The legacy of logging - Estimating arboreal lichen occurrence in a boreal multiple-use landscape on a two century scale

The Legacy o Logging—Es ima ing A bo eal Lichen Occu ence in a Bo eal Mul iple-Use Landscape on a Two Cen u y Scale Tim Ho s ko e 1 *, Jon Moen 1 , Tomas La ¨ma ˚s 2 , Timo Helle 3 1Depa men o Ecology and En i onmen al Science, Umea ˚Uni e si y, Umea ˚, Sweden, 2Depa men o Fo es Resou ce Managemen , Swedish Uni e si y o Ag icul u al Sciences, Umea ˚, Sweden, 3Finnish Fo es Resea ch Ins i u e, Ro aniemi, Finland Abs ac In no he n Sweden, he a ailabili y o a bo eal lichens (B yo ia uscescens, Alec o ia sa men osa) as win e g azing esou ces is an impo an elemen in eindee husband y. Wi h he indus ializa ion o o es y, o es s ich in a bo eal lichens ha e diminished conside ably. He e, we analyze how o es y has impac ed lichen a ailabili y om he 1920’s o he p esen day and model i s u u e de elopmen assuming di e en o es managemen scena ios. We eco ded he cu en occu ence o B. uscescens in 144 sampling plo s, s a i ied by o es age class and dominan ee species in a 26,600 ha bo eal o es landscape ha is used o bo h eindee he ding and o es y. Lichen abundance was isually es ima ed in ou classes: none, spa se, mode a e and abundan . A bina y logis ic model using o es age as he independen a iable was de eloped o p edic he p obabili y o lichens being p esen . Using his model, we ound ha lichens we e p esen in s ands ha a e a leas 63 yea s old. Because o he ela i e pauci y o s ands ich in a bo eal lichens, i was no possible o eliably de e mine how age a ec s he a ia ion in abundance o olde o es s ands. The his o ical de elopmen o o es s whe e a bo eal lichens could po en ially occu was s udied using his o ic o es y eco ds da ing back 80 yea s. Be ween 1926 and he p esen day, o es y has educed he co e o o es s olde han 60 yea s om 84% o 34%. The likely u u e spa ial co e age o hese s ands o e he nex 120 yea s was es ima ed o wo di e en managemen scena ios and an unmanaged e e ence scena io, using he Heu eka s a egic planning p og am. Unde bo h he ‘‘business as usual’’ scena io and ha in ol ing mo e in ensi e o es y, con inued dec eases in lichen a ailabili y a e p ojec ed. Ou esul s emphasize he impo ance o al e na i e o es y p ac ices, such as p olonged o a ion pe iods, o inc ease he a ailabili y o a bo eal lichens as a g azing esou ce o eindee . Ci a ion: Ho s ko e T, Moen J, La ¨ma ˚s T, Helle T (2011) The Legacy o Logging—Es ima ing A bo eal Lichen Occu ence in a Bo eal Mul iple-Use Landscape on a Two Cen u y Scale. PLoS ONE 6(12): e28779. doi:10.1371/jou nal.pone.0028779 Edi o : Da en Ma k E ans, Uni e si y o Hull, Uni ed Kingdom Recei ed July 7, 2011; Accep ed No embe 15, 2011; Published Decembe 16, 2011 Copy igh : ß2011 Ho s ko e 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 un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal au ho and sou ce a e c edi ed. Funding: This wo k was suppo ed by g an s om he Swedish Resea ch Council o En i onmen , Ag icul u al Sciences and Spa ial Planning FORMAS (510057203) (h p://www. o mas.se) o JM and he ‘‘Gunna and Ru h Bjo ¨ kmans und o no he n bo anical esea ch’’ (Dn 223-1 129-09) o T. Ho s ko e. The unde s had no ole in s udy design, da a collec ion and analysis, decision o publish, o p epa a ion o he manusc ip . Compe ing In e es s: The au ho s ha e decla ed ha no compe ing in e es s exis . * E-mail: im.ho s ko [email protected] In oduc ion Many na u al esou ces a e simul aneously exploi ed by se e al ac o s, including indigenous peoples and indus y, some imes wi h ensuing con lic s. Such mul iple-use si ua ions equi e collabo a- i e managemen s a egies in ol ing all o he a ec ed s akehold- e s i he ocal esou ces a e o be sha ed sus ainably and in a way ha minimizes ad e se en i onmen al impac [1], known as co- managemen . Whe e indigenous peoples a e in ol ed, hei poli ical au onomy is seen as equi ed o hei cul u al iden i y and e ec i e con ibu ion esou ce managemen [2]. While indigenous peoples o en accoun o a mino i y o he o al popula ion, Lane [3] a gues ha in wes e n pos -se le s a es, in which he descendan s o o me colonize s now o m he ‘‘na ion’’, he p ima y challenge o esou ce sha ing is less a p oblem o app o ing indigenous igh s and sel -de e mina ion. I is a he he easonable and ealis ic way o alloca ing esou ces be ween di e en claiman s in planning and adminis a i e p ocesses. The con as ing demands placed upon na u al esou c- es, as pe cei ed by indigenous peoples and o he ex ac o s necessi a e a ade-o be ween comme cial uses o a landscape and o he alues, such as social, cul u al and biological alues [4]. This is he si ua ion in he bo eal o es s o No he n Sweden, whe e eindee husband y (Rangi e . a andus L.) is p ac iced by he indigenous Sami people. Th oughou no he n Fennoscandia, eindee husband y is o high cul u al ele ance o he Sami. In Sweden, his o m o pas o alism, mig a ing be ween summe g azing g ounds in he wes e n Scandic moun ains and win e g azing in bo eal o es s nea he Bal ic coas , became es ablished in he 17 h cen u y, albei in a di e en o m han p ac iced oday [5]. E e since, eindee husband y has unde gone nume ous changes and adap a ions [6-8]. Du ing he las cen u y in pa icula , i s demands ha e inc easingly come in o con lic wi h hose o o he o ms o land use. Fo example, he cons uc ions o hyd oelec ic powe plan s and dams subme ged aluable pas u e lands, cal ing g ounds and adi ional mig a ion ou es be ween summe and win e pas u es [9]. Mo eo e , clea -cu ing o o es s was in oduced on a la ge scale in no he n Sweden du ing he 1950’s and was subsequen ly made manda o y by he Fo es y Ac o 1979 [8] o eplace na u al i e-d i en dis u bance dynamics. PLoS ONE | www.plosone.o g 1 Decembe 2011 | Volume 6 | Issue 12 | e28779 Howe e , mode n clea -cu ing is a mo e ex ensi e han any na u al dis u bance pa e n [10]. Selec i e logging o o es s, which used o be ela i ely common, esul s in slowe ee egene a ion han does clea -cu ing wi h subsequen soil p epa- a ion, o ins ance by soil sca i ica ion. As a consequence, managed o es s ha e become much dense , while he mean age o he ees wi hin hem has dec eased conside ably [11]. Al hough hese changes inc ease he p o i abili y o imbe p oduc ion, his indus ialized o es y has had ce ain ad e se ecological conse- quences and has elimina ed ce ain niches [12]. Impo an ly o eindee husband y, his sil icul u al in ensi ica ion has dec eased he abundance o he mos impo an win e o age o eindee , i.e. e es ial and a bo eal lichens [13,14], and deg aded he ca ying capaci y o win e g azing g ounds [15]. In win e , snow co e a ec s he a ailabili y o o age [13,16] and he ene gy expended by he dee while digging o lichens benea h he snow [17]. Unde ce ain ci cums ances, pas u es can become inaccessibly o eindee due o he o ma ion o hick ice c us s. This is pa icula ly common du ing la e win e and ea ly sp ing, a e p olonged ma u a ion o he snow co e o a e wea he e en s such as eeze- haw-cycles [18] o ain-on-snow [19]. The a ailabili y o a bo eal lichens such as B yo ia uscescens and Alec o ia sa men osa is essen ial o he mi iga ion o such c i ical si ua ions, du ing which he wind h ow o lichen-bea ing ees and he li e all o a bo eal lichen agmen s becomes an impo an sou ce o eindee o age [20,21]. A bo eal lichens a e mo e abundan in old-g ow h o es s [22,23], which ha e become signi ican ly less widesp ead in No he n Sweden as a esul o speci ic o es managemen p ac ices [24,22]. Dec eases in he abundance o a bo eal lichens can inc ease he need o supplemen a y eeding and/o necessi a e a educ ion in he d size [25], especially du ing win e s wi h ad e se snow condi ions. As epo ed by Helle & Jaakkola [7], abundan a bo eal lichens ha e his o ically sus ained eindee he ds h oughou he win e , e en unde highly ad e se snow condi ions. Swi ching o a bo eal lichens he e o e was an adap i e s a egy ha allowed eindee o o e come sho ages o g ound lichens. Consequen ly, he capaci y o eindee husband y o adap o ad e se snow condi ions inc eases wi h he ecological a ie y o he a ailable pas u es [26]. The ansi ion o a highly agmen ed and, on he s and le el, less di e se landscape has nega i ely a ec ed he abundance o a bo eal lichens, p ima ily due o hei slow g ow h, low dispe sal a es, and s ong dependence on he a ailabili y o sui able g owing subs a es [27]. Today, he educed a ailabili y o his esou ce has s ong implica ions o he sus ainabili y o eindee husband y [28,7], i.e. he abili y o main ain i s p oduc i i y in he ace o dis u bance e en s [29]. He e, we in es iga e how he ans o ma ion o he bo eal landscape o e he las cen u y (1926–2006) by comme cial o es y has a ec ed a bo eal lichens in a eindee he ding a ea, and simula e possible u u e condi ions. We do his by analyzing how he po en ial habi a o a bo eal lichens has changed o e ime by (i) iden i ying ela ionship be ween o es age and a bo eal lichen abundance, and by (ii) analyzing he change in hese habi a ac o s due o o es managemen o e he las 80 yea s. Finally (iii), we simula e he u u e de elopmen o he po en ial a bo eal lichen habi a o e he nex 120 yea s unde a ious di e en scena ios wi h di e ing in ensi ies o o es y. Ma e ials and Me hods S udy si e Loca ed a la i ude 66u209N, mid-bo eal ege a ion cha ac e - izes he s udy a ea o ca. 26,600 ha (Fig. 1). I s o es s a e domina ed by Sco s pine (Pinus syl es is), wi h a ew s ands domina ed by No way sp uce (Picea abies). Deciduous ees, such as bi ch (Be ula pendula), goa willow (Salix cap ea), and aspen (Populus emula), a e p esen bu no common. The g ound laye consis s o dwa sh ubs (Junipe us communis, Vaccinum my illus, Vaccinum uliginosum, Empe um nig um), mosses, and g ound lichens (Cladonia spp., Ce a ia spp.) g owing on d y glacio- lu ial soils. Se e al small lakes and mi es a e sca e ed ac oss he landscape, co e ing ca. Figu e 1. Loca ion o he s udy a ea. The g ay a ea in he o e iew map illus a es he Swedish eindee he ding a ea. doi:10.1371/jou nal.pone.0028779.g001 E ec s o Logging on A bo eal Lichen Occu ence PLoS ONE | www.plosone.o g 2 Decembe 2011 | Volume 6 | Issue 12 | e28779 24% o he a ea. The i e Pi e A ¨l en uns h ough he a ea, and was o me ly an impo an means o anspo ing imbe . The si e’s ele a ion a ies om 277 m abo e sea le el along he i e alleys o 550 m a mo aine hill ops. The s udy a ea, which consis s o o me Na ional Fo es s Akkajau and Ab au (he ea e e e ed o collec i ely as Akkajau ), is used o yea - ound eindee g azing by he Sami he ding dis ic s O ¨s a Kikkejau and S a˚kke, and as a mig a ion co ido and win e g azing a ea by he Luok a-Ma as he ding dis ic . In 1936, Akkajau was ex ended no hwa ds, inc easing i s a ea by ca. 17% (4,620 ha) o i s cu en size. Da a collec ion A au´jo & Williams [30] iden i ied local-in insic ac o s, such as sui able habi a , and egional-in insic ac o s, such as coloniza ion om o he a eas based on he connec i i y o simila habi a s wi hin he landscape, as essen ial o he pe sis ence o a species. In his s udy, we ocus on local-in insic aspec s, by which we mean he cha ac e is ics o he o es a he s and le el. Field wo k was conduc ed in Augus 2009 and 2010. Using digi ized maps o he s udy a ea in a geog aphical in o ma ion sys em (GIS), we s a i ied o es s ands in o 9 classes acco ding o hei age (young = 1–39 y s, ma u e = 40-120 y s, and old = .120 y s) and species composi ion (pine, sp uce, mixed) on he basis o in en o ies conduc ed by he o es ’s owne , S easkog (Table 1, see also 11). A s and was deno ed as a pine o sp uce s and i he species in ques ion accoun ed o 80% o mo e o he s ems. The sampling uni s we e posi ioned be ween 500 m and 20 m away om he nea es o es oad so as o a oid oad e ec s. Wi hin his bu e zone, o es s ands we e selec ed by andom sampling. Ci cula sampling plo s, co e ing ca 0.1 ha, we e semi- andomly dis ibu ed in he selec ed o es s ands o he 9 s a a (Table 1). Plo s we e loca ed in he ield wi h a GPS ecei e . A ha poin , he basal a ea o he espec i e o es s and was measu ed using a elascope, and he heigh o a ep esen a i e ee in he s and was eco ded. The canopy closu e a each sampling plo was calcula ed om a single digi al hemisphe ic pho og aph aken a each plo wi h a ish eye lens, a e con e sion o bina y images using he p og am Gap Ligh Analyze [31]. As B yo ia uscescens is p e e ed o age by eindee [20], we ocused on his species. Alec o ia sa men osa, which was also p esen in he a ea in low abundance, p e e s la e successional sp uce and mois e habi a s ands han he d y pine o es s o Akkajau [32,33]. The biomass o B yo ia uscescens in he s and was es ima ed isually and classi ied in o one o 4 abundance ca ego ies, ep esen ing mean alues o a bo eal lichen biomass: none (0), spa se (1; #35 kg ha 21 ), mode a e (2; 35–120 kg ha 21 ), and abundan (3; $120 kg ha 21 ). This classi ica ion sys em was based on he indings o Helle e al. [34], who used an o en-d ied lichen clump o known mass o classi y lichen biomass using hese ou ca ego ies. As he clump me hod unde es ima es biomass o a bo eal lichens in ees, he bias in oduced by isual es ima ion has o be co ec ed. In he me hod o Helle e al. [35], his was done by using eg ession equa ions, ha compa e he es ima ed biomass on a b anch o he o en-d ied biomass emo ed om he same b anch. S ands con aining no lichens we e assigned o ca ego y 0. Ca ego y 1 was assigned o s ands in which indi idual lichen halli we e de ec ed on a small numbe o ees and in low quan i ies. S ands in which lichen clumps we e eadily appa en on each ee we e assigned o ca ego y 2; none o he s udied s ands con ained enough lichens o be assigned o ca ego y 3. Lichen abundance was hus a e aged o e he ees wi hin he sampling plo , and his a e age was aken as he s and le el. Using he same me hod, Jaakkola e al. [35], ound a maximum lichen biomass on whole ees o 474 kg ha 21 a esh si es in ma u e sp uce o es s. In o de o o ally elimina e edge e ec s, e.g. om neighbo ing open a eas, i would ha e been necessa y o ocus exclusi ely on sampling plo s wi h a dis ance o 50 m in e e y di ec ion o neighbo ing s ands [33], which would equi e a minimum a ea o 0.79 ha assuming a ci cula o es s and. Howe e , he o es pa ches in he s udy a ea we e no always shaped in a way ha would make his possible, especially in he olde o es s ands. Howe e , sampling plo s in close p oximi y o open places such as la ge lakes o mi es we e no common in ou da ase . In o al, we sampled 144 o es s ands. One hi d o he sample plo s we e in ma u e pine o es s, 19% we e in old pine o es s, and 17% we e in young pine s ands. The emaining plo s we e loca ed in less abundan o es ypes (Table 1). S ands domina ed by sp uce we e especially a e. I has p e iously been epo ed ha he e is a s ong ela ion be ween o es age and a bo eal lichen biomass [33], al hough o he au ho s ha e sugges ed ha he co ela ion wi h olume is s onge [35]. S and age was s ongly co ela ed wi h many o he o es cha ac e is ics speci ied in he o es owne ’s s and egis y (Table 2) and wi h he canopy closu e calcula ed om ou ish – eye pho o analysis. We chose o use age as a p edic o a iable in ou models, since i can easily be compa ed ac oss di e en s udies and is eco ded in mos o es da abases. Pe mission o conduc ing he s udy was gi en by he o es owne S easkog. No endange ed o p o ec ed species we e in ol ed in he s udy. S a is ical analysis We used bina y logis ic eg ession analysis o model he abundance o lichens in ela ion o o es s and age. This was done by con e ing he a bo eal lichen abundance classes in o dicho omous da a. Two di e en app oaches we e adop ed: compa ing plo s in which no lichens we e p esen (i.e. hose in ca ego y 0) o hose wi h a a lichen class o a leas 1, and Table 1. Numbe o sampled o es s ands in s a a, sepa a ed by dominan ee species and age ( o al o es s ands N = 852, sampled o es s ands n = 144). Pine Sp uce Mixed S a um o al sampled % o sampled S a um o al sampled % o sampled S a um o al sampled % o sampled Young 177 24 17 2 1 1 35 10 7 Ma u e 352 48 33 1 1 1 44 12 8 Old 177 28 19 13 7 5 51 13 9 doi:10.1371/jou nal.pone.0028779. 001 E ec s o Logging on A bo eal Lichen Occu ence PLoS ONE | www.plosone.o g 3 Decembe 2011 | Volume 6 | Issue 12 | e28779 compa isons be ween plo s wi h a lichen class o 1 and hose wi h a lichen class o 2. In a logis ic eg ession analysis, he dependen dicho omous a iable x(i.e. he p esence [1] o absence [0] o lichens) is no modeled di ec ly. Ins ead, he eg ession es ima es he p obabili y p(x) ha he ou come will be ‘‘posi i e’’, i.e. ha lichens will be p esen in a pa icula plo o a gi en alue o he independen a iable. A logis ic eg ession akes he o m: p(x)~ eazbx 1zeazbx Bo h he cons an aand he p edic o coe icien ba e i ed by he logis ic model; he coe icien bgi es he slope o he eg ession. The ou pu o he model anges om 0 o 1 o each s and, wi h an inc ease in he p obabili y, p, co esponding o a posi i e ou come. The model’s coe icien s a e mos easily in e p e ed in e ms o odds a ios (OR). An OR measu es how likely an e en (such as he p esence o lichens) is ela i e o i s opposi e (in his case, he absence o lichens) o any gi en alue o he independen a iable. The OR hus indica es he a e o change in p(x) o a one uni change in he independen a iable x[36,37]. An OR la ge han 1 indica es a posi i e ela ionship be ween he dependen and independen a iables; alues below 1 indica e he opposi e. The odds a io is ob ained as ollows: OR~ebx Model i , cu -o selec ion and model alida ion We in es iga ed he models’ goodness-o - i , i.e. he accu acy o hei classi ica ion o occupied and unoccupied si es, using a Recei e Ope a ing Cha ac e is ic cu e (ROC). This cu e is cons uc ed using he ac ion o co ec ly classi ied e en s, i.e. he ue posi i e ac ion (sensi i i y) and ue nega i e ac ion (speci ici y) o e e y h eshold alue ha could po en ially be used o disc imina e be ween occupied and unoccupied si es. Plo ing he sensi i i y agains (1 – speci ici y) o all possible alues calcula ed by he model, a ROC cu e is de i ed [38,39]. This cu e illus a es he model’s abili y o disc imina e be ween occupied and unoccupied si es o e he whole ange o possible h eshold alues ha could be used o seg ega e he da a se . I he model is unable o disc imina e be ween he si e ypes, he ROC - cu e will be a s aigh line wi h a g adien o 45uand he model ou comes a e no mo e eliable han chance. The close he cu e is o he le and op axes, he be e i s disc imina i e abili y and he be e he model i [40]. Pea ce & Fe ie [38] sugges ha he a ea unde he ROC cu e (AUC) is an indica o o he model’s disc imina i e abili y; an AUC be ween 0.5 and 0.7 indica es weak disc imina i e abili y, while a alue be ween 0.7 and 0.9 indica es easonable disc imina ion and alues exceeding 0.9 indica e a e y good abili y o disc imina e be- ween occupied and unoccupied pa ches using he bina y classi ica ion scheme. The model alues calcula ed using bina y logis ic eg ession, ep esen con inuous p obabili ies o occu ence. To classi y a o es ype as being occupied o unoccupied by lichens, he model alues ha e o be spli a a ce ain h eshold alue. The ROC cu e illus a es he comp omises ha a e made be ween he ue and alse posi i es as he decision h eshold is changed, making i possible o iden i y an op imal h eshold using he sensi i i y and speci ici y [38]. Ou s a egy o inding he mos obus cu -o poin was o spli he da a se a he poin whe e he sum o he sensi i i y ( ue posi i es) and he speci ici y ( ue nega i es) is a i s maximum, i.e. he alue ha gi es he highes a es o co ec classi ica ion o occupied and unoccupied si es. The model’s pe o mance was es ed using c oss- alida ion [39,41]. The da a se was andomly di ided in o wo pa s - a aining se consis ing o 80% o he da a and a es se consis ing o he emaining 20%. To assess he consis ency be ween he p edic ions o he aining model and he es model ega ding he p esence o lichens, he co esponding model ou comes we e binned in o 10 equal-sized classes (0-0.1; 0.1-0.2;…; 0.9-1). A s able model would be expec ed o ha e a simila p opo ion o i s o al ou comes in each bin o bo h he aining and he es se . Signi ican Spea man’s Rank co ela ions ( s ) be ween he equencies o he wo se s indica e ha he model is applicable o bo h he aining and he es se [39]. We epea ed he p ocedu e o andom da a-pa i ioning and co ela ion o bin- equencies ou imes, and calcula ed he a e age co ela ion coe icien and signi icance o he i e esul ing co ela ions. Table 2. Two- ailed Spea man’s ank co ela ions ( s ) be ween sample s and cha ac e is ics, n = numbe o sampled s ands. Fo es Age (y s) Volume (m 3 /ha) Basal A ea (m 2 ) Diame e (cm) Heigh (m) Canopy Closu e (%) Fo es Age 1- - --- n144 Volume 0,776 ** 1- --- n144 144 Basal A ea 0,512 ** 0,898 ** 1--- n134 134 134 Diame e 0,898 ** 0,712 ** 0,435 ** 1-- n125 125 125 125 Heigh 0,856 ** 0,904 ** 0,654 ** 0,878 ** 1- n144 144 134 125 Canopy Closu e 0,561 ** 0,611 ** 0,526 ** 0,304 ** 0,556 ** 1 n144 144 134 125 144 144 **Co ela ions a e signi ican a he 0.01 le el. doi:10.1371/jou nal.pone.0028779. 002 E ec s o Logging on A bo eal Lichen Occu ence PLoS ONE | www.plosone.o g 4 Decembe 2011 | Volume 6 | Issue 12 | e28779 Compila ion o his o ical o es age da a His o ical o es s in en o ies o he yea s 1926, 1936, 1960 we e ob ained om a chi al sou ces [11], and he mos ecen in en o y, o 2006, was ob ained om he o es company S easkog. These in en o ies di e in bo h he quali y and quan i y o he epo ed o es cha ac e is ics. Fo ins ance, he way in which o es age is calcula ed di e s, as he scope o o es y shi ed om selec i e cu ing o clea -cu ing in he 1950s [11]. While he mean age o o es s ands is a ailable o he mos ecen su ey yea (2006), olde da a om 1926, 1936 and 1960 had o be in e p e ed in di e en ways. The 1926 and 1936 in en o ies eco ded he age o each o es s and as a composi e o 5 age classes, each co e ing a 40-yea in e al (1-40 yea s, 41-80 yea s, …, 120-160, .161 yea s). These sou ces did no p o ide any g anula i y wi h espec o he ages o s ands olde han 161 yea s because o es s o ha age we e le unmanaged. The olde sou ces p o ide da a on he pe cen age o he a ea o each s and co e ed by ees in each age b acke . This lexible classi ica ion scheme e lec s he a iable age s uc u e ypical o o es pa ches a ha ime. Using hese pe cen ages and he mean o each 40 yea - age class, an a ea-weigh ed a e age age was calcula ed o each o es s and. The hypo he ical maximum age o a s and ha could be calcula ed using his scheme would be 161 yea s, o a s and ha was en i ely co e ed by ees wi hin he highes age b acke . The e o e, all age es ima ions we e cons ained by his uppe limi , e en hose de i ed om mo e ecen in en o ies in which olde mean ages we e epo ed. The o es age da a o he 1960 in en o y a e epo ed using a classi ica ion scheme consis ing o en age classes a he s and le el, each o which spans 20 yea s. This e lec s a change in o es managemen p ac ices ha a o ed mo e e enly-aged s ands [24]. Due o hese es ic ions, o es ages de i ed om he his o ical in en o ies should be ega ded as app oxima ions o he eal si ua ion in hose imes. Simula ion o u u e managemen scena ios We chose h ee possible scena ios o u u e o es managemen , all di e ing in hei impac on o es age s uc u e and hus on po en ial a bo eal lichen abundance. The i s was a ‘‘business as usual’’ scena io (BAU), in which i is assumed ha he o es s will be managed as hey cu en ly a e by hei owne s, wi h he in en o sa is ying con empo a y imbe managemen and na u e conse a ion objec i es. The second in ol es mo e in ensi e o es y (INT), while he hi d in ol es a o al absence o managemen , i.e. no o es y wha soe e (NO). All h ee scena ios we e de eloped using he S andWise so wa e package, which is pa o he Heu eka sys em. This sys em was de eloped by he Swedish Uni e si y o Ag icul u al Sciences (SLU) o o es managemen planning and analysis on se e al scales [42,43]. Fo BAU and INT he economic ou come o o es y was maximized in e ms o he Ne P esen Value (NPV), i.e. he sum o all discoun ed u u e ne cos s and incomes ( o a eal in e es a e o 2,5%). An e en low o imbe ha es s was ensu ed by penalizing luc ua ions in ha es s in he op imiza ion p oblem goal unc ion. The bes managemen al e na i e among he se o po en ial managemen al e na i es o each s and was ound by sol ing he op imiza ion model using linea p og amming. The beha io displayed unde he h ee di e en scena ios was subjec ed o he age dis ibu ion o o es s ands gi en in 2006. The ollowing assump ions we e made o he di e en scena ios: NIn BAU, a eas cu en ly unde p o ec ion we e le unman- aged, as we e o es s olde han 160 y s ( o al co e ed a ea: 2268 ha). Fo s ands olde han 120 yea s, o a ion pe iods we e p olonged and hey we e no allowed o be inal elled un il a e he nex en yea pe iod ( o al co e ed a ea: 1848 ha); comple e inal elling h oughou his second a ea immedia ely a e he end o hese en yea s was p e en ed by he e en imbe low c i e ion in he goal unc ion. O he s ands we e assumed o be managed acco ding o hei cu en managemen egimes in e ms o hinning p og ams and inal elling, including 1-2 hinning e en s and a inal elling age o a ound 100 yea s. The minimum inal elling age was ha s ipula ed by he Fo es y Ac , i.e. ypically 90 yea s. NINT assumes ha all s ands can be cu , including hose in p o ec ed a eas o ha a e mo e han 160 yea s old. S ands olde han 120 yea s we e allowed o be cu immedia ely, which esul ed in sho e ini ial o a ions compa ed o he BAU scena io. NNO is a simple ex apola ion o o es aging in o he u u e, wi hou conside ing any dis u bances o pa ch dynamics, whe he human o na u al in o igin. In eali y, unmanaged o es s a e subjec o na u al dis up ing e en s such as o es i es o insec b eakou s, which mean he o es s a e main ained in a dynamic s a e wi h di e se succession phases. Assessmen o landscape agmen a ion To analyze he deg ee o agmen a ion, we imposed a bu e o 200 m a ound o es s ands iden i ied as h eshold pa ches by he model (see abo e). This bu e was based on he po en ial o dispe sal o a bo eal lichens as epo ed by De ki & Esseen [44]. In o he sou ces, he dispe sal dis ance has been epo ed o ange om 100 m, wi h he numbe o dispe sed halli dec easing by 50% om old g ow h o es s [45], o 350 m om o es s ands wi h high lichen abundance [46]. We also calcula ed he la ges pa ch size index (LPI) in F ags a s [47] o quan i y he change in he a ea co e age o connec ed h eshold s ands iden i ied by he eg ession model. The gene al agmen a ion o all o es s ands p esen in Akkajau due o o es y was calcula ed as he a ea-weigh ed a e age s and size (S a ). This is an app op ia e me hod o a e aging when he equency dis ibu ion o la ge and small s ands is skewed [48]. Using A i o e e o he size o he i- h s and, i is calcula ed as Sa~ S(Ai2) S(Ai) Because we compa ed changes in he age dis ibu ion wi hin he o es o e se e al decades, i was necessa y o ensu e ha he da a o di e en yea s was compa able. Adminis a i e decisions and he e ec s o o es y ha e caused he numbe o o es s ands, hei spa ial con igu a ion, and he size o he s udy a ea o change o e he s udied pe iod o ime. The e o e, changes in he size and quan i y o o es s ands o a gi en age ha e o be conside ed o p ope ly unde s and he o es y-d i en landscape dynamics o he a ea and hei in luence on he beha io o he logis ic eg ession model. Fo ins ance, la ge s ands o young ees con ibu e mo e o he landscape con igu a ion han many small s ands o much olde ees, bu he la e would ha e a mo e p onounced in luence on he eg ession model because o hei age. To accoun o hese empo al changes in he spa ial con igu a ion o Akkajau in e ms o he p opo ion o s ands o di e en ages and hei in luence on he model, we calcula ed an ‘‘a ea-weigh ed model’’. In his model, he aw modeled alue o a gi en s and, being only depended on s and age, is mul iplied by i s ela i e size as a pe cen age o he o al s udy a ea. This scales he pu ely age- dependen model alue o a gi en o es s and acco ding o he E ec s o Logging on A bo eal Lichen Occu ence PLoS ONE | www.plosone.o g 5 Decembe 2011 | Volume 6 | Issue 12 | e28779 magni ude o i s spa ial con ibu ion o he landscape pa e n. Al hough dimensionless, he a ea-weigh ing gi es a be e unde s anding o how he landscape may be ‘‘pe cei ed’’ as an a bo eal lichen habi a . All s a is ics we e compu ed using PASW S a is ics ( e sion 18). Resul s Modeling a bo eal lichen abundance in o es s ands Nea ly hal o he sampled s ands we e wi hou a bo eal lichens (45%). Low a bo eal lichen abundance (abundance class 1) was eco ded in 34% o all s ands, while a bo eal lichens we e mode a ely abundan in he emaining 21% (abundance class 2; Table 3). Bina y logis ic eg ession analysis success ully disc imina ed be ween s ands ha did and did no con ain lichens (i.e. be ween hose in class 0 and hose in ei he class 1 o class 2), using o es age as he independen a iable (Table 4). Exponen ia ing he eg ession coe icien o o es age ga e an odds a io o 1.04. This indica es ha he likelihood o he s and con aining lichens inc eases by 4% pe yea o age. Fig. 2 illus a es he inc ease in he p edic ed p obabili y o lichen occu ence (p) wi h o es age. No model capable o achie ing s a is ically-signi ican disc imina- ion on he basis o age be ween s ands in class 1 and class 2 could be iden i ied; all o he models examined exhibi ed low AUC alues (0.64), indica ing poo disc imina o y abili y (da a no shown). None o he o he o es a iables (Table 2) p oduced signi ican models ei he , p obably due o he low sample size o o es s in which B yo ia was abundan (N = 30, Table 3). All o he emaining analysis and discussion he e o e ocuses on he p esence-absence model. Model i , h eshold alue selec ion, and alida ion The la ge a ea unde he ROC cu e (AUC: 0.92; p,0.001, SE = 0.03) demons a es he model o be well- i ; i co ec ly disc imina es be ween s ands wi h and wi hou lichens on he basis o s and age in 92% o he cases examined. The sum o he sensi i i y ( ue posi i es) and speci ici y ( ue nega i es) peaked a a o es age o 63 yea s, which was used as he cu -o poin (Fig. 2). The eplica ed c oss- alida ion o he logis ic model esul ed in a highly signi ican Spea man-co ela ion (p,0.001; co ela ion coe icien s = 0.945) be ween he es and aining se s, con i ming he model’s obus ness. The abili y o he model o es ima e he his o ical p obabili y o lichen abundance is limi ed by he ac ha he his o ical sou ces used did no eco d he ages o ees olde han 160 yea s, which in oduces some bias in o he model. Howe e , he p obabili y ha a bo eal lichens will be p esen in s ands o ha age is 0.97, and so he a i icial limi imposed by he olde da a has only a small e ec on he es ima ed p obabili y ha lichens we e p esen in his o ical s ands. Age h eshold Ou model iden i ied he s and age h eshold beyond which lichens a e mo e likely o be p esen han absen as 63 yea s (Fig. 2). This cu -o poin is also easonable o apply in ligh o he his o ical da a, since all o he di e en age classi ica ion schemes used in he his o ical in en o ies included an age class o ees ‘‘olde han 60’’. We he e o e used a cu o o 60 yea s in he subsequen analyses, o e ain consis ency wi h he his o ical da a. O e ime, he p opo ion o o es s olde han 60 yea s dec eased, om 86% co e in 1926 o 34% in 2006 (Fig. 3). Consequen ly, he e is li le connec i i y o hese o es pa ches, which es ic ed he scope o lichen dispe sal, especially a e 1960. As hese olde o es s became less common and ceased o occu in close p oximi y o one-ano he , he bu e zone o 200 m a ound he o es pa ches inc eased ela i e o he a ea co e ed by hese olde s ands om 1960 onwa ds (Fig. 3). The la ges connec ed o es pa ch co e ed 50% o Akkajau in 1926, bu only 6% in 2006 (Table 5). The 74% dec ease in a ea-weigh ed mean s and size S a unde lines he inc easing isola ion o hese abo e- h eshold s ands om hei nea es simila ly-aged neighbo s (Table 5). Gene ally, a ia ion in s and numbe and size is a measu e o agmen a ion, since he c ea ion o mo e disc e e s ands in he same a ea spli s he a ea in o smalle pa s. A e he ex ension o Akkajau in 1936, he o e all s and numbe ose by 12% in 1960, and by ano he 28% in 2006 (Table 6). As a esul , he mean s and a ea dec eased by 40% be ween 1936 and 2006. The a ea co e ed by s ands olde han 140 yea s dec eased by 50% be ween 1926 and 2006. Fu u e scena ios We used he 60-yea h eshold o quan i y he consequences o h ee di e en u u e app oaches o o es managemen , one co esponding o in ensi e o es y, one o ‘‘business as usual’’, and one in which no managemen was unde aken. Fig. 3 illus a es he incidence o o es s ands olde han 60 yea s in he h ee scena ios. Unde he ‘‘business as usual’’ scena io, he a ea co e ed by s ands ha p o ide a sui able habi a o lichens will inc ease by abou 30% ela i e o he p esen day a e 40 yea s, as he many young s ands ma u e. Howe e , u u e ha es ing will educe his le el back o ha which cu en ly ob ains a e a pe iod o 120 yea s. Mo e in ensi e o es y is p ojec ed o educe he co e age o lichen s ands, as logging would be ocused on he olde , p e iously-p o ec ed o es s ands. This would only educe he size o he lichen habi a by 5% ela i e o he p esen day, bu would gene ally esul in a smalle lichen habi a han he BAU scena io o e ime. Con e sely, lea ing he a ea unmanaged would inc ease he a ea o he lichen s ands o almos hei o me ex en wi hin 40 yea s. Howe e , i should be no ed ha his scena io does no accoun o na u al dynamics such as o es i es and gap dynamics, which would also a ec lichen dispe sal, es ablishmen and abundance. Changes in age dis ibu ion and o es pa e ns In 1926, Akkajau was domina ed by olde s ands (.140 y s), which co e ed nea ly hal i s o al a ea (Fig. 4). This p opo ion had dec eased o only 9% in 2006, when he landscape was domina ed by s ands younge han 60 yea s (Fig. 4). Be ween 1926 and 1936, selec i e logging o olde ees educed mean s and age (Fig. 4). Ex ensi e clea -cu ing be ween 1936 and 1960 educed he a ea co e ed by all age classes, gi ing ise o a bimodal dis ibu ion o clea cu s/young s ands oge he wi h old s ands (Fig. 4). Table 3. Lichen classes and hei subdi ision in abundance classes as eco ded a sampling plo s in o es s ands, Abundance Classes 1 and 2 bo h belong o Lichen class ‘‘P esen ’’. Lichen Class Abundance Class Mean Biomass (kg/ha) N % Absen 00 6545 P esen 135 4934 2 120 30 21 doi:10.1371/jou nal.pone.0028779. 003 E ec s o Logging on A bo eal Lichen Occu ence PLoS ONE | www.plosone.o g 6 Decembe 2011 | Volume 6 | Issue 12 | e28779 A ea-weigh ed model: accoun ing o landscape changes The model’s ou pu alues a e di ec ly dependen on s and age. Howe e , calcula ions based on s and age alone canno p ope ly ep esen he capaci y o he whole landscape o sus ain a bo eal lichens, because hey do no accoun o he size o indi idual s ands ela i e o he o al a ea. Consequen ly, he a ea-weigh ed model was used o compa e he his o ical ends in he model alues due o o es y in a way ha e lec s bo h s and age and spa ial a ea. Fig. 4 shows he cumula i e dis ibu ion o esul s calcula ed using he a ea-weigh ed model, i.e. he con ibu ion o each o es age class o he sum o he a ea-weigh ed model alues o each yea s udied. In he yea s p io o 2006, o es s olde han 120 yea s con ibu ed subs an ially o he a ea-weigh ed p obabili y o lichen occu ence, p. Landscape agmen a ion, i.e. he dec eases in he numbe and s and size o olde o es s and he dec easing s and ages owa ds 2006 had a nega i e e ec on pa he landscape scale. The cumula i e cu es become less s eep om yea o yea , i.e. he in luence o olde age classes on lichen occu ence (p) dec eases due o he educ ion in he a ea o land hey co e . The cumula i e a ea-weigh ed model indica es ha hei in luence dec eased by 51% be ween 1926 and 2006, because o es s ands wi h a high p obabili y o lichen occu ence became bo h smalle and less abundan (Fig. 4). In con as o he si ua ion in 1960, he high abundance o o es s aged be ween 40 and 60 yea s in 2006 could no compensa e o he loss o e y old o es s. Discussion The cu -o poin and i s p ac ical ele ance The his o ical land usage pa e ns o he Sami people in ex ac ing subsis ence esou ces ha e shaped he bo eal o es s o Sapmi in o a ‘‘cul u al landscape’’ sensu Be kes & Da idson-Hun [49]; such uses include he emo al o he inne ba k om pine ees o ac as a plan ood esou ce [50,51] o cu ing lichen- ich ees as eme gency ood o eindee in ha sh win e s o o o he he ding p ac ices [52]. In con as , comme cial imbe ex ac ion u ned he o es s in o an ‘‘economic landscape’’ ha in which he a ailabili y o a bo eal lichens ha sus ain eindee he ding is limi ed compa ed o he condi ions p io o he in ensi ica ion o o es y. Ou analysis has quan i ied some o he impac s o es y has had on a bo eal lichens in Akkajau . Ra he han modeling B yo ia biomass in ela ion o s and age [33,44], we modeled he p obabili y o lichen occu ence [53]. This iden i ied he alue and impo ance ha an a ea migh possess, making ou app oach analogous o ha adop ed by Juu inen e al. [54] in hei e alua ion o he conse a ion alue o a gi en o es . A h eshold age o 63 yea s was iden i ied o ou s udy a ea, wi h i s pa icula his o y o imbe ha es ing; unde Figu e 2. P edic ed p obabili y o a bo eal lichen p esence and cu -o poin selec ion. Resul s om he logis ic eg ession in ci cles, sensi i i y plus speci ici y as solid line and he chosen o es age cu -o o disc imina ing be ween s ands wi h and wi hou lichen as dashed line. doi:10.1371/jou nal.pone.0028779.g002 Table 4. Reg ession coe icien s o p edic ing he p esence o lichens on he basis o o es age. Coe icien s (b) S.E. d Sig. Nagelke ke R 2 Goodness o i x2 Sig. Exp(b)95% C.I. o Exp(b) Lowe Uppe Fo es Age 0,04 ,0,01 1 ,0,001 0,63 20,116 ,0,001 1,04 1,03 1,06 Cons an -3,11 0,53 1 ,0,001 0,045 doi:10.1371/jou nal.pone.0028779. 004 E ec s o Logging on A bo eal Lichen Occu ence PLoS ONE | www.plosone.o g 7 Decembe 2011 | Volume 6 | Issue 12 | e28779 cu en condi ions, B yo ia is e y likely o be p esen in s ands abo e his age (Fig. 2). This h eshold could concei ably be used as an indica o o e alua e he po en ial o sus ainable eindee husband y. Simila h eshold ages we e epo ed by Gowa d & Campbell [55] o an unmanaged s and o Abies lasioca pa,Picea engelmanni and Pinus con o a in eas -cen al B i ish Columbia. In hese ees, i ook 60 yea s o B yo ia o g ow as a as he ou e ips o he ee b anches, which is he poin when lichen accumula ion begins. Likewise, S one e al. [56] ound a signi ican inc ease in lichen biomass 70 yea s a e ha es ing in eas e n Canada. In a sp uce and pine domina ed o es in sou h-cen al Sweden, Uliczka & Angels am [57] eco ded ha B yo ia agmen s we e p esen on 5% o sampled ees aged be ween 61 and 80 in ci cula plo s o 10 m adius (N = 90). This h eshold needs o be es ed in o he pa s o he Fennoscandian eindee he ding a ea be o e i can be gene ally applied. Fo ins ance, si es wi h highe p oduc i i y a e domina ed by sp uce [22], leading o niche pa i ioning be ween Alec o ia sa men osa, which p e e s he lowe canopy, and B yo ia uscescens in he uppe canopy [58]; his changes he g owing condi ions. We he e o e ecommend ha ou model should be es ed in a di e se se o habi a ypes o iden i y addi ional habi a - ela ed a iables ha in luence he occu ence o a bo eal lichens. As ou esul s show, main aining bo h he spa ial and empo al co e age o a key habi a , i.e. o es olde han 63 yea s, can inc ease he likelihood ha lichens will be p esen as a g azing esou ce o eindee . Because he p obabili y o lichen occu ence inc eases wi h age, ex ending o a ion imes could g ea ly inc ease he abundance o a bo eal lichens [27,44]. Fu he , he con inui y o key o es habi a s is cen al o he main enance o high lichen abundances: Esseen e al. [33] showed lichen biomass o be highe in un-ha es ed old g ow h o es s han in selec i ely logged o es s o he same age The e o e, empo al and spa ial con inui y is especially impo an ; he a ailabili y o sui able o es s ands alone does no gua an ee he p esence o B yo ia, especially when oppo uni ies o coloniza ion ha e been limi ed du ing he o es ’s his o y [59]. Fo es his o y: Loss o impo an landscape elemen s On he landscape scale, o es s ands in Akkajau we e consecu i ely agmen ed in o e e -smalle s ands o e he 80 yea s examined in his s udy. This a ec ed old o es s ands in pa icula (Fig. 4), and he mean s and age dec eased subs an ially (Table 6). As a esul , he p obabili y o a bo eal lichens being Figu e 3. Pas and po en ial u u e changes in he a ea co e ed by s ands olde han 60 yea s. The inc easing agmen a ion om 1920 o 2006 is illus a ed by inc easing a ea isola ing s ands olde han 60 yea s om each o he , exp essed as a ea co e ed by a 200 m-bu e . doi:10.1371/jou nal.pone.0028779.g003 Table 5. La ges pa ch index (LPI) in % and a ea-weigh ed mean s and a ea (ha) o abo e- h eshold s ands. 1926 1936 1960 2006 LPI (%) 50.2 39.8 24.3 5.9 S a (ha) 163 155 160 43 doi:10.1371/jou nal.pone.0028779. 005 Table 6. Changes o e ime in he s and numbe (N), mean a ea (ha), s anda d de ia ion and o al a ea (ha). 1926 1936 1960 2006 N696 575 651 900 Mean a ea (ha) 32 49 44 30 Mean a ea S d. De 62 77 64 30 To al a ea (ha) 21,979 28,283 28,869 26,600 doi:10.1371/jou nal.pone.0028779. 006 E ec s o Logging on A bo eal Lichen Occu ence PLoS ONE | www.plosone.o g 8 Decembe 2011 | Volume 6 | Issue 12 | e28779 p esen in a gi en egion dec eased due o losses o aluable habi a and educed connec i i y be ween abo e- h eshold s ands in space and ime. In Akkajau , he pos -1960 loss o he emaining la ge con inuous old o es s, which had a high p obabili y o lichen occu ence cons i u es a ‘‘bo leneck’’ in habi a a ailabili y and educed he landscape’s abili y o p o ide a bo eal lichens as o age o eindee . The con ibu ion o old o es s ands o he p obabili y o a bo eal lichen being p esen a he landscape le el in 2006 is he e o e low in e ms o a ea (Fig. 4). Since B yo ia is a species wi h limi ed dispe sal abili ies, he spa ial s uc u e o i s habi a is essen ial o i s su i al and abili y o colonize new pa ches [45,60]. Because s ochas ici y in i s su i al and ep oduc ion may a ec popula ion sizes [61], changes in i s spa ial and empo al habi a pa e ns a e especially impo an o unde s anding he impac s o agmen a ion and habi a loss. In pa icula , small and/o isola ed pa ches do no eadily eco e om en i onmen al dis u bances, no do hey acili a e immig a ion and es ablishmen . This is he si ua ion in many bo eal o es s in bo h Scandina ia and Canada [62]. The educed iabili y o he lichen popula ion may he e o e esul in local ex inc ion due o s ochas ic e en s, e.g. when pa ches a e oo isola ed o be colonized [61]. Fu he , a species’ esponse o a speci ic dis u bance may be subjec o ime lag. This makes i pa icula ly impo an o s udy his o ical landscape pa e ns, as hey can explain mode n day species dis ibu ion pa e ns [63]. His o ical landscape ans o ma ions he e o e c ea e a pa h dependence, i.e. u u e op ions and changes depend on decisions aken bo h in he pas and a p esen [64]. Because o he slow dynamics o o es s, managemen decisions will impac hei spa ial and empo al de elopmen on long (mul i-decade) imescales, and will limi o acili a e u u e op ions (Fig. 3). Ou u u e p ojec ions illus a e he possible ou comes o some di e en managemen app oaches. Managemen scena ios The p ojec ed e ec s o he h ee di e en o es y managemen scena ios on he abundance o h eshold s ands p o ides insigh in o he likely u u e de elopmen o he o es s in ques ion, which a e pa icula ly impo an habi a s o a bo eal lichens (Fig. 3). Unde bo h he business-as-usual and he in ensi e scena ios, he co e age o h eshold s ands will ini ially inc ease ela i e o he p esen day si ua ion, because o he cu en p edominance o young and middle-aged s ands oday. The BAU app oach is ocused on main aining a cons an imbe ha es . Howe e , i has di e en consequences o a bo eal lichens. Unde BAU, o es s olde han 160 yea s a e le unmanaged, as a e p o ec ed a eas. These a eas will he e o e se e as ‘‘sou ce habi a s’’ o lichen dispe sal. Due o he dynamic pa e n o eme ging and disappea ing h eshold s ands, lichen es ablishmen and pe sis- ence will be mainly es ic ed by he empo al and spa ial a ailabili y o colonizing habi a s. Unde he in ensi e scena io, all o he a eas ha a e le unmanaged in he BAU scena io a e a ailable o ha es ing. This esul s in a loss o lichen- ich s ands and is also likely o u he dec ease he dispe sal abili y o he lichens due o mo e p onounced agmen a ion. As such, al hough some o es s abo e he h eshold age a e e ained unde his scena io, hey a e less likely o be colonized o he same ex en as would be he case unde BAU. One could easonably specula e ha he age h eshold migh shi upwa ds, al hough such olde ees would be mo e likely o be ha es ed. I should be no ed ha his managemen s a egy is no allowed unde he cu en Swedish Fo es y Ac . The con inui y o o es co e inc eases unde he No Managemen scena io. This s a egy would suppo he sp ead o a bo eal lichen agmen s o dispe sal uni s and consequen ly p omo e hei es ablishmen . A he same ime, lichen abundance would inc ease in al eady-colonized habi a s. This, howe e , does Figu e 4. Changes in age composi ion in Akkajau 1926–2006. A ea in % co e ed by age classes illus a ed by columns, hei cumula i e con ibu ion o he a ea-weigh ed model by he dashed line. doi:10.1371/jou nal.pone.0028779.g004 E ec s o Logging on A bo eal Lichen Occu ence PLoS ONE | www.plosone.o g 9 Decembe 2011 | Volume 6 | Issue 12 | e28779