Chapter 6
Fire and Climate Histories in Sequoia-Mixed Conifer Forests of the Sierra Nevada
Thomas W. Swetnam, Christopher H. Baisan and Anthony C. Caprio
6.1 Introduction
The fire history of Sierra Nevada forests provides a window on past ecological
responses to climatic changes. We have reconstructed detailed chronologies of fire
events within pine and mixed conifer forests by sampling and analyzing hundreds of fire
scarred trees. The frequent surface fires that swept through Sierran forests for millennia
before the arrival of European settlers were generally of low intensity, and so most
mature pine and fir trees survived dozens of these events throughout their lifetimes.
Many giant sequoias are over 2,000 and years old and have survived hundreds of surface
fires. Even though surface fires were apparently widespread and frequent, only a small
proportion of conifer trees consistently incurred fire scars at their bases. The fire scar
record is therefore fragmentary and non-randomly distributed across the landscape, but
by sampling many fire scarred trees widely distributed within stands it is possible to
compile relatively complete chronologies of widespread fires. Because fire ignition and
spread is partly controlled by weather and climatic factors, assessment of long fire
chronologies allows us to detect and evaluate patterns that are indicative of past climatefire relations.
Fire frequency and fire synchrony are two fire regimes parameters that are
particularly useful for assessing past climatic responses of ecosystems. These two
parameters are the main focus of this chapter. A simplistic assumption might be that
increased fire frequency occurs during droughts. However, because fire occurrence is
contingent on both fuel condition (e.g., moisture content) and fuel amount and
distribution, frequent or long lasting droughts may eventually serve to inhibit fire
occurrence through limitation of available fuel for fire ignitions and spread. This may be
especially true in high frequency fire regimes, where surface fires regularly consume
fuels on the forest floor (e.g., grasses and tree leaves), so that ignition and spread of
subsequent fires depends upon adequate re-growth and accumulation of fuels between
fire events. Moreover, different permutations of warmer/drier, cooler/wetter (or
warmer/wetter and cooler/drier) conditions may act to alter fire frequency in different and
complex ways. Climate varies across a range of temporal and spatial scales, e.g., from
seasonal to millennial scales, and from forest stands to regional scales. The variations at
each of these scales may result in different fire regime responses.
Fire synchrony is the temporal pattern of correspondence of fire events among
multiple points in space. This importance of this fire regime parameter has not been as
widely or explicitly recognized by fire ecologists or fire historians (but see Swetnam and
Betancourt 1990, 1998, Swetnam 1993, Grissino-Mayer and Swetnam 1999, Veblen et al.
1999, Kitberger et al. 2001). The analyses of fire synchrony has developed naturally in
tree-ring based fire history studies of surface fire regimes because the primary record –
the fire scarred tree – is by nature a “point record”. Spatial inferences of past fire extent
is derived from evaluations of synchrony of the fire dates recorded at the points within a
network of points. Surface fires are irregularly and incompletely recorded by fire scarred
trees, and therefore it is not possible to reconstruct the precise perimeter of past fires, or
to develop “time since fire maps” that might be used to reconstruct “fire rotation” or “fire
cycle” estimates (e.g., Johnson and Van Wagner 198x). Instead, by sampling widely
within relatively homogeneous stands, and then by compiling composite chronologies of
fire scarred trees, it is possible to infer patterns of relative fire extent by assessing the
degree of synchrony of fire dates among sampled fire-scarred trees. Highly synchronous
dates among sampled points are probably indicative of fire years when a relatively large
proportion of the sampled area burned, whereas fire dates recorded by a small number or
proportion of points probably represents relatively smaller areas burned during those
years. There are a number of important assumptions and potential biases inherent in this
approach to fire history reconstruction and interpretation which we will discuss later in
the chapter. At this point, it is mainly important to note that patterns of fire synchrony
can be assessed at different spatial scales and resolutions, and synchrony at these
different scales leads to different inferences. We argue that patterns of fire synchrony at
the broadest scale (regional, continental, etc.) are most likely to reveal climatic
associations because (1) extreme climatic events that affect fire ignition and spread, such
as drought, are known to occur at regional and broader scales, (2) no controls other than
climatic events are known to operate at these broad scales that would result in annual
synchrony of fire events. Patterns of fire synchrony at the finest spatial scale (between
fire scarred trees, within stands) is most likely to reflect patterns of fire spread and
relative extent within stands, and these patterns are more affected by local ecological
processes and peculiarities of sites, that the composites at broader scales.
In this chapter we present examples from our past and ongoing studies of fire
history and climate-fire associations in pine and mixed conifer forests on the western
slope of the Sierra Nevada. We use these examples to demonstrate how fire frequency
and fire synchrony patterns are useful in identifying responses of fire regimes to past
climate variability at the scales of forest stands to the region. We are developing
extensive networks of fire chronologies along elevational networks in the Sierra Nevada.
The details of elevation/forest type-related fire patterns are described in other works
(Caprio and Swetnam 199x, Swetnam et al in preparation). Here we use examples from
these collections to illustrate fire typical frequency and synchrony patterns at the stand to
watershed spatial scales in pine and mixed conifer forests. Last, we use examples from
our giant sequoia fire history work to show how inter-annual to centennial-scale climatic
changes have affected surface fire regimes in the Sierra Nevada.
6.2 Study Areas
Our fire history collections were obtained along four elevational transects on the
west slope of the Sierra Nevada (Fig. 1). Each of these transects includes 10 to 15 stands
where we have collected partial cross sections from 10 to 50 or more fire scarred trees.
Each of the transects is also anchored by a collection within a giant sequoia grove
(usually within the upper one third of the elevational gradient) (Fig. 2). The
northernmost transect is located in Yosemite National Park, and the southernmost is in
Mountain Home State Forest. The two middle transects are located in Sequoia and Kings
Canyon National Parks. The transects span a broad range of elevations, aspects, slopes,
and forest types. In general, the transects are located in the middle elevations of the west
slope, extending from about 1090 to 2680 m. The transects range in length from 7 to 15
km.
The topography along the length of the transects is relatively steep (up to 100%
slopes), although flat areas are also present within and between many of the collection
sites. The aspects of the sites are primarily towards the west and south, although a few
sites have northern or eastern aspects. The transects are situated on continuous
topographic features, such as along ridge lines, or along the mid-portions of slopes.
These topographic features extend along the elevational gradient without major fire
barriers (e.g., continuous rock escarpments, cliffs, or large rivers) between the collection
sites, except for the Giant Forest transect, which is bisected by a major river canyon
(Marble Fork of the Kaweah River). Because the topography and fuels (at present) are
more-or-less continuous between collection sites, fire spread between most sites along the
transects is possible. Fire spread between the three southern transects is possible but
unlikely because of distances (i.e., up to 30 km) and major barriers such as deep river
canyons and large, continuous rock exposures.
At the lowest end of the transects the dominant over story trees include ponderosa
pine (Pinus ponderosa) and black oak (Quercus kelloggii). Some of these sites are
located on the ecotone of pine/oak forest and foothills chaparral types (see Nates’??
chapter for discussion of Sierra Nevada vegetation types). In the middle portions of the
transects white fir (Abies concolor), ponderosa pine, jeffrey pine (Pinus jeffreyii), and
giant sequoia (Sequoiadendron giganteum) are dominant. This type is commonly
referred to as mixed conifer (Rundel et al. 1977). At the highest end of the transects
white fir, red fir (Abies magnifica), and jeffrey pine are dominant, with western white
pine (Pinus monticola) and lodgepole (Pinus contorta) as occasional dominant trees in
some sites.
The fire season extends from April to November in the southern and central
Sierra, with a maximum area burned in June, and a maximum in numbers of lightning
caused fires in July (ref to Jan’s paper?).
6.3 Methods
6.31 Site and Sample Selection
Our previous fire history research in giant sequoia groves (Swetnam 1993)
provided a database of fire regime reconstructions for the sequoia-mixed conifer portion
of the elevational transects. In our search for suitable collection sites we followed the
major topographical features (ridges and slopes) up and down from a set of four groves:
Mariposa, Big Stump, Giant Forest, and Mountain Home . The transects follow
elevational gradients, with the sequoia groves in the middle or near the upper end of the
transects. We searched for forest stands that were relatively free of major canopy
disturbance, such as complete overstory removal from past logging or recent crown fires
that would have consumed the fire-scar record preserved in living trees, stumps, logs, and
snags. We also searched for forest stands that appeared relatively homogenous
(composition and density) and were not dissected by potential fire barriers (e.g.,
continuous rock outcrops, rivers, etc.).
The selected stands ranged in size from approximately 20 to 50 ha. These areas
were defined by topographic features such as ridge tops and drainage bottoms, or obvious
changes in the overstory structure (e.g., a change from forest to open meadow). We
searched systematically within the stands for fire-scarred trees. Our aim was to
thoroughly search the entire area of the stand and to identify a set of broadly distributed
specimens that contained a well preserved and long historical record of past surface fires.
We examined all potential fire-scarred specimens that we could locate within the selected
stands. This searching and examination involved careful inspection of the fire-scarred
surfaces of all potential specimens, and identification of those with well preserved wood,
datable tree-rings, and maximum numbers of visible scars.
Most of the collected specimens were full cross sections taken with a chainsaw
from stumps, logs, and snags. A smaller number of partial sections were taken from
living trees (Arno and Sneck 1977, Baisan and Swetnam 1990) to extend the fire history
record to the present. A goal of our sampling strategy was to maximize the temporal
completeness and length of the fire event record for each stand. This was not a statistical
sampling of the “population” of fire events (or fire intervals), rather, our goal was to
obtain an inventory of fire events that occurred within the stands that was as complete
and as long as possible, given the fragmentary nature of the record and practical
limitations in recovering it.
6.32 Laboratory Sample Processing and Analyses
Dry cross sections were re-sectioned with a band saw to prepare a smooth surface
for sanding. A belt sander was used with progressively finer grits from 150 to 400. All
tree-rings were crossdated using variations in ring-width, latewood widths, and false ring
patterns to accurately assign calendar years to each growth ring (Stokes and Smiley 1968,
Swetnam et al. 1985). Calendar years and seasonal timing were determined for each fire
scar by observing under magnification (10X to 30X) the relative position of the scars
within the dated annual rings (Dieterich and Swetnam 1984, Baisan and Swetnam 1990,
Ortloff 1996).
Scar position was divided into six categories, dormant (D), early-earlywood (E),
middle-earlywood (M), late-earlywood (L), latewood (A), and undetermined (U). The
early-earlywood, middle-earlywood, and late-earlywood correspond approximately to a
division of the earlywood by thirds (Baisan and Swetnam 1990). Based on our
knowledge of cambial phenology of conifers (Fritts 1976, Baisan and Swetnam
unpublished data, Parsons unpublished data), we estimated the approximate calendrical
dates for each of the scars (except U positions) to within 4 to 6 week periods. The
undetermined (U) category were scars with too much decay, resin, or suppressed (slow)
ring growth in the area of the scar for reliable determination of intra-ring position.
6.33 Compositing Fire Chronologies and Assessing Patterns of Synchrony
The master fire chronology chart is a primary graphical tool that fire historians
use to assess spatial and temporal patterns of fire history. Assembled at the stand level
(Fig. 3) these charts are useful for identifying unusual long or short intervals in fire
occurrence, changes in sample size and replication, and major shifts in fire occurrence.
Software has been developed by Grissino-Mayer (submitted) that produces these graphic
from standardized data input files. This software (FHX2) also performs statistical
description and hypothesis testing of differences in fire interval distributions (between
chronologies or time periods). The composite fire chronology is the summation of fire
events among a set of individual tree records. Composites can also be assembled from
composites at finer scales. In this case we use a composite of fire chronologies from
one of our elevation transects (Giant Forest) to illustrate systematic changes with
elevation, and historical patterns of change in fire regimes across the entire gradient.
Composite fire chronologies can also be assembled from a sub-set of fire dates
that are in common among trees within stands, or sites from within regions. For example,
we have found that fire dates that are common to more than about 10 % of all trees within
stands, or more than about 10% of all sites within a region, tend to be more highly
correlated with climatic patterns than all fire dates recorded by fire scarred trees, or all
fire dates recorded within sites (Baisan and Swetnam 1990, Swetnam 1993). This is
probably due to a stronger relation between inter-annual climate variations and
widespread (extensive and synchronous) fires than numbers of fires ignited. This is also
a general pattern observed in studies of climate correlations with 20th century databases
of numbers of fires ignited and areas burned each season. Area burned time series tend
to correlate more highly with climate variables than time series of numbers of fires
ignited (Swetnam 1990). This is probably due to the fact that many fires are ignited
during some years and seasons because of increased lightning or human-caused fires, but
the ability of fires to spread over large areas is more closely coupled to weather and
climate factors than the rate of fire ignitions.
6.34 Superposed Epoch Analyses, and Other Climate-Fire Assessments
We use the synchronous patterns of fires in giant sequoia groves over the past
2,000 years to illustrate the role of changing temperature and precipitation in controlling
fire regime variability.
Interannual climate-fire relations were evaluated by conducting superposed epoch
analyses (SEA) (Baisan and Swetnam 1990, Swetnam 1993, Veblen et al. 1999). The first
step in this analysis was identifying a set of extreme regional fire occurrence years to
compare with the tree-ring drought reconstructions. To do this we plotted the time series
of numbers of trees and sites recording fires each year from 1700 to 1900. Since there
was a century-scale trend in these series that was related to numbers of fire-scarred trees
sampled through time (due to decay processes), we identified thresholds above and below
certain values (i.e., numbers of trees or sites reecording fires) for identifying the extreme
years. This will be explained more thoroughly in the results section.
We used the extreme low and high fire occurrence years in the SEA. The SEA
computer program (Holmes and Swetnam, unpublished) computed the mean values from
the reconstructions (i.e., PDSI, precipitation, temperature) during the largest and smallest
fire occurrence years (year t). The program also computed the mean climate values for
years t-1 to t-5 (preceding years) and years t+1 and t+2 (subsequent years) for all large
and small fire events. A Monte Carlo simulation (i.e., re-sampling) was used to estimate
confidence intervals around the observed mean values (Mooney and Duvall 1993,
Swetnam and Betancourt 1992, Swetnam 1993, Veblen et al. 1999).
6.4 Fire Frequency and Synchrony Patterns at the Stand and Watershed Scales: The fire
Spread process, and Land Use History
Figs 2, 3, and 4
6.5 Fire Frequency and Synchrony Patterns at the Regional Scale: The Influence of
Climate
Figs 5, 6, 7, 8
6.6 Summary and Conclusions
Figure 1. Map of the western slope of the Sierra Nevada and locations of fire history
transects and giant sequoia groves where fire histories have been compiled.
Figure 2. Examples of a fire scarred giant sequoia in Mariposa Grove (upper left),
sampling a fallen sequoia log in the Giant Forest with chain saw (lower right), fire scars
within sequoia tree rings from the Giant Forest (lower left) and in a ponderosa pine cross
section from Yosemite National park (lower right). [these could be in black & white]
Figure 3. Examples of stand-level fire chronologies from a site on the Giant Forest
transect at xxxxm (upper chart), and on the Mountain Home transect at xxxx m (lower
chart).
Figure 4. A composite fire chronology chart from the Giant Forest transect. The
horizontal lines are the stand-level composites from sites along the transect. The upper 3
sites (LM, CM, and HM) are from groups of fire scarred sequoia trees within the Giant
Forest.
Figure 5. Fire frequency and synchrony in five giant sequoia groves in the Sierra
Nevada, California. Fire frequency is summarized as number of fire events in nonoverlapping century-length periods for each grove (A), and for all groves combined
during moving 50 and 20 year periods, plotted on the central year (B). Changing patterns
of synchrony are show by the events when fires occurred in 3, 4 and 5 of the groves
during the same years (C).
Figure 6. Comparison of tree-ring reconstructed winter-spring precipitation for
California division 5 precipitation (Graybill 2000) with the fire years recorded in four
(filled circles) and five (filled inverted triangles) giant sequoia groves.
Figure 7. Superposed epoch analyses of fire events in five sequoia groves compared to
reconstructed winter-spring precipitation (see figure 6) during the period AD 1050 to
1850. The sets of years with no fires in any grove (0), and synchronous fire events in
multiple groves (1, 2, 3, 4 or 5) were tested and plotted separately.
Figure 8. Comparison of smoothed fire frequency and temperature estimates from AD
500 to the late 20th century. The fire frequency was the summed number of fire events in
20-year non-overlapping periods in all five groves. The short dashed line is summer
temperature (smoothed 20 year non-overlapped means) reconstructed from Sierra Nevada
foxtail pine tree-ring widths (Graumlich 199x) and the long dashed line is ring widths
(smoothed 20 year non-overlapped means) from upper forest border bristlecone pine
from Campito Mountain, White Mountains, CA (LaMarche and Harlan 197x).