Long-term decreasing trend in forest fires in northwestern Canada

Long- e m dec easing end in o es i es in no hwes e n Canada TUOMO H. WALLENIUS, 1, JUHO PENNANEN, 2 AND PHILIP J. BURTON 3,4 1 Finnish Fo es Resea ch Ins i u e, Van aa Resea ch Uni , P.O. Box 18, FI-01301 Van aa, Finland 2 Depa men o Biosciences, Uni e si y o Helsinki, P.O. Box 65, FI-00014 Helsinki, Finland 3 Canadian Fo es Se ice, P ince Geo ge, B i ish Columbia V2N 4Z9 Canada 4 Ecosys em Science and Managemen , Uni e si y o No he n B i ish Columbia, P ince Geo ge, B i ish Columbia V2N 4Z9 Canada Abs ac . The annual a ea o o es bu ned has dec eased in ecen cen u ies o e la ge a eas o Fennoscandia, Sibe ia and empe a e No h Ame ica. To de e mine i his same end ex ends o a spa sely popula ed egion o no he n Canada, i e sca s on li ing and dead ees, o es s and ages and cha ed wood we e sys ema ically sampled in 85 s udy plo s in an a ea o 564 000 km 2 in no hwes e n Canada. A signi ican nega i e end in he occu ence o o es i es was obse ed: a e age a ea bu ned pe yea dec eased om 2.0%in he i s hal o he 19 h cen u y o 0.33%in he la e hal o he 20 h cen u y. Annually bu ned a eas co ela ed signi ican ly wi h a local ee ing based index, July mon hly d ough code and he Paci ic decadal oscilla ion bu no wi h June-Augus mean empe a u e, dis ance o he nea es oad, o he yea o oad building. None o he clima ic indica o s o access his o y (indica i e o he s a o local i e supp ession) could explain he long- e m nega i e end in i es. Ea lie in e p e a ions ha humans domina ed he causes o o es i es in he pas , e en in spa sely popula ed egions, dese e u he a en ion as a possible explana ion o he dec easing end in i es. Key wo ds: annually bu ned a ea; bo eal o es ; Canada, clima e; i e cycle; i e supp ession; o es i e, human in luence. Recei ed 1 Ma ch 2011; accep ed 4 Ma ch 2011; inal e sion ecei ed 18 Ap il 2011; published 4 May 2011. Co esponding Edi o : D. P. C. Pe e s. Ci a ion: Wallenius, T. H., J. Pennanen, and P. J. Bu on. 2011. Long- e m dec easing end in o es i es in no hwes e n Canada. Ecosphe e 2(5):a 53. doi:10.1890/ES11-00055.1 Copy igh : Ó2011 Wallenius e al. This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal au ho and sou ces a e c edi ed. E-mail: Tuomo.W[email p o ec ed] INTRODUCTION The ecological li e a u e ypically desc ibes he No h Ame ican bo eal o es as a biome ha equen ly bu ns in na u al ligh ning igni ed o es i es (Rowe and Sco e 1973, Johnson 1992, Paye e 1992). P io o he mid-19 h cen u y i e cycles ha e a ied om 30 o 130 yea s in Canada’s bo eal o es s (Zhang and Chen 2007). Howe e , he majo i y o i e his o y s udies in No h Ame ica ha e been conduc ed ei he in he sou he n pa o bo eal zone (Wei e al. 2000, Be ge on e al. 2001) o in mo e empe a e coni e ous o es s (Heinselman 1973, Heye dahl e al. 2001). Excep o he esea ch based on i e s a is ics and sa elli e images spanning he las ew decades (S ocks e al. 2002, Kasischke and Tu e sky 2006), ela i ely ew la ge-scale i e his o y s udies (Ya ie 1981, La sen 1997) ha e been conduc ed in no he n bo eal No h Ame - ica o in ‘‘ he ue bo eal’’ as de ined by B and (2009). The gene al iew is ha clima e dic a es he a ea annually bu ned, a leas in emo e a eas such as he bo eal o es s o No h Ame ica. I has been sugges ed ha as a consequence o clima ic wa ming, i es ha e (and will con inue o) become mo e equen (Gille e al. 2004, www.esajou nals.o g 1May 2011 Volume 2(5) A icle 53 Flannigan e al. 2005, Soja e al. 2007), which could be he eason o he inc ease in he a ea annually bu ned seen in he i e s a is ics. Howe e , he e ec o clima e on i e equencies is appa en ly complex and spa ially and empo- ally a iable. Clima e change in he 20 h cen u y has no led o mo e high- isk i e wea he e e ywhe e in he bo eal o es s. Du ing he 20 h cen u y in sou hwes e n and sou heas e n Canada, he clima e has shi ed o condi ions less conduci e o o es i es, bu o e he same pe iod, o example, he Eu asian aiga has become d ie (Gi a din e al. 2009). Se e al wea he ac o s, including empe a- u e, p ecipi a ion and wind, a ec daily i e isk. Fo example, summe empe a u e and p ecipi- a ion oge he wi h o he clima ic a iables explained 79%o he annual a ia ion in bu ned a ea du ing he la e hal o he 20 h cen u y in Alaska (Du y e al. 2005). I has been demon- s a ed ha oscilla ion o he sea su ace empe - a u es could explain he long- e m ends in i e egimes (Ki zbe ge e al. 2001). In bo eal No h Ame ica, he Paci ic Decadal Oscilla ion (PDO) co ela es wi h o es i es (Du y e al. 2005, Fau ia and Johnson 2008). Ano he clima ic index ha has been shown o co ela e well wi h annually bu ned a ea is he mon hly d ough code (MDC) o July in Canada (Gi a din e al. 2009). In addi ion o clima e he e a e also human ac o s ha ha e been hough o a ec i e egimes. No ably, i e supp ession is associa ed wi h a dec ease in i es in Fennoscandia (Zack- isson 1977) and in No h Ame ica (Heinselman 1973). Howe e , in no he n Canada, o es i es ha e no been supp essed as e ec i ely o o as long a pe iod o ime as in sou he n Canada o in Fennoscandia. Fi es in Canada’s emo e no he n egions we e no well documen ed be o e he 1950s o 1960s (S ocks e al. 2002) and o es i es which occu ed ou side o p o ec ion zones be o e abou 1980 we e nei he epo ed no supp essed (Sima d 1997). La ge a eas o Cana- da’s bo eal o es emain zoned o ‘‘ modi ied supp ession’’ by p o incial, e i o ial, and na- ional pa k agencies, meaning ha he o es is allowed o bu n unless communi ies o o he in as uc u e is a isk (S ocks e al. 2002). In con as o ecen i e supp ession e o s, p esumably esul ing in a educ ion in he a eas bu ned, o he human ac ions may ha e aug- men ed he occu ence o i es. In he 19 h cen u y a common no ion was ha humans we e he p edominan cause o o es i es (Blomq is 1888, Bell 1889). Cu en ly, his is no conside ed ue o mos o he bo eal No h Ame ica, as 80%o he bu n a ea and mo e han 70%o he numbe o i es ( hose .200 ha) in compiled s a is ics o no he n ecozones a e caused by ligh ning-igni ed i es (S ocks e al. 2002, Ka- sischke and Tu e sky 2006). Ne e heless, i has been p oposed ha de elopmen ac i i ies and indus ial o es y ha e inad e en ly p omo ed ligh ning i es by inc easing he a ailabili y o lammable uels (A ien i e al. 2009, Lindenmaye e al. 2009). Du ing he las 150 yea s, o es i es and biomass bu ning ha e declined globally (Ma lon e al. 2008). This can be clea ly seen om i e his o y ield s udies in Fennoscandia, sou he n Canada and he wes e n Uni ed S a es, whe e annually bu ned a eas ha e s eeply dec eased o e he las cen u y o longe (Zack isson 1977, Flannigan e al. 1998, Niklasson and G ans o¨m 2000, Zhang and Chen 2007). Howe e , in no he n No h Ame ica he i e his o y is no clea . While some ield s udies show leng hening o he i e cycles (La sen 1997), o he s poin o a conside able empo al a iabili y (Cy e al. 2009). Indi ec econs uc ions sugges ha he e has been no end o e he las h ee cen u ies (Gi a din and Sauchyn 2008), and compiled i e s a is ics a es o conside able sho ening in i e cycles (S ocks e al. 2002, Kasischke and Tu e sky 2006). These os ensible con lic s be ween s udies undoub edly e lec egional di e ences, di e - en ime ho izons o e which ends a e assessed, and p obably some me hodological di e ences. Fu he ield s udies a e clea ly needed o cla i y he complex empo al and spa ial pa e ns in he occu ence o o es i es in he no he n No h Ame ica. The aim o his s udy is o cla i y he i e his o y o he pas wo cen u ies o e a la ge egion o no hwes e n Canada. Speci ic esea ch ques ions a e: (1) How o en ha e o es s bu ned in he pas ? (2) Ha e he e been any signi ican ends in he i e cycle du ing he las wo cen u ies? (3) Wha a e he causes o any obse ed empo al pa e ns in i e cycles? www.esajou nals.o g 2May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL. METHODS S udy a ea The s udy a ea is loca ed in no heas B i ish Columbia, no hwes Albe a, sou heas Yukon Te i o y and sou hwes No hwes Te i o ies in he Canadian No hwes (Fig. 1). The app oxi- ma ely ec angula s udy a ea s e ches abou 730 km om no h o sou h be ween la i udes 578N and 63.58N, and be ween 1258W and 1118W longi ude. The a ea encompasses abou 564 000 km 2 o which G ea Sla e Lake and smalle wa e bodies co e 10%. Topog aphy is mos ly la o gen ly con ou ed excep o he wes e n ma gin o he s udy a ea. Ele a ion anges om 120 m a he Mackenzie Ri e o 2942 m on Moun Syl ia, bu mos ly a ies be ween 150 and 500 m. Clima e o he egion is con inen al. A e age empe a u e in July is be ween 158C and 178C and in Janua y is be ween 198C and 278C (En i onmen Canada, Na ional Clima e Da a and In o ma ion A chi e 2010). A e age annual p ecipi a ion is be ween 280 and 450 mm. The majo i y o he a ea belongs o he Taiga Plains ecozone bu segmen s o he Taiga Shield and Bo eal Plains ecozones a e included in he eas e n pa o he s udy a ea. Picea ma iana (Mill.) B i on, S e ns & Poggen- bu g, Picea glauca (Moench) Voss, Pinus con o a Douglas, Pinus banksiana Lamb., La ix la icina (Du Roi) K. Koch, Populus emuloides Michx. and Populus balsami e a L. a e he mos common ee species. The ield laye is ypically domina ed by dwa sh ubs belonging o genus Ledum,Vacci- nium,andEmpe um. The pan-bo eal mosses Pleu ozium sch ebe i (B id.) Mi . and Hylocomium splendens (Hedw.) B.S.G. and di e en species o Sphagnum ypically co e he o es loo . Based on digi al opog aphic maps om Na u al Resou ces Canada, muskeg and o he ypes o we lands co e app oxima ely 20%o he s udy a ea. Humans ha e inhabi ed no he n Canada o housands o yea s (F ancis e al. 2002). The i s humans in he Ame icas we e hun e -ga he e s who in luenced hei en i onmen by hun ing la ge he bi o es and by ini ia ing a leas some landscape i es. The e ec o hese ea ly people on o es i e egimes is unce ain bu po en ially e y signi ican : e en a small popula ion o humans, i i wishes o do so, is capable o bu ning o es s mo e equen ly han he o es s would na u ally bu n (G ans o¨m and Niklasson 2008). The human popula ion densi y in he s udy a ea is oday p obably a eco d high le els, bu is s ill less han 0.1 inhabi an s pe km 2 . The e we e p ac ically no oads be o e he 1940s. In he No hwes Te i o ies, he i s oads we e buil in he 1960s. A la ge po ion o he oad ne wo k was cons uc ed o suppo he oil, gas and mine al explo a ion ha s a ed in he 1950s and has expanded and con inued o he p esen day. In he sou he n pa o he s udy a ea, a modes le el o comme cial o es y has been p ac iced in he la e hal o he 20 h cen u y (Schneide 2002). Mos o he s udy a ea emains un oaded wilde ness, domina ed by p ime al o es and li le human use. Fi e supp ession ac ions in he a ea ha e a ied h ough ime and space bu he longes policy has been o p o ec p ope y and li es in p oximi y o ci ies, high- ways and majo i e s. Conside ing ou sam- pling s a egy (see Me hods: Fi e his o y econs uc ion), he oad building da es cons i- u e he likely s a o i e supp ession, al hough i may ha e become e ec i e only much la e . Fi e his o y econs uc ion Fi e his o y was econs uc ed using (1) i e sca s on ees, (2) o es s and ages and (3) i e killed and cha ed snags in 85 s udy plo s wi hin a bu e zone o 0.2–1.0 km om he exis ing oad ne wo k in he a ea. The i s plo was andomly loca ed wi h A cGIS so wa e. Addi ional plo s we e loca ed wi h a p ocedu e whe eby he nex plo was placed on he opposi e side o he oad 40 km away as he ae ial poin - o-poin dis ance (Fig. 2). The o hogonal dis ance o each plo cen e om he oad was andomized using 0.1 km s eps o be 0.2–1.0 km. A e he i s 30 plo s he dis ance om he oad was limi ed o 0.5 km a maximum because o ime limi a ions. The plo s we e loca ed wi h he help o a GPS de ice. I a plo cen e ell in a wa e body, i was ejec ed and a new plo was andomized o be 2–9 km o wa d along he oad on he opposi e side o i . The adius o sampling o each ci cula plo was 100 m, encompassing an a ea o 3.1 ha. This plo size was conside ed o be small enough o se e as a poin -wise measu e o he occu ence o i es, ei he bu n o no bu n in any i e, while being la ge enough o ha e a good chance o www.esajou nals.o g 3May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL. include i e sca ed ees. Al oge he 453 samples om li ing ees and 45 samples om dead ees we e collec ed, polished wi h ex a ine sandpa- pe and b ough o he labo a o y o analysis by mic oscope. Fi e sca s we e used as a p ima y sou ce because hey yield ela i ely accu a e da es o i es. Disks o wedges including i e sca s we e sawn om up o i e li ing and dead ees pe plo . Fi e sca s a e ypically loca ed a he base o Fig. 1. Loca ions o he s udy a ea and s udy plo s in No hwes Canada. Fig. 2. Schema ic diag am po aying he loca ing o s udy plo s. www.esajou nals.o g 4May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL. he unk and ha e a sha ply o med inju y showing no lesion ha g ows yea a e yea . The la e is ypical o sca s made by alling ees and ungi. In wo plo s i e sca s we e collec ed mo e han 100 m om he plo cen e . Howe e , in bo h cases i e sca s we e s ill conside ed o ha e o igina ed om he same i e ha had bu ned he plo cen e as he o es age and s uc u e we e simila in he plo cen e and a he addi ionally sampled loca ion u he away. Fi e sca s om li ing ees we e da ed by coun ing he annual ings backwa ds om he las yea o g ow h. Possible missing ings can, howe e , cause an e o o a couple o yea s in da ing. This e o was minimized by selec ing he adius ha ing he b oades annual ings in each sample, and by c osschecking i e sca da es om he same plo . Fi e sca s om 9 dead ees we e dend och onologically c oss-da ed (see he la e pa o his sec ion). Li ing ees we e sampled o hei age in e e y plo . Fi s , h ee dominan ees (no g owing unde he canopy o o he ees) nea es o he plo cen e we e sampled. Secondly, one o h ee ees ha subjec i ely seemed o be he oldes wi hin 20 m om he plo cen e we e sampled. Samples we e ex ac ed om as close o he base o he ee as possible, on a e age less han 0.3 m abo e he g ound le el. Ages o he sampled ees we e es ima ed by coun ing he annual ings o he pi h and by adding a ixed numbe o yea s o accommoda e he ime ha is needed o open-g own ees o each he sampled heigh . Based on ou sampling, one yea was added o all b oadlea ed ees, wo yea s o Pinus and 5 yea s o Picea, La ix and Abies. In a ew cases whe e pi h was no p esen , he missing annual ings we e es ima ed based on cu a u e o he ings nea he pi h (A no and Sneck 1977). The ee ages we e used o in e ing i e da es i no i e sca s co esponding o he ee ages could be ound. Fi e was de e mined o ha e occu ed one yea be o e an es ima ed yea o ge mina ion o he oldes sample ee in he plo . To a oid assigning alse i es, he e had o be a leas one o he sample ee bo n wi hin 30 yea s o indica e a egene a ion coho ollowing a i e. T ees killed and cha ed by i e we e used as addi ional e idence o ensu e ha he s udied o es s we e ac ually bu ned. The samples om dead ees we e dend o- ch onologically c oss-da ed, i possible. Samples om li ing ees wi h he longes sequences o undis u bed g ow h we e used o build mas e ch onologies o Picea glauca and Pinus banksiana. Inaddi ion,da ap o idedbyH.F i s,M. Hu on, T. Knowles, C. La sen, D. Meko, K. Mose , F.H. Schweing ube , C. S ock on and J. Szeicz (Wo ld Da a Cen e o Paleoclima ology, In e na ional T ee Ring Da a Bank 2010) we e used o building he mas e ch onologies. Da es o he samples we e de e mined based on a dis inc ly high co ela ion wi h he mas e se ies using COFECHA so wa e (G issino-Maye 2001) and e i ied ag eemen wi h he known i e yea s and ee age s uc u e in he plo in ques ion. In all cases, he isual appea ance o he sample was also conside ed o exclude e oneous da es. C oss-da ing p o ed o be challenging because we we e dealing wi h mul iple ee species in a e y la ge a ea whe e one mas e ch onology pe species is no enough. O he 45 samples om dead ees only 15 could be dend och onologi- cally c oss-da ed. The samples ha emained unda ed we e mos ly cha ed snags. Cha coal was ound especially on b anch ips bu no on he unk, indica ing ha he ee had p obably been killed by he p e ious i e. Snags ha had been dead o a long pe iod be o e he i e would ha e e y likely been wi hou ba k and d y, and he e o e would ha e bu ned mo e ho oughly. Hence, he dea h o ees wi h cha ed b anch ips was da ed o a pe iod immedia ely (one yea ) be o e he las i e in he plo in ques ion. Ve y low se e i y i es ha do no kill ees, o e en cause sca s on hem, may no be de ec ed wi h he me hods desc ibed abo e. Howe e , such low-se e i y i es end o sp ead slowly and emain small. The g ea majo i y o he a ea bu ned in his egion is om la ge powe ul i es. The econs uc ed i e his o y was compa ed wi h i e s a is ics, including all documen ed o es i es, p o ided by Na u al Resou ces Canada and he i e managemen agencies o B i ish Columbia, Albe a and No hwes Te i- o ies. S a is ical analysis The a e age annually bu ned p opo ion o a pe iod om yea 0 o yea 1 (including 1 )was es ima ed as www.esajou nals.o g 5May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL. a 0; 1¼1 1 0þ1X 1 ¼ 0 bð Þ cð Þð1Þ whe e b( ) is he numbe o bu n plo s and c( ) he numbe o ac i e plo s in yea . A plo was de e mined o be ac i e (i.e., eco ding i es) since he i s de e mined i e yea o since he i s yea in he annual ing ch onology o he plo in ques ion, whiche e came i s . The i e cycle o he same pe iod om yea 0 o yea 1 (including 1 )is 0; 1¼1=a 0; 1ð2Þ The i e cycle equals o he ime needed o bu n a cumula i e a ea equi alen o ha o a la ge e e ence landscape (on a e age once by sepa- a e, small i es) and i is he in e se o he a e age o he annually bu ned p opo ions (Johnson and Gu sell 1994). The i e cycle also equals he expec ed poin -wise a e age i e in e al. We assume ha in a gi en yea , he s udy plo s beha e independen ly. Then, he numbe o obse ed i es o yea ollows a binomial dis ibu ion, b ;B(c( ), p( )), whe e p( ) is he p obabili y ha a plo bu ns, gi en he condi- ions du ing he yea . When b( ) and c( ) a e known, he pos e io p obabili y o p is he be a dis ibu ion Be a(b( )þ1, c( )b( )þ1). The e o e he 0.025 and 0.975 quan iles o he be a dis ibu ion gi e he 95%con idence in e al o p( ). We used logis ic eg ession o e alua e he e ec o se e al ac o s on i e occu ence, and he magni ude o a possible esidual end. The dependen a iable is he obse ed yea ly num- be o i es and he co a ia es a e he yea , oad building da e and a se o clima ic indica o s. The a e age empe a u e o e June–July–Augus (JJA), July mon hly d ough code (MDC), and he annual a e age Paci ic decadal oscilla ion (PDO) index we e e alua ed o bes explain o es i es in ou s udy a ea based on ea lie esea ch in no he n No h Ame ica (Du y e al. 2005, Fau ia and Johnson 2008, Gi a din e al. 2009). We used he annual a e age PDO alue, which is e y s ongly posi i ely co ela ed wi h he Janua y–Feb ua y ( ¼0.79), and sp ing- summe ( ¼0.96) PDO alues used by Du y e al. (2005) and Fau ia and Johnson (2008), espec i ely. In addi ion, an index de i ed om a h ee-cen u y long econs uc ion o ee g ow h in he Taiga Plains eco egion by Gi a din and Sauchyn (2008) was used as a p oxy o clima ic o cing on i e egimes in he egion. Thei econs uc ion (he ea e e e ed o as GiR) is based on he clima ically egula ed componen o a ia ion in ee- ing wid hs in he egion and no di ec ly o bu ned a eas. The GiR index ep esen s an unde ined se o wea he a iables ha co ela e bo h wi h he ee g ow h and he occu ence o i es. His o ical ins umen ally measu ed PDO indi- ces we e acqui ed om he Royal Ne he lands Me eo ological Ins i u e Clima e Explo e (2010). PDO index alues a e sea su ace empe a u e (8C) anomalies oscilla ing a ound ze o; he alues used in ou analysis anged om a minimum o 2.3 o a maximum o 3.08C. July d ough codes (MDC; Gi a din e al. 2009) o he s udy a ea we e compu ed using clima e da a om he Clima e Resea ch Uni o he Uni e si y o Eas Anglia and so wa e SimMDC p o ided by Ma in Gi a din (Canadian Fo es Se ice, Saul S e. Ma ie, On a io). The his o ical June- Augus mean empe a u es in Fo Simpson and Hay Ri e du ing 1895–2008 we e acqui ed om he Royal Ne he lands Me eo ological Ins i u e Clima e Explo e (2010) and he En i onmen Canada, Na ional Clima e Da a and In o ma ion A chi e (2010b). A ew missing da a en ies we e assigned an a e age o he mon h and s a ion in ques ion. Road building da es we e de e mined om maps o a ious da es, o he publica ions (G een- wood 1992), di e en in e ne sou ces, and pe sonal communica ion wi h local au ho i ies. Despi e hese e o s, exac da es could no be ound o e e y oad. Fo hese oads he building da es we e assigned a midpoin da e be ween a known poin o ime when he oad was no buil ( o example, om an old map wi hou he oad) and when i appa en ly was p esen ( o example, a mo e ecen map, o he yea o ield sampling). We model he haza d o bu ning h( ) in a yea o plo by logi ðhð ÞÞ ¼ aþb þq ; þX i cixið Þð3Þ www.esajou nals.o g 6May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL. whe e x i ( ) a e he clima ic co a ia es ques ion, 2 0,1gindica es i he oad nea es o plo was buil be o e , and a,b,q, and c i a e pa ame e s. Fo cla i y, we index he c i co e- sponding o each clima ic indica o as c JJA ,c MDC , c PDO and c GiR . The clima ic indica o s we e no malized so ha he mean alue was 0 and s anda d de ia ion 1. Yea was scaled o ake alues be ween 0, o he i s , and 1, o he las analysed yea . Since we assume ha bo h plo s and yea s a e independen o each o he , he abo e model is a s aigh o wa d gene alized linea model. How- e e , he unce ain y in he da ing o he i es is signi ican , p obably a ies in ime, and di e s be ween he i es da ed om s and age (age- da ed) and i es dend och onologically c oss- da ed om i e sca s (sca -da ed). We he e o e used a Bayesian amewo k o he eg ession modeling, which allows inco po a ing he da ing e o , which is no exac ly known, di ec ly in he analysis. We assume he ac ual da e o a i e eco ded o yea is dis ibu ed acco ding o a no mal dis ibu ion N( þd j , j ( ) 2 ), whe e j¼a and j¼s, o age-da ed and sca -da ed i es, espec i ely. We model a by að Þ¼ðk1ð  0Þþk0ð 1 ÞÞ=ð 1 0Þð4Þ whe e 0 and 1 a e he i s and he las yea s in he da a, espec i ely, and k 0 and k 1 a e pa am- e e s. The sys ema ic e o d j , as well as s, ¼k 2 a e assumed o be independen o and we assume ha a ¼0. To calcula e he likelihood o he obse ed numbe s o each ype o i e obse a ion, we also need o know he p obabili y s( ) o he i e being obse ed only as an age-da ed i e. We assume ha his nuisance pa ame e is a unc ion o ime since he i e in ques ion, and es ima e i by sð Þ¼~ mað Þ=ð~ mað Þþ~ msð ÞÞ ð5Þ whe e s( ) is he p obabili y ha a i e a yea is obse ed only as age-da ed, and ˜ m a ( ) and ˜ m s ( ) a e ke nel-smoo hed yea ly numbe s o obse ed age-da ed and sca -da ed i es, espec i ely. Now he comple e model o he p obabili y o eco ding a i e on a plo a yea is ha; ð Þ¼sð ÞX 1 0¼ 0 Nð þda; að Þ2Þð 0Þlogi 1 ðaþb 0þq ; þX i cixið 0ÞÞ ð6Þ o he age-da ed i es, and hs; ð Þ¼ð1sð ÞÞX 1 0¼ 0 Nð þds; sð Þ2Þð 0Þlogi 1 ðaþb 0þq ; þX i cixið 0ÞÞ ð7Þ o he sca -da ed i es. Finally, he likelihood unc ion is gi en by L¼Y ; q ð Þðga; ð Þha; ð Þþð1ga; ð ÞÞ  ð1ha; ð ÞÞÞ Y ; q ð Þðgs; ð Þhs; ð Þþð1gs; ð ÞÞ  ð1hs; ð ÞÞÞ ð8Þ whe e q ( ) is an indica o unc ion o whe he plo is ac i e in yea , and g a, ( ) and g s, ( ) indica e i an age-da ed (sca -da ed) i e was eco ded on plo in yea . Fo he ac ual pa ame e s o in e es , a,b,q, and c i , we used non-in o ma i e p io s. In o ma- i e p io s we e used o he pa ame e s o he da ing unce ain y ke nel o he age-da ed i es. We used log-no mal p io s o k 0 ,k 1 , and k 2 , wi h meanso 7.4,2.7,and0.6yea s,and95% con idence limi s o 2.7–20.1, 1.0–7.4, and 0.37– 1.0 yea s, espec i ely. A no mal p io was used o d s , wi h mean ¼0 and s anda d de ia ion ¼ 0.5. The pos e io dis ibu ions o he eg ession pa ame e s we e calcula ed wi h an i e a i e me hod using an adap i e Ma ko chain Mon e Ca lo algo i hm (Haa io e al. 2005) implemen ed in Cþþ. RESULTS Signs o i es we e omnip esen in he s udied landscape; all he 85 s udy plo s had been bu ned a leas once du ing he p e ious 220 yea s (Fig. 3). Sca e ed i e sca s could be ound in app oxima ely hal o he s udy plo s, while pos - i e egene a ion coho s and cha ed wood p o ided e idence o s and- eplacing i es in he es o he plo s. None o he sample plo s happened o all in ecen ly bu ned a eas ( hough some we e obse ed be ween plo s), wi h he mos ecen i e sampled occu ing in 1981. The a e age annually bu ned a ea o e he whole www.esajou nals.o g 7May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL. s udy a ea du ing he las wo cen u ies has been 1.3%, co esponding o a i e cycle o 77 yea s (Fig. 4). Theoccu enceo o es i eshas a ied conside ably du ing he las wo cen u ies. The o emos change was he dec easing end in he Fig. 3. Annual ee ing and i e ch onologies o s udy plo s a anged acco ding o he ea lies yea o in e ed i e occu ence. The age o oldes li ing ees is shown by da k g ay ba s, and i e yea s a e indica ed by black diamonds. Ligh g ay ba s deno e dead wood samples wi h cha ed b anch ips belonging o a p e ious ee gene a ion. www.esajou nals.o g 8May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL. annually bu ned a ea o e he whole pe iod. A e age annually bu ned a ea dec eased om 2.0%in he i s hal o he 19 h cen u y o 0.33% in he la e hal o he 20 h cen u y, co espond- ing o i e cycles o 50 and 300 yea s, espec i ely. Fi es dec eased especially a e he mid-19 h cen u y. In he 1940s he e was an in e media e peak in i es a e which a apid u he decline in annually bu ned a eas ollowed (Fig. 4). In he sou he n pa o he s udy a ea (in B i ish Columbia and Albe a) ou econs uc ion is in good ag eemen wi h i e s a is ics o he a ea excep o he 1940s whe e ou econs uc- ion shows conside ably highe bu ned a eas (Fig. 5A). In he whole s udy a ea he 1960s and 1970s s a is ics and econs uc ions show simila alues bu in he 1990s s a is ics epo a e age annually bu ned a eas ha a e highe han ou uppe con idence limi (Fig. 5B). The ull eg ession model was i s i ed o he pe iod 1900–2000, o which all he clima ic ac o s we e a ailable. O he clima ic ac o s, GiR and MDC we e posi i ely, and he PDO index nega i ely, co ela ed wi h he bu ned a ea, indica ed by co esponding eg ession pa ame e s ha we e posi i e (o nega i e, espec i ely) wi h a leas 95%pos e io p oba- bili y (Fig. 6). Since all clima ic p edic o s we e no malized o ha e a a iance o 1, he magni- udes o he co esponding eg ession pa ame- e sshow he ela i eimpo anceo he p edic o s. The GiR index appea s o be he single bes clima ic p edic o o bu ned a ea. Acco ding o he pos e io dis ibu ions, he eg ession pa ame e s o JJA (c JJA ) and oad building da e (q) we e equally likely o be posi i e o nega i e, and hence did no show an e ec on he bu ned a ea (Fig. 6A). Also, sample plo dis ance om oads (0.2–1.0 km) did no co ela e wi h he bu ned p opo ions o he a ea (p¼0.7, Spea man’s ank es ). Despi e he co ela ion be ween he clima ic ac o s and bu ned a ea, he clima ic ac o s did no explain he dec easing end in i es. The esidual end ep esen ed by bwas clea ly nega i e bo h o he pe iod 1900–2000 and 1800–2000 (wi h pos e io p obabili y .0.9999, Fig. 6A and 6B, espec i ely). Indeed, none o GiR, MDC, and PDO co ela e signi ican ly wi h yea (p .0.25 o each, Spea man’s ank es ). I is no ewo hy ha he summe empe a u e index JJA does ha e a signi ican posi i e end (p ,10 6 ), bu i does no ha e explana o y alue in he eg ession model. This indica es ha on a yea ly le el, JJA is no nega i ely co ela ed wi h i es, which would be necessa y o i o explain he end in i es. Fig. 4 shows he a e age yea ly bu ned p opo ion o he pe iod 1800–2000, as p edic ed by he eg ession model. The p edic ion uses he maximum pos e io densi y alues o he pa ame e s, es ima ed o he pe iod 1800–2000 (Fig. 6B). Road building da e and JJA we e le ou o he p edic i e model, based on he analysis o he pe iod 1900–2000, and PDO and MDC we e no used o he 19 h cen u y o which hey we e no a ailable. The p edic ion alls well inside he 95%con idence en elope o he empi ical econs uc ion. Rega ding he da ing e o , ep esen ed by k 0 , k 1, k 2, and d s he analysis ga e clea addi ional in o ma ion only abou he sys ema ic sca da ing e o , d s , ha had i s highes pos e io densi y a 0.5 (Figs. 6A and 6B). This means ha he i es da ed based on i e sca s a e equally likely o ha e occu ed du ing he p e ious as he cu en ly gi en yea . DISCUSSION Signs o pas i es we e ubiqui ous in he la ge ac s udied in no hwes Canada despi e he la ge p opo ion o we lands (20%). In he 19 h cen u y a e age annually bu ned p opo ion o he landscape a ied be ween 3%and 1%in di e en decades, co esponding o i e cycles o 30 o 100 yea s (Fig. 3); o e he en i e 19 h cen u y, he a e age i e cycle was app oxima ely 60 yea s. These alues all well wi hin he ange o his o ical i e cycles epo ed in bo eal No h Ame ica, despi e he mos ly mo e sou he ly loca ion o such s udies (Zhang and Chen 2007). Mos s udies om bo eal Fennoscandia o Cen al Sibe ia ha e also sugges ed pas i e cycles o be ween 30 and 100 yea s, o somewha longe (Niklasson and G ans o¨m 2000, D oby- she e al. 2004, Wallenius e al. 2011). I seems ha despi e di e ences in dominan ee species, his o ical i e e u n in e als ac oss la ge pa s o he ci cumbo eal o es s ha e been less han 100 yea s, wi h he long- e m and la ge-scale a e ages being a ound 50–80 yea s. Howe e , i www.esajou nals.o g 9May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL. Fo es Resea ch 40:2027–2035. Wallenius, T., M. La ja aa a, J. Heikkinen, and O. Shibis o a. 2011. Declining i es in La ix domina ed o es s in no he n I ku sk Dis ic . In e na ional Jou nal o Wildland Fi e in p ess. Wei , J. M. H., E. A. Johnson, and K. Miyanishi. 2000. Fi e equency and he spa ial age mosaic o he mixed-wood bo eal o es in wes e n Canada. Ecological Applica ions 10:1162–1177. Wo ld Da a Cen e o Paleoclima ology, In e na ional T ee Ring Da a Bank. 2010. hh p://www.ncdc.noaa. go /paleo/ ee ing.h mli Ya ie, J. 1981. Fo es i e cycles and li e ables: a case s udy om in e io Alaska. Canadian Jou nal o Fo es Resea ch 11:554–562. Zack isson, O. 1977. The in luence o o es i es in he No h Swedish bo eal o es . Oikos 29:22–32. Zhang, G. F., and W. J. Chen. 2007. Fi e cycle o he Canada’s bo eal egion and i s po en ial esponse o global change. Jou nal o Fo es y Resea ch 18:55– 61. www.esajou nals.o g 16 May 2011 Volume 2(5) A icle 53 WALLENIUS ET AL.