19.2 Dendroecological Methods
some fires may not have been recorded by the trees
sampled (Swetnam et aI., 1999). The length of the
fire scar record can be extended by cross-dating
records from snags and logs. Fire scar records from
individual trees are points in space, and careful
selection of trees for sampling may produce a detailed record of past fires if the sampled trees are
well distributed spatially and have multiple wellpreserved fire scars (Swetnam and Baisan, 1996).
For example, in the southwestern United States, approximately 10 to 30 trees per site were sampled
over a number of sites that generally had spatial
extents of 10 to 100 ha (Swetnam and Baisan,
1996). Reconstruction of the spatial extent of lowintensity fires by means of fire-scar records is difficult because the fires may not affect some trees.
Spatial extents can be estimated, however, by examining patterns of synchrony and asynchrony in
fire dates among trees within and among sites, under the assumption that synchronous fire dates represent larger areas burned than do asynchronous
fire dates (Swetnam and Baisan, 1996). Where trees
have been sampled nonrandomly for fire scars,
though, caution is warranted in extrapolating reconstructed fire histories outside the areas sampled
(T. Veblen, personal communication).
For some ecosystems, such as subalpine forests,
fire histories usually cannot be constructed from
fire-scarred trees because the dominant tree species
are highly fire sensitive and rarely survive fires
(Fall, 1997). In these systems, dendroecological
methods may be used to establish stand age the time
since the last stand-replacing fire (Romme and
Knight, 1981). This method has a less precise temporal resolution than does fire-scar dating because
of uncertainty introduced by the time taken for
trees to establish in a burned area and grow to a
height detectable by dendrochronological methods
(Millspaugh and Whitlock, 1995). The spatial extent of a stand-replacing event can be estimated by
delineating the area occupied by the stand itself, a
more exact determination of spatial scale than is
possible using fire-scar methods. However, uncertainty is introduced by obliteration of the spatial
pattern of older disturbances by more recent fires
(Millspaugh and Whitlock, 1995).
Compilations of fire-scar records include the
proceedings of a fire history workshop held in 1980
(Stokes and Dieterich, 1980), which contains fire
history summaries for many areas in North America and in Sweden. Swetnam and Baisan (1996)
compiled fire-scar chronologies from 63 sites in the
southwestern United States and analyzed regional
variability in fire regimes in a range of vegetation
types to provide baseline information and show as275
sociations with climatic conditions. Barrett et aI.
(1997) acquired long-term records of fire histories
in the interior Columbia River basin and mapped
fire locations as fire episodes (defined as five-year
periods having abundant and widespread fire evidence) in the region from 1540 to 1940. They also
related fire episodes to regional drought and calculated mean annual acreage burned. Heyerdahl et
aI. (1995) assembled a fire history database for the
Pacific Northwest. Maps of twentieth-century fires
(1900-1993) on National Forests in the Sierra
Nevada, California, were digitized (USDA Forest
Service, 1996), and characteristics of mapped fire
patterns were analyzed by McKelvey and Busse
(1996).
Dendroecological methods have been used to reconstruct disturbances other than fire. Flood events
can be detected by examining flood-scarred streambank trees (McCord, 1996; Poff et aI., 1997). Stand
ages can be reconstructed in riparian woodlands to
investigate the effects of flooding (Baker, 1988).
Tree-ring growth patterns associated with known
insect or disease outbreaks can be the basis for reconstructing disturbance histories in forest stands
(Baker and Veblen, 1990; Swetnam and Betancourt, 1998).
In addition to their limitation to wooded areas,
dendroecological methods may not be suitable if
trees have indistinct rings, rings that are not the result of an annual cycle, too little variability among
rings, or too many missing rings (Fritts and Swetnam, 1989). Trees must also attain a sufficient age
to provide useful information. Other important limitations of tree-ring methods are their restriction to
the time periods set by maximum tree lifespans and
the loss of detail in the record with time due to tree
mortality (Clark, 1990). Clark (1990) obtained relatively few fire scars on red pine in northwestern
Minnesota prior to 1800 because of tree mortality,
but concurrent charcoal analysis (see Section 19.5)
indicated that fire was equally frequent prior to
1800 in the area.
EXAMPLES
1. Ful€ et aI. (1997) used fire-scar and tree-ring
dating to quantify the presettlement fire disturbance
regime and forest structure of a ponderosa pine forest in Arizona, including determination of the mean
fire interval, to establish reference presettlement
conditions for comparison with current conditions
and to serve as a goal for restoration.
2. In the Boundary Waters Canoe Area, Minnesota, Heinselman (1973) used fire-scarred trees
and historical records such as General Land Office
notes (see Section 19.6) to determine stand origin
some fires may not have been recorded by the trees
sampled (Swetnam et aI., 1999). The length of the
fire scar record can be extended by cross-dating
records from snags and logs. Fire scar records from
individual trees are points in space, and careful
selection of trees for sampling may produce a detailed record of past fires if the sampled trees are
well distributed spatially and have multiple wellpreserved fire scars (Swetnam and Baisan, 1996).
For example, in the southwestern United States, approximately 10 to 30 trees per site were sampled
over a number of sites that generally had spatial
extents of 10 to 100 ha (Swetnam and Baisan,
1996). Reconstruction of the spatial extent of lowintensity fires by means of fire-scar records is difficult because the fires may not affect some trees.
Spatial extents can be estimated, however, by examining patterns of synchrony and asynchrony in
fire dates among trees within and among sites, under the assumption that synchronous fire dates represent larger areas burned than do asynchronous
fire dates (Swetnam and Baisan, 1996). Where trees
have been sampled nonrandomly for fire scars,
though, caution is warranted in extrapolating reconstructed fire histories outside the areas sampled
(T. Veblen, personal communication).
For some ecosystems, such as subalpine forests,
fire histories usually cannot be constructed from
fire-scarred trees because the dominant tree species
are highly fire sensitive and rarely survive fires
(Fall, 1997). In these systems, dendroecological
methods may be used to establish stand age the time
since the last stand-replacing fire (Romme and
Knight, 1981). This method has a less precise temporal resolution than does fire-scar dating because
of uncertainty introduced by the time taken for
trees to establish in a burned area and grow to a
height detectable by dendrochronological methods
(Millspaugh and Whitlock, 1995). The spatial extent of a stand-replacing event can be estimated by
delineating the area occupied by the stand itself, a
more exact determination of spatial scale than is
possible using fire-scar methods. However, uncertainty is introduced by obliteration of the spatial
pattern of older disturbances by more recent fires
(Millspaugh and Whitlock, 1995).
Compilations of fire-scar records include the
proceedings of a fire history workshop held in 1980
(Stokes and Dieterich, 1980), which contains fire
history summaries for many areas in North America and in Sweden. Swetnam and Baisan (1996)
compiled fire-scar chronologies from 63 sites in the
southwestern United States and analyzed regional
variability in fire regimes in a range of vegetation
types to provide baseline information and show as275
sociations with climatic conditions. Barrett et aI.
(1997) acquired long-term records of fire histories
in the interior Columbia River basin and mapped
fire locations as fire episodes (defined as five-year
periods having abundant and widespread fire evidence) in the region from 1540 to 1940. They also
related fire episodes to regional drought and calculated mean annual acreage burned. Heyerdahl et
aI. (1995) assembled a fire history database for the
Pacific Northwest. Maps of twentieth-century fires
(1900-1993) on National Forests in the Sierra
Nevada, California, were digitized (USDA Forest
Service, 1996), and characteristics of mapped fire
patterns were analyzed by McKelvey and Busse
(1996).
Dendroecological methods have been used to reconstruct disturbances other than fire. Flood events
can be detected by examining flood-scarred streambank trees (McCord, 1996; Poff et aI., 1997). Stand
ages can be reconstructed in riparian woodlands to
investigate the effects of flooding (Baker, 1988).
Tree-ring growth patterns associated with known
insect or disease outbreaks can be the basis for reconstructing disturbance histories in forest stands
(Baker and Veblen, 1990; Swetnam and Betancourt, 1998).
In addition to their limitation to wooded areas,
dendroecological methods may not be suitable if
trees have indistinct rings, rings that are not the result of an annual cycle, too little variability among
rings, or too many missing rings (Fritts and Swetnam, 1989). Trees must also attain a sufficient age
to provide useful information. Other important limitations of tree-ring methods are their restriction to
the time periods set by maximum tree lifespans and
the loss of detail in the record with time due to tree
mortality (Clark, 1990). Clark (1990) obtained relatively few fire scars on red pine in northwestern
Minnesota prior to 1800 because of tree mortality,
but concurrent charcoal analysis (see Section 19.5)
indicated that fire was equally frequent prior to
1800 in the area.
EXAMPLES
1. Ful€ et aI. (1997) used fire-scar and tree-ring
dating to quantify the presettlement fire disturbance
regime and forest structure of a ponderosa pine forest in Arizona, including determination of the mean
fire interval, to establish reference presettlement
conditions for comparison with current conditions
and to serve as a goal for restoration.
2. In the Boundary Waters Canoe Area, Minnesota, Heinselman (1973) used fire-scarred trees
and historical records such as General Land Office
notes (see Section 19.6) to determine stand origin