Doubled volatile organic compound emissions from subarctic tundra under simulated climate warming

Authors

  • Patrick Faubert,

    1. Department of Environmental Science, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland
    Search for more papers by this author
  • Päivi Tiiva,

    1. Department of Environmental Science, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland
    Search for more papers by this author
  • Åsmund Rinnan,

    1. Quality & Technology, Department of Food Science, Faculty of Life Sciences, University of Copenhagen, Rolighedsvej 30, DK-1958 Frederiksberg C, Denmark
    Search for more papers by this author
  • Anders Michelsen,

    1. Terrestrial Ecology Section, Department of Biology, Faculty of Science, University of Copenhagen, Øster Farimagsgade 2D, DK-1353 Copenhagen K, Denmark
    Search for more papers by this author
  • Jarmo K. Holopainen,

    1. Department of Environmental Science, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland
    Search for more papers by this author
  • Riikka Rinnan

    1. Department of Environmental Science, University of Eastern Finland, PO Box 1627, FI-70211 Kuopio, Finland
    2. Terrestrial Ecology Section, Department of Biology, Faculty of Science, University of Copenhagen, Øster Farimagsgade 2D, DK-1353 Copenhagen K, Denmark
    Search for more papers by this author

Author for correspondence:
Patrick Faubert
Tel: +358 40 355 3204
E-mail: patrick.faubert@uef.fi

Summary

  • Biogenic volatile organic compound (BVOC) emissions from arctic ecosystems are important in view of their role in global atmospheric chemistry and unknown feedbacks to global warming. These cold ecosystems are hotspots of climate warming, which will be more severe here than averaged over the globe. We assess the effects of climatic warming on non-methane BVOC emissions from a subarctic heath.
  • We performed ecosystem-based chamber measurements and gas chromatography–mass spectrometry (GC-MS) analyses of the BVOCs collected on adsorbent over two growing seasons at a wet subarctic tundra heath hosting a long-term warming and mountain birch (Betula pubescens ssp. czerepanovii) litter addition experiment.
  • The relatively low emissions of monoterpenes and sesquiterpenes were doubled in response to an air temperature increment of only 1.9–2.5°C, while litter addition had a minor influence. BVOC emissions were seasonal, and warming combined with litter addition triggered emissions of specific compounds.
  • The unexpectedly high rate of release of BVOCs measured in this conservative warming scenario is far above the estimates produced by the current models, which underlines the importance of a focus on BVOC emissions during climate change. The observed changes have implications for ecological interactions and feedback effects on climate change via impacts on aerosol formation and indirect greenhouse effects.

Introduction

The global annual rate of non-methane biogenic volatile organic compound (BVOC) emissions is estimated at 700–1000 Tg (1012 g) carbon (C), which represents over 90% of the total volatile organic compound emissions to the atmosphere (Guenther et al., 1995; Laothawornkitkul et al., 2009). BVOCs are important in atmospheric chemistry as their oxidation leads to the formation of secondary organic aerosols (SOAs; Laothawornkitkul et al., 2009). Moreover, the oxidative reactions between BVOCs, OH radicals and oxides of nitrogen form tropospheric ozone (Laothawornkitkul et al., 2009), one of the most important greenhouse gases after CO2 and methane (IPCC, 2007). BVOCs also compete with methane by scavenging OH radicals, which affects atmospheric methane concentration (Laothawornkitkul et al., 2009). Consequently, BVOC interactions with tropospheric ozone and methane affect climate change, but the feedbacks are still unknown (IPCC, 2007; Peñuelas, 2008). Thus, better knowledge of BVOC emission responses to abiotic factors related to climate change is needed.

A temperature dependence of monoterpene and sesquiterpene emissions has been described for a variety of plant species (Guenther et al., 1995; Duhl et al., 2008). However, most work has focused on the short-term effects of warming on BVOC emissions in laboratory-based facilities (Loreto et al., 1998; Peñuelas & Llusià, 2002) or on modeling (Guenther et al., 1993; Hakola et al., 2001; Helmig et al., 2007). The long-term impact of warming on BVOC emissions was recently studied in a subarctic ecosystem in field conditions (Tiiva et al., 2008). Tiiva et al. (2008) demonstrated that warming increased isoprene emission from the subarctic heath of the present study, but the effect on other BVOCs was not investigated. Emissions of non-isoprene BVOCs (e.g. monoterpenes and sesquiterpenes) have been considered in depth in modeling efforts focusing on BVOCs from the perspective of climate warming (Guenther et al., 1993; Hakola et al., 2001; Helmig et al., 2007). Models failing to consider non-isoprene BVOCs would generate major uncertainties in the extent of warming, as non-isoprene BVOCs are of great importance in the atmospheric chemistry that affects the climate system (Arneth et al., 2008).

Climate warming will be more severe at high latitudes than averaged over the globe. Mean annual temperatures in the Arctic (the area north of 60°N) have already increased by 2–3°C since the 1950s, and climate models project additional increases of 1°C by 2020, 2–3°C by 2050, 4–5°C by 2080 and up to 9°C by 2100 (ACIA, 2005; IPCC, 2007). By lengthening the growing season, these warmer conditions will bring significant changes to the ecosystem dynamics. Plant growth (Myneni et al., 1997) and the abundance of deciduous shrubs (Tape et al., 2006) have already increased. Moreover, the tree line of mountain birch (Betula pubescens ssp. czerepanovii) has moved to higher altitudes (Truong et al., 2007) and migrated northwards (Callaghan et al., 2004). These vegetation changes will lead to increasing leaf litter fall (Cornelissen et al., 2007) and consequently increasing amounts of nutrients in the soil that are available for plants (Rinnan et al., 2007, 2008).

Heaths are important ecosystems in the Arctic, occupying 1.282 × 106 km2, which is 23% of the total ice-free arctic land area (Bliss & Matveyeva, 1992). Subarctic mires are a significant BVOC source (Tiiva et al., 2007a; Bäckstrand et al., 2008), but the emissions from the subarctic heaths have been largely ignored until recently (Tiiva et al., 2008). Many of the typical heath species have already been recognized as BVOC emitters. These include the bryophytes, ericaceous dwarf shrubs, dwarf willows and some Carex species (Isebrands et al., 1999; Klinger et al., 2002; Rinnan et al., 2005; Tiiva et al., 2007a,b, 2008, 2009). Microbial decomposition in the underlying peaty soil is also known to release BVOCs (Beckmann & Lloyd, 2001).

Our aim was to assess the effects of climatic warming on emissions of non-methane BVOCs (excluding isoprene, which is investigated in detail in Tiiva et al., 2008) from a subarctic heath in Abisko, northern Sweden. We performed ecosystem-based chamber measurements and gas chromatography–mass spectrometry (GC-MS) analyses of the BVOCs collected on adsorbent over two growing seasons at a wet subarctic tundra heath hosting a long-term warming experiment, established 7 yr before the commencement of our measurements. We hypothesized, firstly, that simulated climatic warming, although conservative, would modify BVOC emissions by increasing monoterpene and sesquiterpene emissions (Guenther et al., 1995; Duhl et al., 2008; Hartikainen et al., 2009). Secondly, we expected that the mountain birch litter addition would change the emissions, either directly, as a consequence of emissions of some compounds from the litter itself (Isidorov & Jdanova, 2002; Leff & Fierer, 2008), or indirectly, as a consequence of altered microbial processes (Rinnan et al., 2007, 2008), causing the amount of some compounds to increase and that of others to decrease, resulting in a modified emission profile.

Materials and Methods

Study site and experimental design

The experimental site was a wet dwarf shrub/graminoid heath located in Abisko, northern Sweden (68°21′N, 18°49′E; 385 m a.s.l.). The climate is subarctic/alpine and the mean annual temperature and precipitation are −0.5°C and 315 mm, respectively. The vascular vegetation of the heath was mainly characterized by the sedge Carex vaginata Tausch and the evergreen shrubs Empetrum hermaphroditum Hagerup and Andromeda polifolia L. The ground layer was covered by mosses and lichens. The detailed vegetation composition of the heath is presented in Tiiva et al. (2008). The soil had a pH of 6.9 and the uppermost 5-cm layer was highly organic (c. 92% soil organic matter (SOM)), moist (gravimetric water content on average 343%) and had an average ammonium content of 1.2 μg g−1 SOM (Rinnan et al., 2007, 2008). Further soil characteristics have been reported previously (Rinnan et al., 2007, 2008).

The experiment simulating climate warming was established on the heath in 1999 (i.e. 7 yr before the commencement of the measurements) and has been maintained since then (Supporting Information Fig. S1a). The experimental set-up consisted of 24 field plots (120 × 120 cm) including a control, a warming treatment, a litter addition treatment and their combination in a factorial design, each replicated in six blocks (Rinnan et al., 2007, 2008; Tiiva et al., 2008). Warming was provided by open-top chambers (OTCs; Havström et al., 1993) covered with a transparent plastic layer acting as a glasshouse (Fig. S1b). OTCs were installed yearly on the field plots in mid-June and removed in late August. The litter addition treatment consisted of adding 90 g m−2 DW of air-dried mountain birch (Betula pubescens ssp. czerepanovii (N. I. Orlova) Hämet-Ahti) litter yearly in late August (Fig. S1c), which corresponds to the annual leaf litter fall of a nearby Betula pubescens ssp. czerepanovii forest (Bylund & Nordell, 2001). This treatment simulates the change in litter type and quality following the expected increase of deciduous species under climatic warming (Cornelissen et al., 2007) and the expansion of the birch forest surrounding the heath.

Sampling of BVOCs

BVOC emissions were sampled five times during the growing season of 2006 and eight times in 2007 (see Fig. 1 for sampling dates). All samplings were performed without any changes to the experimental set-up described in the previous section except on 17 July 2007. On that date, the OTCs were removed in order to investigate whether the effect of warming was direct and causing an instantaneous increase in the emissions or indirect and acting, for example, via altered vegetation composition. The sampling technique was a conventional push-pull system used for measurement of BVOC emissions from the whole plant/soil system (Fig. S1b; Tholl et al., 2006; Tiiva et al., 2007a, 2008; Ortega & Helmig, 2008). The air was sampled using a transparent polycarbonate chamber (thickness 1.5 mm; 23 × 23 cm; height 25 cm; Vink Finland, Kerava, Finland) placed on an aluminum collar installed in each field plot in June 2002. The collars were equipped with grooves filled with water during sampling to ensure that the chamber headspace was airtight. A small battery-operated sampling pump (12 V; Rietschle Thomas, Puchheim, Germany) pulled the air sample through an Automatic Thermal Desorption (ATD) steel tube (Perkin Elmer, Boston, MA, USA) filled with a combination of Tenax TA and Carbopack B adsorbents (100 mg of each; mesh 60/80; Supelco, Bellefonte, PA, USA).

Figure 1.

 Biogenic volatile organic compound (BVOC) emissions from a subarctic tundra heath subjected to warming and litter addition. (a, b) Mean (a) monoterpene and (b) sesquiterpene emissions (mean + SE; = 5–6) measured for the control, litter addition (L), warming (W) and combined treatments in the growing seasons of 2006 and 2007. The exact mean values of the measurements are presented in Supporting Information Table S1. The P-values from the linear mixed model analysis with W, L and time as fixed factors and block as a random factor are shown for each compound group. The complete linear mixed model analysis tables are presented in Tables S2–S5. One and two asterisks signify treatment effects at < 0.05 and < 0.01 within a date, respectively. Note different y-axis scales. The measurement date with open-top chambers (OTCs) removed, 17 July 2007, is indicated by an arrow, and was not included in the analysis done on the repeated measurements. (c) The mean daily (07:00–19:00 h) ambient air temperature, the mean chamber air temperature (± SE; = 11–12) and the mean photosynthetic photon flux density (PPFD) (± SE; = 22–24) during the measurements.

Air sampling for BVOCs lasted 30 min, representing a sampled air volume of 6 l. The outflow was set to 205 ml min−1 (corresponding to a flow of 200 ml min−1 through the sample tube) with a flow meter (Agilent Flow Tracker 1000; Agilent Technologies Inc., Wilmington, DE, USA). In order to prevent air leakage from outside into the chamber, a slight overpressure was maintained by pumping the inflow air at a rate of 215 ml min−1 (Staudt et al., 2000; Tiiva et al., 2008). The inflow air was not sampled for BVOCs, as the concentrations were considered to be negligible thanks to the purification system, consisting of a charcoal filter to remove BVOC contaminants and a MnO2 scrubber to remove ozone (Ortega & Helmig, 2008). During the sampling period, the chamber air was circulated with a small fan, and the air temperature, humidity (Tinyview Plus; Gemini Data Loggers Ltd, Chichester, UK) and photosynthetically active radiation (quantum sensor; Li-Cor, Lincoln, NE, USA), measured as photosynthetic photon flux density (PPFD), were monitored. Immediately after the sampling, the adsorbent tubes were sealed with Teflon-coated brass caps, and the samples were analyzed within 2 wk at the University of Eastern Finland.

Analysis of BVOCs

The samples were analyzed by GC-MS (Hewlett Packard 6890, MSD 5973; Hewlett Packard, Palo Alto, CA, USA) after thermodesorption at 250°C and cryofocusing at −30°C with an automated thermal desorber (ATD400; Perkin Elmer, Wellesley, MA, USA). BVOCs were separated using an HP-5 capillary column (50 m × 0.2 mm; film thickness 0.33 μm). The carrier gas was helium. The oven temperature was held at 40°C for 1 min, and then raised to 210°C at a rate of 5°C min−1, and finally further to 250°C at a rate of 20°C min−1.

BVOCs were identified according to the mass spectra in the Wiley data library, and quantified using pure standard solutions (Fluka, Buchs, Switzerland) according to total ion counts. The standards used for the quantification were α-pinene, sabinene, terpinolene, limonene, camphene, 1,8-cineole, copaene, trans-β-farnesene, humulene, aromadendrene, δ-cadinene, trans-2-hexenal, cis-3-hexenol and cis-3-hexenyl acetate. The compounds were classified into four groups (Table 1): monoterpenes, sesquiterpenes, other reactive volatile organic compounds (ORVOCs; compounds with a lifetime of < 1 d as a result of reactions with the OH radicals NO3 and O3; Guenther et al., 1995) and other volatile organic compounds (other VOCs; compounds with a lifetime > 1 d; Guenther et al., 1995). Quantification of the compounds for which a pure standard was not available was performed with the pure standard of α-pinene for monoterpenes, copaene for sesquiterpenes and cis-3-hexenyl acetate for the other compounds.

Table 1.   Mean (± SE) biogenic volatile organic compound (BVOC) emissions measured in an unmanipulated (control plots) subarctic tundra heath in the growing seasons of 2006 (= 30) and 2007 (= 42)
Compound (μg m−2 h−1)20062007
  1. Compounds are ordered according to their retention times in each group (monoterpenes; sesquiterpenes; other reactive volatile organic compounds (ORVOCs); other volatile organic compounds (other VOCs)).

  2. nd, not detected.

Monoterpenes
 α-Thujene nd0.02 (0.01)
 (+)-Pin-2(3)-ene nd0.21 (0.06)
 α-Pinene0.13 (0.05)0.13 (0.05)
 β-Phellandrene nd0.05 (0.02)
 Sabinene0.83 (0.32)2.41 (0.74)
 β-Myrcene nd0.19 (0.06)
 α-Terpinolene nd0.53 (0.31)
 Limonene0.25 (0.06)0.26 (0.08)
 Camphene nd0.01 (0.01)
 1,8-Cineole0.32 (0.16)5.83 (1.65)
 3,7-Dimethyl-1,3,6-octatriene nd0.13 (0.06)
 γ-Terpinene nd0.07 (0.02)
 Total monoterpenes1.54 (0.56)9.84 (2.67)
Sesquiterpenes
 Copaene1.06 (0.42)0.79 (0.37)
 Caryophyllene nd0.42 (0.17)
 Aromadendrene0.14 (0.05)0.04 (0.02)
 α-Farnesene nd0.08 (0.06)
 Cubebene0.43 (0.17)0.37 (0.19)
 Germacrene-d0.46 (0.20)0.33 (0.16)
 Selinene0.42 (0.13)0.70 (0.19)
 Humulene0.03 (0.03)0.09 (0.04)
 Eudesma-4(14),11-diene nd0.17 (0.09)
 Cadinene0.11 (0.05)0.22 (0.09)
 Selina-3,7(11)-diene0.06 (0.04)0.20 (0.07)
 Total sesquiterpenes2.72 (0.79)3.41 (0.96)
ORVOCs
 2-Methylfuran0.06 (0.03)0.14 (0.06)
 2-Heptene0.08 (0.03)0.25 (0.08)
 1-Octene nd0.06 (0.02)
 3-Hexenol nd0.08 (0.06)
 2-Hexenal2.35 (1.55) nd
 Xylene0.30 (0.19) nd
 m-Xylene0.41 (0.28) nd
 Styrene nd0.04 (0.01)
 Heptanal nd0.04 (0.02)
 1-Methyl-3-isopropylbenzene nd0.06 (0.03)
 Diisopropylnaphthalene0.08 (0.04) nd
 Dioctyl ether nd0.26 (0.13)
 Total ORVOCs3.27 (1.58)0.93 (0.23)
Other VOCs
 Methylcyclopentane0.24 (0.11) nd
 Hexane1.50 (0.47) nd
 Toluene0.97 (0.21)0.43 (0.07)
 Benzoic acid0.66 (0.55) nd
 Total other VOCs3.38 (0.83)0.43 (0.07)
Total BVOCs10.90 (2.55)14.61 (3.51)

The chromatograms were analyzed using the Enhanced ChemStation software (G1701CA Version C.00.00 21 December 1999; Agilent Technologies, Santa Clara, CA, USA), and information was then extracted and sorted using an in-house function. The data set included compounds that appeared in at least 10% of the measurements and had an identification quality (in the Wiley data library) of > 90%. The compounds 1,4-dioxane, chlorobenzene, methoxy-phenyl oxime, phenol, and nonanal, which qualified for the data set, were not included in the analyses as they were at least partly impurities originating from the sampling system.

The emission rates were calculated using the formula:

image

where the BVOC mass in the ATD tube is divided by the air flow rate and the sampling time, and then multiplied by the chamber volume to obtain the absolute amount of BVOCs in the headspace (Tiiva et al., 2008). The soil surface microtopography was taken into account when determining the chamber headspace volume (on average 14.1 l). The emission rate of BVOCs was finally divided by the surface area of the plot and multiplied by two to give the emission rate per hour.

Statistical analyses

The effect of the warming and litter addition treatments on the quantified BVOC emissions in repeated measurements was tested using the linear mixed model procedure of spss package 14.0 for Windows (SPSS Inc., Chicago, IL, USA). The model included warming, litter addition and time as fixed factors and block as a random factor, with the factor time taking the repeated measurements on the same experimental units into account. If the warming, the litter addition treatment or their interaction had a significant effect in the analysis done on the repeated measurements, linear mixed models were run separately on the measurement dates in order to determine when the treatment had a significant effect. The field plot was used as the unit of replication (= 6).

Principal component analysis (PCA) was performed on the peak areas of the BVOC emissions using simca-p software version 11.5 (Umetrics, Umeå, Sweden). PCA was performed to assess how the treatments affected the BVOC emission profiles, that is, the relative amounts of the different compounds emitted, and how emissions were associated with each other to characterize the treatment effect and potential emission seasonality. PCAs were performed on the repeated measurements and on each date separately. Outliers were removed and the principal components (PCs) were extracted for each model after a unit variance scaling of the variables. The scores of each PC generated by the PCAs were then analyzed for treatment effects by the linear mixed model procedure in spss as described above.

Results and Discussion

Warming doubles monoterpene and sesquiterpene emissions

The warming treatment increased the air temperature in the chamber headspace by 1.9 and 2.5°C (mean increase compared with the non-warmed plots) during the measurements in 2006 and 2007, respectively. The experimental warming consistently increased total emissions of monoterpenes (12 compounds) and sesquiterpenes (11 compounds) throughout the measurement periods (Fig. 1; Supporting Information Table S1). The increases in emission were 85% (2006) and 120% (2007) for monoterpenes and 130% (2007) and 250% (2006) for sesquiterpenes, averaged over the growing seasons. These increases are much greater than expected based on the model fitting an exponential relationship between emission rate and temperature (Guenther et al., 1993). The model yields an 18–25% increase in emission for monoterpenes and a 37–60% increase for sesquiterpenes for a 2°C temperature increase, using a β coefficient of 0.09°C−1 for monoterpenes (Guenther et al., 1993) and 0.17–0.19°C−1 for sesquiterpenes (Hakola et al., 2001; Helmig et al., 2007). The β coefficients inferred from the averages of the measurements from the non-warmed vs warmed plots were 0.33 and 0.32°C−1 for monoterpenes, and 0.67 and 0.34°C−1 for sesquiterpenes in 2006 and 2007, respectively. It therefore appears that the temperature sensitivity of monoterpene and sesquiterpene emissions from subarctic plant communities is much greater than that predicted by current models (Guenther et al., 1993; Hakola et al., 2001; Helmig et al., 2007). The warming-induced increases in monoterpene and sesquiterpene emissions at the plant community-scale were also greater than the increase in isoprene emission in response to the same experimental treatments (Tiiva et al., 2008). In contrast to monoterpenes, sesquiterpenes and isoprene, total emissions of ORVOCs (12 compounds) and other VOCs (four compounds) were not affected by experimental warming (PORVOCs > 0.08; Pother VOCs > 0.16; Table S1).

Our results suggest that summertime sesquiterpene emissions from cold subarctic ecosystems could be much higher after a slight temperature increase than previously thought. Sesquiterpene emissions have been modeled to reach 16% of monoterpene emission rates during July for the contiguous USA (Sakulyanontvittaya et al., 2008), while we observed that sesquiterpenes represented 180% of the monoterpene emissions in 2006 and 30% in 2007. We anticipate further increases in sesquiterpene emissions from the Arctic following both climatic warming and the ongoing expansion of mountain birch into the treeless tundra (ACIA, 2005). More than 60% of the BVOC emissions from mountain birch foliage consist of sesquiterpenes (Mäntyläet al., 2008).

Growth at elevated temperatures increases the so-called ‘isoprene emission capacity’ of leaves (Sharkey et al., 1999; Pétron et al., 2001), and similar long-term effects on production rates and/or pool sizes of other terpenoids are possible. During the measurement without OTCs (17 July 2007), there were no significant differences in the emissions of any BVOC groups between the OTC and control plots (Fig. 1a,b; Tiiva et al., 2008). This indicates that, although the biomass of green vegetation had increased in response to both warming and litter addition (biomass estimated as the normalized differential vegetation index (NDVI); Rinnan et al., 2008), the biomass increase had not yet been translated into higher BVOC emission on a square meter basis; thus there appear to be no long-term effects of warming comparable to those for isoprene (Sharkey et al., 1999; Pétron et al., 2001). There is a possibility that NDVI overestimated the biomass increase in response to warming and litter addition. Nevertheless, this suggests that the effect of warming is direct and acts at the physicochemical level by increasing the volatility and diffusion gradient of monoterpenes and sesquiterpenes from storage or synthesizing pathways (Niinemets et al., 2004).

In general, we recommend sophisticated on-line methods for more accurate quantification of the natural variability of ecosystem emissions. The mean (± SE) monoterpene emissions from the measurements taken during the growing season on the unmanipulated heath were 1.5 ± 0.6 μg m−2 h−1 in 2006 and 9.8 ± 2.7 μg m−2 h−1 in 2007, while the mean sesquiterpene emissions were 2.7 ± 0.8 μg m−2 h−1 in 2006 and 3.4 ± 1.0 μg m−2 h−1 in 2007 (Table 1). The greater emissions in 2007 than in 2006 were probably a result of the higher ambient air temperature (11.0 ± 0.4°C in 2007 in contrast to 9.6 ± 0.3°C in 2006; Fig. 1c). For monoterpene emissions, the 6-fold increase was probably attributable to the contribution of extremely high emission of 1,8 cineole on 6 July 2007, when temperature and light were at their highest (Tables 1, S1; Fig. 1c). The emission of 1,8 cineole is strongly temperature and light dependent (Tarvainen et al., 2005).

Leaf litter is a minor source of BVOCs

Mountain birch leaf litter added to simulate the enhanced litter input expected from the expansion of birch on the subarctic tundra heaths (ACIA, 2005) did not appear to be a significant source of BVOCs. Nevertheless, in the combined warming plus litter addition treatment, there was a tendency toward an increase in monoterpene and especially sesquiterpene emissions (< 0.1 for litter effect and warming × litter interaction in 2006) on a few isolated dates (Fig. 1a,b). For those few cases, we suggest that litter decomposition and concomitant microbial release of BVOCs (Isidorov & Jdanova, 2002; Leff & Fierer, 2008) were accelerated by warming, and that the phosphorus fertilization of the soil by the added litter (Rinnan et al., 2007, 2008) alleviated plant nutrient limitation in the combined treatment, thereby increasing the emissions. Thus, the rich phosphorus status of the combined warming plus litter addition treatment plots has the potential to increase BVOC emissions in the long term. In addition, the increased abundance of mountain birch in the Subarctic under global warming (Callaghan et al., 2004) will increase the density of birch roots in soil, which could increase the BVOC emissions released by roots and microbial communities (Paavolainen et al., 1998; Beckmann & Lloyd, 2001).

BVOC emission associations are seasonal and affected by warming

The PCA revealed a significant seasonality in the BVOC profiles (i.e. the relative amount of the individual compounds emitted), where the emissions of specific compounds were associated with each other to characterize the experimental treatments and a period of the growing season (Figs 2, 3). In 2006, higher emissions of hexane, xylenes, methylcyclopentane and toluene were associated with the beginning of the growing season. By contrast, humulene, selinene and selina-3,7(11)-diene emissions were higher at the end of the growing season (Fig. 2a). Plant phenology might have been partly responsible for this trend over the measurements in the growing season of 2006 (Llusià & Peñuelas, 2000), where leaf budding and development could have influenced the terpenoid emissions (Hakola et al., 1998, 2001). In 2007, 6 July – the date with the highest air temperature (Fig. 1c) – was responsible for most of the variation among the samples and thus also had the greatest influence on the PCA (Fig. 2b). The high air temperature was associated with higher emissions of camphene, limonene, α-thujene, β-myrcene, γ-terpinene and 1,8-cineole. This clearly shows the dependence of terpenoid production on temperature (Guenther et al., 1995).

Figure 2.

 Principal component analyses of the biogenic volatile organic compound (BVOC) emission profiles obtained from measurements in the growing seasons of (a) 2006 and (b) 2007. The mean scores of the principal components (PCs; ± SE; = 5–6) for the control, circle; litter addition, square; warming, triangle apex up; and combined treatments, triangle apex down; and the loading variables are presented. Sampling dates in 2006: open white, 13 June; dotted white, 18–19 June; crossed white, 4 July; open gray, 25 July; dotted grey, 22 August. Sampling dates in 2007: open white, 14 June; dotted white, 17 June; crossed white, 6 July; open gray, 10–11 July; dotted grey, 16 July; crossed grey, 5 August; black, 30 August. The variation explained by each PC is shown in parentheses. The complete linear mixed model analysis table of the PC scores is shown in Supporting Information Table S6.

Figure 3.

 Principal component analysis of the biogenic volatile organic compound (BVOC) emission profiles obtained from measurements on 25 July 2006. The mean scores (± SE; = 5–6) of principal component (PC) 1 and PC2 for the control, litter addition, warming and combined treatments and the loading variables are presented. The variation explained by each PC is shown in parentheses. The complete linear mixed model analysis table of the PC scores is shown in Supporting Information Table S7.

The PCA also revealed that the blend of compounds emitted was altered by warming (< 0.05 for warming effect on PCs on several dates; see Fig. 3 for an example). From the beginning of July onwards, the combined warming plus litter addition treatment in 2006 and both treatments with warming in 2007 were characterized by relatively high emissions of the sesquiterpenes selinene, humulene and selina-3,7(11)-diene (Fig. 3), while the other treatments showed no common associations throughout the measurements. The relatively high selinene emission may mainly have originated from the fungi present on the litter (Jelén et al., 1995; Jelén, 2002). By contrast, humulene may have been released from the mountain birch leaf litter itself, as living foliage emits humulene (Mäntyläet al., 2008).

Increased BVOC emissions under climate warming have implications for biological interactions and atmospheric chemistry

Our results show that BVOC emissions from the subarctic tundra, which is currently exposed to substantial climate change (ACIA, 2005; IPCC, 2007), are more temperature sensitive than the commonly used temperature dependence models (Guenther et al., 1993; Hakola et al., 2001; Helmig et al., 2007) suggest. Our findings of increased quantities and altered compositions of BVOCs emitted from the tundra heath have implications for biological interactions and atmospheric chemistry. BVOCs function in plant-to-plant signaling and tritrophic interactions (Dicke, 2009). Furthermore, BVOC emissions contribute to the formation of SOA (Laothawornkitkul et al., 2009); the new aerosol particle formation above a subarctic area has been observed to peak during periods of high solar radiation in spring and summer (Komppula et al., 2003), that is, when BVOC emissions are also high. Nevertheless, if the proportion of isoprene increases in the BVOC emission mixture, this could suppress SOA formation and consequently mitigate the aerosol negative radiative forcing effect (Kiendler-Scharr et al., 2009). Finally, the oxidative reactions between BVOCs, OH radicals and oxides of nitrogen influence the greenhouse gas concentration as they form tropospheric ozone and affect methane competing for OH radicals (Laothawornkitkul et al., 2009). However, the net impact of BVOCs on climate change remains uncertain (IPCC, 2007; Peñuelas, 2008).

Acknowledgements

We thank T. Oksanen for constructing the equipment and J. Tarhanen for assistance in the GC-MS analyses. We are also grateful to Professors R. Atkinson and M. Claeys for their contributions to the compound classification. The study was financially supported by the Emil Aaltonen Foundation, an EU ATANS grant (Fp6 506004), the Danish Council for Independent Research | Natural Sciences and the Abisko Scientific Research Station. Research fellowships from the FQRNT, Finnish Cultural Foundation and Ella and Georg Ehrnrooth Foundation were awarded to Patrick Faubert.

Ancillary