BMNRI Home > Publications
> Tech Notes > No.
Tech Notes No. 2
Fire Histories: Overview of Methods and Applications
Kathleen R. Maruoka and James K. Agee College of Forest Resources,
University of Washington, Seattle, Washington
Fire has the potential to change the structure and species composition
of a forest and has undoubtedly influenced the development of the
forests we see today in the Blue Mountains. While we can directly
observe the effects of recent fires on present forest structure
and composition, we must infer the effects of previous fires using
current stand structure and composition together with a record of
Fire can be thought of as part of a "disturbance complex"
comprising insects, pathogens, wind, and other disturbances, which
contribute to landscape and species diversity. The interactions
are dynamic with combinations unique to every forest. Because these
factors are interdependent, removing or altering one of them changes
the roles of the other interacting disturbances. For instance, it
has been suggested that fire exclusion has resulted in larger western
spruce budworm (Choristoneura occidentalis) outbreaks than previously
recorded. In the absence of fire, the western spruce budworm plays
a similar role in the forest, killing tree species and age classes
that would have burned previously if wildfire ignitions had not
been successfully suppressed. Because larger outbreaks increase
the number of dead and weakened trees, they influence the magnitude
of subsequent fire and wind events.
Determining the historic fire frequency for a stand helps us understand
the role fire has played in stand development. This information
is important for interpreting several of the current "forest
health" issues in the Blue Mountains, and serves as base information
for forming forest management strategies that incorporate natural
or prescribed fire.
(Return to contents)
Fire History Approaches
The two main approaches to developing a fire history are through
analysis of point frequencies and area frequencies. Point frequencies
assess fire occurrence at one location while area frequencies assess
fire occurrence at the scale of the landscape. Although both yield
a "fire frequency," the frequencies represent different
types of information because of this difference in scale. Selecting
the appropriate method depends primarily on the vegetation types
and physical features of the study area, as well as the type of
fire evidence present.
Vegetation types are important to consider because of different
flammabilities, both in terms of standing fuel, and in rates and
levels of fuel accumulation. Topography and physical features are
important because changes in substrate or terrain can affect vegetation
composition, cover, and continuity. Such changes can serve as effective
barriers to fire spread and deserve consideration when formulating
an approach to constructing a fire history.
(Return to contents)
A point frequency represents repeated occurrences of fires at a
single location and addresses the question "how often did fire
burn across this point?". Point frequency methods are best
used to describe the occurrence of fires in areas that experience
low-severity fires, because such fires commonly leave small, datable
scars on the mature trees. These fires generally consume forest
litter and kill small understory trees without killing larger trees.
This type of fire regime is often associated with ponderosa pine
(Pinus ponderosa) forests and other lower elevation forests where
pine is a codominant species.
Thick bark usually protects larger trees from being killed by low-severity
fires. However, bark fissures and previous wounds offer less thermal
protection and fires may kill localized portions of the cambium
of a tree. Subsequent annual growth rings heal over the damaged
cambial tissues from the edge of the wound inward, resulting in
distinct ring patterns when viewed in cross-section (figure 1).
These scars are the basis of a point frequency.
Figure 1. Fire scar formation. The cambium beneath a thin area
of bark (indicated by arrow) is susceptible to damage from fire
(a). Subsequent annual growth rings heal from the wound edge inward,
producing distinct ring patterns (b, c). Repeated fires may create
multiple scars along a single radius (d).
While a point frequency can be constructed using the scar record
from one tree, a more comprehensive fire record can be compiled
using data from several neighboring trees with fire scars. Because
fires do not burn across landscapes uniformly, and because every
fire does not scar every tree, increasing the number of sampled
trees also increases the likelihood that a fire will be detected.
Caution must be taken, however, to ensure that the sampled trees
are close enough to represent one point on a landscape. If the trees
are too distant, fires will be presumed to have burned across areas
that in fact they may not have, and the frequency estimate will
represent an area frequency rather than a point frequency. In general,
a point estimate should not include samples from an area larger
than the smallest area usually burned. In the Blue Mountains, this
may be from one to several acres. Details about point frequency
sampling methods and compiling fire chronologies may be found in
Arno and Sneck (1977), Dieterich (1980), and Agee (1993).
(Return to contents)
The utility of an area frequency lies not in expressing the fire
frequency at one point, but across a much larger area. It is typically
used in areas where fires are severe and kill most of the trees.
The fire record is represented by different age classes of trees
across the landscape. There are two types of approaches to determine
area frequencies, the first of which is called "natural fire
rotation," and the second called "fire cycle."
Natural Fire Rotation - Natural fire rotation is an area frequency
technique that evaluates fire-return intervals in a study area based
on total area burned over time, usually measured in centuries. It
requires the reconstruction of all past fire events during the period
of study. Unlike the point frequency method, which relies on the
presence of fire-scarred trees, the area frequency method uses stand
ages as well as any fire scars to recreate the extent of past fire
occurrences. Although fire scars are useful when present, stand
ages and charcoal are often the only evidence remaining after severe
fires. In the Blue Mountains, the natural fire rotation method can
be applied to lodgepole pine (Pinus contorta), mountain hemlock
(Tsuga mertensiana), subalpine fir (Abies lasiocarpa), and moist
grand fir (Abies grandis) and Douglas-fir (Pseudotsuga menziesii)
sites because these forest types are often associated with stand-replacing
The natural fire rotation is calculated as the quotient between
the total time period considered and the total proportion of the
landscape burned during that time period. Because some areas may
have burned more than once, the total proportion of the study area
burned may exceed one. For example, if 20,000 acres burned in a
I 0,000 acre study area over a period of I 00 years, the natural
fire rotation calculation would be: 100 years/ [20,000 acres burned/10,000
acre study area] = 50 years.
A natural fire rotation does not imply that every stand in the
study area burns with the same frequency within the calculated time
period. Some stands used in the calculation will have burned more
than once and some not at all. If the study area is topographically
or vegetatively fragmented, it may be useful to divide the study
area into smaller areas to reduce the amount of spatial variability.
Similarly, considering time periods with known differences in human
or climatic influences as separate periods may reduce temporal variability.
Inferences drawn from areas or periods with very different features
may be misleading.
Compiling information to calculate a natural fire rotation begins
with identifying stand ages across the study area. This can be accomplished
using aerial photos and field reconnaissance. Once the stands have
been identified, establishment periods are documented by coring
several trees within each stand. Stand perimeters can then be more
clearly delineated. Sampling density depends on the age, size, species,
and topographic uniformity of the stand.
Establishment and regeneration periods differ with tree species,
location, and fire severity, and may continue for several decades
following a fire. However, establishment periods are usually much
shorter than the fire-free intervals, so using differences in stand
establishment dates serves as a proxy for detecting different fire
events. Because there is often a lag time associated with tree establishment
following a fire, the earliest tree establishment date in a stand
is assumed to be a conservative estimate of the year that fire occurred
in the stand; the fire actually may have occurred several years
prior to the year that the sampled tree became established.
Stands are then combined into a stand age mosaic. Topographic features
such as ridges, streams, rivers, and swamps, as well as vegetative
features such as large areas of discontinuous fuel are often natural
fuel breaks and are used in conjunction with the stand age mosaic
to interpret the fire history of the study area. For example, in
areas without significant fuel breaks, stands with similar establishment
dates separated by a younger stand most likely originated after
the same fire event. The younger stand also burned in the earlier
fire but was created by a subsequent fire burning through a portion
of the earlier burn.
Two aspects of the calculation promote a conservative estimate
of the natural fire rotation period. First, we may be unable to
detect light burns that leave little evidence. Second, the evidence
of some past fires may have been obliterated by more recent fires.
Both may result in a longer estimate of natural fire rotation than
Fire Cycle--The fire cycle is a statistical model of fire history
based on current stand ages (Johnson and Van Wagner 1985). The model
is based on the negative exponential or Weibull distributions, and
best represents fire frequency in large areas where a small propor-
tion of stands burn intensely each year, resulting in a mosaic of
stands with different ages. Certain assumptions are made in fire
cycle calculations regarding ignition patterns, fire size, stand
flammability, topographic uniformity, and stable climate. They assume
ignition patterns are random, that only a small portion of the landscape
burns each year, and that the fire cycle has been relatively constant
over time. These assumptions limit applications in topographically
fragmented terrain, or in areas where the largest fires are a significant
portion of the total study area. The fire cycle best approximates
fire occurrence when very large study areas (200,000-300,000 acres
or more) are used, so that the average fire size is only a small
proportion of the total landscape. The fire cycle model is widely
used in boreal landscapes but has not been applied as much in temperate
(Return to contents)
Sampling and Interpreting Fire Scars
Fire scars are basal wounds which may extend over I0 feet up the
bole of a tree. Large fire scars are often referred to as cat-faces.
Once a tree has been scarred, it is likely to be scarred along the
same radius in subsequent fires. Trees with multiple scars provide
the most information about fire occurrences.
There are several ways to sample fire scars. Collecting a complete
cross-section from a scarred tree is the most definitive method
for identifying fire scars but destroys the tree. Another method
is to collect a cross-section of one-half of the scarred surface.
This can be accomplished by cutting a wedge containing the scars.
Alternatively, the tree can be sampled by making two parallel horizontal
cuts and two plunge cuts, and extracting the section (Arno and Sneck
1977). These sampling techniques have the advantage of, allowing
the tree to survive, but may predispose it to windthrow or pathogens.
Thus, it may be important to consider safety precautions as well
as scar record when selecting trees in certain areas.
In instances where collecting sections is not possible, several
increment cores of a fire-scarred tree may be taken (Barrett and
Arno 1988). One core sample should contain the fire-scarred ring
and the growth ring immediately before it. Another core taken from
an unscarred portion of the tree can then be used as a reference
to match the ring sequences of the other cores and deduce the year
of scarring. This method is the least intrusive, but generally provides
a record limited to the most recent scar. Stumps with fire scars
are another source of information. If a stump is sound, an entire
cross-section can simply be sliced through the stump. After the
scar samples have been collected and airdried, they are sanded with
successively finer grades of sandpaper until the ring patterns become
clear. Counting rings from the bark edge inward using a stereoscopic
microscope results in a single record of fire occurrences. The average
number of years between every pair of sequential scars is the mean
fire-return interval contained in the scar sample. The mean fire-return
interval and the variability of the mean fire-return interval are
important descriptors of fire frequency.
Cross-dating fire scars refines and extends fire scar records.
Cross-dating is accomplished by matching growth ring patterns from
two or more samples. It allows us to distinguish between closely-spaced
fire scars and to determine the actual years that fire burned. Stump
samples must be cross-dated unless the exact year the tree died
Cross-dating fire scars can be accomplished two ways. Unique ring
patterns can be identified by comparing two or more scar samples
Figure 3. Cross-dating fire scar samples. These hypothetical samples
were taken in 1992. Marker years are identified by an asterisk (*).
Fire scar years are indicated using heavy arrows. Note that combining
scar records results in more fire occurrences than represented on
each individual sample.
These marker ring patterns can be used to calibrate the samples
with each other. Alternatively, a master ring chronology can be
compiled using increment core samples from several nearby trees
against which ring patterns from the scar samples can be compared.
The advantage to using a master chronology is that missing rings
can usually be detected and ring counts can be adjusted accordingly.
It should be noted that the ring patterns near the scars are often
distorted and may confound cross-dating efforts.
(Return to contents)
Management Uses and Implications
Interpreting current stand structure with fire history information
can provide a more complete picture of stand development than stand
structure alone. Age cohorts, spatial distributions, and species
distributions of trees within a stand or across a landscape may
reflect fire frequency, extent, and severity.
An area in the Burns Ranger District in the Malheur National Forest
near Myrtle Creek provides a good example of how stand structure
and fire history data can be combined. A sample of fifty trees representative
of the species and age structure at this site shows that most of
the older trees are ponderosa pine, while the younger trees are
primarily grand fir (figure 4).
Figure 4. Tree establishment dates in 10-year-interval classes
and historic fire occurrences (T) at Myrtle Creek. PSME = Douglas-
fir, PIPO = ponderosa pine, ABGR = grandfir. Note the increase in
grandfir after frequent fires ceased.
During the period of 1752 to 1890, there were I0 fire scars at
this site for a mean fire-return interval of 15.3 years and an interval
range from 5 to 23 years. The absence of fires prior to 1752 is
probably more related to the absence of fire evidence than the absence
of fire. Fires were frequent between 1752 and 1890, and were probably
low-intensity. Ponderosa pine was able to withstand the fires, while
thinner-barked seedlings and saplings of grand fir and Douglas-fir
were killed. When the fire frequency decreased in the late nineteenth
century, trees that were unable to survive under a frequent fire
scenario were able to establish and survive. Examining stand development
in the presence and absence of fire may foreshadow changes which
may have future management implications. For instance, extended
periods without fire in forests that historically burned frequently
may result in significant changes in stand composition and structure
and have the potential to change the severity of future fire events
from low to moderate or high. Structural changes might also influence
the magnitude or types of other disturbances in the forest, such
as insect and pathogen outbreaks.
Similarly, fire history studies can be applied to prescribed burning
programs. Mean fire-return intervals are often used as guidelines
for fire frequency in prescribed burning programs. However, the
temporal and spatial variability of fire events within and between
forest types may have more important implications thansimply a mean
fire-return interval and should also be considered. Using the Myrtle
Creek example and assuming a desired prescribed fire plan to burn
the forest at a "natural" fire return interval, the 15.3-year
mean fire-return interval would be recreated by simulating both
the mean and its variances Rather than fires exactly every 15.3
years, a variable set of intervals with the same mean is desirable.
For example, fire return intervals of 9, 20, 22, 5, 12, 20, and
19 would produce a mean fire return interval of 15.3 years and mimic
the variability of historic fire frequency.
(Return to contents)
A fire history documents past fire events, which occurred under
different climatic and floristic conditions than current conditions.
Thus, applying fire history to forest management should be done
with caution after considering these influences. Nevertheless, a
fire history of an area can provide valuable information about the
influence of fire on forest development. This understanding can
be used by managers interested in using fire to achieve management
(Return to contents)
Agee, J.K. 1993. Fire ecology of Pacific Northwest Forests. IslandPress,
Washington, D.C. 493 p.
Arno, S.F. and K.M. Sneck. 1977. A method for determining fire
history in coniferous forests of the mountain west. General Technical
Report INT42. U.S. Dept of Agric., Forest Service, Inter-mountain
Research Station. Ogden, UT. 28 p.
Barrett, S. and S.F. Arno. 1988. Increment-borer methods for determining
fire history in coniferous forests. General Technical Report INT-244.
U.S. Dept. of Agric., Forest Service, Intermountain Research Station,
Ogden, UT. 15 p.
Dieterich, J.H. 1980. The composite fire interval- A tool for more
accurate interpretation of fire history. In: Stokes, M.A, and J.H.
Dieterich, tech. coords. Proceedings of the fire history workshop.
General Technical Report RM-8 1. U.S. Dept. of Agric., Forest Service,
Rocky Mountain Research Station. Fort Collins, CO. p. 8-14.
Johnson, E.A. and C.E. Van Wagner. 1985. The theoryand use of two
fire history models. Can. J. For. Res. 15:214-220.
(Return to contents)
TechNotes are produced in cooperation with the Blue
Mountains Natural Resources Foundation and Institute Partners.