Full text of DOI:10.1016/S0038-0717(01)00058-X

Soil Biology & Biochemistry 33 (2001) 1419±1423
www.elsevier.com/locate/soilbio
Methane ¯ux in grassy marshlands near Kolyma River,
north-eastern Siberia
a,
b
Shiro Tsuyuzaki , Tomoko Nakano , Shun-ichi Kuniyoshi , Masami Fukuda
c
d
*
aGraduate School of Environmental Earth Science, Hokkaido University, Sapporo 060 0810 Japan
bFaculty of Science, Tokyo Metropolitan University, Hachioji 192 0364, Tokyo, Japan
cFaculty of Agriculture, Hokkaido University, Sapporo 060 0810 Japan
dInstitute of Low Temperature Science, Hokkaido University, Sapporo 060 0810 Japan
Received 28 January 2000; received in revised form 30 January 2001; accepted 12 February 2001
Abstract
The relationships between vegetation types and methane ¯ux were evaluated in north-eastern Siberia in the summer of 1995. Based on 71
50 £ 50-cm plots, the vegetation was divided into three types by TWINSPAN cluster analysis: (1) Eriophorum grassland where species
richness was high and moss well-established (Eriophorum grassland), (2) horsetail grassland with moss cover and (3) Carex grassland
without moss. Vascular plant cover in all vegetation types was less than 50%. The methane ¯ux was measured by a static chamber method.
Horsetail grassland emitted the highest amount of methane, 165.5 mg CH4/m2/d, while diverse-sedge grassland emitted less methane, i.e.,
21.9 mg CH4/m2/d. Methane ¯ux positively increased with electric conductivity (EC), water depth, thaw depth, and soil humidity. Of those,
EC was strongly correlated to methane emission (Spearman's rank correlation, r2 ˆ 0.630). The difference in methane ¯ux between horsetail
grassland and sedge-dominated grassland suggested that the morphological characteristics of vascular plants were important on methane
transport, probably because vascular plants, such as horsetail, that developed aerenchyma and intercellular gas space promoted methane
transport from soil to air while mosses do not. q 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Grassy marshland; Methane ¯ux; Siberia; Vegetation type
1. Introduction
wetlands dominated by non-crop species, such as sedges and
grasses (Svensson and Rosswall, 1984; Dise, 1993; Bubier et
al., 1995). Those suggested that the methane ¯ux differed
between plant community types even in the same wetland.
Various determinants on methane ¯ux have been proposed,
e.g., soil temperature (Crill et al., 1988; Morrissey and Living-
ston, 1992), thaw depth (Whalen and Reeburgh, 1992), and
water table, level or position (Dise et al., 1993), and peat and
vegetation types (Dise, 1993). As well as methane ¯ux, the
development patterns of wetlands, such as grassy marshland
and reed swamp, are often determined by the gradients of
multiple environmental factors, including soil pH, water
depth and nutrients (Tsuyuzaki et al., 1990; Tsuyuzaki,
1997). We predict that methane ¯ux differs greatly among
plant communities, and is partly explained by the morpholo-
gical characteristics of dominant plants.
Methane, H2O, CO2, N2O, and O3 are greenhouse gases
(Wang et al., 1976; Zimov et al., 1993), and wetlands have
been identi®ed as a major source of methane (Harriss et al.,
1982; Whalen and Reeburgh, 1992). Tundra occupies 7% of
the land area on the earth and retains about 13% of stocked soil
carbon (Post et al., 1982). Those facts indicate that Siberian
tundra wetlands are a major global source of atmospheric
methane, although there have been few surveys conducted.
Methane is transported from soil to air by diffusion, ebulli-
tion, and conduct through plant stems (Cicerone and Shetter,
1981; Noguchi et al., 1990). In particular, gases, including
methane, are transported well from the rhizosphere to the
atmosphere through aerenchyma and intercellular gas spaces
in emergent aquatic plants (Teal and Kanwisher, 1966; Nogu-
chi et al., 1990). In rice-cultivated ®elds, most methane is
emitted by plant transport (Cicerone and Shetter, 1981;
Holzapfel-Pschorn and Seiler, 1986) and the same is true of
2. Materials and methods
2.1. Study area
* Corresponding author. Tel.: 181-11-706-2283; fax: 181-11-706-4954.
E-mail address: tsuyu@ees.hokudai.ac.jp (S. Tsuyuzaki).
The study site is located at the lower Kolyma River in
0038-0717/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S0038-0717(01)00058-X
-

1420
S. Tsuyuzaki et al. / Soil Biology & Biochemistry 33 (2001) 1419±1423
Chersky, northern Siberia (68.58N, 161.48E), ca 100 km
from the East Siberian Sea (Zimov et al., 1993). This region
is underlain by permafrost, and consists of vegetation
mosaics of taiga, shrub, and herbaceous communities.
Mean annual temperature was 212.48C, and annual preci-
pitation was 174 mm in 1994 (Nakano et al., internal
report).
types, mean differences in environmental factors and vege-
tation characteristics among the vegetation types were
compared by Tukey's HSD tests after ANOVA was
conducted (Zar, 1984).
3. Results
The measurements were conducted from 7th July to 29th
August 1995, in a ¯oodplain of the Kolyma River where
there are waterlogged wetlands, grasslands, and shrublands.
The soils are gley soils. The thickness of organic horizon
ranged from 10 to 40 cm.
Sedges and grasses dominated most plots, with low cover
of forbs (Table 1). From the TWINSPAN cluster analysis,
the vegetation was primarily divided into two types based
on the frequency and dominance of Carex sp. The vegeta-
tion where Carex sp was less frequent was further divided
into two types, based on the presence/absence of Equisetum
limosum.
2.2. Samplings and measurements
In total, 71 50 £ 50 cm plots were set up in ®ve selected
sites on the ¯oodplain. In each plot, the cover of each vascu-
lar plant taxa was visually estimated by 25 10 £ 10 cm
divided frame. Total moss cover, number of tussocks, and
coverage of tussocks, elevation difference within a plot, and
litter thickness were also recorded in each plot.
The vegetation characteristics of three groups 1±3
recognized by TWINSPAN are as follows (Table 1).
Group 1 was dominated by a sedge Eriophorum coreanum,
a grass Calamogrostis purpurascens, and a forb Sangui-
sorba of®cinalis. Total cover of grasses and sedges was ca
31% and E. coreanum showed the highest frequency (79%)
and cover (21%) (hereafter, i.e., Eriophorum grassland).
Moss was also well-established, while a few sedge tussocks
were observed. This group showed high species richness.
Group 2 was dominated by Equisetum limosum and Carex
wiluica. E. limosum established only in this group and
showed the highest frequency (89%) (horsetail grassland).
In this group, tussocks were well-developed, and the domi-
nant sedge was the same found in Group 1. Moss cover was
less than 1/2 of Group 1. Group 3 was a sedge wetland
dominated by Carex sp (Carex grassland), and showed the
lowest plot cover and species richness, i.e., 1.3 species/plot.
Litter accumulation was the deepest (20.3 cm), and moss
did not establish in this group.
In the surveyed plots, methane ¯uxes varied from 29.3 to
229.2 mg CH4/m2/d. The mean rate of methane emission
was signi®cantly highest, 165.5 ^ 6.8 (SE) mg CH4/m2/d,
in the horsetail grassland (P , 0.01) (Table 2), while
methane oxidation often occurred in Eriophorum grassland.
Carex dominated grassland emitted less than half of the
methane (73.2 ^ 4.9 mg CH4/m2/d) emitted from the horse-
tail grassland (P , 0.01).
Ground water did not appear above the ground surface in
Eriophorum grassland, while the other two vegetation types
were waterlogged (Table 2). Soil moisture was also the
lowest in Eriophorum grassland. Topographic variation
increased from vegetation types 1 to 3, due mostly to the
increase of number of tussocks made by Carex spp. The pH
of surface water was slightly acidic ranging from 5.51 to
6.37, and gradually increased from vegetation types 1 to 3
with tussock developments. The Eh did not signi®cantly
differ among the three vegetation types (Table 2). EC was
signi®cantly different among the three vegetation types with
the horsetail grassland showing the highest value and Erio-
phorum grassland showing the lowest one (P , 0.01). Thaw
depths averaged 59.8 ^ 21.4 cm and ranged from 28.2 to
Net methane ¯ux was measured by a static chamber
method (Whalen and Reeburgh, 1988). An open-bottomed
transparent acrylic resin chamber (30 £ 30 £ 30 cm) was put
directly in the soil or surface water on each plot. The base
was inserted into the ground surface not later than 0.5 h
before the start of measurements to obtain the equilibrium
of the system. In plots where the ground surface was not
waterlogged, sticky clay soil was used as collar between
chamber and soil. Headspace samples were extracted from
the chamber every 10 min over a 30-min period using 60-ml
plastic syringes. The extracted gas was stored in 10-ml
vacuum vials. Methane concentration in each sample was
analyzed by a gas chromatograph with a ¯ame ionization
detector. On each plot, methane ¯ux was measured one
time. During the sampling period, air temperature was
recorded and averaged.
To estimate thaw depth, a 1-cm-diameter stainless steel rod
was inserted up to the frozen soil surface. Water depth was
recordedwhenthemethane¯uxwasmeasured. ThepHandEh
of ground water were measured by a portable pH meter (HM-
14P, TOAElectricCo.,Tokyo), andelectricconductivity(EC)
was measured bya portable EC meter (CM-14P, TOA Electric
Co., Tokyo). Volumetric water content was measured by a
time domain re¯ectometry (TDR) (IMKO Trime System),
which is capable of directly measuring the volume of water
content. Spearman rank correlation tests were used to con®rm
the relationship among physical and chemical environmental
factors (Zar, 1984).
TWINSPAN cluster analysis was conducted to classify
the vegetation types, using percentage cover as dominance
parameter of each vascular plant taxa in 71 surveyed plots
(Hill, 1979). TWINSPAN is one of the standard cluster
analysis for the classi®cation of vegetation types based on
the presence/absence and dominance of each species (Kent
and Coker, 1992). After the classi®cation of vegetation
-

S. Tsuyuzaki et al. / Soil Biology & Biochemistry 33 (2001) 1419±1423
1421
Table 1
Vegetation classi®ed by TWINSPAN. Percentage appearance frequencies of vascular plants on each grassland type (total number of species established in the
grassland type/number of plots classi®ed into the grassland type £ 100) are shown with mean cover on each grassland type shown in parentheses. 1: less than
1% cover. 2: no individual observed. Mean differences between vegetation types are indicated by different letter superscripts (Tukey's HSD test, P , 0.01).
Bold letters indicate pseudospecies that classify the vegetation types
Grassland type
Number of plots
Life form
Eriophorum grassland
19
Horsetail grassland
40
Carex grassland
12
All plots
71
Speciesa
Equisetum limosum
Sanguisorba of®cinalis
Carex wiluica
Fern
±
89 (21)
33 (4)
58 (21)
36 (1)
± (±)
24 (1)
4 (1)
±
±
56 (12)
37 (5)
37 (12)
23 (1)
23 (6)
20 (1)
20 (5)
14 (3)
14
Forb
47 (10)
±
17
Sedge
Forb
±
Epilobium palustre
Eriophorum coreanum
Stellaria sp
±
±
Sedge
Forb
79 (21)
16 (1)
±
8
±
Carex sp
Calamoglostis purpurascens
Moss cover
Sedge
Grass
100 (30)
±
0b
1.9b
35b
20.3b
1.3b
31b
53 (10)
24a
0.3a
4a
9.3a
3.3a
44a
11b
Number of tussocks
Coverage of tussocks
Litter thickness (cm)
Mean species richness
Mean plot cover
2.0b
20a
1.6
17
11.7
10.4a
3.2a
2.9
47a
44
a
Other species with less than 10% frequencies were: Caltha sp. (Frequency ˆ 7%), Ranunculus repens (7), Hippuris vulgaris (7), Menyanthes trifoliata (7),
Cerastium sp (6), Chamaenerium angustifolium (4), Salix spp. (4), Vaccinium vitis-idaea (3), Rubus arcticus (3), Eriophorum vaginatum (3), Equisetum
arvense (1), Festuca rubra (1), Pylora incornata (1), Poa glauca (1), Carex limosa (1), Alnaster fruticosa (1).
Table 2
Mean values with standard error of each environmental factor in each cluster group. The different letters upper-right the mean values indicate signi®cantly
different at P , 0.01 (Tukey's HSD test). Except for the measurement of pH, EC, Eh and thaw depth in cluster Group 1 where we did not measure in four plots,
sample numbers are coincidence with number of plots clustered by TWINSPAN (Table 1). On water depth, minus value indicates the depth of soil where the
ground water appeared
Grassland type
Eriophorum grassland
Horsetail grassland
Carex grassland
All plots
Water depth (cm)
PH
2 5.1 ^ 0.8a
5.75 ^ 0.04a
18.8 ^ 0.9
5.2 ^ 0.3b
6.15 ^ 0.03b
17.0 ^ 1.0
8.86 ^ 0.30b
455.7 ^ 2.0a
63.8 ^ 1.4b
69.4 ^ 3.2b
10.1 ^ 0.4a
165.5 ^ 6.8b
5.4 ^ 0.5b
6.25 ^ 0.01c
17.4 ^ 0.3
6.61 ^ 0.13c
454.9 ^ 1.2a
55.5 ^ 1.0a
59.4 ^ 0.4b
14.7 ^ 0.9b
73.2 ^ 4.9c
2.90 ^ 0.6
6.08 ^ 0.03
17.5 ^ 0.6
7.64 ^ 0.26
454.6 ^ 2.2
58.6 ^ 1.2
59.8 ^ 2.5
10.2 ^ 0.4
105.1 ^ 9.5
Air temperature (8C)
EC (mS/m)
Eh (mV)
5.20 ^ 0.29a
451.4 ^ 7.8a
49.8 ^ 2.3a
39.7 ^ 0.8a
7.6 ^ 0.8a
Soil humidity (%)
Thaw depth (cm)
Elevation difference (cm)
Methane ¯ux (mg CH4/m2/d)
2 1.9 ^ 1.6a
98.5 cm in all the plots, and were the lowest in Eriophorum
grassland. During the surveyed period, air temperature
ranged from 8 to 248C.
4. Discussion
In the Siberian wetland, methane ¯ux was correlated to
pH, EC, water depth, thaw depth, and soil humidity. Multi-
ple environmental factors, such as soil temperature (Crill et
al., 1988; Morrissey and Livingston, 1992), thaw depth
(Whalen and Reeburgh, 1992), and water table, level or
position and peat and vegetation types (Dise et al., 1993)
seem to explain methane ¯ux variability (Bubier et al.,
1995). Wetland vegetation patterns are also determined by
multiple environmental factors (Yabe, 1984; ter Braak and
Wiertz, 1994; Tsuyuzaki et al., 1990; Tsuyuzaki 1997). To
explain the difference of methane ¯ux, therefore, not only
chemical and physical factors but also biological factors
Methane ¯ux was positively correlated with pH, EC,
water depth, thaw depth, and soil humidity (r . 10.380,
signi®cant at P , 0.01) (Table 3). Of the ®ve signi®cant
correlations, the relationship between methane ¯ux and
EC was the highest, i.e., 10.794. Thaw depth was related
to EC, Eh, water depth, and soil humidity. The factors which
were correlated to methane ¯ux were roughly related to each
other. Of those signi®cant factors, EC was most strongly
related to methane ¯ux in the surveyed tundra wetland.
Air temperature was correlated to pH, Eh, and thaw depth,
but not to methane ¯ux.
-

1422
S. Tsuyuzaki et al. / Soil Biology & Biochemistry 33 (2001) 1419±1423
Table 3
Matrix of methane ¯ux and other related environmental factors, obtained by Spearman rank correlation coef®cients
Temperature
pH
EC
Eh
Water depth
Thaw depth
Soil humidity
Methane ¯ux
Temperature
PH
10.071
10.546a
10.452a
10.794a
10.071
10.621*
10.237
20.515a
20.029
10.335a
10.538a
10.161
10.682a
10.421a
20.008
10.554a
20.357a
10.487a
10.556a
10.348a
10.233
10.380a
20.780a
20.416a
10.331a
10.331a
10.210
10.396a
EC
Eh
Water depth
Thaw depth
a
Signi®cant at P , 0.01; n ˆ 67.
should be mentioned. We did not measure soil temperature,
although the temperature in¯uence methane ¯ux (Morrissey
and Livingston, 1992; Alford et al., 1997). However, air
temperature, water depth and thaw depth, all of which
were likely to be related to soil temperature, did not differ
between horsetail grassland and Carex grassland.
resistance to the passage of methane, and all methane was
virtually released on the lower 15 cm of the plants as they
emerged from the water (Kelker and Chanton, 1997). There-
fore, vegetation where sedges and grasses are dominant
should show high methane emission. As species in genus
Equisetum develop considerable medullary cavities and
tough root systems (Hauke, 1978), it could transport gases
more effectively than grasses and sedges. Therefore, the
horsetail grassland showed the highest methane emission,
although moss was common. The species composition of
vegetation appeared to greatly in¯uence methane ¯ux,
perhaps due to plant aerenchyma and intercellular gas
spaces, and may have a major role on methane transport
from soil to air. There are differences in methane emissions
between plant communities derived from differences in
substrate or in plant productivity on the global scale (Whit-
ing and Chanton, 1993), while the present study was
conducted in a marshland with rather narrow scale and
those factors may weakly have affected the methane ¯ux.
Eriophorum grassland with the highest moss cover
showed the lowest methane emission, i.e., nearly zero,
while horsetail grassland of which moss cover was a half
of Eriophorum grassland obtained the highest methane
emission. In total, methane emission was low when moss
was dominant. For examples, in Alaskan tundra, methane
emissions from soils averaged 0.5±1.7 mg CH4/m2/d from
mossy areas while 8.7±41.4 mg CH4/m2/d from vegetation
dominated by Eriophorum vaginatum tussocks (Whalen and
Reeburgh, 1992). On the subarctic mire in Sweden, methane
emission varied from 1.1 mg CH4/m2/d at dry moss
hummocks to 360.3 mg CH4/m2/d at wet habitats with Erio-
phorum angustifolium and Sphagnum lindbergii (Svensson
and Rosswall, 1984). Methane emissions ranged from
10 mg CH4/m2/d at a forested bog hummock to 180 mg
CH4/m2/d at an open poor fen in Minnesota peatlands
(Dise, 1992, 1993). Those studies also suggest that soil
dryness restrict methane emission. In another study,
methane was consumed when soils were drying in mountain
meadows (Delgado et al., 1996). In this study, a lowered
water table in the Eriophorum grassland probably resulted
in lowered methane emissions due to oxidation of methane
in the aerobic part of the soil. In conclusion, Eriophorum
grassland obtained the lowest methane emission due to moss
cover and soil dryness. However, Carex grassland where
moss did not establish and water depth did not differ from
horsetail grassland showed lower methane emission
(73.2 mg CH4/m2/d) than horsetail grassland (165.5 mg
CH4/m2/d) of which moss cover was 11%.
Acknowledgements
We express our sincere thanks to S. Zimov, G. Zimova,
and all members concerning with this project (the joint
Siberian permafrost studies between Japan and Russia) for
their invaluable help, and to N. Tanaka and Y.W. Kim for
their help on the measurement of methane concentration.
This work is partly supported by Ministry of Education,
Culture, and Science of Japan.
References
Alford, D.P., Delaune, R.D., Lindau, C.W., 1997. Methane ¯ux from
Mississippi River deltaic plain wetlands. Biogeochemistry (Dordrecht)
37, 227±236.
Aquatic plants developing aerenchyma and intercellular
gas space, including wetland sedges and grasses, have a
great role in gas transport from rhizosphere to the atmo-
sphere (Teal and Kanwisher, 1966; Noguchi et al., 1990;
Neue et al., 1997). Methane emitted to the atmosphere is
transported primarily through grass and secondary by
ebullition and diffusion (Cicerone and Shetter, 1981). The
above-ground portions of Carex offered only slight
Bubier, J.L., Moore, T.R., Bellisario, L., Comer, N.T., 1995. Ecological
controls on methane emissions from a northern peatland complex in the
zone of discontinuous permafrost, Manitoba, Canada. Global Bio-
geochemical Cycles 9, 455±470.
Cicerone, R.J., Shetter, J.D., 1981. Sources of atmospheric methane:
measurements in rice paddies and a discussion. Journal of Geophysical
Research 86C, 7203±7209.
Crill, P.M., Bartlett, K.B., Harriss, R.C., Gorham, E., Verry, E.S., Sebacher,
-

S. Tsuyuzaki et al. / Soil Biology & Biochemistry 33 (2001) 1419±1423
1423
D.I., Mazdar, L., Sanner, W., 1988. Methane ¯ux from Minnesota peat-
lands. Global Biogeochemical Cycles 2, 371±384.
from the rhizosphere to the atmosphere through rice plants. Plant
Physiology 94, 59±66.
Delgado, J.A., Mosier, A.R., Follett, R.H., Follett, R.F., Westfall, D.G.,
Klemedtsson, L.K., Vermeulen, J., 1996. Effects of N management
on N20 and CH4 ¯uxes and 15N. Nutrient Cycling in Agroecosystems
46, 127±134.
Post, W.M., Emanuel, W.R., Zinke, P.J., Stangenberger, A.G., 1982. Soil
carbon pools and world life zones. Nature 298, 156±159.
Svensson, B.H., Rosswall, T., 1984. In situ methane production from acid
peat in plant communities with different moisture regimes in a subarctic
mire. Oikos 43, 341±350.
Dise, N.B., 1992. Winter ¯uxes of methane from Minnesota peatlands.
Biogeochemistry 17, 71±83.
Teal, J.M., Kanwisher, J.W., 1966. Gas transport in the marsh grass,
Saprtina allerni¯ora. Journal of Experimental Botany 17, 356±361.
ter Braak, C.J.F., Wiertz, J., 1994. On the statistical analysis of vegetation
change: a wetland affected by water extraction and soil acidi®cation.
Journal of Vegetation Science 5, 361±372.
Dise, N.B., 1993. Methane emission from Minnesota peatlands: Spatial and
seasonal variability. Global Biogeochemical Cycles 7, 123±142.
Dise, N.B., Gorham, E., Verry, E.S., 1993. Environmental factors control-
ling methane emissions from peatlands in northern Minnesota. Journal
of Geophysical Research 98D, 10583±10594.
Tsuyuzaki, S., 1997. Wetland development in the early stages of volcanic
succession. Journal of Vegetation Science 8, 353±360.
Tsuyuzaki, S., Urano, S., Tujii, T., 1990. Vegetation of alpine marshland
and its neighboring areas, northern part of Sichuan Province, China.
Vegetation 88, 79±86.
Harriss, R.C., Sebacher, D.I., Day Jr, F.P., 1982. Methane ¯ux in the Great
Dismal Swamp. Nature 297, 673±674.
Hauke, R.L., 1978. A taxonomic monograph of Equisetum subgenus Equi-
setum. Nova Hedwigia 30, 385±455.
Holzapfel-Pschorn, A., Seiler, W., 1986. Methane emission during a culti-
vation period from an Italian rice paddy. Journal of Geophysical
Research 91D, 11803±11814.
Wang, W.C., Yung, T.L., Lacis, A.A., Mo, J.E., Hansen, J.E., 1976. Green-
house effects due to man-made perturbations of trace gases. Science
194, 685±690.
Hill, M.O., 1979. TWINSPAN: a FORTRAN program for arranging
multivariate data in an ordered two-way table by classi®cation of the
individuals and attributes. Department of Ecology and Systematics,
Cornell University, Ithaca, New York.
Whalen, S.C., Reeburgh, W.S., 1988. A methane ¯ux time series for tundra
environments. Global Biogeochemical Cycles 2, 399±409.
Whalen, S.C., Reeburgh, W.S., 1992. Interannual variations in tundra
methane emission: A 4-year time series at ®xed sites. Global Biogeo-
chemichal Cycles 5, 261±273.
Kelker, D., Chanton, J., 1997. The effect of clipping on methane emissions
from Carex. Biogeochemistry (Dordrecht) 39, 37±44.
Whiting, G.J., Chanton, G.P., 1993. Primary production control of methane
emission from wetlands. Nature 364, 794±795.
Kent, M., Coker, P., 1992. Vegetation Description and Analysis. A
Practical Approach. CRC Press, Boca Raton.
Yabe, K., 1984. Ecological studies of the Mobara-Yatsumi marsh. Main
physical and chemical factors controlling the marsh ecosystem.
Japanese Journal of Ecology 34, 173±186.
Morrissey, L.A., Livingston, G.P., 1992. Methane emissions from Alaska
Arctic tundra: An assessment of local spatial variability. Journal of
Geophysical Research 97D, 16661±16670.
Zar, J.M., 1984. Biostatistical Analysis. 2nd ednPrentice-Hall, Englewood
Cliffs, NJ.
Neue, H.U., Wassmann, R., Kludze, H.K., Bujun, W., Lantin, R.S., 1997.
Factors and processes controlling methane emissions from rice ®elds.
Nutrient Cycling in Agroecosystems 49, 111±117.
Zimov, S.A., Semiletov, I.P., Daviodov, S.P., Voropaev, Yu.V., Pro-
syannikov, S.F., Wong, C.S., Chan, Y.-H., 1993. Wintertime CO2
emission from soils of north-eastern Siberia. Arctic 46, 197±204.
Noguchi, I., Mariko, S., Aoki, K., 1990. Mechanism of methane transport
-