Historical human influence on forest compositionand structure in boreal Fennoscandia

Torbjorn Josefsson, Bjorn Gunnarson, Lars Liedgren, Ingela Bergman, andLars Ostlund

Abstract: In studies on natural dynamics, biodiversity and reference conditions legacies of preindustrial human land useare often neglected. In this study, using archaeology and dendrochronology combined with field surveys on present forestcharacteristics, we assessed the naturalness of a protected forest landscape and examined the role of indigenous peoples inshaping forest structure in the past. Our results show that the studied Scots pine (Pinus sylvestris L.) forest conforms tothe generally accepted impression of pristine forests and that it has a long history of human utilization. Areas with humanpresence over long time periods, especially in and near settlements, show significant differences in current forest character-istics compared with the rest of the landscape: the forest is younger (mean age 140–190 years compared with >300 years),the volumes of deadwood lower (8–13 m3�ha–1 compared with >20 m3�ha–1), and the tree species composition is substan-tially different from the surrounding forest. We suggest that these disparities are strongly linked to past land use and thatindigenous people can alter ecosystems substantially and that the legacies of their activity may last for centuries. Conse-quently, in ecological research and conservation strategies, forest characteristics should always be considered in the lightof their historical context.

Resume : Dans les etudes portant sur la dynamique naturelle, la biodiversite et les conditions de reference, l’heritagelaisse par l’utilisation preindustrielle des terres par l’homme est souvent neglige. Dans cette etude, nous avons evalue lanaturalite d’un paysage forestier protege et examine l’effet sur la structure de la foret qu’on eu dans le passe les autochto-nes, a l’aide de l’archeologie et de la dendrochronologie combinees a l’inventaire sur le terrain des caracteristiques actuel-les de la foret. Nos resultats montrent que la foret de pin sylvestre (Pinus sylvestris L.) que nous avons etudiee corresponda l’idee generalement acceptee de foret vierge mais qu’elle a egalement une longue histoire d’utilisation par l’homme.Dans les zones ou la presence humaine s’est fait sentir pendant de longues periodes, particulierement dans et pres des en-droits ou l’homme s’est etabli, il y a des differences significatives entre les caracteristiques de la foret actuelle et cellesqu’on observe dans le reste du paysage. La foret est plus jeune (age moyen de 140–190 ans comparativement a > 300ans); les volumes de bois mort sont plus faibles (8–13 m3�ha–1 comparativement a > 20 m3�ha–1) et la composition en espe-ces arborescentes est substantiellement differente de celle de la foret avoisinante. Nous croyons que ces differences sontetroitement reliees a l’utilisation passee des terres mais aussi que les autochtones peuvent modifier substantiellement lesecosystemes et que l’heritage de leurs activites peut se faire sentir pendant des siecles. Par consequent, dans le cadre detravaux de recherche en ecologie et de strategies de conservation, les caracteristiques d’une foret devraient toujours etreexaminees en tenant compte du contexte historique.

[Traduit par la Redaction]

IntroductionExtant forest that has never been logged or cleared for

agriculture has a very high ecological value and is widely

considered to be truly pristine. It is often used both to studynatural dynamics and as an ecological reference for compar-isons with managed forests (Foster et al. 1996; Rhemtulla etal. 2002). Commonly, the criteria for selecting unmanagedforests to designate as nature reserves include prolonged ab-sence of human land use and extended periods where themain disturbance has been natural. Past human land use byindigenous people (Lane 2006) is often neglected in suchcriteria and in ecological studies involving forest structureand biodiversity in supposedly pristine forest ecosystemsand when reference conditions for natural variability in for-ests are defined. Still, past anthropogenic disturbance, bothpulse and press events, may not be different from naturaldisturbance in that it can continue to influence ecosystemstructure and processes at a range of scales up to the presentday (Foster et al. 2003; Schulte et al. 2007). There is an in-creasing body of evidence suggesting that land use by indig-enous people has had (and continues to have) substantialeffects on the structure and composition of temperate forests

Received 21 May 2009. Accepted 31 January 2010. Publishedon the NRC Research Press Web site at cjfr.nrc.ca on 24 April2010.

T. Josefsson1 and L. Ostlund. Swedish University ofAgricultural Sciences, Department of Forest Ecology andManagement, Umea SE-901 83, Sweden.B. Gunnarson. Swedish University of Agricultural Sciences,Department of Forest Ecology and Management, Umea SE-90183, Sweden; Stockholm University, Department of PhysicalGeography and Quaternary Geology, Stockholm SE-106 91,Sweden.L. Liedgren and I. Bergman. The Silver Museum/The Instituteof Subarctic Landscape Research, Storgatan 20, Arjeplog SE-930 90, Sweden.

1Corresponding author (e-mail: torbjorn.josefsson@svek.slu.se).

872

Can. J. For. Res. 40: 872–884 (2010) doi:10.1139/X10-033 Published by NRC Research Press

(Black et al. 2006), especially that of neotropical ecosystems(Thompson et al. 2002; Bayliss-Smith et al. 2003). As a re-sult, the true pristine conditions of many such forests arenow being questioned (Denevan 1992; Heckenberger et al.2003; Gillson and Willis 2004; Willis et al. 2004).

In northern Fennoscandia, supposedly pristine boreal for-ests serve as reference areas for many purposes, includingecological research into forest structure (Lilja andKuuluvainen 2005), coarse woody debris (Rouvinen et al.2002), associated species (Junninen et al. 2006), and the de-velopment of silviculture (Angelstam 1998). However, a fewstudies have questioned the established idea that these pro-tected areas are untouched by humans and therefore consti-tute reference conditions. For example, Uotila et al. (2002)conducted a study in eastern Fennoscandia using current for-est data and historical records to evaluate the level of natu-ralness of protected forests. Although structurallyheterogeneous and species rich at present, these forestswere found to have very different histories, and severalwere found to have been affected by previous human landuse. Similarly, Segerstrom (1997) found clear evidence offormer land use 300 years ago in a supposedly pristine bor-eal forest reserve in northern Sweden. Consequently, thereare reasons to believe that preindustrial land-use legaciesand their effect on forested ecosystems have been muchunderestimated in northern boreal forests as well. In thisstudy, embracing the last four centuries, we assess the natu-ralness of a large protected forest landscape where humanimpact is presumed to be insignificant, and we examine therole of Sami, an indigenous people of Fennoscandia, inshaping forest structure both locally and at larger scales.The main objectives were to (1) analyze differences in the

current forest structure, including composition and variation,in relation to quantities of archaeological remains and cul-turally modified trees (CMT) across the studied landscape;and (2) interpret and discuss the importance of temporaland spatial aspects of historical human land use on ecosys-tem structure and processes in northern boreal forests ingeneral.

Material and methods

Study areaThe study area (668N, 178E) occupies 2700 ha of the

northern part of the Tjeggelvas Nature Reserve situated innorthern Sweden and is a part of a large roadless forestlandscape. Highly appreciated for its intact appearance, con-servation managers as well as people of the scientific com-munity consider this to be one of the areas least affected bymodern forest management in Europe (Fig. 1). The reservelies within a large region called Sapmi, which covers north-ern Fennoscandia including the Kola Peninsula in northwestRussia. Parts of this region have been inhabited for morethan 9000 years, and Sami ethnicity has developed duringthe last few thousand years (Bergman et al. 2003). Thestudy area was selected because it includes a late succes-sional forest with (1) known evidence of Sami settlements,(2) no known history of commercial logging, and (3) nomodern infrastructure or habitation (>5 km to the nearestroad or house).

Tjeggelvas Nature Reserve is located in the upper part ofthe Pite River valley in the northern boreal zone, borderingthe Scandinavian mountain range. The region’s climate ischaracterized by short mild summers and long cold winters

Fig. 1. Location and topographic map of the study area (delimited) showing the location of the two settlement sites Blacka and Munka andthe extent of the decayed reindeer enclosure. The remainder of the study area forms an area with low anthropogenic impact. Reproducedwith permission from Lantmateriverket 1998. From GSD–Roadmap, dnr 507-98-4720.

Josefsson et al. 873

Published by NRC Research Press

with snow cover from October to May. The mean annualprecipitation is approximately 530 mm, and the mean annualtemperature –2 8C (Alexandersson and EggertssonKarlstrom 2001). The local topography is characterized byhilly terrain (450–550 m above sea level), including frequentboulder fields interspersed with small lakes, streams, andwetlands. The soils are mainly coarse tills, underlain bymigmatised granitoids, with younger granite and pegmatiteintrusions. Generally, site conditions are very similarthroughout the forested parts of the study area. The vegeta-tion is dominated by semi-open late-successional Scots pine(Pinus sylvestris L.) forest on dry and nutrient-poor soils,with scattered trees of downy birch (Betula pubescensEhrh.), grey alder (Alnus incana (L.) Moench), goat willow(Salix caprea L.), and Norway spruce (Picea abies (L.)Karst.). Ground vegetation is dominated by dwarf shrubssuch as Vaccinium spp. and Empetrum spp.; lichens such asCladonia spp., Cetraria spp., and Stereocaulon spp.; andmosses, predominately Pleurozium schreberi (Brid.) Mitt.,Hylocomium splendens (Hedw.) Schimp., and Dicranum spp.

Sami low-intensity forest utilizationIn prehistoric times, the Sami primarily relied on hunting

and gathering, characterized by logistic procurement strat-egies, necessary because of the strong seasonal fluctuationsand temporal abundance of resources (Ruong 1969). Some-time after AD 500, a reorientation of the subsistence basistowards reindeer (Rangifer tarandus L.) herding took place,resulting in changes in land use and settlement patterns(Aronsson 1991; Bergman et al. 2008). To a certain extent,reindeer herding formed an integral part of Sami subsistencestrategies, in combination with hunting and fishing, from the11th century onwards (Bergman et al. 2008). By the end ofthe 16th century the socio-economic transfer was complete,as verified by historical records describing intensive reindeerherding characterized by nomadism and seasonal settlements(Lundmark 1982). This form of reindeer herding meant thatthe Sami usually had several camps with temporary or morepermanent huts, used at different times of the year (Ruong1969).

Traditional Sami land use included cutting trees, princi-pally pine and birch, for firewood and construction. Treeson which Bryoria spp. were abundant were felled to allowthe reindeer to feed on the arboreal lichens during periodswhen grazing was severely limited (Berg et al. 2009). Ex-traction of specific resources, e.g., the inner bark from pinefor food, was particularly important (Zackrisson et al. 2000).Trails, routes, and specific locations were often marked byblazes on trees or small piles of stone. In addition, birchwas used for a number of purposes, for example the woodwas used for manufacturing of different utensils and skisand the bark was used for roofing of the huts. In the late19th century, the intensive herding of reindeer was suc-ceeded by the more extensive herding still practiced today.

Sampling and measurementsThe sampling was designed (1) to locate traces of histori-

cal human activity and (2) to record current forest character-istics (cf. Andersson et al. 2008). To detect traces of pasthuman land use, a total of 18 transects (20 m wide), runningin an east–west direction and spaced 500 m apart, were es-

tablished (Fig. 1). The total length of the transects was ap-proximately 40 km, representing ca. 80 ha or about 3% ofthe study area. Evidence of past human land use, i.e., bark-peeled trees, trail markings, cut stumps, and the occurrenceof fire-scarred trees, was carefully recorded along the trans-ects. Scars derived from bark peeling are recognized by theircharacteristic shape and often tool marks are still present.Cut stumps were classified into different categories:(1) stumps created by cutting with an axe to produce fire-wood or timber for various wooden constructions and (2)stumps created as a result of reindeer herding practices(here denoted as lichen stumps). The stumps in the lattercategory are easily differentiated from other cut and naturalstumps, as they are found in clusters and are very distinct incharacter, i.e., high (about 1 m) and thin trunks, commonlywith visible axe marks (cf. Berg et al. 2009). Additionally,the presence of archaeological remains, e.g., hearths alonglakeshores, was derived from a previous land-based archaeo-logical field survey covering the study area (cf. Josefsson etal. 2009; Liedgren et al. 2009). In that field survey, twoabandoned Sami settlements, one on the northern shore ofLake Munkajaure and the other on the western shore ofLake Blackajaure, were recorded. One more important find-ing was the remains of an old reindeer fence, which was al-most completely degraded, and the area originally enclosedwas estimated to encompass ca. 550 ha within the presentstudy area (Fig. 1).

To examine the current forest structure and composition,a total of 74 circular sample plots (0.1 ha each) were estab-lished. Of these, 68 were spaced at 500 m intervals alongthe transects (17 inside the reindeer enclosure and 51 out-side). The remaining six plots were established at the Blackaand Munka settlement sites (three at each site). On the basisof what is known about the spatial extent of forest utiliza-tion near Sami dwellings, settlement sites are thought to en-compass an area of ca. 3 ha (Hicks 1995; Ostlund et al.2003). Therefore the three plots were arranged 100 m apartin an equilateral triangle, with one of the plots centered atthe exact location of a decayed and overgrown Sami hut.We recorded the distribution of tree species, tree height,and the diameter at breast height (DBH) for all trees ‡5 cmDBH, as well as the characteristics of any coarse woody de-bris (CWD) in each sample plot.

The ground vegetation composition was classified accord-ing to Ebeling (1978), based on ground moisture (xeric andmesic) and dominant vegetation (ground lichens, dwarfshrubs, horsetails, herbs). The characteristics associatedwith CWD included tree species, tree height (only standingdead trees) and quality. The minimum diameter for measur-ing standing dead trees was 5 cm DBH. Fallen dead treeswith a diameter less than 5 cm at the base were not meas-ured. Four categories of CWD were designated: (1) standingdead trees, (2) downed logs (including entire fallen trees,pieces of the trunk, and large branches), (3) high stumps(‡1.3 m), and (4) stumps (<1.3 m). Four decay classes wereestablished, based on the hardness of the wood: I — barkand phloem still fresh or wood hard with bark partly loos-ened or detached, II — wood partly decayed from the sur-face with bark usually detached, III — wood fairly softwith no bark left, and IV — wood soft, highly decomposed.Furthermore, confirmatory criteria were that a knife could

874 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

be pushed only a few millimetres, 1–2 cm, and 2–5 cm intothe wood of classes I–III, respectively, while more than5 cm of the blade could easily be pushed into the wood ofclass IV. These decay stage classes were modified fromRenvall (1995).

For age determination and radial growth analysis, we ran-domly selected and cored 659 living trees within the sampleplots, using increment borers (4.5 and 12 mm in diameter).All samples were taken 1.3 m above the ground to avoid rotand irregularities in ring sequences that may be present inthe early part of a tree’s development. To determine thetime needed for a tree to reach breast height, and thus to beable to compensate for this in our calculations, disk samplesfrom 1.3-m-high trees (n = 64) were taken from the base ofthe tree, as near the germination point as possible. In addi-tion, 70 CMTs within the sample plots or along the transectswere sampled using increment cores (from living trees) anda manual handsaw (from dead trees).

Laboratory procedures and data analysisThe volumes (V) of living pine trees and standing dead

trees were estimated using volume functions, including lati-tude, derived from Brandel (1990).

½1� V ¼ 10a � DBHb � ðDBHþ 20Þc � hd � ðh� 1:3Þe

where DBH is diameter at breast height; h is tree height; anda, b, c, d, and e are coefficients. This volume estimate forpine is for volume under bark. Therefore, a deduction forbark volume was applied using data pertaining to bark thick-ness for pine forests in northern Sweden (Ostlin 1963). Forvolume estimates for living birch trees a volume functionfor northern Sweden derived from Naslund (1940) was used.

½2� V ¼ aDBH2 þ bDBH2hþ cDBHh2

where DBH is diameter at breast height, h is tree height, anda, b, and c are coefficients. For downed logs, the length ofthe fallen dead tree (l) and the maximum diameter (Dmax)and minimum diameter (Dmin) were recorded. The volumewas estimated using the formula for the frustum of a circularcone:

½3� V ¼ ðpl=3Þ � ½ðDmax =2Þ2 þ Dmax Dmin þ ðDmin =2Þ2�

The volumes of stumps and high stumps were estimatedon the basis of the length and mid-diameter using the for-mula for a cylinder.

Tree age was determined in the laboratory by annual ringcounts using standard methods. All cores from living treeswere mounted and sanded using two different grades ofsandpaper (100 and 400) to allow the rings to be seenclearly. Ring widths of living trees were measured under amicroscope at 6.4� to 40� magnification. To determinetree age at breast height when increment cores failed to in-tersect the pith, the length of the missing radius was esti-mated by matching the curving of the inner rings toconcentric circles drawn on a transparent plastic sheet. Totaltree age was then determined by adding the average age ofthe sampled 1.3-m-high trees to the estimated ages of treescored at breast height. For dead trees, stumps, and woodenparts from the reindeer fence, two radii were prepared fromeach disc using a scalpel and chalk to obtain a clean surface

and clear ring boundaries. The tree-ring widths of each sam-ple were measured with a precision of 1/100 mm using anAniol device. The two radii from each tree were cross-checked and averaged to produce a mean series from eachindividual sample. The ring width series of CMTs and dead-wood samples was then cross-dated against a master chro-nology from the area using CATRAS software (Aniol1983). All dates were quality checked, both visually and us-ing COFECHA software (Holmes et al. 1986).

To identify possible growth responses that may indicatedisturbance events all samples were further analyzed forabrupt changes in radial increment. A growth release wasdefined as an increase in mean tree-ring width, between twosuccessive 10-year periods, of more than 100% (cf. Grovenet al. 2002). The growth of trees located close to the altitu-dinal tree limit is mainly limited by summer temperaturesand, therefore, is highly responsive to climatic variability(Gunnarson 2008). To distinguish whether the recorded treegrowth responses were due to disturbance events or to cli-mate fluctuations, the number of growth releases were re-lated to a 400-year reconstruction of past summertemperature in north Fennoscandia, based on tree-ring widthand maximum density established by Grudd (2008). Thedata in this study originates from the Tornetrask area innorthern Sweden (350–450 m above sea level), and the sam-ple material was derived from both living and subfossilScots pine. The reconstructed summer temperatures are ro-bust as a result of the incorporation of maximum density.

Statistical analysesTo analyze different degrees of past anthropogenic impact

on present forest structure, we compared subareas displayingdifferent levels of human activity, i.e., containing differentconcentrations of archaeological remains and signs of hu-man activity, such as stumps derived from cutting of fire-wood and stumps produced as a result of reindeer herdingpractices. The northern and eastern parts constitute a singlesubarea with low levels of anthropogenic impact (ca.2150 ha), whereas the settlements of Blacka and Munka(3 ha each) and the decayed reindeer enclosure (550 ha) rep-resent subareas with high levels of anthropogenic impact(Fig. 1).

We used analysis of variance (ANOVA) to examine dif-ferences in forest variables (mean age and volumes of livingtrees and CWD that conformed to requirements of normalityand homogeneity of variance) between areas where therewere different levels of anthropogenic impact. Variablesthat did not conform to the necessary requirements were an-alyzed using a Mann–Whitney U test or a Kruskal–WallisH test, both of which test for differences among groups butare unable to evaluate the significance of individual factorsor interactions between factors. Where a significant differ-ence in the dependent variable was observed, the ANOVAanalysis was followed by pair-wise comparisons using theStudent–Newman–Keuls post hoc procedure. Kruskal–Wallisand Mann–Whitney tests were not followed by post hoc pro-cedures. Test outcomes were considered to be significantlydifferent at a probability of 5%. All statistical analyses wereperformed using SPSS for Windows, version 15.0 (SPSS,Inc., Chicago, Illinois, USA).

Josefsson et al. 875

Published by NRC Research Press

Results

Current forest characteristics and historical land useEnvironmental conditions (edaphic factors and vegetation)

varied only slightly across the study area. In the low-impactarea and in the former reindeer enclosure, the forest wascharacterized by Scots pine with scattered trees of downybirch on dry sites with Empetrum spp. as the most abundantdwarf shrubs. At the two settlement sites, the forest wasslightly more mesic and downy birch was much more abun-dant. Here Vaccinium spp. were the dominant dwarf shrubs.Most variables related to forest structure varied noticeablyacross the study area (Table 1). The mean age of treeswithin sample plots ranged from ca. 100 years old to almost400 years old (for trees ‡5 cm DBH). The mean age was267 years old for pine and 114 years old for birch. However,many pine trees were much older than 400 years and theoldest dated tree germinated in the mid 14th century. Thevolume of Scots pine (all volume data are means) variedconsiderably and ranged from 13 to 203 m3�ha–1 (on average72 m3�ha–1), whereas the volume of downy birch was gener-ally very low (ca. 0.6 m3�ha–1) and varied only slightlyacross the study area. The variation of CWD among differ-ent parts of the study area was large and ranged from 4 to53 m3�ha–1 (representing 6%–43% of the total volume of liv-ing and dead trees), and pine accounted for, on average,97% of the volume of CWD. Most of the CWD was downedlogs (ca. 60%) and more than half of the CWD volume be-longed to decay class I, represented by both standing deadtrees and downed logs.

An analysis of the regeneration pattern of Scots pine indi-cated a continuous and wave-like pattern describing tree re-cruitment, with low recruitment in the early 17th, 19th, and20th centuries and peaks around 1750 and 1850 (Fig. 2). Notree recruitment was recorded after 1940. Few of the coredtrees germinated during the period between the early 15thcentury and the mid 17th century. Following this phase,three periods with specific recruitment patterns were identi-fied: ca. 1660–1720, 1730–1820, and 1830–1910. Fire scarswere very rare and recorded from three small areas: thenortheastern and eastern sides of Lake Munkajaure(<100 ha), dated to 1790 and 1888, and 500 m north ofLake Blackajaure (<75 ha), dated to 1751 and 1888. In addi-tion, one fire scar, found on the southeastern shore of LakeHalkasjaure, was dated to 1634–1635. Downed dead treeswith charred wood were also recorded from an area encom-passing ca. 250 ha between Lakes Halkasjaure and Blacka-jaure. No fire scars suitable for dating were recorded fromthis area.

In total, more than 1100 cultural objects indicating pastland-use practices were recorded during the field inventory(Table 2). A considerable amount of these were bark peel-ings, stumps related to tree cutting for various purposes,and blazes on trees. Most such land-use activities originatedfrom the 18th and 19th centuries and also, to a lesser extent,from the 17th century (Fig. 3). No traces of such activitiesderived from the time period 1910–2000. The bark-peeledtrees were evenly distributed over the whole study area(3.1�ha–1), but higher concentrations were found near Munka(8.7�ha–1) and Blacka (5.0�ha–1) and within the enclosure(4.0�ha–1) (Table 2). The bark peelings originated from the

17th, 18th, and 19th centuries. Lichen stumps were foundthroughout the area (2.7�ha–1) (Table 2) but were stronglyassociated with the area of the former reindeer enclosure(8.1�ha–1). On average, these stumps were 15.7 cm in diam-eter and 1.1 m high and originated from a period lastingca. 200 years (1702–1899) (Table 2, Fig. 3). Stumps relatedto the extraction of firewood and construction timber wereless common (<1�ha–1) (Table 2) and were mainly confinedto the immediate surroundings of Lake Blackajaure andwithin the area of the former reindeer enclosure. Tree blazes(<1�ha–1) occurred throughout the area but were less com-mon in the northeastern part (Table 2). The oldest tree blazewas dated to 1616 and represents the oldest CMT dated dur-ing this study (Table 2). Other types of cultural imprints(7.2�ha–1) that were recorded included wooden constructionsfor storage, stone piles (placed at regular intervals, indicat-ing a route or a trail), and other objects related to huntingand fishing (Table 2).

Table 1. Forest characteristics of the study area.

Variable Mean ± SE Rangeg

Living treesTree heighta 13.7±0.31 8–23Mean tree age (years) 247±7.13 106–389Mean tree age of:

Scots pine (years) 267±7.95 130–519Downy birch (years) 114±3.31 65–181

Basal areab 11±0.43 3–22Diameter of:c

Scots pine 19.6±0.18 5–63Downy birch 7.7±0.05 5–20

Total volumed 72.9±4.00 12.9–203.4Volume of:d

Scots pine 72.3±4.02 12.9–203.3Downy birch 0.6±0.10 0–5.2

Proportion of deciduoustreese

1.0±0.23 0–12.3

Coarse woody debris (CWD)Total volumed 20.1±1.25 4.0–53.4Proportion of CWDf 22.2±1.02 6.0–43.0Volume of:d

Standing dead trees 7.1±0.88 0–38.2Downed logs 11.4±0.87 0.9–32.5High stumps 1.0±0.18 0–10.8Stumps 0.6±0.17 0–10.3

Volume of decay staged

Class I 10.8±1.04 1.1–45.4Class II 2.8±0.31 0–12.2Class III 4.0±0.54 0–23.8Class IV 2.5±0.36 0–15.8

aArithmetical mean tree height in metres for the 100 largest treesper hectare.

bBasal area in m2�ha –1.cMean DBH in cm.dVolumes in m3�ha–1.ePercentage of birch of total living tree volume.fPercentage of CWD of total living and dead tree volumes.gRange is the variation within sample plots.

876 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

The effect of land use at two different spatial scalesTo estimate local effects of past land use on present forest

structure, we compared the two settlement sites (Blacka and

Munka, which were treated as being unrelated) with thelow-impact area. Several significant differences in foreststructure were detected. Mean tree age was higher in the

Fig. 2. Age structure of Scots pine in 10-year intervals for trees ‡5 cm diameter at breast height.

Table 2. Number of culturally modified trees per hectare (no.�ha–1) recorded at the two Sami settlements, inside the decayedreindeer enclosure, in the low-impact area, and in the whole study area.

Type of culturallymodified tree

Blackasettlement

Munkasettlement

Reindeerenclosure

Low-impactarea

Wholestudy area

Totalno.c

Range(years)c

Bark peeled trees 5.0 8.7 4.0 2.8 3.1 249 1632–1888Lichen stumps — — 8.1 0.9 2.7 218 1702–1899Cut stumpsa — 4.0 1.2 0.8 0.9 74 —Tree blazes — — 0.5 0.1 0.3 38 1616–1820Decayed fence — — — — — — 1737–1848Other cultural objectsb 3.3 16.5 13.9 4.9 7.2 580 1644–1890Total cultural objects 8.3 29.2 27.7 9.5 14.2 1159 1616–1899

aCut stumps were not dated.bFor example, wooden constructions and items related to hunting, fishing, and settlement.cTotal numbers and temporal range of culturally modified trees given for the whole study area.

Fig. 3. Number of recorded and dated culturally modified trees per decade divided on different categories.

Josefsson et al. 877

Published by NRC Research Press

low-impact area (249 years) than at both settlement sites(Blacka 193 years and Munka 144 years) (F = 6.044, P <0.01) and the mean age of Scots pine was also higher in thelow-impact area (F = 4.227, P < 0.05); in contrast, the meanage of birch was similar in all three areas (107–114 years)(Table 3, Fig. 3). The volume of living trees differed onlyslightly between areas (Table 3). However, both the volume(F = 10.469, P < 0.01) and the proportion (F = 9.218, P <0.01) of birch were significantly lower in the low-impactarea than at both settlement sites.

The volume of CWD was significantly higher in the low-impact area (ca. 21 m3�ha–1) than in Blacka (ca. 8 m3�ha–1)(F = 3.989, P < 0.05), while intermediate values were re-corded in Munka (ca. 13 m3�ha–1). All the variables describ-ing CWD quality were highest in the low-impact area(Table 3). The volume of CWD in decay classes I and II dif-fered only slightly between the low-impact area and Munka,whereas the volume in later stages of decay was much lowerin Munka. At Blacka, the volume of CWD in all decayclasses was considerably lower than in both the low-impactarea and in Munka and significantly lower for CWD in de-cay stage III (F = 9.056, P < 0.05) (Table 3). In addition, wealso examined the proportions of different forest variables,

since the volume of living and dead trees was extremelyvariable and exceptionally low in some parts of the studyarea. The proportion of CWD differed significantly betweenthe low-impact area (25%) and both settlement sites (Blacka9% and Munka 15%) (F = 6.828, P < 0.01). However, theproportion of downed logs (62%–74%) and high stumps(4%–6%) was similar in the three areas. The proportion ofstanding dead trees was twice as high in the low-impactarea (ca. 30%) than at the settlement sites (ca. 12%–16%)(F = 7.809, P < 0.05). Likewise, the proportion of stumpswas about the same in the low-impact area and in Munka(ca. 3%), but at Blacka, 20% of the CWD volume wasmade up of stumps.

At a larger scale, i.e., comparing the reindeer enclosurewith the low-impact area, notable differences in frequenciesof cultural objects were recorded. The total number of cul-tural objects was ca. 28�ha–1 in the enclosure and ca.10�ha–1 in the low-impact area (Table 2). Substantial differ-ences in forest structure were recorded. The numbers of pinetrees with a diameter 5–20 cm and birch trees 5–10 cm werehigher in the enclosure, whereas there were significantlyfewer large-diameter pine trees (>41 cm) (F = 4.320, P <0.05) (Fig. 4). No significant difference in mean age was de-

Table 3. Comparisons of forest characteristics between the low-impact area and the two unrelatedSami settlement sites Blacka and Munka.

VariableLow-impactarea

Blackasettlement

Munkasettlement

Statisticaltest P

Living treesMean tree age (years) 249±8.11a 193±26.77ab 144±19.32b 6.044d 0.004Mean tree age (years) of:

Scots pine 271±9.69a 222±35.00ab 162±16.90b 4.227d 0.020Birch 114±4.34 107±6.12 115±15.51 0.063d 0.939

Total volumea 68.4±4.39 73.7±18.16 69.5±17.07 0.041d 0.960Volume of:a

Scots pine 68.0±4.41 72.3±18.37 67.0±18.25 0.029d 0.972Downy birch 0.4±0.09 1.4±0.78 2.5±1.36 10.469e 0.005

Proportion of deciduoustreesb

0.8±0.22 1.9±1.3 3.6±3.5 9.218e 0.010

Coarse woody debris (CWD)Total volumea 20.2±1.23a 7.6±2.91b 13.0±5.91ab 3.989d 0.024Proportion of CWDc 23.8±1.10a 8.9±1.47b 15.3±6.11b 6.828d 0.002Volume of:a

Standing dead trees 6.5±0.83 0.9±0.61 2.2±0.78 7.809e 0.020Downed logs 12.2±1.02 4.9±2.32 9.6±5.66 5.018e 0.081High stumps 1.0±0.16 0.3±0.17 0.8±0.38 0.843e 0.656Stumps 0.6±0.24 1.5±0.13 0.4±0.39 7.082e 0.029

Volume of decay stagea

Class I 9.7±0.97 3.5±1.71 10.0±4.33 4.010e 0.135Class II 3.3±0.40 0.6±0.48 2.5±1.90 4.677e 0.096Class III 4.5±0.58 1.0±0.05 <0.1±0.03 9.056e 0.011Class IV 2.7±0.44 2.5±1.52 0.5±0.36 2.835e 0.242

Note: Data represent means and standard errors. Different letters indicate statistically significant differences(P < 5%); for each variable, areas with the same letter are not significantly different. P values in boldface arestatistically significant at P £ 0.05.

aVolumes in m3�ha–1.bPercentage of birch of total living tree volume.cPercentage of CWD of total living and dead tree volumes.dF ratio (one-way ANOVA test, df = 2).ec2 (Kruskal–Wallis H test, df = 2).

878 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

tected. The volume of living trees was considerably higherin the enclosure (87 m3�ha–1) than in the low-impact area(68 m3�ha–1), although the difference was not significant(F = 3.555, P = 0.064), and the volumes of birch and CWDwere similar in the two areas (Figs. 5a and 5b). In the enclo-sure, the volume of standing dead trees was higher (F =3.974, P = 0.050) than in the low-impact area, the volumeof downed logs was somewhat lower, and the volume ofhigh stumps and stumps was quite similar in both areas(Fig. 5c). The volume of CWD in decay class I was signifi-cantly higher (16 m3�ha–1) in the former enclosure than inthe low-impact area (10 m3�ha–1) (F = 5.789, P < 0.05),while the volume of CWD in decay classes II–IV wasslightly higher in the low-impact area (Fig. 5d). The propor-tion of different CWD categories and decay stages exhibiteda pattern similar to that for the mean volumes.

Dating of disturbance eventsA total of 124 growth releases were detected in 82 (13%)

of all trees examined and only in pine. However, to elimi-nate responses to forest fires, five trees that were in areaswith recorded fire scars or traces of fire were excludedfrom the analysis. Consequently, 77 trees overlapping intime (the oldest tree dating back to the mid 14th century)were included in the analysis. Most trees with growth re-leases (approximately 63%) were derived from sample plotswithin the reindeer enclosure. Five 10-year periods with in-creased number of growth releases during the last 400 yearswere identified: 1635–1645, 1745–1755, 1815–1825, 1905–1915, and 1925–1935 (Fig. 6a). The proportion of trees ex-hibiting growth release was most pronounced during the twophases around 1750 and 1820; a total of 25 growth releaseswas detected between 1748 and 1755 and 21 between 1822and 1827. A correspondence between periods with rapidlyincreasing summer temperatures and periods with increasednumber of growth releases was detected for the phases1750, 1910, and 1930 but not for any of the other phases(Fig. 6b).

DiscussionThe heterogeneous, late-successional forest of the Tjeg-

gelvas Nature Reserve resembles other studied old Scotspine forests in northern Fennoscandia (Engelmark 1987;Lilja and Kuuluvainen 2005). The occurrence of CWD com-ponents such as very old trees, large standing and downeddead trees, and high stumps are especially valuable, sincethey provide essential habitats for a range of species groups,including certain species of fungi, lichens, insects, and birds(Siitonen 2001). Furthermore, forests that have high levelsof heterogeneity and no visible signs of logging or forestmanagement are frequently considered to be pristine forests,exclusively shaped by natural processes. However, heteroge-neity within forest ecosystems is created and maintained byseveral different processes over a wide range of temporaland spatial scales. Besides variations in soil properties anddisturbances, such as fire, wind-throw, insects, and patho-gens, the effect of climatic variation and anthropogenic dis-turbance may contribute to current forest structure.

Environmental and anthropogenic impact on forestcharacteristics

It is often suggested that dry lichen-dominated boreal pineforests, where very old trees are present, have been pro-moted by repeated surface fires (Engelmark 1987). In thisstudy, however, only a few fire scars were recorded, andover large parts of the study area there has been an unusu-ally long time since the last fire. Previous estimates of fire-free intervals in northern Fennoscandian pine forests rangefrom 30 to 300 years or even longer (Flannigan et al. 1998;Carcaillet et al. 2007). Although fire is a key feature of theecology of boreal forests, other factors may become increas-ingly important in forests where the long disturbance inter-vals are similar to or exceed the life-span of the dominanttree species, as has been shown by, for example, Mladenoffet al. (1993) in old-growth northern hardwoods in northernWisconsin, USA. For example, the wave-like regenerationpattern in the studied forest, with peaks around 1750 and1850, may have been the result of successful establishmentafter large-scale disturbances such as fire or, as proposed byZackrisson et al. (1995), of periods with good seed produc-tion as a result of favorable climatic conditions. Accordingto Grudd (2008) the decades preceding 1750 and 1850 werecharacterized by increased summer temperatures, and theperiods around 1600, 1800, and 1900 appear to have beencold (Fig. 6b). The peaks in regeneration are also reflectedin an increase in the proportion of trees displaying growthreleases during the same time period (Figs. 2 and 6a). Con-sequently, we believe that climatic variations may have con-tributed not only to the increased number of growth releasesand the concurrent peak in regeneration around 1750 butalso to the lowered regeneration in the early 17th, 19th, and20th centuries.

Time periods with increased number of growth releasesmay also indicate discrete disturbance events, for example,periods with increased human activity. In this study, no cor-relation between increased summer temperatures and periodswith increased growth releases was apparent for the periodsaround 1640 and 1830. Within the study area, most of thedetected growth releases during these time periods were re-

Fig. 4. Diameter distribution of pine (left) and birch (right) trees inthe decayed reindeer enclosure (open bars) and the low-impact area(solid bars). Data are the means and standard errors.

Josefsson et al. 879

Published by NRC Research Press

corded from trees sampled within the enclosure. In addition,in the 1830s a large number of trees seem to have been cutfor foddering of reindeers (Fig. 3). As there are no clearsigns of natural disturbance during the two time periods atthe localities where the growth releases were recorded, wesuggest that variations in summer temperatures alone cannotexplain the patterns of tree growth response; both climaticfluctuations and past human activity may have been impor-tant. However, natural disturbance in the form of forest firecannot be totally ruled out, since not all fires can be de-tected from fire scars.

Traces of historical land useThe study was designed to examine past land use where

anthropogenic impact was assumed to be very low, so thevery large number of detected traces of human land use wasunexpected. The utilization of trees for various purposes wascarried out throughout the study period, and slightly higherfrequencies of such activities were recorded around 1700,1750, 1790, and 1860, indicating a variation in land-use in-tensity over time. Pine trees with bark-peeling scars arecommon, and even though the majority of such trees occurin the most intensively used areas, they are present acrossthe whole study area, indicating a sparse and intermittentuse of the whole landscape. Based on the results of the sur-vey, we estimate that more than 8000 bark-peeled trees existin the study area. This is higher than any concentrations re-corded in Fennoscandia (cf. Zackrisson et al. 2000; Ostlund

Fig. 5. (A) Age of living trees, (B) volume of living trees, coarse wood debris, and birch, (C) volume of coarse wood debris categories, and(D) volume of decay stages, in the decayed reindeer enclosure (open bars) and the low-impact area (solid bars). Data are the means andstandard errors.

880 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

et al. 2003) and North America (cf. Kaye and Swetnam1999; Ostlund et al. 2005).

Both the presence and the size of the old reindeer enclo-sure (>500 ha) is a unique feature. Sami groups in this areaseldom kept reindeer fenced other than in small pens closeto settlement sites or as a temporary measure during activ-ities such as marking calves or slaughtering (Aronsson1991). The dating of the fence indicates that it was estab-lished in the early 18th century and repaired at least once inthe mid 19th century. According to the dating of the lichenstumps sampled inside the enclosure, the fence was in usefor at least 200 years. Interestingly, no bark peelings, lichenstumps, or blazes originated after 1910, which reflects a dra-matic change in resource utilization around that time (cf.Ruong 1969). Previous human land use is therefore obscuredby one century of limited and radically different use of the

area; during this time tree scars have overgrown and woodenartifacts have disintegrated through decay.

Past anthropogenic impact on the forest at two spatialscales

Our results show that the legacies of past human land useis most pronounced with respect to forest structure at the lo-cal scale, around and within the settlements. The forest issignificantly younger and birch much more abundant atboth settlement sites. The mean volume of birch is actuallyfive times higher at the Munka settlement than in the low-impact area. In addition, some deadwood characteristics dif-fer dramatically between the settlement sites and the low-impact area. For example, the volume of standing dead treesis significantly lower at both settlements than in the low-impact area. The cutting of standing dead trees in and

Fig. 6. (A) Number of growth releases recorded in the enclosure (open bars) and the low-impact area (solid bars), and the number of coredtrees overlapping in time displaying growth release (solid line); (B) reconstructed summer temperature (April–Sept.) based on latewoodmaximum density for the period 1600 to 2000 expressed as anomalies (in 8C) from the 1951–1970 mean (thin sold line) and filtered toemphasize climatic variability on a 30-year timescale (thick solid line) (cf. Grudd 2008).

Josefsson et al. 881

Published by NRC Research Press

around the settlement sites is reflected in the significantlylower volumes of this CWD category and decreased vol-umes of fallen logs, and in Munka, in the lower volumes ofCWD at the later decay stages. Two important processes —repeated cutting of trees (both dead and alive) and settle-ment activities (physical disturbance and nutrient addi-tion) — have resulted in there being fewer old trees, lowervolume and proportion of deadwood, and enhanced regener-ation of pine and birch trees, as well as a shift in tree spe-cies composition towards a forest with a distinct element ofbirch, at the settlement sites. Birch was an important re-source for the Sami, but it is not possible to determinewhether the increase in its presence was deliberately pro-moted by humans or whether it was an unintentional resultof settlement activities. Similar changes in forest structurenear Sami settlements was recorded by Hicks (1995) in east-ern Fennoscandia. Similarly, Black et al. (2006), studyingland use by indigenous peoples in northwestern USA sug-gested that these people were capable of producing dramaticchanges in tree species composition.

Comparing the former reindeer enclosure with the low-impact area revealed distinct patterns and differences in hu-man activity also at larger scales. The number of cultural re-mains per hectare inside the enclosure is almost three timeshigher than in the low-impact area. The variety and largenumber of cultural objects recorded in the enclosure indi-cates multiple land-use activities, including reindeer herding,cutting firewood, the use of different natural resources, stor-age, hunting, and fishing. Furthermore, the volume of livingtrees is much higher in the enclosure than in the low-impactarea. Interestingly, the number of small-diameter trees ofboth birch and pine is considerably higher inside the enclo-sure, whereas the number of larger birches and mid-sizedpines are similar in the two areas, and large-diameter pinesare almost absent from the enclosure (Fig. 4). It is probablethat this human-induced disturbance generated a more openforest with increased light conditions and that the positiveeffects of reindeer herding, i.e., grazing (Stark et al. 2000)and manuring in combination with trampling (Aronsson1991), have encouraged the regeneration of trees, and this ismirrored in the higher number of small-diameter pine andbirch trees. Consequently, the higher mean volumes of liv-ing trees recorded in the former enclosure mainly compriselarge numbers of small-diameter trees. Higher tree volumesin old abandoned reindeer pens than in the surrounding for-est were also recorded by Aronsson (1991) and Ostlund etal. (2003). However, in these studies, the high volumeswere related to the presence of large-diameter trees, mostlikely because of too excessive grazing and trampling in asmall area, thus preventing tree regeneration.

The ecological importance of land-use legacies on forestecosystems

One important result from this study is that protected for-ests in remote and inaccessible areas with no recent manage-ment cannot be indiscriminately used to represent pristineforests. Old forests that give the impression of being un-touched by humans might be ecologically valuable in termsof intactness and biodiversity. Yet, as shown in this study,they may also have been used by humans for long time peri-ods, causing changes in forest structure and composition, es-

pecially at local scales. This indicates that many pristineforests are, in fact, cultural landscapes with high ecologicaland cultural values (Gillson and Willis 2004; Burgi andGimmi 2007). We suggest that historical analysis of agestructure, tree growth release, past climatic conditions, andphysical signs of past disturbance is key to understandingwhat processes might have shaped current forest characteris-tics.

In previous studies of preindustrial human land use, it hasbeen shown that modification of the forest land by grazing,felling trees, the use of fire, or a combination of such dis-turbance types may have a great impact on the current forestecosystem (Ericsson et al. 2000; Bayliss-Smith et al. 2003).Within the present study area, it is probable that no largeareas of land were cleared or burned. Still, felling trees forvarious reasons has been carried out repeatedly for manycenturies and is believed to be of major importance for thedifferences in forest structure and composition shown in thestudy, especially at and near settlements. Furthermore, re-peated cutting of trees may not only change forest structureand composition, it may also affect other elements of theforest ecosystem. For example, the effects of tree cutting onmany different species groups can be considerable when thecontinuity and diversity of decaying wood is disrupted(Junninen et al. 2006). Felling standing dead trees may ad-versely affect species that depend on wood in the later decaystages. In addition, there may certainly be pronounced ef-fects of intensive reindeer herding on different ecosystemproperties. As suggested by Aronsson (1991) and Vare et al.(1996), who examined the impact of forest reindeer herdingon vegetation composition, the effect on the ground vegeta-tion can be substantial if reindeer are kept penned or aredriven together. We believe that this kind of disturbance isnot just an important factor affecting vegetation compositionin low-productive ecosystems, presumably, it also has an ef-fect on important ecosystem processes such as soil respira-tion and mineralization, biological productivity, andmaintenance of species diversity, as discussed by Stark etal. (2000).

In accordance with Foster et al. (2003) and Gillson andWillis (2004), we suggest that legacies of human activitycan have a profound effect on the forest ecosystem whencarried out over long time periods. Although a century haspassed since the settlements were abandoned, the legaciesof different kinds of land use, performed several hundredyears ago, are still clear. One important reason for this isthe relatively slow decay rate of deadwood that characterizesdry and low-productive forests, in which trees generally diewhile still standing and can remain upright for many deca-des or even centuries (Niklasson and Granstrom 2000).However, the environmental effects of past human land usedescribed in this study is not unique for this particular area.On the contrary, long-term legacies of human impact on for-est structure and composition have been shown also in otherecosystems, for example in temperate (McLachlan et al.2000) and tropical forests (Thompson et al. 2002). Further-more, the forms of human land use described in the presentstudy are typical for past Sami reindeer herding in northernSweden and, therefore, are expected to have affected foreststhroughout the Scandinavian mountain range. Ignoring thepossible effects of preindustrial land use by indigenous peo-

882 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

ple may result in methodological problems in comparativestudies on biodiversity as well as inappropriate conservationmanagement.

ConclusionThe forest of the Tjeggelvas Nature Reserve conforms to

the generally accepted impression of a pristine forest: a latesuccessional forest containing cohorts of different tree ages(including very old trees), high proportions of deadwood indifferent decay stages, and no history of commercial log-ging. Even so, this area has a long history of human landuse. There are distinct differences in structural componentswithin this landscape and some of these are due to differen-ces in topography, soils, and natural disturbances. Some pat-terns in tree recruitment and growth are related to variationsin past summer temperatures. However, in areas where theevidence of human activity over long periods of time isclear, there is a distinct difference in forest structure andcomposition compared with the rest of the study area. Thisshows that legacies of human activities may last for manycenturies. Not only do we believe that these disparities arestrongly linked to past land use, but we also suggest thatlong-term low-intensity land use may operate at differentspatial scales, especially within and around settlement sites,and also at larger scales as a result of for example reindeerherding for example. These findings are especially relevantwith respect to scientific research (for example obtainingecological baselines for biodiversity), ecological restoration,and ‘‘near-natural’’ management in forestry and conservationstrategies. This implies that the present structure and compo-sition of a forest have to be considered in the light of theirhistorical context before any attempt to assess the degree ofnaturalness (e.g., the selection of areas to represent referenceconditions and the biodiversity of pristine ecosystems) is made.

AcknowledgementsWe are grateful to Hakan Grudd for providing climate

data, Ulf Soderberg for methodological suggestions, SorenHolm and Michael Gundale for statistical support, Sees-editing Ltd. for linguistic corrections, Bjorn Helamb at Arc-tic Air for helicopter transportation, and The RegionalCounty Board of Norrbotten for permission to conduct thefield work in the nature reserve. We also wish to thank Kris-toffer Sivertsson, Asa Fjellborg, and Simon Halling for as-sistance in the field. This study was financially supportedby FORMAS.

ReferencesAlexandersson, H., and Eggertsson Karlstrom, C. 2001. Temperatu-

ren och nederborden i Sverige 1961–1990: Referensnormaler.2 ed. SMHI, Norrkoping.

Andersson, R., Ostlund, L., and Kempe, G. 2008. How to find therare trees in the forest — New inventory strategies for culturallymodified trees in boreal Sweden. Can. J. For. Res. 38(3): 462–469. doi:10.1139/X07-138.

Angelstam, P.K. 1998. Maintaining and restoring biodiversity inEuropean boreal forests by developing natural disturbance re-gimes. J. Veg. Sci. 9(4): 593–602. doi:10.2307/3237275.

Aniol, R.W. 1983. Tree-ring analysis using CATRAS. Dendrochro-nologia, 1: 45–53.

Aronsson, K.A. 1991. Forest reindeer herding A.D. 1–1800: an ar-chaeological and palaeoecological study in northern Sweden.Ph.D. thesis, Department of Archaeology, University of Umea,Umea. p. 125.

Bayliss-Smith, T., Hviding, E., and Whitmore, T. 2003. Rainforestcomposition and histories of human disturbance in Solomon Is-lands. Ambio, 32(5): 346–352. PMID:14571964.

Berg, A., Ostlund, L., and Gunnarson, B. 2009. ‘‘At this point, thelichens in the trees are their only means of survival’’ — the his-tory of tree cutting by native Sami people to feed their reindeer.Environ. Hist. In press.

Bergman, I., Passe, T., Olofsson, A., Zackrisson, O., Hornberg, G.,Hellberg, E., and Bohlin, E. 2003. Isostatic land uplift and Me-solithic landscapes: lake-tilting, a key to the discovery of Meso-lithic sites in the interior of orthern Sweden. J. Archaeol. Sci.30(11): 1451–1458. doi:10.1016/S0305-4403(03)00040-2.

Bergman, I., Liedgren, L.G., Ostlund, L., and Zackrisson, O. 2008.Kinship and settlements: Sami residence patterns in the Fenno-scandian alpine areas around A.D. 1000. Arctic Anthropol.45(1): 97–110. doi:10.1353/arc.0.0005.

Black, B.A., Ruffner, C.M., and Abrams, M.D. 2006. Native Amer-ican influences on the forest composition of the Allegheny Pla-teau, northwest Pennsylvania. Can. J. For. Res. 36(5): 1266–1275. doi:10.1139/X06-027.

Brandel, G. 1990. Volume functions for individual trees. Scots pine(Pinus sylvestris), Norway spruce (Picea abies) and birch (Betulapendula and Betula pubescens). Swed. Univ. Agric. Sci. Dep.For. Yield Res. Rep. 26. [In Swedish with English summary.]

Burgi, M., and Gimmi, U. 2007. Three objectives of historical ecol-ogy: the case of litter collecting in Central European forests.Landsc. Ecol. 22(S1): 77–87. doi:10.1007/s10980-007-9128-0.

Carcaillet, C., Bergman, I., Delorme, S., Hornberg, G., and Zackrisson,O. 2007. Long-term fire frequency not linked to prehistoric occupa-tions in northern Swedish boreal forest. Ecology, 88(2): 465–477.doi:10.1890/0012-9658(2007)88[465:LFFNLT]2.0.CO;2. PMID:17479764.

Denevan, W.M. 1992. The pristine myth: the landscape of theAmericas in 1492. Ann. Assoc. Am. Geogr. 82(3): 369–385.doi:10.1111/j.1467-8306.1992.tb01965.x.

Ebeling, F. 1978. Nordsvenska skogstyper. Sver. Skogsbruksforb.Tidskr. 76: 340–381.

Engelmark, O. 1987. Fire history correlations of forest type and to-pography in Sweden. Ann. Bot. Fenn. 24: 317–324.

Ericsson, S., Ostlund, L., and Axelsson, A.L. 2000. A forest ofgrazing and logging: deforestation and reforestation history of aboreal landscape in central Sweden. New For. 19(3): 227–240.

Flannigan, M.D., Bergeron, Y., Engelmark, O., and Wotton, B.M.1998. Future wildfire in circumboreal forests in relation to glo-bal warming. J. Veg. Sci. 9(4): 469–476. doi:10.2307/3237261.

Foster, D.R., Orwig, D.A., and McLachlan, J.S. 1996. Ecologicaland conservation insights from reconstructive studies of tempe-rate old-growth forests. Trends Ecol. Evol. 11(10): 419–424.doi:10.1016/0169-5347(96)10047-1.

Foster, D.R., Swanson, F., Aber, J.D., Burke, I., Brokaw, N., Til-man, D., and Knapp, A. 2003. The importance of land-use lega-cies to ecology and conservation. Bioscience, 53(1): 77–88.doi:10.1641/0006-3568(2003)053[0077:TIOLUL]2.0.CO;2.

Gillson, L., and Willis, K.J. 2004. ‘‘As Earth’s testimonies tell’’:wilderness conservation in a changing world. Ecol. Lett. 7(10):990–998. doi:10.1111/j.1461-0248.2004.00658.x.

Groven, R., Rolstad, J., Storaunet, K.O., and Rolstad, E. 2002.Using forest stand reconstructions to assess the role of structuralcontinuity for late-successional species. For. Ecol. Manage.164(1–3): 39–55. doi:10.1016/S0378-1127(01)00611-9.

Josefsson et al. 883

Published by NRC Research Press

Grudd, H. 2008. Tornetrask tree-ring width and density AD 500–2004: a test of climatic sensitivity and a new 1500-year recon-struction of north Fennoscandian summers. Clim. Dyn. 31(7–8):843–857. doi:10.1007/s00382-007-0358-2.

Gunnarson, B.E. 2008. Temporal distribution pattern of subfossilpines in central Sweden: perspective on Holocene humidity fluc-tuations. Holocene, 18(4): 569–577. doi:10.1177/0959683608089211.

Heckenberger, M.J., Kuikuro, A., Kuikuro, U.T., Russell, J.C.,Schmidt, M., Fausto, C., and Franchetto, B. 2003. Amazonia1492: Pristine forest or cultural parkland? Science, 301(5640):1710–1714. doi:10.1126/science.1086112. PMID:14500979.

Hicks, S. 1995. The history of a wilderness area in Finnish Laplandas revealed by pollen analysis. Arct. Centre Publ. 7: 126–140.

Holmes, R.L., Adams, R.K., and Fritts, H.C. 1986. Tree-ring chro-nologies of western North America: California, eastern Oregonand northern Great Basin, with procedures used in the chronol-ogy development work, including user manuals for computerprograms COFECHA and ARSTAN. Chronology Series VI. La-boratory of Tree-Ring Research, University of Arizona, Tuscon,Ariz.

Josefsson, T., Hornberg, G., and Ostlund, L. 2009. Long-term hu-man impact and vegetation changes in a boreal forest reserve:implications for the use of protected areas as ecological refer-ences. Ecosystems, 12(6): 1017–1036. doi:10.1007/s10021-009-9276-y.

Junninen, K., Simila, M., Kouki, J., and Kotiranta, H. 2006. As-semblages of wood-inhabiting fungi along the gradients of suc-cession and naturalness in boreal pine-dominated forests inFennoscandia. Ecography, 29(1): 75–83. doi:10.1111/j.2005.0906-7590.04358.x.

Kaye, M.W., and Swetnam, T.W. 1999. An assessment of fire, cli-mate and apache history in the Sacramento mountains, NewMexico. Phys. Geogr. 20(4): 305–330.

Lane, P. 2006. Oral histories and indigenous archaeology: an Afri-canist perspective. Comments on Charlotte Damm (2005): ar-chaeology, ethnohistory and oral traditions: approaches to theindigenous past. Norw. Archaeol. Rep. 39: 70–73.

Liedgren, L.G., Josefsson, T., and Ostlund, L. 2009. Samisk by-ggnadskultur - timrade kator och exemplet Blackajaure. Arkeo-logi i Norr, 11: 115–144.

Lilja, S., and Kuuluvainen, T. 2005. Structure of old Pinus sylves-tris dominated forest stands along a geographic and human im-pact gradient in mid-boreal Fennoscandia. Silva Fenn. 39(3):407–428.

Lundmark, L. 1982. Uppbord, utarmning, utveckling: det samiskafangstsamhallets overgang till rennomadism i Lule lappmark.Arkiv for studier i arbetarrorelsens historia, Lund.

McLachlan, J.S., Foster, D.R., and Menalled, F. 2000. Anthropo-genic ties to late-successional structure and composition in fourNew England hemlock stands. Ecology, 81(3): 717–733. doi:10.1890/0012-9658(2000)081[0717:ATTLSS]2.0.CO;2.

Mladenoff, D.J., White, M.A., Pastor, J., and Crow, T.R. 1993.Comparing spatial pattern in unaltered old-growth and disturbedforest landscapes. Ecol. Appl. 3(2): 294–306. doi:10.2307/1941832.

Naslund, M. 1940. Funktioner och tabeller for kubering av staendetrad: Tall, gran och bjork i norra Sverige. Meddelanden fran sta-tens skogsforsoksanstalt 32(4): 95–132.

Niklasson, M., and Granstrom, A. 2000. Numbers and sizes offires: long-term spatially explicit fire history in a Swedish boreallandscape. Ecology, 81(6): 1484–1499. doi:10.1890/0012-9658(2000)081[1484:NASOFL]2.0.CO;2.

Ostlin, E. 1963. Bark data for pine, spruce, birch, etc. Part 1. Barkdata for provinces and regions. Department of Forest Survey. 5.

Ostlund, L., Ericsson, T.S., Zackrisson, O., and Andersson, R.2003. Traces of past Sami forest use: an ecological study of cul-turally modified trees and earlier land use within a boreal forestreserve. Scand. J. For. Res. 18(1): 78–89. doi:10.1080/0891060310002363.

Ostlund, L., Keane, R.E., Arno, S.F., and Andersson, R. 2005. Cul-turally scarred trees in the Bob Marshall Wilderness, Montana,USA — interpreting Native American historical forest use in awilderness area. Nat. Areas J. 25(4): 315–325.

Renvall, P. 1995. Community structure and dynamics of wood-rotting Basidiomycetes on decomposing conifer trunks in north-ern Finland. Karstenia, 35: 1–51.

Rhemtulla, J.M., Hall, R.J., Higgs, E.S., and Macdonald, S.E. 2002.Eighty years of change: vegetation in the montane ecoregion ofJasper National Park, Alberta, Canada. Can. J. For. Res. 32(11):2010–2021. doi:10.1139/x02-112.

Rouvinen, S., Kuuluvainen, T., and Karjalainen, L. 2002. Coarsewoody debris in old Pinus sylvestris dominated forests along ageographic and human impact gradient in boreal Fennoscandia.Can. J. For. Res. 32(12): 2184–2200. doi:10.1139/x02-144.

Ruong, I. 1969. Samerna. Aldus/Bonnier, Stockholm.Schulte, L.A., Mladenoff, D.J., Crow, T.R., Merrick, L.C., and Cle-

land, D.T. 2007. Homogenization of northern US Great Lakesforests due to land use. Landsc. Ecol. 22(7): 1089–1103. doi:10.1007/s10980-007-9095-5.

Segerstrom, U. 1997. Long-term dynamics of vegetation and distur-bance of a southern boreal spruce swamp forest. J. Veg. Sci.8(2): 295–306. doi:10.2307/3237359.

Siitonen, J. 2001. Forest management, coarse woody debris and sa-proxylic organisms: Fennoscandian boreal forests as an example.Ecol. Bull. 49: 11–41.

Stark, S., Wardle, D.A., Ohtonen, R., Helle, T., and Yeates, G.W.2000. The effect of reindeer grazing on decomposition, minera-lization and soil biota in a dry oligotrophic Scots pine forest. Oi-kos, 90(2): 301–310. doi:10.1034/j.1600-0706.2000.900210.x.

Thompson, J., Brokaw, N., Zimmerman, J.K., Waide, R.B., Ever-ham, E.M., Lodge, D.J., Taylor, C.M., Garcia-Montiel, D., andFluet, M. 2002. Land use history, environment, and tree compo-sition in a tropical forest. Ecol. Appl. 12(5): 1344–1363. doi:10.1890/1051-0761(2002)012[1344:LUHEAT]2.0.CO;2.

Uotila, A., Kouki, J., Kontkanen, H., and Pulkkinen, P. 2002. As-sessing the naturalness of boreal forests in eastern Fennoscandia.For. Ecol. Manage. 161(1–3): 257–277. doi:10.1016/S0378-1127(01)00496-0.

Vare, H., Ohtonen, R., and Mikkola, K. 1996. The effect and theextent of heavy grazing by reindeer in oligotrophic pine heathsin northeastern Fennoscandia. Ecography, 19: 245–253.

Willis, K.J., Gillson, L., and Brncic, T.M. 2004. Ecology: How‘virgin’ is virgin rainforest? Science, 304(5669): 402–403.doi:10.1126/science.1093991. PMID:15087539.

Zackrisson, O., Nilsson, M.C., Steijlen, I., and Hornberg, G. 1995.Regeneration pulses and climate vegetation interactions in non-pyrogenic boreal Scots pine stands. J. Ecol. 83(3): 469–483.doi:10.2307/2261600.

Zackrisson, O., Ostlund, L., Korhonen, O., and Bergman, I. 2000.The ancient use of Pinus sylvestris L. (Scots pine) inner barkby Sami people in northern Sweden, related to cultural and eco-logical factors. Veg. Hist. Archaeobot. 9(2): 99–109. doi:10.1007/BF01300060.

884 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press