Réf. Schumacher & Bugmann 2006 - A

Référence bibliographique complète

SCHUMACHER, S. and BUGMANN, H. 2006. The relative importance of climatic effects, wildfires and management for future forest landscape dynamics in the Swiss Alps. Global Change Biology, 12: 1435-1450.

Abstract: Forest landscape dynamics result from the complex interaction of driving forces and ecological processes operating on various scales. Projected climate change for the 21st century will alter climate-sensitive processes, causing shifts in species composition and also bringing about changes in disturbance regimes, particularly regarding wildfires. Previous studies of the impact of climate change on forests have focused mainly on the direct effects of climate. In the present study, the authors assessed the interactions among forest dynamics, climate change and large-scale disturbances such as fire, wind and forest management. They used the LANDCLIM model to investigate the influence, interactions and the relative importance of these different drivers of landscape dynamics in two case study areas of the European Alps. The simulations revealed that projected future climate change would cause extensive forest cover changes, beginning in the coming decades. Fire is likely to become almost as important for shaping the landscape as the direct effects of climate change, even in areas where major wildfires do not occur under current climatic conditions. The effects of variable wind disturbances and harvesting regimes, however, are less likely to have a considerable impact on forest development compared with the direct effects of climate change coupled with the indirect effects of increased fire activity. The authors conclude that the joint direct and indirect effects of climate change are likely to have major consequences for mountain forests in the European Alps, including their ability to provide protection against natural hazards.

Climate change, Disturbances, Fire, Landscape model, Mountain forest dynamics

Organismes / Contact

Swiss Federal Institute of Technology Zurich, Department Environmental Sciences, Forest Ecology, CH-8092 Zurich, Switzerland - harald.bugmann@env.ethz.ch

(1) - Paramètre(s) atmosphérique(s) modifié(s)
(2) - Elément(s) du milieu impacté(s)
(3) - Type(s) d'aléa impacté(s)
(3) - Sous-type(s) d'aléa
Temperature, Precipitation Forest (tree species, biomass)    

Pays / Zone
Massif / Secteur
Site(s) d'étude
Période(s) d'observation
Switzerland Swiss Alps Dischma valley (Grisons)
Gantertal valley (Valais)
  Dischma: 1550–2550m a.s.l.
Gantertal: 1100–2600m a.s.l.
2000–2100 and 2100-2300

(1) - Modifications des paramètres atmosphériques

A considerable magnitude of anthropogenic climate change is projected for the end of the 21st century (e.g. Schär et al., 2004).

Although the climate change scenario used in this study is based on a sophisticated modeling approach, it is still afflicted with uncertainties (Vidale et al., 2003; Schär et al., 2004); the simple downscaling method used in this study has added to this uncertainty. Although the climatic changes used here are rather drastic, they are certainly not extreme as they are well within the range of other state-of-the art scenarios.


Informations complémentaires (données utilisées, méthode, scénarios, etc.)

Climate data and climate scenario:
In this study, simulations were performed using the observed climate (monthly data), as well as a future climatic scenario:

- data from the climate station Davos–Platz (1960–2000, elevation 1560ma.s.l.), the station Visp (1980–2000, 640ma.s.l.) and two adjacent climate stations (Grächen, 1960–2000, elevation 1617ma.s.l. and Ulrichen, 1981–2000, 1345ma.s.l.) used to derive average altitudinal lapse rates for temperature. Gantertal. The amount of local precipitation varies even more than temperature, particularly in the Valais. Therefore, the annual precipitation lapse rate was estimated for the Gantertal from high-resolution precipitation maps (Baumgartner et al., 1983; Schwarb et al., 2001) rather than from distant climate stations. The annual value was then applied for all months.

- recent climate scenario by Schär et al. (2004) based on the A2 transient greenhouse-gas scenario (IPCC, 2000) and the regional climate model CHRM (Vidale et al., 2003) driven with output from the HadAM3H general circulation model (Pope et al., 2000). The scenario data represent the conditions of the years 2071–2100 on a 56km horizontal resolution. These scenario data are complemented by a ‘current climate’ control dataset covering the period 1961–2000. Downscaling was applied to the results of the regional climate model to overcome the scale gap in resolution between the output of the regional climate model and the LANDCLIM model. A bias correction of precipitation was applied to each of the study areas. The calculation of the monthly bias was based on the ratio of observed monthly precipitation and the downscaled monthly precipitation values of the control simulation.

A comparison of the monthly mean values of the downscaled control simulation (control dataset 1961– 2000) against observed climate data showed a fairly similar pattern for temperature; deviations were always smaller than 1.5 1C (detailed data not shown). While the annual sum of precipitation also displayed a similar pattern (deviation o1%), summer precipitation was underestimated (cf. Frei et al., 2003). Therefore, a bias correction of precipitation was applied to each of the study areas. The calculation of the monthly bias was based on the ratio of observed monthly precipitation and the downscaled monthly precipitation values of the control simulation..

(2) - Effets du changement climatique sur le milieu naturel

Future forest dynamics under varying management regimes:
The simulations of future development of the current forest area in the Dischma valley under a future climate change scenario resulted in a considerable shift of vegetation composition for the year 2100 and beyond. The comparison with simulations under current climatic conditions and under the various harvesting regimes showed that harvest activities, depending on their intensity, can have a considerable impact on vegetation development. However, the impact of climatic change is much stronger.

Climate change between 2000 and 2100 AD was imposed as a linear change; therefore the response of vegetation change is also gradual over the first decades. Compared with the simulations under current climatic conditions, biomass increased during the first decades under the climate change scenario when no (H0) or little (H1) harvesting took place. Under the more intense harvesting regimes (H2 and H3), biomass immediately started to decrease, as the climatically induced increase of biomass production in the first decades after the onset of climatic change was outweighed by harvesting.

The phase of increased growth between 2000 and 2050 in the H0 and H1 simulation scenarios is followed by a phase of strong biomass reduction. It is caused mainly by a drought-induced dieback of today’s forest stands, accompanied by the gradual invasion of new tree species. The simulation results suggest that during this process an upward shift of vegetation will also occur, given that alpine pastures will no longer be used with the same intensity as today, where spontaneous reforestation is almost impossible as long as domestic grazing continues. According to the model, by 2050 the current vegetation cover has already shifted about 100m upwards compared to the situation in 2000 (assuming that the tree line corresponds to a biomass of about 25 t ha–1), while the area above is mainly covered by P. mugo. Compared with other tree line shift studies (e.g. Dullinger et al., 2004), the simulated extent of the P. mugo invasion between 2000 and 2050 is extraordinary (factors other than climate that could restrict dispersal, such as competitive inhibition of recruitment in dense grassland layers, were not taken into account in the simulations. Hence, these model results have to be interpreted with caution.).

During the following decades, the upward shift continued and biomass increased at high elevations. In particular, the seral species L. decidua accumulated a considerable amount of biomass just below and above the current tree line. Thus, the overall decrease in biomass within the current forested area is compensated, to some extent, by the increasing biomass slightly below and above the current tree line.

Effects of future climatic change on potential natural vegetation and the fire regime [see results below]:
In both study areas, the subalpine vegetation zone consisting of P. abies, L. decidua and P. cembra experiences a considerable shift upwards in elevation. The main driver behind this upward shift is temperature increase. However, drought also seems to shape this landscape. This is apparent on the southern slope of the Dischma, where simulated vegetation is dominated by P. cembra at higher elevations, whereas P. abies is seldom found. In contrast, on the northern (and less drought-prone) slope, P. abies is co-dominant, and total biomass is considerably higher than on the southern slope. In the Gantertal, there were also differences in the high elevation vegetation zone of the northern and the southern slopes. The lower limit of the P. abiesL. deciduaP. cembra vegetation is 200m higher on southern than on northern slopes. Simulations with the various wind disturbance regimes did not result in considerably different findings.

In both study areas, lower elevations are mainly dominated by P. silvestris under future climatic conditions. In addition, some deciduous tree species occur there, mainly on the northern slopes in the Gantertal. However, the lower slopes in the Dischma are also codominated by Quercus sp., F. silvatica, and some A. pseudoplatanus. Various factors, such as species composition and biomass values differ considerably at lower elevations between (1) the study areas, (2) the slopes, and (3) simulations with different fire parameter settings. Thus, biomass and species composition at lower elevations are mainly determined by fire occurrence and site-specific differences in the fire regimes.

As expected, simulations using the fire parameter setting F2.0 resulted in more fires than the conservative setting F2.5. Based on previous model tests (Schumacher, 2004), it can be concluded that the most likely case lies somewhere between these two results, (i.e. that the F2.0 simulations are too extreme), whereas the F2.5 simulations are probably overly conservative. Therefore, the results are discussed on the assumption that these two settings define the plausibility range.

As a consequence of the lower summer temperatures and higher summer precipitation, fewer fires were simulated in the Dischma than in the Gantertal study area. Based on simulation results, there would be a major fire (45 ha) in the Dischma every 50–150 years, whereas in the Gantertal this would be the case every 20–50 years. In both study areas, fires occurred mainly on the lower slopes, whereas the frequency and influence of forest fires decreased rapidly at about 2000ma.s.l. The simulated fires spread to higher elevations on southern slopes than on northern slopes, which is quite realistic. By comparing simulated biomass and fire rotation over an elevational gradient, it became apparent that low fire rotation intervals result in a considerable decrease of biomass. The shortest fire rotations simulated were those in the Gantertal at low elevations, where fire caused a parkland-like landscape.

When interpreting the simulation results, it is important to consider the uncertainties inherent in this simulation study.

The following conclusions can be drawn from this study:

Firstly, projected climatic change for the future will have a substantial impact on forest vegetation, particularly on biomass distribution, as well as on species composition along altitudinal gradients. While forest management will still have a visible effect on the landscape in the two case study areas, the vegetation shifts caused by climate change will have a much bigger influence on biomass and species distribution.

Secondly, the projected future climatic changes, which are within the range of many other state-of-the-art climate scenarios for the 21st century, resulted in an increase of summer drought. This in turn increased the probability of fire occurrence. These results suggest that it is important to include fire disturbances in future studies of landscape dynamics in the European Alps and probably also elsewhere in the temperate zone. Studies that focus on the direct effects of climate alone (e.g. Bugmann, 1997) may provide an incomplete picture of future mountain forest trajectories. According to the present simulation results, fire is likely to become an important agent in shaping the landscape in the near future. This is even true in areas where major wildfires do not occur under current climatic conditions, such as in the Dischma valley. Thus, as an indirect effect of climate change, fire is likely to have a great impact on forest cover in addition to direct climatic effects.

Thirdly, the effects of changes in the wind disturbance regime or harvesting activities are likely to be overridden by (1) the direct effects of future climate change on altered species composition and biomass, and (2) the indirect effects of an increased fire activity.

Finally, the results of this study suggest considerable forest cover changes, starting over the coming decades.


The model used does not allow for the simulation of the future interacting effects of elevated atmospheric CO2 levels. Increasing CO2 might partially reduce water use and limit the occurrence of drought, which would in turn reduce fire frequency. Alternatively, CO2-induced increases in productivity might increase fuel loads and exacerbate future fire effects.

Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)

In this paper, the authors explore the direct, as well as the indirect (wildfires) impacts of anticipated future climatic changes on forest vegetation development at the landscape scale, using a model that is capable of predicting the future wildfire regime as a function of climatic conditions and vegetation properties, LANDCLIM (cf. Schumacher, 2004). They evaluate the relative importance of changes in the wildfire regime vs. changes in the wind disturbance regime, as well as different forest management intensities, focusing on two case study areas in the Swiss Alps. They used the model to investigate several questions related to longterm forest dynamics in the Swiss Alps. Specifically, this paper aims to investigate the influence of (1) different disturbance regimes on forest dynamics, (2) a changing climate on forest dynamics, and (3) a changing climate on the fire regime.

The long-term implications of climate and large-scale disturbances for forest vegetation dynamics in the Swiss Alps were analyzed using the LANDCLIM model (Schumacher, 2004; Schumacher et al., 2004). Future forest dynamics were simulated by applying a range of harvesting and wind disturbance scenarios, as well as a future climate change scenario. In addition, based on the simulated climatic effects and the resulting vegetation structure, fire disturbances were simulated as an emergent property of the landscape, which in turn influenced vegetation dynamics.

The LANDCLIM model:
LANDCLIM is a spatially explicit, stochastic model designed to study forest dynamics determined by a set of driving forces including large-scale natural disturbances, land-use, climatic parameters, soil properties and topography. The model operates over long periods of time (hundreds to thousands of years) and large spatial extents (>100 ha) at a relatively fine grain (grid cells of 25m x 25 m).

The modifications include (a) quantitative descriptions of forest structure by including tree cohorts (i.e. groups of trees of the same species and age) that are characterized by aboveground biomass of an individual tree and the number of trees in a cohort; (b) the explicit incorporation of the influence of competition as well as climatic and edaphic parameters on tree population dynamics; and (c) the modeling of fire regimes as an emergent ecosystem property based on climatic parameters and fuel load. The basic structure of LANDCLIM consists of a local model that simulates forest succession for each cell in the landscape on a yearly time step and a landscape model that contains processes operating over several cells, which are simulated in 10-year time steps. The landscape-scale processes included are fire, wind, harvesting and seed dispersal.

Vegetation and climate:
Tree growth is specified as a maximum potential, which is reduced to reflect suboptimal environmental conditions: low temperature, drought, or low light. Low temperature and drought conditions are determined based on monthly temperature and precipitation data. The climatic parameters for each cell are adjusted for elevation using lapse rates. To represent average annual dryness conditions, a seasonal drought index is calculated (Bugmann & Solomon, 2000). This drought index is sensitive to both temperature and precipitation as well as soil type and topographic position of each individual cell. In this way, each cell has unique temperature and drought conditions. Each tree species’ abundance is also dependent on its ability to compete for light. Shading conditions (Monsi & Saeki, 1953) are simulated based on the crown characteristics of the trees (Bugmann, 1994) growing in the cell. The direct effects of enhanced atmospheric CO2 concentrations are not taken into account in the model, because the growth equation is not formulated on a physiological basis.

Tree mortality. In LANDCLIM, a strong reduction of the maximum growth potential leads to an increased mortality probability (Schumacher et al., 2004). This growth-dependent mortality factor is based on the assumption that trees growing slowly due to adverse environmental conditions are likely to be subject to stress-induced mortality (e.g. Shugart, 1984). Other factors that influence tree mortality in the model include (1) a density-dependent mortality, which occurs when total stand biomass exceeds a maximum stand biomass; (2) an age-dependent mortality that is modeled as a constant annual probability of death throughout the lifetime of the tree (Botkin, 1993). In addition to these causes of mortality, fire, wind, and harvesting can cause mortality, as described further below.

Tree establishment. The death of trees allows for more light to reach the forest floor and thus can increase the probability of tree establishment. Light availability, temperature, soil moisture, browsing pressure (Bugmann, 1994) and other user-defined factors are considered to determine the tree establishment potential. Establishment conditions are checked annually; however, the actual establishment occurs once every decade (i.e. establishment of 10-year age cohorts, Schumacher et al., 2004). Regardless of these abiotic and biotic conditions for establishment in every cell, establishment is only possible if propagules are available in the cell; the probability of seed availability depends on the distance to the nearest mature tree of the respective species (He & Mladenoff, 1999a).

[See below Large-scale disturbances...]

The model requires parameter sets describing autecological properties of tree species, physical site conditions, climate and disturbance regimes. In this study, we applied the default parameter sets for species life history parameters and general site parameters as defined and described by Schumacher (2004). Elevation, slope and aspect for every grid cell were derived from a digital elevation model with a resolution of 25m (DEM25, © 2002 Swiss Federal Office of Topography). Soil information for the Dischma was taken from the soil maps by Krause (1986), and for the Gantertal from the Bodeneignungskarte der Schweiz (1980). The map of initial forest stands and stand descriptions were derived from data in Hefti et al. (1986) for the Dischma valley. In the case of the Gantertal, the model was always run starting from an empty landscape .

[See above Climate data and climate scenario...]

Simulation experiments:
Firstly, to explore the influence of various future harvest regimes within the current forest area, simulations starting from the current forest cover under current climatic conditions were started for the Dischma in the year 2000 and were run until 2300 AD. The properties of the stochastic processes embedded in the LANDCLIM model were evaluated by considering the mean values from 25 independent simulations. This number of replicate simulations was found to provide highly robust estimates of landscape properties (cf. Schumacher, 2004), as evidenced by the low standard deviations of simulated biomass values.

Secondly, to simulate possible future development under a changing climate within the current forest area in the Dischma valley, simulations were started from the current forest cover in the year 2000, but with the future climate scenario. For the years 2000–2100 AD, a change of climatic parameters was applied by linearly interpolating the mean values between the current and future climate data sets (Tables 2 and 3). Simulations between 2100 and 2300 AD were then performed based on the assumption that the climate would remain constant after 2100 AD (for the rationale, cf. Bugmann, 1997). The conservative fire parameter set (F2.5) was used for these simulations. Simulations were run for the range of wind disturbance and harvest scenarios described in the study. For each of these combinations, the mean values of 25 simulations were evaluated. As forest cover boundaries are unlikely to remain static over the next couple of decades to centuries (cf. Bebi & Baur, 2002), in a second step we also evaluated the effect of forest establishment in areas above the current tree line.

Finally, to better understand the long-term effect of the projected climate change scenario on forest vegetation, the fire regime, and the feedback of an altered fire regime on vegetation, simulations of the potential natural vegetation (i.e. no forest management and no land cover constraints) under the current climate and the future climate change scenario were carried out in both the Dischma and the Gantertal. Both fire parameter sets and the threewind disturbance regimes were used. These simulations started from an empty landscape, and continued for 1000 years to reach a steady-state forest composition. The mean values over the following 10 000 simulation years under a constant (current or future) climate were evaluated to assess the new equilibrium between climate, wildfire, and vegetation properties..

(3) - Effets du changement climatique sur l'aléa

Effects of future climatic change on potential natural vegetation and the fire regime:
Using the future climate scenario, some fires occurred in the Dischma, mainly at lower elevations and on the southerly slopes. Under the F2.5 fire setting, fires occurred only rarely; a fire of 40.5 ha occurred around every 150 years at the lowest elevations, and every 300 years at about 1800ma.s.l. Fire return intervals rose quickly with increasing elevation. Accordingly, the calculated fire rotations (the time required to burn an area of the size of the entire area for each elevation band) are high; the fire rotation is less than 1000 years only in the lowest elevation band. Using the F2.0 setting, however, more frequent fires were simulated; one to two larger fires (40.5 ha) occurred per century at lower elevations. Fire rotation is about 100 years at the lowest elevations and increases gradually with elevation. At about 1900ma.s.l. on the northern and 2000ma.s.l. on the southern slope, respectively, fire rotation is about 300 years. Above 2050 and 2150ma.s.l., respectively, the fire rotation quickly increases to values larger than 1000 years.

In the Gantertal, fires also occurred mainly at lower elevations, but they extended higher up on the southern slopes. Under the F2.5 fire settings, simulated fire rotations were less than 100 years up to 1300m on the northern and 1500m on the southern slopes, respectively; fire rotations of under 300 years were simulated up to 1500m on the northern and up to 1800m on the northern slopes, respectively. Under the F2.0 setting, the number of fires that occurred within a decade more than doubled at lower elevations compared to the F2.5 setting. Accordingly, fire rotations are low – less than 30 years at the lowest elevations – under the F2.0 setting. On the northern slope, fire rotation increases gradually with elevation; at 1800ma.s.l., fire rotations are still less than 300 years, but they increase quickly above 1800 m. On the southern slopes, fire rotation increases more slowly than on the northern slopes; at 1800ma.s.l., fire rotation is still under 50 years, but above 2000ma.s.l., where fire rotation is still >200 years, it increases rapidly.

Comparing the simulated fires in the two study sites and using both fire settings shows that the simulations using the F2.0 setting not only produced more fires than the simulations under F2.5 setting, but also much larger fires (results not shown). In addition, simulated fires in the Dischma were generally smaller than in the Gantertal; a major fire larger than 5 ha occurred in the Dischma study area one to two times per century, whereas in the Gantertal this occurred two to four times per century. However, the majority of fires simulated in both study sites and under both fire settings were minor fires, smaller than 0.5 ha. On average, over all scenarios, about 70% of the simulated fires were such minor fires.

Changes in the wind disturbance regime did not result in considerably different biomass or species distributions. Generally, biomass decreased somewhat with increased wind disturbance frequency. Within the scenarios of climate change, the increased frequency of wind disturbances reduced total biomass by 6–18% (comparison of W600 scenario with W200 scenario), whereas climate change alone (including the change of wildfire regime) led to considerably more marked changes of biomass relative to current conditions in the vast majority of the simulated cases.


Given the anticipated change of climatic conditions, the frequency, size and intensity of fires are expected to change (e.g. Beniston, 2000; Conedera, 2003).

Paramètre de l'aléa
Sensibilité des paramètres de l'aléa à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)

Under the current climatic conditions, the wildfire regime in the Swiss Alps is characterized by low fire frequencies, although the more continental areas (e.g. the inner alpine dry valleys such as the Valais or the Engadine) exhibit significant exposure to fire under current climatic and land use conditions.

The basic structure of LANDCLIM consists of a local model that simulates forest succession for each cell in the landscape on a yearly time step and a landscape model that contains processes operating over several cells, which are simulated in 10-year time steps. The landscape-scale processes included are fire, wind, harvesting and seed dispersal.

Large-scale disturbances:
The structure of the LANDCLIM fire model is based on the assumption that fire occurrence is primarily responsive to climatic conditions, and is modulated by fuel availability (e.g. Bessie & Johnson, 1995). Climatic conditions are captured by an annual drought index, which mimics average fuel dryness for any given cell. The drought index is calculated for each simulated fire ignition event from the annual sums of potential and actual evapotranspiration (Bugmann & Cramer, 1998). Forest fire occurrence is simulated with a 10-year time step. In each simulation step, several fires (ignitions) may occur. For each of the simulated fires, a year is selected randomly from the 10 years of the past decade, and the fire is then simulated using the weather data of that year. This procedure is repeated for each fire occurring within that decade (cf. Schumacher, 2004)..

Fire ignition and spread: A fire disturbance can only start with the presence of a fire ignition source. A number of cells are randomly selected each decade to mimic lightning-caused fire ignitions (He & Mladenoff, 1999b); the number of lightning strikes is assumed not to be limiting in LANDCLIM. However, a fire is only simulated if fire spread to a neighboring cell is possible. Fire spread probability from each ignited cell to any of its eight neighbors is simulated as a function of the annual drought index (Schumacher, 2004, p. 39). A slope adjustment factor in each cell is used to modify fire spread probability, so as to take into account that fires are more likely to burn up-slope than down-slope (e.g. Rothermel, 1972). Thus, the number and size of the simulated fires depend mainly on climate and topography, and they vary stochastically with inter annual climate variability.

Fire effects: Fire intensity depends on the amount of fuel that is consumed in a fire. Fuel input in each cell results from dead trees, slash and annual litterfall. Fuel decomposition is simulated annually according to climatedependent rates (Meentemeyer, 1978; Mackensen et al., 2003). Typically, only a fraction of the total fuel load will be able to burn during a fire (e.g. Brown et al., 1991). This fraction is estimated based on the annual drought index, assuming that the flammability of fuel components is determined mainly by their moisture content. The amount of fuel consumed is directly linked to fire intensity (Brown, 2000) and, thus, to flame height (Van Wagner, 1973). Based on flame height, in combination with tree size, the percentage of crown scorch in a fire is estimated. The percentage of crown kill, together with bark thickness – which depends on tree size and a species-specific allometric coefficient – determines the probability of mortality of a single tree following a fire (cf. Ryan & Reinhardt, 1988). The modeling of the fire effects has been described in detail by Schumacher et al. (2006).

Wind is simulated in LANDCLIM similarly as in the original LANDIS model (He & Mladenoff, 1999b). Windthrow is simulated stochastically using userdefined mean wind return intervals and windthrow sizes. However, unlike in the original implementation, the susceptibility to windthrow is approximated based on tree size. Trees in different tree size classes have a different susceptibility to be killed by windthrow (e.g. Canham et al., 2001).

Fire parameter sets:
The default fire parameters were used as described by Schumacher (2004) with the following exception: Previous model tests and sensitivity studies revealed that although the fire submodel was able to provide realistic results, some parameters are associated with considerable uncertainty. In particular, the shape of the fire spread probability curve proved difficult to estimate. To take this uncertainty into account, several of the simulations presented in this paper were performed with two fire spread curve shapes by varying the coefficient firePro (Schumacher, 2004), so as to represent (a) a conservative estimate of fire occurrence (fire- Prob = 2.5; referred to as fire parameter set F2.5); and (b) an extreme estimate of fire occurrence [i.e. the upper limit of the plausibility range (fireProb = 2.0; fire parameter set F2.0)].

Wind parameters and scenarios:
The characteristics of the current wind disturbance regime were estimated based on published investigations of storm history in Switzerland (WSL & BUWAL, 2001; Schumacher, 2004). The mean return interval was estimated as 600 years, and the mean disturbance size as 2.5 ha. This default parameter set is called ‘W600’. An increase in storm frequency in Switzerland in the future is possible, but it is rather uncertain if and to what extent this will happen (IPCC, 2001b; Wernli et al., 2003). To investigate the effects of a possible future increase in storm frequency, two alternative storm regimes were defined: a return interval of 400 years (W400) and a return interval of 200 years (W200). We assumed that even though the frequency of windthrow events would increase, their size distribution would remain the same as under current conditions..

(4) - Remarques générales

(5) - Syntèses et préconisations

The results of this study suggest considerable forest cover changes, starting over the coming decades. It is important to understand such changes when planning forest management, considering wood production, sustaining certain forest structures for protection against natural hazards (e.g. avalanches, landslides, etc.), or maintaining landscape characteristics for aesthetic purposes. Such changes are also important for other fields of research (e.g. for conservation efforts that rely on understanding the development of species habitat suitability), or for carbon cycling. Given the importance of such changes, future research into this subject is not only justified, but necessary. Based on this study, the authors conclude that the future ability of many mountain landscapes to provide protection against natural hazards will be dependent on both the direct, as well as the indirect effects of climatic change via changes in the wildfire regime.

Références citées :

Baumgartner A, Reichel E,Weber G (1983) Der Wasserhaushalt der Alpen: Niederschlag, Verdunstung, Abfluss und Gletscherspende im Gesamtgebiet der Alpen im Jahresdurchschnitt für die Normalperiode 1931–1960. R. Oldenburg Verlag GmbH, München.

Beniston M (2000) Environmental Change in Mountains and Uplands. Key Issues in Environmental Change. Arnold, London.

Bessie WC, Johnson EA (1995) The relative importance of fuels and weather on fire behavior in subalpine forests. Ecology, 76, 747–762.

Bodeneignungskarte der Schweiz Massstab 1:200000. (1980) Eidg. Justiz- and Polizeidepartement, Bundesamt für Raumplanung. Bern, 145 S.

Botkin DB (1993) Forest Dynamics: An Ecological Model. Oxford University Press, Oxford.

Brown JK (2000) Ecological principles, shifting fire regimes and management considerations. In: Wildland Fire in Ecosystems: Effects of Fire on Flora. Gen. Tech. Rep. RMRS-GTR-42-Vol. 2 (eds Brown JK, Smith JK), pp. 185–203. US Department of Agriculture, Forest Service, Rocky Mountain Research Station, Ogden, UT.

Brown JK, Reinhardt ED, Fischer WC (1991) Predicting duff and woody fuel consumption in Northern Idaho prescribed fires. Forest Science, 37, 1550–1566.

Bugmann H (1994) On the ecology of mountainous forests in a changing climate: a simulation study. PhD Thesis, No. 10638, Swiss Federal Institute of Technology Zurich, Switzerland.

Bugmann H (1997) Sensitivity of forests in the European Alps to future climatic change. Climate Research, 8, 35–44.

Bugmann H, Cramer W (1998) Improving the behaviour of forest gap models along drought gradients. Forest Ecology and Management, 103, 247–263.

Bugmann HKM, Solomon AM (2000) Explaining forest composition and biomass across multiple biogeographical regions. Ecological Applications, 10, 95–114.

Canham CD, Papaik MJ, Le F (2001) Interspecific variation in susceptibility to windthrow as a function of tree size and storm severity for northern temperate tree species. Canadian Journal of Forest Research, 31, 1–10.

Conedera M (2003) Wäldbrände. In: Extremereignisse und Klimaänderung (ed. Hohmann, R), pp. 57–60. Organe consultatif sur les changements climatiques (OcCC), Bern.

Dullinger S, Dirnböck T, Grabherr G (2004) Modelling climate change-driven treeline shifts: relative effects of temperature increase, dispersal and invasibility. Journal of Ecology, 92, 241–252.

Frei C, Christensen JH, Déqué M et al. (2003) Daily precipitation statistics in regional climate models: evaluation and intercomparison for the European Alps. Journal of Geophysical Research, 108, 4124, doi:10.1029/2002JD002287.

He HS, Mladenoff DJ (1999a) The effects of seed dispersal on the simulation of long-term forest landscape change. Ecosystems, 2, 308–319.

He HS, Mladenoff DJ (1999b) Spatially explicit and stochastic simulation of forest-landscape fire disturbance and succession. Ecology, 80, 81–99.

Hefti R, Schmid-Haas P, Bühler U (1986) Zustand und Gefährdung der Davoser Waldungen. MAB-Schlussberichte23. Bundesamt fûr Umweltschutz, Bern.

IPCC (2000) Emission scenarios. In: Special Report of the Intergovernmental Panel on Climate Change (eds Nakicenovic N, Swart R). Cambridge University Press, Cambridge, NY.

IPCC (2001a) Climate change 2001: synthesis report. In: A contribution of Working Groups I, II and III to the Third Assessment Report of the Intergovernmental Panel on Climate Change (eds Watson RT, Core Writing Team). Cambridge University Press, Cambridge, NY.

IPCC (2001b) Climate change 2001: the scientific basis. In: Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change (eds Houghton JT, Ding Y, Griggs DJ et al.). Cambridge University Press, Cambridge, NY.

Krause M (1986) Die Böden von Davos-Ertragspotential, Belastbarkeit und Gefährdung durch Nutzungsänderungen. Schlussberichte zum Schweizerischen MAB-Programm18, 147 pp.

Mackensen J, Bauhus J, Webber E (2003) Decomposition rates of coarse woody debris – a review wit particular emphasis on Australian tree species. Australian Journal of Botany, 51, 27–37.

Meentemeyer V (1978) Macroclimate and lignin control of litter decomposition rates. Ecology, 59, 465–472.

Monsi M, Saeki T (1953) Über den Lichtfaktor in den Pflanzengesellschaften und seine Bedeutung für die Stoffproduktion. Japanese Journal of Botany, 14, 22–52.

Pope DV, Gallani M, Rowntree R et al. (2000) The impact of new physical parameterizations in the Hadley centre climate model HadAM3. Climate Dynamics, 16, 123–146.

Rothermel RC (1972) A mathematical model for predicting fire spread in wild land fuels. USDA Forest Service, Intermountain Forest and Range Experiment Station Research Paper, INT-115, Ogden, UT.

Ryan KC, Reinhardt ED (1988) Predicting postfire mortality of seven western conifers. Canadian Journal of Forest Research, 18, 1291–1297.

Schär C, Vidale PL, Lüthi D et al. (2004) The role of increasing temperature variability in European summer heatwaves. Nature, 427, 332–336. [Fiche biblio]

Schumacher S (2004) The role of large-scale disturbances and climate for the dynamics of forested landscapes in the European Alps. Ph.D. Thesis, No. 15573, Swiss Federal Institute of Technology Zurich, Switzerland. Available at: http://e-collection.ethbib. ethz.ch/show?type=diss&nr=15573

Schumacher S, Bugmann H, Mladenoff DJ (2004) Improving the formulation of tree growth and succession in a spatially explicit landscape model. Ecological Modelling, 180, 175–194.

Schumacher S, Sibold J, Reineking B et al. (2006) Modeling the impact of climate and vegetation on fire regimes in mountain landscapes. Landscape Ecology, 21, 539–554.

Schwarb M, Frei C, Daly C et al. (2001) Mean annual precipitation throughout the European Alps 1971–1990. Swiss National Hydrological and Geological Survey (ed.), Hydrological Atlas of Switzerland, Plate 2.6. Swiss Federal Office of Topography, Bern, Switzerland.

Shugart HH (1984) A Theory of Forest Dynamics. The Ecological Implications of Forest Succession Models. Springer-Verlag, New York.

Van Wagner CE (1973) Height of crown scorch in forest fires. Canadian Journal of Forest Research, 3, 373–378.

Vidale PL, Lüthi D, Frei C et al. (2003) Predictability and uncertainty in a regional cliamte model. Journal of Geophysical Research, 108 (D.18), paper 4856, doi: 10.1029/2002JD002810.

Walder U (1983) Ausaperung und Vegetationsverteilung im Dischmatal. Mitteilungen der Eidg. Forschungsanstalt für Wald, Schnee und Landschaft, 59, 81–206.

Walther G-R, Grundmann A (2001) Trends of vegetation change in colline and submontane climax forests in Switzerland. Bulletin of the Geobotanical Institute ETH, 67, 3–12.

Wernli H, Bader S, Hächler P (2003) Winterstürme. In: Extremereignisse und Klimaänderungen (ed. Hohmann, R), pp. 79–82. Organe consultatif sur les changements climatiques (OcCC), Bern.

WSL and BUWAL (ed.) (2001) Lothar. Der Orkan 1999. Ereignisanalyse. Eidg. Forschungsanstalt WSL, Bundesamt für Umwelt, Wald und Landschaft BUWAL, Birmensdorf, Bern, 265 pp.