SEARCH

SEARCH BY CITATION

Keywords:

  • climate change;
  • European Alps;
  • ForClim;
  • forest succession model;
  • Holocene;
  • macrofossils;
  • palaeobotany;
  • pollen data;
  • treeline dynamics;
  • vegetation response

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  • 1
    We used the forest succession model ForClim to simulate Holocene treeline dynamics along an elevational transect in the Central European Alps, in order to explore the extent and cause of changes in treeline altitude and composition.
  • 2
    A temperature reconstruction independent of vegetation proxies was used to drive the model, and the simulation results were compared with Holocene pollen and macrofossil records from a nearby site close to the present-day treeline.
  • 3
    The simulation results yielded treeline fluctuations of about ± 100 m (2375–2600 m a.s.l.), confirming earlier palaeoecological studies and quantitatively corroborating the interpretation of most palaeoecologists that decadal- to centennial-scale Holocene fluctuations of pollen and plant macrofossil frequencies reflect treeline shifts rather than productivity changes alone.
  • 4
    The simulated changes in species composition and treeline position show general agreement with palaeobotanical data between 11 000 and 4500 calibrated radiocarbon years BP. In the late Holocene, however, palaeobotanical evidence indicates a distinct lowering of the treeline, while simulation projected continuous forest cover up to an altitude of 2400 m a.s.l.
  • 5
    Our results indicate that changes in temperature alone can account for changes in treeline elevation for the first half of the Holocene. The discrepancy between simulation results and palaeobotanical records since 4500 cal. BP supports the hypothesis of a strong human influence on the Alpine treeline during the late Holocene.
  • 6
    Combining palaeoecological methods with vegetation modelling can disentangle climatic effects and early human impacts on long-term vegetation dynamics. Forest succession models may not only help palaeoecologists to achieve a better understanding of the factors driving past vegetation changes, but their validation with long-term empirical data is also an important step towards applying these models to the assessment of future vegetation dynamics in a changing climate.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Since the elevation of the upper treeline (i.e. the uppermost limit of trees) in mountains is strongly affected by climate, especially by temperature during the growing season (cf. Körner 1998), past changes in treeline location have been widely used to infer past climatic variations (Vorren & Stavseth 1996; Seppä & Birks 2001). Treeline changes during the Holocene (i.e. the past 11 500 years) have been reconstructed in various ways, mainly using pollen analyses, sometimes supplemented by plant macrofossil data (Jalut et al. 1996; Tinner et al. 1996; Tinner & Theurillat 2003). Although such profiles permit a fairly good reconstruction of the vegetation surrounding the site, the causes of vegetational shifts are often ambiguous, because the proxy data reflect the integrated effects of climatic, biotic and anthropogenic impacts. The relative importance of the various drivers of vegetation dynamics in the past (climate, human activity) and the processes affecting palaeobotanical records of past vegetation change (pollen influx, differences in species-specific pollen production) at high elevations is therefore still speculative (Birks et al. 1996).

General trends in the European Alps, such as the reforestation at the beginning of the Holocene and the decline of the timberline (i.e. the uppermost limit of closed forest) by several hundred elevational metres at c. 5000 calibrated radiocarbon years before present (cal. BP), are documented unambiguously in the palaeo records (Burga & Perret 1998). However, uncertainties exist about how to interpret treeline pollen and macrofossil fluctuations at decadal to centennial time scales. Reduced percentages of pollen of treeline species (e.g. Pinus cembra) may reflect either lower pollen productivity in the vegetation belt just below the alpine treeline or a regional decline in treeline elevation, both of which are possible consequences of decreased summer temperatures (Bugmann & Pfister 2000; Heiri et al. 2004). Similarly, the few treeline macrofossil records from the Alps could reflect either decreasing primary net production of (surviving) individual plants or local treeline fluctuations.

A recent pollen and macrofossil study at Gouillé Rion (Central Swiss Alps) (Tinner & Theurillat 2003) supports the widely accepted palaeoecological hypothesis that the upper treeline (but not timberline) in the Alpine region oscillated by no more than ± 100 m throughout the Holocene (e.g. Patzelt & Penz 1975; Burga & Perret 1998; Haas et al. 1998). The uppermost Holocene treeline position is proposed to have been situated c. 100–150 m higher than today, although recent charcoal analyses suggest somewhat higher positions (Thinon & Talon 1998). Indeed, Carcaillet et al. (1998) suggest a treeline in the Vanoise massif (French Alps) that is at least 200 m higher than proposed by palynologists. Dynamic model approaches are needed to disentangle the information contained in palaeoecological records (Prentice 1986; Shugart 1989; Keller et al. 2002).

Among the many approaches to simulating forest dynamics (Lischke 2001), the so-called ‘gap models’ (Shugart 1984) have proved to be particularly suitable for comparison with records of observed changes in species composition, such as those contained in pollen data. These models simulate the establishment, growth and mortality of individual trees on small patches of land as a function of biotic (competition) and abiotic factors (climate).

Most of the studies using forest gap models to simulate Holocene vegetation dynamics (Lotter & Kienast 1992; Solomon & Bartlein 1992; Lischke et al. 1998; Keller et al. 2002; Lischke et al. 2002) have required further assumptions, such as migration times and migration paths of tree species returning from interglacial refugia, prescribed fire intervals or human management to improve the fit between simulation results and pollen data (Keller et al. 2002; Lischke et al. 2002). Furthermore, circularity between cause and effect was often introduced by inferring climate data (e.g. temperature) from the pollen record with which the simulation results were compared (Solomon et al. 1981; Solomon & Shugart 1984; but see Lischke et al. 2002).

We made as few assumptions as possible about the biotic and abiotic factors driving Holocene forest dynamics at Gouillé Rion, using a chironomid-based reconstruction of Holocene summer temperatures from the Bernese Alps (Northern Swiss Alps; Heiri et al. 2003a) to infer past treeline dynamics in the Valais (Central Swiss Alps) using the forest gap model ForClim (Bugmann & Solomon 2000). We evaluated the hypothesis (Tinner & Theurillat 2003) that the upper treeline in the European Alps showed limited variation (± 100 m) by simulating Holocene forest dynamics along an elevational transect. We tested ForClim with respect to its long-term behaviour along an elevational gradient and the accuracy of its simulation results based on a long-term vegetation reconstruction under conditions where human influences can be ruled out (i.e. the early Holocene). We then used this model to isolate the climate-induced effects on treeline dynamics at Gouillé Rion, so as to contribute to the discussion about the onset and extent of human impacts on high-elevation forests and treeline in this region.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

study area

This study was conducted in the Valais, i.e. the Swiss part of the Rhone Valley (Fig. 1). The Valais is characterized by a dry continental climate with an average annual precipitation of about 600 mm at low altitudes (600 m a.s.l.) and about 1000 mm at 2300 m a.s.l., where the annual mean temperature is about 0.5 °C (July 8.6 °C and January −6.4 °C). The natural treeline in the study area occurs at about 2350–2400 m, although human land use means that hardly any forests are found above 2200 m today. Pinus silvestris and Picea abies dominate forests at low altitudes (800 to about 1700 m), followed by nearly pure Picea forests up to about 2000 m. Larix decidua and Pinus cembra become more abundant above 1900 m and attain dominance at elevations above 2000 m. Treeline is formed mainly by Pinus cembra (Hättenschwiler & Körner 1995; Ott et al. 1997), with a narrow belt of Alnus viridis scrub marking the transition from Larix-Pinus cembra forest to meadows and tundra vegetation (Tinner et al. 1996).

image

Figure 1. Topographical map of Switzerland. Low elevations, dark; high elevations, white. The study area in the Valais is encircled (○). © 2005 swisstopo.

Download figure to PowerPoint

forest model

The ForClim model (Bugmann 1996; Bugmann & Cramer 1998; Bugmann & Solomon 2000) was originally developed for Central Europe and based on the FORECE model (Kienast 1987). ForClim has evolved from a simulator of forests in the Swiss Alps to a general model that is applicable to temperate forests of Central Europe (Bugmann & Cramer 1998), eastern North America (Bugmann & Solomon 1995), the Pacific Northwest of the USA (Bugmann & Solomon 2000), north-eastern China (Shao et al. 2001) and the Colorado Front Range in the Rocky Mountains (Bugmann 2001). ForClim consists of three modular submodels: ForClim-E considers the abiotic environment, ForClim-S soil carbon and nitrogen turnover, and ForClim-P tree population dynamics.

The major state variables in ForClim are the diameters of individual trees, and biomass was chosen from a range of possible output variables for comparison against palaeoecological data.

In ForClim, as in most forest gap models, migration is not considered (i.e. it is assumed that all species are potentially available for establishment at all times). Establishment rates are determined from light availability at the forest floor, browsing intensity and winter minimum temperatures. Light availability and winter temperatures are used as discrete, species-specific thresholds that prevent regeneration when it is too dark or too cold for a species, whereas regeneration decreases continuously with browsing rate. Mortality is simulated as a combination of age-related and stress-induced rates (Botkin et al. 1972; Kienast 1987), giving rise to high mortality of both small (due to strong competition for light) and old trees (due to low vigour) (cf. Bugmann 1994). Trees that grow slowly due to adverse environmental conditions are more likely to be subject to stress-induced mortality. Tree growth is determined from a simple carbon budget approach (Moore 1989), taking into account light availability, growing season temperature (degree-days), summer drought and nitrogen availability.

Abiotic driving variables in ForClim are, primarily, the monthly mean temperatures and monthly precipitation sums of every simulation year. These variables can be provided either by measured time-series data, or by using a weather generator based on the long-term statistics. Annual degree-day sum, a summer drought index and a proxy for the winter minimum temperature are then calculated and used to determine tree population dynamics, as well as carbon and nitrogen turnover of the forest.

Rather than being calibrated against large-scale measured data, ForClim uses parameters estimated from the autecological literature (cf. Bugmann 1994; Bugmann et al. 1996). Disagreement between simulation results and measured data has usually been interpreted to indicate an inadequate representation of ecological processes rather than a parameter estimation problem. This approach was essential for deriving a simulator that is capable of running in ecologically and climatically different regions without any structural adjustments (cf. Bugmann & Solomon 2000). We used ForClim Version 2.9.3, as described by Risch et al. (2005), which features an asymptotic relationship between tree diameter and tree height, rather than the parabolic equation originally proposed by Botkin et al. (1972).

current climatic data

Current climate data based on long-term records from Sion Aerodrome and Grande Dixence (Table 1), two nearby weather stations, were used to define an altitudinal transect across the upper treeline near Gouillé Rion. A virtual site was located every 50 elevational metres between 1700 and 2650 m a.s.l. (cf. Bugmann & Pfister 2000) and average monthly temperature and precipitation data from the two stations were inter- or extrapolated linearly to all altitudes, resulting in a lapse rate of 0.5 °C per 100 altitudinal m.

Table 1.  Long-term average climatic data for the SMA weather stations Sion Aerodrome and Grande Dixence, and calculated climate for the simulation sites HITE2000 and Gouillé Rion. At Grande Dixence, the data for October to December of 1985 were missing; the data from October to December 1984 were used to fill this gap. T = temperature; P = precipitation
 Measurement periodElevation (m a.s.l.)T (°C)P (mm year−1)
Sion Aerodrome1978–98 4839.6 623
HITE2000Calculated20002.0 974
Grande Dixence1965–8521661.21013
Gouillé RionCalculated23500.31056

holocene climate scenario

A quantitative, chironomid-based July air temperature reconstruction covering the past c. 12 000 years at Hinterburgsee in the Northern Swiss Alps (Heiri et al. 2003a) is one of few continuous quantitative climate reconstructions. The smoothed, 62-sample July air temperature reconstruction (as described in Heiri et al. 2003a) was used with the revised age-scale for the Hinterburgsee sediment record (described in Heiri et al. 2004) to produce a Holocene temperature scenario for the present study. Hinterburgsee is situated well below the current alpine treeline, and it is therefore unlikely that limnological parameters, and thereby chironomid assemblages in the lake, were affected by Holocene treeline fluctuations. Furthermore, a detailed comparison of palaeobotanical proxies and the chironomid record suggests that vegetation changes during most of the Holocene had only a minor influence on the chironomid fauna of the lake (Heiri et al. 2003b). Distinct shifts in the chironomid assemblages coincide with recent phases of increased human activity inferred from palaeobotanical evidence but, prior to 2000 cal. BP, the Hinterburgsee reconstruction shows good agreement with a number of other palaeotemperature records from the Alpine region and Central/North-western Europe (Heiri et al. 2003a; Heiri et al. 2004) and is therefore assumed to provide a realistic image of summer temperatures.

Quantitative estimates of mean monthly temperatures are not available for the rest of the year. Mean July air temperatures of the smoothed Hinterburgsee reconstruction were therefore converted to temperature anomalies relative to the long-term Holocene mean of 11.9 °C (i.e. slightly lower than the present mean July air temperature of 12.3 °C at Hinterburgsee; Heiri 2001). Assuming a similar historic development of summer and winter temperatures, we added the temperature anomaly to the current monthly temperature for each simulation site. Although this simplistic assumption may introduce a bias, it is quite likely that winter temperature is not as critical as growing-season temperature at high altitudes (e.g. Körner 1998). Finally, due to the lack of an accurate Holocene precipitation scenario for the Valais, we assumed the precipitation to be identical to that of the current climate.

pollen and macrofossil data

We compared our simulation results with well-dated pollen and macrofossil records from Gouillé Rion at 2343 m a.s.l. (Tinner et al. 1996). These suggest the presence of alpine meadows in the lake catchment before 11 300 cal. BP, when Larix decidua forests established in response to climatic warming. Pinus cembra expanded at about 10 200 cal. BP and reached a first maximum at around 9500 cal. BP (Wick & Tinner 1997). Mixed Larix-Pinus cembra forests, with regular Betula presence, persisted until c. 8500–8400 cal. BP, when Larix became less important. Subsequently, Pinus cembra was the dominant tree species until, at c. 3800 cal. BP, forests were substituted by Juniperus nana heaths and Alnus viridis thickets. The establishment of today's pastoral landscape, dominated by herbaceous species and prostrate dwarf shrubs (e.g. Loiseleuria procumbens, Empetrum nigrum, Salix herbacea), occurred at c. 1700 cal. BP (Tinner et al. 1996). Superimposed on these millennial trends are centennial fluctuations of treeline vegetation that have been interpreted as treeline oscillations (Tinner et al. 1996; Wick & Tinner 1997; Haas et al. 1998; Tinner & Theurillat 2003).

Influx from lowland forests often obscures the local vegetation signal in pollen records and therefore prevents reliable reconstructions of treeline elevation (Tinner et al. 1996). Furthermore, pollen percentage sums are difficult to compare with simulation outputs of biomass per unit area. Both problems may be overcome by the additional use of macrofossil records (Tobolski & Ammann 2000). In particular, if sedimentation rates have been fairly constant, macrofossil data may be interpreted as a proxy of past relative biomass (cf. Birks 2001), allowing more direct comparison with our simulation results. Moreover, macrofossil series are hardly affected by long-distance transport. However, pollen data may provide more accurate quantitative estimations of past vegetational changes, as they rely on higher counting sums (Lang 1994).

simulation experiments

Simulation of Holocene forest dynamics at high elevations in the Valais

Simulations were started from bare ground at 12 500 cal. BP, based on the assumption that temperatures at the beginning of the Younger Dryas were too cold for the growth of forests above 1700 m a.s.l. in the Alps (Schwander et al. 2000; von Grafenstein et al. 2000). Macrofossil and pollen records confirm that the treeline was then situated at about 1500–1600 m a.s.l. (Welten 1982; Tobolski & Ammann 2000).

We assumed a nitrogen-rich soil with 100 kg N ha−1 (thus effectively minimizing nutrient effects) and a moderate north-east to east-facing slope (cf. Bugmann 1994) for each of the 20 elevations (‘sites’) along the altitudinal transect. Soil water holding capacity (‘bucket size’) for the entire transect was estimated as 12.1 cm based on water retention potential and soil wetness data from the Swiss soil suitability map (Frei et al. 1980; Löffler & Lischke 2001). For each site, the model was run for 200 independent patches (Bugmann et al. 1996) from 12 500 cal. BP up to the present time, and the output averaged for each elevation. The model has a time step of 1 year, but simulated values were monitored every 50 years only, so as to keep the amount of output data at a reasonable level.

To distinguish between forest and non-forest, above-ground biomass at treeline was taken as 25 t ha−1 (for a further discussion on treeline definitions, see Körner & Paulsen 2004).

Comparison of simulated and measured values at Gouillé Rion

Pollen and macrofossil data from Gouillé Rion (2343 m a.s.l.) were compared with the simulation site at 2350 m a.s.l. To take account of potential pollen transport due to vertical air-mass movements (Tobolski & Ammann 2000), we merged the simulation results within an elevational band of ± 100 m around the target elevation, using weightings of 40% for the simulation site at 2350 m (corresponding to Gouillé Rion), 20% for sites at 2300 and 2400 m (± 50 m) and 10% for sites at 2250 and 2450 m (± 100 m).

As macrofossils allow a direct comparison with the model output from the simulation site at 2350 m a.s.l., no merging with other simulated elevations was performed for this comparison.

Sensitivity study of the precipitation regime

Our assumption that precipitation was constant is almost certainly too simplistic on Holocene time-scales (Guiot et al. 1993; Tinner & Ammann 2001; Magny et al. 2003). As it is known that the competitive relationships between Picea and Pinus cembra are partly determined by moisture conditions (cf. Ott et al. 1997), we ran a sensitivity study for the simulation at 2000 m a.s.l., reducing today's annual precipitation (974 mm year−1) stepwise by 50 mm year−1 up to a maximum reduction of 400 mm year−1, distributed evenly throughout the year. For each value of reduced precipitation, we determined the corresponding equilibrium species composition according to the method described by Bugmann (1997).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

simulation of holocene forest dynamics at high elevations in the valais

Tracking tree species composition along the elevational transect (results for 6400 cal. BP, within the warmest period of the simulation, are shown in Fig. 2), we see a typical sequence from Abies-Picea forests at lower elevations (1700–1900 m a.s.l.) to Pinus cembra-Larix forests at high elevations (2400–2550 m). Nearly pure Picea forest is simulated at mid-elevations (2000–2250 m), where competitiveness of Abies alba, Larix and Pinus cembra is reduced.

image

Figure 2. Simulated tree species composition along the elevational transect at simulation year 6400 cal. BP (warm period).

Download figure to PowerPoint

Figure 3 shows simulated total above-ground biomass from 11 000 to 2000 cal. BP. At all dates, the decline of biomass is quite abrupt as upper treeline is approached and highest biomass values are simulated about 100–200 m below treeline, rather than at the lowest elevations. During the Holocene treeline elevation shifted between 2375 and 2600 m a.s.l. (i.e. only by about ± 100 elevational metres). The response of treeline elevation to changes in temperature is almost immediate, with cross-correlation analysis showing that the highest correlation occurs at a lag of less than 50 years.

image

Figure 3. Contour plot of simulated total biomass (t ha−1) along the elevational transect for the period 10 000–2000 cal. BP.

Download figure to PowerPoint

Focussing on forests just below current potential treeline (i.e. 2200–2350 m a.s.l.), it is remarkable that simulations show a continuously changing tree species composition (Fig. 4 shows data for 2250 m a.s.l.). It is particularly noteworthy that the response of the model is asymmetrical, i.e. there is an immediate response to decreasing temperature, leading to a dieback of Picea and to more cold-adapted species such as Pinus cembra taking advantage of reduced competition, but, unlike treeline position (Fig. 3), there is no immediate recovery as the temperature increases.

image

Figure 4. Non-equilibrium behaviour of forest dynamics at the simulation site 2250 m a.s.l. (just below treeline) for the period 12 000–2000 cal. BP.

Download figure to PowerPoint

comparison of simulated and reconstructed forest dynamics at gouillé rion

The comparison of the simulation results for 2350 m with macrofossil data from Gouillé Rion showed general agreement for the first half of the Holocene, i.e. prior to 4500 cal. BP (Fig. 5). After the initial colonization by trees around 11 500 cal. BP, simulated biomass of Pinus cembra and Larix, as well as their relative importance across time, are reflected by similar patterns in the macrofossil record. Major dieback events, mainly of Pinus cembra, are evident around 8000, 6500 and 4500 cal. BP, in both data sets (Fig. 5) and a smaller dieback event in the macrofossil record at around 9300 cal. BP, corresponds to a simulated moderate dieback at around 9500 cal. BP. Early in the simulation, i.e. at around 10 500 cal. BP, the model produced a dieback that is not evident from the macrofossil record, but it is clearly visible from the pollen data (Fig. 6), with which the simulation results also compare favourably, mainly for the three strong dieback events of Larix and Pinus cembra around 10 500, 8000 and 6000 cal. BP. According to the pollen (Fig. 6) and macrofossil (Fig. 5) data for this location, Betula was present throughout the entire Holocene, but the model predicts only negligible abundance.

image

Figure 5. Macrofossils (a) and simulation results (b) for Gouillé Rion (2350 m a.s.l.). The order of species in the legend corresponds to the order in which they are plotted in the figures.

Download figure to PowerPoint

image

Figure 6. Pollen-percentages (a) and simulation results (b) for Gouillé Rion (2350 m a.s.l.). The simulation results represent the species composition along an elevational belt between 2250 and 2450 m a.s.l. No correction factors were applied to the pollen data. The order of species in the legend corresponds to the order in which they are plotted in the figures.

Download figure to PowerPoint

After 4500 cal. BP, the simulation results diverge from both macrofossil (Fig. 5) and pollen data (Fig. 6). The increasing amount of Alnus viridis and Juniperus nana in the palaeobotanical record suggest a lowering of the treeline, whereas the simulation results show continuous forest cover, dominated by Pinus cembra, at elevations around 2400 m a.s.l. Furthermore, minor amounts of Picea appear in the simulation as early as 10 000 cal. BP, and Picea becomes codominant or even dominant after 7000 cal. BP, while the pollen data contains no Picea until about 5500 cal. BP. Picea is also missing in the macrofossil record, suggesting that this tree species was not important locally throughout the entire Holocene.

sensitivity to the precipitation regime

Under current climatic conditions at 2000 m a.s.l. (i.e. annual precipitation sum = 974 mm) the model simulated a Picea forest (Fig. 7). Reducing precipitation by 150 mm year−1 (i.e. a change of 15% relative to current conditions) shifted tree species composition to a Pinus cembra-Larix forest with only a small fraction of Picea, and a 250 mm year−1 reduction resulted in a nearly pure Pinus cembra forest. Further reduction led to a distinct decrease in total biomass and an increasing share of mountain pine (Pinus montana), but otherwise it did not change species composition any further.

image

Figure 7. Steady-state species composition simulated at 2000 m a.s.l. under different precipitation regimes (current precipitation 974 mm year−1). The precipitation reduction was evenly distributed throughout the year (no seasonality).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

simulation of holocene forest dynamics at high elevations in the valais

The simulated species composition along the elevational transect for the mid-Holocene (Fig. 2) is in good agreement with published qualitative accounts of the vegetation composition of unmanaged forests in the dry valleys of the European Alps (cf. Ott et al. 1997), with Picea (and, locally, also Abies) dominating the lower montane zone, nearly pure Picea forests, with some Larix, at mid-elevations and Larix-Pinus cembra forests at the uppermost sites. According to our simulation (Fig. 3), total above-ground biomass does not decrease gradually towards upper treeline, but high biomass values are maintained until 100–200 m below the upper forest limit. This simulation result is difficult to test against real-world data due to the human-induced depression of treeline elevation today. However, our finding is congruent with dendroecological analyses by Hättenschwiler & Körner (1995), suggesting that high levels of tree growth are maintained in Pinus cembra with increasing elevation, followed by an abrupt decrease just below treeline.

Simulations of the magnitude of treeline fluctuations (± 100 m) and the location of treeline (2375–2600 m a.s.l.) are in excellent agreement with palaeoecological findings based on pollen and macrofossil data and the interpretation of soil biosequences (cf. Tinner & Theurillat 2003), which suggest that treeline varied between 2420 and 2530 m a.s.l. in the early and mid Holocene. Similar estimates of the uppermost position of treeline were obtained by pedological and anthracological studies in the Valais region (Carnelli et al. 2004), and treeline fluctuations of ± 100 m have been reconstructed in several previous studies elsewhere in the Alps (e.g. Patzelt & Penz 1975; Wick 1994; Burga & Perret 1998). The absolute location of the simulated treeline strongly depends on the reference temperature for the climate scenario. However, the simulated range of treeline fluctuations is based on relative temperature changes and therefore independent of this reference level.

Although comparison of the simulated treeline with the Holocene temperature reconstruction (Fig. 3) suggests that changes in above-ground biomass lagged less than 50 years behind the climatic changes, it should not be inferred that treeline location has always been nearly in equilibrium with climate. First, the model does not take into account ecological processes such as migration, soil development or erosion, which may lead to strong lags between climate and vegetation properties in real forests. Secondly, the temporal resolution of the climatic input data is relatively low. Bugmann & Pfister (2000) used a climatic data set with a very high temporal resolution (monthly data) to reconstruct treeline dynamics for a nearby region (Davos) throughout the last 500 years and found that, after clusters of exceptionally unfavourable years, treeline decreased abruptly for about 50 elevational metres, and it took at least 150–200 years for the forest to recover from such dieback events. It is known from other studies that short-term climate extremes can be more important for initiating changes in forest ecosystems than average climatic conditions (cf. Prentice 1986; Innes 1998). Our palaeotemperature reconstruction was smoothed to reduce the high between-sample variability, which is thought to be largely a consequence of noise unrelated to past climatic conditions (Heiri et al. 2003a) and the model cannot therefore reflect high-frequency climatic signals, such as those that led to the abrupt treeline decreases shown by Bugmann & Pfister (2000).

Our modelling results indicate that forests just below treeline (2200–2400 m a.s.l.) seem to be in a continuous state of change (Fig. 4). Temperature decreases lead to a fast biotic response, i.e. a dieback of Picea, after which more cold-adapted species, such as Pinus cembra, can take advantage of reduced competition. Thus, the composition of forests close to treeline (but not those at treeline) seems to react in a particularly sensitive manner to even slight changes in mean summer temperature (< 1 °C). Similar, probably climatically induced, shifts in arboreal vegetation below treeline have been obtained from pollen records in the Alps (Schmidt et al. 2002), and comparable patterns were found in a study with a spatially explicit model (TreeMig; Lischke 2005). It is noteworthy that this feature results mainly from the response of Picea, which is approaching its upper elevational limit and thus is particularly sensitive to temperature-induced growth reductions that may lead to stress-induced mortality (cf. Bigler & Bugmann 2003).

comparison of simulated and reconstructed vegetation at gouillé rion

For the first half of the Holocene, the forest model is capable of simulating the main features of both macrofossil (Fig. 5) and pollen records (Fig. 6), i.e. the importance of Larix decidua and Pinus cembra in this treeline forest and three major treeline decreases around the years 8000, 6000 and 4500 cal. BP. The fact that Larix has a considerably larger relative share in the simulation than in the pollen data is most likely due to its general under-representation (by about a factor of four, Lang 1994) in pollen assemblages. A strong reduction of biomass at 10 500 cal. BP is apparent in the model and in pollen data and a new high-resolution macrofossil series of Gouillé Rion (Tinner & Kaltenrieder 2005) shows a temporary reduction of P. cembra at 10 600–10 200 cal. BP, suggesting that the earlier macrofossil records simply had too low a temporal resolution. It is known (Bugmann 1994) that Betula hardly ever achieves dominance in the ForClim model in the early successional stage. The pollen record suggests presence of Betula, but as the macrofossil record indicates that it is not a major component of these treeline forests, we do not consider that the absence of Betula in the simulations poses a major problem for interpretation of our results.

Simulations based on the prevailing climatic conditions suggest that forest stands dominated by Larix and Pinus cembra would have continued to exist at Gouillé Rion throughout the younger part of the Holocene, i.e. after 4500 cal. BP. The nearly complete absence of evidence for local forest vegetation in both the macrofossil and the pollen record is therefore likely to be a sign of human impact.

In contrast to the simulation results, the palaeobotanical records do not suggest that Picea has been locally significant at Gouillé Rion throughout the Holocene. Picea may be a weaker competitor at low moisture availability, i.e. during the relatively dry early and mid Holocene (see ‘Sensitivity analysis’ below), and human-induced changes in forest composition may have contributed to the almost complete lack of Picea macrofossil deposition during the late Holocene. Our simulation was based on the assumption that dynamics are driven exclusively by climate and migration constraints (which are not considered) may have delayed the arrival of Picea.

The Gouillé Rion macrofossil records are characterized by continuous shifts between Larix and Pinus cembra, with Pinus being more abundant during cold-humid periods (Tinner et al. 1996) but the presence (and behaviour) of Picea abies may have obscured these patterns in the simulation.

sensitivity study of the precipitation regime

The water regime clearly influences the simulated species composition (Fig. 7), with the competitive balance between Picea and Pinus cembra depending strongly on precipitation. There is strong evidence indicating that early to mid-Holocene climatic conditions were considerably drier than current conditions (Guiot et al. 1993), leading to an overestimate of the presence of Picea in the early Holocene in our model runs.

Preliminary simulations with a qualitatively plausible, but quantitatively highly uncertain, precipitation scenario have produced promising results (unpublished data), and reliable scenarios for the Central Alpine region that are independent of the vegetation proxies are now required. Another approach would be to use the forest model in inverse mode, i.e. to determine the climate scenario that produces the best match between simulation results and the palaeo records. However, we would have to be certain that the abiotic and biotic relationships included in the model are fully realistic and robust (cf. Reynolds et al. 2001) and also to define a quantitative relationship between macrofossil and pollen data on the one hand and species-specific biomass values on the other. Furthermore, multiple climate scenarios may result in the same simulated response, so that it would not be possible to unequivocally identify the ‘true’ climate of the Holocene using this approach.

Soil moisture conditions relevant for the performance of tree species depend on both soil properties and precipitation (cf. Bugmann & Cramer 1998). It is clear that soil formation took place at least during some periods of the Holocene, leading to deeper soils with higher water holding capacities over time (Pennington 1986), but we assumed constant (i.e. current) soil properties across the entire simulated period. Lack of accurate information makes it difficult to include these processes in the simulation scenario, in spite of their potential importance (Fig. 7).

Conclusions

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Our confirmation of the conclusions of Tinner & Theurillat (2003), based on vegetation proxy data alone, that Holocene treeline shifts amounted to about ± 100 m has wide implications. It provides independent evidence confirming the intuitive interpretation of most palaeoecologists that decadal- to centennial-scale Holocene fluctuations of pollen and macrofossils at the treeline reflect tree population dynamics and ecotone shifts rather than productivity changes of the (surviving) individuals alone.

The agreement of the simulation results with macrofossil and pollen data for the first half of the Holocene, a period without strong anthropogenic influence on high-elevation forest dynamics, indicates that the succession model is capable of reconstructing long-term vegetation dynamics. The subsequent deviation between palaeo data and the climate-driven simulation results are consistent with the interpretation of Tinner & Theurillat (2003) that the lowering of the treeline since 4500 cal. BP is due to the onset of human impacts such as clearcutting, burning or grazing.

We believe that our study shows that there is a considerable potential in the combination of palaeoecological analyses with modelling approaches and that this can lead to results of relevance to both vegetation modellers and palaeobotanists. Not only can palaeo data be used to test the accuracy of vegetation models, but dynamic vegetation models can also be used to aid the interpretation of palaeoecological records and to assess whether palaeobotanical records are in agreement with independent records of climate change. This approach offers the opportunity of using palaeobotanical records to constrain past climate scenarios in regions where a direct reconstruction of climatic parameters based on pollen assemblages and macrofossil indicator species is problematic.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

We would like to acknowledge the support by and valuable discussions with Professor A. Lotter (Utrecht) that were relevant in shaping this paper. This paper is a contribution to HITE-CH, the Swiss part of the international project ‘Human Impacts on Terrestrial Ecosystems’ (HITE) of the Past Global Changes (PAGES) programme. For more information see http://www.wsl.ch/HITECH/ and http://www.liv.ac.uk/geography/hite.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  • Bigler, C. & Bugmann, H. (2003) Growth-dependent tree mortality models based on tree rings. Canadian Journal of Forest Research, 33, 210221.
  • Birks, H.H. (2001) Plant macrofossils. Tracking Environmental Change Using Lake Sediments (ed. W.M.Last), pp. 4974. Kluwer, Dordrecht.
  • Birks, H.H., Vorren, K.-D. & Birks, H.J.B. (1996) Holocene treelines, dendrochronology and palaeoclimate. Holocene Treeline Oscillations, Dendrochronolgy and Palaeoclimate (ed. B. Frenzel), pp. 118. Gustav Fischer Verlag, Stuttgart.
  • Botkin, D.B., Janak, J.F. & Wallis, J.R. (1972) Some ecological consequences of a computer model of forest growth. Journal of Ecology, 60, 849872.
  • Bugmann, H. (1994) On the ecology of mountainous forests in a changing climate: a simulation study. PhD Thesis. ETH, Zurich, Switzerland.
  • Bugmann, H. (1996) A simplified forest model to study species composition along climate gradients. Ecology, 77, 20552074.
  • Bugmann, H. (1997) An efficient method for estimating the steady-state species composition of forest gap models. Canadian Journal of Forest Research, 27, 551556.
  • Bugmann, H. (2001) A comparative analysis of forest dynamics in the Swiss Alps and the Colorado Front Range. Forest Ecology and Management, 145, 4355.
  • Bugmann, H. & Cramer, W. (1998) Improving the behaviour of forest gap models along drought gradients. Forest Ecology and Management, 103, 247263.
  • Bugmann, H., Fischlin, A. & Kienast, F. (1996) Model convergence and state variable update in forest gap models. Ecological Modelling, 89, 197208.
  • Bugmann, H. & Pfister, C. (2000) Impacts of interannual climate variability on past and future forest composition. Regional Environmental Change, 1, 112125.
  • Bugmann, H.K.M. & Solomon, A.M. (1995) The use of a European forest model in North America: a study of ecosystem response to climate gradients. Journal of Biogeography, 22, 477484.
  • Bugmann, H. & Solomon, A.M. (2000) Explaining forest composition and biomass across multiple biogeographical regions. Ecological Applications, 10, 95114.
  • Burga, C.A. & Perret, R. (1998) Vegetation und Klima der Schweiz Seit Dem Jüngeren Eiszeitalter. Ott Verlag, Thun.
  • Carcaillet, C., Talon, B. & Barbero, M. (1998) Pinus cembra et incendies pendant l’Holocène, 300 m au-dessus de la limite actuelle des arbres dans le massif de la Vanoise (Alpes du nord-ouest). Ecologie, 29, 277282.
  • Carnelli, A.L., Theurillat, J.P., Thinon, M., Vada, G. & Talon, B. (2004) Past upper most tree limit in the Central European Alps (Switzerland) based on soil and soil charcoal. Holocene, 14 (3), 393405.
  • Frei, E., Voegt, U., Flueckiger, R., Brunner, H., Schai, F. & Haeberli, R. (1980) Bodeneignungskarte der Schweiz. Eidg. Drucksachen- und Materialzentrale, Bern.
  • von Grafenstein, U., Eicher, U., Erlenkeuser, H., Ruch, P., Schwander, J. & Ammann, B. (2000) Isotope signature of the Younger Dryas and two minor oscillations at Gerzensee (Switzerland): palaeoclimatic and palaeolimnologic interpretation based on bulk and biogenic carbonates. Palaeogeography, Palaeoclimatology, Palaeoecolgy, 159, 215229.
  • Guiot, J., Harrison, S.P. & Prentice, I.C. (1993) Reconstruction of holocene precipitation patterns in Europe using pollen and lake-level data. Quaternary Research, 40, 139149.
  • Haas, J.N., Rischoz, I., Tinner, W. & Wick, L. (1998) Synchronous Holocene climatic oscillations recorded on the Swiss Plateau and at timberline in the Alps. Holocene, 8, 301309.
  • Hättenschwiler, S. & Körner, C. (1995) Responses to recent climate warming of Pinus sylvestris and Pinus cembra within their montane transition zone in the Swiss Alps. Journal of Vegetation Science, 6, 357368.
  • Heiri, O. (2001) Holocene palaeolimnology of Swiss mountain lakes reconstructed using subfossil chironomid remains: past climate and prehistoric human impact on lake ecosystems. PhD Thesis. University of Bern, Bern.
  • Heiri, O., Lotter, A.F., Hausmann, S. & Kienast, F. (2003a) A chironomid-based Holocene summer air temperature reconstruction from the Swiss Alps. Holocene, 13, 477484.
  • Heiri, O., Tinner, W. & Lotter, A.F. (2004) Evidence for cooler European summers during periods of changing melt-water flux to the North Atlantic. Proceedings of the National Academy of Sciences USA (PNAS), 101, 1528515288.
  • Heiri, O., Wick, L., van Leeuwen, J.F.N., van Der Knaap, W.O. & Lotter, A.F. (2003b) Holocene tree immigration and the chironomid fauna of a small Swiss subalpine lake (Hinterburgsee, 1515 m asl). Palaeogeography, Palaeoclimatology, Palaeoecology, 189, 3553.
  • Innes, J.L. (1998) The impact of climatic extremes on forests: an introduction. The Impacts of Climate Variability on Forests (ed. J.L.Innes), pp. 114. Springer Verlag, Berlin.
  • Jalut, G., Galop, D., Aubert, S. & Belet, J.M. (1996) Late-glacial and postglacial fluctuations of the tree limits in the Mediterranean Pyrenees: the use of pollen ratios. Holocene TreeLine Oscillations, Dendrochronolgy and Palaeoclimate (ed. B.Frenzel), pp. 189201. Gustav Fischer Verlag, Stuttgart.
  • Keller, F., Lischke, H., Mathis, T., Möhl, A., Wick, L., Ammann, B. et al. (2002) Effects of climate, fire, and humans on forest dynamics: forest simulations compared to the palaeological record. Ecological Modelling, 152, 109127.
  • Kienast, F. (1987) FORECE – A Forest Succession Model for Southern Central Europe. Report ORNL/TM-10575. Oak Ridge National Laboratory, Oak Ridge, Tennessee.
  • Körner, C. (1998) A re-assessment of high elevation treeline positions and their explanation. Oecologia, 115, 445459.
  • Körner, C. & Paulsen, J. (2004) A world-wide study of high altitude treeline temperatures. Journal of Biogeography, 31, 713732.
  • Lang, G. (1994) Quartäre Vegetationsgeschichte Europas. Methoden und Ergebnisse. Gustav Fischer Verlag, Jena.
  • Lischke, H. (2001) New developments in forest modeling: convergence between applied and theoretical approaches. Natural Resources Model, 14, 71102.
  • Lischke, H. (2005) Modelling of tree species migration in the Alps during the Holocene: what creates complexity? Ecological Complexity, 2, 159174.
  • Lischke, H., Guisan, A., Fischlin, A., Williams, J. & Bugmann, H. (1998) Vegetation responses to climate change in the Alps – modeling studies. A View from the Alps: Regional Perspectives on Climate Change (ed. C.C.Jaeger), pp. 309350. MIT Press, Boston.
  • Lischke, H., Lotter, A.F. & Fischlin, A. (2002) Untangling a Holocene pollen record with forest model simulations and independent climate data. Ecological Modelling, 150, 121.
  • Löffler, T.J. & Lischke, H. (2001) Incorporation and influence of variability in an aggregated forest model. Natural Resource Modeling, 14, 103137.
  • Lotter, A. & Kienast, F. (1992) Validation of a forest succession model by means of annually laminated sediments. Proceedings of the INQUA Workshop on Laminated Sediments, June 4–6, 1990, Lammi, Finland. Geological Survey of Finland Special Paper Series (eds M.Saarnisto & A.Kahra), pp. 2531. Geologian Tutkimuskeskus, Espoo.
  • Magny, M., Begeot, C., Guiot, J. & Peyron, O. (2003) Contrasting patterns of hydrological changes in Europe in response to Holocene climate cooling phases. Quaternary Science Reviewss, 22 (15–17), 15891596.
  • Moore, A.D. (1989) On the maximum growth equation used in forest gap simulation models. Ecological Modelling, 45, 6367.
  • Ott, E., Frehner, M., Frey, H.-U. & Lüscher, P. (1997) Gebirgsnadelwälder: praxisorientierter Leitfaden für eine standortgerechte Waldbehandlung. Paul Haupt Verlag, Bern.
  • Patzelt, G. & Penz, H. (1975) Unterinntal – Zillertal – Pinzgau – Kitzbühel. Spät- und postglaziale Landschaftsentwicklung, Exkursionsführer Tirol. Innsbrucker Geographyraphische Studien, 2, 309329.
  • Pennington, W. (1986) Lags in adjustment of vegetation to climate caused by the pace of soil development: evidence from Britain. Vegetatio, 67, 105118.
  • Prentice, I.C. (1986) Vegetation responses to past climatic variation. Vegetatio, 67, 131141.
  • Reynolds, J.F., Bugmann, H. & Pitelka, L.F. (2001) How much physiology is needed in forest gap models for simulating long-term vegetation response to global change? Challenges, limitations, and potentials. Climatic Change, 51, 541557.
  • Risch, A.C., Heiri, C. & Bugmann, H. (2005) Simulating structural forest patterns with a forest gap model: a model evaluation. Ecological Modelling, 181, 161172.
  • Schmidt, R., Koinig, K.A., Thompson, R. & Kamenik, C. (2002) A multi proxy core study of the last 7000 years of climate and alpine land-use impacts on an Austrian mountain lake (Unterer Landschitzsee, Niedere Tauern). Palaeogeography, Palaeoclimatology, Palaeoecology, 187 (1–2), 101120.
  • Schwander, J., Eicher, U. & Ammann, B. (2000) Oxygen isotopes of lake marl at Gerzensee and Leysin (Switzerland), covering the Younger Dryas and two minor oscillations, and their correlation to the GRIP ice core. Palaeogeography, Palaeoclimatology, Palaeoecology, 159, 203214.
  • Seppä, H. & Birks, H.J.B. (2001) July mean temperature and annual precipitation trends during the Holocene in the Fennoscandian tree-line area: pollen-based climate reconstruction. Holocene, 11, 527539.
  • Shao, G., Bugmann, H. & Yan, X. (2001) A comparative analysis of the structure and behavior of three gap models at sites in Northeastern China. Climatic Change, 51, 389413.
  • Shugart, H.H. (1984) A Theory of Forest Dynamics. The Ecological Implications of Forest Succession Models. Springer, Berlin.
  • Shugart, H.H. (1989) The role of ecological models in long-term ecological studies. Long-Term Studies in Ecology (ed. G.E.Likens), pp. 90109. Springer-Verlag, New York.
  • Solomon, A.M. & Bartlein, P.J. (1992) Past and future climate change: responses by mixed deciduous-coniferous forest ecosystems in northern Michigan. Canadian Journal of Forest Research, 22, 17271738.
  • Solomon, A.M. & Shugart, H.H. (1984) Integrating forest-stand simulations with paleoecological records to examine long-term forest dynamics. State and Change of Forest Ecosystems – Indicators in Current Research (ed. G.I.Ågren), pp. 333356. Department of Ecology and Environmental Research Report. Swedish University of Agricultural Sciences, Uppsala.
  • Solomon, A.M., West, D.C. & Solomon, J.A. (1981) Simulating the role of climate change and species immigration in forest succession. Forest Succession: Concepts and Application (eds D.C.West, H.H.Shugart & D.B.Botkin), pp. 154177. Springer, New York.
  • Thinon, M. & Talon, B. (1998) Ampleur de l’anthropisation des étages supérieurs dans les Alpes du sud: données pédoanthracologiques. Ecologie, 29, 323328.
  • Tinner, W. & Ammann, B. (2001) Timberline paleoecology in the Alps. PAGES News, 9, 911.
  • Tinner, W., Ammann, B. & Germann, P. (1996) Treeline fluctuations recorded for 12 500 Years by soil profiles, pollen, and plant macrofossils in the Central Swiss Alps. Arctic and Alpine Research, 28, 131147.
  • Tinner, W. & Kaltenrieder, P. (2005) Rapid responses of high-mountain vegetation to early-Holocene environmental changes in the Swiss Alps. Journal of Ecology, 93, 936947.
  • Tinner, W. & Theurillat, J.-P. (2003) Uppermost limit, extent, and fluctuations of the timberline and treeline ecocline in the Swiss Central Alps during the past 11 500 Years. Arctic, Antarctic, and Alpine Research, 35, 158169.
  • Tobolski, K. & Ammann, B. (2000) Macrofossils as records of plant responses to rapid Late Glacial climatic changes at three sites in the Swiss Alps. Palaeogeography, Palaeoclimatology, Palaeoecology, 159 (3–4), 251259.
  • Vorren, K.-D. & Stavseth, A. (1996) Late Holocene vegetation and climate changes interpreted from pollen influx and plant macrofossils from sediments of an alpine lake in north Norway. Holocene TreeLine Oscillations, Dendrochronolgy and Palaeoclimate (ed. B.Frenzel), pp. 257269. Gustav Fischer Verlag, Stuttgart.
  • Welten, M. (1982) Vegetationsgeschichtliche Untersuchungen in den westlichen Schweizer Alpen: Bern–Wallis. Denkschriften Schweizerische Naturforschende Gesellschaft, 95, 1104.
  • Wick, L. (1994) Early-Holocene reforestation and vegetation change at a lake near the alpine forest limit: Lago Basso (2250 m a.s.l.), northern Italy. Festschrift Gerhard Lang. Dissertationes Botanicae, 234, 555563.
  • Wick, L. & Tinner, W. (1997) Vegetation changes and timberline fluctuations in the Central Alps as indicator of Holocene climatic oscillations. Arctic and Alpine Research, 29, 445458.