Observations of the uptake of carbonyl sulfide ( COS ) by trees under elevated atmospheric carbon dioxide concentrations

Introduction Conclusions References


Introduction
Aside from sulfur dioxide (SO 2 ) carbonyl sulfide (COS) is the most abundant sulfur gas in the atmosphere with relative constant concentrations of 450-500 ppt and a lifetime of more than two years (Khalil et al., 1984;Mihalopoulos et al., 1991;Bandy et al., 1992;Barnes et al., 1994;Kjellstr öm, 1998;Montzka et al., 2007;Barkley et al., 2008).Due to this long lifetime, COS can be transported up into the stratosphere where it contributes to stratospheric ozone chemistry (Crutzen, 1976;Andreae and Crutzen, 1997).In times of low volcanic activity COS may serve as a supplier of sulfur to the stratospheric aerosol layer by conversion to sulfuric acid (Junge et al., 1961;Figures Back Close Full Crutzen , 1976) contributing to the backscattering of radiation energy into space.Thus the stratospheric cooling effect by the COS derived sulfate particles can be regarded to approximately cancel the warming tendency as caused by the direct radiative forcing by the trace gas COS within the troposhere (Br ühl et al., 2012).
The global budget of COS has been estimated as being balanced within the ranges of uncertainties (Watts, 2000;Kettle et al., 2002).However, this balance is a matter of debate both for the sources and the sinks, especially with regard to terrestrial vegetation which acts as the main sink for this trace gas and which is reported to be heavily underestimated (Notholt et al., 2003;Mu et al., 2004;Sandoval-Soto et al., 2005;Campbell et al., 2008;Suntharalingam et al., 2008;van Diest and Kesselmeier, 2008).This is valid for the Northern Hemisphere, whereas the Southern Hemisphere seems to be strongly influenced by the oceans (Montzka et al., 2007).
The biological background for the uptake of COS by vegetation is understood to be the combined action of the carboxylation enzymes Ribulose-1,5-bisphosphate carboxylase-oxygenase (Rubisco; EC 4.1.1.39),Phosphoenolpyruvate Carboxylase EC 4.1.1.31)and the key enzyme carbonic anhydrase (CA; EC 4.2.1.1)which were previously reported to be involved in the exchange of carbon dioxide (CO 2 ) and carbonyl sulfide (COS) (Protoschill-Krebs and Kesselmeier, 1992;Protoschill-Krebs et al., 1995, 1996;Schenk et al., 2004;Yonemura et al., 2005;Notni et al., 2007).This enzymatic model assigns a key role for CA and has been confirmed very recently by Stimler et al. (2011).Furthermore, the close relationship between COS and CO 2 uptake enhances discussion to use COS as a tracer for canopy photosynthesis, transpiration and stomatal conductance (Wohlfahrt et al., 2011;Seibt et al., 2010).The role of CA has also been demonstrated in case of lichens and soils (Kesselmeier et al., 1999;Kuhn and Kesselmeier, 2000;Van Diest and Kesselmeier, 2008), thus demonstrating the dominant role of this enzyme which is obviously also responsible for the toxicity of inhaled COS due to metabolization to hydrogen sulfide (Thiess et al., 1968;Chengelis and Neal, 1980).Of special interest within this context are recent findings about the identification of a CS 2 hydrolase acting similarly to carbonic anhydrase by splitting Introduction

Conclusions References
Tables Figures

Back Close
Full CS 2 into H 2 S and CO 2 in a thermophilic Archeon obtaining energy from reduced sulfur compounds (Smeulders et al., 2011).Changes in the enzyme's activities will have consequences for the exchange of CO 2 and COS between plants and the atmosphere.Elevated atmospheric CO 2 can initialize an immediate increase of photosynthetic CO 2 uptake.But on a long term basis this initial stimulation is often followed by a decline of photosynthesis which is obviously caused by a decrease of enzyme activities.Acclimation of Rubisco is well reported (Drake et al., 1997;Moore et al., 1999;Stitt and Krapp, 1999;Possell and Hewitt, 2009), however the mechanism of this kind of adaptation is a matter of debate (Rogers and Ellsworth, 2002).A decrease of Rubisco and PEP-Co activities would lead to a loss of COS uptake capacity related to these enzymes.In contrast, only a few reports are available for CA, though an adaptation of the key enzyme CA might have even stronger impact.High CO 2 levels caused an increase of the CA mRNA steady state level in Arabidopsis (Cervigni et al., 1971) whereas enzyme activities and their transcript levels were reduced in pea plants grown under elevated CO 2 (Majeau and Coleman, 1996).Also, the green alga Chlamydomonas reinhardtii adapts its CA activity to an increase in the environmental CO 2 level with a decrease in the enzyme activity (Spencer et al., 1983;Coleman et al., 1984).Overall, the acclimation of CA is not well documented but it can be predicted (Sage, 2002).Long term observations, however, are not known to us.Besides enzymatic acclimation, a reduction of stomatal conductivity under long term elevated CO 2 enrichment also contributes to the acclimation of photosynthesis (Herrick et al., 2004).Thus the growth of plants under elevated CO 2 may cause an adaptation to the CO 2 availability by reducing the stomatal uptake as well as enzymatic activities.Reduction of stomatal apertures will seriously affect the deposition of COS which is taken up through the stomata (Sandoval-Soto et al., 2005).A decrease in the CA activities as a consequence of elevated CO 2 will affect the metabolic COS consumption by plants as demonstrated earlier with the green alga Chlamydomonas reinhardtii which adapts to high CO 2 levels by decreasing its CA activity (Protoschill-Krebs et al., 1995).Introduction

Conclusions References
Tables Figures

Back Close
Full In view of our knowledge as briefly reviewed above, we may postulate the hypothesis that elevated CO 2 on the long term will lead to a decrease of enzymatic activities and thus to a shift of compensation points, which reflect the ambient concentration at which the consumption balances production resulting in a net flux of zero (Kesselmeier and Merk, 1993;Lehmann and Conrad, 1996;Simmons et al., 1999;Conrad and Meuser, 2000).Elevated CO 2 will trigger a decrease of the enzymatic activities which is balanced by a higher CO 2 availability.Thus, the CO 2 uptake will not decline, but a CA acclimation may lead to a reduction of the COS uptake due to a lower metabolic sink as long as the uptake is not also enhanced by higher substrate (COS) concentration.Furthermore, increased CO 2 without an increase of COS leads to a competitive inhibition of the COS consumption.Thus, changes of the COS uptake capacity should become visible in a shift of the compensation point.Therefore, we investigated the adaption of two European tree species, Fagus sylvatica and Quercus ilex, grown inside chambers under elevated CO 2 and determined the exchange characteristics and the content of CA after a 1-2 yr period of adaption from 350 ppm to 800 ppm CO 2 .

Plant material and growth
The tree species (3-4 yr old) investigated were holm oak (Quercus ilex L.) and European beech (Fagus sylvatica L.).From March 1998 to February 2000 the trees were grown in a greenhouse at 25 • C under a 12/12 h light-dark regime with a light intensity of 600 µmol m −2 s −1 of photons (PAR) and a relative humidity of 70 %.CO 2 concentrations were adjusted using pure CO 2 from commercially available cylinders and held constant at 800 ppm CO 2 (± 20 ppm), or at about 350 ppm (with some variation between 330 to 450 ppm).For details see Peuser et al. (1995) and Peuser and Wild (1996).Introduction

Conclusions References
Tables Figures

Back Close
Full 2.2 Enclosure system (cuvettes) and exchange measurements Measurements of COS exchange was time consuming and had to be spread over several days up to a few weeks to deal with.Table 1 gives an overview of the measurement schedule in order to note potential seasonal effects.Gas exchange of enclosed tree branches was investigated using a dynamic (flowthrough) Teflon-film-cuvette system consisting of a plant measuring and an empty reference cuvette.This cuvette system has been operated in previous studies (Sch äfer et al., 1992;Kesselmeier et al., 1993Kesselmeier et al., , 1996;;Kuhn et al., 2000;Sandoval-Soto et al., 2005).The system was designed for measurements of volatile organics and sulfur compound gas exchange in the laboratory as well as in the field and to have minimal effects on such trace gases.All experiments were performed inside a climate chamber with identical conditions as compared to the growth chamber.Trace gas sampling was accompanied by measurements of ambient CO 2 , CO 2 exchange and transpiration by an infra-red gas analyzer.COS and CO 2 mixing ratios were adjusted by mixing purified compressed air gas mixtures derived from a permeation device (Haunold, Germany) with COS permeation tubes (VICI Metronics, Santa Clara, California) and CO 2 from a pressurized bottle (Messer-Griesheim, Germany).For details see Sandoval-Soto et al. (2005).COS was quantified in the ppt range by an automated analytical system according to Von Hobe et al. (2008) by consecutive sampling at both cuvettes.The exchange rates (F ) were calculated according to the equation: considering the concentration differences between the sample and reference cuvette (∆c = c sample − c ref ) and the chamber flush rate (Q).All exchange rates were related to the enclosed leaf area (A).Leaf area was determined by a calibrated scanner system (ScanJET IICX with DeskSCAN II; both Hewlett-Packard, USA), and SIZE 1.10 (M üller, Germany).For details see Sandoval-Soto et al. (2005).Introduction

Conclusions References
Tables Figures

Back Close
Full

Statistical analysis
The linear relationship between substrate availability and the uptake rates was assessed statistically by the analysis of the Pearson correlation coefficient relating the COS concentration in the reference cuvette to the exchange rate.Further information was obtained regarding the R 2 of the regression analysis of the linear model: with F and c R indicating the exchange rate (dependent variable) and reference cuvette concentration of COS (independent variable), respectively.β 0 and β 1 reflect the regression coefficients and ε the residuals.
For further analysis, the linear model was extended by introducing the CO 2 concentration under which the trees were growing during the experiment (CO 2 ) accompanied by the interaction between c R and CO 2 (c R * CO 2 ) leading to the more complex model: with β 2 and β 3 again reflecting the corresponding regression coefficients.Again, R If the influence of CO 2 is significant, both groups within one data set (one tree) are different.Significance concerning the interaction indicates that both groups within the analyzed data set are significantly different if projected to the y-axis (F ).Finally, interaction may also be indicated in differences of the linear slope.Besides these analyses regarding the linear relationship of exchange rate and COS concentration, several mean value comparisons were performed by the two-sided Student's t-test.The null hypothesis that no difference exists between the two means µ 1 and µ 2 was tested against the alternative with an existing difference, i.e.:

Conclusions References
Tables Figures

Back Close
Full The group comparisons were carried out with each tree and for each measuring period comparing the influence of the two CO 2 concentrations.Thus, means of leaf conductance, CA and the deposition velocities of COS and CO 2 were compared.
All above mentioned statistical analyses were performed with SAS, Version 9.1.The differences of the compensation points were checked by the 95 % confidence intervals of the linear model 1 for F = 0 (Sigma Plot 11).

Leaf conductances and deposition velocities
The data as presented in this study can be discussed on a measurement period related basis and on a long term trend.The latter is, however, a very limited approach as the measurement schedule is biased by seasonal effects because of the labor-intensive and time consuming measurements spread over several weeks for each period.Tables 2-4 give an overview about the development of leaf conductances and deposition velocities of CO 2 and COS as observed in the course of the three year experiment.As expected, for measurements with beech in August 1998 and October 1999 and during the first series for the oak species (June-August 1998) leaf conductances (Table 2) for water vapor were found to be significantly different for plants grown under 350 vs. 800 ppm CO 2 as elevated CO 2 triggers a reduction of stomatal aperture.This observation is in close accordance with earlier interpretations suggesting a reduction in the stomatal opening as a main factor (Paoletti and Gellini, 1993;Ceulemans and Mousseau, 1994;Ainsworth and Long, 2005).We consider the non-significance for the evergreen species, holm oak, within the "winter measurements" to be a seasonal effect.Similarly, the missing significance for the April/July data can be related to different physiological activities of the oak species in the course of this pair of measurements.Introduction

Conclusions References
Tables Figures

Back Close
Full The beech data for summer 1999 showing a significantly higher conductance under elevated CO 2 are not understood, but might be biased by plant development related to the earlier measurements in June as compared to the July measurements under 350 ppm CO 2 .We noted a higher transpiration with all three tree individuals investigated in this case (data not shown).Excluding the June/July 1999 measurements with beech, the leaf conductance data indicated a decreasing trend over time in relation to the growth regime.This behavior is in close accordance with Herrick et al. (2004) who reported a decrease of stomatal conductance for sweetgum leaves under CO 2 enrichment.CO 2 deposition velocities (V dCO 2 ) exhibited a more consistent behavior (Table 3).In all cases we found a much lower V dCO 2 for the measurements under elevated CO 2 .All differences were highly significant with p-values < 0.001.Contrasting the conductance data, a constant development related to elevated CO 2 by adaptation of old leaves or by modified new leaves could not be observed.However, a clear increase was found for beech growing at 800 ppm comparing the measurement period August/September 1998 with June/July 1999, which is in accordance with data for sweetgum as reported by White et al. (2010).
Contrasting the calculation of deposition velocities for CO 2 from the quotients of single measurement points (because of missing CO 2 variation) those for COS (Table 4) could be derived from the slope of the regression line from the plot of the exchange data against the reference gas phase COS concentration data.The results show that in general the V dCOS is lower under elevated CO 2 .However, these differences are not significant with p-values > 0.05 in the case of all beech data and for the spring data in the case of holm oak.A general adaptation trend, i.e. a development of V dCOS with incubation length was not observed for beech, whereas there might be a steady increase in the case of holm oak under elevated CO 2 .This development is in accordance with data observed with sweetgum (White et al., 2010) but contrasts with the behavior of loblolly pine trees as reported by the same authors.However, in our case, the third measurement period for the oak species was scheduled for a winter period, which limits a Introduction

Conclusions References
Tables Figures

Back Close
Full consistent interpretation.For the comparisons within each period we regard the decrease of V dCOS under elevated CO 2 as a consequence of competitive inhibition of the responsible enzymes for COS/CO 2 uptake by the higher number of CO 2 molecules.
As elucidated in the introduction, the uptake of COS is based on the consumption by the enzymatic triad Rubisco, PEP-Co, and carbonic anhydrase (CA) with CA being the key enzyme.Table 5 gives an overview about the amount of CA activity measured within the leaves of the tree individuals growing under elevated CO 2 .Differences in the CA activity were found in the case of holm oak in April/May 1999 and December 1999/January 2000, although these differences were not significant.However, this seems to indicate that long term adaptation may lead to a decrease of CA activity under elevated CO 2 which fits into the overall picture that acclimation of CA can be expected (Sage, 2002).Furthermore, the difference between European beech and holm oak is striking.The oak exhibits a three times higher amount of CA.

Compensation point for COS
An increase of the COS compensation points may be understood as a decrease of metabolic consumption, depending on substrate availability and enzymatic activities.Hence, an increasing compensation point may be understood as a decrease of carbonic anhydrase (CA) activity.As we observed a potential for decrease in the case of Quercus ilex (see above), an analysis of the flux data became highly interesting.Although not significant, CA activity after growth under high CO 2 levels tends to be lower than under normal levels (Table 5) though this adaptation seemed to decrease in the second year.It has to be stated here that we did not observe any emission of COS under our experimental conditions with 350 ppm CO 2 and only a few data points under elevated CO 2 .Hence, we are referring in this paper to a "virtual" compensation point, i.e. the intersection point of the extrapolated regression line of linear model Introduction

Conclusions References
Tables Figures

Back Close
Full for two consecutive measurement periods 1999 and 1999/2000 (Fig. 1; Table 6).European beech (Fagus sylvatica) showed a similar trend to the growing conditions within the measurement period September/October 1999, however the confidence intervals still overlap indicating non-significance.Unfortunately, no data could be reported for a second year for this species due to limited growth and measurement capacities.Nevertheless, it can be noted that the compensation point for beech growing under 350 ppm CO 2 decreases whereas it increases under elevated CO 2 .Table 6 gives an overview of the ranges of compensation points as derived from the regression studies; given are the intersections of the regression line with the x-axis plus the ranges of the 95 % confidence level.The data provide evidence that holm oak adapts to elevated CO 2 levels by shifting compensation points indicating a decrease of the COS uptake capacity induced by high CO 2 levels under long-term conditions.Beech however exhibits only a trend but supports a similar interpretation.

Statistical significance of the differences between flux data sets
Tables 7-8 present an overview on the correlation and regression analyses performed.
As indicated (Table 7) by a Pearson Correlation Coefficient (P c ) < −0.7 except for the data set with beech in 1998 (800 ppm CO 2 ) we observed a strong linear relationship between the exchange flux (F ) and the initial COS concentration as determined within the empty reference cuvette (c R ).Even for the exception, Fagus sylvatica at 800 ppm, with a P c of −0.58 we also detected a correlation.This result demonstrates that the linear model 1 is able to describe the variances well.
Adding the long term growth regimes (CO 2 concentration) as described by linear model 2 (Table 8) R 2 drastically changes.As expected, the new values lie between those separated according to their growth regime (see Table 7).The best description was found in the case of holm oak for the year 1999 (R 2 = 0.91) followed by holm oak in the year 2000 (R 2 = 0.88).For all other data sets the variances are within the range of 51 to 83 %.Regarding the type III SS values, only in the case of holm oak (1999) could a highly significant difference between the 350 to 800 ppm regime be Introduction

Conclusions References
Tables Figures

Back Close
Full found (p < 0.001).In the case of all other measurement sets such an adaptation could not be statistically proved, though sometimes a trend may be discussed.However, often both data clouds overlap at the start of the incubation and in the case of beech during the whole observation time (Fig. 1).All together we may summarize that a statistically sound difference between the exchange behaviors of trees growing under elevated as compared to normal CO 2 was only found for the holm oak after one year of adaptation.In this special case the linear slope is nearly identical indicating very similar deposition velocities, whereas the two other oak data sets exhibit significantly different slopes.For beech trees identical slopes cannot be excluded because of the large p-values.

Global impact
Deposition velocities (V d ) are key for calculating fluxes and for estimating COS uptake versus CO 2 uptake rates to derive GPP related global sink estimates according to Sandoval-Soto et al. (2005).With this approach the flux of COS is related to the uptake of CO 2 as described by the following equation: Thus, the COS/CO 2 V d ratios as derived from our measurements allow the assessment of the effect of elevated CO 2 on vegetation acting as a sink for COS.We performed an Introduction

Conclusions References
Tables Figures

Back Close
Full analysis in close accordance with our previous work (Sandoval-Soto et al., 2005) taking into account the V d values as given in Tables 3 and 4. Table 9 summarizes the result.As these data were derived from the slope of the linear model 1 (linear regression) the new V dCOS data are integrating over the whole range of COS concentrations, contrasting the older calculation based on single data (Flux/concentration) ratios.These new COS deposition velocities were found to range a little higher causing a higher COS/CO 2 V d ratio.
Consequently, the COS sink strength for those ecosystems with trees as measured in the present study (Quercus ilex and Fagus sylvatica) rises from 0.397 to 0.688 Tg (maximum) and would cause a total increase from 1.404 to 1.696 Tg a −1 (maximum) based on NPP, both values to be doubled for GPP.For discussion of the effect of elevated CO 2 we take only those ecosystems into consideration with the two tree species as major contributors, i.e. temperate evergreen and temperate deciduous forests, woodland and scrubland, savannah and desert and semi desert scrub.Furthermore, we assume that the GPP is not altered because of physiological adaption (decrease of enzymatic activities and stomatal aperture).With this approach, we estimate a decrease in the COS sink strength from 0.367-0.687to 0.337-0.542Tg a −1 , representing a decrease of 8-21 %.Based on the few tree species investigated under elevated CO 2 so far this approach has to be regarded as very preliminary.But it stresses how elevated CO 2 might affect the global COS budget and balance.Increase of CO 2 levels, impacting the enzymatic adjustment (CA, RUBISCO, PEP-CO) of plants may cause a decrease of COS uptake as indicated by the V d based estimates and the potential shift of compensation points.As a consequence, the atmospheric COS level may rise and cause an increase of the direct radiative forcing by this trace gas, which is however counterbalanced by the cooling effect of the COS derived stratospheric sulfate aerosol (Br ühl et al., 2012).Introduction

Conclusions References
Tables Figures

Back Close
Full

Conclusions
Growth of two European tree species under elevated CO 2 for nearly two years resulted in some significant changes of the exchange patterns for CO 2 and COS which support the hypothesis that elevated CO 2 may lead to a reduction of the COS uptake capacity.
Beech for example exhibits a significant decrease in leaf conductance under elevated CO 2 , accompanied by a significant decrease of V dCOS for summers 1998 and 1999.
In contrast, holm oak exhibits no significant decrease of leaf conductance for 1999 and the winter values as well, but nevertheless a significant decrease of V dCOS in all cases.Within this picture, it seems to be of interest that holm oak was found to have a 3-4 times higher CA activity and appears to be more sensitive against adaption under elevated CO 2 .Furthermore, in the case of holm oak, a significant shift of the COS compensation point was found.Such a shift may be regarded as a result of a complex mixture of triggers such as substrate availability depending on changes of gas concentrations and leaf conductances, and the influence of enzymatic activities depending on the amount of the enzyme and its availability for the corresponding metabolic step.In the case of the triad CA, Rubisco and PEP-Co, not only a decrease of enzyme activity, but also changes caused by competitive inhibition (increasing CO 2 ) can be expected.
The results support the hypothesis that an adaptation of plants to a higher CO 2 level by decreasing their enzymatic capacity for the CO 2 exchange will affect the COS uptake.Consequently, a visible decrease of the metabolic capacity for consuming COS could be related to an increase of the compensation point.The data presented in our study support this hypothesis though the data base with two tree species is limited, our study was too short, and was biased by plant development due to the time consuming measurements.Furthermore, it is an open question whether this change in COS uptake is caused only by a decrease of CA activity, or also by adaptation of other enzymes such as PEP-carboxylase and Rubisco.Answers to these questions may be found for example by continuous field investigations within Free Atmospheric Carbon Enrichment (FACE) sites, which offer advantages such as more natural conditions as compared

Conclusions References
Tables Figures

Back Close
Full to growth chamber incubation.FACE sites are suitable for more continuous and simultaneous measurements to investigate the relationships between exchange fluxes, atmospheric concentration and incubation history as well as plant physiological background to determine exchange regulations and metabolic capacities.Furthermore, modern online analysis techniques for COS determination (Stimler et al., 2010(Stimler et al., , 2011) ) will be able to add insight into these exchange processes in real time.

Conclusions References
Tables Figures

2
provides information about the quality of the model.Furthermore, p-values indicate the significance for the triggers CO 2 and c R * CO 2 .Here the values of type III Sum of Squares (SS) are taken into account.
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Again, corresponding p-values indicate the significance of the results.A p-values less than 5 % indicates a significant difference.
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Deposition velocity of CO 2 in mm s −1 M COS Molar mass of COS in ambient air (600 ppt equivalent to 1451 ng m −3 at 25 • and 1 atm) M CO 2 Molar mass of CO 2 in ambient air (350 or 800 ppm equivalent to 620 or 1419 mg m −3 at 25 • and 1 atm).

Fig. 1 .
Fig. 1.Linear regression analysis of the relation between the initial COS concentration (c R ) and the uptake by European beech (Fagus sylvatica) and holm oak (Quercus ilex) growing under two different CO 2 regimes (350 and 800 ppm) beginning March 1998 and measured at the indicated time periods.Given are the regression lines (continuous lines) together with their 95 % confidence bands (broken lines).

Table 1 .
Schedule of measurements of trees grown constantly under the indicated CO 2 regimes beginning March 1998.Three individuals of each tree species in each growth regime were consecutively measured.

Table 3 .
CO 2 deposition velocities (V d ; mm s −1 ) as found with trees grown under the indicated CO 2 regimes derived as mean value from the quotient Exchange rate/concentration.Incubation under the indicated CO 2 concentration started in March 1998.Significance of differences of the V d mean values between 350 and 800 ppm growth regimes is indicated by p-values according to a two-sided Student's t-test (SAS Version 9.1).p-values > 0.05 indicate non-significant differences.

Table 5 .
Carbonic anhydrase activities expressed in non-dimensional units(Wilbur and Anderson, 1948)as derived from the pH drops over time with and without enzyme as found with trees grown under the indicated CO 2 regimes.Incubation under the indicated CO 2 concentration started in March 1998.Significance of differences of mean values between 350 and 800 ppm growth regimes is indicated by p-values according to a two-sided Student's t-test (SAS Version 9.1).p-values > 0.05 indicate non-significant differences; "nd" indicates no data.