Tree Physiology 24, 765–773
© 2004 Heron Publishing—Victoria, Canada
Long-term effects of fire frequency on carbon storage and productivity
of boreal forests: a modeling study
J. H. M. THORNLEY1 and M. G. R. CANNELL1,2
1
Centre for Ecology and Hydrology, Bush Estate, Penicuik, Midlothian EH26 0QB, U.K.
2
Corresponding author ([email protected])
Received November 26, 2003; accepted January 26, 2004; published online May 3, 2004
Summary Climate change is predicted to shorten the fire interval in boreal forests. Many studies have recorded positive effects of fire on forest growth over a few decades, but few have
modeled the long-term effects of the loss of carbon and nitrogen to the atmosphere. We used a process-based, dynamic, forest ecosystem model, which couples the carbon, nitrogen and
water cycles, to simulate the effects of fire frequency on coniferous forests in the climate of Prince Albert, Saskatchewan.
The model was calibrated to simulate observed forest properties. The model predicted rapid short-term recovery of net primary productivity (NPP) after fire, but in the long term, supported the hypotheses that (1) current NPP and carbon content
of boreal forests are lower than they would be without periodic
fire, and (2) any increase in fire frequency in the future will tend
to lower NPP and carbon storage. Lower long-term NPP and
carbon storage were attributable to (1) loss of carbon on combustion, equal to about 20% of NPP over a 100–200 year fire
cycle, (2) loss of nitrogen by volatilization in fire, equal to
about 3–4 kg N ha –1 year –1 over a 100–200 year fire cycle, and
(3) the fact that the normal fire cycle is much shorter than the
time taken for the forest (especially the soil) to reach an equilibrium carbon and nitrogen content. It was estimated that a
shift in fire frequency from 200 to 100 years over 1000 Mha of
boreal forest would release an average of about 0.1 Gt C year –1
over many centuries.
Keywords: global warming, nitrogen, soil organic matter, tree.
Introduction
Boreal forests play an important role in the global climate system for many reasons. They represent about one third of the
global forest area and contain about 60% of the global forest
soil carbon (470 Gt C) and 25% of the forest tree carbon
(88 Gt C) (Dixon et al. 1994). Most boreal forests experience
summer droughts and periodic fires, which release NOx and
other gases to the atmosphere, as well as perpetuate their
floristic composition and patchwork of relatively even-aged
stands (Crutzen et al. 1979, Payette 1992, Kimmins 1996). Finally, their growth, structure and net carbon balance with the
atmosphere are highly sensitive to changes in temperature,
precipitation, N input and rates of N mineralization (Chapin et
al. 2000).
We describe the use of a forest ecosystem model to investigate the specific question of whether, over century to millennial timescales, a change in fire frequency is likely to significantly alter net primary productivity (NPP) and the amount of
carbon stored in coniferous boreal forests. To make the results
transparent, we did not include changes in climate or species
composition in these simulations. The question investigated
relates to two hypotheses: (1) current NPP and carbon content
of boreal forests are lower than they would be without periodic
fire; and (2) any increase in fire frequency in future will tend to
lower NPP and carbon storage.
The motivation for the study was twofold. First, the Intergovernmental Panel on Climate Change concluded that global
change is likely to increase the frequency of boreal forest fires
(Watson et al. 1998). The area reported being burned each year
has increased in recent decades—possibly as a result of improved record keeping (Harden et al. 2000). Charcoal deposits
show that, historically, fire frequencies have been positively
correlated with temperature and summer drought occurrence,
which are both likely to increase in many areas (Filion 1984,
Clark 1988, Whitlock et al. 2003). In Canada, it has been predicted that climate change in response to a doubling of pre-industrial atmospheric CO2 concentration will lengthen the fire
season in the boreal region by 30 days (Wotton and Flannigan
1993) and increase the annual area burned by over 40% (Flannigan and van Wagner 1991, Wotton et al. 2003).
The second motivation for the study was that few modeling
studies have considered long-term effects of increased fire frequency on the productivity and carbon balance of boreal forests. Most dynamic global vegetation models include disturbance functions, but simulations have generally focused on the
balance between NPP and heterotrophic respiration over decadal timescales, exploring the roles of changes in water relations and N mineralization rates (Wang and Polglase 1995,
King et al. 1997, Xiao et al. 1998, Keyser et al. 2000, Cramer et
al. 2001). An exception is a study by Kasischke et al. (1995)
who estimated carbon stocks using a model of stand age distribution as a function of fire frequency, but without simulating
the nitrogen or water cycles. They concluded that increased
fire frequency would substantially lower carbon storage in boreal forests, cancelling out any increase associated with en-
766
THORNLEY AND CANNELL
hanced productivity or change in species composition. Similarly, Harden et al. (2000) used a carbon balance model to
show that an increase in fire frequency would make upland boreal forests a carbon source.
Superficially, there is an apparent paradox, in that, in the
short term, fires have positive effects on forest ecosystems by
increasing soil mineral N content, often increasing soil microbial biomass, removing competing shrubs, opening serotinous
cones, increasing soil temperature, and in some cases, promoting the growth of N2-fixing tree species (Kimmins 1996,
DeLuca and Zouhar 2000, Grogan et al 2000). However, in the
long term, fire can have a detrimental effect by depleting carbon and nitrogen stocks. Carbon stocks are depleted by combustion of the trees, slow recovery of biomass after fire and, in
intense fires, combustion of litter. Nitrogen stocks are depleted
by volatilization and particle emission to the atmosphere and
increased leaching and erosion loss.
Clearly, if N2-fixation rates are high following fire, the fire
interval can be short without depleting the site N capital. In
Eucalyptus forests with high symbiotic N2-fixation rates, by
understory vegetation, this interval may be only 9–12 years
(McMurtrie and Dewar 1997). However, in boreal and cool
temperate forests, with low non-symbiotic N2-fixation rates
(1–4 kg N ha –1 year –1; MacLean et al. 1983, Harden et al.
2003), the N capital of the site may be depleted below its
steady-state level with a normal fire cycle of 50–200 years.
Any shortening of the fire interval following climate change
may further deplete the N capital of the site and reduce productivity. Reich et al. (2001) found that NPP and annual N mineralization rate of ecosystems decreased with increase in fire
frequency across a forest–grassland continuum in Minnesota.
Any decrease in NPP as a result of N loss is likely to further deplete the carbon stock, already lowered by more frequent combustion loss.
Materials and methods
The Edinburgh Forest Model
The Edinburgh Forest Model is a process-based evergreen forest-soil ecosystem simulator that couples carbon, nitrogen and
water and runs with a 20-min time-step. It is programmed in
ACSL (advanced continuous simulation language, Aegis Research, Huntsville, AL) and is available from www.nbu.ac.uk.
The source program is FOREST.CSL, which is extensively annotated. The model is generic, assumes horizontal homogeneity and is composed of linked submodels, which are described
elsewhere, for trees (Thornley 1991), soil and litter to a nominal 1 m depth (Thornley 1998a, Chapter 5) and water (Thornley 1996, 1998a, Chapter 6). Flow diagrams of the submodel
structures have been described previously (Thornley and Cannell 2000a) and a synopsis of the processes represented is
given by Thornley and Cannell (1996). For this application,
the model was run in natural forest mode (as opposed to
clear-felled plantation mode) in which (1) stem mortality rate
depends on water stress, leaf area index (LAI) and tree nutrient
status (the product of carbon and nitrogen substrate concentra-
tions) and (2) stem regeneration rate depends on stem mortality (creating gaps), tree nutrient status, irradiance at ground
level and leaf area per stem. The model was calibrated to simulate a coniferous forest in the boreal region of Canada, most
closely resembling the pine/lichen forest type (Harden et al.
1997) with periodic losses of C and N as a result of fire (see below).
Three developments have been made to the model since it
was first described (Thornley 1991, Thornley and Cannell
1996). First, photosynthesis now acclimates to light, nitrogen,
carbon dioxide and temperature (Thornley 1998b). Second,
some of the components of maintenance respiration are separated (Thornley and Cannell 2000b). Third, photosynthesis is
calculated separately for the sun and shade components of the
canopy (Thornley 2002).
One weakness of the model, for this application, was the absence of a moss layer and representation of soil temperature
depth profiles and changes in soil hydrology associated with
permafrost (Bonan and Korzuhin 1989). The likely effect is
overestimation of forest productivity and soil respiration many
decades after fire, when moss and litter insulation allows permafrost to develop and impedes drainage. Quantitatively, this
weakness will affect our results, but we believe the trends relating fire interval to nitrogen loss and long-term carbon
storage to be robust (cf. O’Neill et al. 2003).
Environment
The model was run in a constant annual climate simulating
conditions at Prince Albert, Saskatchewan, 53°13′ N, 431 m
a.s.l. Diurnal data were calculated from daily data assuming sinusoids; daily data were obtained by linear interpolation from
30-year monthly means (Meteorological Office 1982; see
Thornley 1998a for methodology). Wind speed (50 m height)
was assumed constant at 4 m s –1. Other quantities varied
throughout the year: photosynthetically active radiation varied
from a maximum of 11 MJ m –2 day –1 on July 17 to 1 MJ m –2
day –1 on December 16; daily maximum and minimum air temperatures varied from annual maxima of 25.1 and 11 °C on
July 16 to minima of –13.7 and –25.1 °C 6 months later; soil
temperatures were diurnally constant and varied from 18 °C
on July 16 to –19.4 °C 6 months later; precipitation varied
from 0.54 mm day –1 on February 14 to 2.19 mm day –1 on June
16 (totalling 399 mm year –1); daily maximum (dawn) and
minimum (1500h) relative humidities varied from annual
maxima of 0.93 and 0.83 on December 16 to minima of 0.83
and 0.73 on May 16. The forest was assumed to receive 7 kg N
ha –1 year –1 from the atmosphere.
Fire simulation
Two fire regimes were simulated: (1) a high-intensity crown
fire (5000–10,000 kW per m length of fire front) in which all
aboveground biomass and surface litter were burned, 75% of
the N was volatilized to the atmosphere and 25% returned to
the soil ammonium pool; and (2) a low-intensity, mainly surface fire (200–500 kW per m length of fire front) in which
50% of the aboveground biomass and 50% of the surface litter
were burned and 50% of the N was volatilized to the atmo-
TREE PHYSIOLOGY VOLUME 24, 2004
FIRE EFFECTS ON CARBON STORAGE AND PRODUCTIVITY OF BOREAL FORESTS
sphere and 50% returned to the soil ammonium pool.
These two regimes were chosen recognizing that both
crown and surface fires are common in the boreal region (MacLean et al. 1983), that 10–100% of the ground layer can be
burned (Dyrness and Norun 1983, Viereck 1983, Stocks 1989)
and that a large fraction of N is volatilized in intense fires
(Crutzen et al. 1979, Raison et al. 1985). Both regimes probably overestimated the amount of biomass normally combusted, because rarely do all stems burn (Harden et al. 2000);
however, the litter layer simulated by the model probably underestimated the size of the combustible forest floor (see below). Overall, the total fraction of NPP burned during a fire
cycle was similar to that simulated by other models (Harden et
al. 2000, 2003).
In the model, there is the option of moving a fraction of the
burned biomass and litter at each fire event into an inert charcoal pool, the rest being returned to the atmosphere. However,
this precludes the attainment of an equilibrium state, which is
helpful when interpreting simulation results, and therefore this
option was not used.
Following fire, soil temperatures are higher for about
20 years—owing to increased insolation, removal of the insulating litter and lower albedo—deepening the active layer
above permafrost, increasing N mineralization rates and promoting N2 fixation (MacLean et al. 1983, Kimmins 1996,
O’Neill et al. 2003). These processes were simulated crudely
by assuming that soil temperature was raised by 4 °C and that
the mineralization constant was doubled immediately after intense fire—both changes decreasing linearly to normal over
20 years. This modification had a quantitative rather than qualitative impact on long-term ecosystem properties.
The model did not simulate changes in forest species composition following fire (e.g., an increase in deciduous species
following frequent fires) nor ground vegetation, which can, in
some areas, increase N2-fixation rates (McMurtrie and Dewar
1997, Newland and DeLucia 2000).
767
Table 1. Parameters in the Edinburgh Forest Model were calibrated to
give the output values shown below, simulating a generic coniferous
forest at equilibrium in the climate of Prince Albert, Saskatchewan
with a 100-year fire interval. All quantities are annual means for Year
100 (fire does not occur during the year). Abbreviation: NPP = net primary productivity. The model employs a notional soil depth of 1 m.
Parameter
Output value
Carbon budget (kg C m –2 year –1)
Gross canopy photosynthesis
Net photosynthesis (NPP)
Litter flux (above- and belowground)
Soil respiration
0.60
0.30
0.21
0.24
Allocation of NPP to foliage/branches/
stems/coarse roots/fine roots (%)
8/11/31/23/27
Nitrogen budget (kg N ha –1 year –1)
Deposition
N2 fixation
Gaseous emission
Leaching
7
0.7
0.7
0
Water budget (mm year –1)
Annual precipitation
Intercepted and evaporated rain
Plant evapotranspiration
Drainage
399
182
218
0
Forest properties
Leaf area index
Specific leaf area (m 2 kg –1)
Foliage residence time (years)
2.9
5.9
6
Soil properties
Surface litter (kg m –2)
Surface litter lignin residence time (years)
Soil organic matter carbon (kg m –2)
Net N mineralization (kg N ha –1 year –1)
1
26
4.5
0.6
Carbon and nitrogen contents
Tree biomass carbon; nitrogen (kg m –2)
Soil (SOM and litter) carbon; nitrogen (kg m –2)
5.6; 0.07
5.5; 0.44
Model evaluation and calibration
Throughout the development of the Edinburgh Forest Model,
the assumptions and ways in which processes were represented were progressively modified so that the dynamics of the
system were stable and the model was capable of simulating
known trends with forest age in the mass of tree parts, LAI and
main fluxes in C and N budgets.
For this application, the parameters listed by Thornley and
Cannell (1996) were adjusted so that the values of the principal
output variables were within measured ranges for coniferous
forests in the mid-boreal region of Canada. Table 1 presents
some of the output variables at age 100. Thus, simulated net
primary productivity (NPP) of 0.3 kg C m –2 year –1 was in the
range of about 0.4 kg C m –2 year –1 estimated for southern boreal to 0.2 kg C m–2 year –1 for northern boreal forests (Rodin et
al. 1975, Doucet et al. 1976, Cannell 1982, Gower et al 2000,
Chen et al. 2002). High allocation to fine and coarse roots
(about 50% of NPP) in this N-poor environment (simulated
using a dynamic substrate-gradient method) was in agreement
with expectation (Cannell and Dewar 1994, Steele et al. 1997).
The model simulated only low rates of non-symbiotic N2 fixation (0.7 kg N ha –1 year –1) in line with the fixation rates of
1–4 kg N ha –1 year –1 that occur in boreal forests, mainly by
cyanobacteria within blue-green algae associated with lichen
and moss (Cleveland et al. 1999, DeLuca et al. 2002). The simulated LAI of 2.9, specific leaf area of 5.9 m 2 kg –1 and the foliage longevity of 6 years were within measured ranges
(Tyrrell and Boerner 1987, Jarvis et al. 1997). The simulated
surface litter mass of about 1 kg C m –2 was less than the mean
forest floor biomass of 2–3 kg C m –2 (4–5 kg biomass m –2)
measured in boreal and cold temperate forests (Vogt et al.
1986). However, the model simulated 4.5 kg C m –2 of ‘soil’ organic matter, some of which was in an unprotected (fast turnover) organic matter pool, which could be defined as being
within the forest floor. The total tree carbon content in a forest
subject to fire every 100–200 years was simulated as about
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
768
THORNLEY AND CANNELL
5 kg C m –2 (Figure 4C), which corresponds to observed values
(Bonan and Korzuhin 1989).
Model runs
First, the model was run until the forest ecosystem reached a
quasi-equilibrium state (with annual C and N inputs equal to
outputs and stable annual mean pools). A high- or low-intensity fire was then imposed in order to observe effects on ecosystem properties and the time course of recovery. Second, the
degrading effects of repeated fires were simulated by running
the model to a quasi-equilibrium state, then imposing a highintensity fire every 100 years and following changes in annual
mean forest ecosystem properties (averaged over 100 years)
towards a new equilibrium state. Lastly, the model was run to
near equilibrium with high-intensity fire every 2, 5, 10, 20, 50,
100, 200 and 500 years to determine average ecosystem properties over a range of fire cycles (with properties averaged over
2, 5, 10, …, or 500 years).
The latter scenarios included some unrealistic, theoretical
fire cycles in the climate of Prince Albert, Saskatchewan. The
fire return period can be as short as 15–20 years in pine forests
in northern Eurasia (Sannikov and Goldammer 1996), or
50 years in the dry interior of Alaska (Viereck 1983), but, in
most boreal regions, including Saskatchewan, the fire interval
ranges from about 60–70 years in pine-dominated forests to
100–150 years in spruce-dominated forests (Wein and MacLean 1983, Payette 1992). Also, the scenarios were unrealistic
in assuming that high-intensity fires can occur frequently,
when fuel loads remain low (van Wagner 1983).
Results
Boreal forest ecosystem responses to fire
Figure 1 shows the simulated effects of high- and low-intensity fires (at Year 50) on a range of forest properties. Figure 2
shows the simulated effect of soil warming following a highintensity fire (at Year 20).
Although high-intensity fire completely destroyed the trees,
simulated LAI and NPP recovered within about 20 years (Figures 1A and 1B). Rapid recovery was promoted by (1) an increase in leaf stomatal conductance in summer during the
recovery years when LAI was low (a large effect in this dry climate (Figure 1G), and (2) an increase in soil mineral N content
(Figure 1H) as a result of N input from the fire and accelerated
mineralization of soil organic matter in response to soil warming after fire (Figure 1I). Simulated soil warming after fire
advanced the recovery of LAI and NPP by about 5 years (Figures 2A and 2B). Net primary production was enhanced for the
first 10–20 years after fire (Figures 1A and 2A).
In contrast, it took over 300 years after high-intensity fire
for the trees to reach their original carbon content, reflecting
slow build-up of standing biomass in this challenging climate
(Figure 1C shows recovery over 150 years). The trees restored
their original nitrogen content within about 50 years (Figure 1E), a few decades after canopy closure (maximum LAI),
by which time nitrogen was effectively recycled within the
Figure 1. Simulated effect of fire (at Year 50) on boreal forest properties. The model was run to near equilibrium without fire before fire
was applied. Continuous lines = high-intensity fire (with soil warming); broken lines = low-intensity fire (without soil warming). Warming was mimicked by assuming that, immediately after fire, soil
temperatures were raised 4 °C and specific mineralization rates of
protected and stabilized soil organic matter pools doubled. These deviations then returned linearly to normal values over 20 years. Abbreviation: SOM = soil organic matter.
trees and ecosystem.
The soil carbon content fell by about 20% after high-intensity fire (Figure 1D) due to combustion of surface litter, reduced litter input for about 20 years (while NPP was low) and
soil warming (Figure 2D). Soil nitrogen content fell by about
15% after high-intensity fire (Figure 1F). Soil carbon and nitrogen contents did not recover for over 1500 years (Figure 2D
shows recovery over 80 years). Recovery was limited by the
addition of only 7 kg N ha –1 year –1 from atmospheric deposition and only 0.7 kg N ha –1 year –1 from N2 fixation, and gas-
TREE PHYSIOLOGY VOLUME 24, 2004
FIRE EFFECTS ON CARBON STORAGE AND PRODUCTIVITY OF BOREAL FORESTS
769
Figure 2. Simulated effects of soil
warming and permafrost melting during the first 80 years after a high-intensity fire (at Year 20). Continuous lines
= with warming; broken lines = without warming. The model was run to
near equilibrium without any fire or
soil warming before the fire. The
warming effect was simulated as in
Figure 1.
eous and leaching losses, which were high for about 10 years
immediately following fire (Figure 1J).
Low-intensity fire, which had the effect of thinning the forest, temporarily increased NPP because LAI was not lowered
enough to lessen summer water stress (with increased leaf
stomatal conductance), or to drastically lower light interception. However, tree carbon content took about 200 years to recover (Figure 1C) and soil carbon and nitrogen contents took
over 800 years to recover (Figures 1D and 1F).
Degradation of ecosystem in response to repeated fires
If soil carbon and nitrogen contents take 1500 or 800 years to
recover after high- or low-intensity fire, respectively, soil carbon and nitrogen will be progressively depleted when the fire
interval is much shorter. This soil degradation may be expected to lower NPP.
The continuous lines in Figure 3 represent a simulated forest, which began at equilibrium without fire, and was consumed by high-intensity fire every 100 years over four fire
cycles. The broken lines represent a forest that reached equilibrium with fire every 100 years (i.e., was in a steady state
with repeating behavior every 100 years). The continuous
lines progressively approached the broken lines, as the initially
fire-free forest was degraded by fire every 100 years.
As expected, total soil carbon and nitrogen were progressively depleted after each 100-year fire (Figures 3D and 3F).
Recovery after fire was small compared with the loss of carbon
and nitrogen during the fire. Eventually (after more than ten
100-year fire rotations) the soil reached an equilibrium with
only about 25% of the carbon and nitrogen that it had without
fire (from 20 to 5 kg C m –2 and 1.8 to 0.5 kg N m –2).
Both NPP and LAI increased slightly after the first fire cycle, due to release of nitrogen, but then slowly decreased after
each successive fire cycle toward the lower 100-year-fire equilibrium state (Figures 3A and 3B). Tree carbon was com-
pletely removed by fire, recovered to barely half its original
value after the first fire (i.e., the maximum tree biomass was
halved) and reached progressively smaller maximum values at
the end of successive fire cycles (Figure 3C). Tree nitrogen
content recovered almost completely after the first fire and
then slowly decreased after each subsequent fire cycle (Figure 3E).
Table 2 shows the simulated loss of carbon and nitrogen
from the forest with different fire intervals, at equilibrium, expressed as averages per year over the fire cycle. Fire every
100 years volatilized about 360 kg N ha –1, or 3.6 kg N ha –1
year –1 over the fire cycle—equal to about half of the annual input from atmospheric deposition and N2 fixation (Table 1).
Losses at shorter fire intervals were less, simply because the
forest was depleted in nitrogen, whereas losses at longer fire
intervals were less per year because the loss was averaged over
longer periods. It should be noted, however, that a loss of
360 kg N ha –1 represented only about 7% of the 5000 kg N
ha –1 in the soil in a forest subject to fire every 100 years (Figure 3F).
Table 2 also shows the loss of carbon from trees on combustion, equal to total tree carbon content, and loss from litter, representing 10–15% of the total soil carbon at each fire (e.g.,
6300 kg C ha –1 loss every 100-year fire, with total soil carbon
of 50,000 kg C ha –1). The loss of carbon on combustion of
about 500 kg C ha –1 year –1 represented about 17% of the NPP
(3000 kg C ha –1 year –1, Table 1).
Ecosystem properties as a function of fire frequency
Figure 4 shows the effects of different high-intensity fire intervals on the properties of forests when they reach equilibrium.
The infinity fire interval refers to forests with no fire. All values are averaged over the fire cycle.
Equilibrium amounts of carbon and nitrogen in both the soil
and trees decreased as the fire interval shortened (Figures 4C
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
770
THORNLEY AND CANNELL
Figure 3. The gradual degradation of a
forest subject to high-intensity fire every 100 years. Continuous lines = forest starting in a steady state without
fire and then subjected to fire every
100 years. Broken lines = forest in
steady state, with fire every 100 years.
The two lines gradually converge, as
the forest depicted with continuous
lines is degraded by loss of carbon and
nitrogen toward the steady state described by the broken lines. The highintensity fire is accompanied by soil
warming (Figure 1).
and 4D). Forests burned every 500 years had substantially less
carbon and nitrogen than those not burned at all, and forests
burned every 100 years had about half as much carbon and nitrogen as those burned every 500 years.
A surprising result was that gross and net primary production did not fall steeply as the fire interval shortened from infinity to about 200 years. Over such long fire intervals the
average LAI was little affected (Figure 4B) and the mean size
of the soil mineral pool remained high (Figure 4E). The fall in
soil carbon between no fires and 200-year fires was due mostly
Table 2. Loss of carbon and nitrogen in simulated high-intensity fires,
expressed as averages over fire cycles of 5 to 500 years. In each case,
the model was run to equilibrium.
Fire interval
5
10
20
50
100
200
500
C loss by combustion
(kg C ha –1 year –1)
N loss by volatilization
(kg N ha –1 year –1)
Trees
Litter
Trees
Litter
116
214
245
343
459
550
294
7
25
38
47
63
83
47
1.42
2.09
2.74
3.10
3.22
3.14
1.50
0.03
0.10
0.22
0.31
0.42
0.54
0.30
to the loss of litter carbon on combustion (Table 2) rather than
to a decrease in NPP and carbon input to soil. Similarly, the fall
in tree carbon between no fires and 200-year fires was due
mostly to the loss of tree carbon on combustion (Table 2).
The decreases in soil and tree carbon and nitrogen as the fire
interval shortened below 200 years were more clearly attributable to falls in NPP and LAI as well as to losses of carbon and
nitrogen on combustion. The decreases in NPP and LAI were
associated with nitrogen depletion, as a result of volatilization
loss on combustion (Table 2), reflected in low annual net nitrogen mineralization and low soil mineral nitrogen contents
(Figure 4E).
Figure 4 also reveals some non-monotonic trends. In Figure 4B, leaf stomatal conductance (on summer afternoons) decreased with increases in LAI above 1.5 because of increased
water stress, but decreased when fires occurred less than every
20 years because of decreased concentration of osmolytes (in
the model) in young trees. The amount of water draining
through the soil (Figure 4F) decreased with increasing LAI, as
expected, but the amount of N leached per year also depended
on the size of the soil mineral nitrate pool (Figure 4E) at the
time of year when leaching was greatest; this pool was small
both with frequent fires, because the system was nitrogen depleted, and when there were no fires, because of uptake and
immobilization.
TREE PHYSIOLOGY VOLUME 24, 2004
FIRE EFFECTS ON CARBON STORAGE AND PRODUCTIVITY OF BOREAL FORESTS
771
Figure 4. Average forest properties as
affected by frequency of high-intensity
fire (no soil warming). Each property is
the average over the fire cycle for a
forest in a steady state (i.e., the forest
was subjected to repeated fire cycles
until it approached equilibrium). Leaf
stomatal conductance is that at 1500 h
on July 1.
Discussion
This study supports the two hypotheses, namely: (1) current
NPP and carbon content of boreal forests are lower than they
would be without periodic fire, and (2) any increase in fire frequency in the future will tend to lower NPP and carbon
storage.
There are three reasons for the decreases in NPP and carbon
storage with increased fire frequency. First, forests lose substantial quantities of carbon during a high-intensity fire. Over
the common fire cycle of 100–200 years, this study suggested
that fire returned over 20% of the net carbon fixed by photosynthesis over that period to the atmosphere. Harden et al.
(2000) estimated a range of 10–30%. O’Neill et al. (2003,
their Table 6) simulated the same qualitative and quantitative
pattern of decreased carbon storage in soils with decreases in
fire intervals to that shown in Figure 4C. Second, fire volatilizes substantial quantities of nitrogen. Over a fire cycle of
100–200 years, this study suggested that fire releases 3–4 kg
N ha –1 year –1 averaged over the fire cycle. Harden et al. (2003)
estimated that, since the last glaciation, boreal forests have lost
2.5–7.0 kg N ha –1 year –1 as a result of combustion. This exceeds biological N2 fixation in most boreal forests and is about
half of the current atmospheric nitrogen deposition in most areas. Consequently, fires substantially reduce the nitrogen capital, which lowers NPP and carbon storage. Third, the time
constant for a boreal forest to reach equilibrium with its environment after fire is well over 500 years, much longer than the
normal fire interval. O’Neill et al. (2003) estimated that an
Alaskan spruce ecosystem takes 650–800 years to reach an
approximate equilibrium. Most boreal forest growth is severely limited by temperature, nitrogen and summer drought
(Kimmins 1996). Consequently, boreal forests eventually
reach a fire-induced equilibrium, with less carbon and nitrogen, and lower NPP, than they would have without fire.
It should, however, be stressed that the decrease in NPP and
carbon storage with increased fire frequency is not as great as
might be expected. The model suggested little reduction in
NPP with a decrease in fire interval from 500 to 200 years, and
only a 20% reduction in NPP between the 200- and 50-year
fires (Figure 4A). The loss of carbon (from trees and soil)
when the fire interval was halved, from 200 to100 years (an extreme scenario: Wotton et al. 2003) was about 10 kg C m –2 at
equilibrium (Figure 4C). But moving from a 200- to a 100year equilibrium may take over 1000 years (Figure 3). Thus,
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
772
THORNLEY AND CANNELL
the average loss per year was less than 0.01 kg C m –2 year –1. If
this loss occurred over 1000 Mha of boreal forest (in North
America and Asia; Dixon et al. 1994), the total annual loss
would be only 0.1 Gt C year –1. This net loss may be compared
with the estimate of global emissions associated with land use
change in recent decades of about 1.6 Gt C year –1 (Watson et
al. 2000) and a current high-latitude sink of about 0.7 Gt C
year –1 (Apps et al. 1993).
Although fire reduced the nitrogen capital of the site, the
model suggested that it did not greatly reduce NPP unless fires
occurred more often than about once every 200 years, when
soil mineral nitrogen pools became much smaller (at equilibrium) and much less nitrogen was mineralized each year (Figure 4E). Thus, shortening the fire cycle within the commonly
observed range of 100–200 years exacerbates the already severe nitrogen limitation to NPP and growth of most boreal forests (Kimmins, 1996).
The responses of boreal forests to fire over decadal timescales are a poor guide to their long-term responses. The
model simulated rapid recovery of forest NPP after fire, often
to higher values than before for some decades. This response
was the result of increased nitrogen availability, improved water relations and simulated soil warming. However, in the long
term, the loss of carbon and nitrogen to the atmosphere as a result of fire was eventually reflected in lower NPP and carbon
storage.
In the real world, increases in fire frequency in the boreal
zone are associated with climate change, which has been predicted to accelerate soil organic matter decomposition (lowering carbon storage; Jenkinson et al. 1991), increase nitrogen
mineralization and forest NPP (increasing carbon storage;
Xiao et al. 1998, White et al. 2000), alter species composition
and extend the northern limit of the boreal forest (increasing
carbon storage; Neilson et al. 1998). The net effect on carbon
storage is, at present, difficult to predict. This study has demonstrated, using a process-based model that couples the carbon, nitrogen and water cycles, that increased fire frequency, is
likely to have a small but significant negative effect on both
NPP and carbon storage.
References
Apps, M.J., W.A. Kurz, R.J. Luxmoore, L.O. Nilsson, R.A. Sedjo,
R. Schmidt, L.G. Simpson and T.S. Vinson. 1993. Boreal forests
and tundra. Water Air Soil Pollut. 70:39–53.
Bonan, G.B. and M.D. Korzuhin. 1989. Simulation of moss and tree
dynamics in the boreal forests of interior Alaska. Vegetatio 84:
31–44.
Cannell, M.G.R. 1982. World forest biomass and primary production
data. Academic Press, London, 391 p.
Cannell, M.G.R. and R.C. Dewar. 1994. Carbon allocation in trees: a
review of concepts for modelling. Adv. Ecol. Res. 25:59–104.
Chapin, III, F.S., A.D. McGuire, J. Randerson et al. 2000. Arctic and
boreal ecosystems of western North America as components of the
climate system. Global Change Biol. 6:211–223.
Chen, W., J.M. Chen, D.T. Price and J. Chihar. 2002. Effects of stand
age on net primary productivity of boreal black spruce forests in
Ontario, Canada. Can. J. For. Res. 32:833–842.
Clark, J.S. 1988. Effect of climate change on fire regimes in northwestern Minnesota. Nature 344:233–235.
Cleveland, C.C., A.R. Townsend, D.S. Schimel et al. 1999. Global
patterns of terrestrial biological nitrogen (N2 ) fixation in natural
ecosystems. Global Biogeochem. Cycles 13:623–645.
Cramer, W., A. Bondeau, F.I.Woodward et al. 2001. Global response
of terrestrial ecosystem structure and function to CO2 and climate
change: results from six dynamic global vegetation models. Global
Change Biol. 7:357–373.
Crutzen, P.J., L.E. Heidt, J.P. Krasnec, W.H. Pollock and W. Seiler.
1979. Biomass burning as a source of atmospheric gases CO, H2,
N2O, NO, CH3Cl and COS. Nature 282:253–256.
DeLuca, T.H. and K.L. Zouhar. 2000. Effects of selection harvest and
prescribed fire on the soil nitrogen status of ponderosa pine forests.
For. Ecol. Manage. 138:63–271.
DeLuca, T.H., O. Zackrisson, M.-C. Nilsson and A. Sellstedt. 2002.
Quantifying nitrogen-fixation in feather moss carpets of boreal forests. Nature 419:917–920.
Dixon, R.K., S. Brown, R.A. Houghton, A.M. Solomon, M.C. Trexler
and J. Wisniewski. 1994. Carbon pools and flux of global forest
ecosystems. Science 263:185–190.
Doucet, R., J.V. Berglund and C.E. Farnsworth. 1976. Dry matter production in 40-year-old Pinus banksiana stands in Québec. Can. J.
For. Res. 6:357–367.
Dyrness, C.T. and R.A. Norum. 1983. The effects of experimental
fires on black spruce forest floors in interior Alaska. Can. J. For.
Res. 13:879–893.
Filion, L. 1984. A relationship between dunes, fire and climate as recorded in the Holocene deposits of Québec. Nature 309:543–546.
Flannigan, M.D. and C.E. van Wagner. 1991. Climate change and
wildfire in Canada. Can. J. For. Res. 21:61–72.
Gower, S.T., A. Hunter, J. Campbell, J. Vogel, H. Veldhuis, J. Harden,
S. Trumbore, J.M. Norman and C.J. Kucharik. 2000. Nutrient dynamics of southern and northern BOREAS boreal forests. Ecoscience 7:481–490.
Grogan, P., T.D. Bruns and F.S. Chapin. 2000. Fire effects on ecosystem nitrogen cycling in a Californian bishop pine forest. Oecologia
122:537–544.
Harden, J.W., K.P. O’Neill, S.E. Trumbore, H. Veldhuis and
B.J. Stocks. 1997. Moss and soil contributions to the annual net
carbon flux of a maturing boreal forest. J. Geophys. Res. 102
(D24): 28,805–28,816.
Harden, J.W., S.E. Trumbore, B.J. Stocks, A. Hirsch, S.T. Gower,
K.P. O’Neill and E.S. Kasischke. 2000. The role of fire in the boreal carbon budget. Global Change Biol. (Supp.) 6:174–184.
Harden, J. W., M. Mack, H. Veldhuis and S.T. Gower. 2003. Fire dynamics and implications for nitrogen cycling in boreal forests. J.
Geophys. Res. 108 (D3):WFX 4-1 to 4-8.
Jarvis, P.G., J.M. Massheder, S.E. Hale, J.B. Moncrieff, M. Rayment
and S.L. Scott. 1997. Seasonal variation of carbon dioxide, water
vapor and energy exchanges of a boreal black spruce forest. J.
Geophys. Res. Atmos. 102:28,953–28,966.
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.
Kasischke, E.S., N.L. Christensen and B.J. Stocks. 1995. Fire, global
warming and the carbon balance of boreal forests. Ecol. Appl. 5:
437–451.
Keyser, A.R., J.S. Kimball, R.R. Nemani and S.W. Running. 2000.
Simulating the effects of climate change on the carbon balance of
North American high-latitude forests. Global Change Biol. 6:
185–195.
TREE PHYSIOLOGY VOLUME 24, 2004
FIRE EFFECTS ON CARBON STORAGE AND PRODUCTIVITY OF BOREAL FORESTS
Kimmins, J.P. 1996. Importance of soil and role of ecosystem disturbance for sustained productivity of cool temperate and boreal forests. Soil Sci. Soc. Am. J. 60:1643–1654.
King, A.W., W.M. Post and S.D. Wullschleger. 1997. The potential
response of terrestrial carbon storage to changes in climate and atmospheric CO2. Climatic Change 35:199–227.
Maclean, D.A., S.J. Woodley, M.G. Weber and R.W. Wein. 1983. Fire
and nutrient cycling. In The Role of Fire in Northern Circumpolar
Ecosystems (Scope 18). Eds. R.W. Wein and D.A. MacLean. John
Wiley & Sons, Chichester, U.K., pp 111–132.
McMurtrie, R.E. and R.C. Dewar. 1997. Sustainable forestry: a model
of the effects of nitrogen removals in wood harvesting and fire on
the nitrogen balance of regrowth eucalypt stands. Aust. J. Ecol.
22:243–255.
Meteorological Office. 1982. Tables of temperature, relative humidity, precipitation and sunshine for the world. Part I, North America.
HMSO, London, 70 p.
Neilson, R.P., I.C. Prentice, T. Kittel and D. Viner. 1998. Simulated
changes in vegetation distribution under global warming. In The
Regional Impacts of Climate Change: An Assessment of Vulnerability. Eds. R.T. Watson, M.C. Zinyowera, R.H. Moss and D.J.
Dokken. A Special Report of IPCC WG II for the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, pp 440–456.
Newland, J.A. and T.H. DeLuca. 2000. Influence of fire on native nitrogen-fixing plants and soil nitrogen status in ponderosa
pine—Douglas fir forests in western Montana. Can. J. For. Res. 30:
274–282.
O’Neill, K.P., E.S. Kasischke and D.D. Richter. 2003. Seasonal and
decadal patterns of soil carbon uptake and emission along an age
sequence of burned black spruce stands in interior Alaska. J.
Geophys. Res. 108 (D1): FFR 11.1–11.15.
Payette, S. 1992. Fire as a controlling process in the North American
boreal forest. In A Systems Analysis of the Global Boreal Forest.
Eds. H.H. Shugart, R. Leemans and G.N. Bonan. Cambridge University Press, Cambridge, pp 144–169.
Raison, R.J., P.K. Khanna and P.V. Woods. 1985. Transfer of elements
to the atmosphere during low-intensity prescribed fires in three
Australian sub-alpine eucalyptus forests. Can. J. For. Res. 15:
657–664.
Reich, P.B., D.W. Petersen, D.A. Wedin and K. Wrage. 2001. Fire and
vegetation effects on productivity and nitrogen cycling across a forest–grassland continuum. Ecology 82:1703–1719.
Rodin, LE., N.I. Brazilevich and N.N. Rozov. 1975. Productivity of
the world’s main ecosystems. In Productivity of World Ecosystems. National Academy of Science, Washington, DC, pp 13–26.
Sannikov, S.N. and J.G. Goldammer. 1996. Fire ecology of pine forests of northern Eurasia. In Fire in Ecosystems of Boreal Eurasia.
Eds. J.G. Goldammer and V.V. Furayev. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp 151–167.
Steele, S.J., S.T. Gower, J.G. Vogel and J.M. Norman. 1997. Root
mass, net primary production and turnover in aspen, jack pine and
black spruce forests in Saskatchewan and Manitoba, Canada. Tree
Physiol. 17:577–587.
Stocks B.J. 1989. Fire behaviour in mature jack pine. Can. J. For. Res.
17:80–86.
Thornley, J.H.M. 1991. A transport-resistance model of forest growth
and partitioning. Ann. Bot. 68:211–226.
Thornley, J.H.M. 1996. Modelling water in crops and plant ecosystems. Ann. Bot. 77:261–275.
Thornley, J.H.M. 1998a. Grassland dynamics: an ecosystem simulation model. CAB International, Wallingford, U.K., pp xii and 241.
773
Thornley, J.H.M. 1998b. Dynamic model of leaf photosynthesis with
acclimation to light and nitrogen. Ann. Bot. 81:421–430.
Thornley, J.H.M. 2002. Instantaneous canopy photosynthesis: analytical expressions for sun and shade leaves based on exponential light
decay down the canopy and an acclimated non-rectangular hyperbola for leaf photosynthesis. Ann. Bot. 89:451–458.
Thornley, J.H.M. and M.G.R. Cannell. 1996. Temperate forest responses to carbon dioxide, temperature and nitrogen: a model analysis. Plant Cell Environ. 19:1331–1348.
Thornley, J.H.M. and M.G.R. Cannell. 2000a. Managing forests for
wood yield and carbon storage: a theoretical study. Tree Physiol.
20:477–484.
Thornley, J.H.M. and M.G.R. Cannell. 2000b. Modelling the components of plant respiration. II: representation and realism. Ann. Bot.
85:55–67.
Tyrrell, L.E. and R.E.J. Boerner. 1987. Larix laricina and Picea mariana—relationships among leaf life-span, foliar nutrient patterns,
nutrient conservation and growth efficiency. Can. J. Bot. 65:
1570–1577.
van Wagner, C.E. 1983. Fire behaviour in northern conifer forests and
shrublands. In The Role of Fire in Northern Circumpolar Ecosystems (Scope 18). Eds. R.W. Wein and D.A. MacLean. John Wiley
& Sons, Chichester, U.K., pp 65–80.
Viereck, L.A. 1983. The effects of fire in black spruce ecosystems of
Alaska and northern Canada. In The Role of Fire in Northern Circumpolar Ecosystems (Scope 18). Eds. R.W. Wein and D.A. MacLean. John Wiley & Sons, Chichester, U.K., pp 201–220.
Vogt, K.A., C.C. Grier and D.J. Vogt. 1986. Production, turnover and
nutrient dynamics of above- and belowground detritus of world
forests. Adv. Ecol. Res. 15:303–377.
Wang, Y.P. and P.J. Polglase. 1995. Carbon balance in the tundra, boreal forest and humid tropical forest during climate change: scaling
up from leaf physiology and soil carbon dynamics. Plant Cell Environ. 18:1226–1244.
Watson, R.T., M.C. Zinyowera, R.H. Moss and D.J. Dokken. 1998.
The regional impacts of climate change. An assessment of vulnerability. A Special Report of IPCC WG II for the Intergovernmental
Panel on Climate Change. Cambridge University Press, Cambridge, 517 p.
Watson, R.T., I.R. Noble, B. Bolin, N.H. Ravindranath, D.J. Verardo
and D.J. Dokken. 2000. Land use, land-use change and forestry.
Special Report of the IPCC, Cambridge University Press, Cambridge, 377 p.
Wein, R.W. and D.A. MacLean. 1983. An overview of fire in northern
ecosystems. In The Role of Fire in Northern Circumpolar Ecosystems (Scope 18). Eds. R.W. Wein and D.A. MacLean. John Wiley
& Sons, Chichester, U.K., pp 1–18.
White, A., M.G.R. Cannell and A.D. Friend. 2000. The high-latitude
terrestrial carbon sink: a model analysis. Global Change Biol. 6:
227–245.
Whitlock, C., S.L. Shafer and J. Marlon. 2003. The role of climate and
vegetation change in shaping past and future fire regimes in the
northwestern US and the implications for ecosystem management.
For. Ecol. Manage. 178:5–21.
Wotton, B.M. and M.D. Flannigan. 1993. Length of fire season in a
changing climate. For. Chron. 69:187–192.
Wotton, B.M., D.L. Martell and K.A. Logan. 2003. Climate change
and people-caused forest fire occurrence in Ontario. Climatic
Change 60:275–295.
Xiao, X., J.M. Melillo, D.W. Kicklighter et al. 1998. Transient climate
change and net ecosystem production of the terrestrial biosphere.
Global Biogeochem. Cycles 12:345–360.
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com