186
Ecology, 84(1), 2003, pp. 186–201
q 2003 by the Ecological Society of America
HOLOCENE FIRE HISTORY OF A COASTAL TEMPERATE RAIN FOREST
BASED ON SOIL CHARCOAL RADIOCARBON DATES
DANIEL G. GAVIN,
1,3
LINDA B. BRUBAKER,
1
AND KENNETH P. L ERTZMAN
2
1
College of Forest Resources, Box 352100, University of Washington, Seattle, Washington 98195-2100 USA
2
School of Resource and Environmental Management, Simon Fraser University,
Burnaby, British Columbia V5A 1S6 Canada
Abstract. The long-term role of fire in coastal temperate rain forest is poorly under-
stood. To determine the historical role of fire on western Vancouver Island (British Co-
lumbia, Canada), we constructed a long-term spatially explicit fire history and examined
the spatial and temporal distribution of fire during the Holocene. Two fire-history parameters
(time-since-fire [TSF] and fire extent) were related to three landscape parameters (landform
[hill slope or terrace], aspect, and forest composition) at 83 sites in a 730-ha low-elevation
(less than ;200 m) area of a mountainous watershed. We dated fires using tree rings (18
sites) and 120 soil-charcoal radiocarbon dates (65 sites). Comparisons among multiple
radiocarbon dates indicated a high probability that the charcoal dated at each site represented
the most recent fire, though we expect greater error in TSF estimates at sites where charcoal
was very old (.6000 yr) and was restricted to mineral soil horizons. TSF estimates ranged
from 64 to ;12 220 yr; 45% of the sites have burned in the last 1000 yr, whereas 20% of
the sites have not burned for over 6000 yr. Differences in median TSF were more significant
between landform types or across aspects than among forest types. Median TSF was sig-
nificantly greater on terraces (4410 yr) than on hill slopes (740 yr). On hill slopes, all
south-facing and southwest-facing sites have burned within the last 1000 yr compared to
only 27% of north- and east-facing sites burning over the same period. Comparison of fire
dates among neighboring sites indicated that fires rarely extended .250 m. During the late
Holocene, landform controls have been strong, resulting in the bias of fires to south-facing
hillslopes and thus allowing late-successional forest structure to persist for thousands of
years in a large portion of the watershed. In contrast, the early Holocene regional climate
and forest composition likely resulted in larger landscape fires that were not strongly
controlled by landform factors. The millennial-scale TSF detected in this study supports
the distinction of coastal temperate rain forest as being under a fundamentally different
disturbance regime than other Pacific Northwest forests to the east and south.
Key words: British Columbia, Canada; climate–terrain–fire interaction; climate change; coastal
temperate rain forest; disturbance, long-term role; fire history; Holocene fire history; Pacific Northwest
forests; paleoecology; radiocarbon dating; soil charcoal.
INTRODUCTION
The coastal temperate rain forest from southern Brit-
ish Columbia, Canada, north through southeast Alaska,
USA, is noted for the near-absence of recent fire and
the dominance of late-successional forest communities
that experience a disturbance regime of small-scale
tree-fall gaps (Veblen and Alaback 1996, Lertzman et
al. 1996, Wells et al. 1998). These forests are also noted
for the rarity of Douglas-fir (Pseudotsuga menziesii
(Mirb.) Franco), a long-lived seral species that requires
fire for establishment in much of the Pacific Northwest
(Munger 1940, Franklin and Dyrness 1988, Agee
1993). In the windward portions of coastal mountain
ranges in British Columbia, evidence of fire is limited
to the scattered occurrence of Douglas-fir on specific
Manuscript received 31 August 2001; revised 4 April 2002;
accepted 18 May 2002. Corresponding Editor: S. T. Jackson.
3
Present address: Department of Plant Biology, 265 Mor-
rill Hall, University of Illinois, Urbana, Illinois 61801 USA.
E-mail: dgavin@life.uiuc.edu
topographic features such as south-facing hill slopes
(Schmidt 1960, 1970, Veblen and Alaback 1996). This
pattern contrasts with nearby regions, where the pre-
settlement landscape was dominated by forests con-
taining evidence of large fires in the form of even-aged
stands of Douglas-fir (Hemstrom and Franklin 1982).
Thus, the rarity of Douglas-fir and lack of stand-struc-
tural evidence of fire distinguishes coastal temperate
rain forest in the regional context of Pacific Northwest
forests. However, other than these types of observations
about modern forest condition, there is virtually no
documentation of fire history in forests in this region,
and it is not known whether the modern fire regime is
a long-term feature (e.g., .1000 yr) of the coastal tem-
perate rain forest landscape.
Factors affecting fire regimes operate at different
temporal and spatial scales. At broad spatial scales, fire
regimes are influenced by long- and short-term climatic
changes and at more local scales by topographic fea-
tures and vegetation types (e.g., Romme and Knight

January 2003 187FIRE HISTORY OF THE TEMPERATE RAIN FOREST
FIG. 1. Location of the Clayoquot Valley (CV) on Vancouver Island, British Columbia, Canada, showing the extent of
the coastal temperate rain forest (darker areas), defined here as the wettest subzones of the Coastal Western Hemlock
Biogeoclimatic Zone (Very Wet Hypermaritime, Very Wet Maritime–low elevation, and Very Wet Maritime–mid elevation;
Meidinger and Pojar 1991).
1981, Lertzman and Fall 1998, Heyerdahl et al. 2001,
2002). The interaction of large- and small-scale con-
trols causes both the locations and sizes of fires to vary
over time by affecting the distribution of fire-suscep-
tible forests (Turner and Romme 1994). For example,
during generally cool and moist climatic periods, fires
may be restricted to a small portion of the landscape
(e.g., south-facing hill slopes), but during warm and
dry intervals, fires may burn a large portion of the
landscape. If climate undergoes major changes in the
Pacific Northwest, larger areas of coastal temperate rain
forest may become more susceptible to fire. This has
been shown in other lowland regions of the Pacific
Northwest, where several studies have found that large-
scale climate change has greatly altered the fire regime
through the Holocene (Cwynar 1987, Long et al. 1998,
Brown and Hebda 2002). Our goal in this study is to
examine the long-term roles of climate, topography,
and vegetation in the fire regime of a coastal temperate
rain forest that receives some of the highest annual
precipitation in North America.
Fire-history studies must overcome significant chal-
lenges in areas with very long fire intervals. Traditional
fire-history methods focus on particular spatial and
temporal scales that might not be well suited to study
the landscape pattern of past fires in coastal temperate
rain forest (Lertzman and Fall 1998). For example, the
use of tree-ring analysis to understand the pattern of
fire at fine temporal and spatial scales has limited ap-
plication in humid regions, where the time-since-the-
last-fire (TSF) may exceed the ages of trees. Over lon-
ger time scales, fire occurrence may be addressed with
charcoal records in lake sediments. However, this meth-
od is not spatially explicit, because lake-sediment rec-
ords integrate fires within large source areas (Clark
1988). In this study we overcome the spatial and tem-
poral constraints of traditional fire-history studies by
using radiocarbon dates of soil charcoal to describe the
spatial pattern of fire over long time scales (see also
Niklasson and Granstrom 2000). Although charcoal is
inert and should be preserved in stable soils, this meth-
od has rarely been used, and questions remain regard-
ing the taphonomy of charcoal and the temporal res-
olution available from soil-charcoal records. Our over-
all objectives are to examine (1) the utility of radio-
carbon-dated soil-charcoal records for fire history, (2)
the length of time since the last fire in relation to to-
pographic features and forest types, and (3) the tem-
poral pattern of fire over the last ;11 000 years.
STUDY AREA
The research was conducted in the Clayoquot River
watershed, on the western edge of the Vancouver Island
Range and 20 km from the west coast of Vancouver
Island, British Columbia, Canada (498159 N; 1258309
W; Fig. 1 and Plate 1). The 7700-ha, 12-km-long wa-
tershed ranges in elevation from 15 to ;1200 m. High-
elevation tundra and rock covers 16% of the watershed.
Bedrock is of volcanic and sedimentary origin intruded
by numerous granitic batholiths (Muller 1968). The
watershed contains multiple terraces 5–40 m above the
river that abut steep valley walls with slopes of 40%
to .60%. The active floodplain is limited to the lower
4 km of river, and is constricted at several locations by
colluvial fans from tributary valleys. Soils are shallow
(,1 m) or absent on slopes above alluvial fans and
terraces. Soils are mainly Spodosols (Humic Haplor-
thods) (Jungen 1985).
The study area was restricted to low-elevation areas
(less than ;200 m) within the watershed. In the bio-
geoclimatic ecosystem classification, this area is in the
submontane variant of the Very Wet Maritime subzone
of the Coastal Western Hemlock zone (CWHvm1), de-
scribed as having a wet, humid climate with cool sum-