Journalof Ecology(1990), 78, 403-412
THE INFLUENCE OF MAN AND CLIMATE ON
FREQUENCY OF FIRE IN THE INTERIOR WET BELT
FOREST, BRITISH COLUMBIA
E. A. JOHNSON, G. I. FRYER
AND
M. J. HEATHCOTT
of Calgary,
DivisionofEcology,Departmentof BiologicalSciences,University
Calgary,Alberta,Canada T2N IN4
SUMMARY
have been studiedusinghistorical
ofman and climateon firefrequency
(1) The effects
model in GlacierNational Park in BritishColumbia,Canada.
data and a fire-history
a changeinfirecyclein the1760s,whichcould be
(2) GlacierNationalParkexperienced
relatedto theoccurrenceoftheLittleIce Age. Before1760,thefirecyclewas 80 yearsand
after1760itwas 110years.The longerfirecycleafter1760was clearlyrelatedto thecooler,
moisterclimatewhichalso resultedin theadvance of glaciersat thattimein thePark.
of theCanadian PacificRailroad throughthePark corresponded
(3) The construction
to a periodin whichthearea burntby man increased(1883, 1885 and 1886) and many
lightningfiresoccurred.There has not been a decreasein the firefrequencysince the
policy.
establishment
of GlacierNational Park in 1888,despitea fire-suppression
are associatedwith
rapidlyspreadingfirescaused bylightning
(4) Large,high-intensity,
a characteristic
synopticweatherpatternwhichconsistsoftwoparts:(i) intensefueldrying
associatedwitha stationary
high-pressure
systemblockingaccess ofmoistPacificair,and
systeminto low-pressure
(ii) one or more periods of breakdownof the high-pressure
and highwinds.
systems,whichcreatelightning
man must
of firefrequency,
(5) To have supplantedclimateas a major determinant
eitherhave used regularmanagementburningor suppressedfiresduringcriticalweather
periods.Because thishas notoccurredin GlacierNationalPark,thepresentfireregimeis
largelynatural.
INTRODUCTION
Regional climate controlsthe frequencyof firein coniferousforestecosystemsby
oflightning.
climatealso
thefuelmoisturecontentand theprobability
Indirectly
affecting
thefueltypeand therateofdecomposition.AfterEuropeansoccupiedNorth
determines
America,it has generallybeen thoughtthat man supplantedclimateas the primary
determinantof firefrequency.It is commonlybelieved that duringearly European
and development,the frequencyof firesincreased(Nelson &
exploration,settlement
policies,the
Byrne1966;Tande 1979),and in recentdecadeswithincreasingconservation
frequencyof firehas decreasedthroughfiresuppression.Althoughthisinterpretation
to produceevidenceforparticularareas which
seemslogical,it has oftenbeen difficult
methodsofHeinselman(1973) and
supportsit.However,thefire-history
unambiguously
ofbothclimateand
theeffects
Johnson& Van Wagner(1985) can be used to investigate
This methodquantifies
thetemporaland spatialfirefrequency
manon thefirefrequency.
of an area and can establishthe causes of temporaland landscape changes in fire
frequency.
In coniferousforestecosystemsa firehistorycan be producedbymappingstandswith
map givestheage mosiacoftheforestsinceitlast
differing
ages oforigin.The stand-origin
burnedas determinedfromfirescars,tree-ring
growthrelease,total treeages and fire
distribution
is plottedas a negativeexponentialsurvivorship
records.The firefrequency
403
Man, climateandfirefrequency
404
curve,called the time-since-fire
distributionby Johnson& Van Wagner (1985). The
is definedbythefirecyclewhichis thetimetakento burnan area equal to the
distribution
studyarea. Consequently,
duringeach firecyclesomeareasmayburnmorethanonceand
some not at all.
The constructionof the Canadian Pacific Railroad throughRogers Pass in the
ColumbiaMountainsofBritishColumbiaduringthe1880sdefinesan area ofknownage
in whichthe impactof European land use on firefrequencymay be studied.Several
hypothesesmaybe tested:beforetheconstruction
of therailroadthefirecyclewas long,
thefirecyclewas shorter,
duringrailroadconstruction
and aftertheestablishment
ofthe
area as Glacier National Park in 1988,the firecyclewas even longerthan it had been
beforeEuropean occupation.
STUDY AREA
Glacier National Park (1350 km2total,600 km2of forestedland) is in the Columbia
Mountainsof south-easternBritishColumbia (Fig. 1). The Beaver Valley in Glacier
National Park separatesthePurcellMountainsto theeast and theSelkirkMountainsto
thewest.Exceptfortherailroadand highwayin theRogersPass Corridor,therestofthe
Park is accessibleonlyby primitive
trails.
The climatehas a wetand dryseason. In thewetseason,fireis rareand itis in thedry
seasonthatfiresoccur.The wetseason extendsfromOctoberto March(Fig. 2) and is the
resultofa largenumberoflow-pressure
systems
movingeastwardfromthePacificOcean.
These low-pressure
systemsare morefrequentduringthewinterbecause of a stationary
low offthecoast of BritishColumbia. In summer,thisstationarylow movesnorthward
and thestormsare deflectednorthof theColumbia Mountains.This leads to relatively
weatherfromJuneto August(Fig. 2).
long periodsof warm-dry
FireoccurrenceinGlacierNationalParkstartsinApriland endsin September(Fig. 3).
The railroadis themajor sourceof man-madefires(M. J. Heathcott& E. A. Johnson,
unpublishedreportto ParksCanada); thesepeak in May and Junewhilelightning-caused
firespeak in Julyand August.Lightningfiresbetween1950and 1985accountedfor70%
e~~~~
ountainCree
S~~~~~oe
Glacier
National
<z,
-. *
-
Pork
2,(
D
M~~~~
Roes
Pa
V
k xloetre
01
0
kilometres
FIG.
1. The distribution
ofpoint-samples
usedin determining
firedatesin GlacierNationalPark,
BritishColumbia.
E. A. JOHNSON,G. I. FRYER AND M. J. HEATHCOTT
405
-20
400
0~~~~~~~~~~1
T~~~~~~~
E
T
1200
0
II
I
JFMAMJJ A SO0ND
Month
0
-201 J F MAM J J AS 0N D
Month
forGlacierNationalPark from1950
and temperature
FIG. 2. The averagemonthlyprecipitation
to 1980(M. J. Heathcott& E. A. Johnson,unpublishedreportto Parks Canada).
140 20 100
80
E
z
60-
cn~~~~~~ot
4020F M
A M
J
J A
Month
SO0N
D
oflightningdistribution
(0) and railroad-causedfires(0) and (A) all firesin
FIG. 3. The monthly
GlacierNationalPark from1950to 1985(M. J.Heathcott& E. A. Johnson,unpublishedreport
to ParksCanada).
of the firesin Glacier National Park and 72% in the BritishColumbia Nelson Forest
accountedformorethan
Regionwestand southofGlacierPark.In bothcases,lightning
70% of thearea burned.
The InteriorWetBeltForests(Achuffetal. 1984)at loweraltitudescontaintreeswhich
such as Thujaplicata D. Don (westernred cedar),
a coastal distribution
have primarily
Tsuga heterophylla(Raf.) Sarg. (westernhemlock),Pseudotsugamenziesii(Mirbel)
Franco (Douglas fir),and PinusmonticolaDougl. (westernwhitepine).Thesetreesreach
diametersofover 1 m,heightsofmorethan50 m and a forestbiomassofmorethan1000
406
Man, climateandfirefrequency
tonnesha-'. The higher-altitude
forests(above c. 1800 m a.s.l.) are made up of Picea
engelmanni
Parry(Engelmannspruce)and Abieslasiocarpa(Hook.) Nutt.(subalpinefir).
All namesfollowMoss (1959).
METHODS
in GlacierNational Park was estimatedfroma stand-origin
Fire frequency
map (sensu
Heinselman1973),whichis notreproducedherebecauseofitslargesize.The stand-origin
map was constructedby (i) usingaerial photographsto identifyareas whichpossibly
in age, (ii) fieldsamplingto gatherdata on fireages and to delineatethe fire
differed
boundaries,and (iii) compilationof the aerial-photographinterpretations
and field
samplingon to a 1: 50 000 scale stand-origin
map.
Black-and-white
aerial photographsof 1:40000 scale, takenin 1947 and 1978,were
used forinitiallocationofpast fireboundaries.On theaerialphotographs,
in
differences
treeheightwereused to locatefireboundaries,althoughin olderforeststoneand texture
weremoreimportant.The smallscale and oftenpoor qualityof thephotographsmade
interpretation
difficult.
The boundariesofages are inexactifbased onlyon aerialphotographs.Consequently,
more than 500 point-samples(sensuJohnson& Van Wagner 1985) were distributed
thePark to improvetheidentification
throughout
offireboundariesand to date thefires
10ha), thedate ofthelastfirewas obtained
(Fig. 1). In each point-sample(approximately
fromany available reliableevidence.This evidenceconsistedof anycombinationof fire
scars,maximumtreeages,growthreleasein treeswhichsurvivedthefire,dendrochronologicallycross-datedfire-killed
trees,firereportsand historicalaccounts.Attemptswere
made to replicateevidenceof firein and betweenpoint-samples(McBride 1983).
Whole discs fromtreeswereused whenpossible to date firesto ensurethatlocally
missingringswerediscovered.Incrementcoresand wedgeswereused whenwholediscs
werenot available. Both redcedar and westernhemlockwereoftenrottenin thecentre
and rarely dateable. All discs, wedges and cores were prepared by mechanical
sandpaperingand finishedwithfinesandpaperor a razor blade. All sampleswerering
countedat least fourtimesand a fewwerecross-datedagainsta masterchronology(cf.
Fritts1976). The dates of fireson thestand-origin
map wereoftenpreciseto withinthe
yearforrecentfires,whileolderfireswerepreciseto about a decade. The firefrequency
distribution
was estimatedwiththisprecisionin mind.
The stand-origin
map was drawnfromtheaerial-photograph
interpretations
and the
point-sampledata of fireboundaries and dates. The aerial-photographdata were
transferred
to a 1: 50 000 topographicmap usingan Alterbury
Port-a-Scope.Landmarks
visibleon both the aerial photographsand the base-map were used to improvethe
fireages wereestimatedfromthe
positioningofthefireboundaries.The areas ofdiffering
map usinga Zeiss MOP digitizer.
The firefrequency
distribution
was analysedas follows.The time-since-last-fire
dates
on themap weregroupedintotwenty-year
age classes.This groupingtakesintoaccount
thefactthatall thefiredatesare notequallyreliable.For example,samplesdatedfromthe
oldesttreesgiveonlythedate of theearliestsurvivingrecruitment,
i.e. thedata can be
consideredleft-censored
(Lawless 1982).
The stand-origin
datesand area weredividedintoa combinationof
map and resulting
spatialdivisionswhichreflected
altitude,aspect,valleysystemsor climaticboundariesin
the Park. These spatial unitswere thentabulatedand plottedon semilogpaper. The
407
E. A. JOHNSON, G. I. FRYER AND M. J. HEATHCOTT
(b)
(a)
100
_
_
0. *
_
aR
15191760
FC 80
~17591988
FC 110
%
10 _
0~~~~~~~~
1.0
lIe
1-.
0
I
400
200
Standage (years)
I
I
600
0
l
100
I
l I
0
100 200
Time(years)
l
I
300
ofGlacierNationalPark:(a) showinga breakinthe
distribution
FIG. 4. The mixedtime-since-fire
FC is thefire
firecyclein themid-1700sand (b) partitionedintotwohomogeneousdistributions.
distribution
is a multi-modal
distribution
The mixedtime-since-fire
cycle.See textfordefinition.
distributions.
made up of two homogenoustime-since-fire
foreach twenty-year
age class is thecumulativeproportion
estimateofthefirefrequency
map forthatspecificage class.
of thearea on thestand-origin
The semilogplotsofthespatialdivisionsoftheParkshowa temporalbreakintheslope
or a mixedtimeindicatinga changein firefrequency
of thetime-since-fire
distributions
i.e. a mixtureof
distribution,
The latteris simplya multi-model
since-fire
distribution.
distribution.For furtherdiscussionsee
more than one homogeneoustime-since-fire
used a graphic
distribution
of themixedtime-since-fire
Lawless (1982). The partitioning
methoddescribedby Kao (1959). The procedureis as follows:the total time-since-fire
sectionsofthe
distribution
was plottedon semilogpaper,and linestangentialto different
thenewdistributions.
By tracingfrom
weredrawn.These linesrepresented
distribution
of theselinesto theright,thepercentageof samplesin each distribution
theintersection
thisbythetotalnumberofsamples,thenumberofsamplesin
could be read.Multiplying
was thenplottedas
was determined.
Each ofthesenewdistributions
each newdistribution
a cumulativepercentageof theirtotalto givethenew distributions.
The parameterof the negative model was estimatedby Maximum Likelihood.
was testedusinga WE test(Hahn & Shapiro1967).The likelihoodratio
Goodness-of-fit
test(Cox 1953) was used to comparetwo distributions.
thelengthofthefireseason,the
FirerecordsfromtheParkwereexaminedto determine
geographicdistributionand the causes of fire(M. J. Heathcott& E. A. Johnson,
unpublishedreportto ParksCanada).
RESULTS AND DISCUSSION
Changesinfrequencyoffire
from1520
identifies
periodsofchangein firefrequency
distribution
The time-since-fire
forGlacier National Park has a major
distribution
to the present.The time-since-fire
breakin slope in the mid-i700s (Fig. 4a). This changein slope indicatesa mixedtimeTo establishwhetherthebreakin
distribution
and a changein firefrequency.
since-fire
408
Man, climateandfirefrequency
(remember,
slope in the mid-i700s was caused by spatial or altitudinalheterogeneity
can onlymanifest
itselfin thebimodalityofthe
spatialheterogeneity
in thefirefrequency
time-since-fire
distribution,
just as a temporalchangewill),theParkwas dividedspatially
intotheBeaverRiverand MountainCreekWatershedsand RogersPass Corridorand
altitudinally
intotwozonesat 1800m. None ofthesedivisionsor theircombinationsgave
a reasonablegraphicalfitto a negativeexponentialdistribution.
distributions,
Because spatialand elevationalpartitionsstillgavemixedtime-since-fire
thenextpartitionwas a temporaldivisionat 1760wherethebreakin slopeoccurred.Both
periods,beforeand after1760,fita negativeexponential(oc< 0 05), and thetwofirecycles
different
weresignificantly
(oc<0 05). The firecycleforthe period 1760-1988was 110
yearsand thefirecyclefor1520-1760was 80 years.ExaminationofFig. 4(b) revealedthat
besidesthegeneraltrends,thereweresmalldeviationscaused by periodsin whichfires
burntgreaterand smallerareas.For example,in Fig.4 (b) thedecadesaround1880,1730,
1650and 1630appear to have beenperiodswherelargerareas burned.Of theseperiods,
only the 1880s correspondto a period of European activity(railroad construction).
of the Park in
Notice therewas no significant
decreasein firesaftertheestablishment
1888.
A changein thefirefrequency
inthemid-1700shas also beenidentified
infirefrequency
studiesof theMain and FrontRanges of theRockyMountains(Johnson& Fryer1987;
Masters,unpublishedreportto ParksCanada). The period1760-1988,witha firecycleof
110 years,in Glacier National Park is correlatedwiththecooler,wetterperiodduring
whichglaciersshowed major advances and the period 1520-1760,witha firecycleof
a warm,dryperiodprecedingtheseglacialadvances(Osborn 1982;
eightyyears,reflects
Luckman 1986;Osborn& Luckman 1988).
The good fitof the temporaldivisionsto the negativeexponentialdistributionin
Fig. 4(b) indicatesthattheentireParkis subjectto thesamefireregime(althoughthisfire
regimechangesat 1760)and consequentlythereare no topographicor spatialdifferences
ofburning.The changeinfire
infirefrequency.
Thus,everystandhas an equal probability
cyclein themid-1700sappearsto be relatedto a majorchangein climateassociatedwith
theLittleIce Age. DuringtheperiodofEuropeanoccupationafterthe1880s,thereis no
significant
changein thefirecycle.
The restof thispaper will addressthecommonlyacceptedexplanationof European
effectson firefrequencyand explain why it is not applicable here. An alternative
and itseffecton firefrequency
is also presented.
explanationof European intervention
Effectsof theEuropeaninvasiononfires
causes forthe period 1880-1980,
Figure 5 shows the area burnedand the different
obtainedfromfirereports(M. J.Heathcott& E. A. Johnson,unpublishedreportto Parks
Canada). Largefiresofunknownand Europeanoriginburnedonlyin the1880s.Sincethe
has burnedlargeareasin thePark.Duringtheconstruction
phaseof
1890s,onlylightning
theCanadian PacificRailroad(1882-86),severallargefiresoccurred(1883, 1885,1886)in
on thedrier(rainshadow)side
theRogersPass Corridor.Thesefiresall burnedprimarily
whethertheywerecaused by Europeansor
of RogersPass. It is impossibleto determine
fires
is blamed,althoughlightning
railroadconstruction
or both.Traditionally,
lightning,
also occurredin theseyearsand in areas wheretherewas no European activity.For
example,in the Mountain Creek area, therewere firesin 1886 and we know with
reasonablecertainty
thattherewas no European activityin thisvalleyat thattime.
E. A. JOHNSON, G. I. FRYER AND M. J. HEATHCOTT
409
x
20000
c) 10000 _
0
oL
1880
1900
1920
1940
1960
1980
Year
FIG 5. Estimatesofarea burntduringeach decade from1880to 1980in GlacierNationalParkby
cause: (O) railway,(*) lightning,
(x) unknown.Areas are based on M. J. Heathcott& E. A.
Johnson(unpublishedreportto ParksCanada) and shouldbe consideredroughestimates.
clearingand
is blamedforcausingfiresbecause thesurveying,
Railroad construction
forfires.It is coincidentalthatthis
oftherailroadled to moreopportunities
construction
fires;therehad also
correspondedto a periodwithmorelightning
periodofconstruction
been a highincidenceof lightingfiresin 1730, 1650and 1630.However,theimpression
shouldnotbe giventhattherewas a cavalierattitudeoftherailroadtowardsfiresduring
oftherailroadweremade ofwood
structures
thisperiod.Bridgesand otherright-of-way
and wereoftendestroyedby fire.Consequently,mostof therailroad-causedfireswould
set.
and not deliberately
have been unintentional
Clearly,one ofthemostcommoncauses offiremusthave beensparksfromwood and
coal-burning
engines.In recentyears,sparksand hot piecesof brokenbrakeshoeshave
beenthecauses offireon steepgradients(M. J.Heathcott& E. A. Johnson,unpublished
reportto ParksCanada). Railroad-causedfiresoccurearlyin thespringwhenthegrassy
groundfires.The
fuelson therightof way are dry,and willeasilysupportlow-intensity
forestat thistimeis snow-coveredor stilltoo wet to supportfire.Even now, as Fig. 6
along the
shows,man-causedfiresare distributedonlyalong the railroad,particularly
throughoutthe
firesare moreevenlydistributed
steepgradients,whilelightning-caused
Park.
becausetheyhave notconsistently
on firefrequency
Europeanshavehad a smalleffect
firecycleis 110years,in order
burneda largeenougharea. For instance,ifthelightning
forEuropeans to shortenthefirecycleto say eightyyears,an area of 20 km2mustbe
burnedeverytenyearsin additionto the55 km2thatis beingburnedin thesameperiodby
Sincethe1880s,manhas onlycaused 1 5 km2to burn,muchlessthan1% ofthe
lightning.
is thesize of thearea burnedand not
area. Whatis importantin changingfirefrequency
thenumberof fires.
land-useby Europeanswas controlledby theDominion Lands Act of
Furthermore,
1879 which excluded permanentsettlementand ranching,and restrictedloggingto
ofGlacierNationalPark
railroadrightsofway,and thenbythesubsequentestablishment
in 1888.In addition,thePark is ruggedand isolated;consequentlyEuropeanaccess was
discouragedthekindoflanduse thatwouldhaveincreasedfire
limitedand circumstances
and improperslash disposal) or decreasedit (e.g.
(e.g. extensiveclear-cutting
frequency
efforts).
fire-suppression
Man, climateandfirefrequency
410
(a)
(b)
Mountain
Creek
iCreek
ogers
kilomet
Rogerses
FIGlaciherdistributionRogers)man-and(b)lightnng-c ier
NationalP
(JH
Parked
Pass
o
nisona
td
GlacierPass
firearelesstha1hinizPark
N\Z(
0
10
kilometres
0
10
kilometres
firesfrom1960 to 1985in Glacier
of (a) man-and (b) lightning-caused
FIG 6. The distribution
NationalPark(M. J.Heathcott& E. A. Johnson,unpublishedreportto ParksCanada). All mancaused firesare less than 1 ha in size.
The effect
of weatheron thespreadoffire
as suggestedin the
These data showthatEuropeanshave notincreasedfirefrequency
oftherailroad,and
Introduction.
The onlyeffect
Europeanshad was duringconstruction
increaseregionalfirefrequency.
even thisperiodof increasedfiresdid not significantly
of GlacierNational Park,firesuppressionhas not
Furthermore,
sincetheestablishment
weather.Therefore,
itappears
reducedtheareasburnedduringperiodsofseriousfire-risk
that in order for Europeans to increase the firefrequency,they must start fires
systematically
duringall (or a largeportion)of thesedryperiods.Similarly,in orderto
decreasethefirefrequency,
theymustbe able to suppressfiresduringtheseperiods.The
problemis to changeour viewpointof firesso as to understandthatonlyduringcertain
fuel-moisture
conditionsare largerfirespossible.
Fuel drying,ignitionand rapid firespreadare relatedin GlacierNational Park to a
characteristic
synopticweatherpattern(PacificSouthwestForest& Range Experiment
Station 1964; Nimchuk 1983; Fryer& Johnson1988). In Glacier National Park, the
criticalweatheris associatedwitha surfacehigh-pressure
systemwhichestablishesitself
over the interiorof BritishColumbia and Alberta.This surfacehigh-pressure
system
createsa stationaryridgein thewesterly
circulationand blocksthenormalflowofmoist
Pacificairfromthecoast,steering
itnorthand southoftheColumbiaMountains.The fire
season(Fig. 3) correspondsto thedevelopment
ofthisridgeinJuneand itsdisappearance
in autumn.The ridgesbringclear,dry,warmweatherwithverylittlewind.These ridges,
whichblock the westerlyflow,are replacedat intervalsby rapidlymovingupper-level
troughswhichproducesurfacelow-pressure
systems.When weak, thesecold lows can
erraticwindsbut little
a decreasein temperature,
and higher-speed,
producelightning,
precipitation.
Strongerlow-pressure
systemswillproduceabundantprecipitation.
The criticalfireweatherin GlacierNationalParkis markedbyoffueldryingduringthe
Pacificairflowis blockedbythestationary
periodwhenwesterly
ridge.Thiscan lastfrom
a fewdaysto severalweeks,and one or moreepisodesofthisridgebyupper-level
troughs
E. A. JOHNSON, G. I.
FRYER AND M.
J.HEATHCOTT
411
and highwinds.It is duringthebreakdownof theridgethatmost
willcause lightning
seriousfiresigniteand spread rapidly(Nimchuk 1983; Fryer& Johnson1988). The
criticalfireweatheris usuallyended by thedominanceof theupper-leveltroughwhich
thispattern(M. J. Heathcott& E. A. Johnson,
bringsrain.The year 1971 exemplifies
unpublishedreportto Parks Canada).
CONCLUSION
In coniferousecosystemssuch as GlacierNational Park,thespreadof fireis rapidand
almostimpossibleto containduringperiodsof criticalfireweather.Thus, in orderto
belowthatofthenaturalfireregime,firesmustbe suppressedat
reducethefirefrequency
above thenaturalfireregime,
in orderto increasethefirefrequency
suchtimes.Similarly,
areas
burnedsignificant
Europeans would have had to startfireswhichconsistently
duringthesesameweatherperiods.Generally,European-causedfiresin GlacierNational
thefire
affected
or repeated,and havenotsignificantly
Parkareaccidental,notsystematic
frequency.
ACKNOWLEDGMENTS
thestand-origin
map. His crewsconsisted
G. Levorsonled thefieldworkand constructed
ofA. Woo and T. Branscomb.We also thankRogerTurnbull,RogerEddy,Alan Masters
and Dana Wowchuk for helpfuladvice and discussionthroughoutthis study.The
researchwas supportedbygrantsto E.A.J.fromParksCanada and theNaturalSciences
and EngineeringResearchCouncil of Canada.
REFERENCES
of
Achuff,
P. L., Holland,W. D., Coen, G.M. & Van TighemK. (Eds) (1984). EcologicalLand Classification
MountRevelstokeand GlacierNationalParks,BritishColumbia.PublicationM-84-11. AlbertaInstituteof
of Alberta.
Pedology,University
40, 354-360.
Cox, D. R. (1953). Some simpletestsforPoisson variates.Biometrika,
Fritts,H. (1976). TreeRingsand Climate.AcademicPress,New York.
in a subalpineforest.Journalof
firebehaviourand effects
E. A. (1988). Reconstructing
Fryer,G. I. & Johnson,
AppliedEcology,25, 1063-1072.
JohnWiley& Sons, New York.
Hahn, G. J. & Shapiro,S. S. (1967). StatisticalModels in Engineering.
M. L. (1973). Firein thevirginforestsoftheBoundaryWaterCanoe Area,Minnesota.Quaternary
Heinselman,
Research,3, 329-382.
Johnson,
E. A. & Fryer,G. I. (1987). Historicalvegetationchangein theKananaskisValley,Canadian Rockies.
CanadianJournalofBotany,65, 853-858.
E.A. & Van Wagner,C.E. (1985). The theoryand use of twofirehistorymodels.CanadianJournalof
Johnson,
ForestResearch,15, 214-220.
of electrontubes.
Kao, J. H. K. (1959). A graphicestimationof mixedWeibull parametersin life-testing
31, 389-407.
Technometrics,
Lawless,J. F. (1982). StatisticalModels and Methodsfor LifetimeData. JohnWiley& Sons, Toronto.
ofLittleIce Age eventsin theCanadian RockyMountains.Geographie
Luckman,B. H. (1986). Reconstruction
40, 17-28.
physiqueet Quaternaire,
Bulletin,
43, 51-67.
McBride,J.R. (1983). Analysisoftreeringsand firescarsto establishfirehistory.Tree-Ring
of TorontoPress,Toronto.
Moss, E. H. (1959). Flora of Alberta.University
Nelson,J. G. & Byrne,A. R. (1966). Fires,floodsand nationalparksin theBow Valley,Alberta.Geographical
Reviews,56, 226-238.
Nimchuk,N. (1983). WildFire BehaviorAssociatedwithUpperRidgeBreakdown.AlbertaEnergy& Natural
ResourcesForestService.ENR ReportT150. Edmonton,Alberta.
in thesouthernCanadian Rocky Mountains:a
Osborn,G. (1982). Holocene glacierand climatefluctuations
review.Striae,18, 15-25.
412
Man, climateandfirefrequency
in theCanadian Cordillera(Albertaand
Osborn,G. & Luckman,B. H. (1988). Holocene glacierfluctuations
BritishColumbia). QuaternaryScienceReviews,7, 115-128.
PacificSouthwest
ForestandRangeExperiment
Station(1964). SynopticWeatherTypesAssociatedwithCritical
Fire Weather.U.S.D.A. ForestService,PacificSouthwestForest& Range ExperimentStation.
Tande,G.F. (1979). Firehistoryand vegetationpatternofconiferousforestsin JasperNationalPark,Alberta.
CanadianJournalof Botany,57, 1912-1931.
Titus,R. L. (1965). UpperAir Climateof Canada: Average,Extremeand StandardDeviationValues1951-60.
Departmentof Transport,MeteorologicalBranch,Toronto,Ontario.
(Received9 June1989; revisionreceived11 December1989)