Elevated atmospheric CO2 causes seasonal changes in carbonyl emissions from Quercus ilex


Author for correspondence:Jürgen Kreuzwieser Tel: +49 761203 8311 Fax: +49 761203 8302 Email: kreuzwie@uni-freiburg.de


  •  The effect of elevated atmospheric CO2 on the carbonyl emissions of leaves from two Mediterranean oak species (Quercus ilex and Q. pubescens) was analyzed under field conditions.
  •  Physiological and meteorological parameters were determined in parallel with measurements of carbonyl emissions. Gas exchange was quantified in dynamic cuvettes combined with an infrared gas analyzer.
  •  Acetaldehyde and acetone emissions from leaves of Q. ilex were enhanced by elevated CO2 in the autumn (from 14–40 nmol m−2 min−1 and from 2–8 nmol m−2 min−1, respectively), but not in the summer. No significant effects were found for leaves of Q. pubescens. The effects of CO2 on Q. ilex were mainly a result of decreased emissions by control trees under ambient CO2 concentrations in the autumn; emissions from trees exposed to elevated CO2 remained at a high level.
  •  Elevated atmospheric CO2 causes autumnal changes in carbonyl emissions from Quercus ilex. These effects suggest that the production of acetaldehyde and acetone depend on developmental factors. It is not yet clear whether the altered carbonyl emissions are a unique feature of Q. ilex.



2,4-dinitrophenylhydrazine coated silica gel cartridge




volatile organic compound


Vegetation emits a wide range of volatile organic compounds (VOCs) (Kesselmeier & Staudt, 1999), including a large number of oxygenated compounds, into the atmosphere (Fehsenfeld et al., 1992). It is estimated that the emission of VOCs other than methane and isoprenoids, including oxygenated species such as carbonyls, amounts to c. 24% of the total VOC budget in forest ecosystems (Guenther et al., 1994, 1995). Recent studies have indicated that trees, in particular, may act as significant carbonyl emitters (Hahn et al., 1991; Kesselmeier et al., 1997; Kreuzwieser et al., 1999a, 2000, 2001; Janson & de Serves, 2001) by releasing these compounds through their leaves. Under particular conditions carbonyl emission rates can equal the highest rates of isoprene emissions reported (Holzinger et al., 2000; Kreuzwieser et al., 2000).

Together with isoprenoids, carbonyls play a decisive role in the oxidative chemistry of the atmosphere (Thompson, 1992). They may alter the concentrations of hydroxyl radicals and, as a consequence, the lifetime of radiatively active trace gases, including methane, may increase (Brasseur & Chatfield, 1991). Carbonyls such as acetaldehyde, formaldehyde and acetone are also involved in the production of tropospheric ozone and peroxyacylnitrates (PANs) that both negatively affect plant growth and human health (Kotzias et al., 1997; Sakaki, 1998). In addition, the oxidation of carbonyls, particularly of acetaldehyde and formaldehyde, by HO2-radicals causes the generation of acetic and formic acid, which are assumed to contribute significantly to atmospheric acidity (Bode et al., 1997; Kesselmeier et al., 1997; Kotzias et al., 1997).

Due to anthropogenic CO2-emissions, atmospheric CO2 concentration is continuously increasing, which will result in a doubling of the present concentrations within this century (Houghton, 1997). One major goal of present research activities on plants is therefore to study the effects of elevated atmospheric CO2 at all scales, from the physiology of the whole plant up to the ecosystem. However, although the importance of VOCs on tropospheric chemistry is known and the dependence of VOC emissions on environmental conditions has been shown for several VOC-emitting species, there are surprisingly few studies dealing with the effect of elevated CO2 concentrations on VOC emissions. Data available so far focus on the emission of isoprenoids (Monson & Fall, 1989; Loreto & Sharkey, 1990; Loreto et al., 1996; Tognetti et al., 1998; Constable et al., 1999; Loreto et al., 2001; Staudt et al., 2001). There is no information on the effects of CO2 on carbonyl emissions.

In the present study, for the first time, the emission of short-chain carbonyls was investigated with mature tree species grown naturally under elevated atmospheric CO2 concentrations in the field. A CO2 spring in Italy provided a unique opportunity to investigate the effect of life-long exposure to enhanced CO2 levels on the source strength and pattern of VOC emissions of Mediterranean oak species.

Materials and Methods

Experimental field sites

Field studies were performed at a well described site with a natural CO2 vent in Central Italy, Rapolano Terme/Bossoleto, Tuscany (Miglietta et al., 1993; Körner & Miglietta, 1994; van Gardingen et al., 1995). The site is located near Siena (43°17′N; 11°36′Ε) and is covered by a typical Mediterranean forest mainly composed of mature Quercus ilex and Q. pubescens. CO2 emissions occur from the bottom and on the flanks of a circular crater, 80 m in diameter and 20 m in depth. Diurnal CO2 concentrations and gradients are described by Schulte et al. (1999) and Schwanz & Polle (1998). The emitted gas consists of > 99% CO2. In addition, traces of H2S and SO2 are emitted. CO2 accumulates in the crater during the night but disperses early in the morning by convective turbulence so that a CO2 gradient develops from the vents to the outside of the crater. Lowest CO2 concentrations are found in the mid-afternoon whereas in the evening the CO2 concentrations increase again (Schulte et al., 1999). At the study sites selected, mean CO2 concentrations amounted to 600–1200 ppm (Körner & Miglietta, 1994). The H2S and SO2 concentrations measured directly at the vent showed diurnal patterns comparable to CO2 with higher levels during the night (means: H2S, 22 ppb; SO2, 12 ppb) than during day (H2S, c. 9 ppb; SO2, c. 6 ppb). The control site is within c. 1 km distance of the crater and is similar in vegetation, soil properties and climate. Measuring campaigns were conducted in June/July and September of the years 2000 and 2001. The gas exchange measurements were usually performed between 09 : 00 and 16 : 30 h.

Carbonyl sampling and determination

Carbonyl emissions by the leaves of mature trees were determined using a system consisting of two identical cuvettes with 0.5 l volume each of which was constructed of chemically inert perfluorethylenpropylen (FEP) plates and foils (200 µm; Dyneon GmbH, Burgkirchen, Germany) in order to minimize adhesion and reaction of carbonyls with cuvette materials. Between four and eight leaves from sun-exposed twigs were carefully placed into one of the cuvettes, whereas the other was kept empty serving as a control. Both cuvettes were flushed with ambient air at flow rates of 2–4 l min−1. These flow rates were chosen to maintain ambient gas concentrations in the plant cuvette as close as possible. During measurements, flow rates (MAS, Kobold, Germany), PPFD (LI-190SA, Li-Cor Inc., Lincoln, NB, USA), temperature and relative humidity (1400–104, Walz, Germany) as well as the concentrations of CO2 and water (Li-6262, Li-Cor inc.) were measured continuously. Carbonyls were sampled from the air leaving both cuvettes and were quantified by HPLC as 2,4-dinitrophenylhydrazine(DNPH)-derivative. For sampling, air from the cuvettes was pumped for 45 min through DNPH-coated silica gel-cartridges (Sigma, Munich, Germany) with flow rates of 1 l min−1. Carbonyls present in the air reacted with acidified DNPH forming the corresponding hydrazone, which was eluted from the cartridges with 2 ml acetonitrie (ACN) and then quantified by HPLC. For this purpose an aliquot of 100 µl of the eluate was injected into an HPLC-system (Beckman, Munich, Germany). Hydrazones were separated by reversed-phase HPLC (Octa-Decyl-Silicium-column; C-18, 5 µm, 250 × 4.6 mm, UltrasphereTM, Beckman, Munich, Germany) using an ACN/water gradient (20–80% ACN, 0–20 min) at a flow rate of 1 ml min−1. Carbonyl-hydrazones were identified by external and internal standard solutions (Sigma-Aldrich, Germany) and were quantified with a Diode Array detector (Module 168, Beckman, Munich, Germany) at 354 nm.

Carbonyl flux rates as well as the rates of photosynthesis and transpiration were calculated by the concentration differences from the leaf and the control cuvette by taking into account the flow rates through the cuvettes. All gas exchange parameters are based on the projected leaf area which was measured directly after the flux measurements.


The data obtained were subjected either to the Student’s t-test or to analyses of variance and multiple range tests (LSD) by ANOVA (SPSS for Windows, release 10.0, SPSS Inc.). Significant differences at a confidence level of 95% are indicated in the tables and figures.


The meteorological conditions at the control site and CO2 vent during the measuring campaigns were quite similar (Table 1). The only large differences were observed for relative humidities in autumn 2000, with higher values at the CO2 vent, and for temperatures in July 2000 (Q. ilex), with lower values (23 ± 1°C) at the control site than at the CO2 vent (32 ± 1°C). For both Q. pubescens and Q. ilex leaves, the rates of photosynthesis which were analyzed in summer and autumn 2001 did not show significant differences between trees grown and measured under ambient or elevated CO2 levels. The same was true for transpiration rates and stomatal conductance, which did not show differences between the two measuring sites during the same field campaigns. Transpiration rates were higher in summer than in autumn 2001, for both Q. ilex and Q. pubescens, probably as a result of the higher temperatures and light intensities.

Table 1.  Meteorological and gas exchange parameters of Quercus ilex and Quercus pubescens
 Q. ilexQ. pubescens
  1. Twigs with two–eight leaves at a control site (ambient CO2, Am) and a CO2 vent (elevated, El) were placed into cuvettes. After 30 min adaptation, carbonyls were sampled for 45 min (see Figs 1 and 2) and, simultaneously, gas exchange and meteorological data were determined continuously. For each 45 min-period means were calculated. Data given are means (± SD) of these 45-min means of four–eight trees each. Statistically significant differences at P < 0.05 between parameters at the control site and the CO2 vent were calculated with Student’s t-test and are indicated by different letters.

T [°C]  23 ± 3b    32 ± 5a    34 ± 2A    31 ± 1A    37 ± 3a    34 ± 6a    33 ± 1A    34 ± 2A
RH [%]  47 ± 16a    43 ± 24a    20 ± 2B    45 ± 2A    44 ± 9a    27 ± 6b    21 ± 1B    35 ± 3A
PPFD [µmol m−2 s−1]158 ± 277a  573 ± 522a  465 ± 183A  442 ± 68A  186 ± 77a1102 ± 576a  396 ± 148A  566 ± 293A
T [°C]  36 ± 9α    40 ± 3α    24 ± 5A    25 ± 5A    36 ± 6α    38 ± 6α    23 ± 1A    26 ± 7A
RH [%]  29 ± 18α    32 ± 13α    59 ± 19A    53 ± 18A    29 ± 13α    29 ± 12α    61 ± 12A    48 ± 23A
PPFD [µmol m−2 s−1]877 ± 625α  907 ± 454α  486 ± 569A  556 ± 648A1054 ± 518α  938 ± 547α  131 ± 37A  547 ± 462A
A [µmol m−2 s−1]  4.4 ± 2.1α  4.8 ± 3.4α  4.1 ± 1.8A  4.1 ± 2.2A    8.7 ± 3.7α    7.3 ± 1.9α  5.0 ± 2.7A  7.3 ± 5.3A
E [mmol m−2 s−1]  2.1 ± 2.0α0.85 ± 0.44β0.19 ± 0.24A0.23 ± 0.20A    2.0 ± 1.9α    1.9 ± 1.6α0.58 ± 0.69A0.67 ± 0.45A
g(H2O) [mmol m−2 s−1]  42 ± 29α    20 ± 12β    15 ± 12A    14 ± 12A    41 ± 26α    36 ± 21α    46 ± 54A    19 ± 17A

The main carbonyl species, emitted in considerable amounts by the leaves of Q. ilex and Q. pubescens, were acetaldehyde, formaldehyde and acetone with emission rates ranging between c. 1 nmol m−2 min−1 (acetone) and 45 nmol m−2 min−1 (acetaldehyde) (Figs. 1 and 2). In only one case (Q. ilex, summer 2000, elevated CO2) was a deposition of formaldehyde observed (not shown), whereas in all other cases carbonyls were emitted by the plants. For Q. ilex, acetaldehyde and acetone emissions at the control site followed seasonal changes (Fig. 1a,c) with higher emissions (c. 45 nmol m−2 min−1 and c. 12 nmol m−2 min−1) in summer and lower emission rates (c. 14 nmol m−2 min−1 and 2 nmol m−2 min−1) in autumn. By contrast, emissions at the CO2 vent did not show such changes and remained constant in both years at c. 40 nmol m−2 min−1 and 10 nmol m−2 min−1 for acetaldehyde and acetone, respectively, exceeding the level of the controls significantly in autumn by factors of c. 2–5. This pattern was observed in both years for acetaldehyde and acetone. Acetaldehyde emissions in Q. ilex, for example, amounted to 40 ± 26 (CO2 vent) and 16 ± 9 nmol m−2 min−1 (control site) in autumn 2000, and 43 ± 42 (CO2 vent) and 11 ± 8 nmol m−2 min−1 (control site) in autumn 2001; acetone emissions showed a higher interannual variability and were 20 ± 10 (CO2 vent) and 10 ± 3 nmol m−2 min−1 (control site) in autumn 2000 and 8 ± 3 (CO2 vent) and 1.7 ± 1.0 nmol m−2 min−1 (control). By contrast to acetaldehyde and acetone, no such effects were observed for formaldehyde emission (Fig. 1b). However, elevated CO2 seemed to reduce emission rates for formaldehyde in summer. In Q. pubescens leaves a similar tendency to that in Q. ilex was observed, although the differences were not statistically significant (Fig. 2a). Acetaldehyde emissions seemed to be slightly higher in summer than in autumn at the control site, but not at the CO2 vent. No clear seasonal or CO2 effects were obtained for formaldehyde and acetone emissions (Fig. 2b,c).

Figure 1.

Effect of long-term exposure to elevated CO2 on (a) acetaldehyde, (b) formaldehyde and (c) acetone emissions of Quercus ilex. In summer and autumn 2000 and 2001, four measuring campaigns were conducted at the field sites in Italy. Twigs with four–eight leaves of trees at the control site (ambient CO2, white bars) and the CO2 vent (elevated CO2, black bars) were placed into cuvettes and carbonyl emissions were determined after an adaptation time of 30 min by the 2,4-dinitrophenylhydrazine coated silica gel cartridge (DSC) technique. Data shown are means (± SD) of 9–16 trees each. Statistically significant differences at P < 0.05 were calculated with LSD under ANOVA and are indicated by different letters over bars.

Figure 2.

Effect of long-term exposure to elevated CO2 on (a) acetaldehyde, (b) formaldehyde and (c) acetone emissions of Quercus pubescens. As for Q. ilex, four measuring campaigns were conducted at the field sites in Italy in summer and autumn 2000 and 2001. Twigs with two–five leaves of trees at the control site (ambient CO2, white bars) and the CO2 vent (elevated CO2, black bars) were placed into cuvettes and carbonyl emissions were determined after an adaptation time of 30 min by the 2,4-dinitrophenylhydrazine coated silica gel cartridge (DSC) technique. Data shown are means (± SD) of 9–16 trees each. Statistically significant differences at P < 0.05 were calculated with LSD under ANOVA and are indicated by different letters over bars.


The present study is the first investigation of the effect of elevated atmospheric CO2 on the carbonyl exchange between mature Mediterranean tree species and the atmosphere. The most surprising results were the seasonally (autumn) higher acetaldehyde and acetone emissions from Q. ilex at the CO2 vent (compared to controls). The higher autumnal emissions from Q. ilex were a result of decreased emission rates in the controls compared to the summer; such decreases did not occur in trees grown at the CO2 vent. The differences in carbonyl emissions were due to elevated CO2 rather than to other factors since environmental conditions which may have influenced the exchange rates (Kesselmeier et al., 1997) were relatively constant during the flux measurements (Table 1). Since it is assumed that the carbonyl exchange between leaves and the atmosphere occurs via the stomata (Kesselmeier, 2001; Kreuzwieser et al., 2001), carbonyl flux rates may be correlated with physiological parameters such as photosynthesis, transpiration or stomatal conductance. The stomatal conductances measured in the present study were quite low (between 14 and 46 mmol m−2 s−1) probably due to the severe environmental conditions such as high temperature and low relative humidity (Table 1). However, they seem to be in a normal range since only slightly higher values (c. 100 mmol m−2 s−1) have been obtained for the same species but under much less severe conditions 20°C, VPD 0.7 kPa (Polle et al., 2001). A correlation between carbonyl emissions and stomatal conductance was not found in the present study (correlation analysis not shown). This finding is consistent with recent studies showing that acetaldehyde emissions are largely independent of stomatal aperture, but highly dependent on the production rate within the leaves and, as a consequence, by the leaf internal acetaldehyde concentrations (Kreuzwieser et al., 2001).

CO2 effects on VOC emissions have been described so far only for isoprene and monoterpenes with quite different results. Short-term exposures to elevated CO2 did not affect (Loreto et al., 1996) or even reduced isoprenoid emissions (Monson & Fall, 1989; Loreto & Sharkey, 1990). Long-term exposure reduced isoprene emissions in aspen, but caused higher emissions in oaks (Sharkey et al., 1991). Monoterpene emissions were found to be enhanced in Q. ilex (Staudt et al., 2001), but also to be reduced in the same species (Loreto et al., 2001) after long-term exposure to elevated CO2. Whereas altered isoprenoid emissions may be caused by the connection of CO2 assimilation and isoprenoid synthesis, this is unlikely for the modified acetaldehyde flux rates in Q. ilex found in the present study, since CO2 effects only occurred in autumn, but not in summer (Fig. 1). It may rather be assumed that the changed emissions in autumn may be triggered indirectly by developmental factors. The only known pathway for acetaldehyde synthesis in leaves is by oxidation of ethanol, which is produced if other plant organs are exposed to anaerobic conditions; ethanol is transported to the leaves via the transpiration stream (Kreuzwieser et al., 1999a,b, 2000, 2001). Therefore, it appears that CO2 can cause seasonal modifications of processes that contribute to ethanol production, such as prolonged growth in autumn with stimulated cambial activity (MacDonald & Kimmerer, 1991). Acetaldehyde may also arise from various stresses, as shown by Kimmerer & Kozlowski (1982), who fumigated leaves of trees with 200–1000 ppb SO2. However, it seems unlikely that SO2 caused the enhanced acetaldehyde emissions in the present study since SO2 and H2S emissions were quite low even when measured directly at the vent (6 and 12 ppb during the day, respectively). This assumption is supported by the lack of physiological effects, such as altered glutathione levels, in the same trees at the Bossoleto site (Schwanz & Polle, 1998; Marabottini et al., 2001).

In contrast to the situation for acetaldehyde, so far no data are available on the pathway of acetone biosynthesis in plants, whose emission was also affected by exposure to CO2. Recent studies with several woody species including European trees (Picea abies and Pinus sylvestris) indicated that acetone can be the dominant carbonyl compound emitted (Simpson et al., 1999; Janson et al., 1999; Janson & de Serves, 2001; Shao et al., 2001).

The production of formaldehyde in plants, emitted in the present study in considerable amounts from both oak species, is assumed to be closely correlated to methanol, which is synthesized via catalase in the peroxisomes (Halliwell & Gutteridge, 1989) or via pectin methylesterases during seed development (Obendorf et al., 1990), cell expansion, cell wall degradation (Nemecek-Marshall et al., 1995) and senescence processes (see Harriman et al., 1991). Methanol is then converted into formaldehyde (Hourton-Cabassa et al., 1998) and in a subsequent reaction bound to glutathione and oxidised to formate (Giese et al., 1994; Fliegmann & Sandermann, 1997). The latter reaction may have contributed to formaldehyde deposition in herbaceous plants (Giese et al., 1994; Fliegmann & Sandermann, 1997), but also in Q. ilex (summer 2000; data not shown) in the present study.

The emission rates found in the present study with Mediterranean trees (2–50 nmol m−2 min−1 which equals c. 0.8–14 ng g−1 leaf d. wt min−1) were similar to those previously observed with the same and other tree species. Acetaldehyde emissions by poplars amounted to c. 36 ng g−1 leaf d. wt min−1 (Kreuzwieser et al., 1999a), by Pinus pinea to 7–25 ng g−1 leaf d. wt min−1 (Kesselmeier et al., 1997) and by Quercus ilex to 7–15 ng g−1 leaf d. wt min−1 (Kesselmeier et al., 1997) whereas acetaldehyde emissions from flooded poplars were significantly higher (up to c. 280 ng g−1 leaf d. wt min−1) (Kreuzwieser et al., 1999a, 2000). Emission rates of formaldehyde (P. pinea and Q. ilex: 1.7–15 ng g−1 leaf d. wt min−1, Kesselmeier et al., 1997) and acetone (pine c. 0.3–17 ng g−1 d. wt min−1, Janson et al., 2001) were in a similar range.

The present study showed that the carbonyl exchange between trees and the atmosphere can be affected by elevated atmospheric CO2 in some species (Q. ilex), whereas other species do not show any CO2 effects. Future studies have to elucidate whether altered seasonal emissions are a specific feature of: only some Mediterranean species; and/or evergreen tree species. For this purpose, other Mediterranean trees but also trees from temperate forests under long-term exposure to elevated CO2 have to be studied.


The present study was financially supported by the European community under contract No. EVK2-CT-1999-000-42. The authors thank Dr A. Raschi, CNR, Firenze, Italy, for the possibility to perform the measurements at the CO2 vent in Rapolano.