Characterization of the carbon fluxes of a vegetated drained lake basin chronosequence on the Alaskan Arctic Coastal Plain

Characterization of the carbon fluxes of a vegetateddrained lake basin chronosequence on the AlaskanArctic Coastal Plain

D . Z O N A *1 , W. C . O E C H E L w , K . M . P E T E R S O N z, R . J . C L E M E N T S § , K . T . PAW U }and S . L . U S T I N *

*Department of Land, Air, and Water Resources, Center for Spatial Technologies and Remote Sensing (CSTARS), University of

California, Davis, CA 95616, USA, wGlobal Change Research Group, Department of Biology, San Diego State University, San

Diego, CA 92182, USA, zDepartment of Biology, University of Alaska, Anchorage, AK 99508, USA, §School of GeoSciences,

University of Edinburgh, Edinburgh, UK, }Department of Land, Air, and Water Resources, University of California, Davis,

CA 95616, USA

Abstract

Greenhouse gas fluxes from vegetated drained lake basins have been largely unstudied,although these land features constitute up to 47% of the land cover in the Arctic Coastal Plainin northern Alaska. To describe current and to better predict future sink/source activity of theArctic tundra, it is important to assess these vegetated drained lake basins with respect to thepatterns of and controls on gross primary production (GPP), net ecosystem exchange, andecosystem respiration (ER). We measured CO2 fluxes and key environmental variables duringthe 2007 growing season (June through August) in 12 vegetated drained lake basinsrepresenting three age classes (young, drained about 50 years ago; medium, drained between50 and 300 years ago; and old, drained between 300 and 2000 years ago, as determined byHinkel et al., 2003) in the Arctic Coastal Plain. Young vegetated drained lake basins had boththe highest average GPP over the summer (11.4 gCO2 m�2 day�1) and the highest averagesummer ER (7.3 gCO2 m�2 day�1), while medium and old vegetated drained lake basinsshowed lower and similar GPP (7.9 and 7.2 gCO2 m�2 day�1, respectively), and ER (5.2 and4 gCO2 m�2 day�1, respectively). Productivity decreases with age as nutrients are locked up inliving plant material and dead organic matter. However, we showed that old vegetateddrained lakes basins maintained relatively high productivity because of the increaseddevelopment of ice-wedge polygons, the formation of ponds, and the re-establishment ofvery productive species. Comparison of the seasonal CO2 fluxes and concomitant environ-mental factors over this chronosequence provides the basis for better understanding thepatterns and controls on CO2 flux across the coastal plain of the North Slope of Alaska and formore accurately estimating current and future contribution of the Arctic to the global carbonbudget.

Keywords: Arctic Coastal Plain, carbon fluxes, classification of vegetated drained lake basins, climate

feedbacks, global change, old-growth ecosystems, trace gas fluxes, vegetated drained lake basins

Received 19 March 2009; revised version received 7 September 2009 and accepted 1 October 2009

Introduction

Together, thaw lakes and vegetated drained lake basins

comprise the majority of the land surface (50–70%) of

the Arctic Coastal Plain in northern Alaska (Hussey &

Michelson, 1966; Hinkel et al., 2003, 2005). In particular,

vegetated drained lake basins constitute up to 47% of

the land cover in the northern part of the Arctic Coastal

Plain (Hinkel et al., 2005).

Thaw lake formation is a consequence of the com-

bined action of thermal erosion of the edges of small

ponds and permafrost melting (Carson & Hussey, 1962).

The elliptical shape and the orientation of these lakes

are the product of wind-driven circulation cells, result-

ing in the lakes’ long axes being perpendicular to the

Correspondence: D. Zona, tel. 1 001 30 52 092, fax 1 001 30 54 353,

email: [email protected]

1Present address: Department of Biology, University of Antwerp,

Research Group of Plant and Vegetation Ecology, Universiteitsplein

1, B-2610 Wilrijk, Belgium, tel. +32 326 52831, fax +32 32 65227,

e-mail: [email protected]

Global Change Biology (2010) 16, 1870–1882, doi: 10.1111/j.1365-2486.2009.02107.x

1870 r 2009 Blackwell Publishing Ltd

mailto:[email protected]

main wind direction (Mackay, 1956; Rex, 1961; Carson &

Hussey, 1962). These erosion cells lead to expansion of

the lakes and incorporation of nearby ponds or other

lakes (Hinkel et al., 2003). The final stage of the thaw

lake cycle involves drainage due to some combination

of bank overflow, ice wedge erosion, coastal erosion,

headward erosion by streams (Hopkins, 1949; Mackay,

1988) or, in a few cases, human intervention (Hinkel

et al., 2007).

After drainage, vegetation establishes on the drained

lake bed, and organic matter accumulates from the

buildup of plant detritus (Mackay & Burn, 2002; Hinkel

et al., 2003; Bockheim et al., 2004). The thickness of the

organic layer and the fraction of nonextractable carbon

increase with time since drainage, transforming the

organic matter into more recalcitrant forms (Bockheim

et al., 2004; Bockheim & Hinkel, 2007).

Greenhouse gas fluxes from lakes and vegetated

drained lake basins have been largely unstudied. There

has been some research on CO2 flux over lakes (e.g.

Coyne & Kelley, 1974; Kling et al., 1991), but most

research on the patterns and controls of CO2 flux in

tundra regions has been carried out on areas outside of

lakes or vegetated drained lake basins (see, e.g. Oechel

et al., 2000; Kwon et al., 2006). This relative lack of

studies undermines our ability to estimate the future

impact of climate change on land atmosphere CO2

fluxes in the Arctic.

Young vegetated drained lake basins appear to con-

tain the most productive vegetation and the greatest

nutrient availability (K.M. Peterson unpublished re-

sults; Billings & Peterson, 1980; Bliss & Peterson,

1992). Runoff and nutrients move downslope from

more elevated sites (Kummerow et al., 1987), accumu-

late in the thaw lakes, and become available to plants

after the lakes drain. It has been assumed but not

proved that, as nutrients are progressively locked up

in plant material during the terrestrial phase, plant

productivity declines. This decline is further assumed

to cause a slowing of carbon accumulation with age

(Billings & Peterson, 1980; Hinkel et al., 2003; Bockheim

et al., 2004). This general pattern fits Odum’s (1969)

prediction of decreasing net primary production and

net ecosystem exchange (NEE) with maturity of the

ecosystem (Odum, 1969). Therefore, medium and old

vegetated drained lake basins are thought to become

progressively less productive.

Because soil moisture operates as a major control on

ecosystem respiration (ER) and NEE (Billings et al., 1983;

Oechel et al., 1998), it would be expected to play a key

role, together with time after drainage, in predicting

carbon exchange from the vegetated drained lake basins.

The research reported here was designed to investi-

gate some of the patterns of and controls on the

heterogeneity of carbon fluxes in the Arctic at the

landscape scale. In particular we assessed whether the

current classification of the Arctic Coastal Plain, based

on vegetated drained lake age (Hinkel et al., 2003), can

be used to predict the differences in primary produc-

tivity and carbon uptake from the Arctic tundra. We

also investigated the controls of key environmental

variables [e.g. soil moisture, thaw depth, photosynthe-

tically active radiation (PAR) incoming and outgoing,

net radiation, air temperature, vapor pressure deficit,

soil temperature, and precipitation] on NEE, GPP, and

ER in the age-graded vegetated drained lake basins, to

better understand the processes controlling their carbon

fluxes.

Study site

In this study, we selected 12 vegetated drained lake

basins (Table 1) based on the age classification of the

Barrow Peninsula by Hinkel et al. (2003) in the northern

part of the Arctic Coastal Plain. These sites included

four young (draining about 50 years ago), four medium

(draining between 50 and 300 years ago), and four old

(draining between 300 and 2000 years ago) vegetated

drained lake basins (Fig. 1). Hinkel et al. (2003) provides

details on the vegetated drained lake basins classifica-

tion. Figure 2 presents a visual representation of the

vegetation at these sites. Three of the vegetated drained

lake basins (the adjacent Young 1 and Young 2 that form

Dry Lake, and Footprint Lake) were artificially drained

in 1949–1950 to allow access to a natural gas well near

Barrow (Hinkel et al., 2007). In the other cases the cause

of the drainage is unknown or difficult to verify; how-

ever, in lakes near the ocean, coastal erosion is probably

a major factor leading to drainage (Hinkel et al., 2007).

Table 1 Site name, age (young 0–50 years old, medium 50–

300 years old, old 300–2000 years old) and locations of the 12

vegetated drained lake basins measured during summer 2007

Site Age Latitude Longitude

Young 1 Young 71115023.0400N 156137044.4000W

Young 2 Young 71114026.5200N 156126042.6000W

Young 3 Young 71114050.8500N 156136052.0900W

Medium 1 Medium 71115012.6500N 156133051.9300W

Medium 2 Medium 71115057.3200N 156132040.9100W

Medium 3 Medium 71114059.4700N 156122041.4000W

Old 1 Old 71115010.0800N 156132038.4000W

Old 2 Old 71115026.8100N 156131010.5600W

Old 3 Old 71114047.3800N 156123051.8500W

Footprint Lake Young 71116015.5400N 156139049.1900W

Thin Lake Old 70158039.3500N 156155036.8600W

Elbow Lake Medium 711 9028.3900N 156145010.1200W

CHARACTERIZATION ON THE CARBON FLUXES IN THE ARCTIC COASTAL PLAIN 1871

r 2009 Blackwell Publishing Ltd, Global Change Biology, 16, 1870–1882

After drainage, plant succession occurs as edaphic

conditions change. The graminoid Arctophila fulva,

dominant in the wettest areas, in addition to Dupontia

fisheri and Eriophorum scheuchzeri (sedges) are the first

plants to establish. In general, the vascular plants of the

young vegetated drained lake basins are thought to be

more productive than those at older sites (Billings &

Peterson, 1980; Bliss & Peterson, 1992). In medium aged

vegetated drained lake basins, the most characteristic

vegetation type is formed by large, often circular, clones

of Carex aquatilis and Eriophorum angustifolium. Medium

and young vegetated drained lake basins share some

commonality in vegetation (A. fulva, D. fisheri, and

E. scheuchzeri).

After the lakes drain, frozen soil contracts during

winter, cracking the surface into polygons. In spring

and summer, these cracks are filled by water, which

turns into ice during the following fall. Expansion of the

freezing water displaces soil and forms ice-wedges

(Britton, 1957; Billings & Peterson, 1980). The repeated

cracking, inundation, and freezing lead to the expan-

sion of the ice-wedges and to the rising of the rims of

the polygons. The ice-wedge polygons become more

evident as the vegetated drained lake basins age (see

Fig. 2; Hussey & Michelson, 1966; Billings & Peterson,

1980). The rising of the rims creates depressions in the

central areas, which fill with water and form small

ponds. The deeper ponds of older aged drained lakes

are dominated by A. fulva (Billings & Peterson, 1980;

Bliss & Peterson, 1992). In the shallower ponds the

vegetation is mainly composed of C. aquatilis and

Sphagnum ssp., while Dicranum elongatum is typically

present in the ice-wedge polygon rim.

Materials and methods

Period of measurements

Nine of these 12 sites were measured three times each

(Young 1–3, Medium 1–3, Old 1–3, shown in Fig. 1, and

Table 1), with a rotation of three eddy covariance towers

every 4–9 days from the middle of June to the end of

August 2007.

The data collection was organized into weeks (week

1: June 12–18; week 2: June 18–25; week 3: June 25–30;

week 4: June 30–July 9; week 5: July 9–14; week 6: July

Fig. 1 Landsat-7 image from August 30, 2000 of the North Slope of Alaska (displayed is Band 4). Indicated are the vegetated drained

lake basins measured during the summer of 2007.

1872 D . Z O N A et al.


16–24; week 8: August 1–7; week 9: August 7–13; week

10: August 13–17; week 11: August 17–21; week 12:

August 22–28). Data from week 7 (July 24–31) were

not included due to the large data loss. Early season

refers to weeks 1–5, peak season to weeks 6–8, and late

season to weeks 9–12.

Three of the 12 vegetated drained lake basins (Foot-

print Lake, Elbow Lake, and Thin Lake, shown in bold

in Fig. 1), occur in locations without road access and

were measured once with the assistance of a helicopter

between July 16 and 24 (week 6).

Eddy covariance measurements

Fluxes of CO2, H2O, and energy were measured using

eddy covariance, a micrometeorological method that

measures the net exchange of a scalar between the

biosphere and the atmosphere (Swinbank, 1951; Desjar-

dins & Lemon, 1974; Baldocchi, 2003). Three portable

eddy covariance towers were used to measure CO2 and

H2O and energy fluxes in the 12 sites. The CO2 and H2O

fluxes were measured using an open path infrared

analyzer (Li-COR 7500, Li-COR, Lincoln, NE, USA).

Wind speed, direction, and the energy fluxes were

measured using a sonic anemometer (Model CSAT3,

Campbell Scientific, Logan, UT, USA). The CO2 and

H2O fluxes, and sonic windspeeds were sampled and

recorded at a 10 Hz sampling rate using a model CR

3000 datalogger (Campbell Scientific Inc.).

The sonic anemometer and the infrared analyzer

were positioned at a height from 2.0 to 2.7 m, with the

sonic anemometer oriented toward the predominant

wind direction (eastward). The height of the sonic

anemometer was mostly related to the slight differences

in topography at each of the sites. The precise orienta-

tion of the anemometer was calculated using a Trimble

GPS. Orientations were flagged and initial orientations

were replicated during subsequent visits.

CO2 and H2O vapor were calibrated every 2–4 weeks

using ultra high-purity nitrogen for the zero CO2 and

H2O, and a 729 ppm CO2 for the CO2 span (CO2 in air;

certified grade � 1 ppm, Matheson Gas Product, Mon-

tgomeryville, PA, USA). A dew point generator (Li-610,

Li-COR, Lincoln) was used to produce an air stream

with a known water vapor dew point (typically 7 1C

lower than the ambient air temperature) for H2O vapor

calibration.

Postprocessing of the eddy covariance data

Fluxes of CO2 and H2O vapor, sensible heat and

momentum were calculated using the EDIRE software

(version 1.4.3.1169, Robert Clement, University of

Edinburgh). Obvious data outliers were removed,

which were values more than 6 SD from the 30 min

mean for CO2 and H2O vapor and 410 SD from the

30 min mean for the wind velocity components, u, v, and

w. A two components rotation was applied to set mean

vertical (w) and lateral (v) velocity components to zero.

Time delays were calculated using a cross-correlation

function of the scalar fluctuation and the vertical wind

velocity.

Fig. 2 Oblique photographs and characteristic appearance of

the vegetation shown on the bottom right of three vegetated

drained lake basins: (a) young, (b) medium, and (c) old. Notice

the white cottony heads of Eriophorum scheuchzeri in the young

basin (a).



A frequency response correction was applied to the

eddy covariance fluxes following Moore (1986) and

using theoretical attenuation functions and Kaimal

model spectra to account for high-frequency and

low-frequency fluctuations in signal losses. Correction

for density change was applied following Webb et al.

(1980). Further correction using the energy budget

closure ratio, as suggested by some researchers includ-

ing Twine et al. (2000), was not carried out. In fact, we

agreed with the critical analysis of Ham & Heilman

(2003) and Baldocchi (2003) that point out that the

energy budget ratio method is based on assumptions

(not supported theoretically nor empirically) that CO2

and H2O vapor measurement errors are identical and

that therefore, the errors of sensible heat and water

vapor are proportional. The other assumptions required

by the energy budget closure correction and possibly

not met are that there are no adjective errors, as dis-

cussed by Paw U et al. (2000) and that radiation and

ground heat flux measurements are highly accurate and

have the same footprint as the eddy covariance mea-

surements of interest.

The energy budget closure averaged 92%, with a

range of 85–97% on a weekly basis, with no consistent

differences among sites. The efficacy of the corrections

applied was confirmed with the energy budget closure

checks, which revealed a 92% closure average, consid-

ered very good in comparison with many Fluxnet sites

(Wilson et al., 2002). Data quality were assessed by

analysis of energy budget closure and by comparison

of cospectra of w0CO02, w0T0, w0H2O0 (Kaimal et al., 1972).

The data were filtered by wind direction, and only wind

directions within � 901 from the sonic orientation were

used thereby excluding data outside of the footprint of

interest. A friction velocity (u*) threshold of 0.15–

0.3 ms�1, depending on the site, was used as cut off,

and data below the appropriate (u*) values were re-

moved.

Two footprint models were applied to the data (Hsieh

et al., 2000; Kljun et al., 2004; Detto et al., 2006). Both

models indicated that the majority of the fluxes were

coming from the first 150–200 m upwind of the eddy

tower.

Linear interpolation was used to fill short gaps, from

0.5 to 2.5 h (Falge et al., 2001). Larger gaps were not

interpolated and the data from days with these gaps

were not included in the analysis (week 7). After gap

filling, the data from the same week in each of the sites

were averaged for the same time blocks and they were

summed to estimate a typical daily NEE. Because there

was no darkness at the beginning of the season, ER was

calculated at zero PAR from the regression of NEE

against light intensity. Later in the season ER was

estimated as an average of CO2 fluxes at zero PAR.

Environmental variables

Manual data collection. In nine of the 12 sites (the sites

where the sampling was replicated), a square grid of

100 points (100 m� 100 m, with one point every 10 m)

was established to measure thaw depth and soil

moisture over the season. The grids were positioned

depending on the accessibility of the sites, generally

20–100 m south of the eddy covariance tower on the

western edge of the vegetated drained lake basins

(Fig. 3), following procedures described by the Active

Layer Protocol based on the ITEX guidelines

Fig. 3 Quickbird image from August 1 and 2, 2002 (from

Manley et al., 2004), loaded as color infrared, of the vegetated

drained lake Old 1. Note the position of the eddy covariance

tower and one of the transects of spectral measurements (in

yellow) and of the grid (green) where thaw depth and soil

moisture measurements were recorded. Noticeable are the red

ponds with the most productive vegetation. The red and blue

lines are roads.

1874 D . Z O N A et al.


(http://www.geog.ubc.ca/itex/library/). A GPS (Garmin

International Inc., Olathe, KS, USA) was used to

establish the four corners of the grid, and each of the

100 points inside the grid was flagged.

During the period that the eddy covariance tower

was at each site, thaw depth and soil moisture

measurements were taken at all grid points using a

graduated metal rod and a Fieldscout TDR 300 (12 cm

probe). Soil moisture measurements were made when

the thaw depth was � 12 cm (the length of the probe).

Less than 1% of the measurements were excluded as a

result.

Automated data collection. Environmental variables were

also recorded continuously during the period the eddy

covariance towers were at each of the sites. All the

environmental measurements were sampled every

minute, averaged over 30 min, and recorded using the

datalogger CR 3000 (Campbell Scientific Inc.) that was

also used to acquire the eddy covariance data. Soil

moisture was measured continuously at each site

using four probes (each 5 cm long) inserted diagonally

in the 0–10 cm depth profile at four different locations

(ECH2O-5, Decagon Device Inc., Pullman, WA, USA)

and recorded with the CR 3000 datalogger. All soil

moisture probes used were laboratory and field

calibrated during the summer.

A net radiometer (Model NR-Lite, Kipp and Zonen,

Delft, the Netherlands) was used to measure the net

radiation in the spectral range of 0–100 mm. Incoming

and reflected (PAR, 0.4–0.7mm) was measured with two

PAR sensors (Model LI-190SB, LI-COR Inc.). The PAR

sensors were calibrated at the end of the summer, and

appropriate calibration coefficients were retroactively

applied to the data. The net radiometer and the PAR

sensors were installed on a tripod at about 1.5 m height

above the ground, close to the eddy covariance tower.

Air temperature and relative humidity were measured

using a thermistor and capacitive RH probe (Model

HMP45C-L, Vaisala Inc., Helsinki, Finland). Soil

temperatures were measured using type-T

thermocouples at the ground surface (i.e. the top of

the moss layer or, when not present, the top of the

organic litter layer) and at 5 and 10 cm below the

ground surface. Ground heat flux (G) was measured

using five soil heat flux plates (Model HFT3, REBS Inc.,

Seattle, WA, USA) buried between 2 and 5 cm below the

ground surface.

Spectral measurements

Field spectral reflectances were collected at 11 of the 12

vegetated drained lake basins (data at one site is miss-

ing due to harsh weather conditions) mainly during

clear sky conditions from July 20 to August 28 (weeks

6–12). These measurements were collected using a full

range spectrometer (350–2500 nm) (Analytical Spectral

Devices, Boulder, CO, USA) and standardized to Spec-

tralon panel (Labsphere, North Sutton, NH, USA). The

spectra were collected at 10–20 points in the footprint of

the eddy covariance towers in one or two transects

every 10 m from the eddy covariance tower to up to

100–200 m in eastward directions, the prevalent wind

direction. The spectra were recorded at about 1.5 m

above the ground, corresponding to a circular surface

area of about 66 cm in diameter for a field of view of 251.

The aerial extent of these measurements is about 3.4 m2

for 10 measurements and 6.8 m2 for 20 measurements.

Several vegetation indexes [normalized difference

vegetation index (NDVI), Tucker, 1979; normalized dif-

ference nitrogen index (NDNI), Serrano et al., 2002;

normalized difference water index (NDWI), Gao, 1996;

water band index (WBI), Penuelas et al., 1997; normal-

ized phaeophytinization index (NPQI), Barnes et al.,

1992; and cellulose absorption index (CAI), Nagler

et al., 2000) were calculated from the spectra to estimate

biophysical parameters in each of the vegetated drained

lake basins. NDVI is related to plant biomass in the

predominantly herbaceous tundra communities, NDNI

to plant nitrogen content, NDWI and WBI to plant

water content, CAI to litter content, and NPQI to

phaeophytin concentration and chlorophyll degrada-

tion. The two water indexes (NDWI and WBI) had very

similar results, so only the results from the analysis

using NDWI are reported. We choose NDWI because it

is a normalized index and it was proven to be less

sensitive to atmospheric effects (Gao, 1996).

The validity of NDVI has been demonstrated in the

Arctic (Hope et al., 1993; Stow et al., 1998). We assume

that even though NDWI, CAI, and NPQI were not

directly tested for tundra vegetation, because their

calculation is based on the absorption features of water

(NDWI), cellulose (CAI), and phaeophytin (NPQI),

these should be independent from the geographic loca-

tion. On the other hand, nitrogen content (and conse-

quently NDNI) is correlated with the absorption

characteristics of N–H and C–H bonds associated with

leaf proteins (Curran, 1989; Barton et al., 1992). These

features could vary across different plant species, but

they have been found to be valid across broad range of

plant species and ecosystems, such as deciduous, con-

iferous forests, Mediterranean shrubs, and temperate

forests (Matson et al., 1994; Martin & Aber, 1997; Smith

et al., 2002, 2003; Serrano et al., 2002). However, because

most of these indexes were not directly validated in the

Arctic tundra, we are using them as relative, not abso-

lute, indicators of the water, litter, chlorophyll degrada-

tion, and nitrogen content. We look forward to future



http://www.geog.ubc.ca/itex/library/














studies that further validate these indexes for the Arctic

regions.

Figure 3 shows the transect that was used to collect

the spectral reflectance measurements over an old

vegetated drained lake basins (old 1).

Statistical analysis

We analyzed the data, averaged weekly for each age-

graded vegetated drained lake, using a linear mixed

effects model with fixed effects for weeks and age and a

random effect for site, to account for possible depen-

dence of repeated measures (SAS for Windows version

9.2, SAS Institute Inc., Cary, NC, USA). Model fit was

assessed using a graphical analysis of residuals and a

Shapiro–Wilk test for normality. Pairwise comparisons

were made using the Tukey adjustment for multiple

testing. The linear mixed effects model was used to

assess the additional impact of age and time. However,

this model did not capture the importance of all the

variables of interest, and other models, described next,

were applied to the data.

General linear modeling (SYSTAT, version 10, Systat

Software Inc., 2002), with stepwise multiple regression

approach, was used to discriminate between variables

and rank the most important environmental factors

correlated with CO2 fluxes. One way ANOVA (using the

least square mean and Bonferroni’s post hoc test) was

used to test the significance of the differences in CO2

fluxes, soil moisture, thaw depth, and other environ-

mental variables among age-graded vegetated drained

lake basins during the different time periods. These

models were applied to the data both for the entire

season (for the data from weeks 1–12) and divided into

three periods: early season (weeks 1–5), peak season

(weeks 6–8) and late season (weeks 9–12) to explore

differences in controls on the CO2 fluxes at different

times in the growing season.

Results

Gross primary production (GPP), averaged over the

entire summer season, was highest in young vegetated

drained lakes when compared with medium and old

aged vegetated drained lakes (Fig. 4a), which showed

no difference. Only the difference in GPP between

young and old vegetated drained lakes was significant

(one-way ANOVA Fisher’s least significant difference test

P 5 0.037). Young vegetated drained lake basins

showed the highest plant biomass (NDVI), the max-

imum plant nitrogen (NDNI) and the highest plant

water content (NDWI), as deduced from spectral mea-

surements (Fig. 5).

ER (Fig. 4a), averaged over the entire season was also

found to be highest in young vegetated drained lake

basins, followed by medium and then old. The one-way

ANOVA Fisher’s least significant difference test showed a

significant difference in ER between young and med-

ium (P 5 0.008), and slightly significant difference be-

tween young and old vegetated drained lake basins

(P 5 0.059).

Plant biomass and plant nitrogen content are both

related with water availability, as NDNI was correlated

with plant water content (NDWI) (Pearson’s correlation

coefficient of 0.74). NDWI explained 73% of the varia-

bility in NDVI, highlighting the relevance of hydrological

conditions to plant productivity in the Arctic tundra. The

microtopographic controls on establishment of produc-

tive species is demonstrated by the high NDVI values in

the low center ponds in old vegetated drained lake basins

(see Fig. 6). The local mineralization in the ponds allows

the growth of Arctophila, increasing the productivity of

the old vegetated drained lake basins (Fig. 6a). In fact,

Fig. 4 (a) Daily total ecosystem respiration (ER) and gross

primary production (GPP), and (b) net ecosystem exchange

(NEE) (gCO2 m�2 day�1), for age-graded vegetated drained lake

basins averaged over the entire period of measurement (from

June 12 to August 28 2007, which corresponds to weeks 1–12).

Positive values indicate carbon sources, and negative values

indicate carbon sinks. Error bars represent standard error of

the mean.

1876 D . Z O N A et al.


even though on average the NDVI of the old vegetated

drained lake basins is similar to that of the medium

vegetated drained lake basins, the highest NDVI values

were found in ponds where Arctophila occurs and the

lowest values in ice-wedge polygon rims (Fig. 6b). The

medium vegetated drained lake basins were more homo-

geneous in NDVI values (average standard deviation in

medium 0.05, and in old 0.07).

The decrease in NDVI with age is also connected to

accumulation of litter (see increase in CAI, Fig. 5d),

which covers the moss layer, increasing the shadow

fraction. The old vegetated drained lake basins con-

tained slightly less litter than the medium vegetated

drained lake basins (Fig. 5d), probably due to the higher

mineralization in ponds in old vegetated drained lake

basins (Fig. 3). The increase in litter, which tends to be

drier than living vegetation, causes old vegetated

drained lake basins to appear drier than young ones

(as shown by lower NDWI values, Fig. 2c), even when

soil water content is similar (Fig. 7).

GPP and NEE were not significantly correlated with

NDVI, when modeled over the full period of measure-

ments (July 20 to August 28). This result could be due to

the complexity of the controls on plant productivity,

particularly during the limited period of the spectral

measurements, and to the white cottony heads of E.

scheuchzeri (Fig. 2a), which at peak season decreased the

NDVI. Flowering is responsible for the lower NDVI at

10, 20, and 100 m at Young 3 on August 8 (Fig. 6b).

However, more productive young vegetated drained

Fig. 5 (a) Normalized difference vegetation index (NDVI), (b)

normalized difference nitrogen index (NDNI), (c) normalized

difference water index (NDWI), and (d) cellulose absorption

index (CAI) averaged per age-graded vegetated drained lake

basin for the entire measurement period (July 20 to August 28,

2007, which corresponds to weeks 6–12). Error bars represent

standard error of the mean.

Fig. 6 (a) Cross section of the ice-wedge polygon structure with

the low center pond of Old 1, and (b) normalized difference

vegetation index (NDVI) across a transect measured on August

7, 2007 in Med 1 and Old 1, and on August 8, 2007 in Young 3.



lakes had higher NDVI. We used another index, NPQI,

to model primary productivity, which was able to

explain 48% of the variability in GPP (in a general linear

model, with P 5 0.004).

A linear mixed effects model was used to investigate

the differential environmental controls on NEE, GPP,

and ER in the vegetated drained lake basins during the

season. This model was used to assess the additional

impact of age and time. Because of data limitations, the

only conclusion that could be drawn was that season

had a significant impact on NEE, GPP, and ER, and that

the model could not capture the importance of the other

variables. As shown in Fig. 8, young and old vegetated

drained lake basins showed similar seasonal trends in

NEE, while medium vegetated drained lake basins

showed a lower uptake, even during peak season

(weeks 6–8). GPP in young and old vegetated drained

lake basins were similar in week 6 but not in week 8

(Fig. 8). The re-establishment of productive species in

ponds is probably responsible for the higher plant

productivity of old vegetated drained lake basins dur-

ing week 6 (Fig. 8). However, this increase did not

continue to reach the level of the young vegetated

drained lake basins (see week 8, Fig. 8, when old and

medium vegetated drained lake basins have similar

productivity). Overall, old vegetated drained lake ba-

sins’ productivity was similar to that of the medium, as

already described in the previous section (Fig. 4).

As a significant seasonal effect was observed, the

stepwise multiple regression analysis was performed

for three different periods: early, peak, and late season

(respectively, weeks 1–5, 6–8, 9–12; Table 2). NEE,

GPP, and ER were modeled in these three periods as

Fig. 7 Soil moisture (n 5 100; volumetric water content, %) for

age-graded vegetated drained lake basins averaged for the entire

measurement period (from June 12 to August 28, 2007, which

corresponds to weeks 1–12). Error bars represent standard error

of the mean.

Fig. 8 (a) Gross primary production (GPP), and (b) net ecosys-

tem exchange (NEE) in gCO2 m�2 day�1, (c) soil moisture

(VWC%), and (d) thaw depth (cm from surface) for age-graded

vegetated drained lake basins averaged over the weeks of

measurements (week 1: June 12–18; week 2: June 18–25; week

3: June 25–30; week 4: June 30–July 9; week 5: July 9–14; week 6:

July 16–24; week 8: August 1–7; week 9: August 7–13; week 10:

August 13–17; week 11: August 17–21; week 12: August 22–28).

Positive values of NEE indicate carbon sources, and negative

values indicate carbon sinks. Error bars in (c) and (d) are

standard error of the mean.

1878 D . Z O N A et al.


functions of several environmental variables (soil moist-

ure, thaw depth, PAR incoming and outgoing, net

radiation, air temperature, vapor pressure deficit, soil

temperature, and precipitation).

At the beginning of the season (weeks 1–5) NEE was

found to be strongly dependent on depth of thaw

(which explained about 68% of the variability in CO2

fluxes). GPP at this time was also largely explained by

the depth of thaw (which explained 60% of its varia-

bility). A model that includes the interaction term

between thaw depth and vegetated drained lake basin

age explained 84% of the variability in ER (Table 2).

Respiration increases with thaw depth, and the great-

er degree of respiration in young vegetated drained lake

basins compared with medium and older lake basins is

probably due to their higher nutrient content in soil,

denser roots, and richer microbial biomass. D. fisheri, a

dominant species in young vegetated drained lake

basins, has a shallow root system (Shaver & Billings,

1975), and would contribute to the relatively greater

proportion of belowground biomass that may exist in

the early thawed soil fraction near the surface. As a

result, in young vegetated drained lake basins, there

would be a more rapid initial increase in respiration

with thaw than in older vegetated drained lake basins.

Young vegetated drained lake basins have higher soil

moisture, which has been associated with larger heat

conductance and heat transfer in tundra soils (Pavlov,

1975; Hinzman et al., 1991), and therefore a larger depth

of thaw (Fig. 8c and d).

At peak season (weeks 6–8) soil moisture explained

70% of the variability in NEE. Thaw depth was still very

important in driving respiration rates (i.e. alone it was

able to explain 70% of the variability in ER, and a model

that included thaw depth and air temperature was able

to explain 92% of the variability in ER). During peak

season, in absence of other environmental limitations

(e.g. light availability and phenological development),

soil temperature appears to be the limiting factor on

photosynthetic process (explaining 76% of the variabil-

ity in GPP, Table 2).

Later in the season (weeks 9–12), ER was mainly

dependent on the age of the vegetated drained lake

basins (which explained 62% of the variability in ER).

Once the different vegetation communities have

reached maturity in the vegetated drained lake basins,

they are the major predictor of the respiration rates.

The senescence occurring in late summer probably

leads to the dominance of respiration over photosynth-

esis in determining GPP.

Discussion

Even though the controls on CO2 fluxes are complex

and strongly dependent on the season, the classification

of the vegetated drained lake basins according to age

(Hinkel et al., 2003) can provide very useful information

about the landscape and identify the areas with the

highest CO2 uptake. These are areas with higher plant

biomass and plant nitrogen content (NDVI, NDNI),

higher plant water content (NDWI), and lower litter

(CAI).

Medium vegetated drained lake basins had the low-

est soil moisture (Fig. 7) and more soil aeration; there-

fore they would have been expected to have the highest

soil respiration (Billings et al., 1983; Funk et al., 1994;

Oechel et al., 1998). However, ER in medium vegetated

drained lake basins was lower than in young (Fig. 4),

and similar to ER in old vegetated drained lake basins.

Probably plant biomass is a major control on the re-

spiration rates in these ecosystems, and the higher plant

biomass in the young vegetated drained lake basins

more than compensated for the depressive effects of

high soil moisture in these systems (Fig. 5a).

Notably, older vegetated drained lake basins (i.e.

those 300–2000 years old) are nearly as productive as

the medium aged vegetated drained lake basins in

terms of net carbon uptake during the growing season.

The assumption that productivity decreases with ma-

turity of the ecosystem seems not to apply to these

Arctic ecosystems, as medium and old vegetated

drained lake basins showed similar average NEE and

GPP (Fig. 4). These results support recent reports of

sustained ecosystem sequestration in older aged stands

(Paw U et al., 2004; Luo et al., 2007; Luyssaert et al.,

Table 2 Multivariable models results for the weekly averaged CO2 fluxes (gCO2 m�2 day�1) as a function of several environmental

variables during the three periods (early, peak, and late season)

Early season (weeks 1–5) Peak season (weeks 6–8) Late season (weeks 9–12)

NEE Thaw depth (R2 5 68%) Soil moisture (R2 5 70%) –

GPP Thaw depth (R2 5 60%) Soil temperature (R2 5 76%) –

ER Thaw depth� age (R2 5 84%) Thaw depth� air temp (R2 5 92%) Age (R2 5 62%)

Reported are the R2 of the models with the highest explanatory power. No variables was able to significantly explain NEE and GPP

in late season.



2008), even in northern peatlands, more than 2000 years

old (Lafleur et al., 2003; Leppala et al., 2008).

In this study, the ability of vegetated stands to main-

tain summer biomass accumulation over millennia is

probably connected to the increased polygonization and

the development of ponds in old vegetated drained lake

basins (Fig. 3). These older stands are sites of organic

matter mineralization and re-establishment of produc-

tive plant species that tolerate water-logged soils (Fig. 6)

and benefit from increased nutrients from decomposing

peat (Billings & Peterson, 1980; Bliss & Peterson, 1992).

On a larger temporal and spatial scale, the cycling of

carbon by vegetated drained lake basins potentially

makes the ecosystem as a whole a dynamic mosaic of

carbon storage and release that is not simply described

by Odum’s hypothesis, which only includes change in

time as a driving variable. The addition of data from the

entire year would certainly change the estimate of

carbon uptake from these ecosystems (Oechel et al.,

2000; Lafleur et al., 2003) and a full year should be

included in future researches.

The research presented here suggests the importance

of microtopography in controlling primary production

and respiration in the Arctic. The development of ice-

wedge polygons and the formation of ponds, which are

sites of high NDVI (Fig. 6), increase the productivity of

these old successional stages. The strong correlation

between NDVI and NDWI suggests that the most

productive species occur in very wet areas, and these

species appear to occur in young vegetated drained lake

basins and in the low-center ponds of old vegetated

drained lake basins. Considering the importance of

water availability for plant community establishment

and carbon dynamics, permafrost degradation and

change in the hydrology in the Arctic could strongly

affect species composition, productivity, and net carbon

sequestration of the tundra.

The structure and function of the vegetation in the

drained lake basins are very susceptible to thermokarst

(the melting of ground ice in permafrost), because it

affects polygon structure. Shrinking areal extent of

ponds has been observed in areas of discontinuous

permafrost as a consequence of local thawing and

increased active layer depth, which allow increased

drainage (Yoshikawa & Hinzman, 2003; Smith et al.,

2005). Future changes in climate will, without a doubt,

have an effect, especially considering a potential in-

crease in evapotranspiration (and net decrease in soil

water content and/or water table height) in a largely

warming and drying climate (Oechel et al., 2000; Serreze

et al., 2000; Hinzman et al., 2005). In particular, the

increased active layer depth could eliminate ice-wedges

(Billings & Peterson, 1980; Bliss & Peterson, 1992) and

impact the surface drainage structure, leading to the

conversion of some wetlands to upland vegetation

types.

The comparison of the carbon fluxes of these vege-

tated drained lake basins, as well as long-term contin-

uous measurements in other tundra sites, will provide a

more accurate assessment of current Arctic tundra

carbon balance. In turn, it will also give insight into

the potential impact of drying of the Arctic on plant

communities and carbon fluxes across the heteroge-

neous landscape of the wet coastal tundra.

Acknowledgements

Funding for this study was provided by the Departmentof Energy, USA (DOE, DE-FC02-06ER64159) and from NSF(OPP-0436177). Logistic support was provided by Glenn Shee-han, BASC. We would like to thank the SDSU field crew and allthe BASC and UIC workers that helped to deploy and redeploythe multiple eddy covariance towers. This work would have notbeen possible without them. We would also like to thank CraigTweedie and Adrian Aguirre for the GPS support and KennethHinkel for proving the Landsat image and for the very helpfulsuggestion on the paper, Rommel Zulueta, Steven Hastings,Joseph Verfaillie, John Kochendorfer, George Scheer, andMichael Whiting for support during the campaign, and JeromeBraun of the Statistics Laboratory of UC Davis for the statisticsupport. We thank Ellen Abrams for the grammar and stylereview. This manuscript benefited from the input of fouranonymous reviewers.

References

Baldocchi DD (2003) Assessing the eddy covariance technique for eval-

uating carbon dioxide exchange rates of ecosystems: past, present and

future. Global Change Biology, 9, 479–492.

Barnes JD, Balaguer L, Manrique E, Elvira S, Davison AW (1992) A

reappraisal of the use of DMSO for the extraction and determination

of chlorophylls a and b in lichens and higher plants. Environmental and

Experimental Botany, 2, 85–100.

Barton FI, Himmelsbach D, Duckworth J, Smith M (1992) Two-dimen-

sional vibration spectroscopy: correlation of mid- and near-infrared

regions. Applied Spectroscopy, 46, 420–429.

Billings WD, Luken JO, Mortesen DA, Peterson KM (1983) Increasing

atmospheric carbon dioxide: possible effect on arctic tundra. Oecologia,

58, 286–289.

Billings WD, Peterson KM (1980) Vegetational change and ice-wedge

polygons through the thaw-lake cycle in arctic Alaska. Arctic and Alpine

Research, 12, 413–432.

Bliss LC, Peterson KM (1992) Plant succession, competition, and the

physiological constraints of species in the Arctic. In: Arctic Ecosystems

in a Changing Climate: An Ecophysiological Perspective (eds Chapin FS III,

Jefferies RL, Reynolds JF, Shaver GR, Svoboda J), pp. 111–133.

Academic Press, Orlando, FL.

Bockheim JG, Hinkel KM (2007) The importance of ‘‘deep’’ organic carbon

in permafrost-affected soils of arctic Alaska. Soil Science Society of

America Journal, 71, 1889–1892.

Bockheim JG, Hinkel KM, Eisner WR, Dai XY (2004) Carbon pools and

accumulation rates in an age-series of soils in drained thaw-lake basins,

Arctic Alaska. Soil Science Society of America, 68, 697–704.

1880 D . Z O N A et al.


Britton ME (1957) Vegetation of the arctic tundra. In: Arctic Biology, 2nd

edn (ed. Hansen HP), pp. 26–72. Oregon State University Press,

Corvallis.

Carson CE, Hussey KM (1962) The oriented lakes of arctic Alaska. Journal

of Geology, 70, 417–439.

Coyne PI, Kelley JJ (1974) Carbon dioxide partial pressures in Arctic

surface waters. Limnology and Oceanography, 19, 928–938.

Curran P (1989) Remote sensing of foliar chemistry. Remote Sensing of

Environment, 30, 271–278.

Desjardins RL, Lemon ER (1974) Limitations of an eddy-correlation

technique for the determination of the carbon dioxide and sensible

heat fluxes. Boundary-Layer Meteorology, 5, 475–488.

Detto M, Montaldo N, Albertson JD, Mancini M, Katul GG (2006) Soil

moisture and vegetation controls on evapotranspiration in a hetero-

geneous Mediterranean ecosystem on Sardinia, Italy. Water Resources

Research, 42, W08419, doi: 10.1029/2005WR004693.

Falge E, Baldocchi DD, Olson R et al. (2001) Gap filling strategies for

defensible annual sums of net ecosystem exchange. Agricultural and

Forest Meteorology, 107, 43–69.

Funk DW, Pullman ER, Peterson KM, Crill PM, Billings WD (1994)

Influence of water table on carbon dioxide, carbon monoxide, and

methane fluxes from taiga bog microcosms. Global Biogeochemical

Cycles, 8, 271–278.

Gao BC (1996) NDWI — a normalized difference water index for remote

sensing of vegetation liquid water from space. Remote Sensing of

Environment, 58, 257–266.

Ham JM, Heilman JL (2003) Experimental test of density and energy-

balance corrections on carbon dioxide flux as measured using open-

path eddy covariance. Agronomy Journal, 95, 1393–1403.

Hinkel KM, Eisner WR, Bockheim JG, Nelson FE, Peterson KM, Dai X

(2003) Spatial extent, age, and carbon stocks in drained thaw lake

basins on the Barrow Peninsula, Alaska. Arctic, Antarctic, and Alpine

Research, 35, 291–300.

Hinkel KM, Frohn RC, Nelson FE, Eisner WR, Beck RA (2005) Morpho-

metric and spatial analysis of thaw lakes and drained thaw lake basins

in the western Arctic Coastal Plain, Alaska. Permafrost and Periglacial

Processes, 16, 327–341.

Hinkel KM, Jones BM, Eisner WR, Cuomo CJ, Beck RA, Frohn R (2007)

Methods to assess natural and anthropogenic thaw lake drainage on

the western Arctic Coastal Plain of northern Alaska. Journal of Geophy-

sical Research-Earth Surface, 112, F02S16, doi: 10.1029/2006JF000584.

Hinzman LD, Bettez N, Chapin FS et al. (2005) Evidence and implications

of recent climatic change in terrestrial regions of the Arctic. Climatic

Change, 72, 251–298.

Hinzman LD, Kane DL, Benson CS, Everett KR (1991) Hydrologic and

thermal properties of the active layer in the Alaskan Arctic. Cold Region

Science and Technology, 19, 95–110.

Hope AS, Kimball JS, Stow DA (1993) The relationship between

tussock tundra spectral reflectance properties and biomass and vegeta-

tion composition. International Journal of Remote Sensing, 14, 1861–

1874.

Hopkins DM (1949) Thaw lakes and thaw sinks in the Imuruk Lake area,

Seward Peninsula. Journal of Geology, 57, 119–131.

Hsieh CI, Katul G, Chi T (2000) An approximate analytical model for

footprint estimation of scalar fluxes in thermally stratified atmospheric

flows. Advances in Water Resources, 23, 765–772.

Hussey KM, Michelson RW (1966) Tundra relief features near Point

Barrow, Alaska. Arctic, 19, 162–184.

Kaimal JC, Wyngard JC, Izumi Y, Cote OR (1972) Spectral characteristics

of surface-layer turbulence. Quarterly Journal of Royal Meteorological

Society, 98, 563–589.

Kling GW, Kipphut GW, Miller MC (1991) Arctic lakes and streams as gas

conduits to the atmosphere: implications for tundra carbon budgets.

Science, 251, 298–301.

Kljun N, Calanca P, Rotach MW, Schmid HP (2004) A simple parameter-

isation for flux footprint predictions. Boundary-Layer Meteorology, 112,

503–523.

Kummerow J, Mills JN, Ellis BA, Hastings SJ, Kummerow A (1987)

Downslope fertilizer movement in arctic tussock tundra. Holarctic

Ecology, 10, 312–319.

Kwon H-J, Oechel WC, Zulueta RC, Hastings SJ (2006) Effects of climate

variability on carbon sequestration among adjacent wet sedge tundra

and moist tussock tundra ecosystems. Journal of Geophysical Research,

111, G03014, doi: 10.1029/2005JG000036.

Lafleur PM, Roulet NT, Bubier JL, Frolking S, Moore TR (2003) Inter-

annual variability in the peatland-atmosphere carbon dioxide exchange

at an ombrotrophic bog. Global Biogeochemical Cycles, 17, 1036, 1–14.

Leppala M, Kukko-oja K, Laine J, Tuittila ES (2008) Seasonal dynamics of

CO2 exchange during primary succession of boreal mires as controlled

by phenology of plants. Ecoscience, 15, 460–471.

Luo HY, Oechel WC, Hastings SJ, Zulueta RC, Qian Y, Kwon HJ (2007)

Mature semiarid chaparral ecosystems can be a significant sink for

atmospheric carbon dioxide. Global Change Biology, 13, 386–396.

Luyssaert S, Schulze E-D, Bo+rner A et al. (2008) Old-growth forests as

global carbon sinks. Nature, 455, 213–215.

Mackay JR (1956) Notes on oriented lakes of the Liverpool Bay area,

Northwest Territories. Revue Canadienne de Geographie, 10, 169–173.

Mackay JR (1988) Catastrophic lake drainage. Tuktoyaktuk Peninsula area,

District of Mackenzie. Geological Survey of Canada, Paper, 88-1D,

83–90.

Mackay JR, Burn CR (2002) The first 20 years (1978–1979 to 1998–1999) of

ice-wedge growth at the Illisarvik experimental drained lake site,

western Arctic coast, Canada. Canadian Journal of Earth Sciences, 39,

95–111.

Manley WF, Lestak LR, Tweedie CE, Maslanik JA (2004) High-Resolution

Quickbird Imagery and Value-Added Geospatial Layers for Collaborative

Research of Environmental Change at Barrow, Alaska. National Snow and

Ice Data Center. Digital media, Boulder, CO.

Martin ME, Aber JD (1997) High spectral resolution remote sensing of

forest canopy lignin, nitrogen, and ecosystem processes. Ecological

Applications, 7, 431–443.

Matson P, Johnson L, Billow C, Miller J, Pu R (1994) Seasonal patterns and

remote spectral estimation of canopy chemistry across the Oregon

transect. Ecological Applications, 4, 280–298.

Moore CJ (1986) Frequency response corrections for eddy correlation

system. Boundary Layer Meteorology, 37, 17–35.

Nagler PL, Daughtry CST, Goward SN (2000) Plant litter and soil

reflectance. Remote Sensing of Environment, 71, 207–215.

Odum EP (1969) The strategy of ecosystem development. Science, 164,

262–270.

Oechel WC, Vourlitis GL, Hastings SJ, Ault RP Jr, Bryant P (1998) The

effect of water table manipulation and elevated temperature on the net

CO2 flux of wet sedge tundra ecosystem. Global Change Biology, 4, 77–90.

Oechel WC, Vourlitis GL, Hastings SJ, Zulueta RC, Hinzman L, Kane D

(2000) Acclimation of ecosystem CO2 exchange in the Alaskan Arctic in

response to decadal climate warming. Nature, 406, 978–981.

Pavlov AV (1975) Heat transfer of the soil and atmosphere at northern and

temperate latitudes. U.S. Army Cold Regions Research and Engineering

Laboratory Translation, 511, 298.

Paw U KT, Baldocchi DD, Meyers TP, Wilson K (2000) Correction of eddy-

covariance measurements incorporating both advective effects and

density fluxes. Boundary-Layer Meteorology, 97, 487–511.

Paw U KT, Falk M, Suchanek TH et al. (2004) Carbon dioxide exchange

between an old-growth forest and the atmosphere. Ecosystems, 7,

513–524.

Penuelas J, Pinol J, Ogaya R, Filella I (1997) Estimation of plant water

concentration by the reflectance Water Index WI (R900/R970). Interna-

tional Journal of Remote Sensing, 18, 2869–2875.



10.1029/2005WR004693

10.1029/2005JG000036

Rex RW (1961) Hydrodynamic analysis of circulation and orientation of

lakes in Northern Alaska. In: Geology of the Arctic (ed. Rauch GO),

pp. 1021–1043. University of Toronto Press, Toronto.

Serrano L, Penuelas J, Ustin SL (2002) Remote sensing of nitrogen and

lignin in Mediterranean vegetation from AVIRIS data: decomposing

biochemical from structural signals. Remote Sensing of Environment, 81,

355–364.

Serreze MC, Walsh JE, Chapin FS et al. (2000) Observational evidence of

recent change in the northern high-latitude environment. Climatic

Change, 46, 159–207.

Shaver GR, Billings WD (1975) Root production and root turnover in a

Wet Tundra Ecosystem, Barrow, Alaska. Ecology, 56, 401–409.

Smith LC, Sheng Y, MacDonald GM, Hinzman LD (2005) Disappearing

Arctic Lakes. Science, 308, 1429.

Smith ML, Martin ME, Plourde L, Ollinger SV (2003) Analysis of hyper-

spectral data for estimation of temperate Forest Canopy nitrogen

concentration: comparison between an Airborne (AVIRIS) and a Space-

borne (Hyperion) sensor. IEEE Transactions on Geoscience and Remote

Sensing, 41, 1332–1337.

Smith ML, Ollinger SV, Martin ME, Aber JD, Hallett RA, Goodale CL

(2002) Direct estimation of aboveground forest productivity through

hyperspectral remote sensing of canopy nitrogen. Ecological Applica-

tions, 12, 1286–1302.

Stow D, Hope A, Boynton W, Phinn S, Walker D (1998) Satellite-derived

vegetation index and cover type maps for estimating carbon dioxide

flux for arctic tundra regions. Geomorphology, 21, 313–327.

Swinbank WC (1951) The measurement of vertical transfer of heat and

water vapour by eddies in the lower atmosphere. Journal of Meteorology,

72, 135–132.

Tucker CJ (1979) Red and photographic infrared linear combinations for

monitoring vegetation. Remote Sensing of Environment, 8, 127–150.

Twine TE, Kustas WP, Normanc JM et al. (2000) Correcting eddy-covar-

iance flux underestimates over a grassland. Agricultural and Forest

Meteorology, 103, 279–300.

Webb EK, Pearman GI, Leuning R (1980) Correction of flux measurements

for density effects due to heat and water vapour transfer. Quarterly

Journal of the Royal Meteorological Society, 106, 85–100.

Wilson K, Goldstein A, Falge E et al. (2002) Energy balance closure at

FLUXNET sites. Agricultural and Forest Meteorology, 113, 223–243.

Yoshikawa K, Hinzman LD (2003) Shrinking thermokarst ponds and

groundwater dynamics in discontinuous permafrost near Council,

Alaska. Permafrost and Periglacial Processes, 14, 151–169.

1882 D . Z O N A et al.


Characterization of the carbon fluxes of a vegetated drained lake basin chronosequence on the Alaskan Arctic Coastal Plain

Documents