JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Savard, M. M.
Right arrow Articles by Marion, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Savard, M. M.
Right arrow Articles by Marion, J.
Agricola
Right arrow Articles by Savard, M. M.
Right arrow Articles by Marion, J.
Related Collections
Right arrow Global Change
Right arrow Other Environmental Contamination
Right arrow Air Pollution
Right arrow Isotopes
Published in J. Environ. Qual. 33:13-26 (2004).
© ASA, CSSA, SSSA
677 S. Segoe Rd., Madison, WI 53711 USA

TECHNICAL REPORTS

Atmospheric Pollutants and Trace Gases

Effects of Smelter Sulfur Dioxide Emissions

A Spatiotemporal Perspective Using Carbon Isotopes in Tree Rings

Martine M. Savard*,a, Christian Bégina, Michel Parenta, Anna Smirnoffa and Joëlle Marionb

a Natural Resources Canada, Geological Survey of Canada, 880 Chemin Ste-Foy, Bureau 840, QC, Canada G1S 2L2
b INRS-ETE, 880 Chemin Ste-Foy, Bureau 840, B.P. 7500, Sainte-Foy, QC, Canada G1V 4C7

* Corresponding author (msavard{at}nrcan.gc.ca).

Received for publication October 7, 2002.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 REGIONAL SETTING
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
We wanted to test the hypothesis that forest exposure to phytotoxic gases indirectly affects their carbon uptake. We estimated that the reduction of photosynthesis may have reached 20 to 30% at a site located 9 km (test site) from the Horne copper smelter in Rouyn–Noranda, which is a point source of SO2. Twenty-one spruce trees older than 100 yr were selected from seven sites at various distances from the smelter to evaluate conditions prior to and during the periods of smelter operation. The carbon isotope results obtained from spruce tree rings at our test site reveal an unprecedented and abrupt shift of +4{per thousand} after the onset of smelter operations. This large and permanent shift exceeds natural variations in regional pre-smelter series or in the series at a remote control site. All trees up to 116 km downwind from the smelter show {delta}13C positive shifts following the onset of operations. There is also a clear inverse relationship between the amplitude of the first-order trends and distance from the smelter. Those {delta}13C trends indicate that trees exposed to high levels of SO2 decrease their level of CO2 uptake through activation of stomatal closure. This is strongly supported by the significant departure of the Rouyn–Noranda trends from those measured for trees from non-industrialized areas of the Northern Hemisphere, or calculated using global atmospheric conditions. Considering the large number of SO2 point sources in North America, our results imply that CO2 uptake by the boreal forest in the vicinity of these sources may be lower than previously thought.

Abbreviations: DINAMITE, Dendrogeochemical Investigation of Natural and Anthropogenic Metals in the Environment • VPDB, international standard used for reporting the relative abundances of 13C via the delta notation • VPDB, is a carbonate provided by IAEA in Vienna (V) as a replacement of the former international standard Peedee Belemnite (PDB)


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
REGIONAL SETTING
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
GLOBALLY, HUMAN ACTIVITIES have resulted in SO2 concentrations in the atmosphere being up to three orders of magnitude higher than preindustrial concentrations (e.g., Pham et al., 1996). In 1990, the North American atmosphere contained 2000 times more anthropogenic SO2 than under "natural" conditions (Environment Canada, 2002). Few studies have described the potential isotopic {delta}13C changes in growth rings of trees undergoing point-source pollution stress in the natural habitats (Freyer, 1979; Martin and Sutherland, 1990). Most reports on tree-ring {delta}13C values address physiological questions, paleoclimatic records (e.g., Farquhar et al., 1989; Lipp et al., 1996), or higher atmospheric CO2 concentration effects on tree growth (e.g., Bert et al., 1997). The response in terms of carbon uptake of the boreal forest to increasing levels of phytotoxic air pollutants has not yet been examined. Hence, the effect of the increasing number of point sources emitting large quantities of SO2 needs to be assessed in terms of its implication for the present and past carbon cycle.

The boreal forest constitutes a biological pump transferring carbon from the atmosphere to the terrestrial biosphere. Recent biochemical studies suggest an increase of its net photosynthetic production due to the higher concentrations of "fertilizers" such as CO2 and NOx in the atmosphere (e.g., Ladeau and Clark, 2001; Oren et al., 2001; Schimel et al., 2001). However, CO2 uptake has not been widely discussed in terms of forest response to higher concentrations of SO2. The issue can be addressed by examining the response of trees to pollution stress at a regional scale, and by answering the following key questions. Can carbon isotope ratios in stem ring cellulose record ambient atmospheric conditions? How does photosynthesis by trees adjust to pollution-induced stress? Do tree-ring series have the potential to archive a signal of exposure to air pollution? Is the overall rate of C assimilation by trees exposed to air pollution modified or is it maintained at the preindustrial level?

The Dendrogeochemical Investigation of Natural and Anthropogenic Metals in the Environment (DINAMITE) study is part of a multidisciplinary initiative within the Geological Survey of Canada Metals in the Environment (MITE) program on investigating the effects of pollutants from point sources. The main objective of the project is to provide a historical perspective on the changes in the environment generated by the Horne copper smelter emissions in the Rouyn–Noranda region. In DINAMITE, we combined natural and anthropogenic tracers in dendrochronological series to develop an environmental monitoring tool. The DINAMITE study has characterized geochemical soil profiles, dendrochronological, and dendroecological–geochemical series, as well as metal and isotope dendrogeochemistry (Savard et al., 2001, 2002). Having briefly documented the elemental concentrations and only partly the H and C isotope ratios in a previous publication (Savard et al., 2002), we present here the entire carbon isotope data set. This includes data obtained for three trees at each of seven sites selected to document the effect of smelter emissions on regional vegetation, and a full discussion of all aspects of the results and of their effects.

Our working hypothesis states that fractionation of carbon isotopes in tree stems is affected by the presence of potentially phytotoxic concentrations of SO2 in the ambient atmosphere. Specific objectives were to (i) evaluate the potential of carbon isotope dendrogeochemical series as archives of pollutant levels in the atmosphere, (ii) document the effect of elevated atmospheric SO2 concentrations on carbon assimilation by trees, and (iii) evaluate the potential change in the boreal forest CO2 uptake resulting from point-source SO2 pollution.


   REGIONAL SETTING
TOP
 ABSTRACT
 INTRODUCTION
 REGIONAL SETTING
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The Horne smelter (Fig. 1) located in Rouyn–Noranda (northwestern Québec; 48°15' N, 79°01' W) initiated full production in December 1927. Since then, it has been emitting large amounts of SO2 and metals (Table 1). The smelter was servicing mining companies exploiting base and precious metal deposits of northwestern Québec mining camps until 1976. Since then, it has also been processing material from all around the world. Smelting goes on 24 hours a day, all year long, except when SO2 concentrations at monitoring sites in the city of Rouyn–Noranda exceed accepted environmental norms. Operations ceased in 1948 and 1954 for two prolonged periods, due to strikes that took place during winter; the emissions during the growth season of trees therefore remained elevated. Emissions were monitored from 1965 until present (Table 1), and the amount of feed at the smelter is used as a proxy during 1928–1964. Before the implementation of modern control measures, the Horne smelter was one of the largest sources of airborne SO2 in Canada (Environment Canada, 2002). The reported emissions of SO2 ranged annually between 710000 and 90000 Mg from 1965 until 2000 (Table 1). Total emissions have decreased by more than 80% since 1980.



View larger version (61K):
[in this window]
[in a new window]
  		Fig. 1. Location of the studied sites and meteorological stations.

 

View this table:
[in this window]
[in a new window]
  		Table 1. Summary of total annual emissions from the Horne copper smelter in Rouyn–Noranda, Québec, Canada.

 
The Rouyn–Noranda geological domain belongs to the Abitibi Belt in the southeastern margin of the Superior Province. The regional geology includes Precambrian volcanic, sedimentary, and plutonic rocks, all metamorphosed to the greenschist facies, which mainly host volcanogenic massive sulfide and lode gold deposits (Simard et al., 1990). The Rouyn–Noranda region was repeatedly covered by ice sheets that advanced successively in different directions (Veillette, 1989). Hence the tills, which are the parent material of soils at DINAMITE sites, are matrix-dominated silty sand diamictons characterized by well-mixed lithologic and granulometric composition, a situation typical of Shield regions affected by polyphase glacial transport (Parent et al., 1996). The region is covered by an extensive, almost continuous blanket of fine-grained glaciolacustrine sediments deposited in large glacially impounded lakes at the close of the last glaciation (Veillette, 1994).

Wind direction during summer is predominantly from the southwest (Fig. 1). The combined data from five Environment Canada weather stations produce a continuous regional climatic series covering the last century. The five stations (Abitibi Post, Manneville, Amos, Val-d'Or, and Rouyn–Noranda) are all located within a 100-km radius of the smelter. Average climatic and dendrogeochemical series can thus be compared in the Rouyn–Noranda region. Unfortunately, this is not possible at the control site (800 km) as complete long-term meteorological series are not available for the eastern Hudson Bay region. Only discontinuous climatic time series were recorded at the Kuujjuarapik station, 50 km away from the control site, and the quality of the climatic record was not very high until the 1960s when new instrumentation was installed.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 REGIONAL SETTING
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Site Selection and Sampling of Trees
Site selection and sampling protocol constituted key steps of the DINAMITE methodology, leading to optimal dendrogeochemical characterization. Tree species, as well as geological, ecological, pedological, and physiological settings, were analogous at all sites. This procedure allows us to minimize the number of trees per site to be used for dendrogeochemistry. The actual number of trees to be analyzed to get a representative {delta}13C trend was assessed using the expressed population signal (EPS) evaluation (McCarroll and Pawellek, 1998). Following this procedure, black spruce [Picea mariana (Mill.) Britton et al.] and white spruce [Picea glauca (Moench) Voss] trees were sampled during the summers of 1997–2000. These species were selected because of their wood qualities for dendrochronological and dendrogeochemical studies (Cutter and Guyette, 1993; Payette et al., 1989). In addition, their widespread distribution permits comparison at the scale of the North American boreal forest. All stands investigated belong to the Boreal Forest ecozone. In the Rouyn–Noranda region, dominant spruce trees in old-growth stands were first selected in our test site and used to set up the sampling protocol and to assess the validity of the approach. Trees were also selected in five other sites at various distances from the Horne smelter (Fig. 1; Table 2). Although these study sites are located near the smelter, there is presently no obvious sign of contamination in the field. The sites are underlain by Spodosolic soils having similar characteristics (Table 2; Soil Survey Staff, 1998), and formed on unreworked, matrix-dominated basal till derived from local shield rocks. This combination of vegetation and soil is perhaps the most widespread of the Boreal Forest ecozone.


View this table:
[in this window]
[in a new window]
  		Table 2. Characteristics of the investigated sites (test, 1, 2, 3, 4, and 5 along a southwest–northeast transect in Rouyn–Noranda) and the control site near Hudson Bay.

 
The test site is located at 9 km downwind and probably is the nearest old tree stand that survived the first period of smelter operations and urban expansion. The site was chosen to establish the dendrogeochemical {delta}13C response resulting from exposure to smelter emissions (Fig. 1; Table 2). Sites 1 to 5 are all located within 116 km along a southwest–northeast transect running across Rouyn–Noranda (Table 2). These sites were selected to assess the spatial trend of the {delta}13C ratios in exposed trees at increasing distances from the smelter (i.e., to assess the effect of lowering SO2 concentrations on photosynthesis).

The control site is located in the northern Boreal Forest ecozone, in the remote, unpopulated coastal region east of Hudson Bay, approximately 800 km north of Rouyn–Noranda (Fig. 1). This site is considered uncontaminated by point sources; anthropogenic effects, if any, should be minor and result from diffuse pollution. Therefore trees of this site are used to establish the {delta}13C profile of natural or near-natural (background) conditions at the broad regional scale.

More than 15 spruce trees were cored and assessed for dendroecology and dendrochronology before three were selected at each site for dendrogeochemical analysis. All trees were dated by standard dendrochronological methods. The minimum age of the trees selected for dendrogeochemistry was 105 yr to allow for an evaluation of background characteristics during the pre-smelter period. In the first phase of our dendrogeochemical investigation, we carefully established the field sampling protocol for optimal geochemical reproducibility including number, height, and directions of stem cores. Four radial subsamples were cored at 1 m aboveground in four directions at 90° angles. Chronological resolution was set at 2 yr for all trees except for one tree from the control site, which was studied with a 5-yr resolution. Cores were cut and ring pairs separated in the laboratory using ultra-clean blades.

Cellulose Extraction and Analysis
Homogenized wood subsamples of complete ring pairs were split into two parts to make elemental and isotopic analyses on identical samples (Savard et al., 2001). Subsample preparation for carbon stable isotopes was performed at the Delta-Lab of Natural Resources Canada (Geological Survey of Canada, Québec). The isotopic analysis was performed on complete ring pairs rather than on late wood commonly done in paleoclimatic investigations. Ring pairs were assessed to yield a resolution sufficient for this pilot project on pollutant effects. The {delta}13C values were obtained from extracted cellulose to avoid inconsistent isotopic variations caused by changing proportions of wood constituents, such as lipids, resins, carbohydrates, lignin, homo-celluloses, and hemi-celluloses, which have distinct isotopic ratios but can appear in highly different proportions within and between rings (Wilson and Grinsted, 1977). The extraction of cellulose followed a protocol modified from Browning (1952), Green (1963), Epstein et al. (1976), and Sternberg Da Silveira Lobo (1989). Treatment involved the removal of soluble organics using a mixture of benzene–methanol and acetone and bleaching with sodium chlorite and acetic acid. Subsequent alkaline extraction allowed hemi-celluloses to be removed. The final product, {alpha}-cellulose, was thoroughly washed with deionized water, filtered, and dried at 50 to 60°C.

All extracted cellulose subsamples were analyzed for {delta}13C using an elemental analyzer in continuous flow with an isotope ratio mass spectrometer. Calibration was performed using the NBS-19 (National Institute for Standards and Technology, Gaithersburg, MD) and IAEA-CO-9 (International Atomic Energy Agency, Vienna, Austria) standards. The international standard NBS-18 and our internal cellulose standard were used with each run to ensure that there was no analytical drift. The {delta}13C values are all relative to VPDB and the precision on the analyses was better than 0.3{per thousand} (2 standard deviations) over the 1997–2001 period of investigation. The term VPDB indicates the international standard which was used for reporting the relative abundances of 13C via the delta notation; VPDB is a carbonate provided by IAEA in Vienna (V) as a replacement of the former international standard Peedee Belemnite (PDB).

For the purpose of the present research, it was suitable to use the relative isotopic changes rather than the absolute {delta}13C values of tree-ring cellulose, so as to minimize differences induced by individual metabolic effects on 13C–12C fractionation and to compare the isotopic behavior of the various spruce trees (Table 3). Hence, isotopic ratios are reported relative to the result for the ring pair 1920–1921 chosen as an arbitrary 0 value for each specimen. This also allows a comparison through time among specimens and sites, and with various species of Northern Hemisphere trees reported in the literature.


View this table:
[in this window]
[in a new window]
  		Table 3. Carbon isotope ratios for cellulose extracted from ring pairs.

 


View larger version (19K):
[in this window]
[in a new window]
  		Fig. 4. Individual curves of the dendrochronological {delta}13C series for three trees at the control site.

 
Carbon Isotope Fractionation during Photosynthesis and Calculation of Stress Effect
Considerable work has been done to understand the partitioning of carbon isotopes in C3 plants. Photosynthetic production and autotrophic respiration rule the within-tree carbon balance (Pearcy et al., 1987), but according to numerous biogeochemists, respiration seems to have little effect on the final {delta}13C values of trees (e.g., Farquhar et al., 1982). The widely accepted equation of Farquhar et al. (1989) makes the general link between the isotope ratios in plants ({delta}13Cp) and in the source of carbon for photosynthesis, atmospheric CO2 ({delta}13Ca):

where a is the enrichment during gaseous diffusion of CO2, b is the enrichment due to enzymatic activities during carboxylation, and pi and pa are the CO2 partial pressure in the intercellular space within leaves and in the ambient atmosphere, respectively.

Based on theoretical and experimental arguments, the passage of CO2 from the ambient atmosphere into the intercellular space of leaves generates a negative fractionation due to the difference in diffusion rates of 13CO2 and 12CO2 according to Graham's law for gas diffusion:

where m is mass, y and x are isotopes, and v is diffusive velocity.

The theoretical carbon isotope enrichment ({epsilon} = {delta}13Ca – {delta}13Cp) in leaf CO2 is +4.4{per thousand} relative to atmospheric CO2, while some experimental results point toward an enrichment of +4.1{per thousand} (Farquhar et al., 1989). Modification of this value comes from the fact that in polluted or humid areas the air mass differs from the 28.84 mg/mol of pure air, changing the velocity of the heavy and light CO2 molecules accordingly. When air contains high quantities of molecules such as SO2, CO2, or water vapor, the isotopic enrichment due to gaseous diffusion increases or decreases. Also, the isotopic enrichment of leaves relative to stem cellulose in trees is –3{per thousand} due to fractionation of C isotopes during fixation and to the priority of allocation of C in trees (Wilson and Grinsted, 1977; Savard et al., 2000); therefore, the equation of Farquhar et al. (1989) to relate atmospheric and stem cellulose {delta}13C values had to be modified accordingly.

The enzymatic fixation of carbon constitutes the main fractionation step during photosynthesis; 27{per thousand} is generally attributed to the net enrichment b, although reported values range between 24.0 and 38.0{per thousand} (e.g., Farquhar et al., 1982, 1989). In addition, the proportion of carboxylase activity is considered to be constant, but using this approximation over the more rigorous addition of the respective carbon isotope fractionation related to fixation by RubPC (ribulose-1.5-biphosphate carboxylase) and PEPC (phosphoenolpyruvate carboxylase) might not be suitable for modeling photosynthesis under certain conditions (Brugnoli et al., 1998; Saurer et al., 1995). Although criticism is also expressed concerning the use of this equation to evaluate the past atmospheric CO2 concentrations (Hou et al., 2001), most dendrogeochemists agree on its reliability to establish the change of {delta}13C in leaves as induced by relative changes in CO2 pressure within leaves (pi). Therefore, we use here this equation to calculate the expected trend in trees that underwent either variations in atmospheric CO2 in terms of concentrations and {delta}13C values, or pollution stress from the smelter point source (see Discussion, below).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 REGIONAL SETTING
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Carbon Isotope Ratios at the Test Site
The three specimens investigated at the test site presented similar {delta}13CVPDB trends exhibiting coeval positive shifts in the early 1930s (Table 3). The 1930–1931 and 1932–1933 ring pairs recorded major shifts of +4{per thousand} or more, immediately after the initiation of the smelter operations, which was synchronous to ring pair 1928–1929 (Fig. 2) . The shifts translate into a high rate of increase (1{per thousand}/yr). The ring pairs of the 1934–1999 period recorded consistently high {delta}13C values relative to pre-smelter years. The overall first-order trend in trees at the test site, therefore, is a permanent +4.5{per thousand} excursion initiated at the onset of smelter operations.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 2. Carbon isotope results obtained for three specimens selected at the test site.

 
In the three trees, {delta}13C fluctuations of 0.5 to 1.0{per thousand}, which are much less than the first-order excursion, appeared throughout the entire dendrochronological series, during both the 1880–1927 pre-smelter and 1928–1999 smelter-operating periods (Fig. 2). Many of these fluctuations superimposed on the first-order trend are in phase in the three specimens, notably an increase for 1920–1923 and decreases for 1924–1929 and 1934–1935.

The climatic records of the Rouyn–Noranda stations show that the average temperature for June, July, and August remained around 15°C from 1896 until present, with short-term oscillations (maximum 2°C) occurring over the entire period. Average summer precipitation showed a gradual long-term increase from 175 to 220 mm, beginning at the end of the 1930s (Fig. 3) . We investigated the potential link between variations of the {delta}13C values and growing season climate (June, July, and August) for the Rouyn–Noranda region. The {delta}13C values have a very high, positive correlation with the feed at the smelter, and a significant, positive correlation with summer precipitation but no significant correlation with the growth index (GI) or summer temperature (Fig. 3).



View larger version (34K):
[in this window]
[in a new window]
  		Fig. 3. (a) Pearson correlation matrix of the main parameters potentially influencing the {delta}13C of stem cellulose, where GI is growth index (df = 51, critical value = 0.35 at a confidence level of 99%). (b) Smelter feed along with particulate and SO2 emissions. (c) Average {delta}13CVPDB results relative to the 1920–1921 value, and average summer precipitation through time at the test site. The term VPDB indicates the international standard which was used for reporting the relative abundances of 13C via the delta notation; VPDB is a carbonate provided by IAEA in Vienna (V) as a replacement of the former international standard Peedee Belemnite (PDB).

 
Carbon Isotope Ratios at the Control Site
The three spruce trees in the remote region of Kuujjuarapik near Hudson Bay present a flat long-term {delta}13C trend, with no major or abrupt first-order excursion, but with short-term, ±1{per thousand} fluctuations occurring over the entire investigated period (Table 3; Fig. 4) . Short-term fluctuations at the control site are similar to those in ring pairs of the pre-smelter and smelter-operating periods at the test site. In general, {delta}13C values in stem cellulose show inverse variations with average July–August precipitation for the last 35 yr monitored, and no apparent relation with the average temperatures (Fig. 5) .



View larger version (28K):
[in this window]
[in a new window]
  		Fig. 5. Average {delta}13C results of the two trees analyzed at a 2-yr resolution, and average precipitation and temperature during July and August, relative to time at the control site.

 
Spatial Trend in the Rouyn–Noranda Region
The statistics on the {delta}13C results for the six sites of the Rouyn–Noranda region show the high commonality of the within-site signals (Table 4). The expressed population signals all exceed the suggested minimum signal of 0.85 and justify the choice of three trees per site for C isotope dendrogeochemistry.


View this table:
[in this window]
[in a new window]
  		Table 4. Within-site statistics for {delta}13C series.

 
Lasting, first-order, positive {delta}13C excursions with amplitude similar to that found for the test site exist in ring series from Sites 1 and 2, which are within a 15-km radius of the smelter (Fig. 1 and 6) . Positive excursions exist as well at Sites 3 and 5 but their amplitude is lower than that at the test site, with an inverse relationship with distance from the smelter (Fig. 6). The dendroisotopic smelter-operating series at Site 4 has values that are on average higher than at the control site. However, the amplitude of this shift is not as pronounced as those at all other sites of the Rouyn–Noranda region.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 6. Carbon isotope results for all Rouyn–Noranda sites compared with results obtained for the control site and compiled from the literature.

 
Except for Site 4, maximum and average {delta}13C values of the smelter-operating period decrease with distance from the smelter for all sites including the control site (Fig. 7a) . This overall trend shows that the three sites closest to the smelter display the highest average or maximum values and that there is no increase of {delta}13C average values at the control site. We evaluated the promptitude of trees from the Rouyn–Noranda region to reach maximum {delta}13C values by using the first year of 10 consecutive maximal ring-pair values (20-yr duration). Results from the control site were not considered as their average value for the smelter-operating period is 0{per thousand} (see Fig. 7a). We observe a general increase in the post-1928 number of years required to reach the maximum {delta}13C values as the distance from the smelter increases (Fig. 7b). Overall, spatial trends in {delta}13C values differ in two ways with increasing distance from the point source: (i) there is a diminution of the average for the smelter-operating period and (ii) there is an increasing delay in reaching a maximal effect in trees of the Rouyn–Noranda region during the same period.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 7. Representation of changing {delta}13C results in space during the smelter-operating period (1930–2000). (a) Average and maximum values at all sites. (b) First year of the 10 consecutive ring-pairs showing the highest {delta}13C values for each site of the Rouyn–Noranda region.

 
Expected Trends in Trees not Exposed to Phytotoxic Air Pollution
Wood and fossil-fuel combustion constitute the main anthropogenic activities that directly affect the earth's external cycle of carbon. These activities are at least partly responsible for the observed increase of about 70 ppm in atmospheric CO2 since 1880 (Toggweiler, 1995; Siegenthaler and Oeschger, 1987; Houghton et al., 2001). As discussed above (see Carbon Isotope Fractionation during Photosynthesis and Calculation of Stress Effect), vegetation discriminates against heavy isotopes during photosynthesis of tissues; therefore, a lower uptake of CO2 by forests lowers the atmospheric {delta}13CCO2 values. As hydrocarbons are significantly depleted in 13C (approximately –25{per thousand} or less), fuel combustion amplifies this effect by massive transfer of light carbon to the atmosphere. The overall industrial effect is a progressive increase of atmospheric CO2 concentration, from 295 ppm in 1880 to 367 ppm in 1991 (Siegenthaler and Sarmiento, 1993), coincident with a decrease of its {delta}13C from –6.6 to near –8.0{per thousand} (Neftel et al., 1985; Friedli et al., 1986).

The theoretical trend generated if only atmospheric {delta}13C and PCO2 (partial pressure of carbon dioxide) changes were responsible for the variations of {delta}13C in tree-ring cellulose is represented on Fig. 8 . We compiled the carbon isotopic ratios and concentrations of atmospheric CO2 for the last 120 yr from the literature (Carbon Dioxide Information Analysis Center, 2001; Friedli et al., 1986; Francey et al., 1999), and calculated the {delta}13C values for trees theoretically not exposed to air pollution using a modified version of Farquhar's equation (see Carbon Isotope Fractionation during Photosynthesis and Calculation of Stress Effect, above). The decline of the atmospheric {delta}13C values of CO2 coupled with the increase of atmospheric CO2 concentrations creates a hypothetical decline of stem cellulose {delta}13C from –25.3 to –26.5{per thousand}.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 8. Comparison of {delta}13C values for series from unpolluted sites of the Northern Hemisphere drawn from the literature, with theoretical values if changes in isotope ratios and concentrations of atmospheric CO2 were the causes of changes in tree rings.

 
Previous investigations of dendroisotopic series with the purpose of establishing past atmospheric conditions were performed on trees from unpolluted sites of the Northern Hemisphere (Epstein and Krishnamurthy, 1990; Feng and Epstein, 1995; Freyer and Belacy, 1983; Kitagawa and Matsumoto, 1993; Leavitt et al., 1995). We compiled the {delta}13C results of these investigations on stem cellulose and calculated the average using the values relative to 1920 (Fig. 6 and 8). We evaluated the uncertainty on the mean values to be lower than ±1{per thousand} as based on the extreme outliers of the compiled curves (see Savard et al., 2002). Climatic variations at local and global scales, which are not considered in the theoretical calculations, and local ecological conditions, which influence the isotopic ratios of the natural trends, both generate the expected short-term separation between natural and theoretical trends. However there is strong agreement among the natural and theoretical long-term isotopic trends, as shown by a net diminution of 1.5{per thousand} and a coincident change in slope during the 1960s in both curves (Fig. 8). These curves are used for comparative purposes to interpret the dendroisotopic trends obtained in the present study.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 REGIONAL SETTING
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Stem Cellulose Carbon Isotope Variations Created by Natural Conditions in Rouyn–Noranda
Possible causes of changes in {delta}13C values in tree-ring cellulose include natural climatic and ecological conditions in tree stands or variations in air chemistry (PCO2, {delta}13CCO2; see Carbon Isotope Fractionation during Photosynthesis and Calculation of Stress Effect, above). Global warming and higher CO2 pressure may increase temporarily the net CO2 assimilation for regions where trees are not yet under optimal conditions (Saxe et al., 2001; Oren et al., 2001), and can be expected to generate a progressive decrease in {delta}13C values, as represented by the theoretical trend and the curve for trees unexposed to air pollution of Fig. 8. The Rouyn–Noranda dendroisotopic series do not harbor such a trend, particularly not from 1928 to present; that is, the more negative slope of the {delta}13C time series that characterizes background or theoretical global trends is not found in the region (see Fig. 6 and 8).

Natural ecophysiological and climatic conditions include age and radial growth rate and precipitation, air and soil moisture, availability of nutrients, temperature, wind, and light intensity, respectively. Following a strict protocol, we selected trees that are not expected to have recorded major natural geochemical variations, but only short-term, low-amplitude {delta}13C fluctuations of less than ±1{per thousand}, as indicated by ring series from unpolluted sites around the world (Fig. 8). For the Rouyn–Noranda test site, the pre-smelter tree rings display such fluctuations indicating that the pre-smelter period was characterized by naturally varying conditions, but that during the smelter-operating period, short-term fluctuations are superimposed on a long-term positive excursion (Fig. 3 and 6). Similar observations are made for all sites of the Rouyn–Noranda region (Fig. 6). Nutrient enrichment is known to generate increased {delta}13C values in tree rings (Saurer et al., 1995), but there was no fertilization program in the Rouyn–Noranda area during our study period. Trees at the control site essentially show short-term fluctuations throughout the investigated period (Fig. 4), but their flat trend noticeably differs from that expected for trees from unpolluted sites (see Effect on Carbon Dioxide Uptake by the North American Boreal Forest, below).

The Pearson correlation matrix obtained for the test site suggests that, at a confidence level of 99%, summer precipitation represents the second main control on the Rouyn–Noranda dendroisotopic series (Fig. 3). Temperature is considered to be significant at a confidence level of 95% only. These observations support our interpretation of the second-order fluctuations based on the comparison with unpolluted series. We therefore interpret these minor {delta}13C fluctuations as having been induced by fluctuating summer precipitation and, to a minor extent, temperature as well as other natural factors, such as availability of nutrients and light.

Effect of Smelter Emissions on Photosynthesis
First-Order Carbon Isotope Ratios Increase, an Effect of Point-Source Atmospheric Pollution
As discussed in the previous section, the {delta}13C variations expected from natural conditions are short-term and of low amplitude (Fig. 4 and 8). In contrast to these variations, the average first-order 13C enrichment described for the test site extends over more than 50 yr and shows an average amplitude of +4{per thousand}, which is very large in the dendroisotopic domain (Fig. 2, 3, and 6). Coincident and significant changes in {delta}13C values related to natural conditions such as precipitation, temperature, nutrient availability, or light intensity can be expected for trees at a specific site, but such an interpretation is difficult to reconcile for numerous, distant sites. For all the Rouyn–Noranda sites, the long-term positive excursions depart significantly from the persistent, decreasing trends for trees growing in unpolluted stands, or solely undergoing fluctuations due to changes in atmospheric CO2 (Fig. 8). In addition, the positive, first-order, and abrupt {delta}13C excursion at the test site coincides with the onset of smelter operations at the end of the 1920s. Correlation analysis for this site indicates, at a confidence level of 99%, a strong relationship between {delta}13C values in tree-ring cellulose and smelter feed, the proxy for smelter emissions (Fig. 3a). Moreover, the inverse correlations between smelter-operating amplitudes and averages of {delta}13C shifts, and distances from the smelter suggest a primary control of smelter emissions on the isotopic attributes of tree-ring cellulose (Fig. 6 and 7). In essence, the spatiotemporal first-order trends in the Rouyn–Noranda trees cannot be explained by natural factors.

Soil acidification due to deposition of SO2 might have an effect on photosynthesis (Mayo et al., 1992). However, the effect of soil chemical changes on C isotope fractionation is not demonstrated, and there is generally a delay before this process modifies tree attributes through absorption of metals (Savard et al., 2004). Previous works indicate that under polluted atmospheric conditions, tree rings have higher {delta}13C values that are generated by a decrease of stomatal apertures and/or a lower fractionation during enzymatic carboxylation (e.g., Martin et al., 1988; Saurer et al., 1995). In this sense, anthropogenic chemical conditions in the atmosphere of the sites within a 15-km radius of the Rouyn–Noranda smelter can have created the immediate +4{per thousand} shift in the {delta}13C values. The positive {delta}13C shifts of lower amplitude at the other Rouyn–Noranda sites indicate effects of lower intensity due to dilution of pollution (SO2 and perhaps particulate matter) with distance. We therefore contend that the spatiotemporal trends of the Rouyn–Noranda sites are chiefly caused by atmospheric concentrations of SO2.

Airborne Sulfur Dioxide and Reduction of Leaf Carbon Dioxide Uptake in the Rouyn–Noranda Region
Smelters are among the main producers of atmospheric SO2 in North America (Environment Canada, 2002; USEPA, 1999). The Horne smelter has emitted more than 109 Mg of SO2 annually (Table 1). The effect of SO2 on photosynthesis was the subject of numerous laboratory studies that widely recognized its phytotoxicity at varying levels depending on species, as well as on the concentration and duration of fumigation (e.g., Ziegler, 1973). Other pollutants such as O3 or particulate matter probably influence photosynthesis as well; unfortunately, O3 is not monitored in the Rouyn–Noranda area and its potential effect cannot be assessed. The effect of smelter-derived particulates on photosynthesis is not well understood. However, these pollutants may modify photosynthesis especially if they enhance SO2 effects (e.g., Darrall, 1989).

In the previous section we advocated that the spatial trend in tree-ring {delta}13C values was directly linked to pollution effects. Here we suggest that the Rouyn–Noranda trees diminished their uptake of C at a rate depending on the atmospheric amount of smelter-emitted SO2. This interpretation is supported by other findings on C3 plant behavior: (i) photosynthetic processes were demonstrated to be modified similarly by atmospheric pollutants such as O3 or SO2 (Matyssek et al., 1992; Saurer et al., 1995); (ii) significant reduction of net photosynthetic assimilation was documented for trees undergoing fumigation in laboratories with SO2, O3, or various mixtures of both gases (Freyer, 1979; Reich and Amundson, 1985; Lorenc-Plucinska, 1986; Kropff, 1987; Mooney et al., 1988; Darrall, 1989); (iii) significant reduction of growth was documented for trees in natural habitats undergoing SO2 and/or O3 pollution stress (Freyer, 1979; Martin and Sutherland, 1990; Krupa and Legge, 1998; Sakata and Suzuki, 2000); (iv) C allocation was demonstrated to be modified and net tissue production diminished significantly with increasing levels of atmospheric phytotoxic gases (Minchin and Gould, 1986; Mooney et al., 1988; Matyssek et al., 1992); and (v) leaves or tree rings produced under atmospheric pollution stress in greenhouses or in natural habitats all show an increase in {delta}13C values relative to tissues produced under clean air conditions (Martin et al., 1988; Matyssek et al., 1992; Saurer et al., 1995; Niemelä et al., 1997; Sakata and Suzuki, 2000).

The lower C assimilation rate due to noxious gases or reduced amount of available assimilates generally translates into reduced growth of stems (Martin and Sutherland, 1990; Krupa and Legge, 1998; Sakata and Suzuki, 2000), foliage (e.g., Mooney et al., 1999), or roots (e.g., Matyssek et al., 1992). Trees undergoing SO2 concentrations between 2.63 and 10.52 mg/m3 during 18 h were reported to reduce their rate of net photosynthesis by 50 to 80% depending on their relative tolerance to SO2 (Lorenc-Plucinska, 1986). Trees growing in natural habitats undergoing long duration exposure at concentrations as low as 130 to 155 µg/m3 were reported to significantly diminish their ring width (Freyer, 1979).

Away from the Rouyn–Noranda region, the ambient SO2 concentrations at rural and/or remote monitoring sites for air quality were ranging between 0.8 to 6.6 µg/m3 in 1998 (Environment Canada, 2002). Recently, Banic et al. (2001) have measured the concentrations of metals, particulate matter, and gases in the emission plume at different distances from the Horne smelter in Rouyn–Noranda. In February 2000, SO2 concentrations in the plume ranged between 315 and 660 µg/m3 at 10 km from the smelter (Daggupaty et al., 2001). These values imply that during the 30 yr of peak smelter activities, between 1955 and 1985, the atmosphere at the test site, which is 9 km northeast from the smelter, had summer SO2 concentrations ranging between 235 and 2100 µg/m3. Calculations are based on the fact that (i) the amount of emissions in 2000 corresponds to reduction of 85% since 1965, the year of peak emissions (Noranda Inc. [Toronto, ON, Canada] data) and (ii) in Canada, SO2 concentrations in summer air are generally 50% lower than in winter air (Environment Canada, 2002). Current summer concentrations at 100 km, estimated using plume dispersion modeling by Daggupaty et al. (2001) range between 80 and 735 µg/m3, given that sulfur dioxide behaves like fine particles of which more than 33% reach 100 km from the smelter (C. Banic, personal communication, 2001). The plume follows the seasonal dominant wind direction from southwest over the four months of the growth season. Fumigation was practically continuous over the years of smelter operation and the trees, even at distances up to 116 km along the transect, were exposed to SO2 concentrations much higher than the remote background levels. Therefore, the calculated current SO2 concentrations at 100 km toward the northeast are above the U.S. Annual Primary Air Quality Standard of 80 µg/m3 (USEPA, 1999) and well above the Canadian Annual Maximum Desirable National Ambient Air Quality Objective of 30 µg/m3 (Barker and Barker, 1988).

Numerous experiments have demonstrated that trees undergoing SO2 fumigation and showing higher relative {delta}13C values reduce significantly their stomatal aperture as well as their net CO2 assimilation (e.g., Martin et al., 1988; Matyssek et al., 1992). Therefore, we suggest that the net CO2 assimilation was also reduced in the trees of Rouyn–Noranda exposed to SO2. Using atmospheric CO2 concentrations from the literature, the equation of Farquhar (see Carbon Isotope Fractionation during Photosynthesis and Calculation of Stress Effect, above), and our {delta}13C results (Fig. 2), we can estimate the changes in foliar CO2 pressure in those trees. It should be specified that this sort of calculation only evaluates the relative proportion of CO2 used up at the leaf level (use). This proportion constitutes the highest degree of potential changes in net CO2 assimilation by trees (use minus emission) if the amount of CO2 emitted during respiration (emission) cannot be estimated. The more positive cellulose {delta}13C values of the test site trees during the smelter period could have been generated by a decrease in CO2 foliar internal pressure (pi), enzymatic fractionation (b), or both. Here we assume that b is constant and that pi/pa accommodates the major {delta}13C changes. During the peak activities of the smelter, a reduction of foliar uptake of around 30% relative to values obtained for contemporaneous background series is calculated; the reduction reaches 20% relative to the pre-smelter series at the test site (Fig. 9) . The resolution of 5 to 10 yr and high number of unpolluted tree ring series compiled from the literature generated their relatively smooth curve; the 2-yr resolution and the lower number (3) of investigated trees at the control site yield a "spiky" curve. As discussed above, this represents the highest degree of possible change in net CO2 assimilation that the test site trees could have undergone. Given the proposition of a reduction in net photosynthesis by trees around the Horne smelter, the significance of phytotoxic gases in terms of CO2 uptake at a larger scale deserves to be examined.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 9. Potential reduction percentage of photosynthesis through time at the test site, compared with the average percentage calculated for trees from unpolluted sites.

 
Effect on Carbon Dioxide Uptake by the North American Forest
The present atmospheric levels of SO2 in North America are generally under the critical threshold for plant health, but the levels exceed the recommended threshold near metal smelters and coal power plants, and automotive transportation and industrial activities in urban areas add to the annual SO2 budget (USEPA, 1999; Environment Canada, 2002). However, the SO2 emissions were much higher annually than they are now, between the 1940s and 1980s (i.e., before the implementation of environmental controls). The effect of higher SO2 concentrations very likely translated into a lower CO2 consumption by the boreal forest at the time near SO2 sources. At the present time there are hundreds of thousands of SO2 point sources unevenly distributed on the continent as a result of increasing needs for energy production and metal transformation (Environment Canada, 1999, 2002). As a consequence, atmospheric SO2 levels are still significantly higher than preindustrial natural levels (between 10 and 1000 times higher depending on which area is considered; Pham et al., 1996). Moreover, the O3 concentrations, which have also a direct negative effect on photosynthesis, separately or in combination with SO2, are increasing at a rate of 0.5 µg/m3 per year, a trend that could be amplified by global warming. The present atmospheric ozone concentration in the Northern Hemisphere is five times higher than it was in the preindustrial era (Mooney et al., 1999). The role of the forests as a sink in the global carbon cycle may, therefore, have been modified since the last century.

The annual net flux of CO2 from atmosphere to vegetation is estimated to be 1.4 x 109 Mg C (Schimel et al., 2001), therefore small changes in net carbon uptake by land vegetation could significantly modify the cumulative atmospheric CO2 concentration. The net photosynthetic reduction due to atmospheric pollutants such as SO2 and O3 could have contributed to the relative increase of atmospheric CO2. A conservative estimation of the atmospheric CO2 accumulation per year related to phytotoxic pollution is 0.03 x 109 Mg C, based on the distribution of point and area sources (Environment Canada, 1999; USEPA, 2002), the atmosphere–vegetation flux of Schimel et al. (2001), and the following assumptions: the distribution of sources in Asia is estimated to be equivalent to the one in North America; 15% of the north hemisphere forests reduced C uptake by 6% (i.e., in the same proportion than trees at our Site 5, 116 km from the source) and 5% of the forests reduced C uptake by 30%, in proportions equivalent to the Rouyn–Noranda sites located within a 15-km radius to the source.

The atmosphere–hydrosphere–biosphere exchanges in the modern carbon cycle need to be properly understood if the crucial problem of the missing sink for excess anthropogenic carbon dioxide emissions is to be solved. Recent reports on the missing sink search are contradictory, and either suggest that a land C sink, possibly the boreal forest, accounts for the leftover anthropogenic C (e.g., Ciais et al., 1995), or that the missing sink is elsewhere (e.g., Briffa et al., 1998; Schlesinger and Lichter, 2001). This debate and the concerns generated by our results both underline the apparent difficulty to quantify vectors of transfer between C reservoirs such as photosynthesis. Hence, the response of the complex biogeochemical processes to the changing atmospheric chemistry cannot yet be modeled or predicted adequately.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 REGIONAL SETTING
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
This combination of dendrochronological and dendroisotopic investigations of selected sites in the Rouyn–Noranda region and in a remote control site documents an unprecedented major effect of airborne pollutants on photosynthetic processes in natural stands. In addition, this study clearly demonstrates that (i) carbon isotope ratios of stem cellulose record changes in ambient atmospheric conditions; (ii) spatial and temporal isotopic trends reflect the pollution levels of ambient air; (iii) trees responded to phytotoxicity of SO2 and possibly of other pollutants by abruptly lowering discrimination against 13C, through stomatal closure and/or lower enzymatic fractionation; and (iv) exposed trees at 9 km from the point source significantly reduced their C uptake at the leaf level, estimated at 30% in comparison with contemporaneous unpolluted trees.

The inverse relation of average {delta}13Ccellulose values to distance of trees from the smelter suggests that even trees at Site 5, 116 km from the smelter, reacted to airborne pollutants. This suggests that the surface of forests affected by pollution might be much broader than previously thought. The pollution effects documented here potentially have implications for development of tree-ring chronologies to reconstruct the climate of the last century.

The findings of our study indicate that carbon dendroisotopic series constitute archives of the earth's atmospheric changes. Another important implication of this study is that high pollutant concentrations impair the capacity of the boreal trees to accommodate excess anthropogenic CO2. Accurately quantifying the various terrestrial compartments is a crucial step in modeling the global atmospheric carbon budget required to predict CO2–induced climatic changes.


   ACKNOWLEDGMENTS
 
We thank M.R. Luzincourt, M. Heinz, S. Boivin, J. Cloutier, M.-A. Dion, R. Gosselin, and Al. Smirnoff for technical assistance; R. Garrett and C. Banic for helpful discussions; B. Vigneault and R. Prairie of Noranda, Inc. for providing historical data on smelter feed and emissions and for commenting on an earlier draft of the manuscript; and three anonymous reviewers for their constructive comments. The first author wishes to sincerely thank Kevin Percy and Pierre Bernier of the Canadian Forest Service for thoroughly reviewing a pre-submission version of the article. GSC Contribution no. 2002128.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 REGIONAL SETTING
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES