Atmospheric Environment 40 (2006) S128–S137 VOC emissions from Norway spruce (Picea abies L. [Karst]) twigs in the field—Results of a dynamic enclosure study W. Grabmer a , J. Kreuzwieser b , A. Wisthaler a , C. Cojocariu b , M. Graus a , H. Rennenberg b , D. Steigner c , R. Steinbrecher c , A. Hansel a, a Institut fu ¨ r Ionenphysik, Leopold-Franzens-Universita ¨ t Innsbruck, Technikerstr. 25, 6020 Innsbruck, Austria b Institut fu ¨ r Forstbotanik und Baumphysiologie, Professur fu ¨ r Baumphysiologie, Albert-Ludwigs-Universita ¨t Freiburg, Georges-Ko¨hler-Allee Geb. 053/054, 79110 Freiburg i. Br., Germany c Forschungszentrum Karlsruhe GmbH, Institut fu ¨ r Meteorologie und Klimaforschung, Atmospha ¨ rische Umweltforschung (IMK-IFU), Kreuzeckbahnstr. 19, 82467 Garmisch-Partenkirchen, Germany Received 20 June 2005; received in revised form 20 January 2006; accepted 13 March 2006 Abstract During the 2002 summer intensive field campaign of BEWA2000 a proton-transfer-reaction mass spectrometer (PTR-MS) was used for online determination of volatile organic compounds (VOC) emitted by Norway spruce (Picea abies L. [Karst]) twigs in a dynamic sampling enclosure. Emissions of isoprenoids (isoprene and monoterpenes) and oxygenated VOC (OVOC; acetaldehyde, acetone, methanol, and ethanol) were investigated. Emissions showed clear diurnal patterns with high daytime emission rates amounting to 1.8 mgCg 1 dwt h 1 for the sum of monoterpenes and in the range of 0.1 to 0.6 mgCg 1 dwt h 1 for isoprene4acetone4ethanol4methanol. Data were used to validate existing models on isoprene and monoterpene emissions and to discuss environmental and physiological factors affecting VOC emissions. Isoprene and acetaldehyde emission rates were best modelled applying the Guenther 1993 temperature and solar radiation algorithm. Emissions of monoterpenes, acetone and ethanol were best described by a temperature-only exponential algorithm. Using these model approaches a maximum emission variability of 66% was covered (isoprene). Poor r 2 values ranging from 0.15 to 0.42 were typical for oxygenated VOC emission modelling indicating the need for model improvement e.g. development of process-based models describing the emission as a result of biochemical de novo synthesis as well as physico-chemical transport properties inside the leaves. r 2006 Elsevier Ltd. All rights reserved. Keywords: Isoprenoids; OVOC; PTR-MS; BVOC; Emission algorithms 1. Introduction Volatile organic compounds (VOC) play an important role in atmospheric chemistry, particu- larly when abundant together with nitrogen oxides (e.g. Atkinson and Arey, 2003). The atmospheric concentration of VOC influences the production and loss of tropospheric ozone and may control the ARTICLE IN PRESS www.elsevier.com/locate/atmosenv 1352-2310/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2006.03.043 Corresponding author. Tel.: +43 512 507 6245; fax: +43 512 507 2932. E-mail address: [email protected] (A. Hansel). URL: http://www.uibk.ac.at/ionenphysik/umwelt.
10
Embed
VOC emissions from Norway spruce ( Picea abies L. [Karst]) twigs in the field—Results of a dynamic enclosure study
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ARTICLE IN PRESS
1352-2310/$ - se
doi:10.1016/j.at
�Correspondfax: +43512 50
E-mail addr
URL: http:/
Atmospheric Environment 40 (2006) S128–S137
www.elsevier.com/locate/atmosenv
VOC emissions from Norway spruce (Picea abies L. [Karst])twigs in the field—Results of a dynamic enclosure study
W. Grabmera, J. Kreuzwieserb, A. Wisthalera, C. Cojocariub, M. Grausa,H. Rennenbergb, D. Steignerc, R. Steinbrecherc, A. Hansela,�
Volatile organic compounds (VOC) play animportant role in atmospheric chemistry, particu-larly when abundant together with nitrogen oxides(e.g. Atkinson and Arey, 2003). The atmosphericconcentration of VOC influences the productionand loss of tropospheric ozone and may control the
ARTICLE IN PRESSW. Grabmer et al. / Atmospheric Environment 40 (2006) S128–S137 S129
concentration of radicals such as OH, responsiblefor the oxidation of methane and other greenhousegases. VOC can either be of biogenic or anthro-pogenic origin. On a global scale, biogenic volatileorganic compound (BVOC) emissions are estimatedto be an order of magnitude higher than anthro-pogenic VOC emissions (Guenther et al., 1995) withlarge uncertainties particular for the BVOC esti-mates. Guenther (1999) pointed out that great effortis required to reduce these uncertainties.
On a quantitative basis, the most importantBVOC are the isoprenoids isoprene and mono-terpenes. The isoprenoid emission algorithms cur-rently used for model calculations are mainly basedon the approaches reported by Guenther et al.(1995). Strong variations of standard emissionfactors of many plant species with discrepancies ofup to one order of magnitude have been reported(Kesselmeier and Staudt, 1999). This variabilitymay originate from measurements of differentspecies or ecotypes under different environmentalconditions and the limited number of robust datasets available. Norway spruce, for example, con-tributes to 34% and 49% of the total forested areain Germany and Bavaria, respectively (Smiatek andSteinbrecher, 2006), and only few emission datamostly for boreal climates are reported (Steinbre-cher, 1989; Schurmann et al., 1993; Steinbrecheret al., 1999). Even fewer data are available on theemission of oxygenated VOC (OVOC) such asaldehydes or ketones despite their estimated 24%contribution to the total VOC emission from forestecosystems (Guenther et al., 1995).
In the present study a proton-transfer-reactionmass spectrometer (PTR-MS) was used for online-determination of VOC emissions from matureNorway spruce trees in the field. Such high time-resolution measurements are used for calculatingstandard emission factors and validating empiricalmodels designed for predicting VOC emissions fromtrees, which can further be used in models calculat-ing ecosystem VOC exchange.
2. Experimental
2.1. Field site description
A field campaign was conducted in July andAugust 2002 within the framework of theBEWA2000 program. The measurement site was aspruce forest, dominated to 90% by 50–80 year oldNorway spruce (Picea abies [L.] Karst.) trees.
Measurements were performed in a remote forestarea of the German ‘‘Fichtelgebirge’’ (5010803200N,1115200400E, 775m a. s. l.) characterised by an alpinelike climate (Klemm and Mangold, 2001). Ambienttemperatures ranged from about 10 1C to a max-imum of 28 1C indicating a relatively cold weatherperiod with frequent fog or rain events andcorresponding highly variable light conditions. InJuly and August 2002 ambient air mixing rationsranged for isoprene from 0.08 to 0.49 ppb (n ¼ 92),the sum of monoterpenes from 0.02 to 0.69 ppb(n ¼ 92), methanol from 1.31 to 6.56 ppb (n ¼ 77),acetaldehyde from 0.22 to 4.86 ppb (n ¼ 356),acetone from 1.22 to 13.30 ppb (n ¼ 356) asreported by Klemm et al. (2006) and ethanol 0.2to 2.5 ppb (n ¼ 2376). The measurements wereconducted at two 57 year old spruce trees that weredirectly accessible from the 12m platform of ascaffolding tower. Data used for emission modellingand calculating standard emission rates for iso-prene, the sum of monoterpenes, methanol, ethanol,acetaldehyde, and acetone based on the modelapproaches described by Guenther et al. (1993,1999, 2000) were obtained on July 23, 28, 29, 31 andAugust 1 and 2. These days reflect the so-called‘‘Golden Days’’ where all instruments workedproperly and the weather covers sunny and rainyas well as cold and warm days (Klemm et al., 2006).During these days approximately 1100 data pointsfor each compound emitted were recorded and usedin modelling exercises.
2.2. Gas exchange measurements
A dynamic enclosure system described in detailby Cojocariu et al. (2004) was used to determineVOC emissions by spruce twigs. The systemconsisted of two identical 0.5 l volume Teflonenclosures covered with a Teflon film almostcompletely transparent to solar radiation (Kreuz-wieser et al., 1999). To keep the twig environment asclose to ambient conditions as possible bothenclosures were constantly flushed with ambientair at 2 lmin�1 resulting in an air exchange of theenclosure systems on average every 15 s. Fans wereinstalled to ensure homogenous mixing of the air inthe enclosures. All tubing was made of Teflon and,where possible, was kept dark and at constanttemperature (35 1C). The total residence time in thelines downstream of the enclosures was 16 s. Noisoprene oxidation products (methyl vinyl ketoneand methacrolein) were observed, indicating that
ARTICLE IN PRESSW. Grabmer et al. / Atmospheric Environment 40 (2006) S128–S137S130
oxidation processes were negligible in the system. Oneenclosure was kept empty as a reference (‘‘controlenclosure’’), whereas the other (‘‘plant enclosure’’)contained a twig of 6–8 cm length (leaf dry weights(dwt) ranging from 1.2 to 1.7 g). The twig was insertedinto the enclosure and after an adaptation time of 1hemission measurements were started. Care was takenthat the twig was not wounded during fixation;enhanced emission rates due to wounding wereusually not observed during the measurements. Atthe end of each measurement all needles from insidethe plant enclosure were harvested and leaf area anddry weight were determined.
During all measurements, flow rates (MAS,Kobold, Germany), temperature and relative hu-midity (1400-104, Walz, Germany) were recorded.A light sensor (LI-190SA, Li-Cor Inc., Lincoln, NB,USA) was placed on the plant enclosure in closevicinity to the inserted twig to determine photo-synthetic photon flux density (PPFD; 400–700 nm).In addition to meteorological parameters, concen-tration differences of CO2 and H2O (Li-6262, Li-Cor Inc., Lincoln, NB., USA) between bothenclosures were determined continuously. Datawere recorded on an electronic logbook (Log-
Book/300TM —Stand-Alone Data Acquisition sys-
tem, IOtechs, Spectra Computersysteme GmbH,Echterdingen, Germany) connected to the system.The rates of photosynthesis and transpiration werecalculated from concentration differences betweenboth enclosures, taking into account both needleleaf area and flow rate through each enclosure(Cojocariu et al., 2004).
2.3. VOC determination
The analytical technique applied for the determi-nation of VOC emitted by spruce needles in thisstudy was PTR-MS (Hansel et al., 1995; Lindingeret al., 1998a, b). PTR-MS is a chemical ionisationtechnique based on proton transfer reactions fromH3O
+ reagent ions to VOC. At PTR-MS standardoperation conditions (E/N ¼ 120–140 Td; E electricfield strength, N buffer gas number density,1Td ¼ 10�17 cm2 Vmolecule�1) one of the targetmolecules of this study, ethanol, fragments uponprotonation. The PTR-MS instrument used herewas thus operated at reduced E/N of �95Td wherefragmentation of protonated ethanol molecules islargely reduced. The implications of non-standardPTR-MS operation are described in detail by DeGouw et al. (2003) and Hewitt et al. (2003). The
main compounds (and their respective protonatedmasses; in atomic mass units [amu]) detected byPTR-MS were methanol (33), acetaldehyde (45),ethanol (47), acetone (59), isoprene (69), andmonoterpenes (137, fragment on 81). It is importantto note that different types of monoterpenes cannotbe distinguished from each other with PTR-MSsince they are isobaric. Therefore, volume mixingratios (VMR) of monoterpenes in this study alwaysrefer to the sum of all monoterpenes. A possibleinterference from formic acid at mass 47 andpropanal at mass 59 is unlikely to have a significantimpact. Strictly speaking all reported VMR shouldbe seen as upper limits, although no evidence for thepresence of any major interfering species was found.The PTR-MS instrument was calibrated using a gasstandard (E. Apel, Atmospheric Chemistry Divi-sion, NCAR) containing the given target VOC(1–10 ppmv of the respective VOC in nitrogen,accuracy 720%; diluted in humidified syntheticair). Mass signals were monitored on a time-sharedbasis for 5–20 s per compound every 3min. Thecontrol enclosure was measured every hour for9min resulting in a short break in the plantenclosure measurements. Values from the controlenclosure were linearly interpolated and subtractedfrom plant enclosure values.
In addition to PTR-MS analysis, isoprenoids inthe enclosure outlet air were trapped on a three-bed-adsorbent tube (60mg Carbotrap C, 60mg Carbo-trap, 90mg Carbopack X; Supelco, Bellafonte, PA,USA) and analysed using thermo-desorption GC-MS (Schnitzler et al., 2004a).
2.4. Emission modelling
2.4.1. Isoprene emission
The algorithm of Guenther et al. (1993)—here-after referred to as G93—was applied to theemission data to model isoprene emission and tocalculate the standard emission rate. The isopreneemission was calculated using the following equa-tion:
I ¼ IS � CT � CPAR, (1)
where I is the rate of isoprene emission from a leaf(mgC g�1 dwt h�1), IS is the standard emission ratefor isoprene (at 30 1C leaf temperature;1000 mmolm�2 s�1 PPFD), and CT and CPAR areempirical functions taking into account the chan-ging emission rates at different leaf temperaturesand PPFD, respectively. CT and CPAR were used as
Fig. 1. Modelled and measured monoterpene (a) and isoprene (b)
emission rates from Norway spruce as well as PPFD and
temperature (c) and net carbon assimilation and stomatal
conductance for water vapour (d) as measured on July 28,
2002. Isoprene emissions were modelled using the Guenther
algorithms (G93 and G99); the commonly used temperature-only
dependent exponential algorithm (EXP) was used to model
monoterpene emissions.
W. Grabmer et al. / Atmospheric Environment 40 (2006) S128–S137 S131
calculated by Guenther et al. (1993). Thus, the CT
function has a maximum at 40 1C and the CPAR
function reaches 170.05 above 700 mmolm�2 s�1
PPFD. In this approach, differences between plantspecies, but also acclimation to low temperaturesand other effects, are included in the factor IS. Asdeduced from the mechanisms of isoprene synthesisand release, emission is zero in the dark.
Guenther et al. (1999, 2000) pointed out that theisoprene emission potential of a plant not onlydepends on the current leaf temperature, but also onthe temperature-history of the plant. In a differentalgorithm—hereafter referred to as G99—CT andCPAR account for more parameters, e.g. CT includesthe mean temperature of the last 15 days. Further-more, CPAR takes into account shading effects,which can be expressed in the cumulated leaf areaindex (LAI). Here a LAI of 3 was used asmeasurements took place in the upper part of thecanopy. LAI at the ground was 5.5 (Klemm et al.,2006). In the present study, the G99 algorithm wasalso applied to calculate the standard emission rateand to model isoprene emissions.
2.4.2. Monoterpene emissions
Monoterpene emissions are normally consideredto depend on leaf internal pools implying thatsynthesis does not coincide with the emission. In apool model, the actual monoterpene flux is notcontrolled by light, but depends exponentially onleaf temperature, according to the following equa-tion (e.g. Guenther et al., 1993)—hereafter referredto as EXP:
M ¼MS � expðbðT � TSÞÞ, (2)
where M is the rate of monoterpene emission from aleaf (mgC g�1 dwt h�1), MS is the standard emissionrate for monoterpenes at standard leaf temperature(TS ¼ 30 1C), T is the actual leaf temperature, and b isan empirical coefficient. Most monoterpene emittershave a b coefficient of about 0.1 (Guenther et al.,1993) which also was used in the present study.
2.4.3. Emissions of OVOC
For modelling OVOC emissions the EXP algo-rithm is used in general. To test hypotheses onenvironmental factors other than temperature af-fecting OVOC emissions also the G93 and G99models were applied to the experimental emissiondata and tested for highest correlations. Subse-quently, standard emission rates for acetaldehyde,ethanol and acetone were calculated.
3. Results and discussion
3.1. Isoprene
In this study maximum isoprene emission rates ofmature Norway spruce trees were in the order of0.6 mgC g�1 dwt h�1 during periods of high tem-perature and high PPFD (Fig. 1b, c). Under these
ARTICLE IN PRESSW. Grabmer et al. / Atmospheric Environment 40 (2006) S128–S137S132
conditions the rates of net carbon assimilationwere usually high (up to 1.8mgCO2 g
�1 dwt h�1)(Fig. 1d). Compared to isoprene emissions, thediurnal cycle of net assimilation and stomatalconductance was not so pronounced. Measurementstook place relatively high up in the canopy andshading effects on the needle development were oflittle importance. The presented diurnal cycle of gasexchange of a Norway spruce twig is typical for aday with rapidly changing weather conditions. Thisday was chosen out of a set of 6 days to demonstratethe benefits as well as limitations of currentlyavailable models used for predicting biogenic VOCemissions. Applying the G93 algorithm, a standardemission rate IS of 0.32 mgC g�1 dwt h�1 was derived(Table 1; r2 ¼ 0.66).
Norway spruce is known to be a low isopreneemitting species (Steinbrecher et al., 1993; Kessel-meier and Staudt, 1999). The reported standardemission rate is at the lower end from other studies,with values ranging from 0.3 to 1.8 mgC g�1 dwt h�1
(see compilation by Kesselmeier and Staudt, 1999).The G99 algorithm tries to account for the effect
of leaf temperature history on emission rates.Considering a 15 day mean temperature of 15 1Cprior to the measurements, the G99 algorithmyielded a five-times higher standard emission rate(1.7 mgC g�1 dwt h�1; Table 1; r2 ¼ 0.59) as com-pared to the G93 value with significantly lower r2.The G99 algorithm implies that a plant adapted tolower temperatures has a much higher emissionpotential. As in this study low temperaturesprevailed during most days of the field campaign,the G99 resulted in a model specific significantlyhigher standard emission rate.
G93- and G99-derived standard emission rateswere used for predicting isoprene emission rates
Table 1
Standard emission rates calculated from 6 days data obtained in July/A
Compound Temperature PPFD Model
dependency dependency
Isoprene X X G93
Isoprene X X G99
Monoterpenes X EXP
Acetaldehyde X X G93
Ethanol X EXP
Methanol
Acetone X EXP
PPFD: Photosynthetic photon flux density; n: number of data points.aFrom data compilation by Kesselmeier and Staudt (1999).
from Norway spruce needles. Both models showedsome discrepancies between calculated and mea-sured emissions in particular during morning andevening hours. Further the G99 model stronglyunderestimated peak emission rates at noon by afactor of 2 (Fig. 1b). Comparing the results of bothalgorithms, a much better correlation betweenmodelled and measured data was obtained forG93 than for G99 (Table 1; G93, r2 ¼ 0.66; G99,r2 ¼ 0.59). It seems that the G99 overestimates thetemperature history effect on isoprene emission ofNorway spruce in particular during high emissionperiods. As both models have difficulties in predict-ing morning and evening emission rates it may beargued that additional factors have to be consideredin the models to improve the performance, such asplant internal processes, like net carbon assimilationand alternative carbon sources for isoprenoidbiosynthesis (Heizmann et al., 2001; Kreuzwieseret al., 2002; Schnitzler et al., 2004b). For modellingtimely highly resolved emission rates around theyear also the seasonality (phaenology) of emissionfactors may have to be accounted for (Schaab et al.,2003; Schnitzler et al., 1997; Staudt et al., 2000). Butdue to limited data available it is not possible yet toderive a seasonality function for Norway spruceemissions.
3.2. Monoterpenes
The application of the EXP algorithm resulted ina standard emission rate MS of 0.50 mgC g�1
dwt h�1 (Table 1; r2 ¼ 0.42) and a b value of 0.13for the sum of the monoterpenes. The modelcaptures the overall diurnal emission cycle quitewell (Fig. 1a). This is highlighted in Fig. 2 were the 1to 1 line between modelled and measured data is
Fig. 2. Scatterplot of measured monoterpene emission vs.
emissions predicted by model EXP. Only emissions above
detection limit are plotted. Data were obtained on July 23, 28,
29, 31 and August 1,2, 2002.
W. Grabmer et al. / Atmospheric Environment 40 (2006) S128–S137 S133
remarkably good. Yet the data scattering is quitelarge also expressed by the low r2 value. It seemsthat in particular the course of the highly variableemission rates around noon is not captured by themodel calculations (Fig. 1). Other factors thantemperature alone may cause the observed emissionvariability around noon. To check whether emissionmodelling of monoterpenes can be improved, lightas additional controlling factor was considered assuggested by Steinbrecher et al. (1999) and the G93algorithm applied. Yet, that calculation did notimprove the modelling results indicated by an evenweaker correlation (r2 ¼ 0.4) with the experimentaldata. Other processes not ruled by light seem tocontrol short time variation of monoterpene emis-sion from Norway spruce here.
One reason why the temperature and light modeldid not improve the prediction of the observedemission rates may be the time period—end of July/begin of August in the middle of the vegetationperiod—where the measurements were performed.Other studies performed at the beginning of thevegetation period or at young plants showedevidence that at least the emission of some mono-terpene compounds emitted is temperature and lightcontrolled (Steinbrecher, 1989; Schurmann, 1993;Schuh et al., 1997; Steinbrecher et al., 1999; Staudtet al., 2000). As pointed out before, a seasonality of
emission factors is a common feature for manyisoprenoid emitting plant species (Schaab et al.,2003; Schnitzler et al., 1997; Staudt et al., 2000) andthe stage of needle development is likely to impactthe emission potential also for Norway spruce. Thereported standard emission rate is at the lowerend of literature values, ranging from 0.2 to7.8 mgC g�1 dwt h�1 (Kesselmeier and Staudt, 1999).
The sum of the monoterpenes measured by GC-MS showed a similar diurnal emission variation tothat detected by PTR-MS and resulted in a standardemission rate of 0.27 mgC g�1 dwt h�1. Individualmonoterpenes contributing to the sum of mono-terpenes were a-pinene (42%), limonene (27%), b-pinene/sabinene (20%), D3-carene (7%), p-cymene/eucalyptol (4%) and others in traces. These resultsare in accordance with literature data (e.g. Janson,1993; Steinbrecher et al., 1999).
3.3. OVOC
In the present study carbonyls emitted by spruceneedles were mainly acetone and acetaldehyde(Fig. 3a,b). Emissions showed clear diurnal varia-tions with rates around midday of up to approxi-mately 0.5 mgC g�1 dwt h�1 and 0.35 mgC g�1
dwt h�1 for acetone and acetaldehyde, respectivelyand considerably lower rates during night. In 2001,Cojocariu et al. (2004) studied the same trees asinvestigated in this campaign and reported similardiurnal variations and emission rates (0.07–0.7 mgC g�1 dwt h�1). In the latter work carbonyl emis-sions strongly depended on air temperatures, afinding which was also strongly supported bylaboratory studies performed with the same species.Surprisingly, factors affecting stomatal conductancesuch as light and humidity did not influencecarbonyl exchange. The study by Cojocariu et al.(2004) has indicated that the ambient concentra-tions play a crucial role for the direction of trace gasexchange. At high external concentrations, carbo-nyls are taken up by trees. However, the ambientcarbonyl concentrations during the golden days (seeabove), favoured carbonyl emission rather thandeposition as supported by the presented data.According to the findings of this work, a study onSitka spruce (Hayward et al., 2004) also reported astrong dependency of acetaldehyde and acetoneemissions on temperature. Hayward et al. (2004)derived an acetaldehyde standard emission rate of0.37 mgC g�1 dwt h�1 for Sitka spruce. Janson et al.(1999) reported standard total carbonyl emission
Fig. 3. Emissions of methanol and acetone (a), ethanol and
acetaldehyde (b) from Norway spruce as well as PPFD and
temperature (c) and net carbon assimilation and stomatal
conductance for water vapour (d) needles as measured on July
28, 2002.
Fig. 4. Influence of light-dark transitions on acetaldehyde
emission from Norway spruce as measured under field condi-
tions.
W. Grabmer et al. / Atmospheric Environment 40 (2006) S128–S137S134
rates in the range of 0.3 to 4.6 mgC g�1 dwt h�1 fromNorway spruce in a boreal forest. The standardemission rates derived here (acetaldehyde: 0.19 mgC g�1 dwt h�1; acetone: 0.23 mgC g�1 dwt h�1) aresomehow lower than the previously reported values.
Methanol and ethanol emissions showed cleardiurnal variation with highest rates (methanolapproximately 0.15 mgC g�1 dwt h�1; ethanol ap-proximately 0.40 mgC g�1 dwt h�1) around midday(Fig. 3a,b). The observed correlation betweenethanol and acetaldehyde emissions may be due tocommon metabolic pathways leading to the produc-
tion of these compounds. As discussed by Cojocariuet al. (2004), it can be assumed that acetaldehydeemitted by Norway spruce needles is an oxidationproduct of ethanol, which can be derived fromhypoxic parts of the cambium or the roots.Correlations between xylem transported ethanoland acetaldehyde emissions have previously beenreported in several laboratory studies with decid-uous and coniferous species (Kreuzwieser et al.,1999, 2000, 2001). Besides the oxidation of ethanol,rapid light-dark transitions can also lead to a releaseof acetaldehyde by the leaves of trees (Holzingeret al., 2000; Karl et al., 2002; Graus et al., 2004). Ithas been assumed that this acetaldehyde release maycontribute to acetaldehyde emissions in the field,e.g. during light-flecking throughout the day witheach transition from high to low light intensities(Karl et al., 2002). In order to test this assumptionunder field conditions, acetaldehyde emission rateswere determined under conditions of stronglychanging light intensities (Fig. 4). However, undersuch conditions no increased rates of acetaldehydeemissions were observed with a time resolution of3min or lower. In agreement with recent studies ofGraus et al. (2004) it therefore seems unlikely thatlight-dark transitions contribute substantially toacetaldehyde emissions in the field.
Compared to acetaldehyde and ethanol, muchless information is available on production path-ways for acetone and methanol. It was hypothesisedthat acetone is produced in spruce needles by thedecarboxylation of acetoacetate (MacDonald andFall, 1993), whereas the cyanohydrin-lyase cata-lysed reaction found in cyanogenic plant speciesdoes not occur in conifers (Fall, 2003). However, at
ARTICLE IN PRESSW. Grabmer et al. / Atmospheric Environment 40 (2006) S128–S137 S135
present none of these pathways has been proven tobe responsible for acetone released by leaves. Thesame holds for methanol emissions. In the presentstudy the apparent correlation of methanol emissionrates with light and temperature (see Fig. 3a and c),and with transpiration rates (data not shown)suggest that, comparable to ethanol, methanol ora precursor is delivered via the transpiration streamas proposed by MacDonald and Fall (1993). Thefinding of methanol in the xylem sap of Norwayspruce (Cojocariu, unpublished data) supports thisassumption although more studies are required totest this hypothesis.
In order to calculate standard emission factors forOVOC emitted by Norway spruce, modelling wasperformed using the algorithms G93, G99 and EXP.The results of the model approach with the highestcorrelation are shown in Table 1. Whereas acet-aldehyde emission was described well by the G93model, best correlations for ethanol and acetoneemissions were obtained using the EXP algorithmdeveloped for monoterpene emissions. The latterresult indicates that the release of these compoundsoriginated from leaf internal storage pools ratherthan from instantaneous production.
3.4. Carbon budget
In order to estimate the budget of carbonexchanged between Norway spruce and the atmo-sphere, all carbon emitted as VOC was added upand compared to the carbon fixed by photosynth-esis. During daytime values between 0.05 and 0.5%of the photosynthetically fixed carbon was re-emitted into the atmosphere. These changes weredue to both changes in VOC emission rates duringthe day and variation in photosynthesis. It shouldbe mentioned that these values are slight under-estimations due to the exclusion of other VOC suchas sesquiterpenes in the present study. Suchnumbers are in good agreement with valuesobtained in other studies (see compilation byKesselmeier et al., 2002). Enclosure measurementsprovided values of 0.01–0.27% for conifers (Picea
sitchensis, Street et al., 1996), 0.01–0.5% for shrubs(Hansen et al., 1997), up to 1% (of gross carbonassimilation) for Mediterranean trees and approxi-mately 0.12–0.6% (of gross carbon assimilation) forAmazonian trees (Kesselmeier et al., 2002). Some-what higher values (0.18–13%) were observed forpoplar and oak species (see compilation by Kessel-meier et al., 2002); much higher values (15–50%)
can also be reached under stress conditions, whichstimulate isoprenoid emissions but decrease therates of net assimilation at the same time (Sharkeyand Loreto, 1993; Sharkey et al., 1996; Steinbrecheret al., 1997; Harley et al., 1999).
The sum of OVOC emitted by Norway spruceincluding acetaldehyde, ethanol, acetone andmethanol was in the same order of magnitude asfor isoprene and monoterpene emissions. This resultis consistent with the estimate by Guenther et al.(2000), reporting that about one-third of the VOCemitted by forests are OVOC.
4. Conclusion
The present study provides for the first timesimultaneous on-line measurements of the emissionof isoprenoids and OVOC under field conditions.Such measurements are required to develop, im-prove and validate models for the prediction of fluxrates of individual VOC. The results indicated thatthe routinely used models for predicting isoprene(G93 and G99) and monoterpenes (EXP) roughlyreflect the emission source strength measured, butthat in particular isoprene emission modelling fromNorway spruce as well as modelling of OVOCemissions should be considerably improved. Oneapproach could be process based modelling con-sidering the complexity of leaf internal productionas well as its control by environmental parameters.
Acknowledgements
This work was financially supported by theGerman Federal Ministry of Education and Re-search (BMBF) in the frame of BEWA2000, a sub-project of the national joint research project AFO2000(Atmospharenforschungsprogramm 2000). We aregrateful to Otto Klemm and Andreas Held,University of Munster, for their help during thefield measurements. Armin Wisthaler thanks the‘‘Verein zur Forderung der wiss. Ausbildung undTatigkeit von Sudtirolern an der LandesuniversitatInnsbruck’’ for postdoctoral support.
References
Atkinson, R., Arey, J., 2003. Atmospheric degradation of volatile
organic compounds. Chemical Reviews 103, 4605–4638.