The occurrence of thermophilous trees in the Scandes Mountains during the early Holocene: evidence for a diverse tree flora from macroscopic remains



1 Macroscopic remains of the fairly thermophilous tree species Alnus glutinosa, Tilia cordata and Betula pendula were recovered in subalpine and adjacent boreal environments far above and beyond their present-day distributional limits. This establishes that the early Holocene tree flora of the Scandes Mountains in northern Sweden was indeed richer than it is today.

2 Dates ranged between c. 8600 and 7000 radiocarbon years bp. These are much earlier than previous estimates by conventional pollen stratigraphical analyses of the arrival of these species at their maximum geographical limits. This highlights problems in using only pollen data for vegetation reconstruction, and suggests re-evaluation of earlier records.

3 The results, together with similar macrofossils for Picea abies and Larix sibirica in northern Sweden, suggest that many tree species spread rapidly and became established at their most extended range limits during the early Holocene. Abundances have subsequently varied in accordance with the ecology of individual species as well as with climatic change.

4 Palaeoclimatic inferences may suggest a strongly continental climate, i.e. warmer and drier summers and possibly fairly cold winters between 8600 and 7000 bp relative to the present. Some change towards a more oceanic climate regime with less pronounced seasonal contrasts may have occurred towards the end of the period.


Little detail is available concerning the history of forest vegetation of the Swedish Scandes and adjacent regions in the Holocene. Apart from tree-limit studies that draw on megafossil evidence (Kullman 1995a, 1996; Karlén 1976), the only recent evidence comes from a few fairly well dated pollen and macrofossil records (Sonesson 1974; Königsson 1984; Hafsten 1987; Bradshaw & Zackrisson 1990; Berglund et al. 1996; Engelmark 1996). In particular, the question of the presence of tree species that are more thermophilous (temperate) than those that occur today (and their date of arrival) has never developed beyond pure speculation or uncertain inference. Early workers found trace amounts of pollen from, for example, Tilia cordata Mill., Corylus avellana L., Ulmus glabra Huds. and Quercus sp. in peat deposits and raw humus layers (with or without radiocarbon datings) within present-day subalpine–montane regions and inferred their local presence during the early or mid-Holocene (Andersson 1902; Smith 1920; Erdtman 1943; Granlund 1943; Lundqvist 1969; von Post 1906, 1930; Nordhagen 1933, 1943). Others, however, generally dismiss such findings as the result of long-distance transport of extra-regional pollen (Huntley & Birks 1983; Hafsten 1987; Tallantire 1992; Engelmark 1996). Clarification of these matters may add important nuances to vegetational and environmental reconstruction and their methodologies.

A project based on extensive surveys and analyses of plant macro- and megafossils preserved in peat in the alpine and subalpine regions of the Swedish Scandes has yielded entirely new data and a new perspective on the palaeohistory of Picea abies (L.) Karst. (Kullman 1995b, 1996, 1998) and Larix sibirica Ledeb. (Kullman 1998). This led to an increase in attention towards other tree species, thereby establishing the presence of three thermophilous tree species, Alnus glutinosa (L.) Gaertn., Betula pendula Roth and Tilia cordata, during the early Holocene well beyond their modern distributional limits in northern Sweden. The present paper describes these discoveries, and tentatively discusses their palaeoecological and palaeoclimatic context. It is particularly interesting that this raises questions relating to the nature and pliability of communities.

Study area

The area under study is situated in the southern Swedish Scandes (Province of Jämtland) (Fig. 1). Site 1 is on the east-facing flank of Mt Getryggen, 830 m a.s.l. (63°10′N, 12°21′E) and Site 2 is at the Klocka mire, which protrudes into Lake Ånnsjön, 526 m a.s.l. (63°18′N, 12°29′E).

Figure 1.

Location of the study area and the two sample sites (1 and 2). Stippled area, present-day subalpine birch forest; solid line, present-day pine limit. Modern distributional limits of Alnus glutinosa (A), Betula pendula (B), and Tilia cordata (T) are compiled from Hultén (1971).

The climate is moderately oceanic and humid. The mean temperatures for January, July and annually are –7.6, 10.7 and 1.1 °C, respectively. The annual precipitation amounts to 857 mm, of which c. 45% falls as snow. All data are from 1961–1990 records for the official meteorological station at Storlien/ Visjövalen, 642 m a.s.l. and c. 20 km to the north-west of the sites.

Amphibolite and gneisses of Caledonian age form the bedrock, which is mostly well-covered by glacial till, lacustrine or glacifluvial deposits. The regional deglaciation occurred c. 9100 bp (Lundqvist 1986).

At Site 1 a mosaic of subalpine birch forest (Betula pubescens Ehrh. ssp. tortuosa (Ledeb.) Nyman), treeless dwarf shrub heaths and shallow mires prevails at elevations below the alpine tundra. Subordinate members of the tree layer in the birch forest are Sorbus aucuparia L., Alnus incana (L.) Moench and Populus tremula L. At the lower elevations, Site 2, around the limit of the coniferous forest (mainly Picea abies), extensive 2–2.5-m deep mires (bog–fen complexes) are conspicuous elements of the landscape. The local tree limits (i.e. trees > 2 m high) are at c. 900, 800, 700 m a.s.l., for birch, spruce and pine (Pinus sylvestris L.), respectively.

The modern physical and vegetational setting and the Quaternary landscape history are described in more detail by Lundqvist (1969) and Kullman (1995a, 1996).


Macroscopic tree remains were sampled extensively at the exposed peat faces of erosion hollows and at natural erosion fronts of peat bogs. Radiocarbon dating was carried out by Beta Analytic Inc., Miami, Florida. Dates are presented as uncalibrated 14C-years bp (= ad 1950), with a half-life of 5568 ± 30 years. They have been corrected for deviations in ∂13C. Species identification draws on unambiguous fruit and bark characters, by comparisons with reference collections and live material.


Site 1

Female cones (with seeds) of Alnus glutinosa, characteristically on long stalks (Fig. 2), were recovered from c. 0.3 m below the surface of an eroding peat deposit with a maximum depth of 0.5–0.7 m (Fig. 3). Radiocarbon dating (AMS) yielded a date of 7310 ± 70 bp (Beta-99210). This site is c. 550 m higher than the nearest living Alnus glutinosa, which is found c. 70 km to the west-north-west (Gravaå 1970) (Fig. 1). Strong wind and water erosion during the late Holocene has severely affected this and most other peat deposits at similar elevations, and the resulting resorting and mixing of material (Smith 1920) prevents stratigraphical analyses. Although this lack is unfortunate, the dating is clear.

Figure 2.

Subfossil cones from Alnus glutinosa.

Figure 3.

The site where subfossil Alnus glutinosa and Betula pendula were sampled was an erosion hollow strewn with resorted and mixed fragments of subfossil wood.

A few fairly large pieces of bark from Betula pendula were recovered and dated. They were partially exposed at the surface due to redeposition, and although fragments of wood were found beneath, wood anatomy could not distinguish between Betula pendula and B. pubescens (Schweingruber 1990). The bark specimens, however, showed undisputable characteristics of B. pendula in the form of a blackish colour, and deep (5–10 mm) fissures forming rhomboid bosses (Fig. 4) (cf. Atkinson 1992), confirmed by a new simple and accurate chemotaxonomic method (Lundgren et al. 1995). Two pieces, lying c. 15 m apart, were dated to 8570 ± 70 bp (Beta-99218) and 8070 ± 100 bp (Beta-99220). The closest present-day site for B. pendula is c. 35 km to the north-west and c. 600 m lower (Fig. 1).

Figure 4.

Bark from subfossil Betula pendula, with characteristic deep fissures.

Site 2

A well-preserved inflorescence (joint fruit stalk and bract) of Tilia cordata (Fig. 5) was dissected from the base of a wave-eroded peat scarp of a large bog, at the late summer water level of the adjacent lake (Fig. 6). The stratigraphic position was 2.2 m below the peat surface, at the transition between peat and underlying sediments. Surrounding peat was rich in fragments of Phragmites australis (Cav.) Steud., Equisetum sp., mosses and some leaves of Salix spp. and B. pubescens Ehrh., suggesting that the Tilia macrofossil may have been preserved in a fairly wet fen, dominated by birch. A regional pollen diagram, including stratigraphic details concerning the peat composition, is available for this locality (Lundqvist 1969; Fig. 124), and dates the bottom layer of this peat deposit at c. 8400 bp (Lundqvist 1969). Apparently, the site was successively inundated during the course of the Holocene, as the lake has tipped due to differential land uplift in the east and in the west (Lundqvist 1969). As a result, water erosion has steadily cut into the bog and exposed new surfaces in the peat. The peat is generally weakly humified and the bog contains permafrost, which may have improved the preservation of macrofossils.

Figure 5.

Subfossil inflorescence from Tilia cordata.

Figure 6.

The erosion scarp at the Klocka mire, where subfossil Tilia cordata was preserved at the base of the peat accumulation.

AMS radiocarbon dating of the inflorescence indicated 6980 ± 60 bp (Beta-99202). Tilia cordata does not occur spontaneously in the region today: the nearest site is situated 200 km to the west and c. 480 m below Site 2 (Fig. 1).


The fossils described here show that Alnus glutinosa, Betula pendula and Tilia cordata grew at relatively high elevations in the Scandes during the early Holocene, and prompt re-examination of previous accounts of the vegetation history. The local pollen diagram (Lundqvist 1969) shows that trace amounts (c. 1% of the arboreal pollen sum) of Tilia occur discontinuously between 8000 bp and 4000 bp. Earlier undated records also report a few per cent of Tilia pollen deep in upper subalpine mires of this region (Smith 1920) and a subfossil inflorescence of Tilia in the bottom layer of a northern Swedish mire far beyond the the current distributional limit (Erdtman 1943). Since Tilia has fairly large and heavy pollen grains, a few grains may be sufficient to indicate local presence (cf. Erdtman 1943). The macro- and microfossil data cannot be matched for Betula pendula and Alnus glutinosa since the pollen record is not accurate to species level. However, the regional pollen diagram (Lundqvist 1969) shows pollen frequencies of Alnus sp. c. 10–15% and Betula sp. c. 5–10% at the date for the related macrofossils, respectively.

Previous surmises of local presence are now confirmed, suggesting that minor and discontinuous tree pollen frequencies may indeed represent local presence of small populations (cf. von Post 1930; Bennett 1988), as demonstrated by Bush & Hall (1987) and Peteet (1991). However, not all so-called ‘exotic’ pollen grains will represent local presence and it is doubtful whether local presence can ever be confidently inferred from pollen analysis, which draws on subjective assumptions about pollen dispersal and productivity. Nevertheless, palynologists and students of vegetation history who adopt a less conservative and prejudiced attitude with respect to the past tree flora should also realize that pollen analysis alone can only generate hypotheses that need validation from macrofossils (cf. Bradshaw & Zackrisson 1990; Berglund et al. 1996). Re-evaluation of pollen records combined with macrofossil sampling at sites with long and sporadic ‘pollen tails’ of various tree taxa would certainly be rewarding.

If the present findings can be generalized to other species with sporadic pollen representation (Lundqvist 1969), Corylus avellana, Ulmus glabra, and perhaps Acer platanoides L. and Quercus robur L., may also have grown in the study region earlier during the Holocene. Such a situation is palynologically inferred for some montane regions of south-central Norway (Simonsen 1980; Kvamme 1993). The early to mid-Holocene montane–subalpine forest and the forest–alpine tundra ecotone clearly had a much more diverse arboreal flora than at present, as further stressed by the recent finds of macrofossils of Picea abies and Larix sibirica from this period within the upper subalpine and lower alpine belts (Kullman1996, 1998). This mixture of cold-adapted, xeromorphic needle-leaved evergreens and thermophilous, mesomorphic broad-leaved deciduous trees, fits the general conception of the late protocratic phase of ecological processes and patterns during an interglacial cycle (Iversen 1954, 1958).

The disturbed nature of the Betula–Alnus deposit and the regional character of the pollen diagram for Site 2 (Tilia) mean that it is virtually impossible to infer the plant community types in which these species grew. Although Betula pendula and Alnus glutinosa were recovered very close together, they may represent quite different communities within the local vegetation mosaic and might have been brought together by erosional forces. Similarly, Tilia fruits, which are often dispersed by wind, have possibly moved in the order of 0.1–1 km prior to deposition (cf. Sernander 1901).

However reliable, macrofossils provide only possible estimates for local arrival. Even so, dates for Alnus glutinosa and Tilia cordata are several millennia earlier than those from conventional and interpolated isopoll data (Huntley & Birks 1983) or local macrofossil records (Tallantire 1992). Betula pendula is often surmised as one of the first trees to have invaded deglaciated areas in the earliest Holocene (Fries 1976; Huntley & Birks 1983; Pennington 1986), but macrofossils were needed to overcome problems of differentiating pollen of the various potential Betula taxa. It is therefore obvious that all three tree species identified here arrived fairly early during the Holocene. Factors such as migration properties and the position of late-glacial refugia seem not to have been limiting for these species, which must somehow have spread very rapidly at the subcontinental scale. Similar patterns have recently been disclosed for Picea abies and Larix sibirica in the same region (Kullman 1995b, 1996, 1998). Thus, we are confident in the old theory proposed by Rudolph (1930), which implies very rapid and early immigration of most European trees, followed by population expansion (coalescence), decline (fragmentation) or extinction.

The opportunities for quantitative reconstructions of climate and climate change are limited because of the small sample and uncertainty of whether the macrofossils represent contemporary elevational limits. Moreover, comparisons of past and present distributions in climatic terms are confounded by the fact that the nearest modern localities are situated in a decidedly more oceanic climate, which generally implies lower tree limits. However, the elevational retreat since the early Holocene of more than 500 m compares well with that recorded for Pinus sylvestris (Kullman 1995a, 1997; unpublished data) and suggests that this represents a more general, climatically forced, depression of vegetational limits since the earliest part of the Holocene.

The thermophilous character, particularly of Tilia and Alnus (cf. Helland 1912; Hintikka 1963; Pigott & Huntley 1981; Pigott 1981; Prentice & Helmisaari 1991; Tallantire 1992; Atkinson 1992), does suggest that the summers during the early Holocene were much warmer than at present (cf. Berglund 1983; Kullman 1995a; Svendsen & Mangerud 1997). The appearance here of Betula pendula 8600–8000 bp is indicative of a dry continental climate with a low ground-water level, in accordance with the inference from tree-limit history (Kullman 1995a). The later and almost simultaneous presence of Alnus glutinosa and Tilia cordata (7300–7000 BP), which have somewhat more oceanic affinities, might suggest a slight shift towards a less seasonal climate regime with raised ground-water level. Although all three species could well have been present earlier than recorded here, the first appearance of Alnus incana (L.) Moench and a discrete subalpine belt of broad-leaved deciduous forest is dated by macrofossils to the same period of time (Kullman 1995a), and is inferred to be consistent with orbitally forced insolation change and, in consequence, a more oceanic climate (COHMAP Members 1988). In fact, that model of climate forcing throughout the Holocene has been supported by the elevational position of the pine and birch tree limits during the past c. 9000 14C years (Kullman 1995a; L. Kullman, unpublished data).

Winter conditions are less easily inferred from past range limits since the effects of cold temperatures on distribution are poorly understood. However, although Betula pendula and Tilia cordata are fairly tolerant of winters colder than found at present in the region (Prentice & Helmisaari 1991), the appearance of Alnus is not compatible with strong seasonal ground frost (Westman 1985). Alnus glutinosa, however, prefers sites with a high ground water table and may have been able to find suitable conditions for growth and survival at sites with little or no ground frost, e.g. springs and sites with shallow and moving soil water. Picea abies appears to have survived the cold and snow-poor winters of the early Holocene by a similar mechanism (Kullman 1995b, 1996) and still endures in the most continental regions of northern Fennoscandia (Kullman & Engelmark 1997). Thus, the current results do not conflict with previous inferences implying that throughout the North Atlantic region the winters were equally or slightly colder during the early Holocene than they are today (Kullman 1996, 1998; Svendsen & Mangerud 1997).

Received 19 May 1997revision accepted 21 October 1997