United States
Environmental Protection
Agency
Office of Research and
Development
Washington, DC 20460
May 1993
Carbon Cycling in
Boreal Forests and
Sub-Arctic Ecosystems
-------
-------
EPA/600/R-93/084
May 1993
Proceedings
of the
International Workshop
on
CARBON CYCLING IN BOREAL FOREST
AND SUB-ARCTIC ECOSYSTEMS:
Biospheric Responses and Feedbacks to Global Climate Change
Edited by
Ted S. Vinson and Tatyana P. Kolchugina
Department of Civil Engineering • Oregon State University
Corvallis, Oregon
Sponsored by
U.S. Environmental Protection Agency
Global Change Research Program
Robert K. Dixon • Program Leader
Environmental Research Laboratory
Corvallis, Oregon
September 1991
Printed on Recycled Paper
-------
-------
FOREWORD
Global climate is projected to change rapidly during the next half century as a result of alterations
in the chemical composition of the atmosphere. Further, climate change may be more pronounced
in the northern hemisphere. If the projected changes in global climate occur, the impact on
terrestrial ecosystem processes is expected to be substantial.
Considerable uncertainty exists regarding responses and feedbacks of tundra, peat lands and
boreal forest ecosystems to global climate change. Of particular interest is the role of tundra, peat
lands and boreal forests in cycling and sequestering carbon because nearly one-third of the world's
terrestrial carbon may be stored in these ecosystems.
An appreciation of the responses and feedbacks of the carbon cycle in boreal forest and sub-arctic
ecosystems requires an understanding of the carbon pools and fluxes in these ecosystems. The
estimates of current carbon pools and fluxes in the complex of ecosystems that comprise northern
regions may be misleading. This is a result of the fact that the estimates are:
• based upon disparate data that were collected, in part, over the past several decades;
• associated with an uncertain estimate of the aerial extent of major vegetation types;'
• reflective of mainly North American data and very little data from the former Soviet Union,
which has the greatest expanse of boreal forest and sub-arctic ecosystems in the world.
There are also large uncertainties in our ability to project changes in carbon pools and fluxes over
the next several decades.
In recognition of the need to assess the effect of tundra, peat lands and boreal forests on terrestrial
carbon dynamics, an international workshop (with participants from the USA, Canada, the former
Soviet Union and other boreal forest and sub-arctic countries) was convened with the following
objectives:
• identify available tools and methods that may be used to provide extensive, early evaluation
of responses and feedbacks in boreal forest and sub-arctic ecosystems;
• identify available carbon dynamics data and models that may be used to conduct prelim-
inary analyses of carbon cycling and sequestering patterns in boreal forest and sub-arctic
ecosystems and establish carbon budgets for boreal and sub-arctic countries;
• identify the necessary elements of a framework to establish the carbon budget for a boreal
forest and/or sub-arctic country.
The written contributions to the workshop are presented herein.
Ted S. Vinson
Tatyana P. Kolchugina
in
-------
ACKNOWLEDGMENTS
Susan Biskeborn and Daniel Cook served as Production Editors of the Workshop Proceedings.
The Technical Editors gratefully acknowledge their efforts in producing the camera-ready-copy
and their careful proofreading of the manuscripts submitted. The Workshop was funded under the
US EPA Global Climate Research Program, Global Mitigation and Adaptation Program, Robert
A. Dkon, Program Leader.
DISCLAIMER
The information in this document has been funded by the US Environmental Protection Agency
under Agency Agreement CR17682-01 to Oregon State University. It has been subjected to the
Agency's peer and administrative review, and it has been approved for publication as an EPA
document. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
IV
-------
PREFIXES FOR METRIC SYSTEM
peta
tera
giga
mega
kilo
hecto
deca
deci
centi
milli
micro
P
T
G
M
k
h
—
d
c
m
M
IO15
IO12
IO9
IO6
IO3
IO2
10
io-1
io-2
io-3
io-6
COMMON UNITS
1 Teragram (Tg) = IO12 g = IO9 kg = IO61 = 1 Megaton (Mt) = 1 million metric tons
1 Petagram (Pg) = IO15 g = IO12 kg = IO91 = 1 Gigaton (Gt) = 1 billion metric tons
1 kilometer (km) = 1,000 m = 10s cm = IO6 mm
1 hectare (ha) = 0.01 km2 = 10,000 m2
1 km2 = 100 ha = IO6 m2
1 year (yr) = 365 days = 8,760 hours = 525,600 min.
ppm = parts per million
CONVERSION FACTORS
1 ha = 2.48 acres
1 ton (metric) = 1000 kg = 1.1 tons (U.S.) = 2200 pounds (Ib)
1 kg = 2.2 Ib
1 km = 0.621 mile
-------
-------
TABLE OF CONTENTS
WARMING THE NORTH: WHAT HAPPENS?
George M. Woodwell and Richard A. Houghton 1
BOREAL CARBON POOLS: APPROACHES AND CONSTRAINTS
IN GLOBAL EXTRAPOLATIONS
F. Stuart Chapin, III and Elaine Matthews 9
PLANETARY MAXIMUM CO2 AND ECOSYSTEMS OF THE NORTH
Sergey A. Zimov, Sergey P. Daviodov,
Yuri V. Voropaev, and Sergey F. Prosiannikov 21
ARCTIC ATMOSPHERIC CO2 BIMODAL DISTRIBUTION
Igor P. Semiletov, Sergey A. Zimov, Sergey P. Daviodov,
Yuri V. Voropaev, and Sergey F. Prosiannikov 35
MARINE HUMIC ACIDS AS AN IMPORTANT CONSTITUENT OF THE DISSOLVED
ORGANIC CARBON FLUX IN THE BERING AND CHUKCHI SEA ECOSYSTEMS
Irina V. Perminova and Valeriy S. Petrosyan 43
RESIDENCE TIME OF CARBON IN SOILS OF THE BOREAL ZONE
Kira Kobak and Natalia Kondrasheva 51
EQUILIBRIUM ANALYSIS OF PROJECTED CLIMATE CHANGE
EFFECTS ON THE GLOBAL SOIL ORGANIC MATTER POOL
David P. Turner and Rik Leemans 59
DISTRIBUTION AND RENOVATION TIME OF SOIL CARBON
IN BOREAL AND SUBARCTIC ECOSYSTEMS OF EUROPEAN RUSSIA
Alexander E. Cherkinsky and Sergey V. Goryachkin 65
STATE FACTORS, STEADY STATES AND SOIL ORGANIC-
MATTER DYNAMICS AFTER FOREST HARVESTING
Jeffrey G. Borchers 71
CHARACTERISTICS OF FOREST BIOGEOCENOSES
Natalie Pervova 79
DYNAMIC MATHEMATICAL MODEL FOR OXYGEN AND CARBON
DIOXIDE EXCHANGE BETWEEN SOIL AND THE ATMOSPHERE
Larry Boersma and Ying Ouyang 87
vu
-------
TABLE OF CONTENTS
CARBON STORAGE IN PEAT BASED ON REGIONALITY OF RUSSIAN MIRES
MarinaS. Botch 101
METHANE FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
Stephen Frolking 109
METHANE AND CARBON DIOXIDE PRODUCTION
AND UPTAKE IN SOME BOREAL ECOSYSTEMS OF RUSSIA
Nicolay Panikov and Vladimir Zelenev 125
BOREAL FORESTS, THE CARBON CYCLE AND GLOBAL CHANGE:
A CHALLENGE FOR ECOLOGISTS
Gordon B.Bonan 139
CARBON BALANCE IN FOREST ECOSYSTEMS: RESPONSE TO NITROGEN
James A. Entry, Kim G. Mattson, and Mary Beth Adams 155
CONSEQUENCES OF TREE MORTALITY TO THE GLOBAL CARBON CYCLE
Mark E. Harmon, Sandra Brown, and Stith T. Gower 167
FOREST FIRES IN THE FORMER SOVIET UNION: PAST, PRESENT AND FUTURE
GREENHOUSE GAS CONTRIBUTIONS TO THE ATMOSPHERE
OlgaN.Krankina 179
ASSESSING CLIMATE CONTROLS OF NORTHERN FOREST SOILS
AND TREE GROWTH: RECONSTRUCTING SOIL MOISTURE, TEMPERATURE
AND SOLUTION CHEMISTRY FROM MONTHLY WEATHER RECORDS
Paul A. Arp and Xiwei Yin 187
SELECTING A MODEL TO ESTIMATE THE EFFECTS OF CLIMATE CHANGE
AND MANAGEMENT ACTIVITIES ON CARBON CYCLING
IN TEMPERATE FOREST REGIONS
BobZybach 194
ASSESSING THE IMPACT OF CLIMATIC STRESS ON FOREST PRODUCTION
John Runyon, Richard H. Waring, and Richard W. McCreight 203
VEGETATION FEEDBACKS AND THE PREDICTION
OF FUTURE FLUXES OF THE GLOBAL CARBON CYCLE
Robert A. Nisbet and Daniel B. Botkin 209
CARBON BUDGET AND SUCCESSION DYNAMICS
OF CANADIAN VEGETATION
Josef Cihlar and Michael Apps 215
Vlll
-------
TABLE OF CONTENTS
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON
STORAGE BY EXTRAPOLATION FROM LOCAL ECOSYSTEM MODELS
Anthony W. King 221
CARBON POOLS AND FLUX ON FORESTED LANDS
OF THE UNITED STATES
Hermann Gucinski, David P. Turner, Charles Peterson, and Greg Koerper 235
ESTIMATING CARBON BUDGETS OF CANADIAN FOREST
ECOSYSTEMS USING A NATIONAL SCALE MODEL
Michael J. Apps, Werner A. Kurz, and David T. Price 243
REGIONALIZATION AS A TOOL FOR STUDYING
CARBON CYCLING IN THE FORMER SOVIET UNION
Victor Blanutsa 253
FRAMEWORK TO QUANTIFY THE NATURAL TERRESTRIAL
CARBON CYCLE OF THE FORMER SOVIET UNION
Tatyana P. Kolchugina and Ted S. Vinson 259
APPENDIX A: DIRECTORY OF WORKSHOP PARTICIPANTS 277
IX
-------
-------
WARMING THE NORTH: WHAT HAPPENS?
George M. Woodwell and Richard A. Houghton
ABSTRACT
The vegetation and soils of the tundra, the boreal forest and associated montane forests of the northern middle latitudes
contain 500-700 billion tons of carbon, 1/4-1/3 of the total carbon thought to be held in organic form on land. Their total
metabolism is disproportionately smaller, perhaps about 6 billion tons of carbon in a global net primary production of 66
billion tons, or about 9% of the total. Any change in either the total pool of carbon held in the vegetation and soils or in
the rates ofmetabolism has the potential for significantly affectingthe composition of the atmosphere. The most conspicuous,
andpossiblythe mostimportant, short-term change is a change in the area of forest. Harvest and fires are especially important,
but the speed of the warming of the earth maybe great enough to reduce the area of forest significantly over the next decades.
The magnitude of the change is difficult to appraise with accuracy, but the reduction in area seems inevitable in view of the
speed of the warming over the last two decades and the pervasiveness ofotherhuman influences. The net effect of a reduction
in area afforest is almost certainly apositive feedback, afurtherrelease of carbon into the atmosphere due to the combination
of the warming and the continued effects of harvest and fires. Satellite imagery is being used to examine changes in the area
of forests, the factors that affect those changes and to test correlations between the composition of the atmosphere and the
metabolism of forest and tundra. Thepotential of forests andtheir soils for affectingthe composition of the atmosphere makes
such studies urgent.
INTRODUCTION
The boreal forest, the tundra, the montane
coniferous forests and the forests of the transi-
tion zone to middle-latitude deciduous forests
together constitute most of the tundra and
approximately half of the total forested area of
the earth. The area of forest was greater before
humans began their inexorable march across
the landscape, changing forest into farm and
farm into impoverished land (Thomas, 1956;
Woodwell, 1990).Theamountofcarbonheldin
these higher latitude plant communities, in-
cluding their soils, is large, approximately as
much as is held in the atmosphere currently.
The accumulation is the product of an excess of
net production over total respiration (Woodwell
andWhittaker, 1968; Gorham, 1991;Ovenden,
1987; Billings, 1987; Billings et al., 1982) over
thousands of years. The incremental excess
persisted throughout most of the post glacial
period and is assumed to continue, at least in
much of the peatlands of the far north. The most
important question concerning this part of the
world raised by the prospect of a rapid warming
of the earth is how the warming will affect the
region. Will the tendency toward the storage of
carbon be accelerated? Or will the large pools
of carbon in these regions become vulnerable
to decay and the carbon released as carbon
dioxide and methane to speed the warming?
Much hinges on the speed of the warming. Our
current hypothesis, based on first principles of
ecology and simple models, is that the initial
effect of the warming now anticipated over the
-------
WARMING THE NORTH: WHAT HAPPENS?
next decades willbe dominatedby effects on the
forested zone through the impoverishment of
the natural communities of the region
(Woodwell and Houghton, 1990; Woodwell,
1990). The impoverishment will be accompa-
nied by the release of significant, but unpredict-
able, additional quantities of carbon dioxide
and methane into the atmosphere from tundra
peat and forest soils. The evidence is incom-
plete and will remain so until the full course of
the warming has been experienced. Nonethe-
less, experience is accumulating, and new tools
are available to examine various aspects of the
workings of this part of nature.
THEMASS OFCARBONINBIOTICPOOLS
AND PRIMARY PRODUCTION:
THE ROLE OF FORESTS
The total amount of carbon held in plants and
soils globally is widely thought to be about 2000
Pg with 500-600 in plants on land (Olson et al.,
1983; Ajtay et al., 1979) and about 1500 in soils,
including peats (Schlesinger, 1984; Post et al.,
1982). The amount held in the atmosphere
currently as carbon dioxide is about 750 Pg. The
atmospheric burden has been increasing annu-
ally in recent years at a rate between 3 and 5 Pg
(Keeling et al., 1989). More than 3/5ths of the
total carbon on land is associated with forests
and tundra. Gorham's (1991) recent estimate
for the peatlands alone was 455 Pg C, about I/
4 of the global total. The boreal forest probably
contains another 400 Pg or more. It is safe to
assume for the purposes of this analysis that at
least half of the global total of carbon held on
land is held in the vegetation and soils of the
tundra, the boreal forest, montane forests and
the transitional forests of the higher mid-lati-
tudes. These are the areas of the earth that will
be affected most severely and early by the
warming.
An indicationof the importance of these forests
and tundra in determining the composition of
the atmosphere emerges from a recent attempt
by Stone of the Woods Hole Research Center
(unpublished data) to examine correlations
between the oscillation in the carbon dioxide
content of the atmosphere and the metabolism
of the vegetation. He has used data on the CO2
content of the atmosphere at Point Barrow,
Alaska, with NDVI indices (Fung et al., 1987;
Goward et al., 1985) calculated from weather
satellite data with a resolution of 15 km.
The NDVI is defined as a combination of the
reflected radiances measured in the visible
(CVIS) and near infra-red (CNIR) spectral
regions:
NDVI= CNIR - cvis
CNIR + CVIS
(1)
Clouds, water and bare ground result in low
NDVI values, while vegetation produces a range
of values related to the density and vigor of the
vegetation. On regional scales, the NDVI has
been used to monitor temporal variations in
biomass accumulation and productivity in the
semi-arid Sahel (Tucker et al., 1985b; Tucker et
al., 1986b; Hiernaux and Justice, 1986; Justice
et al., 1985). On continental scales, the NDVI
has been used to classify vegetation types
(Norwine and Greegor, 1983; Tucker et al.,
1985a) and as an index of net primary produc-
tivity (Goward etal., 1985). Global distribution
of the NDVI has been used to appraise the
monthly assimilation of CO2 by terrestrial veg-
etation (Tucker etal., 1986a; Fung etal., 1987).
Stone used a principal components analysis
using the NDVI data for the world over 12
months and showed that the first component
was acyclic; it showed little seasonal change.
The second component, in contrast, had a strong
seasonal cycle and appeared to be a measure of
the vigor and seasonally of mid-latitude veg-
etation of the northern hemisphere. The first
four components were used in establishing a
correlation between the principal components
and the annual course of the concentration of
-------
carbon dioxide in the atmosphere as measured
at Point Barrow and at Mauna Loa. The corre-
lation was nearly perfect. The data were then
sorted to determine the locales in which the
NDVI varied as the carbon dioxide content
varied in those places. The answer was that the
NDVI of the mid- and higher-latitude forests of
the northern hemisphere are correlated with
the observations at Barrow of the composition
of the atmosphere. The correlations for Mauna
Loa shift southward to the seasonal tropics.
The existence of large, post-glacial stocks of
carbon in the tundra and taiga is evidence that
net primary production has exceeded total res-
piration in these communities over thousands
of years, and the assumption is that there is a
continuingincrement of accumulation. Gorham
(1991) estimated that the annual increment
added to peatlands was 0.096 Pg C throughout
the post-glacial period and cited Clymo's (1984)
model as the basis for an estimate of the current
annual accumulation of 0.076 Pg. Such incre-
ments are difficult to measure, but there is
reason to expect that as the earth warms, these
large pools of carbon will become vulnerable to
decay, more vulnerable at their warmer and
drier limits of distribution than elsewhere. The
loss may have already begun. Oechel et al.
(1991) have suggested recently that tussock
tundra may be losing 240 g C/m2/yr, and that
the total release may be as much as 0.2 Pg
annually. Because the destruction of forests,
soil organic matter, and peat commonly occurs
more rapidly than the re-establishment of for-
ests, fens or bogs, it is reasonable to expect that
a warming that is as rapid as now anticipated
(Houghton, J. et al, 1990) will release large
quantities of carbon from these pools into the
atmosphere as the simple result of the start of
the transition from one type of vegetation to
another as the earth warms. Precise estimation
of the magnitude of that release is difficult, but
the release might be large, especially if the rate
of warming is measured in a few tenths to one
degree C or more per decade. Thebest measure
G.M. WOODWELL AND RA. HOUGHTON
will be the changes in area of forest (Emanuel
et al., 1985), including the direct changes due to
deforestation or other impoverishment from
human activity.
The rate of the warming is important. A rate of
the order of 1 degree C or less per millennium
or 1/10 degree C or less per century would be
of the general order of change during the glacial
and post-glacial periods. Such rates are slow
enough to allow adjustments in the distribution
of plants and plant communities, adjustments
that include selection for favorable genotypes
over several generations, even for the longer-
lived species. The problems arise when the rate
of change increases, as now seems probable in
these latitudes, to 1 degree per century, or,
possibly, to 1 degree per decade. This change is
occurring at the time when other human-caused
stresses are accumulating rapidly and when
human demands on forests are rising in re-
sponse to the demands of a human population
that has the intrinsic potential to continue dou-
bling every 30-40 years. While no one knows
how rapidly the earth will warm over the next
decades, the speed of the rise in temperature
during the last decade suggests that the higher
rates for the globe as a whole may occur and that
the projections of still higher rates in the higher
latitudes may push temperature changes in
those regions into the range of 1 degree per
decade, possibly more1. The implications of
such rates of change are profound. They fall
well outside the range of experience of science
or scientists. Yet, such changes are well within
the range of possibility. Ignoring that possibility
would seem to be the height of folly. Further-
more, examination of the extremes often offers
insights into patterns to be sought in response to
lesser change.
THE EFFECTS OF RAPID WARMING:
BIOTIC IMPOVERISHMENT
WITH THE RELEASE OF CARBON
A1 degree C change in mean annual tempera-
-------
WARMING THE NORTH: WHAT HAPPENS?
turein the middle and higher latitudes is equiva-
lent to alatitudinal change of 60-100 miles (100-
150 km). The implications are profound. Such
a change in latitudinal position implies not
simply a warmer climate, but a new climate in
that place with a different ratio of precipitation
to evaporation, a different length of growing
season, day length and different temperature
and moisture regimes. Although a plant species
may occur through several degrees of latitude,
the individuals from different parts of that
range are far from identical, each is part of a
sub-population of the species that has been
selected over several, perhaps many, genera-
tions for the particular characteristics required
to survive in a special segment of the range.
Climatic changes that are more rapid than the
generation time of the species may well outrun
the ecological amplitude of the species through-
out its range, not merely at the warmer and
drier margin of its distribution. Vulnerability
appears in various forms, but commonly in
increased susceptibility to insects or disease,
and the morbidity is ascribed to these secondary
symptoms (National Academy of Science, 1989).
The process is insidious, difficult to measure,
but cumulative and can best be described as
systematic impoverishment (Clark, 1988). Car-
ried to an extreme, the forest is replaced by
shrubland or, later, by grassland or other still
more impoverished communities. This pattern
of change has been confirmed for forests from
widely different habitats around the world
(Woodwell, 1990).
The shift from forest to a lesser-statured veg-
etation involves a release of carbon into the
atmosphere as carbon dioxide or methane. The
magnitude of the release depends on the area
affected, the magnitude of the change in stature
and the extent to which the change affects soils.
The most easily measured change is any change
in the area of forest; deforestation can usually
be identified directly onsatellite images and the
area measured. Such changes are underway
now and constitute globally a loss of carbon into
the atmosphere that is commonly thought to be
second only to the release from combustion of
fossil fuels. Estimates for the current release
range between 1 and 3 Pg annually (Houghton,
1991), mostly from deforestation in the tropics.
Increasing pressures on forests globally, espe-
cially the last of the grand forests of the north-
ern hemisphere, are bringing a surge in defor-
estation there as well. This surge remains
unmeasured, despite its importance.
More difficult to measure, or detect, is the
incremental impoverishment of forests and
othervegetation. Such impoverishment is widely
recognized as common in response to chronic
disturbance, including air pollution and the
acidification of rain (National Academy of Sci-
ence, 1989; Woodwell, 1990). The area in-
volved in such incremental degradation may be
very large and the potential releases of carbon
also large. There is no easy approach to such
measurements. Areas that can be used for com-
parison as "controls" may not be available.
Direct effects are confounded by secondary
infections by disease or by insects, and the
causes of the changes in the structure of the
forest remain as obscure as the fact of the
impoverishment. Arapid warming falls into this
category, a cause of impoverishment that will
be obscured by secondary changes that become
the immediate and obvious causes of morbidity
in trees and other plants. Satellite imagery
offers no easy entree to measurement, at least
in the short term, although efforts are being
made to use the higher resolution imagery of
Landsat and SPOT to determine the vigor of
local populations of trees in the eastern decidu-
ous forest (Burke et al., 1990). These efforts are
in an early stage and moving slowly. They offer
no immediate application in the context of this
discussion.
One of the most promising techniques for direct
measurement of metabolism is measurement
-------
of gaseous exchanges between extensive tracts
of forest and the atmosphere. Few such efforts
are underway. One study exists in north central
Maine where an almost continuous forest cover
exists for more than 100 miles to the westward.
A tower that extends well above the tree canopy
has been installed and instrumented. Measure-
ments include a continuous record of carbon
dioxide concentrations at various heights within
and above the forest canopy. The resolution
possible from such measurements is increasing
as experience accumulates. We expect to be
able to define the relationship between respira-
tion and temperature for this forested zone
over time.
One of the effects of the changes in climate may
involve a change in the frequency and severity
of fires (Clark, 1988; Clark, 1989; Flannigan
and Van Wagner, 1991). Uncontrolled fires
now affect large areas of Siberian forests and
have affected boreal forests throughout time.
Satellite imagery offers both the opportunity
for observing current and recent burns, but also
the possibility of tracing past burns on the basis
of appraisals of the stage of successional veg-
etation in fire scars. This technique offers the
possibility of defining the frequency of fires
over time and of tabulating areas affected.
CHANGES IN METABOLISM:
THE PROSPECTS
The combination of effects of a rapid warming
points to the possibility that the predominant
effect will be reduction in the area of forest and
reduction in the carbon content of existing
forest stands as the habitat becomes less favor-
able for existingpopulations of plants and as the
effects of chronic disturbance accumulate. All
suggest a surge in the release of carbon through
accelerated respiration not balanced by any
parallelsurgeinprimaryproduction. The ques-
tion is raised as to why a change in climate,
especially a warming, should be expected to
G.M. WOODWELL AND RA. HOUGHTON
reduce the fitness of species or to cause impov-
erishment, as opposed to enrichment, of natu-
ral communities. The answer lies in the speed of
the changes and in the demands that human
enterprise now makes on the earth as a whole.
We are facing changes in temperature at rates
of tenths of a degree to one degree or more per
decade as opposed to tenths of a degree per
millennium. On the millennial scale, adjust-
ments in the distribution or even the fitness of
populations of species are possible. In a decade
or two they are possible only for species with
short life cycles.
There are abundant data showing that increases
in the carbon dioxide content of the atmo-
sphere can be expected to stimulate the growth
of plants by both speeding the rate of photosyn-
thesis and reducing the rate of dark respiration,
thereby increasing net primary production
(Strain and Cure, 1985; Drake, 1989) and im-
proving efficiency of use of water. The difficulty
is that these effects, however persuasive the
data, do not appear to extend to natural plant
communities. There is no evidence from trees,
for instance, that these effects are appearing as
a result of the 25% increase in the carbon
dioxide content of the atmosphere over the past
century. These effects, if realized, will have to
be very large indeed to exceed the effects out-
lined above and dominate over the next de-
cades.
The changes in the vegetation of the earth
outlined here in response to the warming now
thought to be well underway have a potential
for bringing rapid further change in the compo-
sition of the atmosphere in a period of years to
a few decades. These changes are as large and
threatening as the effects of burning fossil fuels.
There is no guarantee that the effects will be as
outlined here, but the threat they pose is serious
enough to warrant a major effort to appraise
them and to assure that no change occurs in the
earth that might entrain them, for once en-
trained, they are not easily controlled.
-------
WARMING THE NORTH: WHAT HAPPENS?
THE FOREST INVENTORY:
AN ELEMENTARY
BUT IMMEDIATE NEED
Apart from the obvious steps appropriate in
stopping the further accumulation of carbon
dioxide and methane from use of fossil fuels,
the first step in an improved appraisal of the
biotic contributions is an improvement in the
knowledge of the area, changes in area and
changes in carbon content of forests of the
middle and higher latitudes where the effects
are expected to appear first and to be most
severe. Datafor appraisals of changes in area of
forests are available and have been available
for nearly two decades, but virtually no money
has been available for that mundane but impor-
tant purpose. Progress is now being made. The
Woods Hole Research Center, for instance,
using private funds, has recently completed a
map of the vegetation of South America based
on imagery from the NOAA weather satellites
at a resolution of 1 km. The imagery is being
used now as part of aprogram designed to offer
appraisals of rates of deforestation in the re-
gion. A similar map of Canada, not available to
us at the moment, has been prepared by the
Canadian Center for Remote Sensing based on
similar imagery and Canadian data from forest
surveys. Despite their coarse resolution, the
potential ofsuchmapsinadvancingbasicknowl-
edge of existing biotic resources and in moni-
toring changes should not be underestimated.
The imagery offers daily regional coverage.
Higher resolution is obtainable from other sat-
ellites for places of particular interest, which
may be identified on the coarser resolution
imagery.
It is almost inconceivable to a scientist directly
involved in attempting to appraise the biotic
causes and biotic implications of a rapid warm-
ing of the earth that there is no major effort
underway to define and redefine the magnitude
of the major factors that are likely to be en-
trained as the earth warms and affect not only
the course of the warming, but the potential of
the human habitat for supporting people.
CONCLUSIONS
A rapid warming of the earth measured in the
middle and high latitudes as tenths of a degree
C to 1.0 or more degrees per decade will almost
certainly cause a rapid impoverishment of the
vegetation, especially forests, and speed the
decay of organic matter in soils and peat. The
result will be a significant additional release of
carbon as carbon dioxide and methane into the
atmosphere. The magnitude of the release is
difficult to estimate, but the pools of carbon are
large, and the potential releases over years are
in the range of tens to hundreds of Pg of carbon.
Such changes may be underway now.
The potential for a significant positive feedback
from the warming of the earth alone is enough
to justify a greatly increased effort in defining
the details of the distribution of major vegeta-
tion types currently and establishing systems for
monitoring for change. Existing satellite imag-
ery offers significant potential not yet realized
in full, but governmental support for what ap-
pear to be mundane applications is necessary to
provide the basic data on the factors that may be
entrained, the magnitude of the changes and
the potential for affecting the human enter-
prise.
-------
REFERENCES
Ajtay, G.L., P. Ketner, and P. Duvigneaud. 1979. Terres-
trial primary production and phytomass. In: The Global
Carbon Cycle, edited by B. Bolin, E. T. Degens, S. Kempe,
and P. Ketner. SCOPE13, John Wiley and Sons, New
York, pp. 129-182.
Billings, W.D., J.O. Luken, DA. Mortensen, and K.M.
Peterson. 1982. Arctic tundra: a source or sink for
atmospheric carbon dioxide in a changing environment?
Oecologia (Berlin) 53:7-11.
Billings, W.D. 1987. Carbon balance of Alaskan tundra
and taiga ecosystems: past, present and future. Quarterly
Science Reviews 6:165-167.
Burke, M.K., R. A. Houghton, and G.M.Woodwell. 1990.
Progress toward predicting the potential for increased
emissions of CH4 from wetlands as a consequence of
global warming. In: Soils and the greenhouse effect,
editedbyA.F.Bouwman. John Wiley andSons, Chicester,
UK.
Clark, J.'S. 1988. Effect of climate change on fire regimes
in Northwestern Minnesota. Nature. 334:233-235.
Clark, J.S. 1989. Effects of long-term rater balances on
fire regime, Northwestern Minnesota. Journal of Ecology
77:989-1004.
Clymo, R.S. 1984. The limits to peat bog growth.
Philosophical Transactions of the Royal Society of Lon-
don B. 303:606-654.
Cure, B.R. and J.D. Cure. 1985. Direct effects of increas-
ing carbon dioxide on vegetation. U. S. Department of
Energy. National Technical Information Service, Spring-
field, Virginia 22161, p. 286.
Drake, B. 1989. Elevated atmospheric CO2 concentra-
tion increases carbon sequestering in coastal wetlands.
Carbon Dioxide Information Analysis Center, Environ-
mental Sciences Division, ORNL, Oak Ridge, Tennessee.
Emanuel, W.R., H.H. Shugard, andM.P. Stevenson. 1985.
Climatic change and the broad-scale distribution of terres-
trial ecosystem complexes. Climatic Change 7:29-44.
Flannigan, M.D. and C.E. Van Wagner. 1991. Climate
change and wildfire hi Canada. Canadian Journal of
Forest Research 21:66-72.
G.M. WOODWELL AND RA. HOUGHTON
Fung, I.Y., C.J. Tucker and K.C. Prentice. 1987. Appli-
cation of advanced very high resolution radiometer veg-
etation index to study atmosphere-biosphere exchange of
carbon dioxide. Journal of Geophysical Research 92-2999-
3015.
Gorham,E. 1991. Northern peatlands: role in the carbon
cycle and probable response to climaticwarming. Ecologi-
cal Applications 1:182-195.
Goward, S.N., C.J. Tucker, and D.G. Dye. 1985. North
American vegetation patterns observed with the NOAA-
7 advanced very high resolution radiometer. Vegetation
64:3-14.
Hiernaux, P.H.Y., and C.O. Justice. 1986. Suivi Du
developpement vegetal au cours De L'ete 1984 Dans Le
Sahel Malien. International Journal of Remote Sensing
7:1515-1531.
Houghton, RA. 1991. Tropical deforestation and atmo-
spheric carbon dioxide. Climatic Change 19:99-118.
Houghton, J.T., G.J. Jenkins, and J.J. Ephraums. 1990.
Climatic change. The IPCC Scientific Assessment, Cam-
bridge Univ. Press, Cambridge.
Justice, C.O., J. Townshend, B. Holben, and CJ. Tucker.
1985. Analysis of the phenology of global vegetation using
meteorological satellite data. International Journal of
Remote Sensing 6:1271-1318.
Keeling, C.D., S.C. Piper, and M. Heimann. 1989. A three
dimensional model of atmospheric CO2 transport based
on observed winds: 4. Mean annual gradients and interan-
nual variations. Geophysical Monograph 55, American
Geophysical Union, pp. 305-363.
National Academy of Sciences. 1989. Biologic markers
of air-pollution stress and damage in forests. National
Academy Press, Washington, D.C.
Norwine, J. and D.H. Greegor. 1983. Vegetation classi-
fication based on advanced very high resolution radiom-
eter (AVHRR) satellite imagery. Remote Sensing of
Environment 13:69-87.
Oechel, W.C., M. Jenkins, S. J. Hastings, G. Vourlitis, N.
Grulke, and G. Reichers. 1991. Effects of recent and
predicted global change on Arctic ecosystems. Abstract.
Bulletin of the Ecological Society of America 72:199-209.
Olson, J.S., JA. Watts and L.JAllison. 1983. Carbon in
live vegetation of major world ecosystems. DOE-TR004,
U.S. Department of Energy.
-------
WARMING THE NORTH: WHAT HAPPENS?
Ovcndcn, L. 1987. Peat accumulation in northern wet-
lands. Quaternary Research 33:377-386.
Post, W.M., W.R. Emmanuel, P.J. Zinke, and A.G.
Stangenberger. 1982. Soil carbon pools and world life
zones. Nature 298:156-159.
Schlcsinger,W.H. 1984. Soil organic matter: a source of
atmospheric CO2. G.M.Woodwell. Theroleof terrestrial
vegetation in the global carbon cycle: measurement by
remote sensing. SCOPE 23. John Wiley and Sons. Lon-
don, New York.
Thomas, W.L., Jr. 1956. Man's role in changing the face
of the Earth. University of Chicago Press, Chicago, Illi-
nois, p. 1193.
Tucker, CJ., R.R.G. Townshend, and T. Goff. 1985a.
Continental land cover classification using NOAA-7
AVHRR data. Science 227:369-375.
Tucker, CJ., C.L. Vanpraet, MJ. Sharman, and G. Van
Ittersum. 1985b. Satellite remote sensing of total herba-
ceous biomass production in the Senegalese Sahel: 1980-
1984. Remote Sensbg of the Environment 17:233-249.
Tucker, C J., I.Y. Fung, CD. Keeling, and R.H. Gammon.
19S6a. Relationship between atmospheric CO2 variations
and a satellite-derived vegetation index. Nature 319:195-
199.
Tucker, CJ.,C.O. Justice, and S.D. Prince. 1986b. Moni-
toring the grasslands of the Sahel 1984-1985. International
Journal of Remote Sensing 7:1571-1581.
Woodwell, G.M. 1990a. The Earth in transition: patterns
and processes of biotic impoverishment. Cambridge
University Press, New York.
Woodwell, G.M. 1990b. The Earth under stress: a
transition to climatic instability raises questions about
biotic impoverishment. G.M. Woodwell. The Earth in
transition: patterns and processes of biotic impoverish-
ment. Cambridge University Press, New York.
Woodwell, G.M. and R.H. Whittaker. 1968. Primary
production in terrestrial ecosystems. American Zoologist
8:19-30.
Woodwell G.M. and RA. Houghton. 1990. The experi-
mentalimpoverishment of natural communities: effects of
ionizing radiation on plant communities, 1961-1976. G.M.
Woodwell. The Earth in transition: patterns and processes
of biotic impoverishment. Cambridge University Press,
New York.
NOTES
1. The average rate of increase in temperature of the earth as a whole has been estimated by the IPCC (Houghton et
al, 1990) as 0.2-0.5° C per decade. The increase in the higher latitudes will be greater than the average by a factor
that may exceed 2, depending on the latitude.
-------
BOREAL CARBON POOLS: APPROACHES
AND CONSTRAINTS IN GLOBAL EXTRAPOLATIONS
F. Stuart Chapin, III and Elaine Matthews
ABSTRACT
The amount of carbon stored in soils of boreal ecosystems is important because of its potential to act as a carbon source
to the atmosphere-a positive feedback to climatic warming. Any study of boreal carbon storage should begin with
consideration of its relationship to present and future trace-gas flux to the atmosphere. This relationship is currently unclear.
Thereareseveralconceptualfydistinctapproachestoestimatingborealsoilcarbonpools,involvingcorrelationwithclimate
vegetation, soil type, etc. The biggest challenge in estimating boreal soil carbon pools is to determine which combination
of these factors is the most appropriate basis for extrapolating across broad geographic areas. There is much less uncertainty
about the areal extent of different northern vegetation types than about which vegetation types (or indeed which parameters
for extrapolation) are most appropriate. Using vegetation as one mode for extrapolation, there is greater uncertainty in the
accuracy and representativeness ofsoil carbon densities reportedin the literature than inthe arealextent ofeach type Given
our current estimates, approximately 50% of the boreal soil carbon pool is in evergreen and deciduous needle-leaf forests
another 25% in shrub and graminoid tundras and the remaining 25% in assorted other vegetation types.
INTRODUCTION
Northern ecosystems may play a key role in
future climate change because of their poten-
tial to feed back to the global climate. General
circulation models predict that the currently
rising concentrations of greenhouse gases such
as CO2 and CH4 are likely to cause greatest
warming at high latitudes (Maxwell, 1992). The
peats of northern ecosystems comprise 20-30%
of the world's soil carbon (Post et al., 1982,
1985, Billings, 1987). These peats accumulate
because low temperatures limit decomposition
rate through direct temperature effects and
indirect effects of reduced drainage and aera-
tion mediated by the presence of permafrost
(Chapin, 1983). The projected warming of
northern climates could cause northern soils to
be a major source of atmospheric CO2 if decom-
position is stimulated more strongly than pri-
mary production by warmer temperature, melt-
ing of permafrost, and improved drainage.
Moreover, the boreal zone is currently a major
source of methane, a potent greenhouse gas
(Whalen and Reeburgh, 1990).
GENERAL APPROACHES
AND CONSTRAINTS
In this paper we discuss several general ap-
proaches to evaluating the importance of north-
ern regions in climatic feedbacks and consider
the potential errors associated with each ap-
proach. The general problem is to estimate the
future emission of trace gases, particularly CO2
and CH4, from boreal regions:
E = (a) x (f)
(1)
-------
BOREAL CARBON POOLS
where E is the total future northern emission, a
is the future area of northern ecosystems, and f is
the future flux per unit area.
However, this question must immediately be
refined. There are substantial differences
among northern ecosystems hi trace-gas fluxes
(WhalenandReeburgh, 1990; Oechel and Bill-
ings, 1992), so we must consider several distinct
ecosystems:
E =
(2)
where n is the number of future ecosystems with
different rates of trace-gas flux.
The major challenge is to predict the area,
locationandtracegasfluxofeachfuture ecosys-
tem. Predicting the distribution of future eco-
systems isamajor challenge (Davis, 1981; Pren-
tice and Fung, 1990; Overpeck et al., 1991).
Therefore, we should begin by attempting to
estimate current fluxes of trace gases from
northern regions.
There are several perspectives that provide
insights and constraints to estimates of trace-
gas fluxes. Estimates of the carbon budget of
the atmosphere indicate that considerably more
new carbon enters the atmosphere (5.3-7.5 Gt)
from human activity than can be explained by
the current rate of atmospheric CO2 increase
(Tans et al., 1990). Several independent ap-
proaches indicate that the annual carbon flux
into oceans is unlikely to exceed 2 Gt. Con-
straints of seasonal fluctuations in sources and
sinks and inter-hemispheric gradients of CO2
concentration and isotopic composition sug-
gest a terrestrial sink of 0.3-2.5 Gt carbon in
north-temperate regions. Given these atmo-
spheric constraints, boreal ecosystems must
either be a net sink for CO2 or, if they are a net
source, we must find a sink of equivalent mag-
nitude elsewhere among north-temperate
ecosystems.
The paleo-record provides additional hints as
to what changes in flux might be expected from
climatic warming. At the global level, ice cores
indicate that past warming during interglacial
periods is associated with increased concentra-
tions of atmospheric CO2 and CH4 (Barnola et
al., 1987; Raynaud et al., 1988). By contrast,
tundra areas have shown greater peat accumu-
lation during warm periods, suggesting that
low-lying northern areas became a net sink in
warm climates. Moreover, we are moving from
a warm interglacial to a still warmer climate.
With no present or past analogue to the ex-
pected future climate, it is difficult to know how
to use these paleoecological patterns to predict
future trace-gas fluxes from boreal areas.
Patterns of carbon distribution in ecosystems
provide a third basis for predicting changes in
trace-gas flux. Plants typically have a higher
C:N ratio than the soils on which they grow and,
therefore, are capable of storing more carbon
per unit nitrogen than are soils (Rastetter et al.,
1991; Shaver et al., 1992). If climatic warming
caused decomposition of boreal peats, releas-
ing nitrogen, this should allow greater plant
growth, particularly growth of largewoodyplants
like shrubs and trees which have a high C:N
ratio, which would allow northern ecosystems
to store more carbon per unit of nitrogen. This
logic relies on the assumption that climatic
warming would cause a transfer of nitrogen
from soil to plants with no change in total
nitrogen pool of the ecosystem. This latter
assumption might be questioned given the large
changes in ecosystem nitrogen pools that occur
through plant succession (Van Cleve et al.,
1991) and the tendency of tundra ecosystems to
have greater ecosystem nitrogen pools than
more southerly ecosystems (Chapinetal., 1980).
Nonetheless, C:N ratios of plants and soils
provide a strong constraint on the possible
changes in total carbon flux that can occur in
ecosystems.
The final and most direct way to estimate how
10
-------
trace-gas fluxes may change is to examine cur-
rent environmental controls over trace-gas flux
from northern regions. GCM projections of
future climate would then provide a mechanis-
tic basis for predicting future trace-gas fluxes.
Recent studies indicate that upland tundra is
currently a large source of CO2 (Grulke et al.,
1990; Oechel and Billings, 1992), but a weak
sink or source of CH4 (Whalen and Reeburgh,
1990; Torn and Chapin, 1992), depending on
prevailing moisture. By contrast, lowland areas
are strong CO2 sinks and strong CH4 sources.
Seasonal variations in trace-gas fluxes correlate
with both soil moisture and temperature. The
predominance of topographic controls suggests
that soil moisture exerts primary control over
CO2 and CH4 fluxes and that temperature is a
secondary control, a conclusion that is consis-
tent with laboratory incubations (Kielland and
Schimel, 1991). Other controls over net trace-
gas flux may include pH, organic matter quality
and soil nitrogen status, but these possible
controls are less firmly established. There is no
evidence that the soil carbon pool (i.e., quantity
of carbon in the soil) has any influence on trace-
gas flux. Thus, an improved understanding of
soil carbon quality or of the pool size of rapidly
turning-over carbonpools, such as litter, may be
more important than carbon pools in estimat-
ing future trace-gas fluxes. Pool sizes of soil
carbon are important only as a long-term con-
straint on future CO2 flux from northern ecosys-
tems. Fire is another important control over
carbon storage in northern systems. If the
climate warms, dries and supports more plant
biomass, fire frequency may increase, thus re-
ducing the average soil carbon storage.
SOIL CARBON POOLS
Although most attention should be focused on
trace-gas fluxes and their environmental con-
trols, we now discuss estimating the geographic
distribution of soil carbon because this is a
question that has received considerable atten-
F.S. CHAPIN, III AND E. MATTHEWS
tion based on a substantial data base. There-
fore, it provides avenue to explore approaches
and pitfalls to various methods of global ex-
trapolation without immediately being limited
by inadequacy of the data base. As mentioned
above, total soil carbon pools provide a maxi-
mum constraint on the potential feedback of
boreal ecosystems on increasing atmospheric
CO2.
At least three presumed controls over soil car-
bon pools (climate, vegetation and soil type)
have been used to extrapolate soil carbonpools
to the circumpolar Arctic. The most compre-
hensive study has been based on climate (Post
et al., 1982, 1985). Some 4000 soil profiles
world-wide (including about 60 in tundra and
350 in boreal forest) were associated with cli-
matic data either directly or through associa-
tion with vegetation that has a known relation-
ship to climate (Figure 1). When mapped on
climate space defined by biotemperature and
precipitation, carbon density in tundra and bo-
real forest (i.e., those ecosystems at the top of
Figure 1) show a clear pattern of increasing soil
carbon density with decreasing temperature at
any given precipitation (Post et al., 1982). The
large range in soil carbon densities within wet
tundra and wet boreal forest could indicate
strong sensitivity to climate within these zones,
small sample size (n=33 and 24, respectively),
and/or sensitivity of soil carbon density to other
unmeasured factors. Even with this substantial
data base (the largest available), there are large
boreal areas for which no data are available,
particularly in the north (Figure 2a). When we
extrapolate the climate-based carbon densities
of Post et al. (1982), using a high-resolution
data set of temperature and precipitation (Shea,
1986), the broad band of boreal forest has soil
carbon densities of 10-14 kg C/m2 (Figure 3a),
and carbon densities become progressively
higher as one moves north through tundra.
Thus, the climate-based extrapolation predicts
highest carbon densities at the northern-most
latitudes.
11
-------
BOREAL CARBON POOLS
Vegetation provides a second basis for extrapo-
lating soil carbon densities. Beginning with the
topically detailed version of the 1° latitude x 1°
longitude vegetation data set of Matthews
(1983), we simplified the 44 boreal vegetation
types that occur north of 50°N and have a
biotemperature of 6°C or less into three boreal
forest types and seven tundra types (Figure 4).
Averaging measurements available from the
literature (Figure 2b), we made a preliminary
estimate of boreal soil carbon pools. Table 1
lists the newly compiled carbon densities and
resultingcarbonpoolsbyvegetationtype, along
with ecosystem areas. For comparison, Table 1
also includes carbon densities and pools calcu-
lated from the density distributions we derived
following the climate-based extrapolation
method of Post et al. (1982). Note that these
values, stratified by the vegetation distributions
of Matthews (1983) are close but not identical
to those reported by Post for Holdridge (1947)
life zones. Because these averages for vegeta-
tion types are preliminary, we treat this global
extrapolation as an exercise to test alternative
approaches rather than as a definitive carbon
estimate. Soil carbon densities for boreal for-
ests derived from the climate- (Figure 3a) and
vegetation-based (Figure 3b) data sets show
good general agreement (Table 1). However,
for tundra ecosystems, the vegetation-based
distributions indicate decreasing soil carbon
densities with higher latitude, exactly the oppo-
site pattern to that predicted from the climate-
based method of Post et al. (1982). At present,
we cannot say definitively whether this differ-
ence reflects difference in the soil carbon data
bases or differences in modes of extrapolation.
Assuming the latter, this difference in predic-
tion provides an opportunity to evaluate whether
vegetation or climate is the strongest determi-
nant of soil carbon density in northern regions.
This type of analysis would be helped by better
measurement coverage within ecosystems. Al-
ternatively, soil type, parent material, succes-
sional age, topography, or other factors may be
a more effective basis for global extrapolation.
A multi-factor approach, such as the subdivi-
sion of vegetation types according to topogra-
phy into flooded wetlands vs uplands (Table 1)
Polar
Subpolar
Boreal
Cool temperate
Warm temperate ,£
Subtropical
Tropical """
Nival
Alpine
SubcUpine
1.5*
«• -Lf
3 *>
e* *»
6 JL
Montane
^ ^B> — «» •• <•• ^» *
Lower montane
PremonUne
24'
X 4C8 10 1418 22
Carbon in mineral soil (kg m~2)
Figure 1. Relationship of soil carbon density to climate inboreal zones (modified from Holdridge, 1947 and Post et al., 1982)
12
-------
F.S. CHAPIN, III AND E. MATTHEWS
rigure 2. Boreal distribution of soil-carbon measurement sites used for (a) the climate-based extrapolation of Post et
al. (1982) and (b) the vegetation based extrapolation used in this study. Symbols: circle indicates 1-5 measurements in
- 1° land cell, square denotes 6-10 measurements, and star denotes > 10 measurements.
13
-------
BOREAL CARBON POOLS
NASA/GISS
POST ET «... 1962
6 12 18 24 30
Kg/m*
FigureS. Distribution of soil carbon density in boreal regions based on extrapolations from (a) climate (Shea, 1986; Posi
ct al., 1982) and (b) vegetation (this study).
14
-------
F.S. CHAPIN, ffl AND E. MATTHEWS
BOREAL VEGETATION TYPES
EVERGREEN NEEDLELEAF
FOREST/WOODLAND
' DECIDUOUS NEEDLELEAF
FOREST/WOODLAND
\ DECIDUOUS BROADLEAF
I FOREST/WOODLAND
DWARF SHRUB TUNDRA
SHRUBTUNDRA
MOSS/DWARF SHRUB BOG
GRASS SEDGE BOG
GRAMINOID/TUSSOCK
SEDGE TUNDRA
DRY/MESIC MEADOW
POLAR DESERT
Figure 4. Map of boreal vegetation from the data of Matthews (1983).
Table 1. Boreal soil carbon inventory (a) based on vegetation data (Matthews, 1983) and soil carbon measurements
averaged from unpublished compilation of Chapin, and (b) based on climate extrapolations of Post et al. (1982). In
parentheses are wetland portions of total ecosystem areas.
Vegetation Type
1 . Evergreen needleleaf forest/woodland
2. Cold-deciduous needleleaf forest/woodland
3. Cold-deciduous broadleaf forest/woodland
4. Dwarf shrub tundra
5. Shrub tundra
6. Moss/dwarf shrub bog
7. Grass sedge bog
8. Graminoid/tussock sedge tundra
9. Dry/mesic meadow
10. Polar desert
Total
Area CDensity(a) CPool(a) CDensity(b) CPool(b)
lO^m2 kg/m2 1015g kg/m2 1015 g
7.6
4.7
1.2
5.1
0.4
1.1
0.3
1.3
0.4
0.5
22.6
(14)
(4)
(7)
(2)
(10)
(51)
(8)
(15)
(9)
(1)
(10)
11.1
11.1
11.8
4.8
15.0
12.9
30.8
26.8
14.3
0.7
11.0
84.8
51.8
14.3
24.4
6.7
14.3
8.2
33.7
5.8
0.3
244.3
14
13
13
18
18
16
14
16
12
22
15
106
61
16
92
9
' 18
4
21
5
11
343
15
-------
BOREAL CARBON POOLS
Table 2, Carbon inventory of major ecosystems in the
circumpolar boreal zone, as estimated by Post et al.
(1982); Bliss and Matveyeva (1992); Oechel and Billings
(1992); and Chapin and Matthews (this study).
Carbon inventory (x 1015 g)
Post
Tundra
Polar desert
Sena-desert
Low shrub
Tall shrub
Tussock
Mire
Dry/mesio meadow
Total Tundra 192
Boreal Forest
Needle-leaf evergreen
Needle-leaf deciduous
Bread-leaf deciduous
Portlands
Total 202
Boreal Forest
Total 394
Bliss
0.02
3.01
1.28
0.07
0.92
36.9
—
42.2
—
—
—
~
—
—
Oechel
0.07
10.8
4.86
0.09
26.1
13.4
—
55.1
87.5(b)
—
~
122
209.5
265
Chapin
& Matthews
0.3
24.4(a)
—
6.7
33.7
22.5
5.8
93.4
84.8
51.8
14.3
—
151
244
(a) Includes low shrub.
(b) Induifcs all non-peatlands.
may be particularly important in predicting
how carbon storage may change in the future
because tundra uplands are currently carbon
sources and lowlands are carbon sinks (Oechel
and Billings, 1992).
The estimate based on vegetation suggests that
evergreen and deciduous needle-leaf forests
contribute about half of the total boreal soil
carbonpool (Table 1). They are large contribu-
tors because they cover a large geographic area
and have a substantial soil-carbon density. Dwarf
shrub tundra and graminoid/tussock tundra
are also important (another 25% of the total in
this study), but for different reasons: Dwarf
shrub tundra has a low soil-carbon density, but
covers a large geographic area, whereas
graminoid/tussock tundra has a high soil-car-
bon density, but covers a relatively small area.
Table 3. Areal extent of major ecosystems in the boreal
zone, as estimatedby Post et al. (1982); Buss and Matveyeva
(1992); Oechel and Billings (1992); and Chapin and Mat-
thews (this study).
Tundra
Polar desert
Semi-desert
Low shrub
Tall shrub
Tussock
Mire
Dry/mesic meadow
Total Tundra
Boreal Forest
Needle-leaf evergreen
Needle-leaf deciduous
Broad-leaf deciduous
Peatlands
Total Boreal
Forest
Total
Area (x
Post Bliss
0.8
1.4
1.3
0.2
0.9
1.0
—
8.8 5.6
13.1 --
21.9 -
itf5 km2)
Chapin
Oechel & Matthews
0.80
1.50
1.28
0.23
0.90
1.00
—
5.71
10.1(b)
1.1
11.2
16.9
0.5
5.1(a)
-
0.4
1.3
1.4
0.4
9.1
7.6
4.7
1.2
13.5
22.6
(a) Includes low shrub.
(b) Includes all non-peatlands.
Other areas are small contributors to the boreal
soil carbon pool because they have both a small
area and a low soil-carbon density.
There is little information available to evaluate
the accuracy of our estimate of boreal soil-
carbon pools. As a first step, we compared our
estimates with those of previous studies (Table
2). Although the estimates of total soil carbon
in the boreal region agree within 60%, there is
substantial discrepancy among authors for indi-
vidual vegetation types, in part due to differ-
ences in vegetation classification. For example,
Oechel and Billings (1992) distinguish between
boreal peatlands and boreal forest, whereas we
do not. In other cases (e.g., most tundra types),
discrepancies among estimates appear genu-
ine. Concurrency in estimates (e.g., area of
tundra ecosystems; Table 3; Bliss and
16
-------
Table 4. Carbon density of major ecosystems in theboreal
zone, as estimatedby Post etal. (1982); Bliss andMatveyeva
(1992); Oechel and Billings (1992); and Chapin and
Matthews (this study).
Carbon density (kg/m?)
Post
Tundra
Polar desert 21.8(a
Senti-desert
Low shrub
Tall shrub
Tussock
Mire
Dry/mesic meadow
Boreal Forest
Needle-leafevergreen 19.3
Needle-leafdeciduous 19.3
Broad-leafdeciduous 11.6
Peatlands ~
Bliss Oechel
0.02 0.09
1.8 7.20
1.0(c) 3.80
0.4(c) 0.40
1.0(c) 29.00
30.0 13.40
-
- 10.8(d)
..
7.7
111
Chapin
& Matthews
0.7
4.8(b)
-
15(c)
26.8
30.8
14.3
11.1
ll.l(c)
11.8
~
(a) Includes all tundra.
(b) Includes low shrub.
(o) Estimate.
(d) Includes needle-leaf deciduous.
Matveyeva, 1992; Oechel and Billings, 1992)
often reflects citation of the same data rather
than independent estimates. We should focus
attention on ecosystems that show large dis-
crepancies in absolute magnitude. For ex-
ample, authors differ by an order of magnitude
for carbon pools of polar desert and by a factor
of about three for carbon pools of mires. How-
ever, there is agreement that polar desert is not
a large contributor to the total carbon inven-
tory, whereas the discrepancy in carbon inven-
tory for the mire has substantial impact on the
total for tundra. In still other cases (e.g., tussock
and tall shrub), there is a large discrepancy in
both relative and absolute terms.
The discrepancy in carbon pools could reflect
differences in estimates of either areal extent
(Table 3) or soil-carbon density (Table 4).
Surprisingly, all authors agree as to the total
area of boreal vegetation (TableS). Therefore,
estimates of the areal extent of different vegeta-
F.S. CHAPIN, ffl AND E. MATTHEWS
tion types are probably adequate and may not
warrant a major research effort. What is more
questionable is whether these are the most
appropriate vegetation types or even whether
vegetation alone is the most appropriate basis
for extrapolation. Most vegetation maps are
based onphytosociological analyses of vegeta-
tion which may show little correspondence to
the functioning of the ecosystem. For example,
our graminoid/tussock tundra includes both
coastal wet meadows and tussock tundra. Al-
though both these vegetation types have similar
carbon densities, we expect them to respond
quite differently to climate change (Chapin et
al., 1992), so they should probably be separated
in future vegetation classifications, even though
they are grouped in the UNESCO classifica-
tion. Similarly, dwarf shrub tundra should
probably be separated into areas dominated by
deciduous shrubs andthose dominatedby heath
species because the large differences in litter
quality between these types (Shaver and Chapin,
1991) probably cause corresponding differences
in carbon turnover. We conclude that the major
problems associated with estimating area are
conceptual rather than data-based.
Ecosystem differences in carbon density ac-
count for most of the discrepancies in estimates
of boreal carbonpools, reflecting either paucity
of measurements (e.g., tall shrub tundra) or
large differences among estimates despite a
reasonable data base (e.g., mire). There are
several kinds of errors that seriously compro-
mise the validity of current estimates of soil-
carbon density. As with estimates of plant
biomass (Botkin and Simpson, 1990), soil scien-
tists generally measure soil carbonpools in sites
with well-developed mature soil profiles and
ignore rocky, early successional, or disturbed
sites, even though these sites comprise a sub-
stantial proportion of the landscape. More-
over, sites that have been studied are only a
small (and probably biased) sample of the en-
tire boreal region. For these two reasons, the
values reported in the literature may not be
17
-------
BOREAL CARBON POOLS
representative of the vegetation types, even if
the data accurately reflect the sampled site.
There are also many sources of error, such as
incomplete (and often unreported) depth of
sampling, estimation of the amount of soil (vs.
rocks and roots) in aprofile, lack of data on bulk
density and difference in analytical method for
estimating carbon (Zinke et al. 1984) associ-
ated with sampling soil carbon pools. Any data
used in a global extrapolation should be care-
fully scrutinized for problems of this sort before
being included.
CONCLUSIONS
The major challenges in predicting the role of
northern ecosystems in biospheric feedbacks to
global climate are to predict future fluxes of
CO2 and CH4. To do this we must estimate
future fluxes and the distributions of future
ecosystems, which in turn require an under-
standing of the controls over current fluxes,
estimation of the area of current ecosystems
and prediction of the patterns of ecosystem and
climate change. A first step in this direction is
to document current pools and fluxes on a
broad geographic scale. This mustbe done with
a viewto understanding controls and predicting
the future, rather than considering estimation
of pools as an end in itself.
There is a substantial data base from which to
estimate soil carbonpools for the boreal region.
The major stumbling block at present is in
knowing which correlates of soil-carbon density
are most appropriate to use as a basis for
extrapolation. Extrapolations based on climate
and vegetation predict fundamentally different
geographic patterns of soil-carbon density in
the tundra zone. These factors, soil type, some
other correlate or some combination of corre-
lates may prove the most reasonable in extrapo-
latingsoil carbon geographically and in predict-
ing future soil carbon.
In terms of the data, there appears to be much
greater uncertainty about soil-carbon density
for particular ecosystems than in estimating the
areal extent of boreal ecosystems. There has
been no study undertaken to test whether soil-
carbon densities reported in the literature are
representative of the entire extent of vegetation
types, either locally or over broader geographic
ranges. In addition, there are errors inherent in
sampling soil carbon density, many of which are
difficult to evaluate in published reports.
There are some fairly clear next steps that can
be undertaken to improve our estimates of
boreal carbon pools. Extant data should be
rigorously assessed for their validity and then
correlated with several alternative predictors
such as climate, vegetation, soil type and topog-
raphy that are available in global data bases.
These regressions will lead to hypotheses about
which factors control the present pools of soil
carbon. The data should be incorporated into
process-based models developed to predict
present and future fluxes. There should not be
a major field campaign to collect additional
data on soil-carbon density until we know
whether this is a useful parameter in predicting
current trace-gas flux. Other parameters such
as litter pool or the quality of litter and soil
organic matter may be equally or more impor-
tant than soil-carbon density. If an extensive
field campaign is initiated to collect data essen-
tial to the models, it should be undertaken in
regions where information is currently sparse
or lacking and should be carefully designed to
differentiate among alternative controls over
soil-carbon densities. Finally, any estimates of
current or future trace-gas flux should be made
within the constraints provided by other types
of measurements and models, such as the sea-
sonaliry and north-south gradient in concentra-
tion and isotopic composition of trace gases in
the atmosphere, the range of possible C:N
ratios in plants and soil, etc.
18
-------
F.S. CHAPIN, III AND E. MATTHEWS
REFERENCES
Barnola, J.M., D. Raynaud, Y.S. Korotkevitch, and C.
Lorius. 1987. Vostok ice core: a 160,000 year record of
atmospheric CO2. Nature 329:408-414.
Bazilevich, N.I., AA. Tishkov, and G.E. Vilchek. 1991.
Primary productivity and nutrient cycling in Arctic ecosys-
tems. Nauka Publ., Moscow. In press.
Billings, W.D. 1987. Carbon balance of Alaskan tundra
and taiga ecosystems: past, present and future. Quat. Sci.
Rev. 6:165-177.
Bliss, L.C., and N. V. Matveyeva. 1992. Circumpolar arctic
vegetation. In: Arctic Ecosystems in a Changing Climate,
edited by F.S. Chapin, III, R.L. Jefferies, J.F. Reynolds,
G.R. Shaver, and J. Svoboda. Academic Press, San Diego,
pp 59-89.
Botkin, D.B., and L.G. Simpson. 1991. Biomass of the
North American boreal forest: a step toward accurate
global measures. Biogeochemistry. In press.
Chapin, F.S., III. 1983. Direct and indirect effects of
temperature on arctic plants. Polar Biology 2:47-52.
Chapin, F.S., III, P.C. Miller, W.D. Billings, and P.I.
Coyne. 1980. Carbon and nutrient budgets and then-
control in coastal tundra. In : An Arctic Ecosystem: The
Coastal Tundra at Barrow, Alaska, edited by J. Brown,
F.L. Bunnell, P.C. Miller, and L.L. Tieszen. Dowden,
Hutchinson and Ross, Stroudsburg, pp. 458-482.
Chapin, F.S., III, R.L. Jefferies, J.F. Reynolds, G.R.
Shaver, and J. Svoboda. 1992. Arctic plant physiological
ecology in an ecosystem context. In: Arctic Ecosystems in
a Changing Climate, edited by F.S. Chapin, III, R.L.
Jefferies, J.F. Reynolds, G.R. Shaver, and J. Svoboda.
Academic Press, San Diego, pp. 441-451.
Davis, M.B. 1981. Quaternary history and the stability of
forest communities. In: Forest Succession: Concepts and
Applications, edited by D.C. West, H.H. Shugart, and
D.B. Botkin. Springer-Verlag, New York, New York, pp.
132-153.
Grulke, N.E., G.H. Reichers, W.C. Oechel, U. Hjelm, and
C. Jaeger. 1990. Carbon balance in tussock tundra under
ambient and elevated atmospheric CO2. Oecologia83:485-
494.
Holdridge, L.R. 1947. Determination of world plant for-
mations from simple climate data. Science 105:367-368.
Kielland, K. and J.B. Schimel. 1991. Temperature and
moisture controls over carbon flux in arctic tundra soils.
Tenth International Symposium on Environmental Bio-
geochemistry.
Matthews, E. 1983. Global vegetation and land use: new
high-resolution data bases for climate studies. J. Clim
Appl. Meteor. 22:474-487.
Maxwell, B. 1992. Arctic climate: potential for change
under global warming. In: Arctic Ecosystems in a
Changing Climate, editedby F.S. Chapin, HI, R.L. Jefferies,
J.F. Reynolds, G.R. Shaver, and J. Svoboda. Academic
Press, San Diego, pp. 11-34.
Oechel, W.C. and W.D. Billings. 1992. Effects of global
change on the carbon balance of arctic plants and ecosys-
tems. In: Arctic Ecosystems in a Changing Climate,
edited by F.S. Chapin, III, R.L. Jefferies, J.F. Reynolds,
G.R. Shaver, and J. Svoboda. Academic Press, San Diego,
pp. 139-168.
Overpeck, J.T., P.J. Bartlein, and T. Webb, HI. 1991.
Potential magnitude of future vegetation change in east-
ern North America: comparisons with the past. Science
254:692-695.
Post, W.M., W.R. Emanuel, P.J. Zinke, and A.G.
Stangenberger. 1982. Soil carbon pools and world life
zones. Nature 298:156-159.
Post, W.M., J. Pastor, P. J. Zinke, and A.G. Stangenberger.
1985. Global patterns of soil nitrogen storage. Nature
317:613-616.
Prentice, K.C. 1990. Bioclimatic distribution of vegetation
for general circulation model studies. J. Geophys. Res.
95(D8):11811-11830.
Prentice, K.C. and I.Y. Fung. 1990. The sensitivity of
terrestrial carbon storage to climate change. Nature 346:48-
51.
Rastetter, E.B., M.G. Ryan, G.R. Shaver, J.M. Melillo,
K.J. Nadelhoffer, J.E. Hobbie, and J.D. Aber. 1991. A
general biogeochemical model describing the responses
of the C and N cycles in terrestrial ecosystems to changes
in CO2, climate, and N deposition. Tree Physiol. 9:101-
126.
Raynaud, D., J. Chappellaz, J.M. Barnola, Y.S. Korotkevich,
and C. Lorius. 1988. Climate and CH4 cycle implications
of glacial-interglacial CH4 change in the Vostok ice core.
Nature 333:655-657.
19
-------
BOREAL CARBON POOLS
Shaver, G.R., W.D. Billings, F.S. Chapin, HI, A.E. Giblin,
KJ. Nadelhoffer, W.C. Oechel, and E.B. Rastetter. 1992.
Global change and the carbon balance of arctic ecosys-
tems. BioStience.
Shaver, G.R. and F.S. Chapin, III. Production biomass
relationships and element cycling hi contrasting arctic
vegetation types. Ecol. Monogr. 61:1-31.
Shea, DJ. 1986. Climatological atlas: 1950-1979. Tech.
Note, Natl. Center for Atmos. Res., Boulder.p. 35.
Tans, P.P., I.Y. Fung, and T.Takahashi. 1990. Observa-
tional constraints on the global atmospheric CO2 budget.
Science 247:1431-1438.
ACKNOWLEDGMENTS
Torn, M.S. and F.S. Chapin, III. 1992. Environmental and.
biotic controls over methane flux from arctic tundra.
Chemosphere.
Van Cleve, K., F.S. Chapin, III, C.T. Dyrness, and LA.
Viereck. 1991. Element cycling in taiga forests: state-
factor control. BioScience 41:78-88.
Whalen, S.C. and W.S. Reeburgh. 1990. A methane flux
transect along the trans-Alaska pipeline haulroad. Tellus
426:237-249.
Zinke, P.J., A.G. Stangenberger, W.M. Post, W.R.
Emanuel, and J.S. Olson. 1984. Worldwide organic soil
carbon and nitrogen data. Environmental Sciences Div.
Publ. 2212 Dept. of Energy, Oak Ridge.
ThisworkwassupportedbyaNASA grant to E.Matthews andbyNSF grant BSR-8705323 to the University of California.
20
-------
PLANETARY MAXIMUM CO2 AND ECOSYSTEMS OF THE NORTH
Sergey A. Zimov, Sergey P. Daviodov, Yuri V. Voropaev, Sergey F. Prosyannikov, and
Igor P. Semiletov
ABSTRACT
To estimate the contribution of the ecosystems in Northern Siberia to the formation of the high-latitudinal maximum of the
atmospheric CO2 concentration and one season's amplitude, studies ofCO2 balance were carried out in the lowerreaches
of the Kolyma River (69°N) from December, 1989, untilJanuary, 1991. The CO 2 fluxes were obtained at sites in various kinds
of landscapes. The mean annual CO2 release was above 400 g/m2. A range of50-300g C/m2 ofCO2 was released from soil
from September until late December. In October and November, fluxes were greater than in September. We suggest the
following explanation of this phenomenon. It is known that soils of the North are saturated in the summer. During October,
soil freezing occurs. At this time moisture moves to the surface and a deep, highly-porous, well-aerated layer is formed. This
stimulates aerobic organism activity. The following interesting phenomenon is governed by their activity. We identified
ground melting that occurred untilJanuary, and occasionally the temperature rose 3° C. The calculations performed show
that the energy released (about 50-300 g C is oxidized) due to the functioning of soil aerobes is sufficient to provide this
warming. At present, in Northern Siberia, the depth of thawing has increased due to the comprehensive disturbance of the
surface conditions and the aerobic oxidation of carbon, which was previously conserved in permafrost.
INTRODUCTION
The distribution of atmospheric CO2 is impor-
tant for an understanding of the global carbon
cycle. Measurements of atmospheric CO2 con-
centrations show that the CO2 concentration
maximum does not occur between 20° and 60°
N, where 95% of fossil fuel is burned (Rotty,
1983) nor over tropical forests, but between 50°
and 70°N over the tundra and taiga. Therefore,
it is assumed that ecosystems provide the plan-
etary maximum of atmospheric CO2 (Tucker et
al., 1986; Fung et al., 1987; Zavarzin and Clark,
1987).
The presence of a steady maximum at high
latitudes shows the existence of the most pow-
erful CO2 fluxes per unit area. The territory
between 20° and 60° N is 125 x 106 km2. About
5 Gt of carbon are burned annually in this
region (Rotty, 1983), hence the specific anthro-
pogenic emission at these latitudes is about 44
g C/m2/yr. The annual mean CO2 concentra-
tion at latitude 70° N is much higher than at 20-
60° N. Thus, to support this latitudinal concen-
tration gradient, the flux at latitude 70° N must
be larger than 44 g C/m2/year. The question is
whether such apowerful source, in low-produc-
tivity northern ecosystems, and what processes
may have caused such a great carbon imbal-
ance. The CO2 monitored data show also the
greatest season of air CO2 oscillations at the 60-
70th latitudes in the Northern Hemisphere
(Tucker etal., 1986; Fung etaL, 1987; Zavarzin
and Clark, 1987). Annual fluctuations of atmo-
spheric CO2 are believed to be related to sea-
sonal fluctuations in the activity of the conti-
nental biota. A minimum is observed in sum-
mer when absorption of atmospheric CO2 by
photosynthesis exceeds that produced by de-
composition of organic matter in the soil. In
21
-------
PLANETARY MAXIMUM CO2 AND ECOSYSTEMS OF THE NORTH
contrast, maximum CO2 levels occur in winter,
when the photosynthetic activity is least (Tucker
et al., 1986; Fung et al., 1987; Zavarzin and
Clark, 1987; Johnson and Kelly, 1970). How-
ever, the high difference between the summer
and winter extremes at the high latitude zone
(60-7S°N) is difficult to explain by seasonal
changes in photosynthetic activity. Photosyn-
thesis occurs there from June to August, and the
highest soil temperatures were also recorded
during this period. The soils in May are still
frozen, and by September, they are again start-
ing to freeze. The biological activity of soils, in
this case, is assumed to stand still.
Let us assume that there is no gas exchange
betweenthe high latitude atmosphere and other
zones. An air column of 1 m2 and a CO2
concentration of 350 ppm would contain 1440 g
of carbon. Then, approximately 82 g of carbon
would be required to change the CO2 concen-
tration by 20 ppm. However, the atmosphere of
latitude 70°N is not isolated from adjacent
territories with lower concentrations of CO2;
therefore, outflow of CO2 should take place.
Consequently, to produce the observed winter
maximum at latitude 70°N, the production of
carbon for the period from September to Janu-
ary would have to be significantly higher than 82
g (which would correspond to 1521 CO2/m2).
This is really a great amount of carbon and can
be compared with annual carbon accumulation
by Northern ecosystems due to photosynthesis.
Annual fluctuations of atmospheric CO2 have
been observed with a large, variable increase in
the CO2 concentration under the snow over
ambient air CO2 near Barrow, Alaska (Kelly et
al., 1968; Coyne and Kelly, 1974). This indi-
cates the existence of CO2 flux to the atmo-
sphere. It was assumed that CO2 production
occurred during a portion of the fall and spring
associated with biological activity, when media
temperatures are above the minimum physi-
ological threshold, approximately -7°C (Benoit
etal, 1972). Litter decompositionin these soils
generally ceases at -7.5° C (Flanagan and
Scarborough, 1972).
However, what are the values of these fluxes
and are these fluxes able to ensure the planetary
winter maximum of atmospheric CO2? To verify
the hypothesis, we evaluated CO2 fluxes be-
tween the atmosphere and northern ecosys-
tems. Measurements were carried out from
December, 1989, until January, 1991, at the
northeastern station situated at the lower
Kolyma River about 100 km south of the Arctic
coast (69°). Important natural fronts were: the
tree-line, the coast, and a range of hills and
mountains crossing this area, providing in a
relatively small area all the main landscapes
characteristic of Northeastern Siberia.
METHODS
To determine CO2 fluxes from soil, metal reser-
voirs 26 cm in diameter and 25 cm high were
inserted 5 cm deep at different landscape sites.
At the time of an observation, the reservoirs
were closed by transparent lids for 5 or 15
minutes, and then the air was sampled from
under the cover. The CO2 flux from soil was'
determined by a simple calculation, which is
based on the difference between CO2 concen-
trations in air and under the lid, taking into
account the size of a chamber (reservoir) and
the time of exposure (De Jong et al., 1979).
Natural vegetation: moss, lichen, grass and
shrubs was preserved in the majority of the
chambers. Metal cylinders did not shade them,
and that is why we measured the integral flux
between ecosystems and the atmosphere. Also,
CO2 flux from soil was measured only at some
sites with no plants. Analogous portable cham-
bers, which were inserted in the soil at some
fixed points, were also used. Along with these
chambers, at several sites, some metal cylinders
of greater height were inserted. They signifi-
cantly shaded the surface, but allowed us to
determine CO2 fluxes at a much higher snow
level.
22
-------
SA. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
In winter we used a "snow" modification of the
soil profile method (De Jong et al., 1979). This
method is based on Pick's first law;
F=D* C/h
(1)
where FisCO2fluxthroughasnowlayer, his the
thickness of the snow layer, C is the difference
between CO2 concentration at the top and the
bottom of the layer, and D is the CO2 diffusion
coefficient within the snow layer.
The C value was determined as a CO2-content
difference between the bottom of the layer and
the air. For air sampling, a thin tube was
inserted at each point of observation. The
sample value was 300 cm. The diffusion coeffi-
cient in snow was obtained by simultaneous
measurements of CO2 and CO2 fluxes through
a snow layer by means of measuring chambers.
The average value was equal to 0.63 cm2/s. For
a denser snow layer, the diffusion coefficient of
snow was not available, but was obtained by the
following experiment: Barrels of 50 and 30 cm
in diameter were filled with snow. Through the
bottoms of these barrels, a constant flow of CO2
was dispensed. A concentration difference was
determined when the steady state conditions
within the upper 20 cm layer had been reached.
The experiment, carried out in calm weather,
yielded values of CO2 flows of 120,90,60,30 and
15 cm3 CO2/m2 h. The diffusion coefficient in
snow was 0.22 cm2/s in January (t=-40° to -43°
C); 0.33 cm2/s in March (t=-10° to -12° C); and
0.30 cm2/s in April (t=-l° C).
The sample analysis was carried out using a gas
chromatograph, TSVET-530 (detector
catharometer, column with Porapac -T, 0.3 cm
in diam., 250 cm long) and enhanced IR -
analyzer GIAM-SM, which was used in the
discrete measurement mode. The influence of
water vapor on IR absorption was investigated
by means of comparing wet (100% saturation)
and dry (20% saturation) air testing with a
temperature 20°. The discrepancy did not ex-
ceed 7 ppm. With undiluted samples, the analy-
sis error of CO2 concentration was within 2%.
In consequence, the water vapor effect was
neglected.
Most of the CO2 measurements were taken on
the east side of the Kolyma River, along two
profiles crossing a typical gradual slope consist-
ing of loamy soils. The vegetation of this slope
was mosaic and characterized by scattered
larches, 8 -10 m high, shrubs, grass, moss and
lichen in various combinations. The profiles
crossed marshy and steppe areas, old rough-
country roads overrun with weeds and artificial
desert plots. In the winter season 1989-90, the
number of sampled sites on these profiles
reached 80. The measurements were carried
out in the morning and evening, or once in the
daytime. From June, 1990, observations were
made at 15 base profile sites one time per day.
For a detailed investigation of the daily dynam-
ics of CO2 flux, at some sites the daylight and
night observations were carried out with an
interval of two hours. In addition, irregular ,
observations (about 1000 probes) were taken
along six profiles, which crossed: 1) a plot of
forest burned eight years ago; 2) a marshy
thermokarst depression; 3-4) bushy and grassy
associations of the Kolyma River floodplain; 5)
the flat tundra; and 6) tundra on the slope of the
nearest mountainrange. Many sites are located
near monitoring sites for thaw-layer study: ob-
servations of temperature and soil moisture,
the level of frozen subsoil waters, soil freezing-
thawing dynamics. Soil temperature profiles
were determined by means of thermistors with
a precision of 0.1° C. The value of relative
moisture was measured in the same boreholes
at one time with the moisture meter (VGPR,
Russian abbreviation), precision 5%. Dynam-
ics of the freezing was investigated mechani-
cally by means of the insertion of a metal probe
through shallow holes, which were made early
in the upper frozen layer. In winter time, near
23
-------
PLANETARY MAXIMUM CO2 AND ECOSYSTEMS OF THE NORTH
some sites, bore holes were drilled. The bore-
holes were sampled every day.
We also studied the respiration activity of 30
natural soil cores, 25 cm in diameter and 21 cm
deep thatwereplacedinmetalreservoirs. These
cores were cut out in June and studied through-
out the summer to measure the transpiration
activity of various plant species (in loam). In
early February, 1990, these deeply frozen, but
still intact cores were divided into two groups
and placed in two rooms with various controlled
temperatures. To determine the CO2 fluxes
from these cores, each was placed inside a leak-
proof polyethylene bag of 1 m2 total surface
area and 501 in volume. The thickness of the
polyethylene film was 0.02 cm. Every day, air
samples were removed from bags for analysis.
The diffusion CO2 flux through the film was
determined by Pick's law as in Kasparov et al.
(1985). In summer, these monoliths were used
for the following testing: Monoliths were
wrapped in cotton wool and were covered by
transparent lids with a volume about 501 at the
same time the base profile was obtained. Air
samples were taken from under lids after 15-30
min. of exposure. Then, corresponding values
of CO2 release were calculated.
AGO,
PPM
a) 100
im (solid line); soil
'min (dashed line):
ixed grass - cereals
meadow; d) mean CC52 concentration values at 15 observation sites along based profile (see paper); e) mean daily air
temperature, °C, and snow depth, cm. (CO2 = 100 ppm approximately corresponding to a flux of 2 crn3/m2/min/snow depth
of 19 cm; 1 cm3/m2/min/snow depth of 38 cm).
24
-------
S.A. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
26.10.90
29.10.90
16.12.90
178
330
Figure 2. Spatial - time dynamics of CO2 values, ppm: a) the site with shrubby - moss plants (see Figure la, 3b); b)
the site with mixed grass - cereals meadow (see Figure Id, 3c). Each area is 3.3 m. Observation sites are noted by points.
RESULTS
The changes of CO2 subnivean flux are pre-
sented in Figures 1,2 and 3. We'll begin with the
analysis of under-snow fluxes. Our results for
the CO2 concentration beneath the snow layer
are similar to those obtained near Barrow,
Alaska (Kelly et al., 1968; Coyne and Kelly,
1974). However, there was more snow at the
sites measured at Barrow than at our sites and,
therefore, the CO2 flux at Barrow was less than
at some of our sites.
In September, CO2 flux from soil decreased as
the air temperature dropped. At the beginning
of October, the ten-day average along the base
profile CO2 flux was 0.4 g C/m2/day. But at the
middle of October, the flux became larger at
most sites, and one increased up to 1 g C/m2/
day. At the middle of November, it was equal
to 0.8 g C/m2/day, and one fell to minimum
October values only in the second half of De-
cember. In January, flux reached the value of
cni*
m2min
PCOj
27 28.08.90 29 30 31
Figure 3. Daily course of CO2 flux from soil, (cm3/m2/
min): a)broken moss-lichen plants; b) shrubby-moss
plants; c) mixed grass-cereals meadow; d) soil monolith
:rom mixed grass-cereals meadow; e) soil monolith with
leavy lichen; f) air temperature.
25
-------
PLANETARY MAXIMUM CO2 AND ECOSYSTEMS OF THE NORTH
0.16 g C/m2/day. Minimum flux was obtained
atthe end of February, 0.08gC/m2/day. At the
middle of April, after an anomalous early rise in
temperature, the values, averaged over ten
days, of subnivean fluxes sharply increased to
0.8 g C/m2/day. On the whole, within the
territory crossed by the base profile, the release
of CO2 into the atmosphere was equal to: 76 g
C/m2from September to December; 12 g C/m2
from January to March; 43 g C/m2 from April
to May.
We compared the data obtained on the other
profiles with the average values of the base
profile for the same time. Fluxes from soil
under highly productive shrubby and grassy
associations in valleys, marshy fells and tundra
exceeded the fluxes from the soil of the base
profile by 2- to 4-fold from September to De-
cember. High fluxes also were obtained at the
sites with disturbance of soil moss and lichen.
Then, the differences became not so great, and
at the end of December, there was practically
no differencebetween the fluxes. The fluxes on
the top and at the bottom of a mountain range
slope were approximately equal to those of the
base profile, but by December, in comparison,
their values became 1.5 times lower. The CO2
fluxes from soil on the previously burned site
changed very little from the base profile, but
from December to January, their intensity was
1.5 times greater. About 150-200 g C/m2 were
released into the atmosphere from subnivean
soil in the region of the lower Kolyma River
from September until May. This was quite
enough to obtain a winter maximum of air
concentrations.
In summer, the values of CO2 fluxes obtained
are equal to the difference between the soil
respiration and net photosynthesis of the lower
plant layer. At night, when illumination inten-
sity was low, plant photosynthesis did not com-
pensate for soil respiration. Inmost cases, how-
ever, this also occurred during some daylight
periods at sites with an abundance of green
phytomass. In the daytime, photosynthesis pre-
vailed over soil respiration. Both processes
depend upon the temperature (photosynthesis
to a lesser degree); therefore, this occurred in
periods with low air temperatures. In conse-
quence, gross CO2 flux hi ecosystems was ob-
tained more often during low-temperature
times. However, the mean estimations show
that total CO2 flux was directed toward the
atmosphere at all the sites around the clock.
This flux was enhanced as the temperature
increased. Gross CO2 flux to the atmosphere at
the lower plant layer was approximately 280 g/
m2 with a soil flux of about 330 - 360 g/m2.
Moreover, the CO2 flux from the soil under a
highly productive grass community was 2- to 3-
fold more than one under a lichen-moss com-
munity.
DISCUSSION
What is the nature of winter soil fluxes? What
is the source of stratification? Similar fluxes
were obtained in isolated monoliths and in
natural conditions. Thus, we can say that the
observed fluxes are determined by soil respira-
tion, although we do not rule out the possibility
that some portion of CO2 flux is connected with
abiogenic processes: CO2 desorbtion, release
from solutions and a squeezed pocket yield
(Coyne and Kelly, 1971).
The observations of soil monolith respiration
with various temperature regimes allowed us to
conclude that the main process is the biological
decomposition of soil organic matter. If it had
been some physical processes, the soil could
hardly have reacted with a sharp increase when
fertilized. The fluxes decreased sharply after
sterilization of a soil monolith. Then, the CO2
concentration in a bag was close to constant
with changed temperature. It seems that the
heating of monoliths also deactivated extracel-
lular enzymes. The role of these enzymes might
be great in CO2 production processes (Scujins,
1967).
26
-------
S.A. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
In our experiments, a sharp increase of CO2
fluxes from previously deep-frozen soil mono-
liths occurred when they were warmed to -4° C.
Therefore, the growth of subnivean fluxes in
April can be attributed to the warming of the
upper soil horizons. Beneath these horizons
the soil temperature was -8° to -12° C. In
autumn, the vertical temperature gradient is
reversed, upper layers are frozen and greatly
cooled, and deep layers contain melted water
from December until January. Thus, it might
be suggested that autumn CO2 fluxes are con-
nected with these soil layers. Active respiration
of the deep soil was confirmed by measuring
CO2 concentrations in the bore holes we drilled
to this layer. Even in January, they reached a
few thousands of ppm. It should be noted that
the CO2 concentration was much lower in shal-
low surface boreholes.
Northern soils, and especially deep soil layers,
are known to be water-saturated. Oxidative
processes are limited not only due to low tem-
peratures but also with the absence of oxygen.
Therefore, it is commonly assumed that CO2 is
produced by anaerobes (Zavarzin and Clark,
1987). From October until November, how-
ever, observations show that about 3-5 g C/m2
are oxidized daily in deep horizons under grassy
communities. This can be compared with the
CO2 emission from well-aired soils at the middle
latitude belt in summer. Only aerobic organ-
isms can provide such fluxes. There are data
that indicate that deep horizon activity in north-
ern soil is related to the activities of aerobic
organisms. A detailed study of the vertical
distribution of soil invertebrates, carried out in
Taimyr tundra, showed that their maximum
number is found in the uppermost well-aerated
horizons, but with increasing depth, their num-
bers drop off rapidly to zero. However, at the
bottom of the active layer, a number of inverte-
brates have been found. The bottom unex-
plained maximum may be even more than in the
upper layer in sites with highly productive veg-
etation (Chernov, 1978). In this region, a simi-
lar distribution was noted for aerobic microor-
ganisms (Parinkina, 1971).Themaximumnum-
ber of microorganisms at these depths in an
active layer is characteristic of our study area
(Gilichinsky et al., 1988). A high number of
aerobic organisms at these deep horizons sug-
gests the existence of an ecological optimum at
these levels.
We can consider the problem of water-satu-
rated Northern soils. In fact, the deep soil
horizons in the North are poorly aerated. How-
ever, this is only a feature of the summer season,
a fact not appreciated by ecologists. In autumn,
the upper soil horizons freeze. Then, moisture
from deep horizons is attracted to this frozen
upper layer. As a result of this redistribution,
the moisture content (ice content) increases
considerably within the upper horizons while it
decreases at lower levels. A loose, highly po-
rous and relatively dry layer (DL) is thus formed
(Kudriavtsev, 1981). Our investigation indi-
cated that, in many cases, moisture comes to the
surface even from the upper horizons of perma-
nently frozen ground (Figure 4). The upper soil
horizons are usually pierced by a mass of tiny
drying cracks and frost cracks. Because of them,
a porous DL is actively aerated. In the region
under study, the DL is formed in most kinds of
soil, first appearing in October, showing that
oxygen penetrates to the deep horizons. Note
that in our region, at late August, 1990, due to
frequent rains and low transpiration, the level
of subfrost water is almost up to the surface, but
at the beginning of October, it began to fall
down and at the middle of the month, the
majority of the drilled boreholes were already
dry. Then, CO2 fluxes from the soil were sharply
increased.
In autumn (up to December), and sometimes
until January, the temperature of the dry layer
is not less than at the same depth in summer.
However, these horizons have been aerated
since autumn, and therefore, conditions for
biotic activity including that of invertebrates, is
27
-------
PLANETARY MAXIMUM CO, AND ECOSYSTEMS OF THE NORTH
i 11 ill A AAA AA A A A
10 ' 11' 12 1 ' 2 ' 3 ' 4 ' 5 ' 6 ' 7 ' 8 ' 9 '10 ' 11' 12
1989
1990
A A A AAA 1 1 A A A A iA j j A j A A
10 T 11T 12 I 1 ' 2 ' 31 4 ' 5 ' 6 '7 ' 8 ' 9 T10 ' 11' 121
10 ll.
1989
1 A 1 11 1 i A A A A tl A Al i ii U i Ai A 1A ilo
10 f 11 f 121 1 1 2 ' 3 ' 4 '5 '6 ' 7 r 8 ' 9 r!0 ' ll'12l 10 J1 12
1989
\r
_30'
30
AAA AAA/
10 ' 11'12 I
Moisture (%/volume)
Temperature
Figure 4. Temperature dynamics, °C, and soil moisture by % volume: a) mixed grass - cereals meadow; b) shrubby -
moss area; c) lichen - mixed grass area. The depths are marked where the thermistors are located, and the dates when
the temperatures were determined are given. The soil-moisture content was measured by neutron moisture meter; the
measurement increment along the hole is 5 cm.
28
-------
S.A. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
better than in summer. Free water is soaked up,
due to upper-layer freezing. It freezes, forming
ice crystal interlayers, and streaks, loosening
and deforming the soil, killing by pressure or-
ganisms living there. But the DL contains little
free water. Thus, this layer is not deformed by
ice inclusions. Moisture at these depths is
present as film water. Thus, the soils of the DL
are soft even at sub-zero temperatures
(Kudriavtsev, 1981). All of these factors sug-
gest the existence of an ecological optimum for
many organisms at the deep horizons of the
active layer.
We can relate the active CO2 production at
deep-soil horizons to another interesting phe-
nomenon. It is assumed that, in areas where the
mean annual temperature of the permafrost is
lower than -3° to -5° C, the active-layer freezing
in autumn should begin not only at the surface,
but also from the direction of the permafrost. In
our region, the average temperature of perma-
frost is -6° to -8° C. Under these conditions, the
upwards freezing of the active layer must begin
earlier, and its rate should be more than the top-
downward freezing rate (Kudriavtsev, 1981).
However, recent studies have shown that up-
wards freezing from the permafrost layer was
rarely observed. Moreover, according to our
data, at some sites, melting of the permafrost
continued even into October and November.
Freezing dynamics were studied at tens of sites
using thin probes through the holes drilled in
the upper frozen layer and in the some holes
equipped with thermistors. At many sites, this
phenomenon was even recorded in autumn,
1989-1990. We suggest the following explana-
tion for this:
The total autumn CO2 release was found to be
more than 100 g C/m2. The biological forma-
tion of CO2 is an exothermal process. It is
known that during oxidation of 100 g of carbon
in the form of organic matter, about 800 kcal is
released, and this process is inevitable no mat-
ter which method of "burning" is used. If a CO2
flux exists, there is some energy release. The
heat value of about 800 kcal is quite enough for
the melting of 10 kg of ice or a few centimeters
of frozen soil. This heating is quite sufficient to
melt a soil layer 20 cm thick and to warm one to
approximately 15° C. At great depth, under
nearly constant temperatures in the DL, the
oxidation of organic carbon contributes consid-
erably to the heat balance. Thus, it can be
assumed that the high oxidative activity of the
soil biota in the dry layer retards the process of
soil freezing, prolonging their activity. The
importance of this mechanism is amplified by
positive feedback1. To demonstrate this, we
provide some observational data of soil tem-
peratures. Figure 4 shows that at the beginning
of October, there occurs insignificant freezing
of subsoils at the bottom, which after the forma-
tion of the DL, is changed by their thawing. In
this case, in many boreholes, the DL tempera-
ture was recorded to be higher, almost as high
as +3° C. It should be noted that temperature
observations were carried out in inserted metal
tubes. The soil thawing is a faster and deeper
process along these tubes than it is without
inserted tubes. The depth of thawing was
determined by probes in undisturbed sites. In
autumn, 1990, from October to December, the
greatest soil thawing occurred at the previously
burned site and reached 11 cm. It should be
noted that the measurement was carried out
from the fixed mark. Therefore, the vertical
motion of the soil surface was neglected, and
the error in the former measurement was ab-
sent.
To sum up, we can submit the following scheme
of the carbon dynamics in Northern ecosystems
in winter. In September, the Northern soils are
water-saturated and CO2 production occurs
mainlyin theupper horizons. Atdeep horizons,
the aerobic organisms are not active. In Octo-
ber, when the DL is forming, the aerobic organ-
isms in the DL resume their activity. In Decem-
ber, after the DL freezes, biotic activity de-
creases because soil invertebrates have become
29
-------
PLANETARY MAXIMUM CO2 AND ECOSYSTEMS OF THE NORTH
quiescent, and CO2 emission, therefore, de-
creases. However, within thin films of unfrozen
water, some of the microbiota continue func-
tioning, producing CO2. In spring, the upper
soil horizons are warmed up, and rapid CO2
productions notcompensatedforbyphotosyn-
thesis, which is now beginning. This provides
some spring CO2 maximum in the atmosphere.
When the upper soil horizons are thawed, the
temperature gradient changes and the DL is
soaking up melted water from the upper hori-
zons, which brings an abundance of dissolved
nutrients and water-soluble organic compounds
with it.
We found that summer CO2 release to the
atmosphere is about280 g C/m2 at the lowplant
layer. For the estimation of net CO2 release
into the atmosphere, the values of absorbed
carbon from trees and shrubs should be taken
into account. Unfortunately, we couldn't accu-
rately determine this effect, but it is definitely
lower than 280 g/m2. (It is about 500 g of dry
biomass.) Thus, the photosynthesis in our re-
gion did not compensate winter, but summer
fluxes as well, during 1990. Evenin summer, the
ecosystems were the source of carbon, and its
annual imbalance was approximately 350-500
g/m2.
Such large fluxes from soil are perhaps con-
nected with local peculiarities. For example,
the summer of 1990 was anomalously wet (45
days with rain). Perhaps the soil at the sites of
the study was somewhat destroyed, but in any
case, our observations showed that there might
be a great imbalance of carbon in Northern
ecosystems. This agrees with data from atmo-
spheric air CO2 monitoring.
There is considerable evidence that the contri-
bution of natural processes to the atmospheric
CO2 balance is as significant as fossil fuel burn-
ing. For example, fluctuations of anthropo-
genic CO2 emission and atmospheric CO2 con-
centration are not synchronous (Merser, 1978).
Paleomonitoring data in the ancient ice of
Greenland (Oeschger and Stauffel, 1985) and
Antarctica (Semiletov et al, 1989) showed sharp
fluctuations of CO2 concentrations. The dou-
bling of the atmospheric CO2 content was de-
tected for a 100-year period. From this, we may
propose that natural rates of change of CO2
concentrations were comparable to present an-
thropogenic changes in atmospheric CO2.
Ten to 27% of the biospheric carbon store is
concentrated in the ecosystems of tundra and
boreal forests (Oechel and Riechers, 1986).
The carbon, stored in the North as moss sod,
humus and organics in the upper horizons of the
permafrost, is sufficient to support the pro-
posed CO2 flow for many decades. What mecha-
nisms might cause CO2 release? They may be
simple: an increase in the soil temperature and
the depth of seasonal thawing, or the improve-
ment of drainage. But what might cause these
changes? Usually, the scale of changes is com-
pared with the scale of influence, but some-
times an avalanche of processes might be caused
by a small influence.
Consider briefly the process of soil self-warm-
ing as a global process. It is seems that the
contribution of biogenic processes to soil warm-
ing is not comparable with incoming solar ra-
diation. Photosynthesis utilizes the first per-
centage of solar radiation. However, only part
of the total solar radiation is absorbed by soil.
There is an imbalance among incoming solar
radiation and reflected radiation, effective
emission at the surface, heat of evaporation and
heat of turbulent exchange. In summer, this
differs by about 3-10% from the total solar flux.
For example, summer mean inflow total solar
flux at Vorcuta is about 38.2 x 104 kcal/m2 and
heat flux in thesoilis3.2x 104 kcal/m2, at Igarka,
44.2 x 104 kcal/m2 and 2.5 x 104 kcal/m2, at
Yakutsk, 62.4 x 104 kcal/m2 and 2.1 x 104 kcal/
m2 (Pavlov, 1979). These values come from TV
northern USSR. It should be noted that at our
study area, the annual inner soil heating was
30
-------
SA. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
provided by emission of approximately 0.4 x 104
kcal/m2. This value is about 20% of incoming
heat in soil2.
The carbon content of Northern soils is usually
tens of kg/m2, and this is "preserved energy"
equal to hundreds of thousands of kcal. The
decomposition activity of organic matter in soil
depends exponentially upon the temperature.
When the soil temperature is low, this "poten-
tial energy" is preserved. With rising tempera-
tures the soil might be warmed by mechanisms
of self-heating. In Northern ecosystems, there
are two series of processes, each based on
strong positive feedbacks (Zimov and
Chuprynin, 1989; Zimov, 1990).
Northern ecosystems are characterized to a
considerable extent by moss associations. Moss
sod has strong insulation qualities. During moss
accumulation, the soil temperature and depth
of summer thawing are decreased (often, only
the moss sod itself melts). Transpiration from
the moss and lichen surface is low. Therefore, in
places where they are present, the soils are
always water-saturated. This sharply reduces
opportunities for growth of soil organisms un-
der moss sod (Chernov, 1978). Moss-lichen
tissues are decomposed very slowly. In addi-
tion, the moss has bactericidal properties, sup-
pressing microbial activity. Due to these char-
acteristics, the moss associations in the North
are the greatest accumulators of carbon. This
is related to their aggressive influence on envi-
ronmental conditions and the deterioration of
soil conditions, and they have displaced their
rivals, namely the actively transpiring plants. In
northwesternSiberia,60-70%ofthetotalforest
area has disappeared during the last 300 years
owing to competition from moss (Kryuchkov,
1973).
The surfaces covered by moss associations are
thermally unstable. The supercooled soils with
high ice content occur under fragile moss sod.
Therefore, the disturbance of moss sod as a
result of lemming and reindeer browsing, fires,
or cryogenic processes causes the development
of the following processes:
When the moss sod is destroyed, the soil thaws
to a depth two to three times greater than the
depth under moss. The soil is warmed up, and
actively transpiring plants reappear (transpira-
tion among grasses is two to six times more than
for moss and lichen in our study region). The
soil moisture content is reduced, its aeration is
improved, activity of soil organisms rises and
organic matter is more actively utilized. Due to
these changes, the rate of plant transpiration is
increased again, soil moisture is reduced and
the thaw depth is increased. Therefore, more
organic matter, contained in permafrost, is be-
coming involved in the decomposition process.
As a result of this sequence of processes with
positive feedbacks, previously accumulated or-
ganic matter is broken down as a result of
activity of the soil biota (with an appropriate
heat release and warming of soil).
In natural systems where non-linear relation-
ships and bifurcation processes prevail, minor
external effects cause great changes in the natu-
ral process (Zimovand Chuprynin, 1989; Zimov,
1990). Consider the factors that make these
changes and release the carbon stored in the
soil and permafrost. Currently, the Northern
landscapes are actively transformed despite the
low human population density.
First, it is known that lichens and mosses are
more sensitive to overall air pollution. Second,
mining and processing of mineral resources in
Northern Eurasia within the Arctic Circle, on
the Kolsky Peninsula near the towns of Vorcuta
and Norilsk, produces a large amount of air
pollution. The Northern Far East is a zone of
intensive mining of placer deposits, while north-
west Siberia is now the main supplier of petro-
leum and gas. A considerable portion of this
mined fuel is burned in the mining regions.
Third, in all these regions, ecological losses per
31
-------
PLANETARY MAXIMUM CO2 AND ECOSYSTEMS OF THE NORTH
Figure 5. Fires in northeastern Siberia. 1) forest; 2) forest-burnings as shown on topographic maps from the early sixties.
According to the forest inspection after the time the map was made, 50 fires were registered in this territory. These fires
covered more than 3000 km2. Compare this with the square area of the rectangle.
rained resource unit have been much greater
than in other areas, owing to the neglect of
elementary nature-preservation standards
(Zimov, 1990). Now, it is one of the ecologically
distressed regions. Even in the region of our
study, where mineral resources have not been
mined, over half the area has been subjected to
forest fires, which released many millions of
tons of carbon to the atmosphere (Figure 5).
Today, upon old burns, the organic matter
previously preserved in the permafrost is ac-
tively oxidized due to a sharp increase in the
depth of thawing. As a result, the soil depth of
thawing has increased from 0.3-0.6 to 2 m. It
should be noted that the thawing sometimes is
continued until March. Fourth, a large part of
Northern Eurasia is reindeer grazing land. This
area is comparable to that of the Amazon
tropical forests. During the last decades, owing
to the increase in the reindeer population and
violation of the grazing regime, the sharp trans-
formation of plant covering has occurred. Ev-
erywhere, the lichen pastures are on the brink
of overgrazing, and their area andbiomass have
been reduced by more then one-half
(Sioroechkovsky, 1986). In consequence, the
foregoing causes are sufficient to significantly
change the carbon balance in Northern ecosys-
tems.
REFERENCES
Benoit, R.E., W.B. Campbell, and R.W. Harris. 1972.
Decomposition of organic matter in the wet meadow
tundra, Barrow: A revised word model. In: Proceedings
of the 1972 Tundra Biome Symposium, coordinated by J.
Brown, Tundra Biome Center, University of Alaska,
Fairbanks.
Chernov, Yu. I. 1978. Structure of animal population in
Subarctic. Nauka Press, Moscow (in Russian).
32
-------
SA. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
Coyne, P.I. and J J.Kelly. 1971. Release of carbon dioxide
from frozen soil to the Arctic atmosphere. Nature
234(5329):407-408.
Coyne, P.I. and J.J. Kelly. 1974. Variations in carbon
dioxide across an Arctic snowpack during spring. J.
Geoph. Res. 79(6):799-802.
De Jong, E., R.E. Redmann, and EA. Ripley. 1979. A
comparison of methods to measure soil respiration. Soil
Sci. 127(5):300-307.
Flanagan, P.W. and A. Scarborough. 1972. Laboratory
and field studies on decomposition of plant material in
Alaska tundra areas. In: Proceeding of the 1972 Tundra
Biome Symposium, coordinated by J. Brown, Tundra
Biome Center, University of Alaska, Fairbanks.
Fung, I.Y,C.J. Tucker and K.C. Prentice. 1987. Applica-
tion of advanced very high resolution radiometer vegeta-
tion index to study atmosphere - biosphere exchange of
C02. J. Geoph. Res. 92(D3):2999-3015.
Gilichinsky, DA., G.M. Khlebnikova, D.G. Zvyagintsev,
D.G. Fedorov-Daviodov, and N.N. Kudryavtseva. 1988.
Proc. 5 Int. Conf. Permafrost, August 2-5, Ed. Daare
Sennese, Trondheim, Norway.
Houghton, RA., R.D. Boone, J.R. Fruci, T.E. Hobbie,
J.M. Melillo, C A. Palm, BJ. Peterson, G.R. Shaver, G.M.
Woodwell,B.Moore,D.I.Skole,andN. Myers. 1987. The
flux of carbon from terrestrial ecosystems to the atmo-
sphere in 1980 due to changes in land use: geographic
distribution of the global flux. Tellus 398:122-139.
Johnson, P.L. and J.J. Kelly. 1970. Dynamics of carbon
dioxide and productivity in an arctic biosphere. Ecology
51(1):73-80.
Kasparov, S.V., N.S. Panikov, and O.I. Minko. 1985.
Membrane sampler for soil air analysis. Pochvovedenie
11:145-150 (in Russian).
Kelly, J.J., D.F. Weaver, and B.P. Smith. 1968. The
variation of carbon dioxide under the snow on Arctic.
Ecology 49(2):358-361.
Kryuchkov, V.V. 1973. Far North: problems of rational
use of natural resources. Miosl Press, Moscow (in Rus-
sian).
Kudriavtsev, VA., ed. 1981. The permafrost study.
Moscow University Press, Moscow (in Russian).
Merser, J.H. 1978. West Antarctic sheet and CO2 green-
house effect: a threat of disaster. Nature 271:321-325.
Oechel, W.C. and C.H. Riechers. 1986. Impact of
increasing CO2 in natural vegetation, particularly the
tundra. Clim. Veg. Interact. Proc. Workshop, Greenbelt,
Mg, Boulder, Colorado, pp. 36-42.
Oeschger,H.andR.Stauffer. 1985. Review of the history
of atmospheric CO2 recorded in ice cores. In: Proceedings
Sixth ORNL Life Science Symposium, edited by I.R.
Trabalka, D.E. Reichle. Springer Verlag.
Parinkina, O.M. 1971. On microbiological characters of
some Western Taimyr soils. In: Biogeocenoses and Pro-
ductivity of Taimyr Tundra. Nauka Press, Leningrad (in
Russian).
Pavlov, A.V. 1979. Landscape's thermophysics. Nauka.
(in Russian).
Rotty,R. 1983. Distribution of and changes in industrial
carbon dioxide production. J. Geoph. Res. 88:1301-1308.
Scujins,J.J. 1967. Enzymes in soil. In: Soil biochemistry,
edited by A.D. McLaren and G.H. Peterson. Marcel
Dekker, NewYork, pp. 371-414.
Semiletov, I.P., N.I. Barkov, AM. Gusev, N.M. Pozdnyakov,
and V.Ya. Lipenkov. 1989. The CO2 paleovariations in
Antarcticicecore.DokladiAkademiiNaukUSSR309:196-
199 (in Russian).
Sioroechkovsky, E.E. 1986. Northern deer. Agropromizdat,
Moscow, p. 256 (in Russian).
Tucker, C.Y., I.Y. Fung, C.D. Keeling, andR.H. Hammon.
1986. Relationship between atmospheric CO2 variations
and a satellite-derived vegetation index. Nature 319:195-
199.
Zavarzin, GA. and I.J. Clark. 1987. Biosphere and
climate: biologist's view. Priroda. 6:65-77 (in Russian).
Zimov, SA. 1990. Man and Northern Nature: the
harmony of contrasts. Vestnic Akademil Nauk USSR,
2:118-132 (in Russian).
Zimov, S A. and V.I. Chuprynin. 1989. Stable states of
North-East Asian ecosystem. Dokladi Academii Nauk,
USSR 308(6):1510-1514.
33
-------
PLANETARY MAXIMUM CO2 AND ECOSYSTEMS OF THE NORTH
NOTES
1. This process is well-known to people who work on land. The greenhouse effect is provided not only by means of glass
or films, but also with the use of self-heating soil if it contains a lot of organic matter.
2. The foregoing phenomenon exists not only hi the North. At middle latitudes (near Moscow), photosynthesis is more
than in the North, but summer heat absorbed by soil is about 1.3 x 104 kcal/m2.
34
-------
ARCTIC ATMOSPHERIC CO2 BIMODAL DISTRIBUTION
Igor P. Semiletov, Sergey A. Zimov, Sergey P. Daviodov,
Yuri V. Voropaev, and Sergey F. Prosiannikov
ABSTRACT
The observed seasonal cycle of atmospheric CO2 closely follows the biospheric activity of the Northern Hemisphere. The
de-trended atmospheric CO2 time series from monitoring stations such as Alert, Mould Bay, Pt. Barrow and Sable Island,
shows a very broad maximum in winter and a sharp minimum in late summer. During the period of broad maximum,
measurements also show a bimodal feature, with a relative minimum in the early spring. The causes of these features are
as yet uncertain. It has been suggested that the atmospheric CO2 bimodal winter distribution reflects the course of the biotic
carbon flux from the active soil layer, based on the existence of a deep, well-aerated dry layer and on the existence of soil
microbe-decomposers, which are generally adapted to the cold-dominated environment. The present study shows that the
mean averaged by different types of landscapes in northeastern Siberia is about 100 g C/m2/winter. The range ofCO2flux
change is in good agreement with the varied double-peak or near double-peak features in the high-latitude belt.
INTRODUCTION
Recent study of the de-trended atmospheric
CO2 time series from Arctic stations such as
Alert, Mould Bay and Pt. Barrow shows a
prominent seasonal cycle with a broad maxi-
mum in winter and a sharp minimum in late
summer. The amplitude of the cycle is about 15
to 18 ppm. During the period of broad maxi-
mum, the time series of CO2 also shows a
bimodal feature, with a relative minimum in the
early spring. This feature is, however, not
present every year (Higuchi et al., 1990).
Double-peak or near double-peak features at
the winter-spring maxima are observed in all
the Canadian measurements, with signal dilu-
tion towards the south (Wong et al., 1984). It is
assumed that this effect probably reflects the
influences of biological activities, seasonalities
of air-sea fluxes, and the advection of anthropo-
genic CO2.
In this paper, we continue the study of Arctic
CO2 sources and sinks that is based on Barrow
and Arctic Ocean data (Kelley and Gosink,
1979). The resulting CO2-exchange data ob-
tained from northeast Siberia are used jointly
with previous data from the CO2 double-peak
interpretation.
METHODS
The study of CO2 emission from the soil was
carried out from December, 1989, until Janu-
ary, 1991, at the northeastern station situated
about 100 km south of the Arctic coast (70°N,
160°E). Within a relatively small area sur-
rounding the station containing the river valley,
the timberline, meadow tundra and a range of
35
-------
ARCTIC ATMOSPHERIC CO2 BIMODAL DISTRIBUTION
hills and mountains, all the main landscapes
characteristic of northeastern Siberia can be
found.
A detailed description of the methods of the
CO2 flux study is presented by Zimov et al. (this
volume).
RESULTS
The CO2 concentration gradient across the
snowpack, monitored during the winter, shows
that subnivean CO2 was continually higher than
ambient CO2 throughout the periods Decem-
ber, 1989, to late April, 1990, and middle Octo-
ber, 1990, to middle January, 1991. Monthly
mean values for the CO2 gradient, temperature
profile and snow depth are plotted in Figure 1.
The CO2 gradient in parts per million by vol-
ume (ppmv), as used in the discussion, refers to
the concentration differential between the
ground and ambient air along the transects.
The CO2 emission was calculated using values
obtained for the CO2 gradient and the transfer
coefficient. Monthly mean values for CO2 flux,
g C /m2, are presented numerically above the
curve in Figure 1. The mean CO2 gradient of
approximately 10 ppm corresponded to a flux of
4.63 g C /m2/month at a snow depth of 19 cm
and to a flux of 2.31 g C /m2/month at a snow
depth of 38 cm. The foregoing flux values are
based on a CO2 transfer coefficient of 0.63 cm2/
s. The winter CO2 data were considered jointly.
The data plotted in Figure 1 present the aver-
age CO2 gradient and corresponding emission
from all observed locations, which included
areas with shrubby-moss plants, lichen-mixed
grass plants and high-productivity mixed grass-
cereals meadows. Detailed information on all
C02
ppmv
80 H
60-
40-
20-
-10-
-30-
t°C
11.5
.
' s/s.
8.3
1990
.•I
4.1
•„ -n .
3.7
: in..
6.9
IV
-? v
30.0
x
20.9
• XI
14.7
XII
11.7
1991
'/ s 's
v.\t
Figure 1. The CO2 concentration gradient (ppmv) across the snowpack monitored during the winter; the CO2 flux from
tundra soil (g C /m2/month) is indicated numerically; the snow depth data(cm), is indicated with a dashed line, and the
temperature range (°C) with a solid line.
36
-------
I.P. SEMILETOV, S.A. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
locations of the study was presented previously
(Zimov et al., this volume).
The release of CO2 from tundra soils shows a
fall maximum of 30 gC/m2 in October, 1991. It
decreased with air temperature, dropping to
11.7gC/m2inJanuary, 1991. The same relative
decrease of CO2 flux was obtained from De-
cember, 1989 - January, 1990, with this ten-
dency continuing throughout the period of Feb-
ruary to April, 1990.
Plots of the data indicated that values of the
CO2 gradient for December, 1989 - January,
1990, increased about three times during the
winter of 1990/1991 in comparison with the
winter of 1989/1990. The corresponding values
of CO2 flux increased about 20 to 25%. This
may be related to warmer temperatures and
increased snow-fall during the winter of 1990/
1991. These factors provided higher tempera-
tures at the soil layer that can amplify soil biotic
activity. In general, it should be noted that the
winter-observed CO2 gradient and CO2 flux
show fall maxima. The tendency toward spring
maxima occurs only late in April, 1991.
DISCUSSION
The genesis of double-peak or near double-
peak features at the late fall-spring maxima,
may be considered for all of the Alaska and
Canadian measurements.
The release of CO2 from tundra was investi-
gatedpreviously (Coyne and Kelley, 1971; Coyne
andKelley, 1974). It was found from laboratory
measurements that CO2 was released during
the soil-moisture phase change when cores of
tundra soil were freezing. It was hypothesized
that CO2-rich air dissolved in the soil solution is
expelled from the freezing layer into the unfro-
zen portion of the active layer as the soil begins
to freeze in the fall. Part of this air escapes to
the atmosphere immediately, but the remain-
der is contained under pressure and slowly
escapes to the subnivean atmospheric environ-
ment during winter. As the soil begins to warm
differentially from the surface in spring, but
while temperatures are still below -10 °C, pock-
ets of trapped gas may be released quite sud-
denly at the snow-soil interface.
Coyne and Kelley (1971) showed, based on
laboratory experiments, that 500 to 90,000
1 CO2/ha, or 2.7 x lO"2 g C/m2 to 4.8 g C/m2,
could be released to the atmosphere during the
freeze-up. This small quantity could influence
concentrations of atmospheric CO2 near ground
level by a few tenths of a ppm (Halter and
Peterson, 1981). Nevertheless, it is supposed
that the most likely cause of the sudden release
of CO2 during late spring in the Arctic appears
to be the release of trapped gas in the tundra by
physical processes. However, the contribution
of respiratory products and extracellular en-
zyme systems to the injection of CO2 to the
subnivean atmosphere throughout the winter
and spring is considered unresolved (Kelley
and Gosink, 1988).
The foregoing physical mechanism of winter
CO2 emission agrees with the winter bimodal
CO2 distribution, but observed atmospheric
effect is negligible.
Information regarding synoptic meteorological
features and the vertical structure of the atmo-
sphere was used in addition to air-mass trajec-
tory analyses. This study suggests, that the
anthropogenic source regions of Eastern Asia
and North America do not contribute signifi-
cantly to excess winter CO2 at Barrow (Halter
and Harris, 1983). It is difficult to explain the
bimodal distribution in terms of the evolution
of the atmospheric circulation in the Arctic
where the absolute winter maxima of the atmo-
spheric CO2 concentration is observed. Also, it
is important that the Arctic, particularly hi the
winter, is a comparatively closed cyclonic cen-
37
-------
ARCTIC ATMOSPHERIC CO2 BIMODAL DISTRffiUTION
ter belowthe tropopause (Vowinkel and Orvig,
1970). In consequence, relatively small masses
of air are injectedinto or escape from the winter
Arctic troposphere so that it remains a well-
mixed closed system. The CO2 added from
Arctic sources tends to stay there for longer
periods of time, as compared with mixing at
latitudes. Thus, it appears that Arctic Ocean
and terrestrial CO2 sources are a significant
part of the cause of the large variations in the
atmospheric loading of CO2 as recorded at
Barrow, Alert and Sable Island and as derived
from satellite data (Kelley and Gosink, 1979;
Fung et al, 1987). This is supported by a
previous study by Bolin and Bischoff (1970)
indicating that a significant portion of the sur-
face source tends to remain in the lower part of
the troposphere and would thus enhance these
proposed atmospheric CO2 concentration fac-
tors. Also, aircraft data by Gosink and Kelley
(1979) show the repeated presence of a CO2-
enhanced layer below 1 km altitude in the
Arctic. This study of Arctic CO2 sources and
sinks shows that tundra soil is one of the largest
winter sources of CO2,7 Tg CO2, and annual
sea ice is a source of 849 Tg CO2. The total net
winter input is about 905 Tg CO2, or 0.247 Pg C.
The summer tundra soil source, 86 Tg CO2, is
more than 12times the winter tundrasoil source.
This is the second largest summer CO2 source
after tundra ponds / lakes, 111 Tg CO2. It was
shown that the largest summer sink is not vas-
cular plants, 145 Tg CO2 because the ocean
surface absorbs about 778 Tg CO2. In conse-
quence, the Arctic is a net annual source of 180
Tg CO2, or about 0.050 Tg C The foregoing
study shows that the natural Arctic factors can
accountfor6ppmof the 12ppm winter-summer
variations in the Arctic (Kelley and Gosink,
1979). It should be noted that a recent study
shows the season amplitude to be more than 20
ppm (Fung et al., 1987).
There is no clear understanding of CO2 migra-
tion in the Arctic. The present study of winter
CO2 emission from tundra soil shows the great-
est CO2 emission to the atmosphere. So, in the
cold period from January, 1990, until Decem-
ber, 1990, about 100 g C was injected into the
atmosphere. This winter (January-April, Octo-
ber-January), CO2 flux would increase the aver-
age CO2 concentration by approximately 25
ppm, assuming that there is no air exchange in
the atmosphere between the tundra area and
other regions.
An estimate of the percentage of the winter
CO2 amplitude between the period of broad
maximum (October-January; Figure 1) and
minimum in the early spring (February-March)
in terms of the CO2 concentration in high lati-
tudes could be obtained by assuming that there
is no air exchange in the Arctic atmosphere.
For an air column with a base of 1 m2 and with
the CO2 concentration of 350ppm (total amount
about 1,440 g C), the changes of the winter CO2
emission correspond to approximately 2.0 ppm
for December, 1989,11.5 g C/m2, and for March,
1990, 3.7 g C/m2. The difference between
carbon flux in October, 1990, and January,
1991, corresponds to 18.3 g C/m2, or about 4.5
ppm. This range of CO2 change is in good
agreement with the variable double-peak or
near double-peak features at the winter-spring
maxima in all high-latitude measurements.
Thus, we can suggest that the causes of these
features probably reflect the influence of CO2
soil emission.
We suggest that the origin of winter CO2 soil
emission is connected with biotic activity in
tundra soil (Zimov et al., this volume). The CO2
might have been produced from previously
accumulated organic matter. Ten to 27 percent
of the biospheric carbon store is concentrated
in the tundra and boreal forest ecosystems
(Bolin, 1980; Schlesinger, 1989). Despite the
low annual input of energy to tundra, the active
layer of the soil is rich in carbon (9-20 kg/m2 to
a depth of 20 cm) and contains much energy.
38
-------
The energy contents of wet meadow soils range
from 13.8 kJ/g soil in the 0- to 2-cm horizon.
These resources of carbon and energy could
sustain substantial microbial production (Flana-
gan and Bunell, 1980). It is known that under
aerobic conditions, microbes decompose sub-
strate, complex organic molecules to end prod-
ucts that are primarily inorganic (carbon diox-
ide, water and minerals).
It should be noted that the same CO2 gradient
course was obtained at North Meadow Lake on
a beach ridge 2 km from the Arctic Ocean
(Kelley et al., 1968). The snowpack contained
above-ground, coarse-grained snow with a thin
ice layeras lead and 30-60 cmofwindhardpacked
snow. Thus, the CO2 transfer coefficient must
have been very low, but the bimodal winter CO2
gradient course was the same as presented
herein.
The review of the previous study showed that
throughout the entire year, CO2 escaped from
the tundra surface. However, a seasonal oscil-
lation probably was produced principally by
land plants, the Arctic Ocean and annual sea
ice. A study of the amplitudes and phases at
different latitudes suggested Bolin and Keeling
(1963) hypothesis: summer decrease in CO2 is
due to the regional and local weather patterns
and sea vertical mixing. However, the causes of
bimodal winter distribution are difficult to ex-
plain in terms of anthropogenic activities, the
evolution of the atmospheric circulation, air-
sea fluxes, and the photosynthesis-respiration
cycle of land plants.
We suggest that the atmospheric CO2 bimodal
winter distribution reflects the course of the
biotic carbon flux from the active soil layer.
This suggestion is based on the existence of a
deep, well-aerated dry layer (Zimov et al., this
volume) and on the existence of soil microbe-
decomposers, which are generally adapted to
the cold-dominated environment (Flanagan and
I.P. SEMILETOV, S.A. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
Bunell, 1980). Most physical and chemical
changes that occur in and around the decom-
posing substrate cannot be separated from the
effects of microbial activity. Winter measure-
ments of the rates of evolution of CO2 may
represent the rate of mineralization of complex
organic compounds to CO2. Our preliminary
study showed that the mean CO2 emissions
averaged by different types of landscapes in
northeastern Siberia about 100 g C/m2/winter
season, 1990. Due to the bio-oxidation of lOOg
C/m2, about 800 kcal might be released. This
mechanism of soil self-heating might provide
sufficient warming to stimulate the continuous
activity of soil biota during winter. The most
optimal conditions for aerobic activity occurred
during the fall. In consequence, the maxima of
CO2 flux was obtained. The warm, dry aerated
layer cools down during the coldest months, and
that the CO2 flux decreased from the late au-
tumn-winter maxima until March. The spring
carbon maxima might be related partly to the
increasing incoming solar radiation and the
release of trapped gas (Kelley and Gosink,
1988). It should be noted that respiratory
products and extracellular enzyme systems may
contribute to the injection of CO2 to the sub-
nivean atmosphere in a way that is still unre-
solved. Total CO2 winter emission from soil
may be caused by the complicated interaction
of biotic and physical processes. Note that
additional CO2 might be realized during the
spring thaw by emission from numerous lakes
and ponds, which are enriched by CO2 under
the ice cover, as was shown by Kelley and
Gosink (1988). The same effect might occur in
the Arctic Ocean due to the increase of gas
permeability through ice. Insufficient winter
data did not allow the estimation of CO2 emis-
sion in detail.
CONCLUSIONS
The Arctic ecosystem contains vast quantities
of carbon as soil organic matter and, depending
39
-------
ARCTIC ATMOSPHERIC CO2 BIMODAL DISTRIBUTION
on future climatic conditions, has the potential
to act as a major source or sink for atmospheric
CO2. The Arctic ecosystem is one of the few
ecosystems capable of long-term increased car-
bonstorage. Onthe other hand, increases in the
depth of the active layer, melting of the perma-
frost and/or decreases in soil moisture could
result in increased rates of soil decomposition
and net CO2 efflux from the tundra (Oechel,
1990). This effect might be amplified by the
deforestation in the North. Our study indicates
that northeastern Siberia is currently losing
substantial amount of carbon, with the winter
mean rate of net CO2 loss from the tundra on
the order of 100 g C/m2 and more. This rate of
loss, if occurring over the total tundra area of
about 8 x 106 km2 (Schlesinger, 1980), could
account for the net loss of 0.8 Pg C/cold season.
Because of the presence of lakes and ponds
(approximately 10-20% of the total area), the
total area of tundra source mightbe decreased.
In consequence, approximately 0.6-0.7 pg C
was released from northern soil during winter,
1990. In accordance with Bolin et al. (1977), the
area of mountainous tundra is about 2.9 x 106
km2 and the area of grey soils tundra, 3.8 x 106
km2. Winter CO2 emission from the mountain-
ous tundra is approximately 0.29 Pg C and from
the grey soil tundra, 0.35 Pg C, without lakes
and ponds. Both estimations are in good agree-
ment. This is a crude approximation of actual
CO2 emissions, but this value is of one order
with the estimation for the net loss of 0.1 to 0.2
xlOwgC/yearfromtassocktundraonly(Oechel,
1990). Note thattheCO2 tundra signal mightbe
amplified by means of slow winter respiration
of trees and CO2 soil release in the boreal zone.
This possibility is supported by the large litter
content in the soil and a similar regime of solar
radiation, evaporation, precipitation and tem-
perature (Bolin, 1980).
According to some investigators, the amount of
atmospheric CO2 could double by early in the
next century, leading to an average 2 to 3 °C
warming of the world's climate, an increase that
might be three to four times greater at the poles
than in the tropics (Washburn, 1980). The
climatic effects of a doubled CO2 concentration
in the Arctic tundra are expected to include a
rise in summer temperature from 4 to 8 °C.
Consequently, the ecological perturbation ex-
pected to accompany climate warming may
reverse a net annual storage of 0.2 Pg C to a net
release of the same magnitude, or greater
(Houghton et al., 1987). These foregoing esti-
mates agree with our present estimates. On the
other hand, Loomis predicts that 300 to 400 Pg
C might be released to the atmosphere with
each 1 °C increase in global temperature. Pre-
sumably, this release would occur predomi-
nantly in cold regions with large carbon accu-
mulations (e.g., tundra and boreal forests) as a
result of permafrost melting. This estimate is
considered as the possible upper limit of carbon
release from soils (Schlesinger, 1984).
The obtained and predicted rates of net CO2
loss from the tundra differed by the order of one
to two. It reflects the level of present knowledge
of the global CO2 budget. It was found that the
ocean is not a major sink for the CO2 released
as a result of fossil-fuel burning and land-use
modification (Tans et al., 1990). The mecha-
nism of this sink is unknown; its magnitude
appears to be as large as 2.0 to 3.4 Pg C/year,
depending on the sources in the tropical and the
boreal and tundra regions. Northern soil plays
a role in a way that might provide a bimodal
winter distribution of CO2 flux, with a relative
minimum in the early spring.
40
-------
I.P. SEMILETOV, SA. ZIMOV, S.P. DAVIODOV, Y.V. VOROPAEV, AND S.F. PROSIANNIKOV
REFERENCES
Bolin, B. and C.D. Keeling. 1963. Large-scale atmo-
sphericmixing as deducedfrom tie seasonaland meridional
variationsof carbon dioxide. J.Geophys.Res.68(13):3899-
3920.
Bolin, B. and W.Bischoff. 1970. Variations of the carbon
dioxide content of the atmosphere in the northern hemi-
sphere. Tellus 22:431-442.
Bolin, B., E.T. Degens, P. Divigneand, and S. Kempe.
1977. The global biogeochemical carbon cycle. In: The
Global Carbon Cycle, edited by B. Bolin, E.T. Degens, S.
Kempe, P. Ketner. SCOPE 13, John Wiley andSons, N.Y.,
pp. 1-56.
Bonn, B. 1980. Climatic changes and their effects on the
biosphere. WMO-No.542, p. 49.
Coyne, P.I. and J. J.Kelly. 1971. Release of carbon dioxide
from frozen soil to the Arctic atmosphere. Nature
234(5329):407-408.
Coyne, P.I. and J.J. Kelly. 1974. Variations in carbon
dioxide across an Arctic snowpack during spring.
J.Geoph.Res. 79(6):799-802.
Flanagan, P.W. and F.L. Bunnell. 1980. Microflora
activities and decompsition. In: An arcticEecosystem: the
Coastal Tundra at Barrow, edited by J. Brown, P.C.
Miller, L.L. Tieszen, F.L. Bunnell. US/IBP Synthesis
Series 12, Chapter 9, Dowden, Hutchinson and Ross Inc.,
Stroudsburd, Pennsylvania, pp. 291-334.
Fung, I.Y., CJ. Tucker, and K.C. Prentice. 1987. Appli-
cation of advanced very high resolution radiometer veg-
etation index to study atmosphere-biosphere exchange of
CO2. J.GeopkRes. 92(D3):2999-3015.
Halter, B.C. and J.T. Peterson. 1981. On the variability of
atmospheric carbon dioxide concentration at Barrow,
Alaska, during summer. Atmos.Environ. 15:1391-1399.
Halter, B. and J.M. Harris. 1983. On the variability of
atmospheric carbon dioxide concentration at Barrow,
Alaska, during winter. J.Geophys.Res. 88(Cll):6858-6864.
Higuchi, K., S.M. Daggupaty, and N. Trivett. 1990.
Trajectory analysis of the atmosphere CO2 bimodal distri-
bution. In: Abstracts of the International Conference on
the Role of the Polar Regions in Global Change, edited by
G. Weller, Geophysical Institute, University of Alaska,
Fairbanks, p.115.
Houghton, RA., R.D. Boone, J.R. Fruci, T.E. Hobbie,
J.M. Melfflo, CA. Palm, B.J. Peterson, G.R. Shaaver,
G.M. Woodwell, B. Moore, D.I. Skole, and N. Myers.
1987. The flux of carbon from terrestrial ecosystems to the
atmosphere in 1980 due to changes inland use: geographic
distribution of the global flux. Tellus 396:122-139
Kelley, J.J., D.F. Weaver, and B.P. Smith. 1968. The
variation of carbon dioxide under the snow on Arctic.
Ecology 49(2):358-361.
Kelley, J J. and T A. Gosink. 1979. Gases in sea ice, final
report, appendix B3, contract N 00014-76 C-0331, Off. of
Naval Res., Arlington, Va., p.107.
Kelley, J J. and T A. Gosink. 1988. Carbon dioxide and
other trace gases in tundra surface water in Alaska. In:
SCOPE/UNEP sonderbandheft66,Hamburg, February,
pp.117-127.
Levin, I. 1987. Atmospheric CO2 in continental Europe-
an alternative approach to clean air CO, data. Tellus
39B:21-28.
Oeche, W.C. 1990. Effects of global change on net
ecosystem carbon flux of Arctic tussock tundra. In:
Abstracts of the International Conference on the Role of
the Polar Regions in Global Change, edited by G. Weller.
June 11-15, 1990, Geophysical Institute, University of
Alaska, Fairbanks, p.101.
Pearman, G.I. and P. Hyson. 1980. Activities of the global
biosphere as reflected in atmospheric CO2 records. J.
Geophys. Res. 85:4457-4467.
Schlesinger, W.H. 1989. Soil organic matter: a source of
atmospheric CO2. In: The Role of Terrestrial Vegetation
in the Global Carbon Cycle: Measurements by Remote
Sensing, edited by G.M. Woodwell. John Wiley and Sons
N.Y.
41
-------
ARCTIC ATMOSPHERIC CO2 BIMODAL DISTRIBUTION
Tans,PJP., I.Y. Fung, and T. TakahasH. 1990. Observa-
tional constraints on the global atmospheric CO2 budget.
Science 247:1431-1438.
Vowinkel, E. and S. Orvig. 1970. The climate of North
America. In: World Survey of Climatology, vol.14,
Chapter 3. Elsevier Scientific Pub.Co., p.129 and p.186.
Washburn,AX. 1980. Focus on polar research. Science
209(4457):643-52.
Wong, C.S., Y.H. Chan, J.S.Page.and R.D.BeUegay. 1984.
Trends of atmospheric CO2 over Canadian WMO back-
ground station P, Sable Island, and Alert. J. Geophys. Res.
89:9527-9539.
Zimov, S.A., S.P. Davidov, Yu.V. Voropaev, S.F.
Prosyannikov, and I.P. Semiletov. 1991. Planetary maxi-
mum CO2 and ecosystems of the North. This volume.
42
-------
MARINE HUMIC ACIDS AS AN IMPORTANT CONSTITUENT
OF THE DISSOLVED ORGANIC CARBON FLUX
IN THE BERING AND CHUKCHI SEA ECOSYSTEMS
Irina V. Perminova and Valeriy S. Petrosyan
ABSTRACT
The oceans play akey role in maintainingthepresent-day levelofCO2concentration in the atmosphere through active gaseous
exchange at the sea-air interface. The most intense transport of CO2 from the surface to the deep ocean occurs in subarctic
sea ecosystems. To further understand the role of dissolved organic carbon flux in carbon cycling, the humic acid content
distribution patterns in the Bering/Chukchi Seas were considered in the context of physical and biological features intrinsic
to these high-latitude regions. An earlier investigation using the direct fluorometric method indicated that humic acid
concentration was two to three times higher in Chukchi Sea waters than in the photic layer of the Bering Sea. The Humic
acid concentration gradient exhibited strongcoupling to hydrographic as well as biological conditions in the Bering/Chukchi
Seas ecosystem providing a basis to consider HA distribution as a quasi-conservative parameter reflecting long-term
production and decomposition coupling. Datapresented show that carbon cycling in the Bering Sea has global significance.
INTRODUCTION
The oceans play a key role in maintaining the
present-day level of the CO2 concentration in
the atmosphere through the active gaseous
exchange at the sea-air interface. Deep-water
mass formation due to surface cooling could be
a major transport mechanism of atmospheric
CO2 to the deep ocean, especially in high-
latitude seas. For example, the Bering Sea and
Arctic Ocean have extended shallow shelves
that allow rapid winter cooling followed by
sequestering of the surface water supersatu-
rated with CO2 to the adjacent deep basins
(Walsh et al., 1989).
A second transport mechanism, referred tp as a
"biological pump" (Longhurst and Harrison,
1989), occurs through the sinking of particulate
organic matter (POM). This global flux pro-
vides an annual vertical transport of 0.6-2.6 Gt
of organic carbon from the surface to the deep
ocean (Jahnke, 1990). So far, it has been
considered as a major input mechanism of
organic material to the deep sea (Bacastow and
Meier-Reimer, 1990). But recent investiga-
tions of dissolved organic carbon (DOC) verti-
cal profiles in the Northern Pacific using a new
method of high-temperature catalytic oxida-
tion, developed by Sugimura and Suzuki (1988),
have shown the existence of a near 1:1 correla-
tion between decreasing DOC and increasing
apparent oxygen utilization (AOU) with depth.
It suggests that most of the water-column respi-
ration is supported by the direct transit of DOC
from the photic layer and not by sinking particu-
late material. This global DOC flux is esti-
mated at 1-3 Gt C / yr on the basis of polar
deep-water formation rates of 20xl06 m3/ s
(Toggweiler, 1988; Christensenetal, 1989) and
43
-------
MARINE HUMIC ACIDS AS AN IMPORTANT CONSTITUENT OF THE DISSOLVED ORGANIC CARBON FLUX
average DOC concentration in surface waters
of 100-300 umol C / 1 (Druffel et al., 1989;
Sugimura and Suzuki, 1988). Thus, the role of
DOC input could be as important as that of
POM flux.
Moreover, the high concentrations of DOC as
measuredby Sugimuraand Suzuki (1988) could
be interpreted as indicative of a new large class
of organiccompounds inseawater (Toggweiler,
1989; Christensen et al., 1989), which are bio-
logically degradable, but the decomposition
rate is on the order of decades or centuries.
These compounds are located mostly in the
surface (0-200 m) layer; they are younger than
refractory dissolved organic carbon (DOCyy)
according to radiocarbon signatures (Druffel et
al., 1989); and have higher molecular weight
(over 20,000 dalton) (Sugimura and Suzuki,
1988).
The properties of these newly discovered
biopolymers (DOC^,) allow the speculation
that DO GUV is diagenetically downstream of
DOCjjTc. This could be the long-sought pool of
biopolymer precursors of humic acids (HA)
(Hedges, 1988).
To further understand the role of the DOC flux
in global carbon cycling, the distribution and
composition of both of these DOCpools, that is,
the refractory (mostly of humic origin) and the
biodegradable (HAprecursors) shouldbe thor-
oughly studied. This problem is of special
significance in subarctic ecosystems, where the
most intense transport of CO2 from the surface
to the deep ocean occurs.
GLOBAL SIGNIFICANCE OF CARBON
CYCLING WITHIN THE BERING
AND CHUKCHI SEAS
According to the latest estimates, during the
growing season of phytoplankton 0.19 Gt and
8.2 Mt C are produced over the Bering and
Southern Chukchi Seas, respectively (Zeeman,
1992), demonstrating the great importance of
these regions in the global carbon flux. A
substantial portion of this fixed carbon is trans-
ported northward through the Bering Strait
into the adjacent deep basins of the Arctic
Ocean. Recent independent assessments based
on the AOU demands (Walsh et al., 1989) and
the amount of carbon required for dissolution
of calcium carbonate within the basins' halo-
cline (Anderson et al., 1990) suggest a total
carbon input into the Arctic interior of 3.2-115
Mt C/yr. Under present rates of "new produc-
tion" within waters under the ice pack of the
Arctic basins and the shallow Arctic shelves,
such a carbon budget can be balanced only in
case of the input of 0.1 Gt C fixed south of the
Arctic ocean and entrained north through the
Bering Strait, with a concomitant import of
"new" dissolved nitrogen to the Arctic shelves
(Walsh et al., 1989).
The nutrient-laden Bering Strait influx of > 1.0
Sv transits the Arctic Ocean directly as a subsur-
face layer, going under the ice cap (Aagard et
al., 1985). In this transpolar drift, it entrains
about twice its volume of dense and saline water
produced on the shallow Arctic shelves during
seasonal ice formation (Aagard et al., 1981). In
six years, these modified North Pacific waters
crossing Fram Strait enter the Iceland Sea,
contributing to the formation of the North
Atlantic Bottom Water (NABW) (Swift, 1984).
Thus, creation of waters forming North Atlan-
tic Deep Water (NADW) directly depends on
advection of modified Pacific water.
In terms of the present anthropogenic CO2
release of 5.0 Gt C/yr to the atmosphere
(IPCQ1992), an Arctic sink of 0.1 Gt C/yr is
small. But keeping in mind the projected green-
house warming of 4-5° C in this region by 2050,
followed by the melting of the Arctic ice pack
(IPCC, 1992), a future carbon sink may increase
up to 0.5 Gt C/yr as a result of the increase in
primary production over the Bering Sea - Arctic
shelves (Walsh et al., 1989). Besides, through
44
-------
the teleconnection of the Pacific and Atlantic
Oceans via Bering and Fram Straits, such a
change in primary production could greatly
affect today's ventilation of the World Ocean,
which has a strong feedback with global climate
formation.
CARBON FLUXES THROUGH THE
BERING STRAIT
An estimation of the total carbon flux and its
components through the Bering Strait allows
further understanding of the role of subarctic
sea ecosystems in global carbon cycling. Ac-
cording to Zeemans data (1991), 0.82 Gt dis-
solved inorganic C are transported northward
annually through the Bering Strait. But this
estimate could not be considered as a total
carbon flux, taking into account a very high
productivity characteristic of the region under
study. This is demonstrated by a number of
quantitative estimates based on the data on
phytoplankton particulate carbon (POC)
(Walsh etal., 1989; Zeeman, 1992) and the total
POC and DOC content in the water column in
the strait region (Loder, 1971; Lyutsarev et al.,
1988; Glebov et al., 1992; Pershina, 1992), as-
sumingthe average flow-rate through the Bering
Strait of about 1 x 106 m3/s (Walsh et al., 1989).
These estimates are presented in Table 1.
A salient point to consider here is that there is
a rather good coincidence of estimates within a
single approach, but based on different measur-
Table 1. Estimates of POC and DOC fluxes transported
through the Bering Strait.
Flux-value
(MtC/yr)
13
19
13
1.3
3.2
. 42
29-32
Method of POC and DOC
determination
POC by dry combustion
POC by wet combustion
POC by wet combustion
POC by phytoplankton biomass
POC by phytoplankton biomass
DOC by wetcombusion
DOC by HA content
Reference
Glebov etal., 1992
Loder, 1971
Lyutsarev et al., 1988
Walsh etal., 1989
Zeeman, 1992
Loder, 1971
Pershina, 1992
I.V. PERMINOVA AND V.S. PETROSYAN
ing procedures. The low magnitudes of POC
flux, based solely on the content of phytoplank-
ton particulate carbon, are related to underes-
timation of the detritus input into the POC flux.
Subtraction of the value of phytoplanktonflux
of 1.3 Mt C/yr from the total POC flux of 13-19
Mt C/yr allows an estimate of detritus contribu-
tion to the total POC flux of 12-18 Mt C/yr .This
magnitude is tenfold higher than for phyto-
plankton flux, demonstrating a major role of
detrital material in the particulate flux trans-
porting through the strait. North of the Bering
Strait, in the weak-flow regime of the Chukchi
Sea, a substantial, but yet unknown, portion of
this particulate material sinks to the bottom
forming a large carbon depocenter with a high
(6:l)C:Nratio(GrebmeieretaL, 1989), which
fuels enormous benthic populations here. The
remainder of the particulate flux finds its way to
the Arctic Ocean, contributing to "new produc-
tion" in the polar basins.
In connection with the new concept of the role
of DOC in supporting deep-water metabolism,
it is of interest to estimate the role of DOC flux
in the northward carbon transport through the
Bering Strait. According to Table 1, DOC flux
of 29-42 Mt C/yr is two to three times the total
POC flux. Considering the two or three times
higher DOC concentrations measured by
Sugimura and Suzuki (1988) and recently con-
firmed by other investigators (Druffel et al.,
1989; Cauwet etal., 1990), the given value could
be a low estimate. This means that the average
concentration through the water column of 1.22
mg/1 (Loder, 1971) might be increased at least
twice, yielding an annual DOC flux of 80 Mt C/
yr. In addition, Pershina's (1992) results on HA
distribution allow inclusion of refractory or-
ganic carbon into the corrected total DOC flux.
Based on the average HA concentration in the
strait region of 2.0 mg/1 (Pershina, 1992), the
flux of refractory organic carbon is about 23 Mt
C/yr, or 29% of the total DOC flux. This value
is in good agreement with data of Sugimura and
Suzuki (1988), who discovered 35% refractory
45
-------
MARINE HUMIC ACIDS AS AN IMPORTANT CONSTITUENT OF THE DISSOLVED ORGANIC CARBON FLUX
carbon in the DOC samples of the North Pa-
cific.
As it is exported farther north, the DOC is
partitioned over the shallow ( ~ 40 m) extended
area (~lx!012 m2) of the Chukchi/East Sibe-
rian Seas shelves and maybe entrained into the
pycnocline of the Arctic Ocean during intense
deep-water formation. Under the annual 1 Sv
influx of saline-dense (<33.5% ) shelf water to
the Arctic interior (Aagard et al., 1981), it can
provide the total DOC input of 63 Mt C/yr,
consisting of 45 Mt C/yr relatively biodegrad-
able DOC and 18 Mt C/yr refractory (humic
acids) DOC with residence time on the order of
600-6000 years (Druffel et al., 1989;
RomankevichandLyutsarev, 1990; Skopintsev,
1981). Due to the longevity of humics in com-
parison with the Arctic basin's deep-water resi-
dence time of 200 years (Aagard et al., 1981),
these compounds could be transported later-
ally over long distances providing homogenous
vertical DOC distribution below 1500 m.
To further understand the role of the DOC flux
and its components in the carbon cycling within
subarctic sea ecosystems, recent results on the
HA content distribution pattern in the Bering /
Chukchi Seas are considered below in the con-
text of major physical and biological features
intrinsic to these high-latitude regions.
HUMIC ACIDS IN THE BERING
AND CHUKCHI SEAS
The HA distribution in the Bering and Chukchi
Seas was investigated by the direct fluorimet-
ric method (Pershina, 1987) during the grow-
ing season (July - August) of 1988. The results
are presented in Figure 1. Maximum HA
concentrations of 75-80 g/m2 or 1.8-2.0 mg/1
were detected in the Bering Strait. High con-
centrations of 60-70 g/m2 or 1.5-1.7 mg/1 were
characteristic for the Chirikov Basin and the
Southern Chukchi Sea. As a whole, HA con-
centration in Chukchi Sea waters (except for
the northern-eastern part, adjacent to the East
Siberian Sea) was two to three times higher
than it was in the photic layer of the Bering Sea.
The results are rather consistent oceanographi-
cally. More than 60% of the flow passing
through the Bering Strait is dominated by cold
saline water from the Gulf of Anadyr (AW).
The remainder of water on the eastern side of
the strait is a mixture of Yukon River and
Southeastern Bering Shelf Water, termed
Alaska Coastal Water (AC), whereas on the
western side of the strait, it is a mixture of
Siberian Rivers and East Siberian Sea Water,
termed Siberian Coastal Water (SC) (Walsh et
al., 1989). At the stratified interfaces of AC and
SC with AW, referred to as eastern and western
front, respectively, where vertical mixing is in-
hibited, enhanced plant production should be
expected. Indeed, detailed study of nitrogen
cycling in the strait region by Walsh et al. (1989)
showed that the greatest rates of nitrogen up-
take of 1.5-3.0 mg-at N m2/hr (twice that found
in the AW) were observed in these boundary
water parcels. This nitrogen uptake would
result in a daily primary production of 2 to > 10
g C /m2/day at the edges of the AW (McRoy et
al., 1972)versus ~4gC/m2/day within Anadyr
Water and -0.33 g C /m2/day within Alaska
Coastal Water (Walsh et al., 1989). Such high
primary production implies high biosedimenta-
tion rates of phytodetrital material providing a
food source for large benthic populations south
of the Bering Strait as well as north. The
surficial sediments south of the Bering Strait
are characterized by low organic carbon con-
tent and a high C:N ratio exhibiting rapid
mineralization rates of incoming detritus in the
Chirikov Basin (Grebmeieretal., 1989). In the
Chukchi Sea, on the contrary, underlying sedi-
ments are characterized by high organic con-
tent, low C:N ratio (better food quality), high
sulfide concentration (as a result of sulfate-
reduction mineralization mechanisms) and in-
tense urea-fluxes (Walsh et al., 1989). Latter
conditions, concomitant with very high primary
46
-------
I.V. PERMINOVA AND V.S. PETROSYAN
61
165
Figure 1. Depth-integrated distribution of humic acids (g/m2).
47
-------
MARINE HUMIC ACIDS AS AN IMPORTANT CONSTITUENT OF THE DISSOLVED ORGANIC CARBON FLUX
production (at times > 10 gCm2/day), seemed
to be greatly contributed to the intense humifi-
cation processes, providing maximum HA ac-
crual in the region under consideration (Fig-
urel).
It should be pointed out that the axes of west-
ward and northward HA concentration gradi-
ent coincide very well with the western and
eastern fronts, respectively exhibiting strong
coupling to hydrographic as well as biological
conditions in the area. However, despite a good
agreement with the general features of a pri-
mary productivity distribution pattern, direct
comparison manifests a lack of correspondence
between values of primary production and HA
content in the same water parcel (Pershina,
1991; Zeeman, 1991). This finding appears to
be explained by considering the great differ-
ences in the residence time of HA (600-6000
years) (Skopintsev, 1981; Druffel et al., 1989)
versus the ephemeral state of algae populations
that produce them. As a result of such longev-
ity, the content of humic compounds could
serve as a quasi-conservative characteristic of
the water parcel, whereas the HA distribution
pattern could be interpreted as an indicator of
long-term physical-biological coupling in sea
ecosystems.
CONCLUSIONS
The data presented show that carbon cycling in
the Bering/Chukchi Seas has a global signifi-
cance. Through northward transport to the
Arctic Ocean of -50 Mt C/yr of POC and
DOC, respiration can accountfor nearh/50% of
the polar basin AOU demands. Of the total
carbon input, an estimated 30% of the refrac-
tory pool is derived from humic substances.
Humic acid distribution is coupled to the major
physical-biological processes of the Bering/
Chukchi ecosystem and is a quasi-conservative
parameter reflecting long-term production and
decomposition coupling. We can speculate that
the HA biopolymer precursors pool (termed
earlier as DOCHTC, or biodegradable DOC)
may have a direct correlation with the seasonal
changes in the primary productivity regime,
being dependent on the phytoplankton bio-
mass, which contained its chemical precursors.
In other words, this DOC pool could be trace-
able to the short-term variations in primary
productivity. If so, precise determination of
both of these DOC pools is of great importance,
providing means for further understanding the
mechanisms of diagenetic transformations of
biological material during humification pro-
cesses. For this approach to be useful, highly
reliable analytical methods of determination of
the total DOC pool and its constituents should
be used. Studying the DOCpool on the molecu-
lar level with ause of modern physical-chemical
techniques is a next important problem to be
solved here.
REFERENCES
Aagard,K.,L.K. Coachman, and E.G. Carmack. 1981. On
the halocline in the Arctic Ocean. Deep-Sea Res. 28:529-
545.
Aagard.K., J.H.Swift, and E.G. Carmack. 1985. Thermo-
haline circulation in the Arctic Mediterranean Seas. J.
Gcophys. Res. 90:4833-4846.
Anderson, L.G., D. Dyrssen, and E.P. Jones. 1990. An
assessment of the transport of atmospheric CO2 into the
Arctic Ocean. J. Geophys. Res. 95 (C2):1703-1711.
Bacastow, R. and E. Meier-Reimer. 1990. Climate
dynamics 4:95-125.
Cauwet, G., R. Sempere, and A. Saliot. 1990. Dissolved
organic carbon in seawater: confirmation of previous
underestimation. C.R. Acad. Sci Paris, Oceanographie
physique (Physical Oceanography) 311( II):1061-1066.
48
-------
Christensen, J.P., T.T.Packard, F.Q.Dortch.HJ.Minas,
J.C.Gascard,C.Richez,andP.G.Garfield. 1989. Carbon
oxidation in the deep Mediterranean Sea: evidence for
dissolved organic carbon source. Global Biogeochem.
Cycles 3:315-335.
Druffel, E.R.M., P.M. Williams, and Y. Suzuki. 1989.
Concentrations and radiocarbon signatures of dissolved
organic matter in the Pacific Ocean. Geophys. Res. Lett.
16:991-994.
Glebov, B.V., V.I. Medinets, and V.G. Solovjev. 1992.
Intensity of biosedimentation processes. In: Results of the
Third Joint US-USSR Bering and Chukchi Seas Expedi-
tion (BERPAC), summer 1988. PA. Nagel, editor. US
Fish and Wildlife Service, Washington, DC, pp.224-230.
Grebmeier, J.M., C.P. McRoy, and H.M. Feder. 1989.
Pelagic-benthic coupling on the shelf of the northern
Bering and Chukchi Seas. HI. Benthic food supply and
carbon cycling. Mar. Ecol. Progr. Ser. 53:79-91.
Hedges J.J. 1988. Polymerization of humic substances in
natural environments. In: Humic Substances and Their
Role in the Environment, edited by R.F. Christman. John
Wiley and sons, New York, pp. 45-58.
Intergovernmental Panel on Climate Change (IPCC).
1990. T.J. Houghton, G.J. Jenkins, and J.J. Ephraums,
editors. Climate change, the IPCC scientific assessment.
Working group 1 report, WMD & UNEP Univ. Press,
Cambridge, UK.
Jahnke, R A. 19,90. Ocean flux studies: a status report.
Rev. Geophys. 28(4):381-398.
Loder, T.C. 1971. Distribution of dissolved and particu-
late organic carbon in Alaskan polar, subpolar and estua-
rine waters. Ph.D. Thesis, University of Alaska, Fair-
banks.
Longhurst,A.R. and W.G.Harrison. 1989. The biological
pump: profiles of plankton production and consumption
in the upper ocean. Progr. Oceanogr. 22:47-122.
Lyutsarev, S.V., Yu. V. Konnova, and VA. Konnov. 1988.
Composition and distribution of particulate organic mat-
ter in the Bering Sea. Okeanologiya/Oceanology 28:72-
77 (In Russian).
I.V. PERMINOVA AND V.S. PETROSYAN
McRoy, C.P., J.J. Goering, and W.S. Shiels. 1972.
Studies of primary production in the eastern Bering Sea.
In: Biological Oceanography of the Northern Pacific
Ocean. Tokyo, pp. 196-216.
Pershina, I.V. 1987. Determination of fulvic acids in
natural waters. Ph.D. Thesis, Moscow State University.
Pershina, I.V. 1992. Humic acids in the Bering and
Chukchi Seas. In: Results of the Third Joint US-USSR
Bering and Chukchi Seas Expedition (BERPAC), sum-
mer 1988. PA. Nagel, editor. US Fish and Wildlife
Service, Washington, DC, pp. 231-235.
RomankevichjEA. and S.V. Lyutsarev. 1990. Dissolved
organic carbon in the ocean. Mar. Chem. 30:16-178.
Skopintsev. 1981. Water humus. In: Marine Organic
Geochemistry, edited by E.K.Duursma and R.Dowson.
Elsevier oceanography series 31:150-178.
Sugimura, Y. and Y. Suzuki. 1988. A high -temperature
catalytic oxidation method of determination of non-
volatile dissolved organic carbon in seawater by direct
injection of liquid samples. Mar.Chem. 24:105-131.
Swift, J.H. 1984. The circulation of the Denmark Strait
and Iceland-Scotland overflow waters in the Northern
Atlantic. Deep-Sea Res. 31:1339-1356.
Toggweiler,J.R.1988. Deep-sea carbon, a burning issue.
Nature 334:468.
Walsh, J.J., C.P. McRoy, L.K. Coachman, JJ. Goering,
J.J.Nihoul, T.E.Whitledge, T.H. Blackburn, P.L. Parker,
C.D. Wirick, P.G.Shuert, J.M.Grebmeier, A.M. Springer,
R.D.Tripp, D. Hansell, S. Djenidi, E. Deleersnijder, K.
Henriksen, BA.Lund, PAndersen, F.E. Muller-Karger,
and K. Dean. 1989. Carbon and nitrogen cycling within
the Bering/Chukchi Seas: source regions for organic
matter affecting AOU demands of the Arctic Ocean.
Progr. Oceanogr. 22:277-359.
Zeeman, S.I. 1992. The importance of primary produc-
tion and CO2. In: Results of the Third Joint US-USSR
Bering and Chukchi Seas Expedition (BERPAC), sum-
mer 1988. PA. Nagel, editor. US Fish and Wildlife
Service, Washington, DC, pp. 218-224.
49
-------
-------
RESIDENCE TIME OF CARBON
IN SOILS OF THE BOREAL ZONE
Kira Kobak and Natalia Kondrasheva
ABSTRACT
To study the global carbon cycle and its modelling, reliable estimates for the values of the major organic carbon pools and
their transformation rates must be determined. This study of the relationship between the carbon content in organic
compounds in various landbiomes and climatic parameters revealed non-linear connections between the amount of carbon
in soils andphytomass and the aridity index (E^r), where E0 is the potential evapotranspiration and r is annual precipitation.
The improvement of the estimates of the carbon content in biomes makes it possible to determine the turnover rate (the
relaxation time) ofthecarboninphytomass. A distinct correlation wasestablishedbetweenthemeanresidencetimeofcarbon
in organic compoundsin soilandsuch climaticfactorsasradiationbalance andthemean annualairtemperature. Ourglobal
estimate of the stable residence time was 1350 years which is considerably above the majority of the available estimates for
the residence time of carbon in the soil pool that are within the range of 100 years to 500 years.
INTRODUCTION
To study the global carbon cycle and its model-
ling, reliable estimates for the values of the
major organic carbon pools and their transfor-
mation rates must be determined. The bulk of
carbon in organic compounds (Cor) is concen-
trated in sedimentary rocks of the ocean and
continents, about 12,000,000 Gt C, or 99.9% of
the total amount of Corg on the planet. Organic
carbon of the lithosphere, which has been bur-
ied for a hundred million years, is at present of
less importance in the natural carbon cycle; its
significance is felt only on the geologic time
scale.
The most intense transformations of carbon
compounds are primarily associated with the
autotrophic absorption of CO2 by plants (and
phototrophicmicroorganisms) andwith organic
matter destruction in the mixed layer of the
ocean and the upper layer of soils. As a result,
about 8% of atmospheric carbon is extracted
annually and returned. The organic carbon of
the continental biota is contained mainly in the
phytomass of plants (about 560 Gt Cor ). The
results of studies carried out in accordance with
the International Biology Program have re-
cently allowed us to improve global estimates of
the C content in the phytomass and soils of
the main land biomes and of net primary pro-
ductivity (Ajtay et al., 1979; Olson, 1982; Kobak,
1988).
The soil and plant carbon content depends on
climatic conditions. The study of the relation-
ship between the carbon content in organic
compounds in various land biomes and climatic
parameters revealed non-linear connections
between the amount of carbon in soils (Cs) and
phytomass (Cph) and the aridity index (Eo/r),
(where E0 is the potential evapotranspiration, r
51
-------
RESIDENCE TIME OF CARBON IN SOILS OF THE BOREAL ZONE
Cph=5.7e
CgK).3ST
(x+0.17)2
o.n ,byx<0.3
= -0.54x+1.42 0.5
-------
K. KOBAK AND N. KONDRASHEVA
cph/cs
0.3-
0.2-
0.1-
0
-0.8
0
0.8
1.6 logE0/r
Figure 2. Ratio between carbon content in plants and soils depending on the aridity index.
belt: about 740 Gt C, or 35% of the total the tropical soils was minimal compared with
amount. The second largest amount was con- the soils of other belts, the large area of the
centrated in the tropical soil pool, 580 Gt C, or tropical region provides it with a high organic
27%. Despite the fact that C concentration in carbon content (Figure 3).
Cskg/m<
30-
20-
10
Tropical belt
Subtropical
Subboreal
Boreal
Polar
330 Gt
-320Gt-
20 40 60 80
Figure 3. Organic carbon content in soils of different belts.
53
100
120
10
-------
RESIDENCE TIME OF CARBON IN SOILS OF THE BOREAL ZONE
Litter fall
g C/m2»yr
400 • •
200 • •
400 800
Detritus, g C/m2yr
Figure 4. Ratio of Cs in litter fall and detritus in ecosystems of the boreal belt.
1200
The global estimate of detritus mass was ob-
tained based on the empirical data on the litter
and litter-fall reserves in different ecosystems
and biomes. The calculations showed that a
very small portion of the soil carbon, 3.50% of
the total C reserve, or 72 Gt C, is concen-
trated in the litter. In the polar belt, it reaches
the maximum, about 10% of the total C
reserve in these soils; in the tropical belt, it does
not exceed 1.2% of the Coro. in tropical soils. The
data on the litter reserves and the mass of the
annual accumulation of organic substances on
the soil surface (litter-fall mass) as well as the
mass of dyingroots are needed to determine the
mobility of the organic carbon and the resi-
dence time of Corgin the litter pool (Td). The
mean global value Td was two years. It was quite
obvious that in different climatic belts, the ratio
between the masses of the litter and the litter
fall differed considerably. The values of Td
were also dissimilar and decreased naturally
from 30 years in the soils of the polar belt to 0.3
years in the tropical soils. In the boreal belt, the
value of Td was 9.4 years (Figure 4).
The bulk of the soil carbon was represented by
humus and peat substances, in this case the
labile part of the pool, including not fully humi-
fied plant residues, metabolism products and
newly formed humus substances, which made
up 1/3 of its total mass. The stable part of the
pool, where the humus substances were closely
connected with the mineral components of soils,
accumulated about 2/3 of its total mass. The
stable part of the pool was characterized by the
least mobility.
To calculate the intensity of humus formation,
the data on the amount of annual litter fall in
ecosystems of different climatic zones were
used, as well as the experimentally obtained
coefficients taking into account the rate of
organicmattertransformationintohumus, about
6% of annual litter fall (Oberlander and Roth,
1968; Kononova, 1984). The value of the newly
formed humus substances calculated in this way
was 2.5 Gt C/yr, the largest in tropical soils
(1.34 Gt C/yr) and the least in soils of the polar
belt (0.03 Gt C/yr). However, of the total
amount of the newly formed humus substances,
54
-------
K. KOBAK AND N. KONDRASHEVA
kg C/m2 • yr kg C/m2 Td • yr
29X
Tt,103yr Tst,103yr Tt,loV
0.03 -•
0.02 -
0.01 -•
30-
20-
ralO-- F
15"
10-
9X
6-
5-
4-
3
2
1
6-
5-
4-
3-
2-
1-
6-
5-
4-
3"
2-
1"
H Tropical Belt
0 Subtropical
9 Subboreal
£3 Bored
D Polar
n
m
Figure 5. Estimated organic carbon content and the intensity of its transformation in soils of different climatic belts. I
- Formation rate of new organic carbon in soils; II - Cs concentration in soils; III - Residence time of carbon in detritus;
IV - Residence time of carbon in humus (Tt); V - Residence time of carbon in the stable part of the soil humus (Tst);
and VI - Residence time of carbon in the labile part of humus (Tj).
only about 40% (1 Gt C/yr globally, 0.1 Gt C/
yr in soils of the boreal belt) were closely
associated with soil mineral particles producing
stable humus. More than one-half (58%) of this
newly formed humus produced labile, or bio-
logically active humus, which was subjected to
subsequent microbiological processing. In ab-
solute values, this amounted to 1.5 Gt C/yr
globally and 0.15 Gt C/yr in soils of the boreal
belt.
The data on the volumes of soil organic carbon
pools and the annual values of the newly formed
humus substances made it possible to deter-
mine the time of Corg relaxation (residence
time) T and the turnover time (1/T) of carbon
in carbon pools and their parts. The calcula-
tions have shown that, according to mobility,
the soil carbon pool can be divided into three
parts: 1. litter (Td= 2 years); 2. labile humus,
where the mean global residence time (Tl) was
480years; and3. the stable part of the soil pool,
in which the residence time of C (Tst) was
equal, on average, to 1350 years. The residence
time of Corg for the entire soil pool was 850
years.
As one should expect, the values of T were not
the same for soils of different types, zones and
belts (Figure 5). In boreal and polar belts, the
least mobility of organic carbon was typical for
arctic and swamp soils, where the relaxation
time for Corg was not less than 10,000 years.
Flood-land soils in the sub-boreal and boreal
belts were characterized by the largest mobility,
Tt = 630 years, Tst = 980 years, T, = 370 years.
The mean values of Tt for all of the soils of the
boreal belt were about 3000 years, Tst was 4300
years, and T, was approximately 2000 years. In
the tropical belt red-yellow soils were charac-
terized by the largest mobility. In the subtropi-
cal belt, yellow soils and red soils were the most
mobile. The least mobility of organic carbon in
the tropics and subtropics was typical for desert
soils (Tt = 4500 years, Tst = 6000 years, T, =
3000 years). The average values of T changed
appropriately with changing climatic conditions,
being the largest in boreal soils and the smallest
55
-------
RESIDENCE TIME OF CARBON IN SOILS OF THE BOREAL ZONE
E
kcal/c
80-
60-
40-
20-
>
"*»
m2/yr '
T,°C
•24 f
^ 1
•20 \
16 . \
12 2 %^
\
\
V^
U i 1 i v.1 i_
1000 2000 ^ — Tsiyi:
-4
Figure 6. Relationship between residence time of carbon in soils (Tst) and climatic parameters: mean annual air
temperature (1) and the radiation balance (2).
in the tropics. Considerable reserves of organic
carbon in the boreal belt, with a comparatively
smallincome of newly fonnedhumic substances,
existed due to the long residence time of car-
bon. On the other hand, small concentrations
of C in the tropics, where humus formation
intensity was maximum, canbe attributed to the
high rate of organic matter transformation.
The most distinct correlation was established
between the mean residence time of Corg in soil
and such climatic factors as radiation balance
and the mean annual air temperature (Figure
6). With the decrease in the radiation balance
value and the mean annual temperature, the
residence tune of carbon increased. Within the
tropical, subtropical and subboreal belts, when
aridization increased the residence time of car-
bon increased, whereas the amount of newly
formed substances decreased. In the boreal
and polar belts, an increase in the value of T and
decrease in the amount of the newly formed
humic substances correlated with temperature
reduction and moisture rise.
Our global estimates of Tst were considerably
above the majority of the available estimates
for the residence time of carbon in the soil pool
that are within the range of lOOyears (Emanuell
et al., 1984) to 500 years (Goudriaan and Ketner,
1984). Comparison of Tst values obtained for
the different soils with the data on radiocarbon
dating of the age of these soils (Chichagova,
1985) showed good agreement (Figure 7). This
confirmed the correctness of the calculations of
Tst and indicated that the residence time of the
organic carbon in the stable part of the soil pool
was substantial for about 1350 years.
56
-------
K. KOBAK AND N. KONDRASHEVA
Tst,yr
8000 T
6000
4000
2000
o
VI
O
o
VI
O
PH
M
• i-H
O
I
Radiocarbon data
Tst
8
'S
i
i
vt
'o
O
O
E
Figure 7. Comparison between the results of radiocarbon dating of soil ages and T t.
g
a
o
-Cj
O
REFERENCES
Ajtay,G.L.,P.Ketner,andP.Duvigneaud. 1979. Terres-
trial primary productions and biomass. In: The global
carbon cycle. SCOPE 13, New York, pp.129-181.
Chichagova, O.N. 1985. The radiocarbon dating of soils
age. Nauka, Moscow, p.158 (in Russian).
Emanuell, W.R., G.E.G. Killough,W.M. Post, and H.H.
Shugart. 1984. Modelling terrestrial ecosystems in the
global carbon cycle with shifts in carbon storage capacity
by land use change. Ecology 65:970-983.
Goudriaan,J. and PA. Ketner. 1984. A simulation study
for the global carbon cycle, including man's impact on
biosphere. Climatic Change 6:167-192.
Grigorjev,AA.andM.I.Budyko. 1956. Periodical low of
the geographical zonality. Doklady Akademy Nauk USSR
143(N2):391-393 (in Russian).
Kobak, K.I. 1988. Biotical compounds of carbon cycle.
Gidrometeoizdat, Leningrad, p.246 (in Russian).
Kononova, M.M. 1984. The organic matter and the
fertility of soils. Pochvovedenie 8:6-20 (hi Russian).
Oberlander, H. and K. Roth. 1968. Transformation of 14
C-labelled plant material in soils under field conditions.
In: Isotopes and Radiation inSoilOrganicMatter Studies.
InternAtom.EnergAgency, Vienna, p.241.
Olson, J.S. 1982. Earth's vegetation and atmospheric
carbon dioxide. In: Carbon Dioxide Review 1982.
Clarendon Press, New York, pp.388-398.
57
-------
-------
EQUILIBRIUM ANALYSIS OF PROJECTED CLIMATE CHANGE
EFFECTS ON THE GLOBAL SOIL ORGANIC MATTER POOL
David P. Turner and Rik Leemans
ABSTRACT
Increased rates of soil organic matter decomposition may represent a significant positive feedback to global warming. As
a step towards assessing the potential magnitude of this response, an equilibrium analysis was performed in which
representative carbon pools were associated with each vegetation type, and the Holdridge vegetation/climate correlation
system was used to compare distributions of the vegetation types under the current climate and doubled-CO2 climate
scenarios from four general circulation models. Two of the general circulation models predicted a net loss ofbelow-ground
carbon (55-101 Pg) because of large decreases in the areal extent of tundra and boreal ecosystems with high levels ofbelow-
ground carbon storage. Vegetation redistribution projected under the other two general circulation models would result in
the accumulation of carbon (5-41 Pg) in the biosphere; however, this accumulation was driven primarily by an increase in
the areal extent of tropical rain forests that is unlikely given constraints imposed by anthropogenic factors. Additional
considerations not treated by the equilibrium approach support the likelihood of a transient pulse of carbon from the soil
to the atmosphere.
INTRODUCTION
Since at least the time of the 18th century
geographer Alexander von Humbolt, it has
been recognized that vegetation type or physi-
ognomy is correlated with climate. The con-
straints on plant form and microbial metabo-
lism imposed by climate mean that climate also
strongly regulates the typical above-ground and
below-ground carbon in terrestrial ecosystems.
Various compilations of field studies have pro-
vided the basis for estimating representative
above- and below-ground carbon storage for
specific biomes (Olson et al., 1983; Post et al.,
1982), although recent detailed surveys suggest
earlier estimates may be too high (Botkin and
Simpson, 1990). In any case, a change in the
areal extent of different biomes as a result of
global climate change will be likely to result in
fluxes of carbon to or from the biosphere, with
corresponding changes in the atmospheric CO2
concentration. These changes are of interest in
terms of positive or negative feedbacks from
the terrestrial biosphere to climate change.
APPROACH
The specific objective of this study was to inves-
tigate the potential for a net flux of carbon
between the below-ground component of the
terrestrial biosphere and the atmosphere due
to climate change. This was an equilibrium
analysis (i.e., the assumption is made that cli-
mate and vegetation are always in equilibrium)
and abookkeeping approach was used in which:
1) vegetation types are distributed across the
land surface based on the current climate and
double-CO2 climate scenarios using an existing
vegetation-climate correlation system
(Holdridge, 1947); 2) the vegetation types are
59
-------
EQUILIBRIUM ANALYSIS OF PROJECTED CLIMATE CHANGE EFFECTS
assigned representative above- and below-
ground carbonpools; 3) terrestrial below-ground
carbon storage is summed by climate scenario;
and 4) the differences between current storage
and future storage are determined. Detailed
treatment of the analysis is given inDixon and
Turner (1991), and this paper is intended to
discuss the results in light of other recent stud-
ies of the global carbon cycle.
RESULTS AND DISCUSSION
Changes in the area! extent of the biom.es, as
predicted by the four general circulation mod-
els (GCM), showedsomebasicsimilarities (Fig-
ure 1). At middle to high latitudes, tundra and
boreal forests contracted, and temperate forest
(mostly coniferous) expanded. In the tropics,
both the semi-arid woodland and the tropical
rain forest had large increases in area! extent,
mostly at the expense of tropical seasonal for-
ests.
The net change in below-ground carbon stor-
age ranged from a 41 Pguptake predicted by the
OSU model to a 101 Pg release predicted by the
UKMO model (Table 1). The large carbon
releases were associated with reductions in the
areal extent of boreal forests and tundra, while
the carbon uptake was driven by increase in the
areal extent of tropical wet forests. The below-
ground carbon storage is relatively high in the
tropical wet forests, so, in the cases of carbon
uptake by the biosphere, the changes from
tropical dry forest to tropical wet forest tended
to dominate the global trends.
Table 1. Projected changes in terrestrial belowground
carbon storage based on 2XCO2 climate change sce-
narios from four General Circulation Models.
General Circulation Model
Change in Below-
ground Carbon (Pg)
Oregon State University
Ooddard Institute for Space Studies
Geophysical Fluid Dynamics Laboratory
United Kingdom Mcterological Office
+41
+5
-55
-101
It is difficult to evaluate the magnitude of the
feedback to climate change that would be asso-
ciated with these carbon fluxes. With regard to
the positive feedback, some of the CO2 emitted
to the atmosphere would be taken up by the
ocean. However, even if only half of the 101 Pg
were to remain in the atmosphere, the atmo-
spheric CO2 concentration would increase by
about 25 ppm. This compares with an addition
of 340 ppm of CO2-carbon for the doubled-CO2
GCM runs used in this analysis and to the
current rate of accumulation of about 1.5 ppm
per year. Note that some of the carbon would
be emitted as methane, which has a much
stronger radiative forcing per molecule than
does CO2. The 5-41 Pg negative feedback
would decrease the atmospheric CO2 concen-
tration by at most 20 ppm.
A number of other equilibrium analyses have
examined possible changes in below-ground
carbon storage due to climate change. Results
include decreases ( < 50 Pg) in terrestrial be-
low-ground storage (Lashof, 1989; Schlesinger,
1990) or small increases (11-40 Pg) (Prentice
and Fung, 1990). Overall, these analyses sug-
gest the potential for a modest biospheric feed-
back to climate change mediated by changes in
below-ground carbon storage.
Equilibrium analyses suffer from several weak-
nesses in their treatment of factors that could
markedly influence the actual outcome of cli-
mate change. Probably the main limitation is in
the assumption that vegetation and below-
ground carbon storage will change primarily in
response to climate. It seems likely that land
use patterns will be at least as important as
climate-driven vegetation change over the next
few decades. Although there is some hope for
reductions in deforestation and increased re-
forestation (Dixon et al., 1991), the trend in the
areal extent of tropical rain forest is certainly
downward (Houghton, 1991). Thus the pro-
jected carbon sinks in the tropical latitudes may
not appear. If the carbon sources from in-
60
-------
D.P. TURNER AND R. LEEMANS
r>
O
-a
.-a
S 8
1|
0 w
fi
•2 |
o JS
J'-a
^ §
II
•lH-l.ll.h-,
SB
Sil
Q.
O
§
61
-------
EQUILIBRIUM ANALYSIS OF PROJECTED CLIMATE CHANGE EFFECTS
creased decomposition of soil carbon at high
latitudes did meanwhile occur, the global car-
bon flux from soil could be larger than the
equilibrium analyses indicate.
A second weakness of these analyses is the
assumption that climate and below-ground car-
bon storage will stay in equilibrium (Dixon and
Turner, 1991). Schlesinger (1990) reported
strong constraints on rates of below-ground
carbon accumulation (0.2-12 g/m2/yr) based
on chronosequence studies. There is also evi-
dence in the paleo record for disequilibrium
between climate and pedogenic factors
(Pennington, 1986). In contrast, short-term soil
incubations indicate that soil carbon mineral-
izationis very sensitive to temperature increases
(Nadelhofferetal., 1991). The degree to which
recalcitrant fractions of the soil organic matter
would limit the longer term response to warmer
temperature is not known. However, carbon
losses in cultivated fields often stabilize at about
half of the original carbon concentration
(Schlesinger, 1984), suggesting that relatively
large amounts of carbon could be readily oxi-
dized.
The limitations of equilibrium analyses support
the effort to develop process-based models in
which the various soil organic matter fractions
are accounted for, and microbially driven trans-
formations of soil organic matter are controlled
by environmental variables, particularly tem-
perature and soil moisture status. These mod-
els must ideally characterize the behavior of
ecosystems in terms of net primary productiv-
ity, above- and below-ground allocation pat-
terns and decomposition of soil organic matter,
litter and coarse woody debris. Jenkinson et al.
(1991) developed a decomposition model in-
corporating some of these features that was
used to investigate the sensitivity of global soil
carbon storage to temperature increases. Re-
sults indicated a net release of carbon on the
order of 40-100 Pg over 60 years. The CEN-
TURY model of Parton et al. (1987) has also
been used to evaluate climate-change scenarios
over a network of weather stations in the Great
Plains (Schimel et al., 1991). It suggested alow
rate of carbon loss at most sites over the 500
year simulation, due primarily to the large pool
of soil carbon with a long turnover time. Pro-
cess models for below-ground carbon cycling in
forest ecosystems that can be run over large
geographical areas are also under development
(e.g., Running etal., 1989). In combination with
remote-sensing based estimates of initial veg-
etation conditions, these models will provide
increasingly reliable estimates of regional and
global carbon fluxes under the current climate
and future climate scenarios.
CONCLUSIONS
Equilibrium analyses of potential changes in
below-ground carbon storage due to projected
climate change suggest the possibility of a mod-
est positive feedback that is a transfer of carbon
from the soil to the atmosphere. These analyses
do not treat anthropogenic factors, which will
probably serve to increase the soil carbon flux.
Nor do they account for differential rates of
soil-carbon gain and loss that are also likely to
aggravate net-loss rates over the coming de-
cades. Spatially distributed process-based mod-
els, which mechanistically treat above- and be-
low-ground carbon transformations, are needed
to more realistically simulate the transient re-
sponse of biospheric carbon storage to climate
change.
62
-------
D.P. TURNER AND R. LEEMANS
REFERENCES
Botkin, D.B. and L.G. Simpson. 1990. Biomass of the
North American boreal forest: a step toward accurate
global measures. Biogeochemistry 9:161-174.
Dixon, R.K. and D.P. Turner. 1991. The global carbon
cycle and climate change: responses and feedbacks from
below-ground systems. Environ. Poll. 73:245-262.
Dixon, R.K., P.E. Schroeder, and J.K. Winjum. 1991.
Assessment of promising forest management practices
and technologies for enhancing the conservation and
sequestration of atmospheric carbon and then- costs at the
site level. EPA/600/3-91/06. USEPA Environmental
Research Laboratory, Corvallis, OR, p. 138.
Holdridge, L.R. 1947. Determination of world formula-
tions from simple climatic data. Science 105:367-368.
Houghton, R A. 1991. Tropical deforestation and atmo-
spheric carbon dioxide. Clim. Change 19:99-118.
Jenkinson, D.S., D.E. Adams and A. Wild. 1991. Model
estimates of CO2 emissions from soil in response to global
warming. Nature 351:304-306.
Lashof, DA. 1989. The dynamic greenhouse: feedback
processes that may influence future concentrations of
atmospheric trace gases and climatic change. Clim. Change
14:213-242.
Nadelhoffer, K.K., A.E. Giblin, G.R. Shaver, and JA.
Laundre. 1991. Effects of temperature and substrate
quality on element mineralization in six arctic soils. Ecol-
ogy 72:242-253.
Olson, J.S.,JA. Watts, and L.J. Allison. 1983. Carbon in
live vegetation of major world ecosystems. ORNL-5862.
Oak Ridge National Laboratory, Oak Ridge, Tn.. p. 180.
Parton, W.J., D.S. Schimel, C.V. Cole, and D.S. Ojima.
1987. Analysis of factors controlling soil organic matter
levels in Great Plains grasslands. Soil Sci. Soc. Amer. J.
51:1173-1179.
Pennington,W. 1986. Lags in adjustment of vegetation to
climate caused by the pace of soil development. Vegetatio
67:105-118.
Post, W.M., W.R. Emanuel, PJ. Zinke, and A.G.
Stangenberger. 1982. Soil carbon pools and world life
zones. Nature 298:156-159.
Prentice, K.C. and I.Y. Fung. 1990. Bioclimatic simula-
tions test the sensitivity of terrestrial carbon storage to
perturbed climates. Nature 346:48-51.
Running, S.W., R.R. Nemani, D.L. Peterson, L.E. Band,
D.R. Potts, L.L. Pierce, and MA. Spanner. 1989. Map-
ping regional forest evapotranspiration and photosynthe-
sis by coupling satellite data with ecosystem simulation.
Ecology 70:1090-1101.
Schimel, D.S., W.J. Parton, T.G.F. Kittel, D.S. Ojima, and
C.V. Cole. 1991. Grassland biogeochemistry: links to
atmospheric processes. Clim. Change 17:13-25.
Schlesinger, W.H. 1984. Soil organic matter: A source of
atmospheric CO2. In: The Role of Terrestrial Vegetation
in the Global Carbon Cycle: Measurement by Remote
Sensing, edited by G.M. Woodwell. John Wiley & Sons,
New York.
Schlesinger, W.H. 1990. Evidence from chronosequence
studies for a low carbon-storage potential of soils. Nature
348:232-234.
63
-------
-------
DISTRIBUTION AND RENOVATION TIME OF SOIL CARBON
IN BOREAL AND SUBARCTIC ECOSYSTEMS OF EUROPEAN RUSSIA
Alexander E. Cherkinsky and Sergey V. Goryachkin
ABSTRACT
The global distribution of soil organic matter is of great interest, because soil organic matter is one of the main reservoirs of
biospheric carbon. Representative profiles of the main soil types of the subarctic and boreal ecosystems of European Russia
were studied in detail. Carbon content was detected by wet oxidation. The identification of humus type was based on group
composition, and the renovation rate was estimatedfrom radiocarbon data. Based on soil data obtainedfrom other regions
of European Russia and calculations of total soil areas, stores of soil carbon were studied in European Russia north of
approximately 64° north latitude. The results presented are preliminary and give a rather rough estimate, but they do provide
the first opportunity to assess the carbon pool of the territory studied, based on the experimental data.
INTRODUCTION
The global distribution of soil organic matter is
of great interest, as soil organic matter is one of
the main reservoirs of biospheric carbon; how-
ever all of its assessments are not satisfactory.
Dokuchaev, Tyurin and their followers con-
cluded that the main factors regulating organic-
matter content in the soils were the humidity
and temperature. However, parent materials
also play a key role. We attempted to general-
ize all of the various soil organic matter profiles
- organic profiles of natural (undisturbed) soils
of the world, taking into account not only their
humus component, but also the surface accu-
mulation of organic detritus. This consider-
ation is of vital importance for subarctic and
boreal ecosystems where the major part of
organic matter is on the soil surface.
We understand the types of organic profile to
be combinations of relatively homogeneous
conjugate organic and/or organomineral hori-
zons, each of which is formed by a complex of
organic-matter metamorphic processes of dif-
fering intensity. To identify the organic profile
types, we used a well-known humus character-
istics (morphological types of humus, stores of
organic matter, Cha/Cfa etc.) and also proposed
to use temporal parameters, determined by
radiocarbon data and showing the rates of or-
ganic profile formation and organic matter
renovation.
METHODS
We studied representative profiles of the main
soil types of the subarctic and boreal ecosys-
tems of European Russia in detail. The studied
profiles were situated in the Arkhangelsk re-
gion to the north of 64° N latitude. Carbon
content was detected by wet oxidation (Tyurin's
method). The identification of humus type was
based on group composition (Ponomareva and
Plotnikovag, 1980), and the renovation rate was
calculated in accordance with radiocarbon data
65
-------
DISTRIBUTION AND RENOVATION TIME OF SOIL CARBON
(CherkinskyandBrovkin, 1990,1991). Besides
our own experimental data, we used some soil
data obtained from other regions of European
Russia (Ignatenko, 1979). Based on both soil
data sources and calculations of total soil areas
(Dobrovolsky and Urusevskaya, 1984), we esti-
mated the stores of soil carbon in European
Russia to the north of approximately 64° N
latitude.
We consider our results to be preliminary and
giving a rather rough estimation, but they do
provide the first opportunity to assess the car-
bon pool of the studied territory, based on the
experimental data.
DESCRIPTION OF SOIL ORGANIC PRO-
FILES OF THE RUSSIAN EUROPEAN
NORTH
As is widely known, the main processes of
organic profile formation in subarctic and bo-
real ecosystems are detritus accumulation and
humus illuviation. These processes result in a
predominance of accumulative-detritus and
accumulative- detritus/humus- illuvial organic
profile types. However, large-scale surveys
demonstrated that not only these two types, but
awidespectrum of types of soil organic profiles
occurred here (Table 1).
The accumulative-detritus organic profile oc-
cupies the largest area of the studied territory,
especially in the southernmost part of the sub-
arctic zone, as the most common soils here are
peat bog soils (FAO-UNESCO, 1990). This
type is the most inert and is characterized by
peat accumulation and by low renovation rates.
Organic matter stored in this organic profile
type is weakly decomposed and has a low ash
content. The carbon-14 age of the upper 12 cm
of the studied subarctic Belie Histosol is 100
years, and that at a depth of 30-40 cm is 5200
years.
The accumulative-detritus/humus-alluvial or-
ganic profile is characteristic for semi-hydro-
morphic soils (common for subarctic and bo-
real territories of the Russian European North):
bog-podzolic (Gleyic Podzoluvisols), tundra
gleyic (Gleyic Gleysols), and peaty podzols
(Gleyic Podzols). This type is distinguished by
peaty litter (10-20 cm thick) and by spots of
illuvial humus within the profile; the humus has
a fulvatic character, and its content is rather
high (5.5% at a depth of 30 cm). Imperfect
drainage mitigates biochemical activity and
results in low rates of organic matter renovation
(carbon-14 age is 1880± SOyears in the 0-13 cm
layer and 4760 ±110 years in the 21-30 cm
layer).
Table 1. Characteristics of soil organic profiles of subarctic and boreal ecosystems of European Russia.
Organic profile type
Accumulative- detritus
Accumulative-detritus
humus-alluvial
Detritus humus-alluvial
Eluvial-detritus-humus
Detritus accumulative-
humus
Organic profile characteristics
Depth of Degree of
Soils detritus decomposition Humus Type of
horizons of organic content humus Renovation
(cm) matter (%) (Cha/Cfa) time (years)
Peatbog
Bog-podzolic,
tundra-gleyic
Gley-podzolic,
humic podzols
Podzolic, humus
ferrous podzols
Rendzinas, raw-
humus burozems
>50 weak-medium — peat
10-50 medium-well 5-10 fulvatic
5-10 weak-medium E 0.6 fulvatic
Bh2.5
<5 weak-medium E0.2 fulvatic
Bh 0.5-1.5
<5 weak A 4-8 fulvatic
AB2-4
B 1-1.5
nlOOO
detritus nlOO
humus nlOOO
detritus nlOO
humus
nlOO-1000
nlOO
nlOO
66
-------
The detritus/humus-alluvial organic profile is
characteristic for soils with more perfect drain-
age conditions than those of previous types, but
still drainage is not absolutely perfect. The
typical representatives of this organic profile
type are humic podzols (Carbic Podzols) and
gley-podzolic soils (soils of transitional charac-
ter between Dystric and Gleyic Podzoluvisols).
These soils may have a slightly peaty Utter and
a relatively high humus content (more than 2%)
hi the B horizon.
The eluvial-detritus-humus organic profile is
represented by perfectly drained soils devel-
oped from loams, podzolic soils (Dystric and
Eutric Podzoluvisols), and from sands, humus-
ferrous podzols (Haplic Podzols), which are
typical for boreal forests and scarcely distrib-
uted in the subarctic zone. The organic residues
after input on the soil surface have undergone
decomposition and humification, but the prod-
ucts of this humification are not accumulated
within the soil and are eluviated by soil water.
This type of organic profile has a relatively thin
litter (5-10 cm) and fulvatic humus (Cha/Cfa less
than 0.5).
A detritus humus-accumulative organic profile
is very rare for boreal territories and more
common for moderate climatic zones. It is
characterized by the presence of raw-humus as
weh1 as mull-humus horizons. This type occurs
only in places where calcareous rocks or rocks
with high Fe content are the parent materials.
The high degree of the basic saturation of these
parent materials mitigates the processes of
humus eluviation. This organic profile type is
characterized by the thinnest litter for boreal
regions, by a mull-humus horizon with 5%
humus content and by the most intensive reno-
vation rates for this territory. The carbon-14
ageismodernfor the upper horizons and 650±30
years for the 10-20 cm layer. The representa-
tives of this organic profile type are rendzinas
(Rendzic Leptosols) formed oncalcareous rocks
and raw-humus burozems (Eutric Cambisols)
A.E.CHERKINSKY AND S.V. GORYACHKIN
formed on ferrous rocks.
RESULTS AND DISCUSSIONS
In the organic profile types studied, organic
carbon stores in litter or peat horizons and in
mineral layers, in which humus usually had an
illuvial character (Table 2) were estimated.
Table 2 shows that about 70% of the carbon
pool of the territory is the carbon in peat. The
accumulative-detritus organic profile type, in
Table 2. Carbon pool of organic profiles of Russian
European North.
Stores of organic matter Total
(kg/m ) carbon
Organic Detritus
profile type layers
Accumulative- 62.30
detritus
Accumulative-
detritus 9.60
humus-alluvial
Detritus 4.50
humus-alluvial
Eluvial- 3.00
detritus-humus
Detritus 2.23
accumulative-
humus
Total
Mineral Area stores,
layers Total m2 x 1010 1010kg
— 62.30aj 24.90 1551.30
9.20 18.80 24.50 459.99
5.35 9.85 12.20 120.17
3.30 6.30 15.90 100.17
9.5 11.7 <0.10 <1.17
2232.80
(a) Calculated value.based on mean peat stores in the former Soviet
Union (World Peat Resource: Reference book, Nedra, Moscow. 1988).
which the carbon concentration is maximum
(62.3 kg/m2), occupies 32% of the total area of
the Russian European North. Accumulative-
detritus organic profile is formed as a result of
processes of organic matter accumulation on
the mineral surface and then buried by new
organic residuals. In the severe climatic condi-
tions of the North, organic matter is preserved
in a weakly decomposed state in conditions of
impeded drainage. The whole thickness of
accumulated peat is often more than 10 m. The
carbon-14 age of the deepest peat layers corre-
sponds to the Holocene optimum. This proves
that carbon of this organic profile has existed
without visible alterations for a long time. Other
67
-------
DISTRIBUTION AND RENOVATION TIME OF SOIL CARBON
investigations show the changes in peat accu-
mulation rates, but these changes may be caused
not only by climatic fluctuations, but also by
local factors. According to the radiocarbon
data, the contemporary rate of peat accumula-
tion in the southernmost part of the subarctic
zone in the upper 10 cm is 1 mm/year. One may
suppose that climatic wanning would result in
an increase in peat accumulation rates
(BoryachkinandTargulian, 1990), and it would
stimulate the fixation of atmospheric carbon in
lifeless organic matter.
The accumulative-detritus/humus-illuvial or-
ganic profile type is the second largest total
storeof organic-matter carboninthis region. It
contains 16% of the total carbon (Table 2).
This organic profile is characterized by equal
carbon stores in the litter and mineral layers.
This type is differentiated by low carbon reno-
vation rates, not only in mineral horizons, but
also in the litter. The radiocarbon age of the
litter is 100 years, which contradicts the data of
Emanuel et al. (1984,1985) and Kobak (1988),
who have estimated the time of detritus relax-
ation in the subarctic zone as 10-100 years.
Therefore, the rates of detritus carbon turnover
in subarctic and boreal zones, under imperfect
drainage conditions, are lower than previously
known. CUmatic warming would hardly change
the carbon concentration in this organic profile
type, as similar soils with imperfect drainage of
the moderate climatic zone have the same
carbon stores.
A comparable carbon distribution, but with half
the carbon stores, occurs in the detritus humus-
illuvial organic profile type. Minimal carbon
stores occur in the eluvial-detritus-humus or-
ganic profile type, which occupies 20% of the
total area in the subarctic and boreal zones of
the Russian European North. The litter carbon
store and that of the mineral horizons are equal
and are 3 kg/m2 each. This value is half that
estimated by Kobak (1988) for podzolic soils of
the boreal and subarctic zones.
The detritus accumulative-humus organic pro-
file type in these ecosystems is rarely found in
soils formed from parent materials rich in Ca
and/or Fe. But it differs from other types by the
distribution of organic carbon, 80% of which is
in mineral horizons in the form of fulvatic
humus. The role of this organic profile type,
which is more common for more southern ter-
ritories in the carbon pool of the studied region,
is negligible, as the basic and calcareous rocks
occur very seldom here.
CONCLUSIONS
Distinguishing types of organic profiles seem to
provide agood perspective for the estimation of
the stores and turnover rates of carbon in the
soils of the world. This approach gives an
opportunity to determine and evaluate more
exactly the organic carbon pool in the
pedosphere and to assess its stability and/or
changeability in connection with possible envi-
ronmental changes.
The major portion (about 90%) of the organic
carbon in the subarctic and boreal ecosystems
of European Russia is in the accumulative-
detritus and in the accumulative-detritus/hu-
mus-illuvial organic profiles, which are repre-
sented by soils of impeded and imperfect drain-
age. Currently, it is impossible to predict ex-
actly their functions in the case of climatic
warming, but we may suppose that these non-
agricultural soils would not be sources of CO2
emissions, and in fact, more intensive growth of
peat is possible.
This paper is to be considered a preliminary
assessment that allows the correction of previ-
ous contradictory estimations for the practi-
cally unstudied territories of the subarctic and
boreal zones.
68
-------
A.E.CHERKINSKY AND S.V. GORYACHKIN
REFERENCES
Cherkinsky, A.E. and VA. Brovkin. 1990. The use of
radiocarbon method for study of open systems in modern
environment (on examples of contemporary soils). In:
Quaternary Period: Methods of Study, Stratigraphy and
Ecology. Tallinn, pp. 164-166 (in Russian).
Cherkinsky, A.E. and VA. Brovkin. 1991. A model of
humus formation in soils based on radiocarbon data of
natural ecosystems. Radiocarbon 33:186-187.
Dobrovolsky,G.V.andI.S.Urusevskaya. 1984. Soil cover.
In: Soil-geological conditions of non-chernozemic region,
edited by YJvI.Sergeev. pp. 367-464 (in Russian).
Emanuel, W.R., G.E. Killough, W.M. Post, and H.H.
Shugart. 1984. Modeling terrestrial ecosystems in the
global carbon cycle with shifts in carbon storage capacity
by land-use change. Ecology 65:970-983.
Emanuel, W.R., H.H. Shugart, andM.R. Stevenson. 1985.
Climatic change and the broad scale distribution of terres-
trial ecosystem complexes. Climatic Change 7:29-43.
FAO-UNESCO. 1990. Soil map of the world. Revised
legend. Rome.
Goryachkin, S.V. and V.O. Targulian. 1990. Climate-
induced changes of the boreal and subpolar soils. In: Soil
on a Warmer Earth, edited by H.W. Scharpenseel, M.
Schomaker, and A. Ayoub. Proceedings of an Interna-
tional workshop on effects of expected climate change on
soil processes in the tropics and sub-tropics, 12-14 Febru-
ary, Nakobi, pp. 191-209.
Ignatenko, I.V. 1979. Soils of East European tundra and
forest tundra. Nauka, Moscow (in Russian).
Kobak, K.I. 1988. Biota compounds of carbon cycle.
Hydrometeoizdat, Leningrad (in Russian).
OleninA.S. 1988. Peat resources of the world. Reference
book. Nedra, Moscow (in Russian).
Ponomareva, V.V.andTA.Plotnikova. 1980. Humus and
pedogenesis. Nauka, Leningrad (in Russian).
69
-------
-------
STATE FACTORS, STEADY STATES AND SOIL ORGANIC-
MATTER DYNAMICS AFTER FOREST HARVESTING
Jeffrey G. Borchers
ABSTRACT
A "state factor" approach was used to compare the effects of forest harvesting on long-term organic-matter dynamics in soils
that developed from different parent materials. In the geologically diverse region of southwest Oregon, soils that have
developed from granitic substrates are coarser textured than those formed on metavolcanic substrates. The distribution of
carbon and nitrogen in particle-size fractions from soils of old, poorly-vegetated clear-cuts revealed that initial carbon and
nitrogen levels and post-harvest losses were significantly greater in the finer-textured metavolcanic soil type. This behavior
was attributed to more extensive soil aggregation originating from the higher silt and clay contents of the metavolcanic soil.
"Sequestration" of organic matter by soil aggregation is hypothesized to influence theoretical steady-state levels of soil carbon
and nitrogen.
INTRODUCTION
How will terrestrial ecosystems respond to rapid
climate change, and will this further perturb the
climate system? This question invokes amyriad
of complex interactions between biota and cli-
mate (Perry etal., 1991;Kellog, 1983); it can be
resolved only if ecosystem-level studies are
designed to contribute to the solution of large-
scale problems (O'Neill et al., 1991). One
example of this is the study of soil organic
matter (SOM) dynamics. This area of research
has taken on new meaning beyond soil tilth and
fertility because CO2 and CH4 are important
greenhouse gases. Soil-carbon storage in the
northern hemisphere may now be an important
CO2 sink in the global carbon (C) budget (Tans
et al., 1990; Quay et al., 1992). As global
temperatures rise, increased decomposition
rates of SOM are likely to occur (Jenkmson and
Rayner, 1977), but it is difficult to predict if
losses will be offset by increased ecosystem
productivity and larger detrital inputs.
The difficulty of modeling these complex inter-
actions is increased by pervasive changes in
land use practices (Detwiler, 1986; Post et al.,
1990). For example, in Russia, a demand for
hard currency has prompted a significant num-
ber of joint ventures with foreign multi-national
companies to expand Siberian timber opera-
tions (Rosencranz and Scott, 1992). (Because
of the volatile nature of that region's economy,
it may be more challenging to predict the extent
of forestry operations than to forecast the re-
sulting impact on ecosystems and climate.) In
contrast, Pacific Northwest forests have, for the
most part, alreadybeenharvested, and landuse
questions are of a different nature. Although
rapid shifts in management philosophies are
now taking place, it is still useful to look back at
the longer-term changes that previous activities
71
-------
STATE FACTORS, STEADY STATES AND SOIL ORGANIC-MATTER DYNAMICS AFTER FOREST HARVESTING
have wrought, especially in the soil ecosystem.
STATE FACTORS AND
RETROSPECTIVE STUDIES
In one sense, the soil represents a long-term
record of the many biotic and abiotic factors
thatinfluence ecosystembehavior. Jenny (1980)
reduced these forces to a set of relationships
first described by Dokuchaev (Vil'yams, 1968).
Jenny referred to climate, organisms, relief,
parent material and time as the "state factors"
that control soil formation. Other related mod-
els have been proposed: Oades (1988) summa-
rized the factors that affect C retention in soils,
including litter quality, base saturation and soil
parent material. Regardless of the conceptual
model, the complexities of the ecosystem (e.g.,
feedback processes, interactions, hierarchical
structures) will always hinder efforts to isolate
state factors experimentally. Nevertheless, ex-
periments along gradients of state factors have
yielded muchinsight about ecosystemprocesses
(Vitousek, 1991; Van Cleveetal., 1991;Toutain,
1987; Sollins et al., 1983; Dickson and Crocker,
1953). The research reported here represents
a study of two state factors. The first, organisms,
includes human activities, and may represent
the most rapid and pervasive influence on soil
processes. The second, parent material, is the
least mutable, providing a geological context
for more dynamic soil processes.
Differences in soil parent material can strongly
influence soil morphology even where soil de-
velopment has occurred in identical environ-
ments (Harradine and Jenny, 1958). Parent
material can influence the chemical and physi-
cal characteristics of soil in several ways. For
instance, the type of soil minerals and their
weathering rates may provide long-term con-
trol over the availability of certain nutrients to
plants (Anderson, 1988). Eventually, mineral-
ogy and weathering rates also influence the
physical structure of soil. This has several
consequences, not the least of which is an
overriding effect on water-drainage character-
istics (Paton, 1978; Greacan and Williams, 1983).
The physical structure of soil can be defined as
a combination of soil texture and soil aggrega-
tion. Soil texture, defined as the distribution of
material inparticle-size classes, is a feature that
changes slowly relative to the rates of other soil
processes. Except in highly weathered or or-
ganic soils, texture is mainly an expression of
the properties of the parent material. Soil
aggregation, however, is the result of bonding
between particles, organic or inorganic. It is a
much more ephemeral condition than soil tex-
ture and results primarily from the activities of
organisms such as mycorrhizal fungi (Lynch
and Bragg, 1985).
Whereas large soil aggregates ( > 250 mm diam-
eter) are probably short-lived, smaller struc-
tures having a preponderance of clay- and silt-
sized particles maybe quite long-lived (Tisdall
and Oades, 1982; Anderson and Paul, 1984).
Organic matter that binds soil aggregates may
also be long-lived, representing substrates that
are physically less accessible to microbial deg-
radation (Jenkinson and Rayner, 1977; Ander-
son, 1979). Some of the earlier evidence for this
sequestration or physical "protection" of or-
ganic matter consisted of observations that
sieving or grinding soil subsequently increases
microbial activity (Rovira and Greacan, 1957).
Soil texture and aggregation have sometimes
been incorporated into models that simulate
SOM dynamics (Parton et al., 1988; Pastor and
Post, 1986). However, modelers still have much
to gain from a clearer understanding of how soil
aggregation modulates microbial activity
(Rastetter et al., 1991). As we shall see, the
structure of soil can produce some complex
effects on SOM dynamics when ecosystems are
disturbed (Paul, 1984).
72
-------
FOREST HARVESTING, SOIL STRUC-
TURE AND SOIL ORGANIC MATTER
DYNAMICS
The research summarized here is a retrospec-
tive examination of the relationships between
parent material, soil aggregation and SOM
dynamics as affected by forest clear-cutting and
broadcast burning (Borchers and Perry, 1992).
Two sites in southern Oregon (Siskiyou Na-
tional Forest) having similar management his-
tories and topography were selected. However,
soils on the two sites were formed from differ-
ent parent materials, a granitic and a metavol-
canicrockrype (seeHarradine and Jenny, 1958,
for a similar approach to isolating state factors).
Each site consisted of a poorly-revegetated
clear-cut dating from the late 1960s, paired with
an adjacent uncut forest (Figure 1). Because
these sites remain sparsely vegetated, most of
the large, visible soil aggregates have disap-
peared from the clear-cuts (Figure 2; Borchers
and Perry, 1990). In all likelihood, this was a
rapid event, coinciding with the decomposition
of roots and fungal hyphae (e.g., Tisdall and
Oades, 1982). Now, nearly 20 years after har-
vest, observations and SEM micrographs re-
veal a soil almost devoid of roots and fungal
hyphae.
The first experiment in this investigation deter-
mined C and N concentrations in several soil
particle-size classes obtained from a clear-cut-
forest pair representing each of the two parent
materials. Sonication was used to disrupt most
of the soil aggregate structure, followed by
sieving and sedimentation procedures to iso-
late particle-size classes. This revealed that the
metavolcanic soil (silt loam) was finer-textured
than the granitic soil (sandy loam), the greatest
contrast being in the relative proportions of
sand and silt (Table 1).
J.Q. BORCHERS
granitic soil (Table 1). A similar pattern involv-
ing soil texture and total C has been detected in
a number of temperate zone soils converted
from native vegetation to permanent cropping
(Mann, 1986). Analysis by particle-size classes
revealed that most of the significant C and N
losses in the metavolcanic soil were from silt
and clay fractions, whereas the granitic soil
sustained more limited losses from the sand
fraction.
The foregoing suggests that parent material
and soil texture represent important controls
over SOM decomposition and accretion rates.
Anderson and Paul (1988) hypothesized that
the turnover time of labile SOM could be ex-
tended to decades as the result of the physical
protection afforded by complexation between
organics and mineral particles (i.e., aggrega-
tion). Some indirect support for SOM seques-
tration exists in this study. As mentioned pre-
viously, disrupting soil aggregates by sieving or
grinding usually stimulates microbial activity.
To index microbial activity in the metavolcanic
and granitic soils, estimates of net mineraliz-
able N (by anaerobic laboratory incubation)
were made on intact (Nmin) and disaggregated
samples (Nmin*) (Figure 3). The mineralization
response to disaggregation (AN . =N . * -
j. <->*-> &> \ mm mm
Nmin) in the finer-textured metavolcanic soil
was nearly double that of the granitic soil,
indicating substantially greater amounts of physi-
cally-protected substrates. There is also statis-
tical evidence (Borchers and Perry, 1992) that
the N mineralized in response to sonication
originated from SOM associated with the very
particle-size fractions that exhibited significant
post-harvest depletions of C and N (Table 1).
Hence, this experiment may be a microcosm of
the slower process of soil disaggregation and
SOM decomposition that have takenplace since
forest harvesting.
The finer-textured metavolcanic soil had higher Although this experiment measured the miner-
initial C and N concentrations and was also alization of N rather than C, there is little
more prone to post-harvest decreases than the reason to suspect great disparity in their behav-
73
-------
STATE FACTORS, STEADY STATES AND SOIL ORGANIC-MATTER DYNAMICS AFTER FOREST HARVESTING
(a)
Figure 1. Old, poorly-vegetated clear-cuts in the Siskiyou National Forest on soils derived from (a) granitic and (b)
metavolcanic substrates.
74
-------
J.G. BORCHERS
8.0-
4.O-
o-
-4.0-
CHANGE IN DISTRIBUTIONS/
J* CD Ol pj co
o b b b b b
-4.0 -
-8.0 -
-12.0 -
-16.0 -
CEDAR CAMP
9,
jH
HOLCOMB PE^
^^
m
m
m
^X
///// ///// ///// V777/
^K ///////
./X/XI v^ ^ n^
..O-9.2 1.0-2.0 O.5-I.O 0.25-0.5_ 0.05-Q.25
<0.00]W
0.002 (c)
0.073 <">
0.077®
0.163
0.113
0.024(c)
0.032 (c)
a041 p < 0.10.
(c) Probability values p
-------
STATE FACTORS, STEADY STATES AND SOIL ORGANIC-MATTER DYNAMICS AFTER FOREST HARVESTING
Available
Radiation
Precipitation
Vegetation
f+)
Litter
Quality
CH4
production
oxidation
Figure 4. Conceptual model of factors affecting soil C storage.
materials were similar.
THE SOIL C STEADY STATE
The concept of a soil C steady state (Jenny,
1980) evades strict definition, but it still has
intuitive appeal. For example, global terres-
trial C models are simplified by assuming that
soil C levels of relatively undisturbed ecosys-
tems are at steady-state (Lugo and Brown,
1986; also termed soil C carrying capacity;
Johnson et al., 1991). Lugo and Brown (1986)
have questioned the validity of steady-state
assumptions, pointing out that C storage in
forest soils reflects not only current vegetation,
but also the cumulative effects of disturbance
events (e.g., climate change, fire, pathogens,
etc.) on species composition and a variety of
ecosystem processes (Figure 4).
To what extent does a physical protection mecha-
nism mediated by soil structure determine
steady-state soil C levels? The disaggregation
experiment reported here suggests that the
coarser-textured granitic soil is closer to a SOM
steady-state than the metavolcanic soil. The
existence of a strong linear correlation between
^mm anc^ Nmin in the metavolcanic soil (Figure
5) implies that variations in total SOM are
closely linked to variations in the quantity of
physically protected substrates. The absence of
a similar relationship in the granitic soil may
reflect lower silt and clay contents that com-
prise a physical protection mechanism that has
become "saturated".
76
-------
190-
170-
150-
f 110-
.£ 90-
Z 70-
50-
30-
10-
-10-
1
D •
D
• • GRANITIC
• FOREST
t1 O CLEARCUT
(51 O • METAVOLCANIC
n • FOREST
K Q CLEARCUT
° o o I o.
• °* • * *
* 0
0 30 50 70 90 110
Nmin (mg/kg)
Figure 5. Scatterplot of mineralizable nitrogen in soils
from forests and clear-cuts; Nmin, mineralizable nitrogen
in unsonicated soil; ANmin, mineralizable nitrogen in
sonicated soil minus Nmin.
J.G. BORCHERS
The contrast in SOM dynamics between these
two forest ecosystems supports Lugo and Brown
(1986) who questioned current assumptions of
soil C steady-state. A more realistic model of a
soil C steady-state would recognize other con-
straints such as disturbance frequency, soil par-
ent material and other factors that simulta-
neously impinge on ecosystem C flux. The
evidence here suggests that soil parent mate-
rial, texture and aggregation represent signifi-
cant (and possibly quite variable) constraints
on soil C accumulation. But in extremely wet,
cold, or drybiomes, aphysicalprotectionmecha-
nism may be less important to SOM dynamics if
extremes of temperature or moisture impair
the activity of autotrophs or heterotrophs. Under
such conditions, SOM levels may either fail to
approach or exceed the constraint offered by
soil aggregation.
REFERENCES
Anderson, D.W. 1979. Processes of humus formation and
transformation in soils of the Canadian Great Plains. J.
Soil Science 30:77-84.
Anderson, D. W. 1988. The effect of parent material and
soil development on nutrient cycling hi temperate ecosys-
tems. Biogeochemistry 5:71-97.
Anderson, D. W. and E. A. Paul. 1984. Organo-mineral
complexes and then* study,by radiocarbon dating. Soil Sci.
Soc. Amer. J. 48:298-301.
Borchers, J. G. and D. A. Perry. 1990. Organic matter
content and aggregation of forest soils with different
texture in southwest Oregon clear-cuts. In: Maintaining
the Long Term Productivity of Pacific Northwest Forest
Ecosystems, edited by Perry, D. A., R. Meurisse, B.
Thomas, R. Miller, J. Boyle, J. Means, C. R. Perry, and R.
F. Powers. Timber Press. Portland, OR.
Borchers, J. G. and D. A. Perry. 1992. The influence of
soil texture and aggregation on soil carbon and nitrogen
dynamics in southwest Oregon forests and clear-cuts.
Canadian Journal of Forest Research 22:298-305.
Detwiler, R. P. 1986. Land use change and the global
carbon cycle: The role of tropical soils. Biogeochemistry
2:67-93.
Dickson, B. A. andR. L. Crocker. 1953. A chronosequence
of soils and vegetation near Mt. Shasta, California. II. The
development of the forest floors and the carbon and
nitrogen profiles of the soils. J. Soil Science 4:142-154.
Greacan, E. L. and J. Williams. 1983. Physical properties
and water relations. In: Soils: An Australian Viewpoint.
CSIRO/Academic Press, pp. 499-530.
Harradine, F. and H. Jenny. 1958. Influence of parent
material and climate on texture and nitrogen and carbon
contents of virgin California soils. I. Texture and nitrogen
contents of soils. Soil Science 85:235-243.
Jenkinson, D. S. and J. H. Rayner. 1977. The turnover of
soil organic matter in some of the Rothamsted classical
experiments. Soil Science 123:298-305.
Jenny, H. 1980. The Soil Resource. Origin and Behavior.
Springer-Verlag, New York,p. 377.
77
-------
STATE FACTORS, STEADY STATES AND SOIL ORGANIC-MATTER DYNAMICS AFTER FOREST HARVESTING
Johnson, M. G., J. S. Kern, D. A. Lammers, J. J. Lee, L. H.
Licgel, K. Mattson, and P. Shaffer. 1991. Sequestering
carbon in soils: A workshop to explore the potential for
mitigating global climate change. EPA/600/3-91/031.
U.S. Environmental Protection Agency, p. 85.
Kcllog, W. 1983. Feedback mechanisms in the climate
system affecting future levels of carbon dioxide. J. Geo-
physical Research 88:1263-1269.
Lugo, A, E. and S. Brown. 1986. Steady state terrestrial
ecosystem and the global carbon cycle. Vegetation 68:83-
90.
Lynch, J.M.andE. Bragg. 1985. Microorganisms and soil
aggregate stability. Advances in Soil Science 2:133-171.
Mann, L. K. 1986. Changes in soil carbon storage after
cultivation. Soil Science 142:279-288.
McGill, W. B. and C. V. Cole. 1981. Comparative aspects
of cycling of organic C, N, S, and P through soil oganic
matter. Geoderma 26:267-286.
Oades,J.M. 1988. Theretentionoforganicmatterinsoils.
Biogeochemistry 5:35-70.
O'Neill, E. G., R. V. O'Neill, and R. J. Norby. 1991.
Hierarchy theory as a guide to mycorrhizal research on
large-scale problems. Environmental Pollution 73:271-
284.
Parton, W. J., J. W. B. Stewart, and C. V. Cole. 1988.
Dynamics of C, N, and P, in grassland soils: A model.
Biogeochemistry 5:109-131.
Pastor,!, and W.M. Post. 1986. Influence of climate, soil
moisture, and succession on forest carbon and nitrogen
cycles. Biogeochemistry 2:3-27.
Paton, T. R. 1978. The Formation of Soil Material.
George Allen and Unwin, Boston, p. 143.
Paul, E. A. 1984. Dynamics of organic matter in soils.
Plant and Soil 76:275-285.
Perry, D. A., J. G. Borchers, S. L. Gregory, D. P. Turner,
C. R. Perry, R. K. Dixon, S. C. Hart, B. Kaufmann, R. P.
Neilson, and P. Sollins. 1991. Biological feedbacks to
climate change: Terrestrial ecosystems as sinks and
sources of carbon and nitrogen. Northwest Environmen-
tal Journal 7:202-232.
Post, W. M., T.H. Peng, W. R. Emanuel, A. W. King, V.
H. Dale, and D. L. DeAngelis. 1990. The global carbon
cycle. Bioscience 78:310-326.
Quay, P.D., B. Tilbrook, and C. S. Wong. 1992. Oceanic
uptake of fossil fuel CO2: Carbon-13 evidence. Science
256:74-79.
Rastetter, E. B., M. G. Ryan, G. R. Shaver, J. M. Melillo,
K. J. Nadelhoffer, J. E. Hobbie, and J. D. Aber. 1991. A
general biogeochemical model describing the responses
of the C and N cycles in terrestrial ecosystems to changes
in CO2, climate, and N deposition. Tree Physiology 9:101-
126.
Rosencranz, A. and A. Scott. 1992. Siberia's threatened
forests. Nature 355:293-294.
Rovira, A. D. and E. L. Greacan. 1957. The effect of
aggregate disruption on the activity of microorganisms in
the soil. Aust. J. Agric. Res. 8:659-673.
Sollins, P., G. Spycher, and C. Topik. 1983. Processes of
soil organic-matter accretion atamudflowchronosequence,
Mt. Shasta, California. Ecology 64:1273-1282.
Tans,P.P.,I.Y.Fung,andT.Takahashi. 1990. Observa-
tional constraints on the global atmospheric CO2 budget.
Science 247:1431-1438.
Tisdall,J.M. andJ.M. Oades. 1982. Organic matter and
water-stable aggregates in soils. J. Soil Science 33:141-
163.
Toutain, F. 1987. Activite biologique des sols, modalites,
et lithodependence. Biol. Pert. Soils. 3:31-38.
Van Cleve, K., F. S. Chapin III, C. T. Dyrness, and L. A.
Viereck. 1991. Element cycling in taiga forests: State-
factor control. Bioscience 41:78-88.
Vil'yams, V. R. 1968. Basic soil science for agriculture.
Translated from Russian, Israel Program for Scientific
Translations, Ltd., p. 448.
Vitousek, P. M. 1991. Factors controlling ecosystem
structure and function. Agronomy Abstracts. American
Society of Agronomy, Madison, WI, p. 322.
78
-------
CHARACTERISTICS OF FOREST BIOGEOCENOSES
Natalie Pervova
ABSTRACT
One of the most important soil functions is the transformation of organic matter. The research was undertaken to demonstrate
the role of soils in different areas in the formation of gaseous hydrocarbons. The properties of forest biogeocenoses were
studied on sample plots in the Moscow region. Three sites in the Moscow suburbs were chosen: a demo-podzolic soil under
mixed forest; apeatygleyic soil (bog, moss, pine forest); and a humus-gleyic soil (forb meadow) on the border of a raised
bog. The movement of soil solutions was studied during long cycles by means of the lysimeter technique and the superseding
method. The carbon concentration in the soil solution fluctuated from year to year depending upon the composition and
thickness of the leaf'litterandthe intensityof'theleaf-litterdecompositionprocess. Themaximum carbon concentration (84.0-
232.0mg/l) occurredin the waters from forest litter. The determination of the biological activity oflitterby means of cellulose
destruction confirmed the field investigation results. It was shown that biogeocenoses had a definite gas regime. The gas
regime components varied greatly with the diffusion and changing climate conditions. The CO2 discharge from the soil
surface was the essential indicator in soil monitoring together with enzyme activity, nitrification rate and denitrification.
INTRODUCTION
The exploration of the different functions of
soils in ecosystems is an important and compli-
cated task. The soil is a valuable natural source
and storage for many compound substances.
One of the most important soil functions is the
transformation of organic matter. In the carbon
cycle, the most important functions are: the
migration of organic matter in solution through
the soil profile and the emission of carbon
dioxide and hydrocarbons.
The purpose of this paper is to demonstrate the
role of soils in different areas in the formation
of gaseous hydrocarbons.
SITE DESCRIPTION
The properties of forest biogeocenoses were
studied on sample plots in the Moscow region.
This region is situated in the subzone of south-
ern taiga in the western geomorphological area
of moraine plains on terraces of the ancient
Moscow River valley. It is located outside the
boundaries of the last glaciation. Three sites in
the Moscow suburbs were chosen: a derno-
podzolic soil under mixed forest; a peaty gleyic
soil (bog, moss, pine forest); and a humus-gleyic
soil (forb meadow) on the border of a raised
bog.
The climate is moderate-continental; the
monthly temperature in June is +18° C and in
winter, -10° to -11° C. The period of above-zero
temperatures lasts for 206-216 days. The amount
of annual precipitation is 620-640 mm. Most
precipitation occurs in summer. The moisture
coefficient is more than one during this time
period. Parent materials are cover loam,
fluvioglacial and ancient alluvial deposits.
Vegetation is represented by spruce and mixed
forests.
79
-------
aiARACTERISnCS OF FOREST BIOGEOCENOSES
As a result of complex conditions in the studied
region, soils have different structures, com-
pounds and properties. Podzolic and derno-
podzolic are the main soil types, with some
inclusion of brown forest soils. Carbon content
in these soils was 1.6%, 2.53% and 3.06% re-
spectively. All of the soils developed under the
water-washing regime. Other soils, such as hy-
dromorphic soils, also occur. The study covered
the soils of geochemically conjugated land-
scapes. The soil, to a great extent, determines
the productivity, structure and specific compo-
sition of biogeocenoses. Biogeocenoses de-
grades if soil conditions deteriorate. The an-
thropogenic load primarily affects soil biota,
gas and liquid phases of the soil.
LYSIMETRIC STUDIES
The movement of soil solutions was studied
during long cycles by means of the lysimeter
technique and the superseding method. The
last study was carried out with alcohol. In the
taiga zone, where the water-washing regime
predominates, all aspects of changes, transfor-
mations were related to mobile, water-soluble
compounds and the specifics of the biological
cycle. The amount of carbon in soil solutions
was determined in different seasons and years.
The carbon concentration in the soil solution
fluctuated from year to year due to the compo-
sition and thickness of the leaf litter and the
intensity of the leaf-litter decomposition pro-
cess. The variation coefficientis approximately
83-88% because of the extremely uneven tran-
sition of carbon into the solution.
To a great extent, this variation correlated with
the variability of meteorological conditions.
But despite the weather, vegetation and soil
distinctions, major trends appeared. The maxi-
mum carbon concentration (84.0-232.0 mg/1)
occurred in the waters from forest litter. The
amount of water-soluble organic matter in the
soil solutions and lysimetric waters and the
maximum microorganism content in the upper
soil horizon were definitely interrelated. The
highest mineralization occurred in the mixed
forests on the brown forest soils and the lowest
in the spruce forests on the podzolic soils (Table
1). The determination of the biological activity
of litters by means of cellulose destruction
confirmed the field investigation results. There
was 17.9% fabric destruction in the spruce litter
and 33.3% in the brown forest soil litter. It has
been shown by different scientists that the for-
est litter transformation on the soil surface
proceeds in mature coniferous forests with less
intensity than in the Moscow region forests.
Table 1. Statistical parameters of carbon concentra-
tion in lysimetric waters (mg/1).
Sample
plot
spruce forest
with Hylocomium
Dicranum
deep podzol
mixed spruce
forest-low
humic brown
forest soil
straw-yellow
derno-podzolic
soil
Depth
(cm)
0-4
0-35
0-2
0-12
0-2
0-9
x (mean)
spring autumn
63.4 47.4
23.0 26.0
29.2 33.9
25.2 14.0
13.9 30.5
13.2 25.6
Variation
spring autumn
74 83
38 58
74 71
62 32
70 76
55 88
The nature of the products formed was more
stable. In all observation periods, there was
maximum carbon content in the water; mean
concentration was 64.4 mg/1. The participation
of organic matter in plant metabolism was
determined to a considerable extent, by the
formation of the organic matter complex. Al-
most all mobile iron in the podzolic and derno-
podzolic soils was firmly tied up with the or-
ganic matter. There was a linear dependence
between iron content and organic matter con-
tent in the solutions when the iron content in the
soil solution was from 0.6 to 2.0 mg/1. The
mobility of iron-organic complexes in the light
granulometric textured soils of forest landscapes
was higher than in the heavy granulometric
textured soils. The intensity of iron, aluminum
80
-------
N. PERVOVA
and manganese migration was very high in the
complex with organic matter. Micro quantities
of iron, manganese, cobalt and cesium, which
are the radio-nuclead pollution components,
formed soluble complex compounds with soil
organic matter. Bicarbonate ion and calcium
ion predominated in the lysimetric water. The
bicarbonate ion concentration in the waters
from under the spruce litter was 22.4 mg/1 and
from under the mixed forest litter 40.8 mg/1.
The ratio HCO3:Ca2+, in the majority of cases,
was approximately 1 (Figure 1). It was obvious
that there was a buffering system to support the
above-mentioned relationship.
CARBON DIOXIDE EMISSIONS
In view of the fact that soil holds a key position
in biosphere gas-exchange regulation, the soil-
gas function study has particular interest. Soil
can be represented as a biospheric device for
the transformation of elemental compounds.
But, until now, it was not clear to what degree
the soil type determined whether the soil was an
absorber or generator of the gases. Perhaps the
soil can be depicted as a chromatographic col-
umn for the dissolved gases. Observations of
atmospheric gas composition performed on the
biogeocenoses level would reveal the role soil
plays in the biospheric gas exchange.
It was shown that biogeocenoses have a definite
gas regime. The gas regime components fluctu-
ated greatly due to the diffusion and changing
climate conditions. Together with enzyme ac-
tivity, rates of nitrification and denitrification,
the CO2 discharge from the soil surface was the
essential indicator in soil monitoring.
Carbon dioxide emission was studied by means
of the chamber-static method (chamber dimen-
sions were 150x150x150 mm). The investiga-
tion showed large variations in the quantities of
CO2 discharged from soil in the forest biogeo-
cenoses. The amount of CO2 released was 5.5-
8.0 kg/ha/hour (from podzolic and derno-
Ca .HCO, mg/1
Figure 1. Correlation between pH, HCO3^ and
in lysimetric waters.
podzolic soils). Coefficients of variation were:
• deep podzol, 57-70% (spruce forest with
hylocomium, Dicranum)
• deep podzol, 45-55% (bilberry spruce
forest)
• derno-podzolic soil, 38-45% (birch forest)
• shallow meadow soil, 35-40% (forb meadow)
Participation of deciduous species in parcel
formation led to the leveling of the CO2-dis-
charge intensity from soils. The CO2 emission
for the agrophytocoenoses of the Valday region
is estimated at 2-3 kg/ha/hour, and varied
greatly between the parcels. A similar range of
CO2 emission, 0.5-0.8 kg/ha/hour, is reported
by researchers in the Department of Soil Sci-
ence at Moscow State University. The CO2
emission from the soil to the atmosphere was
more evenly distributed in the hardwood for-
ests.
There were big differences in the soil respira-
tion between parcels; the greatest variation was
noted during the period of the most intensive
CO2 emissions. Fluctuations in CO2 emissions
were determined by uneven heating of upper
soil layers and varying thickness and different
biochemical properties of the leaf litters. An
increase of the leaf litter led to the increase in
81
-------
CHARACTERISTICS OF FOREST BIOGEOCENOSES
the variations of CO2 emissions. Researchers
at the Department of Soil Science at Moscow
State University came to the same conclusions.
For example, spruce parcels had a 53% sea-
sonal variation rate of CO2 emission, while oak
parcels had 31%. It is evident that in all cases,
the soil reacted to changes such as the elimina-
tion of the leaf litter, trampling down, decay,
etc. in the biogeocenoses condition within a
parcel. Long-term regime observations are
needed to accumulate correct data for a local
area. Because of different rates of modification
of different soil process components, there is a
need to understand the time limits necessary for
the process to come to equilibrium or quasi-
equilibrium with the environment. For ex-
ample, the monitoring of the hard-phase soil
condition may require long periods of time,
while monitoring of the liquid and the gaseous
phases must be conducted on an annual, sea-
sonal and daily basis.
The gaseous profile is an existing dynamic dis-
tribution in the concentration profile of a gas or
a volatile organic compound in a gaseous state.
The ground layer of the atmosphere, as well as
deep underground gas emanations, play a role
in modifying the environment of the gaseous
soil profile. At the same time, the gaseous soil
profile constitutes an open, unbalanced system
not only of geospheric gas rotation, but also of
intersoil gas redistribution. The study of the
CO2 distribution inside a profile of derno-
podzolicsoils incorporatedmembrane soil-sam-
pling sensors fixedat different depths in the soil.
The composition of soil air had a vertical strati-
fication determined by the production and ki-
netics of gases within the soil profile (Figure 2).
Observations were also done over derno-
podzolic soil at the soil test site at Moscow State
University. The CO2 concentration measure-
ments were taken at fixed time periods for
approximately 30 days inlate October and early
November. The following results were ob-
tained:
Given above-zero temperatures of the ground
layer of the atmosphere and upper soil layers,
maximum CO2 concentrations were found at
depths of 20 to 30 cm. Probably, the CO2
emission is hampered by the cooling of the soil
surface with high temperatures underground.
The process was accompanied by change in the
profile horizon. On the other hand, when the
upper-layer soil temperature fell below 0° C,
the character of the curve noticeably changed
(Figure 2 f, g and h). This may be related to the
change in the direction of biological processes
such as root respiration, fermentative activity,
etc.
The lowest rates of CO2 emission into the
atmosphere from derno-podzolic soils (F=0.26
-0.67 mg CO2/cm2/hr) were observed in the
late fall, while emissions from the same area
when soil and ground atmosphere tempera-
tures were above-zero comprised only 1.11-
2.40 mg CO2/cm2/hr. As the ground atmo-
sphere and soil temperatures approached 0° C,
a certain reduction in CO2 emission was ob-
served (Figure 3). The surge of CO2 emission
intensity was observed when the temperature of
the atmosphere ground layer fell below that of
the upper-layer soil. A linear dependency of
CO2 emitted from the soils into the atmosphere
from the temperature of the upper layer of the
soil took place. Yet such dependency was not
always noticeable. Agrophotocoenose study at
the Valday region has shown that curves which
depict the intensity of soil respiration coincide
with humidity curves and do not coincide with
temperature curves.
GASEOUS HYDROCARBONS
IN THE BIOSPHERE
Attention was focused lately on the study of
tendencies and processes related to global cli-
mate change and the greenhouse effect. High
rates of greenhouse gas accumulation in the
atmosphere was registered. The concentration
of atmospheric methane over the past decade
82
-------
N. PERVOVA
%vo!CO2
1 2
1 1 1 h
20
•$$40
60
80
(a) tair=+2.5° C, t^r+4.00 C (b) tair=+1.0° C, t^+1.50 C (c) tair=+5.5° C, tsoi,=+2.0° C
ar. , ^. air
it - 1 - 1 - 1 - 1 - H - 1 - 1 - 1 - 1
20
€40
60
80
(d) tair=+4.0° C, t^r+1.50 C (e) tair=+1.5° C,
20
•^40
S
60
80
1 - 1 - 1 - 1
1 - -H - 1
r+1.5° C (f) tair=-5.0° C, tsor-1.0° C
H 1 1 1 1 1 1—
(g) tai=-7.0° C, tsoil=-2.3° C (h) tair=-4.0° C, t^p-2.00 C (i) tair=+1.5° C, tsoil=+0.2° C
Figure 2. CO2 in the soil profile.
83
-------
CHARACTERISTICS OF FOREST BIOGEOCENOSES
t° C (soil)
Figure 3. Dependency of CO2 emissions on the temperature of the upper (0-10 cm) soil layer.
was increasing at the rate of 1%/year. The
supposition that maximum CH4 emission is
linked with rice fields has been proved incor-
rect.
Tracing the genesis of gaseous hydrocarbons in
the contemporary biosphere is an important
task, closely linked with studies of the carbon
biogeoehemical cycle and the influence of gas-
eous emissions of terrestrial ecosystems on
their composition. Soil-sampling sensors were
fitted along the profile of the soils, and during
a given time period, samples of soil air were
taken. Sample analysis showed that methane
and its homologues, ethane and propane, were
formedinnon-submerged soils and were present
there at concentration levels of 10"6 to 10'2 %
volume in the proportion of 100:(14-28) (Fig-
ure 4). The studies conducted have shown that
significant concentrations of ethane and pro-
sod-podzolic
hutnic-gleyic
peat-gleyic
j;j;j;!;!;j|j;!i!;!;!;!;i;i;!i!;i;i
•iiiliii-iji-iiHH&H-iHii-i-i:
jijijilHijijjjijIjijijijIjijijijil
! llliliiliill
ill ^^4
SC2H
-------
pane were in the soils, due not to emanation
from deeper layers, but to the processes taking
place in the soil itself.
It was determined that added moisture caused
a rapid increase in the concentration of meth-
ane and its homologues in the soil, with the
CH4, C2H6 and C3Hg concentrations at different
depths changing simultaneously. Evidently,
each gas (CO2, CH4, C2H4) was characterized
by its dimensional-temporal distribution form-
ing rate (it had an individual source functioning
in the soil profile). Apparently, hydrocarbon
emission was a significant source of carbon
depletion of soils. This indicates that methane
and its homologues were formed together in
one or several soil processes, as opposed to
previous theories on the emanation of these
compounds from deeper layers. Evidently, the
synthesis of hydrocarbon gas may occur not only
in an abiogenic way in the deeper layers of the
Earths sedimentary rock under high tempera-
ture and pressure, but also in the course of the
contemporary soil-forming process and the
transformation of carbon-containing substances
of soils.
Methane is practically omnipresent in the bio-
sphere, being a specific life product of a major
group of microorganisms. The distribution of
C2H6, C3H8, etc. in natural space is significantly
N. PERVOVA
smaller. It is considered that gaseous methane
homologues are formed exclusively at great
depths under high temperature and pressures
(mainly in the catagenetic stage of buried or-
ganic matter transformation). The genesis of
the closest methane homologues (primarily
ethane and propane) is less clear. These com-
pounds are not formed in soils, while their
presence in soil air is a consequence of the
emanation process from the lower concentra-
tions of hydrocarbon gas. It is worth mention-
ing, however, that previous studies were related
to humid and subaqual natural systems, charac-
terized by a regeneration regime and the preva-
lence of anaerobic conditions. Agroup of zonal
soils, vast both in their diversity and territory
remains practically unstudied. The soils, char-
acterized by the microheterogenic nature of the
oxidation-regeneration conditions, are of the
peculiar composition and structure of the mi-
crobial Cenoze.
CONCLUSION
It is extremely important to study all of the
components of the gaseous regime. Long-term
stationary observations in biogeocenoses of taiga
forests will contribute to the resolution of a
whole series of problems related to the negative
processes occurring in the biosphere.
85
-------
-------
DYNAMIC MATHEMATICAL MODEL FOR OXYGEN AND CARBON
DIOXIDE EXCHANGE BETWEEN SOIL AND THE ATMOSPHERE
Larry Boersma and Ying Ouyang
ABSTRACT
Exchange of oxygen and carbon dioxide gases between soil and atmosphere is controlled by atmospheric, soil physical and
soil biological conditions. Analysis and evaluation of problems in soil biology and soil ecology would be greatly facilitated
by the availability of a mathematical model that predicts concentrations ofO2 and CO2 as a function of time and soil depth
as well as the rate of gas exchange at the soil surface. For this purpose, we developed a time-dependent mathematical model
for the diffusion of O2 and CO2 through soils, with diffusion affected by changing water content due to rainfall, infiltration
and evaporation and by respiration of roots and soil microorganisms. Because of the large number and complexity of the
processes involved, we could not include all of them in the model. Our immediate goal was to develop a model structure
for analysis, by simulation, of certain scenarios of soil aeration. Of specific concern were root growth and soil aeration under
conditions where deficiency of O2 and/or toxicity of CO2 may occur, the investigation of the role of a crop canopy on
concentrations in the soil and the determination of whether or not CO2 behaves as an ideal gas in the soil.
DEVELOPMENT OF THE FIELD
EQUATIONS
Transport processes and sources and sinks in-
volved in the transport of O2 and CO2 through
soil, shown schematically in Figure 1, include
diffusion through the air-filled pore spaces;
consumption of O2 and production of CO2 by
plant roots and soil microorganisms due to
respiration; dissolution in soil water; physical
adsorption onto colloidal surfaces; and replace-
ment of air by water due to soil wetting or,
conversely, replacement of water by air due to
soil drying.
Factors that affect the rate of diffusion through
soil include soil water content and the soil
temperature; thus, rate of solar radiation, rain-
fall rate and duration, evaporation, air temper-
ature and relative humidity are important vari-
Soil Surface
Oa consumed and COa
produced by microorganisms
Qa and CQa absorbed
onto the colloidal surfaces
of soil particles
Qa consumed and COa
produced by roots
.Oaand COa dissolved
in the soil water
Figure 1. Schematic diagram of processes involved in the
exchange of oxygen and carbon dioxide gases between the
soil environment and the atmosphere.
87
-------
DYNAMICMATHEMATICALMODEL
Solar Radiation
Rainfall
a a
(Atmosphere)
Boundary Layer
6(Z,t) T(Z,t) Co2(Z.t) Cc02(2.t)
(Soil Slab)
n
z = o
Z
AZ + Z
z=;
Figure2. Schematic diagram of the conditions for which the
water.heat, oxygen, andcarbondioxidefieldequations were
developed. Symbols are defined and explained in the text.
ables. Atmospheric conditions control the soil
surface temperature and water content, thereby
affecting the rate of respiration as well as the
rate of diffusion through the soil.
Field equations for the simultaneous transport
of water, heat, O2 and CO2 were developed for
the conditions of Figure 2 showing a soil slab,
bounded below by an impervious layer and
above by the atmosphere. The symbols ©a, T3,
CO2, and Caco refer to relative humidity,
temperature, carbon dioxide concentration and
oxygen concentration of the completely mixed
air at the boundary layer.
Diffusion of Oxygen and Carbon Dioxide
For conditions without mass flow, the diffu-
sional flux of O2 can be expressed by:
where the subscript O2 denotes oxygen, q0 is
total diffusional flux (kg/m2/s), e is the soil
porosity (m3 voids/m3 soil), 6 is the volumetric
water content (m3/m3), qo air is diffusional flux in
the air-filled pore spaces (kg/m2/s), and q0 ,.
is diffusional flux in the soil water (kg/m2/s).
The diffusional fluxes in the air-filled pore
spaces can be expressed by (Stolzy and Fluhler,
1978)
(2)
with
and
(3)
(4)
where Co denotes concentration of oxygen
2
(kg/m3 air), 6Q is a coefficient characterizing
mobility in the soil (mol m2/J/s), |i0 is the
chemical potential (J/mol), u^ is the chemical
potential at standard state (J/mol), R is the gas
constant (J/mol/K), T is temperature (K), a0
is activity (dimensionless), and y0 is the activ-
ity coefficient (m3 air/kg).
Substitution of Eqs. (3) and (4) into Eq. (2)
yields:
.. (5)
By comparing Eq. (5) with Picks firstlaw (Crank,
1975), we have:
502RT,
(6)
where D0 is the diffusion coefficient (m2/s),
which upon substituting into Eq. (5) yields:
'2 3Cr
dz
(7)
88
-------
Because the diffusion of O2 in the liquid phase
is small in comparison with the diffusion in the
air-filled pore spaces (Glinski and Stepniewski,
1985), the liquid phase diffusional flux in Eq. (1)
will be ignored. Substitution of Eq. (7)intoEq.
(1) yields:
r-(e-0)D02[C02.
+1]
(8)
-0-,
Taking into account conservation of mass and
referring to Figure 2, the field equation can be
written as:
da
qo2*5 ds -/a (sro2
Re Co2) d",
(9)
f
for oxygen transport, where t is time (s), C0 is
the concentration in liquid phase water (kg/m3
"f
water), Co is the concentration absorbed onto
the colloidal surfaces (kg/m3 solid), n is the unit
vector going outward normal to the simple
closed surface 2, Q. is the arbitrary volume
element (m3), s is the area element
(m2), S0 and SQ° are the O2 consumption rates
2 2
by roots and soil microorganisms (kg/m3 air/s),
and Re is the change in air-filled pore space due
to replacement of air by soil water during infil-
tration (-) or the replacement of soil water by air
during evaporation drying ( + ) (m3 air/m3 soil/
s).
The concentration Cf in the left hand side of
Eq. (9) can be expressed by Henrys law as:
Cr
(10)
where H0 is the inverse Henrys law constant
(m3 air/m3 water). The concentration Csf in the
L. BOERSMA AND Y. OUYANG
left hand side of Eq. (9) can be expressed using
the Freundlich equation (Hiemenz, 1986) as:
c\ =co^ C1 ^111
O-j O-5 O«5S V /
where tt>0 is the adsorption coefficient (m3 air
/m3 solid).
Using the Gaussian theorem to convert the
surface integrals to volume integrals, multiply-
ing both sides of Eq. (9) by 1/Q, and letting D-
0+, yields limits to the oxygen field equation in
the differential form:
Substitution of Eqs. (8), (10), and (11) into Eq.
(12) yields the field equation for oxygen trans-
port through the air-filled pore spaces as fol-
lows:
.+ 1}
3C,
o,
LC \ + P f~* f11\
^O2J *S; ^O2- (JJ)
The development of the field equation for CO2
is analogous:
ac
" (Sco0
'CO-,
->2 o>_;2/ e i^u2
Transport of Water and Heat
(14)
Field equations for water and heat transport
used in this study were obtained by modifying
equations derived by Lindstrom and Piver
(1985), namely:
89
-------
DYNAMCMATOEMAHCALMODEL
Sz
for water and
4
ot
sat
(15)
cairpairT+0cwpwT]
(16)
for heat, where t is the time (s), pwis the density
of water (kg/m3), 0 is the volumetric water
content (m3/m3), p^J (T) is the density of water
vaporatsaturationattemperatureT(kgvapor/
m3 air), h is the relative humidity (dimension-
less), € is the soil porosity (m3 soil voids/m3
soil),Vj and Vv are the velocity vectors of water
in the liquid and vapor phases (m/s), respec-
tively, pw is the density of water vapor (kg
vapor/m3 air), c^^ is the specific heat of soil
particles (J soil/particle/K), pstdid is the density
of soil particles (kg/m3 solids), T is the tem-
perature (K), cair and cw are specific heats of air
and water (J/kg/K), respectively, pair is the
density of air (kg/m3 air), H^ is the vector of
heat conduction through the soil particles (J/
m2/hr), Hsl is the vector of heat conductionand
convection in the liquid phase (J/m2/s), H^ is
the vector of heat conductionin the vapor phase
and the transport of latent heat (J/m2/s).
Eqs. (13), (14), (15), and (16) describe the si-
multaneous transport of oxygen, carbon diox-
ide, water and heat through unsaturated soils.
ASSUMPTIONS AND BOUNDARY
CONDITIONS
Assumptions Pertaining to the Atmosphere
Daily cycles of air temperature relative humid-
ity and solar radiation can be represented by
mathematical functions for the purpose of
simulations. Air temperature and relative hu-
midity were characterized by Fourier Series,
and solar radiation was characterized by a
Gaussian normal distribution function. The
functions are approximations of the daily
changes as they occur in nature. The coeffi-
cients of the functions can be obtained by fitting
to locally obtained data sets (Ouyang and
Boersma, 1991a,b). When necessary or de-
sired, tabular data could be used. Functions
included in the model are for air temperature,
relative humidity, solar radiation, water loss
from the soil, water potential, hydraulic con-
ductivity and diffusion coefficient.
Assumptions Pertaining to Roots and
Microoganisms
Root respiration and growth rates were as-
sumed to be functions of time and concentra-
tion of O2 or CO2. A Q10 value of 2 was used to
account for the effects of temperature on the
root respiration rate. The rates of respiration
and growth of microorganisms were character-
izing by Monod Kinetics (Molz et al., 1986).
Soil pH and nutrient availability were assumed
to be constant. The effect of water content on
root growth was not included.
Rate of Oxygen Consumption. The rate of O2
consumption by roots was assumed to be a
function of soil O2 concentration and root age
and was represented by the logistic function
SRT(i)=-
A2
.th
(17)
1+B2EXP[-C2(C02-C02)]
where SRT(i) is the root O2 consumption rate
of the ith section of root length converted to kg/
m3 soil/s, A2, B2, and C2 are constants charac-
terizing the shape of the function (Ouyang and
Boersma, 1991a,b), C0 is the soil oxygen
concentration in the ith section of soil depth
corresponding to the ith section of root length
(kg/m3 soil air), and C* is the threshold soil
oxygen concentration below which the root
oxygen consumption rate approaches zero.
Co =50 kg/m3 air according to Glinski and
90
-------
L. BOERSMA AND Y. OUYANG
Stepniewski (1985). Values of Al, Bl, and Cl
were obtained by fitting Eq. (17) to experimen-
tal data reported by Glinski and Stepniewski
(1985). It was further assumed that the rate of
oxygen consumption decreased exponentially
from the root tip to the root base (Lemon, 1962)
according to:
"col
F3=A3 EXP(-B3*L) + C3,
(18)
where F3 is the correction factor, A3, B3, and
C3 are constants characterizing the shape of the
equation, andLis the rootlength (m), assuming
that the factor decreases exponentially with the
root length starting with unity at the root tip.
Multiplying Eq. (17) and Eq. (18) yields:
SRT(02,i)=-
A2
th .
F3, (19)
1+B2EXP[-C2(C02-C02)]
where SRT(O2,i) represents the rate of O2
consumption by roots as a function of soil O2
concentration and root length. Equation (19)
canbe modified to account for the effects of soil
CO2 concentration on the rate of oxygen con-
sumption by roots, by changing CA toClQ and
,th
th
Rates of Microbial Respiration. The rate of
respiration by soil microorganisms was assumed
to be a function of the oxygen concentration in
the soil air. With the assumption that the
oxygen required to produce the energy for gross
biomass production is proportional to substrate
utilization and that the oxygen required for the
energy of maintenance follows Monod Kinet-
ics, the rate of the respiratory use of O2 by soil
microorganisms can be expressed as (Molz et
al., 1986):
Smo=yoY0rs+aok0mc [-
(20)
•col
and
(21)
•col
Substituting Eq. (21) into Eq. (20) and replacing
Ccoibyco2obtains
Smo=
) + a0k0mc][_22_], (22)
D2
where Smo is the rate of respiration (kg/colony/
s), y0 is the oxygen use coefficient for the
synthesis of heterotrophic biomass (dimension-
less), (X0 is the oxygen use coefficient for the
energy of maintenance for heterotrophic mi-
croorganisms (dimensionless), k0 is the micro-
bial decay rate coefficient for aerobic respira-
tion (s"1), mc is the cell mass per colony (kg/
colony), Kg is the oxygen saturation constant for
decay (kg/m3 soil air), |j,0 is the maximum spe-
cific growth rate (s"1), s is the concentration of
substrate within the colony (kg/m3 soil), and
Ksb is the substrate saturation constant (kg/m3).
The diffusion of oxygen from the soil air-filled
pore space into the colony was assumed to be
rapid (Harvey et al., 1984). Therefore, the
concentration of oxygen within the colony (Ccol)
was assumed to be equal to the concentration of
oxygen in the soil air-filled pore space.
SIMULATIONS
Input parameters for the simulations included
the rainfall rate and duration, solar radiation,
air temperature, relative humidity, soil physical
properties and biological activities. These con-
ditions were represented by functions obtained
by fitting to average conditions for Western
Oregon. Values for soil properties that re-
mained constant during all simulations (Table 1)
were obtained from the literature reports cited
in the table. Values for parameters of equa-
tions describing rate of respiration and root
growth were obtained by fitting the equations to
data from literature reports. The initial concen-
trations for O2 and CO2 (Table 2) were set equal
to ambient values for the atmosphere so that
91
-------
DYNAMICMATHEMATICAL MODEL
Table 1. Values of the soil parameters that remained
constant throughout the simulations. Sources for these
values arc indicated.
Table 2. General input parameters for the simulations.
The initial concentrations for O2 and CO2 were set equal
to ambient values for the atmosphere.
Symbol
Psand
Psilt
Pclnv
VI WJ
Pair
Pwaler
c
alort
csand
csiU
cclay
*"*V
\vntcr
cair
<*»&•
"water
"soil
eair
cwatcr
CJ0jj
*solid
'Vatcr
^W
0)Q2
&
C02
Meaning
Density of sand
Density of silt
Density of clay
Density of air
Density of water
Total porosity
Tortuosity factor
Specific heat
of sand
Specific heat
ofsilt
Specific heat
of clay
Specific heat
of water
Specific heat
of air
Albedo of air
Albedo of water
Albedo of soil
Emissivity of air
above the soil
Emissivity of
water
Emissivity of
soil surface
Thermal
conductivity
of solids
Thermal
conductivity
of water
Thermal
conductivity
of air
Solid phase
O2 adsorption
Solid phase
CO? adsorption
Value/Units
2.66 g/cnv>
2.65 g/cm3
2.64 g/cm3
0.00 11 g/cm3
1.00 g/cm3
0.55
0.66
0.175ca1/g/°C
0.175 cal/g/°C
O.I75ca1/g/°C
1.0 cal/g/°C
0.24 cal/g/°C
0.05
0.07
0.09
0.9
0.95
0.5
18.9 cal/cm/hr
5.14 cal/cm/hr
0.2214
cal/cm/hr
3.4xlO'8 cm3
air/cm3 solid
3.4xlO'8 cm3
air/cm3 solid
Reference
Ghildyal and
Tripathi, 1987
Ghildyal and
Tripathi, 1987
Ghildyal and
Tripathi, 1987
Weast, 1986
Weast, 1986
McCoy etal., 1984
Hillel, 1982
Ghildyal and
Tripathi. 1987
Ghildyal and
Tripathi, 1987
Ghildyal and
Tripathi, 1987
Weast, 1986
Weast, 1986
Weast, 1986
Weast, 1986
Ghildyal and
Tripathi, 1987
Weast, 1986
Weast,~1986
Ghildyal and
Tripathi, 1987
Ghildyal and
Tripathi, 1987
Ghildyal and
Tripathi, 1987
Ghildyal and
Tripathi, 1987
Molz et al., 1986
Molzetal., 1986
simulations could show how rapidly equilib-
rium conditions in the soil would be approached.
Consecutive Periods of Infiltration, Evapora-
tion, and Redistribution (Scenario 1)
The first scenario was chosen for evaluation of
changes in oxygen and carbon dioxide concen-
trations as a function of soil depth during a
sequence of infiltration and evaporation events.
The rate of oxygen consumption by roots was
assumed to be a function of root length and
carbon dioxide concentration as described by
Parameter
Simulation
time
Rainfall time
Initial soil
water content
Initial soil
temperature
Initial O2
concentration
Initial CO2
concentration
Unit
hr
mm/hr
cm^/cm3
°C
fig/cir?
/jg/cm3
1
120
5
0.15
8.7
300
0.61
2
120
5
0.15
8.7
300
0.61
Scenario
3
72
5
0.15
8.7
300
0.61
4
72
5
0.25
8.7
300
0.61
5
72
5
0.15
8.7
changes
with soil
depth
changes
with soil
depth
Eq. (20) in Ouyang and Boersma (1991a,b) and
the rate of root elongation was assumed to be a
function of time and carbon dioxide concentra-
tion as described in the same report.
Soil Water Content. Changes in soil water
content and soil temperature during 120 hrs of
simulation are shown in Figure 3A. The simu-
lation started with a rainfall period from 0 to 12
hrs at the rate of 0.5 cm/hr. The rainfall rate
was lower than the infiltration capacity of the
soil, so that the infiltration profiles at 12 hrs did
not show the constant water content of a trans-
mission zone followed by a rapid decrease in
water content in the wetting zone or wetting
front.
Following the 12 hrs with rainfall was a period
of change in the water content of the soil profile
due to evaporation at the soil surface and fur-
ther redistribution of the water into the soil as
shown by the wetting front at 76 hrs. The
evaporative demand was quite low so that the
soil was always able to satisfy the evaporative
demand without substantial drying at the soil
surface. The pattern of rainfall followed by
evaporation and redistribution was repeated,
starting with rain during the period from 76 to
88 hrs. Redistribution continued, increasing
the water content to a depth of 80 cm at 120 hrs.
Soil Temperature. Daily cycles of soil tempera-
92
-------
L. BOERSMA AND Y. OUYANG
0.1
Water Content (cm 3 /cm 3)
0.2 0.3 0.4
£20-
f.40-
Q
5 60
80-
26
O 22-
S 18-
i
14-
8.'
10-
"o 6-
oj
Ohrsj ™*'\ |120hV
:. '. ''.'•'' s
j 12_W«_.«. 7
24
48 72
Time (hrs)
96
120
Figure 3. A. Water content as a Junction of soil depth at 0,12,
76,88and 120hrs. B. Soiltemperatureasafunctionoftimeat
the depths of 0,10 and 20 cm. Two periods of rainfall were
included. The first period of rainfall started at 0 hrs and stopped
at 12 hrs. The second period of rainfall started at 76 hrs and
stopped at 88 hrs. The rainfall rate was 05 on/hr.
ture at several depths are shown in Figure 3B.
This diagram shows that the temperature at the
soil surface was close to the temperature of the
ram during the first 12 hrs. Rain, which com-
pletely wetted the soil surface, decreased the
temperature of the soil surface layer to the
temperature of the rainwater at about 6 hrs.
The rainwater reached the 20 cm depth, de-
creasing the temperature of the soil of the
upper layers. As soon as the rain stopped at 12
hrs, radiation started to warm the soil surface
and then deeper layers. The change in soil
temperature during the period from 12 to 24 hrs
showed the characteristic soil temperature cycle
(Hillel, 1982). Simulations show the character-
istic decrease in amplitude and time lag with
increasing soil depth.
The temperature changes during the second
and third day show the more characteristic soil
temperature behavior, with a decreasing
temperature during the night followed by warm-
ing during the day. These changes more clearly
show the decrease in amplitude and increase in
lag time of the temperature as a function of soil
depth. The fourth day included a rainfall pe-
riod, which started at 76 hrs and ended at 88 hrs.
From 76 hrs to 80 hrs, the soil showed a cooling
profile with the lowest temperature at the soil
surface and the highest temperature at the 20
cm depth. This cooling resulted from the arrival
of the cold rainwater. Then at 88 hrs when
rainfall stopped, the temperature started to
increase. The rain stopped during a time of low
rate of radiation so that the temperature rise
was small. During the fifth day, the typical soil
temperature cycle occurred.
Evaporation. Daily cycles of rate of evapora-
tion are in Figure 4. Condensation of water onto
the soil surface occurred from 6 to 12 hrs during
the period of rainfall. The rate of surface
condensation is controlled by the gradients of
water vapor pressure and temperature between
the atmosphere and the soil surface. In this
simulation, it was assumed that during the rain,
the water vapor in the atmosphere was satu-
rated. The air temperature and solar radiation
were set at 50 percent and 20 percent of condi-
tions without rain (Ghildyal andTripathi, 1987).
The temperature of rainwater was set at 4°C.
When both the pressure of water vapor of the
air is higher and the temperature of the air is
lower in the atmosphere than it is at the soil
surface, water condenses onto the soil surface.
As soon as the rain stopped at 12 hrs, radiation
started to heat the soil. The change in the rate
of evaporation during the period from 12 to 24
hrsfollowedthecharacteristicevaporationcycle.
The rate of evaporation changed during the
second and third day showing the more charac-
teristic surface evaporation behavior, with the
rate decreasing during the night followed by a
gradual increase during the day.
During the fourth day, rain occurred from 76
hrs to 88 hrs. The evaporation rate was close to
93
-------
DYNAMIC MATHEMATICAL MODEL
-0.5
Time (hrs)
24 48 72 96 120
-0.4
j= -0.3-
ii
|| •az;
Si -0.1-
0.0-
0.1
Figure 4. Rate of surface evaporation (-) and condensa-
tion (+) as a function of time. Two periods of rainfall were
included. The first period started at 0 hrs and stopped at
12 hrs. The second period started at 76 hrs and stopped at
88 hrs. The rainfall rate was 0.5 cm/hr.
zero from 82 to 88 hrs due to the low tempera-
ture of the soil surface during the rainfall and
the lack of evaporative energy. The rain stopped
during a time when the rate of radiation was low
so that the rate of evaporation increase was
small. During the fifth day, the typical soil
surface evaporation cycle occurred.
Respiration by Roots and Microorganisms.
Rates of oxygen consumption by roots and
microorganisms (Figure 5) should be evaluated
in combination with Figure 6, showing O2 and
CO2 concentrations. The roots grew from 5 cm
at 12 hrs to 20 cm at the end of the 120 hr
simulation. The oxygen consumption rate was
a function of distance along the root length.
The highest rate was always at the root tip.
Because the rate of oxygen consumption is also
a function of CO2 concentration, the maximum
rate varied according to the carbon dioxide
concentrationin thesoil air. Therateatthe root
tip was highest with the simulation at 12 hrs,
namely 3.8 ug/cm3 air/hr and lowest with the
simulation at 120 hrs, namely 1.0 ug/cm3 air/hr.
Figure 5 shows thattherate of oxygen consump-
Respiration Rate
(ng/crrP air/hr)
01234. 0.31 0.33 . 0.35
0-
5
0.10-
0)
Q
OT 15-
20-
Roots
88 hrs/ ,'
120 hrs,
Microorganisms
Figure 5. Respiration rates by plant roots and soil micro-
organisms as a function of soil depth at times of 0, 12, 76,
88 and 120 hrs. The rate of oxygen consumption in j^g/cm
fresh roots/hr was converted to (J.g/cm3 air/hr.
tion by microorganisms changed with soil depth
and simulation time because the rate was a
function of oxygen concentration, which changed
with soil depth.
Concentrations of Oxygen and Carbon Diox-
ide. Changes in O2 and CO2 concentrations
(Figure 6) show the depletion of O2 due to
respiration by roots and microorganisms and
the concurrent increase of CO2. Oxygen con-
centrations decreased and CO2 concentrations
increased rapidly with soil depth from zero to 12
hrs, when the lowest oxygen concentration was
at 4.5 cm and the highest carbon dioxide con-
centration was at 4.0 cm. This rapid decrease in
oxygen and increase in carbon dioxide
concentrations occurred because the diffusion
rates of both gases were limited because soil
pore space was filled with water (Figure 3).
Aeration conditions improved after rainfall
ceased so that at 76 hrs the lowest oxygen con-
centration was higher and the highest CO2
concentration was lower than at 12 hrs. At
76 hrs, oxygen concentration decreased below
the 10 cm depth and carbon dioxide concen-
94
-------
L. BOERSMA AND Y. OUYANG
Oxygen Carbon Dioxide
(ng/cm3 air) (ng/crrPair)
210 250 290
0-
20-
E
,0,
£
Q.40-
o>
Q
'6
CO
60-
80-
-s***""^*^.
{ ^hrsTs^' 2
\ i *\ o
SShrs^s.^ :X \
s^\. '* \
76 hrs \N \
\\«
120hrs\V.
Ml
\
i;
5
0 20 40 60
- -
- -
*" I I
0 hrV^"^^ox
/•* 76 hrs '• %•
1 ^""'.'' '
•&''? '''
iCK^ j' •
• ** **
//'V* 120 hrs
/•'
I
li
1
Figure 6. Oxygen and carbon dioxide concentrations as a
function of soil depth at 0, 12, 76, 88 and 120 hrs since start
of simulation. Rainfall, evaporation and redistribution of
soil water occurred during the periods as shown in Figures
2 and 3.
Time (hrs)
24 48 72 96 120
'.Si-10-
D.
20-
30-
o 40-1
o
£T
50 V-
Figure 7. Root length as a function of tune since the start
of the simulations for Scenarios 1 and 2.
trations increased. These changes resulted
from the continued growth of roots and micro-
organisms. The root tip was at 5 cm at 12 hrs and
at 17 cm at 76 hrs (Figure 7).
From 76 hrs to 88 hrs, oxygen concentration
decreased dramatically in the upper 20 cm as a
result of rainfall during that period. Water
replaced soil air, decreasing the apparent diffu-
sion coefficient. After the rain stopped at 88
hrs, evaporation of soil water could again occur,
making more air-filled pore spaces available
for diffusion so that oxygen concentrations in-
creased and carbon dioxide concentrations de-
creased.
Effectof Oxygen Concentration onRootGrowth
(Scenario 2)
For this scenario, the assumption was made that
root elongation rate and root oxygen consump-
tion rate are functions of the oxygen concentra-
tion in the soil pores instead of being functions
of carbon dioxide concentration. Figure 7 shows
that the roots grew slower when controlled by
carbon dioxide concentration (Scenario 1). Root
length at 120 hrs was about 21.0 cm in Sce-
nario 1, but about 40.0 cm at the same time in
Scenario 2. As more carbon dioxide was pro-
duced by roots and microorganisms, the rate of
root elongation decreased. The maximum car-
bon dioxide concentration in the soil profile
was about 60 ug/cm3 air at 120 hrs, and the
correction factor for the root elongation rate
was 0.12 at this carbon dioxide concentration
(Figure 8). For Scenario 2, the root elongation
rate was assumed to decrease with oxygen
concentration. The minimum oxygen concen-
tration in the soil profile was 235 (J.g/cm3 air at
120 hrs, and the correction factor for the root
elongation rate was about 0.9 at this oxygen
concentration (Figure 8). The lower the correc-
tion factor became, the slower the roots grew.
Presence of a Crop Canopy (Scenario 3)
This scenario was chosen to evaluate the effects
of the presence of a crop canopy on the O2 and
CO2 concentration at the soil surface and con-
sequently on concentrations in the soil (Figures
9 and 10). Crop leaves fix carbon dioxide and
evolve oxygen by means of photosynthesis and
95
-------
DYNAMICMATHEMATICALMODEL
1.0
50
100
150 200 250 300
Oxygen Concentration (|ig/cm3 air)
O-O-h-
0 20 40 60 80 100
Carbon Dioxide Concentration (ug/cm 3 air)
Figure 8. A. Correction factor for the root oxygen
consumption rate as a function of oxygen concentration.
B. Correcrionfactor for the root oxygen consumption rate
as a function of carbon dioxide concentration.
consume oxygen and produce carbon dioxide
by the processes of photorespiration and respi-
ration. These activities affect the oxygen and
carbon dioxide concentrations at the soil sur-
face. Transport of oxygen and carbon dioxide
within the crop canopy was assumed to occur by
diffusion only. Mass flow of oxygen and carbon
dioxide induced by air movement within the
canopy was not considered. Concentrations of
oxygen and carbon dioxide within the crop
canopy were assumed to be controlledbyphoto-
synthesis, photorespiration and respiration of
crop leaves.
Field Equations. Field equations for the diffu-
sion of oxygen and carbon dioxide within the
crop canopy were obtained by modifying
Eqs. (13) and (14). Eliminating the terms for
soil water content, soil porosity, adsorption of
oxygen and carbon dioxide by the surfaces of
soil particles, dissolution of oxygen and carbon
Oxygen
(|ig/cm3 air)
280 t 290
Carbon Dioxide
(ng/cm3 air)
300 0 10 20 30
Figure 9. Oxygen and carbon dioxide concentrations as a
function of plant height and soil depth at 24,48, and 72 hrs
(midnight). Initial oxygen and carbon dioxide concentra-
tions were 300 and 0.6134 fig/cm3 air, respectively.
dioxide in soil water and sink and source of
roots and microorganisms from Eqs. (13) and
(14), setting the activity coefficient equal to
unity, and adding the terms for photosynthesis,
photorespiration and respiration obtains:
3C,
o,
3z
(23)
for oxygen and
3
f
3t
(24)
for carbon dioxide, where Cn andCro are the
°2 C02
concentration of oxygen and carbon dioxide
within the crop canopy (ng/cm3 air), D0 and
Dco are the diffusion coefficients of oxygen
and carbon dioxide within the crop canopy
(cm2/hr), z is the height of the crop (cm), Pn0
and Pnco are the rates of oxygen evolution and
2 !
carbon dioxide fixation by photosynthesis (ug/
cm3/hr), and Res0 and Resco represent the
rates of oxygen consumption and carbon diox-
ide production by photorespiration and respi-
96
-------
L. BOERSMA AND Y. OUYANG
300
24 36 48
Time (hrs)
60
72
Figure 10. A. Oxygen concentration as a function of time
at the soil surface from 0 to 72 hrs. B. Carbon dioxide
concentration as a function of time at the soil surface from
0 to 72 hrs.
ration (ug/cm3/hr). Eqs. (13) and (14) together
with Eqs. (23) and (24) describe the diffusion of
oxygen and carbon dioxide within the crop
canopy and soil system.
All the input parameters were the same as for
Scenario 1 except solar radiation, air tempera-
ture, relative humidity, wind speed, diffusion
coefficients of oxygen and carbon dioxide and
initial soil water content (Ouyang, 1990). Solar
radiation and wind speed at the soil surface
were set to zero.
Ideal Versus Non-ideal Behavior of Carbon
Dioxide (Scenario 4)
Possible differences in rate of diffusion result-
ing from ideal versus non-ideal behavior of
carbon dioxide were evaluated by two simu-
lations. The first simulation assumed that both
Co"D-CO,SP(ug/crn3 air)
0.020
30
Figure 11. Difference between the carbon dioxide concentra-
tions of the non-ideal and ideal hypothesis as a function of soil
depth. Cco to Cco are the concentrations of carbon
dioxide for the non-ideal and ideal conditions, respectively.
oxygen and carbon dioxide molecules behave as
ideal gases. The second simulation assumed
that the oxygen molecules behave as an ideal
gas, while carbon dioxide molecules behave as
a non-ideal gas. The activity coefficient of
carbon dioxide was chosen to be a function of
carbon dioxide concentration (Ouyang, 1990).
Results (Figure 11) showed that the maximum
difference in carbon dioxide concentration be-
tween the non-ideal versus ideal simulation was
about 0.0175 |-ig/cm3 air at the depth of about
25 cm. This is quite small in comparison with the
carbon dioxide concentration 0.6134 ug/cm3
air in the atmosphere. Such a small difference
would not have a significant effect on root
growth (Nobel and Palta, 1989).
SENSITIVITY ANALYSIS
The sensitivity of the model simulations to
input parameters is of interest if the model is
used at the field scale where flow and transport
parameters change spatially and temporally. A
preliminary sensitivity analysis was performed
with respect to soil temperature and maximum
root respiration rate. Simulations were con-
97
-------
DYNANDCMATOEMA'nCALMODEL
ductedfor48 his with 0 = 0.25 cm3/cm3. No soil
microorganisms were involved. The initial soil
oxygen and carbon dioxide concentration were
thesameasinScenariol. The reference values
for temperature and maximumrespiration rate
were 20°C and 8 ug/cm3 air/hr, respectively
(GUnsMandStepniewski, 1985). AQ10 value of
2 was used for the temperature effect on root
respiration rate.
Results (Table 3) show that by decreasing the
soil temperature 75%, the oxygen concentra-
tion was 16.69% higher, and carbon dioxide
concentration was 98.95% lower. Increasing
the soil temperature by 75% decreased O2
concentration3.4% and increased CO2 concen-
tration 6.5%. These changes occurred as a
result of the temperature effect on root respi-
ration rate. As the temperature decreased, the
root respiration rate decreased resulting in less
use of oxygen and less production of carbon
dioxide.
The temperature increase of 50% increased the
oxygen concentration 0.7% and decreased the
carbon dioxide 3.9%. This trend is opposite to
the results when the temperature was increased
by 75%. This difference in response between
the two temperatures is due to the effect of
temperature on the diffusion coefficients. As
soil temperature increases, the diffusion coeffi-
cients of both gases increases. When the rate of
oxygen diffusion coefficient is higher than the
rate of root respiration at that temperature
(30°C), the above results could be obtained.
The sensitivity of the model was related to
temperature. The percent change in oxygen
concentration was greater than the percent
change in carbon dioxide concentration at low
temperature, however the opposite occurred
with rate of root respiration. These results
point to the need for improving the experimen-
tal measurement techniques so that accurate
field assessment of the processes can be ob-
tained.
Table 3. Changes in oxygen and carbon dioxide concen-
trations resulting from the indicated changes in input
conditions at the soil depth of 20 cm after 48 hrs of
simulation. The base temperature was 20°C and the base
respiration rate was 8.0 Hg/cm3 air/far.
Parameter Percent change
in input value
Temperature
Temperature
Temperature
Temperature
Temperature
Maximum
respiration rate
Maximum
respiration rate
Maximum
respiration rate
Maximum
respiration rate
Maximum
respiration rate
-75
-50
0
SO
75
-75
-50
0
50
75
Percent changes in
Cbcygen
16.69
10.66
0
0.7
-3.4
12.37
8.09
0
-6.85
-13.00
concentration
Carbon dioxide
-98.95
-63.34
0
-3.9
6.5
-73.39
-48.15
0
41.59
78.53
SUMMARY AND CONCLUSIONS
We developed a one-dimensional mathemati-
cal model for the exchange of oxygen and car-
bon dioxide between soil and atmosphere as a
function of time. The model includes effects of
rainfall, infiltration, evaporation, the transport
of soil water and the respiratory activities of
roots and soil microorganisms. This model
consists of four partial differential field equa-
tions that describe the time-dependent simulta-
neous transport of water, heat, oxygen and
carbon dioxide through soils. Mass flow of
oxygen and carbon dioxide through soil was not
included because the theoretical and experi-
mental bases to describe the gas flow velocity in
the natural soil has not been developed. Sce-
narios were chosen for simulations using the
mathematical model. Simultaneous transport
of water, heat, oxygen and carbon dioxide
through a loam soil during infiltration, redistri-
bution and evaporation periods was evaluated,
and diffusion of oxygen arid carbon dioxide
within a crop canopy and soil system was exam-
ined. Several different functions for the root
elongation rate, root oxygen consumption rate
and respiration by microorganisms were used.
Root elongation rate was chosen to depend on
oxygen or carbon dioxide concentrations. Root
98
-------
L. BOERSMA AND Y. OUYANG
oxygen consumption rate was assumed to be a Solutions illustrate that the model may be used
function of root length and oxygen or carbon for an evaluation of water, heat, oxygen, and
dioxide concentrations. carbon dioxide fields, although experimental
validation remains to be done.
REFERENCES
Ghildyal,B.P.andR.P.Tripathi. 1987. Soil physics. John
Wiley & Sons, New York, p. 665.
Glinski, J. and W. Stepniewski. 1985. Soil aeration and its
role for plants. CRC Press, Inc., Boca Raton, Florida,
p. 229.
Harvey, R.W..R.L. Smith, andL. George. 1984. Effect of
organic contamination upon microbial distributions and
heterotrophic uptake in a Cape Cod, Mass., aquifer. Appl.
and Environ. Microbiol. 48:1197-1202.
Hiemenz, P.C. 1986. Principles of colloid and surface
chemistry. Marcel Dekker, Inc., New York and Basel,
p. 815.
Hillel, D. 1982. Introduction to soil physics. Academic
Press, Orlando, FL.
Lemon, E.R. 1962. Soil aeration and plant root relations.
I. Theory. Agronomy J. 54:167-170.
Lindstrom, F.T. and W.T. Piver. 1985. A mathematical
model of the transport and the fate of toxic chemicals in a
simple aquifer. Tech. Rep. No. 52, Oregon State Univer-
sity, Dept. of Mathematics, CorvaUis, Oregon.
McCoy, E.L., L. Boersma, M. J. Ungs, and S. Akratanakul.
1984. Toward understanding soil water uptake by plant
roots. Soil Science 137:69-277.
Molz, F. J., M. A. Widdowson, and L. D. Benefield. 1986.
Simulation of microbial growth dynamic coupled to nutri-
ent and oxygen transport in porous media. Water Resour.
Res. 22:1207-1216.
Nobel, P. S. and J. A. Palta. 1989. Soil O2 and CO2 effects
on root respiration of cacti. Plant and Soil 120:263-271.
Ouyang, Y. 1991. Dynamic mathematical model of oxygen
and carbon dioxide exchange between atmosphere and
soil. Ph. D. Thesis. Oregon State University, CorvaUis, OR
97331.
Ouyang, Y. and L. Boersma. 1992. Dynamic mathemati-
cal model of oxygen and carbon dioxide exchange between
soil and atmosphere: I. Model development. SoilScLSoc.
Am. J.
Ouyang, Y. and L. Boersma. 1992. Dynamic mathemati-
cal model of oxygen and carbon dioxide exchange between
soil and atmosphere: II. Model application. Soil Sci. Soc.
Am.J.
Stolzy, L.H. and H. Fluhler. 1978. Measurement and
prediction of anaerobiosis in soils. In: Nitrogen in the
environment. Volume 1. Nitrogen behavior in field soil, ed.
by D. R. Nielsen and J. G. MacDonald. Academic Press,
New York, pp. 363-426.
Weast, R.C. 1986. Handbook of chemistry and physics.
CRC Press, Boca Raton, FL.
99
-------
-------
CARBON STORAGE IN PEAT BASED ON
REGIONALITY OF RUSSIAN MIRES
Marina S. Botch
ABSTRACT
A great amount of carbon is stored in peat. Mires contribute to global carbon change. Carbon pools in peat depend on mire
types and their productivity. The global mire area may be as much as 500 million ha, and about one-third of this is located
in the former Soviet Union. Large areas ofpeatlands are located in the tundra and forest zones. The regionality ofpeatlands
is represented by seven latitudinal mire zones and several groups of provinces. The estimation of carbon pools in peat must
be based on mire types and their features because they are strongly distinguished, not only by the lands of peat, but also by
the rate of accumulation. The rate of annual peat accumulation can be estimated at 0.7mm and it is one of the possible
mechanisms that could help moderate the disturbed global equilibrium between released and fixed CO?
Non-labile soil carbon makes up more than
one-half of the terrestrial carbon pool in the
former Soviet Union (Kolchugina and Vinson,
1991, this volume). A major amount of carbon
is stored in peat. Mires contribute significantly
to global carbon change. The amount of global
carbon in peat is estimated at 110 Gt (Krapivin
et al., 1982) or 300 Gt (Armentano, 1980).
Carbon pools in peat depend on mire types and
their productivity. This paper is devoted to
peatland regionality in the former Soviet Union
and characteristics of different mire types, which
reflect climatic features.
Mires are peat-forming ecosystems. Mires can
be subdivided into peat-forming bogs, which
are ombrotrophic,/ens that are minerotrophic
treeless mires rich in herbs, sedges and grasses,
marshes, which consist of tall sedges and grasses,
and swamps, or carrs, which are wooded fens
and composed of trees.
The global mire area may be as much as 500
million ha, and about one-third of this is located
in the former Soviet Union. Large areas of
peatlands are located in the tundra and forest
zones. The dry weight of the world's peat is
approximately 600 Gt, and about half of it
consist of carbon. In the former Soviet Union,
there is about 200 Gt of peat. The regionality
ofpeatlands is represented by seven latitudinal
mire zones and several groups of provinces.
These are shown in Table 1 and Figure 1. Every
zone is subdivided into provinces based on non-
zonal features.
The northern zone corresponds to the tundra
area. The main type of mire is a polygonal mire
of complex structure. It consists of polygons,
which are 10-30 m in diameter. They are
characterized by wet hollows dominated by
grasses and sedges and dry ridges around every
hollow. The ridges are covered with dwarf
shrubs and bryales mosses. The polygonal
mires are frozen and have a shallow peat layer,
with a thickness of 0.2-0.5 m. The area of mires
in this zone varies from 10 to 70%.
101
-------
CARBON STORAGE IN PEAT BASED ON REGIONALITY OF RUSSIAN MIRES
f
T-l
W)
a
102
-------
M.S. BOTCH
PH 6 § ^
•rt -J- - ™
•a &, -a ^ 49 >> g fi
=3 tr a "O
1
eg
CO
00
00
8
I H
II
00 OQ
.a .a
t! tJ
H J i
IB g e
00
•S
§
co
103
-------
CARBON STORAGE IN PEAT BASED ON REGIONALITY OF RUSSIAN MIRES
Table 1. Mire zones in Eastern Europe and Western
Siberia and their corresponding vegetation zones.
Mire Zones Vegetation Zones
1. Zone of polygon mires
2. Zone of palsa mires
3. Discontinuous zone of aapa mires
4. Zone of raised string bogs
5. Zone of pine bogs and fens
6. Zone of rccds and sedge fens
7. Zone of fresh and salt water
marshes
Tundra
Forest tundra
Tundra
Forest tundra
Northern taiga
Northern taiga
Northern taiga
Middle taiga
Southern taiga
Southern taiga
Mixed Forest
Decidious forests
Forest steppe
Steppe
Semidesert
Desert
The next zone is characterized by palsa mires,
which are located in the forest-tundra and north-
ern taiga. The area of mires in this zone is
approximately30-50%. Thevegetationofpalsas
is composed of dwarf shrubs, sphagnum and
bryales mosses and lichens. Wet hollows be-
tween them are covered with cotton-grass,
sedges and sphagnum. The depth of peat varies
from 1-5 m, and it consists of sphagnum and
herbs remains.
Aapa-rnires are typical for the northern taiga.
They are located in the Kola and Karelia re-
gions, in the East European part of the former
Soviet Union, west of the Ural mountains and
can be found in the Kamtchatka Peninsula.
Aapa-mires are distinguished by sphagnum
ridges covered with sedges, herbs and bog shrubs
and wet hollows with herbs located between
ridges. Theirpeat depositsbelong mainly to the
eutrophic sedge type; their depth varies from 1-
7m.
Raised-string bogs are the main mire type of
Russia. They cover the boreal zone of the
former Soviet Union. They have a convex
cupola built up of oligotrophic sphagnum peat.
On the surface of bogs, a complicated network
of strings (ridges), hollows and deep pools has
developed. The depth of the peats, which
consist of sphagnum remains, varies from 1-
7(10) m. The percentage of mires is 10-40%.
The next zone is represented by pine bogs and
fens. Bog surface features are poorly devel-
oped. Pine bogs have pine peat 1-2 m in depth.
Eutrophic sedge peat is typical for fens. The
mires here cover only from 3-5%.
The zone of reed and sedge fens is spread over
the southern part of Russian and West Siberian
plains, where steppe vegetation prevails; the
percentage of mires is very low, 0.5-2%.
The last zone is represented by fresh- and salt-
water marshes. It is located in desert- and semi-
desert areas. Scirpus, Phragmites, Typha and
other tall sedges and grasses prevail in marshes.
The group of continental provinces of Siberia
and the Far East includes the mires of the the
East mountain area, which is poor in mires. The
latter belong to the fen type. Special attention
must be paid to Far East mires located in the
region of the Amur River. It is highly paludified
with fens and sphagnum bogs, which are called
mart. They are covered with larch (Lam
gmelinii), bog shrubs, sedges and sphagnum
mosses.
The group of maritime provinces of the Far
East includes mires of the Kamtchatka Penin-
sula, and the Sakhalin and Kurile islands. The
northern parts of Sakhalin and Kamtchatka are
represented by aapa-fens and palsas. Their
middle parts are covered with blanket bogs.
Western Kamtchatka is highlypaludified; mires
cover about 90% of the territory. They are very
wet blanket bogs with deep peat (5-7 m).
High mountain provinces are distributed
through all the mountain regions. Mires are
very rare here, but they canbe found at different
104
-------
Table 2. Plant biomass and annual primary production in
mires of different climatic zones.
MS. BOTCH
Table 3. Plant biomass and annual primary production of
some mire sites.
Phytomass Primary Production
t/ha Mt t/ha/yr Mt/yr
Polar belt (humid 25 470 2.2
and semihumid
mires on
permafrost soils)
Boreal belt 35.0 3892.0 3.5
Subboreal belts 40.0 64.0 25.0
(humid)
Subboreal belt 15.0 85.5 7.0
(semiarid,
herbaceous bogs)
Subtropical belt 200.0 5380.0 130.0
(humid, meadow
bogs and bog
formations)
Tropical belt 300.0 19920.0 150.0
(humid)
41.4
389.2
40.0
39.9
3497.0
9960.0
Data from Rodin et al. (1975).
altitudes and belong to different types, among
them bogs, fens and palsas.
The largest area of peat deposits in the boreal
zone is represented by appa-fens and bogs. It is
strongly paludified and characterized by deep
peat deposits. The productivity of mires in
different zones is shown in Tables 2 and 3. Low
production is typical for the northern mire types
(polygonal, palsas, aappa); swamps and marshes
of the southern regions typically show higher
production data.
The estimation of carbon pools in peat must be
based on mire types and their features because
they are strongly distinguished, not only by the
kinds of peat, but also by the rate of accumula-
tion. Unfortunately, the mires of the largest
area of the former Soviet Union have not yet
been studied (Figure 2). This prevents more
detailed discussion about the peat features of
Russia.
As shown at the beginning of this paper, peat is
very rich in carbon withdrawn from the atmo-
sphere during the post-glacial period. Accord-
Province Mire site
and region
43 Karelia, Aapa-fen(a)
Kndasovo Mesotrqphicswamp(a)
Oligotrophic
string bog (a)
4-6 West-Siberia Eutrophic swamp (b)
Tomsk region Eutrophic swamp (c)
Eutrophic fen (c)
Mesotrophic swamp (b)
Mesotrophic swamp (c)
Mesotrophic fen (b)
Oligotrophic
pine bog (b)
Oligotrophic
pine bog (d)
Oligotrophic
pine bog (c)
Oligotrophic
tree-less bog (b)
SJ Poles'ye Rich fen (e)
Phytomass
(S/m)
1000-1710
2300-4700
1170-1580
17330
17750
900
16160
17700
900
4570
1910-5810
17420
990
—
Net primary
production
(g/m /yr)
390-690
750-1000
440-600
780
720
—
600
640
—
380
210-420
450
—
290-520
(a) Yelina and Kuznetsov, 1977.
(b) Pyavchenko, 1967.
(c) Glebov and Toleiko, 1976.
(d) Valutski and Khramov, 1976.
(e) Parfenov and Kim, 1976.
ing to Armentano (1980), accumulation of car-
bon by peat is 135 Mt/year. Sjors (1980) esti-
mated this accumulation at 90 Mt. He based
this value on an average depth of peat in global
mires of as much as 1 m. The rate of annual peat
accumulation can be estimated at 0.7 mm and
is one of the possible mechanisms that could
help moderate the disturbed global equilib-
rium between released and fixed CO2.
If the world's peatlands were all drained, this
absorption of CO2 would cease. According to
Krapivin et al. (1982) and Masing et al. (1990),
a 10% decrease in wetland area caused a simu-
lated 11 Gt of C stored in peat to be released to
the atmosphere, depending on the rate and type
of mire exploitation. Sjors (1982) believes that
burning peat as fuel and using it for agriculture
has faster effects than drainage for afforesta-
tion. Also, the climate is of importance because
oxidation is much faster in hot climates than in
105
-------
CARBON STORAGE IN PEAT BASED ON REGIONALITY OF RUSSIAN MIRES
1
cs
if
II
'•
106
-------
cool ones. Notmore than20% of virginpeadands
of the world can be drained because of the
inaccessibility of most peatlands. Hie role of
peatlands for short-term turnover of CO2 is
almost negligible.
The mire-preserving program is the best way to
conserve peat. This program has been in effect
from the end of the 60's. The total protected
MS. BOTCH
mire area in the former Soviet Union consists of
about 2% of Russian mires. They are protected
in nature preserving areas, which belong to the
following categories: "zapovedniki," "zakazniki,"
nature monuments and national parks. The
area of protected mires differs in different
nature regions (Masing et al., 1990). These
areas need to be increased.
REFERENCES
Armentano, T.V. 1980. Drainage of organic soils as a
factor in the world carbon cycle. Bioscience Dec 1980:825-
830.
Botch, M.S. and V.V. Masing. 1983. Mire ecosystems in
the USSR. In: Mires: Swamp, Bog, Fen and Moore,
edited by A.P. Gore, Ecosystems of the world, Vol 4B,
Regional studies. Amsterdam, New York, pp. 95-152.
Bramryd,T. 1979. The conservation of peatlands as global
carbon accumulators. In: Classification of Peats and
Peatlands. IPS, Helsinki, pp. 297-305.
Burke, M.J. and C. Stuschnoff. 1979. Carbon dioxide
exchange between peat, vegetation and atmosphere in the
wet coastal tundra of Alaska. In: Stress Physiology in Crop
Plants, edited by H. Mussel etal. New York, pp. 197-225.
Kolchugina, T. and T. Vinson. 1991. Framework to
quantify the natural terrestrial carbon cycle of the former
Soviet Union. This volume.
Rrapivin, V.F., Y.M. Svirezhev, and A.M. Trarko. 1982
Mathematical modelling of global biosphere processes.
Moscow.
Masing, V., Y.M. Svirezhev, M. Loffler, and B.C. Patten.
1990. Wetlands in the biosphere. In: Wetlands and
Shallow Continental Water Bodies, edited by B. Patten
et.al., Vol 1. The Hague pp. 313-344.
Rodin, L.E., NJ. Bazilevich, and N.N. Rozov. 1975.
Productivity of the world's mam ecosystems. In: Produc-
tivity of world ecosystems, edited by D.E. Reichle etal.
Washington, pp. 13-26.
Sjors, H. 1982. The zonation of northern peatlands and
their importnace for the carbon balance of the atmo-
sphere. In: Wetlands:Ecology and Management, edited
by B. Gopal etal. Tjaipur, pp. 11-14.
107
-------
-------
METHANE FROM NORTHERN PEATLANDS
AND CLIMATE CHANGE
Stephen Frolking
ABSTRACT
Methane fluxes from northern peatlands (north of 45° N) are about 32 Tg CH4peryear, about 7% of the total methane flux
to the atmosphere (Harriss et al., 1991; Watson et al., 1990). Measured methane fluxes have shown dependence on both soil
temperature and soil moisture (e.g., Crill et al, 1988; Bartlett et al., 1991), and thus are sensitive to climate change. Ice core
records from Vostok, Antarctica show methane concentrations to be correlated with air temperature over glacial/interglacial
cycles (Chappellaz et al., 1990). A model of warm season methane-flux dependence on air temperature was developed, based
on heat diffusion into the soil and measured relationships between methane flux and soil temperature. Using historical air
temperature records for Bethel, Alaska, and Marcell,Minnesota, the model indicates a strong sensitivity to the range of climate
variation that has occurred this century. For both sites, the modeledfluxfor the warmest summer (JJA) on record is more than
twice as great as the modeledfluxfor the coolest summer on record. The modeled flux for the warmest summer is also more
than 50% greater than the modeled flux for the mean summer (1951-1980). Consideringfive consecutive years, the modeled
flux for the warmest half-decade is more than 50% greater than the modeledfluxfor the coolest half decade at each site. The
observedtemperaturevariationsarewithintherangepredictedforaCO2doublingbycurrentglobalcUmatemodels.Eightmaior
regions of boreal and sub-arctic peatlands were considered in a study of larger scale flux variations. For each, a representative
area, thaw season length and methane flux rate are coupled with reconstructed summer temperature anomaly records for this
century to look at temperature-induced variations in methane fluxes from northern peatlands. In extreme years, the modeled
fluxes vary by up to 30%; more generally, variations are within 0-12% from the normal flux. The model indicates that northern
wetland methane flux response to initial climate change would not be likely to cause a strong positive climate feedback and
thus would not greatly accelerate rates of climate change. However, northern wetland methane fluxes may provide a good early
warning indicator of climate change. Methane fluxes play a significant role in the carbon balance of peatlands and changes
in flux rates will be likely to cause changes in the carbon balance.
INTRODUCTION
Methane is an important constituent in the
global atmosphere, even though it is only present
in trace amounts (1.7 ppmv-parts per million
by volume; Watson etal., 1990). Rodhe(1990)
estimates the greenhouse effect of methane to
be 15% of the total effect, second in strength
only to carbon dioxide. In addition, methane
plays important roles inboth tropospheric (lower
atmosphere) and stratospheric (upper atmo-
sphere) chemistry. Methane emissions to the
atmosphere are the net product of anaerobic
microbial methane production and potential
subsequent oxidation by microbes thatuse meth-
ane as an energy source. For example, in a
wetland environment, production of methane
will take place in water-saturated, anaerobic
zones of the soil. As the methane diffuses to the
atmosphere through aerobic zones in the soil,
or by transport via plants (Dacy and Klug, 1979;
Sebacheretal., 1985; Chanton and Dacy, 1991),
microbial oxidation can occur, reducing the net
flux to the atmosphere (e.g., Whalen and
Reeburgh, 1990a).
109
-------
METHANE FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
Total methane flux to the atmosphere is about
550 Tg/yr; freshwater peatlands are considered
to be a major source (Watson et al., 1990;
Cicerone and Oremland, 1988). Combining
detailed studies of the distribution and total
area of freshwater peatlands with the current,
albeit limited, data on methane emissions from
these environments has produced an estimated
total annualmethane-flux to the atmosphere of
115Tg/yr (Matthews andFung, 1987; Aselmann
and Crutzen, 1989). These studies show large
areas of wetlands concentrated in the boreal/
sub-arctic (45-70°N) and in the tropics (10°N-
20°S). The latest extrapolations of methane
flux studies estimate that the boreal/sub-arctic
methane-flux is about 32 Tg/yr, with slightly
more than hah0 coming from the boreal region
(Harriss et al., 1991). Flux studies to date all
demonstrate at least a qualitative dependence
on climatic factors, particularly soil tempera-
ture and moisture. Global climate model
(GCM) simulations also predict that the ex-
treme air temperature response to a global
warming induced by increased concentrations
of greenhouse gases will occur in northern high
latitude, mid-continent regions (e.g., Mitchell,
1989), where freshwater wetland environments
are common.
Boreal and sub-arctic peatlands have a soil
carbon pool of about 450 Pg C, and the esti-
mated accumulation rate is about 0.075 Pg/yr
(Gorham, 1991). In these units, the methane
carbon flux out of these peatlands is estimated
at 0.024 Pg/yr, and is thus a major component
in the carbon balance of these peatlands. Any
change in the methane flux from northern
peatlands due to climatic change will play an
important role in the carbon balance of these
ecosystems.
A CONCEPTUAL FRAMEWORK
FOR MODELING CLIMATE-INDUCED
METHANE FLUX CHANGES
Thereis intriguing evidence from deep-ice cores
(Chappellaz et al., 1990), that atmospheric
methane concentration is positively correlated
with climate variations, or at least surface
temperature variations. The ice-core record
from Vostok, East Antarctica (Figure 1), ex-
tending back about 160,000 years, shows six
distinct peaks in atmospheric methane concen-
tration (at levels of 0.6 - 0.7 ppmv), separated by
periods of lower concentration (about 0.4 - 0.5
ppmv). The surface-temperature record con-
tained in the ice core, determined by the deute-
rium (2H) to hydrogen (1H) isotopic ratio, is
noisier, but contains four clear peaks at the
same times as four of the methane peaks and
oscillations at the other two methane peaks.
Atmospheric methane concentrations are high-
est during interglacial (warm) periods and low-
est during glacial advance (cold) periods. It is
interesting to note that the ice core data indi-
cate that atmospheric methane concentrations
during past geologic warm periods (and even as
recently as the 1600's) was approximately 0.7
ppmv, compared with the current concentra-
tion of 1.7 ppmv (KhalilandRasmussen, 1987).
The existing ice-core record is not of sufficient
resolution to imply a cause and effect relation-
ship between methane concentration and mean
surface-temperature. However, aspectral analy-
sis of the methane concentration time series
yielded four significant peaks with periods of
122.0,41.7,24.8, and 19.0 thousand years (kyr)
(Chappellaz et al., 1990). These are similar to
the periods of the major astronomical climate
forcings or Milankovitch cycles (Henderson-
Sellers and McGuffie, 1987). These astronomi-
cal climate forcings are clearly external to the
climate system, and cannot be a feedback re-
sponse to another climate forcing.
On the other hand, the atmospheric methane
concentration could change in response to cli-
mate change. A changing climate will have both
an immediate and a delayed impact on a wet-
land ecosystem. The immediate response will
be a change in the wetland soil thermal and
hydrological regimes-the soil climate. Over
110
-------
S. FROLKING
-------
METHANE FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
TERRESTRIAL METHANE FLUX CONTROLS
air temperature
precipitation
radiation
season length
wind speed
SURFACE CLIMATE
(SOIL CLIMATED;
soil temperature
soil moisture
permafrost depth
METHANE PRODUCTION ,
AND OXIDATION
aerobic/anaerobic controls
soil temperature controls
nutrient availability
methane diffusion paths
METHANE FLUX TO
ATMOSPHERE
Figure 2. A schematic model of the climate controls on
terrestrial methane production. Additional controls (e.g.,
nutrient availability) are lumped into the methane produc-
tion and consumption box. The models in this paper focus
only on air temperature, soil temperature and the tem-
perature controls of methane production.
dra ecosystems in Alaska, a continuous perma-
frost region, methane fluxes again show a posi-
tive correlation with soil temperature (Figure
3b) (Bartlett et al., 1991). In addition, Yukon-
Kuskokwim Delta fluxes decreased by a factor
of five as the water table dropped from the soil
surface to a depth of about 20 cm (Bartlett et al.,
1991). At some sites, wetland soils may even
become sinks (rather than sources) for atmo-
spheric methane during dry periods (Whalen
and Reeburgh, 1990a). In contrast, methane
emissions from 'kettle' bog environments in
northern Minnesota showed little response to a
drop in the water table during the summer
period (Grill et al., 1988).
In summary, each regional wetland ecosystem
will have unique characteristics determined by
the interaction of vegetation, topography, hy-
drology, climate and other environmental vari-
ables. A quantitative global inventory of meth-
ane emissions from wetland ecosystems will
require regional studies to determine relation-
ships between the dominant forcing variables
(e.g., vegetation type, soil temperature and
water content) and methane emissions to the
atmosphere.
MODELING THE SENSITIVITY
OF METHANE FLUXES
TO CLIMATE CHANGE
Because there are, as of yet, relatively few
measurements of methane flux from northern
peatlands, none before 1969 (Clymo and
Reddaway, 1971), and none at a given site for
more than a few consecutive years (Whalen and
Reeburgh, 1988; Dise, 1991b), to get a broader
view of the temporal variability of methane
flux, I must resort to modeling. Measurements
from northern peatlands have shown that both
soil temperature and soil moisture, among other
factors, have a strong effect on methane fluxes.
The multi-year studies at a fixed site (Whalen
and Reeburgh, 1988; Dise, 199 Ib) show varia-
tion in annual flux from one year to the next that
will be due in large part to climate variability.
The soil climate is driven primarily by the
atmospheric climate, for which there is some
record over the past century. I have constructed
a model of the effect of air temperature on
methane fluxes, using air temperature records
and heat diffusion principles to calculate soil
temperatures, and empirical relationships be-
tween soil temperature and methane fluxes. In
this model, climate variability and change are
characterized by the historical temperature
record, as opposed to climate model predic-
tions of future climate scenarios.
112
-------
S. FROLKING
1000
100
10
FORESTED BOG
OPEN BOG
"•:i^
10
15
(a) Soil Temperature (°C)
100-
10-
O UPLAND TUNDRA
6 WET MEADOW TUNDRA
8
10
(b) Soil Temperature, 10-20 cm (°C)
Figure 3. Methane flux rates plotted as a function of soil temperature from (a) two sites in Minnesota (from Crill et
al., 1988), and (b) the Yukon-Ruskokwim Delta in Alaska (from Bartlett et al, 1991).
Climate History for Two Northern Sites
The Marcell Experimental Forest (47°32N,
93°28 W) in north-central Minnesota is a south-
ern boreal ecosystem with numerous small for-
ested and non-forested bogs and fens (Crill et
al., 1988). Bethel, Alaska, is in the sub-arctic
coastal tundra (60°45N,161°50W) on the
Yukon-Kuskokwim River Delta (Bartlett et al.,
1991). I have constructed monthly mean air
temperatures for each site by combining 1951-
1980 monthly mean temperatures for Bethel,
AK, and Grand Rapids, MN, (Ruffner, 1985)
with monthly air temperature anomalies for
1900-1987, as compiled by Hansen and Leb-
edeff (1987). The monthly anomalies are for
grid cells of about 200 x 200 km, constructed
froni all available temperature records within
1200 km of the center of the cell. Details are
discussed in Hansen and Lebedeff (1987).
These two sites were chosen because methane-
flux data at each site demonstrated a flux de-
pendence on soil climate (soil temperature
and/or soil moisture) (Crill etal., 1988; Bartlett
et al., 1991). Soil climate, in turn, depends on
the surface climate (air temperature, precipita-
tion, insolation, wind speed, humidity, etc.). I
113
-------
METHANE FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
combined the air-temperature record with ob-
served flux-temperature relationships to con-
struct past flux scenarios. Because northern
wetland methane flux is predominantly a sum-
mer season phenomenon (Whalen and
Reeburgh, 1988), and the flux measurements
occurred during the summer months, I will
focus on the summer temperatures. Figure 4
shows the constructed summer (mean of June,
July and August) temperatures for both Bethel
and Marcell for 1900 through 1987.
In applying this model to Bethel, AK, and
Marcell, MN, I considered five scenarios for
each site:
1) the "mean year" (not an actual year but a year
made up of the 1951-1980 monthly means);
2) the year with the warmest summer (June,
July, August);
3) the year with the coolest summer;
4) The warmest half-decade (five consecutive
years with the warmest average of the five
summers);
5) the coolest half-decade.
I then fit daily mean air temperatures to the
monthly mean temperatures for each of these
years and calculate soil temperatures and meth-
ane fluxes for each year (see the next section for
details of the model).
ASimple Model of Methane-Flux Dependence
on Air Temperature
Soil temperature over a season is driven prima-
rily by the soil surface energy balance. The
surface air temperature is often a good proxy
for the soil surface-temperature (which is rarely
measured), and thus for the surface energy
balance, especially when averaging over many
days for the seasonal signal (Sellers, 1965). I use
a simple model of soil temperature driven by
Summer (mean of JJA) air temperatures
temp. (C)
1900
1920
1940 1960
year
1980
Figure 4. Reconstructed summer air temperatures (mean
of June, July, and August) for Bethel, AK, and Marcell,
MN. Curves are combination of Hansen and Lebedeff
(1987) temperature anomalies and 1951-1980 monthly
means (Ruffner, 1985). The 1951-1980 summermeans are
represented by horizontal lines.
surface air temperature, to be coupled with the
soil temperature-methane flux relationships, to
give us a model of methane flux based on air
temperature.
Vertical heat transfer in soils can be modeled by
the heat diffusion equation
dr/dz = a d2T/dz2
(1)
where T is the soil temperature (° C), t is time
(s), z is depth in the soil (cm) and a is the soil
thermal diffusivity (cm2/s). This equation as-
sumes that the soil thermal properties are uni-
form with depth (up to several meters).
If a sinusoidal soil surface-temperature, Tsurf,
Tsurf(t) = Tmean + T* sin(wt + f) (2)
is applied to a uniform soil profile, then the soil
temperature at depth is given by
T(Z,0 = Tmean
+ T* exp(-z/D) sin(wt + f - z/D) (3)
where Tmean is the mean temperature of the
period (e.g., diurnal or annual), T* is the ampli-
114
-------
S. FROLKING
tude of the temperature oscillation, w is the The phase shifts, fn, are given by the sinusoidal
frequency of the oscillation, f is a phase shift fit to the monthly mean air temperatures, em-
(related to the time of maximum temperature) phasizing a fit to the summer (or high flux)
and D is the damping depth (e-folding depth) of months. Dn is the damping depth for the wave
the oscillation, where with frequency wn.
D = (2a/w)V2
Equation 4 shows that the damping depth of the
surface signal is shorter for higher frequency
oscillations (larger w). Choosing a = 0.001
cm2/s as a typical thermal diffusivity for wet
peat (Hillel, 1980), the damping depth for diur-
nal oscillations is about 5 cm, and for annual
oscillations, it is about 100 cm. Because both
the diurnal and annual temperature oscilla-
tions have about the same amplitude (on the
order of 10°C), below about 10 cm in wet peat,
most temperature variation will be due to the
annual signal.
To make use of the available data, I fit a
sinusoidal function to the mean monthly air
temperature data and consider that to repre-
sent the mean daily air temperature. To more
closely fit the air temperature data during the
summer months, I used an annual sine wave
plus the first two harmonics (periods = 183 days
and 91.5 days). These harmonics are also sine
functions and will propagate into the soil in the
same manner, though with damping depths that
decrease with the square root of the frequency.
The modeled air temperature is given by
T --
~
mean
where Tn is the amplitude of the annual (n= 1)
or harmonic (n=2,3) waves, wn is the frequency
and fn is the phase shift for each wave. The
temperature at a depth d in the soil is then given
by
T(z,t) = Tmean +Zn=i,3Tn
sin(wnt + fn-.z/Dn)
(4) Methane flux measurements near Bethel, AK,
(Bartlett et al, 1991) and in the Marcell Forest
(Crill et al., 1988) correlate with soil tempera-
ture. I used fits to these correlations to empiri-
cally predict methane fluxes from modeled soil
temperatures. The flux/temperature equations
used are
Bethel: log(F) = 0.9594 + 0.1442T15 (7)
Marcell: ln(F) = 67.9 - [17893/(T10+273.2)] (8)
where F is the methane flux (mg CH4/m2/d),
T15 is the daily mean soil temperature (°C) at 15
cm depth and T10is the daily mean soil tempera-
ture (°C) at 10 cm depth. I assume that when the
soil temperature at depth is equal or below 0°C
there is no methane flux. Combining historical
temperature records with equations 6,7, and 8
I can estimate methane fluxes for various years.
DISCUSSION OF MODEL RESULTS
Extreme Years
For both Bethel and Marcell, the annual mod-
eled methane flux for the year with the warmest
summer is more than twice as large as the year
with the coolest summer (Figure 5a). The
warmest summer annual flux is more that 50%
greater than the mean year flux for both sites,
and the coolest summer annual flux is about
70% of the mean year flux for both sites. The
temperature changes from the normal summer
to the warmest summer are 1.6°C for Bethel
and 1.9°C for Marcell; both changes are well
within predicted greenhouse warming over the
next century (Mitchell, 1989). These two wet-
(6) land systems appear to be quite sensitive to
temperature change (both predicted for the
future and realized in the past)~assuming the
115
(5)
-------
METHANE FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
100
SO-
methane (lux
60-
40-
20-
20.3
18.2
Q normal year
El warmest summer
Q coolest summer
15.5
11.5
10.3
Marcell MN Bethel AK
a) Variation in annual methane flux
for normal and extreme summers.
500
400-
methane flux
{g CH4/hi*2)
300-
200-
100-
19.5
E3 warmest half-decade
E) coolest half-decade
12.0
11.0
Marcell MN Bethel AK
b) Mean summer (JJA) temperature
anomaly for eight northern sites.
Figure 5. (a) Modeled methane fluxes (g/m/yr) for
normal and extreme years for sites in Alaska and Minne-
sota. Numbers by each bar are summer (JJA) mean
temperatures (°C). Warm summers are: Marcell, MN,
1936; Bethel, AK, 1977. Cool summers are: Marcell, MN,
1915; Bethel, AK, 1922. (b) Modeled methane fluxes for
warmest and coolest half-decades for sites in Alaska and
Minnesota. Numbers by each bar are mean summer
temperatures for the five consecutive years (°C). Warm
half decades are: Marcell, MN, 1933-37; Bethel, AK, 1977-
81. Coolhalf-decades are: Marcell,MN, 1924-28; Bethel,
AK, 1929-33.
soil moisture regime does not change.
Five Consecutive Years
When the model is run for five consecutive
years (eachyear is modeled individually and the
five years are summed), there is still quite a
large difference in predicted methane flux be-
tween warm and cool periods (Figure 5b). For
Bethel, the modeled warm period (1977-1981)
produced 67% more methane flux than the
modeled cool period (1929-1933). For Marcell,
the modeled warm period (1933-1937) pro-
duced 54% more methane flux than the mod-
eled cool period (1924-1928). As the time
period that the fluxes are integrated over in-
creases, the difference between warm and cool
periods decreases, but it is still quite pronounced
over a five-year period.
SCALING UP FLUX ESTIMATES TO ALL
NORTHERN PEATLANDS
There is much uncertainty about the biospheric
or ecosystem response to climate change and its
potential as a climate feedback mechanism
(e.g., Lashof, 1989). I feel that northern
peatlands and their trace-gas fluxes offer an
important ecosystem in which to investigate
this issue. The physics of soil climate require
that it integrate the more variable surface at-
mospheric climate, damping out the high fre-
quency variability, and responding more strongly
to trends. Wetland methane flux response to
temperature and hydrological change is strong
and rapid, as it is a change in process rates
rather than a change in ecosystem type or
function. Northern peatlands may thus serve as
an early warning indicator of significant climate
change. In addition, predicted climate change
is large for northern high latitudes, and there is
an enormous carbon pool in boreal and tundra
wetland soils (Bouwman, 1990), so their poten-
tial as a greenhouse gas source is large. To
create a strong signal in the global atmospheric
methane budget, large regions of the boreal and
arctic peatlands will have to simultaneously
experience similar climate deviations. For ex-
ample, a summer that is very warm over eastern
Canada and very cool over the Siberian low-
lands may have no net effect on the global
methane budget. Has this happened in the
past?
116
As this is an initial study of the large-scale
-------
S. FROLKING
1) Siberian Lowlands
2) Fenno-Soviet Lowlands
3) Nouveau Quebec
4) Hudson Bay Lowlands
5) Boundary Waters Area
6) Central Alaska
7) North Slope Alaska
8) Yukon-Kuskokwim Delta
Figure 6. Delineation of the eight northern wetland regions modeled.
climate-induced variability of methane fluxes
from northern peatlands, I considered the ef-
fects of temperature variability only. Northern
peatlands may be grouped into eight regions:
the Siberian Lowlands, the Hudson Bay Low-
lands, the Fenno-Soviet lowlands, the Alaskan
North Slope, the Yukon-Kuskokwim Delta, the
boundary waters area of Minnesota, Central
Alaska andNouveau Quebec (Figure 6). These
regions represent most (but not all) of the
freshwater peatlands in the boreal and arctic
regions. They are also regions for which at least
some methane-flux studies have been reported
(except for the Siberian Lowlands, which, be-
cause of their vast size, cannot be ignored).
For each of these regions, a total inundated
wetland area is determined from the l°xl°
fractional inundation data set compiled by
Matthews and Fung (1987), taking into account
that the area of a l°xl° grid cell goes as the
cosine of its latitude. For each of the regions, I
assigned a representative wet-site methane flux
and an estimated flux season length (Table 1).
For the Siberian lowlands, for which I know of
no flux data, I assigned a flux equal to the mean
117
-------
METHANE FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
Table 1. Northern wetland regions data.
Region
Siberian
lowlands
Fcnno-Sov.
lowlands
Hudson Day
lowlands
Boundary
Waters area
North Slope
AK
Central AK
Yukon-Kusk
Delia
N'ouvcau
Quebec
Lat.
Long.
55-73N
60-90E
50-TON
20-50E
50-60N
75-100W
45-SON
85-100W
68-71N
140-165W
60-68N
140-155W
55-65N
155-170W
SQ-STN
60-75W
ArcaW
(109 m2 )
723
634
342
98
71.2
46.7
56.5
30.5
Rep. CH, flux
(mgC^, /m2/d)
80.K")
57-8W
24.0:<1)
188.0W
io3.cco
32.2fe>
86.5W
70.0CO
Season
length
(days)
120
120
120
150
100
120
120
120
Regional
CH4 flux
(Tg)
7.0
4.4
1.0
2.8
0.73
0.18
0.59
0.26
(«} Arcii are Inundation areat from peatlands data base ot Matthews and Fung (1987).
(b) So diia available, value is mean of other 7 values.
(c) Svcnnoa and Rosswall, 19S4; Crill, unpublished data.
( SL is the region's season length (d), AT
118
-------
S. FROLKING
summer temperature
anomaly (C)
R. Barrow AK (70.9N.157.5W)
- Fairbanks AK (64.3N.146.3W)
• Bethel AK (61.9N.159.8W)
Marcell MN (48.4N.92.3W)
Schefferville, PQ (55.1N.65.3W)
Fenno-Sov. lowlands (68.6N.22.5E)
" Hudson Bay lowlands (55.1N.87.8W)
/^jAsA-^A^ - Siberian lowlands (59.6N.87.8E)
1900
1920
1940 1960
year
1980
a) Summer (JJA) air temperature anomalies
from 1951-1980 summer means.
(from NASA GISS data set)
i.o
0.5
summer air
temperature o.o
anomaly (C)
-0.5
-1.0
-1.5
1900
1980
b) Mean summer (JJA) temperature
anomaly for eight northern sites.
Figure 7. (a) Summer (mean of June, July and August) air temperature anomalies (°C) for a site in each peatland region.
Latitude and longitude values are for the center of the grid cell for each anomaly; (b) The mean summer air temperature
anomaly for the eight regions.
119
-------
METHANE FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
is the region's summer temperature anomaly
(°C), and AF is the region's variation from
normal methane flux (Mg CH4), or methane-
flux anomaly, for the year. Figure 8a shows this
methane-flux anomaly for the eight northern
wetland regions modeled. The greatest flux
anomaly occurs in the Siberian Lowlands, due
to their large area and assumed moderate rep-
resentative flux strength. The large Hudson
Bay Lowlands have a much smaller flux anomaly
because of the reported low methane fluxes
from this region (Moore et al., 1991). The
Boundary Waters Area of Minnesota, Ontario
and Manitoba have large flux anomalies due to
their high reported fluxes (Grill et al., 1988;
Dise, 1991b) and their longer season length. In
addition, Dise (1991a) reports significant win-
ter methane fluxes in this region, although these
arenotaccountedforinthismodel. TheFenno-
Soviet Lowlands have a large anomaly due to
their large area and moderate flux strength. As
discussed above, all regions have roughly the
same temperature variations (magnitude, but
not timing) so no one region's methane-flux
anomaly is enhanced or reduced with respect to
the others due to climate variability.
Figure 8b shows the sum of these eight regional
anomalies, or the net northern wetland mod-
eled methane-flux anomaly. Superimposed on
this curve is a nine-year runningmeanof the net
flux anomaly to account for the integrating
effects of methane's lifetime in the atmosphere.
In an individual year, the anomaly (positive or
negative) can be as much as 5 Tg CH4, or 29%
of the total normal flux of the model. Most
variations are about 0-2 Tg, or 0-12% of the
totalmodelflux. The nine-year smoothed curve
is usually within 1 Tg, or about 6% of the
modeled normal flux. Scaling this up to the
current estimate of 32 Tg CH4/yr from north-
ern peatlands, the annual temperature-driven
methane-flux anomaly would be about 0-4 Tg.
The nine-year smoothed anomaly would vary
by up to about 0-2 Tg.
During the initial stages of a climate warming
(i.e., perhaps now or soon) the air temperature
anomalies can be expected to be variable and,
for the northern wetlands regions, probably
only up to about 1°C. This model would thus
predict atemperature-induced increase in meth-
ane flux from northern wetlands of about 5-10
Tg/yr. This is about 1-2% of the total annual
methane-flux to the atmosphere and only about
10-20% of the current rate of increase of meth-
ane in the atmosphere. Thus, northern wetland
methane-flux increases due to the initial stages
of climate warming are not likely to cause a
major feedback, a major acceleration of the
rate of climate change.
FUTURE CLIMATE CHANGE
AND METHANE FLUXES
FROM NORTHERN PEATLANDS
There is general agreement among the current
generation of Global Climate Models that in-
creasing greenhouse gases in the atmosphere
will cause a warmer climate and that high
northern latitudes will experience the greatest
warming (Mitchell, 1989). Table 2 summarizes
the boreal/sub-arctic zone (global 50-70°N)
temperature change predictions of four GCMs
for a doubling of CO2 in the atmosphere. As the
table indicates, summer temperature changes
are generally expected to be less than the an-
nual average (as winter temperature changes
are greater). An average GCM predicted sum-
mer temperature change for the boreal regions
is about 3-5°C, or about 6 to 10 times the
observed maximum temperature anomalies for
the northern peatland regions (Figure 7b). The
model would suggest that for such a significant
temperature change (and all other factors re-
maining unchanged), methane fluxes from
northern wetlands would increase by as much as
50 Tg CH4/yr. This would constitute a signifi-
cant greenhouse gas climate feedback. It would
also dramatically change the carbon balance of
northern peatland soils.
120
-------
S. FROLKING
methane flux
anomaly a
(Tg/year)
o
o-
0
0
-2
-4
1900
methane flux
anomaly (Tg)
-6
1900
- North Slope Alaska
Central Alaska
Yukon-Kuskokwim Delta
Boundary Waters Area
Nouveau Quebec
Fenno-Soviet Lowlands
Hudson Bay Lowlands
Siberian Lowlands
1920
1960
1940
year
Methane flux anomalies for
eight northern wetland regions
1980
1920
1940
year
1960
1980
b) Northern wetlands annual methane flux
anomaly and 9-year running mean
FigureS. (a) The modeled methane-flux anomaly for the eight northern wetland regions (inTgCH4/year). The anomaly
is the modeled temperature-driven difference from the flux for a normal year; (b) The net northern peatlands methane-
flux anomaly is the sum of the eight curves in Figure 8a. The heavy line is a simple nine-year running mean of the net
methane anomaly to account for methane's lifetime in the atmosphere and give some indication of the detectability of
the flux variations.
121
-------
MEIHANB FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
Table 2. GCM predictions for boreal/sub-arctic average
temperature (°C),precipitationandsoilmoisture changes.
Model
GISS
GFDL
NCAR
UKMO
Summer
temp.
(JJA)
+2 to +4
+4 to +8
+0 to 4-4
+5 to +6
Winter
temp.
(DJF)
+5to+12
+6 to +15
+6 to +10
+8 to +10
Annual
temp.
+4.2
+4.0
+4.0
+5.2
Change in
summer
precip.®
+1
-1
+1
—
Seasonal
soil
moisture®
-2 to +2®
-5 to +4®
up to +2^e)
—
Source: MildxU, 1S85.
(•) mm/d for mil-contlncaitl Interiors.
(b) cm of toHwlerslorage.
(c)wtlter la winter, drier la summer.
(d) iacreaiei MUJbera boreal, decreases northern boreal.
(e) year round Increaiei.
However, it is unlikely that all other factors
would remain unchanged. For example, there
is less agreement among the GCM models as to
how precipitation will change for a doubling of
CO2, especially for the summer months (Table
2). Some models predict wetter summers and
some drier summers (Mitchell, 1989). Wetland
soil moisture and water-table level are impor-
tant controls on both methane production and
oxidation, and a more complete model of meth-
ane fluxes will have to incorporate soil moisture
and its changes.
More important than global or latitudinal band
climate change is potential regional climate
change and climate variability. Because the
major northern wetland ecosystems are con-
centrated hi several regions, methane fluxes
from these peatlands will depend on local cli-
mate change and climate variability. At present,
though, there is no reliable method for predict-
ing regional (hundreds of kilometers) scale
climate change due to greenhouse gas warming
(Giorgi and Mearns, 1991; Grotch and Mac-
Cracken, 1991).
CONCLUSIONS: CLIMATE CHANGE
FEEDBACKS AND DETECTION
Freshwater peatlands are a major source of
global atmospheric methane, and these envi-
ronments are concentrated in regions that may
be subject to the extremes of climate change.
Methane emissions from freshwater wetland
ecosystems are influenced by variations in cli-
mate (especially temperature and precipita-
tion). Changes in the emissions of methane
from these regions may be a potential bio-
spheric feedback process, which could influ-
ence the dynamics of the climate system. How-
ever, the results of this preliminary modeling
study indicate that the change in methane flux
fromnorthern peatlands during the initial stages
of a climate warmingwouldnot be large enough
to cause a significant change in atmospheric
methane concentration, and thus would not
likely have a strong feedback effect.
Over the longer term, a significant climatic
change (like those predicted by current GCMs
for a doubling of atmospheric CO2) could cause
a substantial change in the methane flux from
northern peatlands. This would change the
carbon balance of these vast soil carbon pools,
perhaps transforming themfrom sinks to sources
of atmospheric carbon. Of course, many other
ecosystem factors, such as fire frequency, per-
mafrost thawing, plant respiration and produc-
tionrates, may also change with climatic change,
resulting in additional changes to the ecosystem
carbon balance, and perhaps offsetting the tem-
perature effects (Billings, 1987).
It is proposed that in high northern latitudes,
changes in methane emissions from peatlands
may provide a unique early warning indication
of the biospheric response to climate change.
Methane production and emissions respond to
variations in soil temperature and moisture on
time scales of days to weeks. These time scales
avoid the high noise inherent in air tempera-
tures, but do capture intraseasonal patterns of
variability. In addition, methane production
integrates many components of climate change
(e.g., temperature, precipitation, cloud cover)
into one signal.
122
-------
S. FROLKING
REFERENCES
Aselmann, I. and P.J. Crutzen. 1989. Freshwater wet-
lands: global distribution of natural wetlands and rice
paddies, their net primary productivity, seasonality and
possible methane emissions. J. Atmos. Chem. 8:307-358.
Bartlett,K.,P. Grill, R.Sass, R. Harriss, andN. Dise. 1991.
Methane emissions from tundra environments in the
Yukon-Kuskokwim Delta, Alaska. J. Geophys. Res. In
press.
Billings, W.D. 1987. Carbon balance of Alaskan tundra
and taiga ecosystems:, past, present and future. Quater-
nary Sci. Revs. 6:165-177.
Bouwman, A.F. 1990. Soils and the Greenhouse Effect.
John Wiley and Sons, Chichester, UK.
Chanton, J. and J.W.H. Dacy. 1991. Effects of vegetation
on methane flux, reservoirs and carbon isotopic composi-
tion, in environmental and metabolic controls on trace gas
emissions from plants. In press.
Chappellaz, J., J.M. Barnola, D. Raynaud, Y.S.
Korotkevich, and C. Lorius. 1990. Ice-core record of
atmospheric methane over the past 160,000 years. Nature
345:127-131.
Cicerone, R.J. and R.S. Oremland. 1988. Biogeochemical
aspects of atmospheric methane. Global Biogeochemical
Cycles 2:299-328.
Grill, P.M., K.B. Bartlett, R.C. Harriss, E. Gorham, E.S.
Verry, D.I. Sebacher, L. Madzar, and W. Sanner. 1988.
Methane flux from Minnesota peatlands. Global Biogeo-
chemical Cycles 2:371-384.
Clymo, R.S. and E.J.F. Reddaway. 1971. Productivity of
sphagnum (bog-moss) and peat accumulation,
Hidrobiologica 12:181-192.
Dacy, J.W.H. and M. Klug. 1979. Methane efflux from
lake sediments through water lilies. Science 203:1253-
1255.
Dise,N. 1991a. WinterfluxesofmethanefromMinnesota
peats. Submitted to Biogeochemistry.
Dise, N. 1991b. Methane emission from Minnesota
peatlands: spatial and seasonal variability. Submitted to
Global Biogeochem. Cycles.
Fung, I., J. John, J. Lerner, E. Matthews, M. Prather, L.P.
Steele, and P.J. Fraser. 1991. Three-dimensional model
synthesis of the global methane cycle. J. Geophys. Res.,
96:13,033-13,065.
Giorgi, F. and L.O. Mearns. 1991. Approaches to the
simulation of regional climate change: a review. Rev.
Geophys. 29:191-216.
Gorham, E. 1991. Northern peatlands: role in the carbon
cycle and probable responses to climatic warming. Ecol.
Appl. 1(2): 182-195.
Grotch, S.L. and M.C. MacCracken. 1991. The use of
general circulation models to predict regional climate
change. J. Climate 4:286-303.
Hansen, J. and S. Lebedeff. 1987. Global trends of
measured surface air temperature. J. Geophys. Res.
92:13345-13372.
Harriss, R.C., K. Bartlett, S. Frolking, and P. Grill. 1991.
Methane emissions from northern wetlands: a review and
assessment. Submitted to Proceedings of Tenth Interna-
tional Symposium on Environmental Geochemistry, San
Francisco, CA.
Harriss, R.C. and S. Frolking. 1992. The sensitivity of
methane emissions from northern freshwater wetlands to
global change. In: Global Climate Change and Freshwa-
ter Ecosystems, edited by P. Firth and S. Fisher. Springer-
Verlag, New York.
Henderson-Sellers, A. and K.McGuffie. 1987. A Climate
Modelling Primer. John Wiley and Sons, New York.
Hillel,D. 1980. Fundamentals of Soil Physics. Academic
Press, NY.
Khalil,MA.K.andRA.Rasmussen. 1987. Atmospheric
methane: trends over the last 10,000 years. Atmos. Env.
21:2445-2452.
Lashof, D. 1989. The dynamic greenhouse: feedback
processes that may influence future concentrations of
atmospheric trace gases and climatic change. Climatic
Change 14:213-242.
Livingston, G.P. and LA. Morrissey. 1990. Methane
emissions from Alaskan arctic tundra in response to
climatic change. Paper presented at International Conf.
on the Role of Polar Regions in Global Change.
Matthews, E. and I. Fung. 1987. Methane emission from
natural wetlands: global distribution, area and environ-
mental characteristics of sources. Global Biogeochemical
Cycles 1:61-86.
123
-------
METHANE FROM NORTHERN PEATLANDS AND CLIMATE CHANGE
Mitchell, J.F.B. 1989. The greenhouse effect and climate
change. Rev. of Geophysics 24:115-139.
Moore, TJR., N. Roulet, and R. Knowles. 1990. Spatial
and temporal variations of methane flux from subarctic/
northern boreal fens. Global Biogeochem. Cycles 4:29-
46.
Moore, TJR., A. Heyes, S. Holland, W.R. Rouse, N.T.
Roulet, and L. Klinger. 1991. Spatial and temporal
variations of methane emissions in the Hudson Bay Low-
lands. EOS 72:84.
Rodhe, H. 1990. A comparison of the contribution of
various gases to the greenhouse effect. Science 248:1217-
1219.
Ruffner,JA. 1985. Climates of the States, 3rd Ed. Gales
Research Co., Detroit MI.
Sebacher, D.I., R.C. Harriss, and KB. Bartlett. 1985.
Methane emissions to the atmosphere through aquatic
plants. J. Environ. Qual. 14:40-46.
Sebacher,D.I.,R.C.Harriss,K.B.Bartlett,S.M.Sebacher,
and S.S. Grice. 1986. Atmospheric methane sources:
Alaskan tundra bogs, an alpine fen, and a subarctic boreal
marsh. Tellus Ser. B. 38:1-10.
Sellers, W.D. 1965. Physical Climatology. U. Chicago
Press, Chicago.
Svensson B.H. and T. Rosswall. 1984. In situ methane
production from acid peat in plant communities with
different moisture regimes in a subarctic mire. Oikos
43:341-350.
Watson, R.T., H. Rodhe, H. Oeschger, andU. Siegenthaler.
1990. Greenhouse gases and Aerosols. In: Climate
Change, edited by J.T. Houghton, G.J. Jenkins, and J.J.
Ephraums. The IPCC Scientific Assessment,
Intergovernmental Panel on Climate Change. Cambridge
U. Press, Cambridge, England.
Whalen, S. and W. Reeburgh. 1988. A methane flux time
series for tundra environments. Global Biogeochemical
Cycles 2:399-409.
Whalen, S. and W. Reeburgh. 1990a.
atmospheric methane by tundra soils.
162.
Consumption of
Nature 346:160-
Whalen, S. and W. Reeburgh. 1990b. A methane flux
transect along the trans-Alaskanpipeline haul road. Tellus
Ser. B. 42:237-249.
ACKNOWLEDGEMENTS
I would like to thank Robert Harriss, Patrick Grill, and Karen Bartlett for their help with this work. The work was
supported under a NASA Graduate Student Researcher Program Fellowship.
This paper is a summary of two recent works: Harriss and Frolking (1992) and Harriss et al. (1991).
124
-------
METHANE AND CARBON DIOXIDE PRODUCTION AND UPTAKE
IN SOME BOREAL ECOSYSTEMS OF RUSSIA
Nicolay Panikov and Vladimir Zelenev
ABSTRACT
The production, emission anduptake of methane andcarbon dioxide werestudiedin different subarctic and boreal terrestrial
ecosystems of the European plain of Russia. These ecosystems included the arable soils of the Moscow region, the wetlands
and dry forests soils of the Tver (Kalinin) and Syktyvkar regions, as well as tundra soils of the Polar Ural. Two main
approaches were used: 1) the dynamic studies of CO2 and CH4 exchange rates between the soil and atmosphere by the
chamber technique; and 2) the estimation of CO2 and CH4production and uptake in soil cores incubated under aerobic
as well as anaerobic conditions. The highest rates of both CH4 production and utilization were observed in the acid
ombrotrophic bogs of the borealforests. The wetlands were shown to be not only sources, but also sinks for CH4. The methane
cycle for most of the soils studiedwas closed; theprocesses ofmethanogenesis in the lowerhorizons were completely balanced
by methane oxidation in the upperhorizons. The only exceptions were the water-logged depressions of the ombrotrophic bogs
where the activity ofmethanotrophes was absent or poorly expressed. In this case, significant CH4 emission from soils to
the atmosphere (up to 50-120mg CH4 C/hr/m2) was observed. On the basis ofCO2 and CH^fluxdata, the main components
of the carbon balance of the ombrotrophic bog ecosystem were evaluated. The kinetic approach was used to estimate biomass
and in situ activity of both methanogenic and methanotrophic bacteria.
INTRODUCTION
Methane exchange between the soil and the
atmosphere has been considered for the last
decade to be one of the most important prob-
lems of biogeochemistry (Andreae and Shimel,
1989; Bouwman, 1990). Methane is involved in
many chemical reactions with atmospheric gases
(interactions with hydroxyl radicals and strato-
spheric chlorine, formation of tropospheric
ozone and carbon monoxide) and, through its
infrared properties, has an influence on Earth's
energy balance (the greenhouse effect).
The apparent increase in the atmospheric CH4
concentration of 1% per year (Cicerone and
Oremland, 1988) currently is not well under-
stood. The total annual flux of methane to the
atmosphere is estimated to be 374-714 Tg, from
which 110-330 Tg is attributed to different kinds
of wetlands (Stewart et al., 1989). However the
data on natural (non-anthropogenic) sources
and sinks of methane are not reliable, and there
are many contradictions and unproved assump-
tions in this respect. An especially large gap is
evident in the knowledge regarding methane
production and consumption in the soils within
the territory of Russia, which includes about
62% of the total wetland area of the Earth.
Wetlands occupy a considerable part of Earth's
land surface (about 3%), and the dry mass of
carbon in deep peat deposits amounts to 180 Gt
(Kivinen and Pakarinen, 1981). These figures
may be compared with estimates of the total
amount of carbon in the atmosphere (600 - 700
125
-------
METHANE AND CARBON DIOXIDE PRODUCTION AND UPTAKE
Gt), in live plant biomass (800 Gt), or in 'soil
humus,' including peat (700-3000 Gt)
(WoodweU et al., 1978).
About three-quarters of all peat land is distrib-
uted between Canada and Russia. Wetlands
play an essential role in atmospheric CO2 bind-
ing and prolonged carbon conservation. On the
other hand, bogs themselves are the sources of
CO2, CH4 and other volatile organic substances
(VOS).
The objective of the study reported herein is to
characterize quantitative aspects of the spatial
and temporal distribution of sources and sinks
of methane in wetland ecosystems of Russia
and to evaluate the main components of their
carbon balance.
MATERIAL AND METHODS
Site Location
The field stationary observations were carried
out in four geographical points of the European
plain of Russia as follows:
I. Moscow region, Serpukhov district, the right
bank of the Oka river, outskirts of Poushchino
town (55°32'N, 37°39'E). The observations were
carried out during July and August, 1990, on
20x20 m plots of the field experimental station
of the Institute of the Soil Science and Photo-
synthesis, USSR Academy of Sciences.
1. Arable gray forest soil under winter wheat
(cultivar Mironovskaya-808). The following
plots of the field experiments were chosen: a)
control; b) superphosphate (P^K^); c) super-
phosphate + ammonium nitrate (N^P
and d) Nm1?9Q}£9Q+ wheat straw (8 t/ha).
out from July 1990 to July 1991 on the following
sites:
2. The ombrotrophicbog of about 1.7 km2, near
the Sosvyatskoe village, drained in 1972-1973
by means of open 1 m deep trenches, the dis-
tances between them being 105 m. The vegeta-
tion was represented on the periphery of the
bog by pine forest with Oxycoccus-Sphagnum,
and in the central upper part, by pine forest with
the Andromeda-Eriophorum-Sphagnum plant
community. The forest stand was not more than
5 m in height with the crown density of 0.3 - 0.4.
In trenches, plant cover was represented by rare
Eriophorum tussocks. The peat thickness was
about 5.5 m.
3. The Big Rogovskoye ombrotrophic bog (2-3
km from Petrilovo village) of about 7.3 km2 with
the secondary oligotrophic lakes 50 m in diam-
eter. Between the lakes, there was a ridge-
hollow complex under the low shrub-sphag-
nous pine forest. The peat thickness reached 12
m.
4. Theundrained ombrotrophicbog, Lamtevsky
Moch, which was similar to the B. Rogowskoye
wetland in respect to its vegetation type and
hydrological characteristics. The area was 4.1
km2, the peat thicknesses in the central part and
in the periphery were 6 m and 1.5 m, respec-
tively.
The following three sites (5,6,7) were associ-
ated with the Grustinskoe Lake:
5. The floating bog on the lake (-1-3 m in
depth) consisted of a partly decomposed plant
material colonized by a Carex-Sphagnum com-
munity and by a pine-birch forest stand at a
distance of 20-50 m from the water border.
IT. Tver (Kalinin) region. West-Dvinskiy dis- 6. Aminerotrophic groundwater fen adjoining
trict, the field station of the Institute of For- a region with elevated moraine relief. The
estry, USSR Academy of Sciences (56°10'N, observation site was located under a highly
32°12'E). Repeat observations were carried productive black alder-geum-fern forest with
126
-------
N. PANIKOV AND V. ZELENEV
single spruce and birch trees. The area of this
forest was about 0.1 km2; the peat depth was 2.5
m.
7. A mked linden-spruce forest on sod-podzolic
loamy-sandy soil, observation site was located
on the watershed 100 m apart from point 6.
8. The cowberry-moss pine forest Geras'kin
Bor on podzolic sandy loamy soil. The observa-
tions were made at the watershed area and in a
300 m depression occupied by Oxycoccus-Sph-
agnum pine forest. Vegetation type and hydro-
logical characteristics of the wetland were simi-
lar to those in the periphery of Sosvyatskoe bog
(site 5).
HI. Syktyvkar region (61°41'N, 50°45'E). The
observations were carried out from July 25 to
July 30,1990, in three sites:
9. Kuz'-Nur bog near Dodz' village of an
intermediate type of mineral nutrition. It rep-
resented a hummock-depression complex un-
der dwarf shrub-sedge-sphagnum vegetation.
There were some patchy low-stand pine forests.
The hummocks were 0.5-1.0 m in height and
covered mainly by dwarf shrubs as well as Rubus
chamaemorus and Sphagnum sp.', in depres-
sions, there were grass-sphagnum associations.
10. Kiya-Nyur bog represented the transitional
type of wetland with a hummock-depression
complex under dead forest stand. The hum-
mocks were of ~ 1 m in height and covered by
dwarf shrub-sphagnum vegetation with an ad-
mixture of polytrichum mosses. In depressions,
there wereEriophorum-Sphagnum associations
with sedge tussocks.
11. Field forest station Lyali of the Institute of
Biology, Komy Scientific Center, USSR Acad-
emy of Sciences. Sod-podzolic soil under mixed
forest. The measurements were made on two
sites: within the watershed area (woodsorrel
spruce forest) and in the valley of a small forest
brook covered by fern andFillipendula ulmaria
under spruce and alder.
IV. Vorkuta region. Halmer-Yu settlement
(67°59'N, 64°42'E). The observations were
carried out from July 20 to July 25,1990.
12. A plain-hilly swamp of about 5 km with
small lakes 10-20 m in diameter. The peat hills
were 0.5-1.0 min height with the mineral nucleus
inside; permafrost was located at 40-50 cm. The
vegetation on the hills was representedby5efM/a
nana, cowberry, Rubus chamaemorus and Sph-
agnum sp. and had almost 100% covering. The
vegetation in depressions mainly consisted of
Carex aquatilis, C. globularis, Eriophorum
vaginatum, Sphagnum and Polytrichum sp.
Some quantitative characteristics of the studied
soils are shown in Table 1.
Determination of CH4 Emission
The chamber static method (Conrad et al.,
1983; Whalen and Reeburgh, 1988) was used.
The chamber consisted of two parts: 1) a
permanent square stainless steel collar base (a
frame 40x40 cm and 15 cm in height, a water-
tight channel along the upper perimeters of the
base) and 2) a removable Plexiglas® cover (a
hollow bottomless box 40x40 cm and 30-90 cm
in height). The first part of the chamber was
inserted into the soil (depth =10 cm) in advance
(not later than 0.5-1.0 h before the installation
of the second part) to bring the soil-air system
in equilibrium before measurements were be-
gun. To avoid the artifacts of CH4' 'squeezing''
from the soil air, each measuring site was pro-
vided with light bridges (the supports were
placed not closer than 1 m from the chamber)
to accommodate the operator during all ma-
nipulations. The Plexiglas® cover was con-
nected with the base via water seal only during
measuring (usually for 30 min.); the rest of the
time, the chambers were open to air. The air
samples were taken from the isolator by hypo-
127
-------
METHANE AND CARBON DIOXIDE PRODUCTION AND UPTAKE
Table 1. Methane emission from the soil to the atmosphere.
Ground Water
N(a) Site (cm)
Temperature
CO
Eh
(mV)
pH
CHt Flux mg
CH4-C/hr/m2
CH4/CO2
I. Moscow region, Poushchino
1
Field experiment
control
PK
NPK
NPK+strow
<-200(b)
<-200
<-200
<-200
13-15
13-15
13-15
13-15
200
200
200
200
55
55
55
55
0-0.04
0-0.04
0-0.04
0-0.43
0.001-0.005
0.001-0.005
0.001-0.005
0.003-0.011
IL Tver (Kalinin) region, West-Dvinsky District
2
3
4
5
6
7
8
Central part
Between trenches
Trench
Mound
Depression
Lake shore
Hillock
Between them
20 m from the water
100 m from the water
Sphagnum spot:
Sphagnum Carcx sp.
Hillock
Waterlogged depression
Under the trees
Watershed
Depression
-30
+10
-30
+ 1-2
+10
-10
+1
-1-2
-5
-5
-15
+5
<^50
<-50
-20
9
9
12
13
13.5
15
15
8
9
9
8
8
9
10
10
+10-
-160
-200
+15
-15
-100
-
-
-160
-180
-180
-246
-246
+ 170
+ 160
+ 160
3.0
4.0
4.1
4.0
-(<=)
-
-
3.9
35
4.0
7.2
7.2
65
4.0
3.0
0.00-0.08
3.9
0.05-0.25
5.2
0.52
115.2
6.3-84.7
0.0-1.10
4.4
36.6
0.03
0.07
0.00
0.00
0.00
0.00-0.004
0.05
0.002-0.03
0.24
1.24
-
-
0-0.023
0.41
1.74
0.0004
0.009
0.00
0.00
0.00
III. Syktyvkar region, Komi SSR
9
10
11
Under the pine forest
Mound
Depression
Without trees
Hillock
Depression
Watershed
Brook shore
-70
-10
-8
<-50
+1-3
<-50
-2
15
15
15
14
14
12
11
-
-
-
-
-
+ 160
-
4.2
4.2
4.2
-
4.2
5.0
45
0.00
0.00
0.11-1.22
-0.52
0.20-4.23
0.00
0.15
0.00
0.00
-
-
-
0.00
-
IV. Vorkuta region, Halmer-Yu
12
Mound
Mound's base
Waterlogged part
<-70
-20
0
15
14
11
-
-
-
-
6.1
6.1
0.00
0.28-0.38
0.79-7.85
0.00
-
-
(a) Site number (site description given in text).
(b) "+" ground water level is above the soil surface;"-"
(c) "-" not determined
ground water level is below the soil surface.
128
-------
N. PANIKOV AND V. ZELENEV
dermic syringe through a rubber septum three
times: immediately after cover installation and
then at 15 and 30 minutes of exposure. The
samples were stored in glass flasks under salt-
water (10% NaCl) seal. The CH4-emission was
calculated by approximating the dynamics of
CH4 concentration by linear regression.
Determination of CO2 Emission
Determination of CO2 emission was made in
the same chambers. The only exception was
that air samples were taken into 300-500 cc
PCV-bags. The CO2 exchange rates were mea-
sured under dark conditions (dark plant respi-
ration + soil respiration); the Plexiglas® cover
was shaded by a nontransparent screen (alumi-
num foil or black cotton fabric).
Measurement of CH4 Transformation
in Soil (Peat) Cores
The serum glass flasks (500 cc, outer diameter
77 mm) were cut at the bottoms, and glass edges
were sharpened. The borer-flasks used could
be easily inserted into most of the mineral soils
to the depth of 8-10 cm. In the case of easily
deformable peat soils, the preliminary cutting
by a special cylindrical knife (of the same diam-
eter as the sampling flask) was performed.
Then this flask was introduced into the cutting,
and the soil core was carefully removed. The
core height was usually 95-100 cm, and the head
space volume was about 100 cc. At the bottom,
the core was sealed by a tight butyl rubber cap,
and at the top, by a silicone stopper. When core
samples were taken below the level of the
groundwater table, the plot area of 40 x 40 cm
was isolated from water by a stainless steel
frame; water was pumped out, and soil cores
were taken as described above. The flask with
waterlogged soil did not contain any gas-filled
headspace, and all its free volume (from peat
surface to stopper) was occupied by the water
from the same horizons.
Four replicate cores from each horizon were
brought to the field laboratory. The upper
water layer (—100 cc) was carefully sucked off
from the flooded samples and replaced by an
equal volume of atmospheric air. All cores
were incubated outdoors at a temperature close
to those in situ (the differences as a rule did not
exceed 2 ). Periodically withdrawn 0.5 cc gas
probes were analyzed for CH4 concentration.
One to two days later, methane was introduced
into two of the replicate cores (final concentra-
tion 20 -10000 ppm), and incubation was con-
tinued under aerobic conditions. The headspace
in another two replicates was replaced by the
1:1 mixture of H2:CO2 (vol.:vol), and incuba-
tion was carried out under strictly anaerobic
conditions.
Determination of CH4 Concentration
in the Soil Air
Determination of CH 4 concentration in the soil
air was made by membrane probe technique
(Kasparov et al., 1985). The probe was made
from a hollow L-shaped stainless steel tube 10-
20 mm in diameter. The lower perforated part
was covered by a thin ( ~ 100 mm) polyethylene
membrane and inserted horizontally into the
soil at a chosen depth. The vertical part of the
tube coming from the soil upward was sealed by
a rubber septum. The polyethylene membrane
is permeable for many soil gases including CH4
and CO2, but retains water vapors. Hence, this
method is especially valuable for submerged
soils. Steady-state concentrations of soil gases
were usually established in 1-3 days. Then gas
samples were withdrawn via rubber septum and
brought to the laboratory under saltwater seals.
Gas Determinations
The CH4 concentrations were analyzedby means
of a portable gas chromatograph (XPM-4, Rus-
sia) equipped with a flame ionization detector
(1 m columns with aluminum oxide, H2 as gas-
carrier, oven at 100°). The detection limit was
129
-------
METHANE AND CARBON DIOXIDE PRODUCTION AND UPTAKE
0.2 ppm. The CO2 concentration was deter-
mined with a portable infrared gas analyzer.
The detection limit was approximately 5 ppm.
Other Methods
The pH of soil extracts was determined with
glass electrodes on a portable pH-meter (pH-
150, Russia). In wetlands, measurements were
madeinsitu at a depth of 10 cm. In aerobic soils,
the determinations were made in 1 N KC1-
extracts (soilrsolution = 1:1). The redox poten-
tials Eh were measured with platinum elec-
trodes at depths of 5-70 cm. The temperature
was estimated by thermistors and mercury ther-
mometers at depths of 2 and 10 cm. Atmo-
spheric pressure was determined with aneroid,
salinity of rapa solutions by refractometer.
RESULTS AND DISCUSSION
Methane Emission into the Atmosphere
from Different Soils
The results of single-point determinations made
from June-July, 1990, are summarized in Table
1. The highest rates of CH4 emission were
observed in ombrotrophic bogs: Lamtevsky
Moch (up to 115 mg C/hr/m2), the floating bog
of the Grustinskoye Lake (4-37 mg C/hr/m2),
as well as in waterlogged depressions of the
forest wetlands: B. Rogovskoye (5.2 mg C/hr/
m2), Halmer-Yu (up to 7.9 mg C/hr/m2) and
Kiya-Nyur (4.2 mg C/hr/m2). In the drained
Sosviatskoe bog, the CH4 emission was insig-
nificant; during repetitive determinations, it
was detected only occasionally. Intensive meth-
ane emission (about 4 mg C/hr/m2) was steadily
recorded from drainage trenches. Methane
emission was seldom observed from the el-
evated elements of micro- and mesorelief of the
ombrotrophic bogs. Moreover, on the hillocks
and mounds, we sometimes detected negative
emission when CH4 concentration inside the
chamber decreasedfromthe atmospheric back-
ground level to zero (see Table 1, site 6).
Contrary to expectations, we have not found
intensive CH4 emission from highly productive
minerotrophic fen, which seems to be charac-
terized by favorable conditions for
methanogenesis (low Eh values, neutral pH,
high content of organic matter). However, the
CH4 emission from fen rarely exceeded 1 mg C/
hr/m2. The reason for this is probably the
competitive inhibition of methanogenesis by
sulfate reduction (withdrawal of H2 and organic
acids).
In watershed aerobic soils (site 7), CH4 emis-
sion was either absent or extremely low, below
10~2mg C/hr/m2. The detectable methane emis-
sion from control plots of arable gray forest soil
(0.03-0.04 mg C/hr/m2) was monitored only on
two occasions (from more than 10 determina-
tions). After amendments with organic matter
(wheat straw or 200 mg/m2 of glucose), meth-
ane emission increased to 0.4 mg C/hr/m2. In
the forest soils of Lyali station, significant CH4
emission was noticed only in a 1-m zone along
the brook.
The contribution of methane to the total emis-
sion of carbon compounds from soil was evalu-
ated by the CH4:CO2 ratio (Table 1) because
CO2 is the main volatile carbon compound.
This ratio was usually less than 0.001. However,
in waterlogged depressions in wetlands, it in-
creased to 0.2 -1.7. Hence, the contribution of
methane to total carbon flux in some ecosys-
tems can be even higher than that of CO2.
Thus, it follows from our preliminary results,
that CH4 emission from soils varies broadly in
time and space. Even within one particular
location, emission rates were not the same
during several successive measurements and
from several adjacent points. It was impossible
to find one or several parameters (pH, Eh,
temperature, vegetation type, etc.) that would
give high correlations with CH4 emission. High
emission values were not always attributed to
wetlands. Sometimes more CH4 was emitted
130
-------
from aerobic soils (e.g., from arable forest soil
in Poushchino) than from peatlands. There-
fore, it seemed to be worthwhile to carry out an
intensive study of the dynamics and spatial
distribution of CH4 sources and sinks for one
particular ecosystem. For this purpose, we have
chosen the ombrotrophic bog Sosviatskoe of
Tver region.
Dynamics of Gas Emission
We followed the temporal variations in CH4
and CO2 fluxes from soil that occur on 1) daily;
2) weekly (every 5-10 days); and 3) seasonal
cycles. All measurements were made at the
same points of the space (in drainage trenches
and in the undrained central part of the bog),
which were fixed by nonremovable collars of
the chambers. This allowed us to overcome the
problem of spatial variations in CH4 and CO2
emission during dynamic measurements. Be-
cause the upper removable part of the chamber
was connected with the base only for a short
exposition time (30 min.), it did not change the
state of covered soil.
The diurnal cycles of CO2 and CH4 emission
were dependent on the season. In the cold
season, the rate of emission was practically
constant during the all-day span or varied
unpredictably by chance. From April to Sep-
tember, statistically significant increases of CO2
emission were observed in the light part of the
day. In the case of methane, the differences
between light and dark parts of the day were
expressed only as a tendency. It would be
reasonable to assume that acceleration of CO2
and CH4 emission in the middle of the day is
initiated by plant exudations. This mechanistic
explanation was proved (Gorbenko and
Panikov, 1989) for CO2 emission from arable
soil in Poushchino (site 1). Some positive
relationship was also recorded between the
rates of CH4 emission and the soil temperature.
This relationship could be interpreted in two
ways: as a stimulating temperature effect on the
N. PANIKOV AND V. ZELENEV
activity of methanogenic bacteria and as the
positive influence of temperature on the rate of
gas diffusion from the soil to the atmosphere.
Besides, one cannot exclude the possibility that
during the midday warming, the degassing of
waters takes place with the following CH4 evo-
lution.
Bearing in mind the existence of diurnal cycles,
all measurements of weekly (day-to-day) varia-
tions in CO2 and CH4 fluxes were made at a
fixed time of day, from 10:00 until 12:00 a.m.
Both CH4 and CO2 evolution rates exhibited
more or less regular rises and falls every 4-7
days; one maximum in April, three maxima in
June and two maxima in July. However, the
oscillations of CH4 and CO2 fluxes were in
antiphase. Oscillations of CO2 evolution rates
from aerobic arable soil were explained by
periodic changes in root exudation, which re-
flect the changes of plants' photosynthetic ac-
tivity (Gorbenko and Panikov, 1989). Prob-
ably, this conclusion is valid also for wetlands.
Hence, the negative relationship between CO2
and CH4 fluxes would lead to the assumption
that methanogenic microorganisms have no
direct connection with the physiological state of
plants as do aerobic soil microbes.
It is interesting that CH4 emission from the bog
center in July was negative, i.e., atmospheric
methane was consumed (scavenged) on the
interface between air and the bog surface. In
April, net methane flux was steadily positive,
not only from the trenches, but also from the
bog center. In June, the CH4 emission from the
bog center had zero value. The trenches always
were the sources of CH4.
The influence of two recorded environment
factors, temperature and atmospheric pressure,
explained only part of the observed dynamic
patterns of fluxes because there was no com-
plete synchronization of corresponding time
series. As a trend, a positive relationship was
found between the CO2 evolution rate and the
131
-------
METHANE AND CARBON DIOXIDE PRODUCTION AND UPTAKE
temperature, as well as between CH4 emission
from trenches and atmospheric pressure. The
last correlation may be tentatively interpreted
as a result of the barometric squeezing of
accumulated methane from the bog center to
the trenches.
To make a reasonable approximation to the
sizes of gas emissions on annual bases, we took
into account all available data on daily and
weekly variations. Numeric integration gave
the following average values of annual emis-
sion: CH4,0.3 g of C/m2; CO2,142 g of C/m2.
Again as a trend, we can see the positive rela-
tionship between CH4 flux and the soil tem-
perature. However, this relationship is not
simple when we consider details. For instance,
CH4 flux from the bog center reached the maxi-
mal value inApril, when thepeatremained very
cold. In midsummer, when the soil temperature
was maximal, net methane fluxes became nega-
tive. To elucidate the observed phenomena, we
tried to consider separately the processes of
formation and consumption of methane.
Distribution of Methane Sources
and Sinks in the Soil Profile
An example of the monitoring of CH4 produc-
tion and consumption in soil cores from differ-
ent depths of ombrotrophic bog is shown in
Tables 2, 3a and 4a for one sampling datum
(September, 1990). During the first day in all
core samples, except for a sample from the
upper horizon, CH4 emission was observed. Its
rate quickly decreased with time. On the sec-
ond day, after replacement of the gas phase the
conditions were created for either CH4 produc-
tion or aerobic CH4 consumption, respectively
(see Materials and Methods). The reaction of
the microbial community on replacement was
instant. In the upper horizons, CH4 uptake had
quickly started, while in the lower one (below
44 cm), CH4 emission began. There were no
opposite processes. In the upper horizon metha-
Table 2. Vertical distribution of sources and sinks of
methane in wetlands, September 1990. (Fluxes are
expressed in mg CH4-C/h/m2.)
The balance of CH[j as
Rate of evaluated from:
Rate of
Horizon methane methane Soil core CH4 emission by
uptake production experiments chamber method
Big Rogovskoye bog, waterlogged depression
0-15
15-25
0.140
0.000
0.027
3.400
+ 3.29
+5.20
Big Rogovskoye bog, mound
0-17 0.800 0.000
17-25 0.004 0.060
25-35 0.000 1.230
+0.49
+0.25
Floating bog on Grustinskoye Lake, 100m from water
0-15
15-30
2.490
0.000
0.000
3.333
+0.84
+4.4
Lowland groundwater fen, Grustinka
0-15
15-25
0.140
0.310
0.400
1.050
+ 1.00
+ 0.07- +1.60
Mixed forest on the watershed site, 100m from swamp
0-10
10-20
0.290
0.450
0.000
0.000
-0.74
Watershed site (Geraskin Bor) adjacent to Sosviatskoe bog
0-15 0.175 0.000 -0.175 0
Geraskin Bor, local wetland under pine
0-15
15-25
25-35
0.140
0.900
0.190
0.000 -1.23
0.000
0.000
nogenesis was not induced under anaerobic
conditions, and as in the lower layer, CH4 con-
sumption was not observed under aerobic
conditions. In the intermediate horizons (15-44
cm) the coexistence of both sources and sinks of
CH4 was found. Here the anaerobic microsites
are probably distributed in a chaotic way.
On the basis of the kinetic analysis of the
dynamic curves, we can get insight into the
mechanisms of studied processes. The curves
of CH4 consumption were of the hyperbolic
type; they could be approximated by the
Michaels-Menten kinetic equation:
ds/dt = -
or after integration
-S =
(1)
(2)
132
-------
N. PANIKOV AND V. ZELENEV
Table 3. Seasonal dynamics of spatial distribution of CH4 sources and sinks in ombrotrophic Sosvyatskoe bog
(central undrained part of the bog).
(a)
Sampling
date
9/18/90
3/6/91
4/23/91
6/12/91
7/11/91
Horizon
(cm)
0-15
15-28
28-44
44-60
0-10
10-20
20-30
30-40
0-13
13-26
26-43
43-56
0-10
10-20
20-30
30-40
0-10
10-20
20-30
30-40
CHj concentration
in soil air
(ppm)
4.0
7000
12750
38200
3.0
1415
35770
32884
40
1415
36000
33000
250
5650
49454
98047
5
80
270
360
Kinetic parameters
of CH4 uptake
V
(mg C/hr/m2
of soil core)
0.147
0.92
0.71
0.00
0.000
0.280
0.076
0.075
0.072
0.361
0.000
0.000
0.370
0.420
0.950
0.890
11.29
11.65
0.620
0.550
Km
(ppm)
1200
1677
347
-
-
36
51
475
303
1419
-
-
1200(a)
1200
1200
1200
603
1891
1200
1200
(a) Values of Kmmeasured for this horizon were statistically
unreliable; we used average values of parameters found for all
peat samples instead.
w
Sampling
date
9/18/90
3/6/91
4/23/91
6/12/91
7/11/91
Horizon
(cm)
0-15
15-28
28-44
44-60
0-10
10-20
20-30
30^0
0-13
13-26
26-43
43-56
0-10
10-20
20-30
30-40
0-10
10-20
20-30
30-40
Calculated (a)
rate of in situ
CH4uptake
Rate of
methane
production
(mg C/hr/m /horizon depth)
0.0005
0.742
0.691
0.000
0.000
0.270
0.075
0.074
0.008
0.180
0.000
0.000
0.064
0.346
0.765
0.850
0.094
0.601
0.114
0.116
0.000
0.260
0.062
0.420
0.000
0.029
0.054
0.161
0.000
0.021
0.198
0.401
0.000
0.015
0.107
0.120
0.000
0.000
0.383
0.132
Balance of CH4
as evaluated
from soil core
experiments
(mg C/hr/m2)
-0.681
-0.175
+0.432
-1.783
-0.140
(a) Calculated according to the Michaelis-Menten equation with
experimentally determined parameters V and Km.
Table 4. Dynamics distribution of CH4 sources and sinks in the drainage trenches of Sosvyatskoe ombrotrophic bog.
(a)
Sampling
date
4/23/91
6/12/91
7/11/91
Horizon
(cm)
0-10(a)
10-20
O-S(a)
12-20
0-10(b)
11-20
0-10(a)
10-20
CH4
concentration
in soil air
(ppm)
2
35
13
17
2
94
2675
3382
Kinetic parameters
of CH4 uptake
V
(mgC/h/m2
of soil core)
0.102
0.315
6.690
1.720
1.300
3.050
4.400
1.865
Km
(ppm)
63
63
710
427
1200
2251
1469
1200
w
Sampling
date
4/23/91
6/12/91
7/11/91
Horizon
(cm)
0-10(b)
10-20
0-8(b)
12-20'
0-10(c)
11-20
0-10(b)
10-20
Calculated (a)
rate of in situ
CH4 uptake
Rate of
methane
production
(mg C/h/m^/horizon depth)
0.003
0.113
0.120
0.065
0.002
0.122
2.830
1.370
0.000
0.000
0.000
0.000
0.000
0.043
0.000
0.098
Balance of CH4
as evaluated
from soil core
experiments
(mg C/h/m2)
-0.116
-0.185
-0.081
-4.102
(a) Eriophorum hummock.
(b) Part of the trench without plants.
(a) Calculated according to the Michaelis-Menten equation with
experimentally determined parameters V and 1^.
(b) Eriophorum hummock.
(c) Part of the trench without plants.
133
-------
METHANE AND CARBON DIOXIDE PRODUCTION AND UPTAKE
where s is the CH4 concentration, s0 is the s-
value for initial moment of the time, V and Km
are the parameters of Michaelis-Menten equa-
tion and t is a tune. The physical sense of
parameters is the following: V is related to the
biomass of metabolicaUy active methanotrophic
bacteria (V=Qmx, where x is biomass and Qm is
the maximal specific metabolic activity), while
parameter Km characterizes the affinity of bac-
teria to methane.
To find the parameter values, we used the best-
fit computer procedure written in Turbo Pascal
vS.O, program POISK4 (Kratko et al., 1988).
With rare exceptions, experimental curves did
obey equation (2). The numerical values of Km
and V are shown in Tables 3a and 4a for all
incubation experiments. The Km values varied
from36to2251ppmofmethaneinair. Under
equilibrium conditions, these figures are equiva-
lent to 0.2-8.4X106 M of methane dissolved in
peat-pore water. They fell within the interval
(K^ 2-4X106) reported for methanotrophic bac-
teria grown in homogeneous conditions (King,
1990). Maximal activity of methane oxidation V
varied in the range 0.00-11.65 mg CH4- C/hr/
m2. A dramatic increase of V took place in the
central part of the bog in midsummer, from 0.3-
0.4 in June to 11-12 mg CH4-C/hr/m2 in July.
Therefore, we can conclude that intensive
growth of methanotrophic bacteria occurs in
the form of a burst and only during a short
period of seasonal dynamics. High values of V
were always observed in trenches, where both
potential and actual activity of methane oxida-
tion was kept high during the whole year.
Equation (1) gives the possibility of estimating
not only potential, but also actual activity. For
this purpose, we used the results of the determi-
nation of CH4 concentrations in the soil profile
(Tables 3a and 4a) and put obtained values as
the variable s to equation (1). Results of calcu-
lations are presented in Tables 3b and 4b.
The curves of methanogenesis under anaerobic
conditions were often concave, i.e., resembling
growing exponents. Hence, as the first approxi-
mation the following kinetic model may be used
for the quantitative description of methanogens
growth:
dx/dt=M.X, n=p.mc/(Km+ c)~ (o,m (3)
methane production:
dx/dt=pc/Y (4)
where c is the concentration of methanogens
substrates (H^+COj), |am and Km are the pa-
rameters of Monods equations, and Y is the
yields of biomass per unit of methane formed.
After the introduction of an anaerobic gas
mixture into cores, the specific growth rate m
increases up to its upper limit u (u~ u = con-
JL. JL JL. • \* • m
stant at c » Km). For this situation, the dynamics
of CH4 production can be described by the
integrated form of equation (3):
s = (x0/Y)[exp(U.mt)-l]
(5)
where x0 is the initial biomass of methanogens
(i.e., at the moment of substrate enrichment of
the soil). Hence, the exponential curve of CH4
emission can be interpreted as a consequence
of the methanogenic bacteria growth under
favorable unlimiting conditions. Best fit of
equation (5) to experimental data gave the
followingparameter values: um=0.015, h=0.36/
d and x0/ Y =1.0 mg C/m2. If we accept, as a first
approximation, the Y value 0.09 g of biomass C
per g of produced CH4-C (Roberton and Wolfe,
1970), then it appears that x0 = .092 mg of
biomass-C/m2. That is equivalent to the popu-
lation density of 10 cells per g of dry peat. Both
values |im and xo fell within those ranges, which
are reported for pure cultures of methanogens
or for enumeration of these bacteria in soils
(Kalyuzhny et al., 1985; Varfolomeyev, 1987).
However equation (3) was not valid for all
obtained experimental curves. Some of them
134
-------
were biphasic exponents, some linear, some
hyperbolic. The observed deviation should be
considered in future research. So far, we esti-
mated the average rates of CH4 formation dur-
ing the first day of anaerobic incubation (Tables
3b, 4b). These values are close to the initial rates
of methanogenesis after complete establishment
of anoxic conditions.
On the basis of core determinations of CH4
exchange rates, we evaluated two main compo-
nents of CH4 balance (its generation and con-
sumption) for the central part of ombrotrophic
bog (Table 3b, column 4),. It was found that the
algebraic sum of CH4 fluxes for all peat layers
gave values close to the CH4 emission measured
by the chamber technique (Table 3b, column 5).
A small systematic error always had a negative
sign, evidently due to the over estimation of the
CH4 uptake rate. Obviously, equation (1) is too
simple to obtain complete fitness to experimen-
tal data. Probably the most important deviation
comes from the fact that we did not consider the
dependence of the uptake rate on the dissolved
oxygen concentration. However, the main quali-
tative features of net CH4 emission were repro-
duced adequately: the negative emission of CH4
in July and the increased positive emission in
April.
For trenches, soil-core experiments and the
chamber technique gave different results even
from a qualitative point of view. The first showed
net consumption of CH4, while the second showed
net emission (Table 4b). This discrepancy is
explained by the fact that emitted methane was
produced not in trenches, but elsewhere in the
surrounding areas of the bog.
Data on all studied soils are summarized in
Table 2. The general regularity already men-
tioned for ombrotrophic bog was reproduced for
all soils. In the upper parts of soil profiles, the
process of CH4 consumption was prevalent, while
in the lower horizons, methanogenesis occurred.
The rates of CH4 exchange were maximal in
N. PANIKOV AND V. ZELENEV
wetlands: bogs and swamps. Despite wide-
spread opinion (Schutz et al, 1990), CH4 up-
take turned out to be intensive not in aerobic
and dry soils, but in wetlands. In strictly
automorphous soils located far from CH4
sources (the gray forest soil in Poushchino),
not only CH4 production, but also CH4 uptake
proceeded slowly.
The CH4 cycle in most studied soils was found
to be a closed one; the processes of
methanogenesis in the lower horizons were
exactly equal to or less than CH4 consumption.
The exceptions were the trenches and natural
depressions in ombrotrophic bogs, where CH4-
consuming activity was insufficient to compen-
sate the transfer and intensive CH4 generation
in the lower horizons. Special attention should
be paid to the question: What are the limiting
factors for methanotrophe development in
ombrotrophic bogs? Probably there are ad-
equate conditions (low pH and nutrient con-
tent, low temperature) for intensive growth of
these bacteria, but not of methanogenic bacte-
ria. As a result, the methanotrophs are able to
grow only sporadically in midsummer. Tran-
sient (in spring) methane formation becomes
more intensive than methane oxidation.
Under the influence of natural or anthropo-
genic factors (forest cutting, climate change,
pollution of the environment, etc.), the equi-
librium of the CH4 cycle maybe disturbed, and
as a result, CH4 emission into the atmosphere
would take place. The emission could be
expected from both aerobic soils and wetlands.
Therefore, many soils are the potential sources
of CH4. The idea about the exceptional role of
automorphous soils as sinks and wetlands as
the only sources of CH4 should be revised.
Contribution of Gas Fluxes to the Total Car-
bon Balance of Ombrotrophic Bog
Wetlands ecosystems play an essential role as
a CO2 sink. On the other hand, wetlands
135
-------
METHANE AND CARBON DIOXIDE PRODUCTION AND UPTAKE
themselves are the sources of many other green-
house gases such as CH4 and other volatile
organic substances.
The elucidation of a wetlands ecosystem func-
tioning mechanism becomes particularly nec-
essary in the context of the Global Change
Program. Hypothetically, the wetlands, instead
of beinga sink, canbe transformed into a source
of many green-house gases, including CO2. This
may be due to acceleration of the decomposi-
tion of organic matter stored for centuries and
millenniums. In the last part of our paper, we
consider the relationship between gas fluxes
and all main components of carbon-balance for
ombrotrophic bog (Table 5).
The VOS-C fluxes are very difficult to quantify,
but they are known to be less than 10% of CH4-
C flux and by several orders of magnitude less
than CO2-C flux. Thus, VOS could not contrib-
ute significantly to bog carbon balance. How-
ever, VOS composition and steady-state con-
tent do characterize the chemical and micro-
biological processes and reflect the functioning
mechanism. Composition VOS emitted from
Sosviatskoe ombrotrophic bog ecosystem is
shown in Table 6.
The chamber technique can provide only the
integrative C-fluxes evaluation between bog
and the atmosphere. Elucidation of compo-
nent flux sizes and further studies of the bog
ecosystem's functioning mechanism requires a
more detailed investigation and mathematical
modelling approach. Keepingin mind thisidea,
we have determined the main characteristics of
the bog that were recommended for measure-
ment for Northern Wetland studies by the
Dahlem Workshop (1989):
1. moss temperature (Table 5);
2. peat bulk density (Table 7);
3. CH4 profile concentrations for the
top layer (Tables 3a, 4a);
Table 5. Seasonal CO2 and CH4 flux rates for ombro-
trophic bog, 1990-1991.
Month
September
March
April
June
July
Dark
Respiration
Net
Photosynthesis
CH4
Emission
mg C/hr/m2
22.7
0.8
45
73.9
110.0
0.0
80.2
149.3
0.00
0.03
0.24
0.24
Temperature
(°C)
9.0
-1.0
2.0
15.0
20.0
Table 6. Volatile organic substances (VOS) in the air
above the surface of ombrotrophic bog (a).
1. Methane + CO2(b)
2. Ethane
3. Propane
4. Isobutane
5. Butane
6. Butylene
7. Acetaldehyde
8. Propanale
9. Acetone
10. Ethanol
11. -(99) (c)
12. -
13. Benzol
14. Chloroform
15. Isooctane (d)
16. Acetate
17. Siloxanes
18. Toluene
19. Ethylbenzene
20. Hydrocarbon C9H12
21. Hydrocarbon C9H12
22. Cymene
(a) VOS were determined by VA. Isidprov (chromate-
mass spectrometry).
(b) VOS are listed according to increase of their retardation
times.
(c) Unidentified compounds, molecular mass numbers are
indicated in brackets.
(d) Major component.
4. flux measurements by chamber tech-
nique (Tables 3b, 4b, 5);
5. organic matter decomposition rates
(Table 7).
To formalize the bog structure, we used the
concept of 'acrotelm' and 'catotelm,' proposed
by Ingram (1978). The acrotelm is the 'active'
layer of the upper 2 - 50 cm of the bog and is
predominantly aerobic. The catotelm is anoxic
and contains the rest of the peat mass down to
the underlying strata. The boundary between
them is located at the mean depth of the ground-
water table in summer (Ivanov, 1981), but is not
permanent.
136
-------
Table 7. Bulk density and decomposition characteris-
tics of ombrotrophic bog top layer, June 1991.
Horizon
0-10
10-20
20-30
30-40
Bulk
Density
g/cm3
0.06(a)
0.06
0.06
0.06
Respiration
Rate
mg C/hr/m2
23.4
12.1
0.6
13
Specific
Decomposition
Rate
1/yr
1.0*10
5310"1
2.8'IQ"1
5.8*10"2
Half-life
Tune
F
67.8
127.6
2473
1190
(a) Peat layer down to 100 cm has the same bulk density value; 100-250 c
layer has 0.09 g/cm3; 250 cm down to the bottom layer has 0.13 g/cm3.
According to one of the known models (Clymo,
1984), bog development depends on the rela-
tionship between the rates of moss productivity,
as well as peat decomposition in the acrotelm
and catotelm. The kinetics of peat decay were
assumed to obey the first order:
dm(z)/dt = - km(z)
(5)
where m(z) is the mass of the peat at depth z,
and k is the specific decay rate. It was stated
that the k-value is constant within the whole
strata of the acrotelm and catotelm, the acrotelm
value being several orders of magnitude higher.
However, our results (profile distribution of
CO2 and CH4 production rates in peat cores, see
Tables 3b, 4b, 7) proved that k-values are not
constant; the largest decomposition activity
N. PANIKO V AND V. ZELENEV
occurs at the upper 0-10 cm layer. Decompo-
sition within the deeper layers rapidly decreased.
Despite the existence of anaerobic conditions
in deep layers, the total CH4 production didn't
exceed 2-3% of total CO2 production (Tables
3a, 7) and could not be considered as a major
end product of the decompositionprocess. Thus,
the top aerobic layer of the bog (acrotelm)
should be regarded as the main source of car-
bon flux to the atmosphere.
It should be noted that our estimates of gas-
exchange rates by the chamber and core meth-
ods allow the identification of all parameters of
Clymo's model (1984). In the future, this model,
however, should be improved by taking into
account continuous vertical changes in specific
decomposition rates. These changes seem to
reflect the switching of microbiological pro-
cesses from one pathway to another. There-
fore, adequate mathematical models must have
at least two features: 1) the ability to describe
not only temporal, but also spatial variations of
studied variables, and 2) the ability to describe
the metabolic regulation of soil biota as influ-
enced by environmental conditions (concentra-
tions of oxygen, carbon compounds, biogenic
elements etc.).
REFERENCES
Andreae,M.O.andD.S.Shimel. 1989. Exchange of Trace
Gases between Terrestrial Ecosystems and the Atmo-
sphere. John Wiley & Sons,New York, p. 347.
Bouwman,A.F. 1990. Soils and the Greenhouse Effect.
Chichester, New York et al.: John Wiley & Sons, p. 575.
Cicerone, R. and R.Oremland. 1988. Biogeochemical
aspects of atmospheric methane. Global Biogeochem.
Cycles 2:299-327.
Clymo, R.S. 1984. The limits to peat growth. Phil. Trans.
R. Soc. Lond. B 303:605-654.
Gorbenko, AJu. and N.S. Panikov. 1989. Quantitative
description of microbial growth dynamics in soil as related
to process of primary production in ecosystem. Zurnal
obschey biologii (J.of General Biology) 49:38-59.
Ingram, HA.P. 1978. Soil layers in mires: function and
terminology. J.Soil.Sci. 29:224-227.
Ivanov, K.E. 1981. Water movement in mirelands. Trans-
lated by A.Tomson and HA.P. Ingram from Ivanov K.E.
(1975) Vodoobmen v bolotnych landshaftach. London:
Academic Press.
Kalyuzhny, S.V., A.N. Nozhevnikova, and S.D.
Varfolomeyev. 1985. The kinetics of Methanosarcina
vacuolatagrowth on methanol. Microbiologia 54:257-262.
137
-------
METHANE AND CARBON DIOXIDE PRODUCTION AND UPTAKE
Kasparov, S.V., N.S. Panikov, and O.I. Min'ko. 1985.
Membrane probe for analysis of gas composition of soil
air. Pochvovedenije (Soviet Soil Science) 11:145-151.
King, GM. 1990. Dynamics and controls of methane
oxidation in a Danish wetland sediments. FEMS Micro-
bial Ecology 74:309-324.
Kivinen, E. and P. Pakarinen. 1981. Geographical
distribution of peat resources and major peatland complex
types in the world. Suomal. Tiedeakat. Toim., Series A III
Gcologica-Geographica 132:1-28.
Krutko, P.D., AXMaximov, and L.M.Skvortsov. 1988.
Algorithms and programs for automaticsystems construc-
tion. Moscow: Radyo y svias, p. 306 .
Roberton, A.M. and R.S. Wolfe. 1970. Adenosine tri-
phosphate pools in Methanobacterium. J. Bacteriol. 102:
43-53.
Schutz,H.,W.Seiler,andH.Rennenberg. 1990. Soil and
land use related sources and sinks of methane (CH ) hi the
context of the global methane budget. In: Soils and the
Greenhouse Effect. John Wiley & Sons, New York, pp.
269-285.
Stewart, J.W.B., I. Aselmann, A.F. Bouwman, R.L.
Desjardins. 1989. Extrapolation of flux measurements to
regional and global scales. In: Exchange of Trace Gases
between Terrestrial Ecosystems and the Atmosphere.
John Wiley & Sons, New York, pp. 155-174.
Varfolomejev,S.D. 1987. Kinetic regularities of microbial
populations development. In: Contemporary Problems of
Biokinetics (Sovremennye Problemy Biokinetiky), edited
by S.D. Varfolomejev. Moscow Univ.Press, Moscow, pp.
6-77.
Whalen, S.C. and W.S. Reeburgh. 1988. A methane flux
tune series for tundra environments. Global Biogeo-
chemical Cycles 2:399-409.
Woodwell, G.M., R.H. Whittaker, WA. Reimers, G.E.
Likena,C.C.Delwiche,andD.B.Botkin. 1978. The biota
and the world carbon budget. Science 199:141-146.
AKNOWLEDGEMENTS
The authors are indebted to Ms. Marietta Cook for her tuneless efforts to transform the original word processing file
of this paper into a suitable file for use in the Workshop Proceedings.
138
-------
BOREAL FORESTS, THE CARBON CYCLE AND GLOBAL
CHANGE: A CHALLENGE FOR ECOLOGISTS
Gordon B. Bonan
ABSTRACT
The circumpolar boreal forest contains 20 percent of global terrestrial carbon; however, present studies suggesting the
importance of boreal forests in the global carbon cycle have not been based on an ecological, physiological and biophysical
understanding of the factors regulating production and decomposition. In this work, a previously developed mechanistic
model of photosynthesis, respiration, and decomposition that quantifies these relationships was used to examine
atmosphere-biosphere exchange ofCO2 i n the boreal forests of interior Alaska. Daily processes were modeled at the.stand
level. Required stand parameters were: canopy height, leaf area index, foliage nitrogen, forest floor biomass and nitrogen,
moss and humus thickness and green moss, sapwood, and root biomass. Site parameters were: aspect, slope, elevation,
drainage and soil color. Simulation with this model indicated that mature forests near Fairbanks are a significant annual
CO2sink. A forest dynamics model developed earlier that simulates the successional mosaic of forest stands in boreal forests
was combined with the stand-level physiology model to estimate landscape-averaged carbon fluxes. The landscape
surrounding Fairbanks was estimated to absorb 74 g of C/m2/yr. If the landscape-averaged net carbon flux is multiplied
by the areal extent of the circumpolar boreal forest, boreal forest ecosystems absorb 0.85 to 1.11 Gt of C/yr. Carbon fluxes
in this region may be representative of natural stands throughout the circumpolar boreal forest.
INTRODUCTION
The circumpolar boreal forest appears to have
a large role in the global carbon cycle. With
21% of the global soil carbon and 18% of the
carbon stored in terrestrial phytomass, the bo-
real forest is the largest reserve of soil carbon
and is second only to broad-leaved humid for-
ests in terms of terrestrial phytomass (Lashof,
1989). When these two pools are combined, the
circumpolar boreal forest contains 20% of the
global terrestrial carbon. The seasonal ampli-
tude of atmospheric carbon dioxide concentra-
tions increases with higher latitudes in the North-
ern Hemisphere (Kornhyr et al., 1985), and this
is thought to reflect the seasonal cycle of photo-
synthesis and respiration by terrestrial ecosys-
tems (Tucker et al., 1986). This amplitude is
greatest at 65° N, near the northern edge of the
boreal forest (Komhyr et al., 1985). Photosyn-
thesis and respiration by boreal forests may
account for 50% of the seasonal amplitude in
atmospheric CO2 at Barrow, Alaska, and 30%
of the seasonal amplitude at Mauna Loa
(D'Arrigo et al., 1987).
While these studies suggest the importance of
boreal forests in the global carbon cycle, they
have not been based on an ecological, physi-
ological and biophysical understanding of the
factors regulating production and decomposi-
tion. Our current understanding of the ecology
of boreal forests indicates that interactions
among climate, soil temperature, permafrost,
soil moisture, the forest floor, litter quality,
nutrient availability and fire control production
139
-------
BOREAL FORESTS, THE CARBON CYCLE AND GLOBAL CHANGE
and decomposition (Van Cleve and Viereck,
1981; Van Cleve et al., 1983a, 1983b, 1986,
1991; Bonan and Shugart, 1989). Here, I use a
mechanistic model of photosynthesis, respira-
tion and decomposition that quantifies these
relationships to examine atmosphere-biosphere
exchange of CO2intheborealforests of interior
Alaska.
The forest landscape near Fairbanks, Alaska
(64*49'N, 147°52'W), provides a useful opportu-
nity to examine the processes controlling pro-
duction and decomposition in boreal forests
and to develop biogeochemical ecosystem pro-
cess models. The forest landscape is a mosaic of
black spruce (Picea mariana (Mill.) B.S.P.),
white spruce (Picea glauca (Moench.) Voss),
aspen (Populus tremuloides Michx.), birch
(Betulapapyrifera Marsh.) and balsam poplar
(Populus balsamifera L.) forests that reflects
slope, aspect, elevation, soil parent material
and recurring fires and floods (Van Cleve et al.,
1983a, 1983b, 1986,1991).
Along the floodplains of the Tanana River,
erosion and deposition of sediments create a
successional mosaic of productive white spruce
and balsam poplar stands on alluvial deposits
where permafrost is absent (Viereck, 1970;
Van Cleve and Viereck, 1981; Viereck et al.,
1983, 1986). On higher terraces further back
from the river, unproductive moss-dominated
black spruce stands grow on cold, wet, nutrient-
poor soils underlain with permafrost. In the
uplands, recurring fires are common, creating a
successional mosaic of white spruce, black
spruce, birch and aspen stands of all ages
(Viereck, 1973,1983; Van Cleve and Viereck,
1981; Dyraess et al., 1986). In mature stands,
above-ground woody biomass ranges from 2.6
to 24.6 kg/m2, and above-ground tree produc-
tion ranges from 72 to 952 g/m2/yr (Van Cleve
et al., 1983a). Forest floor mass ranges from 1.7
to 10.5 kg/m2 (Van Cleve et al., 1983a), and
forest floor decomposition ranges from 116 to
298 g/m2/yr (Flanagan and Van Cleve, 1983).
Site conditions range from cold, wet, nutrient-
poor soils to warmer, mesic nutrient-rich soils.
Growing season soil temperature sums above
0°C range from 483 on cold soils underlain with
permafrost to 2217 on permafrost-free soils
(Viereck et al., 1983). Soil moisture ranges
from poorly-drained soils underlain with per-
mafrost to well-drained permafrost-free soils to
xeric soils on steep south-facing bluffs along the
Tanana River (Viereck et al., 1983). Nitrogen
mineralization ranges from 8 to 9 kg/ha/yr in
cold, wet black spruce stands (Van Cleve et al.,
1981) to 24 to 58 kg/ha/yr in warmer, perma-
frost-free birch stands (Flanagan and Van Cleve,
1983). The natural fire cycle in the Alaskan
taiga typically ranges from 50 to 200 years
(Dyrness et al., 1986), but can be as short as 26
years in extremely dry deciduous forests (Yarie,
1981). The vast majority of fires in interior
Alaska are small, but in severe fire years indi-
vidual fires can cover 50,000 to 200,000 ha
(Dyrness et al., 1986).
These factors interact to control ecosystem
structure and function (Figure 1). In particular,
decomposition is directly correlated with soil
temperature (Fox and Van Cleve, 1983; Van
Cleve and Yarie, 1986). Cold soil temperatures
in black spruce forests underlain with perma-
frost slow decomposition and nutrient mineral-
ization (Van Cleve et al., 1981, 1983a, 1983b;
Van Cleve and Yarie, 1986). This restricts tree
growth while promoting the accumulation of a
thick forest floor that further reduces soil tem-
perature. Black spruce may have adapted to
these nutrient-poor conditions through high
foliage longevity (Horn and Oechel, 1983) and
low growth potential and nutrient requirement
(Chapin, 1986). In contrast, with their high
growth potentials and nutrient requirements
(Chapin, 1986), white spruce and the succes-
sional hardwoods form productive stands on
warm, mesic, permafrost-free soils where
warmer soil temperatures result in rapid de-
composition and a high rate of nutrient miner-
alization (Van Cleve et al., 1983a, 1983b, 1991;
140
-------
G.B. BONAN
Reduced Organic Matter
Decomposition
Organic Matter
Accumulation
Reduced Forest Floor
and Mineral Soil
Temperature
Reduced
Organic Matter
Quality
Increased
forest Floor and
Mineral Soil H20
Slow
Tree Growth
Reduced Forest Floor
and Mineral Soil pH
I
Moss
Accumulation
Nutrient Conservation
Ecosystem
Process
Figure 1. Ecosystem processes and their controlling environmental variables (after Van Cleve and Viereck, 1981).
Van Cleve and Yarie, 1986). Van Cleve et al.
(1990) confirmed the importance of soil tem-
perature by experimentally heating a cold, black
spruce forest soil 8 to 10° C above ambient
temperatures. Warmer soil temperatures in-
creased decomposition compared with the
unheated control, increasing nutrient availabil-
ity and foliage photosynthesis.
Forest floor nitrogen and lignin concentrations
are also correlated with decomposition rates
(Flanagan and Van Cleve, 1983), and substrate
quality interacts with soil temperature to en-
hance or restrict nutrient availability (Flanagan
and Van Cleve, 1983; Van Cleve et al., 1983a,
1983b; Van Cleve and Yarie, 1986). Cold, wet
black spruce sites have forest floors with low
nitrogen and high lignin concentrations, which
further slows decomposition; warmer sites domi-
nated by white spruce and hardwoods have
forest floors with higher nitrogen and lower
lignin concentrations (Van Cleve et al., 1983a;
Van Cleve and Yarie, 1986). Simulation analy-
ses confirm that the joint effects of substrate
quality and soil temperature control produc-
tion and decomposition (Bonan, 1990a).
Successional differences in soil temperature,
nutrient availability, substrate quality, stand
structure and species composition significantly
affectproductionand decomposition (Van Cleve
and Viereck, 1981; Van Cleve et al., 1983a,
1983b, 1986, 1991). Over time, as the forest
develops, the forest floor becomes the principal
nutrient reservoir as nutrients are tied up in
undecomposed organic matter (Figure 2). The
accumulation of a thick forest floor reduces soil
temperature and increases soil moisture, fur-
ther reducing decomposition, nutrient avail-
ability and stand production. In mature black
spruce stands, moss productivity is similar to
above-ground tree production (Oechel and Van
Cleve, 1986).
Fires interrupt this process, consuming the for-
est floor, mineralizing nutrients contained in
organic matter and improving the soil thermal
regime (Viereck, 1973, 1983; Dyrness et al.,
1986). Mineral soil and organic matter are poor
conductors of heat energy, and direct soil heat-
ing during fire is minimal and has little lasting
effect on soil temperature. Moreover, the lower
portion of the forest floor often remains moist
141
-------
BOREAL FORESTS, THE CARBON CYCLE AND GLOBAL CHANGE
c
0)
E
w
O
Q.
<§
CD
NJ
CO
O)
o:
Forest Floor
I Avoilobll Nutrients
! Totol Nutrients
Nel Primory Production _A»oiloble Soil Nutrient
IVosculor Plonts (Overstory)
I Non-Vosculor Plonts
Phosphorus
Nitrogen
O.
E
o
CO
N
I
o
CO
Time
Figure 2. Hypothesized changes in net primary production, the forest floor and the mineral soil for 150 years following
fire in a black spruce stand (after Van Cleve et al, 1983b).
during burning (Dyrness and Norum, 1983).
However, by reducing the thickness of the for-
est floor, blackening the forest floor and remov-
ing the forest canopy, fire dramatically alters
the soil thermal regime (Dyrness et al., 1986).
A MECHANISTIC MODEL
OF ECOSYSTEM METABOLISM
Bonan (1991a, 1991b) developed a stand-level
biophysical and physiological model of daily
photosynthesis, respiration and decomposition
in boreal forests. Tree photosynthesis is a func-
tion of the CO2 diffusion gradient and the
boundary layer, stomatal and mesophyll resis-
tances. Boundary layer resistance is directly
proportional to leaf size and is inversely pro-
portional to wind speed. Stomatal resistance is
a function of irradiance, foliage temperature,
vapor pressure deficit and foliage water poten-
tial. Mesophyll resistance includes rate limita-
tions imposed by the diffusion of CO2 within
cells and the effects of foliage temperature,
irradiance and foliage nitrogen on the bio-
chemical reactions of photosynthesis. Moss pho-
tosynthesis is limited by irradiance, tempera-
ture and moisture content. Tree foliage, stem
and root respiration and moss respiration are
each partitioned into maintenance and growth
respiration. Maintenance respiration is an ex-
ponential function of temperature; growth res-
piration is a function of the efficiency with
which new tissue is synthesized. Microbial res-
piration is a function of soil moisture, soil
temperature and substrate quality.
Bonan (1991a) described the calculation of
required biophysical parameters such as stoma-
tal and boundary layer resistances, foliage tem-
perature, soil temperature and soil moisture.
The forest canopy is divided into three layers:
anupper canopy, a lower canopy and the ground
surface. The components of the surface energy
budget at each canopy layer are each written in
terms of the upper and lower canopy tempera-
tures and the temperature of the ground sur-
face. At each canopy layer, the energy budget
must sum to zero, and these three equations are
solved simultaneously for the three unknown
temperatures. Then, with ground surface tem-
perature, evapotranspiration and snow melt
known, soil temperature and soil moisture in a
multi-layer soil are updated.
142
-------
G.B. BONAN
i Black Spruce y Balsam Poplai
J White Spruce m Alder. Willow
y Paper Birch wnOwarf Birch
? Aspen
Permafrost
Bedrock
Loess
Gravel
UPLAND
RIVER FLOOD PLAIN
Figure 3. Generalized cross-sectional topography of the 23 forest sites near Fairbanks (after Viereck et al., 1983).
MODEL VALIDATION
Required stand parameters are: canopy height,
leaf area index, foliage nitrogen, forest floor
biomass and nitrogen, moss and humus thick-
ness and green moss, sapwood, and root bio-
mass. Site parameters are: aspect, slope, eleva-
tion, drainage and soil color. Based on observed
descriptions, these parameters were estimated
for 2 aspen, 2 birch, 3 balsam poplar, 5 white
spruce, 9 black spruce and 2 mixed conifer
mature stands located on a soil temperature,
soil moisture and soil nutrient gradient near
Fairbanks (Figure 3). Several simulated state
variables are shown in Figure 4 for a black
spruce stand.
Using climatic data for a "typical meteorologi-
cal year," simulated solar radiation, soil tem-
perature, foliage water potential, evapotrans-
piration and snow melt are consistent with
observed data (Bonan, 1991a). For example,
observed and simulated soil degree-days are
compared in Figure 5. Simulated SDD ac-
counts for 74% of the variation in the observed
SDD, and there is a 1:1 correspondence be-
tween observed and simulated SDD. Simulated
forest floor decomposition and above-ground
tree production are not significantly different
from observed data (Table 1). Simulated tree
photosynthesis is also consistent with observed
data (Table 2).
When averaged over all sites, monthly ecosys-
tem CO2 flux is significantly correlated with
atmospheric CO2 concentration from Barrow,
Alaska, with a lag of two months (r=0.95, alpha
< 0.001, Figure 6). The two-month lag occurs
because monthly ecosystem CO2 uptake is great-
est during June and declines during the remain-
143
-------
BOREAL FORESTS, THE CARBON CYCLE AND GLOBAL CHANGE
60
50
40-
30-
20-
10-
ONDJFMAMJJAS
MONTH
ONDJFMAMJJAS
MONTH
9" 15000
c/T
y 10000-
J2 5000
bj
o:
0
~~*~*Jilf
P 20
g
c> 0
S, -20
x
§ -40
8 -60
ONDJFMAMJJAS
MONTH
ONDJFMAMJJAS
MONTH
Figure 4. Simulated daily snow depth, soil temperature at 10 cm depth, stomatal resistance and ecosystem CO2 flux for
black spruce stand 29. Ecosystem CO2 flux is the sum of the tree, moss and microbial fluxes. A negative flux indicates
COa uptake.
3000
§
§ 2500-
u)
^ 2000-
3
=} 1500-
g 1000-
g 500-
o
0
Y=22.50+1.052-X
..... 45°reference line
R2=0.74
500 1000 1500 2000 2500
SIMULATED SOIL DEGREE DAYS
3000
Figure 5. Relationship between simulated and observed
soil degree days above freezing at a 10 cm depth from May
20 to September 10 (F=52.2, alpha < 0.001).
Table 1. Observed and simulated forest floor decomposi-
tion and above-ground tree production (mean± SE g/m2/
yr). Observed and simulated means for each forest type
are not significantly different at alpha = 0.05. (after
Bonan, 1991b).
Black spruce
White spruce
Aspen
Birch
Balsam poplar
Decomposition (a)
Observed Simulated
149 + 16 104 ±1
140 ±23 197 ±5
267 ±30 182 ±63
233 ±21 295 ±6
314 ±5
Tree Production (b)
Observed Simulated
113 ±17 130 ±38
366 ±63 396 ±36
565 ±199 671 ±243
470 ±61 957 ±222
552 ±206 1003 ±104
ing summer months, whereas CO2 drawdown at
Barrow increases from June to August, reach-
ing maximum drawdown in August. Fairbanks
is several hundred kilometers south of Barrow,
and this lag presumably reflects atmospheric
transport and the fact that the atmosphere acts
to integrate the flux signal, thereby producing a
delayed response. Coupled models of biosphere
CO2 flux and atmospheric transport show a
similar one- to two-month lag between maxi-
mum CO2 uptake by high-latitude ecosystems
(a) Observed data are for four black spruce,two white spruce,two
aspen and six birch stands.
(b) Observed data are for four black spruce,four white spruce,two
aspen,three bitch, and three balsam poplar stands.
and maximum drawdown in atmospheric CO2
concentration at Barrow (Pearman and Hyson,
1980; Fung et al., 1983,1987).
On average, the 23 stands absorb 1173 g CO2/
m2/yr. Net tree photosynthesis is the dominant
CO2 flux (Bonan, 1991b). Monte Carlo simula-
tions reveal that the simulated ecosystem CO2
uptake by the 23 stands is relatively stable based
on two constraints (Bonan, 1991b). When the
144
-------
G.B. BONAN
Table 2. Maximum photosynthetic rates (mg CO2/m2/
hr). Simulated data are maximum rates for conditions
comparable to the observed data, (after Bonan, 1991b).
Black spruce
Aspen
Birch
Balsam poplar
Observed
312-844 (a)
954 (b)
503 (c)
916 (b)
Simulated
479-640
942
498
942
(a) field conditions: foliage N=OJO to 0.74%.
(b) laboratory. 25° C soil and 20° C air temp.,
saturated light.
(c) laboratory: 25° C soil and 20° C air temp., half-
saturation light.
parameters that define tree photosynthesis are
each chosen at random from a uniform prob-
ability density distribution with a specified mini-
mum and maximum, only 11 of 200 parameter
sets produce results in which tree photosynthe-
sis and production are consistent with observed
data. For these 11 parameter sets, mean ecosys-
tem CO2 flux in the 23 stands ranged from -319
to -2286 g CO2/m2/yr. Thus, based on the likely
range of tree physiological parameters, only
5.5% of the possible parameter sets produce
results that are consistent with observed tree
photosynthesis and tree productivity data. Mean
ecosystem CO2 uptake by the 23 stands was less
than the baseline uptake (1173 g CO2/m2/yr)
for four parameter sets and was greater than the
baseline uptake for seven parameter sets. While
the probability that mean ecosystem CO2 up-
take for the 23 stands is less when the baseline
uptake is small, the possibility that one of these
simulations is the correct solution can not be
dismissed.
CARBON FLUXES IN A BOREAL
FOREST LANDSCAPE
The preceding simulations indicate that mature
forests near Fairbanks are a significant annual
CO2 sink. However, this estimate is not a ran-
dom sample of the forest landscape. In particu-
lar, recurring disturbances create a successional
mosaic of vegetation types of all ages in the
10:
5-0
JFMAMJJASOND
-5-
Figure 6. Relationship between monthly ecosystem CO2
flux and atmospheric CO2 concentration at Barrow (after
Bonan, 1991b).
forest landscape. Carbon fluxes vary among
successional stages, and landscape estimates
must be representative of this successional
mosaic.
I have previously developed a forest dynamics
model that simulates the successional mosaic of
forest stands in boreal forests (Bonan, 1989,
1990a, 1990b). The strength of this model is its
simulation of annual production and decompo-
sition in a mosaic of successional stands, but its
weakness is an inability to simulate the seasonal
cycle. This model can be combined with the
stand-level physiological model to estimate land-
scape-averaged carbon fluxes. In the following
analyses, I use the forest dynamics model to
simulate annual production, decomposition and
combustion during fire for 250 forest stands in
the landscape. Specific carbon fluxes are tree
production, moss production, decomposition
and annual carbon losses during fire. I then use
the physiological model to decompose these
annual carbon fluxes into their seasonal cycles.
Annual fire data can be partitioned into sea-
sonal components using long-term fire history
data (Barney, 1969). More detail on these analy-
ses can be found in Bonan (1991d).
The simulated landscape consists of five sites
representative of the landscape near Fairbanks:
well-drained and poorly drained floodplains,
145
-------
BOREAL FORESTS, THE CARBON CYCLE AND GLOBAL CHANGE
TREE
MOSS
E -1000-
0 -1500
en
E -2000
-2500
JFMAMJJASOND
MONTH
MICROBE
JFMAMJJASOND
MONTH
ECOSYSTEM
1000
JFMAMJJASOND
MONTH
JFMAMJJASOND
MONTH
Figure 7. Seasonal dynamics of landscape-averaged tree, moss, microbial and ecosystem carbon fluxes. Data points are
mean daily fluxes for each month. The ecosystem flux is the sum of the tree, moss and microbial fluxes (after Bonan, 1991d).
well-drained and poorly drained south slopes
and a poorly drained north slope (Viereck et al.,
1983). Each site was replicated for 50 stands for
a total of 250 simulated stands. The average
simulated carbon flux for each of the five sites
was weighted by the area! extent of each site in
the landscape based on the soil maps of Rieger
et al. (1979).
The landscape is a small source of carbon from
October through April, when plants do not
grow (Figure 7). In late-spring, as the snow
melts and the soil warms, trees begin to take up
carbon during growth. Microbial respiration
also increases as the soil warms, but the net
effect is that the landscape is a carbon sink
during the summer months. The occurrence of
fires has a pronounced seasonality (Barney,
1969), but in an average fire year, carbon losses
during fire have relatively little effect on the
seasonal dynamics of carbon fluxes (Bonan,
1991d).
The are no measurements of seasonal carbon
fluxes in the forests to validate the simulations.
However, seasonal estimates of the Normal-
ized Difference Vegetation Index (NDVI),
which is thought to be related to canopy photo-
synthesis, provide proxy data that can be com-
pared to simulated data (Running and Nemani,
1988). Observed NDVI values for Fairbanks
show a pronounced seasonal trend with a three
month maximum during June, July and August
(Figure 8). Monthly NDVI is significantly cor-
related with monthly tree carbon flux and ac-
counts for 85% of the variance in tree carbon
flux (F=51, n=12, alpha < 0.001, r2=0.85).
Running and Nemani (1988) also found that
NDVI accounts for 77% of the variance in their
simulated tree carbon flux at Fairbanks.
Simulated annual ecosystem carbon fluxes for
the 250 stands range from a loss of 164 g C/m2/
yr to an uptake of 476 g C/m2/yr; 18 of the 250
stands are an annual source of carbon. The
146
-------
G.B. BONAN
Flux = 134 - 6720*M3VT, r2 = 0.85
-i 1 1 r-
JFMAMJJASOND
MONTH
Figure 8. Seasonal dynamics of net tree carbon flux and
observed NDVI. Data points are mean daily flux and
NDVI for each month (after Bonan, 1991d).
Table 3. Landscape-averaged carbon fluxes (mean± SE g
C/m2/yr). Ecosystem flux is the sum of the tree, moss and
microbial fluxes. A negative flux indicates carbon uptake.
landscape absorbs, on average, 104 g C/m2/yr
during photosynthesis and respiration (Table
3). Carbon uptake by trees is the dominant flux.
Carbon uptake by mosses is of minor impor-
tance. Microbial respiration during decomposi-
tion is 64% of carbon uptake by trees. On
average, 30 g C/m2/yr is lost during combus-
tion, for a net uptake of 74 g C/m2/yr. This
combustion loss represents 29% of the net
annual carbon uptake by ecosystems.
The re-distribution of carbon within the simu-
lated forests indicates that live tree biomass and
undecomposed soil organic matter store 62%
and 37%, respectively, of the 104 g C/m2/yr
absorbed during metabolic activity. Schlesinger
(1990) concluded that the accumulation of re-
fractory humus in mineral soil is low enough
that if terrestrial ecosystems act as a carbon
sink, the carbon should be accumulating in live-
plant biomass and fresh, undecomposed plant
litter on the soil surface. My simulated storage
of carbon within these ecosystems is consistent
with this conclusion.
Based on our understanding of the physiology
and ecology of boreal ecosystems, the land-
scape surrounding Fairbanks absorbs 74 g C/
m2/yr. This estimate appears to represent the
Net Microbial Ecosystem Fire Net
Photosynthesis Respiration
Tree Moss
-225±10 -23±2 144±4 -104±11 30±2 -74
lower bounds of carbon uptake. An analysis of
possible model errors indicates that annual tree
production is slightly underestimated and an-
nual decomposition of coarse woody debris is
overestimated (Bonan, 1991d). If so, the annual
carbon uptake would be greater.
If the landscape-averaged net carbon flux is
multiplied by the areal extent of the circumpo-
lar boreal forest (11.5 to IS.OxlO6 km2), boreal
forest ecosystems absorb 0.85 to 1.11 Gt C/y.
Circumpolar carbon fluxes due to land use
change, which could influence this estimate, are
negligible (Houghton et al., 1987; Melillo et al.,
1988). This global estimate is preliminary, but
it indicates the potential for a moderate North-
ern Hemisphere terrestrial carbon sink in high
latitudes.
Tans et al. (1990) concluded that a Northern
Hemisphere terrestrial carbon sink of 2.0 to 3.4
Gt C /year is required to maintain the observed
latitudinal gradient in atmospheric CO2 con-
centrations. While my analyses indicate that
boreal forest ecosystems could account for part
of this sink, Tans et al. (1990) concluded that the
terrestrial sink occurs at "temperate" latitudes
and that "boreal" regions are a source of car-
bon. Part of this discrepancy could be semantic
in that Tans et al. (1990) provided no latitudinal
definition of "temperate" and "boreal." While
tundra and the unproductive northern taiga
may be carbon sources (Grulke et al., 1990;
King et al., 1990), production data for Fair-
banks are representative of the main and south-
ern taiga (Bonan, 199 Ib), suggesting that these
lower latitude regions are an annual carbon
sink.
147
-------
BOREAL FORESTS, THE CARBON CYCLE AND GLOBAL CHANGE
INTERANNUAL VARIABILITY
IN ATMOSPHERIC CO2
Atmospheric CO2 concentrations show a sea-
sonal cycle that reflects photosynthesis and respi-
ration by terrestrial plants (Tucker et al., 1986).
The amplitude of this seasonal cycle has in-
creased (e.g., by 0.75%/yr at Mauna Loa,
Bacastow et al., 1985), and this is thought to
reflect terrestrial biotic influences such as in-
creased photosynthesis due to higher ambient
CO2 concentrations (i.e., CO2 fertilization) or
increased winter respiration due to warmer tem-
peratures (Bacastow et al., 1985; Keeling et al.,
1985; Gammon et al., 1985). Boreal forests ap-
pear to have a major role in the carbon cycle
(D'Arrigo et al., 1987; Lashof, 1989; Bonan,
1991b, 1991d), and if terrestrial ecosystems are
contributing to the increasing amplitude, such a
signal should appear in the metabolic activity of
boreal forests.
In the following analyses, the stand-level physi-
ological model is combined with observed daily
weather data from 1974 to 1982 and the ob-
served increase in annual atmospheric CO2
concentration to test whether the increasing
amplitude is consistent with the physiology and
ecology of boreal forest ecosystems. From 1974
to 1982, annual CO2 concentration at Barrow,
Alaska, increasedfrom333 to 343 ppm (Peterson
et al., 1986). These annual concentrations were
used to drive the model in response to increas-
ing ambient CO2. Daily air temperature and
precipitation for 1974 to 1982 were obtained
from the National Oceanic and Atmospheric
Administration's co-operative data for Fair-
banks. This record does not include cloudiness,
relative humidity, wind speed and air pressure.
For these parameters, data for a "typical me-
teorologicalyear," as described inBonan (199la,
1990b), were used (cf. Bonan, 1992).
These simulations utilized the 23 stands de-
scribed earlier and in Bonan (1991a, 1991b).
Over the nine years of simulation, the param-
eters that described the biomass and nutrient
status of the forests were held constant at the
values given in Bonan (1991a, 1991b). This is
reasonable because the 23 stands are all mature
forests in which changes in forest structure and
the forest floor occur slowly.
Based on a five-day average after the annual
trend was removed, maximum ecosystem CO2
uptake increased by (mean±SE) 120 ±8 mg/
m2/day/yr over the nine-year period. Maxi-
mum ecosystem CO2 loss increased by 22 ±1
mg/m2/day/yr. The seasonal amplitude, de-
fined as the difference between maximum CO2
uptake and maximum CO2 loss, increased by
142 ± 8 mg CO2/m2/day/yr, or 0.52 ± 0.03 %/yr
relative to 1974. When the simulations were
repeated with ambient CO2 held at 337 ppm for
the nine-year period (i.e., the average of the
observed data), the seasonal amplitude in-
creased by 12 ±2 mg CO2/m2/day/yr, or'only
0.04 ±0.01 %/yr, relative to 1974.
While short-term laboratory experiments show
that photosynthesis increases with higher ambi-
ent CO2, the realization of this growth incre-
ment has been subjected to much debate (Strain
and Cure, 1985;Eamusand Jarvis, 1989;Bazzaz,
1990). The current analyses indicate that the
increased amplitude of seasonal metabolic ac-
tivity in these forests is consistent with the
observed increase in the seasonal amplitude of
atmospheric CO2, and that CO2 fertilization has
a causal role.
Higher tree production caused the increased
seasonal amplitude. However, the growth in-
crease was small: average above-ground tree
dry matter production for the 23 stands in-
creased by only 3 g/m2/yr, or 0.8% of the
observed above-ground tree production (Van
Cleve et al., 1983a). Against the background of
observed above-ground tree biomass and the
standard error in observed estimates of above-
ground treeproduction (Van Cleve etal., 1983a),
the expected increase in tree growth is neither
148
-------
G.B. BONAN
very measurable nor very large.
The small increase in tree growth needed to
produce the amplitude change suggests that
significant changes in atmospheric CO2 con-
centration can occur without large changes in
net primary production, nutrient availability, or
nutrient use efficiency. Moreover, the annual
CO2 growth increment needed to cause the
increased amplitude is small, only 1.35 ppm/yr.
This suggests that more attention should be
focused on the response of vegetation to small
CO2 changes rather than to the 200 to 400%
CO2 increases that have typically been studied
(e.g., Eamus and Jarvis, 1989).
When the annual trends in atmospheric CO2 at
Barrow and the simulated ecosystem flux are
removed, monthly ecosystem flux is signifi-
cantly correlated with atmospheric CO2 with a
lag of two months and accounts for 93% of the
variation in atmospheric CO2 at Barrow from
1974 to 1982 (F= 1304, n= 106, alpha < 0.001).
The correlation between metabolic activity in
the boreal forests near Fairbanks and atmo-
spheric CO2 at Barrow suggests a causal rela-
tionship. However, whereas the seasonal am-
plitude of ecosystem CO2 flux increased signifi-
cantly, two analyses have not found a significant
increase in the amplitude of atmospheric CO2
at Barrow (Peterson et al., 1986; Chan and
Wong, 1990). Moreover, whereas the simulated
ecosystem CO2 fluxes did not exhibit large year-
to-year variation, the annual rate of CO2 in-
crease at Barrow ranged from 0.3 to 2.0 ppm/yr
(Peterson et al., 1986). Also, the Barrow record
shows within-day CO2 variations of up to 50%
of the annual range and day-to-day variations of
15 to 50% of the annual range (Halter and
Peterson, 1981). This suggests that Barrow inte-
grates a large area in addition to interior Alaska.
In fact, analyses of the Barrow CO2 record
indicate that winter and summer variability in
CO2 concentrations reflects synoptic scale at-
mospheric flow that modifies local sources and
sinks of CO2 (Peterson et al., 1980; Halter and
Peterson, 1981; Halter and Harris, 1983; Halter
et al., 1985).
SENSITIVITY OF CARBON FLUXES
TO CLIMATIC PARAMETERS
Boreal forests contain large quantities of soil
carbon, prompting concern that climatic warm-
ing may stimulate microbial respiration and
accentuate increasing atmospheric CO2 con-
centrations. While soil temperature is an im-
portant determinant of decomposition rates in
boreal soils (Van Cleve et al., 1983a; Fox and
Van Cleve, 1983; Schlentner and Van Cleve,
1985; Van Cleve and Yarie, 1986; Van Cleve et
al., 1990), the accompanying increase in nutri-
ent availability may promote tree growth in
these nutrient-poor soils, thus offsetting the
increased CO2 loss by microbial respiration.
For example, Van Cleve et al.'s (1990) experi-
mental heating of a cold, black spruce forest soil
caused greater decomposition, but the result-
ing increase in nutrient availability increased
foliage photosynthesis. How these two carbon
fluxes balance at an annual time scale or longer
is unknown.
The interaction between nutrient availability
and net primary productivity is a major control
over the response of terrestrial ecosystems to
climatic change (Billings et al., 1984; Pastor and
Post, 1988; Schimel et al., 1990). Two simula-
tion analyses indicate that this interaction among
decomposition, production and nitrogen min-
eralization is an important determinant of car-
bon source/sink relationships in boreal forests
and their short-term response to soil warming.
Sensitivity experiments with Bonan's (1990a,
1990b) boreal forest gap model, which includes
nitrogen mineralization, confirm that while a
5°C increase in air temperature warms the soil
and increases decomposition rates, greater ni-
trogen availability stimulates tree growth such
that these forest ecosystems will, in the short-
term (i.e., 25 years after the climatic change),
149
-------
BOREAL FORESTS, THE CARBON CYCLE AND GLOBAL CHANGE
Table 4. Annual CO2 fluxes (g/m2/yr) with climatic
warming. Negative fluxes indicate CO2 uptake.
Air Temperature
Control 1°C 3° C 5° C
Tree
Photosynthesis
Respiration
Moss
Photosynthesis
Respiration
Microbial Respiration
Ecosystem Flux
-1864
1510
-566
366
171
-384
-1916
1604
-579
393
176
-323
-2030
1799
-600
454
184
-192
-2073
2002
-613
529
190
36
Table 5. Annual ecosystem CO2 flux (g/m2/yr) with
climatic warming and as a function of foliage nitrogen.
Negative fluxes indicate CO2 uptake. Current foliage
nitrogen is 0.7%N. Nitrogen indicates the additional nitro-
gen mineralization needed to increase foliage nitrogen
from current levels.
absorb more CO2 (cf. Bonan and Van Cleve,
1992). Whether the increased tree production
that causes the greater CO2 uptake can be
maintained indefinitely is unknown. These
analyses examined only the short-term conse-
quences of below-ground feedbacks that affect
ecosystem CO2 uptake with climatic warming.
The effect of these feedbacks on ecosystem CO2
flux is small, indicating that the direct effects of
climatic warming on tree photosynthesis and
respiration may be more important determi-
nants of short-term carbon source/sink rela-
tionships arising from climatic warming.
Simulation analyses with the stand-level physi-
ological model also highlight the importance of
nutrient availability for the sensitivity of carbon
source/sinkrelationships to climatic change. In
the9blackspruceusedinBonan(1991a,1991b),
1* C, 3" C and 5° C increases in daily air tempera-
ture cause ecosystem CO2 uptake to decrease
(Table 4). While microbial respiration and pho-
tosynthesis increase with warmer air and soil
temperatures, the greatest change occurs in
tree respiration, which increases greatly as air
temperature increases. However, when in-
creased nutrient availability is included in the
simulations, tree photosynthesis increases in
these nutrient-poor conifer stands. The net
effectisthatthesestands take upgreater amounts
ofC02(Table5).
Control
1 C
3 C
5~C
Nitrogen
(g/m2/yr)
0.7%N
-384
-323
-192
36
0.8%N
-539
-482
-359
-134
0.5
0.9%N
-681
-628
-512
-290
1.0
1.0%N
-813
-763
-653
-433
1.6
CONCLUSION
The proceeding analyses indicate that the bo-
real forests of interior Alaska have an impor-
tant role in the global carbon cycle. The long-
term ecological research on the controls of
ecosystem structure and function in interior
Alaska provide a particularly useful database
with which to consider biosphere-atmosphere
carbon exchange. Moreover, site conditions
and forest productivity in interior Alaska span
the range of data for the circumpolar boreal
forest (Bonan, 199 Ic), and ecological relation-
ships developed in interior Alaska are relevant
for other bioclimatic regions of the boreal for-
est (Bonan and Shugart, 1989; Bonan, 1989,
1990b, 1991c). Thus, the carbon fluxes in this
region may be representative of natural stands
throughout the circumpolar boreal forest.
However, many issues remain unresolved, and
these analyses are not meant to be definitive.
For example, long-term changes in the struc-
ture and function of the boreal landscape, due
to either the direct effects of climatic change on
metabolic activity or the indirect effects associ-
ated with changes in the fire regime, have not
been considered. Rather, these analyses are
meant to highlight some of the issues that must
be addressed when considering boreal forests,
the carbon cycle and global change.
150
-------
G.B. BONAN
In particular, with the exception of issues re-
lated to changes in land use, the description of
the role of the terrestrial biosphere in the
carbon cycle has been dominated by scientists
whose training is more geophysical than eco-
logical (e.g., Fung, Keeling, Tans). Studies of
production and decomposition have long been
a staple of ecological research. If ecologists can
not use this knowledge to answer global change
issues, the debate will be defined by those
scientists who are willing to examine the rel-
evant questions.
REFERENCES
Bacastow, R.B., C.D. Keeling, and T.P. Whorf. 1985.
Seasonal amplitude increase in atmospheric CO2 concen-
tration at Mauna Loa, Hawaii, 1959-1982. J. Geophys.
Res. 90:10529-10540.
Barney, RJ. 1969. Interior Alaska wildfires 1956-1965.
U.S. For. Serv., Pacific Northwest Forest Range Exper.
Sta., Juneau, Alaska.
Bazzaz,FA. 1990. The response of natural ecosystems to
the risingglobal CO2 levels. Annu. Rev. Ecol. Syst. 21:167-
196.
Billings, W.D., K.M. Peterson, J.O. Luken, and DA.
Mortensen. 1984. Interaction of increasing atmospheric
carbon dioxide and soil nitrogen on the carbon balance of
tundra microcosms. Oecologia 65:26-29.
Bonan, G.B. 1989. Environmental factors and ecological
processes controlling vegetation patterns inboreal forests.
Landscape Ecology 3:111-130.
Bonan, G.B. 1990a. Carbon and nitrogen cycling in North
American boreal forests. I. Litter quality and soil thermal
effects in interior Alaska. Biogeochemistry 10:1-28.
Bonan, G.B. 1990b. Carbon and nitrogen cycling in North
Americanborealforests.il. Biogeographic patterns. Can.
J. For. Res. 20:1077-1088.
Bonan, G.B. 1991a. A biophysical surface energy budget
analysis of soil temperature in the borealforests of interior
Alaska. Water Resour. Res. 27:767-781.
Bonan, G.B. 1991b. Atmosphere-biosphere exchange of
carbon dioxide in borealforests. J. Geophys. Res. 96:7301-
7312.
Bonan, G.B. 1991c. A simulation analysis of environmen-
tal factors and ecological processes in boreal forests. In:
A Systems Analysis of theGlobal Boreal Forest, edited by
H.H. Shugart, R. Leemans, and G.B. Bonan, Cambridge
University Press, Cambridge, pp. 404-426.
Bonan, G.B. 1991d. Seasonal and annual carbon fluxes in
a boreal forest landscape. J. Geophys. Res. 96:17329-
17338
Bonan, G.B. 1992. Comparison of atmospheric carbon
dioxide concentration and metabolic activity in boreal
forest ecosystems. Tellus 44b:173-185
Bonan, G.B. and H.H. Shugart. 1989. Environmental
factors and ecological processes in boreal forests. Annu.
Rev. Ecol. Syst. 20:1-28.
Bonan, G.B. and K. Van Cleve. 1992. Soil temperature,
nitrogen mineralization, and carbon source-sink relation-
ships in boreal forests. Can. J. for Res. 22:629-639
Chan, Y.H. and C.S.Wong. 1990. Long-term changes in
amplitudes of atmospheric CO2 concentrations at Ocean
Station P and Alert, Canada. Tellus 426:330-341.
Chapin, F.S. III. 1986. Controls over growth and nutrient
use by taiga forest trees. In: Forest Ecosystems in the
Alaskan Taiga, edited by K. Van Cleve, F.S. Chapin, P.W.
Flanagan, LA. Viereck, and C.T. Dyrness, Springer-
Verlag, New York, pp. 96-111.
DArrigo, R., G.C. Jacoby, and I.Y. Fung. 1987. Boreal
forests and atmosphere-biosphere exchange of carbon
dioxide. Nature 329:321-323.
Dyrness, C.T., and RA. Norum. 1983. The effects of
experimental fires on black spruce forest floors in interior
Alaska. Can. J. For. Res. 13:879-893.
Dyrness, C.T., LA. Viereck, and K. Van Cleve. 1986. Fire
in taiga communities of interior Alaska. In: Forest ecosys-
tems in the Alaskan Taiga, edited by K. Van Cleve, F.S.
Chapin, P.W. Flanagan, LA. Viereck, and C.T. Dyrness.
Springer-Verlag, New York, pp. 74-86.
Eamus, D. and P.G. Jarvis. 1989. The direct effects of
increase in the global atmospheric CO2 concentration on
natural and commercial temperate trees and forests. Adv.
Ecol. Res. 19:1-55.
151
-------
BOREAL FORESTS, THE CARBON CYCLE, AMD GLOBAL CHANGE
Flanagan, P.W. and K. Van Cleve. 1983. Nutrient cycling
inrclation to decomposition and organic-matter quality in
taiga ecosystems. Can. J. For. Res. 13:795-817.
Fox, JJF. and K. Van Cleve. 1983. Relationships between
cellulose decomposition, Jennys k, forest-floor nitrogen,
and soil temperature in Alaskan taiga forests. Can. J. For.
Res. 13:789-794.
Fung, I., K. Prentice, E. Matthews, J. Lerner, and G.
Russell. 1983. Three-dimensional tracer model study of
atmospheric CO2: response to seasonal exchanges with
the terrestrial biosphere. J. Geophys. Res. 88:1281-1294.
Fung, I.Y., C J. Tucker, and K.C. Prentice. 1987. Applica-
tion of advanced very high resolution radiometer vegeta-
tion index to study atmosphere-biosphere exchange of
CO2. J. Geophys. Res. 92:2999-3015.
Gammon, RJH., E.T. Sundquist, and P.J. Fraser. 1985.
History of carbon dioxide in the atmosphere. In: Atmo-
spheric Carbon Dioxide and the Global Carbon Cycle,
edited by J.R. Trabalka. Report DOE/ER-0239, US
Dcpt. Energy, Carbon Dioxide Res. Div., Washington,
D.C., pp. 27-62
Grulke, NJE., G.H. Riechers, W.C. Oechel, U. Hjelm, and
C. Jaeger. 1990. Carbon balance in tussock tundra under
ambient and elevated atmospheric CO2. Oecologia83:485-
494.
Halter, B. and J.M. Harris. 1983. On the variability of
atmospheric carbon dioxide concentration at Barrow,
Alaska, during winter. J. Geophys. Res. 88:6858-6864.
Halter.B.C, JM. Harris, andKA.Rahn. 1985. Astudy of
winter variability in carbon dioxide and Arctic haze aero-
sols at Barrow, Alaska. Atmos. Environ. 19:2033-2037.
Halter, B.C. and J.T.Peterson. 1981. On the variability of
atmospheric carbon dioxide concentration at Barrow,
Alaska, during summer. Atmos. Environ. 15:1391-1399.
Horn, J.L. and W.C. Oechel. 1983. The photosynthetic
capacity, nutrient content, and nutrient use efficiency of
differentneedleage-classesofblackspruce(Piceamariana)
found in interior Alaska. Can. J. For. Res. 13:834-839.
Houghton, RA., R.D. Boone, J.R. Fruci, J.E. Hobbie,
J.M. Melillo, C A. Palm, B J. Peterson, G.R. Shaver, G.M.
Woodwell, B. Moore, D.L. Skole, and N. Myers. 1987. The
flux of carbon from terrestrial ecosystems to the atmo-
sphere in 1980 due to changes in land use: geographic
distribution of the global flux. Tellus 39B:122-139.
Keeling, C.D., T.P. Whorf, C.S. Wong, and R.D. Bellagay.
1985. The concentration of atmospheric carbon dioxide at
Ocean Weather Station P from 1969 to 1981. J. Geophys.
Res. 90:10511-10528.
King, A.W., R.V. ONeffl, and D.L. DeAngelis. 1990.
Using ecosystem models to predict regional CO2 ex-
change between the atmosphere and the terrestrial bio-
sphere. Global Biogeochem. Cycles 3:337-361.
Komhyr, W.D., R.H. Gammon, T.B. Harris, L.S.
Waterman, T. J. Conway, W.R. Taylor, and K.W. Thoning.
1985. Global atmospheric CO2 distribution and variations
from 1968-1982 NOAA/GMCC CO2 flask sample data. J.
Geophys. Res. 90:5567-5596.
Lashof, DA. 1989. The dynamic greenhouse: feedback
processes that may influence future concentrations of
atmospheric trace gases and climatic change. Clim. Change
14:213-242.
Melillo, J.M., J.R. Fruci, RA. Houghton, B. Moore, and
D.L. Skole. 1988. Land-use change in the Soviet Union
between 1850 and 1980: causes of a net release of CO2 to
the atmosphere. Tellus 406:116-128.
Oechel, W.C. and K. Van Cleve. 1986. The role of bryo-
phytes in nutrient cycling in the taiga. In: Forest Ecosys-
tems in the Alaskan Taiga, edited by K. Van Cleve, F.S.
Chapin, P.W. Flanagan, LA. Viereck, and C.T. Dyrness.
Springer-Verlag, New York, pp. 121-137.
Pastor, J. and W.M. Post. 1988. Response of northern
forests to CO2-induced climate change. Nature 334:55-58.
Pearman, G.I. and P. Hyson. 1980. Activities of the global
biosphere as reflected hi atmospheric CO2 records. J.
Geophys. Res. 85:4468-4474.
Peterson, J.T., KJ. Hanson, BA. Bodhaine, and SJ.
Oltmans. 1980. Dependence of CO2, aerosol, and ozone
concentrations on wind direction at Barrow, Alaska, dur-
ing winter. Geophys. Res. Lett. 7:349-352.
Peterson, J.T., W.D. Komhyr, L.S. Waterman, R.H. Gam-
mon, K.W. Thoning, andT.J. Conway. 1986. Atmospheric
CO2 variations at Barrow, Alaska, 1973-1982. J. Atmos.
Chem. 4:491-510.
Rieger, S., D.B. Schoephorster, and C.E. Furbush. 1979.
Exploratory Soil Survey of Alaska. U.S. Dept. of Agricul-
ture, Soil Conservation Service, Washington, D.C.
152
-------
G.B. BONAN
Running, S.W. and R.R. Nemani. 1988. Relating seasonal
patterns of the AVHRR vegetation index to simulated
photosynthesis and transpiration of forests in different
climates. Remote Sens. Environ. 24:347-367.
Schimel, D.S., W.J. Parton, T.G.F. Kittel, D.S. Ojima, and
C.V. Cole. 1990. Grassland biogeochemistry: links to
atmospheric processes. Clim. Change 17:13-25.
Schlentner, R.E. and K. Van Cleve. 1985. Relationships
between CO2 evolution from soil, substrate temperature,
and substrate moisture hi four mature forest types hi
ulterior Alaska. Can. J. For. Res. 15:97-106.
Schlesinger, W.H. 1990. Evidence from chronosequence
studies for a low carbon-storage potential of soils. Nature
348:232-234.
Strain, B.R. and J.D. Cure. 1985. Direct Effects of Increas-
ing Carbon Dioxide on Vegetation. Report DOE/ER-
0238, US Dept. Energy, Carbon Dioxide Res. Div., Wash-
ington, D.C.
Tans, P.P., I.Y. Fung, and T. Takahashi. 1990. Observa-
tional constraints on the global atmospheric CO2 budget.
Science 247:1431-1438.
Tucker, C.J., I.Y. Fung, C.D. Keeling, and R.H. Gammon.
1986. Relationship between atmospheric CO2 variations
and a satellite-derived vegetation index. Nature 319:195-
199.
Van Cleve, K. and LA. Viereck. 1981. Forest succession
inrelation to nutrient cyclingintheborealforestof Alaska.
In: Forest Succession: Concepts and Application, edited
by D.C. West, H.H. Shugart, and D.B. Botkin. Springer-
Verlag, New York, pp. 185-211.
Van Cleve, K. and J. Yarie. 1986. Interaction of tempera-
ture, moisture, and soil chemistry hi controlling nutrient
cycling and ecosystem development hi the taiga of Alaska.
In: Forest ecosystems hi the Alaskan Taiga, edited by K.
Van Cleve, F.S. Chapin, P.W. Flanagan, LA. Viereck, and
C.T. Dyrness. Springer-Verlag, New York, pp. 160-189.
Van Cleve, K., R. Barney, and R. Schlentner. 1981. Evi-
dence of temperature control of production and nutrient
cycling hi two interior Alaska black spruce ecosystems.
Can. J. For. Res. 11:258-273.
Van Cleve, K., L. Oliver, R. Schlentner, LA. Viereck, and
C.T. Dyrness. 1983a. Productivity and nutrient cycling in
taiga forest ecosystems. Can. J. For. Res. 13:747-766.
Van Cleve, K., C.T. Dyrness, LA. Viereck, J. Fox, F.S.
Chapin HI, and W. Oechel. 1983b. Taiga ecosystems hi
ulterior Alaska. Bioscience 33:39-44.
Van Cleve, K., F.S. Chapin HI, P.W. Flanagan, LA.
Viereck, and C.T. Dyrness. 1986. Forest ecosystems hi the
Alaskan Taiga. Springer- Verlag, New York.
Van Cleve, K, W.C. Oechel, and J.L. Horn. 1990. Re-
sponse of black spruce (Picea mariana) ecosystems to soil
temperature modification hi ulterior Alaska. Can. J. For.
Res. 20:1530-1535.
Van Cleve, K., F.S. Chapin III, C.T. Dyrness, and LA.
Viereck. 1991. Element cycling hi taiga forests: state-
factor control. Bioscience 41:78-88.
Viereck, LA. 1970. Forest succession and soil develop-
ment adjacent to the Chena River hi ulterior Alaska. Arct.
Alp. Res. 2:1-26.
Viereck, LA. 1973. Wildfire hi the taiga of Alaska. Quat.
Res. 3:465-495.
Viereck, LA. 1983. The effects of fire in black spruce
ecosystems of Alaska and northern Canada. In: The role
of Fire hi Northern Circumpolar Ecosystems, edited by
R.W. Wein and DA. MacLean. Wiley, New York, pp. 201-
220.
Viereck, LA., C.T. Dyrness, K. Van Cleve, and M.J.
Foote. 1983. Vegetation, soils, and forest productivity hi
selected forest types hi ulterior Alaska. Can. J. For. Res.
13:703-720.
Viereck, LA., K. Van Cleve, and C.T. Dyrness. 1986.
Forest ecosystem distribution hi the taiga environment. In:
Forest Ecosystems hi the Alaskan Taiga, edited by K. Van
Cleve, F.S. Chapin, P.W. Flanagan, LA. Viereck, and C.T.
Dyrness. Springer-Verlag, New York, pp. 22-43.
Yarie, J. 1981. Forest fire cycles and Me tables: a case study
from ulterior Alaska. Can. J. For. Res. 11:554-562.
153
-------
-------
CARBON BALANCE IN FOREST ECOSYSTEMS:
RESPONSE TO NITROGEN
James A. Entry, Kim G. Mattson, and Mary Beth Adams
ABSTRACT
The earth's terrestrial ecosystems are enormous processors and storage pools of global carbon. In an undisturbed state,
terrestrial ecosystems exist in equilibrium with the atmosphere; about 100 Gt of carbon are taken up by photosynthesis, and
approximately equal amounts are released by heterotrophic respiration, organic matter decomposition and fires. Increasing
plant growth, carbon accumulation and carbon storage in terrestrial ecosystems by nutrient manipulation may provide a
buffering effect against increasing atmospheric CO2 concentrations. In most temperate and boreal ecosystems, nitrogen
availability is the major/actor limiting plant growth. Although the effects of forest fertilization on the growth of forest trees
have been well studied, there is little information available on long-term effects of nitrogen fertilization on terrestrial
ecosystems, and the full effects of nitrogen fertilization on the amount of carbon sequestered by forest ecosystems cannot be
adequately assessed. The prospects for controttinginsect and disease damage throughfertilization are also largely unknown.
Therefore, long-term studies of ecosystem response and tree physiology are necessary.
INTRODUCTION
The earth's terrestrial ecosystems are enor-
mous processors and storage pools of global
carbon. In an undisturbed state, terrestrial
ecosystems are at equilibrium with the atmo-
sphere; about 100 Gt of carbon are taken up by
photosynthesis and approximately equal
amounts are released by heterotrophic respira-
tion, organic matter decomposition and fires
(Mooney et al., 1987). Carbon cycling within a
forest ecosystem consists of conversion of CO2
to organic carbon via photosynthetic fixation by
autotrophs, transfers of organic carbon among
various pools, and oxidation back to CO2 by
autotrophic and heterotrophic respiration (Fig-
ure 1). Both photosynthesis and CO2-releasing
processes are controlled by temperature and
moisture and are also related to the inherent
productivity of the ecosystem.
Predicted changes in the global climate are
expected to affect the structure of many terres-
trial ecosystems. Major structural changes could
alter the exchange of radiation, water and car-
bon between ecosystems and the atmosphere,
resulting in even greater climate alterations
(King et al., 1989). One of the greatest feed-
backs to the climate system will most likely be
through shifts in the exchange of carbon be-
tween the atmosphere and terrestrial ecosys-
tems.
In many temperate and boreal ecosystems, ni-
trogen availability is the major factor limiting
plant growth. Increasing plant growth, carbon
accumulation and carbon storage in terrestrial
ecosystems by nutrient manipulation may pro-
vide a buffering effect against increasing atmo-
spheric CO2 concentrations. Aside from direct
effects on plant growth, nitrogen may also affect
carbon cycling within the soil. Compared with
other means of altering the carbon cycling pat-
155
-------
CARBON BALANCE IN FOREST ECOSYSTEMS
\
Kt
X
Figure 1. (
are represe
CO, ••
In
1
^ ABOVE-GROUND
C| '
f4
\ BELOW-GROUND
C2
LIVING
CARBON
^F3
^^»" LITTER
C3
F7
F6 ,
^^^ UNPROTECTED
^ SOIL O.M. _
C4
co2
co2
„/
/
' Fll
K. PROTECTED
n *ni| n,n
F12
BELOW-GROUND
F 1 = gross photosynthesis
F2 above-ground respiration
T3 litter deposition
F4 below-ground transport
to rnlzospnere respiration
t-b rhizo-deposltton • exudates
F7 litter Incorporation • leaching
C F8 litter respiration
FIO = leaching and erosion
Fio nc-TD T tits cnemicai ana physical fixation
DETRITAL F 12 - chemical and physical release
T CARBON
ACn = 21 Inputs - Y. outputs
Conceptual model of carbon pools and fluxes in a forest ecosystem. Pools are represented by boxes; fluxes
,nted by arrows.
terns of an ecosystem, management of nitrogen
levels is simple and inexpensive. We will review
wh at is currently known ab out nitrogen controls
on carbon cycling for temperate and boreal
forests.
EFFECT OF NITROGEN
FERTILIZATION ON BIOMASS
ACCUMULATION
Nitrogen fertilization may provide a means to
increase carbon accumulation in forest ecosys-
tems. Nitrogen fertilization has been shown to
increase above-ground tree volume up to 33%
infouryears (Tschaplinskietal., 1991;Bockheim
et al., 1985; Leaf et al., 1975) and to increase
understory biomass in forest ecosystems
(Turner, 1979; Reigel et al., 1991).
In addition, soil nitrogen levels often control
soil microbial activity, carbon turnover and
nutrient cycling (Melillo et al., 1982; Taylor et
al., 1989) thus affecting the balance and storage
of protected carbon and unprotected carbon of
below-ground forest ecosystems (Figure 1).
Nitrogen fertilization of forest soils usually
decreases microbial activity and organic matter
decomposition rates. Nitrogen immobilization
by soil microbes has been shown to be a primary
mechanism of nitrogen retention in forest eco-
systems (Zak, 1990; Entry et al., 1986; Vitousek
andMatson, 1984; Vitousek and Matson, 1985).
On nitrogen-deficient soils, however, nitrogen
fertilization has resulted in higher rates of or-
ganic matter decomposition (Cromack et al.,
1991; Hunt et al., 1988). This paper reviews the
effect of nitrogen fertilization on forest biomass
accumulation in forest ecosystems, including
effects on plant physiology, and on the below-
ground component of forest ecosystems.
Stands of Trees
The effectof nitrogen fertilization on the growth
of stands of forest trees has been intensively
studied (Gillespie and Chancy, 1989; Shumway
and Atkinson, 1978; Lea, 1980). Responses to
nitrogen fertilization will depend on nitrogen
availability, species requirements and specific
site characteristics.
Much of the biomass accumulation in forest
trees that occurs in response to nitrogen fertili-
zation can be attributed to increases in leaf area
(Tschaplinski et al., 1991; Troth et al., 1986;
Linder and Rook, 1984). Nitrogen fertilization
increased needle size, the number of needles
per shoot and the total number of shoots in
Pseudotsuga menziesii (Brix, 1981, 1983). Ef-
fects on leaf area and stem growth per unit of
156
-------
leaf area were evident 20-30 years after fertili-
zation of P. menziesii (Binkley and Reid, 1984).
Aronsson et al. (1977) found a 20% increase in
Pinus sylvestris with three needles per fascicle
rather than two needles per fascicle. Nitrogen
fertilization may also increase the photosyn-
thetic efficiency of plants (Gillespie and Chancy,
1989; Field and Mooney, 1984). Increases in
photosynthetic efficiency were associated with
increases in chlorophylls a and b and reduced
stomatal and mesophyll resistances to CO2
(Under and Troeng, 1980; Natr, 1975).
Nitrogen fertilization of forest trees also may
result in changes in carbon allocation. High
nitrogen concentrations in soils reduced the
rootrshoot biomass ratio of trees (Ahlstrom et
al., 1988; Ericsson, 1981). More than 50% of
the annual photosynthetic production can be
allocated to production of fine roots (Grier et
al., 1981; Agren et al., 1980). Alexander and
Fairley (1983) reported decreased fine root
production and increased root mortality after
nitrogen fertilization. Burke (1989) reported a
negative correlation between soil nitrogen avail-
ability and fine root carbohydrate concentra-
tions.
Understory Plants
Understory plants are in direct competition
with overstory trees for water, light and nutri-
ents. Reigel et al. (1991) found that nitrogen
fertilization increased Understory biomass in a
Pinus ponderosa forest by 17%. Wilson and
Tilman (1991) reported that nitrogen fertiliza-
tion shifted competition of grasses in a Minne
-sota old field from mainly below-ground to
both above- and below-ground. Tilman and
Wedin (199 Ib) found that nitrogen fertilization
changed the competition success among three
native grasses. According to resource competi-
tion theory, the species that can reduce mo-
noculture resource concentration to the lowest
level should eventually displace all other spe-
cies limited by that resource (Tilman, 1990).
JA. ENTRY, K.G. MATTSON, AND M.B. ADAMS
Nitrogen additions to temperate forests in-
creased the nitrogen concentration of the un-
derstory species (Heilman and Gessel, 1963;
Reigel et al., 1991). As a result of increasing
foliar biomass of the overstory trees, there may
be a decrease in light intensity to understory
plants (Turner and Long, 1975). Reigel et al.
(1991) found that nitrogen fertilization plus
irrigation had a synergistic effect on understory
biomass ofaP. ponderosa forest where light was
not limiting. Phumphery (1980) found that the
effects of nitrogen fertilization to the growth of
forest trees is greater when moisture is not
limiting. Therefore, nitrogen fertilization may
be of limited value if the local climate changes
to a more xeric state. Turner (1979) suggests
the following sequence of effects of nitrogen
fertilization on understory plants:
"In stands less than 10 years old, fertilization
usually results in increased growth of under-
story plants and thus increased competition
among species, with a subsequent reduction in
tree growth. When the trees have outgrown the
understory, light intensity reaching the ground
will decrease, producing understory plants with
higher nitrogen concentrations ultimately low-
ering species diversity on the site."
Effects of nitrogen fertilization of mature trees
will vary with the specific site and the tree
species. If trees respond with increased foliar
biomass, reduced light intensity to the under-
story and a reduction in total understory bio-
mass, mainly at the expense of the dominant
vascular plants, will likely occur.
EFFECT OF NITROGEN
ON PLANT PHYSIOLOGY
Photosynthesis
The greatest limitation to the maximum rate of
photosynthesis in C3 plants is the amount of
ribulose bisphosphate carboxylase. The maxi-
mum rate that carbon can be fixed for a C3 plant
157
-------
CARBON BALANCE IN FOREST ECOSYSTEMS
is promoted by a high amount of nitrogen and
carbon to the leaves (Fitter and Hay, 1987).
Nitrogen influences photosynthesis by affect-
ing the chlorophyll concentration and ribulose
bisphosphate carboxylase activity (Beardall et
al, 1991). Nitrogen, chlorophyll content and
photosynthate rates were significantly corre-
lated in pine and spruce seedlings supplied with
various nutrientregimes (KeUerand Wehrmann,
1963). Linder et al. (1981) reported a fivefold
increase in photosynthetic rates when nitrogen
concentration in leaves was increased. Photo-
synthesis in conifers and evergreen shrubs is
less responsive to variation in leaf nitrogen
concentration than deciduous trees (Field and
Mooney, 1984; Linder and Rook, 1984) be-
cause conifers lack the capacity for increasing
photosynthetic enzyme concentrations to the
same extent as deciduous trees, and their leaf
geometry makes CO2 diffusion processes more
likely to be limiting (Waring and Schlesinger,
1985). The effects of fertilizers are most pro-
nounced when other environmental f actors such
as light and temperature are most favorable for
maximum photosynthesis as demonstrated in
Populus spp. (Neuwirth and Fritzsche, 1964).
Finally, fertilization with nitrogen may increase
growth rate and produce a deficiency of other
nutrients, which may eventually result in de-
creased photosynthesis.
NUTRIENT USE EFFICIENCY
Nutrient use efficiency in forests is defined as
the net primary production per unit of nutrient
use by the vegetation (Vitousek, 1982). Annual
nutrient circulation in coniferous forests is much
lower than in deciduous forests, mostly as a
result of lower leaf turnover (Cole and Rapp,
1981). Leaching losses are also lower in conif-
erous forests (Parker, 1983). Cole and Rapp
(1981)foundgreaternutrientreabsorptionfrom
conifer needles than from deciduous leaves.
Nutrient use efficiency is lower in forests with
high soil nitrogen (Vitousek et al., 1982; Pastor
et al., 1984). Net primary productivity is posi-
tively correlated to annual N circulation in both
coniferous and deciduous forests (Cole and
Rapp, 1981). Ingestad (1979) reported that the
growth of nutrient deficient Betula verrucosa
seedlings increased in response to nitrogen
additions, but nutrient use efficiency decreased.
Miller et al. (1976) and Flanagan and Van
Cleave (1983) found a decline in nutrient use
efficiency in Pinus nigra stands and in taiga
ecosystems, respectively, after fertilization.
These analyses suggest that while increased
growth may result with nitrogen fertilization,
the magnitude of the effect will diminish with
increasing nitrogen availability.
INFLUENCE OF NITROGEN FERTILIZA-
TION ON BELOW-GROUND ECOSYSTEMS
Microbial Biomass
Microbial activity in forest soils is primarily a
function of moisture and temperature (Insam
et al., 1989; Donnelly et al., 1990). However,
microbial activity in forest soils usually de-
creases when nitrogen is added (Soderstrom et
al., 1983; Nohrstedt et al., 1989). Nohrstedt et
al. (1989) present two explanations for this
decrease in microbial activity:
1. There is a decreased amount of carbon input
to plant roots and thus to the rhizosphere and
thus decreased root biomass and mycorrhizal
colonization after nitrogen fertilization
(Alexander and Fairly, 1983; Vogt et al., 1985;
Ahlstrometal., 1988). Ectomycorrhizal coloni-
zation of coniferous trees is also often reduced
after nitrogen additions to soil (Myer, 1985).
2. Availability of soil carbon decreases in com-
bination with lignin being rate-limiting for
litter decomposition (Meentemyer, 1978). Re-
pression of ligninolytic enzymes by increased
soil nitrogen will decrease lignin degradation
and limit the rate of cellulose and hemicellu-
lose decomposition in woody materials
(Crawfordand Crawford, 1984; Kirk and Farrell,
158
-------
JA. ENTRY, K.G. MATTSON, AND M.B. ADAMS
1981,1987; Reid, 1979,1991).
A third explanation is that added nitrogen may
upset positive interactions between decompos-
ers (Fog, 1988). Response of microbial activity
and organic matter decomposition after nitro-
gen additions to forest ecosystems over many
years has not been investigated.
Organic Matter Decomposition
Classical thinking would suggest that adding
nitrogen to soils should increase decomposition
rates of soil organic matter by decreasing the
C:N ratio. In a review on the effects of nitrogen
on decomposition, Fog (1988) concluded that
added nitrogen generally has either no effect or
a negative effect on the rate of decomposition
of resistant organic matter. Recalcitrant car-
bon compounds, such as lignin, seem to show
the greatest suppressions in decay rate when
nitrogen is added. However, Gill and Lavender
(1983), Hunt et al. (1988), and Cromack et al.
(1991) found that nitrogen additions to forest
soils increased the rate of forest floor decompo-
sition. Taylor et al. (1989) found that the decay
rates of substances that contain a low concen-
tration of lignin can be predicted by the C:N
ratio; the decay rate of substances that have
high concentrations of lignin are more accu-
rately predicted by the lignin:N ratio.
Nitrogen-induced depression of decomposition
rate has been observed in boreal forests in
Sweden (Nohrstedt et al., 1989). In a study of
carbon cycling in Douglas-fir and western hem-
lock forests on the west side of the Cascade
Mountains in Oregon, Mattson (1991) observed
20-30% lower soil CO2 efflux on sites receiving
360 kg N/ha as ammonium forms when com-
pared with sites receiving no recent nitrogen
additions (Figure 2). In that study, it was not
determined whether the suppression of soil
CO2 efflux was a result of reduced root respira-
tion, reduced microbial activity, or both. How-
ever, the suppression of CO2 efflux demon-
E
o
4 -i
3 -
2 -
1 -
SOIL CARBON EFFLUX
SCIO FIELD SITE
1-4: Double
5-8: Single
9-12: None
1 £
10
Q.
-------
CARBON BALANCE IN FOREST ECOSYSTEMS
have been hypothesized to be causing forest
damage to German spruce forests by either
creating nutrientimbalances (Schulze, 1989) or
by acidifying soils (Ulrich, 1990). Similarly,
excess nitrogen inputs from the atmosphere
have been hypothesized to be "saturating" high-
elevation conifer forests in the northeastern
United States and exceeding the capacity for
uptake and retention by plants, soils and mi-
crobes (Aber et al., 1989).
Long-term inputs of nitrogeninto forest soils by
Alnus rubra have been shown to decrease pH
and base saturation (Binkley and Sollins, 1990).
Coleetal. (1991) not only reported that A. rubra
growing in P. menziesii stands decreased soil
pH and lead to the displacement of base cations
on exchange sites, decreased available phos-
phorus, increased mobilization of aluminum
into solution and declining aider productivity
on the sites. Stone et al. (1991) reported that 20
years of nitrogen fertilization resulted in de-
creased soil pH, decreased availability of phos-
phorus, calcium, magnesium and sodium, but
increased availability of iron, copper and man-
ganese. Moreover, not all nitrogen is absorbed
by the ecosystem. Fertilization has been re-
ported to increase the nitrogen concentrations
of stream water (Hetherington, 1985; Meehan
et al., 1975). In nitrogen-poor ecosystems,
increased nitrogen input to streams may pro-
videbeneficial effects (Stockner and Shortreed,
1978). The type and degree of change to forest
soils that may be caused by excess nitrogen
inputs are largely unquantified as of yet. Quan-
titative relationships would certainly be site-
specific, depending on the availability of nitro-
gen and the system's capacity for uptake.
EFFECTOFNITROGEN FERTILIZATION
ON INSECT AND DISEASE ATTACK
ON FOREST TREES
Although few experiments have been conducted
to determine the influence of fertilization on
forest trees and their susceptibility to insects or
diseases, some research exists. Dendroctonus
ponderosae has evolved a group attack behavior
that maximizes fitness of individual Pinus
contorta (Berryman et al., 1985). Waring and
Pitman (1985) fertilized 120 year-old nitrogen
deficient stands of P. contorta. Treatments
were: 1) 420 kg N fertilizer/ha; 2) 420 kg N/ha
plus 80% canopy reduction; 3) 2,000 plus kg
sawdust/ha (to immobilize soil nitrogen); and
4) untreated control plots. Uniform dispersal
of D. ponderosae throughout treatments was
achieved with equal amounts of pheromones.
Water availability to trees did not vary among
treatments. After two years, the trees in both
fertilized treatments had increased wood pro-
duction/leaf area/year and were able to sur-
vive all D. ponderosae attacks. Dendroctonus
ponderosae continued the high rate of attacks
on stands receiving the sugar + sawdust treat-
ment and on the untreated control stands; thus
high rates of tree mortality continued in these
treatments (Waring and Pitman, 1985). Casotti
and Bradley (1991) found that insect herbivory
was positively correlated with leaf nitrogen
concentration in an Australian eucalypt forest.
Grasshopper (Melanoplus differentialis) con-
sumption of Artemisia tridentata var vaseyana
was reduced when soil nitrogen concentrations
were limiting (Johnson and Lincoln, 1991).
Entry et al. (1991b) inoculated P. menziesii
trees hi northern Idaho wiihArmillaria ostoyae.
The treatments were: 1) thinning of P. menziesii
stands to 5 x 5 m spacing; 2) thinning of P.
menziesii stands to 5 x 5 m spacing and fertiliza-
tion with 360 kg NH4/ha; and 3) an untreated
control. Armillaria ostoyae is a primary patho-
gen, but is known to attack trees under stress
(Entry et al., 199 la). Stands that were thinned
and fertilized had a higher A. ostoyae infection
rate than stands that were thinned or that were
untreated. Pseudotsuga menziesii are highly
susceptible to Armillaria spp. (Hadfield et al.,
1986). Therefore, fertilization of trees that are
less susceptible to Armillaria spp., such asLarix
occidentalis or Thuja plicata, may produce dif-
160
-------
ferent results (Entry et al., 1992).
Phellinus weirii attack has been responsible for
increasing the concentration of soil nitrogen
(Cromack et al., 1991). In a nitrogen-deficient
subalpine T. mertensiana forest 220-year-old
trees were killed by an advancing wave of P.
weirii. As the old growth trees were killed,
regenerating trees replaced them, forming a
younger regrowth forest. Nitrogen availability
and nitrogen mineralization were higher in the
regrowth forest (Matson and Boone, 1984).
The amount of carbon sequestered in a forest
may be substantially reduced by insect and/or
disease attacks. Under endemic conditions,
insects may not reduce the growth of forests, but
may influence only the dominant tree species.
Under epidemic conditions, insects and dis-
eases have been known to drastically reduce the
total amount of biomass on a site. Although
there is an abundance of information available
on the effects of fertilization or the growth of
tree stems, little is known about the effects of
fertilization on tree physiology and tree re-
sponse to insect and disease attack. Further
experimentation is necessary to determine the
influence of fertilization on tree physiology and
the response of the numerous forest pathogens.
IMBALANCES FROM NITROGEN
FERTILIZATION
Nitrogen fertilization will most likely increase
biomass of a site, especially if nitrogen is limit-
ing in that ecosystem. Trees should respond
with increased leaf area and stem growth per
unit leaf area, resulting in reduced light inten-
sity to understory plants, which may in turn
change competition success of understory plants.
Nutrient-use efficiency of plants should de-
cline. Vegetation, especially trees, may shift a
greater proportion of carbon allocation
aboveground. Excess nitrogen may displace
base cations from soil exchange sites and base
XA. ENTRY, K.G. MATTSON, AND MB. ADAMS
cations such as calcium or magnesium may
become limiting. Soil microbial activity may
increase or decrease, depending on initial ni-
trogen concentration of the soil. The amount of
mycorrhizal activity in the soil is usually pre-
dicted to decrease. Organic matter decompo-
sitionrates may remain unchanged or decrease.
Fertilized trees may be more or less susceptible
to insect and/or disease attack, depending on
tree species, the insect or disease and the initial
soil nitrogen concentration. There is little
information available on the long-term impact
of nitrogen fertilization on terrestrial ecosys-
tems. The full effects of nitrogen fertilization
on the amount of carbon sequestered by forest
ecosystems cannot be adequately assessed at
this time.
CONCLUSIONS
Forest fertilization with nitrogen may be an
important strategy to accrete and store atmo-
spheric carbon to mitigate global climate change.
Response of forest ecosystems to nitrogen fer-
tilization will vary with the specific ecosystem
characteristics and the type and amount of
fertilizer. Fertilization can substantially in-
crease the amount of carbon accumulated in
nutrient-deficient terrestrial ecosystems, at least
in the short term, through increases in tree
growth and possibly through suppression of soil
CO2 effluxes. Although the effects of forest
fertilization on the growth of forest trees and
stands have been well studied, the effects on
forest ecosystems such as decomposition and
soil carbon storage are not clearly understood.
The prospects for controlling insect and disease
damage are largely unknown. Long-term stud-
ies of ecosystem response and tree physiology
are necessary. Fertilization is a manipulation
that will most likely produce physiological and
ecological imbalances along with increasing
carbon storage. Particular attention should be
paid to quantifying such relationships and to
determining the mechanisms of control.
161
-------
CARBON BALANCE IN FOREST ECOSYSTEMS
REFERENCES
Aber, J.D., KJ. Nadelhoffer, P. Stevdler, and J.M. Melillo.
1989. Nitrogen saturation in northern ecosystems. Bio-
science 39:378-386.
Agren, G.T., B. Axelsson, J.G.K. Flower-Ellis, S. Linder,
H.Pcrsson,H.Staaf,andE.Troeng. 1980. Annual carbon
budget for young Scots pine. In: Structure and Function
of Northern Coniferous Forests - An Ecosystem Study,
edited by T.Persson. Ecological Bulletin 32:307-313.
Ahlstrom, K, H. Persson, and I. Borjesson. 1988.
Fertilization in a mature Scots pine (Pinus sylvestris L.)
stand: effects on fine roots. Plant and Soil 106:179-190.
Alexander, I J. and R.I. Fairley. 1983. Effects of N
fertilization on: population of roots and mycorrhizas in
spruce humus. Plant and Soil 71:49-53.
Aronsson,A.,S.Elowson,and T.Ingestad. 1977. Elimi-
nation of water and mineral nutrition as limiting factors in
a young Scots pine stand. I. Experimental design and
some preliminaryresults. Swedish Conference Forest
Project. Technical Report 10. Uppsala.
Barclay, HJ. and H.Bria. 1984. Effects of urea and nitrate
fertilizer on growth of a young thinned and unthinned
Douglas-fir stand. Canadian Journal of Forest Research
14:952-955.
BcardalLJ.B.,S.Roberts,andJ.Millhouse. 1991. Effects
of nitrogen limitation on uptake of inorganic carbon and
specific activity of ribulose -1, 5-bisphosphate carboxy-
lase/oxygenase in green microalgae. Canadian Journal of
Botany 67:1146-1150.
Berg, B.,B. Wesson, and G.Ekbohm. 1982. Nitrogen level
and decomposition of Scots pine needle litter. Okios
38:291-296.
Berryman,AA.,B.Dennis,KF.Raffa,andN.C.Stenseth.
1985. Evolution of optimal group attack with particular
reference to bark beetles (Coleoptera:Scofytidae). Ecol-
ogy 66:898-903.
Binkley, D. and P. Reid. 1984. Long-term responses of
stem growth and leaf area to thinning and thinning and
fertilizationinaDouglas-firplantation. Canadian Journal
of Forest Research 14:656-660.
Binkley, D. and P. Sollins. 1990. Factors determining
differences in soil pH in adj acent conifer and alder-conifer
stands. Soil Science Society of America Journal 54:1427-
1433.
Bockheim, J.G., J.E. Leide, and D.S. Tavella. 1985.
Distribution and cycling of macronutrients in a Pinus
resinosa plantation fertilized with nitrogen and potassium.
Canadian Journal of Forest Research 16:778-785.
Brix, H. 1983. Effects of thinning and nitrogen fertiliza-
tion on growth of Douglas-fir: relative contribution of
foliage quantity and efficiency. Canadian Journal of
Forest Research 13:167-175.
Brix, H. 1981. Effects of nitrogen fertilizer source and
application rates on foliar nitrogen concentration photo-
synthesis and growth of Douglas-fir. Canadian Journal of
Forest Research 11:775-780.
Burke, M.K. 1989. Fine root production and turnover in
a northern hardwood forest and the influence of nitrogen
availability. Ph.D. dissertation, State University of New
York, College of Environmental Science and Forestry,
Syracuse, New York, p. 180.
Casotti, G. and J.S. Bradley. 1991. Leaf nitrogen and its
effects on the rate of herbivory on selected eucalypts in the
jarrah forest. Forest Ecology and Management 41:167-
177.
Cole, D.W. and M. Rapp. 1981. Elemental cycling in
forest ecosystems. In: Dynamic Principals of Forest
Ecosystems, edited by D.E. Reichle. Cambridge Univer-
sity Press, London, pp. 341-409.
Cole, D.W., J. Compton, H. VanMiegroet, andP. Hemann.
1991. Changes in soil properties and site productivity
caused by red alder. Water, Air and Soil Pollution 54:231-
246.
Crawford, R.L. and D.L. Crawford. 1984. Recent ad-
vances in studies of the mechanisms of microbial degrada-
tion of lignins. Enzyme and Microbial Technology 6:433-
442'.
Cromack, K. Jr., JA. Entry, and T. Savage. 1991. The
effect of disturbance byPhellinus weirii on decomposition
and nutrient mineralization in a Tsuga mertensiana forest.
Biology and Fertility of Soils. In press.
162
-------
JA. ENTRY, K.G. MATTSON, AND M.B. ADAMS
Donnelly, P.K., J A. Entiy,D.L. Crawford, andK. Cromack,
Jr. 1990. Cellulose and lignin degradation in forest soils:
response to moisture, temperature and acidity. Microbial
Ecology 20:289-295.
Entry, JA., N.M. Stark, and H. Loewenstein. 1986. The
effects of timber harvesting onmicrobialbiomass fluxes in
a northern Rocky Mountain Forest Soil. Canadian Jour-
nal of Forest Research. 16:1076-1081.
Entry, JA., K. Cromack, Jr., E.M. Hansen, and R.H.
Waring. 1992. Response of western conifer ous seedlings
to infection by Armillaria ostoyae under limited light and
nitrogen. Phytopathology 81:89-94.
Entry, JA., K. Cromack, Jr., R.C. Kelsey, and N.E.
Martin. 1991b. Response of Douglas-fir to infection by
Armillaria ostoyae after thinning or thinning plus f ertiliza-
tion. Phytopathology 81:682-689.
Entry, JA., N.E. Martin, R.E. Kelsey, and K. Cromack,
Jr. 1991c. Response of five species of western conifer
saplings to infection by Armillaria ostoyae. Phytopathol-
ogy 82:393-397
Fenn, M. 1991. Increased site fertility and litter decom-
position rate in high-pollution sites in the San Bernardino
Mountains. Forest Science 37:1163-1181.
Field, C. and HA. Mooney. 1984. The photosynthesis-
nitrogen relationship in wild plants. In: On the economy
of plant form and function, edited by T.J. Givnish. Cam-
bridge Univ. Press, pp. 25-55
Fitter, A.M. and R.K.M Hay. 1987. Environmental
Physiology of Plants. Academic Press, New York, p. 421.
Flanagan, P.W. and K. Van Cleve. 1983. Nutrient cycling
in relation to decomposition and organic matter quality in
a taiga ecosystem. Canadian Journal of Forest Research
13:795-817.
Fog, K. 1988. The effect of nitrogen on the rate of
decomposition of organic matter. Biological Review
63:433-562.
Gill, R.S. and D.P. Lavender. 1983. Litter decomposition
in coastal hemlock stands: impact of nitrogen on decay
rates. Canadian Journal of Forest Research 13:116-121.
Gillespie,A.R.andW.R.Chaney. 1989. Process modeling
of nitrogen effects on carbon assimilation and allocation
- a review. Tree Physiology 5:99-112.
Grier, C.C., KA. Vogt, M.R. Keyes, and R.L. Edmonds.
1981. Biomass distribution and above- and below-ground
production in young and mature Abies amdbilis zone
ecosystems of the Washington Cascades. Canadian Jour-
nal of Forest Research 11:155-167.
Hadfield, J.S., DJ. Goheen, G.M. Filip, C.L. Schmitt, and
R.D. Harvey. 1986. Root diseases in Oregon and Wash-
ington conifers. USDA Forest Service, Pacific Northwest
Region, Forest Pest Management Report R6-FPM-250-
86,27 p.
Heilman, P.E. and S.P. Gessel. 1963. Nitrogen require-
ment and the biological cycling of nitrogen in Douglas-fir
stands in relationship to the effects of nitrogen fertiliza-
tion. Plant and Soil 18:386-402.
Hetherington, E.D. 1985. Streamflow nitrogen loss
following forest fertilization in a southern Vancouver
Island watershed. Canadian Journal of Forest Research
15:34-41.
Hunt, H.W., E.R. Ingham, D.C. Coleman, E.T. Elliott,
and C.P.P. Reid. 1988. Nitrogen limitation of production
and decomposition in a prairie, mountain meadow, and
pine forest. Ecology 69:1009-1016.
Ingestad,T. 1979. Nitrogen stress in birch seedlings EN,
K,P,Ca, and Mg nutrition. PhysiologiaPlantarum. 50:373-
380.
Insam, I., D. Parkinson, and K.H. Domsch. 1989. Influ-
ence of macroclimate on soil microbial biomass. Soil
Biology and Biochemistry.
Johnson, R.H. andD.E. Lincoln. 1991. Sagebrush carbon
allocation patterns and grasshopper nutrition: the influ-
ence of CO2 enrichment and soil mineral limitation.
Oecologia 87:127-134.
Keller, T. and J. Wehrmann. 1963. CO2 Assimilation,
Wurzelatmung und Ertrag con Fichen-und
Kefernsamlingen be unterschiedlicher
Mineralstoffenahrung Mitt. Schwerz. Arst. Forstl
Versuchswes 39:217-242.
King, A.W., R.V. O'Neill, and D.L. DeAngelis. 1989.
Using ecosystem models to predict regional CO2 ex-
change between the atmosphere and the terrestrial bio-
sphere. Global Biogeochemical Cycles 3:337-361.
163
-------
CARBON BALANCE IN FOREST ECOSYSTEMS
Kirk,T.K.andR.LFarrell. 1987. Enzymatic Combustion:
The microbial degradation of lignin. Annual Review of
Microbiology 41:465-505.
Lca,R. 1980. Prediction models for N and P fertilization
of loblolly pine plantations. North Carolina Forest Fertili-
zation Cooperative Report No. 9.
Leaf, A.L., R.E, Lenoard, R.F. Wittmer, and D.H.
Bickelhaupt. 1975. Four year growth responses of plan-
tations of red pine to potash fertilization and irrigation in
New York. Forest Science 21:88-96.
Lindcr, S. and DA. Rook. 1984. Mineral nutrition and
CO2. In: Nutrition of Plantation Forests, edited by G.O.
Bowen and E.K.S. Nambiar. Academic Press, London.
pp. 211-236.
Linder, S. and DA. Rook. 1984. Effects of mineral
nutrition on carbon dioxide exchange and partitioning of
carbon hi trees. In: Nutrition of Plantation Forests, edited
by G.D. Bowen and E.K.S. Nambiar, Academic Press,
New York, p. 211-236.
Linder, S.,J. McDonald, and T.Lohammer. 1981. Effect
of nitrogen status and irradiance during cultivation on
photosynthesis and respiration in birch seedlings. Energy
Forestry Project, Swedish University of Agricultural Sci-
ences. Uppsala.
Linder, S. and E. Troeng. 1980. Photosynthesis and
transpiration of 20 year-old Scots pine. In: Structure and
Function of Northern Coniferous Forests - An Ecosystem
Study, edited by T.Persson. Ecological Bulletin 32:165-
181.
Mattson, K.G. 1991. CO2 effluxes at Scio. Report for
Annual OTTER Meetings. Presented at the AMES Re-
search Center, Moffett Field, CA. Jan 16-18,1991.
Meentemyer, V. 1978. MacrocUmate and lignin control
of litter decomposition rates. Ecology 59:465-492.
Meehan, W.R., F.B. Lotsperch, and E.W. Mueller. 1975.
Effects of forest fertilization on two southeast Alaska
streams. Journal of Environmental Quality 4:50-55.
Melillo,J.M.,J.D.Aber,andJ.F.Muratore. 1982. Nitro-
gen and lignin control in hardwood leaf litter decomposi-
tion dynamics. Ecology 59:465-472.
Miller, H.G., JM. Cooper, and J.D. Miller. 1976. Effect
of nitrogen supply on nutrients in litterfall and crown
leaching in a stand of corsican pine. Journal of Applied
Ecology 13:233-248.
Mooney, HA., P.M. Vitousek, and PA. Matson. 1987.
Exchange of materials between terrestrial ecosystems and
the atmosphere. Science 238:926-932.
Myer,F.H. 1985. Einfluss des Stickstoff-Faktors auf den
Mycorrhizabesatz von Fichtensamlingen im Humus einer
Waldschadenflache Allg. Forstztg 9:208-209.
Natr,L. 1975. Influence of mineral nutrition on photosyn-
thesis and the use of assimilates. In: Photosynthesis and
Productivity in Different Environments,edited by J.P.
Cooper. Cambridge University Press, London, pp. 537-
555.
Neuwirth, G. and K.H. Fritzsche. 1964. Untersuchungen
iiber den einfluss verschiedener Diingergaben auf das
gasstoffwechselokologjscheverhalteneinjahriger. Pappel-
Steckholzaufwiichse Arch. Forstwes 13:233-246.
Nohrstedt, H.O., K. Arnebrant, E. Baath, and B.
Soderstrom. 1989. Changes in carbon content, respiration
rate, ATP content and microbial biomass in nitrogen-
fertilized pine forest soils in Sweden. Canadian Journal of
Forest Research 19:323-328.
Pastor, J., J.D. Aber, CA. McClaughtergy, and J.M.
Melillo. 1984. Aboveground production and N and P
cycling along a nitrogen mineralization gradient on
Blackhawk Island, Wisconsin. Ecology 65:256-268.
Paul, E A. and F.E. Clark. 1989. Soil Microbiology and
Biochemistry. Academic Press, Inc., San Diego.
Phumphery, F.V. 1980. Precipitation, temperature and
herbage relationships for a pine woodland site in north-
eastern Oregon. Journal of Range Management 33:307-
310.
Reid,I.D. 1991. Nutritional regulation of synthetic lignin
(DHP) degradation by Phlebia (Merulius) tremellosa:
effects of nitrogen. Canadian Journal of Botany 69:156-
160.
Reid, I.D. 1979. The influence of nutrient balance on
lignin degradation by the white rot fungus Phanerochaete
chrysosporium. Canadian Journal of Botany 57:2050-
2058.
164
-------
J.A. ENTRY, K.G. MATTSON, AND M.B. ADAMS
Reigel, G.M., R.F. Miller, and W.C. Kruger. 1991. Un-
derstory vegetation response to increasing water and
nitrogenlevelsinaP/nu,yp0nderaya forest in northeastern
Oregon. Northwest Science 65:10-15.
Schulze, E.D. 1989. Air pollution and forest decline in a
spruce (Picea abies) forest. Science 244:776-783.
Shumway,J. and W A. Atkinson. 1978. Predicting nitro-
gen fertilizer response in unthinned stands of Douglas-fir.
Communication: A SoilScience and Plant Analysis 9:529-
539.
Soderstrom, B., E. Baath, and B. Lundgren. 1983.
Decrease in soil microbial activity and biomass owing to
nitrogen amendments. Canadian Journal of Microbiology
29:1500-1506.
Stockner,J.G.andK.R.S.Shortreed. 1978. Enhancement
of autotrophic production by nutrient addition in a coastal
rain forest on Vancouver Island. Journal of the Fisheries
Research Board of Canada 35:28-34.
Stone, D.L.R., DA. Whitney, KA. Janssen, and J.H.
Long. 1991. Soil properties after twenty years of fertiliza-
tion with different nitrogen sources. Soil Science Society
of America Journal 55:1097-1100.
Taylor, B.R., D. Parkinson, and W.F. Parsons. 1989.
Nitrogen and lignin content as predictors of litter decay
rates: a microcosm test. Ecology 70:97-104.
Tilman,D.andD.Wedin. 1991a. Plant trails and resource
reduction for five grasses growing on a nitrogen gradient.
Ecology 72:683-698.
Tilman, D. and D. Wedin. 1991b. Dynamics of nitrogen
competition between successional grasses. Ecology
72:1038-1049.
Tilman, D. 1990. Mechanisms for plant competition for
nutrients: elements of a predictive theory of competition.
In: Perspectives in Plant Competition, edited by J. Grace
and D. Tilman. Academic Press, New York> U.S A., pp.
117-141.
Troth, J.L., R.G. Cambell, and H.L. Allen. 1986. Nutri-
ents: use of forest fertilization and nutrient efficient
ecotypes to manage nutrient stress in conifer stands. In:
Stress Physiology and Forest Productivity, edited by T.C.
Hennessy, P.M. Dougherty, S.V. Kossuth, and D.J. John-
son. Martinus Nijhoff Publishers, Dordrecht, pp. 61-99.
Tschaplinski, T.J., D.W. Johnson, R.J. Norby, and D.E.
Todd. 1991. Biomass and soil nitrogen relationships of a
one year-old sycamore plantation. Soil Science Society of
America Journal 55:841-847.
Turner, J. 1979. Effects of fertilization on understory
vegetation. In: Forest Fertilization Conference, edited by
S.P. Gessel, R.M. Kenady and W A. Atkinson. University
of Washington Institute of Resources Contribution No.
40, Seattle, WA.
Turner, J. and J.N. Long. 1975. Accumulation of organic
matter in a series of Douglas-fir stands,, Canadian Journal
of Forest Research 5:681-690.
Ulrich, B. 1990. Waldsterben: forest decline in West
Germany. Environmental Science Technology 24:436-
441.
Van Miegroet, H., D.W. Johnson, and D.W. Cole. 1990.
Soil nitrification as affected by N fertility changes in forest
floor C:N ratio in four forest soils. Canadian Journal of
Forest Research 20:1012-1019.
Vitousek, P.M. 1982. Nutrient cycling and nutrient use
efficiency. American Naturalist 119:553-572.
Vitousek, P.M., J.R. Gosz, C.C. Grier, J.M. Melillo, and
W.A. Reiners. 1982. A comparative analysis of potential
nitrification and nitrate mobility in forest ecosystems.
Ecological Monographs 52:155-157.
Vogt, KA., D.J. Vogt, E.E. Moore, W. Littke,C.C. Grier,
and L. Leney. 1985. Estimating Douglas-fir fine root
biomass and production from bark and starch. Canadian
Journal of Forest Research 15:177-179.
Waring, R.H. and W.H. Schlesinger. 1985. Forest ecosys-
tems: concepts and management. Academic Press. New
York. p. 340.
Waring, R.H. and G.B. Pitman. 1985. Modifying lodge-
pole pine stands to change susceptibility to mountain pine
beetle attack. Ecology 66:889-897.
Williams, W.E., K. Garbutt, FA. Bazzaz, and P.M. Vi-
tousek. 1986. The response of plants to elevated CO2. IV.
Twodeciduous-foresttreecommunities. Oecologja 69:454-
459.
Wilson, S.D. and D. Tilman. 1991. Components of plant
competition along an experimental gradient of nitrogen
availability. Ecology 72:1050-1065.
165
-------
-------
CONSEQUENCES OF TREE MORTALITY
TO THE GLOBAL CARBON CYCLE
Mark E. Harmon, Sandra Brown, and Stith T. Gower
ABSTRACT
Tree mortality and accumulations of woody detritus are important, but unstudied facets of the global carbon (C) cycle.
Ecological studies from undisturbed temperate and tropical forests indicate woody litter inputs associated with tree mortality
range between 0.16-5.0 Mg C/ha/yr. Mortality rates in forests appear to be positively correlated with ecosystem productivity
and time since disturbance. United States Forest Service Continuous Forest Inventory data indicate that tree mortality in
the United States increased over the past five decades from 52 to 65 Tg C/yr. However, mortality is increasingin some regions
(eastern United States), while decreasing in others (Pacific Northwest) because of changes in forest-age structure. Globally,
there are few measurements of tree mortality. Stand and Continuous Forest Inventory data were used to estimate turnover
rates of live woody biomass for forests. Applying these ratios to current live biomass estimates indicates that 1.2-9.1 Pg C/
yr is added globally to detrital pools by dyingtrees within intact forests. An additional 0.9-1.8 Pg of woody Cmay be added
to the detrital cycle by catastrophic disturbances. The combined dead tree input is 7-39% of the 28 Pg C/yr added by fine
litter/all. At these rates of input, coarse woody debris would reach a steady-state mass of 60-232 Pg C. However, because
past land conversions reduced mortality and woody detritus, considerable carbon could be sequestered in forests recovering
from clearing.
INTRODUCTION
Examination of any current global carbon bud-
get (Post et al., 1990; Houghton and Woodwell,
1989) reveals a very startling point: dead trees
do not exist! The tacit assumption in most, if not
all, global carbonbudgets is that although > 80%
of theglobe'slivingbiomassiswoody (Woodwell
et al., 1978), very little of this material becomes
litter. We will demonstrate that this assump-
tion is unfounded and has led to a major under-
estimate of global detrital stores.
Excluding dead trees has several profound im-
pacts upon understanding global carbon dy-
namics. First, the size of the terrestrial detrital
pool has been underestimated (Harmon and
Chen Hua, 1991). If woody detritus was in
equilibrium, this might not be a major concern;
however, land use changes have placed woody
detritus in disequilibrium (Schiffman and John-
son, 1989). For example, logging within Pacific
Northwest forests has halved carbon storage in
this region (Harmon et al., 1990). Fully one-
third of the difference between carbon stores in
a plantation forest and an old-growth forest is
due to the reduction of woody detritus (dead
tree) carbon (Harmon etal., 1990). This reduc-
tion is not unique to this region, but probably
applies to most boreal, temperate and tropical
forests. If this is true, past calculations of
carbon flux from forest clearing are underesti-
mates. Conversely, this would imply that a
large, unaccounted for carbon sink could be
occurring in forests recovering from past har-
167
-------
CONSEQUENCES OF TREE MORTALITY TO THE GLOBAL CARBON CYCLE
vest (Harmon and Chen Hua, 1991; Lugo and
Brown, in press).
Another problem with excluding dead trees
from global carbon budgets involves the future
pattern of detrital stores. Most assessments of
climate change upon detrital stores are com-
parisons of steady-state solutions under current
and projected future climates. These studies
generally indicate a greater storage of carbon
under a warmer climate (Chapin and Matthews,
1992), primarily as aresult of higher productivity.
In contrast, analysis of transient responses to
climate change indicates alarge pulse of carbon
may be injected into the atmosphere during the
transition between these two steady-states
(Neilson, personal communication). To alarge
degree, the temporal dynamics of this transient
pulse willbecontrolledbythedecomposition of
woodyplantskiUedbycatastropMcdisturbances
(i.e., fire), or increased stress.
In the following paper, we will examine the role
dead trees play in the current global carbon
cycle. Coarse and fine woody detritus are
formed from the death of twigs, branches, roots
and boles of trees and other woody plants. The
diameter used to separate coarse- and fine-
fraction woody detritus is 10 cm. For the
purposes of this review, we will exclude fine
root turnover from our analysis, although the
fine roots of many trees and shrubs are in fact
woody and can exceed lead litterfall in some
forest ecosystems (Vogt et al., 1986). We first
review the characteristics of woody detritus that
make its behavior different from other forms of
litter. We then examine factors controlling tree
mortality at the stand and regional levels. Fi-
nally, we estimate the input of carbon via dying
trees to the global detrital system and examine
its implications for the overall global carbon
cycle.
CHARACTERISTICS OF WOODY
DETRITUS
The decay and input dynamics of woody detri-
tus are quite distinct from those of leafy detri-
tus. Decay rates of woody detritus can be more
than an order of magnitude less than leafy litter
(Harmon et al., 1986). This has several impli-
cations for successional dynamics and C stores.
First, fine litter and woody litter will store equal
amounts of carbon at steady-state if woody
inputs are an order of magnitude lower than
fine litter inputs. These differences in decay
rate also imply woody litter will take more than
ten times as long to reach steady-state once
input rates have stabilized.
In addition to differences in decay rates, the
period required to reach maximum input rates
following disturbance is considerably longer for
woody than leafy litter. Input rates for leafy
litter usually peak at the time of crown closure,
which is as little as five years in tropical and
deciduous forests (Bormann and Likens, 1979;
Brown and Lugo, 1990) and up to 50 years in
coniferous forests (Gessel and Turner, 1976).
In contrast, woody litter inputs do not peak until
the old-growth stage of succession, and this may
take > 100 years even in tropical forests (Brown
and Lugo, unpublished). This implies that
stores of woody detritus will not reach steady-
state levels for at least a century following
disturbance.
Despite the fact that disturbances, including
logging, create a large quantity of woody de-
tritus, it is not unusual for this component to be
overlooked in successional or ecosystem pro-
cess models or carbon-sequestration assess-
ments of climate change. We suggest that the
conservation of mass applies even to dead trees.
In the case of natural catastrophic disturbances
(i.e., wind and fire), woody detritus can be the
largest single carbon pool immediately follow-
ing disturbance. In a Douglas-fir forest, for
example, woody detritus comprised 25% of the
total ecosystem carbon before a crown fire and
75% afterward (Agee and Huff, 1987). For
most disturbances, this large increase is obvi-
168
-------
M.E. HARMON, S. BROWN, AND S.T. GOWER
Ponderosa pine
Lodgepole pine
Korean pine
Red spruce-Fir
Western White pine
Virginia pine
Silver-fir
Douglas-fir
Spruce Hemlock
White fir
0
•i MEAN INPUT (Mg/ha/y)
studies with > 5 ha-years of observation
Figure 1. Woody detritus input associated with tree death in conifer forests. Only mature to old-growth forest with
at least five hectare-years of observation are displayed.
ously a result of the large mass in trees being
suddenly transferred to the woody detrital pool.
However, in the case of fires, woody detritus is
less apt to burn than leafy litter. This is es-
pecially true for the coarse- fraction material,
which even under the most severe fire condi-
tions is rarely burned completely (Sandberg
and Ottmar, 1983).
STAND LEVEL STUDIES
Ecologists and foresters have been measuring
forest growth for centuries; however, there are
very few published mortality measurements
based on direct observation (Franklin et al.,
1987). The most reliable data are from large
permanent plots ( > 1 ha area) with individually
marked trees observed for at least five years.
Although numerous data sets meeting these
criteria exist, many of them remain unpub-
lished or analyzed solely from a population
perspective. It is hoped that this trend will not
continue.
Although it is unlikely stand level studies will
ever be numerous enough to estimate global
mortality rates directly, they are quite useful in
understanding the causes of mortality and how
stand structure, age, and productivity affect
rates. Perhaps the best set of stand level plots
available at this time is from the Pacific North-
west, a region where large-scale (>0.5 ha)
permanent plots have been observed for more
than 20 years (Franklin et al., 1987). Woody
detritus inputs (including all tree parts except
leaves) associated with tree death range an
order of magnitude from < 0.16 Mg C ha/yr in
ponderosa pine to >4 Mg C ha/yr in white fir
forests of California (Figure 1). IncontrasVfine
litterfall in these stands only differs by a factor
of two from 1-2 Mg C/ha/yr (Vogt et al., 1986).
Both fine litterfall and tree mortality in mature
to old-growth conifer forests appears positively
correlated with site quality. Plotting total above-
ground NPP (net primary production, equal to
the sum of biomass increment, mortality and
fine litterfall) against tree mortality for mature
to old-growth forests indicates that there is a
general correspondence (Figure 2). For a given
amount of above ground NPP, however, conifer
forests produce far more woody detritus than
deciduous or tropical forests. In particular,
Pacific Northwest conifer forests produce al-
most three times the "expected" amount of tree
169
-------
CONSEQUENCES OF TREE MORTALITY TO THE GLOBAL CARBON CYCLE
10
8
6
4
2
0
(
Figure 2. Woody
The NPP was calc
FREE MORTALITY (Mg/ha/year)
0
0
0
D
0
A 6
0 , g i
i .. I 1
) 5 10 15 20 25 30
ABOVE-GROUND NPP (Mg/ha/year)
0 CONIFER A MIXED x DECIDUOUS ° TROPICAL
detritus input associated with tree death versus above-ground Net Primary Production (NPP).
ulated as the sum of biomass increment, litterfall and tree mortality.
mortality litter. This difference may be attrib-
uted to, in part, from the greater longevity, and
thus biomass, of these forests; a greater amount
of biomass may therefore offset lower NPP.
The lower mortality rate in less productive
forests does not necessarily mean that dead
woody detrital stores will be lower in these
forests. Dead wood stores in tropical forests
might be quite similar to those in boreal forests
because productivity and decomposition rates
are positively correlated onaglobal scale-both
increasing toward the tropics. There are excep-
tions to this pattern as some temperate regions
(i.e., the Pacific Northwest) can also have high
productivity. Woody decomposition rates are
strongly influenced by fungi-toxic compounds
inheartwood and excess moisture, factors that
may not be correlated to latitude. Few boreal
genera contarnnragi-toxiccompounds, whereas
they are extremely common in tropical hard-
woods. Moreover, excess moisture may limit
wood decay in some tropical forests, but it is
frequent in cool humid regions such as the
Pacific Northwest.
Another factor that profoundly influences the
amount of tree mortality is stand age (Figure 3).
Although the number of stems dying is often
highest during the middle stages of succession
(Knox et al., 1989), it would appear that woody
litter input increases with forest age (Harcombe
et al., 1990). This discrepancy may result from
differences of population and detrital input
dynamics. The large number of stems dying
during the middle stages of succession are small
individuals with little mass. In contrast, fewer
trees die in older forests, but they are much
larger and not biased toward the smallest indi-
viduals of the population (Knox et al., 1989).
The non-linear patterns of tree mortality inputs
during succession, coupled with slow decompo-
sition, mean that woody detrital stores rarely
reach a steady-state. Disturbances add a large
amount of woody detritus, yet the following
young forests do not produce much woody
detritus. This leads to alarge decrease in woody
detrital stores early in succession that may
equal or exceed the carbon sink created by
regrowing trees (Harmon et al., 1990). If mor-
tality rates remain low during the middle stages
170
-------
M.E. HARMON, S. BROWN, AND S.T. GOWER
Mortality (Mg/ha/year)
0 20 40
after Harcombe et al. (1990)
60 80
Time (years)
mean
- 1 std
100
+ 1 std
120
140
Figure 3. Changes in tree mortality as a function of stand age for a western hemlock/sitka spruce ecosystem. Two series
of permanent plots were used.
of succession, woody detrital stores may de-
crease below old-growth values (Spies et al.,
1988). Eventually, the mortality rates return to
old-growth amounts, leading to an additional
sink of carbon in the ecosystem. For example,
in the sitka spruce/western hemlock forest,
another 100 to 150 Mg C/ha may be stored in
dead wood in the next 200 to 300 years. This
amounts to an additional sink of 0.5 Mg C/ha/
yr in a forest that has reached a "steady-state"
living biomass (Harcombe et al., 1990).
REGIONAL STUDIES
Stand level ecological plots are not numerous
or representative enough to be unbiased esti-
mates of mortality rates on a regional or conti-
nental scale. Timber inventories generally fit
these criteria; however, few countries appar-
ently include non-catastrophic mortality. This
probably results from a methodological limita-
tion because permanent plots are needed to
estimate mortality, and most timber invento-
ries are one-time surveys. An exception is the
U.S. Forest Service Continuous Forest Inven-
Table 1. Expansion factors used to convert the volume of
growing-stockmortality from Waddelletal. (1987) to total
woody mortality.
Expansion factor
Growing stock to
Total volume (a)
Wood density (Mg/m3 )
Wood to Total mass (b)
Conifer
1.053
0.45
1.62
Deciduous
1.667
0.55
1.62
(a) Based upon ratios of total to growing stock volume (Bechtold, 1984).
(b) Includes bark, branches, roots.
tory (CFI), which has recorded mortality in
permanently marked plots since the 1950's. We
have converted these data (Waddell et al.,
1987) from loss in cubic volume of growing
stock to carbon using various correction factors
for non-merchantable trees, wood density and
non-merchantable parts (Table 1). We have
also used the fine-litter input rates estimated by
Meentemeyer et al. (1982) and the area in
forest land (Waddell et al., 1987) to estimate
fine-litter inputs for the same area examined
for tree mortality.
171
-------
CONSEQUENCES OFTREE MORTALITY TO THE GLOBAL CARBON CYCLE
Total Input USA (Tg carbon/year)
1952
1962 1970
Inventory Year
1976
1986
conifers
hardwoods
Figure 4. Woody detritus input associated with tree death in U.S. forests in the past 50 years.
Continuous Forest Inventory data indicate that
the rate of tree mortality in the United States
has increased in the last five decades from 52 to
65 Tg C/year (Figure 4). This is < 10% of the
total above-ground inputs, which at first glance
would support the notion that tree mortality
input could be overlooked with little conse-
quence. This perspective is very misleading,
however, because woody litter has a very slow
decay rate and consequently can form a large
amount of detrital stores. For example, ap-
plying the known differences in leafy and woody
decay rates for two typical forests (i.e., an east-
ern deciduous and a cold coniferous forest) we
find that woody detritus comprises 40 to 50% of
the above-ground detrital stores in a steady-
state condition.
The cause of the slight increase in tree mortality
in the United States is difficult to unravel given
the large-scale changes in pollution stress and
age-structure that have occurred. Examination
of regions indicates that the overall increase is
driven by increases in the eastern United States
that offset large decreases in the Pacific North-
west. We feel this pattern parallels shifts in
forest age structure; regions with decreasing
mortality rates are being converted to younger
forests, whereas regions with increasing mor-
tality rates are generally increasing in age.
Regardless of the current causes of increasing
mortality, it is important to note that current
tree mortality rates are far belowpre-European
levels. The effect of forest harvest on mortality
rates is most clearly seen in Oregon and Wash-
ington, a region where the area extent of old-
growth forests has decreased in the last 50
years. The CFI data show mortality has de-
clined two-fold during this period (Figure 5).
Extrapolating back to the turn of the century,
when these forests were first beginning to be
developed, indicates there has been a three- to
five-fold decrease in the mortality rate.
GLOBAL ESTIMATES
Ideally, one would base global estimates of tree
mortality upon forest inventories; however, un-
til more government agencies collect or publish
these types of data, direct estimates will not be
possible. We have therefore used an alterna-
tive approach to estimate mortality in intact
stands based upon the ratio of mortality and
growing stock volume (Figure 6). This ratio
amounts to a turnover rate for the living canopy
172
-------
ME. HARMON, S. BROWN, AND S.T. GOWER
FigureS.
Washingt
60
50
40
30
20
10
0
1
ba
Estimated hi
on.
Mortality ( Tg Organic Matter)
^\^___
^"\^
i • / 1 i i i i i i i
} 10 20 30 40 50 60 70 80 90 100
Year
CFI data maximum estimate minimum estimate
sed uporirWaddell et al. (1987) and Harmon et al. (1986)
storical changes in tree mortality during the last 90 years in the Douglas-fir region of Oregon and
that can be multiplied by the living biomass data
to crudely estimate the mass of mortality. Note
that in our current estimates, forest age struc-
ture is not used; more accurate estimates could
be made if age-class specific turnover rates
. were used in conjunction with age-class specific
biomass data. For the temperate and boreal
biomes, we used the U.S. CFI data, and for
humidtropicalbiomes,weuseddatafromBrown
350
300
250
200
150
100
50
0
(
based
Figure 6. Cubic mortalit
the value for a state.
Mortality (million cubic feet/year)
^^
D —
J^ a ^
a^^ ^er^-
3|Prt-P D *^ i t i t i
) 5 10 15 20 25 30 35
Growing Stock Volume (billion cubic ft)
D CFI DATA cold conifer warm conifer
on CFI data (Waddell et al., 1987)
y of growing stock versus growing stock for conifers in the United States. Each point represents
173
-------
CONSEQUENCES OF TREE MORTALITY TO THE GLOBAL CARBON CYCLE
Table 2. Mortality turnover rates of major forest regions
of the world.
Ifcgjon Mortally tumoveriate(%/yr) (a)
mean minimum
CoU confer (b)
Ccid deciduous (b)
Waimconiferfjb)
V&m deciduous (b)
Ttcficalclose(c)
Tropical open (ie)
0.43
0.87
0.95
0.64
1.58
0.22
0.17
0.52
0.52
0.42
0.13
0.08
maximum
1.00
1.48
1.28
0.93
4.93
0.35
(a) Gucub ted as the ratio between the mass orvolume cf mortality and the
growingstock mass orvolume,
(b)EaseduponUSKxcstServfceCH data forindwidualstates; range is the
lowest and highest value for states in given forest type.
(c) Based upon stand data from Down and Lugo (11).
and Lugo (unpublished) to estimate these turn-
over rates. The living biomass data were taken
from WoodweU et al. (1978) and Brown et al.
(1989).
tions: eastern pine-type species with 0.95%/yr
and cold coniferous forests with 0.43 %/yr. This
is consistent with the stand level results that
indicate higher mortality with increasing pro-
ductivity. The opposite appears true for de-
ciduous forests in that cooler, less productive
forests have a higher turnover rate (0.87/yr)
than warmer deciduous forests (0.64/yr). This
pattern may reflect the overall advantage that
conifers have in cooler climates. Alternatively,
this pattern may reflect the switch in succes-
sional status of conifers and deciduous trees
with latitude. That is, conifers are generally
short-lived early serai species in low latitudes,
whereas deciduous trees play this role in higher
latitudes. The tropical Venezuela data indicate
that both dry and moist tropical forests have
perhaps twice the rate of turnover of temperate
forests, suggesting a positive correlation be-
tween turnover rate and productivity.
Examining mortality turnover rates from five Applying these ratios to the biomass data indi-
different regions indicates considerable varia-
tion within and between regions (Table 2).
cates that globally 1.2-9.1 Pg C/yr is added to
detritus via tree mortality within intact stands
Coniferous forests have two distinct popula- (Table 3). By far the largest amount appears to
Table 3. Global detrital input associated with tree mortality.
Ecosystem
Tropical closed
Tropical open
Temperate evergreen
Temperate deciduous
Boreal
Total
Area (a)
(106 km2)
12.0 (h)
7.3 (h)
5.0
7.0
12.0
48.5
Live carbon (a)
(Pg)
122
26
79
95
108
743
Catastrophic
return interval
(centuries)
7-15 (d)
7-15 (d)
2-5 (e)
10-15 (f)
1-2 (g)
3-7
Mortality Input
Normal (b) Catastrophic (c)
1.91 (0.16-6.04)
0.06 (0.02-0.1)
0.75 (0.41-1.01)
0.61 (0.40-0.88)
0.46 (0.18-1.08)
3.79 (1.17-9.11)
0.08-0.18
0.02-0.04
0.16-0.39
0.06-0.09
0.54-1.08
0.86-1.78
(a) Based upon Woodwell et al., 1978; Brown et al., 1989.
(b) Mean (minimum-maximum), calculated using the mean, minimum and maximum
mortality turnover rates Table 2.
(c) Calculated by dividing the range of return intervals in years into the live biomass.
(d) Based on Sanford ct al., 1985.
(e) Based upon Kilgore (1978) and Christensen (1978).
(f) Based upon Scischab and Orwig, 1991.
(g) Based upon Heinsclman, 1978.
(h) Based upon Lanly, 1982.
174
-------
be dying in closed tropical forests that have both
high turnover rates and high biomass stores
(due to extensive areal extent). Boreal forests
are less important globally, but still may be
adding as much as 1 Pg C/yr from intact stands.
These estimates are low, of course, as they do
not include contributions from catastrophic
disturbances. Unfortunately, there are few
estimates of natural rates of catastrophic dis-
turbance on a global scale upon which to base
our calculations. We approach this problem in
two ways. First, one might consider the return
interval required to have catastrophic mortality
equal to within stand mortality. This would
indicate whether setting non-catastrophic and
catastrophic mortality rates equal is a reason-
able assumption. The return interval would be
the reciprocal for the estimated turnover rates
of intact stands. This indicates a catastrophic-
disturbance return interval as short as 43-200
years in tropical forests and as long as 200-600
years in boreal forests. The return interval for
boreal forests seems reasonable in that it im-
plies 2 to 6 x 106 ha/yr are disturbed, less than
the 8 x 106 ha/yr estimated for a decade with
high fire occurrence (Stocks, 1990). In contrast,
the return intervals for temperate and tropical
forests is unrealistically short. A more reason-
able estimate from these regions can be made
by using published return intervals (Table 3).
This indicates catastrophic disturbances may
be adding another 0.85 to 1.78 Pg/yr globally.
Thus total tree mortality input for the globe
would appear to be in the range of 2.03-10.89 Pg
C/yr.
Global estimates of fine litterfall (Meentemeyer
et al., 1982) are about 28 Pg/yr, thus woody
detrital inputs are relatively small compared
with that well-knowninput. Nonetheless, woody
detrital flows are important in two regards. The
M.E. HARMON, S. BROWN, AND S.T. COWER
first is that this flow is about equal to fossil fuel
burning. The idea that flows this large are being
ignored, as ecologists "balance" the global car-
bon budget (Post et al., 1990), is mind-boggling.
The second point is that while the input rate
may be small relative to leafy litter, the storage
in woody detritus may far exceed that of leafy
litter because of the much lower decomposition
rate of woody material. We have calculated the
potential steady-state carbon store in woody
detritus using the range of input rates (Table 3)
and estimated decay rates (i.e., tropical = 10%/
yr, temperate evergreen = 3%/yr, temperate
deciduous = 7.5 %/yr, and boreal 2%/yr). This
calculation indicates that given an input of 2 to
11 Pg C/yr, a steady-state store of 64 to 232 Pg
of woody detritus carbon results. Ecologists
have thus been ignoring a detrital pool that at
least equals and most likely far exceed the 70 Pg
of fine litter carbon (Post et al., 1990).
CONCLUSIONS
Our estimates of the detrital input associated
with tree mortality are preliminary, and to an
order of magnitude only. They do indicate,
however, that a major flow in the global carbon
cycle has been ignored. We question whether
the global carbon budget can be balanced when
flows as large 2-10 Pg C/yr are ignored. Our
analysis indicates global detritus maybe under-
estimated by an amount roughly equal to the
carbon stores in peats. Given that past forest
harvest and clearing have greatly reduced woody
detrital stores globally, this pool may have been
a significant source of carbon in the past and a
potentially significant future carbon sink in the
present. However, until we have better esti-
mates of tree mortality and woody detrital
decay rates, the exact contribution from woody
detritas to the global carbon cycle will remain a
mystery.
175
-------
CONSEQUENCES OF TREE MORTALITY TO THE GLOBAL CARBON CYCLE
REFERENCES
Agec, J.K, and M.H. Huff. 1987. Fuel succession in a
western hemlock/Douglas-fir forest. Canadian Journal of
Forest Research 17:697-704.
Bcchtold.WA. 1984. Forest statistics for North Carolina.
Resource Bulletin SE-RB-78, US. Forest Service, Ashville,
NC,p.60.
Bormann, F. H. and G.E. Likens. 1979. Pattern and
Process in a Forested Ecosystem. Springer-Verlag,Inc.,
New York.
Brown, S. and A.E. Lugo. 1991. Production of coarse
woody debris and its role in organic matter budgets in
tropical forests of Venezuela. Journal of Tropical Ecology
(submitted)
Brown, S. and A.E. Lugo. 1990. Tropical secondary
forests. Journal of Tropical Ecology 6:1-32.
Brown, S. and AJ.R. Gillespie, and A.E. Lugo. 1989.
Biomass estimation methods for trpoical forests wil appli-
cation to forest inventory data. Forest Science 35:881-902.
Chapin,F.S. and E.Matthews. 1992. Boreal carbon pools:
approaches and constraints in global extrapolations, (this
volume).
Christensen, N.L. 1978. Fire regimes in southeastern
ecosystems. In: Fire regimes and ecosystem properties.
General TechnicalReportWO-26, US ForestService, pp.
112-136.
Franklin, JJF., H.H. Shugart, and M.E. Harmon. 1987.
Tree death as an ecological process. Bioscience 37:550-
556.
Gessell, S.P. and J. Turner. 1976. Litter production in
western Washington Douglas fir stands. Forestry 49:63-
72.
Harcombe, PA., M.E. Harmon, and S.E. Greene. 1990.
Changes in biomass and production over 53 years in a
coastalPiceasitcheiisis-Tsugaheterophyllaforestapproach-
ing maturity. Canadian Journal of Forest Research
20:1602-1610.
Harmon, M.E., J.F. Franklin, F J. Swanson, P. Sollins, S.V.
Gregory, J.D. Lattin, N.H. Anderson, S.P. Cline, N.G.
Aumen, J.R. Sedell, G.W. Lienkaemper, K. Cromack Jr.,
and K.W. Cummins. 1986. Ecology of coarse woody
debris in temperate ecosystems. Advances in Ecological
Research 15:133-302.
Harmon, M.E., W.K. Ferrell, and J.F. Franklin. 1990.
Effects of carbon storage of conversion of old-growth
forests to young forests. Science 247:699-702.
Harmon M.E. and C. Hua. 1991. Coarse woody debris
dynamics in two old-growth ecosystems. Bioscience 41:604-
610.
Heinsehnan, M.L. 1978. Fire intensity and frequency as
factors in the distribution and structure of northern eco-
systems. In: Fire regimes and ecosystem properties.
General Technical Report WO-26, US Forest Service, pp.
7-57.
Houghton, RA. and G.M. Woodwell. 1989. Global
climate change. Scientific American 260:36-44.
Kilgore, B.M. 1978. Fire in ecosystem distribution and
structure: western forests and scrublands. In: Fire re-
gimes and ecosystem properties. General Technical Re-
port WO-26, US Forest Service, pp. 58-89.
Knox, R.G., R.K. Peet, and N. Christensen. 1989. Popu-
lation dynamics in loblolly pine stands: changes hi skew-
ness and size inequality. Ecology 70:1153-1166.
Lanely, J.P. 1982. Tropical forest resources.
Forestry Paper 30. FAO, Rome. 106pp.
FAO
Lugo, A.E. and S. Brown Tropical forests as sinks of
atmospheric carbon, in press. Forest Ecology and Man-
agement.
Meentemeyer, V., E.O. Box, and R. Thompson. 1982.
World patterns and amounts of terrestrial plant litter
production. Bioscience 32:125-128.
Neilson, R. 1992. Personal communication.
Post, W.M., T.H. Peng, W.R. Emanuel, A.W. King, V.H.
Dale, and D.L.DeAngelis. 1990. The global carbon cycle.
American Scientist, 78:310-326.
Sandberg,D.V.andR.D.Ottmar. 1983. Slash burning and
fuel consumption in the Douglas-fir subregion. Seventh
Conference on Fire and Forest Meterology, Fort Collins,
Co. pp. 90-93.
Sandford, R.L., jr., J. Saldarriaga, K.E., C. Uhl, and R.
Herrera. 1985. Amazon rain-forest fires. Science 227:53-
55.
Schiffman, P.M. and W.C.Johnson. 1989. Phytomass and
detrital carbon storage during forest regrowth in the
176
-------
southeastern United States. Piemont. Canadian Journal
of Forest Research 19:69-78.
Seischab, F.K. and D. Orwig. 1991. Catastrophic distur-
bances in the presettlement forests of western New York.
Torrey Botanical Club 118:117-122.
Spies, TA., J.F. Franklin, and T.B. Thomas. 1988. Coarse
woody debris in Douglas-fir forests of western Oregon and
, Washington. Ecology 69:1689-1702.
Stocks, B.J. 1990. The extent and impact of forest fires
in northern circumpolar countries. In: Atmospheric,
climatic and biospheric implications, Proceedings of the
Chapman Confernece on global biomass burning.
M.E. HARMON, S. BROWN, AND S.T. GOWER
Vogt,KA.,C.C.Grier,andD.J.Vogt. 1986. Production,
turnover, and nutrient dynamics of above- and below-
ground detritus of world forests. Advances in Ecological
Research 15:302-377.
Waddell, K.L., D.D. Oswald, and D.S. Powell. 1987.
Forest statistics of the United States. Resource Bulletin
PNW-RB-168, U.S. Forest Service, Portland, OR, 1089.
Woodwell, G.M., R.H. Whittaker, WA. Reiners, G.E.
Likens, C.C. Delwiche, and D.B. Botkin. 1978. The biota
and the world carbon budget. Science 199:141-146.
177
-------
-------
FOREST FIRES IN THE FORMER SOVIET UNION:
PAST, PRESENT AND FUTURE GREENHOUSE GAS
CONTRIBUTIONS TO THE ATMOSPHERE
Olga N. Krankina
«
ABSTRACT
Boreal forests of the former Soviet Union play a significant role in the global carbon cycle and the flux of greenhouse gases
to the atmosphere. Because fire and disturbance are ecologically inherent in boreal forests, enormous areas are burned
annually, contributing significantly to the release of greenhouse gases to the atmosphere. Emissions from forest fires were
calculated using data on the area! extent of fires in theyears 1971 to 1990, fuel loads at different forest sites and direct carbon
release/post-fire flux ratio. Extensive forest fires in the former Soviet Union from 1980-1989 occurred concomitant with
average ambient temperatures in the Northern hemisphere of 0.50-0.75 degrees above normal. Crown and understory fires
released 1.5 Gt of carbon to the atmosphere. Post-fire emissions due to microbial activity and respiration resulted in an
additional carbon flux of 0.21 Gt C/year. Forest fires are expected to be a primary mechanism driving vegetation changes
associated with projected global climate warming. Forest fires may occur on 30-50 percent of the land surface area, or 328-
458millionhaofforestlandinthe former SovietUnion, over SOyears. Thus, 6.5to 9.0 million ha offorestwillbum annually.
The flux of carbon from forest fires may reach 0.81-1.14 Gt annually. Such catastrophic fires may increase present-time
biogenicflux from forests by about 20percent and may significantly influence the global carbon cycle. Consistent management
measures and fire suppression activities might help mitigate these impacts.
INTOODUCITON
All aspects of the global carbon budget are
intensively studied and widely discussed in pub-
lications because steadily increasing greenhouse
gas concentrations are predicted to result in
global warming. Forests play a prominent role
in the terrestrial carbon cycle, serving as a
major carbon pool; however, the impact of
boreal and temperate forests on the carbon
cycle remains uncertain. Establishment, func-
tion and succession of boreal forest ecosystems
are inseparably linked with fires, which play an
important role in carbon turnover. Fires cause
pulse-like emissions of carbon into the atmo-
sphere and a considerable post-impact bio-
genic flux, which includes decomposition of
killed but uncombusted trees, increased soil
respiration and CO2 that would have been
accumulated by the forest had the fire not
occurred (Auclair, 1992).
Boreal forests of the former Soviet Union play
a significant role in the global carbon cycle as
they comprise more than one-fourth of the
world's forest resources (Krankina and Dixon,
1991). State forest lands occupied 1254 million
hectares in the former Soviet Union and ac-
counted for 36 percent of the country's terri-
tory. One-fourth of them are located in the
Europe-Ural zone, the rest in Siberia and Far
East (Figure 1). They are mostly boreal and
179
-------
FOREST FIRES IN THE FORMER SOVIET UNION
180
-------
O.N. KRANKINA
subboreal coniferous forests with pine and
spruce dominating the European part; east of
the Ural mountains different species of larch
prevail. Mature and over-mature stands oc-
cupy more than one-half of the total forest area;
the greater part of these stands are located in
the remote areas of Siberia and the Far East
region where only 5.9 percent of the human
population lives. Forest management activities
are essentially concentrated on the European
area of the former Soviet Union and in a few
areas of Siberia where industry and infrastruc-
ture have been developed.
Wildfires annually cause enormous damage to
forests in the Soviet Union, and fire control is
one of the major concerns of state forest man-
agement. Special administration within the
State Committee of Forestry (Goskomles) is
responsible for the protection of forests against
fires, insects and pathogens; however, because
of the lack of investments, this control infra-
structure remains underdeveloped. Fire man-
agement measures include cutting and plough-
ing fire-restricting strips, slash piling and burn-
ing, fire prevention information among the
population and personnel working in the forest,
etc. Fire fighting includes monitoring and in-
formation systems and also special crews
equipped with pumps, ploughs, vehicles and
aircraft. Eighteen percent of forest land is
covered by on-site control, 48 percent is moni-
tored by aircraft, and the remaining 34 percent
has no fire control at all (Korovin, 1989). The
latter are located mainly in Siberia and the Far
East, and it is there that the greater part of
forest fires occur. Fires are expected to be one
of the primary mechanisms driving vegetation
changes and a potentially significant feedback
to global warming.
The objectives of this study are to: (1) assess the
areal extent of forest fires in the former Soviet
Union; (2) estimate their present and expected
Table 1. Areal extent of forest fires.
Years
1971-1975
1976-1980
1981-1985
1985-1989
1980-1989
1987
1987
1989
1990
Area
million ha/yr
Average
0.53 \
0.40 V
0.31 /
0.84
3.00
Selected years
10.0
0.42
2.07
1.30
Reference
-Prilepo, 1988 (a)
Andreev, 1991 (a)
Stocks, 1991
Stocks, 1991
Anonymous, 1988 (a)
Afrin and Bushuj, 1990 (a)
Shubin, 1991 (a)
(a) Monitored territory only.
future contribution to carbon emissions; and
(3) identify mitigation options to reduce their
impact.
METHODS
Three different sources of information were
used to estimate the areal extent of forest fires:
published historical statistics (Table 1), which
account only for fires in monitored territory;
Stocks' (1991) estimate of the areal extent of
forest fires in the former Soviet Union from
1980-1989 based on remote-sensing data; and
forest inventory statistics that report the accu-
mulated total area of burned and dead stands
(Anuchin et al., 1986).
Two major types of forest fires are considered
in this study: crown fires and understory fires.
In years with normal weather conditions, fires
occur mostly in pine and larch stands on dry
sites and on drained peatlands. Understory
fires will account for 82 percent of the burned
area and crown fires for 18 percent (Belov,
1976). In years with dry and hot weather, fires
spread to fresh and wet sites. In such years, the
percentage of devastating crown fires can reach
50 percent of the burned area (Belov, 1976).
The stock of fuel combusted by an average
181
-------
FOREST FIRES IN THE FORMER SOVIET UNION
Table 2. Stock of fuel combusted by an average forest
fire in coniferous forests (Belov, 1976).
Table 3. Carbon flux from forest fires, including
direct and post-fire release.
Canopy
Condition
of Site
Dry
F«sh
Wet
Dry Organic
Matter (t/ha)
10
15
6
Carbon (a)
(t/ha)
5
73
3
Forest Floor
Dry Organic
Matter (t/ha)
15
35
130
Carbon (a)
(t/ha)
73
153
65
(a) Biomass to carbon conversion factor: 03
forest fire depends greatly on the type of site.
Table2shows averaged data compiledby Belov
(1976) from a number of publications. Calcula-
tions of direct carbon release were based on the
following assumptions: 5-year periods with
burned area less than average represented nor-
mal weather conditions and so have an 82:18
understory/crown fire ratio and stock of fuel
combusted corresponding to dry sites; 5-year
periods with burned areas greater than average
were considered as those with peak forest fires
and so have a 50:50 understory/crown fire ratio,
and the amount of fuel combusted was aver-
aged for dry, fresh and wet sites (Belov, 1976).
As a result of these calculations, it was esti-
mated that 1 ha burned in a year, with average
weather conditions, would release, on average,
8.41C, and in peak years, the emissions from 1
hectare would reach 32.51 C. These estimates
generally agree with the 12.8 t C/ha value
suggested by Stocks (1991) that is considered to
be broadly representative of temperate and
boreal conditions.
Post-fire biogenic flux was estimated by A.
Auclair (1992) to be, on average, 2.8 times
greater than direct release. Our calculations of
post-fire flux are based on this ratio and on the
average direct release in the previous decade.
The results of these calculations are presented
in Table 3.
Future forest-firecarbonemissionswere evalu-
ated for two climate-change scenarios under
double CO2 conditions (UKMO and OSU) and
Years
1971-1975
1976-1980
1981-1985
1985-1989
1990-1995
Estimated
Burned Area
(million ha/yr)
2.65
2.00
133
4.20
630
Direct Flux
(GtC/yr)
0.086
0.017
0.013
0.137
0.208
Post-Fire Flux
From Previous
Decade (a)
(GtC/yr)
0.144
0.042
0.210
Total
(GtC/yr
0.157
0.179
0.418
(a) Post-fire flux is 2.8 times greater than direct release (AuClair, 1991).
corresponding vegetation changes (Smith et al.,
1991). These changes are expected to occur on
39.8-56.6 percent of terrestrial ecosystems, or
328-458 million hectares of forest land in the
former Soviet Union. Forest fires can be one of
the primary mechanisms driving projected veg-
etation changes. We assumed that forest fires
will occur once on all of the area with expected
vegetation changes within the 50-year time pe-
riod. Thus, 6.5 to 9.2 million ha of forest land (a
quite realistic figure considering the level of
forest fires in the past few years) will be burned
annually. Using the value of carbon emissions
from one hectare of burned forest calculated
for peak forest-fire years, the future flux of
carbon from forest fires was estimated.
To assess the efficiency of fire control for car-
bon conservation, a comparison with forest
plantations in terms of cost per ton of carbon
sequestered was made. Cost values used in this
study date from 1986 (Sinitsin, 1988). They are
intended for comparison only, and because of
ongoing changes in exchange rates, they were
not converted to dollars.
RESULTS
According to published historical fire statistics,
annually burned forest area, on monitored ter-
ritory, averaged for 5-year periods varies from
0.31 million ha in 1981-1985 (Prilepo, 1988) to
0.84 million ha in 1985-1989 (Andreev, 1991)
(Table 1).
182
-------
O.N. KRANKINA
5 --
£
"8
|4 +
I3 +
2 --
1 - -
H 1990 extrapolation for years 1990-1995
II Estimated total values
CD Historical forest fire statistical data for monitored territory
1970-75
1975-80
1980-85
Years
1985-90
1990-95
Figure 2. Dynamics of burned forest area.
American estimates (Stocks, 1991) indicate as
much as 3 million ha burned annually from
1981-1989, and this value can be supported by
forest inventory data. According to the last
overall forest inventory of 1983, the total area of
burned stands in the former Soviet Union is
30.5 million ha (Anuchin, 1985). Assuming the
average post-fire regeneration period to be 10
years, annually burned territory should be
about 3 million ha. So, historic fire statistics for
monitored territory appear to underestimate
the areal extent of forest fires by a ratio of 1:5.
Based on this ratio, the actual extent of forest
fires for the years 1971-1990 was assessed (Fig-
ure 2), and data for 1990 was extrapolated for
1990-1995. The latter values should be consid-
ered with caution, as should any extrapolation
results. The mean value of burned forest area
for 1971-1990 was 2.5 million ha.
Estimated carbon emissions from forest fires
are comprised of direct flux (13-208 million
tons of carbon per year) and post-fire release.
The total flux of carbonfrom forest fires amounts
to 157-418 million tons of C/year (Table 3).
During the 1980s, forest fires released to the
atmosphere 1.5 Gt C, and their post-fire effect
is expected to increase the flux of carbon in the
1990s by 0.21 Gt C/year. In the future, if
predicted climate and vegetation changes oc-
cur, carbon emissions from forest fires may
reach 0.81-1.14 Gt C/year.
DISCUSSION
Comparison of data from different sources on
the areal extent of forest fires (Table 1) shows
that historical fire statistics tend to underesti-
mate burned area. This underestimation is
explained by the fact that 97 percent of burned
and dead stands are located in East Siberia and
the Far East, regions with enormous territories
not covered by fire monitoring; while statistics
account only for the monitored territory
183
-------
FOREST FIRES IN THE FORMER SOVIET UNION
(Anuchin et al., 1986). This explains the dis-
crepancy between fire statistics and American
estimates.
Three million ha burned annually in 1981-1989
in the former Soviet Union comprise 37.5 per-
cent of all the forest fires in the Northern
Hemisphere (Auclair, 1992). Compared with
other sources of carbon, the contribution of
forest fires is substantial, making up 15-40 per-
centof allindustrial carbon emissions (Makarov
andBashmakov, 1990). The flux of carbon from
forest fires is commensurable with the net sink
of 0.41 Gt of C in former Soviet Union forests
assessed by Sedjo (1992). Our estimate agrees
well with data on forest-fire carbon emissions in
the NorthernHemisphere publishedby Auclair
(1992).
Projected changes in global climate and popu-
lation growth are likely to increase the inci-
dence, extent and intensity of forest fires. For-
est fires are highly sensitive to ch'mate changes;
increasing mean annual temperatures by 0.5-
0.75 degrees in large areas of the boreal zone in
the 1980's (Houghton et aL, 1990) resulted in a
five-fold increase of the area! extent of forest
fires (Auclair, 1991). Besides this direct effect,
ch'mate changes will enhance vegetation die-
back and increase the stock of fuel in forests.
Expected climate warming, with t° several de-
grees C° higher than at present, is likely to result
in catastrophic forest fires, causing economic
and ecological losses. Fires are expected to be
a substantial feedback to global warming con-
tributing about 1 Gt C annually to further CO2
build-up in the atmosphere.
Development of a fire control system can offset
these negative effects to a certain extent. The
cost of forest protection in 1989 was as small as
15 kopecks/ha/year (Korovin, 1989), which is
less than one cent/ha/year. Improvement and
expansion of fire-control systems on all forest
land will require additional state expenditures.
These measures can reduce burned area by 20
percent (Korovin, 1989) and so conserve 30-80
Mt C annually. Assuming that the cost of fire
control must be doubled, the cost of conserva-
tion of 11C is 0.7-1.8 rubles.
Compared with forest plantations, another for-
est management option to sequester carbon,
fire control has certain advantages. The cost of
forest plantations in 1986 was about 77.5 rubles
/ha (Sinitsin, 1988), and in. 50 years, these
stands will sequester 18-39 tons of carbon/ha
(Dixon et al., 1991). The cost of the sequestra-
tion of 1 t C by forest plantations is 2.0-4.3
rubles. Carbon conservation by means of fire
control appears to be more cost-efficient than
plantations. Besides that, it has the advantage
of giving immediate results. From a long-term
perspective, forest fires can be reduced by in-
tensive forest management regulating stocks of
fuel in forest ecosystems and maintaining mixed-
species composition with a considerable share
of deciduous forest in each landscape.
However, the potential for fire control is some-
what limited by the fact that fires are inherent
to boreal forests and so cannot be totally elimi-
nated. Forest management systems aimed at
sequestering carbon should comprise a whole
complex of measures includingplantations, thin-
ning, improvement of harvesting practice and
others (Dixon et al., 1991), but fire control
should be at the top of this list because it is a
cost-efficient option and provides immediate
results. Forest fire management in the former
Soviet Union helps to both reduce carbon emis-
sions and to preserve valuable forest resources.
Unfortunately, the crisis situation in the former
Soviet Union economy makes governmental
investments in large-scale ecological programs
highly unlikely. International efforts to help
Russia control forest fires in its territory prob-
ably should be considered within the frame-
work of agreements on biosphere management
to conserve and sequester carbon.
184
-------
O.N. KRANKINA
CONCLUSIONS
• Enormous areas of boreal forest in the former
Soviet Union are destroyed by fire. The aver-
age area burned annually during the last 20
years is estimated to be 2.5 million ha, and most
of it is located in Siberia and the Far East
region.
• Forest fires make a considerable contribu-
tion to carbon emissions from terrestrial eco-
systems. Emissions comprising immediate di-
rect release and long-term post-fire flux add
0.16-0.42 Gt C to the annual flux of carbon into
the atmosphere.
• In the future, if projected global warming
occurs, the share of forest fires in the efflux of
carbon from terrestrial ecosystems in the former
Soviet Union is expected to increase up to 1 Gt
C/year.
• The area! extent of forest fires and their
contribution to carbon emissions can be re-
duced by fire-control systems. Carbon conser-
vation by means of fire control is more cost-
efficient than sequestration by forest planta-
tions.
• The economic situation in the former Soviet
Union at present makes governmental invest-
ments in carbon sequestration and conserva-
tion programs highly unlikely.
REFERENCES
AfrinM.I.andM.1. Bushuj. 1990. Severe lessons of forest
fires. Forest Industry 5:2-3 (in Russian).
Andreev, NA. 1991. In the Forest Management Ministry
of Russia. Forest Management 4:20-21 (in Russian).
Anonymous. 1988. Stop fire in the forests. Forest
Management 4:2-4 (in Russian).
Anuchin, N.P., et al. 1986. Forest Encyclopedia, Soviet
Encyclopedia, Moscow, USSR. Vol 1 & 2 (in Russian).
Auclair, A.N.D. 1992. Forest wildfire as a recent source
of CO2 at mid-high northern latitudes. Nature (submitted
for publication).
Belov, S.V. 1976. Forest Pyrology. Leningrad Forestry
Academy of the USSR, p.65 (in Russian).
Dixon, R.K., P.E. Schroeder, and J.K. Winjum. 1991.
Assessment of promising management practices and tech-
nologies for enhancing the conservation and sequestration
of atmospheric carbon and their costs at site level. EPA
report 600/3-91/067.
Houghton, J.T., GJ. Jenkins, and J.J. Ephraums. 1990.
Climate Change: The IPCC Scientific Assessment. Cam-
bridge University Press, Cambridge, pp. 352.
Korovin, G.N. 1989. For forests, good protection. Forest
Management 4:2-6 (in Russian).
Krankina, O.N. and R.K. Dixon. 1991. Forest manage-
ment in Russia: challenges and opportunities in the era of
perestroica. Journal of Forestry 90:29-34.
Makarov, AA. and I. Bashmakov. 1990. The Soviet
Union. In: Carbon Emission Control Strategies: Case
Studies in International Cooperation, edited by W.V.
Chandler. World Wildfire Fund Publications, Baltimore,
MD, pp. 35-53.
Prilepo,N.P. 1988. Troubles of the Russian forest. Forest
Management 12:2-6 (in Russian).
Sebjo, RA. 1992. Temperate forest ecosystems in the
global carbon cycle. Ambio. In press.
Shubin, B A. 1991. Chronicles. Forest Management 4:20
(in Russian).
Sinitsin, S.G. 1988. Dynamics of forest management
spendings. Forest Management 6:2-5 (in Russian).
185
-------
FOREST FIRES IN THE FORMER SOVIET UNION
Smith, TM., HJHL Shugart, G.B. Bonan, and Smith J.B.
1991. Modeling the potential response of vegetation to
global climate change. Advances in Ecological Research
22:93-116.
Stocks, B.J. 1991. The extent and impact of forest fires
hi northern circumpolar countries. In: Global Biomass
Burning: Atmospheric, Climatic and Biogenic Implica-
tions, edited by J.S. Levine. MIT Press, Cambridge, MA,
pp. 197-202.
ACKNOWLEDGMENTS
The author gratefully acknowledges assistance and helpful suggestions from A. Auclair and R.K. Dixon.
186
-------
ASSESSING CLIMATE CONTROLS OF NORTHERN FOREST SOILS AND
TREE GROWTH: RECONSTRUCTING SOIL MOISTURE, TEMPERATURE
AND SOLUTION CHEMISTRY FROM MONTHLY WEATHER RECORDS
Paul A. Arp and Xiwei Yin
ABSTRACT
Monthly climate data from a nearby weatherstation were used to simulate, amongothervariables, soil moisture, temperature
and solution chemistry at a northern hardwoods site in Ontario (Turkey Lakes Watershed). According to standardized tree
ring analysis, tree growth was strongly correlated with the reconstructed time series for soil moisture, soil solution nitrate
concentrations and mid-winter temperature. The results are specific to the site, but the methods used (simulation models
to assess forest hydrology and soil temperatures at northern latitudes) are general and should be applicable to examine long-
term effects of climate on forest soils and forest growth across sites and regions.
INTRODUCTION
Forest growth and carbon cycling are con-
strained by local interactions between climate
and soil. For example, soils retain moisture and
nutrients essential for plant growth. Projecting
local soil reactions to climate change, there-
for e, is critical for predicting future forest growth
and carbon cycling rates under various climate
scenarios.
The objective of this paper is to demonstrate a
general approach to simulate dynamic soil vari-
ables using commonly available climate data,
and to partition the contribution of climate and
soil factors to tree growth variations. Specifi-
cally, we summarize (i) two general modules
that simulate soil moisture and temperature
with primary driving input limited to monthly
total precipitation and monthly mean air tem-
perature; (ii) regression models linking soil
solution chemistry to soil moisture and tem-
perature; and (iii) regression results showing
that year-to-year variations in tree growth are
likely attributable more to dynamic soil vari-
ables than to climate variables alone.
Study Site
BasinSl of the Turkey Lakes Watershed (47°03'
N, 84°25' W) in Ontario, Canada, was selected
as the primary study site. The site is part of an
intensive monitoring network for assessing ef-
fects of acidic deposition on forested water-
sheds (Jeffries et al., 1988). Climate records
since 1951 were available from Sault Ste. Marie
(about 60 km south). Local climate and soil
data were available since 1980.
Vegetation is an uneven-aged (40-160 years)
northern tolerant hardwoods stand dominated
by sugar maple (Acersaccharum Marsh.). Cli-
mate, vegetation, hydrology and ion fluxes of
the watershed are described elsewhere (e.g.,
Foster et al., 1986; Nicolson, 1988; Morrison,
1991).
187
-------
ASSESSING CLIMATE CONTROLS OF NORTHERN FOREST SOILS AND TREE GROWTH
Computer Modules Simulating Soil
Moisture and Temperature
Figures 1 and 2 show flowcharts for the hydrol-
ogy and soil temperature modules, respectively.
Figures illustrates datarequirements and types
of simulation output.
The hydrology module simulates all major wa-
ter fluxes through upland forests. It follows the
trickle-down of precipitation water from one
layer to the next (canopy, forest floor, soil, and
subsoil). Effective "permeability11 coefficients
areused for each layer, assuming that water flux
over a given period of time is proportional to
the amount of water available for flow in that
layer.
For the temperature module, heat also flows by
layer, but the direction of flow depends on the
temperature gradient. The model treats heat
transfer by one-dimensional heat conduction.
Latent heat transfer through freezing and thaw-
ing and internal heat sources and heat sinks are
also considered.
The modules are designed to be applicable
across regions by minimizing data requirements
and the number of calibrated parameters. Spe-
cifically, data input is limited to (1) commonly
available climate records (i.e., monthly total
precipitation and monthly mean air tempera-
ture); and (2) readily obtainable descriptive
site information (i.e., latitude, vegetative area
index, dominant tree species, soil thickness and
soil texture). Module parameters, once cali-
brated, should be applicable without modifica-
tion for other sites unless warranted by out-
standing physical differences.
To maximize theirusefulness, the modules gen-
erate output for all major pool and flux vari-
ables significant for ecological investigations,
including (1) throughfall; (2) stemflow; (3) snow-
pack (duration, depth and water equivalent);
(4) water contents (pools) and percolate vol-
Mttt
evapotranspiration
precipitation
(rain, snow)
\ (interception)
(evaporation)
snow
throuphfall.
canopy
snow-
snow- melt
pack }
fnr
flo
^
5
«
es
ar
forest floor percolatior
rain throughfall
7
soil
soil percolation \
subsoil
(subsoil percolation) \ (runoff)
strearhflow
HM
Figure 1. Flowchart for the hydrology module.
air temperature
Figure 2. Flowchart for the temperature module.
188
-------
PA. ARP AND XIWEI YIN
Input
-latitude
_vegetative area index
_air temperature -dominant tree species
_soil thickness (by layer)
_soil texture (by layer)
-rainfall
-snowfall
,
isimulation
_ canopy interception O U t p U t
-throughfall & stemflow
_snowpack: duration, thickness, etc.
-forest floor: water content & percolate
-soil: water content & percolate
-subsoil: water content & percolate
-evapotranspiration
-runoff
-soil temperature (any depth)
-frost: duration & depth
Figure 3. Data requirements and types of simulation output.
umes (fluxes) for forest floor, rooting soil and
subsoil; (5) evapotranspiration; (6) streamflow;
and (7) soil temperature at any depth.
The modules were initially calibrated for north-
ern tolerant hardwoods at Turkey Lakes (see
above). They have since been verified for a
balsam fir forest at Lac Laflamme, Quebec,
mixed woods (maples, birches and red spruce)
at Kejimkujik, Nova Scotia, Canada, and a
watershed with mosaic patches of grasslands,
coniferous or deciduous forests in Colorado,
U.S.A.
Regressions Predicting Soil Solution
Chemistry
We are currently developing ion-flux modules
simulating dynamic aspects of soil solution chem-
istry of forested sites. For the purpose of this
study, we used regression techniques to reveal
factors that contribute significantly to temporal
variations of nutrient concentrations in soil
solution. The resulting regressions were fur-
ther applied to reconstruct solution nutrient
concentrations from simulated soil moisture
and temperature.
In contrast to the simulation modules, regres-
sion analysis for nutrient concentrations was
site specific. For brevity, and because the
concentrations of other ions (sulfate, calcium
and magnesium) were highly correlated with
nitrate (NO3) concentration, only NO3 is de-
scribed here.
Soil solution NO3 concentrations showed two
distinct phases. In particular, NO3 concentra-
tion (umol C/L) during the non-foliage period
(November-May) varied primarily with season,
and was approximately given by:
NO3N_M=-76433 + 22412M-2440.5 M2
+ 117.09 M3 - 2.0873M4,
(1)
where M is digital seasonal month (June=6.5,
July=7.5,....May =17.5).
During the foliage period (June-October), NO3
concentrations correlated to forest floor perco-
late volume PCff (mm/month), soil water con-
tent WC^j, as a fraction of field capacity FCsoil
( = 145 mm), and seasonal month M as follows:
N03J.0 = 18.2 - 0.0190(PCff - 85)2+923(WCsoil/FCsoil
-0.73)2 + 122sin[(M + 7.5)2*PI/12] (2)
Nitrate concentrations in October and Novem-
ber were highly correlated:
N°3.November=74.6-.524N03.0ctober
+ .00660(N03.0ctober)2
(3)
Equations 1-3 were developed from field data
collected between 1982 and 1989 (for data
collection see Foster 1985; Foster and Nicolson
1988), and were used in conjunction with simu-
lation results for forest floor percolate and soil
189
-------
ASSESSING CLIMATE CONTROLS OF NORTHERN FOREST SOILS AND TREE GROWTH
Table 1. Multiple regression coefficients for soil and climate variables (xj) found to contribute significantly to year-to-
year variations of sugar maple radial growth (y) at Turkey Lakes (y= S^ * 3q ).
X
Intercept
soil solution nitrate
air temperature
[soil water content -.88
[_ field capacity(a)
soil solution nitrate
T
Time
(Sept-Oct.)lagged
February
May-August
August
Unit
mmolo/L
°C
mm/mm
mmolc/L
ai(I)#
0.55
1.7
-0.0156
-7.56
1.29
*or>p
0
0.593
-0.465
-0.818
0.544
r partial
0.776
-0.696
-0.792
0.653
Kcumul.
0.249
0.436
0.635
0.790
Note: All coefficients are significant atp<.0001 (n=38; RMS=.048).
ffa j (I): regression coefficients derived from analysis of raw data; ai (II): regression coefficients obtained
after each variable was standardized.
(a) The field capacity for the soil is 110 mm.
water content to reconstruct NO3 concentra-
tions since 1951.
Tree-Ring Analysis Correlating Tree Growth
to Climate and Soil
Year-to-year variations in radial growth be-
tween 1951 and 1990 were measured with tree-
ring data taken from eight sugar maple trees.
Yearly increments represented the average of
ring width measured along the four cardinal
directions for all trees. Increments were con-
verted to a standardized tree-ring growth index
(i.e., the ratio between actual and smoothed
radial increments) to remove stand and age
effects (Fox etal., 1986; Arp and Manasc, 1988).
Multiple regression analysis revealed that about
80 percent of year-to-year variations in tree
growth were explained collectively by (i) soil
water content during the growing season, (ii)
NO3 concentration in soil solution in Septem-
ber and October of the previous year, (iii) NO3
concentration in soil solution in August, and
(iv) mean air temperature in February (Table
1). Regression coefficients derived from the
standardized values for the four variables were
-0.818, 0.593, 0.544 and -0.465, respectively
(Table 1). Hence, soil moisture during the
growing season appeared to have the strongest
impact on tree growth among the four signifi-
cant variables. In contrast, February air tem-
perature appeared to have the least impact.
These results are consistent with our current
knowledge of tree growth in general, and sugar
maple growth in particular. For instance, Spurr
and Barnes (1980) generalized that tree growth
responds more to soil water deficit than to any
other perennial site factor. Yet, excessive soil
water could lead to inadequate oxygen supply,
leading to death of fine roots, a reduced absorb-
ing root surface and decreased tree growth.
Consequently, the partial correlation between
the growth index and soil moisture was strong
and was parabolic with an optimum moisture
level about equal to the long-term average for
the site.
Further, low N availability during dry summers
may aggravate the effect of drought on sugar
maple growth (Mader and Thompson, 1969).
For September and October, monthly precipi-
tation tends to reach a peak. As a result,
increased soil moisture coupled with mild soil
temperatures appears to accelerate nitrifica-
tion and N uptake. Nutritional accumulation by
the tree in autumn is important for tree growth
during the next growing season (Kramer and
Kozlowski, 1979).
Finally, mild mid-winter air temperatures may
trigger early sap flow, which could subsequently
lead to plant-internal frost injuries during the
return of cold air. Auclair (1988) reported
severe sugar maple bud mortality subsequent to
subfreezing temperatures following a February
thaw in 1981 in Quebec.
190
-------
P.A. ARP AND XIWEI YIN
In summary, tree-ring analysis suggests that
year-to-year variations of tree growth for a
given site may be attributed more to dynamic
soil variables than to climate variables alone.
Further, climate change would affect tree growth
through specific, season-dependent responses.
For the study site, mid-winter thaws would
reduce tree growth. Shifts in climate that change
May-August soil moisture in either direction
would also reduce tree growth. Climate affects
solution NO3 concentrations indirectly through
soil moisture. The net effect of climate on tree
growth through dynamic soil variables such as
moisture, temperature and solution composi-
tion would therefore depend on the compara-
tive magnitudes of climate-induced changes
among those variables.
CONCLUSIONS
• Soil moisture, temperature and solution chem-
istry can be effectively simulated with input
limited to monthly records for precipitation
and air temperature, and to readily obtained
site data;
• Climate affects tree growth for a given site
largely through specific responses of dynamic
soil variables; and
• An effective projection of climate-induced soil
changes is essential for simulating forest growth
and carbon cycling under various climate sce-
narios.
REFERENCES
Arp, PA. and J. Manasc. 1988. Red spruce stands down-
wind from a coal-burning power generator: tree-ring
analysis. Can. J. For. Res. 18: 251-264.
Auclair, W.M. 1988. A climate change theory of forest
decline. In: Woody Plant Growth in a Changing Physical
and Chemical Environment, edited by D. Lavender. Proc.
IUFRO Workshop, July 1987, Vancouver. Univ. British
Columbia Press, Vancouver, pp. 80-89.
Foster, N.W. 1985. Acid precipitation and soil solution
chemistry within a maple-birch forest in Canada. For.
Ecol. Manage. 12: 215-231.
Foster, N.W., I.K. Morrison, and J A. Nicolson. 1986. Acid
deposition and ion leaching from a podzolic soil under
hardwood cover. Water Air Soil Pollut. 31: 879-890.
Foster, N.W. and JA. Nicolson. 1988. Acid deposition and
nutrient leaching from deciduous vegetationand podzolic
soils at the Turkey Lakes Watershed. Can. J. Fish. Aquat.
Sci. 45 (Suppl. 1): 96-100.
Fox, CA., W.B. Kincaid III, T.H. Nash, D.L. Young, and
H.C. Fritts. 1986. Tree-ring variation in western larch
(Larix occidentalis) exposed to sulfur dioxide emissions.
Can. J. For. Res. 16: 283-292.
Jeffries, D.S., I.K. Morrison, and J.R.M. Kelso. 1988. The
Turkey Lakes watershed study. In: Canadian Hydrology
Symposium, 9-11 May 1988, Banff, Alberta. Associate
Committee on Hydrology, National Research Council of
Canada, pp. 117-125.
Kramer, P J. and T.T. Kozlowski. 1979. Physiology of
Woody Plants. Academic Press, New York.
Mader, D.L. and B.W. Thompson. 1969. Foliar and soil
nutrients in relation to sugar maple decline. Soil Sci. Soc.
Am. Proc. 33: 794-803.
Morrison, I.K. 1991. Addition of organic matter and
elements to the forest floor of an old-growth Acersaccharum
forest in the annual litterfall. Can. J. For. Res. 21.
Nicolson, JA. 1988. Water and chemical budgets for
terrestrial basins at the Turkey Lakes watershed. Can. J.
Fish. Aquat. Sci. 45 (Suppl., 1): 88-95.
Spurr, S.H. and B. V. Barnes. 1980. Forest Ecology. 3rd ed.
John Wiley, New York.
191
-------
ASSESSING CLIMATE CONTROLS OF NORTHERN FOREST SOILS AND TREE GROWTH
ACKNOWLEDGEMENTS
We thank N. Foster, I. Morrison and J. Nicolson for valuable data that made this work possible. The study was funded
through the joint Project on Atmospheric Acid Sulfate/Nitrate Loading sponsored by Canadian Electrical Association
(CEA), Canadian Pulp and Paper Association (CPPA), Forestry Canada (FC), Environment Canada (EC) and Natural
Science and Engineering Research Council of Canada (NSERC).
192
-------
SELECTING A MODEL TO ESTIMATE THE EFFECTS
OF CLIMATE CHANGE AND MANAGEMENT ACTIVITIES
ON CARBON CYCLING IN TEMPERATE FOREST REGIONS
Bob Zybach
ABSTRACT
TheuseofspecificcomputerizedforestmodelstopredictthepossibleeffectsofglobalclimatechangeontheEarth's temperate
forest regions is examined. The effects are described in two basic categories: 1) abiotic changes resultingfrom global warming
(i.e., increased temperatures, increased atmospheric CO^ changed moisture availability and altered solar radiation); and
2) forest management policies that can be developed in response to climate change projections. This report is arranged in
three parts: 1) a general description of the historical relationship between human activity, temperate forests, increased CO2
emissions and climate change; 2) an outline of forest management strategies that are being considered to reduce or eliminate
net increases of carbon in the atmosphere; and 3) a brief examination of the types of computer models being developed that
are capable of assessing climate and management change impacts upon temperate forests.
INTRODUCTION: FOSSIL FUELS
AND GLOBAL WARMING
The possibility of a change in climate caused by
human burning of fossil fuels was first brought
to the attention of the scientific community in
April, 1896, by the Swedish chemist Svante
Arrhenius, who characterized the problem as
"evaporating our coal mines into the air". At-
mospheric increases of carbon dioxide due to
fossil fuel consumption dating to the beginning
of the industrial revolution were finally proved
"scientifically" by Charles Keeling, through re-
search initiated in the early 1950s. Subsequent
use of computerized Global Circulation Mod-
els (GCMs) by a number of different scientific
teams have generally supported Arrhenius's
original calculations that a doubling of carbon
dioxide in the atmosphere could cause the
average temperature of the earth to increase by
as much as five or six degrees centigrade. Cur-
rent projections call for such a doubling to occur
sometime within the next century; current esti-
mates are that enough coal and oil remain in the
ground to raise the level of carbon dioxide
about ten times higher (Weiner, 1991).
Between 1860 and 1960, it has been estimated
that humans added 80 billion tons (Gt) of carbon
to the atmosphere; from 1960 to 1990, another 80
Gt have been added, and the rate continues to
increase (Weiner, 1991). At the present time, it
is estimated that world combustion of fossil fuels
injects about 5 Gt of carbon into the atmosphere
annually. About 2 Gt is removed by the oceans
(Rosenfeld and Botkin, 1990). The net annual
increase of 3 Gt of carbon is added to the 700 Gt
currently estimated to be contained in the atmo-
sphere and is the basis for present concern;
human activity maybe causing a rapid and perva-
sive warming of the planet.
The possibility of an increase in the average
temperature of the earth by as little as one or
193
-------
SELECTING A PREDICTIVE MODEL TO MEASURE THE EFFECTS OF CLIMATE CHANGE AND MANAGEMENT ACTIVITIES
two degrees centigrade within the next several
decades has given rise to general alarm within
the scientific community. Recentprojections of
resulting discomforts and disasters to the hu-
man community have been publicly delivered
to influential politicians in several nations and
continue to be widely reported in the popular
press. One result of this alarm has been the
development of a number of strategies for de-
creasing the use of fossil fuels and/or increasing
the rate of sequestering atmospheric carbon
(which exists primarily in the form of carbon
dioxide) into a variety of biological "sinks" lo-
cated about the globe. Forest and forest-prod-
uct management options figure prominently in
both strategies.
TEMPERATE FOREST HISTORY
Forests began to appear in the Silurian period,
about 350 million years before the present
(BP). They reached their peak development
during the Carboniferous period, 270 to 220
million years BP, at a time when the earth was
completely frost free. Widespread glaciation
during the ensuing Permian period greatly re-
duced the area of forest cover and resulted in
massive extinctions of both plant and animal
species. By the beginning of the Tertiary pe-
riod, about 50 million BP, modern tree species
had come into existence.
Large-scale human manipulation of forests is
thought to have started between 7000-5000
years BP, with the development of domesti-
cated plants and animals requiring fields and
pasturage. Itis estimated that the earth's forests
have been reduced by at least one-third since
that time, almost entirely due to human activi-
ties (Hermann, 1976). James Watt's develop-
ment of the steam engine in the eighteenth
century became the starting point for the sys-
tematic mining of the buried Carboniferous
forests for fuel. The combination of modern
forest clearing and fossil forest consumption by
humans is thought to be almost entirely respon-
sible for the increase in atmospheric CO2 that
has occurred over the past several millennia.
The rate of increase has accelerated dramati-
cally during the past century and is presumed to
be related to expanding human populations
and the development of the automobile.
Today there are about 8 billion acres of forests
on earth. This figure represents an estimated
reduction of 32-35% of temperate forests, 24-
25% of subtropical forests and savannahs, and
15-20% of tropical forests since the advent of
agriculture (Kauffman et al., 1991). Forests
contain about 90% of the carbon in terrestrial
vegetation, making forest biomassamajor regu-
lator of atmospheric CO2 (Graham et al., 1990).
The purposeful and accidental burning of for-
ests and forest products has been estimated to
cause about 40% of the human-based CO2that
is put into the atmosphere each year (Kauffman
et al., 1991).
Temperate forests primarily exist in the north-
ern hemisphere, between 30° to 60° North lati-
tude. They are estimated to contain about 47%
of the world's forest biomass and can be charac-
terized as consisting of deciduous, mixed de-
ciduous-coniferous, temperate coniferous and
boreal forests. Following tropical forests, they
comprise the largest biotic carbon pool on the
planet (Amentano and Hett, 1980).
This report focuses on the temperate forest
regions of the world for three primary reasons:
1) the Siberian boreal forest of the former
Soviet Union contains the largest undisturbed
extent of temperate forest on the planet (Table
1); 2) most human-caused atmospheric carbon
is the result of industrial development in the
northern latitudes; and 3) many of the predic-
tive models discussed in the following pages are
species-specific and cannot be used effectively
in tropical regions due to the vast amount of
unknown and unmeasured species that charac-
terize the forests of those climates.
194
-------
B. ZYBACH
Table 1. Forest stands of selected European and Asian countries (after Armentano and Hett, 1980).
Ztountry
Finland
Sweden
European Economic
Comittee
Europe (except former
Soviet Union)
Japan
China
Former Soviet Union
Land area
(million ha)
33.70
41.10
152.60
457.00
100.00
956.00
2220.00
Forest area
(million ha)
20.56
23.50
30.92
169.00
25.10
100.00
765.00
Percent
Forest area
(coniferous)
81.00
82.00
42.00
51.00
42.00
75.00
Total timber
resources
(billion m3)
1.50
2.36
3.00
13.40
2.09
79.00
MITIGATING STRATEGIES
In an effort to slow or halt net increases of
carbon in the atmosphere, a number of strate-
gies have been proposed that involve the man-
agement of forests and forest products. A
common denominator of such strategies is that
they require substantial investments of capital,
time and land. Such investments would require
major policy commitments on the part of sev-
eral cooperating forested countries to be effec-
tive. For the temperate regions, agreement to
manage trees and tree products as carbon sinks
would have to be made between the United
States, Canada and the former Soviet Union to
be effective. In addition, China has the poten-
tial for implementing massive afforestation
projects on lands that were deforested centu-
ries in the past. European Economic Commu-
nity (EEC) lands are also a factor, but due to
large, affluent populations and a stabilized for-
est base, these countries are probably better
suited for scientific, technological and political
contributions toward resolving climate change
problems.
Tree Planting
One of the most common strategies considered
for sequestering atmospheric carbon is the re-
forestation and afforestation of lands capable
of supporting temperate forests (Amentano
and Hett, 1980; Moulton and Richards, 1990;
Birdsey, 1991a). Management options typically
consider increasing the area of forest lands by
planting trees or allowing natural regeneration
to occur, thinning overstocked existing stands
to encourage more vigorous growth, planting
gaps in understocked stands and the manipula-
tion of species to encourage greater carbon
sequestering capabilities or to anticipate
changed environments.
Anticipated costs of such a strategy are ex-
tremely high. Moulton and Richards (1990)
estimate that a tree-planting program expected
to reduce net emissions of CO2 by only 10% for
the United States would require 71 million
acres and cost 0.7 billion dollars a year. The
average cost for each ton of sequestered carbon
rises from $9.72 at this level, to $17.91 (7.7
billion dollars per year) for a strategy designed
to reduce net emissions by 30%. The marginal
cost of a strategy to reduce net emissions by
56% rises to an average of $43.30. Rosenfeld
and Botkin (1990) estimate that to offset 100%
of current global fossil fuel consumption, we
would have to double the size of the world's
forests, an area that would be about five times
as large as the entire area of the United States.
195
-------
SELECTING A PREDICTIVE MODEL TO MEASURE THE EFFECTS OF CLIMATE CHANGE AND MANAGEMENT ACTIVITIES
There are other problems as well; we do not
understand how specific species of trees will
respond to climate change, or even know how
individuals or families within a population will
respond. McCreary et al. (1986) suggest that
Douglas-fir viability is significantly altered by a
few degrees of change in winter chilling tem-
peratureswhenthetrees aredormant. Squillace
and Silen (1962) document the great amount of
growth andsurvivalvariationbetween different
races of ponderosa pine planted in experimen-
tal plots throughout their range in 1926.
Aspects of large-scale tree planting projects
that were not discussed in detail within the
reviewed literature include potential extinc-
tions and extirpations of wildlife populations,
and differences in net CO2 emissions caused by
plantation locations (trees planted in urban
environments can reduce fuel uses related to
seasonal heating and cooling requirements of
local human populations, while those located
near industrial developments may experience
poorer health and greater mortality).
Prescribed Fire/Fire Suppression
A significant body of literature exists detailing
the use of prescribed fire in reducing the occur-
renceofperiodicstandreplacementfires(Olson,
1981; Reid and Oechel, 1984, Bergeron and
Brisson, 1990). By removing dead materials at
periodic intervals and suppressing the develop-
ment of ladder fuels, the intensity, range and
frequency of wildfires can be reduced. Such
strategies may become critical if, as projected
by several researchers, the occurrence of global
warming may cause an increase in catastrophic
fires (Graham et al., 1990).
Although prescribed fire is not usually defined
in terms of wildfire control or suppression, that
would prove to be its primary function as a
carbon-sequestering strategy. Auclair (1990)
relates government fire suppression policies to
the capacity of North American forests to func-
tion as an effective carbon sink over the past
several decades. Some of his conclusions may
be debated because the relatively short time
that he analyzes does not take into account the
fuel-clearing catastrophic fires of the mid- 1800s,
the large-scale clear cutting that took place
following World War II, or the relationship
between climate trends and forest fires that
have developed over the past three centuries.
Wood Product Substitutes
The substitution of wood products as a carbon
sequestering strategy is bilateral. By substitut-
ing biomass fuels for fossil fuels, the primary
source of human-caused carbon additions to
the atmosphere is eliminated. By substituting
non-carbon-based products for wood products,
forest harvesting can be reduced. For instance,
the replacement of newsprint by electronic mail
eliminates the use of one of the fastest decom-
posing forest products. Conservation of exist-
ing products by careful storage, reduced use, or
re-cycling is another form of substitution that is
aligned with this strategy.
Wood Preservation Technologies
A basic limitation with most models developed
to track carbon cycling is that they don't ad-
equately assess the "shelf life" of carbon con-
tained in wood products derived from forests.
Often products are simply divided into two or
three basic categories such as paper products
and construction materials, and arbitrary val-
ues are then assigned. A slightly more sophis-
ticated approach to this problem is provided by
Kurz et al. (1990).
Little information exists regarding the life of
wood products after they are processed and
distributed. Preservation can be accomplished
by a number of methods, including distribution
patterns, manufacturing specifications (includ-
ing species selection) and chemical treatment.
Hermann (1976) quotes William Wright when
196
-------
B. ZYBACH
he states that "the Comstock Lode can in truth
be called the tomb of the forests of the Sierra."
It would be interesting to see how much these
mining timbers have decomposed since Wrights
work first appeared in 1876. Pioneer settlers in
the Pacific Northwest were known to prefer
"yellow" (old-growth) Douglas-fir shakes for
home andbarnconstructionbecause they "lasted
longer than cedar."
A primary problem in determining wood prod-
uct life, one shared with chemically-treated
products as well, is the great amount of vari-
abilities encountered following sale and use.
Fire history, local climates and micro-climates,
maintenance schedules and methods, etc., all
profoundly affect the length of time in which
carbon is stored in construction materials, tele-
phone poles, pilings and other long-termuses of
wood.
W.K. Ferrell (Professor Emeritus, OSU Col-
lege of Forestry) is currently engaged in re-
search that attempts to track the life of forest
products in the Douglas-fir region (Harmon,
personal communication, 1991). Such research
should reveal sources of information and de-
velop a methodology that will be of use to other
regions.
Log Banks
One strategy for sequestering carbon that might
be considered is the creation of "log banks,"
essentially the purposeful storage of harvested
wood products in environments that inhibit
decomposition. Biomass could be stored un-
derwater, underground, or in an arctic climate
for decades or centuries. There are two basic
advantages to this strategy. First, dead trees or
underproductive forestlands could be harvested,
allowing for reforestation with a more efficient
cover. Second, future generations could draw
directly from these banks, reducing their need
for harvesting green trees.
USING MODELS TO ANALYZE
PREDICTIONS AND STRATEGIES
Climate changes have effects at a variety of
spatial and temporal scales (Figure 1). Avail-
able models are generally capable of address-
ing problems at only one given scale. An
f Billions
Yrs. j
'•Millions
Millennia
Centuries
Decades
Years
Seasons
Days
Seconds
Figure 1. Ecological scales of
I GLOBE |
§/r
c I CONTINENT]
0>
iREGIQNl
1 "'"•> ' 1 ECOSYSTEM!
/
IPROCESSI
2 2
mm m Ha Basin Region Continent- Planet
3 — Communication Gaps
time and space (after Pastor and Post, 1986).
197
-------
SELECTING A PREDICTIVE MODEL TO MEASURE THE EFFECTS OF CLIMATE CHANGE AND MANAGEMENT ACTIVITIES
Stochastic models are most often used to pre-
dict changes in forest range, structure, or car-
bon-sequestering capabilities in response to
alterations in the climate, but are limited by a
lack of information as to how these processes
work. For instance, very little is known as to
how organic carbon in forest soils has changed
over time (Pastor and Post, 1986). Similarly, it
is impossible to predict how changes in average
global temperature will affect planetary cloud
formations; lower elevation clouds will tend to
cool the earths surface, while higher elevation
clouds will likely exacerbate warming trends
(Weiner, 1991).
additional problem is the gap that exists be-
tween each of the different scales. Existing
models are incapable of bridging these gaps.
The development of computerized forest pre-
dictive models is in its infancy. At the present
time, there are no models that can accurately
reflect the present.or reasonably predict the
future on a regional basis. There are three
primary reasons for this condition: 1) scientists
have only recently started to unravel and under-
stand the complex interdependencies that exist
between the living and non-living components
of the biosphere; 2) computers are technologi-
cally incapable of storing or processing the vast
amount of data necessary to present those in-
terdependencies; and 3) the future is unpredict-
able.
The usefulness of forest predictive models, so
far as climate change is concerned, lies in their
ability to provide a range of possible conse-
quences. Such models can be grouped into two
basic categories: stochastic and non-stochastic.
Stochastic models are those that can accommo-
date random or conjectural events, and thus will
produce a number of theoretically possible
alternatives. It is up to the analyst to then
choose an alternative that most closely re-
sembles a desired result, or else simply average
the total of a number of computer runs to
moderate projections of the more variable fac-
tors. Non-stochastic models tend to apply a set
of given empirical equations to a specifically
defined inventory of values. They are most
often used to determine such things as timber
cutting schedules and anticipated growth rates
for stands of trees, rather than successional
responses to possible or unanticipated changes
to the environment. Non-stochastic models can
also be made to respond to conjectural changes
in temperature, available moisture, or atmo-
spheric CO2 but this process usually involves
the introduction of an entirely new set of equa-
tions.
Predictive models can be used to measure the
impact of climate change by considering the
individual variables most likely to affect forest
growth: CO2, temperature, precipitation and
radiation. Again, a primary problem is our lack
of information. For instance, commercial green-
house operators will sometimes increase the
growth of plants by subjecting them to greater
levels of CO2, but it is not known whether
mature trees will also respond in a like manner,
whether plants can become acclimated to such
changes over time, or if available nutrients
would eventually become depleted through in-
creased levels of plant growth.
The problem becomes more complicated the
more these processes are generalized, particu-
larly over time and space. For instance, a single
temperature event, a region-wide cold snap
occurring over a few hours time in November,
1955, had a severe impact on the viability and
volume of a number of races of 30-year-old
ponderosa pine that had been planted in sepa-
rate plots throughout the region. The results of
this single event were still noticeable during
research measurements taken in 1986 (Silen,
personal communication, 1991). Other prob-
lems include incomplete data regarding stand
fire histories, the impacts of industrial pollut-
ants upon localized tree populations, a lack of
knowledge regarding intra-species genetics and
198
-------
a lack of sophistication regarding the modelling
of biomass decay rates (standing snags vs.
downed logs, north slopes vs. south slopes, etc.).
Despite the limitations posed by predictive
models, they remain our best tool for projecting
the potential impactsuponglobalforests caused
by climate change. They are also our best tool
for measuring the potential results of imple-
menting strategies designed to mitigate such
change.
CURRENT MODELS
There are a number of predictive models being
developed that are capable of measuring pos-
sible impacts of climate change on temperate
forests. A recent technical review classified
several such models according to "levels of
resolution" (Agren et al., 1991).
Physiologically-Based Models
The greatest level of resolution is provided by
models in which plant processes can be de-
scribed in biochemical terms. Predictions of
individual plant responses to environmental
changes can be modelled through an under-
standing of such processes as carbon uptake
through photosynthesis and water-vapor ex-
changes through transpiration and respiration.
The objectives of this approach are to explain
and predict the functioning of plant communi-
ties from a micro-environmental level. Several
forest models exist in this category, including
BACROS, FORGRO, MAESTRO, BIOMASS
and FOREST-BGC. A primary problem with
these models is how to scale them from a plant
physiology level to an ecosystem, regional, or
global level (Figure 1).
Population Models
This level of resolution simulates tree growth as
it is affected by competition between individu-
als and population dynamics. These models
apply to stands of single species and multiple
B. ZYBACH
species, describing volumes spatially, individu-
ally (by tree) through time, or by summarizing
variables (total number of trees, etc.). All
models in this category consider recruitment
(new seedlings), individual growth, factors lim-
iting such growth, mortality, stand structure and
spacing. Most models in this classification are
limited by the assumption that climate is as-
sumed to be the same for an entire region.
Ecosystem Models
At this level of modelling, whole plants or major
plant components are integrated interactively
with the environment, particularly with the soil.
Typically, all green plant biomass is included in
a single compartment. Objectives of these
models include the ability to simulate ecosys-
tem responses to such abiotic drivers as light
intensity, soil water and soil temperature. Dif-
ferent management option impacts are able to
be measured at the ecosystem level. Ecosystem
models can be divided into short (one day or
less) and long (one month to one year) time
steps. Short-time step models require the input
of driving variables on a daily basis and gener-
ally simulate ecosystem dynamics for 2-10 year
time periods, while long-time step models are
designed to predict ecosystem dynamics over
periods of time that can be measured in de-
cades. Short-time step models have been used
largely to predict crop and grassland responses
to such climatic changes as CO2 enhancement,
supersonic-transport-induced weather and hail
suppression. Long-time step models, such as
GEM and CENTURY have been designed to
simulate long-term (10-100 year) responses of
grasslands and forests to differing management
practices and possible climate changes.
Regional Models
This group of models operates at the lowest
level of resolution, with plant populations com-
bined into biomes. At the present time, there
are no reliable models that have been devel-
199
-------
SELECTING A PREDICTIVE MODEL TO MEASURE THE EFFECTS OF CLIMATE CHANGE AND MANAGEMENT ACTIVITIES
oped at this level, although the FOREST-BGC
model is being modified to accommodate re-
gional applications. It is expected that this
model will be validated for general use in about
two years (Running and Gower, 1990).
Brief Description of Select Models
The effects of climate change take place at the
molecular level of individual organisms. How-
ever, it is the cumulative effect, on a regional or
global basis, with which we are most concerned.
The following models were selected for further
description based upon their degree of devel-
opment, their general availability and the fact
that they can be arranged in a hierarchical
fashionfromphysiologicalprocesses to popula-
tion dynamics to ecosystem responses. Theo-
retically, such responses may then be analyzed
in a cumulative and interactive manner to de-
termine regional and global results.
FOREST-BGC (Running)
This is a physiologically based model that is
currently beingupdated. The newer model was
being withheld from general use at the time of
this report and unavailable for further descrip-
tion. Particular attention is paid to water avail-
ability, with descriptions of rainfall, intercep-
tion by canopies, soil evaporation, etc. Re-
quired inputs include daily weather data (tem-
perature, precipitation and light) and such soil
physical characteristics as rooting depth and
water retention functions. The model allows
the examination of temperature and precipita-
tion, with carbon outputs allocated between
three vegetation pools, a detrital pool and a
nondetrital soil pool. The ecosystem objective
is to measure forest growth and water balance,
as evidenced by leaf area production (LEA)
and retention. Photosynthetic rates and tran-
spiration are dependent upon radiation, but
this model doesn't have a mechanism to ac-
countfor increased cloudiness. Prediction times
are limited to months and years; there is no
current method of generalizing predictions to a
regional basis. The earlier version of this model
is described by Running and Coughlin (1988);
the current model is being modified and tested
as outlined in Running and Gower (1990).
JABOWAII(Botkin)
This is a population model with stochastically
simulated processes. Predictions are usually
based upon the mean of several (30-40, or
more) model runs. Simulation is based upon
the birth, growth and death of individual trees
in representative plots. A weakness associated
with averaging runs is that extreme events,
which may significantly affect forested areas,
are averaged out of existence (Graham et al.,
1990). This model is being marketed for $250
in a version that operates on IBM personal
computers (Inouye, 1991). It is of particular
value to studies of temperate forests, both be-
cause of its availability and because of the
principal author's familiarity with forest model-
ling, boreal forests and remote sensing (Botkin
et al., 1972; Botkin et al., 1984; Botkin and
Simpson, 1990). Jabowa II is specifically de-
scribed in Botkin and Nisbet (1991).
CENTURY (Parton)
This is a long-time step ecosystem model that
has been developed to simulate carbon, nitro-
gen, phosphorous and sulphur dynamics for the
plant-soil system inresponse to climate changes
and management practices. Monthly time steps
are used to predict regional trends regarding
ecosystem processes and responses over time
periods that vary from 10-500 years. Driving
variables are daily temperature records, pre-
cipitation and light. Outputs include five car-
bon pools, two allocated to soil and three allo-
cated to vegetation. It features a complex soil
process model with feedback between nitrogen
availability and production.
200
-------
B. ZYBACH
U.S. Carbon Budget Model (Birdsey)
This is a collection of five USD A Forest Service
models: TAMM, ATLAS, FORCARB,
HARVCARB and BICARB. It is essentially an
inventory series that would require the intro-
duction of an entirely new set of equations to
accommodate environmental changes (Birdsey,
1991b). A secondary weakness is that these
models are entirely dependent upon Forest
Service data and are nearly useless for applica-
tions regarding temperate forests outside of the
United States.
CONCLUSIONS
Based on the information presented herein, the
following conclusions are appropriate:
• Selection of a model is dependent upon the
temporal and spatial scale of the question that
is being asked;
• None of the models in current use has demon-
strated an ability to make accurate or reliable
projections;
• Information gaps exist that need to be ad-
dressed to increase model projection accuracy;
• Communication gaps exist between models
that operate at differing scales; and
• The need for predictive models that can pro-
vide insights to biospheric responses to climate
change and to large-scale conifer forest distur-
bances is critical.
REFERENCES
Agren, G.I., R.E. McMurtrie, W.J. Parton, J. Pastor and
H.H.Shugart. 1991. State-of-the-art of models of produc-
tion-decomposition linkages in conifer and grassland eco-
systems. Ecological Applications.
Armentano, T. V. and J. Hett, editors. 1980. The Role of
Temperate Zone Forests hi the World Carbon Cycle-
Problem Definition and Research Needs. CONF-7903105
UC-11: USDE Office of Environment.
Auclair,A. 1990. Forest fires as a recent source and sink
of CO2 at mid-high northern latitudes (unpublished re-
port furnished to the author by Ted Vinson) Draft: Issue
Paper 1: EPA Office of Environmental Processes and
Effects Research, August.
Bergeron, Y. and J. Brisson. 1990. Fire regime hired pine
stands at the northern limit of the species' range. Ecology,
August.
Birdsey, R A. 1991a. Prospective changes inforest carbon
storage from increasing forest area and timber growth.
USDA Forest Service, Washington, D.C., March. In Press.
Birdsey, R A. 1991b. U.S. carbon budget model, (unpub-
lished report provided by Hermann Gucmski of the
CorvalUs EPA), U.S. Forest Service, Washington, D.C.,
March.
Botkin, D.B., J.F. Janak, and J.R. Wallis. 1972. Some
ecological consequencesof a computer model of forest
growth. Journal of Ecology, November.
Botkin, D.B., J.E. Estes, R.M. MacDonald, and M.V.
Wilson. 1984. Studying The earths vegetation from space.
BioScience, September.
Botkin, D.B. and L.G. Simpson. 1990. Biomass of the
North American boreal forest: a step toward accurate
global measures. Biogeochemistry 9.
Botkin, D.B. and RA. Nisbet. 1991. Forest response to
climatic change: effects of parameter estimation and
choice of weather patterns on the reliability of projections.
Climatic Change.
Graham, R.L.,M.G. Turner, and V.H. Dale, 1990. How
increasing CO2 and climate change affect forests.
BioScience, September,
201
-------
SELECTING A PREDICTIVE MODEL TO MEASURE THE EFFECTS OF CLIMATE CHANGE AND MANAGEMENT ACTIVITIES
Harmon, M. 1991. Personal communication, August 30.
Hermann, R.K. 1976. Man and forests~a prodigal
relation. In: Forests And Future Resource Conflicts.
OSU School of Forestry, Corvallis, Oregon.
Inouye, D.W. 1991. Forest growth model Bulletin Of
The Ecological Soc. of American, June.
Kauffman, J. B., KM. Till, and R.W. Shea. 1991. The
biogeochemistry of deforestation and biomass burning.
OSU Department of Rangeland Resources, Corvallis,
Oregon. In press.
Kurz, WA., P J. McNamee, and T.M. Webb. 1990. The
annual carbon budget of the Canadianforest sector. Phase
I: Assessment of the current net balance. DSS Contract
No.: 4Y080-9-0285/01-XSG, Forestry Canada Organiz-
ing Committee, Vancouver, B.C., January.
McCreary, D., D.P. Lavender, and R.K. Hermann. 1986.
Predicted global warming and Douglas-fir chilling re-
quirements (unpublished report provided to the author by
Richard Hermann), Corvallis, Oregon.
Moulton, R J. and KR. Richards. 1990. Costs of seques-
tering carbon through tree planting and forest manage-
ment in the United States. General Technical Report
WO-58: USDA Forest Service.
Olson, J.S. 1981. Carbon balance in relation to fire
regimes. In: Fire Regimes And Ecosystem Properties.
GeneralTechnicalReportWO-26:USDAForest Service.
Pastor,!, and W.M. Post. 1986. Influence of climate,soil
moisture,and succession on forest carbon and nitrogen
cycles. Biogeochemistry 2.
Reid, C. and W. Oechel. 1984. Effect of shrubland
management on vegetation. In: Shrublands in California:
Literature Review and Research Needed for Manage-
ment. Water Resources Center No. 191, Davis California,
November.
Rosenfeld, A.H. and D.B. Botkin. 1990. Trees can
sequester carbon, or die and amplify global wanning:
possible positive feedback between rising temperature,
stressed forests, and carbon dioxide. PhysicsAndSociety,
April.
Running, S.W. and J.C. Coughlin. 1988. A general model
of forest ecosystem processes for regional applications in
hydrologic balance, canopy gas exchange and primary
production processes. Ecological Modelling.
Running, S.W. and S.T. Gower. 1990. FOREST-BGC, a
general model of forest system ecosystemprocesses for
regional applications II. Dynamic carbon allocation and
nitrogen budgets. In press.
Silen, R. 1991. Personal communications.
Squillace, A. E. and R.R. Silen. 1962. Racial variation in
ponderosapine. Society of AmericanForesters, Corvallis,
Oregon.
Weiner, J. 1991. The next one hundred years. Bantam
Books, New York, N.Y.
202
-------
ASSESSING THE IMPACT OF CLIMATIC
STRESS ON FOREST PRODUCTION
John Runyon, Richard H. Waring, and Richard W. McCreight
ABSTRACT
of the United States. Associated with this transect is a wide array of climates. Measurements made across the transect are
serving to test principles on how climate constrains forest productivity. Previous research has demonstrated a linear
relationship between intercepted photosynthetically active radiation and the production of dry matter by vegetation This
^otionshipdidnotholdforthevarietyofforeststandtypesfoundalongthetransecLltwashypothesizedthatthreemaior
chmatic variables were constraining productivity: (1) freezing temperatures; (2) vapor pressure deficits; and (3) drought
^sevanableswereincorporatedintoasimplemodelthateffectivelyreducesannualinterceptedphotosynth^
radiation to account for the environmental constraints. With this information we gauge the relative importance of each
variable and their integrated effect upon forest productivity.
INTRODUCTION
To accurately predict future rates of carbon flux
in the face of uncertain global change requires
that we understand controls onprimaryproduc-
tion. This paper explores a method to explain
differences in above-ground net primary pro-
duction (ANPP) across diverse landscapes by
applying simple relationships between plant
physiology and climate.
The research reported here is from Oregon
Transect Ecosystem Research Project (OT-
TER). The project is a joint Oregon State
University-NASA study emphasizing remote
sensing of ecosystem parameters. In support of
the remote sensing, the project encompasses a
large ground component involving the mea-
surement of a variety of environmental vari-
ables including above-ground biomass, produc-
tion and meteorological values.
STUDY SITES
Oregon provides an ideal location for testing
principles of forest ecosystem function. The
OTTER project focuses on six sites arrayed on
a 300 Km transect across Oregon. This transect
encompasses a wide range of climates and
forest types (Figure 1). The forest stands along
the transect display almost the complete range
of forest net primary production found in North
America (Gholz, 1982; Jarvis and Leverenz,
1984). The coastal site consists of two stands:
an old-growth forest stand of sitka spruce and
western hemlock (site 1), and a deciduous red
alder stand (site 1A). Moving inland along the
transect, the climate becomes increasingly more
continental. Site 4, at an altitude of about 1500
meters, is probably most analogous to the Bo-
real biome. It is characterized by very cold
winters, large amounts of snow, low annual
mean temperatures and abundant moisture
203
-------
ASSESSING THE IMPACT OF CLIMATIC STRESS ON FOREST PRODUCTION
Coast
Range
Washington
Pacific
Ocean
Major Vegetation Zones
Picea sitchensis
heterophvlla
m Transition and
* Subalpine zones
Pinus ponderosa
Juniperus occidentalis
— 46°
— 45°
— 44°
— 43°
— 42°
Idaho
California
I
123' 122°
121°
120°
Nevada
I I
119° 118°
117°
Figure 1. Map of the study area in Oregon showing the location of the study sites and the major vegetation zones.
during the growing season. The sites at the east
end of the transect (sites 5 and 6) are character-
ized by cold winters and hot, dry summers.
METHODS
To accurately gauge the environment along the
transect, each of the sites was equipped with a
meteorological station. Beginning in the sum-
mer of 1989, the stations provided hourly mea-
surements of temperature, precipitation, rela-
tive humidity and incoming short wave radia-
tion. Forest surveys of tree numbers and diam-
eters allowed us to compare biomass across the
transect. In addition, for each stand across the
transect, we measured both dry matter produc-
tion (woody biomass and foh'age) and the inter-
cepted fraction of photosynthetically active ra-
diation (IPAR). To determine ANPP, diam-
eter growth was measured on increment cores
from a random selection of trees at each site.
Percent IPAR was measured using an integrat-
ing ceptometer under the forest canopy. An-
nual IPAR was determined by using the stand
fraction of intercepted photosynthetically ac-
tive radiation (PAR) multiplied by the total
annual incident PAR.
RESULTS AND MODEL DEVELOPMENT
An extreme range of biomass is evident across
the transect: from 471 Mg/ha in the moist
coastal old-growth stand (site 1) to a low of
about 10 Mg/hain ajuniper stand at the dry east
end of the transect. Annual IPAR and ANPP
also varied widely across the transect (Figure
2).
Monteith (1977) has observed a nearly linear
relationship between dry matter production in
well-watered agricultural crops and the amount
of sunlight intercepted by the canopy. How-
ever, for the forest stands along the transect, the
relationship between ANPP and annual IPAR
is not linear (Figure 3). Obviously, annual
IPAR does not account for all of the variability
204
-------
J. RUNYON, R.H. WARING, AND R.W. McCREIGHT
25 -I
r 100
-so
-50
-40
-30
-20
o
H
w
3
H
EH
a
H
W
1A
SITE
Hgure 2. Trends of above-ground net primary production (left axis) and percent intercepted photosynthetically active
radiation (IPAR) (right axis) across the Oregon transect.
in ANPP observed across the transect. This
finding suggests that, in natural ecosystems, the
efficiency of converting solar energy into dry
matter production is a function of both the
amount of intercepted light and the environ-
mental constraints on photosynthesis.
To help explain the range of ANPP values
found along the transect, we examined the
major constraints on the utilization of photo-
synthetically active radiation. The term photo-
synthetically active radiation is somewhat a
misnomer. It is not the radiation that is active,
but rather it is green photosynthetic biomass.
Green biomass has widely variable rates of
photosynthesis, depending, largely, on environ-
mental conditions. Previous research has dem-
onstrated a relationship between patterns of
ecosystem productionand climate (Lieth, 1975).
For this study, we hypothesized that the rela-
tionship between IPAR and dry matter produc-
tion would primarily be affected by climate
controls over tree-level physiological processes.
0 50° 1000 1500 2000 2500
ANNUAL INTERCEPTED PAR (MJ/m2)
Figure 3. Estimates of annual intercepted radiation
(IPAR) were not linearly related to above-ground net
primary production across the Oregon transect.
205
-------
ASSESSING THE IMPACT OF CLIMATIC STRESS ON FOREST PRODUCTION
Table 1. Criteria for reducing intercepted photosyntheti-
callyactive radiation (EPAR) based onphysiological thresh-
olds applicable to all major tree species in Oregon.
-------
J. RUNYON, R.H. WARING, AND R.W. McCREIGHT
T ' r
1500 2000
UTILIZED ANNUAL IPAR(MJ/m2)
Figure 4. After accounting for physiological constraints to
photosynthesis, the estimated utilized annual IPAR was
closely related to observed above-ground net primary
production (ANPP). An old-growth forest (site 1) was
omitted from the regression.
EJ VPD
EZ DROUGHT
• FREEZING
Figure 5. The fraction of intercepted radiation that could
not be utilized by the various forest types because of
freezing temperatures, drought, or excessive vapor pres-
sure deficits (VPD) ranged from less than 10% at the
moist western hemlock/sitka spruce forest to as much as
77% at the dry juniper woodland.
REFERENCES
Gholz,H.L. 1982. Environmental limits on aboveground
net primary production, leaf area, and biomass in vegeta-
tion zones of the Pacific Northwest. Ecology 63:469-481.
Jarvis, P.O. and J.W. Leverenz. 1984. Productivity of
temperate, deciduous and evergreen forests. In: Encyclo-
pedia of Plant Physiology, New Series, Volume 12D.
Physiological Plant Ecology IV. Ecosystem Processes,
Mineral Cycling, Productivity, and Man's Influence.
Springer Verlag, New York, pp. 233-280.
Monteith, J.L. 1977. Climate and the efficiency of crop
production in Britain. Phil. Trans. R. Soc. Lond. B 281
pp. 277-24.
Lieth, H. 1975. Primary production of major vegetation
units of the world. In: Primary Production of the Bio-
sphere. Springer Verlag, New York, pp. 237-264.
Waring, R.H. and W.H.Schlesinger. 1985. Forest Ecosys-
tems: Concepts and Management. Academic Press, Inc.
Running, S.W. and J.C. Couglan. 1988. A general model
of forest ecosystem process for regional applications. I.
Hydrologic balance, canopy gas exchange and primary
production processes. Ecological Modeling 42:125-154.
Ryan, M.G. 1991. A simple method for estimating gross
carbon budgets for vegetation in forest ecosystems. Tree
Physiology 9:255-266.
207
-------
-------
VEGETATION FEEDBACKS AND THE PREDICTION
OF FUTURE FLUXES OF THE GLOBAL CARBON CYCLE
Robert A. Nisbet and Daniel B. Botkin
ABSTRACT
Investigations of climate dynamics and the carbon cycle have been pursued independently of one another, although the two
are certainly interrelated. In the long term, interrelationships between the two must be considered. This paper suggests ways
thesetwoglobalprocesses may interact. Projections of climate change were taken from general circulation models andused
as input for the JABOWA-II computer model of forest growth. A transient from 1980 to 2070 of greenhouse gas conditions
assumed a "business as usual"per capita production of CO2 in the future. The transient was used to modify a local weather
record to project changes in temperature and precipitation in specific locations. Projected temperatures changed; however,
almost all warming occurred in the middle of winter and had little effect on tree growth in the model. During the growing
season, little change occurred. Results suggest there will be important geographical differences in the response afforests to
global warming.
INTRODUCTION
Concern with the possibility of global warming
has led to two pathways of analysis:
1. Investigations of climate dynamics, with the
goal of projecting future changes in climate
under global warming, because of changes hi
atmospheric concentration of greenhouse gases;
2. Investigations of the carbon cycle, with the
goal of projecting the fate of carbon dioxide and
methane added to the atmosphere because of
human activities.
These two lines of research have been pursued
independently of one another, although the two
are certainly interrelated. Changes in climate
can affect carbon storage in the biota, in soils
and in the ocean, while changes in carbon
storage can affect climate. The separation of
the two lines of investigation is understandable.
Both topics, climate dynamics and the global
carbon cycle, are extremely complex, and our
209
understanding of each is meager.
We have had neither time nor resources to
pursue the two topics in combination. But in the
long term, we must consider the inter-relation-
ships between these two global processes. The
purpose of this paper is to suggest a few ways
that these two global processes might interact
and to motivate scientists to begin to study these
interactions.
EVIDENCE FOR A MAJOR DECLINE IN
CARBON STORAGE IN FORESTS
During the past four years, we have been in-
volved in the attempt to project the possible
effects of global warming on forests, under
support from the Office of Policy, Planning and
Analysis of the U.S. EPA. In this work, we have
taken the projections of climate change from
general circulation models and used these as
input for the JABOWA-II computer model of
forest growth. This is the most recent version of
the JABOWA model of forest growth, origi-
-------
VEGETATION FEEDBACKS AND THE PREDICTION OF FUTURE FLUXES OF THE GLOBAL CARBON CYCLE
tr
n
E
CD
*£.
BIOMASS (
Figure 1. Total biomass
80-
60-
40-
20 (
0
19
BOUNDARY WATERS CANOE AREA
VIRGINIA, MINNESOTA
Soil Depth = 1 .0 m Water Table Depth = 1 .2 m
BIOMASS
OO KinPUAl PI IMATF
• • GISS TRANSIENT A CLIMATE
80 2000 2020 2040 2060
i
YEAR
of a balsam fir forest in the Boundary Waters Canoe area of northern Minnesota.
nally developed in 1970 by Botkin et al. (1973).
A forthcoming book provides a complete de-
scription of this new version (Botkin, 1992).
Others have used descendants of this model
(FORET, LINKAGES) (Solomon et al., 1984;
Solomon, 1986; Solomon and West, 1987; Pas-
tor and Post, 1988) for the same purpose.
Another, independent approach to the same
problem has been taken by those who analyze
pollen deposits. They have projected future
steady-state distributions of vegetation, based
on past correlations between climate and veg-
etation (Davis and Botkin, 1985; Zabinski and
Davis, 1988). Results from this approach are
consistent with our results using forest-growth
models, although both approaches provide
unique information.
In our work, we have used the projections of
temperature and precipitation from the NASA
GISS general circulation model for a transient
from 1980 to 2070 greenhouse gas conditions.
The transient used is known as "transient-A"
and assumes a "business as usual" per capita
production of carbon dioxide in the future. We
applied this transient to modify a local weather
record to project changes in temperature and
precipitation in specific locations.
Because carbon is a relatively constant percent-
age of biomass (about 45% by dry weight), the
carbon storage would decline at the same rate
as shown in Figure 1, if these projections occur.
This suggests that the carbon content of a
maturebalsam fir-dominated stand in the south-
ern part of the North American boreal forest
would undergo a decline in 40 years (by year
2020, starting in year 1980) to approximately
23% compared with projections under normal
climate. This loss in carbon storage in live trees
would create a positive feedback, further in-
creasing the rate of global warming.
EXTRAPOLATIONS FROM ONE STUDY,
USING OUR BIOMASS ESTIMATES, TO A
LARGER SCALE
What are the Large-scale Implications of this
Projection?
In other work, also reported in this volume, we
have estimated the carbon content of the boreal
forests of North America to be 9.7 ± 2 Gt. If all
210
-------
RA. NISBETAND D.B. BOTKIN
IT
m
E
O)
V
BIOMASS (1
Erikit Field Camp
Near Ust'-Nera, USSR
Plot 1 1
25 -r
20-
15-
10-
1
5-
0
19
Soil depth = 2.0 m Water tabl« = 1 .2 m Nttrofl«n = 50 kg/ha
BIOMASS
LARCH
O Normal climate, DTMIN = 0.10
• Global warming, DTMIN = 0.10
A A Normal climate, DTMIN = 0.16
i I i I i I
90 2010 2030 2050
I
i
YEAR
Figure 2. Larch biomass at the Erikit Field Camp, near Ust'-Nera, Siberia, USSR.
the North American boreal forest declined at
the same rate as in Figure 1 for a balsam fir
dominated site in the southern part of this
forest, 76.8% of the carbon in live above-ground
biomass would be released in 40 years. This
would represent an average release of 0.19 Gt
of biomass/yr, or 0.08 Gt C/yr.
Extrapolating these carbon storage estimates,
we have estimated that the total carbon content
of the boreal forests of the world would be
approximately 27 Gt. If the entire boreal forest
declined in 40years by the same percentage, the
annual rate of carbon release would be 0.52Gt/
yr. Because the estimated total production of
carbon dioxide from the burning of fossil fuels
is about 5 Gt/yr, the amount of carbon release
from boreal forests might lead to a significant
increase in global warming. Following this pe-
riod of decline, the forest would become a sink
for carbon.
Can We Accept an Extrapolation from One
Site to the Entire Boreal Forest?
But can we expect all forests to respond as this
single site? This is unlikely; however, policy
makers, given political pressures and budget
limitations, face strong pressures that are ex-
erted to use an absolute minimum number of
sites and extrapolate from these to a global
projection. A desire for newness and novelty
reinforces these pressures.
Fortunately, with support from the EPA, the
U.S. Forest Service and the Pew Charitable
Trusts, we have considered some additional
sites. One site is a larch-dominated forest near
Ust'-Nera in eastern Siberia. The results are
surprising because climatologists have pointed
out that the projections from general circula-
tion models suggest the greatest changes in
temperature will occur at mid and high lati-
tudes. From that statement, one would expect
a comparatively large change in temperature in
a.Siberian boreal forest, which is at a high
latitude. One would expect the change to be at
least as great as that for northern Minnesota,
which is at a lower latitude. However, accord-
ing to the JABOWA-II forest model, if the
projections of the NASA GISS transient-A are
correct, nothing will happen to these forests
(Figure 2). This is in striking contrast to our
projections for a balsam fir site in the Superior
211
-------
VEGETATION FEEDBACKS AND THE PREDICTION OF FUTURE FLUXES OF THE GLOBAL CARBON CYCLE
30
MEAN MONTHLY TEMPERATURE FOR
UST'-NERA, ASSR, USSR
• Normal Climate
Global Warming Cllmata
-60
2051
2053
2055
2057
2059
2061
YEAR
Figure 3. Mean monthly temperature projected by the NASA GISS Transient-A climate model for the years 2051-2061.
National Forest. Why is this? When we ob-
tained this result, we were puzzled, so we
graphed the projected change in temperature
for this area from, the GISS model.
True, the projected temperature changes, but
almost all the warming takes place in the middle
of winter, when the temperature warms from
-40°Fto-350F(Figure3). This wouldhave little
effect on tree growth (and has no effect on tree
growth in the model). During the growing
season, little change occurs.
This result suggests that there willbe important
geographic differences in the response of for-
ests to global warming.
Finally, we would like to consider two more
examples, for two sites dominated by sugar
maple near Antigo, Wisconsin, that have differ-
ent soil conditions. We have made projections
at the request of the Wisconsin Department of
Natural Resources. These sites are in a transi-
tion between the boreal forest to the north and
northern hardwoods to the south.
On the firstsite, sugar maple shows little change
in biomass under global warming conditions
(Figure 4), while on the second site, sugar
maple increases in biomass (Figure 5). The
second site is wetter than the first site, too wet
under present climatic conditions for optimum
tree growth. Global warming dries the soil and
improves the site conditions. Furthermore, this
site is at the northern end of the range of sugar
maple, and warming also benefits that species.
Figures 4 and 5 show that there will be soil- and
species-specific differences in responses.
CONCLUSIONS
What General Conclusions Can We Reach
with this Information?
The first conclusion is that if we have to base
policy simply on projections available now, we
would have to say most sites in the boreal forests
will undergo an initial, large decline in carbon
storage, and our best estimate is that the boreal
forests will provide a significant source of car-
bon dioxide to the atmosphere, further acceler-
ating global warming.
212
-------
RA. NISBET AND D.B. BOTKIN
CT
n
CD
to
I
o
m
60
50
40
30
20
S = 0.3m
ANTIGO, MICHIGAN
SUGAR MAPLE
PLOT 2
DT=1.0m T = 200 mm/m
N g 70 kg/ha
O
BIOMASS
-O Sugar maple, normal climate
-• Sugar maple, global warming
1990
2010 2030
YEAR
2050
igure 4. Sugar maple biomass on a relatively dry site near Antigo, Michigan.
tr
n
CO
(ft
<
O
m
50
404-
30
20 4-
ANTIGO, MICHIGAN
SUGAR MAPLE
PLOT 3
Soil depth = 0.4 m Water table = 1.0 m Nitrogen = 85 kb/ha
BIOMASS Texture = 21 Omm/m
O O Sugar maple, normal climate
• • Sugar maple, global warming
1990
2010 2030
YEAR
2050
Figure 5. Sugar maple biomass on a relatively moist site near Antigo, Michigan.
213
-------
VEGETATION FEEDBACKS AND THE PREDICTION OF FUTURE FLUXES OF THE GLOBAL CARBON CYCLE
But more importantly, our second conclusion is
that our results suggest that there should be a
program initiated to obtain projections for a
statistically reliable sample of sites, representa-
tive of the boreal forests. The idea for this
project, known as an "Early Warning Systemfor
Global Climate Change" grew out of our coop-
erative research with the USSR State Forest
Committee, under the auspices of the Center
for Resource Management, and with support
from the U.S.D.A. Forest Service and the EPA.
How Could this be Done?
1. From our biomass studies, we have demon-
strated that we know how to set up a statistical
sampling scheme thatgives a representative esti-
mate with an approximate 20% error (the 95%
confidence interval is within 20% of the mean);
2. At each site, make necessary initial conditions;
3. Project the effects of global warming for
these sites;
4. The result will provide the best projection
that we can do today of the large-scale response
of boreal forests to global warming.
We emphasize that these results show the im-
plications of our present understanding of glo-
bal warming and forest dynamics. Rather than
say that the projections would be "true" we
would say that these projections represent the
implications of what we know today.
We believe that this is an important project,
useful as one of the first steps in linking climate
dynamics and the global carbon cycle. Doubt-
less, other viable approaches could be sug-
gested for investigating the linkages between
the global carbon cycle and climate dynamics.
These, we believe, will be an important next
step in our attempt to understand how human
activities maybe influence the environment at
a global scale.
REFERENCES
Botkin, D.B., J.F. Janak, and J.R. Wallis. 1973. Some
ecological consequences of a computer model of forest
growth. J. Ecol. 60:849-872.
Botkin, D.B. 1992. Forest Dynamics: an Ecological
Model. Oxford University Press.
Davis, M.B. and D.B. Botkin. 1985. Sensitivity of cool-
temperate forests and their fossil pollen record to rapid
temperature change. Quaternary Research 23:327-340.
Pastor, J. and W.M. Post. 1988. LINKAGES: Response
of northern forests to CO2-induced climate change. Na-
ture 334:55-58.
Solomon, A.M.,M.L.Tharp,D.C. West, G.E. Taylor, J.W.
Webb, and J.L. Trimble. 1984. FORET: Response of
unmanaged forests to CO2-induced climatic change: avail-
able information, initial tests, and data requirements. U.S.
Dept. of Energy, Technical Report 009, p. 93.
Solomon, A.M. 1986. Transient response of forests to
CO2-induced climatic change: simulation modeling ex-
periments in eastern North America. Oecologia 68:567-
580.
Solomon, A.M.andD.C. West. 1987. In: The Greenhouse
Effect, Climate Change, and U.S. Forests, edited by W.E.
Shands and J.S. Hoffman. The Conservation Foundation,
Washington, D.C., pp. 189-217. They use climate projec-
tions from J.F.B. Mitchell. 1983. Q.J.R. Meteorl. Soc.
109:113.
Zabinski,C. and M.B. Davis. 1988. Hard times ahead for
Great Lake forests: a climate threshold model predicts
responses to CO2-induced climate change, pp. 5-1-5-19.
214
-------
CARBON BUDGET AND SUCCESSION DYNAMICS
OF CANADIAN VEGETATION
Josef Cihlar and Michael Apps
ABSTRACT
TheNorthemBiosphereObservationandModelmgExperimentisaresearchanddevelopmentefforttoobtainanimproved
understanding of the role of terrestrial vegetation in the total Earth system. The focus is on the vegetation of Canada, and
effective use of space observations is stressed. The project is a cooperative effort involving the Canadian Global Change
Program led by the Royal Society of Canada, universities, the Canada Centre for Remote Sensing, Forestry Canada, and
Agriculture Canada. As a 10-year program NBIOME has established strong links with the international Earth Observing
System which accepted the project for the EOS execution phase in January, 1991. This paper emphasizes the linkage bv
describing the use of remote sensing techniques in the overall NBIOME strategy.
BACKGROUND AND RATIONALE
In response to the Earth Observing System
(EOS) Announcement of Opportunity issued
by the U.S. National Aeronautics and Space
Administration in 1988, a proposal to study the
northern biosphere using satellite observations
and nested modelling was submitted by scien-
tists from several Canadian organizations, with
the Canada Centre for Remote Sensing as the
lead sponsoring agency. The principal objective
of the project called NBIOME (Northern Bio-
sphere Observations andModeling Experiment)
is to increase our understanding of the role of
terrestrial vegetation in the total Earth system
and its changes with tune in a changing environ-
ment. The focus of the project is the vegetation
of Canada and its four major biomes: forests,
agro-ecosystems, wetlands and tundra.
The project is based on four premises:
1. The knowledge of the distribution, character-
istics, and dynamic response of terrestrial veg-
etation is a necessary pre-requisite to the un-
derstanding of the behaviour of the total Earth
system, its component parts, and its changes
with time-both as a result of natural agents and
those caused by the human intervention.
2. Because of the dynamic character of this
living system and the broad spatial distribution,
satellite observations can serve as a principal
measurement technique, supplementedby other
methods as appropriate.
3. Models of vegetation behaviour/response
are a necessary component of the total ap-
proach. This follows from the fact that satellites
observe current conditions, but we are also
strongly interested in future changes that veg-
etation may undergo, as well as in the environ-
mental and socio-economic consequences re-
sulting from such vegetation changes. In addi-
tion, models are required to infer certain sur-
face parameters from remote sensing data.
215
-------
CARBON BUDGET AND SUCCESSION DYNAMICS
4. The present level of knowledge and under-
standing concerning terrestrial vegetation is
inadequate to determine its role in global pro-
cesses and the potential effect of global climatic
change. For example, a study by Tans et al.
(1990) suggested that land ecosystems at tem-
perate and northern latitudes must be substan-
tial sinks of atmospheric CO2 to balance the
global budget. Although the exact location of
such sinks is unknown at the present, it is likely
that a significant portion would exist in the
NBIOME study area.
Among the various techniques that might be
considered for the assessment and monitoring
of vegetation conditions, satellite remote sens-
ing offers a realistic possibility to obtain the
requisite data because of the large areal extent,
strong spatial and temporal dynamics, and lo-
gistical inaccessibility of the vegetation. Al-
though this statement is true globally, it is
particularly relevant for vegetation at northern
latitudes where population density is extremely
low, roads are few, and the growing season is
short. However, remote sensing at,northern
latitudes is hampered by strong cloud cover and
low sun angle, including long periods without
sunlight. Therefore, the development of accu-
rate and reliable techniques for using satellite
data will require substantial research effort.
GOAL AND OBJECTIVE
NBIOME addresses the role of terrestrial veg-
etation of Canada in the total global ecosystem.
Its goal is to obtain an understanding of the
behaviour of terrestrial ecosystems of Canada
that will allow the forecast of changes in their
structure and function resulting from global
environmental changes. To achieve this goal,
an observation and information system will be
developed pertaining to a family of landscape
and ecosystem models, backed by process un-
derstanding. The observations and process-
based models will be used to monitor, evaluate
and project the impact of global change on
boreal ecosystems including forests, agro-eco-
systems, wetlands and tundra.
NBIOME is bounded spatially by the landmass
of Canada that is presently vegetated or which
could be vegetated within the time horizon of
interest. The time horizon is defined as the next
50-100 years, a period over which significant
increase in the concentration of radiatively
active gases is expected and during which long-
term policies may have impact. The specific
project objectives are to assess the likely direct
and indirect interactive effects of global envi-
ronmental change on three aspects of Canada's
terrestrial ecosystems:
1. Disturbance regimes including fire, harvest,
insects, disease, and human intervention;
2. Vegetation change; including growth, regen-
eration and succession; and
3. The size of carbon pools and the net flux of
the important radiatively active gases between
the terrestrial ecosystems and the atmosphere.
NBIOME will encompass a variety of investiga-
tions ranging from process studies to modeling
and remote sensing observations. Modeling
will play a key role in various aspects of the
project, and it will draw on process understand-
ing obtained infield studies and on information
derived through remote sensing. From the re-
mote sensing perspective which is emphasized
in this paper, the following aspects will be
addressed.
1. Development of a vegetation classification
algorithmbased on satellite measurements as a
principal data source. This algorithm will be
used to produce a digital vegetation map of
Canada as a baseline for determining future
changes as well as for use in growth and succes-
sion models.
216
-------
2. Development of a phytomass model for the
vegetation of Canada to produce a map of total
phytomass as input into a growth model.
3. Development/adaptation of remote sensing
inputs into vegetation growth models. These
models should result in digital maps of Canada
showing gross primary productivity and net
change in carbon storage for different years
within the EOS time period.
KEY SCIENCE ISSUES
NBIOME focus has been formulated with ref-
erence to the principal cause-effect links and
feedbacks through which global environmental
change is expected to interact with northern
biomes within the time horizon of interest:
major impact of climate on changes in the
disturbance regime or on the types and rates of
ecosystem processes; ecosystem structure and
function affected by climate change, directly or
through disturbances; carbon fluxes and pools
modified as a function of ecosystem character-
istics; and feedback to climate from the ecosys-
tem through biophysical and biochemical path-
ways. The research agenda of NBIOME will
thus include the following issues.
1. Physiognomic characteristics of the vegeta-
tion of Canada and the spatial distribution of
carbon stored in vegetation. To provide reli-
able answers to basic questions of carbon cy-
cling and vegetation change, we plan to estab-
lish a baseline data base showing the distribu-
tion of vegetation types and biomass across
Canada. Such a data base will initially be
established with 1 km resolution, to be in-
creased to about 250 m in the late 1990s.
2. Annual carbon budget of the vegetation of
Canada. Because of the large area of Canada
and the vast boreal forest biome, Canadian
vegetation accounts for a significant fraction of
the carbon in the terrestrial biosphere. The
J. CIHLAR
carbon budget of Canada has significant uncer-
tainties, particularly at regional and local levels
(Apps and Kurz, 1991). Remote sensing tech-
nology offers the possibility of contributing to
more accurate estimates of above-ground car-
bon storage by vegetation than those provided
by spatially lumped models, and of inferring
accumulation and depletion rates.
3. Year-to-year changes in the carbon budget of
Canada. Re-measurement over several years is
necessary to establish a range of variation of
carbon accumulation and depletion. Once this
range is established it will be possible to evalu-
ate whether the changes are within the normal
range or appear to be abnormal. Changes in
disturbance frequency and distribution (e.g.,
fire), estimated with the help of satellite data,'
will be one of the key variable components of
the annual carbon budget.
4. Relationship of changes in the vegetation
composition to ecological changes, using veg-
etation change as an observable indicator of
ecological processes. These effects may occur
at the site/landscape, regional, and biome lev-
els. A database of Canadian vegetation based
on medium resolution satellite observations
will be established to assist inmonitoring changes
in vegetation distribution, with high resolution
satellite and field observations used where re-
quired for extra detail. With yearly updates, the
observed changes can be compared to expecta-
tions based on ecological and successional
models to look for abnormalities which may be
linked to anthropogenic factors or to climate
change. Regional interactions between vegeta-
tion development and vegetation change which
may be significantly affected by climate change
will be of considerable interest. Such mesoscale
processes are the critical bridge across the gap
between modeling causality at local levels and
at continental scales.
217
-------
CARBON BUDGET AND SUCCESSION DYNAMICS
METHODOLOGY
Because of the diversity of the vegetation of
Canada and the issues addressed in the project,
NBIOME will be implemented as a collabora-
tive effort of researchers from various organi-
zations. The strategy for project implementa-
tion is to address the overall objectives in the
context of each biome (forest, agro-ecosystems
including cropland and grassland, and wetlands/
tundra), using a spatial/temporal ecosystem
framework. It is postulated that the spatial/
temporal space consists of discontinuous clus-
ters in each dimension, separated by transition
regions across which the type or the relative
importance of key environmental variables
change.
Remote sensing will play an important role in
the project. Initial plans call for the develop-
ment of the following components.
Land Cover module will use medium
resolution, daily coverage data (the Advanced
Very High Resolution Radiometer (AVHRR)
before the launch of EOS, MODerate-Resolu-
tion Imaging Spectrometer (MODIS) after the
launch), together with ancillary data and Landsat
Thematic Mapper (TM) data if necessary, as
input into a vegetation classification algorithm.
The output of this module will be a vegetation
map of Canada with equal-area cell sizes, ini-
tially 1km x 1km and decreasing hierarchically
to 250mx 250m for MODIS data. An additional
output of this module at the end of the growing
season willbe temporal curves of greenness and
brightness.
Phytomass module will be developed to
produce a phytomass map of Canada. The
vegetation map from the first module will pro-
vide part of the basic information from which
the phytomass will be determined. The calcu-
lations will rely on published and unpublished
data to estimate thebiomass of cropland, grass-
land.forest, and tundra communities. For above-
ground biomass, satellite data will also be em-
ployed where appropriate. In case of forests,
the potential of radar data will be thoroughly
explored in conjunction with existing invento-
ries. For herbaceous plant communities, the
use of medium resolution optical data will be
investigated. An important issue is the distribu-
tion of below-ground organic matter. This as-
sessment will be based on vegetation cover map
and ancillary data, the latter varying with veg-
etation type (allometric equations, soil analy-
ses, previous field measurements, models).
Annual Organic Matter Budget module
will be designed to produce a map of gross
primary productivity for Canada. The vegeta-
tion map and seasonal growth parameters
(greenness, temperature, length of growing sea-
son, etc.) derived from satellite data will be
employed. After subtracting respiration losses
from the gross primary productivity, a net pri-
mary productivity will be obtained. Depletions
due to fire, harvesting, and insect and disease
damage will then be determined using ecologi-
cal data and models with satellite data, and the
resulting distribution will be translated into a
carbon budget map for Canada. This module
will be the most demanding aspect of NBIOME
from the viewpoint of satellite-derived inputs.
Succession module will use several models (for
forests, grassland, tundra, and wetlands) to
predict changes in the vegetation of Canada
resulting from observed or predicted climate
changes and from changes in management prac-
tices. The vegetation map, together with soil
maps and climate data, will be the principal
inputs for this module. The most challenging
aspect of this portion of the project will be to
understand and model the mesoscale interac-
tions which result in regional distribution of
vegetation types and their spatio-temporal dy-
namics.
218
-------
INFRASTRUCTURE
A key challenge in the NBIOME project is the
design of an integrated research program in
which generally small teams from numerous
geographically dispersed institutions make con-
tributions of the type, comprehensiveness, qual-
ity, and with the timing required to achieve
successful integration at the national level.
Furthermore, it will be necessary to ensure
continuity and consistency over a 10-year pe-
riod or longer and also ensure effective linkages
with other elements of the Canadian Global
Change Program (CGCP). From the NBIOME
team member's perspective, mechanisms and
tools must be put in place that will allow (i) easy
access to the required data, (ii) access to the
results of work by other team members (interim
or final); (iii) maximum efficiency and effec-
tiveness of invested time; and (iv) maximum
potential for true interdisciplinary teamwork.
To fulfil the above requirements, the NBIOME
Information System (NBIS) should have the
following functions:
(i) Provide information about, and access to,
data and products required by investigators
including final products, important raw and
intermediate data sets, and models in use by the
NBIOME team;
(ii) Support communication among investiga-
tors and the conduct of transdisciplinary re-
search in smaller specialty teams. This function
can be fulfilled through GCNet (Fisher et al.,
1991).
It may be feasible to establish a shared digital
database for the project. This database would
initially contain data required by all investiga-
tors (e.g., soil map of Canada) for their data
analysis and modeling research. More specific
information would be added later by individual
investigators to allow the overall work to pro-
ceed efficiently and on schedule. To reduce the
j. CIHLAR
impact of inevitable changes in personnel, all
inputs will be properly documented and regular
updates will be provided.
During its lifetime NBIOME will require a
number of field sites. Some will be set up for a
specific purpose and a limited time period,
others will be monitored on a seasonal and
annual basis over the duration of the program.
Satellite observations will be recorded regu-
larly for selected sites, while the field measure-
ment program may vary from one site to an-
other. Advantage will be taken of existing sites
or other field programs where possible. For
example, BOREAS sites and long-term CGCP
research observatories, when established, will
be included after the NBIOME team actively
participates in the definition of such observato-
ries.
PROGRESS TO DATE
Although the first EOS platform will not be
launched until late 1990s, research with EOS-
like data has already started and will accelerate
during the next few years. The principal data of
interest for vegetation research are provided by
the AVHRR. In addition to remote sensing
activities, an accelerated research on the forest
-climate interactions has commenced at For-
estry Canada. The forestry component of
NBIOME will also strongly benefit from the
proposedBOREAS project (Sellers etal., 1990;
Cihlar et al., 1991). A similar thrust is underway
at Agriculture Canada, with emphasis on the
role of agro-ecosystems in the budget of
radiatively active gases.
A concerted effort is being made to involve the
academic community in NBIOME. A coordi-
nated research proposal is currently in prepara-
tion, for submission to the Natural Sciences and
Engineering Research Council of Canada. It is
thus anticipated that the project team will be
fully formed in late 1992 and the coordinated
research activities will unfold in 1993.
219
-------
REFERENCES
Cihlar, J., M. Apps, R. Desjardins, B. Goodison, D.
Lcckie, and G. Sutler. 1991. Boreal Ecosystems-Almo-
sphereStudy (BOREAS). ReportNo. 5, Canadian Global
ChangeProgram, RoyalSociety of Canada and the Canada
Centre for Remote Sensing. In print.
Fisher, T. A., J. Cihlar, and R.Boudreau. 1991. Global
Change Network (GCNet). Proceedings of the 14th
Canadian Remote Sensing Symposium, 6-10 May, Calgary,
Alberta: 153-157.
Sellers, P J., J. Cihlar, M. Apps, B. Goodison, F. Hall, R.
Harriss, D. Leckie, E. LeDrew, P. Matson, and S. Run-
ning. 1991. BOREAS (Boreal Ecosystem-Atmosphere
Study): Global Change and Biosphere-Atmosphere In-
teractions in the Boreal Forest Biome. Science Plan. 31p.
Tans, P.P., Y. Fung, and T. Takahashi. 1990. Observa-
tional constraints on the global atmospheric CO2 budget.
Science 247 (4949): 1431-1438.
ACKNOWLEDGEMENTS .
The NBIOME project benefited from the involvement of scientists from numerous institutions. Contributions of F.J.
Ahern, C.S. Moiling and T. Fisher to the development of the concepts presented here deserve special acknowledgment.
The project presently is guided by NBIOME Science Steering Committee whose members are: Dr. M. Apps, Forestry
Canada; Dr. G. Baskerville, University of New Brunswick; Dr. J. Cihlar (Chair), Energy, Mines and Resources Canada;
Ms.B.Conway,NaturalSdencesandEngineeringResearchCouncil(Observer);Dr.R.Desjardms, Agriculture Canada;
Dr. C.S. Rolling, University of Florida; Dr. D. Parkinson, University of Calgary; Dr. J. Stewart, University of
Saskatchewan; Mr. G. Sutler (Observer, Canadian Global Change Program).
Note added in proof
The description of the NBIOME proj ect given in this paper is based primarily on the NBIOME Execution Plan submitted
to NASA in August, 1990. As part of project planning, two scientific workshops were organized since that time and a
science plan has been written by the Science Steering Committee. It is expected that the science plan will be published
in early 1993. A proposal for funding university researchers' involvement in NBIOME will be completed in the early 1993
for submission lo the Natural Sciences and Engineering Research Council of Canada.
220
-------
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON
STORAGE BY EXTRAPOLATION FROM LOCAL ECOSYSTEM MODELS
Anthony W. King
ABSTRACT
Ageneralprocedureforquantifyingregionalcarbondynamicsbyspatialextrapolationoflocalecosystemmodelsispresented.
In this procedure, Monte Carlo simulation is used to calculate the expected value of one or more local models, explicitly
integrating the spatial heterogeneity of variables that influence ecosystem carbon flux and storage. These variables are
described by empirically derived probability distributions that are input to the Monte Carlo process. The procedure provides
large-scale regional estimates based explicitly on information and understanding acquired at smaller and more accessible
scales. Results from an earlier application to seasonal atmosphere-biosphere CO2 exchange for circumpolar "subarctic"
latitudes (64° N-90°N) are presented. The results suggest that, under certain climatic conditions, these high northern
ecosystems could collectively release 0.2 Gt of carbon per year to the atmosphere. The results may be interpreted with respect
to questions about global biospheric sinksfor atmospheric CO2 Refinements of the existing application, and an application
to changes in carbon storage for a circumpolar "boreal" zone (53° N~64° N) are also presented.
INTRODUCTION
Understanding boreal forest and subarctic eco-
systems as part of the global carbon cycle re-
quires quantification of their carbon dynamics
over large heterogeneous spatial extents. To be
of global consequence, changes in their carbon
dynamics must occur over large areas. For
example, terrestrial ecosystems store carbon
when whole-system photosynthesis is favored
over ecosystem respiration. When the imbal-
ance (net ecosystem production, NEP) occurs
over a large region, that region becomes a net
sink for atmospheric CO2. If the region and
imbalance are collectively large enough, they
can influence changes in the global concentra-
tion of atmospheric CO2. Assuming an ecosys-
tem imbalance storing 0.150 kg C/m2/year,
(approximating NEP of old-growth Douglas fir
in Oregon (Grier and Logan, 1977)), a bio-
spheric sink of 0.6 Gt C/year (roughly 10% of
1988 fossil fuel emissions (Marlandetal., 1989))
would require an area of 4 x 106 km2 (roughly 2.5
times the area of Alaska).
In this paper, a method for quantifying regional
carbon dynamics by extrapolation from models
of local or site-specific ecosystem carbon dy-
namics is presented. The strengths of the ap-
proach are threefold. First, the regional esti-
mates are based directly on understanding and
information about ecosystem dynamics obtained
by observation and theory at smaller and more
accessible spatial scales. This information is
synthesized and codified in mathematical mod-
els. The models are explicit statements about
how the processes that control ecosystem car-
bon flux and storage respond to temporal and
spatial changes in the environment. The mod-
221
-------
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON STORAGE
els can be evaluated, revised and upgraded as
new information becomes available. Second,
the method explicitly incorporates empirically
defined spatial heterogeneity. Third, the ap-
proach is general and widely applicable to a
large number and variety of ecosystem models
and regions.
THE APPROACH: AN OVERVIEW
A variety of methods are available for translat-
ing local, relatively small-scale, models across
heterogeneous spatial extents (King, 1991).
However, some of these methods are excluded
by the nature of the models used to simulate
ecosystem carbon dynamics or by the large
areas involved. For example, it is difficult to
find ecosystem models written as explicit func-
tions of space where the ecosystem processes
are functions of spatial coordinates or functions
of variables that are themselves functions of
space. This effectively precludes spatial ex-
trapolation using partial differential formula-
tions or explicit analytical integrations over
space (King, 1991). Evenif thespatialfunctions
are available, exact analytical solutions are
generally difficult or impossible even for mod-
els that are simpler than most ecosystem carbon
models (King, 1991). Numerical solutions by
finite element methods or the like are then
required. This latter approach is common in
modeling watershed hydrology, but it is an
uncommon formulation in modeling ecosystem
carbon dynamics.
Large spatial extents also hinder extrapolations
by georeferenced modeling, whichlinks an eco-
system model with an underlying geographical
data base, often through a geographical infor-
mation system (GIS). This method (King's
(1991) "direct extrapolation" or Band's (1991)
"explicitly distributed approach") extrapolates
from points to area by partitioning the hetero-
geneous region into discrete areal subunits
(polygons) of relative homogeneity. TJie local
ecosystem model is applied to each polygon,
and the polygon results are combined by sum-
mation or averaging to provide a regional esti-
mate (King, 1991). The polygons are commonly
the uniform cells of a regular grid (e.g., Sklar et
al., 1985; Running et al., 1989; Bartell and
Brenkert, 1991, among many others), but they
may be nonuniform or irregular polygons (e.g.,
Band et al., 1981; Boumans and Sklar, 1990;
Burke et al., 1990; Band, 1991). This approach
is frequently and successfully used in spatial
modeling of small- to moderate-sized water-
sheds and landscapes (e.g., Bartell and Brenkert,
1991 (11.6 ha); Band, 1991 (14 km2); Running
et al., 1989 (1540 km2); Sklar et al., 1985 (2500
km2)), and it has been used in modeling larger,
regional extents (e.g., Burke et al., 1990 (ap-
proximately 10,000 km2)); Vorosmarty and
Moore, 1991 (1.2 million km2)). However, as
the areas approach those of consequence for
global change, it becomes more and more dif-
ficult to define the underlying georeferenced
data, and computational demands begin to con-
strain applications.
Global biosphere models used in atmospheric
general circulations models (GCMs) are grid-
or lattice-based georeferenced models (e.g.,
Dickinson, 1984; Sellers et al., 1986), but the
spatial resolution is coarse (frequently about
4.5° latitude by 7.5° longitude (e.g., Dickinson
and Kennedy, 1991), but occasionally with finer
resolutions of 2° by 3° (Sato et al., 1989)).
Considerable ecological information is aggre-
gated within these large cells by assumptions of
homogeneity. Even the georeferenced data sets
used to parameterize the models are rather
coarse-grained from an ecological perspective,
usually with resolutions of 1° x 1° or 0.5° x 0.5°
(grid sizes at mid-latitudes of approximately
8740 km2 and 2190 km2, respectively). Subgrid
heterogeneity and the associated variability in
ecological dynamics may have important con-
sequences for the coupling of the atmosphere
and land surface. Representation of these
influences may be improved by explicit extrapo-
lation from local scales where processes are
222
-------
A.W. KING
best understood and more empirically acces-
sible. For example, extrapolations of local
ecosystem models across the spatial extent of a
GCM grid cell (e.g., 4.5° x 7.5°) might be used
to generate better land surface parameteriza-
tions at the scale of coarse-resolution global
biosphere and atmospheric general circulation
models.
At large spatial scales (e.g., the tens or hun-
dreds of thousands of square kilometers in
GCM grid cells), a sampling of the areal extent
can be more efficient than simulating every
homogeneous area (cell) of a spatial grid. The
sampling provides for spatial extrapolation by
calculating a regional (spatial) sample mean to
estimate the population mean provided by the
grid-based or direct georeferenced approach.
This statistical approach is one form of Bands
(1991) "parameter distribution approach." As-
sumptions required by the sampling make it less
general than the georeferenced modeling, and
the approximate nature of the statistical ap-
proach does allow for some error (e.g., sam-
pling error). However, these costs may be
effectively traded against the costs of param-
eterization and computation incurred in the
georeferenced modeling. King (1991) demon-
strated that a sampling approach with large
(96%) reductions in the number of model runs
compared with grid-cell modeling introduced
relative errors of only 1-10% in results for
simple models and landscapes.
FOUNDATIONS OF THE APPROACH
Consider a geographic region R. At any point
or site in R, an ecosystem process (e.g., canopy
photosynthesis) is described by the model
(l)
where;', the localper unit area expression of the
process, is a function of vectors of local states,
x, parameters, p, and driving variables, z. For
clarity I will limit this discussion to vectors of
dimension one (x,p, andz), but the approach is
easily generalized to the common situation of
multiple state variables, parameters and driv-
ing variables.
The regionis a collection of homogeneous sites.
-At any site, the values of the arguments*,/?, and
z are constant with respect to space, but they
vary from site to site across the heterogeneous
region as spatially distributed variables. The
function/holds for all sites in R.
If the sites in R are independent such that, at
time t, the local expressiony at one site is not a
function ofy at another site [i.e.,^. ^/(y •)], then
the regional expression of the local process is
given by
Y =
(2)
where E(f(x,p,z)) is the expected value of the
local models estimate ofy, and A is the area of
the region. For discrete variables,
1 m
i f(xi( Pj) zk) g(x;, Pj ,zk) (3)
where g(x{, p., zk) is the joint probability that at
any point inR, x = Xj, p = p., and z = zk. For
continuous variables,
E[f(x,p,z)] = J J J f(x,p,z) g(x,p,z) dx dp dz (4)
where g(x,p,z) is the joint density function for x,
p, and z. Additional derivation of Equations
(2)~(4) is given in King et al. (1989).
The assumption of independent sites holds for
a large number of ecological processes and the
models describing them. The finer the tempo-
ral resolution of the model, the more valid the
assumption. For example, daily photosynthesis
at a site is virtually independent of concurrent
daily photosynthesis at other sites. Indeed, the
assumption is clearly valid for many of the
process-based models used to simulate site-
223
-------
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON STORAGE
specific ecosystemphotosynthesis. The assump- ferred to as aggregation error (e.g., Rastetter et
tion of independence may also hold as an ap- al., 1991). Much has been written about aggre-
proximation. If actual interactions are not overly gation error in ecological applications (see ref-
strong, the approximation may provide suffi- erences cited in King (1991)). Here I simply
cient accuracy. For example, the biomass dy- note that it is generally incorrect to assume that
namics of forested landscapes can be simulated the regional expression Y is given by the local
effectively by assuming independence between' model evaluated at the spatial means of the
adjacent stands (Shugart, 1984). Elsewhere, I arguments. Unless the local expression is de-
have discussed how time lags in among-site
interactions, and temporal bounds on the simu-
lation can be used to maintain the assumption
of spatialindependence (King etal., 1989,1991).
Here I will simply assume that sites are inde-
pendent, or that the assumption of indepen-
dence is a sufficient approximation. It is also
important to note that correlation among the
arguments x, p, and z at a point or site, a
commonsituation, does not violate the assump-
tion of independence. Correlation will affect
the joint probability distribution for the model
arguments, but it does not preclude use of this
approach for spatial extrapolation. Note also
that among-site spatial correlation in model
arguments also does not preclude use of this
approach. The assumption of independence
applies to the dependent variable, the model
output, and not the model inputs. The ap-
proach assumes that the spatial arrangement or
pattern of model arguments does not affect the
aggregate regional expression (Band, 1991). A
scribed by a linear function, it is generally not
true that
or
= Af[E(x),E(p),E(z)]
Y=-Af(x,p,£)
(6a)
(6b)
In other words, spatial heterogeneity cannot, in
general, be correctly accounted for by first
averaging site-specific arguments across the
heterogeneous region and then evaluating the
local model with those mean values. In prac-
tice, the magnitude of the associated error will
vary with the mathematical form of the model
and the variability (variance) of the model
arguments (O'Neill, 1979; O'Neill et al., 1979;
Kingetal, 1991; Rastetter etal., 1991). Insome
circumstances, this error can be substantial
(O'Neill etal., 1979, King et al., 1991). How-
ever, in other circumstances Equation (6) can
yield anacceptable approximation (King 1991).
The right side of the inequality in Equation (5)
is the first term in a Taylor series expansion
spatialrearrangementthatmaintainedthejoint about the mean argument vaiues (Band, 1991;
probability distribution would not alter Y of Rastetteretal., 1991). The sum of higher order
Equation (2).
It is well known that, in general, the expected
value of a nonlinear function is not equal to the
function evaluated at the expected values or
means of the function's arguments. That is,
E[f(x,p,z)]*f[E(x),E(p),E(z)]
(5)
when f is nonlinear. Failure to recognize this
inequality is sometimes referred to as the "fal-
lacy of averages" (Wagner, 1969; Welsh et al.,
1988). The associated error is sometimes re-
terms can be viewed as an error or correction
(Rastetter et al., 1991) term, the magnitude of
which is partially determined by variances and
covariances of the model arguments. If this
error term is known or believed to be small or
otherwise acceptable, Equation (6b) can be
used to represent the region. However, given
an efficient and accurate means of calculating
the true expected value, the potential error can,
in principle, be avoided entirely and, in prac-
tice, effectively minimized.
224
-------
A.W. KING
CALCULATING THE EXPECTED VALUE
Calculation of the expected value is largely
dependent upon the form of the local model
and the available description of spatial hetero-
geneity. If analytic functions for both the model
and the probability distribution are known,
Equations (3) or (4) can be evaluated directly.
For example, if the model arguments are con-
tinuous random variables, the jointdensity func-
tion is known as an explicit mathematical func-
tion (e.g., bivariate normal), and the local model
and the probability function are relatively simple,
it may be possible to evaluate the integral of
Equation (4) in exact analytical form (see ex-
amples provided by O'Neill (1979), King (1991),
Rastetter et al. (1991)). Band (1991; p. 255)
refers to this approach as "(direct) integration
of a surface function over the parameter distri-
bution function." In many applications the
model orprobabilityfunctionis complex enough
that an exact analytical solution does not exist
or cannot be found. In this case, numerical
quadrature techniques might be used to ap-
proximate the integral.
In many cases, the mathematical form of the
probability distribution function is unknown or
the combination of model and probability func-
tion is very complex, and a precise mathemati-
cal form like Equations (3) or (4) cannot be
defined. This is often true of computer simula-
tion models with many state variables in a
system of differential or difference equations
with complex auxiliary functions defining time-
dependent rate coefficients. Furthermore, in-
formation on spatial heterogeneity may fre-
quently be limited to estimates of distributional
moments (e.g., means, variance and perhaps
covariance) without an explicit formulation of
the distribution function.
In these circumstances, solution of Equations
(3) or (4) (analytically or numerically) is not
possible, and some approximation is required.
In some situations, the expected value can be
estimated by a series expansion about the mean
parameter values (Band, 1991; Rastetter et al.,
1991). Alternatively, Monte Carlo simulations
may be used to estimate the expected value of
the local model. In the Monte Carlo approach,
a collection of N full sets of the models argu-
ments is created by repeated sampling (N times)
from a joint probability distribution defined by
type and the appropriate distribution param-
eters. For example, the arguments may be dis-
tributed according to a multinormal distribu-
tion specified by a mean vector and a covari-
ance matrix (Rubinstein, 1981). The N model
runs are made, each using a different set of
sampled arguments as input, and the results are
saved. The expected value of the distribution of
model output is an estimate of E(f(x, p, z)), and
its standard deviation can be used to calculate
a confidence interval for the regional estimate
given by Equation (2). In effect, the region is
sampled randomly at a large number of sites.
Model arguments (input) at each site are de-
fined by a joint frequency distribution describ-
ing the spatial heterogeneity of model argu-
ments across the region.
Monte Carlo simulation is frequently used in
ecological applications to treat variability in
model arguments (e.g., Bonan, 1991). When
the variability is explicitly tied to spatial hetero-
geneity (i.e., the sampling distributions explic-
itly describe spatially distributed variables), the
simulations are an avenue to spatial extrapola-
tion. For example, Monte Carlo simulation has
been used to extend gap-scale models of forest
dynamics to the landscape (e.g., Shugart, 1984;
Pastor and Post, 1988), and King et al. (1987,
1989) used this approach to estimate regional
CO2 exchange with the atmosphere by extrapo-
lation from site-specific ecosystem production
models. Prentice et al. (1989) proposed Monte
Carlo simulation as an approach to continental-
scale simulations of vegetation dynamics cur-
rently addressed with patch- or stand-scale
models.
225
-------
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON STORAGE
AGENERALPROCEDUREFOR SPATIAL
EXTRAPOLATION BY EXPECTED
VALUE: APPLICATION TO SUBARCTIC
REGIONS
King et al. (1987,1989) defined a general pro-
cedure for spatial extrapolation using a local
models expected value. Implementation re-
quires fourprimary ingredients or components:
(1) the local model (or models), (2) specifica-
tion of the geographic region, (3) the frequency
distribution of spatially varying model argu-
ments for that region, and (4) a procedure for
calculating the expected value of the model(s).
Details of the extrapolation can be found in
King et al. (1987,1989). Here I simply review
that application by sketching the assembly of
the components and presenting a portion of the
results pertinent to regional carbon storage.
The Region
The region of interest is the vegetated surface
of the circumpolar latitude belt between 64° N
and 90° N. This region is dominated by two
biomes or cover types (King et al., 1987). Tun-
dra covers approximately 52% (6,132,182 km2)
of the region; cool coniferous forest (taiga)
covers anadditional 27% (3,137,733 Ion2). Com-
bined, these cover types represent nearly 99%
of the regions vegetated land surface. We
subdivided the latitude belt into two subregions
according to these biome designations. We
assumed that CO2 exchange for the remaining
land cover (mostly ice and polar desert) was
zero and could be ignored.
The Local Models
We assigned a single ecosystem model to each
biome or subregion. The models described the
daily carbon or biomass dynamics of a homoge-
neous stand of vegetation within their respec-
tive biome as a function of environmental vari-
ables such as temperature and available water.
We used ABISKOII, a model of tundra ecosys-
tems described by Bunnell and Scoullar (1975),
for the tundra. For the cool coniferous forest,
we used CONIFER, a model developed for old-
growth Douglas fir ecosystems in the north-
western United States (Coniferous Forest Bi-
ome Modeling Group, 1977). These models are
reviewed in King et al. (1987), and synopses are
provided in King et al. (1989).
Regional Heterogeneity
The theory behind this particular approach to
spatial extrapolation assumes that site proper-
ties influencing model structure do not vary
from site to site. For example, the numbers and
types of state variables, relationships between
variables and the functional representations of
rate processes are assumed to be constant across
sites. All differences among sites are reflected
in the variability of model argument values.
We assumed that each ecosystem model was
structurally an adequate representation of any
site within its respective biome and that with the
proper selection of model arguments, the model
could be used to simulate any site within that
biome.
In principle, the approach allows the treatment
of any model argument as a spatially distributed
random variable. In practice, those spatially
varying arguments that contribute most to vari-
ability in model output require the greatest
attention. We hypothesized that most of the
spatial variability in seasonal CO2 dynamics
across the circumpolar subarctic region could
be explained by spatial variability in climate
and other abiotic factors (King et al., 1987,
1989). Accordingly, we considered only the
spatial heterogeneity of the model's external
driving variables and assumed that each biome
was spatially homogeneous withrespect to model
arguments that were not themselves functions
of the driving variables.
226
-------
A.W. KING
Calculation of the Expected Value
We calculated the expected value of eachmodel
for its respective bioregion using the Monte
Carlo components of PRISM, a program de-
signed for analysis of model sensitivity and
uncertainty (Gardner et al., 1983; Gardner and
Trabalka, 1985). The program uses Latin
hypercube sampling (a constrained randomiza-
tion) (McKay et al., 1979; Iman and
Shortencarier, 1984) to select model arguments
from specified frequency distributions. Input to
PRISM defines the model arguments and the
type of distribution (e.g., normal, triangular,
uniform, lognormal, or constant), its central
tendency (mean or mode), standard deviation,
maximum value and minimum value for each
argument. Nonparametric Spearman rank cor-
relation coefficients are used to specify covari-
ance relationships among the arguments. The
combination of frequency distributions for each
argument, rank order correlations and Latin
hypercube sampling approximate the joint fre-
quency distribution for the model arguments
(Iman and Conover, 1982; Iman and
Shortencarier 1984).
The Monte Carlo approach and the input re-
quirements of PRISM made it necessary to
translate the time series of climatic variables
used in the site-specific implementation of each
model into a set of random variables. The
assumptions and details of the translation are
discussed by King et al. (1987). Briefly, the
translation involved the designation of monthly
climate variables as 12 separate random vari-
ables (e.g., January, February, December air
temperature). The tundra model had 28 input
variables; the conifer model had 96 (King et al.,
1989).
We used climate-station data and International
Biological Program (IBP) site data to describe
the spatial heterogeneity of the input variables
for each model and biome (see King et al.,
(1987,1989) for details). From these data, we
made estimates of minimum values, maximum
values and central tendencies for each of the
input variables. We used frequency histograms
to identify the appropriate type of probability
distribution for each variable (e.g., uniform or
triangular) (see Tables 1 and 2 of King et al.,
1989). We also used the climate and IBP site
data to calculate a matrix of Spearman rank
correlation coefficients for the input variables
of each model. The matrix included correla-
tions among climate variables (e.g., between
June temperature and June radiation) as well
as serial correlations among monthly values of
each variable (e.g., between June and July
temperature). As noted above, the Latin
hypercube sampling uses the correlation matrk
to approximate the joint frequency distribution
from the independently derived frequency dis-
tributions for each variable.
We made 100 runs of each model. Monte Carlo
sampling of the seasonal climate variables de-
fined an annual climate for each run; we as-
sumed the climate to be constant for the dura-
tion of the run. We ran the model with the
constant climate until simulated annual net
CO2 exchange declined asymptotically at a rate
of less than 10 % per year for five consecutive
years. This required runs of up to 50 years for
the tundra model and 100 years for the conifer-
ous forest model. We saved daily values of total
ecosystem photosynthesis, autotroph respira-
tion and heterotroph respiration from the final
year of each run. We calculated total ecosystem
respiration (autotroph plus heterotroph respi-
ration) and net ecosystem-atmosphere CO2
exchange (total ecosystem respiration minus
total ecosystem photosynthesis) from these pri-
mary output variables.
We calculated the means and standard devia-
tions of daily CO2 fluxes from the 100 Monte
Carlo runs. We calculated monthly and annual
CO2 fluxes by summing mean daily fluxes. The
spatially extrapolated fluxes for each biome are
the products of the daily, monthly, or annual
227
-------
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON STORAGE
CO2 fluxes and the area of that biome. The
predicted fluxes for the entire region are the
sums of the biome fluxes.
RESULTS: REGIONAL CARBON
STORAGE
King et al. (1989) compared estimates of sea-
sonal biosphere-atmosphere CO2 exchange
from the extrapolation with estimates from
alternative approaches by Pearman and Hyson
(1980,1981) and Gillette and Box (1986). We
also compared estimates of annual net primary
production from the extrapolation with various
estimates for the region and comparable bi-
omes. (See King etal. (1989) for details of these
results). From these comparisons, we con-
cluded that the extrapolation provided reason-
able estimates of regional CO2 exchange with
the atmosphere, and further concluded that the
general approach and procedure for making
regional estimates by extrapolation from local
site-specific models were valid.
One of the interesting results from the extrapo-
lation, and one found somewhat troubling by
King et al. (1987, 1989) concerns the annual
storage of atmospheric carbon by the terrestrial
ecosystems. The extrapolation predicted that
the region was a small net annual source of
atmospheric CO2 (Table 1).
The tundra appeared as a small sink and the
cool coniferous forest as a larger source (Table
1). As noted by King et al. (1989), this result
must be interpreted in light of the assumptions
of the simulations.
Conventional ecological wisdom assumes that
carbon storage (NEP) in mature ecosystems is
near zero. Uptake of carbon from the atmo-
sphere by photosynthesis is balanced by whole
system respiration, and there is no net carbon
storage or release by the ecosystem. The eco-
system is neither a net annual sink nor source
for atmospheric CO2. Most efforts to estimate
Table 1. Annual Net Atmosphere-Biosphere Carbon
Exchange (Gt C/yr) for the 64° N-90° N Circumpolar
Latitude Belt.
Region
Latitude belt
Tundra
Forest
Net Exchange (a)
0.214
-0.035
0.249
(a) Positive values indicate the region is a net source;
negative values indicate a net sink.
regional carbon exchange with the atmosphere
make this assumption (e.g., Pearman and Hyson,
1980,1981; Fung et al., 1983; Gillette and Box,
1986). A balanced ecosystem carbon budget is
imposed as a constraint on the estimate, and a
near zero estimate of net annual ecosystem
storage or release results. King et al. (1987,
1989) purposively did not make this assump-
tion.
The equilibrium assumption of a balanced eco-
system storage and release implies a relatively
long time scale. Over several years, average net
annual exchange with the atmosphere is ex-
pected to be (approximately) zero. Net re-
leases during years when climate favors decom-
position of litter and soil organic matter are
balanced (at least approximately) by years of
net uptake when climate favors primary pro-
duction and storage. The expectation that net
annual exchange is zero is less compelling over
shorter time scales, and some deviation is ex-
pected. Several authors have reported non-
zero net annual exchange in high-latitudes over
short periods of time (e.g., Grulke et al., 1990;
Tans et al., 1989,1990; Bonan, 1991).
The assumption of a constant climate during
the Monte Carlo simulations of King et al.
(1987,1989) results in an asymptotic (a "quasi-
equilibrium") net annual CO2 exchange for
each ecosystem type. As noted above, this
exchange is not zero (Table 1). However, these
exchanges do not represent long-term averages
for which a near-zero net annual exchange
228
-------
A.W. KING
Table 2. Potential sources of error in the spatial extrapolation of King et al. (1987).
• Failure to include correlations among driving variables
• Insufficient equilibration between climate and model state variables
• Assumption of zero contribution from other ecosystem types in the region
• Error in estimating areal extent of ecosystem types
• Error in estimating probability distributions for model arguments
• Error in matching available climatic data and model driving variables
• Assumption of within-biome homogeneity for biotic model arguments
• Failure to include important constraints on ecosystem carbon dynamics
• Inappropriateness of or error in the local ecosystem models
might be expected. The climate used in the
simulations is best interpreted as that of a
single, albeit characteristic, year that approxi-
mately matches the longer-term average an-
nual climate. The predicted exchanges are
those that would be in quasi-equilibrium with
those meteorological conditions if those condi-
tions persisted for many consecutive years. In
truth, annual climate is not invariant, and this
particular result would not likely be realized.
This arguably "peculiar" result, a consequence
of modeling assumptions and simulation de-
sign, requires careful interpretation. Conserva-
tively, assuming the validity of other model
assumptions, the result suggests that under cer-
tain climatic conditions (specifically those of
the simulation) the region could be a rather
large net source of atmospheric CO2. This is an
intriguing result given current interest in ques-
tions of how terrestrial ecosystems may have
influenced historical changes in atmospheric
CO2 or how they may influence future changes.
However, the implications of this interpreta-
tion are such that an evaluation of the
simulation's assumptions and design and a re-
appraisal of estimated regional carbon storage
are called for. For example, a more exacting
estimate of subarctic annual exchange with the
atmosphere will require multiyear simulations
with realistic interannual variations in climate.
Mean annual exchange for the period of the
simulation could then be evaluated against the
assumption of long-termbalanced, equilibrium,
zero net annual exchange.
REVISIONS AND REFINEMENTS
King et al. (1987) presented nine potential
sources of error in their extrapolated estimate
of high-latitude regional atmosphere-biosphere
CO2 exchange (Table 2).
King et al. (1989) corrected for two of these,
adding correlations among driving variables
treated as random input variables (re: item 1,
Table 2) and addressing the equilibration issue
(re: item 2, Table 2). Any error introduced by
ignoring flux from other ecosystems types (item
3, Table 2) is likely to be insignificant because
their areas are so small. Similarly, unless large
areas classified as tundra are actually cool co-
niferous forest or vice versa, the error in esti-
mating the area of these two ecosystem types
(item 4, Table 2) is likely of little consequence.
Error from inaccurately estimating probability
distributions (item 5, Table 2) could be signifi-
cant, especially if the error is in estimating the
range or endpoints of the distributions or if an
actual modal distribution (e.g., triangular or
normal) is assumed to be uniform (O'Neill et
al., 1982). Treatment of this potential source of
229
-------
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON STORAGE
error will require additional climatic data sta-
tions for the region and/or improvements in the
data obtained from those stations. These
additional data could also improve the correla-
tion estimates used in the Monte Carlo sam-
pling.
The assumption of within-biome biotic homo-
geneity (item 7, Table 2) could be an important
source of error, and relaxation of that assump-
tion could be one of the more interesting exten-
sions of the work described here. The inclusion
of biotic heterogeneity within the current eco-
system or land cover classifications will require
estimates of the spatial distribution of biotic
parameters in the models. At one level, distri-
butions and correlations used in the Monte
Carlo sampling might be inferred or "guessed
at" from auxiliary information on biotic hetero-
geneity in the region. Preferably, spatially
distributed biotic data would be used to empiri-
cally estimate distributions and correlations as
was done for the climatic variables. The num-
ber of data sites will have to be fairly large to
provide good distribution estimates and to avoid
the numerical problems encountered by King
et al. (1989) in estimating correlations among
climate variables.
Biotic heterogeneity could also be addressed by
redefining the land-cover classification and dis-
criminating ecosystem types within the tundra
and cool coniferous forest. For example, up-
land and lowland tundra might be distinguished.
Aset of bioticparametersunique to each subtype
would be defined, and with the assumption that
these parameters are homogeneous across the
subtype, the Monte Carlo treatment of climatic
spatial heterogeneity would be applied to each
subtype.
Items 6, 8, and 9 of Table 2 may best be
addressed by changing the local ecosystem
models used in the exercise. The CONIFER
model may be an inappropriate model for the
forested ecosystems of the region; CONIFER
was developed for the old-growth Douglas fir
forests of Oregon. Consequently, the biotic
parameterization of the model, which was not
changed for its use in the spatial extrapolation,
may very well be inappropriate for high latitude
subarctic forests. Minimally, a reparameter-
ization for a high-latitude site would be a useful
extension or revision. However, the inappro-
priateness of the model may also apply to its
structural features (e.g., the number and type of
state variables or functional representation of
rate processes). Use of a model explicitly
developed for higher latitude forests might
produce more accurate results. At the time of
the extrapolations presented here (i.e., King et
al., 1987,1989), CONIFER was the only model
identified as appropriate for the exercise. Sub-
sequently, Bonan (1991) has published a model
of seasonal carbon dynamics for boreal forests
of high latitudes (e.g., the vicinity of Fairbanks,
Alaska). As an example of differences in struc-
tural representation, Bonan's model includes a
moss layer, which is believed to be an important
constraint on boreal forest production (Bonan
and Korzuhin, 1989) (re: item 8, Table 2); the
CONIFER model does not include a moss
layer. A repetition of the exercise substituting
the Bonan model for CONIFER is appropriate
and could illuminate the extent to which the
results to date are determined by the choice of
coniferous forest model. The Bonan model
also includes a different representation of cli-
matic driving variables that may be more ap-
propriate to high latitudes, and its use might
reduce the error produced in translating from
climate station data to model driving variable
(re: item 6, Table 2; see King et al., (1987)).
This potential error might also be addressed by
building an intermediate model that could more
effectively translate climate station data into
the driving variables of the ecosystem models.
Thetundramodelusedinthis exercise, ABISKO
II, was developed as a general model for com-
parisons of high-latitude tundra sites (Bunnell
and Scoullar, 1975). Thus, the potential for
inappropriate model structure is not as high as
230
-------
A.W. KING
Table 3. Land cover of the 53° N-64° N Circumpolar
Latitude Belt (King et al., 1987).
Table 4. Ecosystem Models that could be used in a spatial
extrapolation for the 53° N-64° N Latitude Belt.
Ecosystem type
Temperate broadleaf
deciduous forest
Cool coniferous forest
Warm coniferous forest
Tundra
Grassland
Cropland
Wetlands
Shores and hinterlands
Ice, sand and polar desert
Area (km2)
1,705,550
7,748,098
27,192
3,272,236
919,541
1,297;270
1,463
59,432
222,211
Percent
11.18
50.80
0.18
21.45
6.03
8.50
0.01
0.39
1.46
Ecosystem type
Model
Temperate broadleaf
deciduous forest
Cool coniferous forest
Warm coniferous forest
Tundra
Grassland
Cropland
Wetlands
Shores and hinterlands
[ce, sand and polar desert
Sollins et al., 1976
Bonan, 1991
Golkin and Ewel, 1984
Bunnell and Scoullar, 1975
Parton and Singh, 1976
De Wit etal., 1978
Kadlec and Hammer, 1988
?
Not Applicable
it is for the CONIFER model. However, the
model might be revised to better represent
tundra in general or to represent different types
of tundra. Minimally, alternatives to the wet
meadow (Point Barrow, Alaska) parameteriza-
tion used in the results to date could be devel-
oped and applied in concert to the appropriate
areas. Alternative models (e.g., ARTUS of
Miller et al. (1984)) might also be substituted.
EXTENSION TO BOREAL REGIONS
An extension of the spatial extrapolation to
boreal forest regions is obvious. The equal-
area latitude belt between 53° N and 64° N
provides a convenient definition of a "boreal"
zone or region. Based on the land-cover classi-
fication of King et al. (1987), the region is
occupied by 10 terrestrial ecosystem types
(Table 3).
Cool coniferous forest (taiga) covers approxi-
mately 51% of the land area, tundra an addi-
tional 21%. Bonan's (1991) coniferous forest
model and the ABISKOII tundra model could
be extrapolated to represent these areas. Mod-
els identified by King et al. (1987) could be used
to represent the remaining vegetated area
(Table 4), but other model choices are of course
possible.
Data on the spatial distribution of primary
climatic data (e.g., weather station data) could
be obtained from the sources used by King et al.
(1987, 1989), but specific data requirements
would be dictated by the particulars of the
chosen ecosystem models. Incorporation of
biotic variability would require additional data
sources. With these data, Monte Carlo extrapo-
lations similar to those described above could
be applied to each ecosystem type. These
results combined with extrapolations for higher
"subarctic" latitudes (64° N--900 N) would pro-
vide an estimate of the characteristic terrestrial
carbon dynamics of the subarctic and boreal
forest region. With predictions or hypotheses
about the spatial distribution of abiotic and
biotic variables under an elevated CO2 climate
(e.g., two times preindustrial atmospheric con-
centrations), a similar extrapolation could be
used to explore changes in regional carbon
storage or other ecosystem carbon dynamics in
response to changes in climate.
CONCLUDING REMARKS
Spatial extrapolation of local models of ecosys-
tem carbon dynamics provides a means of quan-
tifying carbon dynamics at spatial scales that
are of global significance, but relatively inac-
cessible empirically. The statistical and Monte
231
-------
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON STORAGE
Carlo approach described here is theoretically
sound, canbe applied to a variety of models and
regions and is relatively easy to implement.
There are, of course, limitations to the ap-
proach. Beyond the general limitations of any
modeling study, there are questions about the
structural appropriateness of a chosen local
model when applied to diverse sites, and there
are uncertainties and data limitations in defin-
ing probability distributions for spatially dis-
tributed model arguments over large regions.
The approach is also computationally inten-
sive, but often less so than a georeferenced
approach to the same region. These limitations
are not intractable. Estimates can be revised
with refinements in models and data. The
uncertainties and computational demands are
effectively balanced by the ability to make
robust estimates of large scale changes hi re-
gional carbon storage that are firmly rooted in
an understanding of ecosystem carbon dynam-
ics.
REFERENCES
Band,L.E. 1991. Distributed parameterization of com-
plex terrain. Surveys in Geophysics 12:249-270.
Band, L.E., O.B. Elfes, J. T. Hayes, L. O. Mearns, P. A.
O'Rourke, B. J. Stevenson, W. H. Terjung, and P. E.
Todhunter. 1981. Application of a photosynthesis model
to an agricultural region of varied climates. California.
Agricultural Meteorology 24:201-217.
Bartell,S.M.andA.L.Brenkert. 1991. A spatial-temporal
model of nitrogen dynamics in a deciduous forest water-
shed. In: Quantitative Methods in Landscape Ecology.
The Analysis and Interpretation of Landscape Heteroge-
neity, editedby M. G. Turner and R. H. Gardner. Springer-
Verlag,NewYork,pp. 379-398.
Bonan, G.B. 1991. Atmosphere-biosphere exchange of
carbon dioxide in boreal forests. Journal of Geophysical
Research 96:7301-7312.
Bonan, G.B. and M. D. Korzuhin. 1989. Simulation of
moss and tree dynamics in the boreal forests of interior
Alaska. Vegetation 84:31-44.
Boumans,R.MJ.,andF.H.Sklar. 1990. A polygon-based
spatial model for simulating landscape change. Land-
scape Ecology 4(2):83-97.
Bunnell, F.L. and KA. Scoullar. 1975. ABISKO II. A
computer simulation model of carbon flux in tundra
ecosystems. In: Structure and Function of Tundra Eco-
systems, edited by T. Rosswall and O.W. Heal. Ecological
Bulletin 20. Swedish Natural Science Research Council,
Stockholm, pp. 425-448.
Burke, I.C., D.S. Schimel, C.M. Yonker, W.J. Parton, LA.
Joyce, and W.K. Lauenroth. 1990. Regional modeling of
grasslandbiogeochemistryusingGIS. Landscape Ecology
41:45-54.
Coniferous Forest Biome Modeling Group. 1977. CONI-
FER: A model of carbon and water flow through a
coniferous forest. Documentation. Coniferous Forest
Biome Bull. 8. University of Washington, Seattle.
De Wit, C.T., J. Goudriann, H.H. van Laar, F.W.T.
Penning de Vries, R. Rabbinge, H. van Keulen, W.
Louwerse, L. Simba, and C. de Jonge. 1978. Simulation
of assimilation, respiration, and transpiration of crops.
Pudoc, Wageningen, The Netherlands.
Dickinson, R.E. 1984. Modeling evapotranspiration for
three- dimensional global climate models. In: Climate
processes and climate sensitivity, edited by J. E. Hanson
and T. Takahashi. Geophysical Monograph 29, Maurice
Ewing Vol. 5, American Geophysical Union, pp. 58-71.
Dickinson, R.E. and P.J.Kennedy. 1991. Land surface
hydrology in a general circulation model ~ global and
regional fields needed for validation. Surveys in Geophys-
ics 12:115-126.
Gardner, R.H., B. Rojder, and U. Bergstrom. 1983.
PRISM: A systematic method for determining the effect
of parameter uncertainties on model predictions. Re-
port/NW-83/555. Studsvik Energiteknik AB, Nykoping
Sweden.
232
-------
Gardner, R.H. and J.R. Trabalka. 1985. Methods of
uncertainty analysis for a global carbon dioxide model.
DOE/OR/214QO-4. TR024. NationalTechnical Informa-
tion Service, Springfield, Virginia.
Gillette, DA. and E.O. Box. 1986. Modeling seasonal
changes of atmospheric carbon dioxide and carbon 13.
Journal of Geophysical Research 91:5287-5304.
Golkin,K.R.andK.C.Ewel. 1984. A computer simulation
of the carbon, phosphorus, and hydrologic cycles of a pine
flatwoods ecosystem. Ecological Modelling 24:113-136.
Grier,C.C. and R.S. Logan. 1977. Old-growth Pseudotsuga
menziesii communities of a western Oregon watershed:
biomass distribution and production budgets. Ecological
Monographs 47:373-400.
Grulke, N.E., G.H. Riechers, W.C.Oechel, U. Hjelm, and
C. Jaeger. 1990. Carbon balance hi tussock tundra under
ambient and elevated atmospheric CO,. Oecologia83:485-
494.
Iman, R.L. and M J. Shortencarier. 1984. A FORTRAN
77 program and user's guide for the generation of Latin
Hypercube and random samples for use with computer
models. NUREG/CR-3624, SAND83-2365, RG. Sandia
National Laboratories, Albuquerque, New Mexico, and
National Technical Information Service, Springfield, Vir-
ginia.
Iman, R.L. and W.J. Conover. 1982. A distribution-free
approach to inducing rank correlation among input vari-
ables for simulation studies. Communications in Statistics
Bll(3):311-334.
Kadlec, R.H. and D.E. Hammer. 1988. Modeling nutrient
behavior in wetlands. Ecological Modelling 40:37-66.
King, A.W. 1991. Translating models across scales in the
landscape. In: Quantitative Methods in Landscaape
Ecology, edited by M. G. Turner and R. H. Gardner.
Springer-Verlag, New York, pp. 479-517.
King, A.W., A.R. Johnson, and R.V. O'Neill. 1991.
Transmutation and functional representation of hetero-
geneous landscapes. Landscape Ecology 5:239-253.
King, A.W., D.L. DeAngelis, and W.M. Post. 1987. The
seasonal exchange of carbon dioxide between the atmo-
sphere and the terrestrial biosphere: extrapolation from
site specific models to regional models. ORNL/TM-
A.W. KING
10570. Oak Ridge National Laboratory, Oak Ridge,
Tennessee.
King, A.W., R.V.O'Neffl, and D.L. DeAngelis. 1989.
Using ecosystem models to predict regional CO2 ex-
change between the atmosphere and the terrestrial bio-
sphere. Global Biogeochemical Cycles 3:337-361.
Marland, G., TA. Boden, R.C. Griffin, S.F. Huang, P.
Kanciruk, and T.R. Nelson. 1989. Estimates of CO2
emissions from fossil fuel burning and cement manufac-
turing using the United Nations energy statistics and the
U.S.BureauofMmescementmanufacturingdata.ORNL/
CDIAC-25. NDP-030. Oak Ridge National Laboratory,
Oak Ridge, Tennessee.
McKay, M.D., RJ. Beckman, and WJ. Conover. 1979. A
comparison of three methods for selecting values of input
variables in the analysis of output from a computer code.
Technometrics 21:239-245.
Miller, P.C., P.M. Miller, M. Blake-Jacobsen, F.S. Chapin
HI, K.R. Everett, D.W. Hilbert, J. Kummerow, A.E.
Linkins, G.M. Marion, W.C. Oechel, S.W. Roberts, and L.
Stuart. 1984. Plant-soil processesinEriphoriumvaginatiun
tussock tundra in Alaska: a systems modeling approach.
Ecological Monographs 54:361-405.
O'Neill, R. V. 1979. Natural variability as a source of error
in model predictions. In: Systems Analysis of Ecosystems,
edited by G. S. Innis and R. V. O'Neill. International Co-
operative Publishing House, Fan-land, Maryland, pp. 23-
32.
O'Neill, R.V., J.W. Elwood, and S.G. Hilderbrand. 1979.
Theoretical implications of spatial heterogeneity in stream
ecosystems. In: Systems Analysis of Ecosystems, edited
byG. S. Innis and R. V. O'Neill. International Coopera-
tive Publishing House, Fairfield, Maryland, pp. 79-101.
O'Neill, R.V., R.H. Gardner, and J.H. Carney. 1982.
Parameter constraints hi a stream ecosystem model: in-
corporation of a priori information hi Monte Carlo error
analysis. Ecological Modelling 16:51-65.
Parton, WJ. and J. S. Singh. 1976. Simulation of plant
biomass on a shortgrass and a tallgrass prairie with
emphasis on belowground processes. US/IBP Grassland
Biome Tech. Rep. No. 300. Colorado State University,
Fort Collins.
233
-------
QUANTIFYING REGIONAL CHANGES IN TERRESTRIAL CARBON STORAGE
Pastor, J. and W.M. Post. 1988. Response of northern
forests to CO2- induced climatic change: dependence on
soil water and nitrogen availabilities. Nature 334:55-58.
Pearman,G.I.andP.Hyson. 1980. Activities of the global
biosphere as reflected in atmospheric CO2 records. Jour-
nal Geophysical Research 85:4468-4474.
Pearman.GI.andP.Hyson. 1981. A global atmospheric
diffusion simulation model for atmospheric carbon stud-
ies. In: Carbon Cycle Modelling, edited by B. Bolin.
SCOPE 16. John Wiley and Sons, New York, pp. 227-240.
Prentice, I.C., R.S. Webb, M.T. Ter-Mikhaelian, A.M.
Solomon, TM. Smith, S.E. Pitovranov, N.T. Nikolov, A A.
Minin, R. Leemans, S. Lavorel, M.D. Korzukhin, J.P.
Hrabovszky, H.O. Helmisaari, S.P. Harrison, W.R.
Emanuel, and G.B. Bonan. 1989. Developing a global
vegetation dynamics model: Results of anIJASA summer
workshop. RR-89-7. International Institute for Applied
Systems Analysis, Laxenburg, Austria.
Rastetter, E.B., A.W. King, B J. Cosby, G.H. Hornberger,
R.V. O'Neill, and J.E. Hobble. 1991. Aggregating fine-
scale ecological knowledge to model coarser-scale at-
tributes of ecosystems. Ecological Applications. In press.
Rubinstein, R.Y. 1981. Simulation and the Monte Carlo
Method. John Wiley and Sons, Inc., New York.
Running, S.W., R.R. Nemani, D.L. Peterson, L.E. Band,
D.F. Potts, L.L. Pierce, andMA. Spanner. 1989. Mapping
regional forest evapotranspiration and photosynthesis by
coupling satellite data with ecosystem simulation. Ecol-
ogy 70:1090-1101.
Sato, N., P.J. Sellers, DA. Randall, E.K. Schneider, J.
Shukla, J.L.KinterIII, Y-T.Hou, andE.Albertazzi. 1989.
Effects of implementing the simple biosphere model in a
general circulation model. Journal of the Atmospheric
Sciences 46:2757-2782.
SeUers, P.J., Y. Mintz, Y.C. Sud, and A. Dalcher. 1986. A
simple biosphere model (SiB) for use within general
circulationmodels. JournalAtmosphericSciences 43:505—
531.
Shugart, H.H. 1984. A Theory of Forest Dynamics: the
Ecological Implications of Forest Succession Models.
Springer-Verlag, New York.
Sklar,F.H.,R.Costanza,andJ.W.Day,Jr. 1985. Dynamic
spatial simulation modeling of coastal wetland habitat
succession. Ecological Modelling 29:261-2 81.
Sollins, P.,W. F.Harris, and N.T. Edwards. 1976. Simu-
lating the physiology of a temperate deciduous forest. In:
SystemsAnalysis andSimulationinEcology, Vol.4, edited
byB. C. Patten. Academic Press, New York, pp. 173-218.
Vorosmarty.CJ.andB. Moore III. 1991. Modeling basin-
scale hydrology in support of physical climate and global
biogeochemical studies: an example using the Zambezi
River. Surveys in Geophysics 12:271-311.
Wagner, H.M. 1969. Principles of Operation Research.
Prentice-Hall, Englewood-Cliffs., New Jersey.
Welsh, A.H., A. Townsend Peterson, and S. A. Altmann.
1988. The fallacy of averages. American Naturalist
132:277-288.
234
-------
CARBON POOLS AND FLUX ON FORESTED
LANDS OF THE UNITED STATES
Hermann Gucinski, David P. Turner,
Charles Peterson, and Greg Koerper
ABSTRACT
Forest ecosystems contain much of the organic carbon in the terrestrial biosphere andhave the capacity to be a large net source
or sink of carbon. To estimate the carbon pools and net carbon flux at the national scale, information on the age-class
distribution of different forest types and the age-specific pools of carbon in the trees, soils, forest floor, understory and coarse
woody debris is needed. Additional factors in evaluating forest-carbon flux include the quantities of timber removed, the
rate at which harvested carbon returns to the atmosphere and carbon emissions associated with wildfire. With the exception
of a national inventory of forest age-class distribution on public lands, much of the relevant data for evaluating the carbon
budgetforforestedlandoftheUnitedStatesisavailableandisbeingassembledinanationalforestcarbonmodel. Themodel
willprovide the basis for exploring the carbon sequestration potential of these forests under various forest-policy scenarios
and possible impacts of climate change on forest carbon storage.
INTRODUCTION
The potential for global climate change due to
the anthropogenic release of carbon dioxide,
methane and other gases has focused attention
on global forests. Forest ecosystems are a
central component of the global carbon cycle
and are of interest and concern for three rea-
sons. The first is the current contribution of
deforestation, particularly in tropical latitudes,
to the build-up of greenhouse gases (Houghton,
1991). The second is the potential climate-
induced redistribution of the world's forests,
which may result in the release of potentially
large quantities of carbon (Prentice and Fung,
1990; Dixon and Turner, 1991). This may be a
transient release due to changes in disturbance
regimes overbroad spatial scales (Overpeck et
al., 1990; Neilson and King, 1992), or a long
term change in terrestrial carbon storage if
forests are replaced by biomes having relatively
low above- and below-ground carbon concen-
trations (Smith et al., 1991). The third is the
possibility that forests could be managed to
increase the sequestering of carbon (Dixon et
al., 1991) and thereby both delay the onset of
climate change or diminish its magnitude
(Trexler, 1991). Such management could take
advantage of possible benefits that may accrue
from the direct effects of higher ambient CO2
levels (Strain and Cure, 1985; Bazzaz and
Garbutt, 1988), or from the potential expansion
of tropical forests due to the increased water
availability predicted by some general circula-
tion models (Neilson, 1990).
Forest ecosystems dominate the gross annual
biogenic flux of carbon between the atmo-
sphere and the terrestrial biosphere. The glo-
bal terrestrial net primary productivity has been
estimated at roughly 60 Pg/yr. There is a
comparable release through heterotrophic res-
235
-------
CARBON POOLS AND FLUX ON FORESTED LANDS OF THE UNITED STATES
piration and decay of residues. By contrast, atmospheric rise in CO2 (Post et al.,1990). To
fossilfuelreleasesreached5.4Pginl986(Post account for this divergence, Tans et al (1990)
etal 1990) Thus alterations inthe health and have suggested the existence of a possible sink
productivityofforestsduetoclimatechangeor in the temperate latitudes of the Northern
management strategies could have important Hemisphere. This discrepancy suggests we do
consequences for the atmospheric balance of not yet fully understand all the relevant climatic
carbon dioxide. and anthropogenic variables in the carbon cycle
under past and current conditions (Post et al.,
It has been difficult to quantify the annual 1990). Moreover, it appears likely that the
uptake of carbon by terrestrial biota because projected changes in climate will induce changes
quantification requires detailed description of in the biosphere that may dampen or accelerate
vegetation types and soils, the associated bio- the rate of climate change via btosphenc feed-
mass and the rates of change over the seasonal back loops (Houghton, 1990; Lashoff, 1989;
cycle. Disturbances such as conversion of land Prentice and Fung, 1990).
to agriculturaluse, harvesting, biomass burning
and responses to short term climatic fluctua- It may not yet be possible to construct a detailed
tions must also be taken into account. In many and precise carbon budget for terrestrial eco-
Elobal analyses, theassumptionhasbeenmade systems on a global scale. However, regional
thatterrestrialecosystemsareincarbon-excha- carbon budgets may point the way toward im-
nge equilibrium with the atmosphere (Fung et proved flux estimates and help to resolve the
al 1987) Equilibrium conditions require that present contradictions. Such efforts may also
the heterotrophic respiration plus any release help in the evaluation of management prac-
byfiremustbalance, on an annual basis, the net tices, such as large scale reforestation, for se-
primary production (i.e., the amount of carbon questeringatmospheric CO2(Dixonetal., 1991;
actually assimilated into biomass by autotro- Moulton and Richards, 1990; Schroeder and
phs). This is clearly the case in grassland Ladd, 1991; Sedjo, 1989; Trexler et al., 1989)
* / ., * * s» .1 .. _ • 3! A— j.1*«. .4.!.*m •*•»•*-! *-v»-» «-»*i*-1 t-*-» o nr-ni +11r\ f*
and may indicate the direction and magnitude
of biospheric feedbacks to climate change
(Apps and Kurz, 1991).
ecosystems where carbon accumulation (i.e.,
accumulation of standing stock) is minimal on
decadal time scales when compared with for-
ests (Jenkinson et al., 1991; Hall and Scurlock,
1991) By contrast, the latter maybe significant This paper outlines our approach for estimat-
carbon sinks even on annual time scales. Inthe ing the carbon budget for the forest sector of the
case of disturbance such as fire or logging, such United States using existing data. Preliminary
areas may become strong carbon sources to the results are presented; however, improvements
, i-n Q -mi-mV^r nf nr<=«9<: «TY». hf.iriP' Heveloned. as
atmosphere.
Where human land-use impacts alter carbon
storage over large areas, the terrestrial compo-
nent of the biosphere as a whole can become a
net carbon source. Houghton et al. (1983) have
estimated the net flux from terrestrial systems
due to land use changes at 90 to 120 Pg over the
period from 1800 to 1980. However, when
in a number of areas are being developed, as
discussed below.
APPROACH
Acarbonbudget accounts for the pools of carbon
and the net flux of carbon from a selected area
over a specified time interval from biogenic and
anthropogenic sources. An annual time-step is
UV^JLl^JU JUL UlJJl JLWV/ lr\J A.*S\J\J* jt~jL.**t T» v» •»"•«• y " --•—-— IT t?
these estimates are added to the fossil-fuel flux most relevant for characterization of biogenic
over the same period and the ocean uptake is carbon flux because of the seasonal pattern of
subtracted, the net exceeds the observed global photosynthesis and respiration.
236
-------
Forest land area is derived from the latest
inventory of the U.S. Forest Service (Waddell
et al., 1989). Land classification includes the
following categories: pubic or private, reserved
or unreserved and timberland or woodland.
For this paper, total tree biomass data of tim-
berland (growth > 1.4 m3/ha/yr) are derived
from the analyses of Cost et al. (1990). Tree
biomass data are aggregated at the state level,
but do not include understory and stump bio-
mass because of incomplete records. However,
we have used regional stump, foliage and sap-
ling data from Cost et al. (1990) to compute
expansion factors to arrive at an estimate of
total above-ground forest carbon of timber-
lands. Biomass data for woodlands (growth <
1.4 m3/ha/yr), which comprise large areas in
the western U.S., are being compiled from
USFS inventories and resource reports. We
have not attempted to estimate error or confi-
dence limits at this time. While such limits can
be determined for much of the information, the
regionally aggregated data have no associated
sampling error, especially for data from areas
under public ownership (e.g., national forests)
at present. In future reports, we will explicitly
address the problem of sampling error and the
effect on carbon estimates.
Based on preliminary analysis, we assigned the
forest woodlands a carbon density of 1 kg/m2.
Although the use of a single number will intro-
duce some estimation error, the estimate of the
total forest carbon pool will not be changed
appreciably because the low overall contribu-
tion of woodlands will have little effect, even if
the assumed carbon density is in error by as
much as 50%. Carbon pool estimates are
summarized at the regional level for timber-
lands and woodlands, incorporating estimates
of live-root carbon from Koch (1989). Sub-
sequent analyses will allow resolution by state,
forest type, or ownership class. Soil carbon
determination, exclusive of root biomass, are
described elsewhere (Gucinski et al., 1991).
H. GUCINSKI, D.P. TURNER, C. PETERSON, AND G. KOERPER
Forest ecosystem carbon flux is the uptake of
atmospheric CO2 through photosynthesis and
the release of carbon from autotrophic and
heterotrophic respiration. The difference be-
tween these two processes over an annual cycle,
the net flux, is the Net Ecosystem Productivity
(NEP). The NEP of a forest ecosystem varies
by successional stage, with the greatest carbon
accumulation in the mid-successional stage. In
early stages, carbon is lost to the atmosphere
due to the combination of enhanced heterotro-
phic respiration in the soil, low net primary
productivity and, in some instances, coarse
woody debris decomposition. The NEP can be
low or even negative in the late-successional
stage due to a high rate of autotrophic and
heterotrophic respiration (Sprugel, 1985). Dis-
turbances, such as catastrophic fire and forest
harvesting, may release large quantities of car-
bon, particularly at the end of the successional
cycle. Rather than modeling photosynthesis
and respiration explicitly, the approach here
considers NEP based on estimates of carbon
pool accretion as a function of stand age classes
for different forest types and regimes.
Deforestation, harvest and replanting modify
the age distribution of forests. Thus, evaluation
of the biogenic carbon flux requires an estimate
of the age class distribution among stands within
each forest type and region. Because harvest-
ing itself may result in a large carbon flux (e.g.;
the burning of slash), and much of the harvested
carbon may be used in ways which return it to
the atmosphere rather quickly (e.g., paper prod-
ucts), a complete forest-carbon budget must
include harvest-related flux and will undoubt-
edly reveal a story different from one that
considers only biogenic flux (Harmon et al.,
1990). This is the subject of ongoing research.
To estimate carbon-flux, forest-age distribu-
tion data by state and forest type, in 10-year
increments, were combined with age-specific
rates of carbon gains or losses, (i.e., the stand
level carbon budget) estimated for specific for-
237
-------
CARBON POOLS AND FLUX ON FORESTED LANDS OF THE UNITED STATES
300 T
15
25
35
45
55 65
Age Class (yrs)
75
85
95
>105
Figure 1. Distribution of Western Oregon Douglas-fir by age-class on private timberland.
est types. An example of the age class distribu- stand level carbon budget. Extrapolating to
tion for the Douglas-fir forests of western Or- public timberland (e.g., the National Forests)
egon on private lands is shown in Figure 1, and in this manner introduces some estimation er-
the corresponding time course for the carbon ror because of differential harvesting rates and
flux is shown in Figure 2. Empirical studies, differing management practices. We anticipate
ground-based surveys and generic models were attaining better estimates of age class distribu-
used to determine the age-specific rates of tion for public land from both additional inven-
change in above- and below-ground carbon tory data and from remote-sensing analysis.
storage (Birdsey, 1991) and include, for ex- For reserved timberland, e.g., forests in desig-
ample, soil-carbon loss after harvesting; these nated wilderness or in national parks which are
were provided by R. Birdsey, U.S. Forest Ser- withdrawn from timber utilization by statute or
vice. The effects of disturbances such as insect regulation, the assumption of zero NEP, which
outbreaks are implicit in the data base. The implies carbon flux equilibrium, maybe reason-
stand age class distribution by forest type and able because age class distributions include
state was obtained for private timberlands over many older stands. Uncontrolled fires would
the entire United States from the U.S. Forest also contribute to the maintenance of a net
Service (Waddell, pers. comm.). carbon accumulation of zero in these areas.
Similar information of stand age distributions
for public lands is not readily available for the
entire United States. For our initial flux esti-
mate for public timberland, we assumed the
same age class distribution within each forest
type as found on the adjacent private timber-
lands within each state and used the associated
The effort to obtain age class data from remote
sensing for areas of public lands not treated in
current inventories holds considerable promise.
High spatial resolution imagery (-25 m, The-
matic Mapper) has been used to classify stand
structure on some national forests of the Pacific
Northwest (Tepley and Green, 1991;W. Ripple,
238
-------
H. GUCINSKI, D.P. TURNER, C. PETERSON, AND G. KOERPER
80 -r
-T 900
0 5 15 25 35 45 55 65 75 85 95 105 115 125 135 145 155
Stand Age (yrs)
Figure 2. Carbon budget during stand development for Western Oregon Douglas-fir.
Forest Science, Oregon State University, pers.
communication; Cohen and Spies, 1991). In that
scheme, the classifications are 1) pre-canopy
closure; 2) closed canopy; 3) mature; and 4) old-
growth forests. Incorporation of these data and,
over larger spatial scales, the use of imagery from
the Advanced Very High Resolution Radiom-
eter (AVHRR) sensor on the NOAA polar or-
biter satellites will permit improved treatment of
the public land forest sector.
Additional research efforts on the forest-sector
carbon budget include an accounting for coarse
woody debris and differences in the carbon
content due to differences in forest stand origin.
In some forest types, e.g., old growth timber,
significant quantities of dead wood are left after
harvesting, and its decay may release appre-
ciable carbon to the atmosphere (Harmon et
al., 1990). By contrast, the creation of forest
from agricultural land, or lands otherwise re-
verting to forest, occur at significant rates in the
eastern United States. In early succession, such
forests have neither the residue of dead wood
nor loss of soil organic matter commonly seen
in old growth forests (Williams, 1988).
RESULTS
Of 243 Mha (million hectares) of forests in the
conterminous 48 states of the United States, or
roughly 31% of the total land area (Waddell et
al., 1989), 201 Mha is considered timberland.
The remaining 42 Mha include less productive
forests (woodlands) with relatively low NPP. We
estimate above-ground carbon storage of U.S.
forests to be 10.4 Pg (Table 1), of which 10.0 Pg
is in timberlands; woodlands account for the
remainder. Our current estimate for the carbon
in forests, exclusive of understory, of the conter-
minous 48 states, across all ownerships, is 11.5 Pg
when roots are included. Contributions to forest
carbon pools from the understory and the forest
floor are being compiled and will be included in
future reports.
Using the current-stand level carbon budgets
for the base year 1986, the net biogenic flux of
carbon from NEP is approximately 343 Tg/yr
for the conterminous 48 states (Table 1). The
flux into privately owned timberlands is esti-
mated to be 248 Tg/yr, and that into public
timberlands is 95 Tg/yr. The planned inclusion
239
-------
CARBON POOLS AND FLUX ON FORESTED LANDS OF THE UNITED STATES
Table 1. Tree carbon storage and net biogenic carbon flux for forests in the contiguous U.S.(a).
Net biogenic carbon flux1 tTg/y)
Forest area (Mha)
Region
-------
to the year 2040 to make such assessments.
CONCLUSION
The development of a carbon budget for the
United States is important in terms of: 1) com-
paring the relative magnitudes of biogenic and
anthropogenic flux; 2) contributing to the con-
struction of a global carbon budget; 3) evaluat-
ing possible carbon sequestration and conser-
vation strategies; and 4) exploring the magni-
tude and direction of possible feedbacks to
climate change mediated by the terrestrial bio-
sphere. The basic information needed to con-
H. GUCINSKI, D.P. TURNER, C. PETERSON, AND G. KOERPER
struct an estimate for the biogenic carbonpools
and net flux on forested lands includes the areal
extent of the different forest types, the age class
distribution of the forest stands within each
type and the age-specific carbon pools within
each forest type. Regional and national surveys
provide a basis to begin constructing a national
carbon budget, but additional information, de-
rivable in part from satellite remote sensing,
will be needed. Analyses that track the return
of harvested carbon to the atmosphere and
quantify emissions from wildfire are also needed.
REFERENCES
Apps, M J. and W. A. Kurz. 1991. Assessing the role of
Canadian forests and forest sector activities in the global
carbon balance. World Resources Reviews 3(4): 333-344.
Bazzaz, FA. and K. Garbutt. 1988. The response of
annuals in competitive neighbourhoods: effects of elevated
CO2. Ecology 69:937-946.
Birdsey, R. A. 1991. Prospective changes in forest carbon
storage from increasing forest area and timber growth.
USDA Forest Service, Washington D.C. Data on file with
the USDA Forest Service. In press.
Cohen, W.B. and TA. Spies. 1991. Estimating structural
attributes of Douglas-fir/western hemlock forest stands
from Landsat and Spot imagery. Remote Sensing of
Environment In press.
Cost, N.D., J.O. Howard, B. Mead, W.H. McWilliams,
W.B. Smith, D.D.Van Hooser, and EJL Warton. 1990.
The biomass resource of the United States. USDA Forest
Service Gen. Tech. Rep. WO-57.
Dixon, R. K. and D. P. Turner. 1991. The global carbon
cycle and climate change: responses and feedbacks from
below-ground systems. Environ. Pollut. 73: 245-262.
Dixon, R.K., P.E. Schroeder, and J.K. Winjum. 1991.
Assessment of promising forest management practices and
technologies for enhancingthe conservation and sequestra-
tion of atmospheric carbon and their costs at the site level.
U.S.EPA ORD, EPA 600/3-91/67, Washington DC 138 p.
Fung, I.Y., C J. Tucker, andK.C. Prentis. 1987. Application
of advanced very high resolution radiometer vegetation
index to study atmosphere-biosphere exchanges with the
terrestrial biosphere. J. Geophysical Res. 92: 2999-3015.
Gucinski, H., C. Peterson, J. Kern, D. Turner, D. Press, and
GA. King. 1992. Elements of the U.S. carbon budget:
Progress and preliminary results. In: Proceedings of the
PACL1M Symposium, edited by K. Redmond. Monterey
CA, March 10-13,1991.
Hall, D.O. and J.M.O. Scurlock. 1991. Climate change and
productivity of natural grasslands. Annals of botany 67
(Supplement 1): pp. 49-55.
Harmon, M.E., W.K. FerreU, and J.F. Franklin. 1990.
Effects on carbon storage of conversion of old-growth
forests to young forests. Science 247: 699-702.
Houghton, RA., J.E. Hobbie, J.M. Melillo, B. Moore, B J.
Peterson, G.R. Shaver, and G.M.Woodwell. 1983. Changes
in the carbon content of terrestrial biota and soils between
1860 and 1980: net releases of CO2 to the atmosphere.
Ecological Monographs 53: 235-263.
Houghton, RA. 1990. The future role of tropical forests
in affecting the carbon dioxide concentration of the atmo-
sphere. Ambio 19:204-209.
Houghton, RA. 1991. Tropical deforestation and atmo-
spheric carbon dioxide. Climatic Change 19: 99-118.
Koch,P. 1989. Estimates byspeciesgroupandregionin the
USA of below-ground root weight as a percentage of
ovendry complete tree weight and carbon content of tree
portions. Consulting Report to the USDA Forest Service.
Lashoff,DA. 1989. The dynamic greenhouse: feedback
processes that may influence future concentrations of
241
-------
CARBON POOLS AND FLUX ON FORESTED LANDS OF THE UNITED STATES
atmospheric trace gases and climate change. Climatic
Change 14:213-242.
Moulton, RJ. and KJR. Richards. 1990. Costs of seques-
teringcarbon through treeplantingandforest management
in the United States. USDA Forest Service General
Technical Report WO-58, p. 47.
Neilson, RJ>. 1990. Biosphere feedback during climate
change. In: Response and Feedbacks of Forest Systems to
Global Climate Change, edited byGA. King, J.K. Winjum,
RJK. Dixon, and L.Y. Arnaut. U.S. EPA, ORD, EPA/600/
3-90/080 Washington, DC. p. 156.
Neilson, R. P. and GA. King. 1992. Continental scale
biome responses to climate change. In: Ecological Indica-
tors, edited by McKenzie,D., E. Hyatt, J. McDonald.
Proceedings of International Symposium, Ft. Lauderdale,
FL, Oct 16-19,1990. Elsevier Science Publishers, Ltd.
Ov«rpeck,J.T.,D. Rind, and R. Goldberg. 1990. Climate-
induced changes in forest disturbance and vegetation.
Nature 343:51-53.
Prentice,K.C. and I.Y.Fung. 1990. Bioclimaticsimulations
tcstthesensitivity of terrestrial carbon storage in perturbed
climates. Nature 346:48-51.
Post, W.M., T-H. Peng, W.R. Emanuel, A.W. King, V.H.
Dale, and D.L.DeAngelis. 1990. The global carbon cycle.
Am. Scientist 78:310-326.
Schroeder.P. andL.Ladd. 1991. Slowing the increase of
atmospheric carbon dioxide: a biological approach. Cli-
matic Change 19:283-290.
Sedjo, R. 1989. Forests: a tool to moderate global
warming? Environment 31:14-20.
Smith, TJM., H.H. Shugart, G.B. Bonan, and J.B. Smith.
1991. Modeling the potential response of vegetation to
global climate change. Adv. Ecol. Res. In press.
Sprugel, D.G. 1985. Natural disturbances and ecosystem
energetics. In: The Ecology of Natural Disturbances and
Patch Dynamics, edited by S.TA. Picked: and P.S. White.
Academic Press, New York, NY.
Strain, B.R. and J.D. Cure. 1985. Direct effects of
increasing carbon dioxide on vegetation. DOE/ER-0238,
U.S. Dept. of Energy, Carbon Dioxide Research Div.,
Washington, D.C., p. 286.
Tans, PP., I.Y. Fung, and T. Takahashi. 1990. Observa-
tional constraints on the global atmospheric CO2 budget.
Science 247:1431-1438.
Tepley, J. and K. Green. 1991. Oldgrowth forest - how
much remains? Geographic Info. Syst. 1(4): 23-31.
Trexler, M. C., P.E. Faeth, and J.M. Kramer. 1989. Forestry
as a response to global warming: an analysis of the
Guatemala agroforestry and carbon sequestration project.
In: Tropical Forestry Response Options to Global Climate
Change. Conf. Proc., Sao Paulo, Brazil. January 1990.
USEPA/ IBAMA/USP. Rpt by Office of Policy Analysis,
USEPA, Washington, DC. p. 53.
Trexler, M. C. 1991. Minding the Carbon Store: Weighing
U.S. Forestry Strategies to Slow Global Warming. World
Resources Institute, Washington DC. p. 8.
Waddell,K.L.,D.D. Oswald, and D.S. Powell. 1989. Forest
statistics of the United States. USDA Forest Service Res.
Bull. PNW-RB-168,106 p.
Williams, M. 1988. The death and rebirth of the American
forest: clearing and reversion in the United States, 1900-
1980. In: World Deforestation in the Twentieth Century,
edited by J.F. Richards and R. P. Tucker. Duke Univ. Press,
Durham, NC.
ACKNOWLEDGEMENTS
The work presented is a component of the U.S. EPA Global Climate Research Program, Global Mitigation and
Adaptation Program, RA. Dixon, Program Leader, and Jeffrey Lee, Project Officer, at the EPA Environmental
Research Laboratory - Corvallis, OR under contract to ManTech Environmental Technology, Inc. We thank D. Coffey,
W. Cohen, T. Droessler, A. Hairston and E. Vance for their helpful comments and criticism.
The research described in this proceedings paper has been funded by the U.S. Environmental Protection Agency. The
document has been prepared at the U.S. Environmental Research Laboratory in Corvallis, Oregon, through contract
#68-C8-0006 to ManTech Environmental Technology, Inc. It has been subjected to the Agency's peer and administrative
review, and it has been approved for publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
242
-------
ESTIMATING CARBON BUDGETS OF CANADIAN FOREST
ECOSYSTEMS USING A NATIONAL SCALE MODEL
Michael J. Apps, Werner A. Kurz, and David T. Price
ABSTRACT
Forest managers, ecosystem scientists and policymakers are becoming increasingly concerned about possible effects of
predictedchangesin climate on forest carbon budgets, and about how management strategies should be adapted to respond
to these changes. Ttie Canadian boreal forest and sub-arctic ecosystems are carbon repositories of global significance that
nayproveparttcularfysensitivetopossibleclimatechangespredictedfornorthem
^ have developed an integrated model of the processes affectingthecarbonbudget of Canadian forests and forestsector
activities The structure of the carbon budget model and its estimates of forest sector carbon pools and fluxes are reviewed.
Effects of ecosystem disturbances (wildfire, insect attacks causing stand mortality and various harvesting methods) are
simulated by the carbon budget model, allowing sensitivity of the carbon budget to changes in these disturbance regimes to
be investigated The carbon budget model was used to generate a complete carbon budget for the Canadian forest sector
for a single reference year (1986), using disturbance statistics representative of the decade 1980-1989. The contribution of
the Canadian boreal forest regions to the national carbon budget was found to be significant and very sensitive to realistic
changesin the areas burned annually by wildfires. Further development of the carbon budget model is in progress to allow
its use for analysis of possible climate change and management scenarios over periods of several decades
INTRODUCTION
It is an unfortunate paradox that science uses
reductionist techniques to define fundamental
truths, whereas high-level management gener-
ally requires that these truths be molded into
simplistic approximations for broad scale appli-
cation. This is particularly true of the relation-
ship between ecosystem science and forest
management because the fundamental truths
about ecosystem processes are at the base of a
network of great complexity, while the value of
the forest resource (per hectare) is often so low
that generally an extensive form of manage-
ment must be practised. An additional problem
may be that some forest managers are aware of
the inherent complexities of the ecosystems
they are managing, but this complexity discour-
ages them from attempting to incorporate cur-
rent ecological process knowledge into their
long-term planning.
However, it is becoming increasingly apparent
that ecosystem complexities can be of great
economic and political significance, so manag-
ers and policy makers are now being confronted
with forest management problems that require
better understanding of ecosystem responses
than can be achieved through normal line man-
agement. An important example of this con-
cerns the role of northern forests in the global
carbon cycle. Recent work by Tans et al. (1990)
and others (e.g., Zoltai et al., 1991; Gorham,
1991) has drawn attention to the probable sig-
nificance of northern circumpolar terrestrial
vegetation as a major sink for atmospheric CO2.
It now appears possible that a significant pro-
portion of post-industrial anthropogenic CO2
243
-------
ESTIMATING CARBON BUDGETS OF CANADIAN FOREST ECOSYSTEMS
emissions to the atmosphere (Keeling et al., oping the model was intended partially to lo-
1982; Gammon etal., 1985 ;Rotry and Marland, cate and identify weaknesses and gaps in exist-
1986), is sequestered in the biomass and soils of ing data and knowledge. Where possible, these
northernhigh-latitude ecosystems, particularly deficiencies have been remedied using infor-
" " " mation from the relevant scientific literature or
by holding workshops to incorporate current
expert knowledge. In other cases, research
programs to resolve particularly crucial ques-
tions have been initiated. Results obtained
in the vast areas of the circumpolar boreal
forest and sub-arctic vegetation (e.g., Bonan,
1991). Furthermore, recent studies based on
the predictions of atmospheric global circula-
tion models (AGCMs) have indicated that the
increasing atmospheric CO2 concentration will fromrunning the model for a single representa-
have significant impacts on global climate and tive year which will be presented, demonstrate
thatthesesamenorthernforests are likely to be the importance of the boreal and sub-arctic
subjected to biologically significantincreases in ecosystems as major components of the Cana-
mean annual temperatures within the next 50 dian national carbon budget will be presented.
years orso (e.g., Schlesinger and Mitchell, 1987; Results obtained from running the model for a
Houghtonand Woodwell, 1989). Someterres- single representative year will be presented,
trial ecologists have suggested that northern which demonstrate the importance of the bo-
forest ecosystems may be particularly sensitive real and sub-arctic ecosystems as major compo-
to these possible changes hi climate (Zoltai et nents of the Canadian national carbon budget.
al., 1991; Rizzo and Wiken, 1989; Bonan et al.,
1990). RATIONALE
Our current knowledge of the exact nature of
these possible climate changes is limited, as is
our knowledge of how the forest ecosystems
would be likely to respond. However, two
propositions are clear: (1) northern forest eco-
systems occupy such extensive areas that their
possible responses to projected imminent
changes in climate are of major socio-economic
importance; and (2) management decisions
made for these forests today could, therefore,
have very important consequences even within
our own lifetimes.
This paper will briefly discuss the development
and structure of a large-scale carbon budget
model of the Canadian forest sector (hence
referred to as the CBM-CFS). This model is
intended to bridge the gap between the knowl-
edge gained by ecosystem scientists attempting
to understand the processes of critical impor-
tance to the above problems and the resource
managers and policy makers who need that
knowledge now to make better long-term plan-
ning decisions. The approach adopted hi devel-
The total carbon budget of any geographic
region is dependent on the fluxes of carbon into
and out of the landscape within that region.
Changes inboth the inventory and annual fluxes
of forest carbon can result both from natural
causes (e.g., disturbances due to fire, wind and
insect attack and the processes of stand regen-
eration, growth, competition and decay) and
from human actions (e.g., silviculture, logging,
land-use changes and fossil fuel consumption).
These processes potentially can all result in
either positive or negative impacts on ecosys-
tem productivity and atmospheric CO2 ex-
change, corresponding to accumulation or deple-
tion of the carbon inventory. In general, re-
source managers do not require detailed under-
standing of the causes of changes resulting from
possible alternate management actions (or from
a changing climate), but they do need a correct
interpretation of the trends and relative magni-
tudes of ecosystem responses to those changes.
A well-designed large-scale carbon budget
model, built on a correct understanding of the
important ecosystem processes, can therefore
244
-------
M.J. APPS, WA. KURZ, AND D.T. PRICE
Boreal West
Boreal East
Subarctic
igure 1. Outline map of the ecoclimatic provinces (EP) of Canada used in the carbon budget model of the Canadian
approximatelocationsoftheeastandwestborealandsub-articEPs (adaptedfromEcoregiont
assist in deciding how the forest resource (i.e.,
a collection of ecosystems) should best be man-
aged to meet particular objectives under a
range of management and climate scenarios.
MODEL STRUCTURE
The development of the CBM-CFS was planned
in three distinct phases. The objective of Phase
1 was to develop a model to assess the carbon
budget for a single reference year, 1986, se-
lected as representative of the current situa-
tion. Phase 2, currently under development,
will allow the future effects of alternative man-
agement scenarios to be analyzed. Phase 3,
when completed, will allow the national carbon
budget to be analyzed as a function of manage-
ment and climate change scenarios.
A detailed description of the structure of the
Phase 1 CBM-CFS, and the data upon which it
is based, are reported elsewhere (Kurz et al.,
1991; Kurz et al., 1992), so only a brief outline
will be presented here. An unusual feature of
the model is that it links the carbon dynamics of
ecosystem disturbances, growth and decompo-
sition processes, forest management and the
forest industry within a single integrated frame-
work.
The CBM-CFS is based on recent forest and
soils inventory data, other government and
industry statistics for timber harvesting, utiliza-
tion and decay of forest products and losses
from fire and insect attack. From these input
data, using algorithms that attempt to encapsu-
245
-------
ESTIMATING CARBON BUDGETS OF CANADIAN FOREST ECOSYSTEMS
Figur
mode
pool.
A
T
M
O
S
P
H
E
R
E
Net Ecosystem
Productivity
r\
i . ^
— \^
Disturbance
Oxidation
Disturbance
Net Uptake^
! •*
L--
Live Biomass
Ac, c
1 Harvest I
I
Products 1
AC, C A
TOT
T
Soils
AC, C
Peatlands
AC, C
Net Litterfall
+
Disturbance
Transfer
7
s2.Relatiomhipsbetweencarbonpools,changesin these pookandtheinferredfluxes, as used in the carbon-budget
1. The notation C refers to the standing stock of carbon in the pool and AC to the net annual change m that carbon
late the best available knowledge of ecosystem
processes, the model estimates the sizes of
ecosystem carbon (C) pools and fluxes for spa-
tial units based on the Canadian ecoclimatic
classification (Ecoregions Working Group,
1989) shown in Figure 1. The estimates of the
terms in the carbon budget for each spatial unit
are then summed by the model to generate a
national forest sector annual carbon budget.
These ecoclimatic provinces have already been
usedforprojectingfuture changes hi vegetation
coverresultingfromanticipated climate changes
(Rizzo and Wiken, 1989; Zoltai, 1988; Zoltai et
al., 1991) Although subject to debate, these
equilibrium projections indicate that the prai-
rie grassland and cool temperate forests will
migrate northwards into the areas currently
occupied by boreal forest, while the current
boreal forest regions will be greatly reduced. It
is planned that a future version of the CBM-
CFS will attempt to simulate these changes
dynamically in response to the transient stages
of climate change predicted to result from a
doubling of atmospheric CO2 concentration.
The CBM-CFS represents the processes affect-
ing each of the major carbon pools found in
Canadian forest ecosystems: forest biomass,
soils and peatlands. In the Phase 1CBM, only
the first two pools were modeled in any detail,
while peatlands were represented very simply.
However, it was recognized that the boreal
peatlands have historically been a very impor-
tant carbon sink, and that they are potentially
very sensitive to possible climate changes. In
addition to ecosystem carbon pools, biomass
246
-------
carbon transferred to the forest product sector
is tracked until released to the atmosphere.
Figure 2 shows the carbon pools and fluxes
currently accounted for within the CBM-CFS,
where the fluxes are inferred from estimated
annual net changes in the pools. Carbon uptake
is considered to occur solely through ecosystem
net photosynthesis, but there are several routes
by which biomass carbonmay be released to the
atmosphere, including decomposition of litter
fall and coarse woody debris produced through
harvesting, mortality and natural disturbances.
Ecosystem disturbances may transfer some car-
bon to the atmospheric and soil pools (e.g.,
fire), but harvesting is distinct in that it also
exports carbon to the forest products pool.
BIOMASS PRODUCTIVITY
AND DECOMPOSITION
Growth curves for a range of forest types were
constructed from biomass and age-class data
obtained from the Canadian National Forest
Biomass Inventory (Bonnor, 1985) and from
Canada's Forest Inventory (Bonnor, 1982; For-
estry Canada, 1988), respectively. The esti-
mates of biomass carbon per unit land area
derived from these data generally agree very
well with those obtained independently by
Botkin and Simpson (1990). For both the
biomass and soil carbon pools, the carbon in-
ventories were estimated by summing the area-
weighted carbon mass data for each of the
different forest types occurring within each
spatial unit, while net carbon uptake was esti-
mated similarly from the carbon fluxes inferred
by the model.
SOIL CARBON DYNAMICS
Soil carbon and detritus are treated as three
distinct pools with characteristic turnover rates:
slow, medium and fast. Standing debris remain-
ing after a disturbance, such as fire or insect
attack, is added to the medium and fast turn-
over pools. Soil carbon data were obtained
247
M J. APPS, W.A. KURZ, AND D.T. PRICE
from the Oak Ridge National Laboratory data
base (Zinke et al., 1986). The separate simple
model for peatland areas uses historical data to
estimate the annual net carbon sequestration
within Canadian peatlands.
FOREST PRODUCT SECTOR
Forest sector activities affect forest ecosystem
carbon dynamics, both through harvesting and
through the sequestration of carbon in forest
products, including disposal in landfills, while
wood bioenergy is potentially important for
reducing the atmospheric input of CO2 from
burning of fossil fuels. The model of forest
product utilization and decay was developed
using statistics going back to 1947 obtained
from: provincial government records, the Pulp
and Paper Research Institute of Canada, the
Canadian Council of Forest Industries, private
industry sources and Statistics Canada. Forest
products manufactured in previous years are
viewed as belonging to a series of annual co-
horts. The carbon retained by each cohort is
estimated from data on carbon losses due to
initial processing and subsequent changes in
use, including disposal and decay. The total
carbon released in 1986 from decaying wood
products, manufactured from Canadian forest
biomass during the previous 40 years, was then
estimated by summing the 1986 losses from
each annual cohort. This sum was subtracted
from the amount of carbon in new wood prod-
ucts transferred from the forest biomass pool in
1986, to give the net accumulation in the forest
product carbon pool.
DISTURBANCE REGIMES
An important feature of the CBM-CFS model
is that ecosystem disturbances are explicitly
included so that the carbon transfers among
pools resulting from disturbance events may be
tracked. Canadian federal and provincial sta-
tistics were obtained for five distinct types of
forest disturbance: fire, insect attack resulting
in stand mortality and three types of harvesting
-------
ESTIMATING CARBON BUDGETS OF CANADIAN FOREST ECOSYSTEMS
Fij
El
A
T
M
O
S
P
H
E
R
E
>s shown in Fig
sMtCfstorag
92 + 17
i\
i ^
U"-
20
^ 23
15
26 ^
ry of results obta
Tire 1). These re
e in pools') and J
12,OOO
Biomass
A- -28
I 44 I
600
Products
A = +21 i
\
76,000
Soils
A = + 57
135,000
Peatlands
A= +26
UNITS: Mt C
1
17 - Litterfall
I
55 - Disturbances
\
Forests, forest sector
Net sink 51
plus peatlands
Net sink 77
ined with the Phase I carbon budget model of the Canadian forest sector (i.e.,all eleven
suits refer only to the year 1986 and should not be extrapolated to other years. Units
Vlt C/year Cfluxes and changes in pools) (after Apps and Kurz, 1991).
regime. The total forest area annually affected
within each spatial unit by each type of distur-
bance is allocated to stand types, based on
eligibility criteria such as age and forest type
(softwood, hardwood or mixed-wood).
Disturbance matrices define the proportions of
ecosystem carbon transferred between indi-
vidual sources (biomass and soils) and sinks
(soils, atmosphere and forest products) at the
time of disturbance. For example, fire gener-
ally releases only a relatively small amount of
carbon to the atmosphere immediately, but
transfers a larger amount into standing and
fallen woody debris. Debris is treated by the
model as an addition to the soil pools, and
therefore decomposition over the years follow-
ing disturbance is simulated by the soil sub-
model.
RESULTS
Phase 1: Canadian Forest Sector Carbon
Budget 1986
Figure 3 shows the results obtained from the
Phase 1 model, which provides a "snap-shot" of
the Canadian forest sector carbon budget for
the single year 1986 (Apps and Kurz, 1991).
Based on these data and the carbon budget
model's output, standing biomass for the na-
tional forest resource, excluding peatlands, is
12 Gt C, of which about 50% is in the Canadian
boreal and sub-arctic regions. The change in
that pool for the year 1986 after accounting for
disturbances and removal of forest products
material was a net decline of 28 Mt C in above-
ground biomass (Kurz et al., 1992). However,
the model estimates that disturbances trans-
ferred approximately 55 Mt C from the biomass
pool to the soil pool, of which 15 Mt C were
248
-------
Table 1. Carbon budget for Canadian forest biomass
pools, 1986, showingimpacts of carbon transfers related to
disturbances. The western boreal forest is characteristi-
cally drier than the eastern boreal forest, leading to
significantly greater transfers of carbon due to forest fires.
Units are in Mt C/year.
Canada
Area (million ha) 404
Net Primary
Productivity 92.0
(prior to disturbance)
Disturbances
Biomass -> Atmosphere
Wildfire 18.7
nsects 0.1
Slashburning 1.5
Biomass -> Soil
Wildfire 21.0
Insects 12.4
Logging 22.0
West Boreal
98
20.0
11.0
0.0
0.0
11.0
0.0
1.0
Biomass -> Forest Products Sector
Logging 44.2 2.6
Net change -27.9
-55
Bast Boreal
120
29.0
2.2
0.1
0.1
4.0
9.0
9.7
16.3
-11.7
Sub-Arctic
85
6.6
25
0.0
0.0
2.6
0.0
0.0
0.0
1.4
released to the atmosphere, and the remaining
40 Mt C were added to a net gain from litter fall
of approximately 17 Mt C, for a total increase in
the soil pool of 57 Mt C. Meanwhile, the forest
products sector, which is a relatively insignifi-
cant pool of 0.6 Gt C, gained approximately 21
Mt C through harvesting and wood processing.
Peatland areas contain some 135 Gt C, and
after allowing for CO2 and methane releases,
the net increase in the peatland carbon stock
was about 26 Mt C (Gorham, 1991; Zoltai,
1991). Because of the carbon stored hi peatland
and organic soils, the boreal and sub-arctic
ecosystems contain about 85% of Canadian
terrestrial carbon. Canadian forest ecosystems
and the forest industrial sector formed a total
net carbon sink of approximately 51 Mt C, or 77
Mt C if peatland areas are included, of which
the boreal and sub-arctic regions contributed
about 24.5 Mt C, or 49 Mt C if peatlands are
included. It is worth noting that a very different
result would have been obtained if gains in
forest soil carbon due to disturbances and trans-
fers to forest products had not been considered.
M J. APPS, W.A. KURZ, AND D.T. PRICE
Table 2. Summary of carbon budget net pool changes for
Canadian forests, 1986. The forest biomass data are taken
from Table 1, with corresponding totals shown for soils,
peatlands and forest products. Units are inMt C/year '
Area (million ha)
Net Pool Changes
Biomass
Soils
Forest Products
Peatlands
Total (Net Sink)
Canada
404
-27.9
57.4
21.1
26.2
76.8
West Boreal
98
-5.5
7.4
1.2
11.2
14.3
East Boreal
120
-11.7
23.6
6.8
8.4
27.1
85
1.4
1.1
0.0
5.0
7.5
Table 1 shows relative contributions of east and
west boreal and sub-arctic regions to the Cana-
dian national carbon budget, while Table 2
summarizes the 1986 C-budget for these areas.
The boreal forest and sub-arctic regions lost 16
Mt C in biomass, but after accounting for in-
creases in soil and peatland C stocks, they
became a net sink of 49 Mt C, a very significant
proportion of the national total sink for 1986.
Interestingly, disturbance releases were approxi-
mately evenly distributed between eastern and
western boreal forest, but in the east, they were
mainly due to insect-induced stand mortality
and harvesting, while in the west they were
primarily due to fire. Forest products harvested
from the boreal regions created an extremely
large sink compared with the size of the forest
product carbonpool. Hence, carbon sequestra-
tion in forest products is potentially important
when assessing national forest carbon dynam-
ics, particularly if considering future manage-
ment strategies to mitigate CO2 releases to the
global atmosphere.
Sensitivity Tests - Effect of Forest Fires
on the Carbon Budget
Some preliminary sensitivity tests were con-
ducted to examine the effects of changes in
disturbance regimes on the national carbon
budget because these were useful both for veri-
fication of the model, and to get a first assess-
ment of their significance under possible future
249
-------
ESTIMATING CARBON BUDGETS OF CANADIAN FOREST ECOSYSTEMS
management and climate change scenarios, processes for all Canadian forests, parameter-
For example a 200% increase in the area izations will be developed, based on the output
burned annually (comparable to the excep- of a planned smaller-scale (regional) version of
tional 1989 fire season), resulted in a net de- the CBM This will_provide a.
small-scale processes and the total carbon bud-
get for an individual spatial unit. The regional
model is expected to operate at the level of
individual landscapes (or even smaller areas
such as catchments or individual stands) be-
cause many of the important ecophysiological
processes are both nonlinear and highly vari-
able even at these scales.
CONCLUSIONS
crease of 107 Mt C in the biomass pool and a
reduction in the size of the forest sector sink
from 77 to 11 Mt C. Under this extreme fire
scenario, if peatland areas were excluded, the
Canadian forest sector would become a net
annual source of approximately 15 Mt C. It
should be noted that other ecosystem changes,
such as altered production and decomposition
rates that might be expected in years of greater
fire frequency, were not considered in this sen-
si ivity an ysis. ^^ carbon bu(jget model of the Canadian
forest sector provides a framework for estimat-
ing the terms in the Canadian national carbon
budget and allows some preliminary assess-
ment of how that budget may change in the
Oneofthemampurposesrnimtiatingdevelop- future. On the basis of currently available
mentof the national carbon budget model was forest resource statistics and assumptions de-
toallowpolicymakerstheopportunitytoexam- rived from knowledge of ecosystem processes,
ine the potential effects of alternative policy the Canadian forest sector was estimated to be
decisions and climate change scenarios. To a net carbon sink of about 77 Mt C in 1986 of
achieve this objective, several major structural which 26 Mt C were sequestered in peatlands
enhancementstothePhaselmodelareplanned The boreal and sub-arctic regions accumulated
or in progress. First, the Phase 2 and Phase 3 a net total of about 49 Mt C (64% of the
modelswiUbeabletoprojectforwardintimeby Canadian total sink), of which about one half
simulating dynamicresponses to climate change were due to sequestration by peatlands.
and management effects likely to influence the
carbon budget during the next 50-100 years. It must be emphasized very strongly that these
& * • •• . i , _ _t_ 1J x 1~. n ~.,...|..Mn*-h«-i1 *t+£*/A •£r\i-\ir*i-rr\ llffrrv
DISCUSSION
Phases 2 & 3: Future Improvements
\jf^± WSA* +* Vfc **>»»• " •»- — ———^ »>
However, the initial strategy will be to validate
model predictions against responses observed
during the last 40 years and attempt forward
projections for a period of only 10 years. Sec-
ond, for the model to respond realistically to
changes in climate and management, ecosys-
tem processes (particularly disturbance effects
and physiological responses) will be simulated
results should not be extrapolated forward into
other years because the model clearly indicates
that the forest sector carbon budget is poten-
tially very sensitive to changes both in climatic
conditions and management actions. In com-
mon with other studies, the predictions ob-
tained from the current model suffer from large
uncertainties in the soils, peatlands and post-
tillU yiiyolUlUglv<
-------
onstrated by sensitivity tests using plausible
changes in the national forest fire statistics.
Therefore, it should not be assumed that the
sink is sustainable under anticipated climate
change with no changes in current manage-
ment. Improvements to the model are planned
and in process that will allow better assessment
M.J. APPS, W.A. KURZ, AND D.T. PRICE
of ecosystem responses to various future sce-
narios of changing climate and possible alterna-
tive management strategies. These scenarios
must be explored so that policy makers can be
given scientifically based options, from which
they should be able to make better decisions for
the future management of the Canadian forest
resource.
REFERENCES
Apps, M.J. and WA. Kurz. 1991. Assessing the role of
Canadian forests and forest activities in the global carbon
balance. World Resource Review 3:333-344.
Bonan, G.B. 1991. Atmosphere-biosphere exchange of
carbon dioxide in boreal forests. Journal of Geophysical
Research 96(D4):7301-7312.
Bonan, G.B., H.H. Shugart, and D.L. Urban. 1990. The
sensitivity of some high-latitude boreal forests to climatic
parameters. Climatic Change 16:9-29.
Bonnor, G.M. 1982. Canada's Forest Inventory 1981.
Dep. Environ. Can. For. Ser., For. Stat. Syst. Branch,
Ottawa, Ontario.
Bonnor, G.M. 1985. Inventory of forest biomass in
Canada. Can. For. Serv., Petawawa Natl. For. Inst., Chalk
River, Ontario.
Botkin, D.B. and L.G. Simpson. 1990. Biomass of the
North American boreal forest. A step toward accurate
global measures. Biogeochemistry 9:161-174.
Ecoregions Working Group. 1989. Ecoclimatic Regions
of Canada, First Approximation. Ecoregions Working
Group of Canada Committee on Ecological Land Classi-
fication. Ecological Land Classification Series, No. 23,
Sustainable Development Branch, Canadian Wildlife Ser-
vice, Conservation and Protection, Environment Canada,
Ottawa, Ontario.
Forestry Canada. 1988. Canada's Forest Inventory 1986.
For. Can., Ottawa, Ontario.
Gammon, R.H., E.T. Sundquist, and P.J. Fraser. 1985.
History of carbon dioxide in the atmosphere. In: Atmo-
spheric Carbon Dioxide and the Global Carbon Cycle,
edited by J.R. Trabalka. USDA Department of Energy,
DOE/ER-0239, pp. 25-62.
Gorham,E. 1991. Northern peatlands role hi the carbon
cycle and probable responses to climatic warming. Eco-
logical Applications 1(2):182-195.
Houghton, RA. and G.M. Woodwell. 1989. Global
climatic change. Scientific American 260 (4):36-44.
Keeling, C.D., R.B. Bacastow, and T.P. Whorf. 1982.
Measurements of the concentration of carbon dioxide at
Mauna Loa observatory, Hawaii. In: Carbon Dioxide
Review: 1982, edited by W.C. Clark. Oxford University
Press, New York, pp. 377-385.
Kurz, WA., M.J. Apps, T.M. Webb, and P.J. McNamee.
1991. The contribution of biomass burning to the carbon
budget of the Canadian forest sector: a conceptual model.
In: Global Biomass Burning: Atmospheric, Climatic and
Biospheric Implications, edited by J.S. Levine. MIT Press,
Cambridge, MA., pp. 339-344.
Kurz, WA., M.J. Apps, T.M. Webb, and PJ. McNamee.
The Carbon Budget of the Canadian Forest Sector: Phase
1. Forestry Canada Northwest Region Information Re-
port NOR-X-326, Northern Forestry Centre, Edmonton,
Alberta.
Rizzo.B.andE.Wiken. 1989. Assessing the sensitivity of
Canada's ecosystems to climatic change. In: Landscape
Ecological Impacts of Climate Change on Boreal/(sub)
ArcticRegions with Emphasis on Fennoscandia, edited by
EA. Koster and M.M. Boer. LLIC Project, pp. 94-111.
Rotty, R.M. and G. Marland. 1986. Fossil fuel combus-
tion: recent amounts, patterns and trends of CO2. In: The
Changing Carbon Cycle: a Global Analysis, edited by J.R.
Trabalka and D.E. Reichle. Springer-Verlag, New York,
pp. 474-490.
Schlesinger, M.E. and J.F.B. Mitchell. 1987. Climate
model simulations of the equilibrium climatic response to
increased carbon dioxide. Reviews of Geophysics 25:760-
798.
251
-------
ESTIMATING CARBON BUDGETS OF CANADIAN FOREST ECOSYSTEMS
Zoltai, S.C. 1991. Estimating the age of peat samples
from their weight: a study from west-central Canada. The
Holocene 1:68-73.
Tans,P.P.,I.Y.Fung,andT.Takahashi. 1990. Observa-
tional constraints on the global atmospheric CO2 budget.
Science 247:1431-1438.
Zoltai, S.C., T. Singh, and M J. Apps. 1991. Aspen in a
changing climate. In: Aspen management for the 21st
century, edited by S. Navratil and P.B. Chapman. Pro-
ceedings of a symposium held November 20-21, 1990,
Edmonton, Alberta. Forestry Canada, Northwest Region,
Northern Forestry Centre andPopular Council of Canada,
Edmonton, Alberta, pp. 143-152.
Zinke, PJ., A.G. Strangenberger, W.M. Post, W.R.
Emanuel, and J.S. Olson. 1986. Worldwide Organic Soil
Carbon and Nitrogen Data. Report ORNL/CDIC-18.
CarbonDioxidelnformation Centre, Oak Ridge National
Laboratory.
Zoltai, S.C. 1988. Ecodimatic provinces of Canada and
man-induced climatic change. Canada Committee on
Ecological Land Classification, Newsletter No. 17, pp. 12-
15.
ACKNOWLEDGEMENTS
This project was funded by the Canadian Federal Panel on Energy Research and Development (PERD) through the
ENFOR (ENergy from the FORest) program of Forestry Canada. We thank the 26 experts from Forestry Canada,
several U.S. and Canadian universities, and the Canadian forest industry, who freely provided ideas and data at a 3-day
workshop sponsored by Forestry Canada. Joe Lowe and others from Forestry Canada's Petawawa National Forestry
Instituteprovidednationalbiomass data. Tim Webb and Peter McNameehave contributed significantly to the successful
development of the carbonbudget model. Mr. R. Man: provided valuable comments on an earlier draft of this manuscript.
We also wish to acknowledge Dr. J.S. Marni, Assistant Deputy Minister, Forest Environment, whose interest has been
an inspiration for this project. Mr. R. Mair provided valuable comments on an earlier draft of this manuscript.
252
-------
REGIONALIZATION AS A TOOL FOR STUDYING
CARBON CYCLING IN THE FORMER SOVIET UNION
Victor Blanutsa
ABSTRACT
Carbon cyclmgcanbestudiedusingavariety of methods. On a global level, methods of modeling carbon cycling and the
related effect on climate variation predominate. At the local level, investigations are dominated by experimental methods
ofstudymgthebehaviorof carbon in different ecosystems and their components. However, at the regional level, itis difficult
to define a predominant method because of the scarcity of investigations at this level. Therefore, it is necessary to examine
a method of regional investigation and, in the first approximation, to assess its potential uses for studying carbon cvclinsin
extensive territories of the former Soviet Union. "-y^ungm
INTRODUCTION
To date, many interesting results have been
obtained from the study of carbon cycling glo-
bally and locally. However, the scientific knowl-
edge in this field remains incomplete because
investigations on a regional level are nearly
lacking. On many occasions, this lack of infor-
mation leads to attempts to extend the local
regularities of a carbon budget, obtained by
studying a separate ecosystem, at once to a
global level. It seems more appropriate to con-
duct ageneralization of the knowledge and data
through the chain "ecosystem-ecoregion-main-
land-Earth." One of the methods of studying the
carbon budget on aregional levelis regionaliza-
tion. Regionalization becomes of particular
importance when it is necessary to make an
extrapolation beyond the data base for large,
poorly studied territories such as the boreal
forests and subarctic ecosystems of the former
Soviet Union. In this case, the goal of ecological
regionalization is to identify a set of ecoregions
that differ by the character (type, model, main
parameters) of their carbon cycling.
In the former Soviet Union, regionalization
implies a process of many-factor (many-signa-
ture) division of a territory into a set of nonin-
tersecting integral regions, which represent
compact condensations of certain initial cells
(points) both in the three-dimensional physical
and many-dimensional signature spaces. As far
as ecological regionalization is concerned, how-
ever, the direction is toward identifying regions,
the specific character and integrity of each of
them determined by the presence of a special
ecological situation. Depending on how the
"ecological situation" idea is interpreted
(namely, the quality of the environment as a
whole, the geochemical situation, the medico-
geographical situation, the kind of man-bio-
sphere interaction, the structure and intensity
of flows of substance, energy and information in
ecosystems, etc.), a number of subkinds of eco-
logical regionalization are distinguished.
The generally accepted concept of regionaliza-
tion is associated with the ordering of our
knowledge of a territory and with the identifica-
tion of regional blocks of information. The role
253
-------
REGIONALIZATION AS A TOOL FOR STUDYING CARBON CYCLING
of regionalization in geosciences is similar to tal quality, as a sphere with a standard structure
the part played by periodization in history and and intensity of the turnover of substance and
classification in all sciences. This forms the energy, etc.). The latter reflects the spatial
basis for the fact that many scientists perceive formative features of regional ecological con-
regionalization only as a factor in the develop- sciousness of the population and can combine
ment of space-time models (including maps) territories with different environmental re-
for territorially-expressed phenomena. How- sources. For instance, the Baikal region as
ever the potential applications of regionaliza- perceived by the local population, is defined as
tionarefarbroader(Blanutsa,1990;Kagansky, a region of social efforts to salvage Lake Baikal
and covers the territory not only of the lake
basin, but also a significant part of Irkutsk
Oblast, where most social ecological move-
1987).
PECULIARITIES OF THE SOVIET
APPROACH TO ECOLOGICAL
REGIONALIZATION
ments for Baikal protection are operating.
3. There are overwhelming efforts relating
ecological regionalization to problems of eco-
nomics and management. One can even speak
of certain "economization" and "cybernetiza-
Itis impossible to address all of the peculiarities
of ecological regionalization. Therefore, our
treatment here will be based on the North ^ ~«~_~~. .,
American approach to ecological regionaliza- tion" (if cybernetics is regarded as the science of
tion (Bailey et aL, 1985; Omernik, 1987), and management) of ecological regionalization.
weshallmentiononlythesalientfeaturesofthe Thus, emerged natural-economic, economic-
Soviet approach. We shall rely on papers pub- natural resource and ecologo-economic region-
lished in scientific journals and hi the Proceed- alization, as well as problem-oriented regional-
izationf or program-target environmental man-
agement, expert-geographical and some other
"management" kinds of regionalization.
ings of two workshops: "Ecologo-Geographical
Mapping and Regionalization in Siberia"
(Irkutsk, December, 1986) and"Ecoregion'90"
(Irkutsk, October-November, 1990). .
4. First attempts have been and are being made
to humanize ecological regionalization, imply-
ing giving the highest priority to investigations
of man's problems, re-orienting regionalization
from environmental protection problems to
problems of ensuring ecological security of the
population, using sociological information, and
taking into account the properties of percep-
2. Atendency to separate all research work into tion and understanding of ecological problems
several groups has emerged. At present, the by the local population.
mostpronounced differences intheunderstand-
ing of an ecological region exist. To generalize
these differences, we will introduce two new
concepts, namely "region-resource" and "re-
gion-ideal." The former corresponds to all
those ecological regions that are identified ac-
1. Ecological regionalization is presently ha the
developmental stage. Therefore, a generally
accepted conception of regionalization is lack-
ing, theoretical efforts predominate over prac-
tical work and results obtained may to be in-
compatible
5. In a large number of publications, the eco-
logical regionalization problem is formulated
in terms of mathematical theories (probability
statistics, graph theory, non clear-cut sets, etc.)
associated with the attempt to carry out a many-
dimensional-statistical treatment of the "eco-
logical region" notion and is a manifestation of
cording to the presence of some resources (sig-
nificant for man's activity) and conditions of the ~0 0
environment (region as a totality of like ecosys- the mathematization of ecological regionaliza-
tems, as a territory with a definite environmen- tion.
254
-------
6. Ecological regionalization has been deter-
mined on the basis of a geographical informa-
tion system, and first attempts have been made
to develop an expert system: "ECOREGION."
7. During the past several years, regionaliza-
tion was introduced, on a large-scale basis, to
new areas of research. Thus, there emerged
problem-oriented, predictive, solid, fuzzy and
diagnosticregionalization. In the former Soviet
Union, a great need is now felt for two new
kinds of regionalization, namely compensa-
tional (a system of regions in which an unfavor-
able ecological situation in some regions would
be compensated by a good situation in other
regions) and "emergency" (a system of regions,
and a set of measures realizable in them, owing
to which the damage from an ecological disas-
ter would be minimized) (which, by the way,
was not done in the zone of potential effects
from the Chernobyl nuclear power plant). This
list could also be extended, but it is important
for us to stress the presence of new fields of
application of regionalization and, associated
with them, the appearance of new possibilities
of ecological regionalization.
THE SYSTEM OF ECOLOGICAL
REGIONALIZATION METHODS
Currently, it is impossible to conduct a well-
grounded identification of ecoregions using a
single method. This is because, first, such a
versatile method is not yet available and, sec-
ond, it is necessary to invoke a large number of
auxiliary methods to improve the effectiveness
of the definition of a system of ecoregions.
Therefore, Ihave developed a system of region-
alization methods. This system is a conceptu-
ally ordered and informationally interrelated
set of methods of cognizing the object of region-
alization and of developing its space-time mod-
els.
The system consists of four subsystems of meth-
ods: 1) formation of a conceptual model of
regionalization; 2) measurement of parameters
of the regionalization object; 3) identification
255
V. BLANUTSA
of regions; and 4) verification, correction and
interpretation of results obtained. The first
subsystem involves the following groups of
methods: ascertaining the task of regionaliza-
tion and constructing models of the regionaliza-
tion object and the regionalization procedure,
as well as developing a conceptual model and
determining the regionalization principles. The
second subsystem includes three groups of
methods: constructing an operational model of
regionalization, measuring the object and opti-
mizing the set of indices. Identification of re-
gions (the third sub-system) is carried out with
the help of heuristic, expert and automatic
algorithms of regionalization. Methods of the
last subsystem are designed for verifying, cor-
recting and interpreting results of the identifi-
cation of regions.
The system of regionalization methods devel-
oped is characterized by the following features:
1) interdependence, complimentarily, inter-
changeability and mutual controllability of the
methods; 2) combination of qualitative and
quantitative, descriptive and formalized, and
expert and mathematical-statistical methods;
3) flexibility and adaptability to the specific
character of the problem being solved, the
possibility of constructing a set of derivative
(modified) systems of methods, the multi-func-
tionality of regionalization being conducted,
and the orientation toward different principles
and initial information of different quality; 4)
complete mutual coupling of all subsystems; 5)
verifiability and correctability of results ob-
tained; and 6) the possibility of inclusion in
other systems of methods.
The developed system of ecological regional-
ization methods can be used for solving differ-
ent problems (Blanutsa, 1989, 1990). As an
example, I shall give here a list of methods of the
first and third subsystems of methods for evalu-
ating the environmental impact of large re-
gional projects.
The methods of the first subsystem are: (1.1)
analysis of all goals of the implementation of a
-------
REGIONALIZATION AS ATOOLFORSTUDYING CARBON CYCLING
project, including those of low probability and
those that are hypothetical; (1.2) identification
of alternative projects; (1.3) determination of
the theoretical, methodical and informational
support for the conduct of the assessment; (1.4)
specification of the format of presentation of
results of the ecological assessment and of the
timerequiredforits realization; (1.5) definition
of the strategy for conducting the assessment;
(1.6) analysis of possibilities and limitations of
the adopted strategy; (1.7) construction of an
environmental impact model; and (1.8) specifi-
cation of the main concepts.
The main procedures for assessing the environ-
mental impact are performed in the third sub-
system in the following order: (3.1) synthesis of
heterogeneous ecologicalinformationandiden-
tification of existing ecoregions (initial region-
alization); (3.2) prediction of the change in
regional ecological structure, without taking
into account the project being analyzed (pre-
dictive regionalization); (3.3) prediction with
proper accounting for the impact of a project
(evaluating regionalization); (3.4) predictions
taking into account new locations of planned
objects within the territory considered (variants
of alternative regionalization); (3.5-3.6) repeti-
tion of procedures 3.4 and 3.5 when planning a
smaller environmental impact assessment as
compared with the initial variant of the project
and of its location (simulation regionalization).
ECOLOGICAL REGIONALIZATION
OF IRKUTSK OBLAST
A study in this area was made on the basis of a
problem-oriented approach to regionalization
(Blanutsa, 1989), the system of methods as
described briefly above and, on information
characterizing the environmental conditions in
Irkutsk Oblast in 1990.
In the delimitation of ecoregions, the following
sequence of actions was used: identification of
a set of problems to be analyzed; establishment
of the grid of initial operational territorial units
(OTU); determination of the structure of the
set of problems in each OTU; calculation of the
measures of likelihood between the sets of
problems of all OTU; sequential multi-step
(multi-variant) integration of OTU into re-
gions; choice of an optimal variant of regional-
ization; and obtainment of resulting structures
of the sets of problems for the regions identi-
fied.
In terms of the first operation, it was found that
for the study of Irkutsk Oblast, of greatest
importance are the following seven regional
environmentalproblems: 1) technogenic change
hi the geological environment; 2) pollution of
surface waters; 3) air pollution; 4) soil erosion;
5) land use; 6) perturbation of the vegetation
cover; and 7) perturbation of ecological-fauna
complexes. A detailed study of each problem
and mapping of results obtainedpermitted seven
ecological maps of Irkutsk Oblast on a
1:1,500,000 scale to be generated. For the
subsequent handling of these maps, their leg-
ends were unified (via expert examination) with
a view to identifying four levels of criticality of
environmental problems.
The OTU grid was determined by superimpos-
ing the seven maps and determining meshes,
uniform for a given set of problems. In this way,
282 OTU were identified. Subsequently, in
each OTU, the structure of the set of environ-
mental problems was determined. This was
done by arranging the problems in their critical-
ity. As a result, each OTU was characterized by
the arranged sequence of problems. Also, some
OTU did not involve any problems at all while
others included one or two problems. Yet most
of the OTU involved three problems or more.
The subsequent operations were carried out
using a special algorithm (Blanutsa, 1989) with
computer techniques applied. As a result, at the
4th step of unification of the OTU into regions,
with the degree of uniformity 0.80, an optimal
variant of regionalization involving 41 ecore-
gions was identified.
The most critical situation occurs in the Irkutsk
ecoregion (region No. 1 in Figure 1). All of the
256
-------
V. BLANUTSA
igure 1. Ecological regionalization of Irkutsk Oblast, fonner Soviet Union
other regions were assigned a consecutive num-
ber with a decrease in criticality of the environ-
mental problems, and the 41st ecoregion did
not exhibit any of the seven environmental
problems considered.
The main result of the ecological regionaliza-
tion of Irkutsk Oblast has been geographical
synthesis of heterogeneous ecological informa-
tion. Such a generalization of the ecological
knowledge of the region makes it possible to
reproduce the process of spatial differentiation
and integration of the environment through a
system of 41 ecoregions. This opens up the
possibility of disengaging oneself from the huge
number of different facts when making man-
agement decisions to achieve a coherent com-
prehension, understanding and management of
the ecological situation in Irkutsk Oblast. The
integral character of the ecoregions identified
permits them to be used for seeking new links,
dependencies, regularities, etc. In this context,
the identified ecoregions are also useful for
studying the spatial peculiarities of carbon cy-
cling in Irkutsk Oblast.
USE OF REGIONALIZATION
FOR STUDYING CARBON CYCLING
In the former Soviet Union, regionalization is
used mainly for studying energy-production
cycles within the implementation of economic
regionalization (Lavrovetal., 1985). At present,
when ecological regionalization is conducted,
cycles are studied in cascade systems, nodal
regions and ecological "town-region" systems.
Thus, Soviet geographical and ecological sci-
ences have accumulated some experience study-
ing the cycles. At the same time, the study of the
global and regional turnover of substance as
regards separate chemical elements has not yet
been comprehended in terms of ecological re-
gionalization. In this regard, it is proposed to
accommodate the system of methods of eco-
257
-------
REGIONALIZATION AS A TOOL FOR STUDYING CARBON CYCLING
logical regionalization developed by this au- tion (1st sub-system of methods). After the
thor to cognizing the spatial regularities of conceptual image of the ecoregion is created
cycles of separate chemical elements. Ecologi- and its main parameters relating to the specific
cal regionalization is most applicable to the character of carbon balance are established,
study of the carbon cycle because the carbon one can proceed to subsystems of methods of
balance undergoes substantial changes when measurement and identification of ecoregions.
passing from some ecosystems to others and The third sub-system of methods is realizable in
depends on the intensity of man's impact on the automatic mode with the use of computer
ecosystems. As aresult, there emerges arather techniques. Verification and correction of re-
mixed spatial mosaic of the variation in the suits obtained (4th subsystem) also allow imple-
amount of carbon flow between surface ecosys- mentation in the automatic mode;, however,
terns and the atmosphere that requires a sys- interpreting the identified ecoregions requires
tematization and generalization through eco- invoking a large number of specialists in eco-
logical regionalization.
^ - ±_t f.
logical regionalization as well as those engaged
with the study of the carbon balance in surface
ecosystems.
It is possible to identify ecological regions with
different carbonbalances in two ways: 1) on the
basis of empirically established relationships It is anticipated that regionalization can be
between natural-anthropogenic characteristics most effectively used for differentiating a terri-
of the soil-vegetation cover and the intensity of tory into ecoregions with a specific carbon
the carbon flow into the atmosphere (Kobak, balancewhenstudyinglargeareas of the former
1988); and 2) using regional models for the Soviet Union where it is impossible to investi-
carbon cycle variation in the biosphere (Tarko
et al., 1989).
In both cases, attention should be focused on
gate each ecosystem by experimental methods.
Further conclusions about the possibilities and
limitations of regionalization in this field can be
drawn only upon the implementation of par-
generating a conceptual model of regionaliza- ticular projects of ecological regionalization.
REFERENCES
Bailey, R.G., S.C. Zoltai, and E.B. Wiken. 1985. Ecologi-
cal regionalization in Canada and the United States.
Geoforum 16:265-275.
Blanutsa, V.1.1989. A problem approach to regionaliza-
tion: constructing the algorithm and experience of the
realization. Geografiya i prirodnye resursy 1:145-152 (in
Russian).
Blanutsa, V.1.1990. Regionalization as a tool for environ-
mental impact assessment in the USSR. In: IAIA'90.
Proceedings of the 9th Annual Int. Meeting of IAIA,
Lausanne, 27-30 June 1990. EPFL, Lausanne.
Kagansky, V.L. 1987. Methodological problems of region-
alization and its relations to geospace conceptions. In:
Issledovaniye Metodologicheskikh Problem Geografii v
Estonskoi SSR. Estonian Geographical Society, Tallinn,
pp. 89-94 (in Russian).
Kobak, K.1.1988. Biotical Compounds of Carbon Cycle.
Gidrometeoizdat, Leningrad (in Russian).
Lavrov, S.B., V. V. Pokshishevsky, and G.V. Sdasyuk. 1985.
Regionalization for planning in the USSR: concepts,
methods and practice. UNCRD, Nagoya.
Omernik, J.M. 1987. Ecoregions of the conterminous
United States. Ann. Assoc. Am. Geogr. 77:118-125.
Tarko, A.M., B.G. Bogatyrev, A.P. Kirilenko, and M.V.
Udalkina. 1989. Modelling of global and regional changes
of carbon cycle in the biosphere. Computing Centre of the
USSR Academy of Sciences, Moscow (in Russian).
258
-------
FRAMEWORK TO QUANTIFY THE NATURAL TERRESTRIAL
CARBON CYCLE OF THE FORMER SOVIET UNION
Tatyana P. Kolchugina and Ted S. Vinson
ABSTRACT
A framework was created to quantify the natural terrestrial carbon cycle of the former Soviet Union. The organization of the
carboncycleparameterandgeoreferenceddatabasewhichsupporttheframeworkandthecalculationswhicharerequired
to establish the carbon budget are performed with personal computer hardware and commercially available spreadsheet
software. Basedon the framework, net primary productivity (NPP)fortheFSUwas estimated at 6.17 ± 1.65 Gt (Ifftons)
C/yr, the vegetation carbon pool at 118.1 ± 28.5GtC, the litter carbon pool at 18.9 ± 4.4GtC, and total soil carbon pool
at 404.0 ± 38.0 Gt C. The components of the carbon budget obtained with the framework were in good agreement with
estimates from other published sources. Within the framework one may determine: i) the extent of forest and agricultural
ecosystems that can be technically managed on a sustainable basis to conserve and sequester carbon, and ii) the role of the
former Soviet Union in the global carbon cycle.
INTRODUCTION
Natural processes in oceans and terrestrial eco-
systems, together with human activities, have
caused a measurable increase in the atmo-
spheric concentration of CO2. In 1988, the
atmosphere contained 748 Gt of carbon, the
largest amount during the past 160,000 years
(Post etal., 1990). On average, the CO2 concen-
tration is increasing by 3.0 Gt C/yr (IPCC,
1990).
Changes in the chemical composition of the
atmosphere are brought about by three global
system components: industry, oceans and ter-
restrial ecosystems (Bolinetal., 1986; Houghton
and Woodwell, 1989). Burning of fossil fuels
such as natural gas, oil and coal accounts for 5.3
Gt per year and is the major source of CO2
increase. Land-use modificationmay add as much
as 2.6 Gt of carbon per year (Tans et al., 1990).
The long-term ecological consequences of the
change in the chemical composition of the
atmosphere are not fully understood; however,
a warmer global climate is highly probable
(Schneider, 1990). If CO2 concentrations were
to double, the earth's temperatures may rise
between 1° and 5° C (Schneider, 1990). Cli-
matic changes could be more pronounced in the
Northern Hemisphere (Etkin, 1990). Global
warming may accelerate the decay of organic
matter (Dixon and Turner, 1991).
In view of the potential to significantly disrupt
the equilibrium of the natural carbon cycle, it
may be necessary to develop international stra-
tegies to manage terrestrial carbon stores to
offset increased amounts of atmospheric CO2.
Before any international strategy can be formu-
lated, policies aimed at maintaining a desirable
carbon balance within national boundaries
would be required. The determination of a
carbon balance includes the quantification of
259
-------
FRAMEWORKTO QUANTIFY TOE NATURAL TERRESTRIAL CARBON CYCLE OF THE FORMER SOVIET UNION
the natural and anthropogenic contributions to stored in soil is found in boreal ecosystems
the carbon cycle within national bpundaries. (Billings, 1987). Grasslands are also an impor-
The quantification of the carbon cycle follow- tant component of the terrestrial.carbon cycle.
ing an assessment of carbon pools and fluxes is Despite the fact that grasslands do not accumu-
generaUy referred to as the carbon budget, late large quantities of plant mass (compared
Carbonbudgets recentlyhavebeenestablished with forest ecosystems), they exhibit high net
for Sweden (Eriksson, 1991) and the forest primary productivity (NPP) and, therefore, may
sectors of Canada (Apps et al., 1991) and the influence the terrestrial carbon cycle.
United States (Gucinski et al., 1991).
Peatlands, which are wetlands where peat is
TheformerSovietUnion(FSU)wasthelargest accumulating, store a significant amount of
country in the world. It occupied one-sixth of carbon. Organic soil carbon content reaches
thelandsurfaceoftheearth. Anunderstanding 2,000 t/ha (Bonn, 1982). The FSU has the
of the carbon budget of the FSU is essential to greatest expanse of peatlands in the world
the development of international strategies (Tyuremnov, 1976). Wetlands are known to be
aimed at mitigation of the negative impacts of a source of methane to the atmosphere (Bartlett
. -j. _i -inoc~ -inocu. XJ^^^-ioc- o+ ol 1QQ^ A1_
global climate change.
The territory of the FSU is represented by a
variety of climate conditions. The major part of
the territory is in the Boreal and temperate
climatic zones. The climate in the FSU changes
from arctic and subarctic in the North to sub-
tropical and desert in the South. From west to
east, the climate makes a transition from mari-
time to continental to monsoon.
Matthews (1983) identified eightprincipal types
of vegetation in her global data base: forest,
woodland, shrubland, grassland, tundra, desert,
marsh/swamp and cultivated land. The vegeta-
tion of the FSU is represented by a variety of
formations, including all of these major types.
Arctic deserts and tundra formations are found
in the northern regions of the FSU; deserts and
semi-deserts occur in southern regions. A vast
area, the largest of any country in the world, is
occupied by forests and grasslands. The total
area of forest zone under State supervision (in
1983)is l,259millionhectares (Vorobyov, 1985),
which is approximately 56.5 percent of the
territory of the country. About 95 percent of the
forest area is in Russia. Tundra and boreal
forests store a significant amount of organic
et al., 1985a, 1985b; Harriss et al., 1985). Al-
though the atmospheric concentration of CH4
is much lower than the concentration of CO2,
CH4 is 20 times more effective (per molecule)
than CO2 as a greenhouse gas (Blake and
Rowland, 1988).
Anthropogenic carbon emissions in the FSU
have recently been estimated at approximately
!Gt(MakarovandBashmakov, 1990). Despite
an abundance of Soviet data on carbon-cycle
parameters in desert, tundra, forest and grass-
land ecosystems (Bazilevich et al., 1970;
Kazimirovand Morozova, 1973; Aleksandrova,
1977;Tytlyanova, 1977; Vatkovskyi, 1976; Dylis
andNosova, 1977; Kazantseva, 1980; Bazilevich,
1986; Bazilevich et al., 1986; and others), the
national carbon budget of the FSU, specifically
the natural component, has only recently been
estimated, as described herein.
THE NATURAL TERRESTRIAL
CARBON CYCLE
The natural terrestrial carbon cycle consists of
a combination of carbon pools and fluxes as
shown in Figure 1. The pools are carbon stores
in soil and vegetation, including living vegeta-
matter. Twenty-seven percent of the approxi- tion (i.e. phytomass) and plant detritus (i.e.
mately 80 percent of terrestrial organic matter mortmass and litter). In the present study, the
260
-------
T.P. KOLCHUGINA AND T.S. VINSON
Equilibrium
Assumntion:
Ph
PS
Pools
- Phytomass
- Mortmass
- Litter
-Soil
- Protected soil
organic matter
Influxes
Fj -Net Primary
Productivity (NPP)
F2 - Mortmass formation
(stem, branch, & root
detritus)
F4 - Foliage formation
(i.e. leaf litter fall)
F6M L - Total formation of
soil organic matter
(from mortmass[M]
and litter[L])
Fg - Formation of protected
organic matter
Figure 1. Simplified equilibrium natural ecosystem carbon cycle.
261
F9«F7
Effluxes
F3 - Mortmass
decomposition
F5 - Litter
decomposition
F? - Unprotected Soil
organic matter
decomposition
F9 - Protected soil
organic matter
decomposition
F10 - Autotrophic
respiration
-------
FRAMEWORKTO QUANTIFY THE NATURAL TERRESTRIAL CARBON CYCLEOFTHE FORMER SOVIET UNION
term mortmass was used to describe coarse a climatic belt or subbelt. Nine biomes within
above-groundandbelow-groundwoody debris, the FSU were identified: polar desert, tundra,
,™ . ».„ a.i._ j~c:~.<»+v,AimnaT-Grai fr»rpct-tnnHra/snarsetaiea.taisa.mixed-decidu-
The term litter was used to define the upper soil
layer comprised of fine woody debris and
forest-tundra/sparse taiga, taiga, mixed-decidu-
ous forests, forest-steppe, steppe, desert-semi-
icLYwi V»VMU.L/JL.I.OW\JI u *-*-•-«* »»w**^ *.»«— - ' - - —
leaves that are not completely decomposed, desert and subtropical woodlands.
The effluxes are carbon emissions resulting
from plant respiration and decomposition of Three approaches may be used to identify
organic matter. The processes of formation of ecoregions, namely the use of: (1) maps with
new organic matter in soil and vegetation (i.e. specific information on soil and vegetation and
humus, foliage formation and NPP) represent other attributes which relate to carbon cycling
carbon influxes. in ™. ecoregion, (2) satellite imagery and re-
mote sensing techniques, and (3)potentialmap-
The NPP equals the difference between gross ping (i.e. spatially distributed data on soil, cli-
photosynthesis (GPP) and respirotionofoutotro- mate, precipitation and elevation are used to
phic organisms (RA The RA amounts to 44 to predict potential vegetation and biomass). A
* ^ „ " v /~rr i _1_ , 1^,1—„.<-!—*,-. n£ 4-V*e*r*£* *-»t-*r-*r*^\O/>Tl^C TT1Q\/ Cllcr\ h^
52 percent (48 percent on average) (Kobak,
1988) of GPP. Root respiration (R^) comprises
one-third of the RA. The parameter that char-
acterizes carbon storage in ecosystems is called
net ecosystem productivity (NEP). NEP equals
the difference between NPP and carbon loss
resulting from heterotrophic respiration (RH).
Carbon fluxes can be measured or calculated.
However, when carbon effluxes are measured,
the contribution from different processes can-
not be distinguished. For example, when soil
carbon efflux ismeasured, itis difficultto distin-
guish between effluxes resulting from R^ and
RH. The quantitative method allows one to
separate fluxes.
METHODOLOGY TO ESTABLISH
THE NATURAL CARBON BUDGET
combination of these approaches may also be
used. For example, ecoregions identified with
satellite imagery and remote sensing techniques
may be validated with maps of soil and vegeta-
tion.
Carbon cycle parameters have been quantified
by soil, agricultural and forest scientists, ecolo-
gists and botanists for several decades. The
carbon cycle parameters may be expressed in
terms of carbon content (for pools) or rate (for
influxes or effluxes)/ha for a variety of soil-
vegetation complexes. If the soil-vegetation
complexes are related to the natural attributes
identified on maps, which are used to isolate
ecoregions, the carbon budget for an ecoregion
can be established simply by multiplying the
area of the ecoregion (in hectares) by the car-
bon content(s) and flux(es). The carbon con-
tents and fluxes for all the ecoregions may be
summed to arrive at the carbon budget for a
larger region, biome or nation.
Based on the preceding discussion, the frame-
To establish the natural carbon budget, the
geographic area within which it was quantified
was isolated. The tenaecoregion was applied to
the boundaries and area! extent of the geo-
graphic area. The term ecosystem was applied work shown in Figure 2 was created to assess a
to the combination of certain soil-vegetation natural component of the carbon budget. Im-
formationswithinanecoregion.Theconceptof tially, maps were used to isolate ecoregions
an ecosystem is a broad one; its function is to (Frames 1 and 2) and data bases which contain
emphasize obligatory relationships and inter- natural carbon cycle parameters were corn-
dependence (Odum, 1953). The term biome piled (Frame 3). The areal coverage of the
was applied to the complex of ecosystems within ecoregions was integrated with the carbon con-
262
-------
T.P. KOLCHUGINA AND T.S. VINSON
( 1 Collect & Digitize Maps
(Ryabchikov,1988)
(Isachenko,1988)
(Cherdantsev, 1961)
(Kolchugina, 1991;
after Gerasimov, 1933,
and Bazilevich, 1986)
2 Geographic Information System (GIS)
Analysis to Identify
Ecoregions within Georegions
Data Bases for
Natural Carbon Cycle
(Bazilevich, 1986)
(Kobak, 1988)
Correlation and Integration of Carbon Data Bases and GIS Analysis
to estimate Carbon Pools and
Fluxes for Ecoregions
Integration of Ecoregion data to assess
Carbon Pools and Fluxes for Biomes
within Georegions
V
Integration of Ecoregion data to assess
Carbon Pools and Fluxes for Biomes
within the former Soviet Union
ijgure 2. Framework to assess the natural (biogenic) component of the carbon budget.
tent and flux data bases to establish the carbon
budget within the ecoregion (Frame 4). The
organization of the carbon cycle parameter
data base, the hectare data for the ecoregions,
and the calculations that were required to es-
tablish the carbon budget, wereperformed with
personal computer hardware and commercially
available spreadsheet software (Microsoft Cor-
poration, 1991) in a Windows™ environment.
The carbon budgets for the ecoregions were
summed to establish,the carbon budget for a
biome or the entire territory of the FSU (Frames
5 and 6). The specific activities related to the
execution of these steps are discussed in the
following paragraphs.
Isolation of Ecoregions (Frames 1 and 2)
About 95 percent of the territory of the FSU,
including Russia, Ukraine, Belorussia,'
Kazakhstan and the Baltic states, was catego-
rized by the soil-vegetation type of the ecosys-
tem, the presence of wetlands and cultivation
intensity. Maps containing information on the
distribution of zonal soil-vegetation associa-
tions within the FSU (Ryabchikov, 1988), dis-
tribution of wetlands (Isachenko, 1988) and
cultivation intensity of arable lands
(Cherdantsev, 1961) were digitized and com-
puter-superimposed with a geographical infor-
mation system (GIS) (Burrough, 1986). These
maps are shown in Figure 3. The map with the
distribution of soil-vegetation associations
263
-------
FRAMEWORKTO QUANTIFY THE NATURAL TERRESTRIAL CARBON CYCLE OF THE FORMER SO
iVIET UNION
Wetlands
Figure 3. Digitized maps used to define ecoregions within the former Soviet Union.
Arable Lands
(Ryabchikov, 1988) provided thebasis for ecore-
gion isolation. In addition, eight georegions:
Near Ocean; Eastern, Middle and Western
Siberia; Eastern, Central and Western Europe;
and Kazakhstan were defined to accommodate
geodependence of carbon accumulation
(Gerasimov, 1933; Bazilevich, 1986). These
georegions were also mapped, digitized and
computer-superimposed with the GIS. After
computer superimposition of the four maps
noted above, more than 70 ecoregions related
to different ecosystems (i.e. soil-vegetation as-
sociations) were identified. Wetland, flood-
plain, mountain ecosystems and arable land
were isolated within these ecoregions.
TheecosystemspresentedbyRyabchikov(1988)
were aggregated in the nine biomes given in
Table 1. The polar desert biome included areas
covered by ice and stony barrens. The tundra
biome included herbaceous and shrub tundra
formations of polar and subpolar belts on dry
cryic, arctic, peat, turf, gleyic, "podbur" and
podzolized soils. Theforest-tundra/sparse taiga
biome included forest ecosystems within the
subpolar climatic belt and northern areas of the
boreal climatic belts. This biome unified eco-
systems with sparse forest cover on peat-turf,
podzol and "podbur" soils. The taiga biome
included ecosystems within the boreal climatic
belt with light- or dark-crown coniferous forest
vegetation of northern, middle and southern
subzones, mainly on podzol andpodzolic soils.
The mixed-deciduous forest biome unified
mixed (coniferous -broad-leaf or small-leaf) or
broad-leaf forests on podzolic or gray soils
within the subboreal climatic belt. The forest-
steppe biome included mixed coniferous-de-
ciduous forests and grasslands of the subboreal
climatic belt. Gray soils are characteristic for
the northern part of the forest-steppe biome
and chernozem soils for the southern part. The
steppe biome included grasslands on chernozem
and castanozem(i.e. chestnut) soils. The desert-
semidesert biome included shrub-grass and
shrub-tree desert formations of subboreal and
subtropical belts on castanozems, yermosols,
sands and primitive desert soils. The subtropi-
264
-------
T.P. KOLCHUGINA AND T.S. VINSON
Table 1. Terrestrial ecosystems aggregated into major biomes in the former Soviet Union.
Biome
Terrestrial Ecosystems (after Ryabchikov, 1988)
Polar Ice (continental icecaps and glaciers)
desert Stony barrens
Tundra Arcto-tundra on dry cryic arctic soils
Mossy meadows on peat soils
Grass-moss-shrub tundra on gleyic and turf soils
Shrub-moss tundra on gieyic and podbur soils
Shrub-moss and shrub tundra on peat podzolized soils and podburs
Forest- Shrub-Meadow and small-leaf open woodlands on coarse humic-turf soils
tundra/ Shrub and dark-crown coniferous woodlands on gleyic-ferro-humic podzols and podburs
sparse Light- and dark-crown coniferous and shrub woodlands on ferro-humic podzols and podburs
taiga Ug^t-«m coniferous woodlands on ferro-hurnic podzols and podburs
Grass-herb shrub meadows on peat-turf soils
Meadow-tall grass small-leaf open woodlans on turf soils and podburs
Dark-crown coniferous humid taiga boreal forest on illuvial humo-ferric podzols and podburs
Dark- and light-crown coniferous moderately humid taiga: (a) northern subzone on illuvial ferro-humic and cryic podzols;
(b) middle subzone - on podzolic soils and ferric podzds; (c) southern subzone-on turf-podzolic soils and ferro-humic podzols
Light- and dark-crown coniferous moderately humid taiga: (a) northern subzone -on ferro-humic podzols and podburs;
(b) middle subzone - on coarse- humic podzolic and pale soils; (c) southern subzone - on turf podzolic soils
Light-crown coniferous taiga: (a) middle subzone - on cryic taiga and pale soils; (b) southern subzone - on turf-podzolic soils
Mixed- Deciduous-coniferous excessively humid forest on acid cambisok and podzolic soils
deciduous Coniferous-broad-leaf humid forests on ferric podzols and turf- podzolic soils
forest Broad-leaf-coniferous humid forests on cambisols and turf- podzolic soils
Broad-leaf-coniferous moderately humid forests on turf-podzolic soils
Small-leaf-coniferous forests (subtaiga) on turf-podzolic soils and gray luvisols
Broad-leaf humid forests on turf-podzolic soils and orthic cambisols
Broad-leaf moderately humid forests on turf-podzolic soils, gray luvisols and acid cambisols
Broad-leaf moderately humid open forests (with the admixture of conifers) on meado solonetzic soils
Forest- Small-leaf-coniferous forest-steppes on podzolized and luvic chernozems and gray luvisols
steppe Broad-leaf forest-steppes (and praries) on meadow chernozems and chernozemic soils
Broad-leaf forest-steppes on gray luvisols, orthic and luvic chernozems
Deciduous-coniferous forest-steppes on gray luvisols, chernozemic soils and chernozems
Coniferous-small-leaf forest-steppes on chernozems and eutric castanozems
Steppe Grass-herb steppes on orthic and southern chernozems
Grass steppes on orthic and southern chernozems
Shrub-grass-turf setppes on castanozems
Grass-herb steppes on meadow-chernozemic and chernozemic (often solonetzic) soils
Grass steppes on orthic and southern chernozems and eutric (often solonetzic) castanozems
Shrub-grass-turf steppes on castanozems and sdonetzs
Grass-semismaUshrub steppes on orthic and dyrstic castanozems
Desert- Smallshrub-grass semideserts on dyrtric castanozems and cambic yermosols (often solonetzic)
semi Shrub and smallshrub semideserts on dystric castanozems and cambic yermosols
desert Semishrub and shrub semideserts on cambic and orthic yermosols
Shrub and smallshrub semideserts and deserts on cambic and orthic yermosols
Shrub deserts on desert soils
Shrub-tree deserts on sands and primitive soils
Mountainous
subtropical
woodlands
Deciduous foiest-coniferousforest-alpine meadow / meadow
Mediterranean woodlands-mixed forest-coniferous forest-steppe meadow
265
-------
FRAMEWORK TO QUANTIFY THE NATURAL TERRESTRIAL CARBON CYCLE OF THE FORMER SOVIET UNION
Table 2. Mountain ecosystems isolated within
terrestrial biomes
Mountain Ecosystems (after Ryabchikov. 1988)
•Polar desert
-•nrndra/Pdar desert
-Shrub-Polar desert
•Skub-TAmdra
-Forest-ineadcw/Meadow-tuiidra
-Openwoodland/nindra
-Humid taiga/Rmdra
•Taiga/Meadow-tundra
-Taga/I\mdra
•Coniferous Fbrest/Rindra
-Mixed Forest/Coniferous Forest/I\mdra
•Mixed Forest/Coniferous Forest/Shmb-ineadow
-Mixed Forest/Coniferous Forest/Meadow
•Mixed Forest/Coniferous Forest/Alpine meadow
-Deciduous Forest/Coniferous Forest/Meadow
•Broad-leaf Forest/Coniferous-Forest/Alpjne meadow
•Forest-steppe/OoniferousForest/Nfeadow-
•Scmidesert/Steppe/Desert
calbiome included mountainous formations of
the subtropical belt.
The mountain ecosystems isolated by
Ryabchikov (1988) within the nine biomes are
presented in Table 2. Mountain ecosystems are
described by a combination of different vegeta-
tion formations. The vegetation of mountain
ecosystems makes the transition from zonal at
low altitudes to meadow, tundra, or polar desert
at high altitudes.
The map that was used to isolate peatlands
(wetlands) (Isachenko, 1988) allows one to
determine the total area of peatlands, but does
not allow different peatland landscapes to be
distinguished. There are at least three main
classifications of peatland systems in the FSU
(Botch and Masing, 1983). According to the
trophic conditions and the developmental stage,
peatlands can be classified as eutrophic, me-
sotrophic or oligotrophic. According to the
hydrological conditions, peatlands may be di-
vided into minerotrophic (both ground and rain
water supply) and ombrotrophic (rain water
supply). According to the main layer of plant
communities, peatlands can be divided into
moss, graminoid, dwarf-shrub and shrub. Car-
bon accumulation differs depending on peatland
type. Future studies will require the incorpora-
tion of maps that would allow one to isolate
different types of peatlands. However, for the
present, the superimposition of peatland and
soil-vegetation maps allowed the determina-
tion of peatland type within a given ecosystem
and the specification of carbon accumulation
parameters.
Data Bases for Natural Carbon Cycle Param-
eters (Frame 3)
Bazilevich (1986) compiled a data base on
carbon accumulation in vegetation from stud-
ies of 1,500 vegetation complexes in the FSU.
The data base is a comprehensive source of
information on all vegetation formations in the
FSU, namely, 13 polar desert and tundra, 40
forest, 57 grassland, 20 mire ecosystems and
more than 50 desert-semidesert formations.
These vegetation complexes were correlated to
the ecosystems presented by Ryabchikov (1988).
The data base provides site-specific values for
total phytomass content and phytomass pro-
ductivity for all vegetation formations in the
FSU. The data base allows the assessment of
phytomass and phytomass increment alloca-
tion. Phytomass was categorized as green-
assimilating, woody above- (stems andbranches)
and below-ground (roots and buried stems)
parts of plants. Mortmass was categorized as
woody above- (dead stems, branches, grass,
windfall) and below-ground parts of plants and
litter. Productivity of phytomass or NPP was
categorized in the same manner as phytomass.
The net carbon content of plant mass was
assumed to be 50 percent (Kobak, 1988). This
percentage was used to calculate the net carbon
storage and rates of carbon accumulation in
vegetation.
266
-------
T.P. KOLCHUGINA AND T.S. VINSON
Table 3. Vegetation formations associated with major biomes in the former Soviet Union.
Biome
Polar desert
Arctic tundra
Forest-tundra/sparse
taiga
Taiga
Mixed-deciduous forest
Forests-steppe
Steppe
Desert-semidesert
Subtropical woodlands
Vegetation Formations (after Bazilevich, 1986)
Polar arctic deserts (Europe, Siberia); northern and southern (East, Middle and West Siberia) typical
(Middle Siberia), southern (Near Ocean Region; East and West Europe; East and West Siberia);
Peatlands of tundra; Meadows of tundra;
Birch (Western Europe), spruce (Eastern Europe), and larch (Western and
Middle Siberia) forest-tundra;
Peatlands of forest-tundra;
Meadows of forest-tundra
Northern, middle, and southern spruce taiga (Europe); northern (Western and Middle Siberia) and
southern (Western and Middle Siberia; Near Ocean region) polydominant spruce, fir and cedar (Pinus
siberica, P. koraiensis) taiga; northern (Europe; Middle and Eastern Siberia; Near Ocean region),
middle (Siberia), and southern (Middle Siberia; Europe) larch taiga; northern (Western and Central
Europe; Middle Siberia), middle (Eastern and Western Europe, Western and Middle Siberia), and
southern (Europe; Western and Middle Siberia) pine taiga; Peatlands of taiga; Meadows of taiga;
Coniferous broad-leaf: pine, oak, and birch (Europe); coniferous broad-leaf mixed
forests (Near Ocean region); broad-leaf forests (Europe; Near Ocean region);
small leaf: birch (Near Ocean region; Western Siberia); aspen (Europe)
Mires of mixed and broad-leaf forests;
Meadows of mixed and broad-leaf forests
Broad-leaf forests (Europe); aspen (Europe); birch (Western Siberia); and pine
(Europe; Western and Middle Siberia).
Grasslands: meadow (Europe, West and Middle Siberia); real moderately dry (Europe)
Meadow-steppe (West Siberia; Europe)
Zonal, solonchak & solonetz moderately dry steppe (Europe; West and East Siberia; Kazakhstan)
Zonal, solonchak & solonetz northern dry steppe (Europe; West Siberia; Kazakhstan)
Zonal and solonetz dry steppe (Europe; West & East Siberia; Kazakhstan)
Zonal and solonetz southern dry steppe (Europe, Kaiakhstan)
Meadows (continental and floodplain, solonetz & solonchak) of steppe (Europe, Siberia, Kazakhstan)
Steppified semideserts (Europe)
Zonal & solonetz semidesert (Europe, Kazakhstan)
Northern psammophytic semidesert (Europe, Kazakhstan)
Meadows of semidesert (Kazakhstan)
Floodplain solonchak semidesert (Kazakhstan)
Zonal semishrub desert (Kazakhastan)
Northern psammophytic desert (Kazakhstan)
Succulent desert (Turkmenia)
Solonchak meadows of northern desert (Kazakhstan)
Coniferous-broad-leaf forests; oak-lime tree forests (Europe) and meadows of coniferous-broadleaf
forest subzone (Europe)
Broadleaf forests of forest-steppe subzone and real dry steppes (Europe)
Coniferous-broad-leaf forests (Europe, Near Ocean Region) and Meadow-steppe (Europe)
The descriptions of vegetationformations given
by Bazilevich (1986) for the eight georegions
considered are presented in Table 3. Polar
deserts and tundra formations include arctic
and subarctic deserts and tundra. Forest-tun-
dra and taiga forests include forests composed
of spruce, larch, fir, pine and cedar (Pinus
sibirico, P. koraiensis). Mixed-deciduous for-
ests, forests of the forest-steppe zone and sub-
tropical woodlands are represented mainly by
pine, oak, birch and aspen. Steppe formations
are represented by meadow-grasslands, mod-
erately dry and dry steppes. Desert-semidesert
formations include steppified semideserts and
267
-------
FRAMEWORK TO QUANTIFY THE NATURAL TERRESTRIAL CARBON CYCLE OF THE FORMER SOVIET UNION
semishrub and succulent deserts; zonal and
psammophytic deserts andsemideserts, includ-
ing solonchak formations. Carbon accumula-
tion data are also provided for the mires and
continental and floodplain meadows of tundra,
open woodlands, taiga, mixed and broad-leaf
forests, steppes, deserts and semideserts. Data
for meadow-grasslands and moderately dry
grasslands reported by BazUevich (1986) were
used to quantify the carbon cycle in the forest-
steppe biome.
Kobak's (1988) data base was used to charac-
terize the soil component of the carbon cycle.
The data base resulted from the analysis of
published soil data. About 70 different Soviet
and foreign sources were included (Kononova,
1963; Schlesinger, 1977; Bohn, 1982; Post et aL,
1982; Kobak and Kondrashova, 1986; and
others). Soil carbon cycle parameters for more
than 40 soil types of polar, boreal and tropical
belts represent the averages from the empirical
data presented in different sources. However,
in some cases, specific data were selected. For
example, Soviet data were used to characterize
the carbon contents of podzol and chernozem
soils. The data base includes: 1) carbon
contents of soil (total and stable portion to a
depth of one-meter for inorganic soils and for
the entire depth of bog organic soils); 2) the
annual rate of foliage and humus formation
(total and stable portion); and 3) CO2 efflux
from soils. Kobak (1988) reports the following
names of soil types that were related to the
polar desert and tundra biomes: arctic, tundra
gleyic and bog soil. Soils of the forest biomes
were represented by cryic-taiga and podzolic
soils, brown and gray forest soils, chernozems,
peatland soils, mountain-meadow, and "flood-
plain soils. Soils of the steppe biome were
chernozem, chestnut, floodplain, bog and solo-
netz soils. Chestnut (castanozem) and gray-
brown desert soils were related to the desert-
semidesert biome. The combinations of moun-
tain-meadow, podzolic, gray forest and chestnut
soils were used to describe soils of mountainous
formations.
Matthews and Fung (1987) compiled a data
base on methane emissions from natural wet-
lands. Three sources are used to assess the
global distribution of wetlands: 1) vegetation
classified according to Matthews (1983); 2) soil
properties (Zobler, 1986); and 3) fractional
inundation from a global map survey of Opera-
tional Navigation Charts (ONC). Published
data were analyzed to compile: 1) typical
methane emissions for wetland ecosystem type,
and 2) length of the active season (Svensson,
1980; Bartlett et aL, 1985a, 1985b; Sebacher et
al., 1986; and others). Data on methane emis-
sions from wetlands (Matthews and Fung, 1987)
were correlated to the data on wetland distribu-
tion within the nine biomes in the FSU.
Correlation of Data Bases to Mapped Ecosys-
tems (Frame 4)
The maps and data bases are not specifically
designed for carbon cycle quantification. How-
ever, the names of soil-vegetation associations
reported by Ryabchikov (1988) (Table 1) corre-
sponded well with the descriptions of vegeta-
tion formations given by Bazilevich (1986) and
soil types given by Kobak (1988). For example,
for the ecosystem named "Lichen-moss tundra
on gleyic and podbur soils" isolated in the
Central European georegion in accordance with
the Ryabchikov (1988) map, the following
name from Bazilevich's (1986) data base was
used: "Southern subarctic tundra (West and
East Europe);" the soil type was described as
"Tundra-gleyic soil." For the ecosystem named
"Dark- and light-crown coniferous moderately
humid taiga, middle subzone, on podzolic soils
and ferric podzols," the name "Pine and spruce
forests of northern taiga; fir forests of middle
taiga, and larch forests of southern taiga (Cen-
tral Europe)" was used; the soil type was de-
scribed as "Podzolic soil." For the ecosystem
named "Broad-leaf-coniferous moderately hu-
mid forests on rurf-podzolic soils," the name
268
-------
"Coniferous-broad-leaf forests and oak-lime
forests (Europe)" was used; the soil type was
described as "Podzolic soil." For the ecosystem
in the Kazakhstan georegion named "Flood-
plain formations of grass steppes on orthic and
southern chernozems," the name "Floodplain
solonchak meadows and mesophytic meadows
of steppe zone (West Siberia, Kazakhstan)"
was used; the soil type was described as "Cher-
nozem and solonetz soils." For the wetland
formation identified in the ecosystem named
"Broad-leaf moderately humid forests on turf
podzolic soils, gray luvisols and acid cambisols,"
the name Teatlands of broad-leaf-coniferous
forests" was used. When a direct correspon-
dence did not exist, certain extrapolations and
interpolations were made. For example, in the
West Siberian georegion for the ecosystem
called "Mountainous semidesert/steppe/
desert," isolated with the help of the Ryabchikov
(1988) map, the combined name from
Bazilevich's (1986) data base "Subboreal
semideserts; zonal meadows of subboreal
steppified semideserts; northern semishrub
subboreal deserts" was used, the soil type was
described as "Mountain-meadow and gray-
brown desert soils."
Integration of Ecoregion Areas and Carbon
Data Bases (Frame 5 and 6)
Carbon pools and fluxes for natural ecosystems
in the FSU were estimated by integrating the
carbon data bases and the GIS analysis results
(hectare data) using commercially available
spreadsheet software (Microsoft Corporation,
1991). Productivity of green-assimilating parts
of plants was used to characterize the rate of
foliage formation. Data on carbon content in
peatland were integrated with hectare data on
the extent of wetlands within biomes. Low,
mean and high estimates were made for each of
the eight georegions by summing the contribu-
tions from each ecoregion. The carbon pools
and influxes for the biomes in the FSU were
obtained by summing the georegion totals for
T.P. KOLCHUGINA AND T.S. VINSON
the nine biomes. Initially, it was assumed that
1) natural ecosystems are presently in a state of
equilibrium, (i.e. NEP equals 0); 2) forests
totally cover the area of ecosystems (excluding
arable land) within forest-tundra/sparse taiga,
taiga and mixed-deciduous forest biomes; and
3) forests occupy one-half of the area (exclud-
ing arable land) of the forest-steppe biome.
Carbon effluxes were calculated from the in-
fluxes assuming that all ecosystems were ini-
tially in an equilibrium state (NPP equal to RH).
Mortmass decomposition was assumed to be
equal to mortmass production. In turn,
mortmass production was assumed to be equal
to phytomass production (NPP and production
of different parts of plants). Carbon efflux from
litter decomposition was calculated as the dif-
ference between foliage formation (green-as-
similating parts production) and the sum of
total humus formation and peat accumulation.
Carbon efflux from soil organic matter decom-
position were calculated as the difference be-
tween total and stable humus formation. Car-
bon efflux from R^ was calculated from the
NPP, assuming that R^ comprises one-third of
the total RA, and RA comprises 48 percent (on
average) of the GPP; NPP equals the difference
between GPP and RA. The sum of R^ and RH
(below-ground mortmass, litter, and soil or-
ganic matter decomposition) were compared
with field measurements of the surface soil
carbon efflux (Kobak, 1988).
The estimates of carbon cycle parameters ob-
tained in the present study were compared,
where possible, with estimates using other data
bases. The global data base for the NPP of
terrestrial ecosystems, compiled with the help
of advanced very high resolution radiometer
(AVHRR) data (Fung et al., 1987), was incor-
porated in the study. The NPP for natural (non-
arable) ecosystems was calculated by dividing
NPP totals for the ecoregion by the total num-
ber of hectares and multiplying by the number
of hectares of natural ecosystems within the
269
-------
FRAMEWORK TO QUANTIFY THE NATURAL TERRESTRIAL CARBON CYCLE OF THE FORMER SOVIET UNION
100000
10000 • •
1000 • •
100 ••
10 ••
1 -
0.1
I
Is
1^1 Above-ground Phytomus
£
-o
f
Below-ground Phytomass
• Above-ground Mortmass (with 1^1 Below-ground Mortmass (with
Litter) Litter)
Figure 4. Distribution of biomass (phytomass and mortmass) in terrestrial biomes of the former Soviet Union.
ecoregion.
CARBON POOLS AND FLUXES
IN THE FSU - EXAMPLE RESULTS
The vegetation carbon pool of natural terres-
trial ecosystems in the FSU was estimated at
118.1±28.5 Gt C The vegetation carbon pool
included phytomass (91.0±22.0 Gt C) and
mortmass (27.1±6.5 Gt C). The litter carbon
pool was estimated at 18.9±4.4 Gt C. The soil
carbon pool was estimated at 404.0±38.0 Gt C
(including peatlands), with 269.0±25.0 in stable
form. Peatlands accumulated 148.0 Gt C.
The total productivity of phytomass of the nine
biomes was estimated at 6.17±1.65 Gt C/yr.
The productivity of green assimilating parts
based on Bazilevich's (1986) data was esti-
mated at2.51±0.58 Gt C/yr. Humus formation
was estimatedat257.0±94.0 MtC/yr (87.8±15.1
Mt C/yr in stable form). Methane production
in peatlands (77 Mha) was estimated at 7.57
±1.6 Mt C/yr. FSU biomes may be arranged in
four groups based on the distribution of bio-
mass (phytomass and mortmass): 1) polar-desert
biome; 2) tundra and forest-tundra/sparse
taiga biome; 3) taiga, mixed-deciduous forest,
forest-steppe and mountainous subtropical
woodlands biome; and 4) steppe and desert-
semidesert biome (Figure 4).
In the polar desert biome, mortmass was greater
than phytomass and was distributed equally
above and below ground. Above-ground
phytomass was significantly greater than be-
low-ground phytomass. In the tundra and the
forest-tundra/sparse taiga biomes, mortmass
also exceeded phytomass. In the tundra biome,
below-ground biomass was more developed. In
the forest-tundra/sparse taiga biome, phytomass
and mortmass were distributed in equal pro-
portions above and below ground.
In the forest biomes, phytomass was greater
than mortmass. Above-ground parts were
greater than below ground parts in both pools
(Figure 4). Below-ground phytomass was ap-
proximately the same as above-ground
mortmass. In the steppe and desert-semidesert
biomes, phytomass and mortmass did not differ
substantially. Below-ground parts were greater
than above-ground parts in both pools. In the
steppe biome, above-ground mortmass was
270
-------
T.P. KOLCHUGINA AND T.S. VINSON
cC
CD
s
C_3
10000
1000
D NPP (after Fung et al., 1987)
UH Green Parts Production (after
Bazilevich, 1986)
II Above-Ground Phytomass
Production (after Bazilevich,
1986)
• Total Phytomass Production
(after Bazilevich, 1986)
Figure 5. Comparison of NPP of natural ecosystems based on Fung et al. (1987) with phytomass productivity based
on Bazilevich (1986).
greater than above-ground phytomass, while in
the desert-semidesert biome, more living parts
of plants were found above-ground.
The NPP of natural ecosystems estimated with
the data base from Fung et al. (1987) was
compared with the estimate of phytomass pro-
ductivity based on data given by Bazilevich
(1986) (Figure 5). Total phytomass productiv-
ity (6.17 Gt C/yr) estimated from Bazilevich
(1986) was about 2.1 times higher than NPP
(2.98 Gt C /yr) obtained from Fung et al. (1987)
(the desert biomes represented an exception:
NPP obtained from Fung et al. (1987) was
greater than phytomass productivity obtained
from Bazilevich (1986)). The NPP obtained
from Fung et al. (1987) corresponded well only
withtheproductivityofgreen-assimilatingparts
given by Bazilevich (1986). The discrepancy
may be due to the fact that either root produc-
tivity or both root and above-ground woody
phytomass productivity are underestimated by
Fungetal. (1987). Further, a discrepancy in the
results could be related to the use of different
methods to identify carbon-quantifiable regions.
The rate of foliage formation estimated from
Kobak's (1988) data was 3.59±0.64 Gt C /yr.
The value was slightly greater than estimates
based on data from Bazilevich (1986) (Figure
6).
The carbon efflux from the litter decomposi-
tion estimated from the equilibrium analysis
was 2.25 Gt /yr. Soil organic matter decompo-
sition (excluding peatlands) was 0.17 Gt C/yr.
The RAJ. was estimated at 1.84 Gt C /yr. Below-
ground mortmass decomposition was estimated
at 2.47 Gt C/yr.
The carbon efflux resulting from litter and soil
organic matter decomposition (2.42 Gt C/yr)
was less than the soil surface CO2 efflux (3.55±
1.13 Gt C/yr) estimated for the nine biomes
based on data reported by Kobak (1988) (Fig-
ure 7). The aggregated efflux from R^ and
average efflux resulting from the decomposi-
tion of below-ground mortmass, litter, and soil
organic matter was estimated at 6.73 Gt C /yr.
This value was 30 percent greater than the
maximum CO2 efflux (4.68 Gt C /yr) estimated
from Kobak's (1988) data.
271
-------
FRAMEWORK TO QUANTIFY THE NATURAL TERRESTRIAL CARBON CYCLE OF THE FORMER SOVIET UNION
10000-tl
S
o
Foliage Formation (after Kobak,
1988)
Green Parts Production (after
Bazilevich, 1986)
Figure 6. Comparison of foliage formation based on Kobak (1988) and Bazilevich (1986).
Based on Avogadro's number (6.028 x 1023
molecules in a mole), methane emissions from
peatiands were equivalent to 3.76 x 1035 mol-
ecules of CH4. The total efflux from below-
ground mortmass, litter and soil organic matter
decomposition was equivalent to 3,437 x 1035
molecules of CO2. Assuming CH4 is 20-fold
more active than CO2 as a greenhouse gas,
methane emissions represented approximately
2.2 percent of the soil surface efflux estimated
in the present study. However, methane emis-
sions from wetlands could be underestimated
because many wetland areas were not accu-
rately mapped (Botch, 1991).
SUMMARY AND CONCLUSIONS
A framework was created to quantify carbon
pools and fluxes at the ecosystem, biome,
georegional or national scales. The compo-
10000-fl
„ 1000-
Estimated (present study)
Measured (after Kobak, 1988)
Figure 7. Comparison of CO2 soil flux based on field data (Kobak, 1988) with estimate from present study.
272
-------
T.P. KOLCHUGINA AND T.S. VINSON
nents of the carbon budget obtained with the
framework were in good agreement with esti-
mates from other published sources.
The greatest advantage of the framework is that
all elements identified may be easilyupdated in
a computer spreadsheet format, and the carbon
budget can be recalculated immediately there-
after with the quantitative model. Sensitivity
analyses may also be performed with the model.
Further, the model can be extended to accom-
modate changing climate scenarios and mitiga-
tion strategies.
REFERENCES
Aleksandrova,V.D. 1977. Geobotanicalregionalizationof
Arctic and Antarctic. Nauka Press, Leningrad.
Apps,MJ.,A.W.Kurz, and T.D. Price. 1991. Application
of a carbon budget model to strategic planning for the
effects of climate change on the Canadian forest sector.
This volume.
Bartlett, KB., D.S. Bartlett, and D.I. Sebacher. 1985a.
Methane flux from coastal salt marshes. Journal of Geo-
physical Research 90:5710-5720.
Bartlett, K.B., D.S. Bartlett, D.I. Sebacher, R.C. Harris,
and D.P. Brannon. 1985b. Sources of atmospheric meth-
ane from wetlands. Paper presented at the 36th Congress
- of the International Austronautic Federation, Stockholm,
Sweden, Oct. 7-12.
Bazilevich, N.I. 1986. Biological productivity of soil-
vegetation formations in the U.S.S.R. Bulletin of Academy
of Sciences of the U.S.S.R. Geographical Series 2:49-66.
Bazilevich, NX, O.S. Grebenshikov, and AA. Tyshkov.
1986. Natural association of geographical patterns with the
structure and functioning of ecosystems. Nauka Press,
Moscow.
Bazilevich, N.I., L.Ye. Rodin, and N.N. Rozov. 1970. Geo-
graphical aspects of biological productivity studies. Pro-
ceedings of5th Congress of GeographicalSocietyof USSR,
Leningrad.
Billings, W.D. 1987. Carbon balance of Alaskan tundra and
taiga ecosystems: past, present and future. Quaternary
Science Review 6:165 - 177.
Blake, D.R. and F.S. Rowland. 1988. Continuing world-
wide increase in trapospheric methane, 1978-1987. Science
239:1129-1131.
Bohn, H.L. 1982. Estimates of organic carbon in world
soils: H. Journal of Soil Science Society of Arizona 4:1118-
1119.
BolinB.,B.R.Doos,J.Jager,andRA.Warrick. 1986. The
greenhouse effect, climate change and ecosystems. Scien-
tific Committee on Problems of the Environment
(SCOPE)29. John Wiley and Sons, New York.
Botch, M.S. 1991. Carbon storage in peat based on
regionality of Russian mires. This volume.
Botch, M.S. and V.V.Masing. 1983. Mire ecosystemsin the
U.S.S.R. In: Ecosystems of the World Mires: Swamp,Bog,
Fen, and Moor. Vol. 4B, Regional Studies, edited by
AXP.Gore. Elsevier, New York, pp. 95 -152.
Burrough,PA. 1986. Principles of geographical informa-
tion systems for land resources assessment. Clarendon
Press, Oxford, New York, p. 193.
Cherdantsev,GN. 1961. Map: Arable land in the U.S.S.R.
in 1954. In: A Geography of the U.S.S.R.- Background to
a Planned Economy, edited by JP.Cole, F.C. German, p.
290.
Dixon, R.K. and D.P. Turner. 1991. The global carbon
cycle and climate change: responses and feedbacks from
below-ground systems. Environmental Pollution 73:245-
262.
Dylis, N.V. and L.M. Nosova. 1977. Phytomass of forest
biocenosis. Nauka Press, Moscow.
Eriksson, H. 1991. Sources and sinks of carbon dioxide in
Sweden. AMBIO 20(3-4) :146-150.
Etkin, D. 1990. Greenhouse warning: consequences for
Arctic climate. Journal of Cold Regions Engineering,
Amer. Soc. of Civil Engineers 4(l):54-66.
Fung, I.Y., CJ.Tucker, and KLCPrentice. 1987. Applica-
tion of advanced very high resolution radiometer vegeta-
tion index to study atmosphere — biosphere exchange of
CO2.JournalofGeophysicalResearch92(D3):2999-3015.
Gerasimov, I.P. 1933. About soil-climatical faces of the
flatlands of the U.S.S.R. and adjacent countries. Proceed-
273
-------
FRAMEWORK TO QUANTIFY THE NATURAL TERRESTRIAL CARBON CYCLE OF THE FORMER SOVIET UNION
ings of the V.V. Docuchayev Soil Institute, VoL 8, No. 5,
U.S.S.R. Academy of Sciences Press, Leningrad.
Gucinski, H., DJP. Turner, and G. Koerper. 1991. The
carbonfluxonforestedlandsoftheUnitedStates. Proceed-
ings of the IUFRO Conference on Integrating Forest
Information over Space and Time, Canberra, Australia.
Hamss,R.C.,E.Gorham,DJ.Sebacher,K.B.Bartlett,and
PA. Flebbe. 1985. Methane flux from northern peatlands.
Nature 315:652,653.
Houghton, RA. and G.M. Woodwell. 1989. Global
climatic change. Scientific American 260:36-44.
IntergovernmentalPanelonCumate Change (EPCC). 1990.
TJ. Houghton, GJ. Jenkins, and J.J. Ephraums, editors.
Climate change, the IPCC scientific assessment. Working
group 1 report, WMD & UNEP Univ. Press, Cambridge,
UK.
Isachenko,A.G.,ed. 1988. Landscape map of the U.S.S.R.
Institute of Geography, Leningrad State University,
Leningrad.
Kazantseva,TJ. 1980. Productivity and dynamics of above-
ground phytomass of deserts. Problems of Desert Utiliza-
tion 2:76-83.
Kazunirov, NJ. and R.M. Morosova. 1973. Biological
turnover of substances in spruce forests of Karelia, Nayka
Leningrad.
Kobak,KI. 1988. Biotical Compounds of Carbon Cycle.
Hydromcteoizdat Press, Leningrad.
Kobak,KJ.andN.Yu.Kondrashova. 1986. Distribution of
organic carbon in soils of the globe. Trudy GGI320:61-76.
Kononova, MM. 1963. Soil Organic Matter. Nauka Press,
Moscow.
Makarov,AA.andI.Bashmakov. 1990. TheSoviet Union:
carbon emission control strategies. In: Case Studies in
International Cooperation, edited by William U. Chandler.
Matthews, E. 1983. Global vegetation and land use: new
high - resolution databases for climate studies. Journal of
Climate and Applied Meteorology 22:474-487.
Matthews, E. and I. Y.Fung. 1987. Methane emission from
natural wetlands: global distribution, area, and environ-
mental characteristics of sources. Global Biogeochemical
Cycles l(l):61-86.
Microsoft Corporation. 1991. Microsoft Excel, Redmond,
Washington.
Odum, E.P. 1953. Fundamentals of Ecology.
Saunders C., Philadelphia-London.
W.B.
Post, W.M., W.R. Emanuel, P.I. Zinke, and A.G.
Stagenberger. 1982. Soil carbon pools and worldUfe zones.
Nature 298(5870): 156-159.
Post, W.M., Tsung - Hung Peng, W.R. Emanuel, A.W.
King, V.H. Dale, and D.L. DeAngelis. 1990. The global
carbon cycle. American Scientist 78:310-326.
Ryabchikov,A.M.,ed. 1988. Map: Geographical Belts and
Zonal Types of Landscapes of the World. School of
Geography, Moscow State University, Moscow.
Schlesinger, W.H. 1977. Carbon balance in terrestrial
detritus. Ann. Rev. Ecolog. Syst 8:51-81.
Schneider, S.H. 1990. Global Warming. Vintage Books,
New York..
Sebacher, D.I., R.C. Harris, K.B. Bartlett, S.M. Sebacher,
and S.S. Grice. 1986. Atmospheric methane sources:
Alaskan tundra bogs, an alpine fen, and a subarctis boreal
marsh. Tellus 38B:1-10.
Svensson, B.H. 1980. Carbon dioxide and methane fluxes
from the ombrotrophic parts of a subarctic mire. Ecologi-
cal Bulletin(Stockholm) 30:235-250.
Tans, P.P., I.Y. Fung, and T. Takahash. 1990. Observa-
tional constraints of the global atmospheric CO2 budget.
Science 247:1431-1438.
Tytlyanova, AA. 1977. Biological cycle of carbon in
herbaceous formations. Nauka Press, Novosibirsk.
Tyuremnov, S.N. 1976. Torfyanye Mestorozhdeniya, 3rd
ed. Nedra Press, Moscow.
Vatkovskyi,O.S. 1976. The Analysis of Primary Production
of Forest Formations. Nauka Press, Moscow.
Vorobyov, G.I., ed. 1985. Forest encyclopedia. Sobetstaya
Encyclopedia Press, Moscow.
Zobler, L. 1986. A world soil file for global climate
modeling, NASA Technical Memo 87802.
274
-------
ACKNOWLEDGMENTS
The work presented herein was funded by the U.S. EPA Environmental Research Laboratory, Corvallis, OR, under co-
operative Agreement CR17682-01 to Oregon State University. Dr. Jeffrey Lee is the Proj ect Officer for the project entitled
"Carbon Cycling in Sub-Arctic and Boreal Forest Ecosystems of the Former Soviet Union." The work presented is a
component of the U.S. EPA Global Climate Research Program, Global Mitigation and Adaptation Program, R.K. Dixon,
Program Leader. This paper has not been subj ected to the Agency's review and, therefore, does not necessarily reflect the
views of the Agency, and no official endorsement should be inferred.
275
-------
-------
DIRECTORY OF WORKSHOP PARTICIPANTS
APPENDIX A: DIRECTORY OF WORKSHOP PARTICIPANTS
(SENIOR AUTHORS IN BOLD PRINT)
Apps, Mike
Northern Forestry Centre
5320 122nd St.
Edmonton, Alberta
CANADA T6H3S5
Phone (403) 435-7305
Fax (403) 435-7359
Arp, Paul
University of New Brunswick
Faculty of Forestry
Frederickton, New Brunswick
CANADA E3B 6C2
Phone (506) 453-4501
Fax (506) 453-3538
Ball, John (Tim)
Desert Research Institute
P.O. Box 60220
Reno, NV 89506
Phone (702) 673-7447
Fax (702) 673-7397
Barnett, Bruce
University of Alaska
Water Research Center
460 Duckering Bldg.
Fairbanks, AK 99775
Bell, Tom
Oregon State University
Dept. of Forest Science
Corvallis, OR 97331
Phone (503) 737-6571
Fax (503) 737-1393
Binkley, Dan
Colorado State University
Dept. of Forest Sciences
Ft. Collins, CO 80523
Phone (303) 491-6519
Fax (303) 491-6754
Blanutsa, Victor
Institute of Geography, Siberian Branch
Russian Academy of Science
Ulanbatorskaya St., 1, Irkutsk-33
664033 USSR
Bliss, Norman
EROS Data Center
Sioux Falls, SD 57198
Phone (605) 594-6034
Fax (605) 594-6589
Boersma, Larry
Oregon State University
Department of Soil Science
Corvallis, OR 97331
Phone (503) 737-5729
Fax (503) 737-5725
Bonan, Gordon
National Center for Atmospheric Research
P.O. Box 3000
Boulder, CO 80307
Phone (303) 497-1613
Fax (303) 497-1137
Borchers, Jeffrey G.
Oregon State University
Department of Forest Science
Peavy Hall 154
Corvallis, OR 97331
Phone (503) 750-7264
Fax (503) 737-1393
Botch, Marina S.
Komarov Botanical Institute
2 Popova St., St. Petersburg
197022 USSR
Phone 234-8426
Fax (812) 315-1701
Bradley, Peggy
Oregon State University
DepL of Civil Engineering
Apperson 107
Corvallis, OR 97331
Phone (503) 752-2139
Fax (503) 737-3052
Brubaker, Linda
University of Washington
College of Forest Resources
Seattle, WA 98195
Phone (206) 543-5778
Fax (206) 543-3254
277
-------
APPENDIX A
Cairns, Michael
U.S. Environmental Protection Agency
200 SW 35th St.
Corvallis, OR 97333
Phone (503) 754-4777
Fax (503) 754-4799
Chapin, Terry
University of California-Berkeley
Department of Integrative Biology
Berkeley, CA 94720
Phone (510) 642-1003
Fax (510) 643-6264
Cherkinsky, Alexander
Institute of Geography
Russian Academy of Science
Staromonetny St., 22, Moscow
109117 USSR
Cihlar, Josef
Center for Remote Sensing
1547McrivaleRd.
Ottawa, Ontario
CANADA K1AOY7
Phone (613) 952-0500
Fax (613) 952-7353
Cromack, Kermit, Jr.
Oregon State University
Dept. of Forest Science
Corvallis, OR 97331
Phone (503) 737-2244
Fax (503) 737-1393
Denison, Bill
Oregon State University
Dept. of Botany & Plant Pathology
Corvallis, OR 97331
Phone (503) 737-5304
Fax (503) 737-3573
Dixon, Robert K.
U.S. Environmental Protection Agency
200 SW 35th St.
Corvallis, OR 97333
Phone (503) 754-4772
Fax (503) 754-4799
Dubraskih, Mike
206 Cedar Lane
Corvallis, OR 97330
Phone (503) 745-7404
Entry, Jim
Oregon State University
Department of Forest Science
Peavy Hall 154
Corvallis, OR 97331
Phone (503) 737-6099
Fax (503) 737-2668
Frolking, Steve
Complex Systems Research Center
Institute for the Study of Earth, Oceans
and Space, Science and Engineering Bldg.
Durham, NH 03824-3525
Phone (603) 862-4098
Fax (603) 862-1915
Gaston, Greg
U.S. Environmental Protection Agency
200 SW 35th St.
Corvallis, OR 97333
Phone (503) 754-4496
Fax (503) 754-4799
Goetz, Scott
NASA/GSFC, Code 923
Greenbelt, MD 20771
Phone (301) 286-2477
Fax (301) 286-4098
Grulke, Nancy
U.S. Forest Service
3200 SW Jefferson Way
Corvallis, OR 97331
Gucinski, Hermann
USDA Forest Service Pacific NW
Research Station, Forestry Sciences Lab
3200 SW Jefferson Way
Corvallis, OR 97331
Phone (503) 750-7357
Fax (503) 750-7329
Hall, Forrest G.
NASA/GSFC
Greenbelt, MD 20771
Phone (301) 286-2974
Fax (301) 286-5269 RM N22A
Harmon, Mark
Oregon State University
Department of Forest Science
Corvallis, OR 97331
Phone (503) 750-7333
Fax (503) 737-1393
278
-------
DIRECTORY OF WORKSHOP PARTICIPANTS
Haubenstock, Norma
University of Alaska
Water Research Center
460 Duckering Bldg.
Fairbanks, AK 99775
Henderson, Sandra
Environmental Protection Agency
200 SW 35th St.
CorvaUis,OR 97333
Phone (503) 754-4724
Phone (503) 754-4799
Higuchi, Kaz
Research Scientist, ARQM
Atmospheric Research Directorate
Atmospheric Environment Service
4905 Dufferin St.
Downsview, Ontairio
CANADA M3H5T4
Phone (416) 739-4452
Fax (416) 739-5704
Horn, John
U.S.D A., Forest Service
Pacific Southwest Forest and Range
Experiment Station
Forest Fire Lab
4955 Canyon Crest Dr.
Riverside, CA 92507
Phone (714) 276-6545
Fax (714) 276-6426
Homann, Peter
Oregon State University
Dept. of Forest Science
Peavy Hall 154
Corvallis,OR 97331
Phone (503) 737-6584
Fax (503) 737-1393
Johnson, Dale W.
Desert Research Institute
P.O. Box 60220
Reno, NV 89506
Phone (702) 673-7370
Fax (702) 673-7397
Karol, Ronni
Research Forester
Intermountain Research Station
U.S. Forest Service, Box 8089
Missoula, MT 59807
Phone (406) 251-3473
Fax (406) 243-4510
Kern, Jeffrey
ManTech Environmental
U.S. Environmental Protection Agency
200 SW 35th St.
Corvallis, OR 97333
Phone (503) 754-4494
Fax (503) 754-4799
Kimble, John M.
USDA-SCS-NSSC
Federal Bldg., Rm. 152
100 Centennial Mall North.
Lincoln, ME 68508-3866
Phone (402) 437-5363
Fax (402) 437-5336
King, Tony
Oakridge National Laboratory
P.O. Box 2008, Bldg. 1000
Oak Ridge, TN 37831-6335
Phone (615) 576-3436
Fax (615) 574-2232
Kobak, Kira I.
State Hydrological Institute
V.O. Second Line, 23, St. Petersburg
199053 USSR
Phone 554-4146
Kolchugina, Tatyana
Oregon State University
Department of Civil Engineering
Apperson 107
Corvallis, OR 97331
Phone (503) 737-6156
Fax (503) 737-3052
Krankina, Olga
Oregon State University
Department of Forest Products
Corvallis, OR 97331
Phone (503) 750-7403
Fax (503) 737-1393
Kurz, Werner
ESSA
1765 West 8th Ave.
Vancouver, B.C.
CANADA V6J 5C6
Phone (604) 733-2996
Fax (604) 733-4657
279
-------
APPENDIX A
Lcc, Jeff
U.S. Environmental Protection Agency
200 SW 35th St.
Corvallis, OR 97333
Phone (503) 754-4578
Fax (503) 754-4799
Lorcnz, Susanne
Institute of World Forestry
Federal Research Centre for Forestry
and Forestry Products
Lcuschnerstr. 91, D-2050 Hamburg 80
GERMANY
Phone 011-49-407-396-2480
Fax 011-49-407-396-2403
Makundi, Willy
International Energy Studies Group
Lawrence Berkeley Laboratory
1 Cyclotron Rd.
Berkeley, CA 94720
Phone (510) 486-6852
Fax (510) 486-6996
Marrett, David
University of California
Dcpt. of Soil and Environmental Sciences
Riverside, CA 92521
Phone (714) 787-3659
Fax (714) 787-3993
Matthews, Elaine
NASA-GISS
2880 Broadway
New York, NY 10025
Phone (212) 678-5628
Fax (212) 678-5552
Mattson, Kim
Oregon State University
Dept. of Forest Science
154 Peavy Hall
Corvallis, OR 97331
Phone (503) 737-6097
Fax (503) 757-2668
Monlcon-Moscardo, V.
Oregon State University
Dept. of Forest Science
154 Peavy Hall
Corvallis, OR 97331
Phone (503) 737-2244
Nisbet, Robert
Center for the Study of the Environment
7126 Armstrong Road
Goleta, CA 93117
Phone (805) 863-3088
Fax (805) 569-1164
Panikov, Nicolay
Laboratory of Soil Microbiology
Institute of Microbiology
Russian Academy of Science
Pr. 60-Letiya Octyabrya, 7, k.2
Moscow 117811 USSR
Telex 411634INMAN S 4
Telex 213 121SVTSU
Perminova, Irena
Moscow State University
Department of Chemistry
Leninskii Gori, MSU, Moscow
119899 USSR
Telex 411483 MGUSU
Pervova, Natalya
Moscow State University
Department of Soil Science
Leninskii Gori, MSU, Moscow
119899 USSR
Telex 411483 MGUSU
Price, David T.
Forestry Canada, Northwest Region
5320-122 St.
Edmonton, Alberta
CANADA T6H 3S5
Phone (403) 435-7241
Fax (403) 435-7359
Rastetter, Edward B.
Marine Biological Laboratory
Ecosystems Center
Woods Hole, MA 02543
Phone (508) 548-3705
Fax (508) 548-1548
Reeburgh, William
University of Alaska
Institute of Marine Science
Fairbanks, AK 99775
Phone (907) 474-7830
Fax (907) 474-7204
280
-------
DIRECTORY OF WORKSHOP PARTICIPANTS
Rosenbaum, Barb
U.S. Environmental Protection Agency
200 SW 35th St.
Corvallis, OR 97333
Phone (503) 754-4456
Fax (503) 754-4799
Runyon, John
Oregon State University
Department of Forest Science
Corvallis, OR 97331
Phone (503) 737-2244
Fax (503) 737-1393
Shirazi, S.
U.S. Environmental Protection Agency
200 SW 35th St.
Corvallis, OR 97333
Phone (503) 754-4656
Fax (503) 754-4799
Semiletov, Igor
Laboratory of Underwater Cosmic Ray Detection
Pacific Oceanological Institute, Far-Eastern Branch
Russian Academy of Science
7-Radio St., Vladivostok
690032 USSR
Telex 213 121SVTSU
Simpson, Llyod G.
Assistance Reserach Forester
Environmental Studies and Biological Sciences
Santa Barbara, CA 93106
Phone (805) 893-2962
Fax (805) 893-4724
Sommerfeld, Richard
U.S. Forest Service
240 W. Prospect
Ft. Collins, CO 80526
Phone (303) 498-1233
Fax (303) 498-1364
Sundt, Nick
1347 Massachusetts Ave. SE
Washington, DC 20003-1540
Phone (202) 547-0850
Fax (202) 543-8393 c/o Kinko's
Turner, David
ManTec Environmental Technology, Inc.
c/o U.S. EPA Environmental Research Laboratory
Corvallis, OR 97333
Phone (503) 754-4769
Fax (503) 754-4338
Vinson, Ted S.
Oregon State University
Dept. of Civil Engineering
Apperson 107
Corvallis, OR 97331
Phone (503) 737-3494
Phone (503) 737-3052
Visser, Suzanne
Professional Associate
University of Calgary
Calgary, Alberta
CANADA T2NIN4
Fax (403) 2828-1287
Webb, Warren
Oregon State University
Dept. of Forest Science
Corvallis, OR 97331
Phone (503) 737-6558
Winnett, Steven M.
U.S. Environmental Protection Agency
Climate Change Division
PM-221,401MSt.SW
Washington, DC 20460
Phone (202) 260-6923
Fax (202) 260-6405, (202) 260-7884
Woodwell, George
Woods Hole Research Center
P.O. Box 296
Woods Hole, MA 02543
Phone (508) 540-9900
Fax (508) 540-9700
Wyant, Jamie
Project Leader, ManTech Environmental
U.S. Environmental Protection Agency
200 SW 35th St.
Corvallis, OR 97333
Phone (503) 754-4639
Fax (503) 754-4799
Yarie, John
University of Alaska
Forest Soils Lab, UAF
Fairbanks, AK 99775
Phone (907) 474-5650
Fax (907) 474-7439
281
-------
APPENDDCA
Yin,Xiwei
University of Brunswick
Faculty of Forestry
Frcdericton, New Brunswick
CANADA E3B6C2
Phone (506) 453-4501
Fax (506) 453-3538
Yoder, Barbara
Oregon State University
Dcpt. of Forest Science
Corvallis, OR 97331
Phone (503) 737-6110
Zclcncv, Vladimir
Laboratory of Soil Microbiology
Institute of Microbiology
Russian Academy of Science
Pr. 60-Letiya Octyabrya, 7, k.2
Moscow 117811 USSR
Telex 411634INMAN S 4
Telex 213121SVTSU
Zimov, Sergey A.
North-East Scientific Station of the Far-East Scientific
Center, Pacific Ocean Institute of Geography
Russian Academy of Science
P.O. Box 18
Chersky Yakutia
618830 USSR
Telex 213 121SVT SU
Zybach, Bob
Urban Forestry, Inc.
P.O. Box 12384
Portland, OR 97212
Phone (503) 222-1457
282
(1. S. GOVERNMENT PRINTING OFFICE: 1993 - 750-002 / 80230
-------
-------
m
ZR
Sg Q:
-T CD
S o>
i II Ml
o m g. CD
V 9 5T w
33 >
CD (Q
V) CO
CD 3
W .Q
Si
1
5S-2
§1
Is
m^: jo.
3-5^ 9»
M II
3| IS
sg ss.
to l|
1§ ?l
S"O r- 5
8
m
^ _m i
O
m
------- |