Fire as an important factor for the genesis of boreal Picea abies swamp forests in Fennoscandia

Abstract

The initial establishment of Picea abies in Sweden and Norway on a landscape level, between 3000 and 1000 years ago, was often preceded by recurrent fire and thereafter the influence of fire decreased. However, in some swamp forests, the absence of fire over the last 3500 years has promoted the continuous presence of deciduous trees, i.e. Picea has not established although it has been present regionally for over 3000 years. Our objective was to study long-term vegetation development and fire history in a Picea swamp forest located close (c. 600 m) to a deciduous swamp forest with a documented fire-free history in northernmost Sweden. The study included analyses of charred particles, pollen and ignition residues. Principal component analysis was applied to identify major changes in the pollen spectra. Our results showed that the current Picea swamp forest has developed from a deciduous fen and that fires affected the fen between 6700 and 2300 cal. yr BP. Picea abies established on the fen around 2200 cal. yr BP, following the last local on-site fire. The main factors responsible for the local vegetation development have been: fire (6700 to 2300 cal. yr BP); autogenous processes and climate (2300 to 1000 cal. yr BP); autogenous processes or anthropogenic impact (1000 to 300 cal. yr BP); anthropogenic impact through selective cutting and grazing (300 to 100 cal. yr BP); and autogenous processes and grazing (100 cal. yr BP to present). We conclude that fire facilitated the initial Picea abies establishment. Once established, Picea abies created local conditions that in combination with a colder and wetter climate prevented fire and the establishment of other tree species.

Keywords

charred particles fire history loss-on-ignition pollen succession

Introduction

It is generally accepted that boreal forest ecosystems have been characterized by frequent fires throughout the Holocene (Bonan and Shugart, 1989; Heinselman, 1981; Payette, 1992; Schimmel, 1993; Zackrisson, 1977). However, both spatial and temporal variations in fire cycles have been recorded (Niklasson and Granström, 2000), and some recent studies even indicate that fire-free intervals may have been much longer than previously thought (Bergeron et al., 2002; Carcaillet et al., 2007; Pitkänen et al., 2002, 2003). Non-pyrogenic boreal forest stands also seem to have been more common in Fennoscandia in comparison with forests in North America and in Russia (Ohlson et al., 2009, 2011). After stand-replacing fires in northern Europe, post-fire succession generally starts with pioneer tree species such as Betula spp. (birch), Populus tremula (aspen), Salix caprea (goat willow) and Pinus sylvestris (Scots pine) (Engelmark and Hytteborn, 1999; Esseen et al., 1997), nomenclature follows Mossberg and Stenberg (2003). The deciduous trees are, in time, replaced and succession normally results in the establishment of secondary species; in boreal Fennoscandia this tends to be Picea abies (Norway spruce). Picea abies has been an important component of the forests of northeastern Europe throughout the entire Holocene (Huntley and Birks, 1983). During the early Holocene in Fennoscandia, however, Picea abies was present in small, spatially restricted and isolated populations in the Scandinavian mountain range (Kullman, 1995, 2001; Segerström and von Stedingk, 2003). Forests largely dominated by Picea abies developed and spread westwards from the forests of northeastern Europe and became common c. 3400 cal. yr BP in eastern Sweden (Giesecke, 2004; Seppä et al., 2009; Tallantire, 1972, 1977), and c. 2200 cal. yr BP in eastern Norway (Hafsten, 1992). The prerequisites for this successful expansion are still unclear but probably included coincidental climatic changes and biotic factors (Giesecke and Bennett, 2004; Hafsten, 1992; Tallantire, 1972). Today Picea abies is the most common tree species in Sweden partly as a result of modern silviculture, and covers 43% of the forested area (Anonymous, 2004). Because Picea abies is the only Picea species that occurs naturally in Sweden and in Fennoscandia it will hereafter be referred to as Picea in the following text.

Old growth Picea stands on mesic to moist sites usually include trees of varying size and age, and their gap-phase dynamics are mainly driven by small-scale disturbances (Hofgaard, 1993; Hytteborn et al., 1987; Kuuluvainen et al., 1998; Norokorpi, 1979). Thus, as Picea is able to regenerate under a closed canopy it often manages to dominate the tree cover until the next stand-replacing disturbance, when a new succession is initiated (cf. Sirén, 1955; Wallenius et al., 2005). Picea is also the most common tree species in many old-growth boreal swamp forests in Sweden. These swamp forests, especially the more productive ones, have a high biodiversity (Hörnberg et al., 1998; Ohlson et al. 1997), and they have an inverse-J shaped tree-age structure indicating a continuous regeneration over time, in which the oldest trees often exceed 300 years of age (Hörnberg et al., 1995). Such an age structure suggests that they have not been affected by recurrent fires for a long period (cf. Kuuluvainen et al., 1998; Pitkänen et al., 2003).

Picea swamp forests are often located in concave landscape features where fire frequency is generally lower than in forests on convex landscape features (Zackrisson, 1977). Owing to excess water, the swamp forests are characterized by Sphagnum mosses; these have higher moisture content than pleurocarpus mosses that are important for the build up of fuel and the spread of ground fires on drier sites (Schimmel, 1993). Thus, the swamp forests are considered to be one of the least fire-disturbed forest ecosystems in the boreal part of Fennoscandia (Angelstam, 1998; Hörnberg et al., 1995; Pitkänen et al., 2003), although many of them have not been totally unaffected by fire or human impact (Hörnberg et al., 1998; Segerström, 1997; Segerström et al., 1994). However, little is known about the fire history and vegetation development prior to the establishment of Picea in these swamp forests. Some studies have suggested that the initial establishment of Picea was preceded by fire, but thereafter the influence of fire ceased (Molinari et al., 2005; Ohlson and Tryterud, 1999; Ohlson et al., 2011; Segerström, 1997; Segerström et al., 1996; Tryterud, 2003).

In a study of a deciduous swamp forest in the Kuusivaara nature reserve, northern Sweden, it was suggested that the absence of fire over the last 3500 years has promoted the continuous presence of Betula pubescens and Alnus incana, i.e. that Picea has not been able to establish and out-compete the deciduous tree cover in the swamp forest although Picea has been present in the region and within the reserve for over 3000 years (Segerström et al., 2008). Importantly, only c. 600 m to the south of the deciduous swamp forest studied by Segerström et al. (2008), there is a Picea swamp forest, and within a 1 km radius Picea is a common tree in the forest. Thus, the obvious question arises: why have two adjacent swamp forests exposed to the same climatic and geological conditions developed such different vegetation? What factors or processes triggered the vegetation change and the establishment of Picea in the actual swamp forest in this case?

Our objective, therefore, was to study the vegetation dynamics and disturbance history of this boreal Picea swamp forest in the Kuusivaara nature reserve, and to compare the results with the results from the deciduous swamp forest studied by Segerström et al. (2008). Our study includes analyses of charred particles, pollen and loss-on-ignition. Principal component analysis (PCA) was used to identify changes in the pollen spectra. We raise the following questions. (1) How has the Picea swamp forest vegetation changed over time? (2) How often and to what extent has the Picea swamp forest been affected by fire? (3) When did Picea establish in the swamp forest and was there a link between fire and the establishment of Picea? (4) What factors or processes can be linked to the major changes in the Picea swamp forest vegetation in the past? Future effects of predicted climate change on present fire patterns and vegetation composition are also discussed.

Material and methods

Study area

The study area, the Kuusivaara nature reserve (66°40′N: 23°18′E), is situated between Pajala and Övertorneå in the county of Norrbotten, northern Sweden (Figure 1). The bedrock is characterized by granites and gneisses, the soils are formed over till and vast peatlands (From, 1965), and the hills are no more than 325 m above sea level (m a.s.l.). A substantial proportion of land occurs below the highest marine limit in this region, c. 180 m a.s.l., as a result of land uplift. The climate is continental with a mean precipitation of 500 to 600 mm/yr, and 180 days of snow cover (Ångström, 1974). The nature reserve covers c. 1520 ha, including three hills with an altitude of 175 to 325 m a.s.l., and is surrounded by large mires on all sides (Anonymous, 1997). Most (c. 65%) of the forest reserve is characterised by Picea forests on slopes covered by various types of mesic and moist soil. Picea swamp forests occur in depressions, with scattered Betula pubescens, Alnus incana and Pinus sylvestris. Open mires cover c. 20% of the land and a few mesic areas are characterised by deciduous trees. Mixed Pinus–Picea forests with scattered Populus tremula and Sorbus aucuparia trees occur on only c. 12% of the dry and higher elevated areas. Human impact in the reserve before the nineteenth century is largely unrecorded. Only a small area was selectively cut during the late nineteenth century (Anonymous, 1997), and the nearest village is situated on the opposite side of Lake Ruokojärvi (155 m a.s.l.), 2.5 km to the southwest of the edge of the reserve. Owing to the undisturbed structure of the forests, the area was already being preserved in 1957 by the state-owned forest company Domänverket, but it was officially designated a nature reserve in 1997.

Figure 1.

Map of Fennoscandia showing the location of the Kuusivaara nature reserve in NE Sweden (arrow). The Picea swamp forest is marked as a square, the deciduous swamp forest studied by Segerström et al. (2008) is marked as an ellipse

The study site is a Picea swamp forest situated in a narrow depression at c. 230 m a.s.l. between two plateaux (Figure 1), and c. 600 m from the deciduous swamp forest previously studied by Segerström et al. (2008). The swamp forest covers about 1 ha (50 m × 200 m), is south-facing and has no major inflows or outlets. However, further south in the depression a small brook appears which diverts excess water. The site has a closed canopy of Picea trees and scattered specimens of Alnus incana and Betula pubescens. The shrub vegetation is characterized by a number of Salix species and the dwarf-shrubs are all ericaceous species. Comarum palustre, Rubus chamaemorus, Ranunculus lapponica and Cornus suecica are common herb species in the swamp and occur together with sedges (e.g. Carex canescens, C. loliacea and C. globularis) and Equisetum palustre. The moss layer is composed of Sphagnum spp. and Mnium spp. The present swamp forest vegetation is typical of the wet and rather nutrient-poor conditions in this region. The location between the plateaux implies a strong impact of surface run off during snowmelt and after heavy rainfall.

The deepest part of the swamp forest was identified through systematic coring using a ‘Russian’ peat corer (cf. Jowsey, 1966), and entire peat-core sections of 130 cm were collected in duplicate from this basin. The degree of peat decomposition was medium to high throughout the core. The main methods used are the same as the ones used by Segerström et al. (2008) in order to make the results as comparable as possible (cf. Tinner and Hu, 2003).

Radiocarbon dating, age–depth model and charred particle analysis

Selected samples of terrestrial macrofossils (mosses and seeds) from six levels were dated by means of AMS ¹⁴C dating, performed at The Ångström Laboratory in Uppsala, Sweden. Calib 5.01, Intcal 04.14c (Stuiver and Reimer, 1993) was used for age calibrations and the dates are presented as cal. yr BP (before present = ad 1950). Two of the samples for dating were taken from adjacent depths to enhance the credibility of the dating at the point where Picea pollen percentages increased and became a permanent component of the pollen record. Peat accumulation rates were estimated using TILIA software (Grimm, 1991) and all dates were used for the computation of the age–depth models calculated using polynominals (third order) based on median probability age. Charred particles were recorded in each 1 cm layer down the entire 130 cm of the peat core, and concentrations per gram dry peat were calculated. The peat samples were dried at 105°C, and weighed. Thereafter the samples were diluted with KOH (5%) in a petri dish and charred particles counted. Macroscopic charred particles >0.25 mm were recorded using a stereomicroscope with a magnification of ×15 to ×40. The particles were recorded in three size classes; 0.25–0.50 mm, 0.51–1.0 mm and >1.0 mm.

The recording method and size categories used were intended to enable the identification of strictly local, on-site fires, to differentiate these from forest fires in the surrounding vegetation (Ohlson and Tryterud, 2000; Segerström, 1997; Tinner et al., 1998; Whitlock and Millspaugh, 1996), and to make the charcoal results comparable with the study by Segerström et al. (2008). The spatial variations in the occurrence of charred particles in the peat are important (Ohlson et al., 2006). Several studies have discussed the problems in assessing the origin of charred particles on the basis of size (Clark and Royall, 1995; Froyd, 2006; Lynch et al., 2004), since macroscopic charred particles may be transported long distances (Pisaric, 2002; Tinner et al., 2006). According to Segerström et al. (2008), charred particles 10 μm to 1.0 mm are not unambiguous indicators of local, on-site fires because they exhibit variations in abundance and their abundance is not necessarily correlated with that of larger charred particles (>1.0 mm). In the present study we considered that a fire had occurred on-site only when the occurrence of charred particles >1.0 mm coincided with distinct vegetation changes as indicated by the pollen data (cf. Segerström et al., 2008). To test this assumption, the charred particle data (total counts of all size fractions) was also analysed using the program CHARSTER 0.8.3 developed by David Gavin (2006) that identify peaks in the charcoal record that may be the result from nearby fires. The calculations followed the framework of the CHARSTER programme (see Gavin et al., 2006) and were based on concentration values, the records were re-sampled and the background was estimated using LOWLESS. The peak component was calculated as difference and the threshold value was selected from the value given by the zero-mean Gaussian model (95th percentile). No transformation of the data was done.

Pollen analysis, loss-on-ignition analyses and multivariate statistics

For pollen analysis, every second centimetre (each sample comprising 1 cm) was excised from the peat core and analysed (a total of 65 layers). The sample preparation included digestion in KOH (5%), sieving through a 1 mm mesh, and standard acetolysis (Moore et al., 1991). No decanting to remove larger mineral particles was undertaken. The samples were stained with safranine, and mounted in glycerine on microscope slides. From each sample at least 500 pollen grains were counted, and the percentage frequencies were calculated based on the total terrestrial pollen, including Cyperaceae. The key by Moore et al. (1991) was used for pollen identification, and for the determination of critical species, a reference pollen collection was available at the Department of Forest Ecology & Management, SLU, Umeå. Microscopic charred particles between 50 and 150 μm, were also recorded in the pollen samples. The TILIA and TILIA GRAPH programs by Grimm (1991, 2004) were used to draw the pollen diagram. Based on major changes in the pollen record identified by the multivariate analysis, the diagram was divided into pollen assemblage zones (PAZ). The mineral particle content in the peat was determined by loss-on-ignition analysis for each centimetre from the peat core. Each sample was dried at 105°C overnight, weighed, and burned at 550°C for 4 h, and then the ignition residues weighed (Heiri et al., 2001).

The computer program CANOCO version 4.5 (ter Braak and Smilauer, 2002) was used to analyze changes in the pollen spectra from the Picea swamp forest studied here and from the nearby deciduous swamp forest previously studied by Segerström et al. (2008). First, detrended correspondence analysis (DCA) was used to test the gradient length. The options were to use either a linear or a non-linear approach for further multivariate analyses, and the decision was based on the gradient length of the first axis from the DCA analysis (cf. Birks, 1998). The DCA analysis indicated that there was a linear response; the gradient length was 1.4 standard deviations for the deciduous swamp forest data, and 1.27 for the Picea swamp forest. Thus a linear approach (principal component analysis; PCA) was used for the analyses of changes in the pollen spectra in the two swamp forests.

Results

Radiocarbon dating, age–depth model and charred particle analysis

Peat started to accumulate c. 6700 cal. yr BP (Table 1), and the mean accumulation rate has been about 0.2 mm/yr, with lower rates between c. 6700 and 2000 cal. yr BP, and higher rates during the last c. 2000 years (Figures 2 and 3). The average sample resolution is about 50 years. The two samples from 80–81 and 81–82 cm date the establishment of the local Picea abies forest (percentage of Picea pollen increases rapidly). The dates give consistent results: c. 2150–2340 cal. yr BP and c. 2210–2690 cal. yr BP, respectively (Table 1, Figure 2) and are statistically the same at 95% level (Chi-2-test, Calib 5.01, Intcal 04.14c; Stuiver and Reimer, 1993).

Table 1.

Radiocarbon dates of selected samples of terrestrial plant macrofossils (mosses and seeds) from the Kuusivaara Picea swamp forest, and calibrated age according to Calib 5.01, IntCal 04.14c, max and min of 2σ calibrated age range and median values BP, i.e. ad 1950 (Stuiver and Reimer, 1993)

Lab code	Depth (cm)	¹⁴C age±SD (BP)	Calibrated age BP (2 σ)	Calibrated age BP (probability)	Median value cal. yr BP
Ua 19535	34–35	105±40	Present (ad 1950)–271	−4 to −1 (0.01)	134
				11–150 (0.66)
				186–271 (0.33)
Ua 19536	50–51	1095±45	927–1166	927–1089 (0.98)	1046
				1108–1129 (0.01)
				1163–1166 (0.01)
Ua 19537	80–81	2240±45	2152–2341	2152–2341 (1.00)	2246
Ua 20504	81–82	2360±45	2211–2694	2211–2220 (0.01)	2452
				2310–2500 (0.90)
				2596–2613 (0.02)
				2637–2694 (0.07)
Ua 21624	90–91	3649±45	3846–4085	3847–4089 (1.00)	3966
Ua 19538	129–130	5835±55	6496–6777	6496–6752 (0.99)	6636
				6764–6777 (0.01)

Figure 2.

Radiocarbon dating distributions according to depth and age–depth models of the Kuusivaara Picea swamp forest. A third-order polynomial provided the best fit

Figure 3.

Pollen percentages of selected taxa, pollen assemblage zones (PAZ), ignition residue and charred particle occurrences from Kuusivaara Picea swamp forest. Silhouette is 10× the pollen percentage, dashed lines indicate levels where charred particles of all three size classes appear. Only pollen types important for the discussion are shown. Pollen types occurring at low frequencies have been omitted but are still included in the calculated sums

Charred particles 0.25 to 1.0 mm are most frequent in the transition between the mineral and peat soil (c. 6700 cal. yr BP), between 117 and 110 cm (c. 5500–4800 cal. yr BP), and between 45 and 40 cm (c. 800 to 500 cal. yr BP, Figure 3). Charred particles >1.0 mm were only recorded in eight layers in total; these were all between 130 and 84 cm (c. 6700 to 2500 cal. yr BP), and the highest concentrations occurred in the oldest part of the core (Figure 3). All three size fractions of charred particles were present in six of these samples corresponding to two occurrences around c. 6700 and 4700 cal. BP, respectively, and one occurrence c. 3900 and another c. 3500 cal. BP. On the pollen slides only one microscopic (>50 μm) charred particle was recorded (at 5 cm). This was surprising considering the number of charred particles obtained by the other method. Five peaks of charred particles were identified by CHARSTER; at 6710, 5990, 4710, 3460 and 2310 cal. yr BP, respectively.

Pollen analysis, loss-on-ignition analyses and multivariate statistics

The pollen diagrams (Figure 3 and 4) were divided into three main pollen assemblage zones (PAZ I to III), and three subzones (PAZ II a–c). These appear to represent the key changes in the local vegetation composition and are identified by the PCA-analysis. The ignition residues are high at the beginning and at the end of PAZ I, decrease during PAZ IIa and exhibit generally low values during PAZ IIb-c, and PAZ III (Figure 3). The 1st principal component captured 62.8% of the variance mainly emanating from the abrupt change due to local Picea establishment (Figure 5). The 2nd principal component explains another 11.5% of the variance. The second component reflects local on-site changes; variations in Filipendula and Betula pollen, peak in Alnus pollen at 35 cm depth and the rather recent increase in pine pollen.

Figure 4.

Pollen percentage diagram from Kuusivaara Picea swamp forest, silhouette is 10× the percentage

Figure 5.

Principal component analysis (PCA) of the pollen spectra from the spruce swamp forest in Kuusivaara. The 1st principal component captured 62.8% of the variance mainly emanating from the abrupt change because of local spruce establishment c. 2300 cal. BP (Figure 5). The 2nd principal component explains another 11.5% of the variance and reflects local on-site change; mainly the changes in Filipendula, birch and alder pollen, fern spores, and the rather recent increase in pine pollen

PAZ I, Betula–Alnus, c. 6700 to 2300 cal. yr BP (130–82 cm)

The arboreal pollen (AP) recorded are mainly from Betula, Pinus and Alnus, with scattered occurrences of Picea, Populus, Tilia and Ulmus (Figure 4). In the non-arboreal pollen (NAP) Salix and Juniperus are the most common shrub species, and Filipendula, Rosaceae undiff., Saxifraga stellaris-type, Cyperaceae and Poaceae are the most frequent herbaceous pollen types. Pollen from Asteraceae, Melampyrum, Potentilla-type, Galium and Ericaceae are regularly recorded, and Epilobium pollen is identified at three depths. Spores from Polypodiaceae occur very frequently, and spores from Lycopodium annotinum, Equisetum and Sphagnum are present. The ignition residues exceeding 30% are between 130 and 125 cm, and between 106 and 82 cm (Figure 3).

PAZ IIa, Betula–Alnus–Picea, c. 2300 to 1000 cal. yr BP (82–50 cm)

The AP mainly comprises Betula, Alnus, Picea and Pinus (Figure 4). The proportions of pollen from Salix and Ericaceae species, and spores from Lycopodium annotinum increase during this period. Pollen from Melampyrum, Poaceae and Cyperaceae are recorded in approximately the same proportions as in PAZ I, but the proportions of Juniperus, Filipendula, Saxifraga stellaris and Rosaceae undiff. pollen, and spores from Polypodiaceae decrease. The ignition residues decrease to between 16 and 18% (Figure 3).

PAZ IIb, Picea–Betula–Alnus, c. 1000 to 300 cal. yr BP (50–35 cm)

The proportions of Alnus and Betula pollen decrease but the proportions of Pinus and Picea pollen increase (Figure 4). The proportions of Salix, Asteraceae, Poaceae, Cyperaceae, Melampyrum, Potentilla-type and Rosaceae undiff. pollen increase. Occasional occurrences of pollen from Juniperus and Artemisa vulgaris-type are recorded, and one Secale cereale pollen grain is found. The proportion of Polypodiaceae spores decreases markedly. The ignition residues decrease substantially to values below 10% (Figure 3).

PAZ IIc, Betula–Alnus–Picea, c. 300 to 100 cal. yr BP (estimated for 35–15 cm)

Initially, the proportion of Alnus pollen increases and Picea, Betula and Pinus percentages decrease (Figure 4). The proportions of Ericaceae, Poaceae, Melampyrum and Potentilla-type pollen increase, Salix pollen occur in the same proportion as before and pollen from Juniperus, Caryophyllaceae, Chenopodiaceae, Rumex acetosa/acetosella, Artemisia vulgaris-type, Rubus chamaemorus and Epilobium are also recorded. Single pollen grains of the Cannabis-type and Hordeum-type are found. There is also a temporary increase in the percentages of Polypodiaceae spores. The ignition residues are low, except for a minor rise between 28 and 24 cm (Figure 3).

PAZ III, Picea, c. 100 cal. yr BP to the present (estimated for 15–0 cm)

The percentages of Pinus and Picea pollen increase and the proportions of Betula and Alnus decrease (Figure 4). The percentage of Poaceae increases, Salix, Potentilla-type and Rosaceae undiff. remain at the same level as in the previous zone, the proportion of Ericaceae pollen is variable, and Melampyrum pollen and Polypodiaceae spores decrease. Some pollen grains of Artemisia vulgaris-type, Caryophyllaceae and Rumex acetosa/acetosella are recorded. The ignition residues increase slightly but, in general, remain below 12% (Figure 3).

Discussion

The Picea swamp forest studied exhibits a history of vegetation development during the last 2300 years that differs markedly from the nearby deciduous swamp forest (Figure 6, Table 2) studied by Segerström et al. (2008). Climate has probably had the same general impact regarding fire patterns (Carcaillet et al., 2007) and autogenous processes (Kuhry, 1994) for the whole reserve, but local differences and chance have made the output different (cf. Gavin et al., 2006). During the period 6700 to 2300 cal. yr BP, however, both sites were dominated by deciduous trees. We found that the presence of charred particles of 0.25–1.0 mm alone did not correspond with any major changes in the pollen record in the Picea swamp forest (Figure 3). In contrast, charred particles >1.0 mm in size are recorded at eight levels, of which five of these occur at the same levels as changes in the pollen records for Alnus, Betula and finally Picea. The co-occurrence of large charred particles and changes in the pollen record suggests local on-site fires, as was also indicated in the study by Segerström et al. (2008). The identified peaks of charred particle occurrences in CHARSTER support this interpretation in four out of five occasions; i.e. c. 6700, 4700, 3500 and 2300 cal. yr BP. The two sets of charred particle data match each other reasonably well, although not perfectly. The occurrence of local fires is also indicated by Epilobium (Gobet et al., 2003; Tinner et al., 2006). Picea was not locally present and Pinus was unchanged and represented by a regional background component. This interpretation is also supported by the ignition residues data, which indicate increased erosion during this period. Increased erosion can include the inflow of older charred particles that are embedded in the mineral soils. However, in this case the erosion is restricted to small sized, easily eroded material that can be transported by wind or surface water. It is by no means a question of erosion turbidities or mass movements. The analyzed material is peat throughout, just carrying a higher content of mineral matter. Therefore, we consider erosion changes to be the most likely explanation. Consequently the interpretations and discussion below are based on the occurrence of charred particles >1.0 mm and the peaks in the charred particle record. Hence, fires that affected the local vegetation composition, increased erosion and produced a large amount of charred particles >1.0 mm also seen as peaks in the charred particle record are likely to have been identified, whereas low impact fires that did not substantially affect the local vegetation, and only produced a small amount of charred particles, will not have been apparent. The location, size, and intensity of fire, together with wind conditions, and local topographical and geographical factors all influence the production and spread of charred particles (Tinner and Hu, 2003; Tinner et al., 2006). It is therefore difficult to formulate general rules regarding the correlation between charred particles and their origin. The almost total absence of charred particles on the pollen slides is probably the result of a low abundance in general. It is also possible that the number of charred particles has been reduced by the pollen preparation technique as found by Tinner et al. (1998) and Tinner and Hu (2003). Some charred particles may have been attached to the peat substrate and would have been caught in the sieve at the time of preparation.

Figure 6.

Principal component analysis (PCA) including both the deciduous (data from Segerström et al., 2008) and the spruce swamp forest pollen data to reveal the similarities and differences in the long-term development between the two nearby swamp forest sites. The dates included in the PC trajectories are calibrated radiocarbon dates BP. The insert figure shows the species scores (i.e. the loadings) from the PCA analysis

Table 2.

Summary of vegetation development and disturbance history in the Kuusivaara swamp forests

Time (cal. yr BP)	Picea swamp forest development (this study)	Disturbance	Deciduous swamp forest development (after Segerström et al., 2008)	Disturbance
8000	No data	No data	Betula–Alnus wet forest	Local on site fire
7000	Alnus–Betula fen	Local on-site fire
6000	Betula–Alnus fen	Regional fire	Alnus–Betula fen	Local on site fire
5000	Betula–Alnus fen	Regional fire
4000	Alnus–Betula fen	Local on-site fire	Betula fen Betula–Alnus swamp forest^b	Local on site fire Regional fire
3000	Alnus–Betula fen MixedPicea–deciduous swamp forest^a	Local on-site fireRegional fire
2000	Mixed Picea–deciduous swamp forest	Regional fire/trampling, grazing,	Betula swamp forest	Regional fire
1000	Picea swamp forest	Cutting, trampling, grazing No fire	Betula swamp forest	Regional fire/ trampling, grazing No fire
Present

On-site establishment of Picea swamp forest c. 2200 cal. yr BP in this study.

Establishment in the nature reserve of Picea forest c. 3500 cal. yr BP around the deciduous swamp forest studied by Segerström et al. (2008).

The main results suggest that: (1) the current Picea swamp forest developed from a deciduous fen initially dominated by Betula and Alnus. Betula sustained its presence during the fire-free period, 6000–4800 cal. yr BP, whereas Alnus increased during the period of increased disturbance between 4800 and 2300 cal. yr BP. (2) Fires affected the fen c. 6700 cal. yr BP, 6000 cal. yr BP and recurrently between 4800 and 2300 cal. yr BP. Thereafter fires only affected the surrounding plateaux. (3) Picea established in the fen following the last local on-site fire around 2300 cal. yr BP and has characterized the tree layer since c. 2200 cal. yr BP. During 2800 to 2300 BP climate became colder and wetter. (4) The main factors affecting vegetation development in the fen and the swamp forest have been: fire (6700 to 2300 cal. yr BP); autogenous processes and climate (2300 to 1000 cal. yr BP); autogenous processes or anthropogenic impact (1000 to c. 300 cal. yr BP); selective cuttings and forest grazing (c. 300 to 100 cal. yr BP); and autogenous processes and grazing (c. 100 cal. yr BP to the present). These results indicate that periods during which there was little fire promoted the presence of Betula (Segerström et al., 2008), that fire preceded and probably also encouraged Picea establishment (Segerström et al., 1996; Tryterud, 2003), and that fires have not occurred since then (Segerström, 1997). The role of climate change has been discussed in terms of the establishment of Picea in northern Sweden (Giesecke, 2004; Tallantire, 1972, 1977). We cannot rule out the role of climate for Picea here. The proposed cooling between 2800 and c. 2300 BP (Bond et al., 2001; Grosjean et al., 2007; Haas et al., 1998; Heiri et al., 2006; Wanner et al., 2000) may have enhanced the establishment of Picea after the fire, but the effects of this climate cooling are not detectable in our results. The observed 1300 year local lag in the establishment of Picea speaks against climate as the sole trigger (see below).

Deciduous fen; 6700 to 2300 cal. yr BP (PAZ I)

The pollen assemblages suggest that a deciduous fen established in the depression, and that Pinus forest covered the surrounding plateaux (Figures 3 and 4). The fen field layer was characterised by herbaceous species and ferns. Occurrences of charred particles >1.0 mm, charred particle peaks and high loss-on-ignition residues indicate that the deciduous fen and the surrounding terraces were affected by fire around 6700 cal. yr BP and 5990 cal. yr BP (Figure 3). During this period climate was warm (Haas et al., 1998) and fire interval was short in northern Sweden (Carcaillet et al., 2007). These fire events and the high surface run-off caused soil erosion, thus depositing mineral particles and nutrients in the depression. As the impact of fire decreased, peat started to accumulate and the vegetation became denser. A period with no local on-site fire and fewer disturbances covering c. 1300 years (c. 6000–4700 cal. yr BP) and the lack of Picea in the region promoted uninterrupted Betula vegetation in the fen. This development is very similar to the development of the deciduous swamp-forest nearby described by Segerström et al. (2008), where the lack of fire, low levels of disturbance and absence of Picea resulted in a late succession stage characterised by Betula. The similarity between the two sites is verified in the PCA that includes both the deciduous (data from Segerström et al., 2008) and the Picea swamp forest pollen data (Figure 6). The beginning of the development of the deciduous swamp forest (c. 8000 cal. yr BP) was characterized by the absence of ferns, and therefore separating the two sites in the PC chart. However, from about 6700 cal. yr BP (the onset of the Picea swamp forest site) until c. 4000 cal. yr BP the two sites reveal a fairy similar development. The vegetation changes were minor and both sites were characterized by long term stability. About 4000 cal. yr BP in the deciduous swamp forest, the ferns became less important again whereas the importance of birch was even more pronounced. The changes in the Picea swamp forest were continually minor until the local on-site establishment of Picea in the swamp forest site c. 2300 cal. yr BP (Figures 5 and 6). For the two swamp forest sites these changes are well captured by the PC trajectories that indicate the diverging development for the swamp forests, i.e. Picea is present on-site in the Picea swamp forests but not in the deciduous one.

This kind of late deciduous succession stage described here was probably common in the Scandinavian boreal forests on mesic, moist and wet soils before Picea started to appear abundantly across the region. However, between 4700 and 2300 cal. yr BP charred particles >1.0 mm are recorded at four levels while three peaks of charred particles are identified in CHARSTER, and the ignition residues increase, suggesting that the influence of fire increased once again. These fires also caused fluctuations in the fen tree cover and changes in the field layer vegetation including the appearance of Epilobium that is recognized as a fire-promoted indicator species (Tinner et al., 2006). Repeated disturbance and increased erosion favoured the regeneration of Alnus, and the cover of Betula decreased. Thus, although the fen vegetation was characterized by deciduous trees, herbs and ferns suggesting fertile and wet soil conditions, fires that affected the surrounding dry terraces extended into the deciduous fen approximately every 600 to 800 years during this period (cf. Carcaillet et al., 2007).

Mixed Picea–deciduous swamp forest; 2300 to 100 cal. yr BP (PAZ II a–c)

Following the last local on-site fire, c. 2300 cal. yr BP, Picea trees established in the deciduous fen and on the surrounding terraces around 2200 cal. yr BP (Figures 3 and 4). Surprisingly, the establishment of this Picea forest occurred c. 1300 years later than around the deciduous swamp forest studied by Segerström et al. (2008), situated only 600 m north of the present study site (Table 2). However, the two ¹⁴C-datings, for depths of 80–81 and 81–82 cm, are statistically the same, suggesting that they are correct, and that there was a real difference in timing of the local Picea forest establishment between the two sites. This highlights an important spatial variation at the local scale with respect to the spread of Picea forest. Such a spatial variation in Picea establishment has also been recorded e.g. in Norway (Tryterud, 2003), and in the Scandianavian mountains during the early Holocene (Segerström and von Stedingk, 2003). After the establishment of the Picea forest on the study site, there was a marked decrease in the inflow of both charred particles and mineral matter into the depression, suggesting a reduction in the impact of fire and, hence, reduced erosion (Figures 3 and 4). This observation is in line with the general pattern for Picea-dominated forest in Sweden and Norway (Ohlson et al., 2011), but in contrast to the general increase of fire found for the last 2000 years in Canada (Carcaillet et al., 2001) and in northern Sweden (Carcaillet et al., 2007; Granström and Niklasson, 2008). The field layer changed from fen vegetation to swamp forest vegetation as the proportion of herbaceous plants decreased, probably because of a denser tree cover that reduced the amount of light reaching the forest floor. The increased depth of peat changed the hydrological conditions to ombrogenous in the central part of the swamp. Ferns probably decreased in the central depression but they still characterized the field layer on the minerogenous edges of the swamp as well as on the surrounding slopes. On the surrounding drier plateaux Ericaceous and Lycopodium species increased in cover, suggesting that Picea had established on the dry terraces that had formerly been characterized by Pinus. A remarkable local vegetation change took place about c. 2300 and 1000 cal. BP as also indicated by the PCA (both the 1st and the 2nd principal component, Figure 5).

Starting about 1000 years ago the two sites again display some similarities as suggested by combined PCA (Figure 6) most likely due to increasing human impact. Between c. 1000 and c. 300 cal. yr BP (estimated) the proportion of Pinus, Salix species, herbs (e.g. Melampyrum), grasses and sedges increased in and around the Picea swamp forest. These vegetation changes could partly reflect natural processes. In the depression, peat continued to accumulate but the conditions were still suitable for Picea, Salix, grasses and sedges. The abundance of Alnus and ferns, however, decreased along the edges of the swamp and on the surrounding slopes. The occurrence of charred particles < 1.0 mm indicates extra-local fires on the surrounding terraces, resulting in a more open canopy mainly characterised by Pinus. Thereafter, herbs and grasses were able to increase in the field-layer vegetation. However, the vegetation changes that occurred on the surrounding terraces could also, at least in part, be the result of some anthropogenic impact mainly affecting the tree layer, perhaps caused by the introduction of deliberate burning to improve grazing conditions, followed by grazing and trampling by domestic cattle (cf. Behre, 1981; Bjune et al., 2009; Hicks, 1988; Karlsson et al., 2010; Segerström and Emanuelsson, 2002). Such activities on the terraces could also have resulted in an increase in Pinus, Salix species, herbs and grasses, but the swamp forest vegetation was not particularly affected by these low-impact activities. In both the natural disturbance and the human impact scenarios, grasses probably stabilized the mineral soil on the plateaux and slopes more than the earlier herb and fern vegetation were able to, thus reducing soil erosion. The conditions, however, became slowly less fertile in the depression, as peat continued to accumulate (cf. Kuhry, 1994). About 350 cal. yr BP (estimated) a single Secale pollen grain was present, probably the result of temporary local cultivation close to the swamp forest, since Secale pollen is transported only short distances by wind in forested areas (Segerström, 1991; Segerström et al., 1994). These findings are in accordance with other studies indicating settlement establishment in the Torne river area already during the Middle Ages; i.e. ad 1050 to 1525 (Hjelmroos, 1978; Wallerström, 1995).

Around c. 300 cal. yr BP (estimated) Betula, Pinus and Picea decreased, and Alnus increased, probably as a result of some form of disturbance. This cannot however, be directly connected to a local fire, since there were neither charred particles present nor any indication of increased erosion in the peat (Figure 3). The occurrence of Epilobium could suggest fire but no charred particles were recorded, and that may seem puzzling. Epilobium is, however, common also on any disturbed grounds other than after fire; it occurs on clear cut areas, small openings in the canopy, open woods, etc. These vegetation changes were probably the result of cutting selected trees on the terraces, on the slopes and probably also in the swamp forest. This would have resulted in an increase in the groundwater level, creating conditions in the depression that were favourable for Alnus and Rubus chamaemorus. The short period of cutting created disturbed areas with mineral soil where Epilobium could establish, and subsequently grasses, Rumex, Melampyrum and other herbs increased, suggesting that the forest on the surrounding terraces remained semi-open, probably because of grazing by domestic cattle (Figure 4). The oldest living Picea trees in the swamp forest today are c. 270 years old (data not shown) and established after the logging event. Some decades later a number of changes occurred concomitantly: increased erosion, the occurrence of charred particles >0.25 mm and a temporary increase in the proportion of fern spores. These changes were probably the result of fire events that affected the terraces and the slopes that surround the swamp forest. However, the fires did not affect the swamp forest, but they did expose mineral soil on the slopes, caused increased erosion, and made the conditions temporarily suitable for ferns to establish. The fires were probably anthropogenic in origin, since Cannabis, Chenopodiaceae and Hordeum occurred subsequently, thus suggesting some small-scale local cultivation, perhaps in association with a summer farm in the area (cf. Karlsson et al., 2010; Segerström and Emanuelsson, 2002). There are, however, no visible remnants of cultivation or buildings in the vicinity of the swamp forest today.

Picea swamp forest, 100 cal. yr BP (estimated) to present (PAZ III)

During the most recent 150 years (i.e. since c. 100 cal. yr BP), the swamp forest and its surroundings have not been affected by any major disturbances. Picea gained total dominance in the tree layer and the abundance of Betula and Alnus decreased (Figure 4). The development is similar to that described for other Picea swamp forests in northern Sweden (Hörnberg et al., 1998), and the change in tree species composition has also been observed in Picea forests in Norway (Hafsten, 1992). This development follows the general pattern of succession, but contrasts to the development of the deciduous swamp forest nearby where the dominance of Betula has been maintained (Segerström et al., 2008), and also causing the diverging PC trajectories (Figure 6).

The increased cover of Pinus may reflect modern forest management activities, which started in the region during the late nineteenth century. These activities initially included industrial selective felling, targeting large Pinus trees, and suppressing natural fires (Andersson and Östlund, 2004; Östlund, 1993). Subsequently, throughout the twentieth century, management programmes became focused on creating high-yielding conifer forests through thinning, clear-cutting, tree planting that favoured conifers (mainly Pinus), and active suppression of deciduous trees (cf. Hellberg, 2004; Östlund et al., 1997). Today in the county of Norrbotten, 57% of the growing stock on forest land is Pinus, compared with a mean value of 39% over the whole of Sweden (Anonymous, 2004). The influence of grazing by domestic animals has become less intense in the area although grazing and trampling by semi-domestic reindeer still occur. This impact is reflected by the presence of Salix, grasses, Rumex and some other herbs, although the abundance of Melampyrum has decreased (Figure 4).

Processes that have driven swamp-forest succession

The succession identified in the Picea swamp forest studied has been mainly driven by fire, peat accumulation, ecological processes and anthropogenic disturbances. Possible mechanisms that have produced the Picea swamp forest that we see today can be summarized as follows. Before Picea established at the stand-scale in the region, Betula characterised the late successional stages, and the Kuusivaara area (including the preceding deciduous fen) was affected by fire. The establishment of Picea forests may even have been encouraged by fire in combination with a general shift in climate. As soon as Picea had established in the deciduous fen and its surroundings, on-site fires ceased to affect the vegetation. The local conditions changed as the tree cover became denser, and the amount of light that reached the forest floor in the depression decreased. In addition, local temperature and evapotranspiration may have decreased while humidity and peat accumulation increased. There is also a change into wetter and colder climate (Bond et al., 2001; Grosjean et al., 2007; Haas et al., 1998). The field layer changed from minerogenous fen vegetation rich in herbs and ferns to species-poor ombrogenous swamp forest vegetation as peat accumulated. All these changes resulted in wet and cool local conditions that did not promote fire. Picea was still able to regenerate mainly on dead wood and fallen logs, a pattern that is observed in many old-growth Picea swamp forests (Hörnberg et al., 1995, 1998) as well as in old-growth Picea forests in, e.g., northern Sweden (Hofgaard, 1993) and Poland (Zielonka, 2006). Increased concentrations of phenolic compounds both in the peat in the swamp-forest and in the soil on the surrounding terraces were produced from Picea and ericaceous dwarf shrub litter; they probably reduced the likelihood that deciduous species would establish and co-exist with Picea (cf. Wardle et al., 1997). Consequently, although fire may have facilitated the establishment of Picea, the presence of Picea together with a colder and wetter climate has created local conditions that prevent fire (Tryterud, 2003). In addition, litter rich in phenolic compounds slows down decomposition, increases accumulation of organic matter, and reduces nutrient availability (Nilsson and Wardle 2005; Wardle et al., 2003). Taken together, all these factors reduce the chance that other species will establish. Thus, Picea has created conditions that have perpetuated its dominance during the last two millennia (cf. Ohlson et al., 2011).

The current situation, with Picea as one of the most important tree species in Fennoscandia, may change. Picea can be characterized as a continental species with high summer temperature requirements and a low temperature threshold for bud initiation and shoot growth (Skre, 1972, 1979). Picea seedlings need continuous snow cover to ensure a low and stable near-ground temperature to survive the winter, and are vulnerable to early summer drought (Tallantire, 1972). Picea is also known to be more susceptible to fire than Pinus (e.g. Engelmark and Hytteborn, 1999; Wallenius et al., 2005), and is depressed by the occurrence of fires (Bradshaw and Hannon, 1992). Because of global change, the climate in Fennoscandia is expected to become warmer, with increased precipitation; a more maritime climate with milder winters is predicted, including more winter melting and thawing events, and less snow cover (Räisänen et al., 2004; Rowell, 2005). Warmer spring and summer temperatures, and an earlier spring snowmelt have resulted in an increase in wildfires in the western USA during the period 1970 to 2003 (Westerling et al., 2006), and studies in Canada also suggest increasing future areas burned (Flannigan et al., 2005). With such climate changes and repeated fire in the boreal zone of Fennoscandia, the combined effects will probably overrule the internal fire preventing conditions created in the swamp forests by the presence of Picea. The ability of Picea to establish and compete successfully would then be adversely affected, hence making future conditions less favourable for Picea. However, the effects of climate changes on fire pattern are still uncertain and do not necessarily include a general increase in wildfire frequency in Fennoscandia (e.g. Flannigan et al., 1998).

Conclusion

Until Picea abies became present at the stand-scale in Sweden, between 3000 and 2000 yr BP, deciduous trees were common in late successional stages in swamp forests. Fire facilitated the initial Picea abies establishment and, once established, Picea created local conditions in the swamp forests that, in combination with a colder and wetter climate, prevented fire and the establishment of other tree species. It seems as this shift in tree species had a profound influence on local biotic and abiotic conditions, thus changing the long-term fire regime. How these internal fire-preventing conditions will withstand a predicted climate change in the boreal zone of Fennoscandia, associated with an increased occurrence of wildfires, is uncertain.

Footnotes

Acknowledgements

We would like to thank Morgan Karlsson, Matts Lindbladh and Erik Hellberg for assistance in the field, Rikard Andersson for producing the maps, and two anonymous reviewers for providing constructive comments to improve the manuscript.

The project was funded by FORMAS grant no 24.0631/00.

References

Andersson

Östlund

(2004) Spatial patterns, density changes and implications on biodiversity for old trees in the boreal landscape of northern Sweden. Biological Conservation 118: 443–53.

Angelstam

(1998) Maintaining and restoring biodiversity in European boreal forests by developing natural disturbance regimes. Journal of Vegetation Science 9: 593–602.

Ångström

(1974) Sveriges klimat. Stockholm: Generalstabens litografiska Anstalts Förlag.

Anonymous (1997) Bildandet av naturreservatet Kuusivaara domänreservat i Pajala och Övertorneå kommun. Länsstyrelsen i Norrbottens län, beslut 1997-01-10, 2312-11546-96, 2502075.

Anonymous (2004) Skogsstatistisk årsbok. Jönköping: Skogsstyrelsen.

Behre

K-E

(1981) The interpretation of anthropogenic indicators in pollen diagrams. Pollen et Spores 23: 225–245.

Bergeron

Leduc

Harvey

Gauthier

(2002) Natural fire regime: A guide for sustainable management of the Canadian boreal forest. Silva Fennica 36: 81–95.

Birks

HJB

(1998) Numerical tools in palaeolimnology – Progress, potentialities, and problems. Journal of Paleolimnology 20: 307–332.

Bjune

Ohlson

Birks

HJB

Bradshaw

RHW

(2009) The development and local stand-scale dynamics of a Picea abies forest in southeastern Norway. The Holocene 19: 1073–1082.

10.

Bonan

Shugart

(1989) Environmental factors and ecological processes in boreal forests. Annual Review of Ecological System 20: 1–28.

11.

Bond

Kromer

Beer

Muscheler

Evans

Showers

. (2001) Persistent solar influence on north Atlantic climate during the Holocene. Science 294: 2130–2136.

12.

Bradshaw

RHW

Hannon

(1992) Climatic change, human influence and disturbance regime in the control of vegetation dynamics within Fiby Forest, Sweden. Journal of Ecology 80: 625–632.

13.

Carcaillet

Bergeron

Richard

PJH

Frechette

Gauthier

Praires

(2001) Change of fire frequency in the eastern Canadian boreal forest during the Holocene: Does vegetation composition or climate trigger the fire regime? Journal of Ecology 89: 930–946.

14.

Carcaillet

Bergman

Delorme

Hörnberg

Zackrisson

(2007) Long-term fire frequency not linked to prehistoric occupations in northern Swedish boreal forest. Ecology 88: 465–477.

15.

Clark

Royall

(1995) Particle-size evidence for source area of charcoal accumulation in Late Holocene sediments of eastern North American lakes. Quaternary Research 43: 80–89.

16.

Engelmark

Hytteborn

(1999) Coniferous forests. Acta Phytogeographica Suecica 84: 55–74.

17.

Esseen

P-A

Ehnström

Ericson

Sjöberg

(1997) Boreal forests. Ecological Bulletins 46: 16–47.

18.

Flannigan

Bergeron

Engelmark

Wotton

(1998) Future wildfire in circumboreal forests in relation to global warming. Journal of Vegetation Science 9: 469–476.

19.

Flannigan

Logan

Amiro

Skinner

Stock

(2005) Future area burned in Canada. Climatic Change 72: 1–16.

20.

From

(1965) Beskrivning till jordartskarta över Norrbottens län nedanför Lappmarksgränsen. Stockholm: Sveriges Geologiska Undersökningar Serie Ca. 39.

21.

Froyd

(2006) Holocene fire in the Scottish Highlands: Evidence from macroscopic charcoal records. The Holocene 16: 235–249.

22.

Gavin

(2006) CHARSTER version 0.8.3. Paleoecology and Biogeography Laboratory, Department of Geography, University of Oregon, US, http://geography.uoregon.edu

23.

Gavin

Lertzman

Corbett

(2006) Weak climatic control of stand-scale fires history during the Late Holocene. Ecology 87: 1722–1732.

24.

Giesecke

(2004) The Holocene Spread of Spruce in Scandinavia. Comprehensive summaries of Uppsala dissertations from the Faculty of Science and Technology 1027, Uppsala.

25.

Giesecke

Bennett

(2004) The Holocene spread of Picea abies (L.) Karst. in Fennoscandia and adjacent areas. Journal of Biogeography 31: 1523–1548.

26.

Gobet

Tinner

Hochuli

van Leeuwen

JFN

Amman

(2003) Middle to late vegetation history of the Upper Engadine (Swiss Alps): The role of man and fire. Vegetation History and Archaeobotany 12: 143–163.

27.

Granström

Niklasson

(2008) Potentials and limitations for human control over historic fire regimes in the boreal forest. Philosophical Translations of the Royal Society B 363: 2353–2358.

28.

Grimm

(1991) TILIA v. 2.0.b.4. Illinois State Museum, Research and Collections Center.

29.

Grimm

(2004) TILIA GRAPH v. 2.0.2. Illinois State Museum, Research and Collections Center.

30.

Grosjean

Suter

Trachsel

Wanner

(2007) Ice-borne prehistoric finds in the Swiss Alps reflect Holocene glacier fluctuations. Journal of Quaternary Science 22: 203–207.

31.

Haas

Richoz

Tinner

Wick

(1998) Synchronous Holocene climatic oscillations recorded on the Swiss Plateau and at timberline in the Alps. The Holocene 8: 301–309.

32.

Hafsten

(1992) The immigration and spread of Norway spruce (Picea abies L. Karst.) in Norway. Norsk Geografisk Tidskrift 46: 121–158.

33.

Heinselman

(1981) Fire and succession in the conifer forest of northern North America. In: West

Shugart

Botkin

(eds) Forest Successions. New York: Springer Verlag, 374–405.

34.

Heiri

Bugmann

Tinner

Heiri

Lischke

(2006) A model-based reconstruction of Holocene treeline dynamics in the Central Swiss Alps. Journal of Ecology 94: 206–216.

35.

Heiri

Lotter

Lemcke

(2001) Loss on ignition as a method for estimating organic and carbonate content in sediments: Reproducibility and comparability of results. Journal of Paleolimnology 25: 101–110.

36.

Hellberg

(2004) Historical variability of deciduous trees and deciduous forests in northern Sweden. Acta univesitatis agriculturae sueciae, Silvestria 308. PhD-thesis, Swedish University of Agricultural Sciences.

37.

Hicks

(1988) The representation of different farming practices in pollen diagrams from northern Finland. In: Birks

Birks

HJB

Kaland

Moe

(eds) The Cultural Landscape – Past, Present and Future. Cambridge: Cambridge University Press, 189–207.

38.

Hjelmroos

(1978) Den äldsta bosättningen i Tornedalen. En paleoekologisk undersökning. University of Lund, Department of Quaternary Geology. Report 16.

39.

Hofgaard

(1993) Structure and regeneration patterns of a virgin Picea abies forest in northern Sweden. Journal of Vegetation Science 4: 601–608.

40.

Hörnberg

Ohlson

Zackrisson

(1995) Stand dynamics, regeneration patterns and long-term continuity in boreal old-growth Picea abies swamp forests. Journal of Vegetation Science 6: 291–298.

41.

Hörnberg

Zackrisson

Segerström

Svensson

Ohlson

Bradshaw

RHW

(1998) Boresal swamp forests – Biodiversity ‘hotspots’ in an impoverished forest landscape. Bioscience 48: 795–802.

42.

Huntley

Birks

HJB

(1983) An Atlas of Past and Present Pollen Maps for Europe: 0–13000 Years Ago. Cambridge: Cambridge University Press.

43.

Hytteborn

Packham

Verwijst

(1987) Tree population dynamics, stand structure and species composition in the montane virgin forest of Vallibäcken, northern Sweden. Vegetatio 72: 3–19.

44.

Jowsey

(1966) An improved peat sampler. New Phytologist 65: 245–248.

45.

Karlsson

Emanuelsson

Segerström

(2010) The history of a farm-shieling system in the central Swedish forest region. Vegetation History and Archaeobotany 19: 103–119.

46.

Kuhry

(1994) The role of fire in the development of Sphagnum-dominated peatlands in western Canada. Journal of Ecology 82: 899–910.

47.

Kullman

(1995) New and firm evidence for mid-Holocene appearance of Picea abies in the Scandes Mountains, Sweden. Journal of Ecology 82: 923–931.

48.

Kullman

(2001) The geoecological history of Picea abies in northern Sweden and adjacent parts of Norway. A contrarian hypothesis of postglacial tree immigration patterns. Geoöko 21: 141–172.

49.

Kuuluvainen

Syrjänen

Kalliola

(1998) Structure of pristine Picea abies forests in northeastern Europe. Journal of Vegetation Science 9: 563–574.

50.

Lynch

Clark

Stocks

(2004) Charcoal production, dispersal and deposition from the Port Providence experimental fire: Interpreting fire regimes from charcoal records in boreal forests. Canadian Journal of Forest Research 34: 1642–1656.

51.

Molinari

Bradshaw

RHW

Risbøl

Lie

Ohlson

(2005) Long-term vegetational history of a Picea abies stand in south-eastern Norway: Implications for the conservation of biological values. Biological Conservation 126: 155–165.

52.

Moore

Webb

Collinson

(1991) Pollen Analysis. London: Blackwell Scientific Publications.

53.

Mossberg

Stenberg

(2003) Den nya nordiska floran. Tangen, Norge: Wahlström & Widstrand, PDC.

54.

Niklasson

Granström

(2000) Numbers and sizes of fires: Long-term spatially explicit fire history in a Swedish boreal landscape. Ecology 81: 1484–1499.

55.

Nilsson

M-C

Wardle

(2005) Understory vegetation as a forest ecosystem driver: Evidence from the northern Swedish boreal forest. Frontiers in Ecology and the Environment 3: 421–428.

56.

Norokorpi

(1979) Old Norway spruce stands, amount of decay and decay-causing microbes in northern Finland. Communicationes Instituti Forestales Finniae 97: 1–77.

57.

Ohlson

Tryterud

(1999) Long-term spruce forest continuity – A challenge for a sustainable Scandinavian forestry. Forest Ecology and Management 124: 27–34.

58.

Ohlson

Tryterud

(2000) Interpretation of the charcoal record in forest soils: Forest fires and their production and deposition of macroscopic charcoal. The Holocene 10: 519–525.

59.

Ohlson

Brown

Birks

HJB

Grytnes

J-A

Hörnberg

Niklasson

. (2011) Invasion of Norway spruce diversifies the fire regime in boreal European forests. Journal of Ecology 99: 395–403.

60.

Ohlson

Dahlberg

Økland

Brown

Halvorsen

(2009) The charcoal and carbon pool in boreal forests. Nature Geoscience 2: 692–695.

61.

Ohlson

Korbøl

Økland

(2006) The macroscopic charcoal record in forested boreal peatlands in southeast Norway. The Holocene 16: 731–741.

62.

Ohlson

Söderström

Hörnberg

Zackrisson

Hermansson

(1997) Habitat qualities versus long-term continuity of biodiversity in boreal old-growth swamp forests. Biological Conservation 81: 221–231.

63.

Östlund

(1993) Exploitation and structural changes in the north Swedish boreal forest 1800-1992. PhD-thesis, Department of Forest Vegetation Ecology, Swedish University of Agricultural Sciences, Umeå, Sweden.

64.

Östlund

Zackrisson

Axelsson

A-L

(1997) The history and transformation of a Scandinavian boreal landscape since the 19th century. Canadian Journal of Forest Research 27: 1198–1206.

65.

Payette

(1992) Fire as a controlling process in the North American boreal forest. In: Shugart

Leemans

Bonan

(eds) A System Analysis of the Global Boreal Forest. Cambridge: Cambridge University Press, 144–169.

66.

Pisaric

MFJ

(2002) Long-distance transport of terrestrial plant material by convection resulting from fires. Journal of Paleolimnology 28: 349–354.

67.

Pitkänen

Huutunen

Jungner

Tolonen

(2002) A 10 000 year local forest fire history in a dry heath forest site in eastern Finland, reconstructed from charcoal layer records of a small mire. Canadian Journal of Forest Research 32: 1875–1880.

68.

Pitkänen

Huutunen

Tolonen

Jungner

(2003) Long-term fire frequency in the spruce-dominated forests of the Ulvinsalo strict nature reserve, Finland. Forest Ecology and Management 176: 305–319.

69.

Räisänen

Hansson

Ullerstig

Döscher

Graham

Jones

. (2004) European climate in the late twenty-first century: Regional simulations with two driving global models and two forcing scenarios. Climate Dynamics 22: 13–31.

70.

Rowell

(2005) A scenario of European climate change for the late twenty-first century: Seasonal means and inter-annual variability. Climate Dynamics 25: 837–849.

71.

Schimmel

(1993) On fire. Fire behaviour, fuel succession and vegetation response to fire in the Swedish boreal forest. PhD-thesis, Swedish University of Agricultural Sciences, Umeå.

72.

Segerström

(1991) Soil pollen analysis – An application for tracing ancient arable patches. Journal of Archaeological Science 18: 165–175.

73.

Segerström

(1997) Long-term dynamics of vegetation and disturbance of a southern boreal spruce swamp forest. Journal of Vegetation Science 8: 295–306.

74.

Segerström

Emanuelsson

(2002) Extensive forest grazing and hay-making on mires – Vegetation changes in south-central Sweden due to land use since the Medieval Times. Vegetation History and Archaeobotany 11: 181–190.

75.

Segerström

von Stedingk

(2003) Early-Holocene spruce, Picea abies (L.) Karst., in west-central Sweden as revealed by pollen analysis. The Holocene 13: 897–906.

76.

Segerström

Bradshaw

RHW

Hörnberg

Bohlin

(1994) Disturbance history of a swamp forest refuge in northern Sweden. Biological Conservation 68: 189–196.

77.

Segerström

Hörnberg

Bradshaw

RHW

(1996) The 9000-year history of vegetation development and disturbance pattern of a swamp forest in Dalarna, northern Sweden. The Holocene 6: 37–48.

78.

Segerström

von Stedingk

Hörnberg

(2008) Long-term sustainability of a northern boreal deciduous swamp forest in northern Sweden: Succession in the absence of fire. The Holocene 18: 1113–1122.

79.

Seppä

Alenius

Bradshaw

RHW

Giesecke

Heikkilä

Muukkonen

(2009) Invasion of Norway Spruce (Picea abies) and the rise of the boreal ecosystem in Fennoscandia. Journal of Ecology 97: 629–640.

80.

Sirén

(1955) The development of spruce forest on raw humus sites in northern Finland and its ecology. Acta Forestalia Fennicia 62: 1–363.

81.

Skre

(1972) High temperature demands for growth and development in Norway spruce (Picea abies [L.] Karst.) in Scandinavia. Meldinger fra Norsk Landbruks Høgskole 51(7): 1–29.

82.

Skre

(1979) The regional distribution of vascular plants in Scandinavia with requirements for high summer temperatures. Norwegian Journal of Botany 26: 295–318.

83.

Stuiver

Reimer

(1993) Extended ¹⁴C database and revised CALIB 3.0 ¹⁴C age calibration program (rev. 4.3). Radiocarbon 35: 215–230.

84.

Tallantire

(1972) The regional spread of spruce (Picea abies (L.) Karst.) within Fennoscandia: A reassessment. Norwegian Journal of Botany 19: 1–16.

85.

Tallantire

(1977) A further contribution to the problem of the spread of spruce (Picea abies (L.)) in Fennoscandia. Journal of Biogeography 4: 219–227.

86.

ter Braak

CJF

Smilauer

(2002) CANOCO Reference Manual and CanoDraw for Windows User’s Guide: Software for Canonical Community Ordination (Version 4.5). Ithaca NY: Microcomputer Power.

87.

Tinner

(2003) Size parameters, size class distribution and area-number relationship of microscopic charcoal: Relevance for fire reconstruction. The Holocene 13: 499–505.

88.

Tinner

Conedera

Ammann

Gäggeler

Gedye

Jones

. (1998) Pollen and charcoal in lake sediments compared with historically documented forest fires in southern Switzerland since ad 1920. The Holocene 8: 31–42.

89.

Tinner

Hofstetter

Zeugin

Condera

Wohlgemuth

Zimmermann

. (2006) Long-distance transport of macroscopic charcoal by an intensive crown fire in the Swiss Alps – Implications for fire history reconstructions. The Holocene 16: 287–292.

90.

Tryterud

(2003) Forest fire history in Norway: From fire-disturbed pine forest to fire-free spruce forest. Ecography 26: 161–170.

91.

Wallenius

Pitkänen

Kuuluvainen

Pennanen

Karttunen

(2005) Fire history and forest age distribution of an unmanaged Picea abies dominated landscape. Canadian Journal of Forest Research 35: 1540–1552.

92.

Wallerström

(1995) Norrbotten, Sverige och medeltiden. Problem kring makt och bosättning i en europeisk periferi. Del I. Lund studies in Medieval Archaeology 15:1. Lund. Summary: Norrbotten, Sweden and the Middle Ages. Problems concerning power and settlement on a European periphery. Stockholm: Almqvist & Wiksell International.

93.

Wanner

Holzhauser

Pfister

Zumbühl

(2000) Interannual to century scale climate variability in the European Alps. Erdkunde 54: 62–69.

94.

Wardle

Hörnberg

Zackrisson

Kalela-Brundin

Coomes

(2003) Long term effects of wildfire on ecosystem properties across an island area gradient. Science 300: 972–975.

95.

Wardle

Zackrisson

Hörnberg

Gallet

(1997) The influence of island area on ecosystem properties. Science 277: 1296–1299.

96.

Westerling

Hidalgo

Cayan

Swetnam

(2006) Warming and earlier spring increase western U.S. forest wildfire activity. Science 313: 940–943.

97.

Whitlock

Millspaugh

(1996) Testing the assumption of fire-history studies: An examination of modern charcoal accumulation. The Holocene 6: 7–15.

98.

Zackrisson

(1977) Influence of forest fire on the north Swedish boreal forest. Oikos 29: 22–33.

99.

Zielonka

(2006) When does dead wood turn into a substrate for spruce replacement? Journal of Vegetation Science 17: 739–746.