Color profile: Generic CMYK printer profile
Composite Default screen
1285
Process and patterns of duff consumption in the
mixedwood boreal forest
K. Miyanishi and E.A. Johnson
Abstract: Postfire tree recruitment in the boreal forest is restricted to patches from which the duff (organic layer) has
been removed by fire. Duff consumption occurs by smoldering combustion, propagation of which is determined by
bulk density, moisture content, and depth. This study investigated interactions among these factors, their spatial distribution, and spatial patterns of duff consumption in two wildfires. A hypothesized positive relationship between moisture content and depth was supported by a laboratory study. Duff characteristic data were collected from two burns and
comparable unburned areas of mixedwood forest in western Canada to describe and explain patterns of duff consumption within and between top-slope Pinus banksiana Lamb. – Picea mariana (Mill.) BSP stands and lower slope pure
stands of Picea mariana on glaciofluvial hillslopes. In unburned stands, bulk density did not differ significantly between stands, while depths were significantly greater in Picea stands. With moist duff, consumption was significantly
greater in Picea than in Pinus–Picea stands as depth limited smoldering in the thin duff of Pinus–Picea stands. With
dry duff, smoldering was propagated regardless of depth resulting in no differences between stand types. The spatial
correlation between burned patches and fire-killed trees was explained by within-stand spatial variation in duff moisture
due to precipitation interception by tree crowns.
Résumé : La régénération après feu en forêt boréale est limitée aux parcelles où l’humus (horizon organique) a été détruit par le feu. L’humus est consumé par la combustion lente dont la propagation est déterminée par la densité apparente, la teneur en humidité et la profondeur. Cette étude a examiné les interactions entre ces facteurs, leur distribution
spatiale et les patrons spatiaux de destruction de l’humus pour deux feux de forêt. L’hypothèse d’une relation positive
entre la teneur en humidité et la profondeur a été confirmée par une étude en laboratoire. Les données caractéristiques
de l’humus ont été récoltées dans deux brûlis et des superficies non brûlées et comparables de forêt mélangée de
l’ouest du Canada pour décrire et expliquer les patrons de destruction de l’humus à l’intérieur et entre des peuplements
de Pinus banksiana Lamb. – Picea mariana (Mill.) BSP de haut de pente et des peuplements purs de Picea mariana
de bas de pente sur des versants fluvio-glaciaires. Dans les peuplements non brûlés, la densité apparente ne varie pas
significativement entre les peuplements alors que la profondeur est significativement plus importante dans les peuplements de Picea. Dans le cas d’un humus humide, la destruction est significativement plus prononcée dans les peuplements de Picea que dans ceux de Pinus–Picea car la profondeur limite la combustion lente dans l’humus mince des
peuplements de Pinus–Picea. Avec un humus sec, la combustion lente se propage quelle que soit la profondeur,
n’entraînant pas de différence entre les types de peuplement. La corrélation spatiale entre les parcelles brûlées et les arbres tués par le feu s’explique par la variation spatiale dans la teneur en humidité de l’humus à l’intérieur des peuplements due à l’interception des précipitations par la cime des arbres.
[Traduit par la Rédaction]
Miyanishi and Johnson
Introduction
North American boreal forest trees recruit primarily in the
first 6 years after wildfires (DesRochers and Gagnon 1997;
Gutsell and Johnson 2002). The initial post-fire cohort forms
the canopy of the mature forest and is not subsequently replaced by later establishing trees because of the length of the
fire cycle, which is generally shorter than the life-span of the
Received 23 March 2001. Accepted 7 March 2002. Published
on the NRC Research Press Web site at http://cjfr.nrc.ca on
12 July 2002.
K. Miyanishi.1 Department of Geography, University of
Guelph, Guelph, ON N1G 2W1, Canada.
E.A. Johnson. Department of Biological Sciences and
Kananaskis Field Stations, University of Calgary, Calgary,
AB T2N 1N4, Canada.
1
Corresponding author (e-mail: kmiyanis@uoguelph.ca).
Can. J. For. Res. 32: 1285–1295 (2002)
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:13 AM
1295
tree species (Johnson 1992). Therefore, recruitment within
the first few years after a fire is critical in determining the
composition, density, and spatial distribution of trees in mature boreal forest stands. Recruitment of Pinus banksiana
Lamb., Picea mariana (Mill.) BSP, and Picea glauca
(Moench) Voss is inversely related to the amount of duff remaining after the fire (Jameson 1961; Chrosciewicz 1970,
1974, 1976; Zasada et al. 1983; Thomas and Wein 1985;
Weber et al. 1987). Duff is defined as the layer of organic
matter on the forest floor (the F and H layers) between the
litter (L) and the mineral soil (Van Wagner 1972; Canada
Soil Survey Committee 1978). In the boreal forest, duff is
derived from litter (leaves and twigs from the trees, shrubs,
and herbaceous plants) as well as from mosses.
Since successful seedling establishment is virtually restricted to areas from which duff has been removed by fire,
there has long been an interest in duff consumption by forest
fires (e.g., Chrosciewicz 1959; Beaufait 1962; Adams 1966;
DOI: 10.1139/X02-051
© 2002 NRC Canada
Color profile: Generic CMYK printer profile
Composite Default screen
1286
Cayford 1966). Two approaches to modeling duff consumption have been used: (i) empirical regression models which
correlate average duff consumption with various combinations of independent variables including indices of duff
moisture (e.g., Little et al. 1986; Reinhardt et al. 1989;
Brown et al. 1991), and (ii) a process-based model of radiative heat transfer from the flaming front to the duff (Van
Wagner 1972). These approaches predict only the mean
amount of duff removed, despite observations that duff is
consumed by wildfires in distinct deeply burned patches scattered among unburned or scarcely burned duff (Chrosciewicz
1976; Dyrness and Norum 1983; Zasada et al. 1983) (Fig. 1).
Consequently, we propose a third approach here that uses an
understanding of the combustion process by which duff is
consumed to identify the controlling variables and to explain
patterns of duff consumption within stands, among stands
within fires, and between different fires.
Smoldering combustion has been recognized as the major
process by which duff is consumed (Dyrness and Norum
1983; Frandsen 1991; Hungerford et al. 1995; Latham and
Williams 2001). Duff is not consumed to any extent by flaming combustion because of its relatively high content of
lignin, which does not readily release volatiles upon heating,
and its high packing ratio, which prevents rapid enough heat
release to sustain flames (see Miyanishi 2001). Smoldering
combustion differs from flaming combustion in being a
much slower, nonflaming oxidation of a porous charforming solid (Drysdale 1985; Ohlemiller 1985). Smoldering
is generally modeled as a two-step process: an endothermic
process of pyrolysis (thermal degradation) of the solid fuel,
releasing volatiles, and forming char and an exothermic process of char oxidation. Propagation of smoldering depends
on sufficient heat being transferred from the exothermic oxidation zone to the virgin fuel to cause pyrolysis and char
formation. Thus, factors that explain extinction (nonpropagation) of smoldering are those that influence either
the rate of heat generation by oxidation or the rate of heat
transfer from the oxidation zone to the virgin fuel.
Studies of smoldering in various materials (e.g., Moussa
et al. 1976; Ohlemiller 1990) have led to the conclusion that
the rate-limiting factor for smoldering of dry fuels is oxygen
availability (i.e., the rate of diffusion of oxygen to the char
oxidation zone), which determines the rate of oxidation and
hence the rate of heat generation. However, Peter (1992)
found that for smoldering in beds of peat or sawdust with
varying moisture contents, fuel moisture, rather than oxygen
availability, was the limiting factor, since latent heat of vaporization constitutes a significant heat sink and, thus, decreases the heat available for pyrolysis and char formation.
Also, since heat transfer within the fuel bed between the
exothermic oxidation zone and the endothermic pyrolysis
zone is primarily by conduction (Moussa et al. 1976), thermal diffusivity of the fuel, which is a function of fuel density, plays an important role in propagation of smoldering.
Finally, heat losses can occur between the zone of oxidation,
where heat is being generated, and the zone of pyrolysis,
which requires sufficient heat for maintenance of the
pyrolytic process. In particular, with very thin fuel beds,
convective heat losses from the surface of the fuel bed can
be significant enough to decrease effective heat transfer to
the pyrolysis zone. Assuming horizontal propagation of
Can. J. For. Res. Vol. 32, 2002
smoldering, the area of the oxidation zone is larger in
thicker fuel beds, and therefore, proportional convective heat
loss from the fuel bed surface is less (i.e., more of the heat is
trapped within the fuel bed and not lost convectively to the
air above). Thus, Palmer (1957) and Peter (1992) found critical minimum depths of dry fuel beds that could propagate
smoldering.
As explained above, the three variables we would expect
to be controlling smoldering propagation in duff are moisture content, density, and depth. Also, given the roles of fuel
moisture (as a heat sink in evaporation) and depth (which influences the proportion of heat generated by char oxidation
that is lost through convection) in smoldering propagation,
we should expect a positive relationship between fuel moisture content and the minimum fuel depth required for smoldering propagation, i.e., the higher the fuel moisture level,
the greater the minimum depth of fuel required.
Potentially, all three of these variables (moisture content,
density, and depth) might be expected to vary because of
both hillslope position and canopy coverage. Soil moisture
varies along hillslopes (O’Loughlin 1981; Quinn et al. 1992;
Bridge and Johnson 2000) and soil and duff moisture have
been shown to be connected by capillary movement of water
between the duff and mineral soil (Samran et al. 1995). In
the mixedwood boreal forest of western Canada, Bridge and
Johnson (2000) found a moisture–nutrient gradient on
glaciofluvial substrate that ran from drier, less nutrient rich
stands at the top slopes dominated by Pinus banksiana and
Picea mariana to the wetter, more nutrient-rich, almost-pure
stands of Picea mariana on middle to bottom slopes. Thus,
the top slope Pinus–Picea stands would be expected to have
drier duff than the middle to bottom slope Picea stands. Duff
moisture would also be expected to exhibit within stand variability because of interception of precipitation and radiation
by tree crowns. Precipitation interception would reduce input of moisture to the duff directly beneath tree crowns,
while interception of outgoing terrestrial radiation by tree
crowns at night would result in less radiative cooling at the
ground surface and, hence, less dew formation. Thus, we
would expect duff to be generally drier directly beneath tree
crowns than between tree crowns, resulting in a distinct patterning of duff moisture within stands.
Bulk density of duff may also vary along hillslopes because of differences in litter composition, which, in turn,
would be influenced by hillslope changes in species composition. Duff density is not only a factor in heat transfer (in
the smoldering process) but also influences the rate at which
duff dries in the absence of precipitation. Because of differences in drying rates, the Canadian Forest Fire Weather Index System (Van Wagner 1987) distinguishes two categories
of soil organic matter besides litter: (i) a 7-cm layer of
loosely compacted decomposing organic matter with a bulk
density of 0.07 g·cm–3 and (ii) a deep 18-cm layer of compact organic matter with a bulk density of 0.36 g·cm–3. Examples of duff representing the first type include Pinus
banksiana and Pinus resinosa Ait. stands in Ontario (Van
Wagner 1972), while examples of the latter type include
Abies balsamea (L.) Mill. and upland Picea mariana stands
in northern Quebec (Van Wagner 1974).
Moisture content of the first type of fuel is estimated using the duff moisture code (DMC) while moisture content of
© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:14 AM
Color profile: Generic CMYK printer profile
Composite Default screen
Miyanishi and Johnson
1287
Fig. 1. Photograph of duff after the Bittern fire showing the distinct patches of burned duff within a matrix of largely unburned duff.
the latter type is estimated with the drought code (DC). Both
indexes are numerical ratings of mean moisture content, intended to give an indication of the consumption by fire of
these two fuel types (Canadian Forestry Service 1984). In
addition, the DC is an indication of seasonal drought effects
on forest fuels. These indices use noon weather measurements and empirical drying curves and are formulated to increase with increasing dryness (Van Wagner 1987). Because
of differences in drying rates of these two fuel types and the
rainfall required to saturate them, the response rates of the
DMC and DC to weather conditions differ. Furthermore,
both react more slowly to changing weather conditions than
the fine litter fuels whose moisture content largely determines the fire intensity.
Duff depth may also vary both between and within stands.
Duff depth is a product of the relative rates of litter input
and decomposition. Litter input rates would be a function of
stand density and composition. Decomposition rate is both
temperature and moisture dependent, as well as being influenced by litter quality characteristics such as pH, base content, and presence of polyphenols or tannins (Williams and
Gray 1974; Mason 1976). Duff depths in Pinus stands are
generally less than that found in Picea stands in the eastern
Canadian boreal forest (Van Wagner 1972, 1974). Duff
depths may also vary within stands, since needle litter from
the trees would accumulate directly beneath tree crowns and
to a lesser extent between tree crowns.
Therefore, if the three controlling variables of duff moisture, bulk density, and depth vary both along hillslopes and
within stands, we should also expect duff consumption to
vary at both scales. Furthermore, since duff moisture content
is largely influenced by weather (the two duff moisture indexes, DMC and DC, are determined by daily weather data),
it would also be expected to differ temporally and, hence,
vary between different fires in similar stand types. Since few
past studies have addressed the phenomenon of duff consumption from a process approach, one objective of this
study was to apply what is known about the process of smoldering combustion to predict some general patterns of duff
consumption (e.g., the relationship between moisture content
and depth). We conducted a laboratory experiment to test
the hypothesis that critical values of moisture content and
depth required for smoldering propagation are positively related. A second objective was to investigate how duff moisture, bulk density, and depth vary between stand types along
hillslopes on well-drained glaciofluvial substrate in the
mixedwood boreal forest of western Canada. Finally, since
few, if any, past studies of duff consumption have investigated spatial patterning of duff consumption, a third objective was to describe patterns of duff consumption in two
areas burned by natural wildfires and to use the results from
the first two objectives to explain these patterns.
Materials and methods
Study area
The two burns used in this study are located within the
mixedwood boreal forest zone (Rowe 1972) in central
Saskatchewan. The upland forests are dominated by Pinus
banksiana, Picea mariana, Picea glauca, Populus
tremuloides Michx., and Abies balsamea (L.) Mill. (Dix and
Swan 1971; Bridge and Johnson 2000), while the lowland
forests are dominated by Picea mariana and Larix laricina
(Du Roi) K. Koch (Jeglum 1972, 1973).
Most fires in this region occur between April and October
with >95% of the total area burned occurring in April, May,
and June (Johnson et al. 1999; Weir et al. 2000). The Bittern
Creek fire (53°58′ N, 105°55′W) was ignited by lightning on
June 5, 1996, and burned approximately 900 ha of forest
within the Weyerhaeuser Forest Management Licence Area
(FMLA). The Waskesiu Lake fire (54°04′N, 106°02′W) was
ignited by lightning on July 7, 1998, and burned 1738 ha of
forest within Prince Albert National Park and in the
Weyerhaeuser FMLA. In both fires, most of the area burned
© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:15 AM
Color profile: Generic CMYK printer profile
Composite Default screen
1288
Can. J. For. Res. Vol. 32, 2002
Fig. 2. Plots of (a) duff moisture code and (b) drought code for
the period May 3 to August 8, 1996 and 1998. The 1996 Bittern
Creek fire burned June 5–7 (Julian dates 157–159), while the
1998 Waskesiu Lake fire burned primarily July 7–10 (Julian
dates 188–191).
Duff moisture code
100
1998
1996
80
60
40
20
0
700
600
Drought code
a)
120
140
160
180
Julian date
200
220
160
180
Julian date
200
220
b)
500
1998
1996
400
300
200
100
120
140
during the first 3 or 4 days. Both were crown fires (fire-line
intensities of at least 6000 kW·m–1) in which the litter layer
was consumed and all trees were killed. However, duff
moisture conditions differed considerably between the two
fires. Figure 2 plots the DMC and the DC during the fire
season (May 3 to August 8) in 1996 and 1998. The values
for the DMC were 20–28 on June 5–7, 1996 (Julian dates
157–159), during the Bittern fire and 45–59 on July 7–10,
1998 (Julian dates 188–191), during the Waskesiu fire. The
DC values were 252–268 during the Bittern fire and 510–
537 during the Waskesiu fire. Thus, the duff was obviously
much drier during the 1998 Waskesiu fire than during the
1996 Bittern fire.
Data from each of the two burns were collected within a
year after the fires from eight upland stands dominated by
Pinus banksiana – Picea mariana and eight upland stands
dominated by Picea mariana, henceforth referred to as
Pinus–Picea and Picea stands, respectively. Only stands
with no evidence of permanent standing water (i.e., gleyed
or mottled mineral soils) and homogeneous in stem density
and size were selected. All were mature stands when the
fires occurred with ages ranging between 90 and 120 years
(stand age was estimated by obtaining cores as close to the
base as possible of three canopy trees in each stand). We
also collected data from eight comparable unburned Pinus–
Picea and eight unburned Picea stands located adjacent to
the two burns (Table 1).
Laboratory study
To test the hypothesized relationship between duff moisture and the depth of duff at which smoldering combustion
will continue or cease, we used commercial sphagnum peat
in lieu of duff in a laboratory experiment. Previous studies
had found that peat is similar to duff in particle size distribution and bulk density (Frandsen 1987; Hartford 1989;
Hawkes 1993) as well as in chemical composition
(Bracewell and Robertson 1987). The availability of relatively uniform commercial peat made replication of samples
in the tests easier.
The burn box (22 × 22 × 14 cm) was constructed of
1.3 cm thick ceramic board (Cotronics Corp., Brooklyn,
N.Y.). Thermal conductivity of the dry ceramic (0.04 W·m–1·K–1
at 10°C) is similar to that reported by Hillel (1982) for dry
peat (0.06 W·m–1·K–1 at 10°C). The top and one end of the
box were left open. The peat and ceramic board were equilibrated to the moisture content desired for 2 or 3 days before
each test. For each burning trial, the box was filled to a uniform depth, which was varied from 0.2 to 5 cm; the surface
area of the fuelbed was constant at 308 cm2 (22 × 14 cm).
We attempted to maintain as much as possible a relatively
constant dry bulk density of the peat through packing; mean
density was 0.097 g·cm–3. Volumetric moisture content of
the peat was varied from 9 to 20%. Moisture content of each
sample was measured with a ThetaProbe type ML2x and
Thetameter type HH1 (Delta-T Devices Ltd., Cambridge,
U.K.). Using a single two-point gravimetric calibration for
peat, the ThetaProbe had an accuracy of 1%.
The vertical surface of the peat at the open end of the box
was heated with a Bunsen burner until 1 cm of peat was ignited. Then, the peat was left to burn on its own. If the
whole volume of peat was consumed in a trial, self-sustained
smoldering propagation was determined to be successful at
that depth and moisture content. If the peat ceased smoldering once the external heat source was removed, selfsustained smoldering propagation was considered to have
failed at that depth and moisture content. A total of 56 trials
with varying depth–moisture combinations were conducted.
Field study
As explained in the Introduction, two factors that influence propagation of smoldering combustion are bulk density
and depth of the fuel. To determine the importance of differences in these factors in explaining duff consumption patterns at the hillslope scale, we collected duff samples from
the unburned stands and compared depth and bulk density of
duff between Pinus–Picea and Picea stands. Within each of
the two stand types, eight 10 × 10 m plots were established.
At the 81 intersection points of grid lines established at 1-m
intervals within each plot, the litter layer was removed and a
5.4 cm diameter core of the duff was collected down to the
mineral soil. Where the sampling point fell on the location
of a standing or fallen tree or surface tree root, no duff sample could be collected; thus, the number of samples per plot
© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:16 AM
Color profile: Generic CMYK printer profile
Composite Default screen
Miyanishi and Johnson
1289
Table 1. General characteristics of the Pinus banksiana – Picea mariana and Picea mariana dominated stands in the Bittern and
Waskesiu burns and in the comparable unburned stands.
Pinus–Picea
Range of stand ages (years)
Mean tree density (no./100 m2)
Mean tree basal diameter (cm)
Species composition (% of stand basal area)
Pinus banksiana
Picea mariana
Other conifersa
Deciduousb
Unidentifiable
Picea
Bittern
Waskesiu
Unburned
Bittern
Waskesiu
Unburned
90–120
33
15.8
107–111
40
15.7
100–107
40
16.0
90–120
67
11.3
104–111
81
10.4
85–100
81
11.3
67.2
21.1
2.5
2.5
6.7
50.4
49.5
0
0.1
0
62.6
35.0
2.0
0.3
0
1.8
88.7
0.9
2.9
5.7
3.1
96.6
0
0.3
0
2.8
93.5
1.9
1.8
0
Note: All values are means of the eight stands in each group. Stem densities and basal diameters areas are for standing stems only (dead stems in the
two burns and live stems in the unburned stands).
a
Picea glauca and Abies balsamea.
b
Populus tremuloides, Populus balsamifera, and Betula papyrifera.
varied between 70 and 79 for a total of 515 samples from
the Pinus–Picea plots and 606 samples from the Picea plots.
Depth of the duff was used to calculate the sample volume.
The samples were dried at 95°C for 12 h and weighed to obtain dry masses from which the bulk density of each sample
was determined. The bulk density and depth distributions
were plotted separately for the Pinus–Picea and Picea data,
and their means were compared with a t test.
To investigate within-stand variation in duff moisture, we
measured duff moisture in two unburned Picea stands using
the ThetaProbe and Thetameter as in the laboratory study. In
each stand, a 10 × 10 m plot was established and gridded at
0.5-m intervals. At each of the 361 intersection points, the
position was recorded as beneath the canopy or in a gap, and
a moisture reading was taken. Because of the size of the
probe (25 mm diameter × 60 mm length), the duff was at
times too thin to allow insertion of the whole probe vertically; in such cases, the probe had to be inserted either diagonally or horizontally into the duff to obtain a reliable
reading. The ThetaProbe was calibrated for duff using a single two-point gravimetric calibration and had an accuracy of
1%. The rapid readings allowed by this instrument meant
that all of the 361 readings within a stand could be taken
within 4–5 h, making the within-stand readings comparable
(particularly given the response time of duff moisture to
weather). The mean volumetric moisture contents beneath
tree crowns and in gaps were compared using a t test for
each of the two Picea stands separately. Unfortunately, duff
depths in the Pinus–Picea stands were, in many places, too
thin to allow complete insertion of the probe at any angle.
Since data from these stands would be limited (and biased)
to those areas with deep enough duff to obtain a reliable
moisture reading, our objective to record duff moisture variation at a fine scale within individual stands could not be
met; therefore, we did not use Pinus–Picea stands in this
part of the study.
Precipitation interception by tree crowns was measured
with a funnel rain gauge (collecting area of 166.5 cm2)
placed 5 cm from the base of two Pinus banksiana trees
(with basal diameters of 28.2 and 34.0 cm) and three Picea
mariana trees (with basal diameters of 12.4, 17.6, and
23.7 cm). Prevailing winds are from the west, and therefore,
the collectors were placed on the east side of each tree. The
trees were located at different sites and were selected such
that their tree crowns were not touching those of adjacent
trees. A rain gauge was also placed in a large opening at
each site to measure precipitation. Throughfall (and its inverse, interception) were calculated as a percentage of precipitation recorded in the gauge placed in the opening. Data
were collected for 2–5 rain events per tree with a total of 10
rain events per species. The number of rain events differed
for the five trees in the sample, because they were located at
different sites and much of the summer precipitation in this
area is from local convective storms. A t test was conducted
on the arcsine-transformed data to compare interception by
the two tree species.
Duff depths were measured in 10 × 10 m plots in the Bittern and Waskesiu burns in the same manner as described
previously for unburned stands, with eight plots sampled in
Pinus–Picea and eight in Picea stands within each burn. The
distributions of duff depths by site (Bittern, Waskesiu, unburned) and by stand type (Pinus–Picea and Picea) were
plotted, and the normality (skewness) of each distribution
was tested using the g1 statistic and computing the Z g 1 value
(Zar 1996). While the unburned stands are expected to have
a normal distribution of duff depths, patchiness of duff consumption in the burned stands should result in a positively
skewed distribution because of the large number of near-zero
duff depths.
To investigate the pattern of duff consumption in the two
burns, the same 10 × 10 m plots were used as above. Each
plot was subdivided into one hundred 1 × 1 m subplots with
grid lines at 10-cm intervals. Within each subplot, all burned
holes in the duff (hereafter called patches) were mapped,
and their areas were visually estimated using the grid as a
guide. Also, all standing and fallen stems were mapped,
identified to species where possible, and their basal diameters (cm) measured. Both burn sites were examined soon after the fire. Fallen or standing stems could be identified as
having been live or dead at the time of the fire by the extent
and type of charring. A distribution of patch sizes was produced for each burn with data from the Pinus–Picea and
Picea stands separated. The percentage of the area in
patches was compared between the two fires and also be© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:17 AM
Color profile: Generic CMYK printer profile
Composite Default screen
1290
Can. J. For. Res. Vol. 32, 2002
Fig. 3. (a) Relationship between volumetric moisture content (%
of volume) and depth (cm) of fuel at which smoldering combustion fails to propagate in peat and (b) regression line determined
from the fuel–depth combinations that resulted in failure of
smoldering propagation plus the data points for the depth–
moisture combinations that resulted in successful smoldering
propagation.
% moisture content (v/v)
22
a)
20
18
16
14
12
10
0
% moisture content (v/v)
22
1
2
3
Fuel depth (cm)
4
5
b)
20
18
16
14
12
10
y=12.36 + 1.83x
8
0
1
2
3
Fuel depth (cm)
4
5
tween stand types within fires. Chi-square analyses were
used to determine the statistical significance of the spatial
relationship between the burned patches and bases of standing stems by comparing the observed number of stems
within or touching patches with the number expected, given
both the density of stems and the proportion of total stand
area covered by patches (Norcliffe 1982).
Results
Relationship between moisture content and critical fuel
depth
Of the 56 burning trials with varying combinations of fuel
depth and moisture content, 22 resulted in extinguishment or
nonpropagation of smoldering while 34 resulted in selfsustained smoldering propagation. Figure 3a shows a plot of
the 22 depth–moisture combinations that resulted in extinguishment. Depth and moisture content were highly significantly correlated (r = 0.83, p < 0.0001). To use these data to
determine the upper limits of depth–moisture combinations
that allow smoldering propagation, a regression line was
obtained (y = 12.358 + 1.83x, F = 44.57, p < 0.0001) and
plotted (Fig. 3b). We would expect generally that depth–
moisture combinations above the line would result in extinguishment, while those below the line would allow smoldering propagation. As shown in Fig. 3b, 30 of the 34 trials in
which smoldering propagation occurred fell either extremely
close to the line or below and only 4 were above the line.
Patterns of duff bulk density, depth, and moisture in
the field
The results of the laboratory study indicate that if duff
bulk density is similar in field situations, duff moisture and
depth should be the important variables in determining smoldering propagation and, hence, duff consumption. We obtained very similar distributions of duff bulk density for
unburned Pinus–Picea and Picea stands (Fig. 4) and the
means of 0.092 and 0.094 g·cm–3, respectively, were not significantly different (t = 0.66, p > 0.05). However, duff depth
in unburned stands did differ significantly between the two
stand types (Fig. 5) with duff depths of 6.9 ± 0.12 cm (mean ±
SE) for Pinus–Picea stands and 10.3 ± 0.11 cm for Picea
stands (t = 20.31, p < 0.001). Thus, duff depth varies along
hillslopes with top slope Pinus–Picea stands having thinner
duff than middle to bottom slope Picea stands.
As explained in the Materials and methods, we were unable to determine duff moisture differences along hillslopes,
since the moisture probe did not allow measurements in the
thin duff occurring in top slope Pinus–Picea stands. However, an unpublished study by S.T.K. Vo and E.A. Johnson
found that in a hillslope water budget study, duff in top slope
Pinus–Picea stands is drier than duff in Picea stands. We did
look at duff moisture variation within two Picea-dominated
stands in relation to the location of tree crowns, i.e., each of
the 361 sample points in each stand was recorded as in a gap
with no tree crown directly overhead or by the species of
tree whose crown was directly above the sample point. The
first stand was pure Picea mariana, while the second stand
also had some A. balsamea, Populus balsamifera, and
Populus tremuloides. The overall mean volumetric moisture
contents for the duff differed significantly between the two
stands (12.8 vs. 10.8%; t = 7.54, p < 0.001). However, since
the two stands were measured on different days, we were not
interested in comparing among stands but were collecting
these data to investigate within stand variability in duff
moisture. Within each stand, moisture contents of duff beneath tree crowns were significantly lower than in gaps. In
one stand, the moisture contents were 12.4% beneath trees
and 13.5% in gaps (t = 2.68, p < 0.01), based on 230 and
131 measurements, respectively. In the other stand, the moisture contents were 10.4% beneath trees and 11.2% in gaps
(t = 1.69, p < 0.05), based on 179 and 76 measurements, respectively. Thus, there does appear to be variation in duff
moisture based on location either beneath or between tree
crowns.
To partially explain within stand variation in duff moisture, we measured precipitation interception by Pinus
banksiana and Picea mariana. Precipitation from each rain
event, as measured in the gauges in large openings, ranged
from 0.8 to 24.1 mm. Throughfall, as percentage of precipitation, ranged from 0 to 70% (mean 23%) beneath Pinus
© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:18 AM
Color profile: Generic CMYK printer profile
Composite Default screen
Miyanishi and Johnson
1291
Fig. 4. Distributions of duff bulk density (g·cm–3) in unburned
Pinus banksiana – Picea mariana and Picea mariana dominated
stands.
Fig. 6. Distributions of duff depths in Pinus banksiana – Picea
mariana and Picea mariana dominated stands in (a) the Bittern
burn and (b) the Waskesiu burn.
50
Pinus-Picea
Picea
30
25
20
15
10
5
0
Relative frequency (%)
Relative frequency (%)
35
0.02 0.08 0.14 0.20 0.26 0.32 0.38 0.44
20
15
10
5
0
30
20
10
4-6
8-10 12-14 16-18 20-22 24-26 28-30 32-34
Depth classes (cm)
banksiana and from 0 to 41% (mean 16%) beneath Picea
mariana. Thus, both species intercepted approximately 77–
84% of the precipitation, and there was no significant difference found between the two species in their percentage interception of precipitation (t = 1.05, p > 0.05). The
proportion of precipitation intercepted was negatively correlated with the amount of precipitation (r = –0.51, p < 0.05).
Patterns of duff consumption within burns
In unburned Pinus–Picea stands, the distribution of duff
depths (Fig. 5) was normal (symmetric) with a Z g 1 value
of 1.64 (p > 0.05). Although the distribution of duff
depths in the unburned Picea stands also appears normal
(Fig. 5), the Z g 1 value of 9.84, (p < 0.001) indicated skewness. However, this skewness is a result of only two extreme
values. Removing just the two highest (of 621) depth values
changes the Z g 1 value to 1.17 (p > 0.05). Therefore, we conclude that distributions of duff depth in unburned stands are
normally distributed. On the other hand, duff depth distributions for the burned stands at both Bittern and Waskesiu
6-9 12-15 18-21 24-27 30-33 36-39 42-45
b) Waskesiu
50
Pinus-Picea
Picea
40
30
20
10
0
0-2
0-3
-
60
Relative frequency (%)
Relative frequency (%)
Pinus-Picea
Picea
25
Pinus-Picea
Picea
Depth class (cm)
Fig. 5. Distributions of duff depths (cm) in unburned Pinus
banksiana – Picea mariana and Picea mariana dominated
stands.
30
40
0
Bulk density class midpoints (g · cm-3)
a) Bittern
0-2
4-6
8-10 12-14 16-18
Depth class (cm)
20-22
were highly positively skewed (Fig. 6). The Z g 1 values at
Bittern (Fig. 6a) were 11.81 (p < 0.0001) for Pinus–Picea
and 14.15 (p < 0.0001) for Picea stands, while at Waskesiu
(Fig. 6b) they were 8.51 (p < 0.0001) for Pinus–Picea and
7.73 (p < 0.0001) for Picea stands. From these graphs, it is
obvious that the skewness is not due to a couple of extreme
values but to the large number of values in the lowest depth
class. Consequently, assuming that the unburned stands are
representative of the burned stands before the fire, duff
depths before the fire would have been more or less normally distributed, averaging 6.9 cm in Pinus–Picea stands
and 10.3 cm in Picea stands. After both fires, reduction in
overall duff depth was largely due to the high proportion of
zero and near-zero duff depths resulting from complete, or
nearly complete, duff consumption, creating the patches
seen in Fig. 1.
In the Bittern burn, which occurred during wetter duff
conditions (see Fig. 2), most of the area of duff consumption
was attributable to patches in the smallest size classes. The
largest patch recorded was 15 m2 in Pinus–Picea stands and
28 m2 in Picea stands (Fig. 7a). The Waskesiu burn, which
occurred during drier duff conditions, had a wider range of
patch sizes with a maximum of 77 m2 in Pinus–Picea stands
© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:19 AM
Color profile: Generic CMYK printer profile
Composite Default screen
1292
Can. J. For. Res. Vol. 32, 2002
Percentage of total patch area
Percentage of total patch area
Fig. 7. Distribution of percentage of total patch area by patch
size class for Pinus banksiana – Picea mariana and Picea
mariana dominated stands in (a) the Bittern burn and (b) the
Waskesiu burn.
80
a) Bittern
Pinus
Picea
60
40
20
0
40
30
0-2
4-6
Pinus
Picea
2
4
6
8
0
8--1 12-1 16-1 20-2 24-2
2
Patch size class (m )
b) Waskesiu
20
10
0
5
5
5
5
5
5
5
0--5 10-1 20-2 30-3 40-4 50-5 60-6 70-7
2
Patch size class (m )
and 81 m2 in Picea stands (Fig. 7b). Furthermore, 78%
(Pinus–Picea) and 75% (Picea) of the total area of duff consumption in the Waskesiu burn were in patches greater in
size than the largest patches in the Bittern burn (i.e., drier
duff conditions result in larger duff consumption patches).
The patch size distributions differed between stand types in
the wetter duff burn (Bittern); all of the burned duff area in
top-slope Pinus–Picea stands were in patches smaller than
6 m2, while the middle- to bottom-slope Picea stands had
only 22% of the total area of duff consumption in patches
smaller than 6 m2. On the other hand, in the drier duff burn
(Waskesiu), the patch size distributions were more evenly
distributed, irrespective of stand type.
Disregarding stand type, the percentage of the area in
burned patches was significantly less within the Bittern burn
(14.7 ± 2.88%) than within the Waskesiu burn (52.9 ±
4.30%) (t = 7.89, p < 0.001). Within the Bittern burn, the
percent area in patches was significantly less in the Pinus–
Picea stands (4.9 ± 1.40%) than in the Picea stands (24.4 ±
3.39%) (t = 5.33, p < 0.001). However, within the Waskesiu
burn, the percent area in patches did not differ between stand
types: 52.9 ± 6.84% in the Pinus–Picea stands and 58.0 ±
4.99% in the Picea stands (t = 0.67, p > 0.05).
The χ2 values for the spatial relationship between burned
patches and standing stems in the Bittern burn ranged from
50.12 to 381.27 in Pinus–Picea plots and from 0.40 to 52.94
in Picea plots. All were significant at the p < 0.01 level except two Picea plots. In the Waskesiu burn, the χ2 values
ranged from 5.70 to 37.78 in Pinus–Picea plots and from
7.10 to 75.54 in Picea plots. All were significant at the p <
0.01 level except for two of the Pinus–Picea plots, which
were significant at the p < 0.05 level. Thus, 30 of the 32
plots showed a significant spatial relationship between the
burned patches and standing stems. In other words, more
stems were either within patches or touching patches than
would be expected if the stems and patches were independently distributed.
Discussion
Despite the interest of foresters and ecologists in duff consumption because of its importance in post-fire tree regeneration and despite the recognition that the major process
involved in duff consumption is smoldering combustion,
none of the previous duff-consumption studies (see
Miyanishi (2000) for review) have indicated any awareness
of studies on smoldering, such as those by Kinbara et al.
(1967), Moussa et al. (1976), Ohlemiller et al. (1979),
Leisch (1983), Moallemi et al. (1993), Di Blasi (1995),
Buckmaster and Lozinski (1996), and Yi et al. (1998). Although these combustion studies present a number of different models of smoldering, the understanding they all provide
of the smoldering process indicate that the primary variables
that determine duff consumption are bulk density, moisture
content, and depth of duff. Bulk density of the duff is important because it influences diffusivity, which plays a key role
in the rate of heat transfer from the heat generating oxidation zone to the unburned duff. Latent heat of vaporization
of the moisture in the duff acts as a heat sink and, therefore,
determines whether or not sufficient heat is available to
cause pyrolysis of the unburned duff and, thus, propagate
smoldering. Depth of duff determines whether or not the
heat generated by the vertical oxidizing surface is sufficient
to compensate for the convective heat loss from the surface
of the duff. An understanding of the smoldering process also
indicates how these variables are related, e.g., for a given
bulk density, the greater the moisture content, the greater the
critical depth of fuel required to propagate smoldering.
Using commercial peat as a test fuel, we showed experimentally that there was a significant positive linear relationship
between fuel depth and moisture content in determining
smolder propagation. The bulk density (0.097 g·cm–3) of the
peat used in this experiment was very similar to the mean
bulk densities of duff found in Pinus–Picea stands
(0.092 g·cm–3) and in Picea stands (0.094 g·cm–3) and was,
therefore, a reasonable analog of duff.
There was, however, overlap between the points representing depth–moisture combinations that propagated smoldering and those representing extinguishments (Fig. 3a). The
noise in the data suggests that other factors may be playing a
role. In fact, from our understanding of smoldering combustion, we know that bulk density of the fuel is also a control© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:19 AM
Color profile: Generic CMYK printer profile
Composite Default screen
Miyanishi and Johnson
ling variable in determining the limits of smoldering
propagation. Therefore, we would expect there to be a family of curves representing the upper depth–moisture limits
for different bulk densities. Although we attempted to maintain a constant bulk density for all trials, some variability
among trials was inevitable. Any variation in bulk density
would be expected to result in overlap between the propagated and extinguished data sets. Further testing of this relationship with better control of bulk density and use of sets of
trials using different density values would be useful.
In this study, we looked at patterns of duff consumption in
Pinus banksiana – Picea mariana and Picea mariana dominated stands on glaciofluvial substrate in the mixedwood boreal forest of central Saskatchewan with respect to the above
variables and their relationships. Our study found that while
duff bulk density in unburned stands varied, there was no
significant difference between the Pinus–Picea and Picea
stands. Therefore, while bulk density is a factor determining
smolder propagation, we conclude that it would not play an
important role in explaining any differences in duff consumption that may be observed between these two stand
types. It remains to be tested whether or not bulk density differences occur between other stand types.
As found in numerous previous studies (e.g., Van Wagner
1972; Brown et al. 1991; Little et al. 1986), moisture content
of duff was related to duff consumption with significantly
greater area of duff removed by smoldering in the Waskesiu
fire (mean 52.9%) which occurred during drier duff conditions (as indicated by the DMC and DC values) than in the
Bittern fire (mean 14.7%). The present study statistically
confirmed the observation, previously only anecdotally reported, that duff consumption is patchy and occurs preferentially around the bases of trees that had been alive at the
time of the fire. We further showed that in unburned stands,
the duff is generally drier beneath than between tree crowns;
this within stand variation in duff moisture could be explained by interception of precipitation by tree crowns.
Other factors that we did not include in our study, such as
dew formation, would further enhance this effect; tree
crowns intercept long-wave radiation from the ground thus
inhibiting nighttime radiative cooling and dew formation.
Thus, within-stand patchiness of duff consumption could be
explained by the patchiness of duff moisture due to the influence of the tree canopy.
The positive relationship between duff depth and moisture
content in smolder propagation derived from an understanding of the smoldering process and supported empirically in
this study can be used to explain the duff consumption patterns observed in the Bittern and Waskesiu burns. The relationship indicates that at low moisture contents, smoldering
can be propagated in even very thin duff, while at higher
moisture contents, a thicker depth of duff is required to
propagate smoldering. Thus, we can explain why there was
significantly greater duff consumption (as indicated by patch
sizes and total area of duff consumed) in Picea stands than
in Pinus–Picea stands within the Bittern burn but not within
the Waskesiu burn. Converting the DMC to moisture content
(% dry mass) (Van Wagner 1987) and then converting these
gravimetric moisture contents to volumetric moisture, we estimate that the volumetric duff moisture was approximately
12–16% during the Bittern fire and approximately 5–8%
1293
during the Waskesiu fire. Assuming peat is a reasonable analog of duff, we can see from Fig. 3 that duff depth would
not likely have been a limiting factor in smoldering propagation during the Waskesiu fire. However, it is likely to have
played a role in limiting propagation during the Bittern fire
when duff moisture contents were higher, particularly in
Pinus–Picea stands, which have significantly thinner duff
(Fig. 5). Thus, if smoldering extinguished upon encountering
the thin sections of duff in the Pinus–Picea stands, we
would expect that most of the burned patches would be very
small compared with those in the thicker duff of the Picea
stands (Fig. 7a), where smoldering would have been more
likely to be propagated. On the other hand, with the extreme
dry duff conditions during the Waskesiu fire, depth would
not be limiting smolder propagation, and we would not expect to find large differences in either burned patch size
(Fig. 7b) or total area of patches between the two stand
types.
Past duff consumption studies have only considered mean
duff consumption of stands, even though it has been obvious
in boreal forest burns that there is a distinct pattern of
burned and unburned duff within stands. This use of mean
measures of duff consumption have tended to obscure the
patterns apparent both within stands and among stands located on different terrain. Our approach of studying duff
consumption by an understanding of the smoldering combustion process allowed us not only to identify the key factors of moisture, depth, and bulk density determining
propagation of smoldering in duff but also to discover the
nature of the relationships among these variables. The interaction of moisture and depth then provided a reasonable explanation for the observed differences in duff consumption
patterns between stand types in the Bittern fire as well as the
lack of differences between stand types in the Waskesiu fire.
The landscape pattern of duff consumption clearly reveals
the interaction of duff moisture content and duff depth. On
glaciofluvial hillslopes in our study area we know there is a
moisture gradient with wetter areas at the bottom of the
slope (Bridge and Johnson 2000). This moisture gradient is a
function of contributing area, slope angle, and soil
transmissivity. The moisture gradient in turn creates the vegetation composition gradient from the top of hillslopes dominated by Pinus banksiana – Picea mariana grading into
pure Picea mariana stands at the bottom as a result of the
moisture tolerances of these species. Duff depth also varies
along the hillslope, with thin duff in the Pinus–Picea stands
at the tops of glaciofluvial hillslopes and thicker duff in the
Picea stands at middle to bottom slope positions. All of
these environmental gradients (soil moisture, vegetation
composition, duff moisture, and duff depth), thus, are determined ultimately by the hillslope hydrology. Note that the
hydrology is dependent not only on contributing area, slope
angle, and soil transmissivity but also on weather (precipitation and temperature), which was clearly shown by the differences between the two fire events. Consequently, as
concluded by Bridge and Johnson (2000), the hillslope is the
fundamental unit for understanding patterns of recruitment,
not only because it creates the moisture gradient and hence
the species composition gradient but also, as we have shown,
because it determines the patterns of duff consumption
which in turn determine tree seedling recruitment. Thus, to
© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:20 AM
Color profile: Generic CMYK printer profile
Composite Default screen
1294
be able to better predict patterns of duff consumption both
within stands and across the landscape, we need to have
better hydrologic models that predict patterns of duff moisture at varying scales.
Acknowledgements
We gratefully acknowledge M.J. Bajtala, S. Chipman,
S.T.K. Vo, S. Berry, S. Bouchard, T. To, K. Webster, and
M. Wong for their assistance in collection of field data;
S. Vo for conducting the laboratory study on smoldering;
M.J. Bajtala for the spatial analyses; S.L. Gutsell and two
anonymous reviewers for helpful comments on an earlier
draft of the paper; and M. Puddister for drafting the figures.
This project was funded by a grant through the Sustainable
Forest Management Network and a Natural Sciences and
Engineering Research Council operating grant to E.A.J.
References
Adams, J.L. 1966. Prescribed burning techniques for site preparation in cutover jack pine in southeastern Manitoba. Pulp Pap.
Mag. Can. 67: 574–584.
Beaufait, W.R. 1962. Procedures in prescribed burning for jack
pine regeneration. Michigan College of Mining and Technology,
Houghton. For. Cent. Tech. Bull. 9.
Bracewell, J.M., and Robertson, G.W. 1987. Characteristics of soil
organic matter in temperate soils by Curie-point pyrolysis –
mass spectrometry. III. Transformations occurring in surface organic horizons. Geoderma, 40: 333–344.
Bridge, S.R.J., and Johnson, E.A. 2000. Geomorphic principles of
terrain organization and vegetation gradients. J. Veg. Sci. 11:
57–70.
Brown, J.K., Reinhardt, E.D., and Fischer, W.C. 1991. Predicting
duff and woody fuel consumption in northern Idaho prescribed
fires. For. Sci. 37: 1550–1566.
Buckmaster, J., and Lozinski, D. 1996. An elementary discussion
of forward smolder. Combust. Flame, 104: 300–310.
Canada Soil Survey Committee. 1978. The Canadian system of soil
classification. Can. Dep. Agric. Publ. 1646.
Canadian Forestry Service. 1984. Tables for the Canadian Forest
Fire Weather Index System. 4th ed. Environment Canada, Hull,
Que. For. Tech. Rep. 25.
Cayford, J.H. 1966. Some aspects of jack pine regeneration on prescribed burned areas. Can. Dep. For. Bi-Mon. Res. Notes 22(5): 7.
Chrosciewicz, Z. 1959. Controlled burning experiments on jack
pine sites. Forestry Branch, Canadian Department of Northern
Affairs and Natural Resources, Ottawa, Ont. For. Resour. Div.
Tech. Note 72.
Chrosciewicz, Z. 1970. Regeneration of jack pine by burning and
seeding treatments on clear-cut sites in central Ontario. Can.
Dep. Fish. For. Can. For. Serv. Ont. Reg. Inf. Rep. O-X-138.
Chrosciewicz, Z. 1974. Evaluation of fire-produced seedbeds for
jack pine regeneration in central Ontario. Can. J. For. Res. 4:
455–457.
Chrosciewicz, Z. 1976. Burning for black spruce regeneration on a
lowland cutover site in southeastern Manitoba. Can. J. For. Res.
6: 179–186.
DesRochers, A., and Gagnon, R. 1997. Is ring count at ground
level a good estimation of black spruce age? Can. J. For. Res.
27: 1263–1267.
Di Blasi, C. 1995. Smolder spread through thin horizontal fuel layers. In Advanced Computational Methods in Heat Transfer III.
Can. J. For. Res. Vol. 32, 2002
Proceedings of the 3rd International Conference. Edited
by L.C. Wrobel, C.A. Brebbia, and A.J. Nowak. Computational
Mechanics Publications, Southampton, U.K. pp. 323–330.
Dix, R.L., and Swan, J.M.A. 1971 The roles of disturbance and
succession in upland forest at Candle Lake, Saskatchewan. Can.
J. Bot. 49: 657–676.
Drysdale, D. 1985 An introduction to fire dynamics. John Wiley &
Sons, Chichester, U.K.
Dyrness, C.T., and Norum, R.A. 1983. The effects of experimental
fires on black spruce forest floors in interior Alaska. Can. J. For.
Res. 13: 879–893.
Frandsen, W.H. 1987. The influence of moisture and mineral soil
on the combustion limits of smoldering forest duff. Can. J. For.
Res. 17: 1540–1544.
Frandsen, W.H. 1991. Burning rate of smoldering peat. Northwest
Sci. 65: 166–172.
Gutsell, S.L., and Johnson, E.A. 2002. Accurately aging trees and
examining their height growth rates: implications for interpreting forest dynamics. J. Ecol. 90: 153–166.
Hartford, R.A. 1989. Smoldering combustion limits in peat as influenced by moisture, mineral content, and organic bulk density.
In Proceedings of the 10th Conference on Fire and Forest Meteorology. Edited by D.C. MacIver, H. Auld, and R. Whitewood.
Forestry Canada, Ottawa, Ont. pp. 282–286.
Hawkes, B.C. 1993. Factors that influence peat consumption under
dependent burning conditions: a laboratory study. Ph.D. dissertation, University of Montana, Missoula, Mont.
Hillel, D. 1982. Introduction to soil physics. Academic Press Inc.,
Orlando, Fla.
Hungerford, R.D., Frandsen, W.H., and Ryan, K.C. 1995. Ignition
and burning characteristics of organic soils. In Proceedings of the
Tall Timbers Fire Ecology Conference in Wetlands, a Management Perspective. Edited by S.I. Cerulean and R.T. Engstrom. Tall
Timbers Research Station, Tallahassee, Fla. pp. 78–91.
Jameson, J.S. 1961. Observations on factors influencing jack pine
reproduction in Saskatchewan. Can. Dep. For. Res. Div. Tech.
Note 97.
Jeglum, J.K. 1972. Boreal forest wetlands near Candle Lake, central Saskatchewan I. Vegetation. Musk-Ox, 11: 41–58.
Jeglum, J.K. 1973. Boreal forest wetlands near Candle Lake, central Saskatchewan II. Relationship of vegetation to major environmental gradients. Musk-Ox, 12: 32–48.
Johnson, E.A. 1992. Fire and vegetation dynamics: studies from
the North American boreal forest. Cambridge University Press,
Cambridge, U.K.
Johnson, E.A., Miyanishi, K., and O’Brien, N. 1999. Long-term reconstruction of the fire season in the mixedwood boreal forest of
western Canada. Can. J. Bot. 77: 1185–1188.
Kinbara, T., Endo, H., and Sega, S. 1967. Downward propagation
of smoldering combustion through solid materials. In Proceedings, 11th Symposium (International) on Combustion. The Combustion Institute, Pittsburgh, Pa. pp. 525–531.
Little, S.N., Ottmar, R.D., and Ohmann, J.L. 1986. Predicting duff
consumption from prescribed burns on conifer clearcuts in western Oregon and western Washington. USDA For. Serv. Res. Pap.
PNW-362.
Mason, C.F. 1976. Decomposition. Edward Arnold Publishers Ltd.,
London.
Miyanishi, K. 2001. Duff consumption. In Forest fires: behavior and
ecological effects. Edited by E.A. Johnson and K. Miyanishi. Academic Press, San Diego, Calif. pp. 437–475.
Moallemi, M.K., Zhang, H., and Kumar, S. 1993. Numerical modeling of two-dimensional smoldering process. Combust. Flame,
95: 17–182.
© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:21 AM
Color profile: Generic CMYK printer profile
Composite Default screen
Miyanishi and Johnson
Moussa, N.A., Toong, T.Y., and Garris, C.A. 1976. Mechanism of
smoldering of cellulosic materials. In Proceedings, 16th Symposium (International) on Combustion. The Combustion Institute,
Pittsburgh, Pa. pp. 1447–1457.
Norcliffe, G.B. 1982. Inferential statistics for geographers: an introduction. 2nd ed. Hutchinson & Co. Ltd., Essex, U.K.
Ohlemiller, T.J. 1985. Modeling of smoldering combustion propagation. Prog. Energy Combust. Sci. 11: 277–310.
Ohlemiller, T.J. 1990. Smoldering combustion propagation through a
permeable horizontal fuel layer. Combust. Flame, 81: 341–353.
Ohlemiller, T.J., Bellan, J., and Rogers, F. 1979. A model of smoldering combustion applied to flexible polyurethane foams.
Combust. Flame, 36: 197–215.
O’Loughlin, E.M. 1981. Saturation regions in catchments and their relations to soil and topographic properties. J. Hydrol. 53: 229–246.
Palmer, K.N. 1957. Smouldering combustion in dusts and fibrous
materials. Combust. Flame, 1: 129–154.
Peter, S. 1992. Heat transfer in soils beneath a spreading fire.
Ph.D. dissertation, University of New Brunswick, Fredericton,
N.B.
Quinn, P., Beven, K., Chevallier, P., and Planchon, O. 1992. The
prediction of hillslope flow paths for distributed hydrological
modeling using digital terrain models. In Terrain analysis and
distributed modeling in hydrology. Edited by K.J. Beven and
I.D. Moore. John Wiley & Sons, New York. pp. 63–84.
Reinhardt, E.D., Brown, J.K., and Fischer, W.C. 1989. Fuel consumption from prescribed fire in northern Idaho logging slash.
In Proceedings of the 10th Conference on Fire and Forest Meteorology. Edited by D.C. MacIver, H. Auld, and R. Whitewood.
Forestry Canada, Ottawa, Ont. pp. 155–160.
Rowe, J.S. 1972. Forest regions of Canada. Can. Dep. For. Publ.
1300.
1295
Samran, S., Woodard, P.M., and Rothwell, R.L. 1995. The effect of
soil water on ground fuel availability. For. Sci. 41: 255–267.
Thomas, P.A., and Wein, R.W. 1985. The influence of shelter and
the hypothetical effect of fire severity on the postfire establishment of conifers from seed. Can. J. For. Res. 15: 148–155.
Van Wagner, C.E. 1972. Duff consumption by fire in eastern pine
stands. Can. J. For. Res. 2: 34–39.
Van Wagner, C.E. 1974. Effect of duff weight on drying rate. Can.
Dep. For. Bi-Mon. Res. Notes 30(2): 1–12.
Van Wagner, C.E. 1987. Development and structure of the Canadian Forest Fire Weather Index System. Canadian Forest Service, Ottawa, Ont. For. Tech. Rep. 35.
Weber, M.G., Hummel, M., and Van Wagner, C.E. 1987. Selected
parameters of fire behavior and Pinus banksiana Lamb. regeneration in eastern Ontario. For. Chron. 63: 340–346.
Weir, J.M.H., Johnson, E.A., and Miyanishi, K. 2000. Fire frequency and the spatial age mosaic of the mixedwood boreal forest in western Canada. Ecol. Appl. 10: 1162–1177.
Williams, S.T., and Gray, T.R.G. 1974. Decomposition of litter on
the soil surface. In Biology of plant litter decomposition. Edited
by C.H. Dickinson and G.J.F. Pugh. Academic Press, London.
pp. 611–632.
Yi, S-C., Bae, S.Y., and Kim, S.H. 1998. Heat transfer of a smoldering flammable substrate. Part 2. A theoretical model and its
application. J. Fire Sci. 16: 32–45.
Zar, J.H. 1996. Biostatistical analysis. Prentice Hall, Upper Saddle
River, N.J.
Zasada, J.C., Norum, R.A., Van Veldhuizen, R.M., and Teutsch,
C.E. 1983. Artificial regeneration of trees and tall shrubs in experimentally burned upland black spruce/feather moss stands in
Alaska. Can. J. For. Res. 13: 903–913.
© 2002 NRC Canada
I:\cjfr\cjfr32\cjfr-07\X02-051.vp
Tuesday, July 09, 2002 11:44:22 AM