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.
TREE-RING ANALYSIS AS AN AID TO EVALUATING THE EFFECTS OF AIR POLLUTION ON TREE GROWTH Edward Cook Tree-Ring Laboratory Lamont-Doherty Geological Observatory of Columbia University Palisades, New York, 10964 ABSTRACT John Innes Forestry Commission Alice Holt Lodge Wrecclesham Farnham Surrey GU10 4LH United Kingdom Tree-ring analysis has been used to assess the impact of pollution on tree growth near point sources emitting high levels of specific pollutants. However, its use in assessing the impact of lower-level regional air pollutants on forests is more controversial. The plethora of regional pollutants coupled with insufficient physiological understanding species respond to pollutants makes any causal link between regional pollution and tree growth difficult to infer. A variety of statistical analysis procedures are available to search for anomalous behavior in tree rings in the form of ring-width decline anti changes not explained by climate. Neither of these effects is prima facie evidence for pollution stress in trees. However, the discovery and description of anomalous behavior in tree rings is an important step in understanding the epidemiology of forest decline that may ultimately be found to be caused by pollution. of how different tree INTRODUCTION For many years, air pollutants have been recognized as a factor influencing tree growth, and there is now an extensive literature on the subject (Smith, 1981; McLaughlin, 1985~. Since the Industrial Revolution, the presence of industrial plants emitting gases such as sulphur dioxide and hydrogen fluoride or particulates such as soot and heavy metals have caused severe growth reductions and mortality in trees. In extreme cases, large areas have been completely devastated. Such cases are usually well-documented, and the mechanisms and nature of tree growth reductions and mortality are reasonably well understood. Recently, widespread and severe forest declines have been reported in many parts of the world, especially in Europe (Schutt and Cowling, 1985) and North America (Johnson and Siccama, 1983~. These declines cannot be tied directly to any point-source of pollution, but the presence of coincidental high levels of pollutants in the atmosphere suggest that some of the forest declines being observed now have been caused by air pollution. Although many of the currently affected species such as silver fir (babies alba Mill.) and red spruce (Picea rubens Sarg.) have experienced large-scale declines in the past (Cramer, 1984; Weiss et al., 1985; Johnson et al., 1986), those past declines do not appear to have occurred on the same scale as the present declines of the same species (Brand!, 1985; McLaughlin et al., 1987). The apparently unique severity and scale of the current forest declines in North America and Europe have been used to bolster the argument that pollution is a primary contributor to these declines. 157
158 TREE RINGS AS A POTENTIAL INDICATION OF POLLUTION STRESS IN TREES Although there is still little convincing evidence for atmospheric pollution being the cause of regional forest declines (Woodman and Cowling, 1987), the search for indicators or markers of pollution stress in trees is being vigorously pursued. One potential indicator, which is available in virtually all temperate forests of the world, is the annual tree-ring increment. Tree rings are an unique source of information on forests. They are the only widely available source of long-term, baseline data on forest growth and productivity that may predate the present era of elevated atmospheric pollution. Insofar as the year-to-year changes in ring width are an integration and reflection of past environmental influences on tree growth, it may be possible to use tree-ring analysis to detect anomalous changes in tree growth that are characteristic of pollution stress. Thus, by quantitatively comparing past and present tree-ring patterns in forests, an air pollution effect on regional forest productivity may be detectable. There are several ways in which the annual tree-ring increment can be quantified for study. The simplest way, and the way which will be emphasized in this paper, is to measure the radial ring widths sampled from the breast-height region on the bole of a tree. Because these measures of tree growth can be obtained easily and non-destructively from increment cores, they are very practical for studying certain properties of tree growth. However, as expressions of growth, they are not without their interpretational problems. At breast-height, both cambial age and the distance of the cambium from the photosynthetic centers of the canopy increase with time. These effects, coupled with geometric increases in cambial area each year, frequently cause these ring widths to decrease with age in a curvilinear fashion. This decrease, which is an intrinsic property of tree growth, must not be misinterpretated as an indicator of pollution stress. There are alternative measures of tree growth based on the annual increment that can reduce the interpretational problems somewhat. One approach is to transform the breast-height ring widths to basal area increments (BAIs) (Phipps, 1984; Hornbeck and Smith, 1985; Phipps and Whiton, 1988~. Although BAIs are ideally measured from cross-sections, generally less accurate estimates of BAI can also be obtained from increment cores and tree diameters. If estimated from increment cores, the accuracy of BAI estimates will be effected by the circuit uniformity and symmetry of radial growth on the bole. The transformation of ring width to BAI corrects for allometric growth effects associated with increasing cambial area. Phipps ( 1984) suggests that BAI increases linearly with time in healthy stands of trees. If this is the case, then a decline in BAI may be indicative of some abnormal stress on tree growth. However, it is not necessarily an indicator of pollution stress. Other much more informative measures of annual increment can be obtained by detailed stem analysis (Duff and Nolan, 1953; Fritts, 1976, LeBlanc et al., 1 987a). This approach requires destructive sampling and analysis of annual increment both radially and with height. However, it is possible to get accurate estimates of annual volume increment and complete growth layer profiles, which are unobtainable any other way. Using detailed stem analysis, LeBlanc et al. ( 1 987a) compared ring-width patterns obtained from breast-height with those obtained from "ring number sequences" (RNSs) that maintained a constant cambial age and position with respect to the crown. They found that the two measures of tree growth were significantly correlated in cases where the sampled trees were dominants in the stand or growing on good sites. Suppressed trees and trees growing on poor sites showed poorer correlation between the two ring-width sequences. LeBlanc et al. (1987a) also noted that RNSs sometimes showed greater changes in radial growth than breast-height ring widths, which suggests that the
159 former may be more sensitive to environmental changes than the latter in some cases. Detailed stem analysis is the best approach for differentiating allometric growth declines from those due to environmental or pollution effects in individual trees. However, the need for destructive sampling and the enormous increase in measurement effort over breast-height ring-width studies will restrict this approach to only a few intensively studied research sites. Although tree-ring analysis has obvious potential for the study of regional forest decline, its application in such studies is difficult and controversial. The fact that tree rings are an integration of many environmental influences on tree growth means that any pollution signal may be small and embedded in a high level of natural environmental noise, i.e., the signal-to-noise ratio is likely to be low. This problem is especially apparent in the study of regional decline where the level of pollution is comparatively low compared to that near pollution point-sources. Consequently, tree-ring studies of regional forest decline often sample hundreds or even thousands of trees for analysis in order to reduce the noise level (e.g., Schweingruber et al., 1985; McLaughlin et al., 1987~. However, even with very large sample sizes and redundancy in the experimental design, the identification of an unequivocal pollution signal in tree rings is still very difficult. One potential problem in identifying a pollution signal is determining, first, the expectation of normal growth in the absence of pollution (Hyink and Zedacker, 1987~. Developing a useful normal growth expectation is extremely difficult for all but the simplist cases, e.g., even-aged, single-species forest plantations. For the closed-canopy forests of eastern North America, which are typically a mixture of tree species and ages, useful normal growth models do not exist (e.g., Hornbeck and Smith, 1985~. In such environments, the evolution of tree rings through time is fundamentally stochastic (Cook, 1 987a; 1 976b) and, therefore, difficult to predict. Although stochastic forecasting methods, such as those based on autoregressive-integrated moving average (ARIMA) time series models (Box and Jenkins, 1970), could be used to produce an expectation of normal growth in a pollution period, the useful forecasting horizon of such methods is likely to be too short for this application. In addition, the determination of an anomalous ring-width departure from expectations of normal growth would be conditional on the validity of those expectations. It is desirable that such an assessment be made as unconditional as possible, i.e., that it not be based on expectations from a potentially flawed normal growth model. To compound the probable lack of a useful normal growth model, we do not have a good expectation of what a pollution effect looks like in tree rings other than, perhaps, that ring widths should decrease in the presence of air pollution. In itself, an expectation of anomalous ring-width decline is nonunique and, therefore, insufficient for identifying a pollution effect on regional forest growth. Our understanding of the physiological pathways and effects of various pollutants on forests is still poor, especially for relatively low doses of air pollution. We also lack a useful understanding of the interactions between pollution and climate, which may act synergistically or in opposition depending on the way in which climate is affecting tree growth at the time. Thus, without a good pollution effect model for tree rings, there is virtually no possibility of a direct hypothesis test between tree rings and air pollution. Consequently, tree-ring analysis can not be used to prove, in a direct causal sense, that pollution is responsible for forest decline. Rather, it is best suited for eliminating other natural explanations of decline such as climate (Cook, 1 987a) and stand dynamics (McLaughlin et al., 1987), and for discovering relationships that may produce testable hypotheses about the interactions between tree growth, climate, and air pollution (Cook et al., 1987; Johnson et al., 1988~.
160 To this end, we will review some methods of tree-ring analysis based solely on ring widths obtained from the breast-height region of the bole, and have been used to assess forest decline and its possible link to air pollution. These methods fall into two basic categories: testing for anomalous declines in ring width and testing for changes in tree rings that can not be explained by climate. This review will only cover studies of regional forest decline since this problem is much more common than local decline caused by point-sources of pollution (see Thompson, 1981; Fox et al., 1986~. All tree-ring studies should be based on the principles of dendrochronology (Fritts, 1976~. We will assume a basic familiarity with the science, especially the principle of crossdating upon which the integrity of tree-ring analysis rests. For a comprehensive review of the science, see Fritts (1976), and for a detailed examination of the problems associated with analyzing tree rings for pollution effects, see Cook (1987a) and Wigley et al. (1987~. TESTING FOR ANOMALOUS DECLINES IN RING WIDTHS In searching for anomalous declines in ring width, the obvious question arises: What is anomalous? The term "anomalous" implies that we know what is not anomalous, i.e., what is normal growth in tree rings. However, as noted in the previous section, we may not have a useful expectation of normal radial growth except for the simplest cases. In order to avoid this obstacle and make the assessment of anomalous decline unconditional in the sense described above, large-scale forest surveys have been used. In North America, McLaughlin et al. (1987) examined the ring-width patterns obtained from 1012 red spruce trees growing on 48 different sites in the northern and southern Appalachian Mountains. The sites had a wide range of disturbance histories, stocking levels, and age structures. Previous studies of red spruce tree rings (e.g., Johnson and Siccama, 1983; Johnson and McLaughlin, 1986) indicated that red spruce experienced a widespread, synchronous decline in ring width in the northern Appalachians after about 1960. The discovery of this apparently anomalous decline suggested the intervention of some new stress on red spruce growth such as pollution. McLaughlin et al. (1987) sought to test more rigorously and to quantify the existence of this putative anomalous decline using the technique of intervention detection (Downing and McLaughlin, 1987~. This technique searches for the occurrence of step-like changes and gross outliers in time series without regard to cause. An example of a pure stepfunction fitted to a red spruce tree-ring series is shown in Figure 1. McLaughlin et al. (1987) were able to demonstrate conclusively the existence of a widely synchronous decline in red spruce ring widths after about 1960 in the northern Appalachians. Red spruce in the southern Appalachians showed a less distinct synchronous decline after about 1965. Given the heterogeneity in stand characteristics among the 48 analyzed collections, it is highly unlikely that changes in radial increment due to stand maturation (Zedacker et al., 1987) would predict the observed synchronous declines. Therefore, McLaughlin et al. (1987) concluded that red spruce ring widths have declined anomalously over most of that species' range since 1960. An analysis of an independent ring-width data set from 2387 red spruce trees growing on (mostly) previously logged, low-elevation sites in New England supports the existence and timing of this synchronous decline (Hornbeck and Smith, 1985). However, Hornbeck and Smith (1985) cautioned that the maturation of these second-growth forests may be responsible for the synchronous decline because many of the stands were logged at about the same time 80-100 years ago. Recent research by Van Deusen (1987b) also supports the stand maturation hypothesis for low-elevation red spruce decline. However, the coincidental decline of red spruce ring widths at all elevations and for a wide range
161 A. TREERING INDICES WITH THE 1968 STEP FUNCTION INTERVENTION _ I. :~ _ 0.8 n 1890 1900 1910 1920 1930 1940 1950 1960 1970 ~ 98O YEAR S Figure 1. An example of step-like decline in ring width in red spruce. The dashed line is a simple step-function fitted to the tree-ring series by least squares. These kinds of effects can be objectively searched for in tree rings using the method of intervention detection. of stand histories argues for a similar influence operating at all locations. Thus, climate and/or pollution may still be responsible for an anomalous ring-width decline of red spruce even at low elevations. In Europe, Schweingruber et al. (1985) examined the geographic pattern of abrupt, long-term changes in tree rings for several conifer species growing in the Swiss Rhone Valley. The timing of abrupt growth change was determined visually from cross-sections and increment cores of 2500 trees, and the extent of the change was estimated by comparing the mean ring width of the altered growth period with the mean ring width for the same number of rings preceding the period of change. Persistent ring-width changes of 70% or more could be readily determined, although changes of 30% or less could not be reliably identified (Schweingruber et al., 1985~. With the exception of one pine species, all of the conifer species showed abrupt ring-width reductions that strongly clustered in the 1970s decade, a time when regional forest decline also began in Europe (Schutt and Cowling, 1985). Schweingruber et al. (1985) were not able to find any clear relationship between the onset of decline and anomalous climate. In addition, the degree of decline was not correlated with site attributes, except for elevation (less decline above 1500 m) and the proximity of the trees to industrial plants (more decline closer to the plants). Schweingruber et al. (1985) concluded that local and regional pollution was the likely cause of the ring-width declines in the Swiss Rhone Valley. For another example of this kind of abrupt growth change analysis, see Schweingruber (1986~. These North American and European studies avoided the need for a normal growth model by examining a very large number of trees from a large number of sites having many different attributes. In so doing, it was possible virtually to eliminate the probability that some stand-level variables, such as stocking level and disturbance history, could explain the observed synchronous ring-width declines. Under these conditions, the statistics supporting the existence of a synchronous decline can be used to assert that the decline is anomalous, in the sense that stand-level variables are an insufficient explanation. However, air pollution cannot generally be regarded as the
162 probable cause of such anomalous declines because meso-scale and synoptic-scale climatic variables and extremes also have the capacity to produce the observed synchronization. Testing for an anomalous ring-width decline on a single site is much more difficult. Either a justifiable normal growth model or a purely stochastic method, like intervention detection, must be used to test for an anomalous decline. Either way, the possibility that stand-level variables are responsible for the decline will be difficult to eliminate. The conclusion of Schweingruber et al. (1985) that suggested a forest decline-pollution link is tenable to the extent that climate was not correlated with the onset of decline and the Swiss Rhone Valley is a relatively small area with some point-sources of pollution. The McLaughlin et al. (1987) study was more complicated, covering a much larger area with different climatic regimes, airsheds, and pollution levels. For that reason, no pollution effect could be concluded from the existence of synchronous decline. In the next section, some methods will be described for determining if climate can explain such anomalous ring-width declines. TESTING FOR CHANGES IN TREE RINGS THAT CANNOT BE EXPLAINED BY CLIMATE The documentation of an anomalous ring-width decline is an important but insufficient condition for declaring that pollution is contributing to a regional forest decline. Since climate can also reduce the ring widths of trees over broad geographic areas, it is necessary to model and eliminate this potentially confounding effect in the tree rings before a pollution effect can be inferred. In North America, Cook (1987a) devised a method of testing for the intervention of non-climatic effects in tree rings, such as pollution. This method requires that the ring widths first be reduced to a stationary sequence of relative tree-ring indices via some form of standardization (Fritts, 1976~. Cook (1987a) advocated a "stiff' smoothing spline (Cook and Peters, 1981 ~ for estimating and removing the underlying growth trend in ring widths. This method removed very little of the post- 1960 ring-width decline in the red spruce tree rings analyzed in that study. However, the determination of the appropriate spline stiffness has been criticized for its subjectivity. As a purely objective alternative, Van Deusen (1987a) suggested using the first-differences of the logarithmically transformed ring widths as the relative tree-ring indices. This method will largely remove any ring-width decline in the resultant indices. The inherent objectivity of trend removal by first differencing is a compelling argument for its use as long as the decline in the ring widths is of little direct interest. Having standardized the ring widths, Cook ~ 1987a) modeled two slightly different forms of the same red spruce tree-ring chronology with monthly mean temperatures for the years 1890- 1950 using stepwise multiple regression analysis. The regression models were able to explain over 50% of the variance in the red spruce tree-ring indices. He then tested the validity of the regression models by predicting tree-ring indices for the 1951-1960 period, which was assumed to be unaffected by factors related to the post-1960 decline. These validation tests were successful, indicating that the temperature-based models could be used to estimate tree rings from climate in the post- 1960 decline period. A change in the relationship between tree rings and climate after 1960 would be evidence for the intervention of some new unmodeled variables affecting tree growth. In both cases, the monthly temperature models could not predict the behavior seen in the tree-ring indices after 1960. This included an inability to predict both the decline and the yearly pattern of change in the indices. Thus, there appeared to be a change in the
163 red spruce tree rings after 1960 that could not be explained by climate. An example of this effect, redrawn from Cook (1987), is shown in Figure 2. A. ACTUAL AND ESTIMATED UNWHITENED RED SPRUCE INDICES 1.2 bO t: 1.0 Z; i_ 0.8 O.6 18 B) · ~ · =) t ~ 1940 YEARS 1950 1960 1970 1980 Figure 2. An example of using empirical climatic response models to search for anomalous effects in tree rings that are not due to climate. The calibration period is the time period used for developing the regression model between tree rings and climate. The resulting model is then used to predict tree rings from climate in the verification period. Note the good predictions up to about 1967 and the breakdown in the model after that date. With this success, Cook et al. (1987) and McLaughlin et al. (1987) searched for non-climatic changes in tree rings from many red spruce sites. In the northern Appalachians, the results were identical to those just described for all sampled red spruce growing above 800 m elevation. Spruce growing below 800 m elevation sometimes showed a stable response to climate through the post-1960 period. This finding is consistent with data indicating that the red spruce decline is more severe above 800 m (Johnson and McLaughlin, 1986) where, coincidently or not, pollution levels also increase markedly. Cook et al. (1987) also noted a consistent relationship between red spruce growth and climate for all sites above 800 m elevations. They found that red spruce grew better (worse) in the current growing season when August of the previous growing season was cooler (warmer) than average and when the December prior to the current growing season was warmer (cooler) than average. Occurrences of excessively warm Augusts and cold Decembers since the 1820s in the northern Appalachians also correlated with some historical declines of red spruce in that region (Weiss et al., 1985; Johnson et al., 1986). Thus, there may be a relationship between the current red spruce decline and anomalous temperatures, although pollution cannot be eliminated as a contributor. In Europe, essentially the same method of testing for the intervention of non-climatic effects in tree rings was independently developed and used by Eckstein et al. (1983) and Eckstein (1985) to study the decline of several tree species. Their results also indicate that climate alone could not explain the level of decline seen in the ring widths, although the climatic response appears to change less than it does for red spruce in North America.
164 The successful use of empirical climatic response models to test for anomalous changes in tree rings, described here and in other studies (e.g., McClenahen and Dochinger, 1985; LeBlanc et al., 1987b), indicates that this approach has considerable potential for the study of forest decline. However, one assumption of this method needs to be investigated further to insure the proper interpretation of its results. The method assumes a stationary relationship between tree rings and climate when there is no intervention of new variables. For this assumption to be correct, systematic changes in climate must not alter the way in which trees respond to climate. This assumption has not been adequately tested even though it is known that climate has changed significantly in the 20th century (Jones et al., 1986; Bradley et al., 1987~. The possibility that changes in the response of trees to climate may be an indicator of pollution stress has lead to the development of another very powerful statistical method based on the Kalman filter. This method, which explicitly allows for time-dependence in the regression coefficients relating tree rings to climate, was developed independently by Van Deusen (1987a) in North America and Visser (1986) in Europe. The method does not require any prior knowledge about the timing of possible interventions. In addition, the complete yearly time-dependence between tree rings and climate is available, which enables the evolution and timing of change to be readily assessed. Finally, the time-dependence in the regression relationships is available separately for each climatic variable in the model. These are obvious advantages over the previous method, which only produces information about the relationship of tree rings with a composite climatic model that is assumed to be time invariant. A disadvantage of the Kalman filter method is the difficulty in selecting the proper climatic variables for analysis since no subset selection procedures, analogous to stepwise regression, are available. Visser and Molenaar ( 1986) describe a subset selection method that is based on separately screening each candidate climatic variable for statistical significance with tree rings. Those variables that are significant, even in a time-dependent sense, may be retained in the full model. However, the probability of including spurious variables in the model appears to be great, since there is no way of differentiating time dependence due to spurious association from time-dependence due to physical and biological changes in the trees. The possibility that climatic change is creating any observed time-dependence must also be kept in mind when interpreting the results. Powerful statistical tools are now available for searching for non-climatic effects in tree rings that could be due to air pollution. The most daunting tasks are in applying these techniques well and in properly interpreting the results. CONCLUSION As concern about the effects of air pollution on forests continues to grow, it is likely that tree-ring analysis will play an increasingly important role in the assessment of forest health. The historical perspective available from tree rings is unique, but the proper interpretation of this retrospective look at past tree and forest conditions is by no means simple. Still missing is a useful understanding of how ambient levels of various air pollutants affect the growth of different tree species under natural forest conditions. Until this understanding is obtained, any causal link between air pollution and forest decline that is inferred from tree-ring analysis alone will be very difficult to defend. But to the extent that proper care is taken in developing and statistically analyzing tree-ring data, tree-ring analysis should continue to be an important tool for discovering
165 and characterizing anomalous behavior that may be an indicator of pollution stress in trees. ACKNOWLEDGMENTS This research has been supported by a National Science Foundation Division of Climate Dynamics Grant ATM 85-15290. We also acknowledge the support of the Forest Service and National Vegetation Survey in the United States and the Forestry Commission in the United Kingdom. We thank G.C. Jacoby and J. Overpeck for kindly reviewing the manuscript. Lamont-Doherty Geological Observatory Contribution No. 435-3. REFERENCES Box, G.E.P., and G. Jenkins. 1970. Time Series Analysis: Forecasting and Control. Holden-Day, San Francisco. Bradley, R.S., H.F. Diaz, J.K. Eischeid, P.D. Jones, P.M. Kelly, and C.M. Goodess. 1987. Precipitation fluctuations over Northern Hemisphere land areas since the mid-19th century. Science 237:171 - 175. Brandl, H. 1985. Zur Bedeutung bestandesgeschichtlicher Untersuchungen in der Forstgeschichte am Beispiel des Tannensterbens im Schwarzwald. Allgemeine Forst und Jagdzeitung 156: 142- 146. Cook, E.R., and K.P. Peters. 1981. The smoothing spline: a new approach to standardizing forest interior tree-ring width series for dendroclimatic studies. Tree-Ring Bulletin 41:45-54. Cook, E.R. 1987a. The use and limitations of dendrochronology in studying the effects of air pollution on forests. Pp. 277-290 In Hutchinson, T.C. and Meema, K.M., eds. Effects of Atmospheric Pollutants on Forests, Wetlands, and Agricultural Ecosystems. Berlin: Springer-Verlag. Cook, E.R. 1987b. The decomposition of tree-ring series for environmental studies. Tree-Ring Bulletin. 47:37-59. Cook, E.R., A.H. Johnson, and T.J. Blasing. 1987. Forest decline: modeling the effect of climate in tree rings. Tree Physiology 3~1~:27-40. Cramer, H.H. 1984. On the predisposition to disorders of Middle European forests. Leverkusen Pflanzenschutz Nachrichten 37:208-334. Downing, D., and S.B. McLaughlin. 1987. Intervention detection - a systematic technique for examining shifts in radial growth rates of forest trees. Pp. 543-554 in Jacoby, G.C. and Hornbeck, J.W., eds. Proceedings of the International Symposium on Ecological Aspects of Tree-Ring Analysis. United States Department of Energy. Duff, G.H., and N.J. Nolan. 1953. Growth and morphogenesis in the Canadian forest species. I. The controls of cambial and apical activity in Pinus resinosa Ait. Canadian Journal of Botany 31:471-513.
166 Eckstein, D. 1985. On the application of dendrochronology for the evaluation of forest damage. Pp. 287-290 in Schmid-Haas, P., ed. Inventorying and monitoring endangered forests. IUFRO Conference, Zurich 1985. Anstalt fur das forstliche Versuchswesen, Birmensdorf. Eckstein, D., R.W. Aniol, and J. Bauch. 1983. Dendroklimatologische Untersuchungen zum Tannensterben. European Journal of Forest Pathology 13:279-288. Fox, C.A., W.B. Kincaid, T.H. Nash III, D.L. Young, and H.C. Fritts. 1986. Tree-ring variation in western larch (Larix occidentalis) exposed to sulfur dioxide emissions. Canadian Journal of Forest Research 16:283-292. Fritts, H.C. 1976. Tree Rings and Climate. Academic Press, London. Hornbeck, J.W. and R.B. Smith. 1985. Documentation of red spruce growth decline. Canadian Journal of Forest Research 15: 1199-1201. Hyink, D.M., and S.M. Zedacker. 1987. Stand dynamics and the evaluation of forest decline. Tree Physiology 3~1~:17-26. Johnson, A.H., and T.G. Siccama. 1983. Acid deposition and forest decline. Environmental Science and Technology 17~7~: 294A-305A. Johnson, A.H., and S.B. McLaughlin. 1986. The nature and timing of the deterioration of red spruce populations in Appalachian forests. Pp. 200-230 in Monitoring and assessing trends in acidic deposition. Washington, D.C.: National Academy of Sciences Press. Johnson, A.H., A.J. Friedland, and J.G. Dushoff. 1986. Recent and historic red spruce mortality: evidence of climatic influence. Water, Air, and Soil Pollution 30:319-330. Johnson, A.H., E.R. Cook, and T.G. Siccama. 1988. Relationships between climate and red spruce growth and decline in the Northern Appalachians. Proceedings of the National Academy of Sciences. (in press). Jones, P.D., S.C.B. Raper, R.S. Bradley, H.F. Diaz, P.M. Kelly, and T.M.L. Wigley. 1986. Northern hemisphere surface air temperature variations, 1851-1984. Journal of Climate and Applied Meteorology 25:161-179. LeBlanc, D.C., D.J. Raynal, and E.H. White. 1987a. Acidic deposition and tree growth: I. The use of stem analysis to study historical growth patterns. JournaI of Environmental Quality 16~4~:325-333. LeBlanc, D.C., D.J. Raynal, and E.H. White. 1987b. Acidic deposition and tree growth: II. Assessing the role of climate in recent growth declines. Journal of Environmental Quality 16~4~:334-340. McClenahen, J.R., and L.S. Dochinger. 1985. Tree-ring response of white oak to climate and air pollution near the ohio river valley. Journal of Environmental Quality 14:274-280.
167 McLaughlin, S.B. 1985. Effects of air pollution on forests: a critical review. Journal of the Air Pollution Control Association 35(5):512-534. McLaughlin, S.B., D.J. Downing, T.J. Blasing, E.R. Cook, and H.S. Adams. 1987. An analysis of climate and competition as contributors to decline of red spruce in high elevation Appalachian forests of the Eastern United States. Oecologia 72: 487-501. Phipps, R.L. 1984. Ring-width analysis. Pp. 255-271 in Symposium proceedings: Airpollution and productivity of the forest, October 4-5, 1983. Izaak Walton League, Washington, D.C. Phipps, R.L., and J.C. Whiton. 1988. Decline in long-term growth trends of white oak. Canadian Journal of Forest Research 18:24-32. Schutt, P., and E.B. Cowling. 1985. Waldsterben, a general decline of forests in Central Europe: symptoms, development and possible causes. Plant Disease 69:548-558. Schweingruber, F.H., R. Kontic, M. Niederer, C.A. Nippel, and A. Winkler-Seifert. 1985. Diagnosis and distribution of conifer decay in the Swiss Rhone Valley. In: Turner, H. and Tranquillini, W., eds. Establishment and Tending of Subalpine Forest; Research and Management. Eidgenossische Anstalt fur das forstliche Versuchswesen, Berichte 270: 189- 192. Schweingruber, F;H. 1986. Abrupt growth changes in conifers. IAWA Bulletin 7:277-283. Smith, W.H. 1981. Air Pollution and Forests. Springer-Verlag, New York. Thompson, M.A. 1981. Tree rings and air pollution: a case study of Pinus monophylla growing in east-central Nevada. Environmental Pollution, Ser. A 26:251-266. Van Deusen, P.C. 1987a. Some applications of the Kalman filter to tree-ring analysis. Pp. 566-578 in Jacoby, G. C., and Hornbeck, J. W., eds. Proceedings of the Inrernational Symposium on Ecological Aspects of Tree-Ring Analysis. United States Department of Energy. Van Deusen, P.C. 1987b. Testing for stand dynamics effects on red spruce growth trends. Canadian Journal of Forest Research. 17:1487-1495. Visser, H., 1986. Analysis of tree ring data using the Kalman filter technique. IAWA Bulletin 7:289-297. Visser, H., and J. Molenaar. 1986. Time-dependent Responses of Trees to weather variations: an Application of the Kalman Filter. Report 50385MOA 86-3041, N.V. KEMA, Arnhem, The Netherlands. 60 pp. Weiss, M.J., L.R. McGreary, I. Millers, J.T. O'Brien, and M. Miller-Weeks. 1985. Cooperative survey of red spruce and balsam decline and mortality in New Hampshire, New York, and Vermont - 1984. Interim report, USDA Forest Service, Forest Pest Management, Durham, NH. 130 pp.
168 Wigley, T.M.L., P.D. Jones, and K.R. Briffa. 1987. Detecting the effects of acidic deposition and C02-fertilization on tree growth. Pp. 239-254 in Kairiukstis, L., Bednarz, Z., and Feliksik, E., eds. Methods of Dendrochronology - 1. Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland. . Woodman, J.N., and E.B. Cowling. 1987. Airborne chemicals and forest health. Environmental Science and Technology 21~2~: 120-126. Zedacker, S.M., D.M. Hyink, and D.W. Smith. 1987. Growth declines in red Journal of Forestry 85:34-36. spruce.
