 1,', ~ 1900 920 1 940 YEAR Actual ——— Estimated J 1960 1980 _ NEW ENGLAND AVG DEC-FEB TEMPS 11888-89 - 1976-77) 932.33 . 1 976-77 — 1 904-05 _ 1 933-34 _ _ I I i I 1191718 ~ ~ ~ 1888 1896 1 1904 1912 1920 1928 1936 1944 1952 1960 1968 1976 1900 FIGURE 1 .5 (continued ) YEAR .

34 the 100-year record of winter temperature in New England (Chapter 6). Figure l.5(a) demonstrates that the most rapid rate of increase in coal consumption in the eastern United States (Regions B through E) and the northeastern United States (Region B) occurred between about 1880 and 1920. Since 1920, coal consumption in the eastern United States has fluctuated between 300 and 500 million tons, depending on social, political, and economic factors, but in general has shown no substantial increase since the 1920s. How- ever, coal consumption in Region ~ has decreased since 1920. The dramatic rate of increase in consumption around the turn of the twentieth century can be attributed to the increased use of coal to fuel a rapidly expanding industrial economy, including rail transportation. It is reasonable to assume that the decrease in consumption in the 1930s was a reflection of the widespread depression in economic activity, the increase in the early 1940s was associated with World War II, and the decline in the postwar years accompanied the return to a peacetime economy and the increased use of petroleum and natural gas. The increase during the 1960s is attributable to the expansion of coal use in the generation of elec- tricity, and moderating consumption in the 1970s is the result of several economic, political, and social factors that resulted in a reduced coal demand. Emissions of SO2 are estimated from data on the use of fossil fuels, of which coal is the most important contributor. Thus SO2 emissions and coal consumption show similar trends (Figures 1.5(a,b)). The difference in trends, especially after about 1950, is mainly attributable to emissions of SO2 from oil consumption, a source accounting for about 10 percent of national energy consumption in 1920 and for about 50 percent in 1980. Although there is little supporting evidence to verify the estimates of SO2 emissions of a century ago, the evidence that is available is in concert with the general pattern. A rapid rate of increase in light extinction (corresponding to a decrease in visibility) was recorded at the observatory at Blue Hill, Massachusetts, at about the same time period as the most rapid expansion of coal use (Figure 1.5(c)). Since the second decade of the twentieth century, the trend in light extinction at this s ite is somewhat similar to the trends of coal consump- tion and SO2 emissions for Region B. .

35 Multiple lines of chemical evidence from lake sediments of Big Moose Lake in New York indicate a history of changing chemistry paralleling the pattern of changing fuel use over the past 100 years (Figure 1.5(d)). Sediment constituents related to emissions from coal burning (including fly ash, PAHs, certain trace metals, and excess sulfur) dramatically increased in concentration through the first half of the twentieth century. Trace metals associated with combustion of fuel oil (vanadium) and gasoline (lead) increased relatively slowly over the first half of the century and rapidly thereafter, reflecting large-scale changes in fuel use patterns around mid-century. The only data available for reconstructing the historical record of lake-water pH on a 100-year time scale come from measurements of the distributions of diatom and chrysophte taxa present in lake sediments (Figure 1.5(e), for example). The record suggests that after remaining nearly constant over the previous two centuries, the pH of Big Moose Lake decreased rapidly beginning about 30 years ago. Corresponding data on populations of acid-sensitive fish support this conclusion. Figure 1.5(f) shows the trend for values of the tree ring indices for red spruce at Lake Arnold in the Adirondack Mountains (elevation, 1150 m). The dashed line indicates values calculated using a model based on responses to average monthly temperatures. (See Chapter 6.) The model fits the data reasonably well until about 1967, after which index values are considerably less than predicted. This result suggests that although monthly temperature was a major factor governing tree ring width before about 1967, other factors were important after that time. It is possible that a temperature-related factor initiated the decline. Ring width in red spruce is affected by winter temperatures, and the onset of the decline at Lake Arnold and elsewhere coincided with the occurrence of cold winters in the northeastern United States, as demonstrated in the temperature record in New England in Figure 1.5(g). The unusual response in the case of red spruce has been a failure to recover as winter temperatures became more moderate. The cause of the prolonged nature of the tree-ring-width decline and extensive mortality is not clearly understood.

36 The Half-Century Record Lake-Water Chemistry Beginning in the 1920s and 1930s, extensive surveys of the chemistry of lake waters were undertaken in several states. Most of the data were used to monitor water quality of lakes in rural areas used for fishing or other recreational activities. Unfortunately, the surveys were conducted for only a few years, so that there are no reliable long-term continuous records of this type. Coupled with recent data, however, the historical surveys allow us to compare water quality in these lakes at two points in time separated by about 50 years. The parameters of lake-water quality that are most important for acidification are pH and alkalinity. The value of pH gives the concentration of hydrogen ion but does not provide information about the inputs of hydrogen ion that may have been neutralized by basic substances. Thus, it is possible for a highly buffered lake to receive a substantial input of acidity but show little, if any, change in pH. The response of a lake to acid deposition is likely to be more apparent as a change in alkalinity. We reviewed and analyzed available historical data on the alkalinity and pH of lakes in New Hampshire (1930s and 1940-1950), New York (1929 to 1936), and Wisconsin (1925 to 1932) and compared them with recent data (late 1970s and early 1980s). The results of the comparison, detailed in Chapter 7, suggest that some lakes have experienced decreases in alkalinity and pH while others currently have amounts of alkalinity and values of pH higher than they were 50 to 60 years ago. In Wisconsin, more lakes experienced increases in pH and alkalinity than have experienced decreased values. In New Hampshire, the median change in alkalinity is not statistically different from zero, but the change in median pH suggests that more lakes have increased in pH than have decreased. Any interpretation of changes in alkalinity and pH in New York lakes based only on water chemistry data is somewhat equivocal because the calculations are sensitive to the value assumed for the pH of the endpoint of the methyl orange (MO) indicator, a standard chemical used in the determination of historical lake alkalinities. Evidence is substantial in the well-documented Wisconsin survey that the endpoint using standard methods was pH 4.19. Similar documentation does not exist for the New York survey, but a review of the historical literature

37 suggests that the endpoint was bounded in the pH range of 4.19 and 4.04. We compared the 1930s data with two recent New York lake surveys (Pfeiffer and Festa 1980, Colquhoun et al. 1984) and determined that the results depended somewhat on which of the two surveys were included in the calcu- lations. Thus, there are four possible pairings of assumptions as follows: (1) an MO endpoint of 4.04 pH units and comparison of historical data with the 1980 data; (2) an MO endpoint of 4.04 and comparison with the 1984 data; (3) an MO endpoint of 4.19 units and comparison with the 1980 data; and (4) an MO endpoint of 4.19 units and comparison with the 1984 data. Calculations applying assumptions 3 and 4 yield median changes in alkalinity of -69 and -44 peq/L, and median changes in pH of -0.74 and -0.63, respectively. Applying assumptions 1 and 2, the calculations yield median changes in alkalinity of -28 and +1 peq/L, and median changes in pH of -0.14 and -0.12 units, respectively. In each state there are numerous examples of lakes with changes in alkalinity greater in magnitude than 100 peq/L. Since the magnitude of change from acid deposition is estimated at about 100 peq/L or less (Galloway 1984), it is likely that these lakes were affected by factors other than, or in addition to, acid deposition. Relationship of Trends in Diatom-Inferred pH and Fish Populations Analysis of diatoms in sediments of selected lakes offers another indication of long-term acidification. Unlike the lake surveys in Wisconsin, New York, and New Hampshire, the lakes from which diatom data were obtained were selected on the basis of acid sensitivity (i.e., alkalinity less than 200 peq/L) and either little or no disturbance of the watershed or good documentation of disturbance. (See Chapter 9.) Based on paleoecological analysis of the entire history of currently acidic lakes, rates of long-term natural acidification are slow--with decreases of 1 pH unit (from 6.0 to 5.0) occurring over periods of hundreds to thousands of years. In contrast, some lakes in the Adirondack Mountains have apparently experienced decreases in pH on the order of 0.5 to 1.0 pH unit over a 20- to 40-year period in the middle of this century. Diatom data for 31 lakes were evaluated to assess regional trends in lake acidification; data of sufficient

38 quality are available for 11 lakes in the Adirondacks, 10 in New England, 6 in eastern Canada, and a reference set of 4 lakes in the Rocky Mountains. The evaluation suggests that certain poorly buffered eastern lakes have become substantially more acidic during the past 20 to 40 years. The lakes for which evidence is strongest are in the Adirondack Mountains. Of the 11 lakes for which diatom data are available, 6 of the 7 lakes with a current pH at or below about 5.2 show evidence of recent acidification; the seventh is a bog lake. Analyses of chrysophyte scales (mallomonadaceae) agree well with interpretation of the diatom data. None of the 4 lakes with current values of pH above about 5.2 showed strong evidence of a pH decline. Where dating of sediments is available, the most rapid diatom-inferred pH changes (decreases of 0.4 to 1.0 pH units) occurred between 1930 and 1970 beginning in the 1930s to 1950s. Diatom data for the 10 lakes in New England indicate either a slight decrease or no change in pH over the past century. Diatom data for the 6 lakes in eastern Canada indicate no change in pH (4 lakes with current pH greater than 6.0) or a significant decrease in pH that is probably caused by local smelting operations (2 lakes). Rocky Mountain data do not show decreases in pH. Before 1800, several lakes in the Adirondacks and New England had diatom-inferred pH values less than 5.5. These lakes now have a pH of only 0.1 to 0.3 pH units lower, and total aluminum concentrations greater than 100 ug/L. Because of the potential importance of buffering by organic acids, a small decline in pH could be asso- ciated with significant decreases in acid neutralizing capacity. Analysis of the Adirondack data indicates that no other acidifying process except acid deposition has been identified to explain the rapid declines in lake water pH during the past 20 to 40 years. However, watershed disturbances may also play a role, but probably a minor one for the lakes evaluated in this study. Further evidence of acidification trends in lakes of the Adirondacks is provided by comparing measured pH, diatom-inferred pH, sediment chemistry, and fish population data on nine Adirondack Lakes (Table 1.3). They show consistent trends with some exceptions. The lakes with current values of pH of about 5.2 or less have become more acidic in recent times and have lost fish populations, whereas lakes with higher current values of

39 pH show no obvious trend toward acidification or fish declines. In a number of cases (e.g., Woods Lake, Upper Wallface Pond) the observed change in diatom-inferred pa is small, from about pH 5.2 to about pH 4.8, yet major changes in fish populations have occurred. There are two plausible explanations for this phenomenon. First, fish are sensitive to pH in this range, with survival decreasing abruptly over the pH range 5.2 to 4.7. Many lakes in New York and New England with pH 5.0 to 5.2 currently support fish, whereas lakes in this region with pH below 5.0 rarely support fish (Chapter 8). Second, acidification of surface waters from pH 5.2 to 4.7 may be accompanied by a decrease in dissolved organic carbon (Davis et al. 1985), and an increase in dissolved uncompleted aluminum, which is highly toxic to fish. In summary, we analyzed historical and recent data on pH and alkalinity from three large lake surveys in Wisconsin, New York, and New Hampshire. In all cases, some lakes decreased in alkalinity and pH since the 1930s and some increased. In some lakes the magnitude of the change in alkalinity appears to be too large to be explained solely by acid deposition. The evidence further indicates that on average Wisconsin lakes have increased in pH and alkalinity since about 1930, New Hampshire lakes show no obvious change in alkalinity but may have increased in pH, and New York lakes have shown decreases in alkalinity and pH in three of four possible combinations of assumptions, and show no change if one accepts the fourth assumption. Data on diatom-inferred pH for 11 lakes in the Adirondack Mountains (one of them was a bog lake and was disregarded) indicate that 6 of these lakes have become more acidic over the period from about 1930 to the 1970s. All these lakes have current pH values of about 5.2 or lower. The 4 lakes with current pH values above 5.2 show no trend in pH. For 9 of the lakes, concurrent data exist on measured pH, sediment chemistry, and fish populations. The data are generally consistent and support the findings based on diatom analysis. Data for New England indicate slight or no increase in diatom- inferred pH. There is insufficient information to evaluate trends in eastern Canada. The lakes selected in the diatom studies have low alkalinities (less than 200 peq/L) and little disturbance of their watersheds, and hence may be the most likely to show responses to acid deposition.

40 a: . - z me In o .— 'e no a, .~ - ct em . · v, - ~ o hi ·fi ~ so ~ >,o, Cal 'A 3 By o ~ Ct ._ Ct ~ no ~ o ,= m ~ sit Ct ~ no o — i= .= m Ed o - - C: - Ct So Cal S ~ ~ .' . _ ~ V) ~ _% ~ 8 - ° C.) 't Or: = e ~ hi: Ct a ' e ~ e ~ .= ~ ~ o := O e c)— ~ ~ c) ~ ~ ~ ~ ct ~ ~ c) D = ~c ~ ~ O D . ~ ~ ~ = . = = ~ =0 ~ D C' . ~ °O ' ~ ~ ~ ~ — ~ oo . _ r,=~-U, ~ L,2},~ ~,i=,= 82 ~ ' ~ ~ ~ · E , _~ . ~ ' ~ c , ~ , , ~ ~ — ~ ~ ~ - e ~ y ~ =' U ~ _ =,D [L ~1 z) ~ :O ~1 LL CL ~ ~ ~ 1 CL ~ LL LL _1 _) _1 _1 c) ~ O oo ~ 0 ~ 0 ~ 0 u~ O 00 ~ ~ ~ ~ 0 c~ ~ u~ a~ ~ ~ ~ ~ ~ c~ .. ........ ' ' U~ ~ ~ ~ ~—V~ ~ V) ~ ~ U~ U~ ~ ~ U~ O O ~ ~ ~ ~ ~ ~ ~ ~ _ ~ oo o ~ _ ~ ~ _ C-l ~ oo °O ~ ~ r~ oo oo ~ ~ ~ 1— t— oo oo oo ~ \= [— 1— ~ C~ ~ t_ ~ ~ cr~ ~ a~ ~ ~ c~ a~ c~ ~ ~ ~ l—C~ ~ a~ cn _ _ _ ~ _ 0 C ~ ° ~ =~}

41 'e ~ t D O ~ ace c,, .0 ~ ,~ ° fi ~ D O <~ O ~ ~ ~ ~ ~ ~ ~ ~ ~ O ~ ~ O ° '4~ =. ° Ce ~ ~ ~ ells in= ~ == == ~ ~ 0 <~> e A 3 . 5 ~ ~ ~ - ~ . ~ - 3, _ - ~ · < i ~ s ~ ~ ~ ~ , , ~ _ ~ j =~= -<=,3;5='~ =9~= -!; =~ ~ o of 0 0 as, at - , ~ ~ 0 0 ~ of ~ ~ ° - ~ ~ ~ 0 us ~ ~ ~ ~ ~ A, I, lo, lo, ~ lo, A- d- ~ Cat o o oo ~ ~ ~ ~D—~ oo ~ u~ ~ °O - 1 ~ oo oo a~ ~ ~o oo oo ~ ~ ~ ~ oo °O ~ ~D ~ r~ ~ ~ oo a~ ~ ~ ~ a~ ~ ~ [~ ~ a~ c~ ~~a~~~~~ ~~~~<5~~ ~~~~ ~~~ _ _ C~ 0 t4 C~ V, C) ~ 0 C.) ~ O o ~ _ ~ :~; ~ ~ .~ ~ ~ Ce .^ oo ° C~ c' ~ 3 ) Z = —~— ^ V ~ ~ ~ ·o s~ oo - ^ ~ —~ o o~ o ._ ~ .= a~ C) 0= C) O .^ == ~ O — 3 ~ -o ~ ._ C~ ~ .c~ tn ~ V: C~ o 6 5 3 ~= 0- ,0 ~ ;;> ~ ~ =~ _ ~ s~ C) C) ~ ~ =-= ~ ,~ Ct C., D . ~ Z~ ·- =- ~ 3 '5 C., O^ Z ~ o ° ·~= ~ · s _ ~ o D~ D —-- ~s C _1 ~ _ ~ 5

42 The Record Since 1950 General Trends The decades after World War II are characterized by rapidly changing patterns of fuel consumption and fuel use in eastern North America. Before 1945 coal was the dominant source of fuel and consumption was divided among railroads, residential and commercial heating, oven coke and other industrial processes, and electric utilities. The demand for rail transport was particularly high during the war years of the early 1940s. By the end of the 1950s, however, coal consumption by railroads and by the residential-commercial sector essentially vanished. Overall, coal use declined by about 30 percent from 1945 to 1960. However, coal for electric power more than doubled over this same period, and doubled again for the period from 1960 to 1975. Concurrent with the expansion of coal use for electricity was the construction of new power plants with increasingly higher smokestacks, resulting in more than a doubling in average stack height from the mid-1950s to the mid-1970s (Sloane 1983). The seasonal pattern of consumption also changed during this period. In the early 1950s coal consumption in winter, the peak season, was about 27 percent higher than in summer, the season of lowest consumption. By the mid-1970s both winter and summer were peak periods of comparable consumption. Total coal consumption in the eastern United States is currently comparable to consumption during the peak years of the early 1940s, owing to increased consumption after the decline in the 1950s. However, there has been a con- siderable shift in the regional patterns of consumption over the past two decades; some areas currently consume far greater amounts of coal than they did in the early 1940s and some areas consume far less. (See discussion of regional trends below.) Coal provided about 50 percent of the energy needs of the United States in 1945. Currently, it provides about 20 percent as the total energy consumption has more than doubled over this period. The increasing demand for energy has been met largely by natural gas and oil. Natural gas contains little or no sulfur, and oil, after refinement, is of relatively low sulfur content. Thus, SO2 emissions have not increased in proportion to energy consumption over the past four decades. Nitrogen oxides, however, are formed as a by-product of any high-temperature combustion process in the atmosphere, regardless of the cleanness of the fuel. Consequently,

43 nitrogen oxide emissions in eastern North America have more than doubled from the period 1945 to 1980, and have become an increasingly important component of acid deposition. The changes in fuel consumption, fuel use, and fuel type that have occurred over the past four decades have undoubtedly affected the geographic distribution and the composition of acid precursors in the atmosphere in subtle or more obvious ways. Environmental effects (e.g., lake acidification) have occurred in sensitive ecosystems over this same time period. However, if acid deposition is a cause, it may be difficult to determine whether the effect is a consequence of relatively recent changes in fuel use or consumption since 1945, or whether it is a result of cumulative exposure to acid deposition over many decades. Regional Trends Estimates of SO2 emissions in eastern North America suggest that the decade of the 1950s was a period of constant or declining emissions in all of the designated Regions A through E. In contrast, during the 1960s SO2 emissions rose sharply in all regions. The 1970s are characterized by strong regional differences in trends of SO2 emissions. There were differences not only in the magnitude of trends but also (for the first time) in their direction. In the northeastern states (Region B) the trend was distinctly downward. In the southeastern states (Region C) the trend was rapidly upward, continuing the trend that began in the 1960s. the midwestern states (Region D) the trend was upward, but not so rapidly as in Region C. Emissions in the north central states (Region E) remain consistently low. The divergence of trends in SO2 emissions along regional lines in the 1970s provides the opportunity of testing whether other types of data also reveal consistent regional differences. We analyzed two different types of data that have continuous records for a number of years and were collected at numerous sites in different regions. One is the record of light extinction (an inverse measure of visibility) at 35 airports from 1949 to 1983 (Chapter 4). The other is the record of sulfate in Bench-Mark streams from 1965 to 1983 (Chapter 7). We have also analyzed the record of pH, sulfate, nitrate, hydrogen ion, and other ions in bulk precipitation at the Hubbard Brook Experimental Forest, New Hampshire, for the period from 1963 to 1979 (Chapter 5), but we do not include

44 these data in our regional analysis because they represent only one site in one region. However, we note that the observed trend of decreasing sulfate at Hubbard Brook is consistent with our observation of decreasing emissions of SO2 in Region B over this time period. The mechanisms outlined earlier suggest that SO2 emissions, atmospheric sulfate, light extinction, and stream sulfate are related and may exhibit similar temporal trends. We examined this suggestion on a regional basis by testing for associations among the regional trends of these variables by using Friedman's test. For the period l9SO to 1980, only data on SO2 emissions and light extinction are available; the regional rankings for these trends are given in Table 1.4. A ranking of 1 in Region B for the trends of both SC2 emissions and light extinction signifies that this region experienced the lowest rate of increase in each of these parameters over the period from 1950 to 1980. Higher rankings signify greater rates of increase. The p value for Friedman's test in this case is 0.042, signify- ing that if there were in fact no true association between the parameters, then an apparent association to this degree or greater would occur by chance only 4.2 percent of the time. The result is indicative of a strong temporal association between SO2 emissions and light extinction on a regional basis. We applied the same method to test for possible associations among trends in light extinction, sulfate fluxes from Bench-Mark streams, and emissions of SO2 for the period from 1965 to 1980. The regional rankings are shown in Table 1.5. The p value of 0.054 again gives evidence of temporal associations for these data on a regional level. TABLE 1.4 Regional Rankings by Rate of Change in SO2 Emission Densities and Light Extinction for the Period 1950-1980 Region SO2 Emissions Light Extinction - B C D E 4 2 4 2 NOTE: p value of Friedman's test, 0.042.

45 TABLE 1.5 Regional Rankings by Rate of Change of SO2 Emission Densities, Light Extinction, and Stream Sulfate for the Period 1965-1980 Region SO2 Emissions Light Extinction Stream Sulfate B C D E 4 2 2 4 3 2 NOTE: p value of Friedman's test, 0.054. As demonstrated in Chapter 7, changes in the flux of sulfate in soft-water Bench-Mark streams were balanced by changes in alkalinity and base cations. The regional pattern of trends in stream alkalinity for the period 1965 to 1983 was the approximate inverse of that of stream sulfate (see Chapter 7, Figures 7.6 and 7.7) : decreases (or no increases) have occurred at several stations in Region C and Region F while increases (and no decreases) have occurred at stations in Region B. Station-by-station comparisons of alkalinity and sulfate trends, however, do not always show a consistent inverse relationship. Beginning in the 1960s, ring widths of red spruce at high elevations throughout its range in the eastern United States decreased significantly, a change that has persisted to the present. Important regional climatic anomalies occurred when the red spruce decline began and may have been a factor in triggering the response. There is currently no direct evidence linking acid deposition to mortality and decreases in ring width, although this effect has occurred in areas that are receiving relatively large amounts of acidic substances and other types of air pollutants. In summary, based on statistical tests of regional temporal trends, a strong association exists between SC2 emissions and light extinction (30-year records, 1950 to 1980). A similar result is obtained for SO2 emissions, visibility, and sulfate concentrations in Bench-Mark streams (15-year records, 1965 to 1980). Since 1950, the northeastern United States (Region B) nas experienced the smallest rates of change in these parameters. The southeastern United States (Region C) has, in general, experienced the greatest rates of change, and the Midwest (Region D) has experienced rates of change greater than Region B but less than Region C.

46 REFERENCES Colquhoun, J., W. Kretzer, and M. Pfeiffer. 1984. Acidity status update of lakes and streams in New York State. New York Department of Environmental Conservation, report no. WM P-83 (6/84). Albany, New York. Davis, R., D. Anderson, and F. Berge. 1985. Loss of organic matter, a fundamental process in lake acidification: paleolimnological evidence. Nature 316:436-438. Del Prete, A., and C. Schofield. 1981. The utility of diatom analysis of lake sediments for evaluating precipitation effects on dilute lakes. Arch. Hydrobiol. 91:332-340. Driscoll, C. 1980. Chemical characteristics of some dilute acidified lakes and streams in the Adirondack region of New York state. Doctoral dissertation, Cornell University, Ithaca, N.Y. Driscoll, C., and R. Newton. 1985. Chemical character- istics of Adirondack lakes. Environ. Sci. Technol. 19:1018-1023. Galloway, J. N. 1984. Long-term acidification. Pp. 4-48--4-53 in The Acidic Deposition Phenomenon and Its Effects. Critical Assessment Review Papers, Volume II, Effects Sciences. A. P. Altshuller and R. A. Linthurst, eds. Environmental Protection Agency report no. EPA-600/8-83-016BF. Galloway, J. N., C. Schofield, N. Peters, G. Hendrey, and E. Altwicker. 1983. Effects of atmospheric sulfur on the composition of three Adirondack lakes. Can. J. Fish. Aquat. Sci. 40:799-806. Harper, H. J. 1942. Sulfur content of Oklahoma rainfall. Okla. Acad. Sci. Proc. 23:73-82. Hidy, G. M., D. A. Hansen, R. C. Henry, K. Ganeson, and J. Collins. 1984. Trends in historical and precursor emissions and their airborne and precipitation products. J. Air Pollut. Control Assoc. 31:333-354. Lehmann, E. L. 1975. Nonparametrics. San Francisco, Calif.: Holden-Day. MacIntire, W. H., and J. B. Young. 1923. Sulfur, calcium, magnesium, and potassium content and reaction of rainfall at different points in Tennessee. Soil Sci. 15:205-227. Mosteller, F., and J. W. Tukey. 1977. Data Analysis and Regression: A Second Course in Statistics. Reading, Mass.: Addison-Wesley.

47 National Research Council. 1983. Acid Deposition: Atmospheric Processes in Eastern North America. Washington, D.C.: National Academy Press. National Research Council. 1984. Acid Deposition: Processes of Lake Acidification. Washington, D.C.: National Academy Press. Pfeiffer, M. H., and P. J. Festal 1980. Acidity status of lakes in the Adirondack region of New York in relationship to fish sources. New York Department of Environmental Conservation, Albany, N.Y. FW-P168(10/80). 36 pp. Robinson, E. 1984. Natural emission sources. Pp. 2-1--2-52 in The Acidic Deposition Phenomenon and Its Effects. Critical Assessment Review Papers, Volume I, Atmospheric Sciences. A. P. Altshuller and R. A. Linthurst, eds. Environmental Protection Agency report no. EPA-600/8-83-016BF. Schofield, C. L. 1965. Water quality in relation to survival of brook trout, Salveilnus fontinalis (Mitchell). Trans. Am. Fish. Soc. 94:227-235. Seip, H. M. 1980. Acidification of freshwater--sources and mechanisms. Pp. 358-366 of Ecological Impacts of Acid Precipitation, D. Drablos and A. Tollan, eds. Proceedings of an international congress in Sandefjord, Norway. Sur Nedbors Virking Pa Skog Og Fisk (SNSF) Project, Oslo. Sloan, C. S. 1983. Seasonal acid precipitation and emission trends in the northeastern United States. General Motors Research Laboratories, Warren, Michigan. Research publication no. GMR-4456, ENV$16. Wagner, D. P., D. S. Fanning, J. E. Foss, M. S. Patterson, and P. A. Snow. 1982. Morphological and mineralogical features related to sulfide oxidation under natural and disturbed land surfaces in Maryland. Chapter 7 of Acid Sulfate Weathering. J. A. Kittrick, D. S. Fanning, and L. R. Hossner, eds. SSSA Special publication 10. Soil Society of America, Madison, Wis.