1Recent elevational range-margin performance of tree and shrub species was studied at a site in the Swedish Scandes. The methods included comparisons of historical and present-day range-margin records (m a.s.l.) in conjunction with age-determination of newly established saplings.
2Since the early 1950s, the range-margins of Betula pubescens ssp. tortuosa (mountain birch), Picea abies (Norway spruce), Pinus sylvestris (Scots pine), Sorbus aucuparia (rowan) and Salix spp. (willows) have advanced by 120–375 m to colonize moderate snow-bed communities. The non-native Acer platanoides (Norway maple) has become established just below the birch forest-limit. In concert with tree-limit rises by 100–150 m in the same region, the present results suggest a shift in reproduction and a significant break in the late-Holocene vegetation history.
3Ring-counting (in 2000) of a subsample of the recovered saplings revealed that, with one exception, they were aged between 7 and 12 years, i.e. they germinated after 1987. Since 1988 there has been strong and consistent winter warming, with some very warm summers, and this may ultimately have forced the vegetational changes.
4Reduced summer snow-retention has favoured seedling establishment and juvenile growth, and mild winters, with reduced risk of frost-desiccation, have enhanced survivorship and height increment.
5Certain seed-regenerating tree and shrub species have tracked recent climate change quite rapidly and more sensitively than vegetatively propagating field-layer species. Such species-specific responses may give rise to novel high-elevation vegetational patterns in a hypothetically warmer future world.
The main objective of the present study is to assess and quantify range-margin (= species limit) displacements (if any) of some tree and woody shrub species over the past 50 years and to discuss the ecological mechanisms behind them in terms of general theories about responses of tree-limit vegetation to global warming. The study is based on individual age-determination and historical records of the absolute altitudinal limits (m a.s.l.) of certain tree and shrub species within a strictly defined area, which are compared with their present-day equivalents. This approach enables precise comparison with climatic data and range-margin changes in other regions of the world.
Mt. Åreskutan in the southern Swedish Scandes, 63°26′ N; 13°06′ E (Fig. 1) is an isolated and prominent massif with a conspicuous alpine relief, with the highest peak, 1420 m a.s.l., more than 1000 m above the nearby valley bottom. Strongly fissured amphibolite is entirely bare or has just a thin soil cover. The study site is on the south-facing slope between 1360 and 1420 m a.s.l., where the slope is very steep and characterized by an intricate pattern of rock ledges, crevices, block fields and vertical cliff faces (Fig. 2).
Because of the proximity to the Norwegian Sea and the presence of low passes in the mountain chain, the macroclimate is fairly humid and maritime. The closest meteorological station with long-term instrumental records is Storlien/Visjövalen, at 642 m a.s.l. about 50 km west, where mean temperatures (1961–90) for January, July and the year are –7.6, 10.7 and 1.1 °C, respectively, and mean annual precipitation totals 847 mm. From 1980 to 1998, an official (automatic) weather station was operated close to the study site (about 1 km) at an elevation of 1280 m a.s.l. Mean temperatures for December–February, June–August and annually were –8.4, 5.8 and –2.2 °C, respectively, and the temperature range between the warmest and coldest month (13 °C), which indicates a fairly high degree of climatic oceanicity. The mean annual precipitation exceeds 1000 mm. All data are provided by the Swedish Meteorological and Hydrological Institute.
Strong winds and heavy precipitation, superimposed on a rugged local topography, control the accumulation of huge perennial snow drifts (8–10 m deep) in wind-sheltered hollows of the south-facing slope, from 1360 m a.s.l. and upwards.
Picea abies (L.) Karst. (Norway spruce) is totally dominant on the lower slopes. Just a few trees of Pinus sylvestris L. (Scots pine) are known from the entire area. The upper boundary of closed spruce forest is at about 750 m a.s.l. Above this there is a belt (c. 100 m) of Betula pubescens Ehrh. ssp. tortuosa (Ledeb.) Nyman (mountain birch). The current tree-limits, i.e. of trees at least 2 m tall, are situated at 975, 970 and 850 m a.s.l., for birch, spruce and pine, respectively.
In general, the vegetation pattern of the studied slope (1360–1420 m a.s.l.) is shaped by late snow melt, i.e. a patchwork of moderate and extreme snow-bed communities. The most characteristic plants in the field-layer are: Salix herbacea L., Gnaphalium supinum L., Sibbaldia procumbens L., Alchemilla alpina L., Saussurea alpina (L.) DC., Viola biflora L., Polygonum viviparum L., Oxyria digyna (L.) Hill and Silene aucalis. (L.) Jacq. The bottom-layer is dominated by Polytrichum norwegicum Hedw. and Kiaeria starkei (Web. & Mohr) I. Hag. The site is almost exactly at the upper limit of the low alpine belt, i.e. near the last outliers of Vaccinium myrtillus L., Empetrum hermaphroditum Hagerup and Phyllodoce caerulea (L.) Bab.
There are few visible signs of human disturbance and none of fire. Interviews with local residents confirm that the rugged terrain has always provided protection against interference from tourists and grazing and trampling by semi-domestic reindeer. Anyhow, the intensity of any such impacts has not changed over the past few decades.
The nearest meteorological station has continuous instrumental records since 1901, which show a general trend of summer warming amounting to 0.8 °C by 2000 (Fig. 3a). This is a minimum value for the post-Little Ice Age thermal upturn, since most northern Swedish stations with longer records display some additional warming over the period 1860–1900 (Alexandersson & Eriksson 1989). The warmest summer on record in the study area was 1997. Winter temperatures show large inter-annual and periodical fluctuations with a peak in the 1920s–1930s, followed by a cooler climate over much of the 1960s, 1970s and 1980s. The latter phase seems to have been an hemisphere-wide return to ‘normal’ neoglacial cool temperatures, similar to those of the Little Ice Age (Bradley & Miller 1972; Alexandersson & Eriksson 1989), but since 1987/88, the winters have been consistently milder, even than in the 1920s–1930s (Fig. 3b), forcing the mean annual temperature to its highest recorded level. Mean annual precipitation has increased throughout the 20th century.
The large perennial snow fields at the study site were previously suggested as a suitable target for snow phenology monitoring by Högbom (1897), when they may have been even larger and more persistent. Several recent discoveries of Late-Glacial and early Holocene tree megafossils in the forefields of the snow patches, right at their late-summer margins, indicate reduced snow retention into the summer and consequent meso-climate amelioration. The transition from extreme to mainly moderate snow-bed environments during the predominantly warm 20th century (Kullman 2000c; Kullman & Kjällgren 2000) is supported by the existence of broad zones entirely devoid of lichens on rock walls behind the largest snow patches. To judge from the progressively smaller and more scattered lichen thalli, a process of downwards lichen colonization seems to be under way on these faces.
In the late 1940s and early 1950s, Kilander (1955) carefully recorded the altitudinal limits (m a.s.l.) of all vascular plants occurring on the south-facing slope just below the summit of Mt. Åreskutan (Fig. 2). Reconnaissance investigations in 1998 revealed some young and vigorous tree saplings much above Kilander’s uppermost records (Kullman 2001a) and inspired a more thorough survey to be undertaken in 2000, within a plot of c. 300 × 300 m, covering most of the south-facing slope between 1360 and 1410 m a.s.l. During two days, the plot was systematically searched for tree and shrub saplings.
Elevation (m a.s.l.), size and vigour were recorded for each specimen and locations were marked with numbered plastic pegs. Half of the sample for each species was randomly selected and sacrificed for destructive age determination. The saplings were cut off below the root collar and annual ring counts on stem cross sections were made under a dissecting microscope. True age may of course be slightly greater than this.
Modern and historical elevations (m a.s.l.) were determined with a Paulin aneroid altimeter and cited values were rounded off to the nearest 5 m. For the present study, the modern topographic map provided fixed points in the terrain, which were used for calibration of the altimeter at least once an hour. In no case did the control readings differ by more than 5 m from the true values.
Seed viability of Betula pubescens and Picea abies was tested annually in populations growing close to their respective forest-limits in the Handölan Valley, c. 40 km west of the study area. These time-series are updates from a monitoring programme and the details of settings and test procedures are given by Kullman (1993a, 1996).
Altogether 60 saplings of different tree and shrub species were located within the sample plot, at elevations substantially higher than Kilander’s (1955) range-margins in the early 1950s. Seven tree and shrub species were recorded: Betula pubescens ssp. tortuosa, Sorbus aucuparia L., Picea abies, Pinus sylvestris, Salix glauca L., Salix lanata L. and Salix phylicifolia L., with Acer platanoides L. (Norway maple) appearing below the plot. Except for Sorbus aucuparia, which is spread by birds, species are chiefly wind-disseminated.
Betula pubescens ssp. tortuosa
Although Kilander (1955) noted no birch saplings above 1095 m a.s.l., the present study revealed 34 saplings between 1370 and 1410 m a.s.l (Fig. 4). They ranged in size between 13 and 41 cm and grew in seemingly undisturbed snow-bed vegetation. No shoot diebacks or browsing indications from recent years could be seen and the size of leaves was about the same as in trees close to the tree-limit. All specimens, except one, were single-stemmed, erect, and appeared to be quite young (no basal stem swellings) and rapidly growing. These individuals whose ages ranged between 8 and 12 years, and which were growing 435 m above their current tree-limit, may signify advancement of the range-margin amounting to 315 m in the past 50 years.
The birch tree-limit at this site was recorded as 960 m a.s.l. in 1952 (Kilander 1955) and 955 m a.s.l. in 1975 (Kullman unpublished data), but a 2.3-m high and multi-stemmed tree is now located at 975 m a.s.l. The long and slender shoots in its upper crown suggest that it has reached tree-size quite recently. The local tree-limit has thus risen by some tens of metres in the recent past, as in other localities in the same region (Kullman 2000a).
One sapling of Sorbus aucuparia (14 cm high and healthy) was recovered. It grew in snow-bed vegetation on a narrow ledge at 1370 m a.s.l. This is 375 m higher than the range-margin figure determined by Kilander (1955). Eight annual rings were counted at the root neck.
The range-margin of Picea abies was identified at 1141 m a.s.l. by Kilander (1955) in the form of a 30-cm high sapling, but the total of 15 specimens found between 1365 and 1410 m a.s.l. (Fig. 5) suggest an upward displacement of 240 m over the past 50 years.
Fourteen of the spruces were of similar size, general appearance and age, i.e. 8–17 cm high and with 7–11 annual rings discernible at or slightly below ground level. They were all single-stemmed, healthy and possessing 1–1.5 cm long annual shoots. The remaining individual had multiple stems, although no layering, and was somewhat taller (c. 40 cm). The foliage was lush and freshly green, without any signs of recent winter-dieback. This specimen was at least 51 years old, according to the ring count. Subfossil wood remains in the soil beneath the canopy, indicative of an ‘infinite’ Holocene age (cf. Kullman 2000b) could not be found, which supports its establishment in the late-1940s. It was presumably too small to have been noticed by Kilander in the early 1950s. Previous age-structure analyses of tree-limit spruce populations in the region have revealed that during the 20th century establishment peaked during the period 1940–49 (Kullman 1986a).
Adjacent to this spruce, a small cone (4 cm long) was found lying on the ground. After drying in the laboratory, 12 seeds were extracted. Their viability was tested by sowing on moist filter paper in a Petri dish (cf. Kullman 1984 for details) and two of the seeds had germinated after 9 days.
All spruces grew in plant communities dominated by Salix herbacea, Gnaphalium supinum and other species indicative of moderately late snow melt.
As far as is known, the uppermost spruce sapling, 1410 m a.s.l., is the highest individual of this species recorded anywhere in Scandinavia.
Six saplings of Pinus sylvestris were recovered between 1365 and 1375 m a.s.l., i.e. a present-day range-margin that is 340 m higher than the corresponding limit assessed by Kilander (1955). In contrast to the other tree species, the pines have established in sites with an incompletely closed plant cover.
The individual sizes ranged between 4 and 15 cm and the ages between 8 and 11 years. They have grown rapidly, without repeated growth setbacks due to winter frost-desiccation injury except for one specimen, which had lost its 1998-year terminal shoot. The vertical difference between the new range-margin of pine and its current tree-limit is 525 m.
A total of four tiny (c. 10 cm high) and seemingly young specimens of Salix (two S. glauca at 1390 m a.s.l. and one each of S. lanata and S. phylicifolia at 1270 m a.s.l.) were found above Kilander’s records from the 1950s. One of the S. glauca was aged at 11 years. A range-margin rise of 120, 165 and 120 m, respectively, has taken place. All these new individuals grew in snow-bed vegetation.
Surprisingly, a small (25 cm high) individual of Acer platanoides appeared at the base of an almost vertical south-facing cliff face, 905 m a.s.l. The site is right at the upper boundary of scattered birch tree copses and 70 m below the tree-limit. The maple sapling is growing in a fairly intact heath community dominated by Empetrum hermaphroditum, Vaccinium vitis-idaea L. and Vaccinium myrtillus. This specimen was not harvested for age-determination. It had grown rapidly and its age seemed to be less than 10 years.
Acer platanoides is not native to this part of Sweden, although a few young planted trees thrive in the village of Åre, at the base of the studied mountain, c. 400 m a.s.l. This species has never previously been recovered in the Swedish subalpine birch forest.
Other species and regional surveys
There is no evidence of invasion into the sample plot of any other species which, according to Kilander’s (1955) historical records, did not grow there in the early 1950s. The elevational limits of Vaccinium myrtillus, Empetrum hermaphroditum, Betula nana L., Phyllodoce caerulea and Juniperus communis were all found within ±20 m of the old records, which is well within the margin of error.
Reconnaissance studies within a regional monitoring network in the southern Swedish Scandes (cf. Kullman 2001a) corroborate the findings here. For example, a distinct cohort of pine saplings has become established during the 1990s at locations close to the tree-limit, where regeneration had been insignificant for some previous decades (Kullman 1993b). Moreover, at one particular site where historical records were available (Kilander 1955), a 300-m upward range-limit advancement of Picea abies (11-year-old-sapling) was disclosed for the period 1943–2000 (Leif Kullman, unpublished data).
Seed viability time-series
Updates from a long-term monitoring programme (Kullman 1993a, 1996) show that both Betula pubescens and Picea abies display high, and in the case of Betula, rising levels of seed viability over the past few decades (Fig. 6). Germinability correlated significantly with June–August mean temperature, recorded at Storlien/Visjövalen (c. 50 km west).
Range-margin rise and invasion into alpine tundra communities, high above the current tree-limits, is seen for many of the principal tree and shrub species. This very recent phenomenon is preceded and paralleled by a tree-limit and range-margin rise amounting to 100–150 m over the past century more generally in the study region (Kullman 2000a, 2001a). Tendencies of a reversal of this centennial trend were recorded during some colder decades prior to the late 1980s (Kullman 1997). These sharply contrasting developments over just a few past decades bear testament to the great responsiveness of tree-limit vegetation to climate variability. Thus, coordinated monitoring networks would probably reveal that certain parameters of tree-limit vegetation could be just as responsive to climate trends as some physical phenomena, e.g. glaciers and permafrost.
The density of new specimens at the elevated range-margin is so large and the age distribution so narrow that it is unlikely that these saplings represent mere accidental spread, but rather a fundamental change in limiting environmental factors. The combined approach of historical distribution records and age-determination of saplings provides evidence to support colonization being confined to the second half of the 20th century, particularly to the 1990s: regional tree- and range-limit surveys during the 1970s recorded no young saplings at correspondingly high elevations. Scarcity of young seedlings well above the tree-limit until quite recently seems to have been a consistent pattern at many Northern Hemisphere tree-limits (Graumlich 1994; Holtmeier 1995; Woodward et al. 1995; Hessl & Baker 1997). However, in widely separated regions of Europe and North America, tree seedlings and saplings now seem to be emerging further north and at higher elevations than previously known (e.g. Jacoby & D’Arrigo 1995; Hofgaard 1997; Luckman & Kavanagh 2000; Molau & Larsson 2000), suggesting a more general pattern.
The size attained during just one decade by tree saplings growing at a very high elevation sharply contrasts with much slower growth rates of experimental birch populations at the tree-limit over the relatively cool period 1981–92 (Kullman 1993a). Absence of reindeer grazing may have contributed to the fast growth of newly established seedlings, but grazing pressure has remained at a constantly low level.
The new range-margins are situated above any recorded in the studied region during the entire postglacial period. Tree-limits descended continually from their maximum more than 9000 radiocarbon years before the present until the late 19th century (Kullman & Kjällgren 2000; Kullman 2000a) and recent tree-limit and range-margin rises therefore denote a significant break in the trend for late-Holocene vegetation history. The most recent phase of this development, dealt with here, coincides with a period of exceptional winter warming and some fairly warm summers suggesting a causal relationship between climate change and colonization of new sites high above the tree-limit. Earlier and more complete thawing (transition from extreme to moderate snow-bed habitats) on a thermally favourable south-facing slope may be needed for successful establishment.
Propagules from the species concerned are readily spread to the alpine region from surrounding forests, as evident from several studies (Kullman 1979, 2000b; Molau & Larsson 2000). Wind tends to distribute and accumulate snow and seeds to the same places, ensuring that the most suitable sites for establishment (moderately wet soils) are regularly ‘flooded’ with seeds (Kullman 1984). A certain proportion of this seed rain is likely to originate from lower regions where annual variations in seed germinability are relatively small (Kullman 1979). Furthermore, the documented potential for a viable seed bank extending above the tree-limit ensures that shortage of germinable seeds is not a factor limiting elevational expansion (Kullman 1993a; Molau & Larsson 2000). Thus, although the increasing seed viability over the past few decades may be contributory, it may not be the prime reason for the range-margin rise, as confirmed by many saplings having germinated prior to the major rise in the 1990s.
Establishment, survival and growth of seedlings, which depend on local environmental conditions, are the most susceptive phases in the reproductive process (Kullman 1993a) and have benefited most from recent climate change. Reproduction has shifted from a strong reliance on stabilizing clonal reproduction (cf. Kullman 2000b) to more common generative reproduction, allowing elevational expansion. Warm summers (1988, 1996, 1997), are likely to have stimulated tree and shrub establishment from the seed rain or seed bank. Experimental field studies have substantiated the crucial importance of high summer temperatures (air and soil) for initial survival and juvenile growth at the tree-limit (cf. Kullman 1984, 1986b, 1993a; Sveinbjörnsson et al. 1996; Weih & Karlsson 1999). In addition, the most prominent recent climatic event, the run of mild winters after the late 1980s, substantially reduced the hazard of frost-drought injury in the study region (Kullman 2001a), and allowed unchecked height increment of saplings of all species. Young saplings (Figs 4 and 5) therefore lack the compact form characteristic of exposed high-elevation saplings suffering from repeated winter-dieback of the shoots, consistent with the importance of winter severity for growth form at the tree-limit (Payette & Lavoie 1994; Grace 1997; Kullman 1997). Thus, the current range-margin rise seems to depend on complex interactions of general and local factors (over all seasons), resulting in a particularly rapid and clear response at this mountain. It may not be a mere coincidence that this also appears to have been the location for initial immigration of trees (Pinus sylvestris, Picea abies and Betula pubescens) to the Scandes, at the Late Glacial/Holocene transition 11 700–10 000 14C-years bp (Kullman 2000b, 2001b; Kullman & Kjällgren 2000).
The distributional changes reported here largely conform with predictions of performance responses to global warming, although expansions have been much faster than generally anticipated (e.g. Baker & Weisberg 1995; Kupfer & Cairns 1996), i.e. there is practically no dispersal limitation. It is of particular interest that the exotic Acer platanoides has managed to spread upwards, almost to the birch tree-limit and together with other non-native tree species, e.g. Pinus contorta Dougl. and Pinus cembra L., has become established in similar environments elsewhere in the Scandes (Kullman 2000c; unpublished data). The expansion (altitude and latitude) of A. platanoides in various parts of Fennoscandia during earlier warm phases of the 20th century (Erkamo 1956; Aas 1970) makes it a good indicator. To some extent these observations corroborate assumptions that a new and warmer climate regime might facilitate rapid spread (‘jump-dispersal’) of exotic tree species (wildlings from ornamental parks, gardens and managed forests) into natural ecosystems, whose character may become substantially altered (e.g. Beerling & Woodward 1994; Sykes 2001). That would be in line with recent megafossil evidence showing that postglacial afforestation by many Fennoscandian trees occurred surprisingly swiftly (Kullman 1998a,b, 2001b) and analogous responses to climate warming displayed by palaeo-treelines in central Canada (MacDonald et al. 1993). Notably, at lower elevations in the European Alps, natural plant communities are currently invaded by exotic shrub species, possibly in response to the sequence of mild winters that has prevailed during the past decade (Theurillat et al. 1998).
The documented range-margin inertia of certain dwarf-shrubs and other plants indicates that not all species will respond equally to warming. Delayed and species-specific responses (Davis 1986; Callaghan & Jonasson 1995) are empirically substantiated here, because of the high proportion of clonally propagating plant species (perennials) in the Scandinavian high-mountain flora (cf. Jónsdóttir et al. 1996). If the warming is sustained, future subalpine/alpine plant cover is likely, at least transiently, to contain communities without previous analogues, as speculated, e.g. by Davis (1989). This could be further accentuated by atmospheric deposition and increased mineralization of nitrogen in combination with new disturbance regimes (altered land use patterns) (cf. Kullman 2000c, 2001a).
It would be unsafe to conclude, however, that the recent biogeographical processes documented here will inevitably continue on a broad front, leading to tree-limit advance and subsequent afforestation up to the level of the new range-margins. Such a course of development would probably presuppose much warmer summers than at present, and most importantly, substantially reduced wind pressure (cf. Woodward 1993; Crawford 1997). Moreover, the past few decades show that, regionally, short-term cold extremes (e.g. exceptionally cold winters with a poor snow cover) still occur within a generally warmer world. Such climatic disturbances may slow down growth and recruitment and decrease survivorship, thereby retarding or even temporarily reversing the general trend of vegetational landscape transformation in high mountain regions (Kullman 1997). Thus, predictions may be complicated by transient reversals to the natural neoglacial climate forcing (cf. Barry & Chorley 1992; Conway 1998).
Since the mid-20th century, Picea abies has grown at the study site as krummholz (Fig. 4) and has completed all the major stages of its life-cycle more than 400 m above its tree-limit. Data from the nearby meteorological station suggest that spruce can do quite well in locally favourable niches within a landscape with rugged topography with summer temperatures (June–August) around 5 °C and winters with a protective snow cover. Thus, spruce (and perhaps other tree species) may have grown close to the west or south-west edge of the Fennoscandian ice sheet during the Weichselian glaciation, as hypothesized from megafossil evidence (cf. Kullman 2000b, 2001b)
This study is financially supported by the Swedish Natural Science Research Council. Constructive criticism by anonymous referees is gratefully acknowledged.