JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 103, NO. D23, PAGES 31,045-31,056, DECEMBER
20, 1998
Seasonalcourse of isoprene emissionsfrom
a midlatitude
deciduous
forest
Allen H. Goldstein
Divisionof Ecosystem
Sciences,
Department
of Environmental
Science,PolicyandManagement,
Universityof
California,Berkeley
MichaelL. Goulden,
1J. WilliamMunger,andStevenC. Wofsy
Department
of EarthandPlanetary
Sciences,
Divisionof Engineering
andAppliedSciences,
HarvardUniversity
Cambridge,Massachusetts
ChristopherD. Geron
NationalRiskManagement
Research
Laboratory,
U.S. Environmental
Protection
Agency,Research
Triangle
Park, North Carolina
Abstract. Continuous
measurements
of wholecanopyisopreneemissions
overan entire
growingseasonarereportedfromHarvardForest(42ø32'N,72ø11'W). Emissionswere
calculated
fromtheratioof observed
CO2 flux andgradient
multipliedby theobserved
hydrocarbon
gradients.In summer1995,24-houraverageemissions
of isoprene
fromJune1
through
October
31were32.7x 101ø
molecules
cm-2s-1(mgC m-2h-I = 2.8x 10•
molecules
cm-2s-•),andthemeanmidday
mixingratiowas4.4ppbvat24 m. Isoprene
emissions
werezeroatnight,increased
through
themorning
withincreasing
airtemperature
andlight,reached
a peakin theafternoon
between
thepeaksin airtemperature
andlight,and
thendeclined
withlight. Isoprene
emissions
wereobserved
overa shorter
seasonal
period
thanphotosynthetic
carbonuptake.Isoprene
emission
wasnotdetected
fromyoungleaves
andreached
a peakrate(normalized
forresponse
to measured
lightandtemperature
conditions)
4 weeksafterleafoutand2 weeksafteremissions
began.The normalized
emission
rateremained
constant
forapproximately
65 days,thendecreased
steadily
through
September
andintoOctober.
Totalisoprene
emissions
overthegrowingseason
(42 kg C ha-1
yr-l)wereequalto2%oftheannual
netuptake
ofcarbon
bytheforest.Measured
isoprene
emissions
werehigherthantheBiogenic
Emission
Inventory
System-IImodelby at least
40%atmiddayandshowed
distinctly
differentdiurnalandseasonal
emission
patterns.
Seasonal
adjustment
factors(in additionto thelightandtemperature
factors)shouldbe
incorporated
intofutureempiricalmodelsof isoprene
emissions.
Comparison
of measured
isoprene
emissions
withestimates
of anthropogenic
volatileorganic
compound
emissions
suggests
thatisoprene
is moreimportant
for ozoneproduction
in muchof Massachusetts
on
hotsummerdayswhenthe highestozoneeventsoccur.
modeling studieshave shown that ozone productionin the
northeasternUnited States is quite sensitive to biogenic
Emission of isoprene (2-methyl-l,3-butadiene)by NMHC emissions[McKeen et al., 1991; Roselieet al., 1991;
terrestrialvegetationprovidesthe dominantinputof reactive Sillman et al., 1990] and that the uncertainty in those
non-methane
hydrocarbons
(NMHC) to the atmosphere
and emissionsintroduces large uncertainty into predictions of
1. Introduction
influences
tropospheric
chemistry
onbothregionalandglobal emissionsintroduceslarge uncertaintyinto predictionsof
scales[Zimmerman,1979; Chameideset al., 1988; Winer et reductionsthat could be achievedthroughNMHC or NOx
al., 1989; Singhand Zimmerman,1992; Fehsenfeldet al., emissioncontrolstrategies.Hirsch et al. [1996] found that
1992]. Isopreneis principallyremovedfromthe troposphere seasonalityof ozone productionefficiencyper unit NOx
throughoxidationby hydroxylradicals,andin thepresence
of determined
fromthe HarvardForestdatawastightlycoupled
oxidesof nitrogen(NOx = NO + NO2) is closelycoupledto to seasonality of isoprene emissions. Furthermore,
the photochemicalproduction of ozone. Photochemical meteorological
conditionsthat causethe highestisoprene
INow at EarthSystemScience,Universityof California,Irvine.
Copyright1998by theAmericanGeophysicalUnion.
Papernumber98JD02708.
0148-0227/98/98JD-02708509.00
mixing ratios (hot, sunny,and stable)also maximize ozone
production.
Isopreneproductionis closely linked to photosynthetic
processes[Shatkey et al., 1991], and control over its
production
is coupledto isoprenesynthase
activity[Monson
et al., 1992]. Shatkeyand Singsaas[1995] haveshownthat
isoprene
emission
increases
the thermaltoleranceof plants,
31,045
31,046
GOLDSTEIN ET AL.: SEASONAL COURSE OF ISOPRENE EMISSIONS
and suggestthat protectionfrom heat may be the primary
reasonplantsemit isoprene.
Isopreneemissionfrom vegetationhasbeenmeasuredin
natural environmentsusing branch, leaf, and whole tree
enclosure[Zimmerman,
1979; Lamb et al., 1985; Lamb et al.,
1986; Khalil and Rasmussen,
1992; Monson et al., 1994;
Pier, 1995], tracer release[Lamb et al., 1986], mixed layer
and surfacelayer micrometeorlogical
techniques[Knoerr and
Mowry, 1981; Lamb et al., 1985; Baldocchiet al., 1995,
Guentheret al., 1996a, b, c], and in controlledlaboratory
settings[Sanadze,1969; Tingey et al., 1979; Monson and
Fall, 1989; Guentheret al., 1991]. Isopreneemissionrates
were observedto dependon both light and temperatureand
were zero at night. Studies of individual leaves showed
isopreneemissionsto be linearly dependenton the flux of
photosynthetically
activeradiation(PAR) to a saturation
point
of approximately1000 gE m-2 s-l, and exponentially
dependenton temperatureup to approximately35øC, where
emissionsstartto level off and eventuallydeclinenear 40øC.
However,Shatkeyet al. [1996] foundthat isopreneemission
Boston, Massachusetts,
and 100 km northeastof Hartford,
Connecticut. There is a highway • 5 km to the north and a
secondary
road• 2 km to thewest. The siteis accessible
by a
dirt road, which is closed to public vehicle traffic.
Measurements
weremadefrom a 30 m tower,erectedin May
1989, extending7 m abovethe forest canopy. Instruments
were housedin a temperaturecontrolledshacklocated15 m
east of the tower.
The forestis 50 to 70 yearsold with a deciduousleaf area
indexof 3.5 (measured
in 1994by leaf littercollectionby M.
Goulden(unpublisheddata, 1994). The tree speciesin
descending
order of abundanceare oak (mostlyred), red
maple,red pine,hemlock,birch,whitepine,andcherrywith
percent basal areas (stem cross-sectionalarea at 1.37 m
determined
fromplot surveys(M. Gouldenunpublished
data,
1994)of 36.4,21.8, 16.2,15.6,3.5, 2.9, and2.3, respectively.
The terrainis moderatelyhilly (relief • 30 m), but thereis no
evidenceof anomalous
flow patterns
thatwouldmakeeddyflux measurements
at this siteunrepresentative
[Mooreet al.,
1996], and measurements
of the local energybudgetare
fromoaktreesisnotfullylightsaturated
at 1000•E m-2s-1 balancedto 20% [Gouldenet al., 1996a].
underfield conditions. Isopreneemissionratesalso depend
Measurements
of C2-C6NMHCs weremadein air sampled
on leaf age. Measurementson aspenleaves[Monsonet al.,
simultaneously
from2 and7 m abovetheforestcanopyevery
1994] haveshownthat onsetof emissionsbeginsaftera given
number of growing days above 5øC, increases for
approximately3 weeks,holds steadyfor approximately8-10
weeks, and decreases thereafter.
Similar results were
observedusing leaf enclosures[Fuentes et al., 1995] and
whole tree enclosures[Pier, 1995].
Models of troposphericchemistry and photochemical
ozone production require accurate estimatesof biogenic
isoprene emissions. Currently, the U.S. Environmental
Protection Agency uses inventories calculated from the
BiogenicEmissionInventory SystemII (BEIS2) [Geron et
al., 1994]. Thesemodelscalculateemissionsusingdatabases
of foresttype or speciescomposition,biomassdensity,basal
emission rate (species dependent), and the responseof
isopreneemissionto temperatureand PAR from Guentheret
al. [1993]. Leaf temperatureand PAR are derived from
ambientconditionsabovethe ecosystem,
anda canopymodel
is used to estimate their distribution
at five vertical levels in
45 min, commencingJuly 22, 1992 [Goldsteinet al., 1995a,
b, 1996]. Othertracegasmixingratiosand meteorological
conditionshavebeenmeasuredcontinuously
at this sitesince
1990,including
CO,CO2,03,NOx,NOy(NO+ NO2+ NO3
+ N205 + HNO3 + peroxyacetylnitrate
+ other organic
nitrates+ organicaerosols),H20, rainfall,wind speed,wind
direction,temperature,
radiation,and eddy covariancefluxes
of sensible
heat,latentheat,03, NOy,andCO2[Woj•yet al.,
1993, Mungeret al., 1996, 1998; Moore et al., 1996; Goulden
et al., 1996a,b].
2.2. Flux-Gradient Similarity Calculation
We usea similarityapproachfor determiningisopreneflux
for a whole forestecosystem,
basedon other quantitiesfor
which we have both mixing ratio data and direct
measurements
of flux [Goldsteinet al., 1996]. The tracegas
flux (F) is assumedto be proportionalto the time-averaged
the canopy. Two scenariosare used: (1) leaf temperatureis
assumed
to equalabovecanopyair temperature
(BEIS2), (2)
leaf temperatureis estimatedusingthe energybalancemodel
of Lamb et al. [1993] (BEIS2E). The model is designedto
simulatemidsummeremissionsand assumesthat, for a given
ecosystem,emissionrates vary only with temperatureand
PAR. The modeldoesnot includediurnalor seasonalchanges
mixing ratio gradient (dC/dz) above the forest for intervals
longerthanthe timescalefor the slowestsignificantturbulent
events(10 min):
canopyisopreneemissionsover an entiregrowingseason.
[Baldocchiet al., 1995], in agreementwith our afternoon
values.In this study,we computeK usingmeasurements
of
CO2 flux andgradientandtakethe productof thisK with the
isoprenegradientto determinethe isopreneflux (Figure1).
F=K dC/dz,
( 1)
whereK is the exchange
coefficientfor the averaginginterval.
In 1993 we determinedK simultaneouslyfrom fluxes and
verticalgradientsof CO2 and H20 at this site [Goldsteinet
in the basal emission factors.
al., 1996]. CO2 has a sourceat the groundand a sink in the
In orderto quantifywhole ecosystemisopreneemissions, canopy,while H20 and isopreneare both emitted from the
determinethe response
to temperature,light, andphenology, canopy;thusCO2 is not strictlysimilarto theseotherscalars.
assessregionalsignificance,and test the accuracyof current KCO2was 20% lower than K H20 in the afternoon(1200biogenic emission models, we built and deployed an 1800),andwas 20% higherearly in the day (0600 - 0900)
automatedsystemto continuouslymeasureisoprenefluxes and late in the day (1900-2100). Theoretical calculations
and relevant environmental variables which control them. In
suggest
thatK for isopreneandH20 shouldbe equal,butthat
this paper we presenta nearly continuousrecord of whole KCO2 should be 20% lower over a deciduous forest
2. Experiment
2.1.
Site
We chose to use KCO2 becausewe did not have routine
Harvard Forest is located in Petersham, Massachusetts measurements
for H20 or temperature
gradients.Our choice
(42ø32'N, 72ø11'W; elevation 340 m), 100 km west of
of K may introducea systematicunderestimationof the
GOLDSTEIN ET AL.' SEASONAL COURSE OF ISOPRENE EMISSIONS
•o
31,047
o
••
0
O•
0
o
•,
o
,o
•
•
•
o
I
o
o
oo
o
o
o
o
oO
Cb
o o
o
o
øo
o
!
o
o
o
o
o
o
i o
od
o
6
o
o
,
,
182
183
,
o
184
185
186
187
Day of Year 1995
Figure 1. Isoprenemixingratioat 24 m, isoprenegradient,CO2 gradient,CO2 flux, andisopreneflux (July
1 to 6, 1995).
isopreneflux in the afternoonof up to 20%; additionally,we
may overestimatefluxesin the morningand eveningby up to
20% thoughthe fluxes are small at thesetimes of day. The
seasonalcycleof isopreneemissionsis presentedin this paper
asmiddaymeanflux valuesfrom 1000 to 1500, duringwhich
time the mean systematicunderestimateof the flux may be
roughly 12% basedon comparisonof K from H20 and CO2
during this time of day. These systematicerrorsare smaller
than the overall uncertaintyin the individual hydrocarbon
flux measurementsand should not affect the long-term
precision of the measurementsor modify the observed
seasonalcycle, which is the primary contributionof this
study.The coefficientof variation(standarddeviation/flux)
for individual isopreneflux measurementsduring typical
daytime summer conditionsis 28% and is mostly due to
uncertainties
in quantifyingthe smallgradientsof CO2 above
the forest (for detailed discussionof the error analysissee
Goldsteinet al., [ 1996].).
3. Measurements
The analyticalsystemfor automatedin situ measurements
of hydrocarbon mixing ratios and gradients has been
describedin detail elsewhere[Goldsteinet al., 1995a]; thus
only a brief descriptionwill be given here. Recovery of
isoprenein the initial instrumentconfigurationwas low dueto
losson the innersurfacesof stainlesssteeltubing.The lossof
isoprenewas eliminatedin May 1995 by replacingall 1/16
inch stainless steel tubing in the sampling and
preconcentration
systemwith stainlesssteeltubing lined with
fusedsilica(Silcosteel).
Air wasdrawncontinuouslyfrom two inlets(24 and29 m).
Samplesfor analysiswere extractedfrom the inlet lines and
passedthrough glass cold traps at-20øC and Ascarite II
(Thomas Scientific) traps to remove H20, 03, and CO2.
Samples
werecryogenically
preconcentrated
on dualtraps(40
mL min-] of air for 10 minutesonto fused silica lined 1/16
31,048
GOLDSTEIN ET AL ßSEASONAL COURSE OF ISOPRENE EMISSIONS
inch OD stainlesssteeltube, Silcosteel),and injectedinto a
gas chromatograph
with dual flame ionizationdetectors
(HewlettPackard5890 seriesII). Chromatographic
separation
was accomplished
using30 m PLOT GS-AluminaMegabore
capillary columns(J+W Scientific). Every fifth pair of
sampleswas taken from the same altitude (29 m) by
switchinga valve nearthe inlet of the 24 m samplingline in
order to determinethe null for the observedmixing ratio
gradient. The measurement system could operate
continuously
andunattendedfor 2 weeks,althoughdatawere
normallydownloadedat 6 day intervals. Mixing ratios for
most hydrocarbonswere determinedusing relative response
factors[Ackman,1964, 1968; Dietz, 1967] referencedto an
intemal neohexanestandard(Scott-Martin,National Institute
of Standardsand Technology(NIST) traceable+ 2%) added
to every samplenear the sampleinlet by dynamicdilution.
Responsefactorsfor isoprenewere determinedfrom dynamic
dilutionof isoprenestandards(Scott-Martin,NIST traceable+
2%) addedto ambientair samplesnearthe sampleinlet at the
top of the tower and nearthe instrumentat the bottomof the
towerperiodicallyfrom May to November1995.
The analyticalsystemwascheckedfor contamination
daily
by runningzero-airblanks. In addition,the Teflon sampling
tubeswerecheckedfor contamination
andmemoryeffectsby
introducingzero-airat the sampleinletson top of the tower.
No contaminationor memory effects were detected for
isoprene,andtherewas no detectablelossof isoprenein the
Teflon samplingtubes.
The accuracyof the isoprenemeasurements
was estimated
to be +8%, based on the cumulative uncertaintyof the
isoprenestandard,measurements
of standardadditionflows,
and the integrityof isoprenein the samplingand analysis
process.Measurementprecisionwas approximately3% at 1
ppbv,5% at 0.5 ppbv, 10% at 0.2 ppbv,and20% for mixing
ratios less than 0.1 ppbv, as determinedby the variance
between measurements
taken from the samelevel everyfifth
injection.The detectionlimit wasapproximately
0.01 ppbv.
4. Effects of Isoprene Source Spatial
Heterogeneity
An inherent weakness in the gradient approach for
quantifyingisoprenefluxesis that the lowersamplingintake
has a differenteffective footprintthan the upper sampling
intake. If the isoprenesource from the forest canopy is
heterogeneous,
this differencein footprintsfor the sample
intakes could bias the measured fluxes. To determine whether
isopreneemissionwas heterogeneous,
we examinedwhether
fluxes
varied
with
wind
direction
from
the tower.
We
calculateda normalizedemissionrate as measured/(modeled
usingBEIS2E), effectivelyremovingthe influenceof light
and temperature from the emission measurements,as
represented
by the model. Data were parsedinto four wind
quadrants
(0o-90ø (NE), 90ø-180
ø (SE), 180ø-270
ø (SW), and
270ø-360ø (NW)) and the mean normalized emissionrates
were compared.The mean midday(1000 - 1500 EST, June
15 to August31) measuredflux was 35% higher than the
model. The meannormalizedflux from the SW, NW, and NE
wind quadrants
agreedwithin & 6%. The SE wind quadrant
was 20%
lower than the mean of the other three wind
quadrants(NE 1.29, SE 1.16, SW 1.39, NW 1.42). The
percentageof data from eachquadrantwas 10%, 18%, 45%,
and 28%, respectively.This comparisonof normalizedflux
by wind quadrantsuggeststhat systematicisopreneflux
errorsdueto spatialheterogeneity
of isoprenesourcesat our
siteareunlikelyto be higherthan20% andare probablyless
than 10%.
5. Results and Discussion
5.1. IsopreneAmbient Mixing Ratios
Isoprenemixing ratioshad a diurnal cycle with maximum
mixingratiosin the afternoon,andminimummixingratiosat
night (Figure 1). The meansummerdaytimemixing ratio at
24 m was4.4 ppbv(averagedfrom 1200 to 1600 EST, days
170-250);middaymixingratiosexceeded15 ppbvon 3 days.
Mixing ratio gradients of CO2 were measured The daytimeisoprene
mixingratioat 29 m wastypically25%
simultaneouslywith the hydrocarbongradientsusing a lower than the 24 m mixing ratio. Summerdaytimenull
differentialinfraredgas analyzer (LICOR 6251), with air gradientmeasurements
showedexcellentagreementbetween
from 29 m passedthroughthe referencecell and air from 24
the parallelmeasurement
channelswith a meanratio of 1.007
m throughthe samplecell. Water vapor was removedfrom & 0.023 (standarddeviation). Nighttimemixingratiosvaried
the air samplesusing nation dryers,and the sampleswere with atmospheric
stabilityandthe previousday's emissions.
assumedto be at a commontemperaturebeforeanalyzingfor On windy nightsisoprenemixing ratioswere usuallybelow
CO2. The null gradientwas measuredafter everysampling the instrument
detectionlimit (< 10 pptv). On stablenights
period by passingair from 29 metersthroughboth cells. that followeddayswith significantemissions,
mixingratios
Instrumentgain was determinedby sequentialstandard weretypicallybetween0.1 and2 ppbvanddecreased
slowly
additionof CO2 to the sampleandthenthe referenceair. The until isopreneemissionsresumedin the morning.Therewas
standarddeviationsof the zero gradientmeasurements
were no observablegradientfrom 24 to 29 m duringthe stable
determinedby comparingthe null gradientmeasuredevery nights(e.g. night of day 184-185), indicatingthat nighttime
fifth samplingperiod(whenhydrocarbon
null gradientswere isoprene
losswasduemainlyto chemicalreactions
(probably
determined)to the zero measurement
directlyfollowingthat with ozoneor nitrate)ratherthan directdeposition
to the
period. The standarddeviationin the zero measurements
for forest.
CO2 (0.18 ppm) was • 20% of the meanmiddaygradients(The seasonal
coursesof isoprenemixing ratio (24 m),
0.9 ppmCO2). Flux determinations
werenot attemptedwhen carbondioxideflux, percentPAR intercepted
by the canopy
observedgradientswere very small, i.e., within 1 standard (measured
from fixed platformsat 30 and 15 m, aboveand
deviation
of zero. TheCO2 fluxesandgradients,
andall the
NMHC measurements
are reportedas mole fractionsrelative
to dry air at a commontemperature,avoidingthe need for
densitycorrections
dueto gradients
in temperature
or H20
[Webbet al., 1980].
belowthecanopy),
andairtemperature
areshownin Figure2
asmiddaymeanvalues(1000to 1500EST). Percentof PAR
intercepted
indicates
the changein leaf densityin the forest,
and increased
from lessthan 40% as the leavesbeganto
emerge(day 140) to 90% when the canopywas fully
GOLDSTEINET AL.' SEASONALCOURSEOF ISOPRENEEMISSIONS
31,049
o
o' (
o
,
,
o
0
o
e
o
ooo
o
c)
c)
150
200
25O
300
Day of Year 1995
Figure 2. Isoprenemixing ratio at 24 m, CO2 flux, percentphotosynthetically
active radiation(PAR)
interceptedby the canopy,and air temperatureat 29 m (May to November1995, middaymeansfrom 1000 to
1500 EST).
developed(approximatelyday 160). It remainedconstant decreaseas the leavesage, similar to observationsof aspen
throughthe summeruntil the leavesbeganto senescearound leaves by Mortson et al. [1994]. Isoprene mixing ratios
decreased
rapidlyafter day 250, coincidingwith decreasedair
day 240, and continuedto declineslowly asthe leavesfell off
thetrees.CO2 uptakefolloweda similarpatternin thespring temperature,nighttime temperaturesfalling below 10 øC
uptakeby the canopydeclining,and
beginningaboutday 140 and reacheda maximum rate after regularly,photosynthetic
beginning.
day 160. Photosynthetic
uptakedecreasedafter day 230 as leaf senescence
the leavesbeganto senesce,and shut off completelyaround
day 290. The first isopreneemissionsof the growingseason 5.2. Isoprene Emission Rates
were detectedon day 152, when air temperatureexceeded
Plate 1 showsisopreneemissionratesduring midsummer
25øC after two nights in a row with temperaturesremaining (days 165-230) as a function of both light and temperature
above13øC. Isopreneemissionsdid not beginuntil the forest measuredabovethe canopyat 29 m. Isopreneemissionswere
canopyhadreacheda maturestate,2 weeksafter leavesbegan not detectedduringperiodswith low light (PAR < 300 pE mto emerge and photosynthesishad begun. The seasonal 2 s-i) or air temperaturebelowapproximately13 øC.Emission
changein isopreneemissionis not a functionof temperature rates increasedexponentiallywith temperatureand linearly
and light alone. The processes
that controlisopreneemission with PAR up to a saturationpoint of approximately1000 gE
from oaks are apparentlynot active in young leaves and m-2 s-i, as has been observed in other studies. However,
31,050
GOLDSTEINET AL.' SEASONALCOURSEOF ISOPRENEEMISSIONS
o
o
o
o
o
0
200
400
600
10 moleccm-2s-1
o
o
o
-
o
o
-
0
•
10
15
2o
25
3O
Temperature(øC)
Plate 1. Isopreneflux asa functionof lightandtemperature
(June14 to August18, 1995).
has a similar compositionto Harvard Forest
maximumemissionsdid not necessarilyoccursimultaneously Massachusetts
with maximumtemperatureor maximumPAR. On average, (80% is probably more accurate, based on oak tree
isopreneemissions
were zero at night,beganafter sunrise, distributionsin the databaseof Geron et al. [1994]), our
increased
throughthe morningwith increasingair temperature resultssuggestthat 105 metric tons yr-I of isopreneare
and PAR, reacheda peak in the afternoonbetween the emitted from Massachusetts forests. The National Acid
Program(NAPAP)[1985] estimates
maxima in air temperatureand PAR, and declined with PrecipitationAssessment
decreasing
PAR towardthe end of the day (Figure3). The that 1.1 x 105 metric tons of volatile organiccarbon(VOC)
sourcesduringthe summerin
meanisopreneemissionoverthe wholegrowingseason(24 are emittedfrom anthropogenic
hour mean from June to October 1995, calculated by Massachusetts.Thus, on the averageday from Junethrough
integrating
themean
diurnal
flux(Figure
3))was32.7x 10iø October,biogenicisopreneemissionsareapproximatelyequal
moleculescm-2s-i, giving an annualemissionrate of 48 kg to total anthropogenicVOC emissions;however, on hot
isoprene
ha-I yr-1,with emissions
occurring
mainlyfrommid- summerafternoonswhen the highestozone episodesoccur,
isopreneemissions
increasedramatically.For example,mean
Junethroughmid-September.
aroundmidday(PAR > 1000 pE m-2
Previousabove canopyisopreneflux studiesfrom a wide isopreneemissions
cm-2s-1at 20,25,
variety of ecosystems
in the United Statesyield estimates were89, 176,and307x 1010molecules
rangingfrom 30 to 300 x 1010 moleculescm-2 s-I when and 30 + 2 øC, respectively. At 30 øC, isopreneemission
standardized
to air temperatures
of 30øC and abovecanopy rates are approximately 9 times the 24 hour average
PAR of 1000pE m-2s-I [Guenther
et al., 1996a,b,c;Lambet summertimeemissionrate. In addition,isopreneis estimated
al., 1985;Baldocchiet al., 1995;summarized
in Geronet al,
to be 2.5 times as effective for photochemicalozone
1997]. Our measurements
are nearthe high end of these productionper atom of carbonas the typical mix of VOCs
values
(307x 1010molecules
cm-2
s-1at30+ 2 øCwithPAR foundin urbanair [NationalResearchCouncil(NRC), 1991].
> 1000 gE m-2 s-i) and are in reasonable
agreement
with Assuming that anthropogenic VOC emissions are
those at sites with similar oak abundance.
substantially less temperature dependent than biogenic
isopreneemissions,
our measurements
suggestthat isopreneis
5.3. Regional Significanceof Isoprene Emissions
more importantthan anthropogenic
VOCs for photochemical
Redoak is the dominantisoprene-emitting
plantat Harvard ozone production on hot summer days in much of
Forest and in Massachusetts.
If we assume that all of
Massachusetts
(outsideof the largesturbanareas).
GOLDSTEIN ET AL.' SEASONAL COURSE OF ISOPRENEEMISSIONS
Isoprene
Flux
31,051
o
o
Temp
......PAR
.//•
iiII1•'•'•'••.••I•
:
\
/
o
o
--
0
--
0
0
O0
\\
III
',
-o ø
-
\\
_•.
\
\
-
o
--
o
0
-
o
<
\
\
\
O
--
I
/
.- - •-'"
0
,
/
5
10
15
20
Hour of Day
Figure3. Meandiurnalcycleof isopreneflux (+ standard
error)alongwith temperature
at 29 m andincident
photosynthetically
activeradiation(PAR) (June1 to October31, 1995). Data wereparsedinto 2-hourtime
windows.
6. Comparison to Net EcosystemProduction
carbonby a mixtureof plantsthatdo not all releaseisoprene,
and this fraction is certainly higher for plants that emit
The percent of gross carbon uptake returned to the
atmosphereas isopreneis strongly temperaturedependent
(Figure 4, calculatedas isopreneflux divided by gross
ecosystemproductionduring summermidday periods). In
HarvardForest,approximately1% of the grosscarbonuptake
is releasedas isopreneat 25øC, and the percentageincreases
with temperature.Thesemeasurements
reflectthe uptakeof
isoprene.
Annualisopreneemissions
(42 kg C ha-1yr-1)equal2% of
the net carbonuptakeat HarvardForest(2200 kg C ha-1yr-1,
fiveyearmeanvalue[Goulden
et al., 1996b]).Thepercentage
of photosynthesized
carbonthat is releasedas isoprenehas
beenreportedin many studiesto be between0.1 and 3%, and
occasionallyhigherfor leaf level measurements
[Sharkeyet
o
o
O-
15
20
25
Air Temperature (øC)
Figure4. Percentgrosscarbonuptakereleased
asisoprene
asa functionof air temperature.
31,052
GOLDSTEIN ET AL.: SEASONAL COURSE OF ISOPRENE EMISSIONS
to the environmentthat are not capturedby the
al., 1996]. Isopreneemissions
do not represent
a major responses
carbonlossfromthisecosystem
but arepotentiallysignificant
on a shortertime scaleduringhot summerdays.
model.
Figure5a showsan hourlymeanof the measurements
and
the model results for both model scenarios(BEIS2 and
BEIS2E), and 5b showsthe normalizedemissionrate (using
6.1. Isoprene EmissionModel Scenarios
BEIS2E) for summer 1995 (90% confidenceintervalsare
The isopreneemissionmodelusedis BEIS2 as described
indicated). A diurnalcourseof the normalizedemissionrate
by Geronet al. [1994]. Isopreneemissionratesare assumed
is apparent, increasingin the morning,maximizingaround
to be 70 ggC (g-foliardry mass)-Ih-1 standardized
for PAR
values
of 1000•tmolm-2s-•and30øCfortheoaks,and0.1
•tgC(g-foliardry mass)-•h-• for the otherspecies.
Thisforest
average basal emission factor (EF) was determinedby
multiplyingthe fractionof oak and non-oakbasalareasby
their corresponding
EFs, yieldingan averageEF of 25 •tgC
(g-foliardry mass)-•h-1. Sinceoakshavegreatercrownarea
in proportionto basalarea comparedto other genera(e.g.,
birch, pine) this likely representsan underestimate
of oak
foliageon the orderof 5% [Geronet al., 1997].
The empiricalalgorithmsof Guentheret al. [1993] were
usedto adjustemissionratesto ambientPAR andtemperature
conditions.Exponential decay algorithms are applied to
reducePAR and specificleaf weight at lower levelswithin
forestcanopies[Geronet al., 1994]. The measured
LAI of 3.5
was used in the model simulations.
This model assumesthat leaf temperatures
are equivalent
to ambientair temperatures
abovethe forestcanopy. Since
leaf temperatures
candiffer substantially
from surrounding
air
temperature (especially in the upper canopy), the leaf
temperatureenergybalanceof Gatesand Papian [ 1971] used
by Lamb et al. [1993] and Geron et al. [1997] was also used
to estimate leaf temperature in scenario BEIS2E. Vertical
gradientsof humidity, wind speed, solar radiation, and air
temperaturewere estimatedfrom measurements
taken above
the canopy by applyingthe profilesof Lambet al. [1993].
6.2. Comparison With Models
Measuredisopreneemissionswere 40% higher than the
BEIS2E model and 100% higher than the BEIS2 model at
midday and showeddistinctlydifferent diurnal and seasonal
emissionpatterns.The currentoak basalemissionfactorof 70
•tgC g-1 h-• hasan estimateduncertaintyof + 50%. However,
recentscalingstudiesat other siteshave suggestedthat rates
of 100 [tgC g-1 h-1 or more may be more realistic[Guenther
et al., 1996c; Lamb et al., 1996; Geron et al., 1997]. Indeed,
at Harvard Forest, researchers using environmentally
controlledleaf cuvettesystems(for methods,seeHarley et al.
[1997] and Sharkeyet al. [1996]) measuredratesduringJuly
1995 and 1996 rangingfrom approximately70 to over 160
[tgC g-• h-• (C.D. Geron et al., manuscriptin preparation,
1998). Taken together, these results suggestthat a peak
summeroak basalemissionrate of 100 [tgC g-• h-1 + 50%
wouldbe moreappropriate
for the model.Thesehigherbasal
noon,anddecreasing
in the afternoon(the normalizedratesat
night, 2000 - 0500 hours, are not included in the figure
because fluxes are zero for both the measurements and the
model). Most likely, this apparentdiurnalcourseis due to
underestimating
leaf temperatureat high valuesof solarinput
using the energy balancemodel [see Geron et al., 1997].
This wouldresultin the observedpatternfor diurnalchanges
in the normalizedemissionrate and possiblyaccountsfor the
BEIS2E model underestimates of midday fluxes.
Alternatively, the isoprene emitting plants could have
additional physiologicalfunctionsrelated to their diurnal
cycleor incomingPAR that are not represented
by the model,
or the gradientflux methodcouldbe underestimating
fluxes
during the more stable periods in the morning and late
afternoon.It is most importantfor the modelsto accurately
predict isopreneemissionsduring midday, when rates are
highest; thus the rest of our discussionwill focus on
comparingmiddayemissionrates(1000-1500).
Figures 6a and 6b show the seasonal course of the
measuredand modeled isopreneemission rates, and the
normalizedmidday emissionrate (1000-1500 EST, weekly
mean valueswith 90% confidenceintervals). The seasonal
courseof the normalizedemissionrate reachedits peak 4
weeksafterleaf out and2 weeksafteremissions
began.From
day 165 to 230 the normalizedrate remainedrelatively
constant. The normalizedrate decreasedsteadilyafter day
230 as the leavessenesced,and essentiallywent to zero by
day 300. The forestwas at its maximumisopreneemission
potentialfor approximately65 days. The seasonalreduction
in normalizedisopreneemissionafterday 230 coincidedwith
seasonally decreasing air temperature and nighttime
temperaturesfalling below 10 øC. The current model
underpredictsemissionsduring most of the summer and
overpredictsemissions in the spring and fall.
These
predictionscould lead to significanterrors in modeling
photochemical03 productionand in assessing
the need for
legislationto regulateanthropogenicVOC emissions. The
seasonal
durationof strongisopreneemissionlastedonly 100
dayswith muchlower ratescontinuingfor approximately30
days,similarto theCO2 uptakepatternof theoaks[Basso•v,
1995], and shorterthan the approximately150 daysof active
CO2 uptakebythewholeecosystem.
Thisseasonal
emission
cycle needsto be includedin isopreneemissionmodelsto
accuratelysimulateemissionrates.
emission rates could account for the difference between the
Figure 6b comparesthe seasonalchangesin normalized
measuredandmodeledisoprenefluxes.
midday emissionrate to an estimateof the leaf area index
The diurnal and seasonalcyclesof the isopreneemission (LAI) at Harvard Forest (estimated from PAR measured
rate are examinedby normalizingthe measuredisoprene abovethe canopyand at midcanopy,then scaledto summer
emissionsto those calculatedwith the BEIS2 models (as maximumLAI measuredby leaf litter collection),and the
describedabove). This normalizedemissionrate removesthe normalizeddifferencevegetationindex (NDVI, biweekly
effectsof temperatureand PAR as representedby the model. compositeswith 1.1 km resolution,data for Harvard Forest
If the normalized emission rate equals 1 then the extracted
by K. MoorefromU.S. GeologicalSurveyNational
measurements and model are in perfect agreement. MappingDivisionEROSdatacenter).TheNDVI composites
Deviations from 1 indicate changesin the basal EF or can be usedto estimateseasonalvariationin leaf area index,
GOLDSTEINET AL.' SEASONALCOURSEOF ISOPRENEEMISSIONS
o
o
-
31,053
x
x
measured
model(BEIS2E)
model (BEIS)
o
x
,
•..
v
0
i•>
-
i
i
i
0
5
10
15
20
Hour of Day
i
I
10
15
20
Hour of Day
Figure 5. Diurnalpatternof (a) measuredandmodeled(BEIS2 andBEIS2E; seetext for modeldescriptions)
isopreneemissions,
and(b) normalized(measured/modeled
(BEIS2E)) isopreneemissions
(June24 to August
28, 1995, hourlymeanvalueswith 90% confidenceintervals).
chlorophyllcontent,and biomass[Moore et al., 1996]. The
LAI and NDV! data have a seasonalpatternthat is notably
consistent with the normalized midday emission rate.
Isopreneis not emittedfor the first few weeksafter leafout
(LAI < 1.5), then increaseswith increasingLAI and NDVI.
Isopreneemissions
diminishin the fall in a mannerconsistent
with decreasingLAI and NDVI, but isopreneemissionsend
beforethe leaveshave fallen from the trees (LAI < 1.5) and
before photosynthesis
has ceased. Our findings suggestthat
seasonalchangesin isoprene emissionscould be modeled
using NDVI
data, or possibly by finer scale
multispectral/temporal
imagery,with minor adjustmentsfor
emission initiation in the spring and cessationin the fall.
Such metrics should be incorporatedinto future empirical
modelsof isopreneemission.
7. Conclusions
Whole canopy isoprene emissionswere measuredfrom
Harvard Forest over the entire 1995 growing season. The
seasonalduration of isoprene emissionswas substantially
shorterthan net photosyntheticuptake. Comparisonof the
measurementswith the EPA BEIS2E model gave a clear
definitionof temporalchangesin the emissionrate which are
not capturedby the model, and could be used to improve
future models. Young leaves did not emit isoprenefor 2
weeks and did not reach their maximum
rate for another 2 weeks.
normalized
The normalized
emission
emission
rate
remained constant for approximately 2 months, then
decreasedsteadily through Septemberand October before
emissionsceased. These seasonalchanges in isoprene
emissionratesare importantfor two major reasons:(1) they
dramaticallyaffect the regionalozone productionefficiency
per NOx [seeHirsch et al., 1996] and are thus crucialfor the
developmentof air quality control strategies,and (2) they
provide constraints for understandingthe physiological
mechanismof isopreneproduction. Current EPA models
underestimateisoprene emissions at Harvard Forest in
midsummerand overestimatein springand fall. The model
underestimatesin midsummer are likely due to an
31,054
GOLDSTEIN ET AL.' SEASONAL COURSE OF ISOPRENE EMISSIONS
o
iA3
-
x
x
o
o
measured
o
x
model (BEIS2E)
model (BEIS2)
A
ß
o
o
•
A'/
_
,
AA
X
A A
ß
o
,O'
A
O
,
,
O/
A
o
to
X
A
o
-
o
/
A
O
x
X
-
'A
O
A'
X
i
i
i
I
150
200
250
3OO
Day of Year 1995
A
•
x
o
measured/BEIS2E
A
LAI
+
NDVI
+
i
150
Figure 6.
i
200
Day of Year 1995
i
i
250
300
Seasonalcourseof (a) measuredand modeled(BEIS2 and BEIS2E; see text for model
descriptions)
middayisoprene
emissions
(1000to 1500EST,weeklymeanvalues),and(b) normalized
(measured/modeled
(BEIS2E))midday
isoprene
emissions
(with90%confidence
intervals),
andanindexof
LAI and measurements of NDVI.
underestimate
in the basalemission
rate.Futureempirical from anthropogenic
sourcesin Massachusetts
accordingto the
modelsof isopreneemissionshouldincorporate
seasonal NAPAP [1985] emissioninventory(1.1 x 105 metric tons).
adjustmentfactors.
Moreover,the importance
of biogenicisopreneemissions
The relativemagnitude
of anthropogenic
and biogenic increases
dramaticallyon hot summerafternoons,
coincident
VOC emissions
mustbe determined
accurately
to develop with periodsof peakphotochemical
activity. In addition,
usefulcleanair legislation
for reducingtropospheric
ozone. isoprene is approximately2.5 times as effective for
In this study,isopreneemissionsfrom Massachusetts
forests photochemical
ozoneproduction
per atomof carbonas the
were estimatedto be 105 metric tons yr-1, with emissions typicalmix of VOCs foundin urbanair. Assuming
that
occurringmainly from mid-Junethroughmid-September.anthropogenic
VOC emissionsare substantiallyless
This is equivalentto the total summertimeVOC emissions temperature
dependent
thanbiogenicisoprene
emissions,
our
GOLDSTEIN ET AL.: SEASONAL COURSE OF ISOPRENE EMISSIONS
31,055
Guenther, A.B., R.K. Monson, and R. Fall, Isoprene and
monoterpene emission rate variability: Observations with
eucalyptusandemissionrate algorithmdevelopment,
d. Geophys.
Res., 96, 10,799-10,808, 1991.
Guenther,A.B., P.R. Zimmerman,P.C. Harley, R. Monson, and R.
Acknowledgments.This work was supportedby grantsto
Fall, Isopreneandmonoterpene
emissionratevariability: Model
HarvardUniversityfrom the U.S. Departmentof Energy's(DOE)
evaluations and sensitivity analyses, d. Geophys, Res., 98,
National Institute for Global Environmental Change (NIGEC)
12,609-12,617, 1993.
through the NIGEC Northeast Regional Center at Harvard Guenther,A., P. Zimmerman,L. Klinger, J. Greenberg,C. Ennis, K.
University(DOE CooperativeAgreementDE-FC03-90ER61010),
Davis, W. Pollock, H. Westberg, E. Allwine, and C. Geron,
by the NationalAeronauticsand SpaceAdministration(NAGWEstimatesof regional natural volatile organic compoundfluxes
3082), the NationalScienceFoundation(BSR-89-19300),the Long
from enclosureand ambient measurements,
d. Geophys,Res.,
results indicate that isoprene is more important than
anthropogenic
VOCs for photochemical
ozoneproduction
on
hotsummerdaysin muchof Massachusetts.
TermEcologicalResearch
programat the HarvardForestfundedby
the National ScienceFoundation(BSR-88-11764), and by Harvard
University(HarvardForestandDivisionof AppliedSciences).The
authorswish to thank David Fitzjarraldand KathleenMoore of the
Atmospheric
SciencesResearchCenterat the StateUniversityof
New York at Albany for helpful discussions
and for providing
NDVI
data.
References
Ackman, R. G., Fundamentalgroups in the responseof flame
ionizationdetectorsto oxygenatedaliphatichydrocarbons,
J. Gas
Chromatog.,2, 173-179, 1964.
Ackman,R. G., The flame ionizationdetector:Further commentson
molecularbreakdownand fundamentalgroup responses,J. Gas
Chromatog.,6, 497-501, 1968.
Baldocchi,D., A. Guenther,P. Harley, L Klinger, P. Zimmerman,B.
Lamb,andH. Westberg,The fluxesand air chemistryof isoprene
above a deciduous hardwood forest, Philos. Trans. R. Soc.
London.,Ser. ,4 350, 279-296, 1995.
Bassow, S.L., Canopy photosynthesisand carbon cycling in a
decidousforest:implicationsof speciescompositionand rising
concentrations
of CO2, Ph.D. thesis,Harvard Univ., Cambridge,
Mass., 1995.
Chameides,W., R. Lindsay,J. Richardsonand C. Kiang, The role of
biogenichydrocarbons
in urbanphotochemical
smog:Atlanta as
a casestudy,Science,241, 1473-1475, 1988.
Dietz, W. A., Responsefactorsfor gas chromatographic
analysis,J.
Gas Chromatog.,1, 68-71, 1967.
Fehsenfeld,F., et al., Emissionsof volatile organiccompoundsfrom
vegetation and the implications for atmosphericchemistry,
GlobalBiogeochem.
Cycles,6, 389-430, 1992.
Fuentes,J.D., D. Wang, G. Den Hartog, H.H. Neumann, T.F. Dann,
and K.J. Puckett,Modelled and field measurements
of biogenic
hydrocarbon
emissionsfrom a Canadiandeciduousforest,,4tmos.
Environ., 29(21), 3003-3017, 1995.
Gates, D.M., and L.E. Papian, Atlas of Energy Budgets of Plant
Leaves,Academic,SanDiego, Calif., 1971.
Geron, C., A. Guenther and T. Pierce, An improved model for
estimatingemissionsof volatile organiccompoundsfrom forests
in the easternUnited States,J. Geophys.Res.,99, 12,773-12,792,
1994.
101, 1345-1359, 1996a.
Guenther,A., J. Greenberg,P. Harley, D. Helmig, L. Klinger, L.
Vierling, P.Zimmerman,and C. Geron, Leaf, branch,stand,and
landscapescale measurementsof volatile organic compound
fluxesfrom U.S. woodlands, Tree Physiol.,16, 17-24, 1996b.
Guenther,A., et al., Isoprenefluxesmeasuredby enclosure,relaxed
eddy accumulation,surface-layergradient and mass balance
techniques,
J. Geophys,Res., 101, 18,555-18,567,1996c.
Harley, P., A. Guenther,and P Zimmerman,Environmentalcontrols
over isopreneemissionfrom sun and shade leaves in a mature
white oak canopy,Tree Physiol.,17, 705-714, 1997.
Hirsch, A.I., J.W. Munger, D.J. Jacob,L.W. Horowitz, and A.H.
Goldstein,Seasonalvariationof the ozoneproductionefficiency
per unit NOx at HarvardForest,Massachusetts,
J. Geophys.Res.,
101, 12,659-12,666, 1996.
Khalil, M.A.K., and R.A. Rasmussen, Forest hydrocarbon
emissions:
Relationships between fluxes and ambient
concentrations,d. ,4it Water Manage. ,4ssoc.,42, 810-813, 1992.
Knoerr, K.R., and F.L. Mowry, Energy balance/Bowen ratio
techniquesfor estimatinghydrocarbonfluxes, in ,4tmospheric
BiogenicHydrocarbons,edited by J.J. Bufalini and R.R. Arnts,
pp. 35-52, Butterworth,Stoneham,Mass., 1981
Lamb, B., H. Westberg,and G. Allwine, Biogenic hydrocarbon
emissions
from
deciduous
and coniferous
trees in the United
States,d. Geophys.Res.,90, 2380-2390, 1985.
Lamb, B., H. Westberg,and G. Allwine, Isopreneemissionfluxes
determinedby an atmospherictracer technique,,4tmos.Environ.,
20, 1-8, 1986.
Lamb, B., D. Gay, H. Westberg, and T. Pierce, A biogenic
hydrocarbonemission inventory for the U.S. using a simple
forestcanopymodel, ,4tmos.Environ., 27`4, 1673-1690, 1993.
Lamb, B., et al., Evaluationof forestcanopymodelsfor estimating
isopreneemissions,J. Geophys.Res., 101, 22,787-22,797, 1996.
McKeen, S.A., E.Y. Hsie, and S.C. Liu, A studyof the dependence
of rural ozone on ozone precursorsin the easternUnited States,
•. Geophys.Res.,96, 15,377-15,394, 1991.
Monson, R.K., and R. Fall, Isopreneemissionfrom aspenleaves:
Influence of environment and relation to photosynthesisand
photorespiration,
Plant Physiol., 90, 267-274, 1989.
Monson,R.K., C.H. Jaeger,W.W. Adams III, E.M. Driggers,G.M.
Silver, and R. Fall, Relationshipsamongisopreneemissionrate,
photosynthesis,
and isoprenesynthaseactivity as influencedby
temperature,Plant Physiol.,98, 1175-1180, 1992.
Monson, R.K., P.C. Harley, M.E. Litvak, M. Wildermuth, A.B.
Geron, C.D., et al., Biogenic isopreneemission: Model evaluation
in a southeasternUnited States bottomland deciduous forest, J.
Guenther, P.R. Zimmerman, and R. Fall, Environmental and
Geophys.Res.,102, 18,889-18,901,1997.
Goldstein, A.G., B.C. Daube, J.W. Munger, and S.C. Wofsy,
developmentalcontrols over the seasonalpattern of isoprene
Automated in-situ monitoring of atmospheric non-methane
emissionfrom aspenleaves,Oecologia,99, 260-270, 1994.
hydrocarbonconcentrations
and gradients,J. ,4tmos.Chem.,21,
Moore, K.E., D.R. Fitzjarrald, R.K. Sakai, M.L. Goulden, J.W.
43-59, 1995a.
Munger, and S.C. Wofsy, Seasonalvariation in radiative and
turbulent exchange at a deciduous forest in central
Goldstein,A. H., C.M. Spivokovsky,and S.C. Wofsy, Seasonal
Massachusetts,
d. ,4ppl. Meteorol., 35, 122-134, 1996.
variationsof nonmethanehydrocarbonsin rural New England:
Constraints on OH concentrations in northern midlatitudes, J.
Munger, J.W., S.C. Wofsy, P.S. Bakwin, S-M. Fan, M.L. Goulden,
B.C. Daube, A.H. Goldstein,K.E. Moore, and D.R. Fitzjarrald,
Geophys.
Res.,100,21,023-21,033,1995b.
Goldstein,A.H., S.M. Fan, M.L. Goulden,J.W. Munger, and S.C.
Atmosphericdepositionof reactivenitrogenoxidesand ozonein
a temperate deciduous forest and a subarctic woodland, 1,
Wofsy, Emissionsof ethene,propene,and 1-buteneby a
midlatitudeforest,J. Geophys.
Res.,101, 9149-9157,1996.
Measurements
and mechanisms,
d. Geophys.Res., 101, 12,63912,657, 1996.
Goulden,M.L., J.W. Munger, S.M. Fan, B.C. Daube, and S.C.
Wofsy, Measurements
of carbonstorageby long-termeddy Munger, J.W., S-M. Fan, P.S. Bakwin, M.L. Goulden, A.H.
correlation: Methods and a critical evaluation of accuracy,
Goldstein,A.S. Colman, and S.C. Wofsy, Regionalbudgetsfor
nitrogenoxidesfrom continentalsources:Variationsof rates for
GlobalChangeBiol., 2, 169-182, 1996a.
oxidationand depositionwith seasonand distancefrom source
GouldenM.L., J.W. Munger, S.M. Fan, B.C. Daube, and S.C.
regions,d. Geophys.Res., 103, 8355-8368, 1998.
Wofsy, Effectsof interannualclimatevariabilityon the carbon
dioxideexchangeof a temperatedeciduousforest,Science,271, National Acid PrecipitationAssessmentProgram (NAPAP), The
1576-1578, 1996b.
1985 NAPAP emissionsinventory(version2): Developmentof
31,056
GOLDSTEIN ET AL.: SEASONAL COURSE OF ISOPRENE EMISSIONS
the annualdata and modelers'tapes,EPA-600/7-89-OI2a, pp. 219, U.S. Environ. Prot. Agency, Washington, D.C., 1985,
(AvailablefromNatl. Tech.Inf. Serv.,Springfield,Va.).
NationalResearchCouncil(NRC), Rethinkingthe OzoneProblemin
Urban and Regional Air Pollution, Natl. Acad. Press.,
Washington,D.C., 1991.
Pier, P.A., Isopreneemissionratesfrom northernred oak usinga
whole-treechamber,Atmos.Environ., 29(12), 1347-1353, 1995.
Roselle, S.J., T.E. Pierce, and K.L. Schere, The sensitivity of
regionalozonemodelingto biogenichydrocarbons,
d. Geophys.
Res., 96, 7371-7394, 1991.
Sanadze,G.A., Light-dependent
excretionof molecularisopreneby
leaves,Prog.Photosyn.Res.,II, 701-706, 1969.
Sharkey,T.D., and E.L. Singsaas,Why plants emit isoprene,
Nature, 374, 769, 1995.
Webb, E.K., G.I. Pearman, and R. Leuning, Correctionof flux
measurements
for densityeffectsdue to heat and water vapour
transfer,Q. d. R. Meteorol. Soc., 106, 85-100, 1980.
Winer, A., J. Arey, S. Aschman,R. Atkinson,W. Long,L. Morrison,
andD. Olszyk, Hydrocarbonemissionsfrom vegetationfoundin
Califomia'sCentralValley, final report,Calif. Air Resour.Board
contract A732-155, Statewide Air Pollut. Res. Cent., Univ. of
Calif., Riverside, 1989.
Wofsy, S.C., M.L. Goulden,J.W. Munger, S.M. Fan, P.S. Bakwin,
B.C. Daube, S.L. Bassow,and F.A. Bazzaz, Net exchangeof
CO2 in a mid-latitude
forest,
Science,
260,1314-1317,
1993.
Zimmerman,P.R., Determinationof emissionratesof hydrocarbons
from indigenous species of vegetation in the Tampa/St.
Petersburg
Fla. area,Final AppendixC, Rep. EPA-904/9-77-028,
U.S. Environ.Prot.Agency,ResearchTrianglePark,N.C., 1979.
Sharkey,T.D., F. Loreto,and C.F. Delwiche,The biochemistry
of
isopreneemissionfrom leavesduringphotosynthesis,
in Trace
Gas Emissionsby Plants,editedby T.D. Sharkey,E.A. Holland,
C. D. Geron,National Risk ManagementResearchLaboratory,
and H.A. Mooney, pp. 153-184, Academic,San Diego, Calif., U.S. EnvironmentalProtectionAgency,ResearchTrianglePark,NC
1991.
Sharkey,T.D., E.L. Singsaas,P.J. Vanderveer,and C.D. Geron,
Field measurements
of isopreneemissionfrom treesin response
to temperature
andlight, TreePhysiol.,16, 649-654, 1996.
Sillman,S., J.A. Logan,and S.C. Wofsy, The sensitivityof ozoneto
nitrogenoxidesand hydrocarbons
in regionalozoneepisodes,d.
Geophys.Res.,95, 1837-1851, 1990.
Singh,H.B., and P.B. Zimmerman,Atmosphericdistributionsand
sourcesof nonmethanehydrocarbons,in GaseousPollutants:
Characterizationand Cycling, edited by J.O. Nriagu, pp. 177235, JohnWiley, New York, 1992.
Tingey, D.T., M. Manning., L.C. Grothaus,and W.F. Bums, The
influenceof light and temperatureon isopreneemissionrates
from live oak,Physiol.Plant., 47, 112-118, 1979.
27711.
A. H. Goldstein,Division of EcosystemSciences,Departmentof
EnvironmentalScience,Policy and Management,151 Hilgard Hall,
University of
California, Berkeley, CA
94720-3110.
(ahg•nature.berkeley.edu).
M. L. Goulden,Earth SystemScience,Universityof California,
Irvine, CA 92697-3100.
J. W. Munger and S.C. Wofsy Departmentof Earth and
PlanetarySciences,Division of Engineeringand Applied Sciences,
HarvardUniversity,Cambridge,MA 02138.
(ReceivedOctober10, 1997;revisedMay 15, 1998;
acceptedMay 21, 1998.)