LOCATION AND PHYSIOGRAPHIC SETTING

The Sylarna Mountains (63°01′N; 12°13′E) are located in the southern Scandes (Caledonides) of Sweden, right at the border between Sweden and Norway. The highest peak, 1761 m a.s.l., is in Norway, but terrain above 1700 m a.s.l. is also found in Sweden (Fig. 1). The area drains towards the northeast by the Sylälven river.

The entire massif rests on a high-plain at about 1000 m a.s.l. It is characterized by complex alpine relief, with sharp peaks, narrow crests, steep-sided rock slopes and deep cirque valleys with glacier niches (both empty and filled) (Fig. 2). The bedrock is made up of dark and very hard amphibolite, locally intersected with softer schists. Up to around 1200 m a.s.l., the solid rocks are covered by glacial till. At higher elevations, extensive boulder fields dominate the mountain landscape. Rockfalls, gelifluction, and wind erosion are active processes that constantly reshape the local topography.

The deglaciation of the highest peaks in the concerned region was well underway already by 14,000 B.P., whereas thinning residual ice filled the deeper valleys for some further millennia (Kullman and Kjällgren, 2000; Kullman 2002a). Currently, the Sylarna Mountains harbor three small glaciers, situated in east-facing cirques, where the prevailing snow drift from the west accumulates large snow masses. Two of these features may in fact be transitional between true glaciers and large snow/ice patches. The lowest frontal positions are between 1450 and 1600 m a.s.l. Their Holocene history have been studied only tentatively, but has been suggested to be of Neoglacial origin, initiated about 2500 yr ago (Lundqvist, 1969) or perhaps 1000 yr earlier, when substantial cooling and tree-limit recession took place in the region in response to delayed snow melt (Kullman, 2003a). These mountain glaciers are very close to the southern limit of modern montane glaciation in the Swedish Scandes, which may make them susceptible to subtle climate change.

PLANT COVER

A sparse tree cover, exclusively of mountain birch, is found close to the river Sylälven, 6 to 7 km east of the sampling sites. Small birch copses occupy sheltered south-facing slopes of the numerous morainic hillocks up to the tree-limit at 920 m a.s.l. (Fig. 3). During the past century, the local tree-limit advanced by some tens of meters altitudinally (Kullman, 1979). This is much less than the average for the region as a whole, which relates to the flat and wind-exposed (snow-poor) topography at higher elevations. The nearest outposts of mature Pinus sylvestris are found about 20 km northeast (680 m a.s.l.).

Above the tree-limit, the dominant plant cover consists of dwarf-shrub heaths (Betula nana, Empetrum hermaphroditum, Vaccinium myrtillus), snowbed communities, and mires. With increasing elevation, the plant cover becomes sparser and increasingly polarized between depressions with long-lasting snow cover and more or less snow-free crests. The snowbed communities are dominated by plants such as Salix herbacea, Sibbaldia procumbens, Alchemilla alpina, Oxyria digyna, Viola biflora, Bistorta alpina, and Ranunculus acris. The windblown crests display sparsely strewn specimens of Juncus trifidus, Festuca vivipara,Saxifraga cespitosa, Huperzia selago, Cardamine bellidifolia, and Salix herbacea.

A notable feature when discussing past distributional limits is that the Sylarna massif, above 1300–1400 m a.s.l., is characterized by several steep, south-facing rock and scree slopes with associated local climates favorable for plant growth. Therefore, many plant species reach their highest stations in the entire Swedish Scandes region (Kilander, 1955).

MODERN CLIMATE AND CLIMATE CHANGE

The climate is weakly maritime. The nearest meteorological station with long-term data is Storlien/Visjövalen (642 m a.s.l. and 32 km north). The mean temperatures (1961–1990) for January, July, and the year are −7.6, 10.7, and 1.1°C, respectively. The mean annual precipitation amounts to ca. 900 mm (all data from the Swedish Meteorological and Hydrological Institute). An automatic weather station has been operated since 1985 at 1030 m a.s.l. and 3 km northeast of the study site. The mean temperature for July is 7.9°C, while the annual precipitation is estimated to be 1000–1200 mm (Raab and Vedin, 1995).

Standard meteorological records at the Storlien/Visjövalen station reveal summer (June-August) warming by a linear trend (p < 0.01) of 1.1°C over the period 1901–2003. The scatter is large, however, and some significantly cooler decades occurred after the mid-century, with some tendencies for minor tree-limit descent (Kullman, 1997). The warmest and second warmest summers on record were 2002 and 1997, respectively. The winter temperature (December-February) has fluctuated with a larger amplitude on annual to decadal scales and the overall warming trend is less pronounced. Two distinct winter optima can be discerned during the 1930s and 1988–2002. The latter was 1.5 and 0.7°C warmer than the periods 1902–2002 and 1931–1940, respectively. The coldest winter on record was 1965/66, followed by 1969/70. Recent winter warming is a manifestation of a particularly intensive phase of the North Atlantic Oscillation (NAO) (cf. Hurrell and van Loon, 1997). Local precipitation trends are less reliable, although a tendency for annual precipitation to increase has characterized most of the past century, particularly since the 1970s.

ONGOING CHANGES OF ICE, SNOW AND PLANT COVER

Concomitant with the warming climate trend, the mountainscape has changed profoundly over the past century. Advancing natural tree-limits by a maximum of 140 m have, in certain topoclimatic settings, reduced the areal extent of the alpine tundra (Kullman, 1979, 2000, 2001a, 2001c). Recession of glaciers and perennial snowfields has exposed new ground for plant colonization. Geomorphological processes, e.g., many minor and major rock falls, rock slides, debris flows, and landslides at high elevations may relate to recent warming and the reduction of permafrost (cf. Rapp, 1992; Nyberg and Rapp, 1998).

Present-day comparisons with historical terrain photographs from 1908 (Enquist, 1910) and up to the present day reveal areal reduction of the existing glaciers by about 50% (Lundqvist, 1969; Kullman 2003a) (Figs. 2, 4). The recession accelerated during the long and hot summer of 2002, despite above-average winter precipitation that year (Kullman, 2003a). At the very end of the ablation period in early September in 2001, 2002, and 2003, large forefield areas downslope of glaciers and large quasi-permanent snowfields were melted out to a degree that was probably unique for hundreds or thousands of years. The newly exposed ground was entirely devoid of living lichens, bryophytes, and vascular plants. This was the setting of the recovered megafossil birch tree remains, which constitute the core of this paper. More or less complete late-summer thawing of snowfields, i.e. increasing length of the snow-free period, is a characteristic feature in the southern Swedish Scandes since 1996 or even earlier (Kullman, 2001c, unpublished data). As a biological consequence, the altitudinal limits of certain vascular plants have rapidly shifted upwards and species richness has increased in terrain that is not too wind-exposed (Kullman, 2002a, 2003b). For example, young seedlings and saplings of pine, spruce and birch have recently started to invade certain habitats in the alpine tundra up to 1500 m a.s.l. in the Sylarna Mts., where they were never recorded by previous intensive surveys (Kilander, 1955; Kullman, 1979, 1981b). Decreasing meltwater flow from snow patches during the late summer has triggered substantial expansion of grass and bryophyte vegetation at the expense of snowbed plants and bare mineral soil, respectively.

LAND USE

The alpine landscape is extensively grazed by semidomestic reindeer, which have replaced the wild reindeer since the late-19th century. Reindeer numbers have increased throughout the past century. Grazing and trampling has reduced the cover of lichens in the bottom layer of some types of dwarf-shrub heaths and also enhanced the natural erosion processes at wind-exposed sites (Kullman, 1997). On the whole, reindeer exert both positive and negative impacts on tree regeneration and there is no obvious relation between fluctuating reindeer populations and tree-limit evolution during the past century (Kullman, 1979).

SAMPLING AND SPECIES IDENTIFICATION

During the late summers of 2001 and 2002, extensive areas in the forefields of glaciers and large snowfields were intensively searched for subfossil wood remains of any tree species. The elevations of primary concern were between 1200 and 1500 m a.s.l. The lower figure represents the highest published record of subfossil pine in the Sylarna Mountains (Kullman and Kjällgren, 2000). A major part of the surveyed area has been free of glacier ice and late-summer snow since the end of the Little Ice Age in the late 19th century. In addition, large nonglaciated areas (heaths, mires, lakes) at the same elevations have been screened for presence of megafossils. No attempts were made to uncover wood remnants possibly buried in frontal moraines.

Megafossil sampling was carried out at three distinct sites, numbered 1–3 in Figure 1 and described below:

No megafossils emerged outside these three sites. All wood samples that were found in the study area were retrieved, measured and submitted for radiocarbon dating. They were all positively identified as Betula pubescens coll. by the presence of bark fragments and distinguished from Betula pendula by a chemotaxonomical method (Lundgren et al., 1995).

DATING

Radiocarbon dating of the wood samples was performed by Beta Analytic Inc., Miami, Florida. The dates are reported and discussed in terms of conventional radiocarbon years B.P. (=AD 1950), in order to facilitate comparison with previous megafossil dates reported from this specific geographical region. In addition, calendar dates (cal B.P. ± 1 S.D.), calibrated according to Stuiver et al. (1998), are given in Table 1 and occasionally in the text.

HOLOCENE LANDSCAPE HISTORY

The present results provide direct evidence for birch tree growth at unprecedented high altitudes during the early Holocene, and only the highest peaks in the region remained treeless at that time. The uppermost record, at 1500 m a.s.l. (8710 ± 60 B.P.), is 580 m above the nearest present-day birch tree-limit. This is likely to be a minimum figure, however, since the site of recovery was undoubtedly secondary, being a protalus rampart with a considerable downslope transport of blocks, soil and plant debris from adjacent cirque walls. Therefore, it seems reasonable to seek the original growing sites at least 1550 m a.s.l., i.e. beneath present-day permanent snow cover. Thus, tree birch may have grown as much as 630 m above its current tree-limit location.

Comparison with the megafossil tree-limit history of pine in the same region (Kjällgren and Kullman, 2000) reveals that birch trees may have attained positions 350 m higher than the upper limit of pine during the early Holocene. Today, the elevational separation of these specific tree-limits rarely exceeds 200 m in this region of the Scandes. These new results suggest an early Holocene landscape quite different from the present-day treeless and floristically depauperate alpine tundra. It will be an exciting task for future researchers to explore this topic in greater detail, using mega- and macrofossil records.

The present findings support, to some extent, the initial hypothesis that shortly after the regional deglaciation, birch grew above the pine-dominated Caledonian Forest, that initially clothed the valleys. However, the structure and regional coherence of such a belt at this early period cannot be inferred from existing data. Notably, these derive from a very special kind of environment, primarily selected as suitable for long-term preservation of wood remains. It is an open question whether this site was more favorable than surrounding mountain landscapes also for the growth of birch. In fact, the study area seems to possess such qualities. This is explicitly manifested by rapid recent invasion of tree saplings (birch, pine, spruce, and rowan) up to the same elevations as the recovered megafossils in response to warming during the past few decades (Kullman, 2002b, 2003b). It may well be that early postglacial tree growth of mountain birch was consistently restricted to such scattered high-altitude, outlying sites, with favorable topographical and local climatic conditions. In particular, an ample snow cover (cf. Kullman, 2002a) could locally have provided optimal conditions for establishment of metapopulations of mountain birch in a mountainscape otherwise characterized by too little snow for the comfort of this species, generally associated with fairly deep snowpacks and long winters (Kullman, 1981a; Aas and Faarlund, 1996). Moreover, existing megafossil (Kullman, 1995) and pollen data (Segerström and von Stedingk, 2003) from the lower reaches of the same major catchment area as the study site indicate that the mountain birch ecosystem attained its major areal development after approximately 7000 B.P.. Certainly, a general and broad birch belt reaching as high as 1500–1600 m a.s.l. prior to that would have left a fundamentally different fossil record. Likewise, in northernmost Sweden, megafossil data do not support the existence of a vertical (monospecific) birch belt during the time span under discussion here, but rather a tree-limit ecotone composed of mixed pine and birch stands (Kullman, 1999). Farther north (Norway), pollen and some macrofossil data have been interpreted as a pure birch forest belt during the early Holocene (Vorren et al., 1996; Jensen et al., 2002). Notably, however, very little megafossil research has been executed in this region.

Conceivably, the megafossils uncovered in pronounced snow accumulation settings may represent gradual population demise and burial in response to increasing snow cover over the period 8700–6200 B.P.. Concurrently, birch existed, although fairly scarcely at lower elevations, where it gained in landscape-scale dominance following the later part of the above-mentioned time interval (Kullman, 1995). Different lines of evidence from the southern Norwegian Scandes suggest that repeated phases of strong snow-avalanche activity commenced during the period concerned (Blikra and Selvik, 1998; Seierstad et al., 2002).

In conclusion, it is possible that the development of a deeper and more persistent snow cover (8700–6200 B.P.) exterminated the early-Holocene outlying birch populations, but favored growth, densification and coalescence of birch populations at lower elevations. From now on, a coherent birch belt of modern mode may have started to develop above the receding pine tree-limit, as outlined by megafossil data of both birch and pine (Kullman, 1995).

It is an exciting possibility that the early Holocene “birch pockets” at high elevations may rapidly come into existence again in response to predicted future global warming (SWECLIM, 2001), an assumption based on recent observations of tree sapling invasion into high-alpine tundra (Kullman, 2002b, 2003b). Thus, tree-limit adjustments will not necessarily manifest as displacements of distinct bands of woodland (cf. Bush, 2002).

PALEOCLIMATIC INFERENCES

By analogy with modern tree-limit patterns and responses (Malyshev, 1993; Aas and Faarlund, 1996; Kullman, 2001a; Kjällgren and Kullman, 2002), the all-time-high birch tree-limit during the early Holocene indicates warmer and drier conditions than present summers and possibly the warmest summers of the entire postglacial period. Widespread early Holocene tree growth in close spatial association with sites of present-day glaciers and large perennial firn patches suggests that at least the latter landscape element was virtually absent (i.e. higher glaciation limit) for a minimum period bracketed by 8700 and 6200 B.P., as inferred also for other glaciers in the South-Central Scandes (e.g., Nesje and Kvamme, 1991). In addition to warmer summers, this situation may also reflect lower winter precipitation and/or modest snow-accumulating winds. This general character of the early Holocene climate is compatible with paleoclimatic models based on the Milankovitch theory of orbital climate forcing (COHMAP, 1988; Berger, 1989), implying higher summer radiation and thus warmer summers than present. Virtually the same inferences can be made from the tree-limit history of Pinus sylvestris in the study region (Kullman and Kjällgren, 2000) and from combined pollen-macrofossil data in North Norway (Jensen et al., 2002). Under these circumstances, i.e. a stronger insolation regime, the existence of outlying stands in the topoclimatically complex mountainscape makes sense. These inferences concerning the early postglacial climate contrast with the idea of an oceanic climate in northern Scandinavia in the early Holocene (e.g., Seppä and Hammarlund, 2000)

In principle, the vertical difference between the highest postglacial birch record and the present-day tree-limit position, i.e. 630 m, may represent decrease of birch-relevant (summer) temperatures by 3.8°C, applying a crude elevational lapse rate of −0.6°C for 100 m altitudinal increase (Laaksonen, 1976). Tentative correction for glacioisostatic land-uplift (Dahl and Nesje, 1996) reduces this figure to ca. 3.0°C. This calculation suffers from some uncertainties, e.g., possible spatial and temporal variations of the lapse rate (cf. Mook and Vorren, 1996) and the vertical incoherence of the upper birch distribution during the early Holocene. Nonetheless, this lapse rate–based reconstruction gains in credibility by the finding that a maximum of 140 m birch tree-limit rise during the past century corresponds to 1°C summer warming (see above), i.e. near what would be predicted by the lapse rate used above. Moreover, the early Holocene figure obtained compares quite well with independent paleoclimatic reconstruction and models from the Scandes and adjacent regions (COHMAP, 1988; Kullman, 1999; Shemesh et al., 2001; MacDonald et al., 2002; Bigler et al., 2003), but exceeds other reconstructions (Nesje and Kvamme, 1991; Rosén et al., 2001; Bigler et al., 2002). These discrepancies may represent both regional differences and different methodological approaches. The obtained Holocene tree-limit/climate trend in Scandinavia agrees well with analogous reconstructions in various regions of North America and the European Alps (e.g., Ritchie et al., 1983; Luckman, 1990; Clague et al., 1992; Pellat and Matthewes, 1994; Munroe, 2003; Ali et al., 2003), which further supports a causal interpretation in terms of “orbital forcing.”

RECENT GLACIAL AND NIVAL RECESSION IN PERSPECTIVE

Birch wood rapidly decomposes unless it is immediately covered by peat, compacted till, glacier ice or firn. The megafossil birch remnants discussed here were not covered by peat when found. They all occur within the limit of the maximum Little Ice Age glacier extension. It is reasonable to assume that they have been covered by ice/snow and till since their death and burial, i.e. 8700–6200 B.P., and up to the very recent past. Most certainly they would not have “survived” more than a few, possibly no, previous release periods of similar duration as the past century and subsequent reburial and compression by a heavy snow and ice pack. Thus, it is to be concluded that in this part of the Scandes, ice and snow recession during the warm 20th century has recently reached a magnitude that is likely to be exceptional in the perspective of several millennia. Analogous deductions in terms of summer temperature for the same region have been drawn by matching 20th century tree-limit rise of pine and mountain birch with their long-term (Holocene) histories (Kullman and Kjällgren, 2000; Kullman, 2003a). This anomalous recent reduction of snow and ice in the alpine landscape fits a general world-wide trend (Haeberli and Beniston, 1998; Luckman and Kavanagh, 2000; Bradley, 2000; Thompson, 2000; Thompson et al., 2002; Markels, 2002), signifying a fundamental break in the Neoglacial cooling trend (Luckman, 1998; Kullman, 2003a; Mann and Jones, 2003). The present discoveries indicate that the recent warming, if continued, can provide paleoecological science with unique opportunities to date and examine frozen plant remains that are becoming uncovered.