Effects of extended growing season and s
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. D22, PAGES 29,075-29,082, NOVEMBER
27, 1998
Effects of extended growing season and soil warming on
carbon dioxide and methane exchange of tussock tundra
in Alaska
StevenF. Oberbauer,Gregory Starr, and Eric W. Pop
Departmentof BiologicalSciences,FloridaInternationalUniversity,Miami
Abstract. The active seasonof tussocktundrawas extendedduring two growing seasons(1995
and 1996) by snowremovalin early seasonandpreventionof snowaccumulationin late seasonto
testthe effectsof a longergrowingseasonon tundracarbonexchange.Three treatmentswere
established:extendedseason,extendedseason+ soil warming,and controls. Soil warming was
accomplished
usingcold-frame,resistance
heatingwire installedthe yearprior to the initiationof
treatments.Diurnalcoursesof CO2 exchangeweremeasuredweeklyusinginfraredgasanalysis
with enclosedchambertechniques.Methanefluxeswere measuredtwo to threetimesa seasonalso
usingenclosuremethods.In 1995, snowmeltoccurredunusuallyearly, and snowremoval
treatmentsincreasedthe seasononly 9-10 days. In 1996 the early seasonwas increased
approximately24 days. As expected,thawdepth,soil temperature,
andplantgrowthwere greater
earlierin the extendedseasonandextendedseason+ soil heatingplots. Methanefluxes in both
seasons
were low but tendedto be higherin the extendedseasonandsoil heatedplots. Net
ecosystemCO2 fluxeswere similaramongtreatmentsearly in the season,with a tendencytoward
morepositivefluxes(systemloss)for the snowremovaland warmedplots,possiblydue to higher
belowgroundrespiration.Duringmidseason,fluxeswere similaramongthe treatments.Later in
the season,fluxesof extendedseasonandwarmedplotstendedto be lower (lesscarbonloss)than
controls,especiallyin 1995. Totaledover the season,however,the fluxesof the threetreatments
did not statisticallydiffer andrepresented
lossesto the atmosphere.Measurements
of dark
respirationin 1996 indicatethatbothrespirationanduptakewere increasedon the extendedseason
plots, resultingin similar net fluxes to controls.
1. Introduction
Evidencefor climate warming in the Alaskan Arctic is growing [Chapin et al., 1995; Oechel et al., 1993]. Among the
most importantconsequences
of warmingin the Arctic may be
changesin seasonlength resultingfrom shifts in the duration
of snow cover. The duration of snow cover is expected to
change, possibly dramatically, in responseto climate warming, althoughthe direction of changeis uncertain[Maxwell,
1992]. With no change in snowfall, snowmelt will probably
occur earlier.
However, if snowfall increases, the duration of
snowcover may remain the sameor even increase. Becauserelease from snow cover initiates soil thawing and triggers plant
growth, the timing of snowmelt directly affects the season
length. Changesin the start of the growing seasonwill be accompaniedby an extensionof the period of thawed soil at the
end of the seasondue to direct effects of warming [Kane et al.,
1991]. Maxwell [1992] suggeststhat snow-free periods in
the Low Arctic may increase 1 month or more in the next century.
Changes in season length may profoundly affect plant,
community, and ecosystemprocessesin the Arctic, both directly and indirectly. Seasonlength changesand warming di-
Copyright1998 by the AmericanGeophysicalUnion
Papernumber98JD00522.
0148-0227/98/98
JD-00522509.00
rectly affect plant phenologicalpatterns,that is, the timing of
plant developmentalevents. Plant activity is largely initiated
by snowmelt in early season,and dormancyat the end of the
seasonis triggeredby temperature,photoperiod,and/or internal cycles [Shaver and Kummerow, 1992]. Phenologydirectly
affects the overall pattern of plant and therefore ecosystem
carbon uptake [Kikuzawa, 1995]. Nevertheless,relatively little experimental effort has been devoted to the effects of potential phenological changes on ecosystem level processes
[Lechowicz and Koike, 1995]. The objective of this study was
to examine ecosystemcarbon exchangein responseto a growing season experimentally extended by snow removal. We
hypothesized that as a result of increased soil temperatures,
belowground respiration and therefore fluxes to the atmosphere would increase in response to an increased season
length.
Changes in seasonlength are of particular importance to
arctic ecosystemsbecause they are likely very sensitive to
climate changes. Seasonlengthsare short,effects of physical
factors are strong, and environmentalgradientsare very steep
in arctic ecosystems. Furthermore,the climate warming is expected to be greatesttoward the poles [Manabe and Stouffer,
1979; Maxwell, 1992]. Finally, high latitudeshold a significant fraction of terrestrialcarbonin an unstablestate,trapped
in cold soils, providing the potential for greater warming if
the carbon is mineralized [Billings, 1987; Chapin et al.,
1992; Gorham, 1991; Miller,
1981; Oechel et al.,
Oechel and Vourlitis, 1994; Post et al., 1982].
29,075
1993;
29,076
OBERBAUER ET AL.: EXTENDED GR()WlNG SEASONAND SOIL WARMING
2. Methods
2.1.
Study
These
tion of treatments in 1994. Over a 2 hour period each day,
eachplotreceived
anadditional
0.4MJm-e,whichis a signif-
Site
studies
were
conducted
near
the
Toolik
Arctic
ResearchStation (780 m elevation,68ø38'N, 149ø34'W) operated by the Institute of Arctic Biology of the University of
Alaska, Fairbanks. Toolik is located along the Dalton
Highway in the northern foothills of the Philip Smith
Mountains.
2.2.
Vegetation
The experiments were performed in cottongrasstussock
tundra. This tundra communitycoversa large proportionof
the North Slope of Alaska, eastern Canada, and northern
Eurasia. Becauseof its high soil carbon storage,tussocktundra hasbeen identifiedas one of the two mostimportanttundra
vegetation types from the perspectiveof global carbon cycling [Oechel and Billings, 1992]. The dominantgrowth form
is Eriophorum vaginaturnL. tussockswith intertussockspaces
filled with dwarf deciduousandevergreenshrubsandpeat-forming mosses. Several lines of evidence suggestthat cottongrass tundra is currently a source for atmosphericcarbon
[Schell, 1983; Grulke et al., 1990; Oecheland Billings, 1992;
Oechel et al., 1993]. Biomass and structural characteristicsof
this vegetationhave been describedby Hastingset al. [1989]
and Wein and Bliss [1974].
2.3.
Treatments
snow free.
et al., 1990].
The study plots were 1.5 x 1.5 m within six replicate
blocks.
Plot numbers
were limited
because of the intensive
labor required for snow removal. Similar plot sizes have
proven to be sufficient for studiesof ecosystemprocessesin
the Arctic [Chapin and Shaver, 1985; Grulke et al., 1990;
Oechel et al., 1993]. Each block contained a randomly assigned control, snow-removalplot, and snow-removal+ soilheatingplot. Plots were surroundedby elevatedboardwalksto
minimize disturbanceduring the frequentmeasurements.
2.4.
Measurements
Soil temperaturewas monitoredin all plots at 5 cm below
the moss surface with copper-constantan thermocouples.
Thermocouples were measured at 1 min intervals with a
Campbell 21X micrologger and solid-state multiplexer
(Campbell Scientific, Inc., Logan, Utah) and the data storedas
60 min averages. Depth of thaw was measuredweekly or more
frequently at marked intertussocklocations within the plots
using a steel probe. The depth to water table was measured
weekly for each plot in wells made from perforated2.5 cm diameterPVC pipe [Oberbaueret al., 1992].
Bud and leaf status was followed
In manipulatedplots, the growing period was extendedat
both endsof the growing seasonin 1995 and 1996 by removing snow cover and maintainingplots snow free in early season [Galen and Stanton,1993], and preventingsnowaccumulation in late season. Snowmeltusually occursin late May,
but snowmeltbetweenmid-May and early Junehas occurredin
the past 20 years. Snowfall and hard freezes(below -5øC) usually occur after mid-August,and plantsmay be snow covered
frequentlybeginningin late Augustor early September.
Snow depths are relatively shallow at the site (30-70 cm).
As a result, most of the spring snow cover could be removed
manually to near the top of the canopy. The remainderwas
passivelymelted within a few days using open top chambers
[Molau, 1996]. Greatcarewas takento avoiddamageto plants
during snow removal. Following initial snow removal, extended-seasonplots were coveredwith open-endedpolyethylene A-frame tents in the event of advancingstorm fronts to
prevent additional snow depositionuntil the control plots became
icant fraction of the daily groundheat flux for tundra [Weller
and Holmgren, 1974]. Soil heatingexperimentsof this kind
have been successfullycarriedout in borealforest [Van Cleve
Tents
were
installed
the second week
of
August until the end of the extendedseasonto prevent snow
cover and minimize frost damageon treatmentplots. Thus
treatment effects found in late seasonmay be responsesto
early-season manipulations (snow removal), to late-season
manipulations(frost and snow prevention),or both.
Treatmentplots were maintainedsnowfree from May 4 to
September5. Becauseseasonlength extensionfrom climate
warming will occur simultaneouswith increasesin air and soil
temperature,the major effects of which will be belowground
[Shaver and Kurnrnerow, 1992], we included an elevated soil
temperature treatment along with the extended season. Soil
temperatureswere increasedapproximately2øC on a set of the
snow removal plots using thermosratedgreenhouseheating
cablesburied10 cm deepin the soil andpoweredby a portable
generator. Heating cableswere installed1 year prior to initia-
for six shoots of each of
eight dominantspecieson each plot weekly or more frequently
throughout the study period. Leaf status was recorded as
greennesson a qualitative scalefrom 1 to 10. Foliage area index was monitoredmonthlyor more often on the plotsusinga
LAI-2000 canopyanalyzer(LI-COR, Inc., Lincoln,Nebraska).
Five observationswere taken within each plot using marked
locations and sensor alignments.
Table 1. Timing of Snowmeltand Soil Thaw DegreeDays
for Extended Season,ExtendedSeason+ Heating, and
Control
Plots in 1995 and 1996
Control
Extended
Extended +
Heat
1995
Date of first soil thaw
May 11
May 3
May 3*
0
8
17
12
41
57
477
630
725
May 26
May 4
May 4*
0
19
35
Thaw degreedaysJune1
9
34
55
Total seasonthaw degree
465
Thaw degreedays at
snowmelt
Thaw degreedaysJune1
Total seasonthaw degree
days
1996
Date of first soil thaw
Thaw degreedays at
snowmelt
days
557
,
*Heatingbeganon May 10 in 1995 andon May 4 in 1996
646
OBERBAUER
ETAL.:EXTENDED
GROWING
SEASON
ANDSOILWARMING
29,077
15
ß
1995
Control
& Extended
10-
5-
•
•
[]
Extended
+ Heat
_
a•' 1• aee
•
0
-5
125
•
i
•
•
150
175
200
225
250
15
1996
10-
-5
125
5/5
I
150
5/30
I
175
6/24
I
200
7/19
I
225
8/13
250
9/7
Day of year
Figure 1. Seasonal
patterns
of soiltemperature
at 5 cm depthfor extended
season,
extended
season
+ heat,
andcontrolplotsfor 1995and1996. Valuesaremeansof sixplotspertreatment.
Diurnfl coursesof net ecosystem
carbondioxidefluxeswere surementswere made over the courseof 24 hours(4 hour interassessed
at weekly intervalson the studyplots. We useda vals). Measurements
were takenby block to distributeany
modificationof techniques
described
by Bartlettet al. [1989], weather changesduring a measurementperiod among the
Whitinget al. [1991]; Oechelet al. [1993], and Tenhunenet treatments. In 1996, an additional set of measurements was
al. [1995]. In this procedure,
a rapid,transientmeasurement madeunderartificially darkenedconditionsin early morning
of ecosystem
exchange
is madeusinga portablecuvette(0.7 x (0400 hours)to estimateecosystem
dark respiration.
0.7 x 0.5 m) attached
to a polypropylene
chamberbaseperMethaneeffiuxesweremeasured
by gaschromatography
and
manentlyinstalledin the soil. CO2 concentration
changes staticchambertwice in 1995 and threetimesin 1996 during
weremeasured
with a Li-6200 portablephotosynthesis
system middle to late season. Previous studies have revealed that
(LI-COR, Inc, Lincoln, Nebraska). For each measurement, methanefluxesfrom tundraare typicallylow until mid-growthree 30 s incubationperiodswere followed. If the valuesof ing season[Whalenand Reeburgh,1988]. Gas sampleswere
the sequential
readings
differedsubstantially,
the process
was takenfromthechamber
usedforCO2fluxeswith10cm3 glass
repeateduntil a steadyvalue was obtained. Six setsof mea- syringesat 15 min intervalsover a 45 min period. The aver-
29•078
OBERBAUF•ETAL.:EXTENDED
GROWING
SEASON
ANDSOILWARMING
age rate of changein methaneconcentrationover the 45 min
period was used to calculatemethanefluxes [Whalen and
Reeburgh, 1988].
greatestfor the heatingplots,but differenceswere not signifi.cant. In 1996 the differencesamongtreatmentswere considerably greater. Foliage area of the extendedseasonplots increaseddramaticallyover the extended+ soil heatingalthough
2.5.
the reasons for the difference
Data
Analysis
To calculatesoil thaw degreedays,soil temperaturesat 5 cm
depth greater than 0øC were multiplied by days. For CO2
fluxes the mean of the last two 30 s readingswas used as the
plot flux for a samplingperiod. Data were testedfor normality
prior to testing with parametric tests. Respirationdata were
square-roottransformedprior to testingwith repeatedmeasures
analysis of variance. In the casesof soil temperature,thaw
depth, and depth to water table, the data deviated somewhat
from normality and could not be successfullytransformedto
normality. Nevertheless,because analysis of variance is robust to small deviations from normality, we performed repeated measuresanalysisof variance to compareseason-long
treatment effects for these variables. Spearmanrank correlations were used to test correlationsbetween dark respiration
and environmental
variables.
3. Results
3.1.
Site
Environments
In 1995, early-seasontemperatureswere unusually warm,
and completesnowmelton the controlplotsoccurredwithin 910 days of initiation of the treatments. In contrast,in 1996
snowmeltdid not occur until late May, providing24 days of
early-seasontreatment(Table 1). However, low temperatures
during May 1996 resultedin similar total thaw degreedays for
both years by the end of May. Overall, 1995 was warmerthan
1996, especially in late season.
Soil temperature, depth of thaw, plant phenology, and
canopy development were all affected by the treatmentsin
both years. Soil temperatureson treatmentplots were higher,
particularly early in the season(Figure 1). Soil temperatures
on the heated plots remained above the controls throughout
the growing season, resulting in substantially higher soil
thaw degree days for the season(Table 1). Differenceswere
dampenedduring rainy periodsrelative to warmer dry periods.
The seasonalaverage increasein temperatureon the warmed
plots was 1.8øC in both years. Extendedseasonplots averaged about IøC warmer than controls, but the difference was
not statistically significant (Table 2).
Thaw depths also differed among treatments. The control
plots had significantly shallower thaw depths than extended
and heatedplots (Figure 2, Table 2). The extendedplots had
slightly greater thaw depths than heated plots in both years,
apparently because of slight microtopographic differences
among the plots. Mean depth to water table was slightly
greaterfor the extendedand heatedplots,thoughthe difference
was not significant (Table 2).
3.2.
Phenology
and
Canopy
Development
Treatment effects on plant phenologywere readily apparent. For example,the deciduousshrubs,Salix pulchra Cham.
and Betula nana L., broke bud and senescedleaves earlier both
years on the extended treatments (Figure 3). The two
graminoidspecies,E. vaginaturnand Carex bigellowii Tort.,
initiatedleaf growthearlierbut senesced
leavesaboutthe same
time or later on the extended season treatments (data not
shown). Extended seasontreatmentshad higher estimatedleaf
area indices (Figure 4). In 1995 the leaf area indexes were
are not clear.
3.3. CO2 Fluxes
In 1995, seasonalCO2 exchangewas similar among the
treatmentsearly in the seasonbut divergedas the seasonprogressed(Figure5). Soil warmingtendedto reduceuptakeearly
in the season, but it increased late in the season relative to
controls(Figure 5). Extendedseasonplotswere intermediate.
This patternof systemcarbonfluxesappeared
to be a resultof
the interactionbetweenincreasedsoil respirationearly in the
seasonin snow removal plots and retentionof photosynthetic
capacityon these plots late in the season. In 1996, differencesin net seasonalCO2 flux betweentreatmentswere again
relatively small and not statisticallydifferent (Table 3). In
both years,uptakewas evidentvery early in the seasonon the
extendedand extended+ heatedplots.
Dark respirationmeasurementsrevealed that the extended
seasonand heated plots generally had higher respirationrates
than the controlsalthoughonly the extendedplots were statistically differentfrom the controls(Figure 6, Table 3).
Table 2. Resultsfrom RepeatedMeasuresAnalysisof
Varianceon Daily Mean Soil Temperature,Depthof
Thaw, and Water Table for ExtendedSeason,
ExtendedSeason+ Heating,and Control
Plots in 1996
Source
Soil temperature
df
F
2
2.767
0.0949
Time
107
150.02
0.0001
Treatment x Time
214
3.477
Treatment
Mean, øC
Control
Extended
P
0.0001
Extended +
Heat
3.6a
4.6a
5.4b
2
15.4
0.0002
Time
43
304.1
0.0001
Treatment x Time
86
1.58
0.0013
Control
Extended
Extended +
Depth of Thaw
Treatment
Mean, cm
Heat
15.9a
Water
23.2b
21.8b
Table
Treatment
2
1.4
0.2769
Time
14
52.13
0.0001
Treatment x Time
28
1.65
0.026
Control
Extended
Extended +
Mean, cm
Heat
11.7a
14.2a
14.1a
Meansfollowedby differentlettersare significantly
differentat P <
0.05 by Fisher'sLSD test.
OBERBAUER
ETAL.:EXTENDEDGROWINGSEASON
ANDSOILWARMING
29,079
1995
-10
-20
-40
-50
_-
Control
&
Extended
[]
Extended + Heat
-60
I
125
I
I
150
175
I
I
200
225
250
0
-10 -
-20 -
-30
-40.
-50 -
-60
,
,
125
150
175
5/5
5/30
6/24
•
•
200
225
250
7/19
8/13
9/7
Day of year
Figure 2. Seasonalpatternsof depth of thaw for extendedseason,extendedseason+ heat, and control plots
for 1995 and 1996. Values are meansof six plots per treatment. Standarderrorsare shownfor controlsto represent magnitude of variation for treatments.
-Control
•
Extended Fi, Extended
+ Heat
100
100
Salixpulchra
Betula nana
?•
75
o
75
x•
50
25
0
0
125
5/5
150
5/30
175
6/24
200
7/19
Dayofyear
225
8/13
250
9/7
125
5/5
150
5/30
175
6/24
200
7/19
225
8/13
250
Dayofyear
Figure3. Timingof budbreak(left sideof eachpanel)andleafsenescence
(fightsideof eachpanel)for two
deciduous
shrubs,SalixpulchraandBetulanana,in 1995on extendedseason,
extendedseason
+ heat,andcontrolplots. Valuesarebasedonmeanof sixplantsmeasured
oneachof sixplots.
9/7
29,080
OBERBAUERET AL.: EXTENDED GROWING SEASONAND SOIL WARMING
4. Discussion
_
1995
Control
Extended
0.8 -
Extended + Heat
T
0.4-
0.2
_
125
1;0
1•5
2;0
I
225
250
Day of year
996
Snow removal and soil warming significantly affected soil
environment, plant phenology, and canopy characteristics.
Changesin soil environmentincluded increasedsoil temperatures and depth of thaw as well as nonsignificantincreasesin
depth to water table. These factors are among the dominant
controls over carbon effiuxes from tundra systems [Peterson
and Billings, 1975; Peterson et al., 1984; Oberbauer et al.,
1991, 1992, 1996, Oechel et al., 1993]. Our finding of higher
dark respirationon the treatmentplots was consistentwith the
original hypothesis that an extended seasonwould increase
carbon lossesthrough changesin the soil environment.
However, these higher dark respirationrates were not apparent from the net fluxes. Instead, net fluxes were similar
among treatments,and lossesto the atmospheretended to be
slightly lower than those of the control plots. Similar net
fluxes with higher respiratory losses suggestthat the gross
uptakecapacityof the treatmentplots was increased,a finding
consistent
with
the observed
increases
in leaf
area.
Net
ecosystemflux representsa balancebetweenlarge respiratory
lossesand uptake capacity. The effect of the early snow removal and increasedsoil temperatureis apparentlyto increase
both of these factors similarly.
0.8-
0.6-
Control
0.4-
1995
Extended
---I:i---
0'
I
Extended+ Heat
'r
--
5/5
[o
5/30
6/24
7/19
8/13
9/7
Day of year
Figure 4. Leaf area index as measuredby Li-Cor LAI-2000
for extendedseason,extendedseason+ heat, and controlplots
-2
125
for 1995 and 1996. Values (+ standard error) are based on
meansof five measurements
on eachof six plots.
1•0
1'•5 2(•0 2•5
250
Day of year
1996
3.4.
Methane
Measurements
Fluxes
of methane
flux
in middle
and late season
1995 and 1996 suggestthat methane emissionsare enhanced
on the treatmentplots (Table 4). Treatmentcomparisonswere
not significantly different becauseof high variance resulting
from large variation in the amount of E. vaginaturnon the
sample sites. Eriophorum vaginaturnis the major conduit of
methanein thesesoils[Whalenand Reeburgh,1988].
3.5.
Controls
on
Fluxes
-2
Carbondioxideuptakeratesare nonlinearfunctionsof light
125
150
1'15
200
225
250
and temperaturethat change with seasonalchangesin plant
5/5
5/30
6/24
7/19
8/13
9/7
canopy function. Change in uptake capacity in responseto
Day of year
light is easily seen in the light responseof net flux through
the season(Figure 7). In contrast,respirationwas strongly
Figure 5. Seasonalpattern of mean net CO2 exchangefor
correlatedwith soil temperature(rs= 0.89, P < 0.0001) and extendedseason,extendedseason+ heat, and control plots for
significantlycorrelatedwith depthto water table (rs = 0.45, P 1995 and 1996. Values are means of five or six plots.
< 0.0001). Respirationwas not significantlycorrelatedwith Standarderrors are shownfor the last sampledate to represent
depth of thaw (rs= 0.10, P = 0.09).
magnitudeof variation for treatments.
OBERBAUERET AL.' EXTENDEDGROWINGSEASONAND SOILWARMING
29,081
Table 3. Resultsof RepeatedMeasuresAnalysis of Variance
for SeasonalCoursesof Net CO2Flux and Dark Respiration
for ExtendedSeason,ExtendedSeason+ Heating, and
Table 4. Methane Fluxes for Extended Season, Extended
Control
Date
Plots in 1996
Season+ Heating, and Control Plots for 1995 and 1996
Control
Extended
Extended
+
P
Heat
Source
df
F
P
Net Flux 1996
Treatment
Time
Treatment
x Time
2
0.21
0.8128
14
27.03
0.0001
0.8
0.7522
28
MeanFlux,g m-2d-1
Control
Extended Extended
+
Heat
0.37a
0.29a
0.2%
1995
July 21
1.6 + 0.4
6.0 + 2.0
13.8 + 6.9
0.1741
Aug. 6
9.8 + 3.2
21.2 + 0.6
16.2 + 8.6
0.4160
1996
July12
3.6 + 1.6
12.2+ 4.4
6.5 + 1.8
0.1661
Aug. 3
1.4+ 1.4
3.2 + 2.1
6.4 +3.2
0.3594
Aug. 16
4.6 + 3.7
10.6+ 4.4
14.2+ 5.6
0.0942
Dark Respiration1996
Treatment
Time
Treatment x Time
2
a oa
0.0486
14
23.63
0.0001
0.82
0.7301
28
MeanFlux,gmolm-2s-1
Control
Heat
2.7b
2.3a
Means followed by differentlettersare significantlydifferentat P <
0.05 by Fisher'sLSD test.
Recent studiesof net CO2 exchangeof tussocktundra indicate that this vegetationmay currentlybe a sourceof carbonto
the atmosphere [Occhel et al., 1993]. The results t¾omour
control plots are consistentwith this idea, althoughthe losses
are not very large. Summed over the seasonthe net lossesof
carbon as CO2 to the atmosphereduring the bulk of the snow.free season(May 29 to Septen•ber2, 1996) would be 31, 25,
and26 g m-2 for control,
extended
season,
andextended
seasm,
+ heating plots. respectively. The results from our treatment
plots suggestso far that systemCO2 balance will not change
substantiallywith decreasedduration of snow cover.
The patternsof systemflux show dramatic fluctuationsover
_•
arein unitsof (mgm-2d-•)
ExtendedExtended
+
2.0a
Control
Values are means+ one standarderror of three to six plots in 1995
and six plots in 1996. P values indicate results from Kruskal-Wallis
nonparametricanalysisof variance for treatmentcomparisons. Values
Extended •
Extended + Heat
1
2..
the season. These differencesare largely a consequence
of
changesin respiration,as can be seenfrom the 1996 data.
These patterns track fairly well the seasonal pattern of soil
temperature, a finding consistent with previous work in tussock tundra [Oberbauer et al., 1991]. The strong seasonal
temperature responseis consistentwith the finding of higher
respiration rates on the treatment plots.
In 1995, differences in net flux among treatments were
largest at the end of the season,when extendedand extended+
heating treatments had lower carbon losses than the control
plots. We interpret these results to be a prolonging of the
photosynthetic capacity of the graminoids, particularly
Eriophorum vaginaturn. Eriophorum vaginatum tussockshave
relatively high photosynthetic capacity and strong influence
on system fluxes [Tenhunen et al., 1995]. The end-of-season
decline in photosyntheticcapacity of E. vaginatum is thought
to be mediated by low temperatures[Defoliart et al., 1988].
By minimizing frosts and snow cover on the treatmentplots at
the end of the season,we likely reducedthe end of seasondecline in photosyntheticcapacity. A similar pattern was also
seen in early August 1996, but near the end of August, several
very hard freezesaccompaniedsnow storms,againstwhich the
A-frame tents were ineffective protection. Consequently,we
did not see a clear late-seasonpatternof more negative fluxes
for the treatmentplots in that year.
The two years also differed, in that the leaf area index increasedon treatmentplots in 1996 over 1995. The results are
suggestiveof a directional shift in plant community structure
..... ,•.... •,• •,
•n alpine warming experiments[Harte and
Shaw, 1995]. The resultspresentedhere representshort-term
responsesto the treatments. Ecosystem changes that may
take place over the longerterm, suchas shiftsin canopystructure, are likely to result in changesin the carbon balance of
these treatments
0
I
I
125
150
175
200
I
5/5
5/30
6/24
7/19
I
225
2:50
8/13
9/7
Day of year
Figure 6. Seasonalpatternof dark respirationmeasuredat
0400 hours for extendedseason,extendedseason+ heat, and
in the future.
Acknowledgments.
This study was supportedby the National
ScienceFoundationOffice of Polar Programsgrant OPP-9321626 as
part of the International Tundra Experiment (ITEX), an international
effort to assessthe responseof arctic and alpine plants and ecosystems
to climate change. Logistical supportby the staff of the Institute of
Arctic Biology, University of Alaska, Fairbanks, is gratefully
appreciated. Carlo Calandrielloand Brooke Shamblinmade invaluable
contributions
to collections
of field data.
Methane
measurements
were
controlplotsfor 1996.Valuesare meansof five or six plots. •nade possible by collaboration with Bill Reeburgh, University of
Standard
errorsare shownfor the lastsampledateto represent California-Irvine. Two anonymous reviewers made very helpful
magnitude of variation for treatments.
cmnmentson the •nanuscript.
29May
3 July
løss
54
I
ß
.••
ß
ß
ß
.
ß
.d•e
......... ...................
;-. ...........
Iee.....
© ß el....
o•_
,ell.
oo
ß
_
I
0
2 September
:
•• 0
upt•
7 August
500
ß
ß
.
I
1000
I
1500
i
0
500
I
I
1000
1500
!
0
•
,
500
1000
I
1500
I
0
5{X)
I
I(XX)
I
15(X)
2(XX)
Photosynthetic
0hoton
flux
density
(•tmol
n{2s-1)
Figure 7. Light-response
scattergrams
of net CO:zexchange
tbr selecteddatesof all plotscombinedin 1996.
ripariantundrain the northernfoothillsof theBrooksRange,Alaska,
USA, Oecologia,92, 568-577, 1992.
Oberbauer,S. F., J. D. Tenhunen,R. W. Virginia, W. Cheng,and C. T.
Bartlett, D. S., G. J. Whiting, and J. M. Hartman,Use of vegetation
Gillespie, Landscapepatternsof carbon gas exchangein tundra
indexesto estimatesolar radiationand net carbonexchangeof a
ecosystems,in Landscape Function and Disturbance in Arctic
References
grasscanopy, Remote Sens. Environ., 30, 115-128, 1989.
Billings, W. D., Carbonbalanceof Alaskatundraandtaigaecosystems:
Past,present,and future, Q. Sci. Rev., 6, 165-177, 1987.
Chapin,F. S., III, and G. R. Shaver,Individualisticgrowthresponseof
tundra plant speciesto environmentalmanipulationsin the field,
Ecology,66, 564-576, 1985.
Chapin, F. S., III, R. L. Jeffries,J. F. Reynolds,G. R. Shaver,and J.
Svoboda,Arctic plant physiologicalecology:A challengefor the
future,in Arctic Ecosystems
in a ChangingClimate,editedby F. S.
ChapinIII et al., pp. !-8, Academic,SanDiego,Calif.,1992.
Chapin, F. S., III, G. R. Shaver,A. E. Giblin, K. G. Nadelhoffer,and J.
A. Laundre, Responseof arctic tundra to experimentaland observed
changesin climate,Ecology,76, 694-711, 1995.
Tundra, Ecol. Stud., vol. 120, edited by J. F. Reynoldsand J. D.
Tenhunen,pp. 223-257, Springer-Verlag,New York, 1996.
Oechel, W. C., and W. D. Billings, Effects of global changeon the
carbon balance of arctic plants and ecosystems, in Arctic
Ecosystems
in a ChangingClimate,editedby F. S. Chapinet al., pp.
139-168, Academic,San Diego, Calif.,1992.
Oechel, W. C., and G. L. Vourlitis, The effects of climate changeon
arctictundraecosystems,TREE, 9, 324-329, 1994.
Oechel, W. C., S. J. Hastings,M. Jenkins,G. H. Riechers,N. Grulke,
and G. Vourlitis, Recentchangeof arctic tundraecosystems
from a
net carbon dioxide sink to a source,Nature, 361,520-523, 1993.
Peterson,K. M., and W. D. Billings, Carbon dioxide flux from tundra
soilsand vegetationas relatedto temperatureat BarrowAlaska,Am.
Midl. Nat., 94, 88-98, 1975.
Defoliart,L. S., M. Griffith, F. S. ChapinIII, andS. Jonassons,
Seasonal
Peterson,K. M., W. D. Billings, and D. N. Reynolds,Influenceof water
patterns of photosynthesisand nutrient storagein Eriophorum
table and atmosphericCO2 concentration
on the carbonbalanceof
vaginaturnL., an arcticsedge,Funct.Ecol.,2, 185-194,1988.
arctic tundra,Arct. Alp. Res., 16, 331-335, 1984.
Galen, C., and M. L. Stanton,Short-termresponses
of alpinebuttercups
Soil
to experimentalmanipulationsof growing seasonlength,Ecology, Post,W. M., W. R. Emanuel,P. J. Zinke, and A. G. Stangenberger,
carbonpoolsandworld life zones,Nature,298, 156-159, 1982.
74, 1052-1058, 1993.
in Alaskanaquatic
Gorham,E., Northernpeatlands:Role in the carboncycleand probable Schell,D. M., Carbon-13and carbon-14abundances
organisms:Delayed productionfrom peat in arctic food chains,
responses
to climaticwaxming,Ecol. Appl., 1, 182-195,1991.
Science, 219, 1068, 1983.
Grulke, N. E., G. H. Riechers,W. C. Oechel, U. Hjelm, and C. Jaeger,
Shaver,G. R., and J. Kummerow, Phenology,resourceallocation,and
Carbon balance in tussock tundra under ambient and elevated
growthof arctic vascularplants,in Arctic Ecosystems
in a Changing
atmospheric
CO2, Oecologia,83, 485-494, 1990.
Clintate,editedby F. S. ChapinIII et al., pp. 193-212,Academic,San
Harte, J., and R. Shaw, Shiftingdominancewithin a montanevegetation
Diego, Calif., 1992.
community:Resultsof a climate-warmingexperiment,Science,267,
876-880, 1995.
Tenhunen, J. D., C. T. Gillespie, S. F. Oberbauer,A. Sala, and S.
Whalen, Climate effects on the carbon balance of tussock tundra in
Hastings, S. J., S. A. Luchessa,W. C. Oechel, and J. D. Tenhunen,
the Philip Smith Mountains,Alaska,Flora, 190, 273-283, 1995.
Standingbiomassand productionin water drainagesof the foothills
Van Cleve, K., W. C. Oechel,and J. L. Hom, Responseof blackspruce
of the Philip Smith Mountains,Alaska,Holarctic Ecol., 12, 304-311,
1989.
(Picea mariana) ecosystemsto soil temperaturemodificationsin
interior Alaska, Can. J. For. Res., 20, 1530-1535, 1990.
Kant, D. L., L. D. Hinznian, and J.P. Zerling, Thermalresponseof the
active layer to climatic warming in a permafrostenvironment,Cold Wein, R. W., and L. C. Bliss, Primaryproductionin arcticcottongrass
tussocktundracommunities,
Arct.Alp Res.,6, 261-274, 1974.
Reg. Sci. Technol.,19, 111-112, 1991.
Kikuzawa,K., Leaf phenologyas an optimalstrategyfor carbongainin Weller, G., and B. Holmgren,The microclimatesof the arctictundra,J.
Appl. Meteorol.,13, 854-862, 1974.
plants,Can. J. Bot., 73, 158-163, 1995.
Lechowicz, M. J., and T. Koike, Phenologyand seasonalityof woody Whalen, S.C.; and W. S. Reeburgh,A methaneflux time seriesfor
tundraenvironments,Global Biogeochem.Cycles,2, 399-409, 1988.
plants:An unappreciatedelement in global changeresearch?,Can.
J. Bot., 73, 147-148, 1995.
Whiting, G. J., J.P. Chanton, D. S. Bartlett, and J. D. Happell,
Relationships
betweenCH4 emission,biomass,
andCO2 exchange
in
Manabe,S., and R. J. Stouffer,A CO2 climatesensitivitystudywith a
a subtropicalgrassland,J. Geophys.Res.,96, 3067-3071, 1991.
mathematicalmodelof the globalclimate,Nature,28, 491-493, 1979.
Maxwell, B., Arctic climate:Potentialfor changeunderglobalwarming,
in Arctic Ecosystems
in a ChangingClimate,editedby F. S. Chapin
III et al., pp. 11-34, Academic,SanDiego, Calif.,1992.
S. F. Oberbauer,E. W. Pop, and G. Starr,Departmentof Biological
Miller, P. C., Carbonbalancein northernecosystems
and the potential Sciences,Florida InternationalUniversity,UniversityPark, Miami, FL
effect of carbondioxide-induced
climaticchange,CONF-8003118, 33199. (e-mail: oberbaue@fiu.edu)
Natl. Tech. Inf. Serv.,Springfield,Va, 1981.
Molau, U., International Tundra Experiment:ITEX Manual, 2nd ed., (ReceivedAugust 11, 1997;revisedJanuary19, 1998;
DanishPolar Center,Copenhagen,1996.
acceptedFebruary 11, 1998.)
Oberbauer,S. F., J. D. Tenhunen,and J. F. Reynolds,Environmental
effectson CO: efflux from watertrack andtussocktundrain Arctic
Alaska,USA, Arct. Alp. Res.,23, 162-169, 1991.
Oberbauer,S. F., C. T. Gillespie,W. Cheng,R. Gebauer,A. SalaSerra,
and J. D. Tenhunen,Environmentaleffectson CO2 efflux from
27, 1998
Effects of extended growing season and soil warming on
carbon dioxide and methane exchange of tussock tundra
in Alaska
StevenF. Oberbauer,Gregory Starr, and Eric W. Pop
Departmentof BiologicalSciences,FloridaInternationalUniversity,Miami
Abstract. The active seasonof tussocktundrawas extendedduring two growing seasons(1995
and 1996) by snowremovalin early seasonandpreventionof snowaccumulationin late seasonto
testthe effectsof a longergrowingseasonon tundracarbonexchange.Three treatmentswere
established:extendedseason,extendedseason+ soil warming,and controls. Soil warming was
accomplished
usingcold-frame,resistance
heatingwire installedthe yearprior to the initiationof
treatments.Diurnalcoursesof CO2 exchangeweremeasuredweeklyusinginfraredgasanalysis
with enclosedchambertechniques.Methanefluxeswere measuredtwo to threetimesa seasonalso
usingenclosuremethods.In 1995, snowmeltoccurredunusuallyearly, and snowremoval
treatmentsincreasedthe seasononly 9-10 days. In 1996 the early seasonwas increased
approximately24 days. As expected,thawdepth,soil temperature,
andplantgrowthwere greater
earlierin the extendedseasonandextendedseason+ soil heatingplots. Methanefluxes in both
seasons
were low but tendedto be higherin the extendedseasonandsoil heatedplots. Net
ecosystemCO2 fluxeswere similaramongtreatmentsearly in the season,with a tendencytoward
morepositivefluxes(systemloss)for the snowremovaland warmedplots,possiblydue to higher
belowgroundrespiration.Duringmidseason,fluxeswere similaramongthe treatments.Later in
the season,fluxesof extendedseasonandwarmedplotstendedto be lower (lesscarbonloss)than
controls,especiallyin 1995. Totaledover the season,however,the fluxesof the threetreatments
did not statisticallydiffer andrepresented
lossesto the atmosphere.Measurements
of dark
respirationin 1996 indicatethatbothrespirationanduptakewere increasedon the extendedseason
plots, resultingin similar net fluxes to controls.
1. Introduction
Evidencefor climate warming in the Alaskan Arctic is growing [Chapin et al., 1995; Oechel et al., 1993]. Among the
most importantconsequences
of warmingin the Arctic may be
changesin seasonlength resultingfrom shifts in the duration
of snow cover. The duration of snow cover is expected to
change, possibly dramatically, in responseto climate warming, althoughthe direction of changeis uncertain[Maxwell,
1992]. With no change in snowfall, snowmelt will probably
occur earlier.
However, if snowfall increases, the duration of
snowcover may remain the sameor even increase. Becauserelease from snow cover initiates soil thawing and triggers plant
growth, the timing of snowmelt directly affects the season
length. Changesin the start of the growing seasonwill be accompaniedby an extensionof the period of thawed soil at the
end of the seasondue to direct effects of warming [Kane et al.,
1991]. Maxwell [1992] suggeststhat snow-free periods in
the Low Arctic may increase 1 month or more in the next century.
Changes in season length may profoundly affect plant,
community, and ecosystemprocessesin the Arctic, both directly and indirectly. Seasonlength changesand warming di-
Copyright1998 by the AmericanGeophysicalUnion
Papernumber98JD00522.
0148-0227/98/98
JD-00522509.00
rectly affect plant phenologicalpatterns,that is, the timing of
plant developmentalevents. Plant activity is largely initiated
by snowmelt in early season,and dormancyat the end of the
seasonis triggeredby temperature,photoperiod,and/or internal cycles [Shaver and Kummerow, 1992]. Phenologydirectly
affects the overall pattern of plant and therefore ecosystem
carbon uptake [Kikuzawa, 1995]. Nevertheless,relatively little experimental effort has been devoted to the effects of potential phenological changes on ecosystem level processes
[Lechowicz and Koike, 1995]. The objective of this study was
to examine ecosystemcarbon exchangein responseto a growing season experimentally extended by snow removal. We
hypothesized that as a result of increased soil temperatures,
belowground respiration and therefore fluxes to the atmosphere would increase in response to an increased season
length.
Changes in seasonlength are of particular importance to
arctic ecosystemsbecause they are likely very sensitive to
climate changes. Seasonlengthsare short,effects of physical
factors are strong, and environmentalgradientsare very steep
in arctic ecosystems. Furthermore,the climate warming is expected to be greatesttoward the poles [Manabe and Stouffer,
1979; Maxwell, 1992]. Finally, high latitudeshold a significant fraction of terrestrialcarbonin an unstablestate,trapped
in cold soils, providing the potential for greater warming if
the carbon is mineralized [Billings, 1987; Chapin et al.,
1992; Gorham, 1991; Miller,
1981; Oechel et al.,
Oechel and Vourlitis, 1994; Post et al., 1982].
29,075
1993;
29,076
OBERBAUER ET AL.: EXTENDED GR()WlNG SEASONAND SOIL WARMING
2. Methods
2.1.
Study
These
tion of treatments in 1994. Over a 2 hour period each day,
eachplotreceived
anadditional
0.4MJm-e,whichis a signif-
Site
studies
were
conducted
near
the
Toolik
Arctic
ResearchStation (780 m elevation,68ø38'N, 149ø34'W) operated by the Institute of Arctic Biology of the University of
Alaska, Fairbanks. Toolik is located along the Dalton
Highway in the northern foothills of the Philip Smith
Mountains.
2.2.
Vegetation
The experiments were performed in cottongrasstussock
tundra. This tundra communitycoversa large proportionof
the North Slope of Alaska, eastern Canada, and northern
Eurasia. Becauseof its high soil carbon storage,tussocktundra hasbeen identifiedas one of the two mostimportanttundra
vegetation types from the perspectiveof global carbon cycling [Oechel and Billings, 1992]. The dominantgrowth form
is Eriophorum vaginaturnL. tussockswith intertussockspaces
filled with dwarf deciduousandevergreenshrubsandpeat-forming mosses. Several lines of evidence suggestthat cottongrass tundra is currently a source for atmosphericcarbon
[Schell, 1983; Grulke et al., 1990; Oecheland Billings, 1992;
Oechel et al., 1993]. Biomass and structural characteristicsof
this vegetationhave been describedby Hastingset al. [1989]
and Wein and Bliss [1974].
2.3.
Treatments
snow free.
et al., 1990].
The study plots were 1.5 x 1.5 m within six replicate
blocks.
Plot numbers
were limited
because of the intensive
labor required for snow removal. Similar plot sizes have
proven to be sufficient for studiesof ecosystemprocessesin
the Arctic [Chapin and Shaver, 1985; Grulke et al., 1990;
Oechel et al., 1993]. Each block contained a randomly assigned control, snow-removalplot, and snow-removal+ soilheatingplot. Plots were surroundedby elevatedboardwalksto
minimize disturbanceduring the frequentmeasurements.
2.4.
Measurements
Soil temperaturewas monitoredin all plots at 5 cm below
the moss surface with copper-constantan thermocouples.
Thermocouples were measured at 1 min intervals with a
Campbell 21X micrologger and solid-state multiplexer
(Campbell Scientific, Inc., Logan, Utah) and the data storedas
60 min averages. Depth of thaw was measuredweekly or more
frequently at marked intertussocklocations within the plots
using a steel probe. The depth to water table was measured
weekly for each plot in wells made from perforated2.5 cm diameterPVC pipe [Oberbaueret al., 1992].
Bud and leaf status was followed
In manipulatedplots, the growing period was extendedat
both endsof the growing seasonin 1995 and 1996 by removing snow cover and maintainingplots snow free in early season [Galen and Stanton,1993], and preventingsnowaccumulation in late season. Snowmeltusually occursin late May,
but snowmeltbetweenmid-May and early Junehas occurredin
the past 20 years. Snowfall and hard freezes(below -5øC) usually occur after mid-August,and plantsmay be snow covered
frequentlybeginningin late Augustor early September.
Snow depths are relatively shallow at the site (30-70 cm).
As a result, most of the spring snow cover could be removed
manually to near the top of the canopy. The remainderwas
passivelymelted within a few days using open top chambers
[Molau, 1996]. Greatcarewas takento avoiddamageto plants
during snow removal. Following initial snow removal, extended-seasonplots were coveredwith open-endedpolyethylene A-frame tents in the event of advancingstorm fronts to
prevent additional snow depositionuntil the control plots became
icant fraction of the daily groundheat flux for tundra [Weller
and Holmgren, 1974]. Soil heatingexperimentsof this kind
have been successfullycarriedout in borealforest [Van Cleve
Tents
were
installed
the second week
of
August until the end of the extendedseasonto prevent snow
cover and minimize frost damageon treatmentplots. Thus
treatment effects found in late seasonmay be responsesto
early-season manipulations (snow removal), to late-season
manipulations(frost and snow prevention),or both.
Treatmentplots were maintainedsnowfree from May 4 to
September5. Becauseseasonlength extensionfrom climate
warming will occur simultaneouswith increasesin air and soil
temperature,the major effects of which will be belowground
[Shaver and Kurnrnerow, 1992], we included an elevated soil
temperature treatment along with the extended season. Soil
temperatureswere increasedapproximately2øC on a set of the
snow removal plots using thermosratedgreenhouseheating
cablesburied10 cm deepin the soil andpoweredby a portable
generator. Heating cableswere installed1 year prior to initia-
for six shoots of each of
eight dominantspecieson each plot weekly or more frequently
throughout the study period. Leaf status was recorded as
greennesson a qualitative scalefrom 1 to 10. Foliage area index was monitoredmonthlyor more often on the plotsusinga
LAI-2000 canopyanalyzer(LI-COR, Inc., Lincoln,Nebraska).
Five observationswere taken within each plot using marked
locations and sensor alignments.
Table 1. Timing of Snowmeltand Soil Thaw DegreeDays
for Extended Season,ExtendedSeason+ Heating, and
Control
Plots in 1995 and 1996
Control
Extended
Extended +
Heat
1995
Date of first soil thaw
May 11
May 3
May 3*
0
8
17
12
41
57
477
630
725
May 26
May 4
May 4*
0
19
35
Thaw degreedaysJune1
9
34
55
Total seasonthaw degree
465
Thaw degreedays at
snowmelt
Thaw degreedaysJune1
Total seasonthaw degree
days
1996
Date of first soil thaw
Thaw degreedays at
snowmelt
days
557
,
*Heatingbeganon May 10 in 1995 andon May 4 in 1996
646
OBERBAUER
ETAL.:EXTENDED
GROWING
SEASON
ANDSOILWARMING
29,077
15
ß
1995
Control
& Extended
10-
5-
•
•
[]
Extended
+ Heat
_
a•' 1• aee
•
0
-5
125
•
i
•
•
150
175
200
225
250
15
1996
10-
-5
125
5/5
I
150
5/30
I
175
6/24
I
200
7/19
I
225
8/13
250
9/7
Day of year
Figure 1. Seasonal
patterns
of soiltemperature
at 5 cm depthfor extended
season,
extended
season
+ heat,
andcontrolplotsfor 1995and1996. Valuesaremeansof sixplotspertreatment.
Diurnfl coursesof net ecosystem
carbondioxidefluxeswere surementswere made over the courseof 24 hours(4 hour interassessed
at weekly intervalson the studyplots. We useda vals). Measurements
were takenby block to distributeany
modificationof techniques
described
by Bartlettet al. [1989], weather changesduring a measurementperiod among the
Whitinget al. [1991]; Oechelet al. [1993], and Tenhunenet treatments. In 1996, an additional set of measurements was
al. [1995]. In this procedure,
a rapid,transientmeasurement madeunderartificially darkenedconditionsin early morning
of ecosystem
exchange
is madeusinga portablecuvette(0.7 x (0400 hours)to estimateecosystem
dark respiration.
0.7 x 0.5 m) attached
to a polypropylene
chamberbaseperMethaneeffiuxesweremeasured
by gaschromatography
and
manentlyinstalledin the soil. CO2 concentration
changes staticchambertwice in 1995 and threetimesin 1996 during
weremeasured
with a Li-6200 portablephotosynthesis
system middle to late season. Previous studies have revealed that
(LI-COR, Inc, Lincoln, Nebraska). For each measurement, methanefluxesfrom tundraare typicallylow until mid-growthree 30 s incubationperiodswere followed. If the valuesof ing season[Whalenand Reeburgh,1988]. Gas sampleswere
the sequential
readings
differedsubstantially,
the process
was takenfromthechamber
usedforCO2fluxeswith10cm3 glass
repeateduntil a steadyvalue was obtained. Six setsof mea- syringesat 15 min intervalsover a 45 min period. The aver-
29•078
OBERBAUF•ETAL.:EXTENDED
GROWING
SEASON
ANDSOILWARMING
age rate of changein methaneconcentrationover the 45 min
period was used to calculatemethanefluxes [Whalen and
Reeburgh, 1988].
greatestfor the heatingplots,but differenceswere not signifi.cant. In 1996 the differencesamongtreatmentswere considerably greater. Foliage area of the extendedseasonplots increaseddramaticallyover the extended+ soil heatingalthough
2.5.
the reasons for the difference
Data
Analysis
To calculatesoil thaw degreedays,soil temperaturesat 5 cm
depth greater than 0øC were multiplied by days. For CO2
fluxes the mean of the last two 30 s readingswas used as the
plot flux for a samplingperiod. Data were testedfor normality
prior to testing with parametric tests. Respirationdata were
square-roottransformedprior to testingwith repeatedmeasures
analysis of variance. In the casesof soil temperature,thaw
depth, and depth to water table, the data deviated somewhat
from normality and could not be successfullytransformedto
normality. Nevertheless,because analysis of variance is robust to small deviations from normality, we performed repeated measuresanalysisof variance to compareseason-long
treatment effects for these variables. Spearmanrank correlations were used to test correlationsbetween dark respiration
and environmental
variables.
3. Results
3.1.
Site
Environments
In 1995, early-seasontemperatureswere unusually warm,
and completesnowmelton the controlplotsoccurredwithin 910 days of initiation of the treatments. In contrast,in 1996
snowmeltdid not occur until late May, providing24 days of
early-seasontreatment(Table 1). However, low temperatures
during May 1996 resultedin similar total thaw degreedays for
both years by the end of May. Overall, 1995 was warmerthan
1996, especially in late season.
Soil temperature, depth of thaw, plant phenology, and
canopy development were all affected by the treatmentsin
both years. Soil temperatureson treatmentplots were higher,
particularly early in the season(Figure 1). Soil temperatures
on the heated plots remained above the controls throughout
the growing season, resulting in substantially higher soil
thaw degree days for the season(Table 1). Differenceswere
dampenedduring rainy periodsrelative to warmer dry periods.
The seasonalaverage increasein temperatureon the warmed
plots was 1.8øC in both years. Extendedseasonplots averaged about IøC warmer than controls, but the difference was
not statistically significant (Table 2).
Thaw depths also differed among treatments. The control
plots had significantly shallower thaw depths than extended
and heatedplots (Figure 2, Table 2). The extendedplots had
slightly greater thaw depths than heated plots in both years,
apparently because of slight microtopographic differences
among the plots. Mean depth to water table was slightly
greaterfor the extendedand heatedplots,thoughthe difference
was not significant (Table 2).
3.2.
Phenology
and
Canopy
Development
Treatment effects on plant phenologywere readily apparent. For example,the deciduousshrubs,Salix pulchra Cham.
and Betula nana L., broke bud and senescedleaves earlier both
years on the extended treatments (Figure 3). The two
graminoidspecies,E. vaginaturnand Carex bigellowii Tort.,
initiatedleaf growthearlierbut senesced
leavesaboutthe same
time or later on the extended season treatments (data not
shown). Extended seasontreatmentshad higher estimatedleaf
area indices (Figure 4). In 1995 the leaf area indexes were
are not clear.
3.3. CO2 Fluxes
In 1995, seasonalCO2 exchangewas similar among the
treatmentsearly in the seasonbut divergedas the seasonprogressed(Figure5). Soil warmingtendedto reduceuptakeearly
in the season, but it increased late in the season relative to
controls(Figure 5). Extendedseasonplotswere intermediate.
This patternof systemcarbonfluxesappeared
to be a resultof
the interactionbetweenincreasedsoil respirationearly in the
seasonin snow removal plots and retentionof photosynthetic
capacityon these plots late in the season. In 1996, differencesin net seasonalCO2 flux betweentreatmentswere again
relatively small and not statisticallydifferent (Table 3). In
both years,uptakewas evidentvery early in the seasonon the
extendedand extended+ heatedplots.
Dark respirationmeasurementsrevealed that the extended
seasonand heated plots generally had higher respirationrates
than the controlsalthoughonly the extendedplots were statistically differentfrom the controls(Figure 6, Table 3).
Table 2. Resultsfrom RepeatedMeasuresAnalysisof
Varianceon Daily Mean Soil Temperature,Depthof
Thaw, and Water Table for ExtendedSeason,
ExtendedSeason+ Heating,and Control
Plots in 1996
Source
Soil temperature
df
F
2
2.767
0.0949
Time
107
150.02
0.0001
Treatment x Time
214
3.477
Treatment
Mean, øC
Control
Extended
P
0.0001
Extended +
Heat
3.6a
4.6a
5.4b
2
15.4
0.0002
Time
43
304.1
0.0001
Treatment x Time
86
1.58
0.0013
Control
Extended
Extended +
Depth of Thaw
Treatment
Mean, cm
Heat
15.9a
Water
23.2b
21.8b
Table
Treatment
2
1.4
0.2769
Time
14
52.13
0.0001
Treatment x Time
28
1.65
0.026
Control
Extended
Extended +
Mean, cm
Heat
11.7a
14.2a
14.1a
Meansfollowedby differentlettersare significantly
differentat P <
0.05 by Fisher'sLSD test.
OBERBAUER
ETAL.:EXTENDEDGROWINGSEASON
ANDSOILWARMING
29,079
1995
-10
-20
-40
-50
_-
Control
&
Extended
[]
Extended + Heat
-60
I
125
I
I
150
175
I
I
200
225
250
0
-10 -
-20 -
-30
-40.
-50 -
-60
,
,
125
150
175
5/5
5/30
6/24
•
•
200
225
250
7/19
8/13
9/7
Day of year
Figure 2. Seasonalpatternsof depth of thaw for extendedseason,extendedseason+ heat, and control plots
for 1995 and 1996. Values are meansof six plots per treatment. Standarderrorsare shownfor controlsto represent magnitude of variation for treatments.
-Control
•
Extended Fi, Extended
+ Heat
100
100
Salixpulchra
Betula nana
?•
75
o
75
x•
50
25
0
0
125
5/5
150
5/30
175
6/24
200
7/19
Dayofyear
225
8/13
250
9/7
125
5/5
150
5/30
175
6/24
200
7/19
225
8/13
250
Dayofyear
Figure3. Timingof budbreak(left sideof eachpanel)andleafsenescence
(fightsideof eachpanel)for two
deciduous
shrubs,SalixpulchraandBetulanana,in 1995on extendedseason,
extendedseason
+ heat,andcontrolplots. Valuesarebasedonmeanof sixplantsmeasured
oneachof sixplots.
9/7
29,080
OBERBAUERET AL.: EXTENDED GROWING SEASONAND SOIL WARMING
4. Discussion
_
1995
Control
Extended
0.8 -
Extended + Heat
T
0.4-
0.2
_
125
1;0
1•5
2;0
I
225
250
Day of year
996
Snow removal and soil warming significantly affected soil
environment, plant phenology, and canopy characteristics.
Changesin soil environmentincluded increasedsoil temperatures and depth of thaw as well as nonsignificantincreasesin
depth to water table. These factors are among the dominant
controls over carbon effiuxes from tundra systems [Peterson
and Billings, 1975; Peterson et al., 1984; Oberbauer et al.,
1991, 1992, 1996, Oechel et al., 1993]. Our finding of higher
dark respirationon the treatmentplots was consistentwith the
original hypothesis that an extended seasonwould increase
carbon lossesthrough changesin the soil environment.
However, these higher dark respirationrates were not apparent from the net fluxes. Instead, net fluxes were similar
among treatments,and lossesto the atmospheretended to be
slightly lower than those of the control plots. Similar net
fluxes with higher respiratory losses suggestthat the gross
uptakecapacityof the treatmentplots was increased,a finding
consistent
with
the observed
increases
in leaf
area.
Net
ecosystemflux representsa balancebetweenlarge respiratory
lossesand uptake capacity. The effect of the early snow removal and increasedsoil temperatureis apparentlyto increase
both of these factors similarly.
0.8-
0.6-
Control
0.4-
1995
Extended
---I:i---
0'
I
Extended+ Heat
'r
--
5/5
[o
5/30
6/24
7/19
8/13
9/7
Day of year
Figure 4. Leaf area index as measuredby Li-Cor LAI-2000
for extendedseason,extendedseason+ heat, and controlplots
-2
125
for 1995 and 1996. Values (+ standard error) are based on
meansof five measurements
on eachof six plots.
1•0
1'•5 2(•0 2•5
250
Day of year
1996
3.4.
Methane
Measurements
Fluxes
of methane
flux
in middle
and late season
1995 and 1996 suggestthat methane emissionsare enhanced
on the treatmentplots (Table 4). Treatmentcomparisonswere
not significantly different becauseof high variance resulting
from large variation in the amount of E. vaginaturnon the
sample sites. Eriophorum vaginaturnis the major conduit of
methanein thesesoils[Whalenand Reeburgh,1988].
3.5.
Controls
on
Fluxes
-2
Carbondioxideuptakeratesare nonlinearfunctionsof light
125
150
1'15
200
225
250
and temperaturethat change with seasonalchangesin plant
5/5
5/30
6/24
7/19
8/13
9/7
canopy function. Change in uptake capacity in responseto
Day of year
light is easily seen in the light responseof net flux through
the season(Figure 7). In contrast,respirationwas strongly
Figure 5. Seasonalpattern of mean net CO2 exchangefor
correlatedwith soil temperature(rs= 0.89, P < 0.0001) and extendedseason,extendedseason+ heat, and control plots for
significantlycorrelatedwith depthto water table (rs = 0.45, P 1995 and 1996. Values are means of five or six plots.
< 0.0001). Respirationwas not significantlycorrelatedwith Standarderrors are shownfor the last sampledate to represent
depth of thaw (rs= 0.10, P = 0.09).
magnitudeof variation for treatments.
OBERBAUERET AL.' EXTENDEDGROWINGSEASONAND SOILWARMING
29,081
Table 3. Resultsof RepeatedMeasuresAnalysis of Variance
for SeasonalCoursesof Net CO2Flux and Dark Respiration
for ExtendedSeason,ExtendedSeason+ Heating, and
Table 4. Methane Fluxes for Extended Season, Extended
Control
Date
Plots in 1996
Season+ Heating, and Control Plots for 1995 and 1996
Control
Extended
Extended
+
P
Heat
Source
df
F
P
Net Flux 1996
Treatment
Time
Treatment
x Time
2
0.21
0.8128
14
27.03
0.0001
0.8
0.7522
28
MeanFlux,g m-2d-1
Control
Extended Extended
+
Heat
0.37a
0.29a
0.2%
1995
July 21
1.6 + 0.4
6.0 + 2.0
13.8 + 6.9
0.1741
Aug. 6
9.8 + 3.2
21.2 + 0.6
16.2 + 8.6
0.4160
1996
July12
3.6 + 1.6
12.2+ 4.4
6.5 + 1.8
0.1661
Aug. 3
1.4+ 1.4
3.2 + 2.1
6.4 +3.2
0.3594
Aug. 16
4.6 + 3.7
10.6+ 4.4
14.2+ 5.6
0.0942
Dark Respiration1996
Treatment
Time
Treatment x Time
2
a oa
0.0486
14
23.63
0.0001
0.82
0.7301
28
MeanFlux,gmolm-2s-1
Control
Heat
2.7b
2.3a
Means followed by differentlettersare significantlydifferentat P <
0.05 by Fisher'sLSD test.
Recent studiesof net CO2 exchangeof tussocktundra indicate that this vegetationmay currentlybe a sourceof carbonto
the atmosphere [Occhel et al., 1993]. The results t¾omour
control plots are consistentwith this idea, althoughthe losses
are not very large. Summed over the seasonthe net lossesof
carbon as CO2 to the atmosphereduring the bulk of the snow.free season(May 29 to Septen•ber2, 1996) would be 31, 25,
and26 g m-2 for control,
extended
season,
andextended
seasm,
+ heating plots. respectively. The results from our treatment
plots suggestso far that systemCO2 balance will not change
substantiallywith decreasedduration of snow cover.
The patternsof systemflux show dramatic fluctuationsover
_•
arein unitsof (mgm-2d-•)
ExtendedExtended
+
2.0a
Control
Values are means+ one standarderror of three to six plots in 1995
and six plots in 1996. P values indicate results from Kruskal-Wallis
nonparametricanalysisof variance for treatmentcomparisons. Values
Extended •
Extended + Heat
1
2..
the season. These differencesare largely a consequence
of
changesin respiration,as can be seenfrom the 1996 data.
These patterns track fairly well the seasonal pattern of soil
temperature, a finding consistent with previous work in tussock tundra [Oberbauer et al., 1991]. The strong seasonal
temperature responseis consistentwith the finding of higher
respiration rates on the treatment plots.
In 1995, differences in net flux among treatments were
largest at the end of the season,when extendedand extended+
heating treatments had lower carbon losses than the control
plots. We interpret these results to be a prolonging of the
photosynthetic capacity of the graminoids, particularly
Eriophorum vaginaturn. Eriophorum vaginatum tussockshave
relatively high photosynthetic capacity and strong influence
on system fluxes [Tenhunen et al., 1995]. The end-of-season
decline in photosyntheticcapacity of E. vaginatum is thought
to be mediated by low temperatures[Defoliart et al., 1988].
By minimizing frosts and snow cover on the treatmentplots at
the end of the season,we likely reducedthe end of seasondecline in photosyntheticcapacity. A similar pattern was also
seen in early August 1996, but near the end of August, several
very hard freezesaccompaniedsnow storms,againstwhich the
A-frame tents were ineffective protection. Consequently,we
did not see a clear late-seasonpatternof more negative fluxes
for the treatmentplots in that year.
The two years also differed, in that the leaf area index increasedon treatmentplots in 1996 over 1995. The results are
suggestiveof a directional shift in plant community structure
..... ,•.... •,• •,
•n alpine warming experiments[Harte and
Shaw, 1995]. The resultspresentedhere representshort-term
responsesto the treatments. Ecosystem changes that may
take place over the longerterm, suchas shiftsin canopystructure, are likely to result in changesin the carbon balance of
these treatments
0
I
I
125
150
175
200
I
5/5
5/30
6/24
7/19
I
225
2:50
8/13
9/7
Day of year
Figure 6. Seasonalpatternof dark respirationmeasuredat
0400 hours for extendedseason,extendedseason+ heat, and
in the future.
Acknowledgments.
This study was supportedby the National
ScienceFoundationOffice of Polar Programsgrant OPP-9321626 as
part of the International Tundra Experiment (ITEX), an international
effort to assessthe responseof arctic and alpine plants and ecosystems
to climate change. Logistical supportby the staff of the Institute of
Arctic Biology, University of Alaska, Fairbanks, is gratefully
appreciated. Carlo Calandrielloand Brooke Shamblinmade invaluable
contributions
to collections
of field data.
Methane
measurements
were
controlplotsfor 1996.Valuesare meansof five or six plots. •nade possible by collaboration with Bill Reeburgh, University of
Standard
errorsare shownfor the lastsampledateto represent California-Irvine. Two anonymous reviewers made very helpful
magnitude of variation for treatments.
cmnmentson the •nanuscript.
29May
3 July
løss
54
I
ß
.••
ß
ß
ß
.
ß
.d•e
......... ...................
;-. ...........
Iee.....
© ß el....
o•_
,ell.
oo
ß
_
I
0
2 September
:
•• 0
upt•
7 August
500
ß
ß
.
I
1000
I
1500
i
0
500
I
I
1000
1500
!
0
•
,
500
1000
I
1500
I
0
5{X)
I
I(XX)
I
15(X)
2(XX)
Photosynthetic
0hoton
flux
density
(•tmol
n{2s-1)
Figure 7. Light-response
scattergrams
of net CO:zexchange
tbr selecteddatesof all plotscombinedin 1996.
ripariantundrain the northernfoothillsof theBrooksRange,Alaska,
USA, Oecologia,92, 568-577, 1992.
Oberbauer,S. F., J. D. Tenhunen,R. W. Virginia, W. Cheng,and C. T.
Bartlett, D. S., G. J. Whiting, and J. M. Hartman,Use of vegetation
Gillespie, Landscapepatternsof carbon gas exchangein tundra
indexesto estimatesolar radiationand net carbonexchangeof a
ecosystems,in Landscape Function and Disturbance in Arctic
References
grasscanopy, Remote Sens. Environ., 30, 115-128, 1989.
Billings, W. D., Carbonbalanceof Alaskatundraandtaigaecosystems:
Past,present,and future, Q. Sci. Rev., 6, 165-177, 1987.
Chapin,F. S., III, and G. R. Shaver,Individualisticgrowthresponseof
tundra plant speciesto environmentalmanipulationsin the field,
Ecology,66, 564-576, 1985.
Chapin, F. S., III, R. L. Jeffries,J. F. Reynolds,G. R. Shaver,and J.
Svoboda,Arctic plant physiologicalecology:A challengefor the
future,in Arctic Ecosystems
in a ChangingClimate,editedby F. S.
ChapinIII et al., pp. !-8, Academic,SanDiego,Calif.,1992.
Chapin, F. S., III, G. R. Shaver,A. E. Giblin, K. G. Nadelhoffer,and J.
A. Laundre, Responseof arctic tundra to experimentaland observed
changesin climate,Ecology,76, 694-711, 1995.
Tundra, Ecol. Stud., vol. 120, edited by J. F. Reynoldsand J. D.
Tenhunen,pp. 223-257, Springer-Verlag,New York, 1996.
Oechel, W. C., and W. D. Billings, Effects of global changeon the
carbon balance of arctic plants and ecosystems, in Arctic
Ecosystems
in a ChangingClimate,editedby F. S. Chapinet al., pp.
139-168, Academic,San Diego, Calif.,1992.
Oechel, W. C., and G. L. Vourlitis, The effects of climate changeon
arctictundraecosystems,TREE, 9, 324-329, 1994.
Oechel, W. C., S. J. Hastings,M. Jenkins,G. H. Riechers,N. Grulke,
and G. Vourlitis, Recentchangeof arctic tundraecosystems
from a
net carbon dioxide sink to a source,Nature, 361,520-523, 1993.
Peterson,K. M., and W. D. Billings, Carbon dioxide flux from tundra
soilsand vegetationas relatedto temperatureat BarrowAlaska,Am.
Midl. Nat., 94, 88-98, 1975.
Defoliart,L. S., M. Griffith, F. S. ChapinIII, andS. Jonassons,
Seasonal
Peterson,K. M., W. D. Billings, and D. N. Reynolds,Influenceof water
patterns of photosynthesisand nutrient storagein Eriophorum
table and atmosphericCO2 concentration
on the carbonbalanceof
vaginaturnL., an arcticsedge,Funct.Ecol.,2, 185-194,1988.
arctic tundra,Arct. Alp. Res., 16, 331-335, 1984.
Galen, C., and M. L. Stanton,Short-termresponses
of alpinebuttercups
Soil
to experimentalmanipulationsof growing seasonlength,Ecology, Post,W. M., W. R. Emanuel,P. J. Zinke, and A. G. Stangenberger,
carbonpoolsandworld life zones,Nature,298, 156-159, 1982.
74, 1052-1058, 1993.
in Alaskanaquatic
Gorham,E., Northernpeatlands:Role in the carboncycleand probable Schell,D. M., Carbon-13and carbon-14abundances
organisms:Delayed productionfrom peat in arctic food chains,
responses
to climaticwaxming,Ecol. Appl., 1, 182-195,1991.
Science, 219, 1068, 1983.
Grulke, N. E., G. H. Riechers,W. C. Oechel, U. Hjelm, and C. Jaeger,
Shaver,G. R., and J. Kummerow, Phenology,resourceallocation,and
Carbon balance in tussock tundra under ambient and elevated
growthof arctic vascularplants,in Arctic Ecosystems
in a Changing
atmospheric
CO2, Oecologia,83, 485-494, 1990.
Clintate,editedby F. S. ChapinIII et al., pp. 193-212,Academic,San
Harte, J., and R. Shaw, Shiftingdominancewithin a montanevegetation
Diego, Calif., 1992.
community:Resultsof a climate-warmingexperiment,Science,267,
876-880, 1995.
Tenhunen, J. D., C. T. Gillespie, S. F. Oberbauer,A. Sala, and S.
Whalen, Climate effects on the carbon balance of tussock tundra in
Hastings, S. J., S. A. Luchessa,W. C. Oechel, and J. D. Tenhunen,
the Philip Smith Mountains,Alaska,Flora, 190, 273-283, 1995.
Standingbiomassand productionin water drainagesof the foothills
Van Cleve, K., W. C. Oechel,and J. L. Hom, Responseof blackspruce
of the Philip Smith Mountains,Alaska,Holarctic Ecol., 12, 304-311,
1989.
(Picea mariana) ecosystemsto soil temperaturemodificationsin
interior Alaska, Can. J. For. Res., 20, 1530-1535, 1990.
Kant, D. L., L. D. Hinznian, and J.P. Zerling, Thermalresponseof the
active layer to climatic warming in a permafrostenvironment,Cold Wein, R. W., and L. C. Bliss, Primaryproductionin arcticcottongrass
tussocktundracommunities,
Arct.Alp Res.,6, 261-274, 1974.
Reg. Sci. Technol.,19, 111-112, 1991.
Kikuzawa,K., Leaf phenologyas an optimalstrategyfor carbongainin Weller, G., and B. Holmgren,The microclimatesof the arctictundra,J.
Appl. Meteorol.,13, 854-862, 1974.
plants,Can. J. Bot., 73, 158-163, 1995.
Lechowicz, M. J., and T. Koike, Phenologyand seasonalityof woody Whalen, S.C.; and W. S. Reeburgh,A methaneflux time seriesfor
tundraenvironments,Global Biogeochem.Cycles,2, 399-409, 1988.
plants:An unappreciatedelement in global changeresearch?,Can.
J. Bot., 73, 147-148, 1995.
Whiting, G. J., J.P. Chanton, D. S. Bartlett, and J. D. Happell,
Relationships
betweenCH4 emission,biomass,
andCO2 exchange
in
Manabe,S., and R. J. Stouffer,A CO2 climatesensitivitystudywith a
a subtropicalgrassland,J. Geophys.Res.,96, 3067-3071, 1991.
mathematicalmodelof the globalclimate,Nature,28, 491-493, 1979.
Maxwell, B., Arctic climate:Potentialfor changeunderglobalwarming,
in Arctic Ecosystems
in a ChangingClimate,editedby F. S. Chapin
III et al., pp. 11-34, Academic,SanDiego, Calif.,1992.
S. F. Oberbauer,E. W. Pop, and G. Starr,Departmentof Biological
Miller, P. C., Carbonbalancein northernecosystems
and the potential Sciences,Florida InternationalUniversity,UniversityPark, Miami, FL
effect of carbondioxide-induced
climaticchange,CONF-8003118, 33199. (e-mail: oberbaue@fiu.edu)
Natl. Tech. Inf. Serv.,Springfield,Va, 1981.
Molau, U., International Tundra Experiment:ITEX Manual, 2nd ed., (ReceivedAugust 11, 1997;revisedJanuary19, 1998;
DanishPolar Center,Copenhagen,1996.
acceptedFebruary 11, 1998.)
Oberbauer,S. F., J. D. Tenhunen,and J. F. Reynolds,Environmental
effectson CO: efflux from watertrack andtussocktundrain Arctic
Alaska,USA, Arct. Alp. Res.,23, 162-169, 1991.
Oberbauer,S. F., C. T. Gillespie,W. Cheng,R. Gebauer,A. SalaSerra,
and J. D. Tenhunen,Environmentaleffectson CO2 efflux from