Ecology, 89(3), 2008, pp. 705–716
 2008 by the Ecological Society of America
MULTI-SEASON CLIMATE SYNCHRONIZED HISTORICAL FIRES IN DRY
FORESTS (1650–1900), NORTHERN ROCKIES, USA
EMILY K. HEYERDAHL,
1,3
PENELOPE MORGAN,
2
AND JAMES P. RISER II
1
1
USDA Forest Service, Rocky Mountain Research Station, 5775 US West Highway 10, Missoula, Montana 59808 USA
2
Department of Forest Resources, University of Idaho, P.O. Box 441133, Moscow, Idaho 83844-1133 USA
Abstract. Our objective was to infer the climate drivers of regionally synchronous fire
years in dry forests of the U.S. northern Rockies in Idaho and western Montana. During our
analysis period (1650–1900), we reconstructed fires from 9245 fire scars on 576 trees (mostly
ponderosa pine, Pinus ponderosa P. & C. Lawson) at 21 sites and compared them to existing
tree-ring reconstructions of climate (temperature and the Palmer Drought Severity Index
[PDSI]) and large-scale climate patterns that affect modern spring climate in this region (El
Nin˜o–Southern Oscillation [ENSO] and the Pacific Decadal Oscillation [PDO]). We identified
32 regional-fire years as those with five or more sites with fire. Fires were remarkably
widespread during such years, including one year (1748) in which fires were recorded at 10 sites
across what are today seven national forests plus one site on state land. During regional-fire
years, spring–summers were significantly warm and summers were significantly warm-dry
whereas the opposite conditions prevailed during the 99 years when no fires were recorded at
any of our sites (no-fire years). Climate in prior years was not significantly associated with
regional- or no-fire years. Years when fire was recorded at only a few of our sites occurred
under a broad range of climate conditions, highlighting the fact that the regional climate
drivers of fire are most evident when fires are synchronized across a large area. No-fire years
tended to occur during La Nin˜a years, which tend to have anomalously deep snowpacks in this
region. However, ENSO was not a significant driver of regional-fire years, consistent with the
greater influence of La Nin˜a than El Nin˜o conditions on the spring climate of this region. PDO
was not a significant driver of past fire, despite being a strong driver of modern spring climate
and modern regional-fire years in the northern Rockies.
Key words: dendrochronology; El Nin˜o–Southern Oscillation; fire history; fire scars; Idaho; Montana;
Pacific Decadal Oscillation; Palmer Drought Severity Index; spring; summer; temperature.
INTRODUCTION
Annual extremes in fire-season climate synchronized
the occurrence of fires across broad regions of western
North America for at least several centuries in the past
(e.g., Swetnam and Betancourt 1998, Veblen et al. 1999,
Heyerdahl and Alvarado 2003, Swetnam and Baisan
2003, Kitzberger et al. 2007) and continued to do so
throughout the 20th century (e.g., Westerling et al. 2003,
2006, Collins et al. 2006, Morgan et al. 2008). Although
the importance of summer drought in driving regional-
fire years has long been recognized, the potential role of
warm springs in synchronizing regional fire in the
northern United States has been suggested for historical
and modern fires but only recently confirmed for the
20th century (Balling et al. 1992, Heyerdahl et al. 2002,
Hessl et al. 2004, Westerling et al. 2006, Morgan et al.
2008). Late 20th-century trends toward warm, early
springs have been associated with increased forest fire
activity and further increases in temperatures projected
under plausible climatic change scenarios imply a
greater frequency of regional-fire years in the 21st
century (Running 2006, Westerling et al. 2006). Assess-
ing the potential impact of climatic change on forest fire
activity in the western United States requires that we
understand the climate drivers of regionally synchro-
nous fires before any recent changes in climate.
The U.S. northern Rockies in Idaho and western
Montana are a critical region for understanding both
past and present climate drivers of fire. Repeatedly
during the 20th century, this region experienced
regionally extensive fires that led to changes in national
fire policy (Pyne et al. 1996). The late 20th-century
increase in large forest fires was greater in the northern
Rockies than other regions of the western United States
(Westerling et al. 2006), a region where fire and climate-
driven changes in fire have had significant implications
for vegetation composition and structure (Whitlock et
al. 2003). However, this region lacks a network of
annually accurate histories of pre-20th-century fire from
which we could identify past climate drivers, despite
abundant fire-scar evidence in dry forests, i.e., those
dominated or co-dominated by ponderosa pine.
In the northern Rockies, spring and summer climate
have varied through time (Briffa et al. 1992, Cook et al.
2004) and warm-dry summers have been inferred as
Manuscript received 11 December 2006; revised 19 June
2007; accepted 3 July 2007. Corresponding Editor: A. H. Lloyd.
3
E-mail: eheyerdahl@fs.fed.us
705

significant drivers of historical fires in some subalpine
forests of the region (Kipfmueller 2003). The fire season
in dry forests here begins after annual snowpacks have
melted in the late spring to early summer (Serreze et al.
1999). Despite variation in local climate and topogra-
phy, empirical and modeling studies have demonstrated
that relatively warm springs historically resulted in
relatively earlier melting of snowpacks at sites spanning
a broad range of elevations across the region (Hamlet et
al. 2005, Mote 2006, Westerling et al. 2006). These
relatively warm springs are partly driven by large-scale
climate patterns. Following El Nin˜o events, winters and
springs in this region tend to be anomalously dry and
warm which results in anomalously shallow snowpacks
while La Nin˜a events lead to anomalously deep
snowpacks (Redmond and Koch 1991, Gershunov et
al. 1999, Harshberger et al. 2002, Mantua 2002, McCabe
and Dettinger 2002). PDO affects spring climate in the
northern Rockies in a manner similar to ENSO but
varies on a longer time scale (Mantua et al. 1997,
McCabe and Dettinger 2002). PDO was an important
driver of 20th-century fires in the northern Rockies
(Morgan et al. 2008) and elsewhere in the inland
Northwest (Westerling and Swetnam 2003, Gedalof et
al. 2005, Collins et al. 2006). Interannual variations in
PDO, sometimes interacting with ENSO, account for
roughly a third of the variance in modern spring
snowpack in the region (Gershunov and Barnett 1998,
Harshberger et al. 2002, McCabe and Dettinger 2002).
Interactions of ENSO and PDO elsewhere in the interior
West are associated with historical fire activity (Schoen-
nagel et al. 2005, Sibold and Veblen 2006, Kitzberger et
al. 2007). It is likely that annually and decadally varying
spring and summer climate and large-scale climate
patterns synchronized past fire across the northern
Rockies as well. However, the climate drivers of past
fire in this region have not yet been examined across
broad spatial scales using fire records with annual
accuracy.
Our objective was to infer the climate drivers of past
regionally synchronous fires in dry forests (dominated or
co-dominated by ponderosa pine) in the northern
Rockies. We reconstructed past fires from fire scars at
21 sites in Idaho and Montana west of the Continental
Divide, and compared them to existing tree-ring
reconstructions of climate (temperature and PDSI) and
to large-scale climate patterns that affect modern spring
climate in this region (ENSO and PDO).
STUDY AREA
Modern instrumental climate
The climate of the study area is continental, with cold
winters and warm summers. All but one of our sites
(FLA; Table 1) lie within two climate divisions (Idaho
division 4 or Montana division 1; data available online).
4
Mean January temperature in these divisions is58 and
68C and mean July temperature is 188 and 198C,
respectively (1895–2005). Mean annual precipitation is
low, more so in Montana than Idaho (49 and 68 cm,
respectively, 1895–2005) and much of this falls as snow in
winter (62%, 1963–1996; Serreze et al. 1999). Snow-water
equivalent at elevations averaging 1905 m (range 960–
2790 m) in the study area generally peaks in mid-April
TABLE 1. Dry-forest sampling sites and evidence of fire collected at each (1650–1900).
Site
no.
Site
code
No.
trees
Area
(ha)
Elevation
(m)
Latitude
(N)
Longitude
(W)
No. fire
scars
No. years
with
2 scars
Scar date
First Last
1 VIZ 46 13 760 48858
0
15
00
11587
0
33
00
1450 38 1652 1882
2 HUN 25 31 750 48853
0
27
00
115813
0
46
00
308 17 1652 1869
3 LIB 29 36 700 48824
0
58
00
115832
0
16
00
544 36 1661 1889
4 MCM 32 8 830 48818
0
42
00
115829
0
11
00
513 20 1661 1889
5 CRN 18 18 1050 47853
0
46
00
11480
0
58
00
216 16 1655 1892
6 COR 26 3 1190 47836
0
8
00
114855
0
51
00
439 24 1652 1889
7 HOL 19 38 1220 47826
0
21
00
113838
0
48
00
224 16 1671 1882
8 FAV 17 148 1130 4789
0
19
00
114829
0
21
00
259 17 1687 1895
9 BUT 38 2 1400 4787
0
21
00
14823
0
36
00
803 25 1675 1870
10 LUB* 20 70 1270 46853
0
26
00
113827
0
17
00
139 20 1652 1861
11 BMT 36 33 1141 46849
0
27
00
11487
0
35
00
770 40 1665 1889
12 SML 32 53 1520 46827
0
4
00
113853
0
40
00
231 16 1670 1889
13 SHE 41 14 1350 46820
0
16
00
114814
0
25
00
523 22 1665 1898
14 FLA 25 28 950 46850
0
43
00
116853
0
54
00
569 40 1660 1895
15 KEA 22 4 1530 45828
0
1
00
11688
0
58
00
227 17 1671 1889
16 COV 25 12 1670 45825
0
7
00
115831
0
1
00
250 14 1659 1889
17 FRI 31 47 1500 45832
0
54
00
113856
0
33
00
379 20 1656 1889
18 POV 33 18 1270 44849
0
13
00
115842
0
7
00
303 12 1743 1889
19 WSH 25 31 1030 4482
0
41
00
115855
0
27
00
508 21 1652 1863
20 LOW 9 45 1340 4483
0
50
00
115836
0
56
00
81 9 1656 1888
21 WSR 27 15 1530 43849
0
33
00
115853
0
47
00
509 25 1652 1878
Total 576 9245
Note: Number of fire scars includes eroded scars (5% of total).
* Samples from LUB were collected for another study (L. A. Jones, personal communication).
4
hwww7.ncdc.noaa.gov/CDO/CDODivisionalSelect.jspi
EMILY K. HEYERDAHL ET AL.706 Ecology, Vol. 89, No. 3